file
stringlengths
27
28
text
stringlengths
40
2.12M
PMC007xxxxxx/PMC7614773.txt
==== Front 9215515 Neuroimage Neuroimage NeuroImage 1053-8119 1095-9572 36108798 ems177461 10.1016/j.neuroimage.2022.119624 Article Psychedelics and schizophrenia: Distinct alterations to Bayesian inference Rajpal Hardik abf* Mediano Pedro A.M. [email protected] cde Rosas Fernando E. aghp Timmermann Christopher B. g Brugger Stefan k Muthukumaraswamy Suresh m Seth Anil K. no Bor Daniel de Carhart-Harris Robin L. gi Jensen Henrik J. abj a Centre for Complexity Science, Imperial College London, South Kensington, London, United Kingdom b Department of Mathematics, Imperial College London, South Kensington, London, United Kingdom c Department of Computing, Imperial College London, South Kensington, London, United Kingdom d Department of Psychology, University of Cambridge, Cambridge, United Kingdom e Department of Psychology, Queen Mary University of London, London, United Kingdom f Public Policy Program, The Alan Turing Institute, London, United Kingdom g Centre for Psychedelic Research, Department of Brain Sciences, Imperial College London, London, United Kingdom h Data Science Institute, Imperial College London, London, United Kingdom i Psychedelics Division, Neuroscape, Department of Neurology, University of California San Francisco, US j Institute of Innovative Research, Tokyo Institute of Technology, Yokohama, Japan k Cardiff University Brain Research Imaging Centre, School of Psychology, Cardiff University, United Kingdom l Centre for Academic Mental Health, Bristol Medical School, University of Bristol, United Kingdom m School of Pharmacy, The University of Auckland, New Zealand n School of Engineering and Informatics, University of Sussex, United Kingdom o CIFAR Program on Brain, Mind, and Consciousness, Toronto, Canada p Department of Informatics, University of Sussex, Brighton, United Kingdom * Corresponding author. [email protected] (H. Rajpal). 01 11 2022 13 9 2022 04 7 2023 18 7 2023 263 119624119624 https://creativecommons.org/licenses/by/4.0/ This work is licensed under a CC BY 4.0 International license. https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Schizophrenia and states induced by certain psychotomimetic drugs may share some physiological and phe-nomenological properties, but they differ in fundamental ways: one is a crippling chronic mental disease, while the others are temporary, pharmacologically-induced states presently being explored as treatments for mental illnesses. Building towards a deeper understanding of these different alterations of normal consciousness, here we compare the changes in neural dynamics induced by LSD and ketamine (in healthy volunteers) against those associated with schizophrenia, as observed in resting-state M/EEG recordings. While both conditions exhibit in-creased neural signal diversity, our findings reveal that this is accompanied by an increased transfer entropy from the front to the back of the brain in schizophrenia, versus an overall reduction under the two drugs. Furthermore, we show that these effects can be reproduced via different alterations of standard Bayesian inference applied on a computational model based on the predictive processing framework. In particular, the effects observed under the drugs are modelled as a reduction of the precision of the priors, while the effects of schizophrenia correspond to an increased precision of sensory information. These findings shed new light on the similarities and differences between schizophrenia and two psychotomimetic drug states, and have potential implications for the study of consciousness and future mental health treatments. Psychedelics Schizophrenia Information theory Predictive processing ==== Body pmc1 Introduction Classic serotonergic psychedelic drugs have seen a blooming resurgence among the public and the scientific community in recent years, largely driven by promising clinical research into their therapeutic potential Carhart-Harris et al. (2021, 2017). At the same time, and somewhat paradoxically, psychedelics are known to elicit effects that mimic some symptoms of psychosis – earning them the label of ‘psychotomimetic drugs’ Carhart-Harris et al. (2016a). In this context, our aims with this study are twofold: First, to explore the limits of psychotomimetic models of psychosis at a neurophysiological level, thus helping us refine these models. Second, to further our understanding of extended and acute alterations to normal consciousness, which may help the design better mental health therapies. To contrast these conditions in an empirical manner, we compare neuroimaging data from patients suffering from schizophrenia and healthy subjects under the effects of two psychoactive substances: the classical psychedelic lysergic acid diethylamide (LSD) Carhart-Harris et al. (2016b) and the dissociative drug ketamine (KET) Frohlich and Van Horn (2014). Using standardised assessments, it has been claimed that KET reproduces both positive and negative symptoms of schizophrenia in humans Beck et al. (2020), and its mechanism of action – NMDA receptor antagonism – is thought to reproduce a key element of the molecular pathophysiology of schizophrenia Friston et al. (2016); McCutcheon et al. (2020) LSD – in common with all classical psychedelics – is a potent agonist of a number of serotonin receptors, but its characteristic effects depend primarily on 5-HT2A Nichols (2004). These neurotransmitter systems have been linked to symptoms of early acute schizophrenic stages, such as “ego-disorders, affective changes, loosened associations and perceptual alterations” Vollenweider et al. (1998) (see Ref. Carhart-Harris et al. (2013) for a quantitative analysis of these associations). Both psychotomimetic drug states and schizophrenia are also associated with marked changes in large-scale neural dynamics. For both LSD and KET, previous studies have found increased signal diversity in subjects’ neural dynamics Mediano et al. (2020); Schartner et al. (2017) and reduced information transfer between brain regions Barnett et al. (2020). However, in the case of KET, evidence from intracranial recordings in cats suggests a much more complicated picture than that of LSD, with very high variability across individuals, brain regions, and dose levels Pascovich et al. (2021). In a separate line of enquiry, work on EEG data from patients with schizophrenia has also found increased signal diversity Fernández et al. (2011); Li et al. (2008), akin to the effect found under these drugs. Nonetheless, a parsimonious account explaining the similarities and differences between the two states is still lacking. A promising approach to gain insights into the mechanisms driving the core similarities and differences between psychotomimetic drug states and schizophrenia is to leverage principles from the predictive processing (PP) framework of brain function Clark (2015); Rao and Ballard (1999). A key postulate of the PP framework is that the dynamics of neural populations can be viewed as engaged in processes of inference involving top-down and bottom-up signals. Under this framework, brain activity can be viewed as resulting from a continuous modelling process in which a prior distribution interacts with new observations via incoming sensory information. In accordance with principles of Bayesian inference, discrepancies between the prior distribution and incoming signals (called ‘prediction errors’) carried by the bottom-up signals drive revisions to the top-down activity, so as to minimize future surprise. The PP framework has been used to explain perceptual alterations observed in both psychotomimetic drug states Corlett et al. (2009); Leptourgos et al. (2020) as well as in psychiatric illnesses Adams et al. (2016) with a focus on schizophrenia Adams et al. (2013); Brugger and Broome (2018); Fletcher and Frith (2009); Speechley et al. (2010). Most of these accounts of PP are task-based studies, which manipulate stimuli in order to modulate prediction errors. In contrast, here we extend this approach to the resting state, focusing on spontaneous “prediction errors” that arise from naturally occurring neural activity. PP has also been used to understand the action of psychedelics, most notably through the “relaxed beliefs under psychedelics” (or REBUS) model Carhart-Harris and Friston (2019) which posits that psychedelics reduce the precision of prior beliefs encoded in spontaneous brain’s activity. REBUS has also been used to inform thinking on the therapeutic mechanisms of psychedelics, where symptomatology can be viewed as pathologically over-weighted beliefs or assumptions encoded in the precision weighting of brain activity encoding them. To deepen our understanding of the similarities and differences between these conditions, in this paper we replicate and extend findings on neural diversity and information transfer under the two psychotomimetic drugs (LSD and KET) and in schizophrenia using EEG and MEG recordings, and we reproduce these experimental findings as perturbations to a single PP model. Our modelling results reveal that the effects observed under the drugs are indeed reproduced by decreasing the precision-weighting of the priors, while the effects observed under schizophrenia are reproduced by increased precision-weighting of the bottom-up sensory information. Overall, this study puts forward a more nuanced understanding of the relationship between two different psychotomimetic drug states and schizophrenia, and offers a new model-based perspective on how these conditions alter conscious experience. 2 Materials and methods 2.1 Data acquisition and preprocessing Data from 29 patients diagnosed with schizophrenia and 38 age-matched healthy control subjects were obtained from the Bipolar-Schizophrenia Network on Intermediate Phenotypes (BSNIP) database Tamminga et al. (2013). The subjects were selected within an age range of 20–40 years to match the psychedelic datasets described below. Data included 64-channel EEG recordings sampled at 1000Hz of each subject in eyes-closed resting state, along with metadata about demographics (age and gender), patients’ medications and their PANSS symptom scores Kay et al. (1987). The strength of the medication was estimated using the number of antipsychotics taken by each patient (mean: 2.7, range: 0–8), as the dosage of each medication was not available. Data from healthy subjects under the effects of both drugs was obtained from previous studies with LSD Carhart-Harris et al. (2016b) (N = 17) and ketamine Muthukumaraswamy et al. (2015) (N = 19). Data included MEG recordings from a CTF 275-channel axial gradiometer system with a sampling frequency of 600Hz. Each subject underwent two scanning sessions in eyes-closed resting state: one after drug administration and another after a placebo (PLA). Preprocessing steps for all datasets were kept as consistent as possible, and were performed using the Fieldtrip Oostenveld et al. (2011) and EEGLAB Delorme and Makeig (2004) libraries. First, the data was segmented into epochs of 2 seconds, and epochs with strong artefacts were removed via visual inspection. Next, muscle and eye movement artefacts were removed using ICA Winkler et al. (2011). Then, a LCMV beam-former Van Veen et al. (1997) was used to reconstruct activity of sources located at the centroids of regions in the Automated Anatomical Labelling (AAL) brain atlas Tzourio-Mazoyer et al. (2002). Finally, source-level data was bandpass-filtered between 1–100Hz, and downsampled with phase correction to 250Hz (EEG) and 300Hz (MEG), and AAL areas were grouped into 5 major Regions of Interest (ROIs): frontal, parietal, occipital, temporal and sensorimotor (see Fig. 1 and Table D.2 in the Appendix). In the rest of the paper we refer to these 5 areas as “ROIs” and to the AAL regions as “sources.” 2.2 Analysis metrics Our analyses are focused on two complementary metrics of neural activity: Lempel-Ziv complexity (LZ) and transfer entropy (TE). Both metrics are based on the same mathematical framework of information theory, and provide characterisations of different but complementary aspects of neural dynamics: LZ captures aspects of the temporal dynamics of single regions, while TE quantifies how different regions influence each other. Both metrics have a long history, and have been used and robustly validated across a wide range of states of consciousness, including psychedelic states Barnett et al. (2020); Bossomaier et al. (2016); Mediano et al. (2020); Schartner et al. (2017). Lempel-Ziv complexity (LZ) is a measure of the diversity of patterns observed in a discrete – typically binary – sequence. When applied to neuroimaging data, lower LZ (with respect to wakeful rest) has been associated with unconscious states such as sleep Andrillon et al. (2016) or anaesthesia Zhang et al. (2001), and higher LZ with states of richer phenomenal content under psychedelics, ketamine Mediano et al. (2020);Schartner et al. (2017) and states of flow during musical improvisation Dolan et al. (2018). To calculate LZ, first one needs to transform a given signal of length T into a binary sequence. For a given epoch of univariate M/EEG data, we do this by calculating the mean value and transforming each data point above the mean to 1 and each point below to 0. Then, the resulting binary sequence is scanned sequentially using the LZ76 algorithm presented by Kaspar and Schuster Kaspar and Schuster (1987), which counts the number of distinct “patterns” in the signal. Finally, following results by Ziv Ziv (1978), the number of patterns is divided by log2(T)/T to yield an estimate of the signal’s entropy rate Cover and Thomas (2006), which we refer to generically as LZ. This process is applied separately to each source time series (i.e. to each AAL region), and the resulting values are averaged according to the grouping in Table D.2 to yield an average LZ value per ROI. In addition to LZ, our analyses also consider transfer entropy (TE) Bossomaier et al. (2016) — an information-theoretic version of Granger causality Barnett et al. (2009) — to assess the dynamical in-terdependencies between ROIs. The TE from a source region to a target region quantifies how much better one can predict the activity of the target after the activity of the source is known. This provides a notion of directed functional connectivity, which can be used to analyse the structure of large-scale brain activity Barnett et al. (2020); Deco et al. (2021). Mathematically, TE is defined as follows. Denote the activity of two given ROIs at time t by the vectors Xt and Yt, and the activity of the rest of the brain by Zt. Note that Xt, Yt, and Zt have one component for each AAL source in the corresponding ROI(s). TE is computed in terms of Shannon’s mutual information, I, as the information about the future state of the target, Yt+1, provided by Xt over and above the information in Yt and Zt : (1) TEY→X|Z=I(Xt;Yt−1−|Xt−1−,Zt−1−), where Xt− refers to the (possibly infinite) past of Xt, up to and including time t (and analogously for Yt and Zt). This quantity can be accurately estimated using state-space models with Gaussian innovations Barnett and Seth (2015) implemented using the MVGC tool-box Barnett and Seth (2014) Note that, when calculating the TE between ROIs, we consider each ROI as a vector — without averaging the multiple AAL sources into a single number. The result is a directed 5×5 network of conditional TE values between pairs of ROIs, which can be tested for statistical differences across groups. 2.3 Statistical analysis For both LSD and KET datasets, since the same subjects were monitored under both drug and placebo conditions, average subject-level differences (either in LZ or TE) were calculated for each subject, and one-sample t-tests were used on those differences to estimate the effect of the drug. For the data of patients and controls in the schizophrenia dataset, group-level differences were estimated via linear models. These models used either LZ or TE as target variable, and condition (schizophrenia or healthy), age, gender, and number of antipsychotics (set to zero for healthy controls) as predictors. Motivated by previous work suggesting a quadratic relationship between complexity and age Gauvrit et al. (2017), each model was built with either a linear or quadratic dependence on age, and the quadratic model was selected if it was preferred over a linear model by a log-likelihood ratio test (with a critical level of 0.05). Finally, multiple comparisons when comparing TE values across all pairs of ROIs were addressed by using the Network-Based Statistic (NBS) Zalesky et al. (2010) method, which identifies ‘clusters’ of differences – i.e. connected components where a particular null hypothesis is consistently rejected while controlling for family-wise error rate. Our analysis used an in-house adapted version of NBS that works on directed networks, such as the ones provided by TE analyses. 2.4 Computational modelling A computational model was developed in order to interpret the LZ and TE findings observed on the neuroimaging data. Building on predictive processing principles Rao and Ballard (1999), we constructed a Bayesian state-space model that provides an idealised common ground to contrast the three studied conditions – the psy-chotomimetic drug states, schizophrenia, and baseline (i.e. healthy controls). Our modelling is based on the postulate that the activity of neuronal populations across the brain can be interpreted as carrying out inference on the causes of their afferent signals. Following this view, the proposed model considers the following elements:the internal state of a low-level region (i.e. near the sensory periphery), denoted by st ; the internal state of neural activity taking place functionally one level above, denoted by ht ; the signal generated at the high-level region in the form of a prediction of the low-level activity, denoted by s^t; the signal generated at the low-level region in the form of a prediction error ξt ; and the precision of the prior λp and precision of sensory/afferent information λs. This model represents neural activity within a larger hierarchical processing structure, as illustrated in Fig. 2. The key principle motivating this model is that minimisation of prediction error signals throughout the hierarchy, by updating top-down predictions, implements a tractable approximation to Bayesian inference.2 Within this model, we represent the schizophrenia and psychedelic conditions as different types of disruption to Bayesian inference. To describe the psychedelic state, we build on the REBUS hypothesis Carhart-Harris and Friston (2019), which posits a reduced precision-weighting of prior beliefs, leading to increased bottom-up influence. Conversely, to describe schizophrenia we build on the canonical predictive processing account of psychosis in schizophrenia Sterzer et al. (2018), which postulates an increased precision of sensory input, along with decreased precision of prior beliefs Adams et al. (2016); Fletcher and Frith (2009). Therefore, both conditions are similar in that there is a relative strengthening of bottom-up influence, although instantiated in different ways – which, as shown in Section 3.3, bears important consequences for the behaviour of the model. It is important to note that predictive processing accounts of schizophrenia remain hotly debated, with other works proposing an increase of prior precision (instead of decrease) as a model of auditory and visual hallucinations Corlett et al. (2019); Teufel et al. (2015). Recent reviews Sterzer et al. (2018) have attempted to reconcile both views by suggesting that sensory hallucinations may be caused by stronger priors, while hallucinations related to self-generated phenomena (like inner speech or self-attention Schneider et al. (2008)) may stem from weaker priors. Here, we base our modelling of SCZ on the weak prior hypothesis, as described above – we return to this issue in this discussion. To simulate the aberrant dynamics of the inference process, as described above, we consider a given afferent signal (st) and construct the corresponding activity of a higher area (ht), prediction (s^t), and prediction error (λt), building on the rich literature of state-space models in neuroscience Dayan and Jyu (2003); Dayan et al. (2000); Ratcliff and Rouder (1998). Specifically, we use the linear stochastic process: (2a) ht=aht−1+εt (2b) st=bht−1+vt where a, b are weights, and εt, vt are zero-mean Gaussian terms with precision (i.e. inverse variance) λp and λs, respectively. Note that this formulation is equivalent to (3a) ht|ht−1~N(aht−1,λp−1) (3b) st|ht~N(bht,λs−1). As we show in the following, λp corresponds to the precision of the prior and λs to the precision of sensory/afferent information. The dynamics of this system can be described as a recurrent update between predictions and prediction errors as follows. Eq. (3b) implies that the internal state ht, generates a prediction about the low-level activity given by s^t=bht. At the same time, the dynamics of the high-level region can be seen as a Bayesian update of ht, given st and ht+1. Under some simplifying assumptions, the mean of the posterior distribution of ht+1 (denoted by h^t+1) is equal to (see Appendix A) (4) h^t+1=ah^t+βξt which effectively combines a prior ah^t (which is the optimal prediction of ht+1 given only h^t, as seen from Eq. (3a)) and a likelihood given by the prediction error ξt=st−bh^t that is precision-weighted via ß, a parameter known as the Kalman gain Durbin and Koopman (2012). In our simulations, the model is first calibrated using as afferent signals (i.e. st) data from the primary visual cortex, corresponding to epochs randomly sampled from the placebo conditions in the LSD and KET datasets. This calibration results in the estimation of the model parameters acon, bcon, λpcon,λscon for the control condition, which is done us- p s ing the well-known expectation-maximisation algorithm Moon (1996). With these, the schizophrenia condition is then modelled by setting (5) λpscz=λpconandλsscz=ηλscon, where η > 1 is referred to as a noise factor, This increase of λs induces a strengthening of bottom-up prediction errors, and makes the posterior of ht, excessively precise. Conversely, the drug condition is modelled by setting (6) λppsy=λpηconandλspsy=λscon, Reducing λp also increases the influence of prediction errors, but reduces the precision of the posterior of ht,. Subsequently, for both conditions the parameters a, b are retrained with another pass of the expectation-maximisation algorithm on the placebo trials. Finally, to compare the model with the empirical M/EEG data, the LZ of the neural activity elicited in the low-level area (i.e. the prediction errors, ξt) and the top-down transfer entropy (from the high-level activity s^t towards the low-level activity ξt) are calculated for each of these three models (control, schizophrenia, and drug). 3 Results 3.1 LSD, KET and schizophrenia all show increased LZ We begin the analysis by comparing changes in signal diversity, as measured by LZ, across the LSD, ketamine (KET), and schizophrenia (SCZ) datasets. Our results show strong and significant increases in LZ in all three datasets (Fig. 3), in line with previous work Fernández et al. (2011); Li et al. (2008); Mediano et al. (2020); Schartner et al. (2017). In all three cases the LZ increases are widespread throughout the brain, with the effects in schizophrenia patients being more pronounced in frontal and parietal regions. While the t-scores are higher in LSD and KET than schizophrenia, this could be due to the within-subjects design of both drug experiments – which are more statistically powerful than the between-subjects analysis used on the schizophrenia dataset. Interestingly, we found that controlling for the medication status of each schizophrenia patient was crucial to obtain results that match prior work Fernández et al. (2011). A direct comparison of LZ values between patients and controls yielded no significant differences (t = −0.38, p = .70); however, when using a linear model correcting for age, gender, and number of antipsychotics, the antipsychotics coefficient of the model reveals a negative effect on LZ (ß = −0. 016, t = −2. 3, p = .021). Additionally, a two-sample t-test calculated between the corrected LZ values of patients and controls yields a substantial difference (t = 3.4, p = .001). Nonetheless, the sensitivity of this result to these preprocessing steps, as well as the lack of detailed dosage data for each medication, mean it should be considered preliminary and could only be properly interpreted after further investigation in future research (see the corresponding discussion in Section 4.3). 3.2 Opposite effect of psychotomimetic drugs and schizophrenia on information transfer We next report the effects of LSD, KET, and schizophrenia on large-scale information flow in the brain, as measured via transfer entropy (TE). The TE between each pair of ROIs (conditioned on all other ROIs) is calculated for each subject, and used to build directed TE networks. The resulting networks were tested for differences between the drug states and placebo conditions (for LSD and KET), and between patients and controls (for SCZ), correcting for multiple comparisons via cluster permutation testing (see Section 2.3). We found a ubiquitous decrease in the TE between most pairs of ROIs under LSD and KET (Fig. 4), which is consistent with previous findings Barnett et al. (2020). In contrast, SCZ patients exhibit marked localised increases in TE – and no decreases – with respect to the control subjects. Notably, most increases in TE originated in the frontal ROI, and are strongest between the frontal and occipital ROIs. The increase of information transfer seen in schizophrenia patients therefore takes place “front to back” – aligned with the pathways thought to carry top-down information in the brain from highly cognitive, decision-making regions to unimodal regions closer to the sensory periphery. As was the case for LZ, controlling for antipsychotic use was key to revealing differences between the healthy controls and schizophrenia patients. In addition, we found a small negative correlation between antipsychotic use and TE between certain ROI pairs – but, unlike for LZ, this effect did not survive correction for multiple comparisons. Although we find significant increase in both LZ and TE between certain regions among the schizophrenia patients when compared to healthy controls, these are not correlated with the symptom scores within the schizophrenia cohort. (see the corresponding discussion in Section 4.3). 3.3 Computational model reproduces experimental results So far, we have seen that subjects under the effects of two different psychotomimetic drugs display increased signal diversity and reduced information flow in their neural dynamics. In comparison, schizophrenia patients display increased complexity but also increased information flow with respect to healthy controls. We now show how complementary perturbations to the precision terms of the predictive processing model introduced in Section 2.4 reproduce these findings. We compared the basline model against the drug and schizophrenia variants by systematically increasing the noise factor η, which results in reduced prior precision in the drug model, and increased sensory precision in the schizophrenia model. We then computed the corresponding LZ and TE based on the model-generated time series ξt, s^t as per Section 2.4 (Fig. 5). Results show that the proposed model successfully reproduced the experimental findings of both LZ and TE under the two different psychotomimetic drugs and schizophrenia (Fig. 5). Interestingly, the model also shows (Fig. 5b) that a relative strengthening of sensory information (via either increased sensory precision, or decreased prior precision) can trigger either an increase or a decrease (respectively) of top-down transfer entropy. This suggests that transfer entropy changes cannot be directly interpreted as revealing the changes in any underlying predictive processing mechanisms (see Discussion). Finally, as a control, we repeated the analysis on the model but exploring the variation of the precision terms in the two unexplored di-rections — either reducing λp or increasing is (see Section 2.4). Neither of these changes reproduced the experimental findings (Supp. Fig. B.6), which highlights the specificity of the modelling choices. 4 Discussion In this paper we have analysed MEG data from healthy subjects under the effects of the psychotomimetic drugs LSD and ketamine, as well as EEG data from a cohort of schizophrenia patients and healthy control subjects. We focused on signal diversity and information transfer, both widely utilised metrics which provide a complementary account of neural dynamics. We found that all datasets show increases in signal diversity, but diverging changes in information transfer, which was higher in schizophrenia patients but lower for subjects under the effects of either drug. In addition to replicating previous results reporting signal diversity and information transfer under the effects of both drugs Barnett et al. (2020); Mediano et al. (2020); Schartner et al. (2017), we described new findings applying these metrics to schizophrenia. Using a computational model inspired by predictive processing principles Keller and Mrsic-Flogel (2018) Rao and Ballard (1999), we showed that this combination of effects can be reproduced via specific alterations to prediction updating, which can be interpreted as specific forms of disruption to Bayesian inference. Critically, the effects of both psychotomimetic drugs and schizophrenia, on both signal diversity and information transfer, are explained by a relative strengthening of sensory information over prior beliefs, although triggered by different mechanisms – a decrease in the precision of priors in the case of psychotomimetic drugs (consistent with Ref. Carhart-Harris and Friston (2019)), and an increase in the precision of sensory information for schizophrenia. 4.1 Increased sensory precision in schizophrenia The idea that the symptoms of schizophrenia can be understood as alterations to processes of Bayesian inference has been particularly fertile in the field of computational psychiatry Adams et al. (2016). In particular, various studies based on PP have related psychosis to decreased precision of prior beliefs and increased precision of the sensory inputs Corlett et al. (2009); Fletcher and Frith (2009); Friston et al. (2014); Notredame et al. (2014); Sterzer et al. (2016). These computational models have been supported by a growing number of related experimental findings, including an enhanced confirmation bias Balzan et al. (2013), impaired reversal learning Leeson et al. (2009); Waltz and Gold (2007), and a greater resistance to visual illusions Silverstein and Keane (2011). For instance, schizophrenia patients are less susceptible to the Ebbinghaus illusion, which arises primarily from misleading prior expectations, suggesting that patients do not integrate this prior context with sensory evidence and thus achieve more accurate judgements Horton and Silverstein (2011). Most of the above mentioned studies are task-based, focusing on differentiating perceptual learning behaviours between healthy controls and schizophrenia patients. Though these studies provide a range of experimental markers, the corresponding methodologies cannot be applied to resting-state or task-free conditions, under which it is known that certain behavioural alterations (e.g. delusions, anhedonia, and paranoia) persist Northoff and Duncan (2016); Northoff and Qin (2011). The findings presented in this paper provide a step towards bridging this important knowledge gap by providing empirical and theoretical insights into resting-state neural activity under schizophrenia. Although we build on and replicate results related to signal diversity, we are not aware of previous studies of information transfer on schizophrenia in resting state. 4.2 Beyond unidimensional accounts of top-down vs bottom-up processing The findings presented here link spontaneous brain activity to the PP framework using empirical metrics of signal diversity and information transfer. In the psychotomimetic drug condition, the former increases while the latter decreases; in schizophrenia, both increase – in both cases as compared to baseline placebo or control. The explanation for this pattern of results, articulated by our computational model, is based on the idea that a bias favouring bottom-up over top-down processing can be triggered by changing different precision parameters, which can give rise to opposite effects in specific aspects of the neural dynamics. This observation, we argue, opens the door to more nuanced analyses for future studies. The increased transfer entropy from frontal to posterior brain areas observed under schizophrenia could be naively interpreted as supporting increased top-down regulation; however, neither the empirical analysis nor the computational model warrant this conclusion. Transfer entropy simply indicates information flow and is agnostic about functional role. Our model-based analyses illustrate how aberrant Bayesian inference in which bottom-up influences become stronger can trigger either an increase or a decrease in transfer entropy from frontal to posterior regions, depending on which precision terms are involved. An interesting possible explanation for this divergence between mechanisms and TE is provided by recent results that show that TE is an aggregate of qualitatively different information modes Mediano et al. (2021b). Future work may explore if resolving TE into its finer constituents might provide a more informative mapping from observed patterns to underlying mechanisms, as well as how these quantities may be related to other consciousness-related electrophysiology metrics Nilsen et al. (2020) ; Sitt et al. (2014). Taken together, these findings suggest that conceiving the bottom-up vs top-down dichotomy as a single-dimensional trade-off might be too simplistic, and that multi-dimensional approaches could shed more light on this issue. In particular, our results show how such a simplistic view fails to account for the rich interplay of similarities and differences between schizophrenia and psychosis. 4.3 Limitations and future work While our empirical and modelling results agree with the canonical PP account of psychosis Sterzer et al. (2018), some reports have suggested a stronger influence of priors over sensory signals – especially in some cases of hallucinations Alderson-Day et al. (2017); Cassidy et al. (2018); Powers et al. (2017). It is important to remark that the ‘strengthened prior’ interpretation put forward by these task-based studies cannot be accounted for by the simple computational modelling developed here. At the same time, the resting-state model presented here relates spontaneous activity, and our results cannot be directly generalised to task-based settings. Future work may investigate whether a richer hierarchical model is able to reproduce both rest and task data, bridging between these results and prior work. Regarding the empirical analyses, it is important to note that our analyses are subject to a few limitations due to the nature of the data used. First, the analyses used only 60 AAL sources across 5 ROIs (due to the spatial resolution limitations of EEG), and therefore may neglect potential PP effects that may exist at smaller spatial scales. In addition, the studies on both drugs and schizophrenia used different imaging methods (MEG vs EEG), sampling rate, and experiment designs (within vs between subjects), complicating direct comparisons. Finally, future work should examine how power spectra across the different conditions relate to the findings presented, in terms of both their effect on LZ Mediano et al. (2021c), and their relationship with top-down and bottom-up signalling, for example using band-limited Granger causality Bastos et al. (2015), as well as how directed functional connectivity measures relate to undirected measures such as mutual information and coherence Barnett et al. (2020). Similarly, while the measures discussed here capture significant differences between schizophrenia patients and healthy controls, more work needs to be done to further characterise the differences within the schizophrenia spectrum, which features a heterogenous array of symptoms and states, e.g. at different phases of the so-called ‘psychotic process’ Brouwer and Carhart-Harris (2021). A crucial part of this research will be to analyse the clinical symptom scores of the patients and their relationship to both medication and neural dynamics, which was not possible here due to the lack of appropriate metadata on the dosage of antipsychotics. In our preliminary analysis we use the number of antipsychotics as a proxy to the missing dosage data. This proxy measure was found to be negatively correlated with the positive symptom scores of the PANSS scale (see Appendix C) among the schizophrenia patients, suggesting that the symptom scores are confounded by antipsychotic use - but without dosage data it is difficult to disentangle this effect from potential confounds. An interesting possibility is that the neural underpinnings of positive and negative symptoms could be different Fletcher and Frith (2009), and investigating these differences may yield further insight into schizophrenia itself and its relationship with the psychotomimetic drug states. Moreover, both schizophrenia and drug-induced states can be conceived of as dynamic states of consciousness, comprised of several sub-states and/or episodes with hallucinations, delusions and negative symptoms varying widely between and within individuals. Future studies could explore these finer fluctuations in conscious state Mediano et al. (2021a), as well as what features or episodes overlap in the neural and psychological levels between psychotomimetic drug states and schizophrenia. Finally, recall that (as described in Section 2.1) we used the number of antipsychotic medications being used by each patient as a proxy measure for their medication load. This is a significant oversimplification, as it ignores the specifics of all drugs and their dose-response effects, and future work with richer datasets should explore in more detail the effects of each particular medication – which would potentially bring more nuance to these analyses. Also, the models used for statistical analysis (as per Section 2.3) are linear and may not capture possible non-linear dependencies between antipsychotic use and its effect on neural dynamics (in our case, LZ or TE). Bearing this caveat in mind, our tentative results in the schizophrenia group suggest that antipsychotic use may bring the patients’ neural dynamics closer to the range of healthy controls. This finding should be replicated with more detailed analyses involving dosage information and clinical symptom scores, and, if robust, could potentially be used to investigate the mechanism of action of current antipsychotic drugs. 4.4 Final remarks In this paper we have contrasted changes in brain activity in individuals with schizophrenia (compared to healthy controls) with changes induced by a classic 5-HT2A receptor agonist psychedelic, LSD, and an NMDA antagonist dissociative, ketamine (compared to placebo). Empirical analyses revealed that both schizophrenia and drug states show an increase in neural signal diversity, but they have divergent transfer entropy profiles. Furthermore, we proposed a simple computational model based on the predictive processing framework Rao and Ballard (1999) that recapitulates the empirical findings through distinct alterations to optimal Bayesian inference. In doing so, we argued that both schizophrenia and psychotomimetic drugs can be described as inducing a stronger “bottom-up” influence of sensory information, but in qualitatively different ways, thus painting a more nuanced picture of the functional dynamics of predictive processing systems. Crucially, the proposed model differs from others in the literature in that it is a model of resting-state (as opposed to task-based) brain activity, bringing this methodology closer to other approaches to neuroimaging data analysis based on complexity science Turkheimer et al. (2021). Overall, this study illustrates the benefits of combining information-theoretic analyses of experimental data and computational modelling, as well as of integrating datasets from patients with those from healthy subjects. We hope our findings will inspire further work deepening our understanding about the relationship between neural dynamics and high- level brain functions, which in turn may accelerate the development of novel, mechanism-based treatments to foster and promote mental health. Supplementary Material Supplementary Material Acknowledgements H.R. is supported by the Imperial College President’s PhD Scholarship. P.M. and D.B. are funded by the Wellcome Trust (grant no. 210920/Z/18/Z). F.R. is supported by the Ad Astra Chandaria foundation. C.T. is funded by the Psychedelic Research Group, Imperial College London. R.C.-H. is the Ralph Metzner Chair of the Psychedelic Division, Neuroscape at University of California San Francisco. A.K.S. is supported by the European Research Council (Advanced Investigator Grant 101019254.) This work was supported in part by grant MR/N0137941/1 for the GW4 BIOMED MRC DTP, awarded to the Universities of Bath, Bristol, Cardiff and Exeter from the Medical Research Council (MRC)/UKRI (S.B.). The LSD research described here was supported by the Beckley Foundation and Walacea Crowd Funding campaign. We thank Imperial College London’s Research Computing Service for the computing facilities. Parts of this work were carried out using the computational facilities of the Advanced Computing Research Centre, University of Bristol. Data and code availability statement Raw MEG data for the LSD and KET conditions (and their respective placebo controls) is available in the Harvard Dataverse repository. EEG data from schizophrenia patients and healthy controls was obtained from the BSNIP study, accessed via the NIMH Data Archive. Open-source implementations are available online for all tools used in the study, including LZ (link), TE (link), and NBS (link). Fig. 1 Regions of interest (ROIs) represented on the MNI-152 standard template. Each ROI is comprised of several regions of the AAL atlas, as per Table D.2. Fig. 2 Graphical illustration of the predictive processing model. The activity of a high-level neural population is represented as a prediction s^t and the activity of a low-level population as a prediction error ξt. The internal states of the high- and low-level regions are captured by st and ht, respectively. Fig. 3 Increased signal diversity in subjects under the effects of psychotomimetic drugs and in schizophrenia patients. LZ changes are widespread across all ROIs in the three datasets. For the schizophrenia dataset, LZ values shown are corrected for age, gender, and the number of antipsychotic medications taken by each patient using a linear model. Fig. 4 Lower information transfer under LSD and ketamine but higher information transfer for schizophrenia patients. Transfer entropy (TE) shows a strong widespread decrease in subjects under the effect of LSD or ketamine (KET), compared to a placebo. Conversely, schizophrenia (SCZ) patients show an increase in TE with respect to controls (CTRL), especially from the frontal region to the rest of the brain (controlling for age, gender and antipsychotic use). Links shown are significant after multiple comparisons correction. Fig. 5 A computational model based on predictive processing principles reproduces experimental findings in the LSD, ketamine and schizophrenia datasets (a) By increasing the sensory precision (for schizophrenia; blue), or reducing the prior’s precision (for LSD and KET; orange) by a given ‘noise’ factor η, the model can reproduce the experimental findings of (b) increased in LZ in both conditions, and (c) opposite changes in TE in both conditions, compared to a baseline (grey). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) 2 Note that in the simple scenario of Eqs. (3), with a single level and Gaussian distributions, inference is in fact exact. In larger or more complicated models inference is often carried out only approximately Gershman (2019). Ethics statement The LSD study Carhart-Harris et al. (2016b) was approved by the National Research Ethics Service committee London-West London and was conducted in accordance with the revised declaration of Helsinki (2000), the International Committee on Harmonization Good Clinical Practice guidelines, and National Health Service Research Governance Framework. Imperial College London sponsored the research, which was conducted under a Home Office license for research with Schedule 1 drugs. The KET study Muthukumaraswamy et al. (2015) was was approved by a UK National Health Service research ethics committee. The schizophrenia data was collected by the B-SNIP consortium Tamminga et al. (2013). Credit authorship contribution statement Hardik Rajpal: Conceptualization, Methodology, Software, Formal analysis, Writing – original draft, Writing – review & editing. Pedro A.M. Mediano: Conceptualization, Methodology, Software, Formal analysis, Writing – original draft, Writing – review & editing. Fernando E. Rosas: Conceptualization, Writing – original draft, Writing – review & editing. Christopher B. Timmermann: Writing – review & editing. Stefan Brugger: Data curation, Writing – review & editing. Suresh Muthukumaraswamy: Data curation, Writing – review & editing. Anil K. Seth: Writing – review & editing. Daniel Bor: Writing – review & editing. Robin L. Carhart-Harris: Data curation, Writing – review & editing. Henrik J. Jensen: Supervision, Writing – review & editing. ==== Refs Adams RA Huys QJ Roiser JP Computational psychiatry: towards a mathematically informed understanding of mental illness J Neurol Neurosurg Psychiatry 2016 87 1 53 63 26157034 Adams RA Stephan KE Brown HR Frith CD Friston KJ The computational anatomy of psychosis Front Psychiatry 2013 4 47 23750138 Alderson-Day B Lima CF Evans S Krishnan S Shanmugalingam P Fernyhough C Scott SK Distinct processing of ambiguous speech in people with non-clinical auditory verbal hallucinations Brain 2017 140 9 2475 2489 29050393 Andrillon T Poulsen AT Hansen LK Léger D Kouider S Neural markers of responsiveness to the environment in human sleep J Neurosci 2016 36 24 6583 6596 10.1523/JNEUROSCI.0902-16.2016 27307244 Balzan R Delfabbro P Galletly C Woodward T Confirmation biases across the psychosis continuum: the contribution of hypersalient evidence-hypothesis matches British Journal of Clinical Psychology 2013 52 1 53 69 23398112 Barnett L Barrett AB Seth AK Granger causality and transfer entropy are equivalent for gaussian variables Phys Rev Lett 2009 103 23 238701 10.1103/physrevlett.103.238701 20366183 Barnett L Muthukumaraswamy SD Carhart-Harris RL Seth AK Decreased directed functional connectivity in the psychedelic state NeuroImage 2020 209 116462 31857204 Barnett L Seth AK The MVGC multivariate granger causality toolbox: a new approach to granger-causal inference J Neurosci Methods 2014 223 50 68 10.1016/j.jneumeth.2013.10.018 24200508 Barnett L Seth AK Granger causality for state-space models Phys Rev E 2015 91 4 040101 10.1103/PhysRevE.91.040101 Bastos AM Vezoli J Bosman CA Schoffelen J-M Oostenveld R Dowdall JR De Weerd P Kennedy H Fries P Visual areas exert feedforward and feedback influences through distinct frequency channels Neuron 2015 85 2 390 401 10.1016/j.neuron.2014.12.018 25556836 Beck K Hindley G Borgan F Ginestet C McCutcheon R Brugger S Driesen N Ranganathan M D’Souza DC Taylor M Association of ketamine with psychiatric symptoms and implications for its therapeutic use and for understanding schizophrenia: a systematic review and meta-analysis JAMA Netw Open 2020 3 5 e204693 32437573 Bossomaier T Barnett L Harré M Lizier JT An introduction to transfer entropy Springer 2016 65 Brouwer A Carhart-Harris RL Pivotal mental states J Psychopharmacol 2021 35 4 319 352 10.1177/0269881120959637 33174492 Brugger S Broome M Computational Psychiatry The Routledge Handbook of the Computational Mind Routledge 2018 452 468 Carhart-Harris RL Brugger S Nutt D Stone J Psychiatry’S next top model: cause for a re-think on drug models of psychosis and other psychiatric disorders J Psychopharmacol 2013 27 9 771 778 10.1177/0269881113494107 23784738 Carhart-Harris RL Friston K REBUS And the anarchic brain: toward a unified model of the brain action of psychedelics Pharmacol Rev 2019 71 3 316 344 31221820 Carhart-Harris RL Giribaldi B Watts R Baker-Jones M Murphy-Beiner A Murphy R Martell J Blemings A Erritzoe D Nutt DJ Trial of psilocybin versus escitalopram for depression N top N Engl J Med 2021 384 15 1402 1411 Carhart-Harris RL Kaelen M Bolstridge M Williams T Williams L Underwood R Feilding A Nutt DJ The paradoxical psychological effects of lysergic acid diethylamide (LSD) Psychol Med 2016 46 7 1379 1390 10.1017/S0033291715002901 26847689 Carhart-Harris RL Muthukumaraswamy S Roseman L Kaelen M Droog W Murphy K Tagliazucchi E Schenberg EE Nest T Orban C Neural correlates of the LSD experience revealed by multimodal neuroimaging Proc Natl Acad Sci 2016 113 17 4853 4858 27071089 Carhart-Harris RL Roseman L Bolstridge M Demetriou L Pannekoek JN Wall MB Tanner M Kaelen M McGonigle J Murphy K Psilocybin for treatment-resistant depression: fmri-measured brain mechanisms Sci Rep 2017 7 1 1 11 28127051 Cassidy CM Balsam PD Weinstein JJ Rosengard RJ Slifstein M Daw ND Abi-Dargham A Horga G A perceptual inference mechanism for hallucinations linked to striatal dopamine Curr Biol 2018 28 4 503 514 29398218 Clark A Surfing uncertainty: Prediction, action, and the embodied mind Oxford University Press 2015 Corlett PR Frith CD Fletcher PC From drugs to deprivation: a bayesian framework for understanding models of psychosis Psychopharmacology (Berl) 2009 206 4 515 530 19475401 Corlett PR Horga G Fletcher PC Alderson-Day B Schmack K Powers AR III Hallucinations and strong priors Trends Cogn Sci (Regul Ed) 2019 23 2 114 127 Cover TM Thomas JA Elements of information theory Wiley Hoboken 2006 Dayan P Jyu A Uncertainty and learning IETE J Res 2003 49 2-3 171 181 Dayan P Kakade S Montague PR Learning and selective attention Nat Neu-rosci 2000 3 11 1218 1223 Deco G Vidaurre D Kringelbach ML Revisiting the global workspace orchestrating the hierarchical organization of the human brain Nat Hum Behav 2021 5 4 497 511 10.1038/s41562-020-01003-6 33398141 Delorme A Makeig S EEGLAB: An open source toolbox for analysis of single-trial EEG dynamics including independent component analysis J Neurosci Methods 2004 134 1 9 21 10.1016/j.jneumeth.2003.10.009 15102499 Dolan D Jensen HJ Mediano PA Molina-Solana M Rajpal H Rosas F Slo-boda JA The improvisational state of mind: amultidisciplinary study of an improvisatory approach to classical music repertoire performance Front Psychol 2018 9 1341 10.3389/fpsyg.2018.01341 30319469 Durbin J Koopman SJ Time series analysis by state space methods Oxford University Press 2012 Fernández A López-Ibor M-I Turrero A Santos J-M Morón M Hornero R-D Gómez C Méndez MA Ortiz T López-Ibor JJ Lempel-ziv complexity in schizophrenia: a MEG study Clin Neurophysiol 2011 122 11 2227 2235 21592856 Fletcher PC Frith CD Perceiving is believing: a bayesian approach to explaining the positive symptoms of schizophrenia Nat Rev Neurosci 2009 10 1 48 58 19050712 Friston K Brown HR Siemerkus J Stephan KE The dysconnection hypothesis (2016) Schizophr Res 2016 176 2-3 83 94 27450778 Friston KJ Stephan KE Montague R Dolan RJ Computational psychiatry: the brain as a phantastic organ Lancet Psychiatry 2014 1 2 148 158 26360579 Frohlich J Van Horn JD Reviewing the ketamine model for schizophrenia J Psychopharmacol 2014 28 4 287 302 24257811 Gauvrit N Zenil H Soler-Toscano F Delahaye J-P Brugger P Human behavioral complexity peaks at age 25 PLoS Comput Biol 2017 13 4 e1005408 10.1371/journal.pcbi.1005408 28406953 Gershman SJ What does the free energy principle tell us about the brain? Neurons Behav Data Anal Theory 2019 Horton HK Silverstein SM Visual context processing deficits in schizophrenia: effects of deafness and disorganization Schizophr Bull 2011 37 4 716 726 21700590 Kaspar F Schuster H Easily calculable measure for the complexity of spatiotemporal patterns Phys Rev A 1987 36 2 842 10.1103/PhysRevA.36.842 Kay SR Fiszbein A Opler LA The positive and negative syndrome scale (panss) for schizophrenia Schizophr Bull 1987 13 2 261 276 3616518 Keller GB Mrsic-Flogel TD Predictive processing: a canonical cortical computation Neuron 2018 100 2 424 435 30359606 Leeson VC Robbins TW Matheson E Hutton SB Ron MA Barnes TR Joyce EM Discrimination learning, reversal, and set-shifting in first-episode schizophrenia: stability over six years and specific associations with medication type and disorganization syndrome Biol Psychiatry 2009 66 6 586 593 19576575 Leptourgos P Fortier-Davy M Carhart-Harris R Corlett PR Dupuis D Halberstadt AL Kometer M Kozakova E Larøi F Noorani TN Preller KH Hallucinations under psychedelics and in the schizophrenia spectrum: an interdisciplinary and multiscale comparison Schizophr Bull 2020 46 6 1396 1408 10.1093/schbul/sbaa117 32944778 Li Y Tong S Liu D Gai Y Wang X Wang J Qiu Y Zhu Y Abnormal EEG complexity in patients with schizophrenia and depression Clin Neurophysiol 2008 119 6 1232 1241 18396454 McCutcheon RA Krystal JH Howes OD Dopamine and glutamate in schizophrenia: biology, symptoms and treatment World Psychiatry 2020 19 1 15 33 31922684 Mediano P Ikkala A Kievit RA Jagannathan SR Varley TF Stamatakis EA Bekinschtein TA Bor D Fluctuations in neural complexity during wakefulness relate to conscious level and cognition bioRxiv 2021 10.1101/2021.09.23.461002 Mediano PA Rosas FE Luppi AI Carhart-Harris RL Bor D Seth AK Barrett AB Towards an extended taxonomy of information dynamics via integrated information decomposition arXiv preprint 2021 arXiv:2109.13186 Mediano PA Rosas FE Timmermann C Roseman L Nutt DJ Feild-ing A Kaelen M Kringelbach ML Barrett AB Seth AK Effects of external stimulation on psychedelic state neurodynamics bioRxiv 2020 10.1101/2020.11.01.356071 Mediano PAM Rosas FE Barrett AB Bor D Decomposing spectral and phasic differences in nonlinear features between datasets Phys Rev Lett 2021 127 12 124101 10.1103/PhysRevLett.127.124101 34597101 Moon TK The expectation-maximization algorithm IEEE Signal Process Mag 1996 13 6 47 60 Muthukumaraswamy SD Shaw AD Jackson LE Hall J Moran R Saxena N Evidence that subanesthetic doses of ketamine cause sustained disruptions of NMDA and AMPA-mediated frontoparietal connectivity in humans J Neurosci 2015 35 33 11694 11706 26290246 Nichols DE Hallucinogens Pharmacol Therapeutic 2004 101 2 131 181 Nilsen AS Juel B Thürer B Storm JF Proposed EEG measures of consciousness: A systematic, comparative review 2020 Northoff G Duncan NW How do abnormalities in the brain’s spontaneous activity translate into symptoms in schizophrenia? from an overview of resting state activity findings to a proposed spatiotemporal psychopathology Prog Neurobiol 2016 145 26 45 27531135 Northoff G Qin P How can the brain’s resting state activity generate hallucinations? a ‘resting state hypothesis’ of auditory verbal hallucinations Schizophr Res 2011 127 1-3 202 214 21146961 Notredame C-E Pins D Deneve S Jardri R What visual illusions teach us about schizophrenia Front Integr Neurosci 2014 8 63 25161614 Oostenveld R Fries P Maris E Schoffelen J-M Fieldtrip: open source software for advanced analysis of MEG, EEG, and invasive electrophysiological data Comput Intell Neurosci 2011 2011 1:1 1:9 10.1155/2011/156869 21837235 Pascovich C Castro-Zaballa S Mediano PA Bor D Canales-Johnson A Torterolo PD Bekinschtein TA Ketamine and sleep modulate neural complexity dynamics in cats bioRxiv 2021 10.1101/2021.06.25.449513 Powers AR Mathys C Corlett PR Pavlovian conditioning-induced hallucinations result from overweighting of perceptual priors Science 2017 357 6351 596 600 28798131 Rao RP Ballard DH Predictive coding in the visual cortex: afunctional interpretation of some extra-classical receptive-field effects Nat Neurosci 1999 2 1 79 87 10.1038/4580 10195184 Ratcliff R Rouder JN Modeling response times for two-choice decisions Psychol Sci 1998 9 5 347 356 Schartner MM Carhart-Harris RL Barrett AB Seth AK Muthukumaraswamy SD Increased spontaneous MEG signal diversity for psychoactive doses of ketamine, LSD and psilocybin Sci Rep 2017 7 46421 28422113 Schneider F Bermpohl F Heinzel A Rotte M Walter M Tempelmann C Wiebking C Dobrowolny H Heinze H Northoff G The resting brain and our self: self-relatedness modulates resting state neural activity in cortical midline structures Neuroscience 2008 157 1 120 131 18793699 Silverstein SM Keane BP Perceptual organization impairment in schizophrenia and associated brain mechanisms: review of research from 2005 to 2010 Schizophr Bull 2011 37 4 690 699 21700589 Sitt JD King J-R El Karoui I Rohaut B Faugeras F Gramfort A Cohen L Sigman M Dehaene S Naccache L Large scale screening of neural signatures of consciousness in patients in a vegetative or minimally conscious state Brain 2014 137 8 2258 2270 24919971 Speechley WJ Whitman JC Woodward TS The contribution of hypersalience to the “jumping to conclusions” bias associated with delusions in schizophrenia J Psychiatry Neurosci: JPN 2010 35 1 7 20040242 Sterzer P Adams RA Fletcher P Frith C Lawrie SM Muckli L Petrovic P Uhlhaas P Voss M Corlett PR The predictive coding account of psychosis Biol Psychiatry 2018 84 9 634 643 30007575 Sterzer P Mishara AL Voss M Heinz A Thought insertion as a self-disturbance: an integration of predictive coding and phenomenological approaches Front Hum Neurosci 2016 10 502 27785123 Tamminga CA Ivleva EI Keshavan MS Pearlson GD Clementz BA Witte B Morris DW Bishop J Thaker GK Sweeney JA Clinical phenotypes of psychosis in the bipolar-schizophrenia network on intermediate phenotypes (B-SNIP) Am J Psychiatry 2013 170 11 1263 1274 23846857 Teufel C Subramaniam N Dobler V Perez J Finnemann J Mehta PR Goodyer IM Fletcher PC Shift toward prior knowledge confers a perceptual advantage in early psychosis and psychosis-prone healthy individuals Proc Natl Acad Sci 2015 112 43 13401 13406 26460044 Turkheimer FE Rosas FE Dipasquale O Martins D Fagerholm ED Expert P Váša F Lord L-D Leech R A complex systems perspective on neuroimaging studies of behavior and its disorders The Neuroscientist 2021 10.1177/1073858421994784.1073858421994784 Tzourio-Mazoyer N Landeau B Papathanassiou D Crivello F Etard O Delcroix N Mazoyer B Joliot M Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain NeuroImage 2002 15 1 273 289 11771995 Van Veen BD Van Drongelen W Yuchtman M Suzuki A Localization of brain electrical activity via linearly constrained minimum variance spatial filtering IEEE Trans Biomed Eng 1997 44 9 867 880 9282479 Vollenweider FX Vollenweider-Scherpenhuyzen MF Bäbler A Vogel H Hell D Psilocybin induces schizophrenia-like psychosis in humans via a serotonin-2 agonist action Neuroreport 1998 9 17 3897 3902 9875725 Waltz JA Gold JM Probabilistic reversal learning impairments in schizophrenia: further evidence of orbitofrontal dysfunction Schizophr Res 2007 93 1-3 296 303 17482797 Winkler I Haufe S Tangermann M Automatic classification of artifactual ica-components for artifact removal in eeg signals Behav Brain Funct 2011 7 1 1 15 21205317 Zalesky A Fornito A Bullmore ET Network-based statistic: identifying differences in brain networks NeuroImage 2010 53 4 1197 1207 20600983 Zhang X-S Roy RJ Jensen EW EEG Complexity as a measure of depth of anesthesia for patients IEEE Trans Biomed Eng 2001 48 12 1424 1433 10.1109/10.966601 11759923 Ziv J Coding theorems for individual sequences IEEE Trans Inf Theory 1978 10.1109/TIT.1978.1055911
PMC009xxxxxx/PMC9384215.txt
==== Front ACS Agric Sci Technol ACS Agric Sci Technol as aastgj ACS Agricultural Science & Technology 2692-1952 American Chemical Society 10.1021/acsagscitech.2c00084 Article Highly Efficient and Reproducible Genetic Transformation in Pea for Targeted Trait Improvement Kaur Rajvinder ‡ Donoso Thomas † Scheske Chelsea ‡ Lefsrud Mark ‡ https://orcid.org/0000-0002-1139-9251 Singh Jaswinder *† † Department of Plant Science, McGill University, 21111 Rue Lakeshore, Sainte-Anne-de-Bellevue, Quebec, Montreal H9X 3V9, Canada ‡ Department of Bioresource Engineering, McGill University, 21111 Rue Lakeshore, Sainte-Anne-de-Bellevue, Quebec, Montreal H9X 3V9, Canada * Email: [email protected]. 19 07 2022 15 08 2022 19 07 2023 2 4 780787 © 2022 The Authors. Published by American Chemical Society 2022 The Authors https://creativecommons.org/licenses/by-nc-nd/4.0/ Permits non-commercial access and re-use, provided that author attribution and integrity are maintained; but does not permit creation of adaptations or other derivative works (https://creativecommons.org/licenses/by-nc-nd/4.0/). A reproducible tissue culture protocol is required to establish an efficient genetic transformation system in highly recalcitrant pea genotypes. High-quality callus with superior regeneration ability was induced and regenerated on optimized media enriched with copper sulfate and cytokinins, 6-benzylaminopurine and indole-3-acetic acid. This successful regeneration effort led to the development of a highly efficient transformation system for five pea genotypes using immature and mature seeds. The new transformation protocol included the addition of elevated glucose and sucrose concentrations for cocultivation and inoculation media to improve callus induction and regeneration, thus resulting in consistent transformation frequencies. Using the Agrobacterium strain AGL1, a transformation frequency of up to 47% was obtained for the pea genotype Greenfeast, using either of two different selection marker genes, PAT or NPT, sourced from two different vectors. Sixty-two transgenic pea events were able to survive kanamycin and phosphinothricin selection. A total of 30 transgenic events for Greenfeast, 15 for CN 43016, 9 for snap pea, and 5 for CN 31237 are reported herein. Two additional transgenic events were recovered from particle gun bombardment experiments. Quantitative RT-PCR analysis confirmed the transgenic status of pea plants, indicating elevated expression of relevant genes cloned into the transformation constructs. pea transformation tissue culture agrobacterium biolistic transgene Natural Sciences and Engineering Research Council of Canada 10.13039/501100000038 CRDPJ 477365-14 Lefsrud Seed and Processors Ltd. NA NA Harbin Seed Farm Ltd. NA NA Agrocentre Belcan Inc. NA NA document-id-old-9as2c00084 document-id-new-14as2c00084 ccc-price ==== Body pmcIntroduction Plant-based diets have risen in popularity in recent years, and evidence suggests that swapping meat servings with protein-rich crops could shape the future of food systems and the environment worldwide.1 Field pea (Pisum sativum) belongs to the Leguminosae family and it is an important cool season grain legume of the Galegoid clade, which also includes lentils (Lens culinaris), chickpeas (Cicer arietinum), and faba beans (Vicia faba). These legumes are good sources of dietary protein and rich in other nutrients such as isoflavonoids, iron, and dietary fibers2,3 that are considered beneficial to human health through anticancer and other health-promoting activities.4 Legumes provide major benefits to cropping systems and the environment because of their ability to perform symbiotic nitrogen fixation in their root nodules, which enhances both soil fertility and water-use efficiency.4−6 Given that pulses can be utilized as human food and animal feed while improving soil and environmental health, they are considered sustainable crops. Those with high protein content are currently being exploited as alternatives to meat worldwide.3,4 Canada is the largest producer and exporter of dry pea, with an annual production of 4.9 million tonnes.7 However, field pea production is constrained by diseases and pests, resulting in significant economic losses.8 Improvement by conventional breeding is as old as agriculture itself, but, until recently, it has been limited to agronomic traits to achieve the desired results. Crop breeding faces a number of challenges due to undesirable genetic linkages and limited knowledge exists for the genetic basis of disease resistance. Genetic transformation has additionally been used to increase crop productivity with the introduction of foreign genes,4 yet gene transfer to higher plants requires robust tissue culture and transformation systems for the generation of viable and fertile transgenic plants. Efficient genetic transformation systems play a key role in molecular breeding and understanding the genetic control of physiological and developmental traits. Nevertheless, field pea is a known recalcitrant plant species for genetic manipulation.9 Pea tissue culture, including callus induction and plant regeneration, has been successful with certain explants. However, genetic transformation of pea genotypes remains limited, likely due to a lack of effective protocols and appropriate plant materials. Earlier efforts using different explant types such as protoplasts, cotyledon, epicotyl, and embryonic segments, yielded low-transformation efficiencies.9−13 Puonti-Kaerlas et al.10 were the first to generate fertile transgenic plants in two pea genotypes, Stivo and Puget, using hygromycin as a selection agent in the medium, and the obtained plants were tetraploids. Three cultivars, Greenfeast, Rondo, and Puget, were transformed using the BAR gene with phosphinothricin (PPT) as the selection agent, with transformation frequencies of 1.5 and 2.5%.9,12 More cultivars, Bolero, Hadlee, Courier, and Crown, were transformed using kanamycin selection, and the transformation frequency ranged from 0.9 to 3.4%.14 Two other cultivars, Crown and Lincoln, were transformed using hygromycin selection with an efficiency of 0.6 to 2.8%.15,16 Since the work of Gregor Mendel, pea has provided a superb understanding of genetic traits.17 Access to its completely sequenced genome18 will certainly expedite our understanding of the genetic and molecular bases of agronomically important traits while enabling progress toward the pea crop improvements to nutritional value and food security.18 However, questions still remain on how to exploit sequenced genes for specific trait improvements in pea, particularly through genetic transformation. The absence of an efficient genetic transformation system has hampered the progress in pea genetics and functional genomics. Therefore, refinement in genetic transformation is required to revolutionize pea breeding with advanced gene-editing methods such as CRISPR-Cas. The present study aimed to develop a highly efficient transformation system in field pea. Modifying media formulations could substantially improve pea transformation, as previously observed in response to increased levels of copper and 6-benzylaminopurine (BAP) in the medium used for cereals transformation.19,20 Herein, we present the optimization of all media required, from callus induction to rooting medium and other transformation variables, to develop an efficient and reproducible system in pea. Materials and Methods Plant Materials Six pea genotypes were used in this study: CN 42819, CN 43016, CN 31237, Greenfeast (TMP 25969), Rondo (TMP 21877), and Snap. These genotypes were selected on the basis of seed color, as light-colored seeds produce fewer phenolic compounds, a desirable trait for tissue culture. The seed coat color was light green for all the selected genotypes, and all genotypes had similar flowering times. Two genotypes, CN 42819 and Rondo, produced round seeds while all others produced wrinkled. Seeds were acquired from Plant Gene Resources of Canada (Saskatoon, SK, Canada). All pea donor plants were grown in pots filled with G6 Agro Mix (Teris, Laval, QC, Canada) and a sand mixture (10:1) in the research greenhouse at McGill University’s Macdonald Campus. Every 2 weeks, plants were fertilized (20:20:20) with all-purpose fertilizer (Plant Products Inc., Leamington, ON, Canada). Temperatures were kept at approximately 25 °C during the day and 20 °C at night. Plants were grown at a light intensity between 1000 and 1400 μmol/m2 s with a 16 h photoperiod. Green Callus Induction and Its Further Proliferation Seed pods were collected when grown to maximum size but still green. Immature green seeds were isolated from pods and surface sterilized with 20% bleach for 20 min, followed by 3–4 washes with autoclaved water. Seeds were cut into two halves. The embryonic axis and cotyledon were used for callus induction. Callus was initially induced on different media (D′, CIM, and P1) as described previously9,21,22 and from mature seeds. Callus initiated on D′ and CIM media was further proliferated and maintained on DBC3 media by moving the calli to fresh media every 2 weeks.21 Similarly, callus initiated on P1 was further maintained on P1. The calli were maintained under dim light conditions (approximately 10–30 μmol/m2 s and a 16-h photoperiod). Calli were additionally induced on a new medium, “KREG”. The composition of all media used in this study is described in Table 1. Callus was multiplied until enough explant was acquired for genetic transformation and regeneration experiments. Table 1 Composition of Media for Tissue Culture D′21 4.4 g/L MS salts (Phytotechnology Laboratories, Shawnee Mission, KS, catalogue number M524), 0.25 g/L myo-inositol, 1.0 g/L casein hydrolysate, 1.0 mg/L thiamine HCL, 2.0 mg/L 2,4-D, 30 g/L maltose, 0.69 g/L l-proline, 0.025 mg/L cupric sulfate, 3.5 g/L phytagel, pH 5.8 CIM22 4.4 g/L MS salts, 0.5 mg/L nicotinic acid, 0.5 mg/L pyridoxine HCl, 1.0 mg/L thiamine HCl, 0.1 g/L myoinositol, 2.0 mg/L 2,4-D, 30 g/L sucrose, 3.5 g/L phytagel, pH 5.8 P19 4.4 g/L MS salts, 1.0 mg/L nicotinic acid, 1.0 mg/L pyridoxine HCl, 10.0 mg/L thiamine HCl, 0.1 g/L myoinositol, 30g/L sucrose, 2 mg/L BAP and 2 mg/L NAA, 3.5 g/L phytagel, pH 5.8 P29 P1 medium with changes, 4.5 mg/L BAP and 0.02 mg/L NAA DBC321 4.4 g/L MS salts, 0.25 g/L myo-inositol, 1.0 g/L casein hydrolysate, 1.0 mg/L thiamine HCL, 1.0 mg/L 2,4-D, 30 g/L maltose, 0.69 g/L l-proline, 1.22 mg/L cupric sulfate, 0.5 mg/L BAP, 3.5 g/L phytagel, pH 5.8 LREG23 4.4 g/L MS salts, 1.0 mg/L nicotinic acid, 1.0 mg/L pyridoxine HCl, 10.0 mg/L thiamine HCl, 0.1 g/L myoinositol, 30g/L sucrose, 1 mg/L BAP, 1 mg/L IAA and 0.16 mg/L CuSO4, 3.5 g/L phytagel, pH 5.8 KREG 4.4 g/L MS salts, 0.5 mg/L nicotinic acid, 0.5 mg/L pyridoxine HCl, 1 mg/L thiamine HCl, 30 g/L sucrose, 0.16 mg/L CuSO4, 1 mg/L BAP and 1 mg/L IAA, 0.5 g/L phytagel, pH 5.8 KREG-OM KREG medium with 0.2 M Sorbitol, 0.M Mannitol, no sucrose B59 4.4 g/L MS salts, 1.0 mg/L nicotinic acid, 1.0 mg/L pyridoxine HCl, 10.0 mg/L thiamine HCl, 0.1 g/L myoinositol, 10 g/L glucose, 20 g/L sucrose, 2 g/L MES, with 100 μM acetosyringone pH 5.8 KREG-IN KREG medium with 68.5 g/L sucrose, 36 g/L glucose, with 100 μM acetosyringone added before use KROOTING KREG medium without BAP with 1 mg/L, NAA, 1 mg/L IBA Standardization of Regeneration of Plantlets from the Green Callus To ensure the quality of callus for regeneration, 10 pieces of 2-month old green tissues were placed on three different regeneration media, LREG,23 P2,9 and KREG, and exposed to higher light intensity (45–55 μmol/m2 s). KREG media was used as the sole media for callus induction to the regeneration of the full plants with the different pea genotypes. Rooting media (KROOTING) was optimized by modifying KREG with respect to MS salt strength and growth hormone (IBA, IAA, and NAA) concentrations. Four different rooting media formulated and investigated: LMS, L1/2MS, PMS, and P1/2MS (P with BAP). Two media, L1/2MS and P1/2MS, contained MS salts at half strength, the others contained MS salts at full strength, with and without 1 mg/L BAP. For root induction, well-formed shoots were sliced off the callus and planted in solidified rooting media (KROOTING, Table 1) in plastic sundae cups. After approximately 4–5 weeks, plantlets with well-established roots were transferred to G6 Agro Mix potting mix and grown in a CMP4030-Conviron growth chamber with 23 °C day/18 °C night under 16-h photoperiod. Plants were covered for 4–5 days with plastic domes to acclimatize and grow there until maturity at a light intensity of 200–250 μmol/m2 s. Gene Constructs and Their Confirmation Two binary vectors pJP3502 and pJP367924 that contain the DGAT1 and WRI1 genes from Arabidopsis thaliana as well as OLEOSIN from Sesamum indicum for co-expression were kindly provided by Dr. Surinder Singh (CSIRO Canberra, Australia). Genes in pJP3502 are driven by leaf-specific promoters and a double enhancer region 35S promoter controls the NPTII kanamycin resistance gene. Genes in pJP3679 are driven by seed-specific promoters and a double enhancer region 35S promoter controls the PAT bialaphos and PPT resistance gene. Both plasmids were cloned into Escherichia coli (DH5α cells) and grown overnight at 37 °C in Luria–Bertani (LB) media with 100 mg/L kanamycin to isolate DNA for bombardments. The plasmids were confirmed by restriction digestion with enzymes and PCR amplification. Standardization of Appropriate Selectable Markers in Genetic Transformation Calli (25–30 pieces each) from all pea genotypes were subcultured on DBC3 media with 2 mg/L PPT, DBC3 with 20 mg/L hygromycin and DBC3 with 20 and 40 mg/L kanamycin. Concentrations of PPT and kanamycin were further investigated using different media (LREG, P2, and KREG) with PPT (2.5, 5.0, and 7.5 mg/L) and kanamycin (50, 100, and 150 mg/L). Particle Gun Bombardment Five pea genotypes (CN 43016, CN 31237, Snap, Greenfeast, and Rondo) were used for particle gun bombardment. Highly regenerative green calli proliferated on KREG media for pea were placed on osmotic media (KREG-OM), KREG without sucrose and enriched with 0.2 M (36.4 g/L) mannitol and 0.2 M (36.4 g/L) sorbitol prior to bombardment for 4 h. Approximately, 25 callus pieces were used per plate, and 4–6 plates were used for each pea genotype and each plasmid. A total of 25 μg plasmid DNA was coated on 0.6 μm gold particles (INBIO GOLD, Hurstbridge, Victoria, Australia) and treated with spermidine and calcium chloride prior to bombardment. Bombardment was carried out with a PDS-1000He biolistic device (Bio-Rad, Hercules, CA) at 1100 psi.20 Bombarded calli were moved to KREG 16–18 h after bombardment and grown at 24 °C under higher intensity light (45–55 μmol/m2 s). After 6–7 days, tissues were transferred to the KREG medium with selection. The selection system was 5–10 mg/L PPT (Phytotech Labs, Lenexa, KS, US) in the KREG regeneration medium for pJP3679 and 100–150 mg/L kanamycin (Bio Basic, Markham ON, Canada) for pJP3502. Calli were subcultured at 2–3 weeks intervals on the same selective medium. After 3–4 selection rounds, surviving shoots were placed on rooting media with 50–100 mg/L kanamycin. Plantlets with developed roots were transferred to the soil and grown under similar conditions as described above. Agrobacterium Transformation Five pea genotypes (CN 43016, CN 31237, Greenfeast, Rondo, and Snap pea), two constructs (pJP3679 and pJP3502), and the Agrobacterium tumefaciens strain AGL1 were used for Agrobacterium transformation experiments. Agro-transformation was also performed with a GFP reporter gene containing plasmid (pActinIntGFPnos25). Seeds were sterilized and embryonic axes were excised from cotyledons to use as explants. The root part was cut off and the remaining region (shoot) was used for agro-infections. Mature seeds were soaked in water for 1–2 days before inoculation. AGL1 was grown overnight at 28 °C in the LB medium supplemented with 100 mg/L kanamycin with shaking at 200 rpm. The bacterial culture was centrifuged at 5000 rpm for 5 min and the pellet was suspended (A600 = 0.4) in an inoculation medium. Agrobacterium transformation was performed by two different methods. In initial experiments, embryonic axes were sliced longitudinally into three to four segments and plated on a B5-H medium, after inoculation with the A. tumefaciens suspension in B5-I.9,26 Both media (B5-I and B5-H) were supplemented with 100 μM acetosyringone after autoclaving. After 4 days of cocultivation, explants were subcultured on the callus induction medium P1, after washing with sterile water containing 300 mg/L Timentin to remove overgrown AGL1.9 In the majority of agro-infection experiments, the bacterial pellet was suspended in the inoculation liquid medium, KREG with 36 g/L glucose, 68 g/L sucrose, and 100 μm acetosyringone without a solidifying agent (KREG-In). The embryonic axis was cut into two cross segments and plated on the KREG medium. Sliced segments were immersed in the bacterial suspension. For each agro-inoculation, embryonic regions of approximately 20 seeds/genotype were soaked in the Agrobacterium culture for 15 min and placed on co-cultivation agar media, KREG-Co (solidified KREG-In), for 2–3 days at 24 °C under light. After 2–3 days of co-cultivation, the segments were transferred to KREG media containing the agro-inhibiting antibiotic Timentin (150 mg/L). After 6–7 days, initiating young shoots were removed and subcultured along with calli on the same media amended with Timentin and 3–4 mg/L PPT for pJP3679, and kanamycin (100–150 mg/L) for pJP3502. Shoots surviving after two to three rounds of selection were subcultured on the rooting medium with kanamycin (50–100 mg/L) and 3 mg/L PPT for pJP3502 and pJP3679 putative events, respectively. The surviving plants with roots from different transformation events were transferred to pots. In pots, they were grown in a CMP4030-Conviron Growth Chamber at 23 °C day temperature and 18 °C night temperature with a 16-h photoperiod and 50–60% relative humidity. Most plantlets transferred to soil in a controlled environment survived. Characterization of Transgenic Events PCR Analysis DNA from leaf samples of the putative transgenic plants was extracted and analyzed by PCR to verify the presence of selection marker genes PAT (in pJP3679) and NPT (in pJP3502). This amplification was done using either a Phire Plant Direct PCR Master Mix (F-160L Thermo Fisher Scientific, Baltics UAB, Vilnius, Lithuania) or Go Taq G2 Green Master Mix (M7822, Promega, Madison, WI, US) according to the manufacturer’s instructions. The PCR reaction volume was 15 μL, including 7.5 μL Phire master mix, 0.25 μM both forward and reverse primers (PATF and PATR), 0.5 μL DNA obtained with dilution buffer, and 6 μL molecular biology grade water. Amplification conditions for the PAT gene were 98 °C for 5 min, followed by 40 cycles at 98 °C for 5 s, 58 °C for 5 s, and 72 °C for 20 s, with final extension on 72 °C for 1 min. Temperatures and timing were controlled using a Thermal cycler C1000 from Bio-Rad. The NPT gene was amplified with primers NPTII-11F and NPTII-488R under similar conditions, except that annealing was done at 56 °C. PCR amplifications were also conducted using the reverse primer JIPR11 with PATF and NPTII-11F as forward primers. Sequences of the PCR primers are listed in Table 2. PCR products were analyzed by gel electrophoresis in 0.8% agarose gels. Table 2 Primers Used for PCR and qRT-PCR Amplification primer sequence 5′ to 3′ PATF GACCTGCTACTGCTGCTGATATGG PATR TCGTTAGGGAGTCCGATAACAGC NPTII-11F AAGATGGATTGCACGCAGGT NPTII-488R TGATGCTCTTCGTCCAGATC JIPR11 GCGCGCGATAATTTATCC HMGF ATGGCAACAAGAGAGGTTAA HMGR TGGTGCATTAGGATCCTTAG RTDGATF GCTGGTGTTACTACCGTTAC RTDGATR CTGCTGGAGAACCAGAAATC RTOLEOF GAGAACTGCTCAGCCCTATC RTOLEOR TGTGCTTTCCGGACTTACTC HMGIIIF AGGGGTAGGCCGAAGAAGAT HMGIIIR TGAGGCTTCACCTTAGGAGG qRTDGATF CTAACCTCGCTGGTGATAAC qRTDGATR TGGCCTGTAGGTGAAAGTAG qRTOLEOF GGTTCCAGCAAGAGTAAGTC qRTOLEOR GGCTCCACATCTTCAACTTC PPTF ATACAAGCACGGTGGATGG PPTR TGGTCTAACAGGTCTAGGAGGA In the case of Go Taq G2 Green Master Mix, PCR reactions contained 50 ng (1 μL) genomic DNA, 0.5 μM each primer (forward and reverse), 10 μL Master Mix, and ddH2O to reach a final volume of 20 μL. PCR was performed with an initial denaturation at 95 °C for 2 min, followed by 35 cycles of 95 °C for 30 s, 58 °C for 45 s, and 72 °C for 45 s, with a final extension on 72 °C for 5 min. Expression Analysis Primary screening of the transformation events obtained with pJP3679 was conducted by leaf painting with the Liberty herbicide (20% glufosinate ammonium; 0.25% v/v with 0.1% Tween) as described previously by Singh et al.25 For gene expression analysis, total RNA was isolated from young leaves using the TRIzol (Ambion RNA by life technologies Carlsbad, Ca, US) method. RNA integrity and concentration were assessed by spectrophotometry with ND1000 (NanoDrop, Wilmington, DE) and a total of 1 μg RNA was treated with DNase I (Catalogue number M198A, Promega, Madison, WI, US) prior to cDNA synthesis using the AffnityScript cDNA Synthesis Kit (Agilent technology, Cedar Creek, Texas, US). RTPCR and qRTPCR were carried out to detect the expression of DGAT, OLEO, and WRI genes as described by Singh and Singh.27 HMG-I/Y was used as a reference gene. For RTPCR primers, HMGF and HMGR were used to amplify 350 bp and 570 bp fragments (DNA/cDNA).28 RT-PCR amplification with cDNA was carried out for the DGAT gene using primers RTDGATF and RTDGATR, and the OLEO gene was amplified with primers RTOLEOF and RTOLEOR (Table 2). 12 μL of the RT-PCR products were analyzed by electrophoresis on 1.2% agarose gels. Primer pairs qRTDGAT, qRTOLEO, and PPT (Table 2) were used to amplify the DGAT, OLEO, and PPT genes to analyze quantitative transgene expression in the transgenic plants. To normalize expression of these transgenes, the pea housekeeping gene, HMG-I/Y was used as a reference with primers HMGIII-F and HMGIII-R amplifying a 164-bp product for the quantitative expression.28 qRT-PCR was performed in optical strip tubes using the Mx3000 qPCR system (Stratagene, US). qRT-PCR was performed in optical strip tubes using the Mx3000 qPCR system (Stratagene, US). The qRTPCR mixture (20 μL) contained 0.8 μL from 10 μm stock of both forward and reverse primers, 10 μL of Advanced qPCR Master Mix (Cat: 800-435-UL Wisent Inc., St-Bruno, QC, Canada), and 2 μL cDNA of each sample. Amplification conditions were an initial denaturation step at 95 °C for 2 min, followed by 40 cycles of 95 °C for 5 s, 60 °C for 30 s, and 72 °C for 30 s, with a final melting curve analysis step of 95 °C for 1 min, 60 °C for 30 s, and 95 °C for 30 s. The qRTPCR data were automatically collected and analyzed. The relative level of gene expression was analyzed using the 2–ΔΔCq method.29 Expression values were represented as the average values of three biological repeats and error bars represented standard deviations. Asterisks indicated significant differences revealed by the Student’s t-test using JMPpro 16 software. Results Standardization of Tissue Culture Process For any successful tissue culture experiment to produce reliable embryogenic callus, the choice of an appropriate explant is extremely crucial. Callus initiation was observed 7–10 days from when cotyledons and embryonic regions of immature/mature seeds were excised for all the genotypes used in this study (Figure 1A,B). However, the growth and proliferation behavior of calli from different genotypes was dramatically different. Calli for genotype CN 42819 was light green in color and the fastest growing. Calli from all other genotypes were slow in growth, with a compact nodular shape and a darker green color (Figure 1C). To check regenerability, calli obtained from different genotypes were transferred to known regeneration media including LREG and P2.9,23 No shoot regeneration was observed on LREG, but root production was observed (Figure 1D). With the P2 medium, very few calli did demonstrate some shoot growth (2–3 shoots per callus) in the genotype Greenfeast. We observed that calli from almost all genotypes could not survive more than 3–4 weeks on the P2 medium. These findings prompted us to identify new media components for the regeneration of pea calli and new medium, “KREG”, was formulated by modifying concentrations of vitamins and growth regulators and supplementing with copper sulfate (Table 1). Callus initiation, multiplication, and regeneration on the KREG medium was superb for all pea genotypes (Figure 1E). Once shoots emerged to 1–2 cm, they were transferred to rooting media. Figure 1 Standardization of tissue culture steps for regeneration of field pea plants. (A) Embryonic axis attached to cotyledon from immature seeds; (B) callus initiation from cotyledon and embryonic segments on the D medium in 2–3 weeks; (C) development and proliferation of green callus on DBC3; (D) failed green callus regeneration on LREG; (E) regenerating callus on KREG; and (F) plantlets with roots on rooting media. Different rooting media were modified by altering the composition of growth hormones and MS salt strength and then subsequently investigated. Shoots could not survive on 1/2 MS strength media. Shoots multiplied on MS media with BAP, but root initiation was hampered. Roots were successfully generated on the rooting medium with full strength MS salts without BAP (Figure 1F). The best response was observed in Greenfeast, followed by CN 43016, Rondo, and CN 31237. Reliability of Appropriate Selectable Markers in Genetic Transformation For any successful genetic transformation system, reliable selection markers are required for the recovery of viable transgenic plants. Three selective agents, hygromycin, kanamycin, and PPT were optimized for the pea genotypes. All green calli began to display hygromycin stress symptoms on DBC3 with 20 mg/L hygromycin within 1 week of culturing. Calli became necrotic, the media turned brown in 2–3 weeks, and calli were unable to survive. Contrarily, calli proliferated slowly on the DBC3 medium containing 40 mg/L kanamycin and 2 mg/L PPT. Concentrations of PPT and kanamycin were further optimized in LREG, P2, and KREG media by supplementing with 2.5–7.5 mg/L PPT and 50–150 mg/L kanamycin. It was noted that PPT between 2.5–5.0 mg/L and kanamycin 100–150 mg/L were effective as selective agents for all pea genotypes. Recovery of Transgenic Events via Agrobacterium Transformation and Particle Gun Bombardment Poor callus response was observed when longitudinally cut embryonic axes were used for inoculation and subcultured on the B5 medium. Heavy AGL1 growth was observed, which covered all segments in 4 days. Therefore, calli was unable to establish on the P1 induction medium. Interestingly, callus response was superb when embryonic segments were cross-sectioned (cut horizontally) and inoculation and co-cultivation media were supplemented with copper sulfate, glucose, and a higher concentration of sucrose in all agro-transformation experiments (Figure 2A,B). Callus proliferated faster and was highly regenerable on media enriched with copper sulfate (0.16 mg/L) in combination with cytokinins (1 mg/L BAP and 1 mg/L IAA) (Table 1), leading to the regeneration of 5–6 plantlets per putative transgenic event. This observation was made for all pea genotypes used in this transformation experiment (Figures 1E and 2D,E). Upon transfer to the regeneration medium with a selective agent and with the surviving shoots to rooting medium with selective agents, different genotypes produced transformants at varying efficiencies (Figure 2E,F). The frequency of transformed embryos with pActinIntGFPnos expressing GFP was examined 2 d, 14 d, and 28 d after co-cultivation. A negative correlation was observed between the number of GFP-expressing calli and days after co-cultivation. As expected, the size of the GFP-expressing spots became larger as the number of days increased after inoculation (Figure 2C). This expression pattern indicated the presence of stable transformation events. Sixty-two putative transgenic events in pea for both plasmids (pJP3502 and pJP3679) through agro-infection were able to survive through kanamycin and PPT selection (Figure 2E,F). Figure 2 Agrobacterium-mediated transformation in field pea. (A) Embryonic axis of immature seeds after 2 days of inoculation with Agrobacterium strain AGL1 on KREG-CO; (B) callus initiation of inoculated embryonic on KREG after 1 week; (C) GFP expression in 4 weeks old callus; (D) selection of transgenic calli on KREG using 100–150 mg/L kanamycin; (E) transgenic plants on rooting media; and (F) healthy transgenic plant. Transformation efforts through particle gun bombardment were successful but low transformation frequency was recorded. In different transformation experiments, where 1500 pieces from three genotypes (CN 43016, Greenfeast, and Rondo), were bombarded, only 16 calli survived through selection and only two transgenic plants from CN 43016 were obtained (Figure 2F). Molecular Confirmation of Transgenic Events PCR analysis confirmed the presence of the NPT and PAT transgenes in 58 pea events (T0) for 4 pea genotypes, and for 15 events in T1 to T3 generations (Figures 3, 4 and 5). Most tissues were resistant to kanamycin or PPT and the regenerated T0 plants were positively amplified for their respective genes. PCR analyses of T1, T2, and T3 plants using the same set of primers confirmed stable inheritance of the transgene. Transformation frequency varied strikingly for different pea genotypes (Table 3). The maximum number of transgenic events was obtained in Greenfeast with a transformation frequency of 47% (30 events), followed by snap peas (22%; 9 events), CN 43016 (13%; 14 events), and CN 31237 (4%; 5 events). No events were recovered in Rondo (Table 3). transformation frequency via bombardment was extremely low; only two transgenic events were confirmed in the CN 43016 pea genotype. Figure 3 PCR amplification of the KAN gene from plants pJP3502; (A) T0 events. (B) T1 plants. Lane 1: 1 kb plus ladder; lanes 2–17: transformants; lane 18: non-transgenic control; lane 19: water; lane 20: plasmid. Figure 4 PCR amplification of the PAT gene from Greenfeast transformed with pJP3679. (A) T0 events. Lane 1: Marker; lanes 2–16: transgenic events; lane 17: non-transgenic control; lane 18: +ve transgenic control; lane 19: water; lane 20: plasmid DNA. (B) T1 plants. Lane 1: marker; lanes 2–17: transgenic plants; lane 18: non-transgenic control, lane 19: water; and lane 20: plasmid DNA. Figure 5 PCR amplification of the KAN gene from CN 43016 transformed with particle gun bombardment. (A) T0 events. Lane 1. Marker; lanes 2–11 transgenic events; lane 12: plasmid DNA: lane 13: non-transgenic control; lane 14: water. (B) Progeny of T1 transgenic plants. Lane 1: marker; lanes 2–15 T2 plants; lanes 16–17 T1 transgenic plants; lane 18: non-transgenic control; lane 19: plasmid DNA; and lane 20: water. Table 3 Frequency of Transformation in Different Genotypes genotype construct delivery system no of explants no of transgenic events transformation frequency (%) CN 31237 agro-infection 120 5 4.1 CN 43016 agro-infection 113 15 13.2   particle gun bombardment 500 2 0.4 Greenfeast agro-infection 64 30 46.8 snap agro-infection 40 9 22.5 Transcript Abundance of Transgenes Expression of the PAT gene using pJP3679 in transformed events and their subsequent generations was examined using the leaf paint assay with the Liberty herbicide. Plants containing the PAT gene remained healthy after herbicide treatment (Figure 6). The plasmids used to develop transgenic plants also contain the OLEO gene, in addition to the PPT selective marker. The selected T2 transgenic plants from Greenfeast were further analyzed by qRT-PCR and semi-quantitative RT-PCR. The transcript of both genes was detected in the RT-PCR. Transcript abundance for both genes were further observed by qRT-PCR in the selected T2 transgenic and non-transgenic (control) plants. Expression folds for PPT and OLEO were normalized with the HMG reference gene. Analysis of the selected transgenic events indicates a 3- to 8-fold transcript abundance of PPT and OLEO, respectively (Figure 7). The expressions of PPT and OLEO were significant different from the expression in the wild-type control (Greenfeast) using the student t-test. This analysis further verified the transgenic status of the pea genotypes. Figure 6 Leaf paint assay of the transgenic events (pJP3679) showing Liberty herbicide tolerance, with a representative image showing the reaction of Liberty herbicide on pea leaves after 1 week. (A) Plants resistant to Basta and (B) susceptible (control). Figure 7 Expression analysis of different genes in T2 transgenic Greenfeast plants using real-time quantitative PCR; (A) expression of selectable marker PPT; (B): expression of the DGAT gene; and (C): expression of the OLEO gene. The statistical significance of relative expression in transgenic plants and Greenfeast wild control was determined with a student t-test (*p < 0.05). Discussion Efforts to improve field pea through genetic modifications are hindered because of poor transformation efficiency. Choosing the best explant to produce reliable embryogenic callus and an appropriate selection strategy both play critical roles in genetic transformation. A maximum frequency of transformation of 3.4% has been reported in some pea genotypes, using embryonic regions and cotyledons as explants with the selection agents hygromycin, PPT, and kanamycin.9,14,30 Other studies that indicate a higher frequency of transgenic calli that could be selected failed to regenerate calli into transgenic plants.30 Therefore, regenerating plants from the selected calli is one of the most critical factors for obtaining transgenic plants. Since 2000, there has not been any known constructive effort made to improve transformation efficiency in pea. However, new approaches for precise manipulation, through gene-editing for crop improvement, have reinvigorated the need and potential of plant tissue culture and transformation. This study has made substantial progress in devising new media formulations for high-quality callus induction, efficient regeneration from cotyledon and embryonic segments, and efficient genetic transformation in five different pea genotypes, a plant species that is historically recalcitrant to transformation. Shiny compact green callus with highly embryogenic structures were originally obtained on three different callus-inducing media, after culturing embryonic regions and cotyledons from five pea genotypes in a few weeks. Calli were unable to regenerate on these previously optimized media. Similar media have been successfully utilized in other crop plants, including barley, oat, and sorghum crops.21,22,31 For cereals, calli initiated and maintained on these media are highly regenerable with large numbers of shoots that can be maintained for more than a year with minimal loss in regenerability. Previous efforts for the regeneration of the callus in pea have unsuccessful32−35 or with very limited success.30 Optimization of the media for callus induction, regeneration, and rooting was critical for transformation success presented in this work. The effective regeneration of calli in our experiments and seems dependent on new media formulation (Figures 1 and 2). This high-quality callus was highly regenerable on media enriched with copper sulfate, in combination with the cytokinins BAP and IAA (Table 1). This observation was true for all pea genotypes used for transformation in this study. The importance of BAP and NAA in legumes to increase shoot bud differentiation and regeneration has been reported previously.36 Similarly, high levels of copper present in the medium enhance the callus induction and regeneration ability in cereals.19,20 An efficient regeneration ability in different pea genotypes allowed us to perform effective and efficient genetic transformation in the field pea. High-efficiency pea transformation was obtained, between 4% (CN 31237) to 47% (Greenfeast). The number of these events in different pea genotypes includes 30 for Greenfeast, 14 for CN 43016, 9 for snap pea, and 5 for CN 31237 (Table 3). In agro-transformation experiments, only 1–2 cross-sectioned segments of the embryonic axis were used (next to the root shoot node) for agro-infection over 3 d on the co-cultivation medium KREG-Co, followed by regeneration on KREG media. Our co-cultivation media contains higher amount of sugars (glucose and sucrose) and CuSO4, which seems crucial for calli health during agro-infection. Similar observations for callus response with high concentrations of sugars are reported in cereals.22,37,38 Agro-mediated transformation using different types of pea explants such as epicotyl segments, protoplasts, lateral cotyledonary meristems, nodulus explants, and embryonic axis, have been reported previously.9,14 However, low transformation frequency has continued to be a major hurdle.9,14,30 The reproducibility of this protocol was demonstrated by the consistently high transformation frequencies generated from embryonic regions across multiple experiments with five pea genotypes. Using this optimized protocol with Agrobacterium strain AGL1, the transformation frequency was as high as 47% for the pea genotype Greenfeast, which is substantially higher than previously reported efficiencies.12,13 Nevertheless, our transformation experiments with biolistic have limited success, where only two stable transgenic plants were generated. To our knowledge, this is the first study where particle gun bombardment has been used for transformation in field pea. Out of five pea genotypes used in the transformation experiment, only one genotype, CN 43016, was successfully transformed. This was surprising, as bombardment is considered to have less genotypic dependence;39 as such, this particle gun bombardment strategy needs further refinement. By using modified media and honing our transformation protocol, 62 independent transgenic pea events were obtained, 15 of which were advanced to the T3 generation. PCR and qRT-PCR confirmed transgene inheritance. Although genotype selection is of the utmost importance to efficient tissue culture systems, we have established a highly efficient reproducible tissue culture system for recalcitrant pea genotypes, now available for further pea improvements through modern genetic editing tools. This work was supported by the Natural Sciences and Engineering Research Council of Canada (NSERC CRDPJ 477365–14), Lefsrud Seed and Processors Ltd, Harbin Seed Farm Ltd. and Agrocentre Belcan Inc. The authors declare no competing financial interest. Acknowledgments Special thanks to Dr. Surinder Singh from CSIRO, Plant Industry Canberra, Australia for providing specific constructs. Also acknowledge the assistance of Dr. Rajiv Kumar Tripathi and Sukhjiwan K Kadoll, McGill University, for their help in QRT-PCR. The authors gratefully acknowledge proofreading of the manuscript by Dr. Sarah MacPherson. ==== Refs References Robinson G. H. J. ; Balk J. ; Domoney C. Improving Pulse Crops as a Source of Protein, Starch and Micronutrients. BNF Nutr. Bull. 2019, 44 , 202–215. 10.1111/nbu.12399. Kouris-Blazos A. ; Belski R. Health Benefits of Legumes and Pulses with a Focus on Australian Sweet Lupins. Asia Pac. J. Clin. Nutr. 2016, 25 , 1–17. 10.6133/apjcn.2016.25.1.23.26965756 Maphosa Y. ; Jideani V. A. The Role of Legumes in Human Nutrition. In Functional Food-Improve Health through Adequate Food, 1; Hueda M. C. , Ed.; IntechOpen: London, England, 2017; Vol. 1 , p 13. Sahruzaini N. A. ; Rejab N. A. ; Harikrishna J. A. ; Khairul Ikram N. K. ; Ismail I. ; Kugan H. M. ; Cheng A. Pulse Crop Genetics for a Sustainable Future: Where We Are Now and Where We Should Be Heading. Front. Plant Sci. 2020, 11 , 531 10.3389/fpls.2020.00531.32431724 Graham P. H. ; Vance C. P. Legumes: Importance and Constraints to Greater Use. Plant Physiol. 2003, 131 , 872–877. 10.1104/pp.017004.12644639 Stagnari F. ; Maggio A. ; Galieni A. ; Pisante M. Multiple Benefits of Legumes for Agriculture Sustainability: An Overview. Chem. Biol. Technol. Agric. 2017, 4 , 2 10.1186/s40538-016-0085-1. Statistics Canada. Table 32-10-0359-01. Estimated Areas, Yield, Production, Average Farm Price and Total Farm Value of Principal Field Crops, in Metric and Imperial Units, 2020. Nikolova I. M. Response of Pea Varieties to Damage Degree of Pea Weevil, Bruchus pisorum L. Scientifica 2016, 2016 , 8053860 10.1155/2016/8053860.27042379 Schroeder H. E. ; Schotz A. H. ; Wardley-Richardson T. ; Spencer D. ; Higgins T. Transformation and Regeneration of Two Cultivars of Pea (Pisum sativum L.). Plant Physiol. 1993, 101 , 751–757. 10.1104/pp.101.3.751.12231726 Puonti-Kaerlas J. ; Eriksson T. ; Engström P. Inheritance of a Bacterial Hygromycin Phosphotransferase Gene in the Progeny of Primary Transgenic Pea Plants. Theor. Appl. Genet. 1992, 84 , 443–450. 10.1007/BF00229505.24203206 Jordan M. C. ; Hobbs S. L. A. Evaluation of a Cotyledonary Node Regeneration System for Agrobacterium-mediated Transformation of Pea (Pisum sativum L.). In Vitro Cell. Dev. Biol.: Plant 1993, 29 , 77–82. 10.1007/bf02632256. Bean S. J. ; Gooding P. S. ; Mullincaux P. M. ; Davies D. R. A Simple System for Pea Transformation. Plant Cell Rep. 1997, 16 , 513–519. 10.1007/bf01142315.30727570 Polowick P. L. ; Quandt J. ; Mahon J. D. The Ability of Pea Transformation Technology to Transfer Genes into Peas Adapted to Western Canadian Growing Conditions. Plant Sci. 2000, 153 , 161–170. 10.1016/s0168-9452(99)00267-8.10717322 Grant J. E. ; Cooper P. A. ; Gilpin B. J. ; Hoglund S. J. ; Reader J. K. ; Pither-Joyce M. D. ; Timmerman-Vaughan G. M. Kanamycin is Effective for Selecting Transformed Peas. Plant Sci. 1998, 139 , 159–164. 10.1016/s0168-9452(98)00184-8. Grant J. E. ; Thomson L. M. J. ; Pither-Joyce M. D. ; Dale T. M. ; Cooper P. A. Influence of Agrobacterium tumefaciens Strain on the Production of Transgenic Peas (Pisum sativum L.). Plant Cell Rep. 2003, 21 , 1207–1210. 10.1007/s00299-003-0640-7.12819922 Nadolska-Orczyk A. Agrobacterium-Mediated Transformation of Pea (Pisum sativum L.). Acta Physiol. Plant. 2008, 187–202. 10.1201/9781439801352.ch13. Reid J. B. ; Ross J. J. J. Mendel’s Genes: Toward a Full Molecular Characterization. Genetics 2011, 189 , 3–10. 10.1534/genetics.111.132118.21908742 Kreplak J. ; Madoui M.-A. ; Cápal P. ; Novák P. ; Labadie K. ; Aubert G. ; Bayer P. E. ; Gali K. K. ; Syme R. A. ; Main D. ; Klein A. ; Bérard A. ; Vrbová I. ; Fournier C. ; d’Agata L. ; Belser C. ; Berrabah W. ; Toegelová H. ; Milec Z. ; Vrána J. ; Lee H. ; Kougbeadjo A. ; Térézol M. ; Huneau C. ; Turo C. J. ; Mohellibi N. ; Neumann P. ; Falque M. ; Gallardo K. ; McGee R. ; Tar’an B. ; Bendahmane A. ; Aury J.-M. ; Batley J. ; Le Paslier M.-C. ; Ellis N. ; Warkentin T. D. ; Coyne C. J. ; Salse J. ; Edwards D. ; Lichtenzveig J. ; Macas J. ; Doležel J. ; Wincker P. ; Burstin J. A Reference Genome for Pea Provides Insight into Legume Genome Evolution. Nat. Genet. 2019, 51 , 1411–1422. 10.1038/s41588-019-0480-1.31477930 Wu E. ; Lenderts B. ; Glassman K. ; Berezowska-Kaniewska M. ; Christensen H. ; Asmus T. ; Zhen S. ; Chu U. ; Cho M.-J. ; Zhao Z.-Y. Optimized Agrobacterium-mediated sorghum transformation protocol and molecular data of transgenic sorghum plants. In Vitro Cell. Dev. Biol.: Plant 2014, 50 , 9–18. 10.1007/s11627-013-9583-z.26316679 Cardinal M.-J. ; Kaur R. ; Singh J. Genetic Transformation of Hordeum vulgare ssp. spontaneum for the Development of a Transposon-Based Insertional Mutagenesis System. Mol. Biotechnol. 2016, 58 , 672–683. 10.1007/s12033-016-9967-5.27480175 Cho M.-J. ; Jiang W. ; Lemaux P. G. Transformation of Recalcitrant Barley Cultivars through Improvement of Regenerability and Decreased Albinism. Plant Sci. 1998, 138 , 229–244. 10.1016/s0168-9452(98)00162-9. Gurel S. ; Gurel E. ; Kaur R. ; Wong J. ; Meng L. ; Tan H.-Q. ; Lemaux P. G. Efficient, Reproducible Agrobacterium-mediated Transformation of Sorghum using Heat Treatment of Immature Embryos. Plant Cell Rep. 2009, 28 , 429–444. 10.1007/s00299-008-0655-1.19115059 Liu G. ; Godwin I. D. Highly Efficient Sorghum Transformation. Plant Cell Rep. 2012, 31 , 999–1007. 10.1007/s00299-011-1218-4.22234443 Vanhercke T. ; El Tahchy A. ; Liu Q. ; Zhou X. R. ; Shrestha P. ; Divi U. K. ; Ral J. P. ; Mansour M. P. ; Nichols P. D. ; James C. N. ; Horn P. J. ; Chapman K. D. ; Beaudoin F. ; Ruiz-López N. ; Larkin P. J. ; Feyter R. C. ; Singh S. P. ; Petrie J. R. Metabolic Engineering of Biomass for High Energy Density: Oilseed-like Triacylglycerol Yields from Plant Leaves. Plant Biotechnol. J. 2014, 12 , 231–239. 10.1111/pbi.12131.24151938 Singh S. ; Tan H. Q. ; Singh J. Mutagenesis of Barley Malting Quality QTLs with Ds Transposons. Funct. Integr. Genomics 2012, 12 , 131–141. 10.1007/s10142-011-0258-8.22105861 Brown D. C. W. ; Atanassov A. Role of Genetic Background in Somatic Embryogenesis in Medicago. Plant Cell, Tissue Organ Cult. 1985, 4 , 111–122. 10.1007/bf00042269. Singh M. ; Singh J. Seed Development-related Expression of ARGONAUTE4_9 Class of Genes in Barley: Possible Role in Seed Dormancy. Euphytica 2012, 188 , 123–129. 10.1007/s10681-012-0624-1. Teressa Negawo A. ; Baranek L. ; Jacobsen H.-J. ; Hassan F. Molecular and Functional Characterization of cry1Ac Transgenic Pea Lines. GM Crops Food 2016, 7 , 159–174. 10.1080/21645698.2016.1240148.27764552 Tripathi R. K. ; Overbeek W. ; Singh J. Global Analysis of SBP Gene Family in Brachypodium distachyon Reveals its Association with Spike Development. Sci. Rep. 2020, 10 , 15032 10.1038/s41598-020-72005-7.32929136 De Kathen A. ; Jacobsen H.-J. Agrobacterium tumefaciens-mediated Transformation of Pisum sativum L. using Binary and Cointegrate Vectors. Plant Cell Rep. 1990, 9 , 276–279. 10.1007/bf00232301.24226826 Cho M.-J. ; Jiang W. ; Lemaux P. G. High-frequency Transformation of Oat via Microprojectile Bombardment of Seed-derived Highly Regenerative Cultures. Plant Sci. 1999, 148 , 9–17. 10.1016/s0168-9452(99)00082-5. Hussey G. ; Gunn H. V. Plant Production in Pea (Pisum sativum L. cvs. Puget and Upton) from Long-term Callus with Superficial Meristems. Plant Sci. Lett. 1984, 37 , 143–148. 10.1016/0304-4211(84)90217-7. Puonti-Kaerlas J. ; Stabel P. ; Eriksson T. Transformation of Pea (Pisum sativum L.) by Agrobacterium tumefaciens. Plant Cell Rep. 1989, 8 , 321–324. 10.1007/bf00716664.24233266 Lulsdorf M. M. ; Rempel H. ; Jackson J. A. ; Baliski D. S. ; Hobbs S. L. Optimizing the Production of Transformed Pea (Pisum sativum L.) Callus using Disarmed Agrobacterium tumefaciens Strains. Plant Cell Rep. 1991, 9 , 479–483. 10.1007/BF00232100.24213784 Krejčí P. ; Matušková P. ; Hanáček P. ; Reinöhl V. ; Procházka S. The Transformation of Pea (Pisum sativum L.): Applicable Methods of Agrobacterium tumefaciens-mediated Gene Transfer. Acta Physiol. Plant. 2007, 29 , 157–163. 10.1007/s11738-006-0020-3. Gulati A. ; Jaiwal P. K. Culture Conditions Effecting Plant Regeneration from Cotyledons of Vigna radiata (L.) Wilczek. Plant Cell, Tissue Organ Cult. 1990, 23 , 1–7. 10.1007/bf00116082. Zhao Z.-y. ; Cai T. ; Tagliani L. ; Miller M. ; Wang N. ; Pang H. ; Rudert M. ; Schroeder S. ; Hondred D. ; Seltzer J. ; Pierce D. Agrobacterium-mediated Sorghum Transformation. Plant Mol. Biol. 2000, 44 , 789–798. 10.1023/a:1026507517182.11202440 Gao Z. ; Xie X. ; Ling Y. ; Muthukrishnan S. ; Liang G. H. Agrobacterium tumefaciens-mediated Sorghum Transformation using a Mannose Selection System. Plant Biotechnol. J. 2005, 3 , 591–599. 10.1111/j.1467-7652.2005.00150.x.17147630 Wan Y. ; Widholm J. M. ; Lemaux P. G. Type I Callus as a Bombardment Target for Generating Fertile Transgenic Maize (Zea mays L.). Planta 1995, 196 , 7–14. 10.1007/bf00193211.
PMC009xxxxxx/PMC9595388.txt
==== Front Sports Med Health Sci Sports Med Health Sci Sports Medicine and Health Science 2666-3376 Chengdu Sport University S2666-3376(22)00063-4 10.1016/j.smhs.2022.10.004 Original Article Athletes with mild post-COVID-19 symptoms experience increased respiratory and metabolic demands: Α cross-sectional study Stavrou Vasileios T. [email protected] ab∗ Kyriaki Astara ac Vavougios George D. ad Fatouros Ioannis G. e Metsios George S. f Kalabakas Konstantinos b Karagiannis Dimitrios b Daniil Zoe a I. Gourgoulianis Konstantinos a Βasdekis George b a Laboratory of Cardio-Pulmonary Testing and Pulmonary Rehabilitation, Respiratory Medicine Department, Faculty of Medicine, University of Thessaly, Larissa, Greece b The Medical Project, Prevention, Evaluation and Recovery Center, Larissa, Greece c Department of Neurology, 417 Army Equity Fund Hospital (NIMTS), Athens, Greece d Department of Neurology, Faculty of Medicine, University of Cyprus, Lefkosia, Cyprus e School of Physical Education and Sport Sciences, University of Thessaly, Trikala, Greece f Department of Nutrition and Dietetics, University of Thessaly, Trikala, Greece ∗ Corresponding author. Laboratory of Cardio-Pulmonary Testing and Pulmonary Rehabilitation, Respiratory Medicine Department, Faculty of Medicine, University of Thessaly, Larissa, 38221, Greece. [email protected] 23 10 2022 6 2023 23 10 2022 5 2 106111 22 7 2022 9 10 2022 17 10 2022 © 2022 Chengdu Sport University. Publishing services by Elsevier B.V. on behalf of KeAi Communications Co. Ltd. 2022 Chengdu Sport University https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Coronavirus Disease 2019 (COVID-19) has significantly affected different physiological systems, with a potentially profound effect on athletic performance. However, to date, such an effect has been neither addressed nor investigated. Therefore, the aim of this study was to investigate fitness indicators, along with the respiratory and metabolic profile, in post-COVID-19 athletes. Forty male soccer players, were divided into two groups: non-hospitalized COVID-19 (n = 20, Age: [25.2 ± 4.1] years, Body Surface Area [BSA]: [1.9 ± 0.2] m2, body fat: 11.8% ± 3.4%) versus [vs] healthy (n = 20, Age: [25.1 ± 4.4] years, BSA: [2.0 ± 0.3] m2, body fat: 10.8% ± 4.5%). For each athlete, prior to cardiopulmonary exercise testing (CPET), body composition, spirometry, and lactate blood levels, were recorded. Differences between groups were assessed with the independent samples t-test (p < 0.05). Several differences were detected between the two groups: ventilation (V˙E: Resting: [14.7 ± 3.1] L·min−1 vs. [11.5 ± 2.6] L·min−1, p = 0.001; Maximal Effort: [137.1 ± 15.5] L·min−1 vs. [109.1 ± 18.4] L·min−1, p < 0.001), ratio VE/maximal voluntary ventilation (Resting: 7.9% ± 1.8% vs. 5.7% ± 1.7%, p < 0.001; Maximal Effort: 73.7% ± 10.8% vs. 63.1% ± 9.0%, p = 0.002), ratioVE/BSA (Resting: 7.9% ± 2.0% vs. 5.9% ± 1.4%, p = 0.001; Maximal Effort: 73.7% ± 11.1% vs. 66.2% ± 9.2%, p = 0.026), heart rate (Maximal Effort: [191.6 ± 7.8] bpm vs. [196.6 ± 8.6] bpm, p = 0.041), and lactate acid (Resting: [1.8 ± 0.8] mmol·L-1 vs. [0.9 ± 0.1] mmol·L-1, p < 0.001; Maximal Effort: [11.0 ± 1.6] mmol·L-1 vs. [9.8 ± 1.2] mmol·L-1, p = 0.009), during CPET. No significant differences were identified regarding maximal oxygen uptake ([55.7 ± 4.4] ml·min−1·kg−1 vs. [55.4 ± 4.6] ml·min−1·kg−1, p = 0.831). Our findings demonstrate a pattern of compromised respiratory function in post-COVID-19 athletes characterized by increased respiratory work at both rest and maximum effort as well as hyperventilation during exercise, which may explain the reported increased metabolic needs. Keywords Cardiopulmonary exercise testing Infected with COVID-19 Respiratory work ==== Body pmcAbbreviations COVID-19 Coronavirus Disease 2019 CPET cardiopulmonary exercise testing FEV1 forced expiratory volume in the 1st second HR heart rate MVV maximal voluntary ventilation PASC post-acute coronavirus disease 2019syndrome PSQI Pittsburg sleep quality index RER respiratory exchange ratio SARS-CoV-2 severe acute respiratory syndrome coronavirus 2 VE ventilation VE/MVV breathing reserve, ventilation/maximal voluntary ventilation ratio V˙O2max maximal oxygen uptake Δchest chest circumference difference between maximal inhalation and exhalation ΔSpO2 difference values between resting and end of test in oxygen saturation measurement with pulse oximetry % predicted percent of predicted values bpm beats per minute cm centimeters kg kilograms kg·m-2 kilogram per square meter km·h−1 kilometers per hour L·min−1 liters per minute m2 square meter mmol·L-1 millimoles per liter μL microliter ml·min−1·kg−1 milliliter per minute to kilograms Introduction The Coronavirus Disease 2019 (COVID-19) pandemic has led to an increase in morbidity and mortality globally.1 After five pandemic waves, attention has shifted to the post-COVID-19 era, in which residual or nascent syndromes formulate the spectrum of long-COVID-19.2 Emerging data support the existence of systematic long-lasting symptoms in COVID-19 survivors, beyond the respiratory system, which is collectively described under the term of post-acute COVID-19 syndrome (PASC).3 PASC may manifest as breathlessness, impaired breathing, increased oxygen requirements, post-viral cough, cardiovascular muscular changes,4 sleep disorders, chronic fatigue, cognitive impairment,5 and sarcopenia.2,6 The multi-organ sequelae may extend up to 6 months post-COVID, making the prioritization of follow-up essential,7 regardless of comorbidity risk status and severity of illness. Along with PASC, a rather sedentary lifestyle prevailed due to local prohibiting legislation to prevent the spread of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), which worsened the disease burden.8 Recent studies have shown that even post-COVID-19 athletes, who have enhanced physical condition, as well as, no previous history of illness or comorbidities, may, face challenges upon their return to their training programs.9 In the absence of relevant data, the aim of this study was to investigate fitness indicators, along with their respiratory and metabolic profile, in post-COVID-19 athletes. Materials and methods Participants Forty male professional soccer players from the Greek Super League 1 and 2 volunteered for this study and were divided into two groups: previously infected with SARS-CoV-2, but non-hospitalized (i.e. mild COVID-19) versus healthy ones (Table 1). All athletes previously infected with SARS-CoV-2 (Omicron variant), were without symptoms (e.g. chest pain, fever, runny nose, cough, sore throat, headaches, muscle pain, fatigue, etc.) and included in our study two days after polymerase chain reaction negative test. The total duration of virus positivity in athletes was (6.1 ± 1.1) days. Athletes were recruited between November 2021 to January 2022. For all athletes' inclusion criteria were: age ≥ 20 to ≤ 30 years, training age ≥ 6 years, without recent injury (> 12 months). Exclusion criteria were: the lack of medical history and recent athletes' transcripts (< 30 days),10 while for post-COVID-19 athletes were difference (Δ) value in blood oxygen saturation (SpO2) between resting and at the end of cardiopulmonary exercise testing (ΔSpO2 > 4%), hospitalization and self-reported symptoms (chest pain, fatigue and/or dyspnea).11 The study's protocol was approved by the Institutional Review Board/Ethics Committee of the University Hospital of Larissa, Greece (approval reference number: N◦ 13463). All athletes provided written informed consent, in accordance with the Helsinki declaration.Table 1 Athletes’ characteristics. Data are expressed as mean ± standard deviation (SD). Table 1 Post-COVID-19 Healthy p value Age years 25.2 ± 4.1 25.1 ± 4.4 0.971 Body mass kg 75.0 ± 7.6 77.7 ± 7.4 0.257 Body mass index kg·m-2 23.2 ± 1.8 23.9 ± 1.3 0.147 Body surface area m2 1.9 ± 0.2 2.0 ± 0.3 0.333 Body fat % 11.8 ± 3.4 10.8 ± 4.5 0.448 Muscle mass kg 62.4 ± 4.3 64.1 ± 3.7 0.174 Lean body mass kg 59.3 ± 4.2 60.6 ± 4.7 0.388 Total body water % 63.0 ± 3.0 62.4 ± 3.4 0.574 Δchest cm 7.0 ± 1.5 6.6 ± 1.5 0.301 FEV1 % predicted 109.2 ± 4.5 111.7 ± 5.8 0.139 PSQI score 6.1 ± 3.2 3.0 ± 1.7 < 0.001 Abbreviations: % = percentage, % predicted = percent of predicted values, cm: centimeters, COVID-19 = Coronavirus Disease 2019, FEV1 = forced expiratory volume in the 1st second, kg·m-2 = kilogram per square meter, kg = kilograms, m2 = square meter, PSQI = Pittsburg sleep quality index, Δchest = chest circumference difference between maximal inhalation and exhalation. Data collection The study protocol initiated with the assessment of anthropometrical characteristics (i.e. body height (Seca 700, Seca Deutschland, Hamburg, Germany), chest circumference difference between maximal inhalation and exhalation (Δchest, Seca 201, Hamburg, Germany), body composition (Tanita MC-980, Tanita Europe BV, Amsterdam, The Netherlands) and calculated the percentage of body fat (from seven skinfold points measurement, Harpenden, Baty International Ltd, Burgess Hill, UK)12 and body surface area according to Mosteller's formula.13 All participants underwent standard spirometry and lung volume measurements in the sitting position using the MasterScreen-CPX pneumotachograph (VIASYS HealthCare, Germany). For each pulmonary function test, three maximal flow-volume loops were obtained to determine forced expiratory volume in the 1st second (FEV1) according to the American Thoracic Society/European Respiratory Society guidelines.14 Prior to the procedures, all athletes answered the Pittsburgh Sleep Quality Index (PSQI) questionnaire10,15 and it was recorded in their medical history. Cardiopulmonary exercise testing (CPET) was performed on a treadmill (h/p/Cosmos, Nussdorf-Traunstein, Germany). All participants initiated the CPET at a speed of 7 km·h−1. Thereafter the speed of the treadmill increased by 1 km·h−1 every minute until volitional exhaustion was reached. Following CPET, all participants engaged in a 3-min active recovery i.e. walking on the treadmill. Prior to testing, 2-min familiarization sessions were provided for all participants; after the end of the maximal test (start of test with 7 km per min and increase 1 km per min until 18th km per min), they performed a 3-min walking (3 km·h−1) for recovery purposes. Analysis of breathing gases (Fitmate MED Cosmed, Italy) was used for all respiratory parameters while heart rate was recorded via Polar H10 (USA).10 Predicted values for oxygen uptake at peak (V˙O2max) was calculated according to Wasserman et al.'s formula V˙O2max (mL·min-1) = (Height [cm] – Age [years]) x 20 and maximum heart rate (HR) was calculated according the formula (HRmax [bpm] = 207 − 0.7 × Age [years]).16, 17, 18 Each trial was terminated when the participant reached symptom-limited maximum exercise, which was confirmed by the presence of respiratory exchange ratio (RER) > 1.10, HR ≥ 80% of predicted HRmax, and/or plateau of oxygen consumption with increasing workload.19 A sample of 0.5 ​μL of peripheral blood taken from the fingertip was drawn from each participant before, at the end and the 1st minute of recovery after the CPET for the evaluation of blood lactate levels. Blood lactate concentrations were evaluated with enzymatic amperometry detection method (Lactate Scout+, EKF diagnostic, Leipzig, Germany). All sessions were performed at The Medical Project Center (Larissa, Greece), with the environmental temperature at (22.1 ± 1.1) °C and humidity at (32.6% ± 4.1%). The evaluation of patients was performed between 11:00 a.m. to 1:00 p.m. Statistical analysis Data are presented as mean ± standard deviation (SD) and percentage (%) where appropriate. Data normality was assessed via the Kolmogorov-Smirnov One Sample test. Independent Samples t-Test was used to assess differences between groups (post-COVID-19 versus healthy controls). For all tests, a p-value of < 0.05 was considered statistically significant. The IBM SPSS 21 statistical package (SPSS inc., Chicago, Illinois, USA) was used for all statistical analyses. Results Table 1 presents athletes’ characteristics while the results of respiratory parameters during CPET are presented in Fig. 1, Fig. 2, Fig. 3. HR in the maximal effort was significantly different between the groups. COVID-19 athletes demonstrated significantly lower HR during maximal effort ([191.6 ± 7.8] bpm versus [196.6 ± 8.6] bpm, t[38] = −2.120, p = 0.041, Fig. 4) compared to the healthy group. However, mean arterial blood pressure did not reveal significant differences between groups during resting, maximal effort, or the 1st min of recovery (p > 0.05).Fig. 1 Ventilation alteration in cardiopulmonary exercise testing during resting phase, main test and recovery, between the groups. ∗p < 0.05. Abbreviations: COVID-19 = Coronavirus Disease 2019, km·h−1 = kilometers per hour, L·min−1 = liters per minutes. Fig. 1 Fig. 2 Breathing reserve alteration in cardiopulmonary exercise testing during resting phase, main test and recovery, between the groups. ∗p < 0.05, #p < 0.001. Abbreviations: % = ventilation/maximal voluntary ventilation ratio, COVID-19 = Coronavirus Disease 2019, km·h−1 = kilometers per hour. Fig. 2 Fig. 3 Ventilation to body surface area ratio alteration in cardiopulmonary exercise testing during resting phase, main test and recovery, between the groups. ∗p < 0.05. Abbreviations: % = ventilation/body surface area ratio, COVID-19 = Coronavirus Disease 2019, km·h−1 = kilometers per hour. Fig. 3 Fig. 4 Heart rate alteration in cardiopulmonary exercise testing during resting phase, maximal effort and recovery, between the groups. ∗p < 0.05. Abbreviations: bpm = beats per minutes, COVID-19 = Coronavirus Disease 2019. Fig. 4 Blood lactate concentration was also significantly different between the groups. In specific, post-COVID-19 athletes showed higher values of blood lactate concentration in resting ([1.8 ± 0.8] mmol·L-1 versus [0.9 ± 0.1] mmol·L-1, t[38] = −4.695, p < 0.001, Fig. 5), during maximal effort ([11.0 ± 1.6] mmol·L-1 versus [9.8 ± 1.2] mmol·L-1, t[38] = 2.742, p = 0.009, Fig. 5) and the 1st min of recovery ([10.0 ± 1.6] mmol·L-1 versus [8.9 ± 1.2] mmol·L-1, t[38] = 2.441, p = 0.019, Fig. 5) compared to the healthy group.Fig. 5 Blood lactate concentration alteration during the cardiopulmonary exercise testing between the groups. #p ​< ​0.001, ∗p < 0.05. Abbreviations: COVID-19 = Coronavirus Disease 2019, mmol·L-1 = millimoles per liter. Fig. 5 Oxygen uptake was not significantly different between groups in the resting condition (post-COVID-19: [5.4 ± 0.6] ml·min−1·kg−1 versus Healthy [5.1 ± 0.8] ml·min−1·kg−1, p > 0.05, Fig. 6): or maximal effort (post-COVID-19: [55.7 ± 4.4] ml·min−1·kg−1, 131.4% ± 9.3% of predicted versus Healthy: [55.4 ± 4.6] ml·min−1·kg−1, 131.8% ± 9.0% of predicted, p > 0.05, Fig. 6), between groups.Fig. 6 Oxygen uptake differentiation during cardiopulmonary exercise testing between the groups. Abbreviations: COVID-19 = coronavirus disease 2019, ml·min−1·kg−1 = milliliter per minute to kilograms. Fig. 6 Sleep quality, as assessed by PSQI, revealed significant differences between the athlete groups with COVID-19 survivors demonstrating higher values compared to the healthy group (post-COVID-19: [6.1 ± 1.7] score versus Healthy: [3.0 ± 1.7] score, t[38] = 3.814, p < 0.001). Oxygen saturation, self-assessed dyspnea, and leg fatigue also did not show significant differences between groups. Discussion To our knowledge, this is the first study to investigate differences in respiratory and metabolic parameters in post-COVID-19 athletes. Our study has indicated that, despite the non-significant differences in performance, it becomes evident that athletes surviving COVID-19 exhibit significant respiratory and musculature strain during exercise. Despite mild illness, these athletes display significant aerometric burdens, in order to achieve the same training performance, in contrast to non-infected ones. This may be due to the fact that COVID-19 transitions to long-COVID-19 regardless of the severity of the illness.20 Even in athletes, who are considered to have greater cardiorespiratory system capacity when compared to their age-matched sedentary controls, studies have reported a regression in the onset of the aerobic threshold, as well as lower V˙O2max as a result of COVID-19 infection.21,22 In our study, despite the relatively equal V˙O2max in both groups, increased respiratory work at both rest and maximum effort as well as hyperventilation during exercise, were documented. CPET was integrated with spirometry to offer a deeper understanding of lung function. Moreover, blood lactate concentration was found significantly increased in post-COVID-19 athletes. In buffering potential hypoxia, hyperventilation is expected to occur at the expense of CO2. However, as relevant studies show, diffusing capacity of the lungs for carbon monoxide is impaired in post-COVID-19 patients.23 As the physiological process of gas exchange is hindered, hypoxia is perpetuated resulting in the recruitment of the anaerobic metabolism. The anaerobic energy pathways have higher rates of adenosine triphosphate production, but a smaller amount of total adenosine triphosphate production, compared to the aerobic ones.24 Anaerobic metabolism yields excessive levels of blood lactate concentration, disproportionately to pyruvate's (lactate/pyruvate ratio > 10).25 Therefore, we hypothesize that there exists a systematic deceleration in O2 utilization, amenable to impaired gas exchange capacity, secondary to SARS-CoV-2 infection. Hypoxia depends on two elements: tissue-level oxidative metabolism and the supply of oxygen in the circulation (hyperventilation and tachycardia). As mentioned above, the shift to dominant-energy pathways via aerobic metabolism is delayed, while ventilatory response amplified metabolic demands and stress in post-infected athletes. However, the cardiovascular response did not yield similar differences between the two groups. One would expect that heart rate would be concomitantly increased, in line with ventilation. In a recent study, post-COVID-19 cases of previously hospitalized patients exhibited significantly increased heart rates during exercise assessment.11 Athletes with increased cardiac capacity, along with no previous history of cardiovascular pathology that could undermine ejection fraction (e.g. infection), may, indeed, require screening before returning to their prior training program. Małek et al.26 demonstrated that in a small proportion of professional post-COVID-19 athletes with mild or even asymptomatic infection, non-specific cardiac abnormalities could be identified by magnetic resonance imaging. Despite the overall low risk for cardiac involvement,27 engaging in competitive sports increases the risk of fatality and as such, guidelines for the safe return of athletes after COVID-19 infection have been published.28 It is worth noting that screening for such abnormalities should be streamlined with a personalized rehabilitation regimen for post-COVID-19 patients. Due to local restrictive legislation for COVID-19, medicine has shifted towards rehabilitation remotely supervised or even unsupervised.11,29 A consequence of the harmful effects of the SARS-CoV-2 infection, in conjunction with the aforementioned cardiorespiratory complications, is sleep disturbances. In the long-COVID-19 setting, athletes report higher scores in PSQI, suggesting the presence of persistent underlying sleep disorders. Sleep quality is essential for maximal performance, as it has been implicated in cognitive implications.30 Sleep-deprived athletes are prone to both injuries and affected perceptual ability with slower reaction times.10,31 It becomes obvious that long-COVID-19 manifests as a chronic burden, in which non-invasive means, like exercise, could be proved beneficial.32 Rehabilitation is advisable to extend to sleep hygiene intervention in the setting of holistic approaches. Limitations, strengths, and context Our study should be interpreted within the context of its potential limitations. The study included solely soccer athletes, whose sport combines the capability to perform in aerobic conditions for prolonged periods of time. The context and potential limitation here is that this conditioning provides a reserve against hypoxia and other noxious effects on stamina and oxygenation affected by COVID-19. The latter was reflected in our findings, indicating increased respiratory work at rest and maximum effort as well as hyperventilation. These effects would impact the performance of athletes in similar sports, and less so than others. Another caveat stemming from this is that other athletes from sports that require strength such as weightlifting would not be represented by our population. Another potential limitation is that omicron was associated with less severe respiratory illness, and thus cannot account for COVID-19 survivors infected with other variants. As a final potential limitation, the higher PSQI scores may indicate that sleep disturbances were an intermediate step in retracting from the athletes’ maximum capabilities, and thus may not be a specific effect of SARS-CoV-2. Conclusion In conclusion, a phenotype of post-COVID-19 implications was outlined in mild cases of previously infected athletes. The post-COVID-19 pattern was characterized by increased respiratory work at both rest and maximum effort as well as hyperventilation during exercise, which may have increased metabolic needs and mechanical stress. Such implications are not benign and require a carefully curated rehabilitation program, which could take into consideration principally the hindered oxygen supply, as well as the asymptomatic cases of myocarditis, which are gradually revealed in the post-COVID-19 era. Submission statement All authors have read and agree with manuscript content. In addition, as long as this manuscript is being reviewed for this journal, it will not be submitted elsewhere for review and publication. Ethical approval statement All participants provided informed consent, and the study's protocol was approved by the Institutional Review Board/Ethics Committee of the University Hospital of Larissa, Greece (approval reference number: N◦ 13463). Authors’ contributions VTS, IGF, GSM, KK and DK collected the data, VTS ran statistical analyses, VTS, KA, GDV, IGF and GSM drafted the manuscript and ZD, KIG and GB supervised the whole protocol. All authors reviewed the paper and agreed on the final form of submission. Funding The current research protocol was not financially supported. Conflict of interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements We thank all the participants from the Medical Project, Prevention, Evaluation and Recovery Center, for volunteering in the current research protocol. ==== Refs References 1 Gebru A.A. Birhanu T. Wendimu E. Global burden of COVID-19: situational analyis and review Hum Antibodies 29 2 2021 139 148 10.3233/HAB-200420 32804122 2 Deer R.R. Rock M.A. Vasilevsky N. Characterizing long COVID: deep phenotype of a complex condition EBioMedicine 74 2021 103722 10.1016/j.ebiom.2021.103722 3 Carfì A. Bernabei R. Landi F. Gemelli against COVID-19 post-acute care study group. Persistent symptoms in patients after acute COVID-19 JAMA 324 6 2020 603 605 10.1001/jama.2020.12603 32644129 4 Stavrou V.T. Vavougios G.D. Boutlas S. Physical fitness differences, amenable to hypoxia-driven and sarcopenia pathophysiology, between sleep apnea and COVID-19 Int J Environ Res Publ Health 19 2 2022 669 10.3390/ijerph19020669 5 Vavougios G.D. Stavrou V.T. Papayianni E. Investigating the prevalence of cognitive impairment in mild and moderate COVID-19 patients two months post-discharge: associations with physical fitness and respiratory function Alzheimers Dement 17 suppl 6 2021 e057752 10.1002/alz.057752 6 Greenhalgh T. Knight M. A'Court C. Buxton M. Husain L. Management of post-acute covid-19 in primary care BMJ 370 2020 m3026 10.1136/bmj.m3026 32784198 7 Nalbandian A. Sehgal K. Gupta A. Post-acute COVID-19 syndrome Nat Med 27 4 2021 601 615 10.1038/s41591-021-01283-z 33753937 8 Woods J.A. Hutchinson N.T. Powers S.K. The COVID-19 pandemic and physical activity Sports Med Health Sci 2 2 2020 55 64 10.1016/j.smhs.2020.05.006 34189484 9 Hull J.H. Wootten M. Moghal M. Clinical patterns, recovery time and prolonged impact of COVID-19 illness in international athletes: the UK experience Br J Sports Med 56 1 2022 4 11 10.1136/bjsports-2021-104392 34340972 10 Stavrou V.T. Astara K. Daniil Z. The reciprocal association between fitness indicators and sleep quality in the context of recent sport injury Int J Environ Res Publ Health 17 13 2020 4810 10.3390/ijerph17134810 11 Stavrou V.T. Tourlakopoulos K.N. Vavougios G.D. Eight weeks unsupervised pulmonary rehabilitation in previously hospitalized of SARS-CoV-2 Infection J Personalized Med 11 8 2021 806 10.3390/jpm11080806 12 Santos D.A. Dawson J.A. Matias C.N. Reference values for body composition and anthropometric measurements in athletes PLoS One 9 5 2014 e97846 10.1371/journal.pone.0097846 13 Mosteller R.D. Simplified calculation of body-surface area N Engl J Med 317 17 1987 1098 10.1056/NEJM198710223171717 3657876 14 Miller M.R. Hankinson J. Brusasco V. Standardisation of spirometry Eur Respir J 26 2 2005 319 338 10.1183/09031936.05.00034805 16055882 15 Stavrou V. Vavougios G.D. Bardaka F. Karetsi E. Daniil Z. Gourgoulianis K.I. The effect of exercise training on the quality of sleep in national-level adolescent finswimmers Sports Med Open 5 1 2019 34 10.1186/s40798-019-0207-y 31392589 16 Dafoe W. Principles of exercise testing and interpretation Can J Cardiol 23 4 2007 274 17 Stavrou V. Boutou A.K. Vavougios G.D. The use of cardiopulmonary exercise testing in identifying the presence of obstructive sleep apnea syndrome in patients with compatible symptomatology Respir Physiol Neurobiol 262 2019 26 31 10.1016/j.resp.2019.01.010 30684645 18 Tanaka H. Monahan K.D. Seals D.R. Age-predicted maximal heart rate revisited J Am Coll Cardiol 37 1 2001 153 156 10.1016/s0735-1097(00)01054-8 11153730 19 American Thoracic Society;American College of Chest Physicians ATS/ACCP Statement on cardiopulmonary exercise testing Am J Respir Crit Care Med 167 2 2003 211 277 10.1164/rccm.167.2.211 12524257 20 Townsend L. Dowds J. O'Brien K. Persistent poor health after COVID-19 is not associated with respiratory complications or initial disease severity Ann Am Thorac Soc 18 6 2021 997 1003 10.1513/AnnalsATS.202009-1175OC 33413026 21 Anastasio F. LA Macchia T. Rossi G. Mid-term impact of mild-moderate COVID-19 on cardiorespiratory fitness in élite athletes J Sports Med Phys Fit 62 10 2022 1383 1390 10.23736/S0022-4707.21.13226-8 22 Vonbank K. Lehmann A. Bernitzky D. Predictors of prolonged cardiopulmonary exercise impairment after COVID-19 Infection: a prospective observational study Front Med 8 2021 773788 10.3389/fmed.2021.773788 23 Fuschillo S. Ambrosino P. Motta A. Maniscalco M. COVID-19 and diffusing capacity of the lungs for carbon monoxide: a clinical biomarker in postacute care settings Biomarkers Med 15 8 2021 537 539 10.2217/bmm-2021-0134 24 Sahlin K. Tonkonogi M. Söderlund K. Energy supply and muscle fatigue in humans Acta Physiol Scand 162 3 1998 261 266 10.1046/j.1365-201X.1998.0298f.x 9578371 25 Alberti K.G. The biochemical consequences of hypoxia J Clin Pathol Suppl 11 1977 14 20 10.1136/jcp.s3-11.1.14 26 Małek Ł.A. Marczak M. Miłosz-Wieczorek B. Cardiac involvement in consecutive elite athletes recovered from COVID-19: a magnetic resonance study J Magn Reson Imag 53 6 2021 1723 1729 10.1002/jmri.27513 27 van Hattum J.C. Spies J.L. Verwijs S.M. Cardiac abnormalities in athletes after SARS-CoV-2 infection: a systematic review BMJ Open Sport Exerc Med 7 4 2021 e001164 10.1136/bmjsem-2021-001164 28 Greene D.N. Wu A.H.B. Jaffe A.S. Return-to-play guidelines for athletes after COVID-19 infection JAMA Cardiol 6 4 2021 479 10.1001/jamacardio.2020.5348 29 Wang T.J. Chau B. Lui M. Lam G.T. Lin N. Humbert S. Physical medicine and rehabilitation and pulmonary rehabilitation for COVID-19 Am J Phys Med Rehabil 99 9 2020 769 774 10.1097/PHM.0000000000001505 32541352 30 Astara K. Siachpazidou D. Vavougios G.D. Sleep disordered breathing from preschool to early adult age and its neurocognitive complications: a preliminary report Sleep Sci 14 Spec 2 2021 140 149 10.5935/1984-0063.20200098 35082983 31 Stavrou V.T. Astara K. Tourlakopoulos K.N. Sleep quality's effect on vigilance and perceptual ability in adolescent and adult athletes J Sports Med 2021 2021 5585573 10.1155/2021/5585573 32 Anderson E. Durstine J.L. Physical activity, exercise, and chronic diseases: a brief review Sports Med Health Sci 1 1 2019 3 10 10.1016/j.smhs.2019.08.006 35782456
PMC009xxxxxx/PMC9856796.txt
==== Front J Cell Biol J Cell Biol jcb The Journal of Cell Biology 0021-9525 1540-8140 Rockefeller University Press 36656648 jcb.202301035 10.1083/jcb.202301035 Spotlight Mammalian fertilization: Does sperm IZUMO1 mediate fusion as well as adhesion? Mammalian fertilization https://orcid.org/0000-0001-8124-7328 Bianchi Enrica 1 https://orcid.org/0000-0003-0537-0863 Wright Gavin J. 1 1 Department of Biology, Hull York Medical School, York Biomedical Research Institute, University of York , York, UK Correspondence to Enrica Bianchi: [email protected] Gavin J. Wright: [email protected] 06 2 2023 19 1 2023 222 2 e202301035© 2023 Bianchi and Wright 2023 Bianchi and Wright https://creativecommons.org/licenses/by-nc-sa/4.0/ http://www.rupress.org/terms/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms/). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 4.0 International license, as described at https://creativecommons.org/licenses/by-nc-sa/4.0/). Brukman and colleagues provide evidence that the sperm surface protein IZUMO1, which is essential for mammalian fertilization, can induce membrane fusion in cultured cells. The molecular mechanism of sperm–egg fusion is a long-standing mystery in reproduction. Brukman and colleagues (2022. J. Cell Biol. https://doi.org/10.1083/jcb.202207147) now provide evidence that the sperm surface protein IZUMO1, which is essential for mammalian fertilization, can induce membrane fusion in cultured cells. ==== Body pmcThe basic genetic principles of sexual reproduction are remarkably constant across a wide range of different organisms: diploid cells partition their genetic material into two haploid cells that, at some stage of their lifecycle, are segregated into different mating types or sexes which recognize one another and fuse at fertilization to recreate a single diploid cell (1). In mammals, the haploid cells differ significantly in their size and shape: the female egg is large and spherical, whereas the male sperm is small, elongated, and highly motile. For successful fertilization, these two cells must first recognize and adhere to each other before finally fusing. For many mammals—including humans—fertilization can successfully be performed in vitro, meaning that we have a good cellular description of this process; however, we are only just learning which molecules are involved and what their biological roles are. IZUMO1 was the first sperm cell surface protein shown to be essential for mammalian fertilization and importantly passed the stringent in vivo test of demonstrating that male mice lacking a functional Izumo1 gene were infertile (2). Prior to this, targeted knock out of sperm cell surface proteins that had been suggested to be required for fertilization resulted in male mice that remained fertile (3). In the case of IZUMO1 knock outs, the sperm were ostensibly normal: they looked and moved like wild-type sperm but were unable to finally fuse and fertilize the egg. Since this founding discovery, the molecular mechanism of IZUMO1 function as either mediating adhesion between the sperm and egg membranes or actively driving the fusion of the membranes has been a topic of debate. These two processes are likely to be distinct and ordered: adhesion must first align areas of apposing gamete plasma membranes to within around 10 nm, followed by the action of a fusogen to overcome the thermodynamic barrier that normally prevents inappropriate membrane fusion between neighboring cells (4). Perhaps surprisingly, experimentally distinguishing between an adhesive and fusogenic role is not trivial: they require specifically designed assays and their interdependence means that even when only adhesion is defective, cellular fusion will not occur. It was conceivable that IZUMO1 could act independently of other factors because some fusogens—such as those used by viruses—are able to act in a unidirectional manner meaning they only need to be present on one of the two fusing cells. By contrast, proteins mediating adhesion usually require a specific binding partner displayed on the apposing cell; however, identifying these extracellular interactions which are typified by very weak binding affinities is technically challenging (5). Clues that IZUMO1 had an adhesive role came when a receptor displayed on eggs called JUNO was identified (6). Confirmatory studies have shown that both mice and rats lacking IZUMO1 produce normal-looking sperm that are unable to bind eggs (7). There is evidence, however, that IZUMO1 could have additional roles other than just JUNO binding (8). Could IZUMO1 also function as a fusogen? In this issue, Brukman and colleagues showed that IZUMO1 could act directly as a fusogen in the Baby Hamster Kidney fibroblast (BHK) cell line (9). When IZUMO1 was overexpressed in BHKs, they observed cells with two or more nuclei, an effect that could not be ascribed to a failure in cytokinesis, and they also confirmed that cell fusion occurred using a content-mixing assay. The level of fusion events was low but comparable with that obtained with the established fusogen HAP2. Using structure-guided mutagenesis, the authors separated the adhesive and fusogenic activities of IZUMO1: the single mutant W148A prevented binding to JUNO but did not affect fusion whereas the triple mutant (F28A, W88A, and W113A) retained the ability to bind JUNO but did not induce fusion. These data indicated that the adhesive and fusogenic roles are performed by two distinct and functionally separable regions of the IZUMO1 protein. While the results obtained with the hamster cell line were clear, the picture became more complex when cells and gametes were mixed in fusion assays. The authors found that sperm fused only to JUNO-expressing BHK cells, indicating that the adhesion step driven by JUNO-IZUMO binding was required and that IZUMO1 was not sufficient to induce cell fusion in this assay. Remarkably, while IZUMO1 behaved as a unilateral fusogen in the transfected cells, the authors found that the heterologous expression of the protein was not sufficient to fuse cells and eggs. This is in agreement with previous reports that showed that IZUMO1 ectopically expressed on cell lines adhered to—but did not fuse with—eggs (10). In summary, the mixed cell–gamete assays suggested that JUNO-mediated binding was required for fusion of sperm and cells, but for eggs to fuse with IZUMO1-expressing cells, an additional factor is required. Sperm and eggs are terminally differentiated cells that must fuse with each other, but it is equally important that they do not fuse with other cells in the body. In sperm, this is probably achieved by sequestering IZUMO1 within the cell and making it available at the sperm surface only following the acrosome reaction as the sperm approaches the egg. The extracellular matrix surrounding the egg, the zona pellucida, shields the egg membrane from coming into direct contact with other cells, but it is conceivable that a molecule or perhaps intrinsic feature, such as the thick actin cortex that lines the oolemma, inhibit cell fusion. The successful manipulation of murine and human gametes in vitro in the 1960s and 1970s (11) brought the promise of quick discoveries, but the dissection of the molecular mechanisms of fertilization has proven more difficult than anticipated (5). While we can biochemically investigate the interactions of cell surface proteins, studying their mechanisms of action in the context of the cell membrane remains challenging. Here, Brukman and colleagues use content mixing and multinucleation as a proxy for measuring cell fusion; consequently, fusion events could be the result of other phenomena such as endocytosis, trogocytosis, nondisjunction of sister cells, and transfer of cargo molecules via special structures like tunneling nanotubes (12, 13). The development of better techniques to monitor cell adhesion, cell fusion, and the dynamics of membrane proteins in real time will shed further light on the function of the sperm proteins during fertilization (14). Crucially, the lack of a suitable cell line for mammalian gametes represents a major limitation within the field meaning that transgenic animals remain necessary to confirm the validity of observations obtained from in vitro assays. For example, the generation of an IZUMO1 transgenic mouse line carrying the triple mutation that retained JUNO binding but lost fusogenic activity would provide important data to support the new role proposed for IZUMO1. The work from Brukman and colleagues represents a valuable contribution to untangle the molecular mechanisms of fertilization. By unveiling a novel role of the sperm protein IZUMO1, it prompts the deployment of similar approaches to better understand the function of other essential sperm surface proteins such as the structurally similar SPACA6 and TMEM95 (15, 16) and to understand the mechanisms of cell fusion in somatic cells such as skeletal muscle and the placenta. ==== Refs References 1 Bianchi, E., and G.J. Wright. 2016. Annu. Rev. Genet. 10.1146/annurev-genet-121415-121834 2 Inoue, N., . 2005. Nature. 10.1038/nature03362 3 Okabe, M. 2015. Asian J. Androl. 10.4103/1008-682X.153299 4 Hernández, J.M., and B. Podbilewicz. 2017. Development. 10.1242/dev.155523 5 Wright, G.J., and E. Bianchi. 2016. Cell Tissue Res. 10.1007/s00441-015-2243-3 6 Bianchi, E., . 2014. Nature. 10.1038/nature13203 7 Matsumura, T., . 2022. Front. Cell Dev. Biol. 10.3389/fcell.2021.810118 8 Inoue, N., . 2015. Nat. Commun. 10.1038/ncomms9858 9 Brukman, N.G., K.P. Nakajima, C. Valansi, K. Flyak, X. Li, T. Higashiyama, and B. Podbilewicz. 2023. J. Cell Biol. 10.1083/jcb.202207147 10 Noda, T., . 2020. Proc. Natl. Acad. Sci. USA. 10.1073/pnas.1922650117 11 Bavister, B.D. 2002. Reproduction. 10.1530/rep.0.1240181 12 Dagar, S., . 2021. Biochem. J. 10.1042/BCJ20210077 13 Rustom, A., . 2004. Science. 10.1126/science.1093133 14 Nakajima, K.P., . 2022. Sci. Rep. 10.1038/s41598-022-13547-w 15 Tang, S., . 2022. Proc. Natl. Acad. Sci. USA. 10.1073/pnas.2207805119 16 Vance, T.D.R., . 2022. Commun. Biol. 10.1038/s42003-022-03883-y
PMC009xxxxxx/PMC9893807.txt
==== Front Sports Med Health Sci Sports Med Health Sci Sports Medicine and Health Science 2666-3376 Chengdu Sport University S2666-3376(23)00005-7 10.1016/j.smhs.2023.01.005 Opinion Sport and physical exercise in sustainable mental health care of common mental disorders: Lessons from the COVID-19 pandemic Lange Klaus W. [email protected] ∗ Nakamura Yukiko Reissmann Andreas Faculty of Human Sciences (Psychology, Education and Sport Science), University of Regensburg, Regensburg, Bavaria, Germany ∗ Corresponding author. University of Regensburg, 93040 Regensburg, Germany. [email protected] 02 2 2023 6 2023 02 2 2023 5 2 151155 2 9 2022 25 1 2023 29 1 2023 © 2023 Chengdu Sport University. Publishing services by Elsevier B.V. on behalf of KeAi Communications Co. Ltd. 2023 Chengdu Sport University https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). The large-scale disruptions to physical activity during the coronavirus pandemic have been found to be a leading predictor of common mental disorders. In addition, regular physical exercise has been found to alleviate anxiety, sadness and depression during the pandemic. These findings, together with numerous studies published before the pandemic on the effects of physical activity on mental health, should be considered in the provision of mental health care following the pandemic. Cross-sectional research has revealed that all types of exercise and sport are associated with a reduced mental health burden. Therefore, the effectiveness of exercise and sport participation in sustainable mental health care as well as the causal relationship between exercise, psychosocial health and common mental disorders merit further investigation. Physical activity and sport, with their global accessibility, significant and clinically meaningful efficacy as well as virtual absence of adverse effects, offer a promising option for the promotion of mental health, including the prevention and treatment of common mental disorders. Physical exercise and sport are likely to become valuable public mental health resources in the future. Keywords Anxiety Depression Mental health Physical exercise COVID-19 ==== Body pmcAbbreviation COVID-19 Coronavirus disease caused by SARS-CoV-2 virus In addition to serious pulmonary and extrapulmonary manifestations of the infectious coronavirus disease caused by the SARS-CoV-2 virus (COVID-19), the pandemic has resulted in significant mental health effects.1 Public health measures introduced to curb the spread of the coronavirus, such as self-isolation, quarantine, lockdown and social distancing, have restricted physical activity and social contact. As a consequence, many people have experienced a decline in well-being, an increase in psychological stress as well as feelings of isolation, depressed mood and anxiety,2,3 with common mental disorders being major adverse effects.1,4 Common mental disorders, comprising different types of depression and anxiety, cause marked emotional distress and interfere with daily functioning, while they do not usually affect cognition or insight. The mental health effects of the pandemic may shape the health of entire populations for many years to come. Failure to address these mental health issues is likely to prolong their impact. Mental health experts have therefore emphasized the importance of developing approaches to supporting mental health at the population level. Available evidence suggests that dietary factors and physical exercise should be considered in the promotion of mental health and the prevention of mental illness.5 These factors may also contribute to shifting the population distribution of infection risk and preventing severe outcomes of COVID-19.6, 7, 8 For example, a systematic review and meta-analysis of observational studies has found evidence for associations between ultra-processed food consumption and adverse mental health, with greater consumption of ultra-processed food being associated with elevated odds of depressive and anxiety symptoms.9 Furthermore, various nutrients and food bioactive have been proposed as effective measures in the management of common mental disorders.10,11 However, evidence of their clinical efficacy is lacking12,13 and their long-term administration may even be harmful.14 The large-scale disruptions to physical activity during the pandemic have been found to be a leading predictor of depression. At the onset of the pandemic, average steps per day in young adults plunged from around 10 000 to 4 60015 and the proportion of young adults at risk for clinical depression ranged from 46% to 61%, which represented an increase by 90% compared to the pre-pandemic rates.15 Beneficial acute and chronic effects of exercise on the immune system have been demonstrated,16 and regular physical activity has been promoted by the World Health Organization as a non-pharmacological and cost-effective approach to supporting mental health during the COVID-19 pandemic.6,17 Empirical evidence suggests that regular exercise is effective in maintaining or improving well-being and mental health18, and exercise has also been found to alleviate anxiety, sadness and depression during the coronavirus pandemic.19,20 This, together with the large number of studies published before the pandemic on the effects of physical activity on mental health,21, 22, 23, 24, 25 should be considered in the provision of mental health care following the pandemic. The enhancement of physical health affected by exercise has been widely investigated, and regular exercise has been shown to be a key factor in good health.26 Physically active individuals have been found to benefit from improved levels of health-related fitness, reduced risks of developing a wide range of disabling chronic conditions, including obesity, diabetes, cardiovascular disease, stroke and cancer, and a decrease in overall mortality.27 The health benefits of exercise have also been widely acknowledged to encompass mental health. A cross-sectional study of more than 1.2 million adults in the United States, matched for various sociodemographic and physical health factors, found that people who exercised reported 1.49 (43.2%) fewer days of poor mental health in the past month than people who did not exercise, with all types of exercise and sport being associated with a reduced mental health burden.28 The largest significant associations were for periods of exercise lasting 45 min and exercise performed 3–5 times per week as well as for popular team sports, cycling, aerobic and gym activities.28 Engaging in sports involving interaction with other people may improve social confidence and skills and thus promote positive interpersonal relationships and psychosocial development. While the association between physical activity and mental health is well established, causation remains a matter of debate. Unmeasured factors closely related to physical activity may play a role. In view of the social context of physical activity, the interplay between exercise and social interactions or other lifestyle habits may impact mental health. Disturbances of both mental well-being and physical activity may be influenced by a third variable, such as people's response to the pandemic. Individuals capable of maintaining their lifestyle during the pandemic may be better able to preserve their mental health and well-being. Before the pandemic, these individuals may have been more resistant to stress and less prone to depressed mood or anxiety.29 Reverse causation is also possible, with physical activity being driven by mental health rather than lifestyle. For example, changes in lifestyle could be early signs of depression. Establishing a causal relationship between exercise and mental well-being and health requires randomized controlled trials demonstrating the beneficial effects of exercise interventions. Depression has become the leading cause of the burden of disability globally.30 Since the ability of pharmacological and psychological therapies to mitigate the cumulative burden of depression on society is limited,31 there is an urgent need to find modifiable factors that could become targets of health campaigns at the population level. Since adults with major depressive disorder have been reported to engage in low levels of physical activity and high levels of sedentary behavior,32 lifestyle interventions targeting the adoption and maintenance of exercise may be warranted. The debate surrounding the effects of physical exercise on depression has often been contentious, leading to uncertainty regarding the magnitude of these effects. Meta-analyses of randomized controlled trials have reported moderate to large antidepressant efficacy of exercise.33, 34, 35, 36 The varying effect sizes reported may have been influenced by inclusion criteria, heterogeneity of samples and diagnostic criteria as well as publication bias. Furthermore, various studies compared exercise in addition to an established therapy versus an established therapy alone.33 This approach assumes that the efficacies of treatments can be added and subtracted algebraically. However, this assumption does not take into account that exercise may at least partially overlap and interact with the mechanisms underlying pharmacological and psychological therapies of depression.37 In a meta-analysis adjusting for publication bias, large antidepressant effects of exercise on depression were found in comparison with non-active control conditions.24 The effect was particularly high for studies including people with major depressive disorder. Larger effect sizes were also found in patients with a diagnosis of depression but without other clinical co-morbidities. In the context of the high risk of social isolation in older individuals with depression,38 group exercise appears to be particularly effective. Future studies should therefore address exercise as a means of promoting social interaction. The question of whether exercise in combination with a conventional treatment produces better results than exercise or the other treatment alone remains at present a matter of debate owing to the limited data available. Furthermore, the currently established meta-analytic evidence showing the antidepressant efficacy of exercise in clinical settings, with relatively low attrition in people with major depressive disorder, needs to be complemented by randomized controlled trials providing ecological (i.e. real-life) evidence. These trials should examine the effectiveness of exercise interventions, defined as benefits produced in daily clinical practice, using wide inclusion criteria and functional health outcomes, such as quality of life. Exercise programs have also been proposed as a promising means of protecting against the development of depressive symptoms in young people,39 which may be particularly helpful since they are non-stigmatizing and have few side effects.40 Anxiety and depressive symptoms in adolescents have been found to be higher for low physical activity compared to moderate and high physical activity.41 Furthermore, some evidence from randomized controlled trials suggests a moderate positive effect of exercise interventions on the severity of adolescent depression.42,43 Anxiety disorders are the most prevalent mental disorders in the general population44 and have a significant societal and economic impact.45 Epidemiological findings suggest that individuals who are more physically active may be less likely to suffer from anxiety disorders.41,46 Current evidence, presented in several systematic reviews and meta-analyses, suggests that engaging in physical activity and sports may protect against anxiety symptoms and disorders across the lifespan.23,47, 48, 49, 50, 51, 52 Higher-intensity exercise regimens may be more effective than low-intensity exercise.49 Physical exercise as an add-on to cognitive behavioral psychotherapy appears to be feasible and beneficial, especially when performed regularly several times per week over extended periods of time.52 While exercise may be a viable option for the therapy of anxiety, the evidence for its effectiveness in anxiety disorders is equivocal and limited by a lack of data from methodologically rigorous randomized controlled trials with large sample sizes. Major limitations of the available studies include non-representativeness of samples, inadequate assessment of fitness and adherence levels, inconsistent adjustment for putative confounders, attrition bias and issues regarding exercise exposure and outcome measures. The available evidence is therefore insufficient to warrant a recommendation of exercise as a treatment in people with anxiety disorders. Nevertheless, in view of preliminary findings suggesting exercise as a useful, accessible and affordable therapeutic option for anxiety, further high-quality, long-term randomized controlled trials using robust experimental designs are called for. The lack of large samples and statistical power does not allow for the examination of exercise form, frequency, duration or intensity. It would be important to know whether certain types of exercise have advantages over others or whether an individual's favored sport can be chosen. Mental health benefits of exercise and sport may be related more to the context of physical activity than the behavior of being active itself.53, 54, 55, 56 For example, participation in team sports and informal group activities has been reported to be inversely associated with depressive symptoms in comparison with individual physical activity.57 Furthermore, team sports appear to be associated with positive mental health regardless of the volume of physical activity.57 This suggests that features of the team sport context provide benefits to mental health and that increasing the volume of physical activity has no beneficial effects. The context of team sport is thought to be beneficial to mental health since it offers an opportunity for social interaction, thereby strengthening social networks and perceived feelings of support and integration.57, 58, 59 An umbrella review of systematic reviews and meta-analyses found a small positive impact of participation in organized sports on mental health in children and adolescents.60 This was not related to any specific type of organized sport activity. The findings of a recent study from Canada using a large and diverse sample of high school students suggest that sport participation should be promoted for the prevention and management of adolescent mental illness.61 Students participating in varsity sports were found to have lower symptoms of anxiety and depression in comparison with nonparticipants, with associations being strongest in those who participated in both varsity sports and sports outside school. Another recent study including data from over 11 000 children and adolescents in the United States reported that participation in team sports compared to no sport participation was associated with fewer mental health difficulties, while participation in exclusively individual sports was associated with greater difficulties.62 These findings suggest that team-based organized sports may be a means to support youth mental health. However, given the cross-sectional design of the study, the causal relationship between participation in organized sport activities and mental health remains to be established. The physiological and biochemical mechanisms discussed to be involved in mediating the effects of exercise on mental health include, among others, endorphins,63 myokines,64 inflammation65 and stress modulated by the hypothalamic-pituitary-adrenal axis. The impact of stressful life events on mental health in general and major depressive episodes, in particular, has been firmly established.66,67 An increase in the prevalence of stress, anxiety and depression in the general population has been observed during the COVID-19 pandemic.68,69 Exercise interventions may improve the ability to deal with psychological stressors during and after the pandemic, since increased physical activity and fitness levels appear to be associated with an attenuated response to psychosocial stress.70 Exercise may therefore exert its beneficial effects on mental health through stress-buffering mechanisms.71 Furthermore, physical activity seems to be capable of modulating stress-induced changes in DNA methylation and gene expression, with positive epigenetic influences of exercise counteracting the negative influences of stress.72 Changing an inactive lifestyle and increasing physical activity is a challenge for many people. The outcomes of introducing a physical exercise regimen may be improved when various strategies are combined.73,74 Such strategies include starting with short periods of physical activity,75 improving self-regulation,76,77 strengthening non-conscious processes,78 improving accessibility to environments and facilities77,79 and using internet and smartphone apps.80 Maintaining a physical exercise regimen presents a further problem, since adherence to short-term prescribed exercise has been found to drop to 50% within six months.81,82 Failure of practitioners to provide specific exercise recommendations is common.83 Rates of adherence to exercise regimens can be enhanced when written prescriptions provide a detailed exercise plan, and suggestions for overcoming anticipated potential barriers are given.84 The effects of exercise counseling may be facilitated through pre-intervention educational programs, continued motivational support and improving the convenience of exercise facilities and opportunities.85,86 Furthermore, increased drop-out rates have been found in groups undertaking high-intensity training compared to low-intensity programs.87 Given the higher efficacy of high-intensity programs, exercise regimens need to be individually tailored in order to maximize the beneficial effects of exercise while minimizing the drop-out rate. Exercise has long been underestimated as a variable contributing to mental health, and lessons from the current pandemic regarding mental health care include the value of physical exercise and sport in the management of common mental disorders. The findings of numerous studies support the antidepressant and anxiolytic efficacy of physical exercise and sport, indicating their promise as additional or alternative options for managing anxiety and depressive disorders. Physical exercise as a therapeutic option is relatively free from side effects and produces additional benefits for physical health, such as a decrease in blood pressure and weight loss. The potential improvement in mental health resulting from lifestyle changes incorporating increased physical activity should therefore be considered in the therapy regimens of individuals and in population-wide public health programs. Further interventional study of sustainable exercise effects in individuals with depression and anxiety is required to identify the specific types and optimum frequency, intensity and duration of physical activity. The financing of high-quality studies investigating the effects of exercise on mental health will pose challenges, since immediate pecuniary returns from such investigations cannot be expected. However, long-term gains include amelioration of the burden of mental illness and additional physical health benefits in respect of cardiovascular and metabolic diseases. Political interventions, public funding and support through non-profit organizations will be needed to help counter the challenges faced in the fostering and maintenance of public health. In addition to specialist care, task-shifting may be capable of providing cost-effective and sustainable mental health care worldwide, especially where health services are overwhelmed by the pandemic.88 In conclusion, the effectiveness of exercise and sport participation in sustainable mental health care as well as the causal relationship between exercise, psychosocial health and common mental disorders merit further investigation. Physical activity and sport, with their global accessibility, significant and clinically meaningful efficacy as well as the virtual absence of adverse effects, offer a promising, evidence-based option for the prevention and treatment of common mental disorders. Any increase in physical activity should therefore be encouraged by health practitioners in order to improve mental health. Physical exercise and sports are likely to become valuable public mental health resources in the future. Submission statement The work described has not been published previously, is not under consideration for publication elsewhere, its publication is approved by all authors and tacitly or explicitly by the responsible authorities where the work was carried out, and, if accepted, it will not be published elsewhere including electronically in the same form, in English or in any other language, without the written consent of the copyright-holder. Authors' contributions KWL contributed substantially to the conception of the work, drafted the work, provided approval for publication of the content and agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. YN contributed substantially to the conception of the work, revised it critically for important intellectual content, provided approval for publication of the content and agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. AR contributed substantially to the conception of the work, revised it critically for important intellectual content, provided approval for publication of the content and agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. Conflict of interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. ==== Refs References 1 Lange KW. Coronavirus disease 2019 (COVID-19) and global mental health Glob Health J 5 1 2021 31 36 10.1016/j.glohj.2021.02.004 33614179 2 Chtourou H Trabelsi K H’mida C Staying physically active during the quarantine and self-isolation period for controlling and mitigating the COVID-19 pandemic: a systematic overview of the literature Front Psychol 1708 2020 11 10.3389/fpsyg.2020.01708 3 Gupta A Puyat JH Ranote H Vila-Rodriguez F Kazanjian A. A cross-sectional survey of activities to support mental wellness during the COVID-19 pandemic J Affect Disord Rep 100167 2021 5 10.1016/j.jadr.2021.100167 4 Pfefferbaum B North CS. Mental health and the covid-19 pandemic N Engl J Med 383 6 2020 510 512 10.1056/NEJMp2008017 32283003 5 Lange KW. Diet, exercise, and mental disorders - public health challenges of the future Mov Nutr Health Dis 2 2018 39 59 10.5283/MNHD.12 6 Lange KW Nakamura Y. Lifestyle factors in the prevention of COVID-19 Glob Health J 4 4 2020 146 152 10.1016/j.glohj.2020.11.002 33520339 7 Lange KW. Food science and COVID-19 Food Sci Hum Wellness 10 1 2021 1 5 10.1016/j.fshw.2020.08.005 8 Lange KW. The contribution of food bioactives and nutrition to the management of COVID-19 J Future Foods 2 1 2022 13 17 10.1016/j.jfutfo.2022.03.012 9 Lane MM Gamage E Travica N Ultra-processed food consumption and mental health: a systematic review and meta-analysis of observational studies Nutrients 14 13 2022 2568 10.3390/nu14132568 35807749 10 Lange KW. Omega-3 fatty acids and mental health Glob Health J 4 1 2020 18 30 10.1016/j.glohj.2020.01.004 11 Lange KW Lange KM Nakamura Y Reissmann A. The use of vitamin D in depression J Food Bioact 15 2021 13 20 10.31665/JFB.2021.15278 12 Lange KW. Lack of evidence for efficacy of omega-3 fatty acids in depression Acta Psychiatr Scand 144 4 2021 415 416 10.1111/acps.13357 34351615 13 Lange KW Nakamura Y Zhao H Bai D Wang H. Are omega-3 fatty acids efficacious in the treatment of depression? A review J Food Bioact 14 2021 10 19 10.31665/JFB.2021.14265 14 Lange KW Nakamura Y Gosslau AM Li S. Are there serious adverse effects of omega-3 polyunsaturated fatty acid supplements? J Food Bioact 7 2019 1 6 10.31665/JFB.2019.7192 15 Giuntella O Hyde K Saccardo S Sadoff S. Lifestyle and mental health disruptions during COVID-19 Proc Natl Acad Sci U S A 118 9 2021 e2016632118 10.1073/pnas.2016632118 16 Nieman DC Wentz LM. The compelling link between physical activity and the body's defense system J Sport Health Sci. 8 3 2019 201 217 10.1016/j.jshs.2018.09.009 31193280 17 World Health Organization. HealthyAtHome - Physical Activity. World Health Organization. Accessed January 25, 2023. https://www.who.int/news-room/campaigns/connecting-the-world-to-combat-coronavirus/healthyathome/healthyathome--physical-activity. 18 Maugeri G Castrogiovanni P Battaglia G The impact of physical activity on psychological health during Covid-19 pandemic in Italy Heliyon 6 6 2020 e04315 10.1016/j.heliyon.2020.e04315 19 Ai X Yang J Lin Z Wan X. Mental health and the role of physical activity during the COVID-19 pandemic Front Psychol 759987 2021 12 10.3389/fpsyg.2021.759987 20 Wolf S Seiffer B Zeibig J-M Is physical activity associated with less depression and anxiety during the COVID-19 pandemic? A rapid systematic review Sports Med 51 8 2021 1771 1783 10.1007/s40279-021-01468-z 33886101 21 Ströhle A. Physical activity, exercise, depression and anxiety disorders J Neural Transm 116 6 2009 777 784 10.1007/s00702-008-0092-x 18726137 22 Wegner M Helmich I Machado S Nardi AE Arias-Carrion O Budde H. Effects of exercise on anxiety and depression disorders: review of meta- analyses and neurobiological mechanisms CNS Neurol Disord: Drug Targets 13 6 2014 1002 1014 10.2174/1871527313666140612102841 24923346 23 Stonerock GL Hoffman BM Smith PJ Blumenthal JA. Exercise as treatment for anxiety: systematic review and analysis Ann Behav Med 49 4 2015 542 556 10.1007/s12160-014-9685-9 25697132 24 Schuch FB Vancampfort D Richards J Rosenbaum S Ward PB Stubbs B. Exercise as a treatment for depression: a meta-analysis adjusting for publication bias J Psychiatr Res 77 2016 42 51 10.1016/j.jpsychires.2016.02.023 26978184 25 Harvey SB Øverland S Hatch SL Wessely S Mykletun A Hotopf M. Exercise and the prevention of depression: results of the HUNT cohort study Am J Psychiatr 175 1 2018 28 36 10.1176/appi.ajp.2017.16111223 28969440 26 Warburton DER Nicol CW Bredin SSD. Health benefits of physical activity: the evidence CMAJ (Can Med Assoc J) 174 6 2006 801 809 10.1503/cmaj.051351 16534088 27 Lee I-M Shiroma EJ Lobelo F Puska P Blair SN Katzmarzyk PT. Effect of physical inactivity on major non-communicable diseases worldwide: an analysis of burden of disease and life expectancy Lancet 380 9838 2012 219 229 10.1016/S0140-6736(12)61031-9 22818936 28 Chekroud SR Gueorguieva R Zheutlin AB Association between physical exercise and mental health in 1·2 million individuals in the USA between 2011 and 2015: a cross-sectional study Lancet Psychiatr 5 9 2018 739 746 10.1016/S2215-0366(18)30227-X 29 Teychenne M Ball K Salmon J. Physical activity and likelihood of depression in adults: a review Prev Med 46 5 2008 397 411 10.1016/j.ypmed.2008.01.009 18289655 30 Friedrich MJ. Depression Is the leading cause of disability around the world JAMA 317 15 2017 1517 10.1001/jama.2017.3826 31 Casacalenda N Perry JC Looper K. Remission in major depressive disorder: a comparison of pharmacotherapy, psychotherapy, and control conditions Am J Psychiatr 159 8 2002 1354 1360 10.1176/appi.ajp.159.8.1354 12153828 32 Schuch F Vancampfort D Firth J Physical activity and sedentary behavior in people with major depressive disorder: a systematic review and meta-analysis J Affect Disord 210 2017 139 150 10.1016/j.jad.2016.10.050 28033521 33 Cooney GM Dwan K Greig CA Exercise for depression Cochrane Database Syst Rev 9 2013 CD004366 10.1002/14651858.CD004366.pub6 34 Josefsson T Lindwall M Archer T. Physical exercise intervention in depressive disorders: meta-analysis and systematic review Scand J Med Sci Sports 24 2 2014 259 272 10.1111/sms.12050 23362828 35 Kvam S Kleppe CL Nordhus IH Hovland A. Exercise as a treatment for depression: a meta-analysis J Affect Disord 202 2016 67 86 10.1016/j.jad.2016.03.063 27253219 36 Krogh J Hjorthøj C Speyer H Gluud C Nordentoft M. Exercise for patients with major depression: a systematic review with meta-analysis and trial sequential analysis BMJ Open 7 9 2017 e014820 10.1136/bmjopen-2016-014820 37 Honey Ekkekakis P. I shrunk the pooled SMD! Guide to critical appraisal of systematic reviews and meta-analyses using the Cochrane review on exercise for depression as example Ment Health Phys Act 8 2015 21 36 10.1016/j.mhpa.2014.12.001 38 Schuch FB Vancampfort D Rosenbaum S Exercise for depression in older adults: a meta-analysis of randomized controlled trials adjusting for publication bias Br J Psychiatr 38 3 2016 247 254 10.1590/1516-4446-2016-1915 39 Pascoe MC Parker AG. Physical activity and exercise as a universal depression prevention in young people: a narrative review Early Interv Psychiat 13 4 2019 733 739 10.1111/eip.12737 40 Chu I-H Buckworth J Kirby TE Emery CF. Effect of exercise intensity on depressive symptoms in women Ment Health Phys Act 2 1 2009 37 43 10.1016/J.MHPA.2009.01.001 41 McDowell CP MacDonncha C Herring MP. Brief report: associations of physical activity with anxiety and depression symptoms and status among adolescents J Adolesc 55 2017 1 4 10.1016/j.adolescence.2016.12.004 27988443 42 Axelsdóttir B Biedilae S Sagatun Å Nordheim LV Larun L. Review: exercise for depression in children and adolescents - a systematic review and meta-analysis Child Adolesc Ment Health 26 4 2021 347 356 10.1111/camh.12438 33277972 43 Wang X Cai Z-D Jiang W-T Fang Y-Y Sun W-X Wang X. Systematic review and meta-analysis of the effects of exercise on depression in adolescents Child Adolesc Psychiatr Ment Health 16 1 2022 16 10.1186/s13034-022-00453-2 44 Kessler RC Aguilar-Gaxiola S Alonso J The global burden of mental disorders: an update from the WHO World Mental Health (WMH) surveys Epidemiol Psichiatr Soc 18 1 2009 23 33 10.1017/s1121189x00001421 19378696 45 Hu T. Perspectives: an international review of the national cost estimates of mental illness, 1990-2003 J Ment Health Pol Econ 9 1 2006 3 13 46 McDowell CP Gordon BR Andrews KL MacDonncha C Herring MP. Associations of physical activity with anxiety symptoms and status: results from the Irish longitudinal study on ageing Epidemiol Psychiatr Sci 28 4 2019 436 445 10.1017/S204579601800001X 29382402 47 Gordon BR McDowell CP Lyons M Herring MP. The effects of resistance exercise training on anxiety: a meta-analysis and meta-regression analysis of randomized controlled trials Sports Med 47 12 2017 2521 2532 10.1007/s40279-017-0769-0 28819746 48 Stubbs B Vancampfort D Rosenbaum S An examination of the anxiolytic effects of exercise for people with anxiety and stress-related disorders: a meta-analysis Psychiatr Res 249 2017 102 108 10.1016/j.psychres.2016.12.020 49 Aylett E Small N Bower P. Exercise in the treatment of clinical anxiety in general practice - a systematic review and meta-analysis BMC Health Serv Res 18 1 2018 559 10.1186/s12913-018-3313-5 30012142 50 McDowell CP Dishman RK Gordon BR Herring MP. Physical activity and anxiety: a systematic review and meta-analysis of prospective cohort studies Am J Prev Med 57 4 2019 545 556 10.1016/j.amepre.2019.05.012 31542132 51 Carter T Pascoe M Bastounis A Morres ID Callaghan P Parker AG. The effect of physical activity on anxiety in children and young people: a systematic review and meta-analysis J Affect Disord 285 2021 10 21 10.1016/j.jad.2021.02.026 33618056 52 Frederiksen KP Stavestrand SH Venemyr SK Sirevåg K Hovland A. Physical exercise as an add-on treatment to cognitive behavioural therapy for anxiety: a systematic review Behav Cognit Psychother 49 5 2021 626 640 10.1017/S1352465821000126 33678210 53 Brunet J Sabiston CM O’Loughlin E Chaiton M Low NCP O’Loughlin JL. Symptoms of depression are longitudinally associated with sedentary behaviors among young men but not among young women Prev Med 60 2014 16 20 10.1016/j.ypmed.2013.12.003 24333606 54 Doré I O’Loughlin JL Beauchamp G Martineau M Fournier L. Volume and social context of physical activity in association with mental health, anxiety and depression among youth Prev Med 91 2016 344 350 10.1016/j.ypmed.2016.09.006 27609745 55 Evans MB Allan V Erickson K Martin LJ Budziszewski R Côté J. Are all sport activities equal? A systematic review of how youth psychosocial experiences vary across differing sport activities Br J Sports Med 51 3 2017 169 176 10.1136/bjsports-2016-096725 27986759 56 Lubans D Richards J Hillman C Physical activity for cognitive and mental health in youth: a systematic review of mechanisms Pediatrics 138 3 2016 e20161642 10.1542/peds.2016-1642 27542849 57 Doré I O’Loughlin JL Schnitzer ME Datta GD Fournier L. The longitudinal association between the context of physical activity and mental health in early adulthood Ment Health Phys Act 14 2018 121 130 10.1016/j.mhpa.2018.04.001 58 Eime RM Harvey JT Brown WJ Payne WR. Does sports club participation contribute to health-related quality of life? Med Sci Sports Exerc 42 5 2010 1022 1028 10.1249/MSS.0b013e3181c3adaa 19996991 59 Eime RM Young JA Harvey JT Charity MJ Payne WR. A systematic review of the psychological and social benefits of participation in sport for children and adolescents: informing development of a conceptual model of health through sport Int J Behav Nutr Phys Activ 98 2013 10 10.1186/1479-5868-10-98 60 Boelens M Smit MS Raat H Bramer WM Jansen W. Impact of organized activities on mental health in children and adolescents: an umbrella review Prev Med Rep 101687 2022 25 10.1016/j.pmedr.2021.101687 61 Murphy J Patte KA Sullivan P Leatherdale ST. Exploring the association between sport participation and symptoms of anxiety and depression in a sample of canadian high school students J Clin Sport Psychol 15 3 2021 268 287 10.1123/JCSP.2020-0048 62 Hoffmann MD Barnes JD Tremblay MS Guerrero MD. Associations between organized sport participation and mental health difficulties: data from over 11,000 US children and adolescents PLoS One 17 6 2022 e0268583 10.1371/journal.pone.0268583 63 Dishman RK O’Connor PJ. Lessons in exercise neurobiology: the case of endorphins Ment Health Phys Act 2 1 2009 4 9 10.1016/j.mhpa.2009.01.002 64 Kim S Choi J-Y Moon S Park D-H Kwak H-B Kang J-H. Roles of myokines in exercise-induced improvement of neuropsychiatric function Pflügers Archiv 471 3 2019 491 505 10.1007/s00424-019-02253-8 30627775 65 Mikkelsen K Stojanovska L Polenakovic M Bosevski M Apostolopoulos V. Exercise and mental health Maturitas 106 2017 48 56 10.1016/j.maturitas.2017.09.003 29150166 66 Hammen C. Stress and depression Annu Rev Clin Psychol 1 2005 293 319 10.1146/annurev.clinpsy.1.102803.143938 17716090 67 Tibubos AN Burghardt J Klein EM Frequency of stressful life events and associations with mental health and general subjective health in the general population J Publ Health 29 5 2021 1071 1080 10.1007/s10389-020-01204-3 68 Salari N Hosseinian-Far A Jalali R Prevalence of stress, anxiety, depression among the general population during the COVID-19 pandemic: a systematic review and meta-analysis Glob Health 16 1 2020 57 10.1186/s12992-020-00589-w 69 Racine N McArthur BA Cooke JE Eirich R Zhu J Madigan S. Global prevalence of depressive and anxiety symptoms in children and adolescents during COVID-19: a meta-analysis JAMA Pediatr 175 11 2021 1142 1150 10.1001/jamapediatrics.2021.2482 34369987 70 Mücke M Ludyga S Colledge F Gerber M. Influence of regular physical activity and fitness on stress reactivity as measured with the Trier Social Stress Test Protocol: a systematic review Sports Med 48 11 2018 2607 2622 10.1007/s40279-018-0979-0 30159718 71 Gerber M Pühse U. Review article: do exercise and fitness protect against stress-induced health complaints? A review of the literature Scand J Publ Health 37 8 2009 801 819 10.1177/1403494809350522 72 Archer T. Physical exercise as an epigenetic factor determining behavioral outcomes Clin Exp Psychol 1 1 2015 e101 10.4172/2471-2701.1000e101 73 Rhodes RE Janssen I Bredin SSD Warburton DER Bauman A. Physical activity: health impact, prevalence, correlates and interventions Psychol Health 32 8 2017 942 975 10.1080/08870446.2017.1325486 28554222 74 Geller K Lippke S Nigg CR. Future directions of multiple behavior change research J Behav Med 40 1 2017 194 202 10.1007/s10865-016-9809-8 27785652 75 Goenka S Lee I-M. Physical activity lowers mortality and heart disease risks Lancet 390 10113 2017 2609 2610 10.1016/S0140-6736(17)32104-9 28943265 76 Michie S Abraham C Whittington C McAteer J Gupta S. Effective techniques in healthy eating and physical activity interventions: a meta-regression Health Psychol 28 6 2009 690 701 10.1037/a0016136 19916637 77 Schwartz J Rhodes R Bredin SSD Oh P Warburton DER Effectiveness of approaches to increase physical activity behavior to prevent chronic disease in adults: a brief commentary J Clin Med 8 3 2019 295 10.3390/jcm8030295 30832260 78 Kaushal N Rhodes RE. Exercise habit formation in new gym members: a longitudinal study J Behav Med 38 4 2015 652 663 10.1007/s10865-015-9640-7 25851609 79 Bauman AE Reis RS Sallis JF Wells JC Loos RJF Martin BW. Correlates of physical activity: why are some people physically active and others not? Lancet 380 9838 2012 258 271 10.1016/S0140-6736(12)60735-1 22818938 80 Vandelanotte C Müller AM Short CE Past, present, and future of ehealth and mhealth research to improve physical activity and dietary behaviors J Nutr Educ Behav 48 3 2016 219 228 10.1016/j.jneb.2015.12.006 e1 26965100 81 Blair S.N. Exercise, health, and longevity Lamb DR Murray R. Perspectives in exercise science and sports medicine 1988 Benchmark Press 443 489 82 Dishman RK. Compliance/adherence in health-related exercise Health Psychol 1 3 1982 237 267 10.1037/0278-6133.1.3.237 83 Williford HN Barfield BR Lazenby RB Olson MS. A survey of physicians’ attitudes and practices related to exercise promotion Prev Med 21 5 1992 630 636 10.1016/0091-7435(92)90070-x 1438111 84 Eakin EG Glasgow RE Riley KM. Review of primary care-based physical activity intervention studies: effectiveness and implications for practice and future research J Fam Pract 49 2000 158 168 10718694 85 Jakicic JM Winters C Lang W Wing RR. Effects of intermittent exercise and use of home exercise equipment on adherence, weight loss, and fitness in overweight women: a randomized trial JAMA 282 16 1999 1554 1560 10.1001/jama.282.16.1554 10546695 86 Gorin A Phelan S Tate D Sherwood N Jeffery R Wing R. Involving support partners in obesity treatment J Consult Clin Psychol 73 2 2005 341 343 10.1037/0022-006X.73.2.341 15796642 87 Sexton H Maere A Dahl NH. Exercise intensity and reduction in neurotic symptoms. A controlled follow-up study Acta Psychiatr Scand 80 3 1989 231 235 10.1111/j.1600-0447.1989.tb01332.x 2816466 88 Lange KW. Task sharing in psychotherapy as a viable global mental health approach in resource-poor countries and also in high-resource settings Glob Health J 5 3 2021 120 127 10.1016/j.glohj.2021.07.001
PMC009xxxxxx/PMC9942802.txt
==== Front Proc Natl Acad Sci U S A Proc Natl Acad Sci U S A PNAS Proceedings of the National Academy of Sciences of the United States of America 0027-8424 1091-6490 National Academy of Sciences 36656854 202213727 10.1073/pnas.2213727120 research-articleResearch Articlebiophys-bioBiophysics and Computational Biology408 Biological Sciences Biophysics and Computational Biology Fine structure and assembly pattern of a minimal myophage Pam3 Yang Feng a b Jiang Yong-Liang [email protected] a 1 Zhang Jun-Tao a Zhu Jie a https://orcid.org/0000-0001-9884-0355 Du Kang a Yu Rong-Cheng a Wei Zi-Lu a Kong Wen-Wen a https://orcid.org/0000-0001-6955-3517 Cui Ning a Li Wei-Fang a Chen Yuxing a https://orcid.org/0000-0002-7560-1922 Li Qiong [email protected] a 1 https://orcid.org/0000-0003-2838-9380 Zhou Cong-Zhao [email protected] a 1 aSchool of Life Sciences, Division of Life Sciences and Medicine, University of Science and Technology of China, Hefei 230026, China bResearch Center for Intelligent Computing Platforms, Zhejiang Lab, Hangzhou, Zhejiang 311121, China 1To whom correspondence may be addressed. Email: [email protected], [email protected], or [email protected]. Edited by Stephen Harrison, Boston Children's Hospital, Boston, MA; received August 25, 2022; accepted December 12, 2022 19 1 2023 24 1 2023 19 7 2023 120 4 e221372712025 8 2022 12 12 2022 Copyright © 2023 the Author(s). Published by PNAS. 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This article is distributed under Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND). Significance The cyanophage Pam3 has a substantially simpler molecular organization than the prototypical, contractile-tail bacteriophage (“myophage”), T4, but it retains the essential elements required for DNA injection. The 3D structure of an intact Pam3 particle, described in this paper, shows the essential, conserved modules of a minimal myophage. Pam3 harbors only one type of tail fiber, in alternating up and down configurations, facilitating its search for host cell receptors. Disulfide bonds between the tail fibers and the baseplate wedge suggest a redox-dependent mechanism of baseplate assembly and tail-sheath contraction. The structure lays an empirical foundation for practical engineering of cyanophages for synthetic biology applications. The myophage possesses a contractile tail that penetrates its host cell envelope. Except for investigations on the bacteriophage T4 with a rather complicated structure, the assembly pattern and tail contraction mechanism of myophage remain largely unknown. Here, we present the fine structure of a freshwater Myoviridae cyanophage Pam3, which has an icosahedral capsid of ~680 Å in diameter, connected via a three-section neck to an 840-Å-long contractile tail, ending with a three-module baseplate composed of only six protein components. This simplified baseplate consists of a central hub-spike surrounded by six wedge heterotriplexes, to which twelve tail fibers are covalently attached via disulfide bonds in alternating upward and downward configurations. In vitro reduction assays revealed a putative redox-dependent mechanism of baseplate assembly and tail sheath contraction. These findings establish a minimal myophage that might become a user-friendly chassis phage in synthetic biology. cyanophage cryo-EM structure a minimal myophage redox-dependent mechanism Ministry of Science and Technology of the People's Republic of China (MOST) 501100002855 2018YFA0903100 Cong-Zhao Zhou National Natural Science Foundation of China (NSFC) 501100001809 U19A2020 Cong-Zhao Zhou Strategic Priority Strategic Priority Research Program of the Chinese Academy of Sciences XDB37020301 Yong-Liang JiangQiong LiCong-Zhao Zhou Youth Innovation Promotion Association of Chinese Academy of Sciences 2020452 Yong-Liang JiangQiong LiCong-Zhao Zhou the Fundamental Research Funds for the Central Universities WK2070000195 Yong-Liang JiangQiong LiCong-Zhao Zhou ==== Body pmcBacteriophages are widespread viruses that specifically infect bacteria and play major roles in the dynamics and genetic diversity of bacterial communities (1). They commonly contain an icosahedral capsid encapsulating a double-stranded DNA (dsDNA) genome that is sealed with a tail at the portal vertex (2). The tail is a key macromolecular machinery responsible for securing genomic DNA, recognizing the host, penetrating the cell envelope, and injecting the phage genome (3). Based on the morphology of the tail, bacteriophages are commonly classified into three families, Podoviridae, Siphoviridae, and Myoviridae, which have a short, long noncontractile, or long contractile tail, respectively. The contractile tail of myophage is of the most complexity, which undergoes a substantial conformational switch upon attachment to the host cell: the external tail sheath contracts and drives an inner rigid tube to penetrate the host cell membrane, forming a channel that enables the injection of phage genome into the host cytosol (4). In fact, several protein machineries, such as type VI secretion systems (T6SSs) (5), R-type pyocins (6), metamorphosis-associated contractile arrays (7), and Photorhabdus virulence cassettes (PVCs) (8), also use a similar contractile mechanism to penetrate the bacterial or eukaryotic cell envelope. The baseplate, which is the most complicated part of these contractile injection systems (CISs), is responsible for coordinating host recognition or other environmental signals with sheath contraction (9). As a well-investigated model of contractile phages, Escherichia coli phage T4 possesses a baseplate composed of at least 15 various components, which assemble into two large intermediates: the hub and wedge (10). Surrounding the hub at the center, the sixfold symmetric wedge forms a ring that provides a platform for connecting tail fibers at the periphery. Upon attachment to the host cell, the tail fibers of phage T4 reorient to better bind to the host receptors, accompanied by conformational changes at the inner components of the baseplate, eventually enabling the release of the hub and simultaneously initiating sheath contraction (11). Recently, we isolated a Myoviridae cyanophage termed Pam3 from Lake Chaohu in China, which specifically infects the host cyanobacteria Pseudanabaena mucicola (12). Genomic annotation indicated that, compared with phage T4, Pam3 harbors a much simpler baseplate composed of only six components gp19–gp24 encoded by the putative open reading frames (Fig. 1A). In contrast to the T4 baseplate of a much more complicated structure, the baseplate of Pam3 is similar to that of E. coli phage Mu, which also harbors a baseplate of six conserved components (13). Notably, these phage baseplates of a simpler composition resemble the CISs of known structures, such as type VI secretion systems (14) and R-type pyocins (6), indicating their similarity in evolution and working mode. In fact, a simpler baseplate will enable the relative simplicity and ease of engineering of these phages, thus providing an ideal model system for synthetic engineering and biotechnological applications. However, the lack of atomic structures of a simple myophage largely precludes understanding of its assembly pattern and the mechanism of tail sheath contraction. Fig. 1. Overall architecture of the Pam3 virion. (A) A schematic diagram of the organization of Pam3 structural genes. The genes in white encode the nonstructural proteins that are not presented in the intact Pam3 virions. (B) The overall cryo-EM map of the intact mature Pam3 virion. The structural proteins are colored the same as their encoding genes. The sizes of the head, neck, tail, and baseplate are shown in Å. Here, we report the cryo-electron microscopy (cryo-EM) structure of the intact myophage Pam3 and reveal the assembly pattern of the capsid, neck, tail, and especially a minimal baseplate composed of six essential components. Notably, we found that the twelve tail fibers, which adopt alternating upward and downward configurations, are bridged to the six heterotriplex units of the wedge via disulfide bonds. These fine structures provide insights into the assembly pattern and infection model of a minimal myophage. Results Overall Architecture of Pam3. Genomic analysis indicated that the structural genes of Pam3 are clustered within the region from 2,272 to 19,635 bp in the genome that harbors four operons encoding the proteins for the assembly of the head, neck, tail, and baseplate (Fig. 1A). We purified Pam3 particles using density gradient centrifugation and solved the intact structure by cryo-EM (SI Appendix, Figs. S1 and S2 A–D). Generally, Pam3 has an icosahedral capsid with a diameter of ~680 Å that encapsulates a highly compacted 54.5-kb genome, followed by an ~840-Å long contractile tail with a baseplate at the distal end (Fig. 1B). The icosahedral capsid shell consists of 60 hexons and 11 pentons, or in total, 415 copies of capsid protein gp7. Additionally, there are 140 trimers of the cement protein gp6 at the icosahedral threefold and quasi-threefold axes. The dodecameric portal protein gp3 replaces a penton at one of the 12 fivefold vertexes, forming a channel for DNA entry and exit (Fig. 1B). The dodecameric adaptor gp9, the hexameric connector gp11, and the hexameric terminator gp12 [also termed completion protein in some cases (15, 16)] sequentially constitute the neck of Pam3, which forms a joint to connect the head and tail and functions as a gatekeeper to prevent leakage of the genome (Fig. 1B). The long contractile tail contains 22 helically stacked hexamers of sheath protein gp13 and 22 hexamers of tube protein gp14, growing from the sheath initiator gp21 and tube initiator gp17, respectively (Fig. 1B). Notably, Pam3 possesses a greatly simplified baseplate with only six components that assemble into three modules: the central hub-spike complex surrounded by the wedge and 12 tail fibers. The hub-spike complex comprises a trimeric hub gp19 and a trimeric spike gp20, whereas the wedge contains six heterotriplex units, each of which consists of two gp22 subunits and one gp23 subunit. Six gp18 monomers (termed plug) insert into the cleft between the wedge and hub-spike complex. At the periphery of the wedge, 12 tail fibers (each composed of a trimeric gp24) of 110 Å in length link to six gp23 subunits in alternating upward and downward configurations. The Symmetry-Mismatched Assembly of the Capsid and Portal. The Pam3 capsid is organized in an icosahedral lattice with a triangulation number T of 7. The major capsid protein gp7, which adopts a canonical HK97 fold (17), consists of four distinct domains (Fig. 2A and SI Appendix, Fig. S3A). In contrast to previously reported HK97-like proteins (17), gp7 contains a unique insertion segment (residues Arg214–Arg224) positioned at the inner surface of the capsid (SI Appendix, Fig. S3A). Due to the variations in the peripheral segments, the gp7 subunits oligomerize into two forms of capsomers, hexons, and pentons, which further assemble into the icosahedral capsid (SI Appendix, Fig. S3B). Fig. 2. The symmetry-mismatched assembly of the capsid and portal. (A) Structures of the hexon, penton, and cement of the Pam3 capsid, shown in top and side views, respectively. The major capsid proteins gp7 in hexon and penton are colored blue, and yellow, respectively. The cement protein of the gp6 trimer is colored red. (B) Longitudinal cut views of the cryo-EM map of the portal vertex. The portal, DNA, major capsid, and cement are colored magenta, gray, light blue, and red, respectively. The diameters of the portal are labeled. (C) The interface between the portal and the initiating terminus of genomic DNA. The loops involved in the interaction with DNA are colored green. (D) Top view of the portal vertex (seen from the inside of the capsid). The surrounding capsomers are shown as cartoons, whereas the 12 subunits of the portal are displayed as black circles and sequentially labeled from 1 to 12. The surrounding subunits (S1–S5) and distal subunits (D1–D5) of the capsomers are colored cyan and magenta, respectively. The Inset shows an enlarged view of the interface between the capsomers and portal. The secondary structure elements of the capsomers involved in the interaction with portal are colored blue. The cement protein gp6 consists of an N-terminal β-tulip domain followed by a C-terminal domain of seven antiparallel β-strands (SI Appendix, Fig. S3C). The β-tulip domain is structurally conserved among phages that harbor trimeric cement proteins (18), whereas the C-terminal domains vary greatly (SI Appendix, Fig. S3C), indicating that the β-tulip domain is a general module for protein‒protein interactions in phages. Three gp6 subunits form a trimer in a “head-to-tail” manner. One gp6 trimer sits on the threefold axis and interacts with three neighboring capsomers. The N-terminal 23-residue loop (termed the N-loop) of gp6 contributes the majority of the interactions with gp7 (SI Appendix, Fig. S3D). Notably, the N-loop of gp6 protrudes out of the core domains and interacts with the N-arm and P-domain of one gp7 subunit, in addition to the E-loop and A-domain of another gp7 subunit, further strengthening the interface (SI Appendix, Fig. S3D). The N-loop of gp6 adopts a conformation distinct from those of previously reported phage structures (18). It runs perpendicularly across the E-loop of gp7, makes a 90° turn to align parallel to the N-arm of gp7, and finally inserts into the interface between two adjacent gp7 subunits. The gp6 N-loop and gp7 N-arm form a unique ring-like structure that lies above the gp7 E-loops (SI Appendix, Fig. S3D). This interlocked ring-like architecture at the threefold axis probably provides plasticity of the capsid structure in response to fluctuations in the living environment and thus maintains the integrity of the phage. Similar to other phages (19), Pam3 contains a special portal vertex that acts as a platform for the assembly of the phage DNA-translocating motor. The portal gp3 dodecamer adopts a canonical structure that resembles a flying saucer with a height of 126 Å and an external diameter of 130 Å. It has a central channel of 24 to 64 Å in diameter, which is filled with a rope-like DNA chain, as shown in the cryo-EM maps (Fig. 2B). Each gp3 subunit consists of five different parts arranged as follows (from inside to outside): barrel, crown, wing, stem, and clip (Fig. 2C and SI Appendix, Fig. S4A). The 12-fold portal dodecamer contacts five copies of hexons that possess a fivefold axis, showing a symmetry-mismatched interaction pattern (Fig. 2D). In total, the gp3 dodecamer is surrounded by 10 gp7 subunits, which are contributed by five surrounding hexons. These 10 subunits can be further classified into two groups according to their relative positions to the portal: five closely surrounding subunits (labeled S1–S5) form a ring around the portal like a nest, whereas five distal subunits (labeled D1–D5) interact with the portal’s wing (Fig. 2D). In the circular cleft between the wing of the portal and the inner surface of the capsid, we observed a clear and continuous cryo-EM density, which could be fitted with an ~160-bp DNA segment (SI Appendix, Fig. S4B). It forms a lasso structure tightly anchored to the portal. The DNA forms direct contacts with several basic residues from a protruding loop of the portal on one side and the surrounding major capsid subunits on the other side (SI Appendix, Fig. S4 C and D). This lasso structure enables anchoring of the genome at the portal and offsets the angular momentum of DNA during packaging into the capsid. The Adaptor, Connector, and Terminator Are Sequentially Interlocked to Form the Neck. The dodecameric portal is connected with the neck, which comprises a dodecameric adaptor gp9, a hexameric connector gp11, and a hexameric terminator gp12, forming a continuous DNA injection channel at the center (Fig. 3A). In this channel, we observed a clear cryo-EM density of the DNA segment extending to the junction between the connector and terminator (Fig. 3A). This DNA segment corresponds to the ending terminus of the genome, which is plugged by the tape measure protein (TMP) at the bottom, as shown in the cryo-EM density (Fig. 3A). Fig. 3. Assembly and interfaces of the Pam3 neck. (A) Surface representation of the overall structure of the neck. The neck comprises a dodecameric adaptor (cyan), a hexameric connector (salmon), and a hexameric terminator (blue). The ending terminus of genomic DNA (gray) in the channel is sealed by TMP (purple). (B–D) Magnified views of the pairwise interfaces among neck proteins: (B) portal–adaptor; (C) adaptor–connector; (D) connector–terminator. The secondary structure elements involved in the interactions are labeled. The neck is connected to the portal via direct interactions between the C-terminal hook of adaptor gp9 and the portal clip (Fig. 3B). Each gp9 subunit consists of two domains, the conserved α-helix bundle domain and a unique β-hairpin domain (SI Appendix, Fig. S5A). Twelve gp9 subunits form a dodecamer, with the β-hairpin domains forming a β-barrel of 24 Å in diameter with a positively charged inner surface, possibly interacting with DNA via a cluster of lysine residues (SI Appendix, Fig. S5B). This β-barrel is docked on a β-barrel formed by six subunits of the connector gp11, surrounded by a unique N-terminal helix α1 of the connector (Fig. 3C). Beyond the β-barrel, each connector subunit contains two downward-extending β-hairpins (β2-β3 and β5-β6) (SI Appendix, Fig. S5C). Two β-hairpins from the adjacent gp11 subunits clasp the N-terminal loop of the terminator gp12 (Fig. 3D). Following this N-terminal loop, each terminator subunit consists of a central conserved globular domain of five β-strands and one helix, in addition to a protruding domain composed of β-strands β2-β3-β8 and β9 (SI Appendix, Fig. S5D). Six subunits of gp12 form a hexameric terminator, which links to the connector at the top and terminates the growth of the tail tube/sheath at the bottom. Structural analysis showed that the pairwise interfaces between these components of the neck are complementary in charge and shape (SI Appendix, Fig. S6), which enables the sequential joining and efficient assembly of the interlocked neck. The Tail Is a Three-Layered DNA Injection Machine Composed of a Contractile Sheath and a Rigid Tube Surrounding the TMP. Compared with the well-known myophage T4 (9), Pam3 possesses a shorter contractile tail ending with a greatly simplified baseplate and is connected to the neck via the single terminator protein gp12. The tail of Pam3 is ~840 Å in length and comprises 22 hexamers of the sheath gp13, 22 hexamers of the tube gp14, one hexamer of the sheath initiator gp21, and one hexamer of the tube initiator gp17, surrounding a flagstaff-like TMP hexamer fixed on the baseplate (Fig. 4 A, Left). Fig. 4. Structure of the tail. (A) Overall architecture of the tail, TMP, and baseplate. The structures of the terminator (blue), tube (light green and dark green), and sheath (red and orange) proteins are shown as cartoons, whereas the TMP (purple) and baseplate (cyan) are shown as cryo-EM maps. The helical structures of the tube and sheath surrounding the TMP are shown on the right. (B) The bottom and side views of the cryo-EM map of TMP, fitted with a predicted model. (C) The cryo-EM maps showing the interfaces between the terminator and sheath/tube, with a zoomed-in view shown as an Inset. The secondary structure elements involved in the interactions are labeled in the Inset. (D) The cryo-EM maps showing the interfaces between the sheath initiator and sheath, with a zoomed-in view shown as an Inset. The secondary structure elements involved in the interactions are labeled in the Inset. The tube gp14 subunits are organized into a six-entry helix around the TMP, with a helical rise of 37.2 Å and a twist of 33.8°, as shown in one helical structure of gp14 subunits (Fig. 4 A, Right). Each tube subunit has a central globular domain that consists of one helix and a β-barrel composed of two four-stranded β-sheets. These two β-sheets, which are perpendicular to each other, are located at the inner and outer surfaces of the tube, respectively. Beyond the central globular domain, a β-hairpin of the tube subunit protrudes toward the next layer, which mediates the intersubunit interactions between hexamers (SI Appendix, Fig. S7A). Notably, the inner four-stranded β-sheet of the tube is structurally similar to the central β-sheet of the terminator gp12 (SI Appendix, Fig. S7B), both of which form a channel of ~40 Å in diameter. The sheath gp13 also adopts a helical structure similar to the tube (Fig. 4 A, Right). Each sheath subunit contains two globular domains with two protruding termini (SI Appendix, Fig. S7C). The N-terminal domain adopts an α/β structure that forms a prominent ridge on the surface of the sheath, whereas the C-terminal domain consists of three α-helices in addition to two β-strands (β11 and β12) that mediate the interhexamer interactions of the sheath. The N terminus of one subunit and the C terminus of the neighboring subunit from the same hexamer, together with the two β-strands of the C-terminal domain from the proceeding hexamer, form an interlaced four-stranded β-sheet across two hexamers (SI Appendix, Fig. S7D). This β-sheet contributes the majority to both interhexamer and intrahexamer interactions among sheath subunits. It is known that the length of the tail is determined by the TMP (20), which has been proposed to assemble into a trimer (21) or hexamer (20, 22, 23) but lacks direct structural evidence. In our cryo-EM maps of the tail, we found a continuous density filling the tail tube (Fig. 4A), which showed clear features of six-entry helical structures of ~21 Å in diameter from the bottom view (Fig. 4B). Given that each TMP subunit of 588 residues is predicted to fold into an extended all-α structure, it should theoretically have a straight length of 882 Å, which is consistent with the 840-Å long six-helical flagstaff-like density observed in our structure. Moreover, the mass spectrometry analysis also identified the TMP proteins in the intact Pam3 virions (24). Therefore, the TMP of Pam3 most likely adopts a sixfold helical bundle structure that provides a scaffold for the assembly of the tail. Beyond TMP, which determines the tail length, the terminator gp12 contributes to stopping the growth of the tail tube and sheath. The protruding β2–β3–β8 of the terminator interacts with the loops from two tube subunits, thus disrupting the interhexamer interface of the tube and terminating its growth (Fig. 4C). Meanwhile, the C-terminal β9 of the terminator and β11–β12 of the sheath C-terminal domain form an extended β-sheet, thus altering the interhexamer β-sheet structure of the sheath and eventually terminating the growth of the tail sheath (Fig. 4C). The growth of the tail tube and sheath is initiated on the baseplate by the tube initiator gp17 and sheath initiator gp21, respectively (Fig. 4D). The tube initiator gp17 possesses an N-terminal domain of the β-barrel fold, which is structurally similar to the tube gp14, in addition to a C-terminal domain composed of two α-helices and connecting loops (SI Appendix, Fig. S8 A and B). The β-barrel domains of six tube initiator subunits form a ring-like structure to initiate the extension of the tube, whereas the C-terminal domain protrudes outward and interacts with the inner surface of the baseplate wedge. The sheath initiator gp21 consists of a core domain that is structurally similar to the C-terminal domain of sheath gp13, in addition to the N-terminal six β-strands, forming a triangle-like structure (SI Appendix, Fig. S8 C and D). β7-β8 from the sheath initiator, together with β1 of one subunit and β12 of the neighboring subunit from the same sheath hexamer, form a similar interlaced four-stranded β-sheet to initiate sheath extension (Fig. 4D). Pam3 Has a Minimal Baseplate Composed of Six Essential Components. The baseplate of Pam3 contains six protein components: the trimeric spike gp20, the trimeric hub gp19, the baseplate wedge (six heterotriplexes of gp22-gp23), the plug of six gp18 monomers, and the fibers (12 trimers of gp24) (Fig. 5A). The trimeric spike gp20 docks into the trimeric hub gp19, forming a central hub-spike complex (Fig. 5 A and B). The hub consists of three segments: the four-stranded β-sheet at the N terminus, followed by an α/β domain and a C-terminal β-barrel domain that resembles the β-barrel of tube gp14 (SI Appendix, Fig. S9A). Three subunits form a hub that interacts with the hexameric tube initiator via their C-terminal β-barrel and N-terminal β-sheet domains. Three α/β domains of the hub trimer form triangular claws that embrace the three N-terminal helices of the spike trimer. The spike comprises four segments: an N-terminal α-helix, a conserved domain of an oligonucleotide/oligosaccharide-binding fold that is commonly shared by myophages and CISs (6, 8, 11, 25), a three-stranded β-sheet and two α-helices at the C terminus (Fig. 5B and SI Appendix, Fig. S9B). Three β-sheets of the Pam3 spike trimer form a β-helical structure, which is much shorter than that of T4 (11), R2 (6), PVC (8), or the antifeeding prophage (25). Notably, the C-terminal α-helices of Pam3 spike form an α-helical bundle located at the distal end (SI Appendix, Fig. S9B). This distal helical bundle of the spike has not been found in the structure-known spikes of phages or CISs (SI Appendix, Fig. S9C), indicating a distinct mechanism for Pam3 to penetrate the host cell envelope. Fig. 5. Structure of the baseplate. (A) Longitudinal cut (Up) and bottom views of the baseplate (Down). The protein components of the baseplate are colored differently in the cryo-EM map. (B) Cartoon presentation of the central hub-spike complex. The hub and spike are colored green and red, respectively. (C) Organization of the baseplate wedge. The trifurcation unit is denoted by a red triangle, whereas the extrusions of one trifurcation unit are denoted by red rectangles and a circle. The intersubunit disulfide bonds are shown as an Inset and are colored yellow. (D) A zoomed-in view of the interactions between the sheath and the wedge heterotriplex mediated by the sheath initiator. The sheath initiator gp21 is shown in cartoon representation, whereas the sheath gp13, tube initiator gp17, and wedge heterotriplex gp22-gp23 are shown on the surface. (E) A zoomed-in view of the interactions between the hub and the wedge heterotriplex. The tube initiator gp17 and the plug gp18 are shown in cartoon representation, whereas the wedge heterotriplex gp22-gp23 and the hub gp19 are shown on the surface. Similar to phage T4 (11), the baseplate wedge of Pam3 consists of six heterotriplexes, each of which is composed of two gp22 subunits of distinct conformations and one gp23 subunit at the periphery, forming a core six-helical bundle and a trifurcation unit (Fig. 5C and SI Appendix, Fig. S10 A and B). Each trifurcation unit has four docking sites: two from gp22 subunits are used for interacting with gp22 of adjacent trifurcation units, and the other two from gp23 are required for the attachment of the tail fiber (Fig. 5C). Notably, unlike T4 wedge (11), gp22 and gp23 of Pam3 are much simpler and lack extra insertion domains (SI Appendix, Fig. S10 C and D). Moreover, different from other phages, we observed two intersubunit disulfide bonds (gp22ACys187-gp22ACys215, gp22BCys215-gp23Cys78) between two trifurcation units and three intra-gp23 disulfide bonds (Cys5-Cys48, Cys76-Cys122 and Cys80-Cys237) of one trifurcation unit to stabilize the wedge (Fig. 5C and SI Appendix, Fig. S11A). The wedge anchors to the central hub-spike complex via extensive interactions mediated by the sheath initiator gp21 and the plug gp18. The sheath initiator firmly clasps the tip of the core helical bundle of each wedge heterotriplex via its N-terminal six β-sheets (Fig. 5D). The plug consists of two-layered β-sheets and a C-terminal α-helix, resembling the immunoglobin fold. Each plug monomer is embedded between the two C-terminal helices of the tube initiator, forming a “plug” to fix the core helical bundle of the wedge onto the hub (Fig. 5E). Twelve Tail Fibers Anchor to the Wedge via Disulfide Bonds in Alternating Upward and Downward Configurations. Around the baseplate wedge, 12 tail fibers (a gp24 trimer for each) are arranged in alternating upward and downward configurations. Each subunit of the trimeric fiber is composed of an N-terminal α-helical domain, a tumor necrosis factor-like (TNF-like) domain and a glycine-rich domain (SI Appendix, Fig. S11B). The α-helical domain is responsible for attaching the fiber to the wedge, whereas the other two domains point outward, which might contribute to recognizing the host receptors. The Pam3 fiber gp24 shows a similar architecture to the S16 receptor-binding adhesion gp38, which caps the trimeric gp37 β-helix at the tip of long tail fiber (26). Compared with the β-helix domain of S16 gp38, the middle TNF-like domain of gp24 shows a different topology and lacks the α-helix. In addition, both the glycine-rich domains of Pam3 gp24 and S16 gp38 are rich in polar and aromatic residues, but differ in the number of glycine-rich motifs (GRMs). Pam3 gp24 has 12 GRMs forming a four-layer lattice, whereas S16 gp38 possesses a three-layer lattice of 10 GRMs. Moreover, Pam3 gp24 has a much longer N-terminal α-helical bundle, compared with that of S16 gp38. Each heterotriplex of the wedge contains two cysteine-rich regions, which bridge an upward and a downward fiber via disulfide bonds (Fig. 6A). In detail, Cys90, Cys104, and Cys113 of one region in a loop of the wedge subunit gp23 form disulfide bonds with the three Cys28 residues protruding to the outermost of the α-helical domains of the upward trimeric fiber (SI Appendix, Fig. S12). The other cysteine-rich region comprising Cys222 and Cys228 of gp23, and Cys187 of one gp22 subunit (gp22B) of the wedge also bridges the three Cys28 residues of the downward tail fiber (SI Appendix, Fig. S12). To further identify these disulfide bonds, the intact Pam3 virions were applied to the mass spectrometry analyses. Despite a rather low abundance of tail proteins in the intact Pam3 virions, three of the six disulfide bonds could be clearly assigned (SI Appendix, Table S1). Generally, the two cysteine-rich regions on the wedge provide two triangular platforms that facilitate the docking of two tail fibers to the wedge in alternating upward and downward configurations. Fig. 6. Tail fibers are attached to the baseplate wedge via disulfide bonds. (A) The interface between wedge and tail fibers with magnified views (Left panel for the upward fiber and Right panel for the downward fiber) shown as an Inset. One upward fiber and one downward fiber are shown in cartoon, whereas the other fibers and the wedge are shown in surface. The disulfide bonds are colored yellow. (B) Negative-staining electron microscopy of Pam3 virions (Left) or treated with the reducing agent dithiothreitol for 1 h (Middle) or 4 h (Right). To test whether the disulfide bonds are necessary for the assembly of the baseplate, Pam3 virions were treated with the reducing agent dithiothreitol and subjected to transmission electron microscopy. Almost all tail fibers of Pam3 were dissociated from the wedge upon dithiothreitol treatment (Fig. 6B). In addition, once treated with dithiothreitol for 4 h, the tails of a large fraction of Pam3 virions adopt a postcontracted state (Fig. 6 B, Rightmost). Therefore, we propose that these disulfide bonds are indispensable for stabilizing the interactions between the tail fibers and the wedge. Discussion Here, we solved the intact structure of the freshwater Myoviridae cyanophage Pam3, which possesses an icosahedral capsid of ~680 Å in diameter stabilized by cement proteins, followed by a contractile tail with a minimal baseplate. Analyses of these structures (in total 16 components) and their interfaces enabled us to better understand the structural protein assembly of a minimal Myoviridae phage. Assembly of the mature Pam3 phage particle involves the individual assembly of the capsid and the tail, which are subsequently joined together via the neck proteins (Fig. 7). Similar to other phages (27), assembly of the Pam3 capsid is initiated by the portal dodecamer, followed by assembly of an empty icosahedral capsid. Once genomic dsDNA is packaged into the capsid by the terminase complex (Fig. 7a), 12 adaptor subunits are recruited to the portal clip (Fig. 7b). Upon the recruitment of six subunits of the connector surrounding the exposed ending terminus of the genome, a capsid full of genomic DNA is formed (Fig. 7c). Fig. 7. A putative model for the assembly of Pam3. The capsid and tail of Pam3 are individually assembled. (a–c), After the completion of DNA packaging, 12 adaptor subunits (a) and six connector subunits (b) are recruited surrounding the exposed DNA terminus (c). (A and B) The hub-spike complex (A) assembles first, followed by the docking of the tube initiator gp17 and plug gp18 (B). (C) The core of the baseplate recruits sheath initiator gp21 and the wedge heterotriplex gp22–gp23. (D) Twelve trimeric tail fibers bind to the periphery of the wedge via disulfide bonds. (E) The baseplate recruits the six-stranded TMP. (F) A total of 132 subunits of the tube proteins from the six-entry helical structure surrounding TMP. (G) Six terminator subunits are recruited to terminate tube elongation. (H) Assembly of 132 sheath subunits along the inner tube. (I) The mature Pam3 tail is formed. Tail maturation initiates with the assembly of the baseplate, which grows on the hub-spike complex gp19–gp20 (Fig. 7A), followed by docking of the tube initiator gp17 and plug gp18 (Fig. 7B). Once this inner core of the baseplate was formed, the wedge heterotriplex gp22–gp23 and the sheath initiator gp21 were sequentially attached to the periphery, forming a relatively rigid wedge (Fig. 7C). Around the wedge, 12 tail fibers of trimeric gp24 covalently bind to wedge proteins gp22&gp23 via disulfide bonds, forming a mature baseplate with 12 tail fibers arranged in alternating upward and downward configurations (Fig. 7D). Afterward, TMP, which is a sixfold helical bundle vertically fixes on the baseplate and provides a measure and scaffold for the growth of the tail (Fig. 7E). Surrounding the standing TMP, 138 tube subunits form the six-entry helical structure, starting from the tube initiator (Fig. 7F). Once the tail tube reaches the end of TMP, six subunits of the terminator are recruited and eventually terminate elongation of the tube (Fig. 7G). Outside of the tail tube, the sheath subunits are recruited to the sheath initiator and sequentially form a similar six-entry helical structure of a total 132 subunits (Fig. 7H). Upon reaching the terminator at the top of TMP, an intact tail is eventually formed (Fig. 7I). Finally, accompanied by docking the tail to the capsid via interactions between the terminator and the connector, a mature virion of Pam3 is completely assembled. Structural comparison revealed that the contractile tail of Pam3 resembles the CISs of known structures (6, 8, 25), with a similar composition of protein components (SI Appendix, Fig. S13 A–C). Structure-based phylogenetic analysis indicated that the Pam3 tail and CISs are grouped into the same branch (SI Appendix, Fig. S13 D and E), suggesting that they may have evolved from a common ancestor. Different from phage T4 (11), Pam3 possesses a minimal baseplate of only six components, and the domain compositions of these components, such as the wedge proteins gp22 and gp23, are also simplified, lacking extra insertion domains (SI Appendix, Fig. S10). In addition, the elongation of the sheath and the tube of Pam3 is terminated by a single protein, the terminator gp12, whereas in phage T4, two proteins, gp3 and gp15, are responsible for terminating the tube and the sheath, respectively (16). Furthermore, the sheath subunit gp13 of Pam3 contains only two domains (SI Appendix, Fig. S7C), compared with other phage sheath subunits that contain three or four domains (28). All these observations indicated that Pam3 harbors the essential and conserved components that are required for building an effective and functional myophage. As Pam3 infects ancient cyanobacterial hosts, which have lived on Earth for over 3.5 billion years (29), it might be developed into an ideal chassis phage for future applications in synthetic biology. The present structure of Pam3 also provides insights into the mechanism of sheath contraction during phage infection. Previous studies on extracellular CIS (eCIS) showed that reorientation and dissociation of wedge heterotriplexes initiates a cascade of events and relays the contraction signal to the sheath (6, 25). Given the high resemblance of the wedge between Pam3 and the eCIS, we speculate that the Pam3 baseplate and tail sheath should also undergo sequential conformational changes upon host recognition and infection. Attachment of Pam3 to the host is mediated by 12 tail fibers, which are covalently linked to the baseplate and adopt alternating upward and downward configurations. Notably, the mass spectrometry analysis only identified three of the six disulfide bonds, indicating that the cross-linking event between tail fiber and baseplate is sporadic. In addition, compared with the rigid baseplate, the 12 tail fibers are relatively flexible and dynamic, as reflected by rather low local resolutions of the corresponding cryo-EM maps (SI Appendix, Fig. S2). These flexibly tethered tail fibers of Pam3 should increase the searching space for the host receptors and thereby promoting the capability of Pam3 to recognize and infect the host. Once Pam3 baseplate approaches the cell surface, a signaling cascade is triggered to drive the sheath contraction in a way similar to other myophages and CISs (11, 30, 31). For Pam3, redox signaling should play a role in reorienting the wedge heterotriplexes and eventually initiating sheath contraction. Reduction of the disulfide bonds within the subunits of the wedge, especially those between the heterotriplexes, most likely leads to the dissociation of the ring structure of the wedge. Consequently, six plug gp18 proteins are released, which further promotes the dissociation of the central hub-spike. Without an interlocking anchor of the hub-spike, the sheath is contracted toward the neck, accompanied by the injection of genomic DNA to the host cytosol through the tube channel. The detailed mechanism of sheath contraction and whether the cross-linking physiologically relates to Pam3 assembly and infection need further investigations. In summary, we solved the intact structure of a freshwater cyanophage, which provides the basis for understanding the assembly pattern of a minimal myophage. The structures reveal the essential and highly conserved modules required for the assembly and action of a functional myophage, which might be a simplified and user-friendly model applied in synthetic biology. Methods Isolation and Purification of Pam3 Virions. The Pam3 virions were amplified by infecting the host Pseudoanabaena 1,806 isolated from Lake Chaohu. The Pam3 virion particles were purified by the methods described previously (32–34). Briefly, the crude lysate after phage infection was harvested by PEG-NaCl precipitation and differential centrifugation. This crude sample was further purified using a CsCl density gradient (1.25, 1.30, 1.35, 1.40, and 1.45 g/mL) at 100,000 × g for 4 h at 4 °C. The Pam3 band was dialyzed against the SM buffer (50 mM Tris-HCl pH 7.5, 100 mM NaCl, 10 mM MgSO4). For the reduction assays, we treated Pam3 with 0.5 M dithiothreitol at 25 °C for 0, 1, and 4 h, respectively. The purity and integrity of isolated Pam3 virions were checked by the negative-stain EM. Cryo-EM Sample Preparation and Data Collection. The cryo-EM samples were prepared with a Vitrobot (FEI). 3.5 μL Pam3 samples were applied on a Quantifoil R2/1 Cu 200 mesh grid pretreated with 1 mg/mL poly-L-lysine for 10 min. Then the grids were blotted for 2 s at a blot force of −3 in 100% humidity. The cryo-EM data were collected under 300 kV FEI Titan Krios microscope equipped with a K3 Summit direct electron camera in counting mode under defocus range of −1 to −2 μm at University of Science and Technology of China. Automated data acquisition was performed with EPU software (Thermo Fisher Scientific), with a nominal magnification of 81,000×, which yielded a final pixel size of 1.0 Å. A total of 6,087 movies (32 frames, each 0.16 s, total dose 50 e−/Å2) were collected, and the defocus value for each micrograph was determined using CtfFind (35). Cryo-EM Data Processing. A total of 130,253 particles (800 × 800 pixels) were manually picked and extracted from the 6,087 micrographs and background-normalized using RELION 3.1 (36). After several rounds of 2D and 3D classifications, 129,239 good particles were selected for the further 3D refinement, yielding a final resolution of 3.17 Å with icosahedral symmetry imposed. The structures of the portal vertex and neck were solved by the sequential localized classification and symmetry relaxation, as reported previously (34, 37). The brief protocols are as follows (SI Appendix, Fig. S1): i) using the relion command “relion_particle_symmetry_expand,” we expanded the I3 symmetry of the particles generating 60 orientations for each particle. ii) We defined the orientation for each of the 12 vertices from the 60 orientations using the script developed by Liu (37). The d parameters (the distance from the center of the reconstructed capsid to the location of the subparticles to be reached) of the portal vertex and neck are 300 and 600 pixels, respectively. Then, the subparticles containing the 12 vertices were further extracted from the phage particle image. iii) We classified all vertex subparticles with fivefold symmetry without a rotational orientation search. Among the five classes, one class that contains 8.3%, or approximately 1/12 of the total subparticles, exhibits markedly different structures, which obviously should be the portal vertex (SI Appendix, Fig. S1). iv) We expanded the fivefold symmetry of the portal vertex, generating five unique orientations for each subparticle. Then, we applied 3D classification with 12-fold symmetry imposed but without rotational orientation search, yielding five classes of similar structures, each of which contains approximately one-fifth of the symmetry expanded subparticles. We chose the class with ~20% of symmetry-expanded subparticles for local 3D refinement with 12-fold symmetry imposed, yielding a 3.57 Å map of the portal dodecamer. By imposing the C1 symmetry, we obtained the structure of the portal vertex complex at 3.97 Å resolution. v) The neck is reconstructed in a similar way as the portal vertex. The image center was moved 600 along the portal axis to subtract the subparticles of the neck. The following 3D refinement was performed by imposing C6 symmetry, yielding final resolutions of 3.66 for the neck. The structures of the tail tube and sheath were determined by helical reconstructions in RELION 3.1. Images of 226,730 tail segments were extracted with an inter-box distance of approximately one helical rise (~40 Å). After several rounds of 2D and 3D classifications, 200,090 good tail segments were eventually selected to perform the 3D refinement imposed helical C6 symmetry, yielding a final resolution of 2.90 Å. The refined helical twist and rise are 33.8° and 37.2 Å, respectively. For the baseplate region, the corresponding segments were extracted and subjected to 2D and 3D classifications. Finally, 45,155 good particles were selected to perform the 3D refinement with C3 symmetry, yielding a final resolution of 3.19 Å. To further improve the resolution of the fiber, the center of the 3D volume was adjusted, and the segments were reextracted with a box size of 200 × 200 pixels. The following 3D refinement was performed by imposing C1 symmetry, yielding a final resolution of 3.96 Å. Model Building and Refinement. The initial models for the protein components of Pam3 were generated by AlphaFold2 (38). Then the models were automatically fitted into the corresponding maps using Chimera (39). Afterward, the models were manually adjusted and rebuilt by COOT (40) followed by the automatic refinement using the real-space refinement in PHENIX (41). After several rounds of iterations, the final models of each protein component perfectly match the cryo-EM maps (SI Appendix, Fig. S14), which were further evaluated by Molprobity (42). The cryo-EM parameters, data collection, and refinement statistics are summarized in SI Appendix, Table S2. The structure figures were prepared using Chimera (39), ChimeraX (43) or PyMOL (https://pymol.org/2/). Mass Spectrometry Analysis of Disulfide Bonds. The LC/tandem mass spectrometry (MS/MS) analyses were performed to identify the disulfide bonds formed between tail fiber and baseplate using the intact Pam3 virions. The Pam3 virions were precipitated with acetone, and the protein pellet was dried using a SpeedVac for 1 to 2 min. The pellet was subsequently dissolved in 8 M urea, 100 mM Tris-HCl, pH 7.5. The protein mixture was diluted four times and digested overnight with chymotrypsin at 1:50 (w/w) (Promega). The digested peptide solutions were desalted using a MonoSpinTM C18 column (GL Science) and dried with a SpeedVac. The peptide mixture was analyzed by a home-made 30-cm-long pulled-tip analytical column (75 μm ID packed with ReproSil-Pur C18-AQ 1.9 μm resin, Dr. Maisch GmbH, Germany), the column was then placed in-line with an Easy-nLC 1200 nano HPLC (Thermo Scientific) for mass spectrometry analysis. Data-dependent tandem mass spectrometry (MS/MS) analysis was performed with a Q Exactive Orbitrap mass spectrometer (Thermo Scientific). Peptides eluted from the LC column were directly electrosprayed into the mass spectrometer with the application of a distal 2.5-kV spray voltage. A cycle of one full-scan MS spectrum (m/z 300 to 1,800) was acquired followed by top 20 MS/MS events, sequentially generated on the first to the twentieth most intense ions selected from the full MS spectrum at a 28% normalized collision energy. Peptides containing the disulfide bonds were identified using pLink2 software (pFind Team) as described previously (44, 45). Supplementary Material Appendix 01 (PDF) Click here for additional data file. We thank Dr. Yong-Xiang Gao for technical support on cryo-EM data collection of Pam3 at the Cryo-EM Center, University of Science and Technology of China. We thank Yue Yin and Dr. Chao Peng of the Mass Spectrometry System at the National Facility for Protein Science in Shanghai (NPFS) for providing technical support and assistance in data collection and analysis. This research was supported by the Ministry of Science and Technology of China (https://www.most.gov.cn/index.html; project number 2018YFA0903100), the National Natural Science Foundation of China (http://www.nsfc.gov.cn; grant number U19A2020), the Strategic Priority Research Program of the Chinese Academy of Sciences (http://www.cas.cn; grant number XDB37020301) and the Fundamental Research Funds for the Central Universities (grant number WK2070000195). Y.-L.J. thanks the Youth Innovation Promotion Association of Chinese Academy of Sciences for their support (Membership No. 2020452). Author contributions Y.-L.J., Y.C., Q.L., and C.-Z.Z. designed research; F.Y., J.-T. Z., J. Z., K.D., R.-C.Y., Z.-L.W., W.-W.K., N.C., and W.-F.L. performed research; F.Y. and Y.-L.J. analyzed data; and F.Y., Y.-L.J., and C.-Z.Z. wrote the paper. Competing interest The authors declare no competing interest. Data, Materials, and Software Availability All study data are included in the article and/or SI Appendix. Atomic coordinates and EM density maps of the capsid (PDB: 8HDT; EMBD: EMD-34680), portal-adaptor (PDB: 8HDS; EMBD: EMD-34679), neck (PDB: 8HDR; EMBD: EMD-34678), sheath-tube (PDB: 8HDW; EMBD: EMD-34681), baseplate (PDB: 7YFZ; EMBD: EMD-33802), fiber (PDB: 7YFW; EMBD: EMD-33799) in this paper have been deposited in the Protein Data Bank and the Electron Microscopy Data Bank, respectively. Supporting Information This article is a PNAS Direct Submission. ==== Refs 1 R. W. Hendrix, M. C. M. Smith, R. N. Burns, M. E. Ford, G. F. Hatfull, Evolutionary relationships among diverse bacteriophages and prophages: All the world’s a phage. Proc. Natl. Acad. Sci. U.S.A. 96 , 2192–2197 (1999).10051617 2 A. Fokine, M. G. Rossmann, Molecular architecture of tailed double-stranded DNA phages. Bacteriophage 4 , e28281 (2014).24616838 3 D. Veesler, C. Cambillau, A common evolutionary origin for tailed-bacteriophage functional modules and bacterial machineries. Mol. Biol. Rev. 75 , 423–433 (2011). 4 B. Hu, W. Margolin, I. J. Molineux, J. Liu, Structural remodeling of bacteriophage T4 and host membranes during infection initiation. Proc. Natl. Acad. Sci. U.S.A. 112 , E4919–E4928 (2015).26283379 5 O. Salih , Atomic structure of Type VI contractile sheath from Pseudomonas aeruginosa. Structure 26 , 329–336.e323 (2018).29307484 6 P. Ge , Action of a minimal contractile bactericidal nanomachine. Nature 580 , 658–662 (2020).32350467 7 N. J. Shikuma , Marine tubeworm metamorphosis induced by arrays of bacterial phage tail-like structures. Science 343 , 529–533 (2014).24407482 8 F. Jiang , Cryo-EM structure and assembly of an extracellular contractile injection system. Cell 177 , 370–383 (2019).30905475 9 M. L. Yap , Role of bacteriophage T4 baseplate in regulating assembly and infection. Proc. Natl. Acad. Sci. U.S.A. 113 , 2654–2659 (2016).26929357 10 P. G. Leiman , Morphogenesis of the T4 tail and tail fibers. J. Virol. 7 , 355 (2010). 11 N. M. Taylor , Structure of the T4 baseplate and its function in triggering sheath contraction. Nature 533 , 346–352 (2016).27193680 12 K. Du , Comparative genomic analysis of five freshwater cyanophages and reference-guided metagenomic data mining. Microbiome 10 , 128 (2022).35974417 13 C. R. Buttner, Y. Wu, K. L. Maxwell, A. R. Davidson, Baseplate assembly of phage Mu: Defining the conserved core components of contractile-tailed phages and related bacterial systems. Proc. Natl. Acad. Sci. U.S.A. 113 , 10174–10179 (2016).27555589 14 Y. Cherrak , Biogenesis and structure of a type VI secretion baseplate. Nat. Microbiol. 3 , 1404–1416 (2018).30323254 15 Y. Chaban , Structural rearrangements in the phage head-to-tail interface during assembly and infection. Proc. Natl. Acad. Sci. U.S.A. 112 , 7009–7014 (2015).25991862 16 L. Zhao, S. Kanamaru, C. Chaidirek, F. Arisaka, P15 and P3, the tail completion proteins of bacteriophage T4, both form hexameric rings. J. Bacteriol. 185 , 1693–1700 (2003).12591887 17 R. L. Duda, C. M. Teschke, The amazing HK97 fold: Versatile results of modest differences. Curr. Opin. Virol 36 , 9–16 (2019).30856581 18 C. L. Dedeo, C. M. Teschke, A. T. Alexandrescu, Keeping it together: Structures, functions, and applications of viral decoration proteins. Viruses 12 , 1163 (2020).33066635 19 M. Feiss, V. B. Rao, The bacteriophage DNA packaging machine. Adv. Exp. Med. Biol. 726 , 489–509 (2012).22297528 20 R. Linares, C. A. Arnaud, S. Degroux, G. Schoehn, C. Breyton, Structure, function and assembly of the long, flexible tail of siphophages. Curr. Opin. Virol. 45 , 34–42 (2020).32777752 21 J. L. Kizziah, K. A. Manning, A. D. Dearborn, T. Dokland, Structure of the host cell recognition and penetration machinery of a Staphylococcus aureus bacteriophage. PLoS Pathog. 16 , e1008314 (2020).32069326 22 I. Katsura, Determination of bacteriophage-lambda tail length by a protein ruler. Nature 327 , 73–75 (1987).2952887 23 J. Xu, R. W. Hendrix, R. L. Duda, Chaperone-protein interactions that mediate assembly of the bacteriophage lambda tail to the correct length. J. Mol. Biol. 426 , 1004–1018 (2014).23911548 24 Z. L. Wei , Structural insights into the chaperone-assisted assembly of a simplified tail fiber of the Myocyanophage Pam3. Viruses 14 , 2260 (2022).36298815 25 A. Desfosses , Atomic structures of an entire contractile injection system in both the extended and contracted states. Nat. Microbiol. 4 , 1885–1894 (2019).31384001 26 M. Dunne , Salmonella phage S16 tail fiber adhesin features a rare polyglycine rich domain for host recognition. Structure 26 , 1573–1582.e1574 (2018).30244968 27 H. E. White Bacteriophages: Their structural organisation and function. In Bacteriophages: Perspect and future (eds. Savva, R.) 1-32 (IntechOpen London, 2019). 28 P. G. Leiman, M. M. Shneider, Contractile tail machines of bacteriophages. Viral Mol. Mach. 726 , 93–114 (2012). 29 W. Altermann, J. Kazmierczak, A. Oren, D. T. Wright, Cyanobacterial calcification and its rock-building potential during 3.5 billion years of Earth history. Geobiology 4 , 147–166 (2006). 30 S. Bhatt, P. G. Leiman, N. M. I. Taylor, Tail structure and dynamics in Encyclopedia of Virology, B. Dennis, Z. Mark), Eds. Elsevier Netherlands, 2020), pp. 186–193. 31 R. C. Guerrero-Ferreira , Structure and transformation of bacteriophage A511 baseplate and tail upon infection of Listeria cells. EMBO J. 38 , e99455 (2019).30606715 32 F. Yang , Genomic analysis of Mic1 reveals a novel freshwater long-tailed cyanophage. Front. Microbiol. 11 , 484 (2020).32322241 33 H. Jin , Capsid structure of a freshwater Cyanophage Siphoviridae Mic1. Structure 27 , 1508–1516.e1503 (2019).31378451 34 J. T. Zhang , Structure and assembly pattern of a freshwater short-tailed cyanophage Pam1. Structure 30 , 240–251 (2022).34727518 35 A. Rohou, N. Grigorieff, CTFFIND4: Fast and accurate defocus estimation from electron micrographs. J. Struct. Biol. 192 , 216–221 (2015).26278980 36 S. H. W. Scheres, Amyloid structure determination in RELION-3.1. Acta Crystallogr. D 76 , 94–101 (2020). 37 Y. T. Liu, J. Jih, X. Dai, G. Q. Bi, Z. H. Zhou, Cryo-EM structures of herpes simplex virus type 1 portal vertex and packaged genome. Nature 570 , 257–261 (2019).31142842 38 J. Jumper , Highly accurate protein structure prediction with AlphaFold. Nature 596 , 583–589 (2021).34265844 39 E. F. Pettersen , UCSF Chimera–a visualization system for exploratory research and analysis. J. Comput. Chem. 25 , 1605–1612 (2004).15264254 40 P. Emsley, K. Cowtan, Coot: Model-building tools for molecular graphics. Acta Crystallogr. D 60 , 2126–2132 (2004).15572765 41 P. D. Adams , PHENIX: A comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr. D 66 , 213–221 (2010).20124702 42 V. B. Chen , MolProbity: All-atom structure validation for macromolecular crystallography. Acta Crystallogr. D 66 , 12–21 (2010).20057044 43 E. F. Pettersen , UCSF ChimeraX: Structure visualization for researchers, educators, and developers. Protein. Sci. 30 , 70–82 (2021).32881101 44 B. Yang , Dong identification of cross-linked peptides from complex samples. Nat. Methods 9 , 904–906 (2012).22772728 45 S. Lu , Dong mapping native disulfide bonds at a proteome scale. Nat. Methods 12 , 329–331 (2015).25664544
PMC009xxxxxx/PMC9942836.txt
==== Front Proc Natl Acad Sci U S A Proc Natl Acad Sci U S A PNAS Proceedings of the National Academy of Sciences of the United States of America 0027-8424 1091-6490 National Academy of Sciences 36656858 202208536 10.1073/pnas.2208536120 videoVideoresearch-articleResearch Articlebiophys-bioBiophysics and Computational Biology408 Biological Sciences Biophysics and Computational Biology Twist response of actin filaments Bibeau Jeffrey P. a Pandit Nandan G. a Gray Shawn a Shatery Nejad Nooshin a https://orcid.org/0000-0002-1013-7540 Sindelar Charles V. a https://orcid.org/0000-0002-6646-7776 Cao Wenxiang a De La Cruz Enrique M. [email protected] a 1 https://orcid.org/0000-0003-4798-2892 aDepartment of Molecular Biophysics and Biochemistry, Yale University, New Haven, CT 06520 1To whom correspondence may be addressed. Email: [email protected]. Edited by Dimitrios Vavylonis, Lehigh University, Bethlehem, PA; received June 1, 2022; accepted December 16, 2022 by Editorial Board Member Yale E. Goldman 19 1 2023 24 1 2023 19 7 2023 120 4 e22085361201 6 2022 16 12 2022 Copyright © 2023 the Author(s). Published by PNAS. 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This article is distributed under Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND). Significance How actin filaments respond to mechanical loads is central to understanding cellular force generation and mechanosensing. While there is consensus on the actin filament bending stiffness, reported values of the filament torsional stiffness vary by almost 2 orders of magnitude. We used magnetic tweezers and hydrodynamic flow to determine how filaments respond to applied twisting and pulling loads. Twisting causes filaments to adopt a supercoil conformation. Pulling forces inhibit supercoil formation and fragment filaments. These observations explain how contractile forces generated by myosin motors accelerate filament severing by cofilin regulatory proteins in cells. Actin cytoskeleton force generation, sensing, and adaptation are dictated by the bending and twisting mechanics of filaments. Here, we use magnetic tweezers and microfluidics to twist and pull individual actin filaments and evaluate their response to applied loads. Twisted filaments bend and dissipate torsional strain by adopting a supercoiled plectoneme. Pulling prevents plectoneme formation, which causes twisted filaments to sever. Analysis over a range of twisting and pulling forces and direct visualization of filament and single subunit twisting fluctuations yield an actin filament torsional persistence length of ~10 µm, similar to the bending persistence length. Filament severing by cofilin is driven by local twist strain at boundaries between bare and decorated segments and is accelerated by low pN pulling forces. This work explains how contractile forces generated by myosin motors accelerate filament severing by cofilin and establishes a role for filament twisting in the regulation of actin filament stability and assembly dynamics. actin torsion plectoneme cofilin severing HHS | NIH | National Institute of General Medical Sciences (NIGMS) 100000057 R35-GM136656 Enrique M De La Cruz NIGMS R01 GM 110530 Charles V. Sindelar DOD | Multidisciplinary University Research Initiative (MURI) 100014036 W911NF1410403 Enrique M De La Cruz ==== Body pmcCells sense, respond, and adapt to internal and external forces (1, 2). The actin cytoskeleton, a dynamic, branched, and cross-linked network of protein filaments (3) that behave as semiflexible polymers on cellular length scales (4–6), mediates many of these cellular responses. Cellular actin networks are pulled (7, 8), squeezed (9, 10), and twisted (11, 12) during growth and remodeling and through interactions with contractile and regulatory binding proteins (13). These physical forces can stall network growth (14), alter the filament structure (15, 16), modulate interactions among filaments (17) and with regulatory proteins (7, 8, 18, 19), and induce filament fragmentation (15, 19–23), all of which influence network remodeling and mediate cellular “mechanosensing” (24, 25). The capacity for actin networks to respond to force is dictated by the mechanical properties of filaments. Relaxed (i.e., resting) filaments are straight but helical with an intrinsic twist (26). The forces required to twist and bend a filament scale with the filament mechanical properties (4, 27), specifically their bending and torsional stiffness, which are commonly represented in terms of bending and twisting persistence lengths (LB and LT; we note these are effective persistence lengths because filaments are not homogeneous, isotropic materials). Filaments with larger persistence lengths are stiffer and require more force to deform than those with shorter persistence lengths. Similarly, stiff filaments store more elastic strain energy for any given deformation than more compliant ones. The elastic free energy (i.e., strain energy) stored in the filament shape (16) can generate force and work when relaxing to the resting configuration. It can also fragment filaments (15, 19–23, 28) and mediate interactions with binding partners (7, 8, 18, 19). Dissipation of elastic energy in bent filaments contributes to force generation at the leading edge of migrating cells (10, 29) and during essential cellular processes such as endocytosis (9). Twisted filaments are also strained. Such twisting has been implicated in symmetry breaking (30, 31), network chirality (11), and the buckling of actin networks in filopodia (12). Quantitative knowledge of filament bending and twisting mechanics is therefore critical to reliably account for and model complex cellular behaviors. The bending mechanics of actin filaments have been extensively characterized. There is general agreement that filaments have a bending persistence length (LB) of ~10 µm (32–34), which can be modulated by regulatory proteins (35, 36) and ligands (5, 33, 37, 38). A consensus on the filament twisting stiffness is lacking, with torsional persistence lengths reported from 0.5 to 20 µm (20, 39–44). In addition, it is not known how filament twisting and bending are coupled (16, 27) or how filaments respond to combinations of twisting, bending, and pulling forces, as experienced in cells. Here, we use a magnetic tweezers apparatus coupled with microfluidics to evaluate how single actin filaments respond to applied twisting and pulling loads. Our results and analyses provide multiple, independent determinations of the filament bending and twisting stiffness, demonstrate how bending and twisting are coupled, and show how this coupling is affected by pulling and filament fragmentation. These findings have implications for actin cytoskeleton mechanosensing and network force generation and remodeling. Results Twisting and Pulling Actin Filaments. We developed an assay to twist actin filaments about their long axis with magnetic tweezers while simultaneously visualizing them by a TIRF microscope (Fig. 1). Short, Alexa 488–labeled actin filament seeds were tethered to the surface of the microscope coverslip and elongated from the barbed end with purified, Alexa 647–labeled actin monomers. These filaments were further elongated from their barbed ends with digoxigenin-conjugated actin monomers to which we attached a paramagnetic bead. The barbed-end–conjugated paramagnetic bead was twisted at a constant rate (0.31 rot s−1) with permanent magnets mounted on a stepper motor (Fig. 1B). Filament-attached beads and filaments rotated in phase with the permanent magnets (Fig. 1 B–D and Movies S1 and S2), indicating that no slipping occurs during manipulation. Filaments attached to two beads (Fig. 1B) were used only to determine whether rotations were in phase with the permanent magnet (Fig. 1C); filaments attached to a single bead were used in all subsequent experiments. Fig. 1. Twisting actin filaments with magnetic tweezers. (A) Cartoon schematic of the experimental setup. Alexa 488–labeled actin filament seeds (green) were attached to a Biotin-PEG-Silane surface through biotin (yellow circles) and neutravidin (black diamonds) interactions and elongated from the barbed ends with Alexa 647–labeled actin (red). Filaments were further elongated from their barbed ends with digoxigenin (DIG)-labeled actin (purple). Paramagnetic beads (2.8 μm in diameter; gray) coated with DIG antibodies (purple) were attached to filaments at or near their barbed ends. Filament-attached beads can be rotated by a permanent magnet (blue and red rectangles) and pulled by buffer flow (black arrow). Relative filament and bead sizes are not drawn to scale. Twisting clockwise or counterclockwise corresponds to under- or overtwisting, respectively. (B) Rotation of a phalloidin-decorated filament attached to two paramagnetic beads. (C) Cosine of the rotational angle of the second paramagnetic bead (black trace) and the magnet (red trace) indicates that the bead and magnet rotation are in phase. (D) Rotation of an Alexa 647–labeled actin filament with visible attachment to paramagnetic bead indicates that the bead and filament rotation are in phase. Pulling forces exerted by the permanent magnet are negligible in our experimental setup (SI Appendix, Fig. S1), so they were applied with fluid flow using a microfluidic device. The force applied on the filament scales with the size of the bead (d = 2.8 µm) and the fluid velocity (SI Appendix). This force is constant throughout the filament (at a constant flow rate) and is independent of the filament length in contrast to the much smaller and negligible forces exerted by flow on tethered filaments without conjugated beads (17, 22, 45, 46). Actin Filament Force–Extension Response. To establish the mechanical properties of actin filaments can be reliably determined with our experimental conditions, we measured the bending persistence length (LB) in the absence of twisting from the force–extension response of filaments undergoing thermally driven shape fluctuations (Fig. 2 and Movie S3). The pulling force required to straighten thermally bent filament scales with the bending stiffness (LB). The filament end-to-end distance (R), defined as the linear distance from the bead and surface attachment points, depends on the long-axis pulling force (Fig. 2). In the absence of fluid flow (i.e., pulling force ~ 0) filaments undergo thermally driven bending, so R is shorter than the filament contour length (L), i.e., R/L < 1. At ~ 0.02 pN pulling force, the ratio of the filament end-to-end distance and the contour length (R/L) was ~0.92 (Fig. 2). The end-to-end distance approached the contour length R/L ~ 1 at >3 pN pulling force (Fig. 2). The best fit of the force dependence of the filament end-to-end distance (Fig. 2) to a worm-like chain model [Eq. 1; (47)] yields an actin filament bending persistence length (LB) of 10.7 (±1.0) µm (Table 1), consistent with previous wet lab and computational model determinations of bare filaments [i.e., without phalloidin or other binding partners (19, 32–35, 42, 48–50)]. The forces in these experiments are calculated using a bead-center distance from the surface (d) equal to the bead radius (r; d = r) since the bead was at the surface of coverslip (SI Appendix). Fig. 2. Actin filament force–extension response. (A) Representative fluorescent images of Alexa 647–labeled actin filaments (magenta) conjugated to a paramagnetic bead (cyan) under fluid flow. No magnetic field is applied. (B) Force–extension curves for actin filaments of varying lengths (colored points) with the global best fit to Eq. 1 (colored lines) with pulling forces at d = r (Methods). (C) Average force–extension curves, normalized to filament lengths, for 7 actin filaments and corresponding theory with the global best fit persistence length of 10.7 (±1) µm (solid red line) (Eq. 1). Theoretical force–extension curves (Eq. 1) in descending order with LB = 100, 5, 1, and 0.1 µm (dashed red lines). Uncertainty bars indicate standard error of the mean (SEM). Table 1. Actin filament bending and torsional persistence lengths Persistence Length (µm) Assay Bending (LB) 10.7 (±1.0) Force–extension, Fig. 2 10.9 (±2.6) Twist–extension, Fig. 3, f at d = r Twisting (LT) 11.7 (±2.2) Twist–extension, Fig. 3, f at d = r 12.9 (±2.4) Twist fluctuations, Fig. 4 8.2 (±0.2) Cryo-EM, refinement volume of 5 subunits, histogram fit, Fig. 5 5.5 (±0.2) Cryo-EM, refinement volume of 5 subunits, MLE* 4.4 (±2.7) Cryo-EM, refinement volume of 5 subunits, MCMC† 5.8 (±0.3) Cryo-EM, refinement volume of 1 subunit, histogram fit 7.0 (±1.1) Cryo-EM, refinement volume of 1 subunit, MLE* 6.3 (±3.3) Cryo-EM, refinement volume of 1 subunit, MCMC† *Maximum likelihood estimation (MLE, see Methods). †Markov chain Monte Carlo (MCMC, see Methods). Twisted Filaments Bend and Supercoil. Filaments undergo a series of shape transitions when continuously twisted with magnetic tweezers (Movie S4). In the absence of applied twist (and with or without pulling loads), filaments bend randomly due to thermally driven forces. Applied twisting (twist density <0.8 rot µm−1) causes filaments to bend further in a nonrandom manner, provided long-axis pulling forces are weak (≤0.03 pN; Fig. 3A), and the value of R continues to shorten gradually with twisting until a critical twist density (σs) is reached, at which point filaments form a looped segment. Additional twisting causes the linear, nonlooped filament segments to wind up and twist around each other, yielding an interwound, actin filament supercoil (called a plectoneme) like those observed with twisted DNA (51, 52). This transition is detected as a dramatic and abrupt reduction in R (Fig. 3B). Only a single loop (i.e., one plectoneme) per filament was observed in our experiments. Plectoneme formation is reversible and relaxed (i.e., straight) filaments can be recovered with untwisting (Movie S5). We note that rhodamine phalloidin–decorated actin filaments supercoil when subjected to high-intensity laser light (53), presumably due to photo-induced torsional strain. Fig. 3. Actin filament twist–extension response and supercoiling. (A) Representative fluorescent images of actin filament twist–extension with 0.01 (Top), 0.03 (Middle), and 0.25 (Bottom) pN pulling force (SI Appendix, Eq. S1 with d = r). (Scale bar, 5 µm.) (B) Twist–extension curves for actin filaments under 0.01 (white circles), 0.03 (gray circles), and 0.25 (black circles) pN pulling forces with the global best fit to Eq. 3 (red lines). Solid and dashed red lines differentiate model before and after plectoneme formation, respectively. Rotations along the positive x-axis indicate filament overtwisting, and rotations along the negative x-axis indicate undertwisting. The complete dataset represents 50 filaments and n > 3 for each experimental condition. Uncertainty bars represent SEM. The asymmetric look of red dashed fitting lines for over- and undertwists under the same pulling force is due to the different fixed parameters 1L¯ in the fitting (Methods) and not because of differences in response to applied over- and undertwist. Actin filaments are helical and can be described as having an intrinsic right-handed twist. Therefore, counterclockwise rotations increase the intrinsic twist (“overtwisting”), whereas clockwise rotations lower the intrinsic twist (“undertwisting”). Binding of cofilin/ADF regulatory proteins, for example, also undertwists filaments (54, 55). Filament plectoneme formation depends on the pulling force but not on the twisting direction (i.e., over- versus undertwisting), as indicated by the symmetry of the twist–extension response curves (Fig. 3B). The lack of a detectable asymmetry with over- and undertwisting is surprising given the intrinsic filament twist. However, this observed behavior likely arises from the fact that the deformations associated with plectoneme formation are modest at the subunit level and not in the regime in which differences between the two directions could be detected (16). Filaments adopt a plectoneme configuration at low twist densities (~0.2 rot µm−1) when the pulling force is low (<0.01 pN) but require more twist (~0.8 rot µm−1) at higher pulling forces (0.03 pN; Fig. 3). Plectonemes did not form when the pulling force was 0.25 pN, even up to twist densities of ~1 rot µm−1. The reversibility of plectoneme formation was also independent of the twisting direction. The plectoneme loop size also depends on the pulling force (SI Appendix, Fig. S2). The average loop radius was ~400 to 500 nm under 0.01 pN, whereas it was ~200 nm under 0.03 pN. The local filament curvatures at these radii are small compared with those predicted to significantly accelerate filament fragmentation (15), consistent with these plectoneme loops being stable throughout the duration of our experiments. Filament Twist–Extension Response. The experimental data presented thus far demonstrate that twisted actin filaments bend and adopt supercoiled plectoneme structures when pulling forces are low (<0.25 pN). This response originates from the intrinsic filament bending and twisting mechanics and the coupling between these two deformations (16, 52, 56). Accordingly, the actin filament mechanical properties can be extracted from the data with appropriate theory, analysis, and modeling. A two-state model used to describe plectoneme formation in DNA (52) (modified to include polymers with L~LB, such as actin filaments; Methods; Eq. 3) accounts for the twist–extension response and plectoneme formation of actin filaments over the range of pulling forces evaluated here (Fig. 3B). This model considers actin filament segments as semiflexible rods in either “straight” or plectoneme states. Pulling forces favor the straight configuration. Twisting introduces strain energy, which is dissipated by bending and subsequent plectoneme formation. Prior to reaching a critical twist density (σs) for plectoneme formation, twisted filaments bend to dissipate torsional strain, shortening R. Once a plectoneme has formed, applied twisting strain is dissipated through conversion of strained, straight state segments to more relaxed plectoneme configurations (i.e., straight segments shorten, while plectoneme segments elongate linearly with applied twisting loads). Since only the linear, nonlooped filament segments contribute to the R value (plectoneme segments do not), the R value decreases linearly with the applied twist until R = 0, at which the entire filament is in a plectoneme configuration (middle line in Eq. 3; Fig. 3B). The applied twist at R = 0 is referred to as σp. The best fit of the filament twist–extension data to this model (Eq. 3; Fig. 3B) yields actin filament persistence lengths for bending (LB) and twisting (LT) of 10.9 (±2.6) and 11.7 (±2.2) µm, respectively (Table 1). This value of the bending persistence length (LB) is comparable with the value of 10.7 (±1.0) µm obtained from the force–extension response (Fig. 2). Filament torsional persistence lengths an order of magnitude longer or shorter do not account well for the observed experimental data (SI Appendix, Fig. S5). We note that some twisted filaments “wobbled” slightly during twisting manipulations (Movie S2). To estimate the maximum possible error introduced by bead wobbling in these cases, we analyzed the data assuming a larger force as expected if the bead moved away from the surface during rotation. Deviations far greater than observed for wobbling (i.e., one full bead height deviation, such that d = 2r; SI Appendix) yield essentially identical LT values (LT = 11.7 ± 2.2 µm versus LT = 11.4 ± 2.1 µm for beads wobbling an entire bead height) but ~twofold lower LB values (10.9 ± 2.6 µm versus 6.5 ± 1.5 µm for beads wobbling an entire bead height). Thermally Driven, Filament Twist Fluctuations. We also determined the filament torsional stiffness by directly visualizing spontaneous, thermally driven, twist fluctuations (Fig. 4). A hollow, cylindrical magnet was positioned above the tethered filament to generate a magnetic field (perpendicular to the surface of the sample chamber) that held the filament orthogonally to the surface without constraining its rotation (57) (Fig. 4 A and B and Movie S6). The pulling force generated by the closely positioned hollow magnet maintains the filament relatively straight, thereby eliminating contributions from filament bending to the observed dynamics and allowing determination of the true filament torsional stiffness (SI Appendix, Fig. S3, (39, 56)). Fig. 4. Direct visualization of actin filament twisting fluctuations. (A) Cartoon schematic of the experimental setup. A cylindrical magnet with a center hole was positioned 5 mm above the coverslip surface–anchored actin filament (i.e., in the z direction perpendicular to the surface). A DIG-coated marker bead was added to the paramagnetic bead to track rotations. No flow was applied during experiments. Filament length and bending are not to scale. (B) (a) Assay is set up by identifying a filament attached to a paramagnetic bead (large dim bead) with identifiable marker beads (small bright bead) under fluid flow. (b) Fluid flow is turned off. (c) Cylindrical magnet is lowered into position. (d) Filament is pulled out of the focal plane. (e) Focal plane is adjusted to observe the rotational fluctuations of both beads. Images were taken every 5 s. (C) Example images of the angular fluctuations of the filament visualized by the absolute angle of a line connecting the two beads to the x direction. (D) The mean-subtracted absolute angle (SI Appendix, Eq. S25) of the marker beads over time. Black trace indicates the angular fluctuations from the filament tracked in (C). Gray traces represent four other sample traces from different experiments performed at different times. Histogram represents the distribution of absolute angles from the black trace. The actin filament torsional persistence length of 12.9 (±2.4) μm is an average (n = 5) of separate measurements, each was determined from the value of the variance at long times (see Panel E) and the filament length according to Eq. 4. (E) Time-dependent variance of the traces in D. It demonstrates that the measurement time of whole filament angular fluctuation has to be long enough for the variance to reach equilibrium such that experiencing all possibilities. Filament twisting fluctuations were monitored by tracking a smaller fluorescent marker bead attached to the paramagnetic bead at the filament barbed end (Fig. 4 C and D and Movie S7). The angular fluctuations of the marker bead reflect the cumulative rotational fluctuation of all subunits between the surface and paramagnetic bead attachment points (L = 8 to 19 μm, ~365 subunits µm−1; SI Appendix, Fig. S6). The rotation angle of the marker beads around the filament center axis fluctuates randomly (Fig. 4 C and D). Time courses of the observed marker bead angle variance (Fig. 4E) plateau at times >1,000 s, indicating the entire accessible diffusive space of the filament-attached bead had been sampled (58). The twist persistence length for each set of data in Fig. 4D was determined from the angular variance calculated by directly averaging the set of data and the filament length according to Eq. 4 (Methods). The averaged (n = 5) value from separate experiments yields a filament torsional persistence length of 12.9 (±2.4) µm (Table 1) comparable with the value of 11.7 (±2.2) µm determined from the filament twist–extension response and plectoneme formation (Table 1) and with reported values of ~16 µm (20) and ~6 µm (39). Single Subunit Twisting Fluctuations. We measured the filament twisting persistence length a third way, from the variance of twisting angles between subunits, as visualized by electron cryomicroscopy (Fig. 5). Alignment parameters output from the 3D structure refinement yield estimates of filament subunit orientations and hence the twisting angle between them (54). Deviations of the observed intersubunit twist angle from the intrinsic (average) filament twist reflect thermally driven twist fluctuations. The distribution (i.e., width) of these deviations scales with the filament torsional stiffness, such that filaments displaying a narrow distribution are less compliant in twisting than those with a broader distribution. Fig. 5. Torsional persistence length of actin filaments determined by electron cryomicroscopy. (A) Cartoon schematic of measured filament subunit twisting fluctuations. The top cartoon depicts an actin filament (gray) with the average, intrinsic twist (Δφ1intrinsic) between adjacent subunits i and i+1 illustrated as a red curved arrow. The observed twist deviates from the intrinsic twist, either over or under, because of thermal fluctuations. The middle and bottom cartoons illustrate an undertwisted filament (light gray) overlaying a canonical filament with an intrinsic twist (dark gray). Blue arrows illustrate the observed twist between subunits i and i+1 (Middle) or i and i+3 (Bottom), which differs from the intrinsic twist by Δφ’n (illustrated by black arrows). (B) Histograms of actin filament subunit twist fluctuations (Δφ’n, in degrees) estimated from cryo-EM alignment parameters (reference volume of 5 subunits) for n = 1, 10, and 50 subunits. Red lines represent fits to a normal distribution with mean zero and variance (σobs,n+12). (C) n dependence of the twist variance (σobs,n+12). The solid red lines represent the best fit to SI Appendix, Eq. S21. A challenge with accurately calculating this angle distribution by cryo-EM is the low signal-to-noise ratio associated with the images. The noise introduces uncertainty in the angle measurements that can exceed the true intersubunit angle variance. Since the torsional stiffness and persistence length (LT) are determined from the distribution variance (SI Appendix, Eq. S15), and large uncertainties in individual intersubunit angle measurements (SI Appendix, Eq. S17) yield a larger variance than the true variance, neglecting contributions from these uncertainties causes filaments to appear more compliant than they actually are. We therefore developed an analysis method (SI Appendix, Eqs. S17–S21) that addresses the uncertainty in angle measurement to accurately measure LT from cryo-EM micrographs. Because each filament subunit is independently subject to thermally driven torsional angle fluctuations in a scale dictated by the torsional stiffness, the width of the true angular distributions σn+12, measured across filament segments, increases linearly with the number of subunits according to σobs,n+12 = nΔsLT+σε2 (SI Appendix, Eq. S21). In contrast, the uncertainty in the estimated twist angle σε2 remains constant. Our cryo-EM images confirmed this behavior (Fig. 5C), yielding estimates of the true intersubunit torsional variance and persistence length (σn+12 and LT, respectively; SI Appendix, Eq. S21) from the slope of the line relating the observed variance (σε2) to n (Fig. 5C). The intercepts of these lines reflect the contribution of noise to σobs2 (SI Appendix, Eq. S21). The best linear fit of the n-dependent variance, obtained by Gaussian distribution fit to the angle Φ histogram measured from a refinement volume of 5 subunits, to SI Appendix, Eq. S21 yields an LT of 8.2 (±0.2; ± indicates SDs of the fit) comparable with the LT values of 11.7 (±2.2) and 12.9 (±2.4) µm determined by plectoneme formation (Fig. 3) and thermally driven filament torsional fluctuations (Fig. 4), respectively. We include two additional analysis methods, directly averaging (MLE) and MCMC (Methods), to independently determine n-dependent variance and thus LT to see if one is more robust than the others (Table 1). The differences among the three different methods are not significant. We also repeated the analysis to the angles measured from a refinement volume of 1 subunit (Table 1). These values are not significantly different from those from a refinement volume of 5 subunits. Twist-Induced Filament Fragmentation. Filaments often fragmented during twist–extension manipulations (Fig. 3). At 0.25 pN pulling force, filaments fragmented before forming a plectoneme, indicating that intersubunit bonds rupture if strain from applied twisting is not dissipated (15, 19). Most filaments (83/96) fragmented at the bead or surface attachment sites (twist density = 1.2 (±0.6) rot µm−1; discussed below), suggesting that the weakest mechanical elements are filament attachment points. The remaining events (13/96) could be reliably discerned as fragmentation within the filament. This occurred at a twist density of ~1.1 (±0.5) rot µm−1 (Movie S8). At low pulling forces (≤0.03 pN), filaments adopt a plectoneme configuration, which dissipates the torsional strain from applied twisting. Accordingly, no fragmentation was observed, and all filaments formed a plectoneme under 0.01 pN pulling force. At a pulling force of 0.03 pN, some fragmentation events were observed. These occurred at the onset or during plectoneme formation [twist density = 0.83 (±0.17); Fig. 6 and Movie S9], with 9/17 filaments fragmenting exclusively within the filament. Fig. 6. Twisted cofilactin filaments fragment more easily than twisted bare actin filaments. (A) Representative images of twist-induced fragmentation for undertwisted (UT) bare, overtwisted (OT) bare, undertwisted cofilin saturated, and overtwisted cofilin saturated filaments. (Scale bar, 4 µm.) (B) Survival analysis from the experiments in (A) at a pulling force of 0.03 pN. Log-rank test comparing UT bare to UT cofilin (P < 0.0001) and OT bare and OT cofilin (P = 0.0094). Both log-rank tests and Gehan–Breslow–Wilcoxon tests yielded similar P values, which conclude that the observed twisting response of bare and cofilin-decorated filaments is statistically different. Cofilin Promotes Twist-Induced Filament Fragmentation. Filaments saturated with the actin regulatory protein, cofilin, referred to as cofilactin [cofilin] = 2 μM, which is saturating for this yeast isoform under our conditions (35, 38), did not form a plectoneme, even at low (0.03 pN) pulling force, because they fragmented. Cofilactin filament fragmentation occurred at a lower twist density than fragmentation of bare filaments (Fig. 6 and Movies S10 and S11). The twist density dependence of the filament survival probability decay (Fig. 6) indicated a midpoint of ~ 1 rot µm−1 for fragmentation of bare actin, while the midpoint for cofilactin filaments was significantly lower (0.64 (±0.28) and 0.43 (±0.12) rot µm−1 for overtwisting and undertwisting, respectively; Fig. 6 and Movies S10 and S11). The observed fragmentation originates from twisting strain rather than flow-mediated forces as neither untwisted bare nor cofilactin filaments fragmented on the timescales of these experiments (SI Appendix, Fig. S4 and Movie S12). Discussion Actin Filament Torsional Persistence Length is ~10 μm. Here, we have shown through twist–extension (Fig. 3), filament rotational fluctuations (Fig. 4), and single subunit fluctuations (Fig. 5) that actin filaments have a torsional persistence length of ~10 μm, comparable with their bending persistence lengths (Fig. 2) (19, 32–35, 38, 42, 48–50, 59). The reported torsional persistence length of actin filaments varies significantly from 0.5 to 20 µm (20, 22, 39–44). Our measured LT is consistent with reported values determined in optical traps (20, 39), fluorescent polarization microscopy (22), and electron microscopy of filaments straightened with hydrodynamic flow (43) but differs from the shorter LT values determined with fluorescent polarization microscopy (41), negative stain electron microscopy (44), phosphorescence anisotropy (40), and molecular dynamics simulations (42). Our actin persistence lengths are about two orders of magnitude more rigid than DNA, which has an LT and LB of 100 and 50 nm, respectively (52), although these values depend greatly on solution conditions (e.g., salt composition and concentration). The large uncertainties in the measured rotation angles of individual actin subunits could contribute to the short LT values determined by electron microscopy (44). Uncertainty in subunit rotation angles overestimates the angular fluctuations of adjacent filament subunits and yields an artificially short twisting persistence length. Using an analysis method as in this work (Methods) that accounts for these uncertainties in rotational angles yields a larger (i.e., stiffer) torsional persistence length (Fig. 5). The discrepancy in the filament twisting persistence length values determined by fluorescence polarization and phosphorescence anisotropy (40, 41) may be due to the independent movement of protein side chains or subunit domains to which the spectroscopic probe is conjugated. Actin Filaments Fragment at a Twist Density of ~1 to 2 deg sub−1. In our twist–extension experiments with 0.25 pN pulling force, actin filaments fragmented at a twist density of 1 to 2 rot μm−1 (Figs. 3B and 7A). A twist density of 1 rot μm−1 is equivalent to ~1 deg rotation per actin subunit, which corresponds to a 1-deg change in relative twist between two laterally adjacent actin subunits and a 2-deg change in twist between two longitudinally adjacent actin subunits. This twist density introduces only 0.65 and 1.3 kBT of strain energy (SI Appendix, Eq. S4) at the lateral and longitudinal contacts of actin subunits (Fig. 7B), respectively, which is considerably less than the estimated bond energies associated with lateral (4 to 8 kBT) and longitudinal (12 to 20 kBT) filament contacts (26). Fig. 7. Modeling twist-induced fragmentation of actin filaments. (A) Experimental actin filament survival curves for undertwisted (blue) and overtwisted (black) bare actin filaments at 0.25 pN pulling force (2 µL min−1 flow rate) and a twisting rate of ω = 0.3 rot s−1. Data include instances where fragmentation occurs close to the bead or surface interfaces. For comparison, the plot includes simulations of filament survival curves as a function of twist density (SI Appendix, Eq. S31) at the same twist rate of 0.3 rot s−1 with a length of L = 15 µm (red trace), as that typical in our twist–extension experiments in this study, and a twist rate of ω = 2.2 rot s−1 with a length L = 0.1 µm (gray trace). Inset image is an example of twist-induced fragmentation. Inset graph is the model-predicted filament torque (SI Appendix, Eq. S9). (B) Model-predicted twisting strain energy per subunit (left y-axis, SI Appendix, Eq. S4) and the relative increase in fragmentation rate constant of strained relative to relaxed, native filaments (right y-axis, SI Appendix, Eq. S28). Dashed lines indicate the model-predicted twisting strain energy for the twist density imposed at boundaries of human cofilin clusters (blue) and by singly isolated bound human cofilin (red). Why then do actin filaments fragment at such low twist densities? For an actin filament to fragment, three intersubunit interfaces—two longitudinal and one lateral—must rupture simultaneously (15, 19). The combined twist strain energy in these three bonds of a filament twisted to a density of ~1 deg rotation per subunit is only ~3.2 kBT, more than an order of magnitude lower than the 44 kBT subunit−1 activation energy for fragmentation (15). Although the imposed twist strain energy does not directly overcome the activation energy for filament fragmentation, it does accelerate the filament fragmentation rate constant ~twofold (calculated from exp(Estrain/kBT); SI Appendix, Eq. S28) (15). While this effect may seem small, it accounts for the rapid fragmentation of twisted filaments observed in our experiments when other factors contributing to severing are considered. Two additional factors that contribute to the rapid fragmentation of twisted filaments are the number of potential severing sites and the time duration of the applied twisting load. Filament severing occurs at subunit interfaces, so long filaments have more potential fragmentation sites than shorter ones. That is, the severing reaction is a microscopic process, but observed filament severing is a macroscopic process that scales with the filament length, a collective effect happening at individual subunits. A typical filament in our experiments is >10 µm in length (>3,700 subunits), which means the observed, macroscopic severing rate constant is the microscopic severing rate constant times 3,700. In terms of fragmentation probability, if P is the microscopic fragmentation probability expressed in units subunit−1, the macroscopic probability for fragmentation of a filament that comprised n subunits is given by 1−(1−P)n ~ nP + 0(P2) (SI Appendix, Eq. S30). The duration (Δt) of applied twist deformation is a second critical factor contributing to the observed fragmentation. Twist loads introduce strain energy and accelerate filament fragmentation according to the Arrhenius equation (SI Appendix, Eq. S28) (15) from which the twofold acceleration is calculated above. However, the filament fragmentation probability (P) scales with the fragmentation rate constant (kfrag) and the duration of the applied load (Δt) according to P = 1− exp(−kfragΔt) (SI Appendix, Eq.  S29) (15). Therefore, fragmentation will not be significantly affected if the duration of the applied twist is short relative to the characteristic severing time (Δt  ≪ 1/kfrag). We derived the filament survival probability as a function of time for a given rate of applied twisting strain (SI Appendix, Eq. S31). Survival curves, as a function of twist density, were simulated according to SI Appendix, Eq. S31 under two different conditions: 1) slow rotation (~0.3 rot s−1 or 3.2 s for 1 rot) of a 15-μm filament, as carried out in our experiments, and 2) rapid rotation (2.2 rot s−1) of a short filament (0.1 μm), as previously modeled (15) (Fig. 7A). Long filaments with slow rotation break at ~2 rot μm−1 (duration 96 s), whereas the short filaments with fast rotation break at 5 rot μm−1 (duration 0.22 s). This behavior explains why filaments do not break under thermal twist fluctuation of ~1 deg sub−1 (Fig. 7A) and suggests that filaments can undergo rather large structural changes without fragmenting, provided the duration of these shape changes are short. Local Twist Strain Drives Filament Severing by Cofilin. The actin filament severing protein, cofilin, binds between two adjacent longitudinal actin subunits and undertwists the filament ~4.3 deg sub−1 (54, 55). The filament twist changes abruptly, within ~1 to 2 subunits, at boundaries between bare and cofilin-decorated segments (54, 55, 60). Filaments preferentially sever at these boundaries (45, 61, 62) within the bare actin side of the boundary (60). If we assume that the ~4.3-deg twist strain spreads evenly (54) among the two subunits at the boundary such that each experiences a twist change of ~2.1 deg, a twist density of ~2.1 deg sub−1 introduces strain energy of ~2.9 kBT sub−1 (12LTsσ2, SI Appendix, Eq. S4, Fig. 7B). The strain energy of this magnitude is predicted to accelerate the bare actin intrinsic severing rate constant ~18-fold (exp(2.9) using SI Appendix, Eq. S28; Fig. 7B). This value agrees with estimates of a boundary severing rate constant that is ~10 to 25 times faster than that of bare actin (32, 35, 38). It has been reported that boundary severing rate constants vary among cofilin isoforms [e.g., Saccharomyces cerevisiae severs more rapidly than human cofilin (32, 35, 38, 45, 61, 63, 64)]. Subtle differences in cofilin-induced twist could potentially account for large variability in severing, given the quadratic dependence of the severing rate constant on twist density (Fig. 7B). We note that the filament twist is constant and does not change within bare and cofilin-decorated segments and therefore does not introduce strain between subunits in those regions as it does at boundaries where twist discontinuities exist (62). A singly bound [i.e., isolated (65)] cofilin also changes the filament twist, although less than at a boundary and only at the one subunit to which it directly binds (55). Accordingly, a singly bound cofilin severs filaments but does so more slowly than boundaries (32, 38). Assuming the twist induced by isolated bound cofilin is ~ 1 to 2 deg, the local twist strain should accelerate fragmentation ~2 to 18-fold (Fig. 7B), consistent with the reported ~fivefold acceleration (32). Cofilin Renders Twisted Actin Filaments Brittle. Cofilactin filaments (i.e., saturated with cofilin) break at lower twist densities than bare actin filaments (Fig. 6). Several factors can potentially contribute to this mechanical instability. A twisted cofilactin filament could store more elastic strain energy than a bare filament for a given twist load, thereby resulting in more rapid fragmentation. While conceivable, cofilactin filaments are thought to be more compliant in bending (16, 19, 34, 42) and twisting (40, 42) than bare actin filaments so they have less strain energy for an identical shape deformation. It is also possible that cofilactin filaments are more fragile and fragment more easily than bare actin filaments. While this is also conceivable, cofilin bridging interactions stabilize cofilactin filaments and protect them from fragmentation (15, 60). These bridging interactions render fully decorated cofilactin filaments comparably stable to bare actin filaments (61, 62, 65–68). A third more likely possibility is that cofilin dissociation from actin filaments transiently introduces a boundary, which fragments more easily under twisting (and bending) loads (15, 19). Spontaneous cofilin dissociation can introduce this effect but it would be more prominent if twisting loads weakened cofilin binding and/or accelerated cofilin dissociation, as predicted from modeling studies (19). Pulling Accelerates Cofilactin Filament Fragmentation. Surface tethering and cross-linking constrain filament bending and twisting, which prevents the dissipation of cofilin-induced torsional strain, thereby enhancing cofilin severing activity (15, 19, 22, 67). Long-axis pulling forces on actin filaments, of magnitude comparable with those exerted by myosin motors [3 to 5 pN per ATP hydrolyzed (69)], also dampen thermally driven (Fig. 2) and twist-induced filament bending (Fig. 3) and dramatically accelerate filament fragmentation. This behavior supports models in which contractile forces generated by myosin motors rapidly sever twisted filaments such as those with bound cofilin (15, 19, 28). Recent studies show that contractile forces produced by myosin motors “catalyze” cofilin-mediated actin filament disassembly and turnover in Aplysia neurons (70) and also contribute to actin filament turnover during contractile ring constriction in S. pombe (71), demonstrating how combinations of pulling, bending, and twisting forces can dramatically accelerate actin filament network fragmentation and turnover in cells. Twisting a Filament Bundle. With the torsional persistence length measured in this study, it is possible to make predictions about the behavior of bundled actin filaments. We can model a perfectly cross-linked actin filament bundle as a filament with different dimensions such that the radius of the bundle cross-section (R) is determined by the radius of a single actin filament (r) and the number (n) of filaments comprising the bundle. The area of the cross-section of the bundle is the sum of the cross-sectional areas of each filament forming the bundle (i.e., πR2 = nπr2). Therefore, R2 = nr2, and the bundle’s torsional and bending persistence lengths become LT,bundle = G(πR4/2)/kBT = n2 G(πr4/2)/kBT = n2LT and LB,bundle = E(πR4/4)/kBT = n2 E(πr4/4)/kBT = n2LB, where G is the shear modulus, and E is the Young’s modulus of an actin filament, respectively. Therefore, a filament bundle’s torsional and bending persistence lengths and corresponding strain energies scale with n2. This suggests that twisted bundles are a result of very large applied torques as twisting a three-filament bundle requires nine times as much torque to twist compared with a single filament. The authors of this study (12) concluded that these torsional loads on actin bundles in filopodia are driven by myosin contractility, indicating the off-axis torques generated by myosin motors (72–74) are sufficiently strong to twist filament bundles. Materials and Methods A brief description of the experimental materials and methods used is provided here, and for more details, see SI Appendix. Protein Purification. Actin was purified from rabbit skeletal muscle and labeled on surface lysines with NHS ester derivatives of Alexa 488, Alexa 647, biotin, or digoxigenin (17). Alexa 488 phalloidin was purchased from Thermo Fisher (catalog# A12379). Ca2+-actin monomers (5 µM) were converted to Mg2+-actin by addition of 50 µM MgCl2 and 0.2 mM EGTA and equilibrated for 5 min on ice immediately before use (75). Saccharomyces cerevisiae cofilin with a surface-engineered cysteine was purified and labeled with Alexa 488 (61). Microscope Sample Preparation. Surface functionalization and passivation of microscope coverslips with 2 to 5% Biotin-labeled PEG-Silane slides were adapted from elsewhere (76). Microfluidic chambers were assembled as described (17, 77). Superparamagnetic Dynabeads™ M-270 Epoxy (2.8 µm in diameter, Thermo Fisher catalog #14301) were conjugated to antidigoxigenin antibody following the company-provided protocol. Samples were prepared and experiments carried out in KMI buffer (10 mM fluorescence-grade imidazole pH 7.0, 50 mM KCl, 2 mM MgCl2, 0.2 mM ATP, and 2 mM DTT) supplemented with 15 mM glucose, 0.02 mg mL−1 catalase, and 0.1 mg mL−1 glucose oxidase. Actin polymerization was done as previously described (65). Filament–bead conjugation in sample chambers is described in detail in SI Appendix. Microscopy. Imaging was conducted on a Till iMic digital total internal reflection fluorescence (TIRF) microscope equipped with a 100× objective (Olympus) and an Andor iXon897 electron-multiplying charge-coupled device (EMCCD) camera (17, 33, 38). Filament Force–Extension. In the force–extension experiments without twist, filament end‐to‐end length (R) was measured as a function of the tensile force (f) applied with buffer flow. Images were recorded at each unique buffer flow (tensile force) with an acquisition rate of 1 frame s −1. The positions of the paramagnetic beads were tracked with the TrackMate plugin in ImageJ (NIH, USA) (78). The value of R was determined as the direct straight-line distance between the two attachment points at the bead and surface. A filament contour length (L) was measured as R at a high flow rate to make the filament straight. The measured force dependence of R (i.e., force–extension curve) was fitted to the following equation describing the force–extension behavior of semiflexible polymers (6, 47):[1] R=L−ΔL=L1−12kBTLBfcothLfLBkBT−1LLBkBTf=→L→0L→L→∞L1−12kBTLBf, where LB is the filament bending persistence length, kB is Boltzmann constant, and T is the room temperature (296 K). ΔL is the deviation of R from L of a semiflexible polymer due to thermally driven random bending, and it does not scale with L linearly. For the polymers with very short L, ΔL ~ 0, the deviation per L is negligible and R ~ L, whereas with very long L, the deviation of R from L reaches the maximum per unit length 12kBTLBf. This nonlinear L dependence of ΔL is consistent with the conclusion by another study (79), which claims a popular force–extension equation (80) with ΔLL=const is overly simplified. Using Origin software (Originlab, Northampton, MA, USA), experimental replicates were globally fitted with LB as a global fitting parameter and L unique to each individual filament dataset (Fig. 2B). In addition to globally fitting individual filament force–extension curves, we compiled and fitted data for filaments of different lengths by averaging the normalized filament end‐to-end length (R/L), which is given as follows:[2] RL¯ = 1n∑i=1nRiLi= 1−12kBTLBf1+2n∑i=1ne−2LifLBkBT1−e−2LifLBkBT−LBkBTf1n∑i=1n1Li, ∼1-12kBTLBf+kBT2f1L,¯ because the term 2e-2LifLBkBT1-e-2LifLBkBT is <0.08 and can be ignored since it is ≪1 and more than 1 order of magnitude smaller than LBkBTf1Li in our experimental force and filament length ranges. Filament Twist–Extension. Filament images in the twist–extension experiments were recorded with an acquisition rate of 1 frame s −1 while twisting filament at a constant rate of 0.31 rot s−1 and applying a given tensile force (f) with hydrodynamic flow. The position of the paramagnetic beads was tracked with the same procedure, and filament end‐to‐end distance (R) and contour length (L) were determined in the same manner as those in the preceding Force–Extension section. R/L is a function of twist density, σ (unit: rot μm−1), and L. Filament contour lengths in our experiments range from 7 to 20 μm. Normalized R/L data were averaged by binning according to the filament twist density (rot μm −1) with a bin size of 0.025 rot µm −1. To describe the twist–extension behavior of actin filaments, an analytical two-state model developed for DNA (52) was modified for actin filaments. In the model, it is assumed that a filament subunit exists in one of two states when a filament is twisted: plectonemic or nonplectonemic (linear). A filament with a plectoneme consists of a mixture of plectonemic and nonplectonemic regions. The fraction of plectonemic regions increases linearly with twist density and does not contribute to R/L. Eq. 3 (SI Appendix, Eq. S10) gives the force dependence of R/L contributed from only the fraction in the nonplectonemic state (52) (SI Appendix) as follows:[3] RL=1−12kBTLBf+kBT2f1L¯−ω02σ24LBLTfLBkBT+12kBTLB3f, σ<σsσp−σσp−σs1−12kBTLBf+kBT2f1L¯−ω02σs24LBLTfLBkBT+12kBTLB3f, σs≤σ<σP0, σp≤σ. In the equation, σs is a special applied twist density at which filaments form plectonemes, and σp is applied twist density when the entire filament is plectonemic. The fraction of subunits in the nonplectonemic conformation is given by σp-σσp-σs. A custom routine was written with Origin software to globally fit the experimental data of R/L as a function of applied twist at different pulling forces to Eq. 3 (Fig. 3) with LB and LT as shared unconstrained parameters. σs for pulling force of 0.01 and 0.03 pN were fixed parameters from averaging experimentally observed values, but the other σs for pulling force of 0.25 pN and value of σp for all forces were also unconstrained during the fitting procedure but were not shared because they depend on and vary with the pulling force. The value of 1L¯ was constrained during fitting to the values calculated from the actual filament contour lengths measured under a high pulling force (to straighten). Design of Freely Orbiting Magnetic Tweezers. The rotational fluctuations of actin filaments were measured with freely orbiting magnetic tweezers (57). Five cylindrical magnets (R422-N52, K&J Magnetics) arranged in series were mounted on a linear XY micrometer stage (XR25, Thorlabs) and centered directly above the microscope objective while visualizing with wide-field microscopy. The procedure for conjugating antidigoxigenin paramagnetic beads to digoxigenin marker beads for these experiments is given in detail in SI Appendix. Analysis of Filament Rotational Fluctuations. In whole filament twisting fluctuation experiments (Fig. 4A), we monitor how the angle θ(t) between the two filament attachment points (at the surface and on the bead, length L) changes over time, and the variance (σθ2) of the angle fluctuation is given by Eq. 4 (SI Appendix, Eq. S22):[4] σθ2 = <θt-<θt>2>t→∞ =LLT+σε2∼LLT. Here, σε2 is the variance of the uniform randomly distributed noise in a relative twist angle between two adjacent filament subunits, and we neglect it because it is much smaller than L/LT (SI Appendix). Eq. 4 indicates that at times sufficiently long to sample the equilibrium distribution, indicated by plateau of the variance (Fig. 4E), the variance of the angle fluctuations at equilibrium is inversely proportional to LT, and scales with L linearly (SI Appendix, Fig. S6), which has been demonstrated for actin filaments by experimental observation (20, 39). Analysis of Filament Subunit Angular Fluctuations from cryo-EM Data. From the electron micrographs of actin filaments, which were classified as described previously (54), we defined Δs as the length of an actin subunit and Φ as the rotational Euler angle around the filament centerline. Any rotation angle difference (twist) between two subunits spaced n subunits apart (SI Appendix, Eq. S20) was calculated by the sum of all the observed relative twists between 2 adjacent subunits and it displays a Gaussian distribution with a variance that scales linearly with n (SI Appendix, Eq. S21). Fitting SI Appendix, Eq. S21 to the n dependence of the observed variance (σobs,n+12) yields Δs/LT from the slope of the best linear fit of the data and the noise variance (σε2) from the y intercept. The variance of the angle distribution was extracted from the data using three different analysis procedures. In one approach, the histogram of ΔΦI,obs,n+1 for every value of n was independently fitted to a normal distribution, yielding σ2i,obs,n+1. Then, σ2i,obs,n+1as a function of n was fitted to SI Appendix, Eq. S21. In addition, in maximum likelihood estimation (MLE), each variance with spacing of n was directly calculated from the square of the SD as follows:[5] σobs,n+12=〈Δ∅i,obs,n+1-〈Δθobs,n+1〉2〉. The n dependence of σ2i,obs,n+1 was then fitted to SI Appendix, Eq. S21. In the third method, we applied Bayesian inference using Markov chain Monte Carlo (MCMC; Metropolis–Hastings algorithm) to sample the posterior probability distribution of the true variance (81). The method was coded in language R (www.r-project.org), and the most likely LT and σε2 values were determined from the peaks of their probability distributions. These three methods of analyses were repeated for the rotation Euler angle Φ determined from a refinement volume of 1 subunit instead of 5 with comparable results (Table 1) (54, 55). Supplementary Material Appendix 01 (PDF) Click here for additional data file. Movie S1. Phalloidin decorated actin filament attached to two paramagnetic beads is rotated (left). Cosine of the rotation angle of bead (black dots) and the magnet (red line) are in phase (right). Note that all movies are not played in real time. Movie S2. Alexa-647 actin filament (magenta) conjugated to a paramagnetic bead (cyan) rotates around the paramagnetic bead as the permanent magnet is rotated. Movie S3. Alexa-647 actin filament (magenta) conjugated to a paramagnetic bead (cyan) is subjected to a range of flow induced pulling forces, 0.007pN to 2.9pN. Movie S4. Alexa-647 actin filament (magenta) conjugated to a paramagnetic bead (cyan) is pulled with 0.03 pN of tension and twisted (left). Filament end-to-end distance is determined by tracking of the paramagnetic bead (right). Movie S5. Alexa-647 actin filament (magenta) conjugated to a paramagnetic bead (cyan) is pulled with 0.03 pN of tension and over-twisted until it supercoils. After supercoiling filament is under twisted until the filament twist is neutral. Movie S6. Alexa-647 actin filament (magenta) conjugated to a paramagnetic bead (cyan) is pulled with 0.03 pN of tension and does not fragment. Movie S7. Alexa-647 actin filament (magenta) conjugated to a paramagnetic bead (dim magenta) and a nonmagnetic marker bead (bright magenta). Flow is turned off, a cylindrical magnet is lowered into place, and the focal plane is adjusted to visualize the paramagnetic and marker beads. Movie S8. Alexa-647 actin filament (magenta) conjugated to a paramagnetic bead (cyan) is pulled with 0.25 pN of tension and over-twisted until it fragments. Movie S9. Alexa-647 actin filament (magenta) conjugated to a paramagnetic bead (cyan) is pulled with 0.03 pN of tension and over-twisted until it fragments. Movie S10. Alexa-647 actin filament (magenta) saturated with cofilin (cyan) is conjugated to a paramagnetic bead (cyan) and over-twisted until the filament fragments. Movie S11. Alexa-647 actin filament (magenta) saturated with cofilin (cyan) is conjugated to a paramagnetic bead (cyan) and under-twisted until the filament fragments. Movie S12. Rotational fluctuations of a surface tethered Alexa-647 actin filament conjugated to a paramagnetic bead (dim magenta) and a nonmagnetic marker bead (bright magenta). Filament is pulled straight perpendicular to the coverslip. This research was supported by the NIH through R35-GM136656 (awarded to E.M.D.L.C.), J.P.B. was supported in part by the Department of Defense Army Research Office through a multidisciplinary university research initiative grant W911NF1410403 (awarded to E.M.D.L.C.). S.G. was supported by R35GM136656-S1. C.V.S. was supported by R01 GM 110530. Author contributions J.P.B., N.G.P., and E.M.D.L.C. designed research; J.P.B., N.G.P., and N.S. performed research; J.P.B., N.G.P., S.G., N.S., C.V.S., W.C., and E.M.D.L.C. analyzed data; and J.P.B., N.G.P., W.C., and E.M.D.L.C. wrote the paper. Competing interest The authors declare no competing interest. Data, Materials, and Software Availability All study data are included in the article and/or SI Appendix. Supporting Information This article is a PNAS Direct Submission. D.V. is a guest editor invited by the Editorial Board. ==== Refs 1 P. A. Janmey, C. A. McCulloch, Cell mechanics: Integrating cell responses to mechanical stimuli. Annu. Rev. Biomed Eng. 9 , 1–34 (2007).17461730 2 I. Schoen, B. L. Pruitt, V. Vogel, The Yin-Yang of rigidity sensing: How forces and mechanical properties regulate the cellular response to materials. Annu. Rev. Mater Res. 43 , 589–618 (2013). 3 T. D. Pollard, J. A. Cooper, Actin, a central player in cell shape and movement. Science 326 , 1208–1212 (2009).19965462 4 J. Howard, Mechanics of Motor Proteins and the Cytoskeleton (Sinauer Associates, Publishers, Sunderland, Mass., 2001), p. xvi, 367 p. 5 F. Gittes, B. Mickey, J. Nettleton, J. Howard, Flexural rigidity of microtubules and actin filaments measured from thermal fluctuations in shape. J. Cell Biol. 120 , 923–934 (1993).8432732 6 F. C. Mackintosh, J. Kas, P. A. Janmey, Elasticity of semiflexible biopolymer networks. Phys. Rev. Lett. 75 , 4425–4428 (1995).10059905 7 J. D. Winkelman, C. A. Anderson, C. Suarez, D. R. Kovar, M. L. Gardel, Evolutionarily diverse LIM domain-containing proteins bind stressed actin filaments through a conserved mechanism. Proc. Natl. Acad. Sci. U.S.A. 117 , 25532–25542 (2020).32989126 8 X. Y. Sun , Mechanosensing through direct binding of tensed F-actin by LIM domains. Dev. Cell 55 , 468–482 (2020).33058779 9 M. Akamatsu , Principles of self-organization and load adaptation by the actin cytoskeleton during clathrin-mediated endocytosis. Elife 9 , e49840 (2020).31951196 10 E. S. Welf , Actin-membrane release initiates cell protrusions. Dev. Cell 55 , 723–736 (2020).33308479 11 Y. H. Tee , Cellular chirality arising from the self-organization of the actin cytoskeleton. Nat. Cell Biol. 17 , 445–457 (2015).25799062 12 N. Leijnse , Filopodia rotate and coil by actively generating twist in their actin shaft. Nat. Commun. 13 , 1636 (2022).35347113 13 N. A. Medeiros, D. T. Burnette, P. Forscher, Myosin II functions in actin-bundle turnover in neuronal growth cones. Nat. Cell Biol. 8 , 215–226 (2006).16501565 14 S. H. Parekh, O. Chaudhuri, J. A. Theriot, D. A. Fletcher, Loading history determines the velocity of actin-network growth. Nat. Cell Biol. 7 , 1219–1223 (2005).16299496 15 A. C. Schramm, G. M. Hocky, G. A. Voth, J. L. Martiel, E. M. De La Cruz, Plastic deformation and fragmentation of strained actin filaments. Biophys J. 117 , 453–463 (2019).31301801 16 E. M. De La Cruz, J. Roland, B. R. McCullough, L. Blanchoin, J. L. Martiel, Origin of twist-bend coupling in actin filaments. Biophys. J. 99 , 1852–1860 (2010).20858430 17 N. G. Pandit , Force and phosphate release from Arp2/3 complex promote dissociation of actin filament branches. Proc. Natl. Acad. Sci. U.S.A. 117 , 13519–13528 (2020).32461373 18 V. I. Risca , Actin filament curvature biases branching direction. Proc. Natl. Acad. Sci. U.S.A. 109 , 2913–2918 (2012).22308368 19 A. C. Schramm , Actin filament strain promotes severing and cofilin dissociation. Biophys. J. 112 , 2624–2633 (2017).28636918 20 Y. Tsuda, H. Yasutake, A. Ishijima, T. Yanagida, Torsional rigidity of single actin filaments and actin-actin bond breaking force under torsion measured directly by in vitro micromanipulation. Proc. Natl. Acad. Sci. U.S.A. 93 , 12937–12942 (1996).8917522 21 M. P. Murrell, M. L. Gardel, F-actin buckling coordinates contractility and severing in a biomimetic actomyosin cortex. Proc. Natl. Acad. Sci. U.S.A. 109 , 20820–20825 (2012).23213249 22 H. Wioland, A. Jegou, G. Romet-Lemonne, Torsional stress generated by ADF/cofilin on cross-linked actin filaments boosts their severing. Proc. Natl. Acad. Sci. U.S.A. 116 , 2595–2602 (2019).30692249 23 Y. Arai , Tying a molecular knot with optical tweezers. Nature 399 , 446–448 (1999).10365955 24 P. Bieling , Force feedback controls motor activity and mechanical properties of self-assembling branched actin networks. Cell 164 , 115–127 (2016).26771487 25 J. Mueller , Load adaptation of lamellipodial actin networks. Cell 171 , 188–200.e16 (2017).28867286 26 A. Jegou, G. Romet-Lemonne, The many implications of actin filament helicity. Semin. Cell Dev. Biol. 102 , 65–72 (2020).31862222 27 L. D. Landau, E. M. Lifshitz, Theory of Elasticity (Pergamon Press, Oxford, New York, 1986). 28 E. M. De La Cruz, J. L. Martiel, L. Blanchoin, Mechanical heterogeneity favors fragmentation of strained actin filaments. Biophys. J. 108 , 2270–2281 (2015).25954884 29 A. Mogilner, G. Oster, Force generation by actin polymerization II: The elastic ratchet and tethered filaments. Biophys. J. 84 , 1591–1605 (2003).12609863 30 A. Davison , Formin is associated with left-right asymmetry in the pond snail and the frog. Curr. Biol. 26 , 654–660 (2016).26923788 31 S. R. Naganathan, S. Furthauer, M. Nishikawa, F. Julicher, S. W. Grill, Active torque generation by the actomyosin cell cortex drives left-right symmetry breaking. Elife 3 , e04165 (2014).25517077 32 W. A. Elam , Phosphomimetic S3D cofilin binds but only weakly severs actin filaments. J. Biol. Chem. 292 , 19565–19579 (2017).28939776 33 H. R. Kang , Identification of cation-binding sites on actin that drive polymerization and modulate bending stiffness. Proc. Natl. Acad. Sci. U.S.A. 109 , 16923–16927 (2012).23027950 34 B. R. McCullough, L. Blanchoin, J. L. Martiel, E. M. De la Cruz, Cofilin increases the bending flexibility of actin filaments: Implications for severing and cell mechanics. J. Mol. Biol. 381 , 550–558 (2008).18617188 35 B. R. McCullough , Cofilin-linked changes in actin filament flexibility promote severing. Biophys. J. 101 , 151–159 (2011).21723825 36 H. Kojima, A. Ishijima, T. Yanagida, Direct measurement of stiffness of single actin-filaments with and without tropomyosin by in-vitro nanomanipulation. Proc. Natl. Acad. Sci. U.S.A. 91 , 12962–12966 (1994).7809155 37 H. Isambert , Flexibility of actin-filaments derived from thermal fluctuations–effect of bound nucleotide, phalloidin, and muscle regulatory proteins. J. Biol. Chem. 270 , 11437–11444 (1995).7744781 38 H. Kang , Site-specific cation release drives actin filament severing by vertebrate cofilin. Proc. Natl. Acad. Sci. U.S.A. 111 , 17821–17826 (2014).25468977 39 R. Yasuda, H. Miyata, K. Kinosita Jr., Direct measurement of the torsional rigidity of single actin filaments. J. Mol. Biol. 263 , 227–236 (1996).8913303 40 E. Prochniewicz, N. Janson, D. D. Thomas, E. M. De la Cruz, Cofilin increases the torsional flexibility and dynamics of actin filaments. J. Mol. Biol. 353 , 990–1000 (2005).16213521 41 J. N. Forkey, M. E. Quinlan, Y. E. Goldman, Measurement of single macromolecule orientation by total internal reflection fluorescence polarization microscopy. Biophys. J. 89 , 1261–1271 (2005).15894632 42 J. Fan , Molecular origins of cofilin-linked changes in actin filament mechanics. J. Mol. Biol. 425 , 1225–1240 (2013).23352932 43 T. Fujii, A. H. Iwane, T. Yanagida, K. Namba, Direct visualization of secondary structures of F-actin by electron cryomicroscopy. Nature 467 , 724–728 (2010).20844487 44 E. H. Egelman, D. J. Derosier, Image-analysis shows that variations in actin crossover spacings are random, not compensatory. Biophys. J. 63 , 1299–1305 (1992).1477281 45 J. P. Bibeau, S. Gray, E. M. De La Cruz, Clusters of a few bound cofilins sever actin filaments. J. Mol. Biol. 433 , 166833 (2021).33524412 46 N. Courtemanche, J. Y. Lee, T. D. Pollard, E. C. Greene, Tension modulates actin filament polymerization mediated by formin and profilin. Proc. Natl. Acad. Sci. U.S.A. 110 , 9752–9757 (2013).23716666 47 J. S. Palmer, M. C. Boyce, Constitutive modeling of the stress-strain behavior of F-actin filament networks. Acta Biomater. 4 , 597–612 (2008).18325860 48 J. S. Graham , Multi-platform compatible software for analysis of polymer bending mechanics. PLoS One 9 , e94766 (2014).24740323 49 H. Isambert, A. C. Maggs, Dynamics and rheology of actin solutions. Macromolecules 29 , 1036–1040 (1996). 50 M. J. Greenberg, C. L. Wang, W. Lehman, J. R. Moore, Modulation of actin mechanics by caldesmon and tropomyosin. Cell Motil. Cytoskeleton 65 , 156–164 (2008).18000881 51 M. T. van Loenhout, M. V. de Grunt, C. Dekker, Dynamics of DNA supercoils. Science 338 , 94–97 (2012).22983709 52 J. F. Marko, Torque and dynamics of linking number relaxation in stretched supercoiled DNA. Phys. Rev. E Stat. Nonlin. Soft Matter Phys. 76 , 021926 (2007).17930084 53 T. Sanchez, I. M. Kulic, Z. Dogic, Circularization, photomechanical switching, and a supercoiling transition of actin filaments. Phys. Rev. Lett. 104 , 098103 (2010).20367015 54 A. Huehn , The actin filament twist changes abruptly at boundaries between bare and cofilin-decorated segments. J. Biol. Chem. 293 , 5377–5383 (2018).29463680 55 A. R. Huehn , Structures of cofilin-induced structural changes reveal local and asymmetric perturbations of actin filaments. Proc. Natl. Acad. Sci. U.S.A. 117 , 1478–1484 (2020).31900364 56 J. D. Moroz, P. Nelson, Entropic elasticity of twist-storing polymers. Macromolecules 31 , 6333–6347 (1998). 57 J. Lipfert, M. Wiggin, J. W. Kerssemakers, F. Pedaci, N. H. Dekker, Freely orbiting magnetic tweezers to directly monitor changes in the twist of nucleic acids. Nat. Commun. 2 , 439 (2011).21863006 58 A. J. Hunt, J. Howard, Kinesin swivels to permit microtubule movement in any direction. Proc. Natl. Acad. Sci. U.S.A. 90 , 11653–11657 (1993).8265603 59 E. M. De La Cruz, M. L. Gardel, Actin mechanics and fragmentation. J. Biol. Chem. 290 , 17137–17144 (2015).25957404 60 G. M. Hocky, C. V. Sindelar, W. Cao, G. A. Voth, E. M. De La Cruz, Structural basis of fast- and slow-severing actin–cofilactin boundaries. J. Biol. Chem. 296 , 100337 (2021).33508320 61 C. Suarez , Cofilin tunes the nucleotide state of actin filaments and severs at bare and decorated segment boundaries (vol 21, pg 862, 2011). Curr. Biol. 21 , 862–868 (2011).21530260 62 E. M. De La Cruz, How cofilin severs an actin filament. Biophys. Rev. 1 , 51–59 (2009).20700473 63 H. Wioland , ADF/cofilin accelerates actin dynamics by severing filaments and promoting their depolymerization at both ends. Curr. Biol. 27 , 1956–1967 (2017).28625781 64 K. Hayakawa, S. Sakakibara, M. Sokabe, H. Tatsumi, Single-molecule imaging and kinetic analysis of cooperative cofilin-actin filament interactions. Proc. Natl. Acad. Sci. U.S.A. 111 , 9810–9815 (2014).24958883 65 E. M. De La Cruz, Cofilin binding to muscle and non-muscle actin filaments: Isoform-dependent cooperative interactions. J. Mol. Biol. 346 , 557–564 (2005).15670604 66 E. Andrianantoandro, T. D. Pollard, Mechanism of actin filament turnover by severing and nucleation at different concentrations of ADF/cofilin. Mol. Cell 24 , 13–23 (2006).17018289 67 D. Pavlov, A. Muhlrad, J. Cooper, M. Wear, E. Reisler, Actin filament severing by cofilin. J. Mol. Biol. 365 , 1350–1358 (2007).17134718 68 W. A. Elam, H. Kang, E. M. De La Cruz, Biophysics of actin filament severing by cofilin. FEBS Lett. 587 , 1215–1219 (2013).23395798 69 E. M. De La Cruz, E. M. Ostap, Relating biochemistry and function in the myosin superfamily. Curr. Opin. Cell Biol. 16 , 61–67 (2004).15037306 70 X. F. Zhang , Regulation of axon growth by myosin II-dependent mechanocatalysis of cofilin activity. J. Cell Biol. 218 , 2329–2349 (2019).31123185 71 M. Malla, T. D. Pollard, Q. Chen, Counting actin in contractile rings reveals novel contributions of cofilin and type II myosins to fission yeast cytokinesis. Mol. Biol. Cell 33 , mbcE21080376 (2021). 72 G. Lebreton , Molecular to organismal chirality is induced by the conserved myosin 1D. Science 362 , 949–952 (2018).30467170 73 J. F. Beausang, H. W. Schroeder III, P. C. Nelson, Y. E. Goldman, Twirling of actin by myosins II and V observed via polarized TIRF in a modified gliding assay. Biophys. J. 95 , 5820–5831 (2008).18931255 74 S. Pyrpassopoulos, E. A. Feeser, J. N. Mazerik, M. J. Tyska, E. M. Ostap, Membrane-bound myo1c powers asymmetric motility of actin filaments. Curr. Biol. 22 , 1688–1692 (2012).22863317 75 E. M. De La Cruz, T. D. Pollard, Nucleotide-free actin: Stabilization by sucrose and nucleotide binding kinetics. Biochemistry 34 , 5452–5461 (1995).7727403 76 Y. Gidi, S. Bayram, C. J. Ablenas, A. S. Blum, G. Cosa, Efficient one-step PEG-silane passivation of glass surfaces for single-molecule fluorescence studies. ACS Appl Mater Interfaces 10 , 39505–39511 (2018).30346695 77 E. M. Johnson-Chavarria, U. Agrawal, M. Tanyeri, T. E. Kuhlman, C. M. Schroeder, Automated single cell microbioreactor for monitoring intracellular dynamics and cell growth in free solution. Lab. Chip 14 , 2688–2697 (2014).24836754 78 J. Y. Tinevez , TrackMate: An open and extensible platform for single-particle tracking. Methods 115 , 80–90 (2017).27713081 79 Y. Seol, J. Li, P. C. Nelson, T. T. Perkins, M. D. Betterton, Elasticity of short DNA molecules: Theory and experiment for contour lengths of 0.6-7 microm. Biophys. J. 93 , 4360–4373 (2007).17766363 80 J. F. Marko, E. D. Siggia, Stretching DNA. Macromolecules 28 , 8759–8770 (1995). 81 M. K. Cowles, Applied Bayesian statistics: With R and OpenBUGS Examples (Springer, New York, 2013), p. xiv, 232 pages.
PMC009xxxxxx/PMC9942854.txt
==== Front Proc Natl Acad Sci U S A Proc Natl Acad Sci U S A PNAS Proceedings of the National Academy of Sciences of the United States of America 0027-8424 1091-6490 National Academy of Sciences 36656855 202120869 10.1073/pnas.2120869120 research-articleResearch ArticleecoEcology414 Biological Sciences Ecology Three decades of increasing fish biodiversity across the northeast Atlantic and the Arctic Ocean Gordó-Vilaseca Cesc [email protected] a 1 https://orcid.org/0000-0001-5992-218X Stephenson Fabrice b c Coll Marta d e https://orcid.org/0000-0001-6235-5868 Lavin Charles a https://orcid.org/0000-0002-2068-1080 Costello Mark John a https://orcid.org/0000-0003-2362-0328 aFaculty of Biosciences and Aquaculture, Nord University, Bodø 1049, Norway bNational Institute of Water and Atmosphere, Hamilton 3251, New Zealand cSchool of Science, University of Waikato, Hamilton 3216, New Zealand dDepartment of Marine Renewal Resources, Institute of Marine Science - Consejo Superior de Investigaciones Científicas, Barcelona 08003, Spain eEcopath International Initiative, Barcelona 08193, Spain 1To whom correspondence may be addressed. Email: [email protected]. Edited by Andrea Belgrano, Sveriges lantbruksuniversitet, Lysekil, Sweden; received November 16, 2021; accepted December 17, 2022 by Editorial Board Member Pablo A. Marquet 19 1 2023 24 1 2023 19 7 2023 120 4 e212086912016 11 2021 17 12 2022 Copyright © 2023 the Author(s). Published by PNAS. 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This article is distributed under Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND). Significance Global modeling studies suggest increased species arrivals from lower latitudes and local extirpations at high latitudes due to global warming. Our analysis of 20,670 standardized scientific trawls with 193 fish species from the northeast Atlantic and Arctic Oceans found an increase in species richness in the region correlated with an increase in sea bottom temperature. Some Arctic species declined in probability of occurrence over time, but some increased. This, together with the increase in southern-latitude species led to an enrichment of the Arctic and sub-Arctic marine fauna attributed to climate change from 1994 to 2020. Observed range shifts of numerous species support predictions of climate change models that species will shift their distribution northward into the Arctic and sub-Arctic seas due to ocean warming. However, how this is affecting overall species richness is unclear. Here we analyze 20,670 scientific research trawls from the North Sea to the Arctic Ocean collected from 1994 to 2020, including 193 fish species. We found that demersal fish species richness at the local scale has doubled in some Arctic regions, including the Barents Sea, and increased at a lower rate at adjacent regions in the last three decades, followed by an increase in species richness and turnover at a regional scale. These changes in biodiversity correlated with an increase in sea bottom temperature. Within the study area, Arctic species’ probability of occurrence generally declined over time. However, the increase in species from southern latitudes, together with an increase in some Arctic species, ultimately led to an enrichment of the Arctic and sub-Arctic marine fauna due to increasing water temperature consistent with climate change. climate warming species richness biodiversity Arctic Ocean demersal fish MEC | Agencia Estatal de Investigación (AEI) 501100011033 PID2020-118097RB-I00 Marta Coll ==== Body pmcClimate warming constitutes one of the main faces of climate change and is having a direct impact on species, communities, and ecosystems (1, 2). In the oceans, the average increase in temperature in the last 140 y has been of 1 °C (3, 4). Marine ectotherms occupy most of their potential latitudinal range with regard to thermal tolerance and therefore move to higher latitudes following the displacement of their thermal niche (5, 6). However, these changes can occur at different rates across species, depending on traits such as their dispersal potential, thermal niche and capacity to exploit new resources, which may lead to changing community composition (1, 7). Understanding how these changes occur is crucial for effective conservation and management strategies. Yet, to date, consistent empirical evidence of a generalization of these shifts in the Arctic fish community is lacking. Arctic and sub-Arctic ecosystems are among the most rapidly warming regions in the world, with some areas warming four times faster than, and seas warming twice, the global average (4, 8, 9), and their species composition could be changing accordingly (10). Until now, a doubling in species richness has been reported in some areas of the North Sea, though not in others, and increases of smaller magnitude have also been reported around North America (11–13). However, fish community analyses are mostly restricted to nonpolar latitudes, and they rarely exceed the 62°N of the north Bering’s Sea. Studies focusing on areas above 62°N only exist in the Barents Sea, where some boreal species arrived recently (10, 14, 15) (SI Appendix, Table S1). Of these, one study examined distributional shifts in a fish community of 49 species from 2004 to 2017, and it reported an increase of less than 50% in species richness (15). This represents the only work reporting empirical evidence of a regional increase in marine fish species richness with climate change in Arctic latitudes. Thus, although an increase in species richness is predicted into the Arctic Ocean, and several model projections exist in the literature, including species extirpations (16–18), the empirical evidence is limited temporally and taxonomically, and lacks a long-term correlation with climate warming. In the Norwegian and Barents Seas, recent warming and increased Atlantic water inflow events have been recorded, with a decline in sea-ice cover in the northern Barents Sea (19–21). As a consequence, profound effects on the geographical distribution and productivity of commercial fishing stocks are expected (22–24). In fact, species turnover was projected to increase in the area in the next decades, resulting from some local species extirpations and the arrival of warmer-water species (17, 25, 26). Here, we report three decades of field data to test these predictions. Recent changes in species distributions in the Norwegian Sea include consistent northward expansions of the Atlantic mackerel (Scomber scombrus) and the European Hake (Merluccius merluccius), and the predicted northward expansion of bluefin-tuna (Thunnus thynnus), among other species (27–30). Even stronger are the changes reported in the Arctic region of the Barents Sea, where at least 11 boreal (sub-Arctic) species have been recently recorded (14, 31). Similarly, studies in the Bering Sea found localized increases in biodiversity and suggested that areas with increased species richness and climatic stability were climate refugia (32). Several recent expansions of benthic species distributional ranges, such as that of the red king crab (Paralithodes camtschaticus) or the snow crab (Chionoecetes opilio), are rapidly altering benthic communities, and are also affecting demersal fish trophodynamics (33, 34). However, how these examples may materialize into a wider trend in the demersal fish fauna of the Arctic region due to climate change has not been investigated. Moreover, accounting for both species gains and losses needs to cover a sufficiently large region to avoid boundary effects. Here, we test the commonness of these shifts into the Arctic across the demersal fish community of a wider latitudinal range (from 56 to 82°N), for more species and a longer time period (27 y) than previous studies. We developed our analysis in Norwegian-Barents Seas, and the adjacent areas in the North Sea and around Svalbard with the aim of understanding changes in biodiversity across the whole region. To do so we explored changes in three scales of biodiversity: alpha diversity (average local species richness), beta diversity and its component turnover (excluding species richness effect), and gamma diversity (total regional species richness) (35–37). Each of these measures provides information on biological diversity at a particular scale, and when combined provide a complete measure of biological diversity across the whole study area. To analyze the change in local diversity (alpha diversity) at each sampling site over time, we used generalized additive models (GAMs) (a semi-parametric statistical modeling technique that allows for nonlinear effect of variables, as is the case of time). We then explored the contribution of environmental variables to local species richness and made annual projections of changes in local species richness using the nonparametric algorithm of boosted regression trees (BRTs) (selected because BRT models can easily accommodate the inclusion of a large number of environmental variables, including those with relatively high level of collinearity) (38, 39). To analyze the change in beta diversity, which accounts for the variability between sites, we calculated the pairwise mean diversity using the Bray–Curtis dissimilarity metric. However, this difference can arise from simply having different species richness (i.e., number of species at each site) or from having different species composition (i.e., different species at each site), and to discern among these two sources of variability between sites, we also calculated Nestedness and Turnover, which are the two components of beta diversity (37). Finally, to calculate gamma diversity, or the overall species richness in the study area, we used rarefaction species accumulative curves, which consider the relationship between sampling effort and species found during this sampling (40). Results Alpha Diversity. A total of 193 demersal fish species were recorded between 1994 to 2020 across the whole study area (Fig. 1 and SI Appendix, Table S2). To study the change in species richness per trawl considering differences in sampling effort across time, GAM models with year, latitude, and sampling effort were fitted at each study area to species count data. In the main study area (Norwegian-Barents Seas), the model predicted a 66% increase in the average number of species per trawl (alpha diversity), from 8.0 species/trawl in 1994 (95% confidence interval CI 7.8, 8.1), to 13.4 species/trawl in 2020 (95% CI 13.1, 13.6) (P < 0.05, deviance explained (DV) explained = 14%) (Fig. 2 and Table 1). The increase in species richness was correlated with changes in sea bottom temperature (Pearson r = 0.59, P < 0.05, SI Appendix, Supplementary Results). Increasing species richness was also found for the adjacent areas in the North Sea and around Svalbard, though no significant correlation was detected with sea bottom temperature (SBT) at those regions (Table 1 and SI Appendix, Fig. S1). Fig. 1. Study area and histograms of the temporal distribution of the number of trawls in the present study. A: Norwegian & Barents Sea; B: Svalbard; C: North Sea. Dashed lines are latitude and longitude, and thick black dashed line is the polar circle (66°N). Fig. 2. Average number of species per trawl (alpha diversity) across the Norwegian-Barents Sea. Black line represents mean species richness per trawl with 95% CI in blue. Grey smoothed line and light grey 95% CI represents the marginal effect of Year for constant sampling effort from a GAM model using Year and swept area as an offset (Table 1). Red line indicates changes in mean sea bottom temperature across the main study area (correlation with mean alpha diversity r = 0.59). Table 1. Average species richness per trawl (alpha diversity) in the first and last year of sampling predicted from a GAM with Year and Latitude as smooth predictors, and swept area as a logarithmic predictor (%DV = % Deviance explained) Region First year of data Richness first year [lower, upper 95% CI] Last year of data Richness last year [lower, upper 95% CI] % increase [lower, upper 95% CI] No. of trawls % DV Norwegian & Barents Sea 1994 8.0 [7.8, 8.1] 2020 13.4 [13.1, 13.6] 66 [61, 75] 16,283 16 North Sea 1998 11.2 [10.6, 11.8] 2020 16.4 [15.6, 17.2] 46 [32, 6] 2,338 42 Svalbard 1996 6.9 [6.3, 7.5] 2020 12.8 [12.2, 13.5] 87 [63, 115] 2,065 12 The more relaxed assumptions of the nonparametric BRT allowed the inclusion of the whole study area in one model (which facilitates spatiotemporal projections), and the inclusion of 17 explanatory variables. The resulting model presented a good fit to the data (correlation with independent data r = 0.63, DV explained = 38%, P < 0.05) and predicted local increases in species richness up to 125% in some regions from 1994 to 2019. Among the explanatory variables included in the BRT model, “Depth” was the variable that explained the most deviance in the data, followed by “year” and “bottom temperature” (SI Appendix, Fig. S2). The model projected an increase in species richness at higher latitudes, especially high in the Arctic region of the Barents Sea, where species richness more than doubled in some regions, during the study period (Fig. 3 and SI Appendix, Fig. S3). Fig. 3. Difference between mean species richness from 1994 to 1996 and 2017 to 2019 expressed as percentage of change. The orange polygon is the study area boundary. Dashed lines are latitude and longitude. Gamma Diversity. The overall species richness in the main study area (gamma diversity) increased 45% during the study period, from 65 species in 1994 to 1994 species in 2020, following a similar temporal pattern to alpha diversity (Fig. 4 and Table 2 and SI Appendix, Fig. S4). Very similar increases were reported in adjacent areas (47% in the North Sea and 45% in Svalbard, Table 2 and SI Appendix, Fig. S4). GAM models for total species richness at each region, with year and both annual mean trawl-swept area and total mean swept area only selected year as relevant explanatory variables. All additional estimators, including Chao index, incidence based index, and first- and second-order jackknife estimators, calculated across the study area, detected an increase in gamma diversity across time, of similar magnitude (SI Appendix, Fig. S5). Fig. 4. Left: Annual species accumulation (rarefaction) curves in the main region Norwegian-Barents Sea. Dashed line marks the minimum common number of trawls. Right: Annual mean accumulated species richness within the minimum common number of trawls. Table 2. Total species richness predicted from GAM with year as the only explanatory variable at each region (gamma diversity) in the first and last years of data, at minimum common annual number of trawls (%DV = %Deviance explained) Region First year of data Richness first year [95% CI] Last year of data Richness last year [95% CI] % increase [95% CI] Annual trawls % DV Norwegian & Barents Sea 1994 66 [61, 71] 2020 96 [91, 101] 45 [29, 65] 361 83 Svalbard 1996 33 [29, 37] 2020 48 [44, 52] 45 [20, 77] 30 69 North Sea 1998 40 [37, 42] 2020 58 [51, 57] 47 [32, 65] 37 81 Beta Diversity. Pairwise mean total beta diversity in the main study area did not significantly change with time (Fig. 5). However, the turnover contribution to beta diversity, which is not affected by species richness, increased until 2008, declined until 2014, and increased afterward, with an overall increase of 16% (95% CI 6, 27) (Fig. 5 and SI Appendix, Fig. S6 and Table 3). Similarly, total beta diversity in the adjacent region of Svalbard increased until 2005 and declined afterwards. However, the turnover increased linearly across time in this region. In the North Sea, both the total beta diversity and turnover declined with time. Fig. 5. Annual mean beta diversity and turnover (richness independent), red line) in the main region across time, calculated from mean pairwise Jaccard dissimilarity index. Table 3. Mean total beta diversity and turnover predicted from GAM at first and last year of data at each region, at minimum common annual number of trawls (%DV = %Deviance explained) Region P First year of data Estimation first year [95% CI] Last year of data Estimation last year [95% CI] % increase [95% CI] Annual trawls % DV Norwegian & Barents Sea B 1994 0.35 [0.34, 0.36] 2020 0.36 [0.35, 0.37] 4 [−3, 9] 361 54 T 0.63 [0.60, 0.66] 0.73 [0.70, 0.76] 16 [6, 27] 75 Svalbard B 1996 0.32 [0.31, 0.33] 0.34 [0.32, 0.35] 5 [−4, 15] 30 69 T 0.61 [0.57, 0.64] 0.75 [0.71, 0.79] 24 [15, 38] 50 North Sea B 1998 0.34 [0.32, 0.36] 0.32 [0.29, 0.34] −7 [−19, 6] 37 55 T 0.68 [0.63, 0.74] 0.62 [0.55, 0.69] −9 [−25 ,9] 63 The GAM models for temporal change in beta diversity and turnover did not select trawl-swept area as a relevant explanatory for the main study area or the adjacent regions. Thus, when the increasing species richness is accounted for, species turnover, as well as alpha and gamma diversity, increased over the years. Species’ Trends. From the 193 species in our study, 99 species showed a significant temporal trend (increasing or decreasing) in their probability of occurrence with time, in at least one of the studied areas (Fig. 6). Of these, 71 species showed only positive trends, and 23 species showed only negative trends in at least one studied area, while five species showed positive and negative trends depending on the study area (Fig. 6). Thus, while no trend was detected for about half of the species, 76% of species showing a significant consistent trend increased. The number of species increasing was consistently higher than the number of species decreasing over time, across all regions. Species declining mostly inhabited high mean latitudes, and linear regression identified a negative effect of mean latitude on the temporal change in species probability of occurrence in the main study area, and around Svalbard (Fig. 6). Among the 67 species only found in the main region and/or Svalbard (Arctic region), 18 species showed a significant change in their probability of occurrence with time, six increased and 12 declined. Fig. 6. Change in the probability of species’ occurrence with time. Length of the bars correspond to the Slope of the effect of Year on species probability of occurrence, using a GAM model including swept area, latitude, and depth as smooth predictors, and Year as a linear predictor. Species are ordered by their mean latitude of occurrence during the study period. Linear regression on the effect of Year with mean latitude was conducted and plotted, and asterisks indicate significance (P < 0.05). Discussion Following global trends in warming, sea bottom temperature data suggests an increase of 0.3 °C per decade in our study area from 1993 to 2019, and this increase was several times higher in some regions of the Arctic Barents Sea (9). Here we show that this has been accompanied by a significant increase of the demersal marine fish biodiversity during the last three decades, in alpha, gamma, and turnover beta diversity. Considering previous studies focusing on smaller areas and fewer species (15, 32), this study indicates an ecologically significant increase in species richness in the Atlantic side of the Arctic in concert with climate warming. Consequently, species are expanding their range poleward, directly increasing the local species richness in the Barents Sea, in line with other results in near-Arctic regions (11), and with predicted poleward shifts in the North and Arctic Seas (16). This increase in biodiversity was concentrated in two different periods, and interrupted by a decline period in alpha, beta and gamma diversity between 2007 to 2014, as suggested by Ellingsen et al. (41). Even though there have been studies showing consistent changes in biodiversity patterns and beta diversity at a global scale, whether a systematic loss of local (alpha) diversity has or will take place remains uncertain (42–44). Several species are changing their distributional range, mostly expanding poleward, which leads to redistributions of global biodiversity and local increases in biodiversity (11, 42). In an analysis of 50,000 marine species, Chaudhary et al. (5) found that thousands of species had shifted from equatorial to mid-latitudes, and we confirm that this shift has continued into higher northern latitudes, leading to two times more demersal fish species in the study area. Increases in alpha diversity were already reported in some regions of the Arctic Ocean, albeit smaller areas and/or shorter time periods (15, 32). With this study, we show that the increase in species richness is not localized, but widespread across the region and concomitant with the increase in bottom temperature. We identified sea bottom temperature as the third most relevant variable in our models, after depth and year. The fact that year remains an important variable, and that the BRT model explains 36% of the overall deviance in species richness, suggests that other processes and variables are needed to fully explain the changes in species richness across the study area during the study period. For example, our analysis is based on presence-absence data, and for this reason alpha diversity is statistically dependent on abundance, because increased abundance in the sampling tends to result in more species observed as a result of increased sampling effort. Although we do take into account the sampling effort with every trawl swept area, this remains a potential source of uncertainty that needs to be considered. Most studies on species richness consider only gamma (total) or alpha (average) diversity. When beta diversity is considered it is most commonly measured as Jaccard’s coefficient, which is not independent of species richness (45). Here, we find that while both alpha and gamma diversity are significantly increasing, so is species turnover, but not total beta diversity (Fig. 5). Thus, the spatial heterogeneity of biodiversity is increasing as well as biodiversity overall. However, the beta-diversity decomposition also have been affected by varying sampling efficiency across sites and years, which we did not test in this study. Over the study area, we found that 71 species showing a significant increase in probability of occurrence with time. When species’ mean latitude was used as an indicator of the Arctic affinity of each species, it proved statistically significant to model the effect of year in species probability of occurrence in the Norwegian and Barents Seas, and also around Svalbard. Although this a clear indicator of a decline of several Arctic species, not all species occupying high mean latitudes showed a decline in probability of occurrence with time. Some high-latitude species increased substantially, perhaps because they benefited from changes in food-web interactions. This also suggests a partial coexistence between boreal and Arctic species that, together with the increase of lower-latitude species, leads to a consistent increase in biodiversity, in line with previous results on fishes and crustaceans (46, 47), but not with a previous hypothesis of a synchronous species extirpation in the western Barents and Norwegian Seas (17, 25). If this trend is maintained in the following decades, we may observe enriched communities in the Arctic, which may be sustained by the projected increase in net primary productivity of up to 50% (26, 48). Our results have direct implications for the management of marine resources in the region, where proactive mitigation and adaptation actions to changes in species occurrence can help prevent future negative impacts in the socioeconomic activities of the area, or inform about new opportunities. However, the wider effects of changing fish biodiversity on marine food webs and socioecological system remain to be investigated. Our results have direct implications for the management of marine resources in the region, where proactive mitigation and adaptation actions to changes in species occurrence can help prevent future negative impacts in the socioeconomic activities of the area, and inform about new opportunities. Closer monitoring of individual species decreasing in abundance is already necessary to detect these changes as soon as possible. However, the wider effects of changing fish biodiversity on marine food webs and socioecological system remain to be investigated. Materials and Methods Study Area. The data analyzed here were selected from a research trawl surveys database published by the Institute of Marine Research between 1980 and 2020 in an attempt to gather all the scientific trawl surveys conducted by the institute in a single and open access resource (49). These trawls were mostly restricted to the continental shelf and slope from the north of the North Sea into the Arctic Ocean, mostly from 56°N to 81°N Latitude, and from 2°W to 50°E Longitude (49). The whole area had a marked temperature gradient, with average bottom water temperatures over 8 °C in the North Sea to −1 °C in the northern region of the Barents Sea. Because of variation in the temporal and spatial distribution of the data, we considered one main study area, with the longest time period (Norwegian & Barents Sea) and two adjacent areas (Svalbard and North Sea), which contain a shorter temporal period (Fig. 1). The database initially contained 60,355 research trawls from several surveys between 1980 and 2020. We excluded data associated with broken gear, had incomplete metadata (data without reporting depth, or coordinates, or type of gear), or were questionable (e.g., shrimp trawl opening of several km). We restricted the analysis to shrimp trawling data using Campelen 1800 shrimp trawl with 20 mm mesh size, a maximum of 5 km trawling distance and 60 m of trawl opening, from 30 m to 700 m depth. We only included fish species (classes Actinopterygii, Elasmobranchii, Holocephali, Myxini and Petromyzonti). Invertebrates were not included in the study. After data standardization and selection, 20,670 surveys from 1994 to 2020 remained. Depending on the data spatial coverage over time, and based on the biogeographical realms that have been described in the area (50), we divided the study area into a main area containing most of the data and the longest temporal coverage, the Norwegian-Barents Sea, and two adjacent areas with more limited data: the area around Svalbard and the North Sea (Fig. 1). Although the latitudinal and longitudinal distribution of trawls across time was relatively homogeneous at each region, the swept area of each trawl presented a weak negative trend with time for the Norwegian-Barents Sea and Svalbard regions, which was accounted for within the statistical modelling (SI Appendix, Figs. S7 and S8). Environmental Variables. Spatially explicit environmental co-variates were collated from three sources: Copernicus Marine Service database, Bio-Oracle and MARSPEC database (51–53). Some of the environmental variables were available as annual estimates [e.g., sea surface temperature (SST)] whereas others were available as long-term averages (e.g., bottom nitrate concentration) (SI Appendix, Table S3). Variables from the Copernicus Marine Service were available for each year from 1993 to 2019. These include estimates of SST, SBT, sea surface salinity, ice concentration and sea surface currents (northward and eastward components), and were obtained from the “GLOBAL_REANALYSIS_PHY_001_031” dataset (53) (SI Appendix, Table S3). Because these layers were only available from 1993 to 2019, we limited the analyses to this period. Long-term averages (2000 to 2014) of nitrate, iron, dissolved oxygen, surface and bottom productivity, and temperature range were obtained from Bio-Oracle. Bathymetry was obtained from the MARSPEC database (51, 52). Variables from Bio-Oracle and MARSPEC were downloaded using the “sdmpredictors” package from R software (54). A distance to coast layer was created using the “raster” package, also in R (55). All the environmental layer resolutions were matched by downscaling to the lowest original resolution (0.083° Longitude × 0.083° Latitude). Variables from year, latitude, longitude, and annual number of trawls were also manually created. The shapefiles for the ocean and land were obtained from the Marine Regions database (56) and maps were created using QGIS software (57). The change in SBT with time was calculated as the annual mean across each study area (Fig. 2 and SI Appendix, Methods). SBT thus reflects the likely average temperature individual fish may have experienced over time rather than temperature when sampled. Biodiversity Measures. Species richness is typically measured as alpha (local) and gamma (regional) diversity, while the extent of differentiation of communities along habitat (spatial and/or temporal) gradients, is beta diversity (36,  58). The most commonly used measures of beta diversity, such as the Jaccard index, are influenced by species richness, which should be accounted for where species richness significantly varies. We thus report both total beta diversity and a species-richness-independent index here called turnover, following (35). The mathematical difference between beta diversity and turnover is called nestedness, and arises where sites with less species are a subset of species from neighboring sites with more species. Thus, we report four measures of diversity: alpha, gamma, total beta, and turnover. Alpha Diversity. We assessed alpha diversity in the study area as the annual mean species richness per trawl (species/trawl) using presence data only. GAMs were used to explore the changes in species richness with time, correcting for the different swept area by adding it as a logistic explanatory variable. The model construction was done using the package “mgcv” within the R software (59, 60). We used BRTs to explore the drivers of these changes, and to make spatial projections of the species richness in the study area. A threshold of 0.9 correlation coefficient was selected due to the robustness of the procedure to collinearity (38) (SI Appendix, Methods). The model calibration was conducted using 75% of the trawls, and the remaining 25% were used for model validation. To improve the model calibration and reduce computational requirements, we conducted a 50 times bootstrapping process (SI Appendix, Methods). Each model was limited to 20,000 trees, and tree complexity was set to 3 to allow for interactions between the explanatory variables. The learning rate was adjusted to 0.01, bag fraction to 0.3, and other parameters were set to their defaults (38). Pearson correlations were used for model validation. Species richness was projected annually for each bootstrap, taking the mean as the final annual projection. To study the effect of each environmental variable on species richness, we built partial dependence plots using the R package “devEMF” and basic R from the BRT model. These showed the mean species richness predicted across each environmental variable gradient, while all the other variables were kept at their means (59, 61). To explore the spatial changes in alpha diversity over time, we statistically predicted the geographic distribution of species richness in each year from 2017 to 2019 and 1994 to 1996 from the BRT model, and calculated the difference between the mean of each period. We used the “predict” function of the “dismo” package, and the package “raster” from R (38, 55, 59, 62). Maps were created with QGIS (57). Beta Diversity. Beta diversity and its turnover component were calculated using the mean pairwise Jaccard dissimilarity index (ßJ) from presence-only data, which can be divided into two components: species replacement (Repl.) and nestedness (Nest.) (35, 37, 63):ßJ=b+ca+b+c=2 min(b,c)a+b+c+b-ca+b+c→Repl.=2 min(b,c)a+b+c; Nest.=b-ca+b+c, where a refers to the number of species present in both sites, and b and c are the species present only in either site. Pairwise means each trawl is compared with every other trawl in a year and the average dissimilarity calculated. Here we report total beta diversity (ßJ) and the relative contribution of species replacement, to beta diversity, which we refer to as turnover:Turnover=Repl.ßJ. To account for differing sample sizes between years, we conducted a bootstrapping process. That is, we randomly selected a subsample of the smallest sample size over years, calculated its total beta diversity and turnover, and repeated this process 200 times for each year (35). We report the mean of these bootstrapping calculations in Fig. 5 and SI Appendix, Fig. S6 (35). All calculations were carried out using the functions beta.div.comp, from the adespatial package (64). To explore the trend of these changes with time and to account for the negative trend of swept area with time (SI Appendix, Fig. S3), we fitted a GAM model with year and annual mean swept area as explanatory variables to the annual mean pairwise total beta diversity and turnover. Gamma Diversity. To study changes in gamma diversity, we constructed species accumulation curves (SAC) for each year across our study regions (40, 65). To eliminate the bias that a selected order of the locations may have on the overall SAC, we used the function “specaccum,” in the “vegan” package, set the parameter method = “random” to randomize the order of site addition, and permutated the process 200 times (65, 66). We report the mean species richness per trawl of all random permutations. We then assessed the change in gamma diversity with time, standardized to the minimum common number of sites, fitting a GAM model with year and annual swept area as the explanatory variables. To test whether the results were robust to alternative measures of richness, we used the package “SpadeR” to calculate nine different species richness indices with time, each of them with different statistical assumptions, including Chao indices, incidence-based indices, and first and second order jackknife estimators (67, 68). Individual Species Contributions. To study which species drove the changes in biodiversity, we fitted GAMs to the presence data of each species with smooth parameters for depth, swept area and latitude, and fixed effect of year, using a binomial distribution. Each species biomass-weighted mean latitude was calculated from the complete area to arrange the species by their Arctic affinity, and simple linear regression was used to assess the effect of mean latitude on the temporal change in species probability of occurrence. Supplementary Material Appendix 01 (PDF) Click here for additional data file. C.G.-V. would like to thank Eric Molina for his interest in the work and helpful discussion. M.C. would like to acknowledge partial funding from the Spanish National Project ProOceans (PID2020-118097RB-I00) and institutional support of the “Severo Ochoa Centre of Excellence” accreditation (CEX2019-000928-S). F.S. would like to acknowledge partial funding from the National Institute of Water and Atmospheric Research (NIWA) Coasts & Oceans Research Programme 5 (SCI 2020/21). Author contributions C.G.-V., F.S., M.C., and M.J.C. designed research; C.G.-V. performed research; C.G.-V. analyzed data; F.S. contributed to and critically revised the manuscript; M.C. and M.J.C. contributed to and critically revised the manuscript. Supervised the whole process.; C.L. contributed to and critically revised the manuscript; and C.G.-V. wrote the paper. Competing interest The authors declare no competing interest. Data, Materials, and Software Availability Previously published data were used for this work (Ove Djupevåg (2021) IMR bottom trawl data 1980 to 2020 10.21335/NMDC-328259372). All the code and subset of data is available through GitHub (https://github.com/CescGV) Supporting Information This article is a PNAS Direct Submission. A.B. is a guest editor invited by the Editorial Board. ==== Refs 1 E. S. Poloczanska , Global imprint of climate change on marine life. Nat. Clim. Change 3 , 919–925 (2013). 2 F. T. Dahlke, S. Wohlrab, M. Butzin, H. O. Pörtner, Thermal bottlenecks in the life cycle define climate vulnerability of fish. Science (80-) 369 , 65–70 (2020). 3 Boyin Huang , (2017): NOAA Extended Reconstructed Sea Surface Temperature (ERSST), Version 5. NOAA National Centers for Environmental Information. doi:10.7289/V5T72FNM [1/07/2021]. 4 H.-O. Pörtner, “Climate change 2022: Impacts, Adaptation, and Vulnerability. Contribution of Working Group II to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change”, H.O. Pörtner , (Cambridge Univ. Press. Press, 2022). 5 C. Chaudhary, A. J. Richardson, D. S. Schoeman, M. J. Costello, Global warming is causing a more pronounced dip in marine species richness around the equator. Proc. Natl. Acad. Sci. U. S. A. 118 (2021). 6 J. M. Sunday, A. E. Bates, N. K. Dulvy, Thermal tolerance and the global redistribution of animals. Nat. Clim. Change 2 , 686–690 (2012). 7 M. L. Pinsky, R. L. Selden, Z. J. Kitchel, Climate-driven shifts in marine species ranges: Scaling from organisms to communities. Ann. Rev. Mar. Sci. 12 , 153–179 (2020). 8 M. R. England, I. Eisenman, N. J. Lutsko, T. J. W. Wagner, The recent emergence of Arctic amplification. Geophys. Res. Lett. 48 , e2021GL094086 (2021). 9 K. Isaksen , Exceptional warming over the Barents area. Sci. Rep. 12 , 1–18 (2022).34992227 10 M. Fossheim , Recent warming leads to a rapid borealization of fish communities in the Arctic. Nat. Clim. Change 5 , 673–677 (2015). 11 R. D. Batt, J. W. Morley, R. L. Selden, M. W. Tingley, M. L. Pinsky, Gradual changes in range size accompany long-term trends in species richness. Ecol. Lett. 20 , 1148–1157 (2017).28699209 12 J. G. Hiddink, R. ter Hofstede, Climate induced increases in species richness of marine fishes. Glob. Change Biol. 14 , 453–460 (2008). 13 A. E. Magurran, M. Dornelas, F. Moyes, N. J. Gotelli, B. McGill, Rapid biotic homogenization of marine fish assemblages. Nat. Commun. 6 , 1–5 (2015). 14 L. Pecuchet , Novel feeding interactions amplify the impact of species redistribution on an Arctic food web. Glob. Chang. Biol. 26 , 4894–4906 (2020).32479687 15 A. Frainer , Increased functional diversity warns of ecological transition in the Arctic. Proc. R. Soc. B Biol. Sci. 288 : 202100547 (2021). 16 H. M. Pereira, Scenarios for global biodiversity in the 21st century. Science (80-) 330 , 1496–1501 (2010). 17 W. W. L. Cheung , Projecting global marine biodiversity impacts under climate change scenarios. Fish Fish. 10 , 235–251 (2009). 18 J. G. Molinos , Climate velocity and the future global redistribution of marine biodiversity. Nat. Clim. Change 6 , 83–88 (2016), 10.1038/nclimate2769. 19 J. C. Comiso, Large decadal decline of the arctic multiyear ice cover. J. Clim. 25 , 1176–1193 (2012). 20 K. A. Mork, Ø. Skagseth, H. Søiland, Recent warming and freshening of the Norwegian Sea observed by Argo data. J. Clim. 32 , 3695–3705 (2019). 21 S. Lind, R. B. Ingvaldsen, T. Furevik, Arctic warming hotspot in the northern Barents Sea linked to declining sea-ice import. Nat. Clim. Change 8 , 634–639 (2018). 22 K. F. Drinkwater, T. Kristiansen, A synthesis of the ecosystem responses to the late 20th century cold period in the northern North Atlantic. ICES J. Mar. Sci. 75 , 2325–2341 (2018). 23 S. Koenigstein , Forecasting future recruitment success for Atlantic cod in the warming and acidifying Barents Sea. Glob. Change Biol. 24 , 526–535 (2018). 24 E. K. Stenevik, S. Sundby, Impacts of climate change on commercial fish stocks in Norwegian waters. Mar. Policy 31 , 19–31 (2007). 25 W. W. L. Cheung, R. Watson, D. Pauly, Signature of ocean warming in global fisheries catch. Nature 497 , 365–368 (2013).23676754 26 D. P. Tittensor , Next-generation ensemble projections reveal higher climate risks for marine ecosystems. Nat. Clim. Chang. 11 , 973–981 (2021).34745348 27 X. Cormon, A. Kempf, Y. Vermard, M. Vinther, P. Marchal, Emergence of a new predator in the North Sea: Evaluation of potential trophic impacts focused on hake, saithe, and Norway pout. ICES J. Mar. Sci. 73 , 1370–1381 (2016). 28 J. N. Druon , Habitat suitability of the Atlantic bluefin tuna by size class: An ecological niche approach. Prog. Oceanogr. 142 , 30–46 (2016). 29 G. Skaret, E. Bachiller, H. Langøy, E. K. Stenevik, Mackerel predation on herring larvae during summer feeding in the Norwegian Sea. ICES J. Mar. Sci. 72 , 2313–2321 (2015). 30 R. M. Wienerroither , The marine fishes of Jan Mayen Island, NE Atlantic - past and present. Mar. Biodivers. 41 , 395–411 (2011). 31 S. Kortsch, R. Primicerio, M. Fossheim, A. V. Dolgov, M. Aschan, Climate change alters the structure of arctic marine food webs due to poleward shifts of boreal generalists. Proc. R. Soc. B Biol. Sci., 282 , 20151546 (2015), 10.1098/rspb.2015.1546. 32 I. D. Alabia , Marine biodiversity refugia in a climate-sensitive subarctic shelf. Glob. Chang. Biol. 27 , 3299–3311 (2021).33899298 33 R. E. Holt , Snow crab (Chionoecetes opilio), a new food item for North-east Arctic cod (Gadus morhua) in the Barents Sea. ICES J. Mar. Sci. 78 , 491–501 (2021). 34 G. Lorentzen , Current status of the red king crab (Paralithodes camtchaticus) and snow crab (Chionoecetes opilio) industries in Norway. Rev. Fish. Sci. Aquac. 26 , 42–54 (2018). 35 A. Baselga, Partitioning the turnover and nestedness components of beta diversity. Glob. Ecol. Biogeogr. 19 , 134–143 (2010). 36 R. H. Whittaker, Evolution and measurement of species diversity. Taxon 21 , 213–251 (1972). 37 P. Legendre, Interpreting the replacement and richness difference components of beta diversity. Glob. Ecol. Biogeogr. 23 , 1324–1334 (2014). 38 J. Elith, J. R. Leathwick, T. Hastie, A working guide to boosted regression trees. J. Anim. Ecol. 77 , 802–813 (2008).18397250 39 S. N. Wood, Generalized Additive Models: An Introduction with R, C. and Hall/CRC, Ed. (2017). 40 J. Soberón, J. Llorent, El uso de funciones dc acumulación de especies para la prediccion de la rigueza de especies. Conserv. Biol. 7 , 480–488 (1993). 41 K. E. Ellingsen , The rise of a marine generalist predator and the fall of beta diversity. Glob. Chang. Biol. 26 , 2897–2907 (2020).32181966 42 M. Dornelas, Assemblage time series reveal biodiversity change but not systematic loss. Science (80-.) 344 , 296–299 (2014). 43 S. H. M. Butchart, Global biodiversity: Indicators of recent declines. Science (80-.) 328 , 1164–1168 (2010). 44 F. Pilotto , Meta-analysis of multidecadal biodiversity trends in Europe. Nat. Commun. 11 , 1–11 (2020).31911652 45 H. Kreft, W. Jetz, A framework for delineating biogeographical regions based on species distributions. J. Biogeogr. 37 , 2029–2053 (2010). 46 P. E. Renaud , Is the poleward expansion by Atlantic cod and haddock threatening native polar cod, Boreogadus saida? Polar Biol. 35 , 401–412 (2012). 47 J. M. Węsławski, J. Legeżyńska, M. Włodarska-Kowalczuk, Will shrinking body size and increasing species diversity of crustaceans follow the warming of the Arctic littoral? Ecol. Evol. 10 , 10305–10313 (2020).33072260 48 D. P. Tittensor , Global patterns and predictors of marine biodiversity across taxa. Nature 466 , 1098–1101 (2010).20668450 49 O. Djupevåg, IMR bottom trawl data 1980-2020. Institute of Marine Research (2021), 10.21335/NMDC-328259372. 50 M. J. Costello , Marine biogeographic realms and species endemicity. Nat. Commun. 81 , 1–10 (2017). 51 E. J. Sbrocco, P. H. Barber, MARSPEC: Ocean climate layers for marine spatial ecology. Ecology 94 , 979 (2013). 52 J. Assis , Bio-ORACLE v2.0: Extending marine data layers for bioclimatic modelling. Glob. Ecol. Biogeogr. 27 , 277–284 (2018). 53 A. Gounou, M. Drévillon, M. Clavier, PRODUCT USER MANUAL For Global Ocean Reanalysis Products. E.U. Copernicus Mar. Serv, Inf. [Online] (2020) (August 18, 2021). 54 C. F. Dormann , Collinearity: A review of methods to deal with it and a simulation study evaluating their performance. Ecography (Cop.) 36 , 27–46 (2013). 55 R. J. Hijmans (2015). Raster: Geographic Data Analysis and Modeling. R Package Version 2.4-15. http://CRAN.R-project.org/package=raster 56 Flanders Marine Institute, Maritime Boundaries Geodatabase: Maritime Boundaries and Exclusive Economic Zones (200NM). version 11le , 10.14284/386 (2019). 57 QGIS Development Team (2021). QGIS Geographic Information System. Open Source Geospatial Foundation Project. http://qgis.osgeo.org 58 P. Legendre, D. Borcard, P. R. Peres-Neto, Analyzing beta diversity: Partitioning the spatial variation of community composition data. Ecol. Monogr. 75 , 435–450 (2005). 59 R Core Team (2022). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/ 60 S. N. Wood, N. H. Augustin, GAMs with integrated model selection using penalized regression splines and applications to environmental modelling. Ecol. Modell. 157 , 157–177 (2002). 61 P. Johnson (2020). _devEMF: EMF Graphics Output Device_. R package version 4.0-2. http://CRAN.R-project.org/package=devEMF 62 R. J. Hijmans, S. Phillips, J. Leathwick, J. Elith (2022). _dismo: Species Distribution Modeling_. R package version 1.3-9. http://CRAN.R-project.org/package=dismo 63 J. Podani, D. Schmera, A new conceptual and methodological framework for exploring and explaining pattern in presence - absence data. Oikos 120 , 1625–1638 (2011). 64 S. Dray , (2022). _adespatial: Multivariate Multiscale Spatial Analysis_. R packageversion 0.3-19. 10.1890/11-1183.1 http://CRAN.R-project.org/package=adespatial 65 K. I. Ugland, J. S. Gray, K. E. Ellingsen, The species-accumulation curve and estimation of species richness. J. Anim. Ecol. 72 , 888–897 (2003). 66 J. Oksanen, Package “vegan” title community ecology package (Version 2.5-7, 2020). 67 A. Chao, C.-H. Chiu, “Nonparametric estimation and comparison of species richness” in ELS (JohnWiley & Sons Ltd., 2016), pp. 1–11. 68 A. Chao, K. H. Ma, T. C. Hsieh, C.-H. Chiu, SpadeR: Species-richness prediction and diversity estimation with R. CRAN (2016) (April 20, 2022).
PMC009xxxxxx/PMC9942871.txt
==== Front Proc Natl Acad Sci U S A Proc Natl Acad Sci U S A PNAS Proceedings of the National Academy of Sciences of the United States of America 0027-8424 1091-6490 National Academy of Sciences 36656863 202208749 10.1073/pnas.2208749120 research-articleResearch Articlecell-bioCell Biology409 Biological Sciences Cell Biology Soluble cyclase-mediated nuclear cAMP synthesis is sufficient for cell proliferation Pizzoni Alejandro a https://orcid.org/0000-0002-7714-9968 Zhang Xuefeng a https://orcid.org/0000-0002-1932-1598 Naim Nyla a 1 https://orcid.org/0000-0002-2564-9725 Altschuler Daniel L. [email protected] a 2 https://orcid.org/0000-0001-7599-9865 aDepartment of Pharmacology and Chemical Biology, University of Pittsburgh School of Medicine, Pittsburgh, PA 15261 2To whom correspondence may be addressed. Email: [email protected]. Edited by Robert Lefkowitz, HHMI, Durham, NC; received May 24, 2022; accepted December 9, 2022 1Present Address: Scientific Support, Addgene, Watertown, MA 02472. 19 1 2023 24 1 2023 19 7 2023 120 4 e220874912024 5 2022 09 12 2022 Copyright © 2023 the Author(s). Published by PNAS. 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This article is distributed under Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND). Significance GPCRs are the largest family of mammalian receptors. GsPCRs signaling via cAMP initiates at the plasma membrane (first wave) and continues upon receptor internalization from an intracellular compartment (second wave), whose physiological role is not yet clearly established. Utilizing the TSHR-thyroid system, we show that TSH triggers an internalization-dependent accumulation of nuclear cAMP mediated solely by local soluble cyclase (sAC) activation, rather than cAMP diffusion from the cytosol. We present a new “three-wave model of cAMP signaling” proposing that the role of the sustained cAMP (second wave) is the mobilization of Ca2+ as an intermediate diffusible factor that enters the nucleus and activates sAC. This nuclear sAC-generated cAMP (third wave) is sufficient and rate-limiting for thyroid cell proliferation. cAMP, a key player in many physiological processes, was classically considered to originate solely from the plasma membrane (PM). This view was recently challenged by observations showing that upon internalization GsPCRs can sustain signaling from endosomes and/or the trans-Golgi network (TGN). In this new view, after the first PM-generated cAMP wave, the internalization of GsPCRs and ACs generates a second wave that was strictly associated with nuclear transcriptional events responsible for triggering specific biological responses. Here, we report that the endogenously expressed TSHR, a canonical GsPCR, triggers an internalization-dependent, calcium-mediated nuclear sAC activation that drives PKA activation and CREB phosphorylation. Both pharmacological and genetic sAC inhibition, which did not affect the cytosolic cAMP levels, blunted nuclear cAMP accumulation, PKA activation, and cell proliferation, while an increase in nuclear sAC expression significantly enhanced cell proliferation. Furthermore, using novel nuclear-targeted optogenetic actuators, we show that light-stimulated nuclear cAMP synthesis can mimic the proliferative action of TSH by activating PKA and CREB. Therefore, based on our results, we propose a novel three-wave model in which the “third” wave of cAMP is generated by nuclear sAC. Despite being downstream of events occurring at the PM (first wave) and endosomes/TGN (second wave), the nuclear sAC-generated cAMP (third wave) is sufficient and rate-limiting for thyroid cell proliferation. GPCR nuclear calcium nuclear cAMP nuclear PKA proliferation HHS | NIH (NIH) 100000002 GM099775 Daniel L. Altschuler HHS | NIH (NIH) 100000002 GM130612 Daniel L. Altschuler ==== Body pmcCyclic adenosine monophosphate (cAMP), the first second messenger described (1, 2), is a key intermediate in signaling pathways controlling many cellular processes including proliferation, differentiation, survival, and metabolism (3). The players involved in its synthesis (adenylyl cyclases, ACs) and degradation (phosphodiesterases, PDEs) are relatively well-defined; however, the mechanisms cells utilize to compute (code/decode) the relay of the cAMP signal (i.e., fidelity, specificity, efficiency) are still not understood. In the canonical cAMP pathway, an activated GPCR (G-protein coupled receptor) couples to heterotrimeric Gs to stimulate one or more isoforms of the transmembrane adenylyl cyclases (tmAC, AC1-9) at the plasma membrane (PM). The synthesized cAMP directly binds and activates a set of effectors (4–10), including protein kinase A (PKA), whose catalytic subunit (PKA-C) can diffuse to the nucleus, phosphorylate transcription factors and initiate the transcription of cAMP-specific genes (11, 12). An additional intracellular source of cAMP is soluble adenylyl cyclase (sAC, AC10) which, unlike tmACs that are regulated by Gs and forskolin (FK), is activated by bicarbonate, Ca2+, and ATP (13). In addition to the cytosol, sAC was found at discrete cellular locations including the nucleus (14), where it represents the only cAMP-synthesizing activity. These findings showed that cAMP can be synthesized deep in the cell and ruled out the PM as the only cAMP source. TSHR (thyroid-stimulating hormone receptor) is a class A GPCR activated by TSH, a glycoprotein synthesized by the thyrotrophs in the anterior lobe of the pituitary (adenohypophysis) that is the main regulator of the thyroid gland (15). TSHR activation leads to membrane G-protein activation and stimulation of secondary messenger pathways, mainly Gs-AC-cAMP (16, 17). Downstream activation of cAMP effector pathways, e.g., Epac1 and PKA, works synergistically toward thyrocyte proliferation (18, 19), establishing thyroid/TSHR as a bona fide model to study cAMP signaling. The spatiotemporal properties of signaling intermediates impart specificity on biological outputs (20–24). According to the classical view, cAMP signaling was considered to originate solely from the PM, with receptor endocytosis mediating signal termination, leading to a transient cAMP response (25). This original view was recently challenged by observations, first reported for TSHR (26), showing that upon internalization GPCRs can sustain cAMP signaling from intracellular membranes. In this novel “two-wave paradigm,” upon the initial transient PM-generated cAMP wave, a sustained second wave is observed following agonist-induced receptor internalization into the endocytic compartment (27–33). This temporal bias might convey informational content if cells are able to discriminate cAMP production from different sources and trigger unique biological outcomes differentially associated with these two phases. In this context, a sustained cAMP elevation was reported to be specifically required in several physiological responses, such as chronic inflammation and pain (34–37), to resume meiosis in the oocyte (38), and several reports indicate its role in cAMP-dependent nuclear transcription (30, 31, 39–44). Similarly, internalization-mediated sustained cAMP elevation can aid in producing stable responses to pulsatile or low-concentration hormones such as PTH. In this respect, fast cytosolic and nuclear cAMP elevations were observed upon the synergistic action of adrenergic and PTHR receptors in an internalization-dependent manner (39). More conclusively, it was recently demonstrated that only a photoactivated bacterial cyclase (bPAC) targeted to the outer mitochondria or endosome but not to the PM can induce a light-dependent CREB phosphorylation and nuclear transcription (42). Thus, it is becoming clear that endocytosis, via altering the subcellular location of cAMP production rather than changing the total amount of cAMP produced, might facilitate the selective activation of distinct PKA pools (40, 43). Furthermore, recent studies suggest the existence of a pool of PKA holoenzyme in the nucleus (39, 45, 46), adding another challenge to the classical cAMP model. The existence of a buffering capacity (47, 48) and of PDE nanodomains (47, 49) at different cellular locations can explain the “barriers” that prevent cAMP produced at different locations from transducing its full repertoire of downstream effects; for example, it is known that in some cells PM-generated cAMP is unable to reach the nuclear compartment (50), unless PDE activity is inhibited (42). Moreover, the kinetic discrepancies between nuclear cAMP accumulation and nuclear PKA activation also indicated the existence of cAMP nanodomains where AKAP-PKA-PDE complexes constrain cAMP levels and control nuclear PKA activation (45). Thus, despite the controversy of whether cAMP diffusion or PKA-C translocation represents the rate-limiting step, it is becoming increasingly clear that the location, localization, and duration of signals are key factors determining nuclear PKA activation and cAMP-dependent transcription. However, whether cAMP diffuses from the cytosol or is synthesized locally in the nucleus is still unknown. In the present study, we demonstrate using a combination of pharmacological, genetic, and optogenetic manipulations that local sAC activation is the only source of TSH-mediated nuclear cAMP production/elevation. While sAC inhibition minimally affected the cytosolic cAMP levels, it blunted nuclear cAMP accumulation and PKA activation. Moreover, our results show that sAC-mediated nuclear accumulation is not only necessary but sufficient and rate-limiting for cAMP-dependent proliferation. Results TSH increases Nuclear cAMP in an sAC-dependent Manner. To monitor cAMP dynamics in live cells, we exploited the TSH/PCCL3 thyroid model system we have extensively used in the past (18, 19, 51–54). Real-time cAMP recordings were performed by transfecting PCCL3 thyroid follicular cells with cytosolic and nuclear-targeted FRET-based cAMP sensors, as we reported before (52) (SI Appendix, Fig. S1 A and B). As shown in Fig. 1 A and B, TSH stimulated a sustained cAMP increase in both cytosolic and nuclear compartments. Nuclear events are frequently associated with a sustained elevation of cytosolic cAMP that is dependent on receptor-mediated endocytosis. Accordingly, pharmacological (Dyngo-4a and Dynasore) and genetic (dn K44A) dynamin inhibitors blocked the TSH-mediated sustained phase in a dose-dependent manner and blunted nuclear cAMP accumulation (Fig. 1C and SI Appendix, Fig. S2 A–G). Similarly, barbadin, a novel dynamin-independent endocytosis inhibitor that targets β-arrestin/AP2 (55) only affected the TSH-mediated sustained cytosolic phase and nuclear cAMP accumulation (Fig. 1D and SI Appendix, Fig. S2 F and G). None of the dynamin inhibitors affected the cAMP response to direct activation of sAC using bicarbonate (SI Appendix, Fig. S3 A and B). Fig. 1. TSH elevates nuclear cAMP in an sAC-dependent manner. Real-time FRET-based measurements of cAMP levels using the NLS-H188 (nuclear) and H188 (cytosolic) sensors in live PCCL3 cells. FRET ratios were normalized to the maximal response to forskolin (20 μM) [cAMP (% of FK)]. Nuclear (A) and cytosolic (B) cAMP measurements of cells stimulated with TSH (1 mIU/mL). Nuclear (Left) and cytosolic (Right) cAMP measurements of cells preincubated for 15 min with Dyngo-4a (D4a, 30 μM) (C), barbadin (Barba, 100 μM) (D), or vehicle (Veh; DMSO). Nuclear (E) and cytosolic (F) cAMP measurements after 30 min of incubation with an sAC inhibitor (LRE1; 100 μM) or vehicle (Veh; DMSO). Quantitative analysis of the same data (E and F) as the % of their respective controls after 15 min of incubation with TSH (G). Concentration-dependent inhibition of the TSH response by LRE1. Measurements shown in this panel were performed with the NLS-H188 (nuclear) sensor and are shown as the % of the maximum response to TSH (H). Cyclase activity and cAMP steady-state measurements (pmol/mL) were performed after labeling the cellular pool of ATP with [3H]-adenine in TSH-stimulated (+) cells or unstimulated (−) cells pretreated either with vehicle (Veh; DMSO), KH7 (100 μM), or LRE1 (100 μM) (I). Data are expressed as mean ± SEM (for n cells/samples and independent experiments, see Statistics). Significance was tested using one-way ANOVA with Tukey multiple comparison tests (*P < 0.05; **P < 0.01; ****P < 0.0001; ns, not significant). The increase in nuclear cAMP could represent diffusion from the cytosol or local synthesis by nuclear sAC. sAC is expressed in PCCL3 cells, as evaluated by immunoblotting and immunofluorescence (IF), confirming its presence in the nucleus (SI Appendix, Fig. S4A). To evaluate the role of sAC in TSH-mediated nuclear cAMP accumulation, we preincubated cells with LRE1, a selective sAC inhibitor that does not affect the activity of tmACs (56). While LRE1 minimally affected the cytosolic component, it significantly inhibited nuclear cAMP accumulation (Fig. 1 E–G and SI Appendix, Fig. S4C). Dose-response experiments showed that LRE1 inhibits nuclear cAMP signals with an EC50 ~ 18 µM (Fig. 1H and SI Appendix, Fig. S4D). Importantly, LRE1 preincubation did not affect the cAMP response to FK confirming its specificity for sAC (SI Appendix, Fig. S4 E and F) To validate the FRET results, total cAMP was assessed biochemically from cell lysates preincubated with LRE1 and KH7, another sAC selective inhibitor (57), confirming a component of sAC activity in TSH-mediated total cAMP accumulation (Fig. 1I). Thus, these combined results indicate that TSH stimulation triggers a nuclear sAC-mediated cAMP accumulation that is dependent on receptor internalization. sAC Activity Is Required and Rate-Limiting for Cell Proliferation. Next, we decided to test whether sAC inhibition has an impact on TSH-mediated cell proliferation. Pharmacological inhibition of sAC by LRE1 or KH7 blocked cell proliferation, assessed by both 3H-thymidine and BrdU incorporation assays (Fig. 2A). To confirm the pharmacological results, we generated an shRNA for rat sAC. sh-sAC-mediated sAC downregulation but not sh-Vector blocked TSH-mediated cell proliferation (Fig. 2B). sAC 1-469 (sACt) is a truncated and highly active soluble cyclase variant (58). Mutations were introduced to generate a sh-resistant sACt (sACtR), and IF staining confirmed both sACtR and sACt are present in cytosol and nucleus (SI Appendix, Fig. S5 A and B); however, only expression of sACtR but not sACt was able to rescue sh-sAC-mediated inhibition (Fig. 2B). Interestingly, expression of sACtR/sACt consistently showed a proliferative response in TSH-stimulated samples (~60% BrdU/tag) above the usual values observed in non-transfected or sh-vector controls (~30 to 40% BrdU/tag). This observation prompted us to study this effect in more detail confirming that expression of sACtR/sACt increased TSH-mediated cell proliferation in a dose-dependent manner without losing the sensitivity to LRE1 (Fig. 2C). Moreover, consistent with the pharmacological inhibition by LRE1, sAC knockdown, and sACt, overexpression affected TSH-mediated nuclear cAMP accumulation without interfering with the cytosolic cAMP response (Fig. 2 D and E and SI Appendix, Fig. S5C). Thus, these results confirmed that sAC activity is required for the TSH-mediated proliferative response and suggest that local sAC-generated cAMP signals in the nucleus could represent a rate-limiting step for cell proliferation. Fig. 2. sAC activation is necessary and rate-limiting for G1/S progression. (A) Concentration-dependent inhibition of the TSH-triggered (1 mIU/mL) proliferation in PCCL3 cells by KH7, KH7.15, and LRE1 (30 min preincubation). Proliferation was assessed by 3H-thymidine (Left) or BrdU (Right) incorporation in the presence of 5% FBS as a co-mitogen. BrdU data are expressed as % of BrdU+ nuclei among the DAPI+ nuclei. (B) Cells were transfected with pSiren-red-sh-sAC1 (sh-sAC), empty vector (sh-Vec), sACt, or an sh-resistant sACt (sACtR). BrdU data are expressed as % of BrdU+ nuclei in the dsRed+ population. Broken lines represent the average %BrdU/DAPI values of control (~10%) and TSH-stimulated parental cells (~40%). (C) Cells were transfected with increasing doses (25, 50, 100, 200, and 400 ng) of sACt/sACtR plasmids, and proliferation was assessed using BrdU incorporation in the presence of vehicle (Veh; DMSO) / LRE1 (100 μM) (Left), sh-Vec / sh-sAC (Middle), or sh-sAC (Right). (D) Real-time FRET-based measurements of cAMP levels using the NLS-H188 (nuclear; Left) and H188 (cytosolic; Right) sensors in live PCCL3 cells. Cells were transfected with sACt, sh-sAC, or sh-Vec). Quantitative analysis of the same values shown in panel D after 15 min of incubation with TSH (E). FRET ratios were normalized to the maximal response to forskolin (20 μM) [cAMP (% of FK)]. Data are expressed as mean ± SEM (for n cells/samples and independent experiments, see statistics). Significance was tested using one-way ANOVA with Tukey multiple comparison tests (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns, not significant). Nuclear sAC Activity Is the Rate-Limiting Step for Cell Proliferation. To assess whether sAC expression in the nuclear compartment is relevant for cell proliferation, we transfected PCCL3 cells with either nuclear (NLS) or PM (Lyn) targeted versions of a mCherry-tagged sACt (46). Like the non-targeted version of sACt (Fig. 2B), NLS-sACt but not Lyn-sACt showed a dose-dependent effect on TSH-mediated proliferation (Fig. 3A and SI Appendix, Fig. S6A). Like LRE1 and sh-sAC, CRISPR/Cas9-mediated sAC deletion (SI Appendix, Fig. S6 B–D) blocked TSH-mediated proliferation and reduced nuclear but not cytosolic cAMP dynamics (Fig. 3 B and C). Only an sg-resistant NLS-sACt (NLS-sACtR) was able to rescue sg-sAC-mediated inhibition, and like untagged sACtR/sACt (Fig. 2B), NLS-sACtR/NLS-sACt expression consistently showed a high proliferative response in TSH-stimulated cells (Fig. 3B). Thus, these results show that a nuclear-targeted sAC can sustain cell proliferation even in an sAC-knockdown context and indicate that nuclear sAC activation is a limiting factor controlling TSH-mediated proliferation. Fig. 3. Nuclear-targeted sACt stimulates proliferation in a dose-dependent manner. (A) PCCL3 cells were transfected with increasing doses of the nuclear-targeted NLS-mCherry-sACt (Left) or the membrane-targeted Lyn-mCherry-sACt constructs (Right) and stimulated with TSH (1 mIU/mL). Proliferation is expressed as the % of BrdU+ nuclei in the mCherry+ or GFP+ populations. (B) Proliferation in cells expressing the NLS-mCherry-sACt or Lyn-mCherry-sACt plasmids (Left). CRISPR/Cas9-mediated sAC deletion: cells were transfected TLCV2-sg243 (sAC-KO) or TLCV2-vector. Rescue experiments: cells were transfected with sg-resistant NLS-mCherry-sACtR, sg-sensitive NLS-mCherry-sACt, Lyn-mCherry-sACt, or Lyn-mCherry-sACtR (Right). (C) Real-time FRET-based measurements of cAMP levels using the NLS-H188 (nuclear) and H188 (cytosolic) sensors in live PCCL3 cells. Nuclear (Left) and cytosolic (middle panel) cAMP measurements in TSH-stimulated (1 mIU/mL) cells transfected with TLCV2-vector (TLCV2) or TLCV2-sg243 (sAC-KO). Quantitative analysis of cAMP values after 15 min of TSH stimulation (Right). FRET ratios were normalized to the maximal response to forskolin (20 μM) [cAMP (% of FK)]. Data are expressed as mean ± SEM (for n cells/samples and independent experiments, see statistics). Significance was tested using one-way ANOVA with Tukey multiple comparison tests (**P < 0.01; ***P < 0.001; ****P < 0.0001; ns, not significant). sAC Inhibition Cannot Be Rescued by a Membrane-Permeable cAMP Analog. 8-Bromoadenosine 3′,5′-cyclic monophosphate (8-Br-cAMP) is a membrane-permeable cAMP analog capable of activating both Epac1 and PKA (59), cAMP effectors involved in the TSH mitogenic response (19). Accordingly, 8-Br-cAMP mimicked TSH effects and was sufficient to trigger proliferation in PCCL3 cells assessed by both 3H-thymidine and BrdU incorporation assays (Fig. 4A). Although we initially reasoned that by mimicking the cyclase reaction product, 8-Br-cAMP should bypass the negative effects of sAC inhibition, preincubation of cells with LRE1 or KH7 consistently reduced cell proliferation (Fig. 4B). Although 8-Br-cAMP incubation was able to raise nuclear cAMP levels, this effect was totally dependent on nuclear sAC; sAC downregulation (sh-sAC) and LRE1-mediated sAC inhibition blocked 8-Br-cAMP action. However, incubation with FK plus 3-isobutyl-1-methylxanthine (IBMX; an inhibitor of most PDE isoforms) generated further increases in nuclear cAMP in an LRE1-resistant manner (Fig. 4 C–E). Next, we took advantage of the rat hepatoma HC-1 cell line that has no detectable tmAC activity and shows very low basal cAMP levels (60). These cells do not respond either to GsPCR ligands or FK unless transfected with a tmAC (SI Appendix, Fig. S7 A and B) (60, 61). However, HC-1 cells endogenously express nuclear sAC as manifested by their LRE1-sensitive cAMP response to bicarbonate (SI Appendix, Fig. S7 C and D), making them a suitable model to dissect the mechanism of action of 8-Br-cAMP. Incubation with 8-Br-cAMP manifests a slow increase in the cytosolic cAMP levels that is almost absent in the nucleus for at least 30 min (SI Appendix, Fig. S7 F and G). However, upon AC9 transfection, which in the absence of stimulation does not increase basal cAMP levels (SI Appendix, Fig. S7E), 8-Br-cAMP incubation dramatically increases the rate of cytosolic cAMP accumulation with a kinetic profile comparable to an agonist-mediated sustained cAMP response (SI Appendix, Fig. S7F). Moreover, 8-Br-cAMP incubation in AC9-transfected cells reveals a nuclear LRE1-sensitive accumulation (SI Appendix, Fig. S7H). Therefore, these combined results show that 8-Br-cAMP itself is unable to reach the nuclear compartment even at concentrations where it effectively stimulates cell proliferation in PCCL3 cells. These results indicate that its effects, like those of TSH, require a sustained cAMP-dependent activation of nuclear sAC, a necessary factor for cell proliferation. Fig. 4. 8-Br-cAMP requires an active sAC to stimulate proliferation. (A) Proliferation assessment of PCCL3 cells by stimulation with increasing doses of 8-Br-cAMP. 3H-thymidine (Left) and BrdU incorporation (Right) measurements. BrdU data are expressed as % of BrdU+ nuclei among the DAPI+ nuclei. (B) Cells were incubated with 3 mM 8-Br-cAMP and increasing doses of LRE1. Proliferation was assessed by 3H-thymidine (Left) or BrdU incorporation (middle panel). Cells were incubated with 8-Br-cAMP (0.3 or 3 mM) and increasing doses of KH7 (Right). (C) Real-time FRET-based measurements of cAMP levels using the NLS-H208 (nuclear) sensor. Traces are the averaged FRET ratios normalized to the resting values (R/R0). cAMP measurements in cells preincubated with vehicle (Veh; DMSO, Left) or LRE1 (100 μM; Right). Cells were stimulated sequentially with 8-Br-cAMP (3 mM) and FK (20 μM) plus IBMX (250 μM). (D) cAMP measurements in cells transfected with pSiren-red-sh-sAC1 (sh-sAC). (E) Quantitative analysis data shown in panels C and D after 8 min of 8-Br-cAMP incubation (shown as % of Vehicle). Data are expressed as mean ± SEM (for n cells/samples and independent experiments, see Statistics). Significance was tested using one-way ANOVA with Tukey multiple comparison tests (****P < 0.0001; ns, not significant). Nuclear sAC Is Required for Local PKA Activation and CREB Phosphorylation. PKA and its substrate, and CREB are the immediate downstream players in the cAMP pathway required for cell proliferation of PCCL3 cells (19). To evaluate if sAC is the source of cAMP activating these effectors, we first assessed if nuclear PKA activity increases after TSHR activation. In cells transfected with NLS-AKAR4, a nuclear-targeted FRET-based PKA activity sensor (46), TSH incubation elicited a rapid increase in nuclear PKA activity, with a fast initial peak consistent with the nuclear cAMP kinetics, followed by a slower sustained phase (Fig. 5A and SI Appendix, Fig. S8A). Both pharmacological and downregulation experiments performed with LRE1 and sh-sAC, respectively, showed inhibition of local PKA activity, mainly affecting the fast initial peak (Fig. 5 A–C). To evaluate CREB phosphorylation, cells were starved for 20 h in a TSH-free and reduced FBS (0.5%) medium and then stimulated with TSH at saturating concentrations. pCREB levels were evaluated at different times by IF and immunoblotting, showing a positive signal in most cells (~80%) after 10 min of TSH incubation (Fig. 5D and SI Appendix, Fig. S8 B and C). Pharmacological (LRE1) and genetic (sh-sAC) inhibition of sAC impeded CREB phosphorylation (Fig. 5 D–F and SI Appendix, Fig. S8 B–D). These results show that nuclear sAC activity is essential for the activation of two successive elements of the cAMP pathway, PKA and CREB. In sum, the evidence presented so far strongly suggests that the levels of cAMP in the nuclear compartment determine G1/S progression. Fig. 5. Nuclear PKA and pCREB are activated in a sAC-sensitive manner. Real-time FRET-based measurements of nuclear PKA activity levels using the NLS-AKAR4 sensor in live PCCL3 cells. Traces are the normalized FRET ratios (R/R0) in cells stimulated with TSH (1 mIU/mL). (A) Nuclear PKA activity measurements in cells preincubated with either vehicle (Veh; DMSO) or LRE1 (100 μM) 30 min before stimulation. (B) PKA activity measurements in cells transfected with either sh-sAC or sh-Vec. (C) Quantitative analysis of either maximum (Max) or sustained (Sust.) R/R0 values in the initial 2 min of stimulation. The sustained cAMP measurements are the R/R0 values after 5 min of stimulation. (D) CREB phosphorylation was assessed by IF at 0, 3, 5, 10, 20, 30, 45, and 60 min after stimulation. Data are the % of pCREB+ nuclei among the DAPI+ nuclei at each time point. Cells were preincubated with vehicle (Veh, DMSO) or LRE (100 μM). (E) CREB phosphorylation in cells transfected with either sh-Vec or sh-sAC. Data are the % of pCREB+ nuclei among the dsRed+ nuclei (F) Quantitative analysis of nuclear CREB phosphorylation after 60 min of TSH stimulation. Data are the % of pCREB+ nuclei among DAPI or dsRed+ nuclei at each time point. Data are expressed as mean ± SEM (for n cells/samples and independent experiments, see Statistics). Significance was tested using one-way ANOVA with Tukey multiple comparison tests (*P < 0.05; **P < 0.01; ****P < 0.0001; ns, not significant). An Internalization-Dependent Calcium Increase Controls Nuclear sAC Activation. Compelling evidence shows that sAC activity can be directly stimulated by Ca2+ increases (62, 63). To evaluate if Ca2+ plays a role in nuclear sAC activation, we first assessed if TSH was able to increase Ca2+ levels. Simultaneous measurements of Ca2+ levels in both the nuclear and cytosolic compartments using compatible FRET sensors showed that TSH stimulation generates sustained Ca2+ elevations in both compartments. These elevations were blocked by preincubation with Dyngo-4a, showing their dependence on internalization (Fig. 6A). Preincubation of cells with the membrane-permeable Ca2+ chelator BAPTA-AM blunted not only the nuclear TSH-triggered Ca2+ increase but also the nuclear cAMP elevation (Fig. 6B and SI Appendix, Fig. S9A). Utilizing pharmacological inhibitors, we found that while both the cAMP and the Ca2+ responses were insensitive to JTV-519 fumarate, a ryanodine receptor (RyR) inhibitor (64), they were sensitive to the inhibition of phospholipase C (PLC) and inositol 1,4,5-trisphosphate receptor (InsP3R) [U73122 and Xestospongin C (65), respectively] (Fig. 6 C and D and SI Appendix, Fig. S9 B and C). Importantly, FR-900359, a cyclic depsipeptide that selectively inhibits Gαq (66), did not affect Ca2+ or cAMP responses (Fig. 6E and SI Appendix, Fig. S9D). Thus, we concluded that the nuclear cAMP response depends on the PLC-InsP3-InsP3R pathway, but it is not mediated by the canonical Gαq-PLCβ pathway. Next, to determine the relative involvement of nuclear and/or cytosolic InsP3 elevations, we used InsP3-buffer constructs that contain the binding domain of type I human InsP3 receptor (67–69). Both cAMP and Ca2+ elevations were nearly abolished in cells expressing the untargeted but not the nuclear-targeted InsP3-buffer construct (70) (Fig. 6 F and G and SI Appendix, Fig. S9 E and F). Finally, we measured Ca2+ and cAMP dynamics in cells transfected with nuclear- and cytosolic-targeted constructs of the high-affinity Ca2+-binding protein parvalbumin [PV; PV-dsRed-NLS and PV-dsRed-NES (71)]. Both constructs were able to suppress the TSH-triggered nuclear and cytosolic Ca2+ elevations which in turn, lead to the inhibition of the cAMP response (Fig. 6 H–L and SI Appendix, Fig. S9 G and H). Thus, these combined results indicate that TSH triggers an internalization-dependent Gαq-independent elevation of cytosolic Ca2+. Given that in some cells the nucleus contains all the machinery required for an InsP3R-mediated Ca2+ release (i.e., intranuclear PLC and InsP3R isoforms), experiments with PV and InsP3-buffer constructs strongly suggest that activation of the PLC-InsP3-InsP3R pathway and Ca2+ release events from the endoplasmic reticulum (ER) occur in the cytosol. Contrary to cAMP, Ca2+ can diffuse into the nucleus and mediate the activation of sAC and downstream events. Fig. 6. TSH triggers Ca2+ signaling using the PLC-InsP3-InsP3R pathway. Simultaneous intensiometric measurements of nuclear (NLS-R-GECO) and cytosolic (GCaMP6f-NES) Ca2+ in live PCCL3 cells. Traces are the normalized fluorescence values (F/F0) in cells stimulated with TSH (1 mIU/mL). Cells were preincubated with vehicle (Veh; DMSO) or Dyngo-4a (D4a; 30 µM). At the end of the experiment, the Ca2+ ionophore ionomycin (Iono, 10 μM) was introduced. Real-time FRET-based measurements of cAMP levels using the NLS-H188 (nuclear) sensor. Cells were preincubated with BAPTA-AM (100 μM; B), U17322 (20 μM; C), Xestospongin C (Xesto C, 20 μM; C), JTV-019 fumarate (10 μM; D), and FR-900359 (1 μM; E), or vehicle (Veh; DMSO except for FR-900359: methanol). Cells were transfected with non-targeted (InsP3-Bf; F) or nuclear-targeted (InsP3-Bf-NLS; G) InsP3-buffer constructs. Cells were also transfected with the high Ca2+ affinity cytosolic-targeted (PV-NES; H) or nuclear-targeted (PV-NLS; I) parvalbumin constructs. Nuclear but Not Cytosolic cAMP Levels Control G1/S Progression. To assess whether nuclear cAMP levels alone can control cell cycle progression, we exploited bPAC-nLuc, an optogenetic tool recently developed in our laboratory (52, 72), that combines a blue light-activated (445 nm) adenylyl cyclase (bPAC) and luciferase (nLuc). Stably transfected NLS-bPAC-nLuc cells responded to TSH as well as to Fz-4377 (a degradation-resistant substrate for nLuc derived from Furimazine) and blue light stimulation (see Methods). Compared to TSH, and like our results with the untargeted and NLS-sACt constructs (Figs. 2B and 3B), proliferation rose to 60% in cells stimulated with Fz-4377 or blue light (Fig. 7A). Although preincubation of cells with LRE1 consistently blocked proliferation triggered by NLS-sACt, the sAC inhibitor did not affect NLS-bPAC-nLuc stimulation (Fig. 7B). Consistent with its inhibitory action on nuclear cAMP and Ca2+ dynamics (SI Appendix, Figs. S6 H and I and S9H), PV-NLS also generated a decrease in cell proliferation that could be rescued by NLS-bPAC-nLuc stimulation (Fig. 7E). Moreover, a single blue light pulse (0.5 s) triggered CREB phosphorylation, but, unlike TSH, it was insensitive to LRE1 (SI Appendix, Fig. S10 A and B). These results demonstrated that NLS-bPAC-nLuc-mediated nuclear cAMP generation is sufficient for proliferation in an LRE1-insensitive manner. However, even though we confirmed that NLS-bPAC-nLuc expression is restricted to the nuclear compartment and that it generates cAMP signals in the nucleus, we also found a minor cAMP “leak” toward the cytosol (SI Appendix, Fig. S11 A and B) (72). To rule out that this leak was responsible for activating one or more cytosolic effectors, we utilized ΔRI-PDE8, a PDE8-derived construct with an increased hydrolytic activity (73) that does not bind to AKAPs (SI Appendix, Fig. S10D) (see Methods). We reasoned that targeting ∆RI-PDE8 to the ER membrane facing the cytosol with a P450 sequence (74) will rapidly inactivate any cAMP diffusing out of the nucleus into the cytosol. Both the nuclear (NLS) and the ER-cytosolic (P450) targeted versions of ∆RI-PDE8 were successful in specifically blocking cAMP local elevations without affecting the other compartment response (SI Appendix, Fig. S10C). In addition, to rule out that the cAMP “leak” was not actually cytosolic NLS-bPAC-nLuc expression, cells were photostimulated with a 445 nm laser controlled by a point scanning system that allowed us to stimulate a small area of ~1 μm2 within the cytoplasm (SI Appendix, Fig. S10E) (see Methods). While laser pulses directed to the nuclear area were able to generate cAMP elevations (detected both in the nuclear and cytosolic compartments), pulses directed to the cytosol of the same cells were not (SI Appendix, Fig. S11 A and B). In this context, NLS- and P450-∆RI-PDE8 expression successfully abolished nuclear and cytosolic cAMP elevations generated by nuclear-directed light pulses (SI Appendix, Fig. S11 C–E). Finally, while both ∆RI-PDE8 constructs were able to reduce proliferation in WT-PCCL3 cells, light-stimulated NLS-bPAC-nLuc expressing cells proliferated in the presence of P450-∆RI-PDE8 but not in the presence of NLS-∆RI-PDE8 (Fig. 7 D and E). These results conclusively demonstrate that in thyroid cells nuclear but not cytosolic cAMP is the key factor defining cell cycle progression. Fig. 7. Nuclear cAMP is sufficient for cell proliferation. (A) PCCL3 cells were stably transfected with NLS-bPAC-nLuc, and proliferation was assessed using BrdU incorporation. In the absence of light stimulation, proliferation was TSH-dependent (1 mIU/mL). In the absence of TSH, cells proliferated when exposed to chemical stimulation (Fz-4377; Fz, 1:1,000) or light (440 nm, 0.5s, 15 min interval) stimulation for 24 h. (B) Cells were transiently transfected with NLS-mCherry-sACt and stimulated with TSH or stably transfected with NLS-bPAC-nLuc and stimulated with light (440 nm, 0.5 s, 15 min interval) in the absence of TSH. Proliferation was assessed in the presence of vehicle (Veh; DMSO) or LRE1 (100 μM). (C) WT-PCCL3 cells were transfected with the NLS-∆RI-PDE8, P450-∆RI-PDE8 or empty vector (Vec), and proliferation was assessed in the presence of TSH. Cells stably transfected with NLS-bPAC-nLuc were transiently transfected with an empty vector (Vec), NLS-∆RI-PDE8, or P450-∆RI-PDE8 constructs and stimulated with light (440 nm, 0.5s, 15 min interval) (D). Experiments were repeated in the presence of the nuclear-targeted (PV-NLS) parvalbumin constructs in cells stimulated with TSH or Light (E). Data are the % of pCREB+ nuclei among DAPI+/myc+/mCh+ nuclei. Data are expressed as mean ± SEM (for n cells/samples and independent experiments, see Statistics). Significance was tested using one-way ANOVA with Tukey multiple comparison tests (****P < 0.0001; ns, not significant). Discussion Using a combination of pharmacological, genetic, and optogenetic approaches, we show that agonist-dependent TSHR-Gs-tmAC activation at the PM, in an internalization-dependent manner, results in fast nuclear sAC-mediated cAMP production, PKA activation, and CREB phosphorylation, critical for maintaining cell proliferation. We also show that small variations in the nuclear expression levels of sAC have a large impact on the proliferative response, consistent with nuclear sAC activation representing a rate-limiting step. Furthermore, our findings show that an elevation of cAMP in the nucleus, independent of a cytosolic cAMP component, is sufficient to stimulate cell proliferation. For many years, PKA-C translocation to the nucleus upon cAMP-mediated activation was thought to represent the rate-limiting step for subsequent nuclear transcriptional events. A diffusive mechanism was proposed (75) and recently estimated to represent ~1 to 2% of total cellular PKA-C (40). However, its slower kinetics compared to nuclear cAMP accumulation and PKA activation rates suggested that cAMP diffusion into the nucleus rather than PKA-C translocation might represent the rate-limiting step, acting instead on a local nuclear-resident PKA pool (39). However, our 8-Br-cAMP results indicate that despite its small molecular weight and diffusivity, cAMP is not itself entering the nucleus, it is most likely restricted in the cytosol by binding and/or PDE activity. Instead, we propose that in PCCL3 cells, intracellularly synthesized cAMP is rather required to mobilize Ca2+ as an intermediate regulator responsible for an LRE1-sensitive nuclear sAC activation. Only upon non-physiological strong stimulation with FK and/or IBMX, cAMP concentration overcomes the buffering capacity and/or PDE nanodomains, allowing it to accumulate in the nucleus in an LRE1-insensitive manner. Upon activation, nuclear PKA activity is terminated by binding to PKI, resulting in PKA-C inhibition and nuclear export (76). TSH-mediated nuclear PKA activation showed a fast sAC-dependent initial phase, followed by a delayed sAC-independent sustained phase (Fig. 5 A–C). Whether this delayed phase represents the slow PKA-C translocation responsible to replenish the local nuclear PKA pool deserves further investigation. However, our combined results utilizing sAC inhibition/downregulation and the effects of 8-Br-cAMP clearly indicate that in thyroid cells, neither PKA-C translocation nor cAMP diffusion, but rather a nuclear sAC activation is the rate-limiting step responsible for the fast nuclear cAMP accumulation, PKA activation, and CREB phosphorylation. Recent works have shown that sAC may be an alternative cAMP source amplifying the GsPCR signaling in many cell types via Ca2+ and/or HCO3−. Prostaglandins via EP1 and EP4 receptors in bronchial cells can activate sAC in a mechanism that depends on Ca2+ but not HCO3− (77, 78). Corticotropin-releasing hormone receptor 1 (CRHR1) in hippocampal neurons can activate sAC in an internalization-dependent manner, generating a sustained cAMP signal; CRH-mediated sAC activation depends on both Ca2+ and HCO3– (79). Follicle-stimulating hormone receptor (FSHR) activation in the granulosa cells in the ovary stimulates sAC in a mechanism that depends on Cystic Fibrosis Transmembrane Regulator (CFTR)-mediated HCO3– influx (80). Similarly, luteinizing hormone receptor (LHR) stimulation in testicular Leydig and Sertoli cells activated sAC, in a mechanism proposed to be dependent on CFTR/HCO3− (81), consistent with the involvement of a CFTR/HCO3−/sAC-dependent cAMP signaling in spermatogenesis (82). Interestingly, Epac-Rap has been linked to both CFTR (77, 83–85) and Ca2+ influx and mobilization from intracellular stores (reviewed in ref. 86). Our studies with TSHR are consistent with Ca2+ as the main intracellular factor regulating nuclear sAC-dependent cAMP accumulation using a Gαq-independent PLC-InsP3-InsP3R pathway. Among the PLC isoforms, a likely candidate is PLCε, a GPCR-activated isoform that is modulated by Rap1-GTP, that binds directly to and allosterically modulates PLCε (87), but not Gαq (88, 89). Furthermore, it was reported that PLCε resides in perinuclear membranes near the nuclear envelope (90, 91) and has previously been shown to mediate Epac1-Rap1 stimulatory effects on InsP3 production and Ca2+ release (92–95). Finally, in this work, we also demonstrated that cAMP generated in the nucleus is sufficient to trigger cell proliferation. Utilizing novel compartment-specific cAMP actuators to synthesize (NLS-bPAC-nLuc) or hydrolyze (ΔRI-PDE8) cAMP, we show that light-dependent nuclear cAMP synthesis is sufficient for CREB phosphorylation and G1/S entry in a sAC-independent manner. Moreover, while the expression of a cytosolic ΔRI-PDE8 blocked TSH-mediated cell proliferation, it was unable to affect light-dependent proliferation triggered by NLS-bPAC-nLuc, thus ruling out concerns of a putative cAMP “leak” activating cytosolic effectors and stimulating proliferation. Therefore, we propose that a nuclear PKA pool is rapidly activated downstream of nuclear sAC activation, thanks to local production of cAMP and without any diffusion of cAMP into the nucleus. This sAC-mediated nuclear cAMP production is sufficient to sustain proliferation regardless of upstream events occurring in the membrane or cytosolic compartments. Based on our results on the TSHR/thyroid system, we propose a new “three-wave model of cAMP signaling” (Fig. 8). Ligand-mediated GsPCR-Gαs-tmAC activation is responsible for the initial canonical cAMP synthesis generated at the PM (first cAMP wave). The internalization of the signaling complex allows the sustained production of cAMP away from the PM. Our combined results suggest that cAMP generated from this endocytic compartment (second cAMP wave) facilitates the mobilization of Ca2+ that, unlike cytosolic cAMP, can reach the nuclear compartment and rapidly activates nuclear sAC. This nuclear sAC-mediated cAMP production (third cAMP wave) quickly activates a local PKA pool which leads to CREB phosphorylation, transcription of cAMP-dependent genes, and finally cell proliferation. Future studies are needed to test whether this three-wave model is present downstream of other GsPCR-activated pathways. Recent findings on the complexity of intracellular GPCR signaling and the therapeutic potential of cAMP production outside the PM highlight the therapeutic importance of seeking new ways to target these signaling pathways for the development of drugs with improved specificity and efficacy (34, 43, 96, 97). Fig. 8. Soluble cyclase-mediated nuclear cAMP synthesis is required and sufficient for cell proliferation. Ligand-mediated GPCR activation triggers a Gαs-mediated cAMP synthesis by one or more tmACs generating the first cAMP wave in the PM compartment. Ligand binding also triggers the internalization of the signaling complex that drives cAMP production into the endocytic compartment. Our combined results suggest that cAMP generated in the endocytic compartment (second wave) triggers PLC-mediated Ca2+ release from the ER through the InsP3 receptor. Unlike cytosolic cAMP, Ca2+ can reach the nuclear compartment and rapidly activate local sAC. Local cAMP production (third wave) quickly activates a local PKA pool which leads to CREB phosphorylation activating the transcription of cAMP-dependent genes and enhancing cell proliferation. Our results also suggest that proliferation can be sustained solely by the artificial elevation of nuclear cAMP (bPAC-nLuc), independent of the events occurring in the membrane or the cytosolic compartments. Methods Materials. Forskolin (F6886), IBMX (I7018), LRE1 (SML1857), 8-Br-cAMP (B7880), BrdU (B5002), dynasore (D7693), (−)-isoproterenol (I6504), barbadin (SML3127), BAPTA-AM (A1076), ionomycin (I0634), and Dowex–Alumina resins were from Sigma. Dyngo-4a (D4a; ab120689) was from Abcam. [2,8-3H]-Adenine (NET063001MC) was from Perkin Elmer. TSH (TSH-1315B) was from Creative Biomart. U73122 (1268), JTV-519 fumarate (4564), and Xestospongin C (1280) were from Tocris. FR-900359 (33666) was from Cayman Chemical. The Nano-Glo Endurazine Live Cell Substrate prototype (Fz-4377), a degradation-resistant derivative of Furimazine, was from Promega. KH7 and its inactive analog KH7.15 were provided by Drs Buck-Levin (Cornell). Antibodies. See SI Appendix, Supporting Methods. DNA Constructs. The cAMP FRET sensors H188 and H208 (YFP-EPAC-Q270E-mScarletI) were kindly provided by Dr. Jalink and were reported before (52). NLS-H188 and NLS-H208 were constructed using PCR to subclone an SV40-NLS motif to the N-terminal by using HindIII. pcDNA3-AKAR4-NLS (Addgene plasmid #138217), pcDNA3-sACt-mCherry-NLS (Addgene plasmid #138214), and pcDNA3-Lyn-sACt-mCherry (Addgene plasmid #138216) were gifts from Dr. Jin Zhang. K44A HA-dynamin 1 pcDNA3.1 was a gift from Dr. Sandra Schmid (Addgene plasmid # 34683). CMV-NLS-R-GECO was a gift from Robert Campbell (Addgene plasmid #32462). pcDNA5-TO-PV-NES-DsRed (PV-NES; Addgene plasmid #16339) and pcDNA5-TO-PV-NLS-Dsred (PV-NLS; Addgene plasmid #16341) were a gift from Anton Bennet. The InsP3-buffer constructs were a gift from Dr. Péter Várnai. The GCaMP6f-NES sensor was a gift from Dr Robert Grosse. The NLS-∆RI-PDE8 and P450-∆RI-PDE8 plasmids contain a deletion of aa1-74 in RI from the original template (73) and were generated by the addition of 3xSV40 NLS-FLAG and P450 2C1 ER (74) targeting-FLAG to their N termini, respectively. Expression was assessed with anti-FLAG antibodies. pcDNA3.0 AC2-HA was a gift from Dr. Cooper (Cambridge University). Flag-hAC9 was kindly provided by Lily Jiang (UT Southwestern). The shRNA-sAC(1 to 2), sh-scramble (sh-Vec), and expression plasmids for sACt and sACtR were described in (98) and provided by Dr. Muller (Northwestern University). sh-Vec and sh-sAC were transferred to pSIREN-DNR-dsRedExpress. sACt and sACtR were transferred to pCMV-myc. Cell Lines and Transfections. PCCL3, a normal TSH-dependent rat thyroid follicular cell line, was grown in Coon’s Media: Nutrient Mixture F-12 Ham (Coon’s modification; Sigma-Aldrich, F6636) supplemented with 2.68 g/L sodium bicarbonate, 5% fetal bovine serum (FBS; Corning, MT35011CV), 1% penicillin/streptomycin, 2 mM L-glutamine, insulin (1 μg/mL), apo-transferrin (5 μg/mL), hydrocortisone (1 nM), and thyroid-stimulating hormone (TSH 1 mIU/mL). The rat hepatoma clonal cell line HC1 (kindly provided by Dr. Elliot Ross, University of Texas Southwestern Medical Center) was grown in DMEM (Corning, 10013CV) supplemented with 10% FBS, penicillin (100 IU/L), and streptomycin (100 mg/L), and L-glutamine. Cells were grown to ~90% confluency before passaging every 2 to 4 d at 37 °C in 5% CO2, 95% humidified air. Transient transfections were performed using Lipofectamine 3000 Transfection kit (Invitrogen) for 24 h. NLS-bPAC-nLuc stable cell lines were generated by lentiviral infection with a multiplicity of infection (MOI) of 80 and 5 μg/mL of polybrene and selected with puromycin as described before (52). Generation of sAC-KO Cells. sAC-KO PCCL3 cells were generated by CRISPR/Cas9 genome editing system. Several sg sequences were identified with CHOPCHOP (https://chopchop.cbu.uib.no/). sgRNAs were synthesized and subcloned into the BsmBI site of the all-in-one (dox-inducible Cas9-2A-eGFP and constitutive U6 promoter) TLCV2-vector (Addgene #87360). sAC-targeted sgRNAs: sg243 to exon 2 of the C1 domain and sg396 to exon 4 of the C1 domain (SI Appendix, Supporting Methods). For lentivirus production, HEK293T cells were seeded on 10 cm tissue culture dishes and incubated until cells reached ~70% confluence. TLCV2, TLCV2-sg243, or TLCV2-sg396 constructs were mixed with packaging vectors pMD2.G and pCMV-delta-R8.2 and transfected using X-tremeGENE HP (Roche). The virus-containing medium was collected and filtered. Lentivirus was concentrated using the Lenti-X concentrator (TaKaRa). Cells were selected for 6 d in puromycin followed by 3 d of induction with doxycycline. Gene editing efficiency in the puro-resistant clones was assessed by a T7e assay (CRISPR Genomic Cleavage Detection Kit, Abm). The sg243-resistant sACt-mCherry plasmids were made by introducing the following (underlined) mutations (CTATTATATCTCCGCCATCGTC). FRET and Intensiometric Measurements. Cells were seeded on 25 mm glass coverslips and transfected with the H188, H208, NLS-H188, NLS-H208, or NLS-AKAR4 FRET-based sensors or the intensiometric GCaMP6f-NES and the NLS-R-GECO sensors. Transfected cells were given fresh media for 48 h and hormone starved for 3 h in Coon’s media with 5% FBS and lacking TSH, insulin, and hydrocortisone before measurements. Cells were washed once in PBS and imaged in OptiMEM media lacking phenol red (Gibco). Two microscope setups were used; Setup A: Olympus IX70 microscope equipped with a Till Polychrome V monochromator, a 60×/1.4 NA oil objective, and a Hamamatsu CCD camera (Photonics Model C4742-80-12AG; 8 × 8 binning). Setup B: Olympus IX83 Motorized two-deck Microscope equipped with a 6-line multi-LED Lumencor Spectra X, a Prior Emission Filter Wheel, Prior Proscan XY stage, an ORCA-Fusion Digital CMOS camera (C14440-20UP; 4 × 4 binning), a 40×/1.4 NA oil objective (UPLXAPO30X). Images were acquired every 5 or 10 s depending on the experiment, using Slidebook 6 software (Intelligent Imaging Innovations Inc.). The NLS-AKAR, H188, and NLS-H188 sensors were excited at 440 nm, and fluorescence was collected using emission filters 470/30 nm and 535/30 nm with a 69008bs dichroic (Chroma Technology Corp.). The H208 and NLS-H208 sensors were excited at 500 nm with emission filters 510/20 and 620/52 nm and two dichroic beam splitters: 468/526/596 (Semrock) or 69008bs (Chroma Technology Corp) for setups A and B, respectively. The GCaMP6f-NES and the NLS-R-GECO sensors were excited at 488 and 550 nm, and fluorescence was collected using emission filters 520/20 and 620/60, respectively, with an 89402 dichroic (Chroma Technology Corp). Changes in background-subtracted FRET ratios (R) were normalized to ratios at resting conditions (R/R0). In most of the cAMP experiments, FRET ratios were normalized to the maximal response of each cell after incubation with forskolin (20 μM). Changes in background-subtracted fluorescence measurements (F) were normalized to values at resting conditions (F/F0). No significant photobleaching was observed during the time lapses. Optogenetic Stimulation. Light stimulation was achieved using a custom-built, Arduino-controlled system capable of regulating the duration, frequency, and intensity of light exposure (52, 72). In some experiments, cells were stimulated with a solid-state laser illuminator (445 nm, LDI-7, 89 North) connected to a UGA-42 Geo (Rapp OptoElectronic) point scanning system that allowed us to simulate very small circular areas (~1 μm2) inside the cells (SI Appendix, Supporting Methods). Adenylyl Cyclase Activity in Cells. Labeling of the cellular ATP pool was performed by overnight incubation with 1 mCi/mL [3H]-adenine in complete Coon’s medium. The next day cells were washed twice and incubated for 1 h in Coon’s starvation medium. Cells were then stimulated with TSH together with either vehicle or inhibitors. Reactions were stopped in trichloroacetic acid (7.5% w/v). Product [3H]-cAMP was separated from the substrate [3H]-ATP by sequential column chromatography over Dowex and Alumina, as previously described (99). IF and BrdU Labeling. Cells were grown to 70% confluency on glass coverslips, incubated with Lipofectamine 3000 Transfection kit (Invitrogen) and the corresponding plasmids for 24 h Then after 24 to 48 h cells were made quiescent by TSH starvation in Coon’s (5% FBS) media for 16 h. Upon agonist stimulation (TSH; 1 mUI/mL) for 8 h, cells were labeled for 16 h with BrdU (Sigma, 100 µM). At the end of the labeling period, incorporated BrdU was detected by indirect IF. Unless otherwise noted, IF protocol started with fixation in 4% paraformaldehyde (10 min, room temperature) and permeabilization with 0.5% Triton X-100 (10 min, room temperature). After washing, cells were stained for 30 min at 37 °C with the primary antibody diluted in PBS/3% BSA/0.05% Tween 20 (see Antibodies section). After washing, samples were incubated for 30 min at 37 °C with the secondary antibody diluted in PBS/3% BSA/0.05% Tween 20 (see Antibodies section) and DAPI (DAPI 0.125 mg/mL, Invitrogen) in PBS/3% BSA/0.05% Tween 20. After extensive washes, samples were mounted in Vectashield Vibrance Mounting Medium (Vector Labs; H-1700-10). For BrdU experiments, a sheep anti-BrdU antibody was used in the presence of RQ1 DNase (Promega, 10 units/mL). and after washing with conjugated anti-sheep antibodies (see Antibodies section). Data are expressed as the % of BrdU+ cells among the total population (DAPI) or a particular population of cells expressing different fluorescent (dsRed, mCherry, GFP) or non-fluorescent (myc, FLAG) tags. The IF images were acquired in setups A or B described above or in an Olympus Fluoview FV1000 confocal imaging system. 3H Thymidine Proliferation Assay. Thymidine incorporation assay was performed as described before (53). Briefly, cells were plated into 96-well plates (10,000 cells/well). On the next day, cells were made quiescent by serum starvation in Dulbecco’s modified Eagle’s medium, 0.2% BSA for 20 h. Upon agonist stimulation (16 h), cells were labeled with [methyl-3H]-thymidine (Amersham Biosciences; 1 μM, 1 μCi/mL), and 24 h later, samples were collected by using a cell harvester. Filters were dried and analyzed by scintillation counting. 3H-thymidine data are expressed as the averaged counts per minute measured in each well (cpm/well). Statistics. Comparisons between multiple samples were made using one-way ANOVA with Tukey's multiple comparison test or an unpaired Student's t test with Welch's correction when comparisons were made only between two samples. The number of cells/samples and independent experiments (n/N, respectively) for each panel is described below. For panels with more than one time point/concentration, we report the one with the fewest number of independent experiments. Statistical analyses and curve fitting were performed using Graph Pad Prism 9 (GraphPad Software Inc.). P values less than 0.05 (P < 0.05) were considered significant. Fig. 1: A (16/5); B (12/3); C: Nuc Veh (5/4), Nuc D4a (4/3), Cyto Veh, Cyto D4a (5/3); D: Nuc Veh (5/4), Nuc Barba (5/3), Cyto Veh (6/3), Cyto Barba (7/3); E: Nuc Veh (14/4), Nuc LRE1 (5/3), F: Cyto Veh (8/3), Cyto LRE1 (6/4); G: same as E and F; H (5/3); I: Veh− (3/3), Veh+ (3/3), KH7− (3/3), KH7+ (3/3), LRE1− (3/3), LRE1+ (3/3). Fig. 2:  A: Left and Right Middle panels: KH7 (3/3), KH7.15 (3/3), and LRE1 (3/3); Right panel: no TSH (11/5), sh-Vec (9/5), sACt (7/4), sh-sAC+sACt (12/5), sh-sAC (9/4), sh-Vec+sACtR (15/5), sh-sAC+sACtR (16/6). C: Left: Veh (7/4), LRE1 (7/3), Center: sh-Vec (4/3), sh-sAC (5/3), Right: sACt (3/3), sACtR (3/3); D/E: Nuc sh-Vec (6/4), Nuc sACt (17/6), Nuc sh-sAC (12/5), Cyto sh-Vec (7/5), Cyto sACt (18/6), Cyto sh-sAC (32/6). Fig. 3: A: Left and Right panels (3/3); B: Left and Right panels (3/3); C: Nuc TLCV2 (7/5), Nuc sAC-KO (9/6), Cyto TLCV2 (7/4), Cyto sAC-KO (14/6). Fig. 4 : A: Left and Right panels (3/3); B: Left, Center, and Right panels (3/3); C: Veh (26/5), LRE1 (13/4), sh-sAC (6/3). Fig. 5: A, B, and C: Veh (7/3), LRE1 (8/4), sh-Vec (6/3), sh-sAC (8/3). C, D, and E: Veh (3/3), LRE1 (3/3), sh-Vec (3/3), sh-sAC (3/3). Fig. 6: A: Nuc Ca2+ (7/3), Cyto Ca2+ (6/3); B: Veh (8/3), BAPTA (9/3); C: Veh (8/4), U73122 (8/3), Xesto c (5/3), D: Veh (8/3), JTV-519 (7/3); E: Veh (6/3), FR-900359 (8/3); F: Vec (7/3), InsP3-Bf (8/3); G: Vec (10/3), InsP3-Bf-NLS (6/3); H: Vec (8/4), PV-NES (9/4); I: Vec (7/3), PV-NLS (8/3). Fig. 7: A: TSH-/Stim- (22/7), TSH+/Stim− (22/6), TSH-/Fz (25/7), Light (22/5); B: NLS-sACt (18/4), NLS-sACt+LRE1 (17/4), bP-nL (18/5), bP-nL+LRE1 (20/4); C: Vec (25/5), NLS-PDE8 (12/6); P450-PDE8 (16/5); D: Vec (25/4), NLS-PDE8 (18/6); P450-PDE8 (22/4); E: Vec (27/3), PV-NLS + TSH (25/3), PV-NLS + Light (26/3). SI Appendix, Fig. S2: A: Nuc Veh (5/3), Nuc Dyna 15 μM (14/5), Nuc Dyna 15 μM (6/3); B: Cyto CTRL (5/3), Cyto Dyna 15 μM (6/3), Cyto Dyna 15 μM (5/3), C: (4/3); D: Nuc Vec (5/3), Nuc K44A (4/3); E: Cyto Vec (4/3), Cyto K44A 0.75 μg (4/3), Cyto K44A 1.5 μg (7/3). SI Appendix, Fig. S3: A: Veh (7/3), Dyna (5/3), D4a (5/3); B: Veh (6/3), Dyna (8/4), D4a (5/3). SI Appendix, Fig. S4: C: Nuc Veh (6/3), Nuc LRE1 (8/4), Cyto Veh (8/3), Cyto LRE1 (6/4); E and F: same as Fig. 1 E and F. SI Appendix, Fig. S5: C: sh-Vec (4/3), sACt (13/5). SI Appendix, Fig. S6: C: TLCV2 (3/3), sg243 (3/3); D: (3/3) for all traces/bars. SI Appendix, Fig. S7: A (6/3); B (7/3); C (6/4); D: Veh (7/3), LRE1 (8/4); E: Nuc WT (20/5), Nuc AC9 (16/6), Cyto WT (8/4), Cyto AC9 (8/3); F: WT (7/3), AC9 (4/3); G: Veh (7/3), LRE (10/4); H Veh (8/4), LRE (6/3). SI Appendix, Fig. S8: D: Veh (3/3), LRE1 (3/3). SI Appendix, Fig. S9: A: Veh (6/3), BAPTA (8/4); B: Veh (5/3), U73122 (7/3), Xesto c (11/4), C: Veh (7/4), JTV-519 (7/3); D: Veh (8/4), FR-900359 (8/3); E: Vec (9/3), InsP3-Bf (7/3); F: Vec (7/3), InsP3-Bf-NLS (7/4); G: Vec (6/3), PV-NES (8/3); H: Vec (7/4), PV-NLS (7/3). SI Appendix, Fig. S10: A and B: Dark (3/3), Light (3/3), Light + LRE1 (3/3); C: Nuc Vec (13/5), Nuc P450 (12/4); Nuc NLS (19/6); D: Cyto (46/6), Cyto P450 (22/5); Cyto NLS (19/6) SI Appendix, Fig. S11: C: Left panel (5/3), Right panel (3/3); E: NLS-H208 (6/4), H208 (6/4). Supplementary Material Appendix 01 (PDF) Click here for additional data file. We thank Drs. Cooper (Cambridge University) and Lily Jiang (UT Southwestern) for the AC2 and Flag-hAC9 expression plasmids, respectively, Dr. Anand (Pennsylvania State University) for the RI-PDE8 construct, Dr. Muller (Northwestern University) for the shRNA and sh-resistant sAC plasmids, Dr. Várnai (Budapest, Hungary) for InsP3-buffer plasmids, Dr. Grosse (Freiburg, Germany) for the GCaMP6f-NES plasmid, and Dr. Ross (University of Texas Southwestern Medical Center) for the HC1 cells. Funding was provided by the NIH grants R01 GM099775 and GM130612 to D.L.A. Author contributions D.L.A. designed research; A.P., X.Z., and D.L.A. performed research; N.N. contributed new reagents/analytic tools; A.P. and D.L.A. analyzed data; and A.P. and D.L.A. wrote the paper. Competing interest The authors declare no competing interest. Data, Materials, and Software Availability All study data are included in the article and/or SI Appendix. Supporting Information This article is a PNAS Direct Submission. ==== Refs 1 T. W. Rall, E. W. Sutherland, Formation of a cyclic adenine ribonucleotide by tissue particles. J. Biol. Chem. 232 , 1065–1076 (1958).13549487 2 E. W. Sutherland, T. W. Rall, Fractionation and characterization of a cyclic adenine ribonucleotide formed by tissue particles. J. Biol. Chem. 232 , 1077–1091 (1958).13549488 3 M. Zaccolo, A. Zerio, M. J. Lobo, Subcellular organization of the cAMP signaling pathway. Pharmacol. Rev. 73 , 278–309 (2021).33334857 4 H. M. Berman , The cAMP binding domain: An ancient signaling module. Proc. Natl. Acad. Sci. U.S.A. 102 , 45–50 (2005).15618393 5 J. de Rooij , Epac is a Rap1 guanine-nucleotide-exchange factor directly activated by cyclic AMP. Nature 396 , 474–477 (1998).9853756 6 H. Kawasaki , A family of cAMP-binding proteins that directly activate Rap1. Science 282 , 2275–2279 (1998).9856955 7 R. F. Schindler, T. Brand, The Popeye domain containing protein family–A novel class of cAMP effectors with important functions in multiple tissues. Prog. Biophys. Mol. Biol. 120 , 28–36 (2016).26772438 8 F. Zufall, G. M. Shepherd, C. J. Barnstable, Cyclic nucleotide gated channels as regulators of CNS development and plasticity. Curr. Opin. Neurobiol. 7 , 404–412 (1997).9232810 9 A. M. Krahling , CRIS-a novel cAMP-binding protein controlling spermiogenesis and the development of flagellar bending. PLoS Genet. 9 , e1003960 (2013).24339785 10 S. S. Taylor , From structure to the dynamic regulation of a molecular switch: A journey over 3 decades. J. Biol. Chem. 296 , 100746 (2021).33957122 11 P. Sassone-Corsi, The cyclic AMP pathway. Cold Spring Harb. Perspect. Biol. 4 , a011148 (2012).23209152 12 K. Benchoula, I. S. Parhar, P. Madhavan, W. E. Hwa, CREB nuclear transcription activity as a targeting factor in the treatment of diabetes and diabetes complications. Biochem. Pharmacol. 188 , 114531 (2021).33773975 13 T. Rossetti, S. Jackvony, J. Buck, L. R. Levin, Bicarbonate, carbon dioxide and pH sensing via mammalian bicarbonate-regulated soluble adenylyl cyclase. Interface Focus 11 , 20200034 (2021).33633833 14 J. H. Zippin , Bicarbonate-responsive "soluble" adenylyl cyclase defines a nuclear cAMP microdomain. J. Cell Biol. 164 , 527–534 (2004).14769862 15 J. E. Dumont, F. Lamy, P. Roger, C. Maenhaut, Physiological and pathological regulation of thyroid cell proliferation and differentiation by thyrotropin and other factors. Physiol. Rev. 72 , 667–697 (1992).1320763 16 T. Kimura , Regulation of thyroid cell proliferation by TSH and other factors: A critical evaluation of in vitro models. Endocr. Rev. 22 , 631–656 (2001).11588145 17 M. Rivas, P. Santisteban, TSH-activated signaling pathways in thyroid tumorigenesis. Mol. Cell Endocrinol. 213 , 31–45 (2003).15062572 18 D. Hochbaum, G. Barila, F. Ribeiro-Neto, D. L. Altschuler, Radixin assembles cAMP effectors Epac and PKA into a functional cAMP compartment: Role in cAMP-dependent cell proliferation. J. Biol. Chem. 286 , 859–866 (2011).21047789 19 D. Hochbaum, K. Hong, G. Barila, F. Ribeiro-Neto, D. L. Altschuler, Epac, in synergy with cAMP-dependent protein kinase (PKA), is required for cAMP-mediated mitogenesis. J. Biol. Chem. 283 , 4464–4468 (2008).18063584 20 G. Balazsi, A. van Oudenaarden, J. J. Collins, Cellular decision making and biological noise: From microbes to mammals. Cell 144 , 910–925 (2011).21414483 21 A. Ganesan, J. Zhang, How cells process information: Quantification of spatiotemporal signaling dynamics. Protein. Sci. 21 , 918–928 (2012).22573643 22 S. Jeknic, T. Kudo, M. W. Covert, Techniques for studying decoding of single cell dynamics. Front. Immunol. 10 , 755 (2019).31031756 23 Y. Jiang, N. Hao, Memorizing environmental signals through feedback and feedforward loops. Curr. Opin. Cell Biol. 69 , 96–102 (2021).33549848 24 G. Micali, G. Aquino, D. M. Richards, R. G. Endres, Accurate encoding and decoding by single cells: Amplitude versus frequency modulation. PLoS Comput. Biol. 11 , e1004222 (2015).26030820 25 I. Sutkeviciute, J. P. Vilardaga, Structural insights into emergent signaling modes of G protein-coupled receptors. J. Biol. Chem. 295 , 11626–11642 (2020).32571882 26 D. Calebiro , Persistent cAMP-signals triggered by internalized G-protein-coupled receptors. PLoS Biol. 7 , e1000172 (2009).19688034 27 S. Ferrandon , Sustained cyclic AMP production by parathyroid hormone receptor endocytosis. Nat. Chem. Biol. 5 , 734–742 (2009).19701185 28 R. Irannejad , Conformational biosensors reveal GPCR signalling from endosomes. Nature 495 , 534–538 (2013).23515162 29 T. N. Feinstein , Noncanonical control of vasopressin receptor type 2 signaling by retromer and arrestin. J. Biol. Chem. 288 , 27849–27860 (2013).23935101 30 R. S. Kuna , Glucagon-like peptide-1 receptor-mediated endosomal cAMP generation promotes glucose-stimulated insulin secretion in pancreatic beta-cells. Am. J. Physiol. Endocrinol. Metab. 305 , E161–E170 (2013).23592482 31 A. Godbole, S. Lyga, M. J. Lohse, D. Calebiro, Internalized TSH receptors en route to the TGN induce local Gs-protein signaling and gene transcription. Nat. Commun. 8 , 443 (2017).28874659 32 L. A. Merriam , Pituitary adenylate cyclase 1 receptor internalization and endosomal signaling mediate the pituitary adenylate cyclase activating polypeptide-induced increase in guinea pig cardiac neuron excitability. J. Neurosci. 33 , 4614–22 (2013).23467377 33 S. J. Kotowski, F. W. Hopf, T. Seif, A. Bonci, M. von Zastrow, Endocytosis promotes rapid dopaminergic signaling. Neuron 71 , 278–90 (2011).21791287 34 D. D. Jensen , Neurokinin 1 receptor signaling in endosomes mediates sustained nociception and is a viable therapeutic target for prolonged pain relief. Sci. Transl. Med. 9 , eaal3447 (2017).28566424 35 N. N. Jimenez-Vargas , Endosomal signaling of delta opioid receptors is an endogenous mechanism and therapeutic target for relief from inflammatory pain. Proc. Natl. Acad. Sci. U.S.A. 117 , 15281–15292 (2020).32546520 36 Q. N. Mai , A lipid-anchored neurokinin 1 receptor antagonist prolongs pain relief by a three-pronged mechanism of action targeting the receptor at the plasma membrane and in endosomes. J. Biol. Chem. 296 , 100345 (2021).33515548 37 R. E. Yarwood , Endosomal signaling of the receptor for calcitonin gene-related peptide mediates pain transmission. Proc. Natl. Acad. Sci. U.S.A. 114 , 12309–12314 (2017).29087309 38 S. Lyga , Persistent cAMP signaling by internalized LH receptors in ovarian follicles. Endocrinology 157 , 1613–1621 (2016).26828746 39 F. G. Jean-Alphonse , beta2-adrenergic receptor control of endosomal PTH receptor signaling via Gbetagamma. Nat. Chem. Biol. 13 , 259–261 (2017).28024151 40 G. E. Peng, V. Pessino, B. Huang, M. von Zastrow, Spatial decoding of endosomal cAMP signals by a metastable cytoplasmic PKA network. Nat. Chem. Biol. 17 , 558–566 (2021).33649598 41 M. J. Roberts , Inhibition of the proliferation of human lung fibroblasts by prostacyclin receptor agonists is linked to a sustained cAMP signal in the nucleus. Front. Pharmacol. 12 , 669227 (2021).33995100 42 N. G. Tsvetanova, M. von Zastrow, Spatial encoding of cyclic AMP signaling specificity by GPCR endocytosis. Nat. Chem. Biol. 10 , 1061–1065 (2014).25362359 43 A. D. White , Spatial bias in cAMP generation determines biological responses to PTH type 1 receptor activation. Sci. Signal 14 , eabc5944 (2021).34609896 44 J. M. B. Cajulao, E. Hernandez, M. E. von Zastrow, E. L. Sanchez, Glucagon receptor-mediated regulation of gluconeogenic gene transcription is endocytosis-dependent in primary hepatocytes. Mol. Biol. Cell 33 , ar90 (2022).35767325 45 T. Clister, E. C. Greenwald, G. S. Baillie, J. Zhang, AKAP95 organizes a nuclear microdomain to control local cAMP for regulating nuclear PKA. Cell Chem. Biol. 26 , 885–891.e4 (2019).30982750 46 V. Sample , Regulation of nuclear PKA revealed by spatiotemporal manipulation of cyclic AMP. Nat. Chem. Biol. 8 , 375–382 (2012).22366721 47 A. Bock , Optical mapping of cAMP signaling at the nanometer scale. Cell 182 , 1519–1530.e17 (2020).32846156 48 J. Z. Zhang , Phase separation of a PKA regulatory subunit controls cAMP compartmentation and oncogenic signaling. Cell 182 , 1531–1544.e15 (2020).32846158 49 S. E. Anton , Receptor-associated independent cAMP nanodomains mediate spatiotemporal specificity of GPCR signaling. Cell 185 , 1130–1142.e11 (2022).35294858 50 C. P. O’Banion, B. M. Vickerman, L. Haar, D. S. Lawrence, Compartmentalized cAMP generation by engineered photoactivated adenylyl cyclases. Cell Chem. Biol. 26 , 1393–1406.e7 (2019).31353320 51 M. M. Edreira , Phosphorylation-induced conformational changes in Rap1b: Allosteric effects on switch domains and effector loop. J. Biol. Chem. 284 , 27480–27486 (2009).19651783 52 N. Naim , Luminescence-activated nucleotide cyclase regulates spatial and temporal cAMP synthesis. J. Biol. Chem. 294 , 1095–1103 (2019).30559293 53 F. Ribeiro-Neto, J. Urbani, N. Lemee, L. Lou, D. L. Altschuler, On the mitogenic properties of Rap1b: cAMP-induced G(1)/S entry requires activated and phosphorylated Rap1b. Proc. Natl. Acad. Sci. U.S.A. 99 , 5418–5423 (2002).11959997 54 X. Zhang , Cyclase-associated protein 1 (CAP1) is a prenyl-binding partner of Rap1 GTPase. J. Biol. Chem. 293 , 7659–7673 (2018).29618512 55 A. Beautrait , A new inhibitor of the beta-arrestin/AP2 endocytic complex reveals interplay between GPCR internalization and signalling. Nat. Commun. 8 , 15054 (2017).28416805 56 L. Ramos-Espiritu , Discovery of LRE1 as a specific and allosteric inhibitor of soluble adenylyl cyclase. Nat. Chem. Biol. 12 , 838–844 (2016).27547922 57 K. C. Hess , The "soluble" adenylyl cyclase in sperm mediates multiple signaling events required for fertilization. Dev. Cell 9 , 249–259 (2005).16054031 58 B. S. Jaiswal, M. Conti, Identification and functional analysis of splice variants of the germ cell soluble adenylyl cyclase. J. Biol. Chem. 276 , 31698–31708 (2001).11423534 59 A. E. Christensen , cAMP analog mapping of Epac1 and cAMP kinase. Discriminating analogs demonstrate that Epac and cAMP kinase act synergistically to promote PC-12 cell neurite extension. J. Biol. Chem. 278 , 35394–35402 (2003).12819211 60 E. M. Ross, A. C. Howlett, K. M. Ferguson, A. G. Gilman, Reconstitution of hormone-sensitive adenylate cyclase activity with resolved components of the enzyme. J. Biol. Chem. 253 , 6401–6412 (1978).210183 61 C. W. Dessauer, A. G. Gilman, Purification and characterization of a soluble form of mammalian adenylyl cyclase. J. Biol. Chem. 271 , 16967–16974 (1996).8663304 62 B. S. Jaiswal, M. Conti, Calcium regulation of the soluble adenylyl cyclase expressed in mammalian spermatozoa. Proc. Natl. Acad. Sci. U.S.A. 100 , 10676–10681 (2003).12958208 63 T. N. Litvin, M. Kamenetsky, A. Zarifyan, J. Buck, L. R. Levin, Kinetic properties of "soluble" adenylyl cyclase. Synergism between calcium and bicarbonate. J. Biol. Chem. 278 , 15922–15926 (2003).12609998 64 X. H. Wehrens , Protection from cardiac arrhythmia through ryanodine receptor-stabilizing protein calstabin2. Science 304 , 292–296 (2004).15073377 65 J. Gafni , Xestospongins: Potent membrane permeable blockers of the inositol 1,4,5-trisphosphate receptor. Neuron 19 , 723–733 (1997).9331361 66 R. Schrage , The experimental power of FR900359 to study Gq-regulated biological processes. Nat. Commun. 6 , 10156 (2015).26658454 67 E. Guimaraes , Inositol 1, 4, 5-trisphosphate-dependent nuclear calcium signals regulate angiogenesis and cell motility in triple negative breast cancer. PLoS One 12 , e0175041 (2017).28376104 68 G. Gulyas , Measurement of inositol 1,4,5-trisphosphate in living cells using an improved set of resonance energy transfer-based biosensors. PLoS One 10 , e0125601 (2015).25932648 69 D. A. Gomes , c-Met must translocate to the nucleus to initiate calcium signals. J. Biol. Chem. 283 , 4344–4351 (2008).18073207 70 A. V. Zima, D. J. Bare, G. A. Mignery, L. A. Blatter, IP3-dependent nuclear Ca2+ signalling in the mammalian heart. J. Physiol. 584 , 601–611 (2007).17761776 71 M. A. Rodrigues , Nucleoplasmic calcium is required for cell proliferation. J. Biol. Chem. 282 , 17061–17068 (2007).17420246 72 N. Naim, J. M. Reece, X. Zhang, D. L. Altschuler, Dual activation of cAMP production through photostimulation or chemical stimulation. Methods Mol. Biol. 2173 , 201–216 (2020).32651920 73 N. K. Tulsian, S. Krishnamurthy, G. S. Anand, Channeling of cAMP in PDE-PKA complexes promotes signal adaptation. Biophys. J. 112 , 2552–2566 (2017).28636912 74 K. Ahn, E. Szczesna-Skorupa, B. Kemper, The amino-terminal 29 amino acids of cytochrome P450 2C1 are sufficient for retention in the endoplasmic reticulum. J. Biol. Chem. 268 , 18726–18733 (1993).8360166 75 A. T. Harootunian , Movement of the free catalytic subunit of cAMP-dependent protein kinase into and out of the nucleus can be explained by diffusion. Mol. Biol. Cell 4 , 993–1002 (1993).8298196 76 C. Liu, P. Ke, J. Zhang, X. Zhang, X. Chen, Protein kinase inhibitor peptide as a tool to specifically inhibit protein kinase A. Front. Physiol. 11 , 574030 (2020).33324237 77 P. Ivonnet, M. Salathe, G. E. Conner, Hydrogen peroxide stimulation of CFTR reveals an Epac-mediated, soluble AC-dependent cAMP amplification pathway common to GPCR signalling. Br. J. Pharmacol. 172 , 173–184 (2015).25220136 78 P. Ivonnet, H. Unwalla, M. Salathe, G. E. Conner, Soluble adenylyl cyclase mediates hydrogen peroxide-induced changes in epithelial barrier function. Respir. Res. 17 , 15 (2016).26857816 79 C. Inda , Different cAMP sources are critically involved in G protein-coupled receptor CRHR1 signaling. J. Cell Biol. 214 , 181–195 (2016).27402953 80 H. Chen, H. C. Chan, Amplification of FSH signalling by CFTR and nuclear soluble adenylyl cyclase in the ovary. Clin. Exp. Pharmacol. Physiol. 44 , 78–85 (2017).28345252 81 T. M. D. Nguyen , Effect of soluble adenylyl cyclase (ADCY10) inhibitors on the LH-stimulated cAMP synthesis in Mltc-1 leydig cell line. Int. J. Mol. Sci. 22 , 4641 (2021).33924969 82 W. M. Xu , Defective CFTR-dependent CREB activation results in impaired spermatogenesis and azoospermia. PLoS One 6 , e19120 (2011).21625623 83 J. C. Domingue, M. Ao, J. Sarathy, M. C. Rao, Chenodeoxycholic acid requires activation of EGFR, EPAC, and Ca2+ to stimulate CFTR-dependent Cl- secretion in human colonic T84 cells. Am. J. Physiol. Cell Physiol. 311 , C777–C792 (2016).27558159 84 M. J. Lobo, M. D. Amaral, M. Zaccolo, C. M. Farinha, EPAC1 activation by cAMP stabilizes CFTR at the membrane by promoting its interaction with NHERF1. J. Cell Sci. 129 , 2599–2612 (2016).27206858 85 J. D. Santos , Cytoskeleton regulators CAPZA2 and INF2 associate with CFTR to control its plasma membrane levels under EPAC1 activation. Biochem. J. 477 , 2561–2580 (2020).32573649 86 R. Kosuru, M. Chrzanowska, Integration of Rap1 and calcium signaling. Int. J. Mol. Sci. 21 , 1616 (2020).32120817 87 M. Sieng , Functional and structural characterization of allosteric activation of phospholipase Cepsilon by Rap1A. J. Biol. Chem. 295 , 16562–16571 (2020).32948655 88 T. D. Bunney, M. Katan, Phospholipase C epsilon: Linking second messengers and small GTPases. Trends Cell Biol. 16 , 640–648 (2006).17085049 89 T. K. Harden, S. N. Hicks, J. Sondek, Phospholipase C isozymes as effectors of Ras superfamily GTPases. J. Lipid. Res. 50 , S243–S248 (2009).19033212 90 C. A. Nash, W. Wei, R. Irannejad, A. V. Smrcka, Golgi localized beta1-adrenergic receptors stimulate Golgi PI4P hydrolysis by PLCepsilon to regulate cardiac hypertrophy. Elife 8 , e48167 (2019).31433293 91 C. Song , Differential roles of Ras and Rap1 in growth factor-dependent activation of phospholipase C epsilon. Oncogene 21 , 8105–8113 (2002).12444546 92 S. S. Dusaban, J. H. Brown, PLCepsilon mediated sustained signaling pathways. Adv. Biol. Regul. 57 , 17–23 (2015).25453218 93 T. G. Jin , Role of the CDC25 homology domain of PLC {epsilon} in amplification of Rap1-dependent signaling. J. Biol. Chem. 6 , 30301–30307 (2001). 94 G. G. Kelley, S. E. Reks, A. V. Smrcka, Hormonal regulation of phospholipase cepsilon through distinct and overlapping pathways involving G12 and Ras family G-proteins. Biochem. J. 378 , 129–139 (2004).14567755 95 E. A. Oestreich , Epac-mediated activation of phospholipase C(epsilon) plays a critical role in beta-adrenergic receptor-dependent enhancement of Ca2+ mobilization in cardiac myocytes. J. Biol. Chem. 282 , 5488–5495 (2007).17178726 96 D. S. Eiger , Location bias contributes to functionally selective responses of biased CXCR3 agonists. Nat. Commun. 13 , 5846 (2022).36195635 97 M. Shimizu , Pharmacodynamic actions of a long-acting PTH analog (LA-PTH) in thyroparathyroidectomized (TPTX) rats and normal monkeys. J. Bone Miner. Res. 31 , 1405–1412 (2016).26865415 98 R. L. Watson , Endothelial CD99 signals through soluble adenylyl cyclase and PKA to regulate leukocyte transendothelial migration. J. Exp. Med. 212 , 1021–1041 (2015).26101266 99 X. Zhang , CAP1 binds and activates adenylyl cyclase in mammalian cells. Proc. Natl. Acad. Sci. U.S.A. 118 , e2024576118 (2021).34099549
PMC009xxxxxx/PMC9942881.txt
==== Front Proc Natl Acad Sci U S A Proc Natl Acad Sci U S A PNAS Proceedings of the National Academy of Sciences of the United States of America 0027-8424 1091-6490 National Academy of Sciences 36656852 202208275 10.1073/pnas.2208275120 research-articleResearch Articlebiophys-physBiophysics and Computational Biology408 Biological Sciences Biophysics and Computational Biology Physical Sciences Biophysics and Computational Biology De novo protein fold design through sequence-independent fragment assembly simulations Pearce Robin a https://orcid.org/0000-0001-6402-734X Huang Xiaoqiang a https://orcid.org/0000-0002-1005-848X Omenn Gilbert S. a b c d http://orcid.org/0000-0002-8976-6074 Zhang Yang [email protected] a e f g 1 https://orcid.org/0000-0002-2739-1916 aDepartment of Computational Medicine and Bioinformatics, University of Michigan, Ann Arbor, MI 48109 bDepartment of Internal Medicine, University of Michigan, Ann Arbor, MI 48109 cDepartment of Human Genetics, University of Michigan, Ann Arbor, MI 48109 dSchool of Public Health, University of Michigan, Ann Arbor, MI 48109 eDepartment of Biological Chemistry, University of Michigan, Ann Arbor, MI 48109 fDepartment of Computer Science, School of Computing, National University of Singapore 117417, Singapore gCancer Science Institute of Singapore, National University of Singapore 117599, Singapore 1To whom correspondence should be addressed. Email: [email protected]. Edited by William DeGrado, University of California San Francisco, San Francisco, CA; received May 17, 2022; accepted December 22, 2022 19 1 2023 24 1 2023 19 7 2023 120 4 e220827512017 5 2022 22 12 2022 Copyright © 2023 the Author(s). Published by PNAS. 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This article is distributed under Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND). Significance Despite the vast sequence space, only a tiny fraction of possible folds and functions achievable by proteins have been realized in nature, probably due to the selective pressures by environmental constraints during evolution. There is considerable interest in de novo protein design to engineer artificial proteins with novel structures and functions beyond those created by nature. However, the success rate of computational de novo design remains low and frequently requires extensive user intervention and large-scale experimental optimization. To address this issue, we developed an automated open-source program, FoldDesign, which shows improved performance in creating high-fidelity stable folds compared to other state-of-the-art methods. The success of FoldDesign should enable the creation of desired protein structures with promising clinical and industrial potential. De novo protein design generally consists of two steps, including structure and sequence design. Many protein design studies have focused on sequence design with scaffolds adapted from native structures in the PDB, which renders novel areas of protein structure and function space unexplored. We developed FoldDesign to create novel protein folds from specific secondary structure (SS) assignments through sequence-independent replica-exchange Monte Carlo (REMC) simulations. The method was tested on 354 non-redundant topologies, where FoldDesign consistently created stable structural folds, while recapitulating on average 87.7% of the SS elements. Meanwhile, the FoldDesign scaffolds had well-formed structures with buried residues and solvent-exposed areas closely matching their native counterparts. Despite the high fidelity to the input SS restraints and local structural characteristics of native proteins, a large portion of the designed scaffolds possessed global folds completely different from natural proteins in the PDB, highlighting the ability of FoldDesign to explore novel areas of protein fold space. Detailed data analyses revealed that the major contributions to the successful structure design lay in the optimal energy force field, which contains a balanced set of SS packing terms, and REMC simulations, which were coupled with multiple auxiliary movements to efficiently search the conformational space. Additionally, the ability to recognize and assemble uncommon super-SS geometries, rather than the unique arrangement of common SS motifs, was the key to generating novel folds. These results demonstrate a strong potential to explore both structural and functional spaces through computational design simulations that natural proteins have not reached through evolution. de novo protein design structural design novel fold of protein structural motif replica-exchange Monte Carlo simulation HHS | NIH | National Institute of General Medical Sciences (NIGMS) 100000057 GM136422 Yang Zhang HHS | NIH | National Institute of General Medical Sciences (NIGMS) 100000057 S10OD026825 Yang Zhang Division of Intramural Research, National Institute of Allergy and Infectious Diseases (DIR, NIAID) 100006492 AI134678 Yang Zhang National Science Foundation (NSF) 100000001 IIS1901191 Yang Zhang National Science Foundation (NSF) 100000001 DBI2030790 Yang Zhang National Science Foundation (NSF) 100000001 MTM2025426 Yang Zhang HHS | NIH | National Cancer Institute (NCI) 100000054 U24CA210967 Gilbert S Omenn HHS | NIH | National Institute of Environmental Health Sciences (NIEHS) 100000066 P30ES017885 Gilbert S Omenn ==== Body pmcProteins are important biological molecules that perform the majority of cellular functions in living organisms. Their unique and varied functions are made possible by the diverse structural folds adopted by different protein molecules. However, despite the enormous conformational space available, only a tiny portion appears in nature following billions of years of evolution, probably due to the selective pressures exerted by environmental constraints upon organisms (1). For example, there have been just under 1,500 protein folds classified in the SCOPe database (2), and studies have indicated that the current PDB is nearly complete, representing the vast majority of natural folds (3, 4). Given the vital importance of proteins to living organisms, there has been growing interest in designing artificial proteins with enhanced functionality beyond their native counterparts. However, many of the attempts have focused on generating new protein sequences starting from the structures of experimentally solved proteins (5–8). While this may be effective in certain cases, protein design starting from solved structures is severely limited as nature has essentially sampled from an insignificant portion of the possible structure and function space, thereby greatly limiting the number of design applications. Given these limitations, de novo protein design, which aims to create not only artificial protein sequences, but also novel structures tailored to specific design applications, e.g., with specific fold types or binding pockets, has gained considerable traction in recent years. For instance, approaches such as Rosetta have been applied to design proteins with promising therapeutic potential (9–11), novel ligand-binding activity (12, 13), and complex logical interactions (14). The core protocol that has enabled Rosetta to design new protein folds is fragment assembly, which involves the identification of small structural fragments from experimentally solved structures that match a desired fold definition and the assembly of the identified fragments to produce full-length structural folds (15–17). Notably, fragment assembly was adapted from the related field of protein structure prediction, where it has been among the most successful classical approaches to template-free structure modeling (17–20). Despite the successes, de novo protein design remains somewhat of an art form, where large-scale experimental optimization is often required to generate successful designs (9, 11). In particular, extensive user intervention during scaffold creation and selection is frequently necessary (12, 21). Nevertheless, automated fold design tailored to specific applications is highly non-trivial because traditional homologous structure assembly programs often create folds that are similar to the template structures even when distracted with strong external spatial restraints (22, 23). Although ab initio fragment assembly approaches, such as QUARK (19) and Rosetta (17), can create template-free models, they need to start from specific natural sequences and often create conformations that either converge to specific folding clusters or are not protein-like (24). Most recently, Anishchenko et al. performed an interesting study that combined deep neural-network training with structural refinement simulations to “hallucinate” proteins; it could create novel protein sequences but the structural folds were generally close to PDB structures (with an average TM-score = 0.78) (25). Meanwhile, the resulting protein folds were largely randomized depending on the stochastic process of the design iterations, where the method was further extended to allow for the incorporation of specific functional sites or structural motifs (Smotifs) (26). In another recent approach, Huang et al. combined a neural network-derived, sidechain-independent potential (SCUBA) with stochastic dynamics simulations and demonstrated an impressive ability to generate successfully folded designs (27). Notably, the method should be used in tandem with 3D backbone sketches adapted from a ‘periodic table’ of protein structures (28) through manual manipulation and thus the conformational space of the final structures is limited to the topological area defined by the initial backbone sketches. Similarly, extensions of the Rosetta fragment assembly protocol such as TopoBuilder require pre-definition of a target fold in the form of sketches that specify the 3D arrangement of the desired secondary structure (SS) elements, where the sketches are first parametrically optimized based on matching the desired fold with analogous structures in the PDB and then assembled from fragments that match the fold definition using Rosetta (29). Other methods like SEWING (30) have been successful at producing stable designs by reassembling relatively large helical substructures identified from the PDB; however, the approach is limited to the conformations adopted by large substructures present in the PDB and has been benchmarked only on helical folds (30, 31). Additionally, most of the successful de novo designs have highly idealized structural folds with optimized SS compositions that lack the complex irregularities often present in native proteins, where a significant portion of the designed folds are well represented in nature or may be described through ideal parametric geometries (32–36). Thus, the development of automated algorithms capable of precisely designing any required fold type, including those without structure analogs in the PDB or idealized SS compositions, with limited human intervention is critical to improve the scope and success rate of de novo protein design. Toward this goal, we proposed an automated pipeline, FoldDesign, to create desired protein folds starting from user-specified restraints, such as the SS topology and/or inter-residue contact and distance maps, through sequence-independent replica-exchange Monte Carlo (REMC) simulations. Since the designed folds do not necessarily have experimental counterparts, we designed several objective assessment criteria based on the satisfaction rate of the input requirements and the folding stability of the designs, as outlined in SI Appendix, Fig. S1. The results showed that FoldDesign is capable of producing protein-like structural folds that closely recapitulate the input features with enhanced folding stability, significantly outperforming other start-of-the-art approaches on the large-scale benchmark tests. Importantly, this was demonstrated on a set of non-idealized, complex SS topologies and roughly 1/4 of the designs possessed novel folds that were not represented in the PDB, illustrating an important ability of the program to explore the areas of protein fold space unexplored by natural evolution. The online server, which presently supports fold design targets up to 1,500 residues long, and the standalone package for FoldDesign are freely available to the community at https://zhanggroup.org/FoldDesign/ and https://github.com/robpearc/FoldDesign, respectively. Results and Discussion FoldDesign is an automated algorithm for sequence-independent, de novo protein fold design, where the flowchart is outlined in Fig. 1. The program takes as input the SS topology for a designed structure scaffold, which includes the length, order, and composition of the SS elements. A set of structural fragments with lengths between 1 and 20 residues is then collected from the PDB library by scoring the similarity between the input SS and the SS of the PDB fragments. These fragments are finally reassembled by REMC folding simulations to generate protein-like structural scaffolds that satisfy the input constraints, where the lowest energy structure is subjected to further atomic-level refinement to produce the final structural design (see Methods). Fig. 1. Overview of the FoldDesign pipeline. Starting from a user-defined SS topology as well as any further design constraints such as inter-residue contacts or distances, FoldDesign identifies 1 to 20 residue structural fragments from the PDB with SSs that match the input constraints. These fragments are then assembled together along with 10 other conformational movements during the REMC folding simulations under the guidance of a sequence-independent energy function that accounts for the fundamental forces that underlie protein folding. The lowest energy structure produced during the folding simulations is selected for further atomic-level refinement by ModRefiner to produce the final designed structure. Auxiliary Movements Improve the Folding Simulation Efficiency and Ability to Identify Low-Energy States. Fragment substitution is the predominant movement used by FoldDesign, which involves the replacement of a selected region of a decoy structure with the structure from one of the identified fragments collected from the PDB. However, fragment substitution may cause large conformational changes that prevent the movement from being accepted. To improve the simulation efficiency, FoldDesign introduces 10 auxiliary movements, including bond length and angle perturbations, segment rotations, torsion angle substitutions, and those that form packing interactions between specific SS elements (SI Appendix, Text S1 and Fig. S2). Fig. 2A displays the FoldDesign energies of the lowest energy structures produced for each of the 354 test SS topologies (see Methods), either using the full set of 11 conformational movements or only using fragment substitution. Of note, the 354 test SS sequences were derived from native proteins, which include irregularities and non-ideal compositions, making it a rigorous test set to determine if a method can design stable structures given non-ideal SS definitions. It can be observed that the auxiliary movements enabled the simulations to find structures with significantly lower energies than those found using fragment substitution alone. Overall, the average FoldDesign energy of the best structures produced using the full movement set was −529.5 kBT compared to −449.7 kBT when using only fragment substitution, where the difference was statistically significant with a P-value of 2.1E-66 as determined by a paired two-sided Student’s t test. In addition to the improved ability to sample low-energy states, the auxiliary movements reduced the simulation times required to fold the proteins. Fig. 2B plots the simulation time versus the protein length for each of the test topologies. From the figure, a clear reduction in the simulation time required can be seen across all protein lengths, where the average time for the simulations with the full movement set was 9.6 h compared to 22.8 h for the simulations that used only fragment substitution. This reduction in simulation time is due to the fact that fragment substitution is computationally expensive and requires additional loop closure to ensure that it does not cause large downstream perturbations, while the auxiliary movements are comparatively fast. Fig. 2. Importance of the auxiliary conformational movements. (A) Energy distributions for the designs produced by the FoldDesign simulations using the full movement set and using only fragment assembly. (B) Simulation time required versus protein length for FoldDesign using the full movement set and fragment assembly alone. (C and D) Two representative case studies that demonstrate the dynamics of the folding simulations without (C) and with (D) the auxiliary movements. The y-axis displays the TM-score between the decoy at REMC cycle i compared to the decoy at cycle i-1. In Fig. 2 C and D, we further present a representative case study for the topology from the PDB protein 1ec6A, which adopts an α/β fold. Fig. 2C shows the conformational dynamics of the decoys produced during the lowest-temperature replica of the simulations using only the fragment substitution movement, while Fig. 2D uses the full movement set. Specifically, the figures plot the TM-score between the decoy at REMC cycle i compared to cycle i-1 from cycles 50 to 100. In Fig. 2C, there are several plateaus where no movement could be accepted, leading to identical conformations between a number of cycles, where the most notable plateau lasted for 12 cycles (cycles 59 through 70). On the other hand, with the full movement set in Fig. 2D, no such plateaus were observed. Although several cycles had very similar folds, which may be caused by subtle conformational refinements such as bond length perturbation, none of the cycles had identical structures. As a result, the simulations using the full movement set generated a structure with an energy of −346.2 kBT in 4.7 h compared to a structure with an energy of −224.3 kBT in 14.2 h using only fragment substitution. As a comparison, SI Appendix, Fig. S3 depicts the native 1ec6A structure, which had a higher FoldDesign energy (−145.5 kBT) than either of the simulated designs in Fig. 2 C and D. This is expected as de novo protein design methods optimize the structure of a design with respect to their own energy functions and the native proteins from which an SS topology was derived will most likely never be the lowest energy conformation that the sampling procedures could/should achieve. Moreover, since many natural proteins with divergent global folds may adopt similar SS types, a given natural protein, such as 1ec6A, may not necessarily represent the most optimal fold or the lowest energy structure for a given SS composition, even with a perfect energy force field. In fact, it has been shown that many de novo designed proteins have increased stability compared to their native counterparts (35, 36). This is a departure from the scenario of protein structure prediction, in which the native structure, with some caveats, should lie at the global free energy minimum for a given protein sequence following Anfinsen’s thermodynamic hypothesis (37); however, the same is not necessarily true for protein structure design given just the SS composition. FoldDesign Scaffolds Closely Match the Input Constraints. To assess its ability to design structural folds that possess the desired SS topologies, we list in Table 1 a summary of the FoldDesign results in terms of the average Q3 scores on the 354 test topologies. As a comparison, we also list the results from the state-of-the-art Rosetta method (32), which similarly starts from the desired SS of a designed scaffold, where a detailed description of the procedures used to run Rosetta is given in SI Appendix, Text S3. Here, the Q3 score is defined as the fraction of positions with SS elements that are identical to that of the input topology. Following fold generation, the SSs of the designed scaffolds for both FoldDesign and Rosetta were assigned using DSSP (38) and compared to the input for each protein. Table 1. Comparison of the Q3 scores for the structures produced by FoldDesign and Rosetta on the 354 test SS topologies Method Q3 Score All (P-value) Q3 Score α-proteins (P-value) Q3 Score β-proteins (P-value) Q3 Score α/β proteins (P-value) FoldDesign 0.877 (*) 0.934 (*) 0.863 (*) 0.875 (*) Rosetta 0.833(1.7E-08) 0.828 (5.4E-05) 0.829 (0.10) 0.835 (4.5E-06) Here, the Q3 score is defined as the fraction of positions in the designed structures whose SSs were identical to the input SSs. The results are further separated based on the fold type (α, β, and α/β) and the P-values were calculated using paired, two-sided Student’s t tests. Overall, FoldDesign achieved an average Q3 score of 0.877 compared to 0.833 for Rosetta with a P-value of 1.7E-08. When considering the Q3 scores for α-proteins, β-proteins, and α/β-proteins separately, FoldDesign achieved Q3 scores of 0.934, 0.863, and 0.875, compared to 0.828, 0.829, and 0.835, respectively, for Rosetta. Thus, across all fold types, FoldDesign was able to generate structures that more closely matched the input topologies than Rosetta. This partially reflects the advanced dynamics of the folding simulations as well as the effectiveness of the optimized energy function in FoldDesign. Although no user-defined distance restraints were included in the above tests, these are still important in many design cases where recapitulation of specific folds is desired. In SI Appendix, Table S1, we extracted the pairwise Cα distances from the native structures in the test set and used them as restraints during the design simulations. From the table, it can be seen that FoldDesign was able to generate designs that closely matched the native structures with average TM-scores/RMSDs of 0.993/0.31 Å, 0.993/0.27 Å, 0.992/0.32 Å, and 0.994/0.31 Å for all, α, β, and α/β topologies, respectively. Here, TM-score (39) is a structure comparison metric that takes a value in the range (0, 1], where a value of TM-score =1 indicates an identical match between two structures and a TM-score ≥0.5 signifies that two proteins share the same global fold (40). Therefore, the FoldDesign structures nearly perfectly recapitulated the desired folds when guided by user-defined distance restraints. Additionally, the mean absolute errors between the Cα distance maps extracted from the designed folds and native structures were 0.148, 0.115, 0.130, and 0.154 Å for all, α, β, and α/β topologies, respectively, confirming that the generated structures closely satisfied the given distance restraints. FoldDesign Generates Low-Energy, Native-Like Protein Structures. While an important metric, the Q3 score is unable to provide a complete picture of the physical quality of the designs. In theory, a method could produce trivial or even unfavorable folds that satisfy the desired SS definitions. Thus, a more detailed analysis of the energetics and physical characteristics of the produced structures had to be performed (SI Appendix, Fig. S1). As the designed scaffolds for FoldDesign and Rosetta are both sequence-independent and many of the traditional scoring and assessment tools are sequence-specific, the sequence for each scaffold had to be designed before further quantitative analysis could be conducted. To design the sequences for each scaffold, two sequence design methods were used, namely EvoEF2 (41) and RosettaFixBB (42), where the backbone structures of the designed scaffolds were kept fixed during the sequence design to ensure a fair comparison of the scaffolds that were directly output by FoldDesign and Rosetta. Here, RosettaFixBB and EvoEF2 are sequence design methods that perform Monte Carlo sampling in sequence space guided by combined physics- and knowledge-based energy functions. A total of 100 sequences were designed for each scaffold, and the average results from the 10 lowest energy sequences were reported for both FoldDesign and Rosetta in the following analyses. First, Fig. 3A shows that the percent of buried residues for the FoldDesign scaffolds closely resembled the native protein structures from which the input SSs were extracted. For example, in the native structures, 19.2% of the residues were buried in the hydrophobic core, compared to 20.2% and 17.2% for the FoldDesign scaffolds whose sequences were designed by EvoEF2 and RosettaFixBB, respectively. However, for Rosetta, only 9.8% and 7.5% of the residues were buried in the hydrophobic core. Additionally, the solvent accessible surface area (SASA) for the native proteins was 7081.8 Å2 compared to 6964.9 Å2 and 7376.3 Å2 for the FoldDesign scaffolds whose sequences were designed by EvoEF2 and RosettaFixBB, while the average SASA for the corresponding Rosetta scaffolds was 8721.2 Å2 and 8944.2 Å2, respectively. These results suggest that the FoldDesign scaffolds possessed more compact hydrophobic cores and less solvent-exposed area than the Rosetta scaffolds and shared a higher similarity to the native structures for these characteristics. The difference is in part due to the fact that FoldDesign includes a number of energy terms that promote the formation of well-packed SS elements; these include specific fragment-derived distance and solvation potentials, generic backbone atom distance energy terms, and SS-specific fragment packing terms (SI Appendix, Text S2). In addition, the energy weights were carefully optimized using the results of the design simulations to ensure the formation of well-folded globular proteins (see Methods). Fig. 3. Comparison of the physical characteristics and energies for the designed folds by FoldDesign and Rosetta on the 354 test proteins, where the sequence for each scaffold was designed by EvoEF2 and RosettaFixBB, respectively. The native designation represents the proteins from which the SSs of the designed folds were derived. (A) Percent of buried residues is plotted for each protein, where a buried residue was defined as having a relevant SASA <5%. (B) SASA for each protein. (C and D) Energies for each protein calculated by GOAP and ROTAS. In Fig. 3 C and D, we further display the energies of the designed scaffolds by FoldDesign and Rosetta, as assessed by two leading third-party atomic-level statistical energy functions, GOAP (43) and ROTAS (44). For the sequences designed by EvoEF2 and RosettaFixBB, the FoldDesign scaffolds had average GOAP energies of −9736.9 and −10166.7, which were significantly lower than the GOAP energies of −8174.5 and −8838.8 for the Rosetta scaffolds with P values of 3.4E-13 and 4.3E-10, respectively. Similar trends were observed for ROTAS. For the sequences designed by EvoEF2 and RosettaFixBB, the FoldDesign scaffolds had average ROTAS energies of −6110.3 and −4446.5 compared to −4360.8 and −3281.5 for the corresponding Rosetta designs; the differences were statistically significant with P values of 6.8E-27 and 1.3E-15. Overall, the FoldDesign scaffolds possessed more tightly packed hydrophobic cores and were energetically more favorable than the Rosetta scaffolds, with GOAP energies that were 19.1% and 15.0% lower than the Rosetta scaffolds and ROTAS energies that were 40.1% and 35.5% lower than the Rosetta scaffolds depending on the sequence design method that was used. Importantly, neither FoldDesign nor Rosetta used any of the third-party energy functions for optimization. It is noted that introduction of ABEGO bias (45) during the Rosetta fragment selection protocol and enabling sub-rotamer sampling during the RosettaFixBB sequence design did not alter the above conclusions (SI Appendix, Text S4 and Figs. S4 and S5). Furthermore, despite the fact that Valine was used as the generic center of mass in FoldDesign and Rosetta (see Methods), neither method demonstrated a bias toward scaffolds that favored Valine as described in SI Appendix, Text S5 and Fig. S6, and all allowable regions of the Ramachandran plot were well represented in the FoldDesign scaffolds (SI Appendix, Fig. S7). The FoldDesign Force Field Plays an Important Role in Promoting the Structural Design Performance. As shown in Eq. 1 in the Methods section, FoldDesign utilizes a number of newly introduced energy terms, including fragment-derived distance and solvation potentials (Efrag_dist_profile and Efrag_solv) and detailed SS-specific packing potentials (Ehhpack, Esspack, and Ehspack), as well as generic atomic contact- and distance-based terms that promote the formation of compact, globular structures (Egeneric_dist and Econtact_num). Moreover, these terms were optimally combined with other more routine energy terms using an extensive weight optimization protocol based on the 107 training proteins (see Methods). To examine the impact of the FoldDesign force field and to probe the reason for the performance difference from the control method, we present in SI Appendix, Fig. S8 the comparative results for the physical characteristics of the Rosetta-designed scaffolds when the final models were selected using either the Rosetta or FoldDesign energy functions. It is noted that for this test we had to disable the fragment-derived distance and solvation potentials for FoldDesign as these are specific to the fragments generated by the FoldDesign program, which were not used to assemble the Rosetta designs given the differences in the fragment databases and identification protocols for the two methods. The data showed that selecting the Rosetta decoys according to their FoldDesign energies led to a significant improvement in the compactness of the folds as well as the GOAP and ROTAS energies compared to the designs selected using their original Rosetta energies. For example, the selection using the FoldDesign energy function increased the percent of buried residues by 31.5% for the EvoEF2 sequence designs and 39.3% for the RosettaFixBB sequence designs, compared to selection by the Rosetta centroid energy function, where the differences were statistically significant with P-values of 1.6E-13 and 1.5E-14, respectively. Similarly, improvements were observed in the third-party energies of the designed scaffolds. For example, the average GOAP energy improved by 9.2% and 7.6% for the EvoEF2 and RosettaFixBB sequence designs, respectively, where the differences were significant with P values of 3.1E-04 and 1.5E-03. In SI Appendix, Fig. S9, we present a similar comparative result for the FoldDesign scaffolds when the final designs were selected by either the Rosetta or FoldDesign energy functions. For this test, an opposite trend was observed, where the selection of the FoldDesign scaffolds using the alternative force field from Rosetta resulted in a reduced performance compared to the original FoldDesign force field. For instance, the Rosetta energy-based selection led to a 43.2% and 49.4% decrease in the percent of buried residues for the EvoEF2 and RosettaFixBB sequence designs, compared to the models selected using the original FoldDesign energy function; these differences were statistically significant with P values of 8.2E-79 and 5.8E-86, respectively. Furthermore, the GOAP energies were 26.7% and 25.2% worse for the EvoEF2 and RosettaFixBB sequence designs with P values of 5.8E-35 and 9.8E-34, respectively. Based on the data shown in the above section, apart from the extensive REMC searching simulations, the optimized force field of FoldDesign, with newly introduced energy features, plays another critical role in creating compact and physically sound structure designs that outperform those from other state-of-the-art design methods. FoldDesign Generates Stable Structures with Novel Folds. To further assess the stability of the designed structures, molecular dynamics (MD) simulations were run starting from the designed scaffolds produced by FoldDesign and Rosetta. MD is a useful tool as it allows for the study of protein motion and stability beyond static measurements such as energy calculations, where 20 ns unconstrained MD simulations were carried out using GROMACS (46) with the CHARMM36 force field (see Methods). Following the simulations, the final MD structures were obtained by clustering the 1,000 trajectories from the last nanosecond of each simulation using the GROMOS method with an RMSD cutoff of 2 Å, where the representative structure for each design was taken from the largest cluster center. To determine the stability of the structures, the TM-scores (39) between the initially designed scaffolds and the final clustered MD structures were calculated, where the results are depicted in Fig. 4 A and B for the structures whose sequences were designed by EvoEF2 and RosettaFixBB, respectively. Fig. 4. Analysis of the FoldDesign and Rosetta scaffolds using MD (A and B) and protein structure prediction by AlphaFold2 (C and D). (A and B) TM-scores of the FoldDesign and Rosetta scaffolds relative to their final structures following 20 ns MD simulations, where the sequence for each scaffold was designed by EvoEF2 (A) and RosettaFixBB (B). (C and D) TM-scores of the FoldDesign and Rosetta scaffolds relative to the structures predicted by AlphaFold2 starting from the EvoEF2 (C) and RosettaFixBB (D) sequences designed for each scaffold. (E) TM-score distribution between the FoldDesign structures and their closest native analogs obtained by searching the designed scaffolds through the PDB using TM-align. From the figures, it can be seen that the TM-scores between the initial FoldDesign scaffolds and the final MD structures were higher than those for the Rosetta scaffolds, indicating a closer match and thus more stable conformations for the FoldDesign scaffolds against MD-based perturbations. For instance, the average TM-score between the FoldDesign scaffolds and final MD structures for the EvoEF2 sequence designs was 0.645 compared to 0.584 for the corresponding Rosetta scaffolds (Fig. 4A), where the difference was statistically significant with a P value of 7.4E-19. A similar trend was observed for the scaffolds whose sequences were designed by RosettaFixBB, where the average TM-score between the initial FoldDesign structures and the final MD structures was 0.602 compared to 0.525 for the Rosetta scaffolds with a P-value of 4.6E-26 (Fig. 4B). Furthermore, when considering a cutoff TM-score of 0.5, 93.7% and 87.9% of the FoldDesign scaffolds whose sequences were designed by EvoEF2 and RosettaFixBB, respectively, shared the same global folds as their final MD structures, compared to 77.1% and 54.8% of the corresponding Rosetta structures. Fig. 5A shows three examples selected from among the most stable FoldDesign scaffolds, where the TM-scores were all greater than 0.8 and the RMSDs were less than 2Å, indicating a close atomic match between the designed scaffolds and the final MD structures. Overall, the vast majority of the FoldDesign scaffolds possessed stable global folds, outperforming the state-of-the-art Rosetta protocol across the test set. Fig. 5. Examples of stable, well-folded FoldDesign scaffolds as assessed by MD (A) and AlphaFold2 (B), where the sequences for each scaffold were designed by EvoEF2. (A) The initial FoldDesign structures (yellow) superposed with the final MD structures (blue). (B) The FoldDesign scaffolds (yellow) superposed with the AlphaFold2 models (blue). Interestingly, despite the high fold stability with local structural features that were highly similar to the native proteins, a large portion of the FoldDesign scaffolds adopted novel folds that were different from what exists in the PDB. In Fig. 4E, we present a histogram distribution of the TM-scores between the FoldDesign scaffolds and the closest structures identified by TM-align (47) from the PDB, where the average TM-score of 0.551 was relatively low given the searching power of TM-align and the near completeness of the PDB (3, 47). Of the 354 designs, 79 had a TM-score below 0.5 to any structure in the PDB, indicating they possessed novel folds, while the remaining 275 designs had analogous structures in the PDB with the same global folds (TM-scores ≥ 0.5). Furthermore, 74 of the 79 novel structures whose sequences were designed by EvoEF2 had stable folds with TM-scores ≥0.5 to their final structures output by the MD simulations. Moreover, there was no obvious difference between the novel folds and other folds in terms of stability, as the TM-score distributions between the designs and the final MD structures were quite similar (SI Appendix, Fig. S10), where their average TM-scores were 0.647 and 0.645, respectively. These results demonstrate that FoldDesign is capable of producing compact and stable scaffolds, while allowing for the exploration of novel areas of protein fold space. Protein Structure Prediction Indicates FoldDesign Produces Well-Folded Structures. As additional proof of the foldability of the designed structures, we examined the structural similarity between the designed scaffolds and the predicted models generated by the state-of-the-art AlphaFold2 program (48) starting from the designed sequences for each scaffold. As protein structure prediction is essentially the inverse problem of protein design, it would stand to reason that well-formed structure designs should be able to be recapitulated starting from their corresponding designed sequences. However, given that AlphaFold2 is a deep learning-based modeling program, its performance largely depends on collecting meaningful MSAs (48), yet de novo designed proteins almost always lack natural sequence homologs. To illustrate this, in SI Appendix, Fig. S11 we plot the number of Blast hits that were detected from the nr sequence database (E-value < 1E-5) when starting from either a single designed sequence or from jumpstarting the Blast search using an alignment of all 100 designed sequences for each FoldDesign scaffold. As shown in SI Appendix, Fig. S11A, no Blast hits were detected when starting from a single EvoEF2 sequence design and jumpstarting the Blast search from the alignment of designed sequences only picked up 1 to 2 hits for 4 of the 354 designs. For the RosettaFixBB designs, neither the single-designed sequence searches nor the jumpstarted Blast searches yielded any detectable homologs (SI Appendix, Fig. S11B). In SI Appendix, Table S2, we also list the structure prediction results by AlphaFold2 for the 354 native protein structures starting from the MSAs generated by the DeepMSA program (49) compared to the results starting from the single designed sequences. As expected, AlphaFold2 created excellent models with an average TM-score of 0.913 when starting from the native MSAs; but starting from the single designed sequences by either EvoEF2 or RosettaFixBB produced significantly less accurate models, where the average TM-scores were only 0.506 and 0.482 for the EvoEF2 and RosettaFixBB sequence designs, respectively, and nearly (or more than) half of the cases had TM-scores below 0.5. This result is in line with previous studies that have indicated that single sequence-based modeling using deep learning approaches for non-ideal folds is significantly less accurate than that for idealized de novo designed folds (50). This is likely due to the fact that most of the computationally designed structures have relatively simple global folds with optimized SS compositions that lack the irregularities that exist in native proteins (33, 35, 36). Since the 354 SS topologies in the benchmark dataset were derived from native protein structures, which contain numerous irregularities, the above results indicate that single sequence-based AlphaFold2 modeling may not be reliable for the FoldDesign and Rosetta scaffolds. Interestingly, when starting from artificial MSAs collected from the 100 designed sequences for the native structures, AlphaFold2 could generate reasonable folding results, where more than 97% of the cases had TM-scores >0.5, which was close to the modeling results obtained when starting from the DeepMSA MSAs (SI Appendix, Table S2). This demonstrates that the MSAs collected from sequence design simulations contain some level of evolutionary information that can facilitate deep learning-based structure prediction. Thus, given the lack of natural sequence homologs and the difficulty of AlphaFold2 to model complicated folds from single sequence designs, we constructed the input MSAs for AlphaFold2 by taking the 100 sequences designed by EvoEF2 and RosettaFixBB for each of the FoldDesign/Rosetta scaffolds. As shown in SI Appendix, Table S3, when starting from the sequences designed by EvoEF2, the average TM-score between the AlphaFold2 models and the FoldDesign scaffolds was 0.714 compared to 0.663 for the Rosetta scaffolds, where the difference was statistically significant with a P-value of 4.6E-09. In Fig. 4C, we present a head-to-head TM-score comparison, where the FoldDesign scaffolds had higher TM-scores than the corresponding Rosetta scaffolds for 211 cases, while Rosetta did so for 133 of the 354 cases. If we consider the number of designs with TM-score ≥0.5, 324 (or 91.5%) of the FoldDesign scaffolds shared the same global folds as the AlphaFold2 models compared to 301 (or 85.0%) of the scaffolds by Rosetta. These results demonstrate that the FoldDesign scaffolds more closely resembled the AlphaFold2 models than the Rosetta scaffolds did, indicating their enhanced stability/foldability. Similar patterns were observed for the sequences designed by RosettaFixBB, where the average TM-score between the FoldDesign scaffolds and AlphaFold2 models was 0.696 compared to 0.670 for Rosetta with a P-value of 3.0E-04 (SI Appendix, Table S3). Moreover, 208 of the 354 FoldDesign scaffolds had higher TM-scores than the Rosetta scaffolds and 315 (or 89.0%) of the designs had TM-scores ≥0.5 (Fig. 4D). Fig. 5B presents three examples from some of the closest matches between the FoldDesign scaffolds and AlphaFold2 models, where each had a TM-score greater than or close to 0.9 and RMSDs below 2.25 Å, indicating close atomic matches between the designed scaffolds and predicted models. Notably, these cases came from designs with some level of analogous structural information in the PDB, although the TM-scores between the designed scaffolds and their closest native analogs (0.517 to 0.611, see SI Appendix, Fig. S12) were much lower than those between the designed scaffolds and the AlphaFold2 predicted models (0.889 to 0.909, Fig. 5B). To further examine the foldability of the novel structures produced by FoldDesign, SI Appendix, Fig. S13 plots the AlphaFold2 TM-score distributions for the FoldDesign scaffolds that lacked or possessed native analogs, where the novel designs (with TM-score = 0.723/0.718 for the EvoEF2/RosettaFixBB sequences) were found to be as foldable or even more so than those with native analogs (with TM-score = 0.711/0.689 for the EvoEF2/RosettaFixBB sequences). Overall, these tests demonstrated that the FoldDesign scaffolds more closely matched the predicted models than the Rosetta scaffolds did, and the overwhelming majority of the designs shared the same global folds as the AlphaFold2 models. This structural consistency may suggest that FoldDesign captures some structural characteristics that have been integrated in the AlphaFold2 learning process. Assembling Uncommon Smotifs Is Essential to Produce Novel Fold Designs. Given the high population of novel folds produced by FoldDesign starting from native SS compositions, it was of interest to quantitatively examine the structural characteristics of these folds and determine how they deviate from native protein structures. Toward this goal, we first examined their local structural quality using MolProbity (MP) (51), where the results are summarized in SI Appendix, Table S4. It was observed that the novel designs possessed favorable MP-scores, with an average MP-score of 1.66 compared to 1.57 for the designs that had identifiable native analogs, where both scores were comparable to (or only slightly higher than) those of the corresponding native structures (1.19). Meanwhile, the novel folds had very few Ramachandran outliers, atomic clashes, or deviations in bond lengths and angles, largely comparable to (or slightly better than) the native and analogous designs. This result provides support that the novel folds possessed favorable local geometries and physical realism that resembled native proteins, although they had completely different global folds. To further probe the source of the distinct structural folds adopted by the novel designs, following the idea of previous studies (52–54), we investigated the local geometries of the associated super-SS elements by decomposing the global folds into their local Smotifs. Briefly, a Smotif is composed of two adjoining regular SS elements, either helices or strands, that are linked by a loop region (52). As shown in SI Appendix, Fig. S14, the geometry of a Smotif is specified by four spatial characteristics, including the distance (D) between the bracing SS elements and the three angles formed between them (hoist δ, packing θ, and meridian ρ). The overall fold of a protein can then be broken down into the basic SS building blocks, where a total of 540 Smotif types can be obtained by splitting the four-dimensional (D-δ-θ-ρ) space into 4-3-3-6 intervals and only ~320 to 330 Smotif geometries can be used to describe all existing protein structures (53). In Fig. 6, we present the relative frequency of Smotifs in the 79 novel folds and 354 native proteins in the test set versus the normalized background frequency of the Smotifs calculated from 51,094 non-redundant full-chain structures in the I-TASSER template library (55, 56), where the relative frequency values were normalized for each protein across the four background frequency bins in the plot (see SI Appendix, Eq. S16 in SI Appendix, Text S10). Fig. 6. Relative frequency of Smotifs found in the 354 native protein structures and 79 novel folds produced by FoldDesign vs. the normalized background frequency of the Smotifs calculated from the 51,094 non-redundant full-chain structures in the I-TASSER template library (SI Appendix, Text S10). Two motifs are considered as identical if they fall into the same bin in the four-dimensional (D-δ-θ-ρ) space (53). The mean values of the distributions are shown by the white circles, where a point with 0-frequency indicates that a Smotif with the indicated background frequency did not appear in one of the tested structural folds. It can be observed from Fig. 6 that compared to the native proteins, the novel designs by FoldDesign were highly enriched for rare or uncommon Smotifs, where 24.5% and 70.8% of the Smotifs in the novel designs had normalized background frequencies in the range [0, 1E-3] and (1E-3, 1E-2], respectively, compared to just 4.5% and 29.7% for the 354 native proteins. Additionally, 50.6% of the Smotifs from the native folds were common with background frequencies >1E-1, while just 4.1% of the Smotifs from the novel designed folds were commonly found. Of note, the vast majority of the Smotifs in the novel designs were found in nature, with the exception of one geometry that did not appear in the proteins from the PDB as shown in SI Appendix, Fig. S15. Thus, the novelty of the designed folds by FoldDesign may largely be a consequence of the combination of rare/uncommon local super-SS geometries, rather than the creation of new local geometries or a unique arrangement of common Smotifs. Furthermore, given the computationally assessed stability of the novel folds, these results support the claim that FoldDesign is able to produce stable designs for non-idealized SS elements, as the majority of the super-SS geometries were rarely observed in nature. Fig. 7 highlights two design cases with novel folds whose SS compositions were taken from the PDB proteins 1id0A and 2p19A, where the designed scaffolds are shown superposed with their AlphaFold2 models and closest native analogs from the PDB. It can be observed that the AlphaFold2 models closely resembled the designed scaffolds with TM-scores of 0.809 and 0.811 for the 1id0A and 2p19A designs, respectively, indicating they were foldable by the deep learning program. Interestingly, the clusters that these designs were selected from were highly conserved with average TM-scores of 0.769/0.826 between the cluster members and 1id0A/2p19A, pointing to a clear evolutionary relationship between the SS topologies and the native folds. Despite this, FoldDesign generated novel scaffolds for these two topologies, which had low TM-scores (0.467 and 0.451) to their closest structures in the PDB, again demonstrating an ability to explore structure space unexplored by nature even for highly conserved clusters. Fig. 7. Case study of two novel designed folds for the SS topologies taken from 1id0A (A) and 2p19A (B). The designed structures are shown on the left-hand side of the figure in yellow superposed with their AlphaFold2 models and closest native analogs in blue. Additionally, each native structure in the same SS cluster as 1id0A (A) and 2p19A (B) are shown aligned with their respective cluster centers, where the average TM-scores were calculated based on the alignment of each structure in the cluster to the cluster center. Lastly, the right-hand side of the figure illustrates the Smotif geometries found in the novel folds, where the depicted frequencies for each Smotif represent the relative background frequencies calculated from the representative structures in the PDB. In the right column of Fig. 7, we illustrate the Smotifs that the two designs were composed of, where the Smotifs for the two native structures are shown in SI Appendix, Fig. S16. For the 1id0A topology design, the global structure was composed of eight Smotifs, where all eight were rare with a background frequency ≤1E-3, while the corresponding native structure was composed of 7 common Smotifs with a high background frequency of ~3E-1 and 1 Smotif that was less common with a background frequency of ~1E-2. Similar trends were observed for 2p19A, where the designed structure was composed of eight uncommon Smotifs with a background frequency ≤1E-2, while the native structure was composed of eight common Smotifs with a background frequency of ~3E-1. Thus, from these cases, it can be seen that the combination of rare or uncommon local super-SS geometries gave rise to new global folds, which was observed across the 79 novel designs. Concluding Remarks Protein design generally consists of two steps of structural fold design and sequence design. Many protein design efforts have focused on the second step of sequence design with input scaffolds taken from existing protein structures in the PDB. Despite the success, such experiments constrain design cases to the limited number of folds adopted by natural proteins, while curtailing the exploration of novel areas of protein structure and biological function. In this work, we developed a pipeline, FoldDesign, for de novo protein fold design. Different from traditional protein folding simulations which start from native sequences and therefore, as expected, often result in folds that are similar to what exists in the PDB library, FoldDesign starts from structural restraints (e.g., SS assignments and/or inter-residue distance restraints) and performs folding simulations under the guidance of an optimized sequence-independent energy function. Large-scale tests on a set of 354 unique, non-ideal fold topologies demonstrated that FoldDesign could create protein-like folds with a closer Q3 score similarity to the desired structural restraints than the state-of-the-art design program, Rosetta. Meanwhile, the FoldDesign scaffolds had well-compacted core structures with buried residue rates and solvent-exposed areas that more closely matched those of native proteins, while MD simulations showed that the folds were more stable than those produced by Rosetta. Importantly, FoldDesign is capable of designing folds that are completely different from the native structures in the PDB, highlighting its ability to explore novel areas of protein structure space despite the high fidelity to the input restraints and the native-like local structural characteristics. Detailed data analyses showed that the major contributions to the success of fold design lie in the optimal energy force field, which contains a balanced set of energy terms that account for fragment and SS packing, as well as the efficient exploration of conformational space through REMC simulations assisted with a composite set of efficient movements. It was also found that the ability to identify and assemble less common super-SS geometries from the PDB, rather than creating new motifs or the unique arrangement of common SS motifs, represents the key for FoldDesign to create novel fold designs. Although the FoldDesign server outputs both the designed fold and the lowest energy designed sequences when combined with the EvoDesign/EvoEF2 programs (5, 41), the validation of the designed sequences remains to be experimentally examined. However, complete experimental validation requires both designed structures and designed sequences, where the latter is out of scope of the present study, and we leave this important work to future investigation. Nevertheless, the findings presented here have shown that FoldDesign can be used as a robust tool for generating high-quality, stable structural folds when applied to the very challenging task of completely de novo scaffold generation without human-expert intervention. This therefore provides a strong potential for the experimental protein design to effectively explore both structural and functional spaces which natural proteins have not reached despite billions of years of evolution. Methods FoldDesign aims to automatically design desired protein structure folds starting from user-specified rules such as SS composition and/or inter-residue contact and distance maps. The pipeline consists of three main steps, including fragment generation, REMC folding simulations, and main chain refinement and fold selection (Fig. 1). Fragment Generation. Starting from a user-specified SS, high-scoring fragments are identified from a fragment library, which consists of structural fragments collected from a non-redundant set of 29,156 high-resolution PDB structures used by QUARK (19, 57). The fragments were collected from structures deposited on or before 4/3/2014 and shared <30% sequence identity to each other (19, 57). Notably, this library has been extensively validated in the related field of protein structure prediction in the most recent CASP experiments (58, 59). Gapless threading through the library is performed to generate 1 to 20 residue fragments, where the fragments are scored based on the compatibility of their torsion angles and SS similarity to the desired SS at each position. The top 200 fragments are generated for each overlapping 1 to 20 residue window. The information for each fragment includes the backbone bond lengths, bond angles, and torsion angles, as well as other useful data such as the position-specific solvent accessibility and Cα coordinates, which are later used to derive distance and solvation restraints. REMC Folding Simulations and Refinement. Following fragment generation, REMC folding simulations are performed in order to assemble full-length structural models, where each simulation uses 40 replicas and runs 500 REMC cycles (see SI Appendix, Text S1 for a full description of the REMC parameters and movements). The protein conformation in FoldDesign is represented with a coarse-grained model, which specifies the backbone N, Cα, C, H, and O atoms as well as the Cβ atoms and an atom that represents the side-chain center of mass (SI Appendix, Fig. S17). To allow for a less biased exploration of structure space, the energy terms used by FoldDesign are sequence-independent, where the side-chain center of mass for Valine is used as the generic center of mass for each residue to minimize steric clashes. The initial conformations are produced by randomly assembling different high-scoring 9 residue fragments and then minimized using a set of 11 movements. Here, the major conformational movement is fragment substitution, which involves swapping a selected region of a decoy structure with the structure from one of the fragments randomly selected from the fragment library. Next, cyclical coordinate descent loop closure (60) is used to minimize the structural perturbations downstream. Since FoldDesign uses 1 to 20 residues fragments, larger fragment insertions are typically attempted during the initial REMC cycles, while smaller ones are attempted during the later steps of the simulations to improve its acceptance rate when the protein is more globular and well-folded. In addition to fragment insertion, 10 other conformational movements are attempted throughout the course of the simulations, including perturbing the backbone bond lengths, angles or torsion angles, segment rotations, segment shifts, and movements that form specific interactions between different SS elements, where these are described in SI Appendix, Text S1 and Fig. S2. The movements are accepted or rejected using the Metropolis criterion (61), where the energy for each conformation is assessed by the following energy function:[1] EFoldDesign=EHB+Ess_satisfaction+Erama+Ehhpack+Esspack+Ehspack+Eev+Egeneric_dist+Efrag_dist_profile+Efrag_solv+Erg+Econtact_num. Here, EHB, Ess_satisfaction, Erama, Ehhpack, Esspack, Ehspack, Eev, Egeneric_dist, Efrag_dist_profile, Efrag_solv, Erg, and Econtact_num are terms that account for backbone hydrogen bonding, the satisfaction rate of the input SS, Ramachandran torsion angles, helix-helix packing, strand-strand packing, helix-strand packing, excluded volume, generic backbone atom distances, fragment-derived distance restraints, fragment-derived solvent accessibility, radius of gyration, and expected contact number, respectively. A more detailed explanation of these terms is given in SI Appendix, Text S2. After the REMC simulations are completed, the design with the lowest energy is selected for further atomic-level refinement, for which sequence design and structural refinement are performed iteratively using EvoDesign (5) and ModRefiner (62), respectively. Training and Test Dataset Collection. To test FoldDesign’s ability to perform de novo protein fold design, we collected a non-redundant set of SS sequences. This was accomplished by extracting the three-state SSs from 76,166 protein domains in the I-TASSER template library (55, 56) using DSSP (38). All of the pairwise SS alignments were obtained using Needleman–Wunsch dynamic programming to align the three-state SS sequences. The target sequences were then clustered based on the distance matrix defined by their SS identities, i.e., the number of identical SSs divided by the total alignment length, where an identity cutoff =70% was used to define the clusters. The identified clusters were further refined by eliminating atypical SS topologies (clusters with less than 10 members) and by selecting only those clusters where a clear relationship existed between the SS and the tertiary structure adopted by the cluster members. The latter requirement was accomplished by using TM-align (47) to perform structural alignment between each cluster member and the cluster center, where conserved clusters were required to have an average TM-score ≥0.5 between the members and cluster center. Finally, we obtained 461 clusters; 107 and 354 SS sequences were used for the training and test sets, respectively. The training set was composed of 22 α, 25 β, and 60 α/β topologies, while the test set was composed of 24 α, 55 β, and 275 α/β topologies. FoldDesign Energy Function Optimization. In order to ensure proper structure generation, each energy term must be carefully weighted in the FoldDesign energy function. This was done on the 107 training topologies. Briefly, a grid searching strategy was used to optimize the weights, where all weights were initially assigned as 0, except for the weight for the steric clash term, which was set to 1.0. Then the values for each weight were adjusted one-at-a-time around the grid values and the FoldDesign simulations were run to produce scaffold structures using the new weight set. After structure generation, the sequences for each scaffold were designed using EvoEF2 (41) and the designed structures were assessed based on:[2] Eaccept=-ΔEvoEF2+100∗ΔBuriedResidues+100∗ΔQ3Score. where, ΔEvoEF2, ΔBuriedResidues, and ΔQ3Score are the changes in the average EvoEF2 energy, percent of buried residues, and SS Q3 score, respectively, between the structures produced by the old and new weight sets. If the new weighting parameter increased the value of Eaccept, the weights were accepted. Once the initial weights for each energy term were determined, many more iterations were conducted to precisely fine-tune their values based on Eq. 2 as well as by hand inspection of the structures. Although time-consuming, the process of directly optimizing the weights based on the results of the folding simulations resulted in high-quality scaffolds with physical characteristics that resembled native proteins. MD Simulation for Examining Fold Stability. To examine the stability of the FoldDesign scaffolds, we performed MD simulations starting from the designed structures. For each simulation, a dodecahedron box was constructed with a distance of 10 Å from the solute and filled with TIP3P water molecules, where Na+ and Cl− ions were used to neutralize the charge of the system. Following this, energy minimization was carried out using the steepest descent minimization with a maximum force of 10 kJ/mol. The system was then equilibrated at 300 K using 100 ps NVT simulations and 100 ps NPT simulations with position restraints (1,000 kJ/mol) on the heavy atoms of the protein. After the two equilibration phases, the system was well-equilibrated at the desired temperature and pressure, and unconstrained MD simulations were performed at 300 K for 20 ns. During the simulations, non-bonded interactions were truncated at 12 Å and the Particle Mesh Ewald methods was used for long-range electrostatic interactions. Lastly, the velocity-rescaling thermostat and Parrinello–Rahman barostat were used to couple the temperature and pressure, respectively. A total of 1,000 structures were collected from the MD trajectories during the final nanosecond of the simulations. This ensemble was then clustered using the GROMOS method with an RMSD cutoff of 2 Å, and the final MD structure for each simulation was collected from the cluster center. Supplementary Material Appendix 01 (PDF) Click here for additional data file. This work used the Extreme Science and Engineering Discovery Environment, which is supported by the NSF (ACI-1548562). This work was supported in part by the NIGMS (GM136422, S10OD026825), the NIAID (AI134678), the NSF (IIS1901191, DBI2030790, MTM2025426), the NCI (U24CA210967), and the NIEHS (P30ES017885). Author contributions Y.Z. designed research; R.P. performed research; X.H. contributed new reagents/analytic tools; R.P. analyzed data; and R.P., X.H., G.S.O., and Y.Z. wrote the paper. Competing interest The authors declare no competing interest. Data, Materials, and Software Availability All study data are included in the article and/or SI Appendix. The online server, stand-alone program and benchmark data for FoldDesign are available at https://zhanggroup.org/FoldDesign/, while the stand-alone program may also be downloaded from https://github.com/robpearc/FoldDesign. Supporting Information This article is a PNAS Direct Submission. ==== Refs 1 C. R. Darwin, The Origin of Species (John Murry, Landon, 1859). 2 N. K. Fox, S. E. Brenner, J. M. Chandonia, SCOPe: Structural Classification of proteins–extended, integrating SCOP and ASTRAL data and classification of new structures. Nucleic Acids Res. 42 , D304–D309 (2014).24304899 3 Y. Zhang, I. A. Hubner, A. K. Arakaki, E. Shakhnovich, J. Skolnick, On the origin and highly likely completeness of single-domain protein structures. Proc. Natl. Acad. Sci. U.S.A. 103 , 2605–2610 (2006).16478803 4 Y. Zhang, J. Skolnick, The protein structure prediction problem could be solved using the current PDB library. Proc. Natl. Acad. Sci. U.S.A. 102 , 1029–1034 (2005).15653774 5 R. Pearce, X. Huang, D. Setiawan, Y. Zhang, EvoDesign: Designing protein-protein binding interactions using evolutionary interface profiles in conjunction with an optimized physical energy function. J. Mol. Biol. 431 , 2467–2476 (2019).30851277 6 B. I. Dahiyat, C. A. Sarisky, S. L. Mayo, De novo protein design: Towards fully automated sequence selection. J. Mol. Biol. 273 , 789–796 (1997).9367772 7 G. A. Lazar, J. R. Desjarlais, T. M. Handel, De novo design of the hydrophobic core of ubiquitin. Protein Sci. 6 , 1167–1178 (1997).9194177 8 G. Dantas, B. Kuhlman, D. Callender, M. Wong, D. Baker, A large scale test of computational protein design: Folding and stability of nine completely redesigned globular proteins. J. Mol. Biol. 332 , 449–460 (2003).12948494 9 L. Cao , De novo design of picomolar SARS-CoV-2 miniprotein inhibitors. Science 370 , 426–431 (2020).32907861 10 D.-A. Silva , De novo design of potent and selective mimics of IL-2 and IL-15. Nature 565 , 186–191 (2019).30626941 11 A. Chevalier , Massively parallel de novo protein design for targeted therapeutics. Nature 550 , 74–79 (2017).28953867 12 J. Dou , De novo design of a fluorescence-activating β-barrel. Nature 561 , 485–491 (2018).30209393 13 C. E. Tinberg , Computational design of ligand-binding proteins with high affinity and selectivity. Nature 501 , 212–216 (2013).24005320 14 M. J. Lajoie , Designed protein logic to target cells with precise combinations of surface antigens. Science 369 , 1637–1643 (2020).32820060 15 B. Kuhlman , Design of a novel globular protein fold with atomic-level accuracy. Science 302 , 1364–1368 (2003).14631033 16 E. Verschueren , Protein design with fragment databases. Curr. Opin. Struct. Biol. 21 , 452–459 (2011).21684149 17 K. T. Simons, C. Kooperberg, E. Huang, D. Baker, Assembly of protein tertiary structures from fragments with similar local sequences using simulated annealing and Bayesian scoring functions. J. Mol. Biol. 268 , 209–225 (1997).9149153 18 J. U. Bowie, D. Eisenberg, An evolutionary approach to folding small alpha-helical proteins that uses sequence information and an empirical guiding fitness function. Proc. Natl. Acad. Sci. U.S.A. 91 , 4436–4440 (1994).8183927 19 D. Xu, Y. Zhang, Ab initio protein structure assembly using continuous structure fragments and optimized knowledge-based force field. Proteins 80 , 1715–1735 (2012).22411565 20 R. Pearce, Y. Zhang, Deep learning techniques have significantly impacted protein structure prediction and protein design. Curr. Opin. Struct. Biol. 68 , 194–207 (2021).33639355 21 P.-S. Huang , De novo design of a four-fold symmetric TIM-barrel protein with atomic-level accuracy. Nat. Chem. Biol. 12 , 29–34 (2016).26595462 22 A. Sali, T. L. Blundell, Comparative protein modelling by satisfaction of spatial restraints. J. Mol. Biol. 234 , 779–815 (1993).8254673 23 Y. Zhang, J. Skolnick, Automated structure prediction of weakly homologous proteins on a genomic scale. Proc. Natl. Acad. Sci. U.S.A. 101 , 7594–7599 (2004).15126668 24 H. Deng, Y. Jia, Y. Zhang, 3DRobot: Automated generation of diverse and well-packed protein structure decoys. Bioinformatics 32 , 378–387 (2016).26471454 25 I. Anishchenko , De novo protein design by deep network hallucination. Nature 600 , 547–552 (2021).34853475 26 J. Wang , Scaffolding protein functional sites using deep learning. Science 377 , 387–394 (2022).35862514 27 B. Huang , A backbone-centred energy function of neural networks for protein design. Nature 602 , 523–528 (2022).35140398 28 W. R. Taylor, A “periodic table” for protein structures. Nature 416 , 657–660 (2002).11948354 29 Z. Harteveld , A generic framework for hierarchical de novo protein design. bioRxiv [Preprint] (2022). 10.1101/2022.04.07.487481 (Accessed 22 October 2022). 30 T. M. Jacobs , Design of structurally distinct proteins using strategies inspired by evolution. Science 352 , 687–690 (2016).27151863 31 S. L. Guffy, F. D. Teets, M. I. Langlois, B. Kuhlman, Protocols for requirement-driven protein design in the rosetta modeling program. J. Chem. Inf. Model 58 , 895–901 (2018).29659276 32 N. Koga , Principles for designing ideal protein structures. Nature 491 , 222–227 (2012).23135467 33 X. Pan, T. Kortemme, Recent advances in de novo protein design: Principles, methods, and applications. J. Biol. Chem. 296 , 100558 (2021).33744284 34 W. Zhou, T. Smidlehner, R. Jerala, Synthetic biology principles for the design of protein with novel structures and functions. FEBS Lett. 594 , 2199–2212 (2020).32324903 35 P. S. Huang, S. E. Boyken, D. Baker, The coming of age of de novo protein design. Nature 537 , 320–327 (2016).27629638 36 D. Baker, What has de novo protein design taught us about protein folding and biophysics? Protein Sci. 28 , 678–683 (2019).30746840 37 C. B. Anfinsen, Principles that govern the folding of protein chains. Science 181 , 223–230 (1973).4124164 38 W. Kabsch, C. Sander, Dictionary of protein secondary structure: Pattern recognition of hydrogen-bonded and geometrical features. Biopolymers 22 , 2577–2637 (1983).6667333 39 Y. Zhang, J. Skolnick, Scoring function for automated assessment of protein structure template quality. Proteins 57 , 702–710 (2004).15476259 40 J. Xu, Y. Zhang, How significant is a protein structure similarity with TM-score = 0.5? Bioinformatics 26 , 889–895 (2010).20164152 41 X. Huang, R. Pearce, Y. Zhang, EvoEF2: Accurate and fast energy function for computational protein design. Bioinformatics 36 , 1135–1142 (2020).31588495 42 R. F. Alford , The rosetta all-atom energy function for macromolecular modeling and design. J. Chem. Theory Comput. 13 , 3031–3048 (2017).28430426 43 H. Zhou, J. Skolnick, GOAP: A generalized orientation-dependent, all-atom statistical potential for protein structure prediction. Biophys. J. 101 , 2043–2052 (2011).22004759 44 J. Park, K. Saitou, ROTAS: A rotamer-dependent, atomic statistical potential for assessment and prediction of protein structures. BMC Bioinformatics 15 , 307 (2014).25236673 45 Y. R. Lin , Control over overall shape and size in de novo designed proteins. Proc. Natl. Acad. Sci. U.S.A. 112 , E5478–E5485 (2015).26396255 46 M. J. Abraham , GROMACS: High performance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX 1–2 , 19–25 (2015). 47 Y. Zhang, J. Skolnick, TM-align: A protein structure alignment algorithm based on the TM-score. Nucleic Acids Res. 33 , 2302–2309 (2005).15849316 48 J. Jumper , Highly accurate protein structure prediction with AlphaFold. Nature 596 , 583–589 (2021).34265844 49 C. Zhang, W. Zheng, S. M. Mortuza, Y. Li, Y. Zhang, DeepMSA: Constructing deep multiple sequence alignment to improve contact prediction and fold-recognition for distant-homology proteins. Bioinformatics 36 , 2105–2112 (2020).31738385 50 M. Baek, D. Baker, Deep learning and protein structure modeling. Nat. Methods 19 , 13–14 (2022).35017724 51 V. B. Chen , MolProbity: All-atom structure validation for macromolecular crystallography. Acta Crystallogr. D Biol. Crystallogr. 66 , 12–21 (2010).20057044 52 N. Fernandez-Fuentes, B. Oliva, A. Fiser, A supersecondary structure library and search algorithm for modeling loops in protein structures. Nucleic Acids Res. 34 , 2085–2097 (2006).16617149 53 N. Fernandez-Fuentes, J. M. Dybas, A. Fiser, Structural characteristics of novel protein folds. PLoS Comput. Biol. 6 , e1000750 (2010).20421995 54 S. Wu, Y. Zhang, Recognizing protein substructure similarity using segmental threading. Structure 18 , 858–867 (2010).20637422 55 S. Wu, J. Skolnick, Y. Zhang, Ab initio modeling of small proteins by iterative TASSER simulations. BMC Biol. 5 , 17 (2007).17488521 56 J. Yang , The I-TASSER Suite: Protein structure and function prediction. Nat. Methods 12 , 7–18 (2015). 57 D. Xu, Y. Zhang, Toward optimal fragment generations for ab initio protein structure assembly. Proteins 81 , 229–239 (2013).22972754 58 W. Zheng , Deep-learning contact-map guided protein structure prediction in CASP13. Proteins 87 , 1149–1164 (2019).31365149 59 W. Zheng , Protein structure prediction using deep learning distance and hydrogen-bonding restraints in CASP14. Proteins 89 , 1734–1751 (2021).34331351 60 A. A. Canutescu, R. L. Dunbrack Jr., Cyclic coordinate descent: A robotics algorithm for protein loop closure. Protein Sci. 12 , 963–972 (2003).12717019 61 N. Metropolis, A. W. Rosenbluth, M. N. Rosenbluth, A. H. Teller, E. Teller, Equation of state calculations by fast computing machines. J. Chem. Phys. 21 , 1087–1092 (1953). 62 D. Xu, Y. Zhang, Improving the physical realism and structural accuracy of protein models by a two-step atomic-level energy minimization. Biophys. J. 101 , 2525–2534 (2011).22098752
PMC009xxxxxx/PMC9942887.txt
==== Front Proc Natl Acad Sci U S A Proc Natl Acad Sci U S A PNAS Proceedings of the National Academy of Sciences of the United States of America 0027-8424 1091-6490 National Academy of Sciences 36656857 202209329 10.1073/pnas.2209329120 research-articleResearch ArticleneuroNeuroscience424 Biological Sciences Neuroscience Arginine–vasopressin-expressing neurons in the murine suprachiasmatic nucleus exhibit a circadian rhythm in network coherence in vivo Stowie Adam a Qiao Zhimei a https://orcid.org/0000-0001-8674-2846 Buonfiglio Daniella Do Carmo a Beckner Delaney M. a https://orcid.org/0000-0001-9302-2249 Ehlen J. Christopher a https://orcid.org/0000-0003-3223-9262 Benveniste Morris a https://orcid.org/0000-0001-7070-1521 Davidson Alec J. [email protected] a 1 https://orcid.org/0000-0003-4205-1968 aNeuroscience Institute, Morehouse School of Medicine, Atlanta, GA 30310-1495 1To whom correspondence may be addressed. Email: [email protected]. Edited by Joseph Takahashi, The University of Texas Southwestern Medical Center, Dallas, TX; received June 1, 2022; accepted December 19, 2022 19 1 2023 24 1 2023 19 7 2023 120 4 e22093291201 6 2022 19 12 2022 Copyright © 2023 the Author(s). Published by PNAS. 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This article is distributed under Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND). Significance This work is the first to employ two novel in vivo recording techniques, miniaturized calcium microscopy and optogentically targeted single-unit activity recording, to examine the rhythmic behavior of AVP-expressing neurons both at the individual neuronal and network level. These results suggest that while AVP neurons are important for organismal rhythmicity, individual cellular rhythms are unstable and diverse. However, we observed correlated activity among these neurons which appears more reliably rhythmic, suggesting that emergent network properties of the SCN may be more relevant for organismal rhythmicity than individual neuronal characteristics. The suprachiasmatic nucleus (SCN) is composed of functionally distinct subpopulations of GABAergic neurons which form a neural network responsible for synchronizing most physiological and behavioral circadian rhythms in mammals. To date, little is known regarding which aspects of SCN rhythmicity are generated by individual SCN neurons, and which aspects result from neuronal interaction within a network. Here, we utilize in vivo miniaturized microscopy to measure fluorescent GCaMP-reported calcium dynamics in arginine vasopressin (AVP)-expressing neurons in the intact SCN of awake, behaving mice. We report that SCN AVP neurons exhibit periodic, slow calcium waves which we demonstrate, using in vivo electrical recordings, likely reflect burst firing. Further, we observe substantial heterogeneity of function in that AVP neurons exhibit unstable rhythms, and relatively weak rhythmicity at the population level. Network analysis reveals that correlated cellular behavior, or coherence, among neuron pairs also exhibited stochastic rhythms with about 33% of pairs rhythmic at any time. Unlike single-cell variables, coherence exhibited a strong rhythm at the population level with time of maximal coherence among AVP neuronal pairs at CT/ZT 6 and 9, coinciding with the timing of maximal neuronal activity for the SCN as a whole. These results demonstrate robust circadian variation in the coordination between stochastically rhythmic neurons and that interactions between AVP neurons in the SCN may be more influential than single-cell activity in the regulation of circadian rhythms. Furthermore, they demonstrate that cells in this circuit, like those in many other circuits, exhibit profound heterogenicity of function over time and space. in vivo calcium imaging SCN AVP HHS | NIH | National Institute of General Medical Sciences (NIGMS) 100000057 R35GM136661 J. Christopher EhlenAlec J Davidson HHS | NIH | National Institute of General Medical Sciences (NIGMS) 100000057 SC1GM127260 J. Christopher EhlenAlec J Davidson HHS | NIH | National Institute on Aging (NIA) 100000049 SC1AG046907 Morris Benveniste HHS | NIH | NHLBI | NHLBI Division of Intramural Research (DIR) 100017540 T32HL007901 Delaney M Beckner HHS | NIH | National Institute of Neurological Disorders and Stroke (NINDS) 100000065 R21NS108197 Alec J Davidson ==== Body pmcIn mammals, circadian rhythms are synchronized by the suprachiasmatic nucleus (SCN) of the ventral hypothalamus (1, 2). At the molecular level, circadian rhythms are driven by a transcription-translation feedback loop in which the core clock genes CLOCK and BMAL1 promote the transcription of the CRY and PER genes, which in turn inhibit their own transcription in a process that takes approximately 24 h (3). While most mammalian cells contain this molecular machinery (4), with a few notable exceptions (5, 6), rhythmicity is lost after only a few cycles when isolated from the SCN (7). SCN neurons, cultured in isolation or experimentally separated, exhibit weak or unstable rhythms (4, 8–10). Thus, although molecular clock machinery is necessary for the generation of circadian rhythms, individual neurons are insufficient to maintain longitudinal rhythms. There is increasing evidence that the neurons in the SCN act as a neural network in which interactions between neurons within the circuit may be responsible for regulation circadian rhythms. SCN brain slices exhibit robust rhythms in core clock gene expression and intracellular calcium dynamics ex vivo for weeks to months (7, 11), and these rhythms are reversibly dampened when cellular communication is disrupted with TTX (11–13). In addition, longitudinal bioluminescent slice recordings reveal the existence of an organized wave of clock gene expression which propagates through the SCN in a stereotypic fashion, for both mPer1 (14) and mPer2 (11, 15). This temporal organization is modified in a predictable and reversible way by alterations in day length, suggesting that this emergent property of the network encodes important information about the external world (16, 17), and that rhythmic function among this population is typified by heterogeneity. Most SCN neurons are GABAergic (18), but can be subdivided based on other expressed proteins. Distinct populations of vasoactive intestinal polypeptide (VIP)-expressing, gastrin releasing peptide (GRP)-expressing and arginine vasopressin (AVP)-expressing GABAergic neurons are somewhat divided anatomically (19). In addition, approximately 40% of SCN neurons express the neuropeptide Neuromedin S (NMS) and approximately 60% express the dopamine receptor DRD1 (20); however, these gene products are not expressed in distinct neuronal populations and these neurons may also express AVP or VIP (21). The activity of these neuronal subpopulations may correspond to different roles within the circadian circuit. For instance, loss of NMS neurons abolishes rhythmicity entirely (21). VIP- and GRP-expressing neurons likely receive input from the retina and play a role in entrainment of the SCN to the light/dark cycle (22, 23). Also, VIP and GABA act together to maintain synchrony between SCN neurons, particularly under changing environmental lighting (16). However, the role of AVP neurons is less clear. Disrupting the circadian clock by selectively knocking out either bmal1 or casein kinase 1 delta in AVP neurons lengthened the period of oscillation (24–26), although selective destruction of AVP neurons themselves does not affect LMA rhythms (27). Taken together, available data suggest that rhythmicity of the SCN network is rather resistant to the loss of specific neuronal populations of the SCN network. Multi-unit neuronal activity (MUA) recordings show that the intact SCN network exhibits a robust firing rhythm peaking during the day, even in constant darkness (28–31). Cell-type-specific contributions to circadian timing in the intact SCN is less understood. Measurement of VIP-neuron gene expression and intracellular calcium via fiber photometry (FP) show robust population level rhythms in calcium as well as Per1, Per2, and Cry1 expression in vivo (23, 32). This approach highlights the importance of VIP neurons for photic entrainment (23, 33), and the importance of GABA release from AVP neurons in regulating the timing of firing of other SCN neurons (34). However, such population measures do not provide the cellular resolution needed to elucidate how individual neurons interact in neural networks to exact function and behavior. Without the measure of individual neuron activity, we cannot know how diverse neuronal behavior may be within a network. Studies on neural networks have indicated that cohesive function of the circuit can occur with significant heterogeneity of function among individual neurons (35–38). New in vivo fluorescent approaches have been developed in which individual neurons of specific cell types within a network can be targeted and recorded within an intact circuit of an awake, behaving animal. Recording from brain slices which lack inputs and outputs, or recording population-level behavior, even in vivo (e.g., MUA or FP), could mask potential heterogeneity which may be important for circuit function. State changes in such circuits are a point at which heterogeneity of function can switch to more homogenous, coordinated function. One example is cortical reorganization during sleep, where diverse signaling in wake changes to more coordinated activity during sleep (39). The mammalian SCN undergoes reliable, repeating state changes each day making it an excellent model for determining how individual neurons contribute to a cohesive behavioral outcome. In slice recordings, significant heterogeneity in the phase of individual cells is observed (40) and this heterogeneity in vivo may influence behavior-level functional outcomes (41). Different cell types exhibit diverse characteristics (26, 42) and a single cell can exhibit stochastic rhythmicity (43), another type of heterogeneity observed within individual cells over time. However, whether such diversity of function across cells and time exists in vivo, in an unperturbed SCN circuit is unknown. In the present work, we characterize bursting electrical behavior of AVP neurons within the intact SCN network using optogenetically targeted single-unit activity recording (SUAAVP). Further, we employ miniaturized fluorescence microscopy to characterize a portion of the SCN network by conducting longitudinal recording of intracellular calcium dynamics in AVP-expressing neurons in the fully intact SCN of freely behaving mice. We report that individual AVP neurons appear to be relatively unstable circadian oscillators in vivo, exhibiting a stochastic pattern of rhythmicity in calcium dynamics over time. However, correlational network analysis reveals a strong, stable circadian pattern of coherence among AVP neuron pairs at the population level, peaking during the circadian daytime. This study is the first to undertake cell-type-specific in vivo recordings of individual SCN neurons, demonstrating that AVP neuronal rhythms emerge largely at the network level. Methods Animals. For in vivo calcium imaging experiments, 10-wk-old heterozygous male AVP-IRES2-Cre-D (Jackson Laboratory 023530) (44) was used. Cre-mediated recombination in these mice specifically targets SCN AVP+ neurons in several studies to date (45, 46). Control experiments were conducted using male Ai6 mice which express robust ZsGreen1 fluorescence following Cre-mediated recombination (Jackson Labs 007906) crossed with AVP-IRES2-CRE-D knock-in mice (Jackson Labs 023530). For in vivo optogenetically targeted single-unit recording, mice were bred by crossing AVP-Cre with Ai32 (RCL-ChR2(H134R)/EYFP,Stock # 024109) mice. In vivo electrophysiological recordings were performed on adult male mice (3 to 4 mo) housed in 12-h light/dark cycle from birth. Food and water were available ad libitum. All procedures described were approved by the IACUC committee at Morehouse School of Medicine. In vivo Miniaturized Fluorescent Calcium Imaging. Chronic lens implantation surgeries were done as previously described (47). Briefly, using a stereotaxic apparatus, AVP-cre mice were injected (ML: +0.750 mm, AP: −0.200 mm, DV: −5.8 mm) with an adeno-associated viral vector conveying the genetically encoded calcium sensor jGCaMP7s (#104491, Addgene) into the SCN and a gradient index lens 0.50 mm in diameter (1050-004611 Inscopix) was implanted above the SCN (ML: +0.750 mm, AP: −0.200 mm, DV: −5.7 mm; Fig. 1A). Minimitter G2 Emitters were implanted at the same time into the peritoneal cavity. Post-hoc histology verified lens placement near/above the SCN and away from other AVP-expressing cell populations in hypothalamic paraventricular nucleus, or supraoptic nucleus. This placement was targeted since GRIN lens focal distance is typically ~200 µm below the bottom surface. Fluorescence in ventral hypothalamus, but outside the SCN borders appeared noncellular in nature, while cell bodies are clearly observable within the SCN. Six weeks after surgery, mice were assessed for evidence of discrete, dynamic calcium signals and a baseplate was attached over the gradient index lens to establish a fixed working distance with the microscope. Baseplated mice (n=4 mice) were acclimated to the isolated recording chamber and the presence of the microscope for 2 d before recordings were initiated. The recording schedule consisted of 5-min recordings where mice were imaged at 6.67 Hz, every 3 h for 48 consecutive hours for a total of 16 timepoints under both constant darkness and 12:12LD. An example field-of-view and recording traces (5 min) from two different regions-of-interest (ROIs) and a background region containing no visible cell are shown in Fig. 1B. The two 48-h recordings did not result in photobleaching or any other rundown of GCaMP in the neurons. Specificity of GCaMP infection was verified by colocalization of GCaMP and AVP via IHC (Fig. 1C). Colocalization was difficult to detect, as expected due to the reduced expression of the AVP peptide in AVP-cre mice (48). GCaMP7s never colocalized with neurons expressing VIP, supporting AVP-expressing neuronal specificity. Locomotor activity (LMA) during a fixed, then a shifted LD cycle was measured in one mouse after imaging to verify that implanted mice had unimpaired circadian rhythms, including proper entrainment to phase shifts (Fig. 1D). Coincident recording of LMA was also used to verify rhythmicity during Ca2+ recordings (SI Appendix, Fig. S1). Proper functionality of the calcium sensor was established by comparing dynamic GCaMP7s recordings with green fluroescent protein (GFP)-expressing AVP neurons used as a control (Fig. 1E). The GFP signal never exhibiting dynamics like those observed using the calcium indicator. Fig. 1 F and G show three more example traces to highlight the features of the traces used for the analysis described in more detail below. Fig. 1. In Vivo Characterization of AVP Neurons in the SCN Network. (A) Left: Illustrations describing AAV infection of AVP neurons, GRIN implantation above the SCN, and tethered recordings using microendoscope. Right: example of histological validation of lens placement. Typical focal distance below the lens is ~200 μm. (B) Representative still image demonstrating how AVP calcium dynamics are observed in vivo. Green polygons are drawn around AVP neuronal ROI and the red polygons are non-AVP neuronal spaces used for background subtraction. Sample traces on the right illustrate examples of 5-min recordings obtained from AVP neurons in the same field of view. (C) Image of the SCN taken with a confocal microscope (63× lens) showing neuronal colocalization (yellow) of GCaMP7s (green) and AVP antibody (orange) as well as the exclusion of GCaMP7s from VIP expressing neurons (white). Red boxes indicate neurons with clear colocalization. (D) Representative actogram demonstrating that mice with a GRIN lens implanted above the SCN appear to have normal daily circadian behavior which shifts normally in response to a 6-h phase advance. (E) Representative calcium traces recorded from the SCN of AVPcre x GFP mice and AVPcre mice infected with GCaMP7s demonstrating that while GFP does not exhibit acute calcium dynamics, intracellular calcium dynamics are detected by GCaMP7s. F–H: Representative 5-min calcium traces demonstrating how mean fluorescence intensity (red dotted line, F), acute events (red stars, G), and calcium waves (onsets green; offsets red, H) were quantified for the purposes of determination of single-cell rhythmicity in Fig. 2. Fig. 2. Circadian Rhythmicity in individual AVP Neurons is stochastic. (A) Heatmaps illustrating mean fluorescence intensity across the 5-min recording (see Fig. 1F) of individual AVP neurons/ROIs (rows) by recording timepoint (columns) in both DD and LD. Rows are normalized to their means. ROIs which exhibited significant circadian rhythms according to cosinor (P < 0.05) for mean intensity appear above the dashed line and triangle. ROIs that were not rhythmic are plotted below the dashed line. Plotted beneath the heatmap is the population average and standard errors for each timepoint. When the population measure is rhythmic (P < 0.05), the cosine fit is superimposed on Top of the bar graph. (B) Polar plots illustrating the phase distribution of ROIs exhibiting significant circadian rhythms in mean fluorescence in both DD and LD. The direction of the bar represents hour in CT for DD or ZT for LD. The length of the bar represents the number of ROIs. Circular mean and SEM are represented in green. Pie charts indicate the proportion that exhibited a statistically significant rhythm across the 48 h of DD and LD. (C) Phenotype tracking plots indicating the stability/change of rhythmic state of individual AVP neuron ROIs between the two recording days (DD day 1 and DD day 2) under constant darkness (Left) and between the 48 h recordings in each lighting condition (Right). Blue lines indicate stable rhythmic or arrhythmic ROIs, while red lines indicate a loss of rhythmicity between conditions, and green lines a gain of rhythmicity. Black triangles on the axes indicate the division between rhythmic (Above) and arrhythmic (Below) ROIs. (D–F) Rhythmicity measures for unsupervised acute events as quantified as described in Fig. 1G. Same conventions as A–C, above, except that raw event number is plotted in the heatmaps rather than normalized counts. (E–I): Rhythmicity measures for calcium waves, as quantified as described in Fig. 1H. Same conventions as A–C, above. In Vivo Optogenetically Targeted Single-Unit Activity Recording (SUAAVP). Optrode implantation. Optrode implantation was done as previously described (49). Briefly, a stereotaxic apparatus (KOPF, model 1900) was used for implanting custom 12-channel optrodes above the SCN (AP: +0.38 mm, ML: +0.15 mm, DV: −5.5 mm). Optrodes were drivable via manual screw-drive, and once implanted were slowly advanced into the SCN over days/weeks (see below). Optrodes were constructed by combining a microelectrode bundle (insulated nichrome wire, 30 μm diameter, Stablohm 675, California Fine Wire, USA) with a custom-made optical fiber (200 μm diameter, Doric Lenses). The entire site was then sealed and secured with dental acrylic (Fig. 3A). Fig. 3. Electrical Characterization of Calcium Waves using Optogenetically Tagged Single Unit Activity Recording In Vivo. (A) Schematic illustrating how laser stimulation is used to identify and record from AVP neurons in freely behaving mice. (B) Photomicrograph of representative histological section illustrating the location of the fiber optrode above the SCN. Blue = DAPI; Green = EYFP expression tag for channelrhodopsin. OC = optic chiasm. (C) Blue light stimulation (blue bar) yielded an increase in firing rate in AVP+ neurons containing channelrhodopsin. (D) Histogram showing short (~2.25 ms) latency between blue light stimulation and action potential firing in AVP neurons containing channelrhodopsin. (E) Action potential firing measured from the optrode of a representative AVP neuron (Top) with the smoothed firing rate plotted beneath. (F) Representative 5-min calcium trace of GCaMP mediated fluorescence from an AVP neuron. Note the similarity between the fluorescence calcium waves to the smoothed firing rate dynamics measured by SUAAVP plotted in E. (G) Example histogram of firing rate for an AVP neuron. The black line indicates the fit to the log transform of the exponential fit with the yellow lines being the individual exponential components from the fit. (H) Example histogram from the same neuron showing the log transform of burst length. (I) Summary of average burst duration recorded with SUAAVP (Left) compared with the duration of calcium waves measured via miniaturized fluorescent microscopy (Right). There was no significant difference in duration between recording modalities and lighting conditions (ANOVA P > 0.05). Identification of ChR2-expressing neurons and data acquisition. After recovery from electrode implantation surgery, the mice were handled to adapt to extracellular recording procedures (30 min) and the optrode is gradually lowered through the brain toward the SCN (40 μm, once a day) until light stimulation through the optrode yielded single unit responses. For optogenetic identification of ChR2-expressing (AVP+) neurons, pulses of blue light (473 nm, 2 to 10 ms duration, 20 Hz, 10 to 15 mW) were delivered to the SCN through the optical fiber. A single unit was identified as a ChR2-expressing neuron when action potential spikes were repeatedly and reliably evoked by the blue light pulses (>50% occurrence) with a short first-spike latency (<7 ms), low jitter (<3 ms) and with waveforms consistent with prior-recorded spontaneous spikes (correlation coefficient between waveforms of spontaneous and optically evoked spikes >97%). The latency to action potential firing after light stimulation was 2.26 ms ± 0.02 ms (n = 100 randomly selected APs; Fig. 3D). Thirteen AVP neurons were recorded from five mice in total. However, for only three of them (AVP neuron #3, 4, 13), recorded from two mice, were we able to obtain complete recordings of a successive 96 h (48 h under LD and 48 h under DD). One neuron (AVP neuron #9) from one mouse was lost during recording under DD. So four neurons in LD and three neurons in DD were used for further data analysis. Electrophysiological data were recorded with a sampling rate of 20 KHz by a 16-channel head-stage and controller (RHD2000, Intan Technologies). Recorded raw voltage data were high-pass filtered (>250 Hz) and spikes were identified by applying thresholding. Single units were identified through offline sorting and then analyzed using NeuroExplorer and MATLAB software. All electrode locations were verified histologically (Fig. 3B). Due to the relative scarcity of the neurons recorded, longitudinal (i.e., circadian) analysis was not performed for SUAAVP data; instead, the data were only used to analyze bursting behavior. Image Analysis and Statistics. A complete description of our data analysis approach is provided in SI Appendix, Supplemental Methods. A brief description follows. Image correction, ROI definition, and production of intensity traces. Raw image stacks acquired at 6.67 Hz were motion-corrected using Inscopix Data Processing software and 5-min recordings of images were exported as TIFF stacks. The TIFF stacks were imported into Igor Pro 8 (Wavemetrics Inc, Lake Oswego, Oregon) for further analysis. Regions of interest (ROI) were identified from maximum projection images. Using the background subtracted image stack, the average intensity for each ROI was determined for each image in the stack. This produced a line trace of varying intensities with time over the 5-min period for each ROI (e.g., Fig. 1 F–H). Mean intensity, intensity correlation, and events analysis. To determine whether individual ROIs varied with circadian rhythmicity, we first characterized each ROI for its average intensity within the 5-min trace at each 3-h time point (Figs. 1F and 2 A–C). Fluorescence intensity traces for each ROI were also cross-correlated with each other yielding a Pearson coefficient for each ROI–ROI interaction at each circadian time point (Fig. 4A). In addition, we performed event analysis on the calcium traces. This was accomplished in two different ways. An unbiased analysis of acute events was conducted in which the onset of an event was found if the amplitude increased by more than two standard deviations above the intensity of the previous 20 points (3 s) (Figs. 1G and 2 D–F). A second method was used to determine event parameters of slow calcium waves (Figs. 1H and 2 G–I). In a first pass, automatic analysis was done in which the raw fluorescence line trace for each ROI was smoothed with a 50-point box window. Then, the first derivative was calculated from the smoothed raw data trace. The first derivative trace was also smoothed using a 100-point box window followed by recording the time when a threshold level on the first derivative was crossed in an increasing manner when searching from the start of the first derivative trace to the end for either positive or negative slopes. This approximated the start of slow calcium fluorescence wave rise and the end of the calcium wave. The value of the threshold of the first derivative was determined heuristically (usually 0.05). These time points could then be manually corrected for missing and/or spurious events. Wave durations, inter-event intervals, and wave number could then be calculated from the start and end times of each event in each fluorescent intensity line trace. Fig. 4. Circadian Rhythmicity in Correlation among AVP Neuron Pairs. (A) Schematic illustrating correlational (a.k.a. coherence) analysis among AVP neuron pairs. Calcium traces of individual SCN neurons across a field of view within a 5-min recording were cross-correlated, then the Pearson’s coefficient for cell pairs was plotted for each time point over 48 h and subjected to rhythmicity analysis. High correlation is interpreted as active interaction or coordination among cell pairs and referred to as coherence. Coherence among some pairs was strongly rhythmic. (B) Heat maps illustrating Pearson coefficients of single AVP neuronal pairs (rows) by timepoint (columns) in both DD and LD. Neuronal pairs which exhibited significant circadian rhythms according to cosinor (P < 0.05) in correlated activity appear above the dashed line and triangle. Plotted beneath the heatmap is the population average and standard errors for each timepoint. Where the population measure is rhythmic (or trending in that direction), the cosine fit is superimposed on Top of the bar graph (DD cosinor P = 0.004; LD cosinor P = 0.057). (C) Polar plots illustrating the phase distribution of ROI pairs exhibiting significant circadian rhythms in coherence in both DD and LD. The direction of the bar represents hour in CT for DD or ZT for LD. The length of the bar represents number of ROI pairs. Pie charts indicate the proportion that exhibited a statistically significant rhythm across the 48 h of DD and LD. (D) Histogram illustrating the frequency contribution to neuronal pair correlation. A low-pass filter (<0.25 Hz) applied to raw data yielded correlations with Pearson Coefficients approximating the unfiltered data. On average, over 90% of the power was carried by these low frequencies in comparison to when a high-pass filter (>1 Hz) was applied to the raw data. (E) Phenotype-tracking plots indicating the stability/instability of rhythmic state of AVP neuron ROI pair coherence between the two recording days (DD day 1 and DD day 2) under constant darkness (Left) and between the 48 h recordings in each lighting condition (Right). Blue lines indicate stable rhythmic or arrhythmic pairs, while red lines indicate a loss of rhythmicity between conditions, and green lines a gain of rhythmicity. Black triangles on the axes indicate the division between rhythmic (above) and arrhythmic (below) ROI pairs. Circadian rhythm analysis. Circadian rhythmicity was tested by fitting data (mean intensity, acute events, calcium wave variables including number, duration and inter-event interval, and Pearson correlations among ROI pairs) collected at 3-h intervals over 24 or 48 h with a cosine function. Fit optimization was based on Levenberg–Marquardt least-squares method constraining the results for the period between 21 and 28 h (50, 51). The P values from these fits were determined using a nonparametric Mann–Kendall tau test. A particular ROI was deemed rhythmic if the P value was below 0.05. The 24- or 48-h data were then ranked by P value and displayed as heatmaps. Data which could not be fit was given a P value of 1 in the resulting heatmaps and is not ranked (Figs. 2 A, D, and G and 4B). Those non-rhythmic ROIs are plotted below the dashed line in each heatmap. To determine whether AVP neurons displayed any rhythmicity as a population, the columns of the heatmaps were averaged, and these values were subjected to the circadian fit. For population averages, fitting was also weighted by the reciprocal of the SDs determined for each column of the heatmap. When the population measure was rhythmic, the cosine fit is displayed on the mean bar graph below each heat map (Figs. 2 A, D, and G and 4B). Phases resulting from the fits were recorded for each ROI or ROI–ROI interaction with significant rhythmicity. Histograms of the phases were then calculated utilizing 3-h bins and then plotted on polar coordinate graphs (Figs. 2 B, E, and H and 4C) with associated circular means and SEM (in green). Calculation of firing rate and duration of bursting for single-unit analysis. After spike sorting, traces dedicated to a particular single unit were analyzed for their firing rate (Fig. 3E). Time stamps of action potential firing were subjected to histogram analysis with a bin size of 3 s over the whole recording and then divided by the bin size to get the firing rate per 3 s time point. A histogram of this firing rate time course was then produced, indicating the prevalence of firing rates. Burst lengths were determined by rebinning the time stamp data per 10 s bins and determining the duration of the burst by analyzing the points when the firing rate had risen above and then returned below 1.5 Hz (Fig. 3G). Because firing rates and burst lengths could have a range of several decades, these histograms were converted to a log transform where the abscissa was the log of the firing rate and the ordinate was the square root of the number of events per bin (e.g., Fig. 3 G and H) (52). These data were then fit with a log transformation of a sum of exponential components (53) to determine the firing rate or the burst length for Fig. 3 G and H, respectively. Raw data from this study can be found in Dryad (https://doi.org/10.5061/dryad.2ngf1vhpz) (84). Results Circadian Rhythmicity in Mean Fluorescence of AVP Neurons in the SCN. Circadian analysis of mean fluorescence intensities could indicate if basal levels of calcium, a surrogate for overall levels of neuronal activity, would change over the course of the day. Average fluorescence intensity was determined from each 5-min recording at each 3-h time point for each ROI (e.g., Fig. 1F). The mean fluorescence per 3-h time point was fit with Equation 2 for each ROI to determine whether the ROI exhibited a circadian rhythm. Heatmaps of these rhythms in mean intensity are shown in Fig. 2A. In constant darkness, 24% of AVP neurons exhibited significant circadian rhythmicity (cosinor, P < 0.05) with mean peak phase at CT 8.69 ± 0.92 h; n = 6 ROIs (Fig. 2A, Left, above the dotted line; Fig. 2B, Upper); whereas in 12:12 LD, 53% of the AVP neurons demonstrated significant circadian rhythmicity with a mean peak phase at ZT 9.11 ± 0.80 h; n = 18 ROIs (Fig. 2A, Right, above the dotted line; Fig. 2B, Lower). For each individual 24-h day, approximately 35 to 50% of recorded cells were rhythmic, but the set of cells that made up that rhythmic subpopulation was dynamic, with neurons gaining or losing rhythmicity in the two 24-h periods (Fig. 2C, Left and SI Appendix, Fig. S4A). Also, there was no discernable pattern between lighting conditions except for an overall increase in the likelihood of rhythmicity in LD compared with DD (Fig. 2C, Right). When mean fluorescence values for all ROIs were averaged for each 3-h time point, the population did not display circadian rhythmicity in constant darkness (Fig. 2A; cosinor P = 0.546; n = 26 ROIs). In contrast, under a 12-h light/dark cycle the population average resulted in a significant population rhythm in mean fluorescence (Fig. 2B; cosinor P = 0.005; n = 34 ROIs), peaking during late day. Bootstrapping analysis was also performed to address the unequal contribution of cells from each mouse to the population measure (SI Appendix, Fig. S7 and Table S2) for this and other variables reported below. A majority of bootstrap trials in both lighting conditions suggested a population-level rhythm regardless of # of ROIs sampled per mouse. Rhythmicity of Acute Calcium Events in AVP Neurons. Although slower variants of GCaMP including the GCaMP7s used in this study are not capable of resolving distinct rapid events such as action potentials, significant changes in fluorescence within a recording are indicative of neuronal dynamics in these AVP neurons. Within the 5-min recordings, events were detected automatically by finding fluorescence that crossed a threshold that was two standard deviations above a sliding 3-s window (e.g., red diamonds, Fig. 1G). In DD, 50% of AVP neurons exhibited a circadian rhythm (cosinor P < 0.05) in the number of dynamic calcium events per 5-min recording session compared with 44.1% of neurons in LD (Fig. 2D). The subpopulation average phase for these dynamic events for these rhythmic cells in DD was at CT 14.21 ± 1.30 h (n = 13 ROIs) with a similar population average phase for LD at ZT 14.10 ± 1.68 h (n = 15 ROIs) (Fig. 2E). However, when considering all characterized AVP neurons (both rhythmic and non-rhythmic), the population average exhibited no significant rhythm in DD (Fig. 2D, left below the heatmap; cosinor P = 0.322; n = 26 ROIs) or LD (cosinor P = 0.652; n = 34 ROIs). This was largely due to the wide-phase distribution apparent in the polar plots for the rhythmic ROIs (Fig. 2E). Bootstrapping results confirmed this lack of population rhythmicity as a significant minority of trials resulted in the detection of a population rhythm (SI Appendix, Fig. S7 and Table S2). As was observed for mean fluorescence (Fig. 2C), there was a fairly consistent ~30% of neurons that were rhythmic for any given 24-h period, but the members of that rhythmic subset, and sometimes the phase of those rhythms (SI Appendix, Fig. S4H), were highly dynamic across the two recording days of DD (Fig. 2F, Left) and LD (SI Appendix, Fig. S4B), and between 48-h series from each lighting conditions (Fig. 2F, Right). AVP Neurons in the SCN Exhibit Distinctive Calcium Waves In Vivo. Slow, high-amplitude calcium waves appear to be a characteristic feature of most AVP neurons in the SCN. Nearly all ROIs in three of the four mice recorded for this study exhibited unambiguous waves distinct from the other dynamics apparent in the signal (SI Appendix, Table S1). For the fourth mouse (AVP63), the waves were less distinct from the faster dynamics. Such waves were never observed in control experiments in which AVP neurons expressing GFP under the CAG promoter were recorded using the same methods, in either DD or LD (Fig. 1E). The waves occurring within each 5-min recording were counted (Fig. 2H) across the circadian day to determine whether the frequency of these events is rhythmic. To be conservative, wave counting and analysis for rhythmicity were not performed for AVP63 (SI Appendix, Table S1). In DD, calcium waves were observed in most AVP neurons during at least one (usually most) time point(s), though a significant circadian rhythm in their counts was observed in only 16.7% of AVP neurons with a peak phase at CT 14.24 ± 0.96 h; n = 5 ROIs (Fig. 2 G and H). Under 12-h light/dark cycle, calcium waves were also observed in most analyzed neurons, and significant circadian rhythms were detectable in 26.5% of the cells with a peak phase at ZT 20.06 ± 1.79 h; n = 9 (Fig. 2 G and H). Averaging the number of waves by timepoint for all AVP neurons (or via bootstrapping simulations (SI Appendix, Fig. S7 and Table S2) did not reveal a significant circadian population rhythm in DD or in LD [Fig. 2G below the heatmaps; P = 1 (n = 26 ROIs); P = 0.133 (n = 34 ROIs) respectively]. As reported above for both mean fluorescence and acute calcium events, very little stability in the subpopulation of rhythmic ROIs was observed for the number of calcium waves over 2 d in DD (Fig. 2I, Left) or LD (SI Appendix, Fig. S4C), with a variable ~15 to 30% of neurons rhythmic for 24-h periods. And again, for calcium waves as reported above for other measures, largely different populations of cells were rhythmic in DD or LD (Fig. 2I, Right). In addition to the analysis of wave number here, wave duration (SI Appendix, Fig. S2) and interevent interval between these calcium waves (SI Appendix, Fig. S3) were also calculated. Neither additional measure provided the evidence of stable circadian rhythmicity or common phasing, and both exhibited similar state-switching (SI Appendix, Figs. S2D, S3D, S4, and S5) as reported above. Calcium Waves Likely Correspond to Bursting Activity in AVP Neurons. Optogenetically targeted Single-Unit Activity Recording (SUAAVP) was employed to determine if the timing of the slow wave phenomenon observed in calcium imaging (e.g., Fig. 1H) corresponded to increased action potential firing activity in individual AVP neurons within the intact SCN (Fig. 3 A and B). Mice expressing channelrhodopsin in AVP neurons were implanted with optrodes and neurons were verified as AVP positive if an increase in firing rate was elicited by blue light stimulation through the optrode (Fig. 3C). The latency to action potential firing after light stimulation was 2.26 ms ± 0.02 ms (n = 100 randomly selected Aps; Fig. 3D). When the firing rate was considered for 5-min epochs, a distinct pattern of bursting was observed (Fig. 3E) which corresponded closely with the calcium imaging data observed from SCN AVP neurons in other animals (Fig. 3F). Both firing rate (Fig. 3G) and burst duration (Fig. 3H) were calculated using the unbiased method described above. Average burst length was 13.41 ± 2.6 s and 8.92 ± 2.11 s in DD (n = 3 cells) and LD (n = 4 cells) respectively compared with the calcium wave duration of 14.20 ± 0.50 s and 13.97 ± 0.32 s in DD and LD respectively (Fig. 3I). Two-way ANOVA indicated that there was no main effect of method of observation (DF = 1, F1,677 = 1.27, P = 0.26; n = 7) or lighting condition (DF = 1, F1,677 = 0.827, P = 0.37, n = 7) as well as no interaction (DF = 1, F1,677 = 0.67, P = 04.1, n = 7). Rhythmicity of Correlated Activity Between AVP Neurons. Raw fluorescence across all cell/ROI pairs within a recording (Fig. 4A) was subjected to correlational analysis for each timepoint to quantify the degree of coherence, or coordinated activity among AVP neuron pairs. We sought to determine if the degree of coherence varies over circadian and diurnal time, which would indicate changes in the state of network signaling or output beyond those appreciated by single-neuron analysis, or by measurement of bulk fluorescence. We observed that many ROI pairs exhibited strong positive correlations between their calcium dynamics during the middle/late day/subjective day and either no correlation or were inversely correlated during the subjective night (Fig. 4A, Right). In DD, 34.33% of AVP neuron pairs exhibited a circadian rhythm in correlated activity (Fig. 4B, Left; ROI pairs above the dotted line) with an average peak correlation for rhythmic pairs at a phase of CT 9.83 ± 0.25 h; n = 149 ROI pairs (Fig. 4C, Upper). In LD, 34.70% of ROI pairs demonstrated significant rhythmicity of coherence with a mean peak phase at ZT 9.37 ± 0.24 h; n = 161 ROI pairs (Fig. 4B, Right and Fig. 4C, Lower). In both cases, the phase distributions were tight compared with measures of single-cell rhythms in Fig. 2. The population average of Pearson coefficients by timepoint exhibited a significant rhythm at the population level (all cells/ROIs) in DD and in LD (Fig. 4B, below the heatmap; DD: P = 0.004; n = 434 ROI pairs; LD: P = 0.024; n = 451 ROI pairs). However, bootstrapping simulations suggest that the population rhythm in LD is less circadian than under DD, when equal samples are drawn from each mouse (SI Appendix, Fig. S7 and Table S2). Most ROIs in the study had at least one rhythmic correlation with another ROI, with the modal # of rhythmic connections being from two to five (SI Appendix, Fig. S6C). We wanted to know how much of the power in the correlations resulted from low-frequency fluorescence changes (<0.25 Hz) or if higher-frequency calcium dynamics were mostly responsible for the coherence between AVP neuron ROIs. Frequency contribution analysis on a subset of interactions indicated that wave components slower than 0.25 Hz accounted for 99.2 ± 0.40% of the correlation between AVP neurons (n = 10 ROI pairs; Fig. 4D), indicating that an increased synchronous activity was arising from coincidence in calcium waves or other slow dynamics in AVP neuronal pairs rather than due to shared high-frequency dynamics. Rhythmicity in correlated activity consistently occurred in ~25 to 35% of AVP neuron pairs across all four of the 24-h periods for which we recorded (Fig. 4E and SI Appendix, Fig. S4), but the membership of that rhythmic subset of cell pairs was highly variable. Cell pairs either gained, lost, or maintained their rhythmic relationship across the 48-h recordings (Fig. 4E, Left and SI Appendix, Fig. S4) and between DD and LD (Fig. 4E, Right) with no predictability. The strength of correlation as a function of distance between neurons was also measured. In DD, there was a weak but significant relationship between correlated activity and the distance between AVP neurons, such that the further one neuron was from another for both positively (R2 = 0.007, P < 0.001; n = 160 ROI pairs) and negatively correlated relationships (R2 = 0.01, P < 0.004; n = 160 ROI pairs), Pearson coefficients decreased. Correlated activity also weakly decreased with distance between neurons for 12:12LD, for both positively correlated (R2 = 0.02, P < 0.001; n = 188 ROI pairs) and negatively correlated (R2 = 0.04, P < 0.001;n = 188 ROI pairs). Multimodal Analysis of Rhythmicity. SI Appendix, Table S1 provides a summary of rhythmic characteristics across all of our primary measures for each ROI. The final two columns indicate how many rhythmic coherence relationships each ROI has within that mouse. Likelihood of rhythmicity across variables appears to be stochastic, such that no one parameter predicts the value of another, and no single-cell variable predicts the number of rhythmic relationships. Results from each mouse in the study share these general characteristics. SI Appendix, Fig. S6 summarizes the multimodal analyses with histograms that indicate shared and unique rhythmic features within ROIs. About 60% of cells exhibited at least one rhythmic single-cell parameter in both DD and LD (SI Appendix, Fig. S6 A and B), and nearly all cells had at least one rhythmic relationship, with the modal # of rhythmic relationships of two to four. Discussion Individual Neuron Heterogeneity of Function in the Context of an Intact Circuit. The master circadian pacemaker in mammals in the SCN is comprised of groups of neurons which express specific neuropeptides and these cells may exert distinct influences over the regulation of circadian rhythms. It is still poorly understood how this network generates a unified output in which different neurons within the network interact to regulate circadian rhythmicity. Imaging of brain slices, using bioluminescent and fluorescent gene-reporter mice, demonstrates the existence of robust network organization that is predictably and reversibly altered by changing light conditions (11–13, 54). However, this approach is limited by the need to dissect the SCN network for observation, severing all inputs, outputs, and most intra-network connections. Multi-unit recording offers a window into the electrical activity of the intact SCN network in vivo, and such studies have been fundamental in developing our understanding of the SCN as a master pacemaker for circadian rhythmicity (28–31, 55), but are unable to provide additional information regarding cell-type-specific population or individual neuronal contribution to the regulation of circadian rhythmicity within the SCN. Intersectional genetics and novel recording techniques have dramatically improved our ability to conduct population level, cell-type-specific studies within the intact circadian network. For example, recent studies in vivo demonstrate the necessity and sufficiency of VIP-expressing neurons in synchronizing the SCN to light-induced phase resetting (23, 32, 33). In vivo FP is a powerful technique, but one that can be enhanced with the ability to observe individual neurons and the interactions between these neurons within the fully intact SCN network. Thus, here we present work involving two novel techniques: optogenetically targeted single unit activity recording (SUAAVP) and in vivo calcium imaging of AVP-expressing neurons in the intact SCN of awake, freely moving mice under different lighting schedules using miniaturized fluorescence microscopy. Targeted SUA has been used for cell-type specific electrophysiological recording in the medial prefrontal cortex (56), and the striatum (57), but applied here for the first time in the SCN. In vivo calcium imaging has been used to measure calcium dynamics in many brain areas including the frontal cortex (58), hippocampus (59), and cerebellum (60, 61). This technique permits the characterization of individual cells within the SCN and how they contribute to communal network behavior, and how these contributions are modulated over circadian and diurnal time. Our data indicate that most AVP neurons in the SCN exhibit slow calcium waves, approximately 20% of which show a modulatory rhythm of their frequency over a 24-h period (Fig. 2 G–I). While increases in intracellular calcium can be indicative of neuronal firing (62), because of the slow decay time of the variant of GCaMP7 we employed, temporal discrimination of individual action potentials within bursting events was not possible. Thus, we employed SUAAVP to better characterize the firing behavior of a small number of AVP neurons within the SCN in vivo. The burst duration resulting from firing rate of AVP neurons by SUAAVP is similar to the duration of calcium waves measured by fluorescence (Fig. 3I). This suggests that the waves we observe in our calcium imaging data are likely the result of bursting activity exhibited by AVP neurons. Previous studies observed similar bursting activity in AVP neurons of the supraoptic nucleus (63) and paraventricular nucleus (64). Although the specific cell types were not identified, such bursting was also reported within the SCN (65). The present results are the first to bridge these findings and confirm that AVP neurons within the SCN exhibit phasic firing, or bursting activity, in vivo. Interestingly, although bursting activity is present in nearly all SCN AVP neurons recorded, a minority of neurons exhibit circadian rhythmicity in bursting. This suggests that while bursting is an identifying characteristic of these neurons, it may not play a significant role in the regulation of circadian rhythms. Although circadian or diurnal regulation of bursting is uncommon in SCN AVP+ neurons (Fig. 2 G–I and SI Appendix, Figs. S2 and S3), the well-known role of these cells as regulators of the circadian clock suggests that some aspect of their activity should exhibit reliable rhythmicity. AVP-specific Per2::luc rhythms in SCN slices illustrate the presence of functional molecular clocks (26). We have observed similar results using a D site-binding protein (DBP) reporter in AVP-expressing SCN cells (42); however, our selection of cell-based ROI used rhythmicity as a criterion and thus excluded non-rhythmic cells which may have been present. Imaging from SCN slices from AVPEluc/+ mice demonstrates the rhythmic transcription of avp in the SCN (66), and there is a circadian rhythm in the secretion of vasopressin (67, 68) and GABA (34) from AVP neurons in the SCN. Using longitudinal calcium imaging, we find that stable circadian rhythmicity is not observed in all AVP neurons. Only a minority of these neurons exhibited rhythmicity for each of the three main parameters characterized (mean fluorescent intensity, calcium events, or calcium waves). While individual neurons sometimes exhibit circadian rhythmicity in various measures (sometimes several measures; SI Appendix, Table S1 and Fig. S6), these rhythms are not stable from day to day nor is there a reliable trend in rhythmicity between lighting conditions (Fig. 2, and SI Appendix, Figs. S2–S4). Furthermore, the phases of the calcium rhythms that we did observe vary among cells as well as across time and lighting condition such that rhythms do not emerge at the population level in some of these measures. This was not equally true among the three measures, as mean intensity was rhythmic in LD at the population level and was trending that way in DD. Bootstrapping simulations suggested that population rhythms existed in both lighting conditions (SI Appendix, Table S2). But overall, while AVP neurons exhibit strong rhythms in gene transcription with coherent phasing in slice recordings, such rhythmicity may not accurately reflect the temporal, and electrical activity of a majority of these neurons at all times in vivo. Instead, we suggest that stable rhythmic activity of individual neurons is not needed. Instead, perhaps a generous (and ever-changing) subset (i.e., ~30% of cells perhaps) need only be rhythmic, electrically, in order to drive rhythmic outputs. These data may suggest a novel view of network rhythmicity for AVP neurons, where all cells contribute to the generation of a population rhythm, but not all the time. A very similar phenomenon has been reported for hippocampus, where the dynamic sharing of the responsibility for encoding place memory among neurons was described (69). At first blush, this view of SCN rhythmic organization may seem at odds with what has been reported prior, that a large majority of SCN neurons exhibit rhythms in gene expression and bulk Ca2+ which reflect underlying molecular rhythms in these cells (7, 8, 11, 40). We argue that our findings reflect properties only appreciable through longitudinal in vivo recordings with cellular resolution. It is important to distinguish how these cells might behave in the presence of dynamic input from outside of the SCN from how they behave in a slice. It is also important to distinguish how a cell behaves over time, versus how the population does. Previous studies utilizing diverse techniques have had either single-cell resolution in a reduced prep, OR in vivo bulk recordings, OR in vivo single cell recordings without cellular specificity. No studies have had all of these advantages combined, and thus could not have observed the stochasticity and dynamically shared function reported here. The closest comparison is in vivo MUA/SUA (70, 71), which (until now) lacks cell specificity, but nonetheless has revealed non-rhythmic cells, cells with diverse phases, and acute cellular behavior which reflects dynamic behavioral input, which are observations one would predict given our current results. Thus, the suggestion of diversity among neurons and stochastic behavior of rhythms of individual neurons in the SCN circuit in vivo is not surprising. Such diversity of rhythmicity (43) and phase (40) both among cells and over time within cells has been reported for SCN. Phasic diversity among cells in vivo may also play an important role in plasticity and robustness of the circuit (41). Recordings of other brain circuits using the same technique employed in our study suggest that heterogeneity of function, even among cells that share many other features is the norm, not the exception(35–38), and is likely a fundamental property of neural circuits. Correlated Activity among Cells during Only the Daytime Is an Emergent Network Property of AVP Neurons. Recent work suggests that emergent properties of the SCN network play an important role in the generation of robust and stable rhythmicity. Emergent properties are observations that cannot be appreciated by studying single cells alone, and that may have functional significance for how the network operates. The wave observed in bioluminescent slice recordings reveal a spatiotemporal organization within the SCN (11, 14, 15) which is predictably and reversibly altered by changes in environmental lighting (16, 17). Synchronization between cells may be dependent on this organization (72, 73) driven by region-specific circadian programming within the SCN network (54, 74, 75). However, to date, it is unknown if such an organization observed ex vivo extends to an intact SCN network in vivo. Previous in vivo studies were not capable of single-cell discrimination and/or have not been able to observe multiple cells simultaneously, and therefore have not addressed the issue of network organization. Here, while single-cell rhythms appear unstable, we observed an underlying circadian structure to coordinated activity among neuron pairs. Calcium dynamics in many AVP cell pairs are more highly correlated (i.e., coherent) during the day than during the night, indicating either coordinated function intrinsic to those cells, or a shared acute response to a common input of either intrinsic or extrinsic origin (Fig. 4). This coherence in activity is dependent on distance such that the further two neurons are from each other in the same field of view, the less correlated they are likely to be. Generally, the relationship is absent during the night such that the cells behaved entirely independently, or sometimes are inversely correlated. About 33% of all AVP neuron pair relationships have significant circadian rhythms, and nearly every cell we recorded has at least one such relationship (most have more) (SI Appendix, Fig. S6C). The average of these correlations across all cell pairs, a surrogate population measure, in both DD and LD exhibit a significant circadian rhythm with a (subjective) daytime peak between CT/ZT 6 to 9. This in vivo observation of rhythms in coherence may reflect time-of-day dependent signaling dynamics in the SCN previously reported. GABA is known to switch between being inhibitory and excitatory in a time-of-day dependent manner (76–79), regulated via neuronal intracellular chloride concentration and expression of the Na+-K+-2Cl− cotransporter (NKCC1). The significance of this is demonstrated by time-dependent blockade of the NKCC1 transport, which results in diminished phase-resetting to light exposure during the early subjective night, but not during the late subjective night or during the light phase (80). Changing network dynamics may also be driven thru non-synaptic paracrine signaling (e.g., by AVP itself). The relevance of this time-of-day dependent rhythm in neuronal coherence for circadian function needs further study. Study Limitations. There are technical limitations to this body of work that place some constraints upon the conclusions. Due to the difficulty in cell-type specific targeting of a relatively small neuronal population (~2,000 neurons) within a very small, deep brain region, the number of AVP neurons which were observed in a longitudinal fashion was somewhat limited. Now that feasibility of this technique has been established by us and others (81), other cell-specific populations can be examined. The AVP-IRES2-Cre mouse used in this study has been reported to express reduced levels of AVP protein and mRNA, with no change in circadian behavior or AVP cell number (48). This fact limited our ability to demonstrate robust colocalization of AVP and gcAMP7 expression using immunofluorescence. However, the model is well-documented to selectively target AVP neurons in the SCN and elsewhere (42, 44–46). Additionally, there is the possibility that a minority of AVP-expressing neurons dorsal to, but not part of, the SCN that are marked by the cre- line may have been recorded (82). It was not possible to distinguish between these and AVP-expressing neurons within the SCN in the present study. Since the vast majority of AVP-expressing neurons in the ventromedial hypothalamus are within the SCN (19), it is almost certain that most cells recorded are within its borders. No lens was found to be dorsal enough to have recorded from the hypothalamic paraventricular nucleus, and lateral placements that captured supraoptic neurons were unambiguous and omitted from this analysis. In Fig. 1A, the fluorescence that is present dorsal to the SCN is largely non-cellular in appearance. And since the heterogeneity we report is longitudinal across time in addition to across cells, ectopic AVP-neurons present in our recordings would not account for such diversity. Future studies will take advantage of novel cell position identification techniques such as light-guided sectioning (83) to identify precisely where, and what, recorded neurons are post hoc, allowing us to determine whether they exhibit different behavior inside versus outside the SCN. Another potential limitation is that damage caused by implantation of the GRIN lens could have altered the function of the SCN. Tissue damage is also a limitation for other techniques such as slice recording (neo-natal or adult), in vivo MUA and FP approaches. The question being addressed with SUAAVP in the present study did not require recording from many neurons, resulting in a relatively low number of observations. This precluded us from doing an accurate longitudinal analysis. Future studies will include full 48-h observations of a larger number of cells in both in DD and LD using this parallel technique. Finally, since the changing behavioral state of animals may modify cellular function acutely, behavioral and sleep measures will be added to future cellular imaging experiments to assess this possibility and to better understand the functions of these neurons within the context of behavior. The population measure of rhythmic coherence among AVP cell pairs highlights the importance of emergent characteristics in the SCN neural circuit as well as the potential limitations of studying population-level dynamics which ignore individual neuronal activity. While the use of cell-type specific AVP recording in this study helped to ensure we were observing network activity mostly within the SCN, the relative sparsity of this neuronal subtype made observing a large number of neurons prohibitive. Furthermore, conducting such analysis on only a single cell type within a heterogeneous network is inherently limiting. Therefore, future studies will focus on investigating the role of network organization in a more diverse, SCN-specific cell population. Given what we have now learned about AVP cell behavior in vivo, recording a broader cohort of cell types inclusive of, but extending beyond just AVP neurons, will reveal whether our findings are reflective of behavior unique to AVP cells, or reflect general properties of SCN neuronal function. Supplementary Material Appendix 01 (PDF) Click here for additional data file. This work was supported by NIH grants R21NS108197 and R35GM136661 to A.J.D., SC1AG046907 to M.B., and SC1GM127260 to J.C.E., NHLBI-funded predoctoral fellowship to D.M.B. (T32HL007901) and an equipment grant to MSM by the W.M. Keck Foundation. We also thank Ivory Ellis and Inscopix, Inc., for technical assistance with the project. Author contributions A.S., Z.Q., J.C.E., M.B., and A.J.D. designed research; A.S., Z.Q., and D.D.C.B. performed research; M.B. and A.J.D. contributed new reagents/analytic tools; A.S., Z.Q., D.M.B., J.C.E., M.B., and A.J.D. analyzed data; and A.S., D.M.B., J.C.E., M.B., and A.J.D. wrote the paper. Competing interest The authors declare no competing interest. Data, Materials, and Software Availability Raw data from this study can be found in Dryad (https://doi.org/10.5061/dryad.2ngf1vhpz) (84). Supporting Information Preprint: https://doi.org/10.1101/2021.12.07.471437. This article is a PNAS Direct Submission. ==== Refs 1 M. R. Ralph , Transplanted suprachiasmatic nucleus determines circadian period. Science 247 , 975–978 (1990).2305266 2 D. R. Weaver, The suprachiasmatic nucleus: A 25-year retrospective. J. Biol. Rhythms 13 , 100–112 (1998).9554572 3 J. S. Takahashi, Transcriptional architecture of the mammalian circadian clock. Nat. Rev. Genet 18 , 164–179 (2017).27990019 4 D. K. Welsh , Individual neurons dissociated from rat suprachiasmatic nucleus express independently phased circadian firing rhythms. Neuron 14 , 697–706 (1995).7718233 5 J. Su Terman, C. E. Reme, D. K. Terman, Rod outer segment disk shedding in rats with lesions of the suprachiasmatic nucleus. Brain Res. 605 , 256–264 (1993).8481775 6 K. B. Koronowski , Defining the independence of the liver circadian clock. Cell 177 , 1448–1462.e14 (2019).31150621 7 S.-H. Yoo , PERIOD2::LUCIFERASE real-time reporting of circadian dynamics reveals persistent circadian oscillations in mouse peripheral tissues. Proc. Natl. Acad. Sci. U.S.A. 101 , 5339–5346 (2004).14963227 8 E. D. Herzog , Circadian rhythms in mouse suprachiasmatic nucleus explants on multimicroelectrode plates. Brain Res. 757 , 285–290 (1997).9200759 9 S. Honma , Circadian periods of single suprachiasmatic neurons in rats. Neurosci. Lett. 250 , 157–160 (1998).9708856 10 E. D. Herzog, J. S. Takahashi, G. D. Block, Clock controls circadian period in isolated suprachiasmatic nucleus neurons. Nat. Neurosci. 1 , 708–713 (1998).10196587 11 T. Noguchi , Calcium circadian rhythmicity in the suprachiasmatic nucleus: Cell autonomy and network modulation. eNeuro 4 , 4 (2017). 12 C. Schmal, E. D. Herzog, H. Herzel, Measuring relative coupling strength in circadian systems. J. Biol. Rhythms 33 , 84–98 (2018).29219034 13 S. Yamaguchi , Synchronization of cellular clocks in the suprachiasmatic nucleus. Science 302 , 1408 (2003).14631044 14 M. H. Hastings , Differential regulation of mPER1 and mTIM proteins in the mouse suprachiasmatic nuclei: New insights into a core clock mechanism. J. Neurosci. 19 , Rc11 (1999).10366649 15 R. Enoki , Topological specificity and hierarchical network of the circadian calcium rhythm in the suprachiasmatic nucleus. Proc. Natl. Acad. Sci. U.S.A. 109 , 21498 (2012).23213253 16 J. A. Evans , Dynamic interactions mediated by nonredundant signaling mechanisms couple circadian clock neurons. Neuron 80 , 973–983 (2013).24267653 17 K. E. Rohr , Seasonal plasticity in GABA(A) signaling is necessary for restoring phase synchrony in the master circadian clock network. Elife 8 , e49578 (2019).31746738 18 R. Y. Moore, J. C. Speh, GABA is the principal neurotransmitter of the circadian system. Neurosci. Lett. 150 , 112–116 (1993).8097023 19 E. E. Abrahamson, R. Y. Moore, Suprachiasmatic nucleus in the mouse: Retinal innervation, intrinsic organization and efferent projections. Brain Res. 916 , 172–191 (2001).11597605 20 Q. J. Meng , Setting clock speed in mammals: The CK1 epsilon tau mutation in mice accelerates circadian pacemakers by selectively destabilizing PERIOD proteins. Neuron 58 , 78–88 (2008).18400165 21 I. T. Lee , Neuromedin s-producing neurons act as essential pacemakers in the suprachiasmatic nucleus to couple clock neurons and dictate circadian rhythms. Neuron 85 , 1086–1102 (2015).25741729 22 J. Park , Single-cell transcriptional analysis reveals novel neuronal phenotypes and interaction networks involved in the central circadian clock. Front. Neurosci. 10 , 481 (2016).27826225 23 J. R. Jones , SCN VIP neurons are essential for normal light-mediated resetting of the circadian system. J. Neurosci. 38 , 7986 (2018).30082421 24 M. Mieda , Cellular clocks in AVP neurons of the SCN are critical for interneuronal coupling regulating circadian behavior rhythm. Neuron 85 , 1103–1116 (2015).25741730 25 M. Mieda, H. Okamoto, T. Sakurai, Manipulating the cellular circadian period of arginine vasopressin neurons alters the behavioral circadian period. Curr. Biol. 26 , 2535–2542 (2016).27568590 26 Y. Shan , Dual-color single-cell imaging of the suprachiasmatic nucleus reveals a circadian role in network synchrony. Neuron 108 , 164–179.e7 (2020).32768389 27 W. D. Todd , Suprachiasmatic VIP neurons are required for normal circadian rhythmicity and comprised of molecularly distinct subpopulations. Nat. Commun. 11 , 4410 (2020).32879310 28 S. T. Inouye, H. Kawamura, Persistence of circadian rhythmicity in a mammalian hypothalamic “island” containing the suprachiasmatic nucleus. Proc. Natl. Acad. Sci. U.S.A. 76 , 5962–5966 (1979).293695 29 S. Yamazaki , Rhythmic properties of the hamster suprachiasmatic nucleus in vivo. J. Neurosci. 18 , 10709 (1998).9852606 30 W. Nakamura , In vivo monitoring of circadian timing in freely moving mice. Curr. Biol. 18 , 381–385 (2008).18334203 31 N. N. Takasu , In vivo monitoring of multi-unit neural activity in the suprachiasmatic nucleus reveals robust circadian rhythms in Period1−/− mice. PLoS One 8 , e64333 (2013).23717599 32 D. Ono, K. Honma, S. Honma, Circadian and ultradian rhythms of clock gene expression in the suprachiasmatic nucleus of freely moving mice. Sci. Rep. 5 , 12310 (2015).26194231 33 C. Mazuski , Entrainment of Circadian Rhythms Depends on Firing Rates and Neuropeptide Release of VIP SCN Neurons. Neuron 99 , 555–563.e5 (2018).30017392 34 T. Maejima , GABA from vasopressin neurons regulates the time at which suprachiasmatic nucleus molecular clocks enable circadian behavior. Proc. Natl. Acad. Sci. U.S.A. 118 , e2010168118 (2021).33526663 35 B. C. Jarvie , Satb2 neurons in the parabrachial nucleus mediate taste perception. Nat. Commun. 12 , 224 (2021).33431851 36 J. H. Jennings , Visualizing hypothalamic network dynamics for appetitive and consummatory behaviors. Cell 160 , 516–527 (2015).25635459 37 P. Viskaitis , Modulation of SF1 neuron activity coordinately regulates both feeding behavior and associated emotional states. Cell Rep. 21 , 3559–3572 (2017).29262334 38 C. A. Zimmerman , A gut-to-brain signal of fluid osmolarity controls thirst satiation. Nature 568 , 98–102 (2019).30918408 39 M. Bellesi , Enhancement of sleep slow waves: Underlying mechanisms and practical consequences. Front. Syst. Neurosci. 8 , 208 (2014).25389394 40 J. A. Evans , Intrinsic regulation of spatiotemporal organization within the suprachiasmatic nucleus. PLoS One 6 , e15869 (2011).21249213 41 J. A. Evans , Neural correlates of individual differences in circadian behaviour. Proc. Biol. Sci. 282 , 1810 (2015). 42 C. B. Smith , Cell-type-specific circadian bioluminescence rhythms in Dbp reporter mice. J. Biol. Rhythms 37 , 53–77 (2022).35023384 43 A. B. Webb , Intrinsic, nondeterministic circadian rhythm generation in identified mammalian neurons. Proc. Natl. Acad. Sci. U.S.A. 106 , 16493–16498 (2009).19805326 44 H. Pei , AVP neurons in the paraventricular nucleus of the hypothalamus regulate feeding. Mol. Metab. 3 , 209–215 (2014).24634830 45 Y. Shan , Dual-color single-cell imaging of the suprachiasmatic nucleus reveals a circadian role in network synchrony. Neuron 108 , 164–179.e7 (2020).32768389 46 J. Whylings , Reduction in vasopressin cells in the suprachiasmatic nucleus in mice increases anxiety and alters fluid intake. Horm. Behav. 133 , 104997 (2021).34062279 47 S. L. Resendez , Visualization of cortical, subcortical and deep brain neural circuit dynamics during naturalistic mammalian behavior with head-mounted microscopes and chronically implanted lenses. Nat. Protoc. 11 , 566–597 (2016).26914316 48 A. H. Cheng, S. W. Fung, H. M. Cheng, Limitations of the Avp-IRES2-Cre (JAX #023530) and Vip-IRES-Cre (JAX #010908) models for chronobiological investigations. J. Biol. Rhythms 34 , 634–644 (2019).31452438 49 J. A. Cardin , Targeted optogenetic stimulation and recording of neurons in vivo using cell-type-specific expression of Channelrhodopsin-2. Nat. Protoc. 5 , 247–254 (2010).20134425 50 W. H. Press,B. P. Flannery,S. A. Teukolsky,W. T. Vetterling, Modeling of data in numerical recipes in C (Cambridge University Press, 1992), pp. 668–672. 51 D. W. Marquardt, An algorithm for least-squares estimation of Nonlinear Parameters. J. Soc. Ind. Appl. Math. 11 , 431–441 (1963). 52 F. Sigworth, S. Sine, Data transformations for improved display and fitting of single-channel dwell time histograms. Biophys. J. 52 , 1047–1054 (1987).2447968 53 D. Colquhoun, F. Sigworth, Fitting and statistical analysis of single-channel records in Single-Channel Recording, B. Sakmann, E. Neher, Eds. ( Springer, Boston, MA, 1995), pp. 483–587. 54 R. Enoki , Dual origins of the intracellular circadian calcium rhythm in the suprachiasmatic nucleus. Sci. Rep. 7 , 41733 (2017).28155916 55 S. Michel , Electrophysiological approaches to studying the suprachiasmatic nucleus. Methods Mol. Biol. 2130 , 303–324 (2021).33284454 56 P. Anikeeva , Optetrode: A multichannel readout for optogenetic control in freely moving mice. Nat. Neurosci. 15 , 163–170 (2012). 57 A. V. Kravitz, S. F. Owen, A. C. Kreitzer, Optogenetic identification of striatal projection neuron subtypes during in vivo recordings. Brain Res. 1511 , 21–32 (2013).23178332 58 R. Prevedel , Fast volumetric calcium imaging across multiple cortical layers using sculpted light. Nat. Methods 13 , 1021–1028 (2016).27798612 59 T. Indersmitten , Utilizing miniature fluorescence microscopy to image hippocampal place cell ensemble function in Thy1.GCaMP6f transgenic mice. Curr. Protoc. Pharmacol. 82 , e42 (2018).30129249 60 M. R. Sullivan , In vivo calcium imaging of circuit activity in cerebellar cortex. J. Neurophysiol. 94 , 1636–1644 (2005).16079125 61 J. Díez-García, W. Akemann, T. Knöpfel, In vivo calcium imaging from genetically specified target cells in mouse cerebellum. NeuroImage 34 , 859–869 (2007).17161628 62 W. Fleming , Inferring spikes from calcium imaging in dopamine neurons. PLOS One 16 , e0252345 (2021).34086726 63 L. Gouzènes , Vasopressin regularizes the phasic firing pattern of rat hypothalamic magnocellular vasopressin neurons. J. Neurosci. 18 , 1879 (1998).9465012 64 M. Dewald , Phasic bursting activity of paraventricular neurons is modulated by temperature and angiotensin II. J. Thermal Biol. 24 , 339–345 (1999). 65 J. D. Miller, C. A. Fuller, Isoperiodic neuronal activity in suprachiasmatic nucleus of the rat. Am. J. Physiol. Regul. Integr. Comp. Physiol. 263 , R51–R58 (1992). 66 T. Yoshikawa , Spatiotemporal profiles of arginine vasopressin transcription in cultured suprachiasmatic nucleus. Eur. J. Neurosci. 42 , 2678–2689 (2015).26342201 67 P. Södersten , A daily rhythm in behavioral vasopressin sensitivity and brain vasopressin concentrations. Neurosci. Lett. 58 , 37–41 (1985).4047474 68 A. Kalsbeek , In vivo measurement of a diurnal variation in vasopressin release in the rat suprachiasmatic nucleus. Brain Res. 682 , 75–82 (1995).7552330 69 Y. Ziv , Long-term dynamics of CA1 hippocampal place codes. Nat. Neurosci. 16 , 264–266 (2013).23396101 70 J. H. Meijer , Multiunit activity recordings in the suprachiasmatic nuclei: In vivo versus in vitro models. Brain Res. 753 , 322–327 (1997).9125419 71 J. Schaap , Heterogeneity of rhythmic suprachiasmatic nucleus neurons: Implications for circadian waveform and photoperiodic encoding. Proc. Natl. Acad. Sci. U.S.A. 100 , 15994–15999 (2003).14671328 72 P. Paszek , Population robustness arising from cellular heterogeneity. Proc. Natl. Acad. Sci. U.S.A. 107 , 11644 (2010).20534546 73 C. Gu, H. Yang, Differences in intrinsic amplitudes of neuronal oscillators improve synchronization in the suprachiasmatic nucleus. Chaos 27 , 093108 (2017).28964140 74 R. Enoki , Synchronous circadian voltage rhythms with asynchronous calcium rhythms in the suprachiasmatic nucleus. Proc. Natl. Acad. Sci. U.S.A. 114 , E2476 (2017).28270612 75 S. Pauls , Differential contributions of intra-cellular and inter-cellular mechanisms to the spatial and temporal architecture of the suprachiasmatic nucleus circadian circuitry in wild-type, cryptochrome-null and vasoactive intestinal peptide receptor 2-null mutant mice. Eur. J. Neurosci. 40 , 2528–2540 (2014).24891292 76 S. Wagner , GABA in the mammalian suprachiasmatic nucleus and its role in diurnal rhythmicity. Nature 387 , 598–603 (1997).9177347 77 H. J. Choi , Excitatory actions of GABA in the suprachiasmatic nucleus. J. Neurosci. 28 , 5450–5459 (2008).18495878 78 M. D. Jeu, C. Pennartz, Circadian modulation of GABA function in the rat suprachiasmatic nucleus: Excitatory effects during the night phase. J. Neurophysiol. 87 , 834–844 (2002).11826050 79 R. P. Irwin, C. N. Allen, GABAergic signaling induces divergent neuronal Ca2+ responses in the suprachiasmatic nucleus network. Eur. J. Neurosci. 30 , 1462–1475 (2009).19821838 80 J. K. McNeill, J. C. Walton, H. E. Albers, Functional significance of the excitatory effects of GABA in the suprachiasmatic nucleus. J. Biol. Rhythms 33 , 376–387 (2018).29974800 81 L. E. Cheikh Hussein , Nested calcium dynamics support daily cell unity and diversity in the suprachiasmatic nuclei of free-behaving mice. bioRxiv [Preprint] (2021). Accessed 20 October 2022. 82 K. E. Rohr , Vasopressin regulates daily rhythms and circadian clock circuits in a manner influenced by sex. Horm. Behav. 127 , 104888 (2021).33202247 83 A. Kahan , Light-guided sectioning for precise in situ localization and tissue interface analysis for brain-implanted optical fibers and GRIN lenses. Cell Rep. 36 , 109744 (2021).34592157 84 A. Stowie , Arginine-vasopressin expressing neurons in the murine suprachiasmatic nucleus exhibit a circadian rhythm in network coherence in vivo. Dryad. 10.5061/dryad.2ngf1vhpz. Deposited 3 January 2023.
PMC009xxxxxx/PMC9942908.txt
==== Front Proc Natl Acad Sci U S A Proc Natl Acad Sci U S A PNAS Proceedings of the National Academy of Sciences of the United States of America 0027-8424 1091-6490 National Academy of Sciences 36656861 202217840 10.1073/pnas.2217840120 research-articleResearch ArticlegeneticsGenetics419 Biological Sciences Genetics BAP1 is a novel regulator of HIF-1α Bononi Angela a 1 Wang Qian b c 1 Zolondick Alicia A. a d Bai Fang b e f https://orcid.org/0000-0003-1468-5568 Steele-Tanji Mika a Suarez Joelle S. a https://orcid.org/0000-0003-1528-2701 Pastorino Sandra a Sipes Abigail a Signorato Valentina a Ferro Angelica a Novelli Flavia a https://orcid.org/0000-0002-3746-7478 Kim Jin-Hee a Minaai Michael a d Takinishi Yasutaka a Pellegrini Laura a Napolitano Andrea a Xu Ronghui a https://orcid.org/0000-0002-9842-2628 Farrar Christine a https://orcid.org/0000-0001-8561-2257 Goparaju Chandra a Bassi Cristian g Negrini Massimo g Pagano Ian a https://orcid.org/0000-0002-6248-0113 Sakamoto Greg a Gaudino Giovanni a https://orcid.org/0000-0002-1572-6571 Pass Harvey I. h Onuchic José N. [email protected] b 2 http://orcid.org/0000-0002-9448-0388 Yang Haining [email protected] a 2 https://orcid.org/0000-0003-1417-2420 Carbone Michele [email protected] a 2 https://orcid.org/0000-0001-8928-8474 aThoracic Oncology, University of Hawaii Cancer Center, Honolulu, HI 96813 bCenter for Theoretical Biological Physics, Rice University, Houston, TX 77005 cHefei National Laboratory for Physical Sciences at the Microscale and Department of Physics, University of Science and Technology of China, Hefei, Anhui 230026, China dDepartment of Molecular Biosciences and Bioengineering, University of Hawaii at Manoa, Honolulu, HI 96822 eShanghai Institute for Advanced Immunochemical Studies, ShanghaiTech University, Shanghai 201210, China fSchool of Life Science and Technology, ShanghaiTech University, Shanghai 201210, China gDepartment of Translational Medicine LTTA Centre University of Ferrara, Ferrara 44121, Italy hDepartment of Cardiothoracic Surgery, New York University, New York, NY 10016 2To whom correspondence may be addressed. Email: [email protected], [email protected], or [email protected]. Contributed by José N. Onuchic; received October 18, 2022; accepted December 22, 2022; reviewed by Pier Paolo Pandolfi and I. Le Poole 1A.B. and Q.W. contributed equally to this work. 19 1 2023 24 1 2023 19 7 2023 120 4 e221784012018 10 2022 22 12 2022 Copyright © 2023 the Author(s). Published by PNAS. 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This article is distributed under Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND). Significance BAP1 modulates crucial cellular pathways that regulate genomic stability and cell death. BAP1 mutations on the one hand favor malignant transformation and mesothelioma development; on the other hand, they reduce mesothelioma aggressiveness. Investigating this apparent paradox, we discovered that BAP1 deubiquitylates and stabilizes HIF-1α in hypoxia; thus, BAP1 inactivating mutations significantly reduce HIF-1α. Given the critical role of HIF-1α in promoting tumor invasion, we propose that: 1) Reduced BAP1 in the tumor cells and tumor microenvironment of individuals carrying germline BAP1 mutations may contribute to the reduced invasion and the significantly improved prognosis of mesothelioma; 2) targeting wild-type BAP1 after tumor development could be a novel effective strategy to reduce HIF-1α protein levels in hypoxic tissues and impair tumor growth. BAP1 is a powerful tumor suppressor gene characterized by haplo insufficiency. Individuals carrying germline BAP1 mutations often develop mesothelioma, an aggressive malignancy of the serosal layers covering the lungs, pericardium, and abdominal cavity. Intriguingly, mesotheliomas developing in carriers of germline BAP1 mutations are less aggressive, and these patients have significantly improved survival. We investigated the apparent paradox of a tumor suppressor gene that, when mutated, causes less aggressive mesotheliomas. We discovered that mesothelioma biopsies with biallelic BAP1 mutations showed loss of nuclear HIF-1α staining. We demonstrated that during hypoxia, BAP1 binds, deubiquitylates, and stabilizes HIF-1α, the master regulator of the hypoxia response and tumor cell invasion. Moreover, primary cells from individuals carrying germline BAP1 mutations and primary cells in which BAP1 was silenced using siRNA had reduced HIF-1α protein levels in hypoxia. Computational modeling and co-immunoprecipitation experiments revealed that mutations of BAP1 residues I675, F678, I679, and L691 -encompassing the C-terminal domain-nuclear localization signal- to A, abolished the interaction with HIF-1α. We found that BAP1 binds to the N-terminal region of HIF-1α, where HIF-1α binds DNA and dimerizes with HIF-1β forming the heterodimeric transactivating complex HIF. Our data identify BAP1 as a key positive regulator of HIF-1α in hypoxia. We propose that the significant reduction of HIF-1α activity in mesothelioma cells carrying biallelic BAP1 mutations, accompanied by the significant reduction of HIF-1α activity in hypoxic tissues containing germline BAP1 mutations, contributes to the reduced aggressiveness and improved survival of mesotheliomas developing in carriers of germline BAP1 mutations. BAP1 HIF-1α hypoxia mesothelioma cancer syndrome HHS | NIH | National Institute of Environmental Health Sciences (NIEHS) 100000066 1R01ES030948-01 Haining YangMichele Carbone HHS | NIH | National Cancer Institute (NCI) 100000054 1R01CA237235-01A1 Haining YangMichele Carbone HHS | NIH | National Cancer Institute (NCI) 100000054 1R01CA198138 Haining YangMichele Carbone U.S. Department of Defense (DOD) 100000005 W81XWH-16-1-0440 Harvey PassHaining YangMichele Carbone National Science Foundation (NSF) 100000001 PHY-2019745 Qian WangFang BaiJosé N. Onuchic National Science Foundation (NSF) 100000001 PHY-2019745 Qian WangFang BaiJosé N. Onuchic Welch Foundation (The Welch Foundation) 100000928 Grant C-1792 Qian WangFang BaiJosé N. Onuchic ==== Body pmcBRCA1-associated protein 1 (BAP1) is a deubiquitylase that modulates DNA repair by homologous recombination, chromatin assembly, transcription, intracellular calcium (Ca2+) homeostasis, different mechanisms of cell death, and mitochondrial metabolism (1–3). The cells of carriers of heterozygous germline BAP1 mutations (BAP1+/−) contain about 50% of the amount of BAP1 found in BAP1 wild-type (BAP1WT) individuals, levels that are insufficient for the normal biological activities of BAP1 (4, 5). BAP1+/− carriers are therefore affected by the BAP1 cancer syndrome, and close to 100% of them develop one or more cancers during their lifetime (1–3, 6). About 30% of BAP1+/− carriers developed diffuse malignant mesothelioma, a malignancy of the pleura, peritoneum, and/or, rarely, pericardium (1). Germline BAP1 mutations are transmitted in a Mendelian fashion; hence, multiple cases of mesothelioma are seen in affected families (1, 6–9). The critical causative role of BAP1 mutations in mesothelioma is underscored by the finding that acquired (somatic) mutations are found in ~ 60% of sporadic mesotheliomas (1, 10). Although mesothelioma can develop in patients affected by other tumor predisposition syndromes caused by inactivating heterozygous germline mutations of TP53, BRCA2, BLM, etc., mesotheliomas in these syndromes are rare (1, 11–14). Instead, 30% of carriers of germline BAP1 mutations have developed mesothelioma, underscoring the key role of BAP1 in preventing the malignant transformation of mesothelial cells (1, 6). In addition to mesothelioma, carriers of germline BAP1 mutations develop other malignancies, among them uveal and cutaneous melanomas, and clear cell renal cell carcinomas (ccRCC) are the most frequent. Indeed, several patients develop multiple malignancies during their life. For a detailed description of the BAP1 cancer syndrome as well as of the molecular pathways altered by BAP1 mutations, please see ref. 1. Sporadic mesotheliomas are polyclonal malignancies (15) not linked to germline mutations, mostly caused by exposure to asbestos, and have a dismal median survival of 6 to 24 mo from diagnosis (1, 16–18). Asbestos-induced signature mutations have not been demonstrated in mesothelioma. Mesothelioma is largely resistant to current therapies (16–18). Therapies based on promising experiments in rodents have not been successfully translated to patients (19–23). Therefore, there is an urgent need to identify novel effective therapies (16). Intriguingly, patients with sporadic mesothelioma whose cancer cells carry somatic biallelic BAP1 mutations may have improved survival of about 1 y compared to mesotheliomas with BAP1WT (24–26). Moreover, when carriers of germline BAP1 mutations develop mesothelioma, they have a significantly improved median survival of 6 to 7 y; some of them have survived mesothelioma and died of other causes 20+ y later (1, 6, 12, 13, 16, 27, 28). In summary, germline BAP1 mutations predispose carriers to developing mesothelioma; however, these same mutations, especially when present in both the tumor cells (biallelic mutations) and in the non-malignant cells that include the tumor microenvironment (heterozygous mutations), render mesotheliomas less aggressive and possibly more sensitive to chemotherapy (26). Why? The answer to this question is critical to design novel effective targeted therapies for all mesothelioma patients. Mesothelioma causes patient demise largely by invading nearby tissues and organs and compromising vital functions; metastases occur late in the course of the disease and are rarely the cause of death (16, 29). Mesotheliomas developing in carriers of germline BAP1 mutations characteristically grow over the surface of the lungs and nearby organs: invasion is limited and occurs late in the course of the disease (6). Tumor invasion requires that malignant cells acquire the ability to grow in conditions of hypoxia, a process mainly regulated by the hypoxia-inducible factor-1 (HIF-1). The activity of HIF-1 is dependent on a heterodimer formed by an oxygen-dependent α (HIF-1α) subunit, and an oxygen-independent constitutively expressed β subunit (HIF-1β). In normoxia, HIF-1α is targeted by prolyl hydroxylases; once hydroxylated, HIF-1α is recognized by the von-Hippel Lindau E3 ligase (VHL), ubiquitinylated, and targeted for proteasomal degradation. VHL modulates the rapid (~5 min) clearance of HIF-1α in normoxia, while an oxygen-independent slower degradation of HIF-1α further regulates HIF-1α, mainly in hypoxia (30). Hypoxia stabilizes HIF-1α resulting in its nuclear translocation where it forms an active heterodimer with HIF-1β (31). The HIF-1α/HIF-1β dimer (HIF-1) modulates transcription of over 1,000 genes, including anti-apoptotic and pro-angiogenic factors (31) that promote mesothelioma growth (32). As a result of HIF-1 transcriptional activity, cells undergo metabolic reprogramming from oxidative phosphorylation to glycolysis (Warburg effect) and produce biosynthetic intermediates required for the synthesis of NADPH, nucleotides, lipids, and ATP that support tumor cell growth (33, 34). In summary, HIF-1α activity facilitates the invasion of nearby tissues and metastases by allowing cancer cells to grow and survive in a hypoxic environment (34). The oxygen-dependent mechanisms that cause HIF-1α degradation and the genes that suppress HIF-1α in hypoxia have been studied in detail (10, 34), while the gene products that by facilitating HIF-1α expression and activity in hypoxia influence tumor invasion and metastases, remain largely unknown. We reported that reduced BAP1 levels increase aerobic glycolysis (5). Because aerobic glycolysis is strongly linked to HIF-1α activation (33, 34), we investigated whether BAP1 inactivating mutations might induce HIF-1α activity, which in turn promotes glycolysis and tumor cell growth. We found the opposite to be true: BAP1+/− primary cells had reduced HIF-1α levels in hypoxia, and mesothelioma cells carrying biallelic BAP1 mutations almost constantly lost nuclear HIF-1α. We discovered that BAP1 binds, deubiquitylates, and stabilizes HIF-1α, an effect best seen in hypoxia. In summary, we discovered that BAP1 is a critical positive regulator of HIF-1α activity in hypoxia; therefore, when BAP1 is mutated the levels of HIF-1α are significantly reduced. Our results suggest that the improved prognosis observed in mesotheliomas carrying biallelic BAP1 mutations, and particularly in those developing in carriers of germline BAP1 mutations, may be linked to reduced HIF-1α levels in the tumor cells and in the microenvironment. Results Reduced BAP1 Activity Causes Decreased HIF-1α Protein Levels. Immunostaining is considered the most sensitive and specific methodology to detect biallelic BAP1 mutations, and it is widely used in the differential diagnosis of mesothelioma (10). Nuclear staining is evidence of BAP1WT, while the absence of nuclear staining is evidence of mutated, inactive BAP1 (1, 6, 16, 24, 35). We analyzed BAP1 and HIF-1α using immunohistochemistry in 49 human mesothelioma biopsies obtained from the National Mesothelioma Virtual Bank (NMVB) (Fig. 1A). BAP1 nuclear staining was present in 14 BAP1WT biopsies and absent in 35 mutated biopsies. HIF-1α nuclear staining was present in 12 (86%) WT biopsies and absent in 26 (74%) mutated biopsies [χ2(1) = 14.7, P = 0.0001] (SI Appendix, Fig. S1A). In the same biopsies negative for BAP1 and HIF-1α nuclear staining, the non-malignant nearby mesothelial cells, forming the single cell layer known as “pleura,” showed positive nuclear staining for both BAP1 and HIF-1α (internal positive control) (Fig. 1A). These findings suggested that loss of BAP1 might result in loss of HIF-1α nuclear functions. Additional staining of the only two available mesothelioma biopsies from patients carrying germline BAP1 mutations revealed the absence of nuclear staining for both BAP1 and HIF-1α and also reduced to undetectable expression of HIF-1α in the tumor microenvironment compared to tumors with BAP1WT (SI Appendix, Fig. S1B). Because IHC is not a precise test to quantify differences in protein expression, these findings will need to be validated by Western blot analyses of frozen biopsies of mesotheliomas developing in carriers of germline BAP1 mutations, which were not available to us. Fig. 1. Reduced BAP1 protein levels correlate with reduced HIF-1α protein levels. (A) Representative HIF-1α immunostaining in human pleural mesothelioma (MM) biopsies. Left: Note the nuclear staining for HIF-1α and BAP1 in BAP1 WT MM cells infiltrating the chest wall, black arrows. Right: Note the absence of HIF-1α and BAP1 nuclear staining in infiltrating BAP1-mutated MM cells, red arrows. Note that the normal nearby mesothelial cells (pleura) retain BAP1 and HIF-1α nuclear staining, green arrows. Magnification 400×; (Scale bar: 50 μm.) (B) Immunoblot, showing that BAP1 silencing in primary HM cells leads to reduced HIF-1α protein levels after 12 h incubation in 1% O2. Primary human mesothelial (HM) cells were transfected with control scrambled siRNA or siBAP1 (a pool of four different siRNAs targeting BAP1 mRNA: siBAP1#1, siBAP1#2, siBAP1#3 and siBAP1#5) and incubated in normoxia (N) or hypoxia (1% O2) for 12 h. (C) Densitometric analysis of HIF-1α protein levels normalized to α-tubulin in BAP1-silenced HM relative to scrambled (scr) control (100%); data shown as mean ± SD of n = 4 biological replicates, from four independent experiments. (D) Immunoblot: time course showing reduced HIF-1α protein levels in BAP1-silenced primary HM incubated in 1% O2 for the indicated time. (E and F): Immunoblot: reduced amounts of HIF-1α protein in carriers of germline BAP1 mutations. Total cell lysates of primary fibroblasts from 6 W (E) and 6 L (F) family members with or without BAP1 mutations, matched by gender and age. The W and L families are two families carrying BAP1+/− that we have been studying for ~20 y (8) (G) Densitometric analysis of HIF-1α protein levels normalized to α-tubulin from (E and F); densitometry of bands in BAP1+/− fibroblasts is expressed relative to BAP1WT fibroblasts (100%), matched by gender and age; data shown as mean ± SEM of n = 6 biological replicates per condition, representative of six independent experiments. (H) Quantitative PCR analysis of HIF1A mRNA expression levels normalized using the geometrical mean of 18S and ACTB reference genes, in BAP1WT and BAP1+/− fibroblasts. mRNA expression levels in BAP1+/− fibroblasts are expressed relative to BAP1WT. Data shown as mean ± SD of n = 6 biological replicates per condition, representative of six independent experiments. (I) Immunoblot: time course analysis showing reduced HIF-1α levels in BAP1+/− compared to BAP1WT fibroblasts incubated in 1% O2 for the indicated amount of time. In B, D, E, F, and I, decimals indicate the amounts of HIF-1α relative to α-tubulin as per densitometry. P value calculated using two-tailed unpaired Welch's t test, **P < 0.01. Because mesothelioma cells carry many gene mutations and gene rearrangements (36) that could influence these results, we studied the possible link between BAP1 and HIF-1α in primary human mesothelial cells (HM). We incubated primary human mesothelial (HM) cells in 1% oxygen (O2) for 12 h, which is the hypoxic conditions to induce HIF-1α and found that HM cells transfected with siRNA targeting BAP1 mRNA (siBAP1) contained a significantly reduced amount of HIF-1α protein compared with control HM transfected with scrambled siRNA (Fig. 1 B and C). Reduced HIF-1α protein levels in BAP1 silenced HM were reproducibly observed at 3, 6, 12, and 24 h of incubation in 1% O2 (Fig. 1D). In addition, we observed a direct correlation between reduced BAP1 and HIF-1α protein levels in primary fibroblast cells we established from skin biopsies from six individuals carrying inherited heterozygous germline BAP1-inactivating mutations (BAP1+/−), compared to six age- and sex-matched wild-type BAP1 (BAP1WT) control family members, from two separate families: the Wisconsin (W) family and the Louisiana (L) family (4). When incubated in 1% O2 for 12 h, fibroblasts from BAP1+/− carriers from the W family (Fig. 1E) and the L family (Fig. 1F) contained significantly less HIF-1α protein compared with their age- and sex-matched BAP1WT controls from the same families, respectively (Fig. 1G). This mechanism was not regulated transcriptionally (Fig. 1H). Time course experiments in which total cell homogenates and RNAs were collected in parallel after 3, 6, 12, and 24 h of incubation in 1% O2 confirmed that HIF-1α protein levels were always reduced in fibroblasts from BAP1+/− carriers incubated in 1% O2 compared to the BAP1WT controls (Fig. 1I), while no significant changes were detected in HIF1A mRNA levels (SI Appendix, Fig. S1C). Reduced HIF-1α protein levels in BAP1+/− carriers were also observed in fibroblasts incubated in 0.1% O2 for 24 h (SI Appendix, Fig. S1D). To confirm that BAP1 regulates HIF-1α, we transduced BAP1+/− fibroblasts with human adenoviruses expressing GFP and BAP1WT for 24 h and cultured these cells in normoxia and in hypoxia for 6 h. Compared to the BAP1+/− fibroblasts transduced with the Ad-GFP control, which maintain about 50% of BAP1 activity, BAP1+/− fibroblasts transduced with Ad-BAP1 restored fully functional BAP1 and these cells displayed similar levels of HIF-1α as those observed in BAP1WT fibroblasts in hypoxia (SI Appendix, Fig. S1E). Therefore, BAP1 modulates HIF-1α expression in hypoxia. BAP1 Interacts with HIF-1α. Co-immunoprecipitation (CoIP) and proximity ligation assay (PLA) experiments revealed that 1) HIF-1α and BAP1 bind to each other and co-precipitate (Fig. 2A), and 2) the nuclei of BAP1WT cells contained significantly more PLA positive signals—evidence of BAP1 and HIF-1α interaction—than BAP1+/− cells (Fig. 2 B and C). Fig. 2. BAP1 binds HIF-1α. (A) HIF-1α and BAP1 co-precipitate. CoIP of endogenous HIF-1α and BAP1, in BAP1WT fibroblasts grown in normoxia (N) or hypoxia (1% O2) for 4 h, using BAP1 as a bait. (B and C) PLA: red dots demonstrate the BAP1–HIF-1α interaction in the nuclei of BAP1WT and BAP1+/− fibroblasts incubated in 1% O2 for 6 h. Nuclei stained blue with DAPI (B); (Scale bar: 5 μm.) Bar graph: quantification of PLA red dots per cell showing reduced BAP1–HIF-1α interaction in BAP1+/− fibroblasts. Data shown as mean ± SD (n = 20 cells per condition) (C). (D and E) Mapping of the BAP1–HIF-1α interaction. The deletion of the CTD-NLS BAP1 domain –as observed in individuals of the W and L families, greatly reduces the interaction with HIF-1α. (D) CoIP of HIF-1α and BAP1 in homogenates from HEK-293 co-transfected with HA-tagged HIF-1α and Myc-tagged [displayed on top (4)], BAP1, catalytic inactive (C91S), L family truncated mutant, W family truncated mutant, using anti-HA resin (E) The CTD-NLS domain of BAP1 is the major contributor to the interaction with HIF-1α, while the fragment consisting of the UCH together with the NORS domains binds to a minor extent. CoIP of HIF-1α and BAP1 in homogenates from HEK-293 co-transfected with HA-tagged HIF-1α and Myc-tagged BAP1, and the Myc-tagged BAP1 fragments displayed on top (4) using anti-Myc resin. We further investigated the specificity of the BAP1 interaction with HIF-1α in HEK-293 cells expressing Myc-BAP1 and HA-HIF-1α, using HA-Tag as bait. We found that the Myc-tagged truncated mutant proteins BAP1(W) and BAP1(L) (8) lose the ability to bind HIF-1α completely (W) or almost completely (L), while the full-length BAP1 and the catalytically inactive BAP1 mutant (C91S) (37) interact with HIF-1α (Fig. 2D). Deletion fragments of BAP1 (4) revealed that its C-terminal portion, consisting of the C-terminal domain (CTD) and nuclear localization signal (NLS), is key to the interaction with HIF-1α (Fig. 2E). The fragment consisting of the ubiquitin C-terminal hydrolase (UCH) with the non-regular secondary structure (NORS) domains binds to a minor extent (Fig. 2E), explaining why the BAP1(L) and BAP1(W) truncated mutants have lost or have reduced ability to bind HIF-1α, respectively. We established a computational model of the binding complex of BAP1 and HIF-1α. The structural predictions of the BAP1(CTD-NLS) are highly converged with three different methods: coarse-grained molecular dynamic simulations (38), the I-TASSER web server (39–41) and the RaptorX web server (42) (SI Appendix, Fig. S2A). For all models, residues 637 to 698 in the CTD form three consecutive helical fragments; in contrast, the full NLS domain is highly disordered (SI Appendix, Fig. S2A). To study the binding between BAP1 and HIF-1α, first, a rigid docking protocol was applied to model the binding complex of the CTD of BAP1 (the NLS domain is removed due to its flexibility) and HIF-1α by using the ClusPro server (43–45) (SI Appendix, Fig. S2B). We identified residues 1 to 73 of HIF-1α as the main binding interface for BAP1. Consistently, RaptorX, a server utilizing co-evolutional information of proteins and deep learning techniques (46), also predicts that residue 1 to 73 of HIF-1α can form contacts with BAP1 with high probability (SI Appendix, Fig. S2C). Therefore, we focused on this region (noted as HIF-1α-r73). We used coarse-grained molecular dynamic simulations to model the binding complex of BAP1 (CTD-NLS) and HIF-1α-r73 (Fig. 3A), as well as understanding the binding kinetics (Fig. 3B). The NLS domain of BAP1 binds to HIF-1α-r73 and the thermodynamic stability of the binding complex increases through electrostatic interactions with DNA. Because the NLS domain is highly disordered, it appears as an extended structure and thus greatly increases the searching range of BAP1 during the binding process. For all simulated trajectories that successfully lead to the correct binding complex, the NLS domain of BAP1 binds to HIF-1α-r73-DNA ahead of the CTD. This suggests that the binding between BAP1 and HIF-1α is facilitated by the “fly-casting” mechanism (47, 48). Once the NLS domain of BAP1 binds to the DNA, it serves as an anchor to increase the local concentration of the CTD of BAP1 near HIF-1α, which helps the CTD bind sequentially (Fig. 3B). Notably, BAP1 binding to HIF-1α-r73-DNA does not require HIF-1β for the interaction. The critical role of the NLS domain of BAP1 is supported by the experimental fact that removing this domain greatly decreases the binding to HIF-1α (Fig. 2 D and E). Fig. 3. The CDT-NLS domain of BAP1 interacts with residues 1 to 73 of HIF-1α. (A) Structural modeling for the binding complex of BAP1(CTD-NLS) and HIF-1α (1-73) (residues 1 to 73 of HIF-1α) in the presence of DNA. The CTD of BAP1 is colored in blue, the NLS domain of BAP1 is colored in green, HIF-1α is colored in red, DNA is colored in orange and grey; three interacting regions are marked by light silver circles. (B) Coarse-grained molecular dynamic simulations to model the binding complex of BAP1(CTD-NLS) and HIF-1(1-73). The NLS domain (colored in green) is extended to increase the searching range of BAP1 to bind to HIF-1α. The NLS domain of BAP1 binds to HIF-1α first. Then the CTD binds sequentially. (C) HA-tagged HIF-1α fragments and HIF-1α domains: basic-helix-loop-helix motif (bHLH) protein, two Per and Sim (PAS) domain (A and B), oxygen-dependent degradation domain (ODDD), two transactivation domains (TAD): NH2-terminal (N-TAD) and COOH-terminal (C-TAD), intervening inhibitory domain (ID). (D) BAP1 binds to the N terminus region of HIF-1α [HIF-1α(2-400)]. Residues 1 to 73 of HIF-1α are essential for the interaction because HIF-1α(74-826) did not bind BAP1. CoIP of BAP1 and HIF-1α in homogenates from HEK-293 co-transfected with Myc-BAP1 and HA-tagged HIF-1α or the HA-tagged HIF-1α fragments displayed in (C), or the empty vector (mock); anti-Myc resin was used as bait. (E) Point Mutations of residues I675, F678, I679, and L691 of BAP1 abolish the interaction with HIF-1α. CoIP of BAP1 and HIF-1α in homogenates from HEK-293 co-transfected with Myc-BAP1 or Myc-BAP1(mut) (in which residues I675, F678, I679, L691 of BAP1 are mutated to Alanine), and HA-tagged HIF-1α or empty vector (mock); anti-Myc resin was used as bait. (F) The simultaneous mutation of residues I675, F678, I679, L691 of BAP1 abolish the interaction with HIF-1α, while single-point mutations decrease but do not abolish this interaction. CoIP of BAP1 and HIF-1α in homogenates from HEK-293 co-transfected with HA-tagged HIF-1α and Myc-BAP1, or Myc-BAP1(mut) (in which residues I675, F678, I679, L691 of BAP1 are mutated to Alanine), or Myc-BAP1 mutants carrying each individual point mutation; anti-Myc resin was used as bait. CoIP experiments in cells co-transfected with full-length Myc-tagged BAP1 (Myc-BAP1) and HA-tagged full-length HIF-1α, or HIF-1α fragments covering residues 74 to 826 [HIF-1α(74-826)], 2-400 [HIF-1α(2-400)], 401-826 [HIF-1α(401-826)] (Fig. 3C), confirmed that BAP1 binds to the N terminus region of HIF-1α [HIF-1α(2-400)] (Fig. 3D). As predicted by our computational model, residues 1 to 73 of HIF-1α are essential for the interaction because HIF-1α(74-826) did not bind BAP1 (Fig. 3D). The binding interfaces between BAP1(CTD-NLS) and HIF-1α-r73 in the presence of DNA include three parts: 1) residues K656, R657, K658 and K659 of BAP1(CTD-NLS) insert into the major groove of DNA through electrostatic interactions; 2) residues I675, F678, I679 and L691 of BAP1(CTD-NLS) form the hydrophobic core with residues F37, L40, Q43, L44 of HIF-1α-r73; 3) the NLS domain of BAP1 inserts into the major groove of DNA through electrostatic interactions. Among those residues, we found that the ones in BAP1 are more critical. Mutating those residues to A [BAP1(mut)] significantly decreases the binding stability; in contrast, mutating F37, L40, Q43, L44 of HIF1α-r73 to A [HIF1α(mut)] only has a minor effect (SI Appendix, Fig. S2D). The accuracy of this model is supported by CoIP experiments revealing that mutations of residues I675, F678, I679, and L691 of BAP1 (CTD-NLS) to A abolish the interaction with HIF-1α (Fig. 3E), while point mutations of residues F37, L40, Q43, L44 of HIF-1α-(r73) to A did not affect the binding with BAP1 (SI Appendix, Fig. S2E). All four residues forming the hydrophobic core of BAP1 must be mutated to completely abolish the binding of HIF-1α, while in the presence of single point mutations BAP1 interaction with HIF-1α is decreased but not entirely abolished (Fig. 3F). We verified that mutating four residues will not significantly change the structure of BAP1 (CTD-NLS) (SI Appendix, Fig. S2F). We concluded that the hydrophobic core formed by I675, F678, I679, L691 of BAP1 and F37, L40, Q43, L44 of HIF1α is sufficient to maintain the binding between the two proteins. BAP1 Interacts with HIF-1α and HIF-1β Independently of DNA. Aligning the crystal structure of HIF-1α-HIF-1β complex (PDB ID: 4zpr) (49) to our structural model for the binding complex of BAP1-HIF-1α reveals the significance of residue 1 to 73 of HIF-1α, as this region binds to both BAP1 and HIF-1β (Fig. 4A). Therefore, we checked whether BAP1 could also bind to HIF-1β. The deletion fragments of BAP1 (4) revealed that its NORS and CDT-NLS domains are the major contributors to the interaction with HIF-1β (Fig. 4B). CoIP experiments in cells co-transfected with full length Myc-tagged BAP1 (Myc-BAP1) and Flag-tagged full-length HIF-1β, or HIF-1β fragments covering residues 2 to 470 [HIF-1β(2-470)], 142-470 [HIF-1β(142-470)], 471-789 [HIF-1β(471-789)], 582-789 [HIF-1β(592-789)] (Fig. 4C), showed that BAP1 binds to the N terminus region of HIF-1β, specifically to the DNA binding and dimerization region [HIF-1β(2-470) and HIF-1β (142-470)] (Fig. 4D). Fig. 4. BAP1 binding to HIF-1α and HIF-1β does not require DNA. (A) Shared binding region among BAP1, HIF-1α, and HIF-1β. The CTD of BAP1 is colored in blue, the NLS domain of BAP1 is colored in green, HIF-1α is colored in red, DNA is colored in orange and grey, and HIF-1β (colored in yellow) is docked onto the binding complex of BAP1 and HIF-1α by utilizing the crystal structure of HIF-1α–HIF-1β (PDB ID: 4zpr) (49). Missing residues of the crystal structure are added by the SWISS-MODEL server (50). (B) HIF-1β interacts with the NORS and CTD-NLS domain of BAP1. CoIP of HIF-1β and BAP1 in homogenates from HEK-293 co-transfected with Flag-tagged HIF-1β and Myc-tagged BAP1 and the Myc-tagged BAP1 fragments displayed in Fig. 2E (4), using anti-Myc resin. (C) Flag-tagged HIF-1β fragments and HIF-1β domains: basic-helix-loop-helix motif (bHLH) protein, two Per and Sim (PAS) domain (A and B), and COOH-terminal transactivation domain (C-TAD). (D) CoIP of BAP1 and HIF-1β in homogenates from HEK-293 co-transfected with Myc-BAP1 and Flag-HIF-1β or the Flag-HIF-1β fragments displayed in (C), or the empty vector (mock); anti-Myc resin was used as bait. (E) HEK-293 cells were grown in normoxia (N) or hypoxia (1% O2) for 4 h. Cell homogenates were collected, treated with benzonase for 15, 30 or 60 min (SI Appendix, Fig. S3), and then used to co-immunoprecipitate endogenous HIF-1α and HIF-1β using BAP1 as bait. (F) Computational binding free energy profile between BAP1 and HIF-1α in the absence of DNA; the result indicates that the binding complex formed by BAP1 and HIF-1α can still hold when DNA is absent (the binding free energy ~ 3 kcal/mol). We tested the hypothesis that although DNA facilitates the binding between BAP1 and HIF-1α, this binding complex still holds in the absence of DNA. Total cell homogenates of cells grown in normoxic or 1% O2 (hypoxic) conditions were incubated with benzonase for 15, 30, or 60 min, to achieve complete DNA degradation (SI Appendix, Fig. S3). Subsequently, endogenous BAP1 was used as bait to co-immunoprecipitate endogenous HIF-1α and HIF-1β (Fig. 4E). These results show that BAP1 can interact with HIF-1α and HIF-1β even without DNA. The computational analysis of the binding free energy profile between BAP1 and HIF-1α in the absence of DNA confirmed that the complex of BAP1-HIF-1α holds even without DNA, with a binding free energy of ~ 3 kcal/mol (Fig. 4F). Identification of HIF-1α as a Substrate of BAP1. It has been reported that the ubiquitin-proteasome pathway regulates the degradation of HIF-1α (51, 52). Since BAP1 is a member of the UCH subfamily of deubiquitylating enzymes (1), we investigated whether BAP1 deubiquitylates and stabilizes HIF-1α. We measured the ubiquitylation levels of exogenously expressed HIF-1α in cells co-transfected with HA-tagged ubiquitin (HA-Ub), Flag-HIF-1α and Myc-BAP1. CoIP of Flag-HIF-1α showed reduced ubiquitin levels when cells overexpressed BAP1, but not in cells overexpressing the catalytic inactive BAP1(C91S), compared to mock control (Fig. 5A). In vitro de-ubiquitylation assays using purified recombinant proteins confirmed increased deubiquitylation of HIF-1α in the presence of BAP1, while in the presence of BAP1(C91S), BAP1(L), and BAP1(W) HIF-1α deubiquitylation was comparable to mock control (Fig. 5B). Together, these results demonstrated that BAP1 deubiquitylates and thus stabilizes HIF-1α. Fig. 5. BAP1 Deubiquitylates HIF-1α. (A) Reduced endogenous ubiquitylation of HIF-1α in HEK-293 cells co-transfected with Flag-tagged HIF-1α and Myc-tagged BAP1, catalytic inactive (C91S), or mock. Cells were treated with 10 µM MG-132 for 3 h, then total cell homogenates were collected and HIF-1α immunoprecipitated using anti-Flag resin. Ubiquitylation levels of the immunocomplexes were detected using an anti-Ub-HRP antibody and normalized on the total amount of Flag-HIF-1α immunoprecipitated (decimals indicate the ratio as per densitometric analysis). (B) Western blot analysis of in vitro ubiquitylation/de-ubiquitylation assay. HA-HIF-1α ubiquitylated in vitro, and subsequently incubated with immunopurified Myc-BAP1, Myc-BAP1(C91S), Myc-BAP1(L), Myc-BAP1(W), or mock, for 1 h. Ubiquitylation levels were detected using an anti-Ub-HRP antibody and normalized on the total amount of HA-HIF-1α (decimals indicate the ratio as per densitometric analysis). Discussion We discovered that BAP1 binds and deubiquitylates HIF-1α, contributing to the high levels of HIF-1α in hypoxia (Figs. 1–5). Accordingly, primary cells we derived from carriers of germline heterozygous BAP1 mutations, as well as cells in which we downregulated BAP1 using siRNA, and mesothelioma biopsies containing tumor cells with biallelic BAP1 inactivation, displayed significantly reduced levels of HIF-1α and loss of nuclear HIF-1α compared to normal cells or tumor cells with BAP1WT (Fig. 1). Therefore, our data suggest that BAP1 is a key regulator of HIF-1α and its tumor-promoting activities. In previous studies performed in normoxic conditions, we demonstrated that BAP1 regulates intracellular Ca2+ flux by binding and deubiquitylating, and thus stabilizing the IP3R3 receptor (4). Therefore, BAP1 deubiquitylating activity appears to remain active in both conditions, normoxia and hypoxia. We found that BAP1 also binds to the N terminus region of HIF-1β, specifically to the DNA binding and dimerization region (Fig. 4). The crystal structure of HIF-1α-HIF-1β complex (PDB ID: 4zpr) (49) shows that without BAP1, HIF-1α, and HIF-1β bind to DNA (49), thus BAP1 is not required for HIF-1α-HIF-1β complex formation. Aligning the crystal structure of HIF-1α-HIF-1β complex (PDB ID: 4zpr) (49) to our structural model for the binding complex of BAP1-HIF-1α showed that both BAP1 and HIF-1β bind to the same residues of HIF-1α (1-73) on the DNA; however, in Fig. 3B, we demonstrate that BAP1, HIF-1α and the DNA form a complex without HIF-1β. In addition, we show that after total degradation of DNA, BAP1 remains bound to both HIF-1α and HIF-1β (Fig. 4A). Therefore, our data indicate that BAP1 is not required for HIF-1α-HIF-1β complex formation to functionally bind to DNA, that HIF-1β is not required for BAP1-HIF-1α complex formation to functional binding to DNA, and that although DNA facilitates the binding of BAP1 and HIF-1α, it is not required to maintain the binding of both BAP1-HIF-1α and BAP1-HIF-1β. In summary, our data suggest that BAP1 directly binds and stabilizes both HIF-1α and HIF-1β increasing their intra-nuclear availability for dimer formation, thus fine-tuning HIF activities required to support malignant cell growth. So far, the pathogenic variants reported in ClinVar for both HIF-1α and HIF-1β are not located among the crucial residues of HIF-1α-r73 where HIF-1α can bind to BAP1, HIF-1β and DNA, or of HIF-1β (2-470) where HIF-1β can bind to BAP1, HIF-1α, and DNA. This finding suggests that tumor cell clones that may acquire HIF-1α and/or HIF-1β mutations that impair their binding to BAP1 may be negatively selected compared to tumor cell clones expressing HIF-1α and HIF-1β that maintain the capacity to bind to BAP1. HIF-1α is the master regulator of cell growth in hypoxia (33, 34). HIF1 activity is regulated by the interaction of HIF-1α with >100 other proteins (53). Among them, VHL plays a key role by recruiting an E3-ubiquitin ligase complex to mediate HIF-1α protein degradation in normoxia. Biallelic BAP1 mutations occur in several human cancers (1); their tumor cells, based on our data studying mesothelioma, should contain reduced HIF-1α levels. However, this might not remain true in malignancies in which the VHL gene (34)—or other genes that suppress HIF-1α (10)—are also mutated and thus display constitutively high levels of HIF-1α, which may overrun the fine-tuning BAP1 deubiquitylating activity. In addition to VHL, which is active in normoxia, other proteins mediate the ubiquitylation of HIF-1α in hypoxia. The UCH-L1 (UCHL1) is a deubiquitylase that has been shown to positively modulate HIF-1α levels (54). Our data identified BAP1 as a deubiquitylase that binds and inhibits the degradation of HIF-1α, an effect best observed in hypoxia. BAP1 shares 23% sequence homology with UCHL1 (55). UCHL1 hydrolyzes the C-terminal peptide tails of small ubiquitin derivatives but cannot hydrolyze large ubiquitin chains because of short active site crossover loops. BAP1 instead has long crossover loops and thus can process polyubiquitin chains (55, 56). Thus, UCHL1 and BAP1 are both independently required for HIF-1α stabilization and activities. UCHL1 and BAP1 were both identified as deubiquitylases for γ-tubulin through screening a siRNA library of deubiquitylases; however, when both UCHL1 and BAP1 were depleted using siRNA, the degradation of γ-tubulin was comparable to the γ-tubulin levels after either BAP1 or UCHL1 silencing alone (57). Future studies shall address whether these two ubiquitin hydrolases interact in modulating HIF-1α levels in hypoxia and whether their effects are cell type specific. It has been proposed that targeting UCHL1 might reduce HIF-1α stabilization and impair tumor growth (54). Our data point to BAP1 as a novel target to reduce HIF-1α tumor-promoting activity in malignancies with elevated HIF-1α levels and intact VHL. We identified the nucleotides responsible for the binding between BAP1 and HIF-1α and BAP1 and HIF-1β. Previous studies using the HIF-1α inhibitor YC-1 (58) or siRNAs targeting HIF-1α in mesothelioma cells in tissue culture (59), revealed increased apoptosis; however, the authors suggested that an additional blockade was required to inhibit growth signals completely. We are designing small molecules to test the hypothesis that their intra-pleural administration alone or together with HIF-1α inhibitors, will interfere with the binding among BAP1 and HIF-1α and cause HIF-1α degradation. It is hoped that reduced HIF-1α activity will impair mesothelioma growth and increase susceptibility to therapy, as observed in patients carrying germline BAP1 mutations and tumors with biallelic BAP1 inactivating mutations (6). Mesotheliomas have large areas of hypoxia (60). The activity of HIF-1α-induced metabolic reprogramming provides malignant cells with maximal growth support in a hypoxic tumor microenvironment. Therefore, the reduced levels of HIF-1α in BAP1-mutated tumor cells may contribute to the reduced tumor aggressiveness of BAP1-mutant mesotheliomas, compared to mesotheliomas with BAP1WT (6, 12, 13, 27, 28). Mesotheliomas developing in carriers of germline BAP1 mutations invariably carry biallelic inactivating BAP1 mutations (BAP1−/−), easily detectable by the absence of nuclear BAP1 staining, while the cells forming the tumor microenvironment carry heterozygous germline BAP1 mutations (BAP1+/−) (1, 8, 14). About ~60% of sporadic mesotheliomas carry somatic (acquired) biallelic inactivating BAP1−/−; however, the cells forming the tumor microenvironment are BAP1WT (1, 8, 14). Our hypothesis is that in sporadic BAP1−/− mesotheliomas, BAP1 loss results in reduced HIF-1α in the malignant cells; however, the surrounding hypoxic tumor cell microenvironment comprised of BAP1WT cells will maintain stable HIF-1α levels that sustain tumor cell invasion. Conversely, the hypoxic tumor microenvironment of mesotheliomas developing in patients carrying germline BAP1 mutations express reduced HIF-1α. Accordingly, these patients have less invasive tumors. This hypothesis, based on our in vitro experiments (Fig. 1 E–I), was supported by IHC analyses in which we studied mesothelioma biopsies from patients carrying germline BAP1 mutations. In their mesothelioma biopsies, IHC showed undetectable HIF-1α expression in the tumor cells and reduced HIF-1α expression in the cells forming the tumor microenvironment (SI Appendix, Fig. S1B), compared to sporadic BAP1−/− mesothelioma biopsies which maintained HIF-1α expression in the surround BAP1WT cells (Fig. 1A). Altogether, these data suggest that reduced HIF-1α levels may contribute to the reduced aggressiveness of mesothelioma in carriers of germline BAP1 mutations (6, 12, 13, 27, 28). Reduced HIF-1α levels may also render mesothelioma cells more susceptible to cell death in hypoxia and this could contribute to the reported increased response to chemotherapy of BAP1 mutated mesotheliomas (26), and in those that develop among carriers of germline BAP1 mutations, in particular (6). BAP1 mutations are not associated with an improved prognosis in uveal melanoma (UVM) and ccRCC, the other two malignancies that, together with mesothelioma, most often carry BAP1 mutations (1). Moreover, the loss of BAP1 expression has been detected together with increased expression of HIF-1α in UVM (61) and ccRCC (62, 63). These results appear to contradict our findings. However, in about 90% of ccRCC, the initiating event is the inactivation of the VHL gene located on chromosome 3 (64). Physiologically, VHL binds to HIF-1α targeting it for ubiquitylation and proteasomal degradation, therefore, once VHL is lost, HIF-1α ubiquitylation is markedly reduced, and its levels are significantly increased (33, 34) an effect that should render the reduced deubiquitylation of HIF-1α by BAP1 in BAP1 deficient cells physiologically less relevant. The study in UVM (61) measured HIF1A mRNA levels and their relationship to BAP1 transcription. Here we report that BAP1 modulates the stability of the HIF-1α protein and that it does not regulate HIF-1α gene transcription. Moreover, similarly to ccRCC, deletions of chromosome 3 are frequent in UVM (65) with subsequent loss of VHL. In mesothelioma, instead, nucleotide level deletions as well as minute deletions of 100 to 300 bp are frequent throughout the BAP1 gene located on chromosome 3p, and nearby SETD2, SMARCC1, PBRM1 genes, but deletions extending to the VHL gene are very rare (66). Therefore, the effects of reduced deubiquitylating activity of BAP1 mutations on the HIF-1α protein may be more relevant in mesothelioma compared to ccRCC and UVM in which the very frequent inactivation of VHL may result in elevated levels of HIF-1α independently from BAP1 deubiquitylating activity. Overall, our findings may help explain the opposite effects on survival in BAP1-mutated mesothelioma compared to BAP1-mutated ccRCC and UVM. Further studies in renal cell carcinomas and UVMs, compared to mesothelioma, are necessary to fully address the mechanisms and the possible relationship with HIF-1α expression in these malignancies. In summary, we report that BAP1 deubiquitylates and thus stabilizes HIF-1α in hypoxia, and, therefore, BAP1 mutations significantly reduce HIF-1α protein levels. Given the well-established role of HIF-1α in promoting tumor growth in hypoxia, we propose that the reduced aggressiveness and improved prognosis of mesothelioma in carriers of germline BAP1 mutations may result, at least in part, from the combined reduced HIF-1 activity caused by biallelic BAP1 mutations in mesothelioma cells and the presence of heterozygous BAP1 mutations in the cells that form the tumor microenvironment. Materials and Methods Subjects. BAP1+/− mutant carriers and unaffected controls were recruited from the L and W families and provided informed written consent allowing their specimens to be used for this project. The collection and use of patient information and samples were approved by the Institutional Review Board (IRB) of the University of Hawaii (IRB no. CHS14406). Technical Procedures. Cell cultures, immunohistochemistry, gene silencing, qPCR, western blot (WB), Co-IP, in vitro ubiquitylation and de-ubiquitylation assays, Duolink PLA, and computational modeling were performed according to standard techniques and are described in SI Appendix. Statistics and Reproducibility. P values were calculated using two-tailed unpaired Welch’s t test, unless otherwise specified. P values < 0.05 were considered statistically significant and marked with asterisks (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001), as indicated in the figure legends. All data collected met the normal distributions assumption of the test. Data are represented as mean ± SD, unless otherwise specified.in the figure legends. The exact sample size (n) for experimental groups/conditions and whether samples represent technical or cell culture replicates are indicated in the figure legends. The results shown are representative of experiments independently conducted three times that produced similar results. Supplementary Material Appendix 01 (PDF) Click here for additional data file. We are in debt to the patients who donated their specimens for research. We would like to acknowledge the UH Cancer Center Microscopy, Imaging, and Flow Cytometry Core and the Leica Thunder Live Cell 3D microscope SIG: NIH S10ODO028515-01 for supporting this work. We acknowledge the NMVB for providing the mesothelioma biopsies. M.C. and H.Y. have a patent issued for “Using Anti-HMGB1 Monoclonal Antibody or other HMGB1 Antibodies as a Novel Mesothelioma Therapeutic Strategy” and a patent issued for “HMGB1 As a Biomarker for Asbestos Exposure and Mesothelioma Early Detection.” M.C. and H.Y. report funding from the National Institute of Environmental Health Sciences (NIEHS) 1R01ES030948-01 (M.C. and H.Y.), the National Cancer Institute (NCI) 1R01CA237235-01A1 (M.C. and H.Y.) and 1R01CA198138 (M.C.), the US Department of Defense (DoD) W81XWH-16-1-0440 (H.Y., M.C., and H.I.P.), and from the UH Foundation through donations from the Riviera United-4-a Cure (M.C. and H.Y.), the Melohn Family Endowment, the Honeywell International Inc., the Germaine Hope Brennan Foundation, and the Maurice and Joanna Sullivan Family Foundation (M.C.). H.I.P. and H.Y. report funding from the Early Detection Research Network NCI 5U01CA214195-04. H.I.P. reports funding from Genentech, Belluck, and Fox LLP. J.N.O., F.B., and Q.W. work was supported by NSF (Grants PHY-2019745 and PHY-2019745) and by the Welch Foundation (Grant C-1792). J.N.O. is a CPRIT Scholar in Cancer Research sponsored by the Cancer Prevention and Research Institute of Texas. Author contributions J.N.O., H.Y., and M.C. designed research; A.B., Q.W., A.A.Z., F.B., M.S-T., J.S.S., S.P., A.S., V.S., A.F., F.N., J.-H.K., M.M., Y.T., L.P., A.N., R.X., C.F., C.G., G.S., G.G., and H.I.P. performed research; C.B., M.N., and I.P. analyzed data; G.S. and H.I.P. contributed surgical specimens; and A.B. and M.C. wrote the paper. Competing interest The authors have organizational affiliations to disclose. M.C. is a board-certified pathologist who provides consultation for pleural pathology, including medical-legal. The authors have patent filings to disclose. M.C. has a patent issued for “Methods for Diagnosing a Predisposition to Develop Cancer”. Data, Materials, and Software Availability All study data are included in the article and/or SI Appendix. Supporting Information Reviewers: P.P.P., Renown Institute for Cancer; and I.L.P., Northwestern University. ==== Refs 1 M. Carbone , Biological mechanisms and clinical significance of BAP1 mutations in human cancer. Cancer Discov. 10 , 1103–1120 (2020).32690542 2 E. Conway , BAP1 enhances polycomb repression by counteracting widespread H2AK119ub1 deposition and chromatin condensation. Mol. Cell 81 , 3526–3541.e8 (2021).34186021 3 L. Masclef , Roles and mechanisms of BAP1 deubiquitinase in tumor suppression. Cell Death Differ. 28 , 606–625 (2021).33462414 4 A. Bononi , BAP1 regulates IP3R3-mediated Ca(2+) flux to mitochondria suppressing cell transformation. Nature 546 , 549–553 (2017).28614305 5 A. Bononi , Germline BAP1 mutations induce a Warburg effect. Cell Death Differ. 24 , 1694–1704 (2017).28665402 6 M. Carbone , Medical and surgical care of patients with mesothelioma and their relatives carrying germline BAP1 mutations. J. Thorac. Oncol. 17 , 873–889 (2022), 10.1016/j.jtho.2022.03.014.35462085 7 M. Carbone , Combined genetic and genealogic studies uncover a large BAP1 cancer syndrome kindred tracing back nine generations to a common ancestor from the 1700s. PLoS Genet. 11 , e1005633 (2015).26683624 8 J. R. Testa , Germline BAP1 mutations predispose to malignant mesothelioma. Nature Genet. 43 , 1022–1025 (2011).21874000 9 S. Walpole , Comprehensive study of the clinical phenotype of germline BAP1 variant-carrying families Worldwide. J. Natl Cancer Inst. 110 , 1328–1341 (2018).30517737 10 R. Bernardi , PML inhibits HIF-1alpha translation and neoangiogenesis through repression of mTOR. Nature 442 , 779–785 (2006).16915281 11 A. Bononi , Heterozygous germline BLM mutations increase susceptibility to asbestos and mesothelioma. Proc. Natl. Acad. Sci. U.S.A. 117 , 33466–33473 (2020).33318203 12 S. Pastorino , A subset of mesotheliomas with improved survival occurring in carriers of BAP1 and other germline mutations. J. Clin. Oncol. 36 , JCO2018790352 (2018), 10.1200/JCO.2018.79.0352.30376426 13 V. Panou , Frequency of germline mutations in cancer susceptibility genes in malignant mesothelioma. J. Clin. Oncol. 36 , 2863–2871 (2018).30113886 14 M. Carbone , Tumour predisposition and cancer syndromes as models to study gene-environment interactions. Nat. Rev. Cancer 20 , 533–549 (2020).32472073 15 S. Comertpay , Evaluation of clonal origin of malignant mesothelioma. J. Translat. Med. 12 , 301 (2014). 16 M. Carbone , Mesothelioma: Scientific clues for prevention, diagnosis, and therapy. CA Cancer J. Clin. 69 , 402–429 (2019).31283845 17 A. S. Tsao , Current and future management of malignant mesothelioma: A consensus report from the national cancer institute thoracic malignancy steering committee, international association for the study of lung cancer, and mesothelioma applied research foundation. J. Thorac. Oncol. 13 , 1655–1667 (2018).30266660 18 L. Mutti , Scientific advances and new frontiers in mesothelioma therapeutics. J. Thorac. Oncol. 13 , 1269–1283 (2018).29966799 19 H. Yang , Aspirin delays mesothelioma growth by inhibiting HMGB1-mediated tumor progression. Cell Death Dis. 6 , e1786 (2015).26068794 20 Z. Rivera , CSPG4 as a target of antibody-based immunotherapy for malignant mesothelioma. Clin. Cancer Res. 18 , 5352–5363 (2012).22893632 21 C. M. Goparaju , Onconase mediated NFKβ downregulation in malignant pleural mesothelioma. Oncogene 30 , 2767–2777 (2011).21317924 22 M. Nasu , Ranpirnase interferes with NF-κB pathway and MMP9 activity, inhibiting malignant mesothelioma cell invasiveness and xenograft growth. Genes Cancer 2 , 576–584 (2011).21901170 23 H. I. Pass , Inhibition of hamster mesothelioma tumorigenesis by an antisense expression plasmid to the insulin-like growth factor-1 receptor. Cancer Res. 56 , 4044–4048 (1996).8752177 24 D. B. Chapel, J. J. Schulte, A. N. Husain, T. Krausz, Application of immunohistochemistry in diagnosis and management of malignant mesothelioma. Transl. Lung Cancer Res. 9 , S3–S27 (2020).32206567 25 M. Farzin , Loss of expression of BAP1 predicts longer survival in mesothelioma. Pathology 47 , 302–307 (2015).25938359 26 A. Louw , BAP1 loss by immunohistochemistry predicts improved survival to first-line platinum and pemetrexed chemotherapy for patients with pleural mesothelioma: A validation study. J. Thorac. Oncol. 17 , 921–930 (2022).35489694 27 F. Baumann , Mesothelioma patients with germline BAP1 mutations have 7-fold improved long-term survival. Carcinogenesis 36 , 76–81 (2015).25380601 28 R. Hassan , Inherited predisposition to malignant mesothelioma and overall survival following platinum chemotherapy. Proc. Natl. Acad. Sci. U.S.A. 116 , 9008–9013 (2019).30975761 29 R. M. Flores , Extrapleural pneumonectomy versus pleurectomy/decortication in the surgical management of malignant pleural mesothelioma: Results in 663 patients. J. Thorac. Cardiovasc. Surg. 135 , 620–626 (2008), 626.e1-3.18329481 30 E. Moroz , Real-time imaging of HIF-1alpha stabilization and degradation. PLoS One 4 , e5077 (2009).19347037 31 N. Ferrara, T. Davis-Smyth, The biology of vascular endothelial growth factor. Endocr. Rev. 18 , 4–25 (1997).9034784 32 P. Cacciotti , SV40 replication in human mesothelial cells induces HGF/Met receptor activation: A model for viral-related carcinogenesis of human malignant mesothelioma. Proc. Natl. Acad. Sci. U.S.A. 98 , 12032–12037 (2001).11572935 33 M. S. Nakazawa, B. Keith, M. C. Simon, Oxygen availability and metabolic adaptations. Nat. Rev. Cancer 16 , 663–673 (2016).27658636 34 G. L. Semenza, HIF-1 mediates metabolic responses to intratumoral hypoxia and oncogenic mutations. J. Clin. Invest. 123 , 3664–3671 (2013).23999440 35 M. Nasu , High incidence of somatic BAP1 alterations in sporadic malignant mesothelioma. J. Thorac. Oncol. 10 , 565–576 (2015).25658628 36 R. Bueno , Comprehensive genomic analysis of malignant pleural mesothelioma identifies recurrent mutations, gene fusions and splicing alterations. Nat. Genet. 48 , 407–416 (2016).26928227 37 D. E. Jensen , BAP1: A novel ubiquitin hydrolase which binds to the BRCA1 RING finger and enhances BRCA1-mediated cell growth suppression. Oncogene 16 , 1097–1112 (1998).9528852 38 A. Davtyan , AWSEM-MD: Protein structure prediction using coarse-grained physical potentials and bioinformatically based local structure biasing. J. Phys. Chem. B 116 , 8494–8503 (2012).22545654 39 A. Roy, A. Kucukural, Y. Zhang, I-TASSER: A unified platform for automated protein structure and function prediction. Nat. Protoc. 5 , 725–738 (2010).20360767 40 J. Yang , The I-TASSER Suite: Protein structure and function prediction. Nat. Methods 12 , 7–8 (2015). 41 J. Yang, Y. Zhang, I-TASSER server: New development for protein structure and function predictions. Nucleic Acids Re.s 43 , W174–W181 (2015). 42 M. Kallberg , Template-based protein structure modeling using the RaptorX web server. Nat. Protoc. 7 , 1511–1522 (2012).22814390 43 S. Vajda , New additions to the ClusPro server motivated by CAPRI. Proteins 85 , 435–444 (2017).27936493 44 D. Kozakov , The ClusPro web server for protein-protein docking. Nat. Protoc. 12 , 255–278 (2017).28079879 45 D. Kozakov , How good is automated protein docking? Proteins 81 , 2159–2166 (2013).23996272 46 S. Wang, S. Sun, Z. Li, R. Zhang, J. Xu, Accurate de novo prediction of protein contact map by ultra-deep learning model. PLoS Comput. Biol. 13 , e1005324 (2017).28056090 47 B. A. Shoemaker, J. J. Portman, P. G. Wolynes, Speeding molecular recognition by using the folding funnel: The fly-casting mechanism. Proc. Natl. Acad. Sci. U.S.A. 97 , 8868–8873 (2000).10908673 48 E. Trizac, Y. Levy, P. G. Wolynes, Capillarity theory for the fly-casting mechanism. Proc. Natl. Acad. Sci. U.S.A. 107 , 2746–2750 (2010).20133683 49 D. Wu, N. Potluri, J. Lu, Y. Kim, F. Rastinejad, Structural integration in hypoxia-inducible factors. Nature 524 , 303–308 (2015).26245371 50 A. Waterhouse , SWISS-MODEL: Homology modelling of protein structures and complexes. Nucleic Acids Res. 46 , W296–W303 (2018).29788355 51 L. E. Huang, J. Gu, M. Schau, H. F. Bunn, Regulation of hypoxia-inducible factor 1alpha is mediated by an O2-dependent degradation domain via the ubiquitin-proteasome pathway. Proc. Natl. Acad. Sci. U.S.A. 95 , 7987–7992 (1998).9653127 52 G. L. Semenza, HIF-1 and mechanisms of hypoxia sensing. Curr. Opin. Cell Biol. 13 , 167–171 (2001).11248550 53 G. L. Semenza, A compendium of proteins that interact with HIF-1alpha. Exp. Cell Res. 356 , 128–135 (2017).28336293 54 Y. Goto , UCHL1 provides diagnostic and antimetastatic strategies due to its deubiquitinating effect on HIF-1alpha. Nat. Commun. 6 , 6153 (2015).25615526 55 P. Hanpude, S. Bhattacharya, A. Kumar Singh, T. Kanti Maiti, Ubiquitin recognition of BAP1: Understanding its enzymatic function. Biosci. Rep. 37 , BSR20171099 (2017).28935764 56 J. C. Scheuermann , Histone H2A deubiquitinase activity of the polycomb repressive complex PR-DUB. Nature 465 , 243–247 (2010).20436459 57 R. Zarrizi, J. A. Menard, M. Belting, R. Massoumi, Deubiquitination of gamma-tubulin by BAP1 prevents chromosome instability in breast cancer cells. Cancer Res. 74 , 6499–6508 (2014).25228651 58 E. J. Yeo , YC-1: A potential anticancer drug targeting hypoxia-inducible factor 1. J. Natl. Cancer Inst. 95 , 516–525 (2003).12671019 59 T. Shukuya , Hypoxia inducible factor-1alpha Inhibition in Von Hippel Lindau-mutant malignant pleural mesothelioma cells. Anticancer Res. 40 , 1867–1874 (2020).32234874 60 R. J. Francis , Characterization of hypoxia in malignant pleural mesothelioma with FMISO PET-CT. Lung Cancer 90 , 55–60 (2015).26259878 61 N. J. Brouwer , Ischemia is related to tumour genetics in uveal melanoma. Cancers (Basel) 11 , 1004 (2019).31323773 62 S. S. Wang , Bap1 is essential for kidney function and cooperates with Vhl in renal tumorigenesis. Proc. Natl. Acad. Sci. U.S.A. 111 , 16538–16543 (2014).25359211 63 Y. F. Gu , Modeling renal cell carcinoma in mice: Bap1 and Pbrm1 inactivation drive tumor grade. Cancer Discov. 7 , 900–917 (2017).28473526 64 T. J. Mitchell , Timing the landmark events in the evolution of clear cell renal cell cancer: TRACERx renal. Cell 173 , 611–623.e17 (2018).29656891 65 J. W. Harbour , Frequent mutation of BAP1 in metastasizing uveal melanomas. Science. 330 , 1410–1413 (2010).21051595 66 Y. Yoshikawa , High-density array-CGH with targeted NGS unmask multiple noncontiguous minute deletions on chromosome 3p21 in mesothelioma. Proc. Natl. Acad. Sci. U.S.A. 113 , 13432–13437 (2016).27834213
PMC010xxxxxx/PMC10028936.txt
==== Front bioRxiv BIORXIV bioRxiv Cold Spring Harbor Laboratory 36945387 10.1101/2023.03.07.531546 preprint 2 Article INTERPRETING GENERATIVE ADVERSARIAL NETWORKS TO INFER NATURAL SELECTION FROM GENETIC DATA Riley Rebecca 1 Mathieson Iain 2 Mathieson Sara 1† 1 Department of Computer Science, Haverford College, Haverford PA, 19041 USA. 2 Department of Genetics, Perelman School of Medicine, University of Pennsylvania, Philadelphia PA, 19104 USA. † Corresponding author: [email protected] 09 7 2023 2023.03.07.531546https://creativecommons.org/licenses/by-nc/4.0/ This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License, which allows reusers to distribute, remix, adapt, and build upon the material in any medium or format for noncommercial purposes only, and only so long as attribution is given to the creator. nihpp-2023.03.07.531546.pdf Understanding natural selection in humans and other species is a major focus for the use of machine learning in population genetics. Existing methods rely on computationally intensive simulated training data. Unlike efficient neutral coalescent simulations for demographic inference, realistic simulations of selection typically requires slow forward simulations. Because there are many possible modes of selection, a high dimensional parameter space must be explored, with no guarantee that the simulated models are close to the real processes. Mismatches between simulated training data and real test data can lead to incorrect inference. Finally, it is difficult to interpret trained neural networks, leading to a lack of understanding about what features contribute to classification. Here we develop a new approach to detect selection that requires relatively few selection simulations during training. We use a Generative Adversarial Network (GAN) trained to simulate realistic neutral data. The resulting GAN consists of a generator (fitted demographic model) and a discriminator (convolutional neural network). For a genomic region, the discriminator predicts whether it is “real” or “fake” in the sense that it could have been simulated by the generator. As the “real” training data includes regions that experienced selection and the generator cannot produce such regions, regions with a high probability of being real are likely to have experienced selection. To further incentivize this behavior, we “fine-tune” the discriminator with a small number of selection simulations. We show that this approach has high power to detect selection in simulations, and that it finds regions under selection identified by state-of-the art population genetic methods in three human populations. Finally, we show how to interpret the trained networks by clustering hidden units of the discriminator based on their correlation patterns with known summary statistics. In summary, our approach is a novel, efficient, and powerful way to use machine learning to detect natural selection. Population genetics Machine learning Generative adversarial networks Natural selection Interpretability ==== Body pmc1 Introduction In the last several years, numerous deep learning methods have been proposed as solutions to population genetic problems (see [1] for a recent review), in part due to the ability of deep neural network architectures to recover evolutionary signals from noisy population-level data. Convolutional neural networks (CNNs) have been particularly effective and allowed the field to move away from summary statistics and towards analyzing haplotype matrices directly. The broad idea behind CNNs for genetic data is to treat these matrices (with individuals/haplotypes on the rows and sites/SNPs on the columns) as analogous to images, with convolutional filters picking up on correlations between nearby sites. Population genetic CNNs have now been developed for a variety of applications, including recombination hotspot identification [2], introgression, recombination rates, selection, and demography [3], natural selection [4, 5], adaptive admixture and introgression [6, 7], balancing selection [8], dispersal inference [9], and the task-agnostic software dnadna [10]. New architectures (including those built upon CNNs) have also been proposed, including graph neural networks (GNNs) [11], and U-Nets [12], and recurrent neural networks (RNNs) [13, 14]. One drawback of these machine learning approaches is that they require simulated training data, since labeled examples (i.e. where the historical evolutionary “ground truth” is known) are limited. If the simulated training data is not a good match for the real genetic data, inference results may not be robust. This is a particular problem for understudied populations and non-model species, where less is known about their evolutionary parameter space. Several solutions to this “simulation mis-specification” problem have been proposed, including adaptive re-weighting of training examples [15] and domain-adaptive neural networks [16]. Another strategy is to create custom simulations using Generative Adversarial Networks (GANs) [17], which have recently been developed for population genetic data. Our previous work [18] created a GAN (pg-gan) that fits an evolutionary model to any population – as training progresses, the generator produces synthetic data that is closer and closer to the real data, confusing the discriminator. The key innovation of pg-gan is that it learns an explicit evolutionary generative model, in contrast to other GANs which generate sequences that look like real data from random processes with no underlying model [19, 20]. Recent work has improved GAN-style approaches for population genetics using adversarial Monte Carlo methods [21]. GANs have two components, a generator which creates synthetic examples, and a discriminator which classifies examples as real or fake. After training, the focus is often on the generator, which can be used to produce novel results (i.e. faces, text, genetic data) from a vector of random noise. In pg-gan the generator is not a neural network but instead an evolutionary model. The fitted generator from pg-gan is therefore useful in its own right, as it is directly interpretable in terms of population genetic parameters. Because the specification of such models is relatively standardized (population size changes, splits, migrations, admixtures, etc), we can easily use the resulting model in downstream applications or make it available to other researchers in a standard format (see [22]). For example, we could use the demography as a null model or as training or validation data for a second machine learning method. Although well-trained discriminators can be used downstream to validate the examples created by other generators (e.g. [23]), there is typically less emphasis on using or analyzing the trained discriminator. In this work we develop a new approach that uses the trained discriminators of pg-gan to identify non-neutral regions. The major advantage of this approach over existing approaches is that it requires very few simulated selected regions during training. Despite recent advances (e.g. [24]), simulating realistic natural selection is very difficult, with many parameter choices that make mirroring real data challenging. These include the type of selection, time of onset of selection, selection strength, and frequency of the selected variant at the present. For models with exponential growth or large population sizes, forward simulations become prohibitively computationally expensive. Our approach bypasses the need to simulate large numbers of selected regions by instead detecting regions of real data that do not conform to an inferred (neutral) demographic model. This approach is similar to traditional population genetic approaches to detecting selection by looking for regions that are outliers in terms of divergence, diversity or haplotype structure [25]. However instead of looking at one or a combination of summary statistics [26], our approach operates directly on the observations, and in principle is able to use all the information present in the data. Understanding what neural networks actually learn has been a major goal of interpretability work for many years. Despite progress in the image domain (e.g. [27–29], less is known about what CNNs for population genetic data are learning (see [30] for recent progress). Our interpretability analysis links traditional summary statistics (which are known to be informative for selection and other evolutionary parameters) with the hidden nodes of the discriminator network. This gives us a window into the network’s ability to compute existing summary statistics, importance and redundancy of different statistics, and differences between discriminators. Overall our method represents a novel approach to detect natural selection and other types of non-neutrality, while avoiding both demographic parameter mis-specification and extensive selection simulations. Our software is available open-source at https://github.com/mathiesonlab/disc-pg-gan. 2 Materials and Methods 2.1 GAN training Our existing GAN application, pg-gan, consists of a CNN-based discriminator and an msprime-based generator that are trained in concert. During training, the generator selects parameters to use for the simulation software msprime [31, 32], creating “fake” regions. The discriminator learns to distinguish these simulated neutral regions from the real training data, which contains both neutral and selected regions. With feedback from the discriminator, the generator learns which parameters will produce the most realistic simulated data, and the discriminator gets better at picking up on differences between the two types of data. pg-gan is implemented in tensorflow [33] and can be run on a GPU using CUDA [34]. As the discriminator is a CNN, we are able to train it using traditional gradient descent approaches (i.e. backpropagation). However, since we currently cannot pass gradients through the coalescent with recombination, we train the generator using a gradient-free approach (in this case simulated annealing, but other methods are possible). At the end of this optimization process, we obtain a generator that is good at creating realistic simulated data, and a discriminator that can identify real data accurately [18]. In this work we focus on the trained discriminator rather than the generator. We modify pg-gan to save the trained discriminator network. The input to this CNN is a genomic region represented by a matrix of shape (n, S, 2) where n is the number of haplotypes, S is the fixed number of consecutive SNPs, and there are two separate channels: one for the haplotype data and one for the inter-SNP distances (duplicated down the rows). These two channels are analogous to the three RGB color channels for images. For one-population evolutionary models, this network has two convolutional layers (32 and 64 filters each of dimension 1 × 5), two fully connected layers (each with 128 units), and an output of a single scalar value (probability of the example being real). We use binary cross-entropy as the loss function, with Adam optimization. With 1 × 2 max-pooling after each convolutional layer and a permutation invariant function before the fully connected layers, this results in 76,577 trainable parameters, all of which are saved after training. 2.2 Discriminator evaluation In a small number of cases, GAN training (which is often subject to difficulties due to the minimax nature of the optimization problem) fails. In all our experiments, this manifested as discriminators predicting the same value for all regions (i.e. the network did not learn anything). We therefore discard these discriminators from further analysis. 2.3 Discriminator fine-tuning Since the real data, unlike the simulated data, include regions that have experienced selection, the intuition is that regions that the discriminator predicts to be real with high probability correspond to non-neutral regions, since those have characteristics found in real but not simulated data. One concern is that there are processes other than selection that are found in the real but not simulated data (heterogeneity in parameter values, genotype or reference errors and so on) which might be identified by the discriminator. We therefore incentivize the discriminator to focus on selection in particular by fine-tuning it using a small number of forward simulations incorporating selection (see Section 2.6 below for details). Specifically, we use 3000 neutral regions and 2400 selected regions (600 each of selection strengths s = 0.01, 0.025, 0.05, 0.1). 20% of these regions are reserved for validation, and the discriminator is training on the remaining 80% for 1000 mini-batches, each of size 50 (roughly 12 epochs). We again use binary cross-entropy as the loss function, except now a label of 0 corresponds to neutral regions and a label of 1 corresponds to selected regions. This procedure fine-tunes the weights of the network such high discriminator predictions can be more reliably interpreted as evidence for selection. 2.4 Prediction The fine-tuned discriminator can then be used to predict outcomes for new genetic regions. Values closer to 1 indicate more similarity to the real training data, and values closer to 0 indicate more similarity to the simulated data. A schematic of the entire workflow can be found in Figure 1. We use ROC curves to evaluate the improvement in performance created by fine-tuning. Notably, in 9/10 cases of failed training, fine-tuning also fails to improve the discriminator – prediction results are the same before and after. In one case, a failed discriminator’s predictions (on selection simulations) were improved from fine-tuning, but we still discard it because it was not able to learn to generate realistic simulations initially. Below we describe additional method details and validation approaches, as well as an ensemble approach for successfully trained discriminators. 2.5 Population data and demographic models To ensure that we are training and testing on different datasets, we use three pairs of similar populations from the 1000 Genomes Project [35]: CEU and GBR (Northern European), CHB and CHS (East Asian), and YRI and ESN (West African). These pairs were selected since they are the most similar in terms of genetic distance (see [35], Figure 2). Specifically, we train discriminators using CEU, CHB, and YRI then test on GBR, CHS, and ESN respectively. Although the number of samples may differ between the populations, this is not a problem due to the permutation-invariant architecture of the discriminator. Population samples sizes and number of regions are given in Table 1. For all populations we used an exponential growth model and fitted five parameters using pg-gan: the ancestral population size (N1), the timing of a bottleneck (T1), the bottleneck population size (N2), the onset of exponential growth (T2), and the growth rate (g). For all simulations we used a constant mutation rate of µ = 1.25 × 10−8 per base per generation and recombination rates drawn from the HapMap recombination map [36]. We simulated L = 50kb regions and then extracted the middle S = 36 SNPs. To avoid assuming an ancestral state, for each SNP we encode the minor allele as 1 and the major allele as −1. We found that centering the data around zero worked best with typical neural network activation functions, and also allowed us to zero-pad a small number of regions with fewer than S SNPs. For real data we filter non-biallelic SNPs and ensure that at least 50% of the bases are inside callable regions (see [35]), so that the discriminator avoids making real vs. fake distinctions based on artifacts. 2.6 Selection simulations for fine-tuning To create the selection simulations used for fine-tuning, we use SLiM [24, 37], which can model a variety of types of natural selection. For the demography we use the same evolutionary model (exponential growth) as we used during pg-gan training. For European populations we use the parameters shown in Table 2, which were obtained from a run of pg-gan with CEU training data. During this training run, we capped the exponential growth parameter at 0.01, as allowing this parameter to be too large led to SLiM computations being too slow. See Figure 2 for a visualization of how well this simulated data matches CEU. Under this combination of parameters, we produced neutral regions, as well as regions with a single mutation undergoing positive selection with selection coefficients of of 0.01, 0.025, 0.05, and 0.1. For each selection coefficient we simulated 100 regions and recorded how many of them were classified as “fake”. We used a similar procedure to obtain a neutral demographic model for CHB and YRI. Due to the effect of large population sizes on SLiM runtimes, we again capped g at 0.01, and additionally capped T2 (the onset time of exponential growth) at 750 generations for YRI. The resulting parameters are shown in Table 2 and a visualization of the match is shown in Figure 2. Selection is introduced by adding the mutation to one individual in the population, 1000 generations before the present. If the mutation is lost in any generation, we restart the simulation from the previous generation until we have reached the present – the resulting tree sequence can then be stored for further processing. First, the tree sequence is recapitated using coalescent simulations for efficiency (using the pyslim module). This step ensures that we have a complete tree sequence with all necessary ancestral nodes. Next, we sample the population down to n = 198 haplotypes, to match the real data. This allows us to prune and simplify the tree sequence before adding neutral mutations with msprime. For each region we store arrays with S SNPs, as well as arrays with all the SNPs, in order to compute Tajima’s D. 2.7 Validation on known selected regions To validate predictions after fine-tuning, we examine each discriminator’s performance on genomic regions previously identified as targets of selection by Grossman et al [26] using a combination of different summary statistics. We use the 153 regions from their Table S1, including genes involved in metabolism, disease resistance, and skin pigmentation. We converted the start and end positions of each region to hg19 coordinates and sorted by population (CEU, CHB/JPT, and YRI). This resulted in a comparison between three types of data: Regions simulated using msprime under the demographic parameters inferred by pg-gan (corresponding to the current discriminator) Regions under positive selection as identified in [26] All other real regions (mostly neutral) 2.8 Discriminator ensemble method to detect selected regions To assess the discriminator’s predictions genome-wide, we iterate through the entire genome (using non-overlapping 36-SNP windows) and make a prediction (probability of selection) for each region, then smooth the results by averaging probabilities in five consecutive windows. To visualize the results, we create modified Manhattan plots by plotting the probabilities on the y-axis (on a log scale). Individual discriminators can vary in their performance, so we also created an ensemble classifier by averaging predictions for each region over all successfully trained discriminators. 2.9 Interpretability analysis Finally, to understand what the discriminator network is learning, we investigated whether the network had learned how to compute summary statistics that are known to be informative for evolutionary parameters (demography, selection, etc). To this end, we computed the correlation between the node values of the last hidden layer and various summary statistics. The motivation for analyzing the last hidden layer is that deep into the network, information from the haplotypes should be greatly distilled and processed, with informative high-level “features” extracted. The remaining step in the network is a linear combination of these hidden node values to create a logit value, which can be converted into a probability. We computed pairwise correlations between 128 hidden values and the following 61 summary statistics: The first 9 entries of the folded site frequency spectrum (SFS), excluding non-segregating sites. The 35 inter-SNP distances between the 36 SNPs of each region. These distances are rescaled to fall between 0 and 1. 15 linkage disequilibrium (LD) statistics (see [18] for details of the computation). Pairwise heterozygosity (π). The number of unique haplotypes across the 36 SNPs. Note that the inter-SNP distances are also an input to the network. We calculated Pearson correlation coefficients and plotted them in a heatmap with summary statistics along rows and hidden values along the columns. To better visualize relationships between the hidden units, we clustered the columns according to their similarity using agglomerative clustering. The final visualization groups hidden nodes that are performing similar computations – since we perform a dropout (with rate 0.5) after each fully connected layer, we expect some redundancy in node behavior. We experimented with reducing the number of hidden units in the last few layers, as well as adding three fully connected layers (instead of two). However, all these modifications to the discriminator CNN resulted in degenerate training results (e.g. predicting all regions as real) so we pursued only the original architecture. 3 Results 3.1 Summary of discriminator training For each training population (CEU, CHB, and YRI) we ran pg-gan 20 times to obtain a set of 60 discriminators. For each run we used a different seed, although even runs with the same seed can produce different results depending on the version of tensorflow and CUDA. In ten cases, training failed and the resulting discriminator predicted the same value for all regions (in other words, it ignores the input data – see Supplementary Figures S1 and S2 for an example). After we discarded these discriminators we were left with 50 (17 for CEU, 18 for CHB, and 15 for YRI). Below we discuss the outcomes of fine-tuning and prediction with three representative discriminators (one for each population pair), then explain how we used all successfully trained discriminators in an ensemble method to predict natural selection. 3.1.1 Fine-tuning and validation on simulated data In (A) of Figures 3, 4, and 5 we show the impact of fine-tuning on individual discriminators (seed 19 for CEU, seed 9 for CHB, and seed 4 for YRI). In the ROC curves for simulated data, a positive represents a region simulated under any strength of selection, and a negative represents a neutral region. For real neutral vs. selected data, we use the regions identified in [26] as the truth set. Dashed lines represent discriminator predictions before fine-tuning and solid lines represent predictions after fine-tuning. Fine-tuning improved discriminator predictions in all cases, although in a few cases the improvement was marginal. For YRI-trained discriminators tested on ESN, improvements were generally more marginal than for CEU- and CHB-trained discriminators. After fine-tuning we run selection simulations (again created with SLiM) through each discriminator. In (B) of Figures 3, 4, and 5, we visualize these results through violin plots with 1000 neutral regions and 100 regions of each selection strength (all unseen during fine-tuning). Most discriminators group neutral regions and those with selection strength s = 0.01 together, with probabilities around 0.2 − 0.3. Regions under stronger selection have much higher probabilities (around 0.9 − 1). Although the network was fine-tuned with only a few simulated regions under selection, the real data includes regions of selection, so this general pattern exposes the ability of most discriminators to see selection as realistic. 3.1.2 Validation on known selected regions To understand how the discriminator behaves when presented with real data from a test population, we produced predictions for regions known to be under selection (from [26]) and compared to predictions from the rest of the genome. We also included data simulated under the demography inferred by the pg-gan training run that produced the given discriminator. In theory this is the data which the discriminator thinks is most similar to real data. Overall we find significant differences in the predictions for selected regions vs. neutral, as seen in the second rows of Figures 3, 4, and 5 (C). For example, in Figure 3, we show the performance of a CEU-trained discriminator (seeds 19) on real data from GBR. Horizontal bars represent means (solid) and 0.05 − 0.95 quantiles (dashed). In this case, the t-test p-value for a difference in means between selected (0.423) vs. neutral (0.261) probabilities is 1.63 × 10−50. Of the 50 successfully trained discriminators, the p-value is significant (at the 0.05 confidence level) in all cases. Predictions for simulated data (neutral) and neutral real data typically have very similar distributions, indicating that the generator’s demographic model trained successfully. For each discriminator, we also performed a genome-wide scan over real test regions (see Figures 3, 4, and 5, E). In the case of CEU (seed 19) for example, we computed predictions for GBR data in 36-SNP regions, then averaged 5 such regions (in overlapping windows) to create the final predictions for these modified Manhattan plots. Similarly, we ran CHS through the CHB-trained discriminators (seed 9 shown) and ESN through the YRI-trained discriminators (seed 4 shown). 3.1.3 Interpretability analysis After performing correlations between the hidden units of each discriminator and traditional summary statistics (both calculated on the test data), we find some common patterns. Our hierarchical clustering approach shows extensive redundancy in the hidden units, which is to be expected given the dropout computations included in the last fully connected layers. We note that none of the correlation values are particularly high – the largest (in magnitude) are around ±0.4. Frequently the highest correlations are with rare variants (first entries of the SFS), LD statistics, and π (pairwise heterozygosity). For some discriminators, inter-SNP distances also seem to be important. In Figures 3, 4, and 5 (D) we show example heatmaps of correlation values between hidden values of the last layer and summary statistics. 3.2 Ensemble results For the successful discriminators of each training population, we average the probabilities for each test region to create ensemble results. To smooth the results, we further average predictions over five consecutive 36-SNP region (in overlapping windows). Across populations, the mean probability of selection is 0.25, and approximately 1% of the genome has a probability of selection >50% (1.1% for GBR, 1.3% for CHS, and 0.45% for ESN; Figure 6). We confidently identify well-known targets of selection, including LCT (probability of selection, p=0.79) [38], SLC24A5 (p=0.94) [39], KITLG (p=0.79) [40, 41] and OCA2/HERC2 (p=0.86) [42] in GBR ; EDAR (p=0.55) in CHS [43]; and APOL1 (p=0.83) [44] in ESN. We also assign high probabilities of selection to several potentially novel targets, including APPBP2/PPM1D (p=0.92) in GBR; CENPW (p=0.95) and EXOC6B (p=0.95) in CHS; and EHBP1 (p=0.88), WHSC1L1 (p=0.87), ARID1A (p=0.85) in ESN (Figure 7). Full predictions are given in Supplementary Table S1. To more systematically validate our results, we obtained selected regions from Grossman et al [26] (Supplementary Figure S3), which uses a composite of multiple classical selection statistics to identify selected regions. We calculated our mean prediction probabilities inside and outside these regions and then performed permutation testing by shuffling the locations of the selected regions from [26] randomly throughout the genome. For each of 100 such permutations we re-computed the mean probability of our regions overlapping with these permuted regions. In all cases the mean selection probability of the true known selected genes (D(Ys)) was much higher than the permutated values. p-values were 0.0 for all populations and our mean was 8–13 standard deviations above the permutation testing mean (Table 3). 3.3 Runtime analysis Overall the runtime of our method is dominated by the initial training of pg-gan, which takes 5–7 hours per discriminator. After the initial training, fine-tuning is very fast, about 1 minute per discriminator. Selection simulations with SLiM were completed in advance of fine-tuning, and were highly sensitive to the demography used to mirror each population. YRI simulations took about 4–5 hours per 100 regions, CHB simulations took about 1 hour per 100 regions, and CEU simulations took about 4–8 hours per 100 regions, depending on the strength of selection (all simulations were parallelized). We hypothesize that CEU took the longest due to the highest inferred exponential growth rate. CHB had the lowest average effective population size, likely resulting in faster simulations. 4 Discussion This work develops three novel aspects of machine learning in population genetics. First, we show how a discriminator trained as part of a GAN can be used as a classifier independent of the generator. Second, we show how the discriminator can be further incentivized to give high probabilities to selected regions in particular through fine-tuning on a small number of selection simulations. Finally, we show how the hidden units of the discriminator can be interpreted in terms of known summary statistics. In particular, we find that rare variant counts, LD statistics, and pairwise heterozygosity are implicitly computed by hidden units in the discriminator network. A major advantage of our approach is that we do not need to simulate large numbers of selected regions across a high-dimensional space of selection parameters. Training pg-gan is not instantaneous, but is much faster than it would be if we used selection simulations throughout the entire training procedure. A typical training run of pg-gan uses around 1.5 million neutral simulated regions depending on the number of demographic parameters. In contrast, we only use around five thousand selection simulations for fine-tuning. By adding this step, we are able to quickly create discriminators that have the ability to pick up on real selected regions. In order to validate our approach in other species, it would be helpful to identify a set of regions under positive selection (for testing). This could be bypassed with high-quality selection simulations, but these might be difficult to obtain depending on the species. Here, we used known selected regions to validate our approach but an alternative would be to instead incorporate these into pg-gan training as an additional fine-tuning step. Currently, our interpretability work is on a global scale, meaning that we analyze genome-wide patterns. Future work could examine why the discriminator makes predictions for particular regions, especially outliers with high probabilities. A caveat of our approach is that the non-neutral regions identified by the discriminator cannot be directly interpreted in terms of selection parameters. Simulations suggest that we can easily detect hard sweeps with selection coefficients greater than around 1%, but may not be able to detect weaker selection (similar to state-of-the art population genetic methods for detecting selection [41, 45]). It is also possible that, despite the fine-tuning step, the regions we identify actually reflect different types of deviation from the generator model, such as heterogeneity in mutation or recombination rates, or structural variation. However, we note that almost all approaches to detecting to selection are sensitive to these effects to some extent and our results, like those of all selection scans, should be treated as candidate regions requiring validation. Another limitation is that we only train and model positive selection. However, our fine-tuning approach could be extended to other selection regimes, for example background and balancing selection and balancing selection. In fact, we could modify the last layer of the discriminator network to output a multi-class prediction that incorporates these different modes of selection. More generally, we view our work as the beginning of an exploration of how the trained discriminator can be used in transfer learning approaches – for these and other evolutionary applications. Supplementary Material Supplement 1 Supplement 2 Acknowledgements SM is funded in part by NIH grant R15HG011528. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Figure 1: Selection inference workflow. First we run pg-gan many times (i.e. 20) to obtain a collection of discriminators. In the evaluation step, we discard any discriminator that does not pass quality checks (i.e. predicting the same value for every region). Retained discriminators (roughly 85%) are fine-tuned with a modest number of selection simulations. Finally, real data from a new but similar population is fed through each fine-tuned discriminator to obtain a probability. Regions where a large fraction of the discriminators produced a high probability are candidate selected regions. Figure 2: Summary statistic comparison of real training data (CEU, CHB, and YRI) to data simulated under SLiM with parameters inferred by pg-gan. The number of regions used for comparison was 5000. Figure 3: Example of a CEU-trained discriminator with test data from GBR (seed 19). A) ROC curve showing performance on selected regions before and after fine-tuning. All predictions (both simulated and real) are made for unseen (test) data. SLiM indicates simulated data with s = xx and REAL indicates selected regions from [26]. B) Performance of discriminator on unseen simulated data under various selection strengths. C) Performance of discriminator on selected regions from [26]. D) Correlation heatmap between discriminator hidden units (x-axis) and classical population genetics summary statistics (y-axis). The columns (hidden units) were clustered according to their similarity in terms of summary statistic correlation profiles. E) Genome-wide Manhattan plots of discriminator predictions on real test data. Figure 4: Example of a CHB-trained discriminator with test data from CHS (seed 9). A) ROC curve showing performance on selected regions before and after fine-tuning. All predictions (both simulated and real) are made for unseen (test) data. SLiM indicates simulated data with s = xx and REAL indicates selected regions from [26]. B) Performance of discriminator on unseen simulated data under various selection strengths. C) Performance of discriminator on selected regions from [26]. D) Correlation heatmap between discriminator hidden units (x-axis) and classical population genetics summary statistics (y-axis). The columns (hidden units) were clustered according to their similarity in terms of summary statistic correlation profiles. E) Genome-wide Manhattan plots of discriminator predictions on real test data. Figure 5: Example of a YRI-trained discriminator with test data from ESN (seed 4). A) ROC curve showing performance on selected regions before and after fine-tuning. All predictions (both simulated and real) are made for unseen (test) data. SLiM indicates simulated data with s = xx and REAL indicates selected regions from [26]. B) Performance of discriminator on unseen simulated data under various selection strengths. C) Performance of discriminator on selected regions from [26]. D) Correlation heatmap between discriminator hidden units (x-axis) and classical population genetics summary statistics (y-axis). The columns (hidden units) were clustered according to their similarity in terms of summary statistic correlation profiles. E) Genome-wide Manhattan plots of discriminator predictions on real test data. Figure 6: Ensemble probability results. On the x-axis is the probability of selection, and the y-axis shows the fraction of the genome classified as selected if we used the given probability threshold (log scale). Across populations the average probability of selection is 0.25, and approximately 1% of the genome has a probability of seleciton >50%. Figure 7: Ensemble results. Genome-wide selections scans for CEU/GBR (A), CHB/CHS (B), and YRI/ESN (C). In each case, the x-axis represents genomic position, the y-axis represents the probability of selection (log scaled) and each point represents the average of five consecutive 36-SNP windows. In black text we highlight known and novel regions with a high probability of selection. Table 1: Populations analyzed along with the number of individuals in each population. The number of regions is determined using non-overlapping 36-SNP windows, where at least 50% of bases must be in the accessibility mask. Code Population description # individuals # regions CEU Utah residents with Northern and Western European ancestry 99 292,936 GBR British in England and Scotland 91 277,038 CHB Han Chinese in Beijing, China 103 280,215 CHS Han Chinese South 105 270,351 YRI Yoruba in Ibadan, Nigeria 108 473,493 ESN Esan in Nigeria 99 460,180 Table 2: Parameters of the exponential growth model, inferred for each population through a training run of pg-gan. There parameters are used for the demography in our SLiM selection simulations. population N 1 N 2 g T 1 T 2 CEU 22552 3313 0.00535 3589 1050 CHB 24609 3481 0.00404 4417 1024 YRI 23231 29962 0.00531 4870 581 Table 3: Ensemble results. For each population we report the mean predicted probabilities for regions that are selected vs. neutral (according to [26]), and the p-value for the difference in mean probabilities between these two sets of regions. CEU/GBR CHB/CHS YRI/ESN # successful discriminators 17/20 18/20 15/20 mean neutral D(Yn) 0.262 0.237 0.252 mean selection D(Ys) 0.386 0.385 0.350 p-value for difference in means 8.87 × 10−47 2.26 × 10−84 4.92 × 10−83 # std dev above mean (permutation testing) 8.75 9.47 12.61 Data availability statement All human genomic data used in this study is publicly available [35]. Our software is also publicly available open-source at https://github.com/mathiesonlab/disc-pg-gan. ==== Refs References [1] Korfmann Kevin , Oscar E Gaggiotti, and Matteo Fumagalli. Deep learning in population genetics. Genome Biology and Evolution, 15 (2 ):evad008, 2023.36683406 [2] Chan Jeffery , Perrone Valerio , Spence Jeffery , Jenkins Paul , Mathieson Sara , and Song Yun . A likelihood-free inference framework for population genetic data using exchangeable neural networks. Neural Information Processing Systems, pages 8594–8605, February 2018. [3] Flagel Lex , Brandvain Yaniv , and Schrieder Daniel . The unreasonable effectiveness of convolutional neural networks in population genetic inference. Molecular Biology and Evolution, 36 (2 ):220–238, December 2018. [4] Torada Luis , Lorenzon Lucrezia , Beddis Alice , Isildak Ulas , Pattini Linda , Mathieson Sara , and Fumagalli Matteo . Imagene: a convolutional neural network to quantify natural selection from genomic data. BMC Bioinformatics, 20 (337 ):2560–2569, November 2019. [5] Qin Xinghu , Charleston WK Chiang, and Oscar E Gaggiotti. Deciphering signatures of natural selection via deep learning. Briefings in Bioinformatics, 23 (5 ):bbac354, 2022.36056746 [6] Gower Graham , Pablo Iáñez Picazo Matteo Fumagalli , and Racimo Fernando . Detecting adaptive introgression in human evolution using convolutional neural networks. Elife, 10 :e64669, 2021.34032215 [7] Hamid Iman , Korunes Katharine L , Schrider Daniel R , and Goldberg Amy . Localizing post-admixture adaptive variants with object detection on ancestry-painted chromosomes. Molecular Biology and Evolution, 40 (4 ): msad074, 2023.36947126 [8] Isildak Ulas , Stella Alessandro , and Fumagalli Matteo . Distinguishing between recent balancing selection and incomplete sweep using deep neural networks. Molecular Ecology Resources, 21 (8 ):2706–2718, 2021.33749134 [9] Chris CR Smith Silas Tittes , Ralph Peter L , and Kern Andrew D . Dispersal inference from population genetic variation using a convolutional neural network. bioRxiv, pages 2022–08, 2022. [10] Sanchez Théophile , Erik Madison Bray Pierre Jobic , Guez Jérémy , Letournel Anne-Catherine , Charpiat Guillaume , Cury Jean , and Jay Flora . dnadna: a deep learning framework for population genetics inference. Bioinformatics, 2022. [11] Korfmann Kevin , Sellinger Thibaut , Freund Fabian , Fumagalli Matteo , and Tellier Aurélien . Simultaneous inference of past demography and selection from the ancestral recombination graph under the beta coalescent. bioRxiv, pages 2022–09, 2022. [12] Dylan D Ray Lex Flagel , and Schrider Daniel R . Introunet: identifying introgressed alleles via semantic segmentation. bioRxiv, 2023. [13] Adrion Jeffrey R , Galloway Jared G , and Kern Andrew D . Predicting the landscape of recombination using deep learning. Molecular biology and evolution, 37 (6 ):1790–1808, 2020.32077950 [14] Hussein A Hejase Ziyi Mo , Campagna Leonardo , and Siepel Adam . A deep-learning approach for inference of selective sweeps from the ancestral recombination graph. Molecular Biology and Evolution, 39 (1 ):msab332, 2022.34888675 [15] Klara Elisabeth Burger Peter Pfaffelhuber , and Baumdicker Franz . Neural networks for self-adjusting mutation rate estimation when the recombination rate is unknown. PLOS Computational Biology, 18 (8 ):e1010407, 2022.35921376 [16] Mo Ziyi and Siepel Adam . Domain-adaptive neural networks improve supervised machine learning based on simulated population genetic data. bioRxiv, pages 2023–03, 2023. [17] Goodfellow Ian , Jean Pouget-Abadie Mehdi Mirza , Xu Bing , David Warde-Farley Sherjil Ozair , Courville Aaron , and Bengio Yoshua . Generative adversarial networks. Communications of the ACM, 63 (11 ):139–144, 2020. [18] Wang Zhanpeng , Wang Jiaping , Kourakos Michael , Hoang Nhung , Hyong Hark Lee Iain Mathieson , and Mathieson Sara . Automatic inference of demographic parameters using generative adversarial networks. Molecular ecology resources, 21 (8 ):2689–2705, 2021.33745225 [19] Yelmen Burak , Decelle Aurélien , Ongaro Linda , Marnetto Davide , Tallec Corentin , Montinaro Francesco , Furtlehner Cyril , Pagani Luca , and Jay Flora . Creating artificial human genomes using generative neural networks. PLoS genetics, 17 (2 ):e1009303, 2021.33539374 [20] Booker William W , Ray Dylan D , and Schrider Daniel R . This population doesn’t exist: learning the distribution of evolutionary histories with generative adversarial networks. bioRxiv, 2022. [21] Gower Graham , Pablo Iáñez Picazo Finn Lindgren , and Racimo Fernando . Inference of population genetics parameters using discriminator neural networks: an adversarial monte carlo approach. bioRxiv, pages 2023–04, 2023. [22] Adrion Jeffrey R , Cole Christopher B , Dukler Noah , Galloway Jared G , Gladstein Ariella L , Gower Graham , Kyriazis Christopher C , Ragsdale Aaron P , Tsambos Georgia , Baumdicker Franz , A community-maintained standard library of population genetic models. Elife, 9 , 2020. [23] Salimans Tim , Goodfellow Ian , Zaremba Wojciech , Cheung Vicki , Radford Alec , and Chen Xi . Improved techniques for training gans. Advances in neural information processing systems, 29 , 2016. [24] Haller Benjamin C and Messer Philipp W . Slim 3: forward genetic simulations beyond the wright–fisher model. Molecular biology and evolution, 36 (3 ):632–637, 2019.30517680 [25] Akey J. M. Constructing genomic maps of positive selection in humans: where do we go from here? Genome Res, 19 (5 ):711–22, 2009. ISSN 1088–9051 (Print) 1088–9051 (Linking). doi: 10.1101/gr.086652.108. URL https://www.ncbi.nlm.nih.gov/pubmed/19411596. 19411596 [26] Grossman Sharon R , Andersen Kristian G , Shlyakhter Ilya , Tabrizi Shervin , Winnicki Sarah , Yen Angela , Park Daniel J , Griesemer Dustin , Karlsson Elinor K , Wong Sunny H , Identifying recent adaptations in large-scale genomic data. Cell, 152 (4 ):703–713, 2013.23415221 [27] Simonyan Karen , Vedaldi Andrea , and Zisserman Andrew . Deep inside convolutional networks: Visualising image classification models and saliency maps. arXiv preprint arXiv:1312.6034, 2013. [28] Marco Tulio Ribeiro Sameer Singh , and Guestrin Carlos . “ why should i trust you?” explaining the predictions of any classifier. In Proceedings of the 22nd ACM SIGKDD international conference on knowledge discovery and data mining, pages 1135–1144, 2016. [29] Montavon Grégoire , Samek Wojciech , and Klaus-Robert Müller. Methods for interpreting and understanding deep neural networks. Digital signal processing, 73 :1–15, 2018. [30] Nait Saada J. , Hu A. , and Palamara P. F. Inference of pairwise coalescence times and allele ages using deep neural networks. NeurIPS Workshop of Learning Meaningful Representations of Life (LMRL), 2021. [31] Kelleher Jerome , Etheridge Alison M , and McVean Gilean . Efficient coalescent simulation and genealogical analysis for large sample sizes. PLoS computational biology, 12 (5 ):e1004842, 2016.27145223 [32] Baumdicker Franz , Bisschop Gertjan , Goldstein Daniel , Gower Graham , Aaron P Ragsdale Georgia Tsambos , Zhu Sha , Eldon Bjarki , Ellerman E Castedo , Galloway Jared G , Efficient ancestry and mutation simulation with msprime 1.0. Genetics, 220 (3 ):iyab229, 2022.34897427 [33] Abadi Martín , Agarwal Ashish , Barham Paul , Brevdo Eugene , Chen Zhifeng , Citro Craig , Corrado Greg S. , Davis Andy , Dean Jeffrey , Devin Matthieu , TensorFlow: Large-scale machine learning on heterogeneous systems, 2015. URL https://www.tensorflow.org/. Software available from tensorflow.org. [34] NVIDIA Péter Vingelmann , and Fitzek Frank H.P. Cuda, release: 10.2.89, 2020. URL https://developernvidia.com/cuda-toolkit. [35] 1000 Genomes Project Consortium A global reference for human genetic variation. Nature, 526 (7571 ):68, 2015.26432245 [36] International HapMap Consortium A second generation human haplotype map of over 3.1 million snps. Nature, 449 (7164 ):851, 2007.17943122 [37] Benjamin C Haller Jared Galloway , Kelleher Jerome , Messer Philipp W , and Ralph Peter L . Tree-sequence recording in slim opens new horizons for forward-time simulation of whole genomes. Molecular ecology resources, 19 (2 ):552–566, 2019.30565882 [38] Bersaglieri T. , Sabeti P. C. , Patterson N. , Vanderploeg T. , Schaffner S. F. , Drake J. A. , Rhodes M. , Reich D. E. , and Hirschhorn J. N. Genetic signatures of strong recent positive selection at the lactase gene. Am J Hum Genet, 74 (6 ):1111–20, 2004.15114531 [39] Lamason R. L. , Mohideen M. A. , Mest J. R. , Wong A. C. , Norton H. L. , Aros M. C. , Jurynec M. J. , Mao X. , Humphreville V. R. , Humbert J. E. , Sinha S. , Moore J. L. , Jagadeeswaran P. , Zhao W. , Ning G. , Makalowska I. , McKeigue P. M. , O’Donnell D. , Kittles R. , Parra E. J. , Mangini N. J. , Grunwald D. J. , Shriver M. D. , Canfield V. A. , and Cheng K. C. Slc24a5, a putative cation exchanger, affects pigmentation in zebrafish and humans. Science, 310 (5755 ):1782–6, 2005.16357253 [40] Stern A. J. , Wilton P. R. , and Nielsen R. An approximate full-likelihood method for inferring selection and allele frequency trajectories from dna sequence data. PLoS Genet, 15 (9 ):e1008384, 2019.31518343 [41] Field Y. , Boyle E. A. , Telis N. , Gao Z. , Gaulton K. J. , Golan D. , Yengo L. , Rocheleau G. , Froguel P. , McCarthy M. I. , and Pritchard J. K. Detection of human adaptation during the past 2000 years. Science, 354 (6313 ): 760–764, 2016.27738015 [42] Lao O. , de Gruijter J. M. , van Duijn K. , Navarro A. , and Kayser M. Signatures of positive selection in genes associated with human skin pigmentation as revealed from analyses of single nucleotide polymorphisms. Ann Hum Genet, 71 (Pt 3 ):354–69, 2007.17233754 [43] Voight B. F. , Kudaravalli S. , Wen X. , and Pritchard J. K. A map of recent positive selection in the human genome. PLoS Biol, 4 :e72, 2006.16494531 [44] Genovese G. , Friedman D. J. , Ross M. D. , Lecordier L. , Uzureau P. , Freedman B. I. , Bowden D. W. , Langefeld C. D. , Oleksyk T. K. , Uscinski Knob A. L. , Bernhardy A. J. , Hicks P. J. , Nelson G. W. , Vanhollebeke B. , Winkler C. A. , Kopp J. B. , Pays E. , and Pollak M. R. Association of trypanolytic apol1 variants with kidney disease in african americans. Science, 329 (5993 ):841–5, 2010.20647424 [45] Palamara P. F. , Terhorst J. , Song Y. S. , and Price A. L. High-throughput inference of pairwise coalescence times identifies signals of selection and enriched disease heritability. Nat Genet, 50 (9 ):1311–1317, 2018.30104759
PMC010xxxxxx/PMC10037123.txt
==== Front Health Serv Manage Res Health Serv Manage Res sphsm HSM Health Services Management Research 0951-4848 1758-1044 SAGE Publications Sage UK: London, England 36952623 10.1177_09514848231165894 10.1177/09514848231165894 Theoretical or Conceptual Development When caring breeds contempt: The impact of moral emotions on healthcare professionals’ commitment during a pandemic https://orcid.org/0000-0002-3890-414X Davidson Morgan https://orcid.org/0000-0002-4713-508X Andiappan Meena 7938 University of Toronto, Institute of Health Policy, Management and Evaluation, Toronto, ON, Canada Morgan Davidson, University of Toronto, Institute of Health Policy, Management and Evaluation, 155 College St 4th Floor, Toronto, ON M5T 3M6, Canada. Email: [email protected] 23 3 2023 8 2023 23 3 2023 36 3 215227 © The Author(s) 2023 2023 SAGE Publications https://creativecommons.org/licenses/by-nc/4.0/ This article is distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 License (https://creativecommons.org/licenses/by-nc/4.0/) which permits non-commercial use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access page (https://us.sagepub.com/en-us/nam/open-access-at-sage). The novel coronavirus (COVID-19) pandemic is a major heath crisis that continues to impact healthcare organizations worldwide. As infection rates surged, there was a global shortage of personal protective equipment, critical medications, ventilators, and hospital beds, meaning that healthcare professionals faced increasingly difficult workplace conditions. In this conceptual study, we argue these situations can lead to healthcare professionals experiencing moral emotions - defined as specific emotions which relate, or occur in response, to the interest or welfare of others - towards their organizations. This paper explores the three moral emotions of contempt, anger and disgust, and their potential influence on healthcare professionals’ workplace commitment in the context of a pandemic. Drawing from the moral emotions and organizational commitment literature, we develop a process model to demonstrate how healthcare professionals’ affective and continuous commitment are likely to decrease while, paradoxically, normative, and professional commitment may become amplified. The possible potential for positive outcomes from negative moral emotions is discussed, followed by theoretical and practical contributions of the model, and finally, directions for future research. Moral emotions workplace commitment healthcare professionals COVID-19 typesetterts10 ==== Body pmcIntroduction The COVID-19 pandemic has become an ongoing crisis for healthcare organizations since it emerged in 2019. 1 Healthcare organizations were uniquely affected by the pandemic as they had to not only care for those infected with COVID-19 but contend with the challenges of keeping their workplaces and employees safe. Healthcare organizations have had to cope with surges in patient numbers despite significant efforts to slow the spread of infection2,3 as well as critical shortages of physical space, equipment (e.g., ventilators), intensive care beds and key medications-particularly through the first wave of the of the pandemic.4–6 Frontline healthcare professionals, who have described themselves as “unwilling martyrs”, 7 have been heavily affected by these conditions and thrust into a distressing new normal in their workplaces. Using the COVID-19 pandemic crisis as an illustrative example of an external event that resulted in organizational crises (where organizational crisis is defined as an extraordinary condition that is disruptive and damaging to the existing operating state of an organization), 8 we examine how employees’ responses to their organizations’ management of crisis events affects their commitment. Specifically, in this paper we explore the potential for moral emotions to influence professionals’ commitment, behaviours, and performance in their workplace by asking: How do healthcare professionals’ crisis-driven moral emotions affect their organizational and professional commitment? Moral emotions are emotions linked to the interest or welfare of others and are experienced in situations that are morally relevant. 9 Individuals perceive their workplaces as morally relevant when they, or their organizations, act in ways that influence the well-being and rights of others. 10 The moral nature of the workplace is highlighted for those working at the front of the COVID-19 pandemic, such as healthcare professionals, as it has been during past pandemics. 11 Moral emotions should be particularly salient in the context of the current pandemic for two reasons. First, although global pandemics have occurred in the past, the scale and scope of the current pandemic has left healthcare organizations in crisis, facing a lack of resources and escalating demands. However, the lack of planning displayed by healthcare organizations was somewhat surprising in light of guidance available from previous outbreaks such as SARS, MERS, H1N1 and Ebola in the last decade. 12 Healthcare professionals in these organizations have had to face the brunt of these effects, through no fault of their own, making them especially likely to experience negative affect towards their workplaces. Second, frontline healthcare professionals working in healthcare organizations such as hospitals means seeing the direct effects of (in)adequate care on patients. Although there are many healthcare professionals working globally during this pandemic, we focus on the organizational and political landscape affecting physicians and nurses in North America. In this conceptual study, we examine the potential influence of healthcare organizations’ responses during the first wave of the COVID-19 pandemic, from March to August 2020, 13 on healthcare professionals’ three moral emotions: contempt, anger, and disgust. We argue that experiencing these three negative emotions will serve to decrease affective and continuance organizational commitment but has the potential to lead to an increase in normative commitment and the commitment to one’s profession. Organizational commitment, a construct defined as the psychological state which characterizes an employee’s relationship with their organization, has been examined for its ability to promote effective organizational functioning. 14 While there have been some studies into how employees’ organizational commitment may change in the face of external events such as economic crises,15, 16 there has not yet been considerable research into how, and by which mechanisms, employees’ organizational commitment may change during a pandemic. In terms of employees’ moral emotions, theoretical frameworks have conceptualized affective commitment as being motivated by emotions and feelings rather than individuals’ rational judgements 17 while other studies have found that moral emotions play a key role in the maintenance of physicians’ professional values and may drive institutional work (e.g., the effort to create and maintain one’s institution) at the micro level. 18 Antecedents of moral emotions in the workplace (e.g.,) 10 have been investigated and theories suggesting that emotions can energize (or de-energize) employees’ behaviours in organizational contexts have been proposed (e.g.,),19, 20 but little research has examined specifically how moral emotions could affect organizational commitment specifically. Professional commitment, the strength of an individual’s identification with and involvement in their profession, is often used in reference to self-regulated professions such as physicians, accountants, and engineers. 21 Tensions between professional and organizational commitment persist, and the effect of moral emotions and COVID-19 have not been investigated as factors affecting either type of commitment. This study aims to contribute to two streams of management literature in three ways. First, we expand the moral emotions literature by attempting to understand how experiencing moral anger, contempt, and disgust may affect perceptions of one’s workplace and behaviours within it, and in doing so, respond to recent calls to further understand moral emotions in management scholarship (e.g.,). 22 Second, we attempt to contribute to the organizational commitment literature by illustrating how different types of commitment (affective, continuance, and normative) may be differentially affected by the same organizational crisis event. Third, we propose to contribute to the broader workplace commitment literature by examining how moral emotions may have opposing effects on professional and organizational commitment - suggesting that negative moral emotions may, in some cases, result in positive outcomes. Our process model is depicted in Figure 1.Figure 1. Process model illustrating the impact of healthcare professionals’ more emotions (contempt, anger, and disgust) on their professional and organization commitment during the COVID-19 pandemic. The impact of negative moral emotions on healthcare professionals’ workplace commitment Moral emotions While there is no neat division between moral emotions and their nonmoral counterparts, Haidt explains that emotions become moral when experienced in the specific material conditions of moral issues linked to the interest or welfare of others, or to society as a whole. 9 These emotions serve to efficiently alert individuals to when they, or others, have not upheld the moral standards of society.9,22 Moral emotions have origins in biology and previous experiences 23 and can be learned via social situations. 22 Individuals experience moral emotions in response to immoral events and the mistreatment of themselves or others 24 – e.g., in the case of healthcare workers, mistreatment may be suffered by employees, patients, or their families. There is evidence to suggest that the capacity to experience moral emotions may be automatic 25 and that moral emotions serve as motivators for individuals’ behaviours, such as preventing unethical conduct 26 and promoting prosocial behaviour. 22 Moral emotions are those that respond to moral violations or motivate moral behaviour. 9 For example, moral anger would be the feeling of anger resulting from experiencing a moral violation, such as anger in response to witnessing discrimination, versus anger resulting from a non-moral violation (e.g., a flight cancellation). We use this terminology in order to denote the emotions discussed occur in response to questions of morality.27,28 Moral emotions are grouped into families of emotions which share similar characteristics. These families are differentiated based on the focus of the emotion (i.e., the transgressor or the victim) and whether the behaviour violates or transcends moral standards. 9 These families include the following moral emotions: other-praising, other-suffering, other-condemning and, self-condemning. 9 In this paper, we focus on other-condemning emotions, which are defined as those that are experienced in response to others’ moral transgressions and are commonly considered to be composed of the trifecta emotions of contempt, anger, and disgust.9,22,25,29 We take the view of the victim – that is, the healthcare worker – where the transgressor is the healthcare organization. We concentrate on other-condemning emotions because healthcare professionals are likely to feel that they are tied to the choices of their organizations which have resulted in subjecting them to detrimental workplace conditions. Contempt is defined as a feeling of disdain towards those who have violated a moral norm or the ethic of a community, 29 in this case, the norms of the health services community. Moral anger describes the indignation that occurs when a moral standard has been violated without justification. 9 Disgust has been conceptualized as a defensive emotion that protects individuals against potential sources of contamination and is often linked to moral judgment. 30 We focus on these three emotions because they are the negative moral emotions experienced in response to others’ moral transgression, thus making them particularly relevant for healthcare professionals during the COVID-19 pandemic – we suggest that such workers would be experiencing emotions related to circumstances largely outside of their control and dictated by their organizations’ responses. These three emotions, named the CAD triad, have been proposed as guardians of the moral order as they motivate people to change their relationships with those that have violated it. 29 Although there has been empirical support for combining contempt, anger, and disgust into a single construct of negative moral emotions, 31 we take the view of Greenbaum et al. (2020), to consider and examine each emotion in isolation since they each have unique antecedents and outcomes. 22 Moral Emotions during Pandemics Contempt Contempt involves a judgement of another party’s moral character and values, where the evaluator finds the other party lacking. 9 Contempt can be experienced in nonmoral situations, 22 but the pandemic creates a moral context in the workplace by creating moral issues for healthcare professionals as they are required to make decisions and perform actions that may harm or benefit others. 32 Greenbaum and colleagues (2020) suggest this can frame feelings of contempt in a moral light. We suggest that healthcare professionals will experience contempt towards their organizations’ management of the pandemic for three reasons. First, studies have shown the performance of undesirable job tasks acts as a driver of moral contempt.10,22 Healthcare professionals are asked to care for patients infected with COVID-19 under undesirable conditions since hospitals are overcrowded and resource-scarce, leading to unsafe environments. 33 A lack of critical resources (e.g. PPE, ventilators, medications, hospital beds) as experienced during the pandemic34,35 makes for distressing work environments where ethical and moral dilemmas create unnecessarily traumatizing experiences that can breed contempt for their organization. Second, poor compensation is known to act as a source of moral contempt in the workplace.10,22 Pay inequities evoke feelings of frustration and resentment in employees because they feel mistreated and that their employers are defaulting on their obligations. 36 The situation for healthcare professionals, particularly in the United States, continues to be paradoxical. As COVID-19 cases rose, many healthcare professionals initially found themselves either out of work, furloughed or taking significant pay cuts of up to 70% of their salary. 37 While this was not the case for all organizations (for example, some organizations offered pandemic pay or increased employee bonuses),38–40 healthcare organizations that did make cuts attributed them to the financial difficulties caused by the cancellation of elective and non-emergent procedures. 37 However, some large healthcare companies, such as HCA Healthcare in Tennessee, received billions of dollars in federal emergency stimulus funding while simultaneously warning their nurses that they might face layoffs if they did not agree to wage freezes and reductions. While HCA Healthcare was cutting the pay of its front-line healthcare providers, it did not reduce its executives’ million-dollar salaries. 41 We argue that such wage cuts are demoralizing to healthcare professionals who were still required to put themselves at risk for lower pay, fewer benefits, and less security 37 while hospital executives remain fully remunerated, leading to feelings of contempt against such organizations. Third, healthcare professionals are likely to experience contempt targeted at healthcare organizations due to organizations’ lack of preparedness for COVID-19. Although the impact and severity of the pandemic could not have been predicted, the significant lack of foresight displayed by hospitals may be considered surprising in light of guidance stemming from previous outbreaks such as SARS, MERS, H1N1 and Ebola in the last decade. 12 Further, the World Health Organization (WHO) had been warning of a “Disease X”, a new and unknown pathogen capable of an outbreak of epidemic or pandemic proportions since 2016 and advising governments and healthcare systems to start creating preparedness plans. 42 Hospitals in the United States were struggling to prepare for the potential onslaught of COVID-19 patients and at that time, only 30% of nurses surveyed reported that their organizations had sufficient equipment inventory to respond to a COVID-19 surge. 12 Moreover, fewer than half of nurses felt that their employer had provided them with necessary information about how they should respond to the virus. 12 One physician likened her experience to being “played for a fool” because as infection rates and case severity rose, she was told to use lower levels of protection (e.g., using a surgical mask instead of an N95) and re-use single-use PPE. 43 Further, some healthcare workers in the United States stated that they faced termination for publicizing their poor working conditions and inadequate personal protective equipment due to their corporations’ efforts to maintain a corporate image. Physicians voiced that they have a duty to speak up about threats to patient safety and were unfairly punished by their organizations for doing so. 44 Given that contempt arises when a party’s (in this case, the healthcare organization) values and priorities are deemed lacking, we argue that each of these failing actions (i.e., forcing undesirable job tasks upon staff; providing low compensation; and lack of preparation) will incite feelings of contempt towards these workplaces. Proposition 1 Healthcare professionals’ will experience moral contempt related to their organizations’ response to the COVID-19 pandemic. Anger Employees experience moral anger when their organizations break with moral standards or commit transgressions without justification. 22 Throughout the pandemic, healthcare professionals have raised the alarm about inadequate and improper PPE that puts themselves and their patients at risk. 34 We argue that if healthcare professionals are going to risk their lives, there exists a reciprocal obligation, both on the principle of fairness and legislated through occupational health and safety acts, that their healthcare organizations will keep them safe. This perspective is supported by social exchange theory, which proposes that powerful, albeit ambiguous, rules dictate behaviour in situations where there are exchanges of resources such as time, money, energy, and other intangibles. We posit that healthcare professionals are likely to experience moral anger due to their organizations continuing to expect them to perform their duties with inadequate personal protection. For example, during the SARS pandemic, one study found that nurses expressed less anger when appropriate PPE was supplied as it was understood to be a display of organizational preparedness and support. 45 Although it may be argued that organizations are justified in their transgressions due to the scarcity of resources across the industry, we suggest that workers will perceive themselves to be unfair victims of this transgression, thus making the transgression unjustified. For example, when Kious Kelly, a nurse manager at a Manhattan hospital, succumbed to COVID-19 after having to wear a garbage bag in lieu of proper protective equipment, his colleagues were angry, calling his death preventable, and blaming the hospital system for failing to protect their innocent colleague. 46 Research has suggested that perceptions of unfairness can lead to moral anger. 22 Quarantine requirements of up to 14 days had been a near universal recommendation for COVID-19 positive individuals. While guidance from public health departments have evolved over the course of the pandemic, one thing has remained constant: if you feel sick, stay home. CDC recommendations included self-isolation mandates for those who are sick, those with known or suspected exposure to COVID-19, as well as individuals who have tested positive. 47 However, some hospitals forced COVID-19 positive healthcare professionals back to work sooner than public health guidelines suggested may be safe by suggesting that they would be fired if they did not return to work. 48 Further, employees who were sick with COVID-19 were instructed to return to work as long as they did not have a fever. 48 In another example of unfairness, when COVID-19 vaccines became available, there were instances in the rollout where non-patient facing staff, administrators or those working from home received vaccines before frontline doctors and nurses. In the case of Stanford Health Care, young frontline physicians and trainees were largely excluded from the first round of vaccinations in favour of senior staff or specialists who were not directly involved in patient care. 49 Similarly, in Toronto, a public relations specialist posted a photo of themselves getting a vaccine, which lead to widespread accusations of “queue jumping” by administrators ahead of those at the highest risk – frontline healthcare providers. 50 We argue that such inequality imposed upon employees by their healthcare organizations - particularly those that appear contrary to public health and CDC guidelines - are likely to lead healthcare workers to experience a sense of broken moral standards, resulting in moral anger. Proposition 2 Healthcare professionals will experience moral anger related to their organizations’ response to the COVID-19 pandemic. Disgust Disgust, in its primordial form, is a defense against contaminants. 9 Although healthcare professionals are potentially exposed to pathogens in the everyday practice of patient care, the working conditions surrounding COVID-19 are unique and have served to elevate these exposures. Healthcare professionals are trained in universal precautions (i.e., standard guidelines aimed at preventing infection), which involves the routine use of PPE and handwashing to prevent the transmission of disease. 51 However, during the current pandemic, a lack of adequate PPE left healthcare providers unable to protect themselves in keeping with their knowledge and training of effective infection control protocols. 11 For example, healthcare providers have been asked to re-use PPE intended for one-time use until it is “visibly soiled” or “no longer retains structural integrity” in efforts to conserve dwindling supplies, 34 effectively increasing their risk of infection. Researchers have illustrated that concerns regarding hygiene, cleanliness, or death, act as antecedents to disgust, 22 suggesting that the conditions created by healthcare organizations are likely to incite feelings of disgust related to physical contamination. Second, the sociomoral aspect of disgust is elicited by hypocrisy and disloyalty. 29 We argue that demonstrations by healthcare organizations and governments alike touting the heroism of healthcare workers while expecting them to work with inadequate PPE 52 is likely to result in employee perceptions of organizational hypocrisy and betrayal. The prevailing narrative of healthcare heroism centers around healthcare workers sacrificing themselves, like soldiers, for the good of others. 53 Healthcare workers have expressed that the more dangerous the working conditions, the more the “hero” moniker has been used. 54 As suggested in the popular press,12,54 we argue that this moniker draws attention to the bravery and altruism of healthcare workers rather than the continued existence of hazardous working conditions that may have been avoided with proper planning, adequate PPE, and staffing. Proposition 3 Healthcare professionals will experience disgust related to their organizations’ response to the COVID-19 pandemic. Influences of moral emotions on organizational commitment Organizational commitment: Affective Affective organizational commitment relates to employees’ emotional attachment and identification with an organization. 14 Healthcare professionals, notably nurses, are increasingly exhibiting low levels of affective commitment towards their healthcare organizations, 55 making them more likely to suffer decreased affective commitment when faced with negative moral emotions. Given that moral anger is an emotion that condemns others’ behavior and judges that behavior offensive, 56 we suggest that this experience will decrease any felt affection or warmth towards one’s organization. Second, Haidt 9 describes contempt as a cool emotion where the object of contempt is treated with less warmth, respect, and consideration in future interactions. Contempt leads individuals to distance themselves from the source of this emotion, particularly when they believe that the source is to blame for its actions. 56 This leads us to suggest that experiencing contempt towards one’s organization, particularly when healthcare professionals blame their organization for an inability to deliver the quality of patient care they desire, is likely to reduce their sense of organizational identification. Lastly, studies have shown that disgust often triggers avoidance behaviours since individuals do not want to be in proximity, physically or otherwise, to a disgusting object or idea. 9 We suggest that the negative moral emotions of anger, contempt and disgust will lead healthcare professionals to lose their emotional attachment to their organizations. Proposition 4 Healthcare professionals’ moral emotions of anger, contempt and disgust will result in reduced affective organizational commitment. Organizational commitment: Continuance Continuance commitment is often seen in employees who have few or no other job alternatives. 14 Based on the idea that employees have made an investment in the organization and leaving would result in losses, 57 continuance commitment occurs most commonly for those working in low demand sectors and is focused on the employer-employee economic exchange relationship. 58 Given that hospitals have suffered from understaffing for some time, and the pandemic has only served to exacerbate these effects, 59 continuance commitment is likely to have been low for healthcare professionals even without resource deficiencies and mismanagement. Moral anger is associated with a tendency to want to punish transgressors (in this case, one’s place of employment). 56 We argue that this desire to punish may result in the consideration of leaving one’s organization, or at least reducing commitment to working within that organization long-term. Further, since morally relevant feelings of contempt can arise in response to poor compensation and benefits, 10 we argue that when positions offering higher compensation both in terms of monetary and non-monetary benefits arise, healthcare professionals are likely to view their own organizations’ values as lacking by choosing to prioritize other concerns over their well-being. Proposition 5 Healthcare professionals’ moral emotions of contempt, anger and disgust will result in reduced continuance commitment. Organizational commitment: Normative Normative organizational commitment is a rational attachment to an organization characterized by employees’ feelings of obligation to remain. 14 These feelings of obligation can result from the internalization of normative pressures exerted on individuals prior to or after entry into an organization and generally denote an acceptance and identification with organizational goals. 17 The goal of most healthcare organizations is the care of the patients that they serve 60 and this organizational goal overlaps with the ethical duties and responsibilities of healthcare providers to treat patients, even during extreme circumstances. 61 While there has been some debate about the limits of healthcare providers’ duty to serve during a pandemic, 11 most healthcare providers believe they should continue to provide care even in highly adverse conditions as a result of their specialized knowledge and training.62,63 This is particularly salient for physicians, who take an oath to serve the public good. 11 Healthcare professionals have consistently demonstrated their normative commitment by continuing to perform their duties and treat patients regardless of hospital working conditions. Negative emotions play a critical role in individual’s cognitive activation process - raising awareness and directing an individual’s attention. 64 We argue that the negative emotions of contempt, disgust and anger will increase commitment to the organizational goal of patient care, and that healthcare professionals will be increasingly aware of the difficulty faced by these patients in pandemic conditions. We suggest that the lack of resources (in terms of time, energy, and protective equipment) which results in negative feelings will also act to increase empathy and awareness of patients’ suffering due to this same lack of resources. This experience could result in increased commitment to those being cared for by healthcare professionals, and thus, increase normative commitment to organizations which allow them, to the best of their ability, to provide care to patients. Proposition 6 Healthcare professionals’ moral emotions of contempt, anger and disgust will result in higher normative organizational commitment. Influences of moral emotions on professional commitment Professional commitment Professional commitment describes the congruency between an individual’s personal beliefs and the goals of their profession, where greater congruence leads to greater efforts to help the profession and its members. 65 We suggest that negative feelings towards one’s organization will serve to increase engagement within one’s profession. First, negative emotions direct behaviour towards victims, and anger in particular leads to actions of victim-directed support. 25 Healthcare professionals have demonstrated considerable support for each other during the pandemic since they are all similarly victims of the same situation. For example, with PPE in critically short supply, healthcare professionals took it upon themselves to either make PPE (e.g. fabric masks, eye-protection and scrub caps), petition their governments, or organize public drives for protective equipment on social media.34,66 Healthcare professionals are also mindful of ways that they can protect each other from unnecessary exposure to the virus. This has included using unconventional protective equipment, such as covering patients with large plastic bags during aerosolizing procedures 67 or changing established procedures. 68 The latter has included limiting the frequency of high-risk procedures (e.g., bronchoscopy), where medically appropriate, to limit disease spread and exposure to their colleagues. 68 Nurses have likened the COVID-19 experience to that of working on a battlefield and acknowledge the importance of caring for their coworkers and sharing the load. 69 Research has shown that shared difficult experiences bring people closer together. 70 COVID-19 has presented a collective challenge for all healthcare professionals such that nurses in Wuhan reported a heightened sense of professional collegiality during COVID-19. 62 Increased collegiality between nurses is consistent with reports from previous pandemics.71,72 Extant research has shown that stronger collegial ties result in stronger professional commitment. 21 Knowing that others within their profession are suffering from the same negative pandemic organizational experiences, we would expect that healthcare professionals would feel more strongly tied to their colleagues and feel a stronger sense of commitment towards them. Being part of profession and/or professional association brings certain privileges, resources, and powers.73,74 For example, studies have found that being part of a nursing association brings nurses social benefits and support for their ideas.75,76 As noted earlier, the moral anger, contempt and disgust experienced by healthcare workers is likely to, at least partially, emerge from feeling undervalued and unappreciated by their organizations. Thus, we secondly argue that if one’s sense of value is diminished within one’s own place of work, one may search for other sources of strength – such as within one’s profession. In terms of the effects of moral anger, the experience of anger as an emotion is subject to moderation. For example, when witnessing a mobbing in the workplace, the degree to which others experience moral anger was moderated by the how much they believed the victim was deserving of the mistreatment. 77 Given that healthcare professionals are unlikely to view others in their profession as deserving of their difficult workplace conditions, we suggest that anger will make suffering colleagues appear more deserving of sympathy and thus increase felt commitment towards them. Contempt can arise due to disdain of those who have violated the norms or ethic of a community. 29 In this case, workers who feel that their organizations have breached the norms of the medical and health community may search for a community that retains these ethical norms – most immediately, their peers. In this way, we suggest that the emotion of contempt towards one’s organization may lead to increased commitment to one’s profession. Finally, in terms of disgust, caring for another’s personal hygiene or bodily functions elicits this emotion. 78 Disgust is often applicable to occupations that are considered to be dirty, as is in the case of coroners or undertakers. 22 Most healthcare professionals are accustomed to disgust in their everyday work life, 79 as part of their profession – however, they are less likely to be familiar with suffering from sociomoral disgust arising from the hypocrisy and betrayal shown by their organizations. We argue that this novel experience is likely to cause such individuals to seek support from those who are in similar, sympathetic situations – their professional peers. Proposition 7 Healthcare professionals’ moral emotions of contempt, anger and disgust will increase their professional commitment to each other. Discussion Theoretical implications The current study contributes to our understanding of the antecedents and consequences of workplace emotions during times of crisis. We build on, and extend, existing literature in emotions and workplace commitment in three ways. First, this study contributes to our knowledge of how moral emotions may directly affect workplace attitudes and outcomes. The impact of employees’ affective commitment to organizations has been investigated, as well as the influence of employees’ moral judgements and virtues. 17 However, to our knowledge, moral emotions have not yet been specifically considered as potential antecedents of organizational commitment. There has been much ongoing debate about the emotions of anger, disgust, and contempt as separate constructs,29,31 our study advances the line of thought that each of these emotions can be examined in isolation as we show that each may result from unique workplace antecedents. For example, we suggest that ignoring public health rulings and asking employees to continue to work when rules are transgressed may trigger anger; that showing weak character during times of crisis may result in contempt; and that the use of the ‘hero’ moniker to describe workers forced to operate in unsanitary conditions may lead to disgust. Second, the current conceptual study contributes to a growing body of literature that illustrates how “feeling bad can do good”. 80 Previous literature around the benefits of negative emotions in the workplace showed that the sharing of negative events lessened and mitigated individuals’ unpleasant feelings. 81 Negative moral emotions have been frequently studied and cited for their deleterious effects, but our study argues that they can also have positive consequences for group members. For example, negative moral emotions could be a mechanism through which in-group loyalty, solidarity and connection is fostered in employees. Third, our model contributes to the workplace commitment literature with separate implications for affective, normative and continuance commitment. There have been studies demonstrating the impact of moral distress and moral injury on healthcare professionals’ affective commitment, 82 but little consideration of the impact of moral emotions on commitment. Our study offers a contribution to the affective commitment literature by way of exploring additional antecedents to healthcare professionals’ affective commitment, which are of particular salience to organizations during the COVID-19 pandemic. Studies have found that breaches in psychological contracts or exchange relationships between employees and their organizations have resulted in decreased affective commitment.83,84 In our model, we used the failure of some healthcare organizations to provide the PPE for their healthcare workers as an illustrative example of a breach in their psychological contract and would predict ensuing decreases in healthcare professionals' affective commitment. Current understanding and research of normative commitment has highlighted the importance of employees’ feelings of obligation to their organization14,85 with investigations into the effects of organizational trust, managerial relationships, work structure and justice. 86 However, our study suggests healthcare professionals’ normative commitment can be increased, not necessarily as a result of intra-organizational factors, but as a result of their professional commitment and their duty to the public. Healthcare professionals need the environment and resources that healthcare organizations provide in order to continue to serve the public. Thus, they remain at their organizations, not from a perceived obligation to the organization itself, but rather because it is a way for them to maintain their duty to the public. Healthcare organizations may benefit from increased normative commitment in this way, as they will maintain a normatively committed workforce despite less-than-ideal organizational conditions. Our study also has the potential to contribute to the organizational citizenship behaviour (OCB) literature. Previously investigated antecedents to OCBs included employee, task, organizational and leadership characteristics. 87 There has been little investigation to date on how moral emotions might lead to OCBs. Researchers have investigated the effects of employees’ discrete emotions, moods, and dispositional traits on organizational outcomes (e.g., performance, creativity, group dynamics and leadership). 88 These investigations showed that positive moods and positive affect demonstrated greater influences on workplace outcomes than their negative counterparts.88,89 The influence of negative affect on organizational life is complex and substantial given how strongly negative emotions are experienced. Our study would argue that negative moral emotions may have opposing effects on OCB-I (individual-focused OCBs) and OCB-O (organization-focused OCBs), 90 such that healthcare workers are more likely to act pro-socially towards fellow employees within their organization or profession and less likely to engage in positive, voluntary behaviors that help their organizations during a pandemic. Practical implications This paper has important practice implications for healthcare organizations during COVID-19, as well as for any future health crises. Some sources of healthcare workers’ negative moral emotions cannot be easily or immediately fixed (e.g., global PPE shortages), but managers within healthcare settings can take concrete steps to temper the sequalae that these moral emotions have on their employees and stakeholders. Managers should take decisive steps to remedy the underlying causes of their healthcare professionals’ negative moral emotions. Healthcare management learnings from previous pandemics have touted the importance of providing emotional and psychological support to healthcare professionals to foster emotional resilience and the ability to deal with stressful situations. 91 We have three suggestions for managers. First, the provision of positive feedback – both at the direct and senior manager level - can act as a proxy for organizational support to help diminish employees’ feelings of anger. 45 Second, contempt may be diminished through the clear communication of the values of the organization and detailing how the division of resources aligns with these values. Third, changing the rhetoric used by hospital administrators concerning the heroism of healthcare professionals, and instead acknowledging the difficulties they face, and outlining steps that the organization has, and will, take to reduce these difficulties, may serve to decrease feelings of disgust. Our study suggests that the strongest risk of current pandemic working conditions is to employees’ continuance commitment. This is demonstrated by 22% of nurses indicating that they intend to leave their current position, which is up from 15.9% in 2019 prior to the pandemic. 92 With lucrative travel contracts available to healthcare professionals, coupled with low or decreased pay at certain organizations, healthcare professionals’ perceived costs of leaving their current organizations will likely be lower than pre-pandemic conditions. Additionally, having experienced workplace exposures to COVID-19, particularly without adequate PPE, leaves healthcare workers unwilling to share physical space with their families and loved ones after work62,93 due to fears of being contaminated and acting as a disease vector. This is likely to also contribute to healthcare professionals’ inclinations to leave frontline workplaces. As research has shown repeatedly that perceived inequities result in greater employee attrition and turnover 94 and pay cuts and job losses have been reported to disproportionately affect front-line workers during the pandemic,37,41 managers must become cognizant that these healthcare professionals are likely to take concrete actions to minimize these occurences or leave the frontlines altogether. Another practical implication of the current study is that healthcare organizations and healthcare leaders should recognize and harness the paradoxical benefits that shared negative experiences may have for their employees in terms of group cohesion commitment. In order to do this, healthcare organizations should understand the actions that are needed to achieve positive outcomes out of negative work events. This includes reinforcing the pattern of professional commitment that healthcare workers develop with one another. Commitment to a professional group does not eliminate the possibility of concurrent organizational commitment 21 and rather, should be beneficial to the organization as well, particularly when employees perceive low conflict between the two.95–97 Building on Hadley’s (2014) findings on emotional disclosure at work, healthcare organizations should foster environments and opportunities in which their workers can process and share their negative moral emotions. This could include promoting weekly or monthly staff meetings that allow for a diverse display of emotions and supporting those displays in a non-punitive manner. Working on the frontlines of a pandemic is distressing for healthcare workers and they experience high rates of fear, sadness, depression and anxiety. 98 Healthcare professionals are at high risk of poor mental health outcomes and may need psychological resources and support. 99 In previous studies, actions directed towards colleagues were found to negatively impact job stress, increase job satisfaction, and act as a surrogate for organizational support. 100 Performing altruistic or helping acts generates positive emotions in individuals who receive help and creates an urge to reciprocate in a similar fashion.101,102 Experiences in helping or gratitude form a self-renewing basis for cooperative relationships and communal support. 102 Healthcare professionals helping one another during COVID-19 is a way in which they can create positive emotions for themselves. In the broaden-and-build model of positive emotions, it was proposed that positive emotions act as an antidote to negative emotions by dismantling their psychological and physiological effects. 102 Healthcare professionals could use the positive emotions they receive from helping their colleagues as a way to cope and recover from their psychological and emotional distress induced by the COVID-19 pandemic. Managers should foster and encourage opportunities for colleagues to support one another, perhaps in the form of peer-support groups. Haidt (2003) described anger as the most underappreciated moral emotion because it is typically associated with motivations to attack, humiliate, or get revenge. Anger is frequently used as an exemplar of a negative emotion due to its connection with adverse effects in the workplace such as aggression, violence, bullying and, deviance. 103 However, anger can serve as an important catalyst for prosocial change. Moral anger propels individuals to stop perpetrators of misconduct, 31 promotes victim-directed support, leads to reconciliation, improves relationships, and can lead to whistleblowing.22,103 Managers should try to harness the power of this emotion to create an ethical organizational climate that creates opportunities for voice and speaking up to make positive changes, particularly in times of crises. Lastly, managers should address the inequities in the exchange relationship between themselves and their employees that can lead to negative moral emotions. Since the psychological contract inherent in organizational commitment involves equal exchanges, healthcare organizations could implement strategies that would help to allay their workers’ fears around COVID-19 infection. This could involve policies that redeploy healthcare workers at elevated risk of COVID-19 infection away from high-risk areas or creating programs that provide supplementary housing to healthcare workers wishing to protect their families, 104 thus providing an increased sense of balance between the efforts and inputs asked of employees and their working conditions and compensation. Limitations and future directions The work presented here is theoretical in nature. Empirical testing of this model is required to determine its validity and ability to predict how healthcare workers organizational commitment has changed due to COVID-19. Since we suggest that it is possible for healthcare professionals to have reasons to increase their normative and professional commitments while also experiencing potentially large decrease in their affective and/or continuance commitments, an avenue for future research would be to determine the relative size and magnitude of these relationships and how those leaving their organizations or professions experienced reductions or increases in their types of commitment. In particular, given the duration of the COVID-19 pandemic and its prolonged impact on healthcare organizations, investigating how organizational and professional commitment changes over the course of the same crisis event should be studied. We also consider only healthcare professionals, primarily in hospital environments, as the experience of caring for COVID-19 patients has been largely experienced there or other acute care settings; we acknowledge that professionals' experience in other healthcare or non-healthcare settings may be different. Cultural variations on the experience of emotions and organizational commitment were not explored here. There is evidence to indicate that the experience of emotions, particularly moral ones, is cross-cultural, 22 but this is not sufficient to conclude that changes in organizational commitment will be uniform worldwide. Similarly, how employees may be affected by the pandemic as individuals outside of the employment context (e.g., suffering from anxiety, worry or depression in their personal lives 105 ) and how those experiences would then impact their professional experience, was beyond the scope of the current study, but interactive effects would provide a worthwhile venue for examination. Moral events may simultaneously activate multiple emotions from different moral emotion families, so future studies should investigate the impact of other moral emotions such as guilt or shame. Pandemics have also been shown to evoke non-moral emotions such as fear, sadness, anxiety, and depression. Subsequent studies can examine the effects of these nonmoral emotions on healthcare professionals’ organizational commitment. Exploring the differences between moral and non-moral anger in organizational contexts is another promising direction for future research. From a standpoint of theoretical contributions, we suggest that exploration of pandemic-specific situations such as the impact of unpredictability and munificence would be worthwhile avenues of future consideration. Exploring other organizational outcome variables would also be valuable. For example, the impact of the pandemic on unethical behaviours (e.g., theft) as well as the effects of solidarity between specific roles on interprofessional teams on professional commitment may be worth further study. Increases in professional commitment and in-group solidarity may have unintended negative consequences (e.g., tribalism), which warrant investigation. These variables can be investigated at the individual, group, and organizational level. Understanding the outcomes of healthcare professionals' moral emotions will provide practical information for organizations in order to sustain and improve their performance through not only the COVID-19 pandemic but various other unpredictable, complex organizational crisis events in the future. ORCID iDs Morgan Davidson https://orcid.org/0000-0002-3890-414X Meena Andiappan https://orcid.org/0000-0002-4713-508X The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding: The author(s) received no financial support for the research, authorship, and/or publication of this article. ==== Refs References 1 World Health Organization . WHO Director-General's opening remarks at the media briefing on COVID-19–11. Geneva: World Health Organization, 2020. Available from https://www.who.int/dg/speeches/detail/who-director-general-s-opening-remarks-at-the-media-briefing-on-covid-19–-11-march-2020 2 Cowling BJ Aiello AE . Public Health Measures to Slow Community Spread of Coronavirus Disease 2019. J Infect Dis 2020; 221 (11 ): 1749–1751.32193550 3 Coronavirus disease . 2019 (COVID-19): situation Report — 201 [press release]. Geneva: World Health Organization, 2020. 4 Ranney ML Griffeth V Jha AK . Critical Supply Shortages — The need for ventilators and personal protective equipment during the Covid-19 Pandemic. N Engl J Med 2020; 382 (18 ): e41.32212516 5 Grasselli G Pesenti A Cecconi M . Critical care utilization for the COVID-19 Outbreak in Lombardy, Italy: Early experience and forecast during an emergency response. Journal of the American Medical Association 2020; 323 (16 ): 1545–1546.32167538 6 Kurihara H Bisagni P Faccincani R , et al. COVID-19 outbreak in Northern Italy: Viewpoint of the Milan area surgical community. J Trauma Acute Care Surg 2020; 88 : 719–724.32267661 7 Mann B . Nurses left vulnerable to COVID-19: 'we're not martyrs sacrificing our lives'. National Public Radio News, 2020. Available from https://www.npr.org/2020/05/02/848997142/nurses-left-vulnerable-to-covid-19-we-re-not-martyrs-sacrificing-our-lives 8 Snyder P Hall M Robertson J , et al. Ethical Rationality: A Strategic Approach to Organizational Crisis. J Bus Ethics 2006; 63 (4 ): 371–383. 9 Haidt J . The Moral Emotions. In: Davidson RJ Scherer KR Goldsmith HH (eds). Handbook of Affective Sciences, 2003 . Oxford: Oxford University Press, 2003, pp. 852–870. 10 Ford MT Agosta JP Huang J , et al. Moral emotions toward others at work and implications for employee behavior: A qualitative analysis using critical incidents. J Bus Psychol 2018; 33 (1 ): 155–180. 11 Schuklenk U . What healthcare professionals owe us: why their duty to treat during a pandemic is contingent on personal protective equipment (PPE). J Med Ethics 2020; 46 (7 ): 432–435.32444425 12 Mason DJ Friese CR . Protecting Health Care Workers Against COVID-19—and Being Prepared for Future Pandemics. JAMA Health Forum 2020; 1 (3 ): e200353. Available from https://jamanetwork.com/channels/health-forum/fullarticle/2763478 36218598 13 Hoogenboom WS Pham A Anand H , et al. Clinical characteristics of the first and second COVID-19 waves in the Bronx, New York: A retrospective cohort study. The Lancet Regional Health – Americas 2021; 3 : 100041.34423331 14 Meyer JP Allen NJ . A three-component conceptualization of organizational commitment. Human Resource Management Review 1991; 1 (1 ): 61–89. 15 Markovits Y Boer D van Dick R . Economic crisis and the employee: The effects of economic crisis on employee job satisfaction, commitment, and self-regulation. European Management Journal 2014; 32 (3 ): 413–422. 16 Meyer JP Morin AJS Wasti SA . Employee commitment before and after an economic crisis: A stringent test of profile similarity. Human Relations 2017; 71 (9 ): 1204–1233. 17 González TF Guillén M . Organizational commitment: a proposal for a wider ethical conceptualization of 'Normative Commitment'. J Bus Ethics 2008; 78 (3 ): 401–414. 18 Wright AL Zammuto RF Liesch PW . Maintaining the Values of a Profession: Institutional Work and Moral Emotions in the Emergency Department. Acad Manage J 2017; 60 (1 ): 200–237. 19 Quinn RW Spreitzer GM Lam CF . Building a sustainable model of human energy in organizations: Exploring the critical role of resources. Acad Manag Ann 2012; 6 (1 ): 337–396. 20 Collins R . Emotional Energy as the Common Denominator of Rational Action. Rationality and Society 1993; 5 (2 ): 203–230. 21 Wallace JE . Organizational and Professional Commitment in Professional and Nonprofessional Organizations. Administrative Science Quarterly 1995; 40 (2 ): 228–255. 22 Greenbaum R Bonner J Gray T , et al. Moral emotions: A review and research agenda for management scholarship. J Organ Behav 2020; 41 (2 ): 95–114. 23 Ekman P . An argument for basic emotions. Cogn Emot 1992; 6 (3–4 ): 169–200. 24 Huang Y-SS Greenbaum RL Bonner JM , et al. Why sabotage customers who mistreat you? Activated hostility and subsequent devaluation of targets as a moral disengagement mechanism. J Appl Psychol 2019; 104 (4 ): 495–510.30321032 25 Tangney JP Stuewig J Mashek DJ . Moral emotions and moral behavior. Annu Rev Psychol 2007; 58 : 345–372.16953797 26 Treviño LK Den Nieuwenboer NA Kish-Gephart JJ . (Un) ethical behavior in organizations. Annu Rev Psychol 2014; 65 : 635–660.23834354 27 Prinz JJ Nichols S . Moral emotions. In: Doris JM (ed). The Moral Psychology Handbook. Oxford, UK: Oxford University Press, 2010. 28 Turner JH Stets JE . Sociological theories of human emotions. Annu Rev Sociol 2006; 32 : 25–52. 29 Rozin P Lowery L Imada S , et al. The CAD triad hypothesis: a mapping between three moral emotions (contempt, anger, disgust) and three moral codes (community, autonomy, divinity). J Pers Soc Psychol 1999; 76 (4 ): 574–586.10234846 30 Oaten M Stevenson RJ Williams MA , et al. Moral violations and the experience of disgust and anger. Front Behav Neurosci 2018; 12 : 179.30214400 31 Hutcherson CA Gross JJ . The moral emotions: A social–functionalist account of anger, disgust, and contempt. J Pers Soc Psychol 2011; 100 (4 ): 719–737.21280963 32 Borges LM Barnes SM Farnsworth JK , et al. A commentary on moral injury among health care providers during the COVID-19 pandemic. Psychol Trauma 2020; 12 : S138–S140.32496101 33 Ripp J Peccoralo L Charney D . Attending to the emotional well-being of the health care workforce in a New York City health system during the COVID-19 Pandemic. Academic Medicine 2020; 95 (8 ): 1136–1139.32282344 34 Jacobs A Richtel M Baker M . ‘At war with No ammo’: doctors say shortage of protective gear is dire New York city: the New York times; 2020, 2020. Available from https://www.nytimes.com/2020/03/19/health/coronavirus-masks-shortage.html 35 Vesoulis A . I’m mentally and physically exhausted.’ healthcare workers battling coronavirus are running out of protective gear: time, 2020. Available from https://time.com/5808992/healthcare-workers-lack-ppe/ 36 Li Z Daspit JJ Marler LE . Executive pay dispersion: Reconciling the differing effects of pay inequality and pay inequity on firm performance. The International Journal of Human Resource Management 2021; 33 : 3056–3084. 37 Gabler E Montague Z Ashford G . During a pandemic, an unanticipated problem: out-of-work health workers New York city the New York times, 2020. Available from https://www.nytimes.com/2020/04/03/us/politics/coronavirus-health-care-workers-layoffs.html 38 Froese I . Bonus pay on the way for nearly 80,000 Manitoba front-line pandemic workers CBC News: Canadian Broadcasting Company, 2020. Available from https://www.cbc.ca/news/canada/manitoba/manitoba-covid19-pandemic-risk-payment-1.5667312 39 The Canadian Press . Pandemic pay premium has not yet been given to health-care workers, Ontario government says. The National Post 2020. Available from https://nationalpost.com/news/pandemic-pay-premium-has-not-flowed-to-workers-ontario-government-says 40 Shepherd L . Front-line workers receive bonus payments: society for human resource management, 2022. Available from https://www.shrm.org/resourcesandtools/legal-and-compliance/employment-law/pages/frontline-worker-pandemic-bonus-hero-pay.aspx 41 Silver-Greenberg J Drucker J Enrich D . Hospitals got bailouts and furloughed thousands while paying C.E.O.s millions New York: the New York times, 2020. Available from https://www.nytimes.com/2020/06/08/business/hospitals-bailouts-ceo-pay.html 42 Simpson S Kaufmann MC Glozman V , et al. Disease X: accelerating the development of medical countermeasures for the next pandemic. Lancet Infect Dis 2020; 20 : e108–e115.32197097 43 Time . 'Please and God. Just Cover Me.' Health Care Workers Are Risking Their Lives Daily in the Fight Against Coronavirus. Time 2020. Available from https://time.com/collection/coronavirus-heroes/5816883/health-care-workers-coronavirus/ 44 Carville O Court E Brown KV . Hospitals tell doctors they’ll Be fired if they speak out about lack of gear: bloomberg news; 2020, 2020. Available from https://www.bloomberg.com/news/articles/2020-03-31/hospitals-tell-doctors-they-ll-be-fired-if-they-talk-to-press 45 Fiksenbaum L Marjanovic Z Greenglass ER , et al. Emotional Exhaustion and State Anger in Nurses Who Worked During the Sars Outbreak: The Role of Perceived Threat and Organizational Support. Canadian Journal of Community Mental Health 2006; 25 (2 ): 89–103. 46 Sengupta S . A N.Y. Nurse dies. Angry Co-workers blame a lack of protective gear. New York City: The New York Times, 2020. Available from https://www.nytimes.com/2020/03/26/nyregion/nurse-dies-coronavirus-mount-sinai.html 47 Centers for Disease Control and Prevention . When to quarantine 2020, 2020. Available from https://www.cdc.gov/coronavirus/2019-ncov/if-you-are-sick/quarantine.html 48 Huetteman E . US hospitals pressure healthcare staff to work even if they have Covid symptoms: the Guardian, 2020. Available from https://www.theguardian.com/us-news/2020/aug/13/us-nurses-doctors-covid-19-symptoms-working 49 Bernstein L Beachum L. Knowles H . Stanford apologizes for coronavirus vaccine plan that left out many front-line doctors: the Washington Post, 2020. Available from https://www.washingtonpost.com/health/2020/12/18/stanford-hospital-protest-covid-vaccine/ 50 Forani J . Why some hospital PR executives and staff on leave have received COVID-19 vaccines before front-line staff: CTV News: Canadian Television News, 2021. Available from https://www.ctvnews.ca/health/coronavirus/why-some-hospital-pr-executives-and-staff-on-leave-have-received-covid-19-vaccines-before-front-line-staff-1.5262295 51 Ndu AC Arinze-Onyia SU . Standard precaution knowledge and adherence: Do Doctors differ from Medical Laboratory Scientists? Malawi Med J 2017; 29 (4 ): 294–300.29963283 52 Kuznia R Black N Ortega B , et al. Health care workers on frontlines feel like 'lambs to the slaughterhouse'. Cable News Network 2020. Available from https://www.cnn.com/2020/03/23/us/coronavirus-healthcare-supply-shortage-workers-at-risk-invs/index.html 53 Perez M . Trump on healthcare workers: 'They're running into death just like soldiers run into bullets'. Forbes 2020. Available from https://www.forbes.com/sites/mattperez/2020/05/14/trump-on-healthcare-workers-theyre-running-into-death-just-like-soldiers-run-into-bullets/#51793bfa4a00 54 Wexler A . The trouble with calling health care workers ‘heroes’ the Boston Globe, 2020. Available from https://www.bostonglobe.com/2020/04/10/opinion/trouble-with-calling-health-care-workers-heroes/ 55 Brunetto Y Farr-Wharton R Shacklock K . Communication, training, well-being, and commitment across nurse generations. Nurs Outlook 2012; 60 (1 ): 7–15.21703652 56 Fischer AH Roseman IJ . Beat them or ban them: The characteristics and social functions of anger and contempt. J Pers Soc Psychol 2007; 93 (1 ): 103–115.17605592 57 Sabir MS Sohail A Khan MA . Impact of leadership style on organization commitment: in a mediating role of employee values. Journal of Economics and Behavioral Studies 2011; 3 (2 ): 145–152. 58 Taing MU Granger BP Groff KW , et al. The multidimensional nature of continuance commitment: Commitment owing to economic exchanges versus lack of employment alternatives. J Bus Psychol 2011; 26 (3 ): 269–284. 59 Lasater KB Aiken LH Sloane DM , et al. Chronic hospital nurse understaffing meets COVID-19: an observational study. British Medical Journal Quality & Safety:bmjqs-2020–011512. 60 Spencer EM Mills AE Rorty MV , et al. Organization ethics in health care. Oxford, UK: Oxford University Press, 2000. 61 American Medical A . AMA code of medical ethics: guidance in a pandemic, 2020. Available from https://www.ama-assn.org/delivering-care/ethics/ama-code-medical-ethics-guidance-pandemic]. 62 Liu Q Luo D Haase JE , et al. The experiences of health-care providers during the COVID-19 crisis in China: a qualitative study. Lancet Glob Health 2020; 8 : e790–e798.32573443 63 Sokol DK . Virulent Epidemics and Scope of Healthcare Workers' Duty of Care. Emerg Infect Dis 2006; 12 (8 ): 1238–1241.16965703 64 Costello B (eds). Lingering serious experience as trigger to raise awareness, encourage reflection and change behavior. Persuasive Technology. Springer Berlin Heidelberg, 2013. 65 Chang HY Shyu YIL Wong MK , et al. Which aspects of professional commitment can effectively retain nurses in the nursing profession? J Nurs Scholarsh 2015; 47 (5 ): 468–476.26219346 66 Elgin B Tozzi J . Hospital workers make masks from office supplies amid U.S. Shortage: Bloomberg News, 2020. Available from https://www.bloomberg.com/news/articles/2020-03-18/hospital-makes-face-masks-covid-19-shields-from-office-supplies 67 Favaro A Phillip ES Cousins B . Doctors are finding creative ways to treat patients with COVID-19: Canadian Television News, 2020. Available from https://www.ctvnews.ca/health/coronavirus/doctors-are-finding-creative-ways-to-treat-patients-with-covid-19-1.4870525 68 Pritchett MA Oberg CL Belanger A , et al. Society for Advanced Bronchoscopy Consensus Statement and Guidelines for bronchoscopy and airway management amid the COVID-19 pandemic. J Thorac Dis 2020; 12 (5 ): 1781–1798.32642084 69 Chung BPM Wong TKS Suen ESB , et al. SARS: caring for patients in Hong Kong. J Clin Nurs 2005; 14 (4 ): 510–517.15807759 70 Bastian B Jetten J Ferris LJ . Pain as social glue: Shared pain increases cooperation. Psychol Sci 2014; 25 (11 ): 2079–2085.25193943 71 Chung C Cooper SJ Cant RP , et al. The educational impact of web-based and face-to-face patient deterioration simulation programs: An interventional trial. Nurse Educ Today 2018; 64 : 93–98.29459198 72 Fernandez PR Lord H Halcomb PE , et al. Implications for COVID-19: a systematic review of nurses’ experiences of working in acute care hospital settings during a respiratory pandemic. International Journal of Nursing Studies 2020; 2020 : 103637. 73 Ki EJ Wang Y . Membership benefits matter: Exploring the factors influencing members’ behavioral intentions in professional associations. Nonprofit Management and Leadership 2016; 27 (2 ): 199–217. 74 Mata H Latham TP Ransome Y . Benefits of professional organization membership and participation in national conferences: Considerations for students and new professionals. Health Promot Pract 2010; 11 (4 ): 450–453.20689051 75 White MJ Olson RS . Factors affecting membership in specialty nursing organizations. Rehabil Nurs 2004; 29 (4 ): 131–137.15222095 76 Yeager SJ Kline M . Professional association membership of nurses: Factors affecting membership and the decision to join an association. Res Nurs Health 1983; 6 (2 ): 45–52.6553942 77 Mitchell MS Vogel RM Folger R . Third parties’ reactions to the abusive supervision of coworkers. J Appl Psychol 2015; 100 (4 ): 1040–1055.25243999 78 Moll J de Oliveira-Souza R Moll FT , et al. The Moral Affiliations of Disgust: A Functional MRI Study Cognitive and Behavioral Neurology. 2005;18 (1 ). 79 Hadjittofi M Gleeson K Arber A . The experience of disgust by healthcare professionals: A literature review. Int J Nurs Stud 2020; 110 : 103720.32755714 80 Lindebaum D Jordan PJ . When it can be good to feel bad and bad to feel good: Exploring asymmetries in workplace emotional outcomes. Hum Relat 2014; 67 (9 ): 1037–1050.25418996 81 Hadley CN . Emotional roulette? Symmetrical and asymmetrical emotion regulation outcomes from coworker interactions about positive and negative work events. Human Relations 2014; 67 (9 ): 1073–1094. 82 Shoorideh FA Ashktorab T Yaghmaei F , et al. Relationship between ICU nurses' moral distress with burnout and anticipated turnover. Nurs Ethics 2015; 22 (1 ): 64–76.24948793 83 Coyle-Shapiro JAM Conway N . Exchange relationships: examining psychological contracts and perceived organizational support. J Appl Psychol 2005; 90 (4 ): 774–781.16060794 84 Quratulain S Khan AK Crawshaw JR , et al. A study of employee affective organizational commitment and retention in Pakistan: the roles of psychological contract breach and norms of reciprocity. The International Journal of Human Resource Management 2018; 29 (17 ): 2552–2579. 85 Meyer JP Stanley DJ Herscovitch L , et al. Affective, Continuance, and Normative Commitment to the Organization: A Meta-analysis of Antecedents, Correlates, and Consequences. Journal of Vocational Behavior 2002; 61 (1 ): 20–52. 86 Tekingündüz S Top M Tengilimoğlu D , et al. Effect of organisational trust, job satisfaction, individual variables on the organisational commitment in healthcare services. Total Quality Management & Business Excellence 2017; 28 (5–6 ): 522–541. 87 Podsakoff PM MacKenzie SB Paine JB , et al. Organizational Citizenship Behaviors: A Critical Review of the Theoretical and Empirical Literature and Suggestions for Future Research. Journal of Management 2000; 26 (3 ): 513–563. 88 Barsade SG Gibson DE . Why Does Affect Matter in Organizations? Acad Manag Perspect 2007; 21 (1 ): 36–59. 89 Thoresen CJ Kaplan SA Barsky AP , et al. The Affective Underpinnings of Job Perceptions and Attitudes: A Meta-Analytic Review and Integration. Psychol Bull 2003; 129 (6 ): 914–945.14599288 90 Podsakoff NP Whiting SW Podsakoff PM , et al. Individual- and organizational-level consequences of organizational citizenship behaviors: A meta-analysis. J Appl Psychol 2009; 94 (1 ): 122–141.19186900 91 Restubog SLD Ocampo ACG Wang L . Taking control amidst the chaos: Emotion regulation during the COVID-19 pandemic. J Vocat Behav 2020; 119 : 103440.32390659 92 Berlin G Lapointe M. Murphy M. , et al. Nursing in 2021: retaining the healthcare workforce when we need it most. New York: McKinsey & Company, 2021. Available from https://www.mckinsey.com/industries/healthcare-systems-and-services/our-insights/nursing-in-2021-retaining-the-healthcare-workforce-when-we-need-it-most 93 Guarino B . Nurses, doctors take extreme precautions to avoid infecting family members: the Washington Post, 2020. Available from https://www.washingtonpost.com/health/2020/04/16/nurses-doctors-are-taking-extreme-precautions-avoid-bringing-coronavirus-home/ 94 Greenberg J . Employee theft as a reaction to underpayment inequity: The hidden cost of pay cuts. Journal of Applied Psychology 1990; 75 (5 ): 561–568. 95 Meyer JP Allen NJ Smith CA . Commitment to organizations and occupations: Extension and test of a three-component conceptualization. Journal of Applied Psychology 1993; 78 (4 ): 538–551. 96 Norris DR Niebuhr RE . Professionalism, organizational commitment and job satisfaction in an accounting organization. Accounting, Organizations and Society 1984; 9 (1 ): 49–59. 97 Shafer WE Park LJ Liao WM . Professionalism, organizational-professional conflict and work outcomes. Accounting, Auditing & Accountability Journal 2002; 15 (1 ): 46–68. 98 Khee KS Lee LB Chai OT , et al. The psychological impact of SARS on health care providers. Critical Care and Shock 2004: 100–106. 99 Lai J Ma S Wang Y , et al. Factors Associated With Mental Health Outcomes Among Health Care Workers Exposed to Coronavirus Disease 2019. JAMA Netw Open 2020; 3 (3 ): e203976.32202646 100 Tsang S-S Chen T-Y Wang S-F , et al. Nursing work stress: The impacts of social network structure and organizational citizenship behavior. J Nurs Res 2012; 20 (1 ): 9–18.22333962 101 Isen AM . Positive affect, cognitive processes, and social behavior. In: Advances in Experimental Social Psychology, 20 . Amsterdam, Netherlands: Elsevier, 1987, pp. 203–253. 102 Fredrickson BL . What good are positive emotions? Rev Gen Psychol 1998; 2 (3 ): 300–319.21850154 103 Lindebaum D Geddes D . The place and role of (moral) anger in organizational behavior studies. J Organ Behav 2016; 37 (5 ): 738–757.27773966 104 Adams JG Walls RM . Supporting the Health Care Workforce During the COVID-19 Global Epidemic. JAMA 2020; 323 (15 ): 1439–1440.32163102 105 Le K Nguyen M . The psychological burden of the COVID-19 pandemic severity. Econ Hum Biol 2021; 41 : 100979.33497964
PMC010xxxxxx/PMC10153209.txt
==== Front bioRxiv BIORXIV bioRxiv Cold Spring Harbor Laboratory 37131784 10.1101/2023.04.17.536926 preprint 2 Article Loss-of-function mutation in Omicron variants reduces spike protein expression and attenuates SARS-CoV-2 infection Vu Michelle N. 1Conceptualization Formal analysis Funding acquisition Investigation Methodology Project Administration Supervision Visualization Writing – original draft Writing – review and editing Alvarado R. Elias 12Investigation Writing – review and editing Morris Dorothea R. 13Investigation Lokugamage Kumari G. 1Investigation Zhou Yiyang 4 Morgan Angelica L. 5Investigation Estes Leah K. 1Investigation McLeland Alyssa M. 1Investigation Schindewolf Craig 1Investigation Writing – review and editing Plante Jessica A. 167Investigation Ahearn Yani P. 1Investigation Meyers William M. 5Investigation Murray Jordan T. 1Investigation Crocquet-Valdes Patricia A. 5Investigation Weaver Scott C. 1678Funding acquisition Supervision Writing – review and editing Walker David H. 58Investigation Supervision Visualization Writing – review and editing Russell William K. 4 Routh Andrew L. 4Funding acquisition Investigation Plante Kenneth S. 167Investigation Methodology Supervision Menachery Vineet 1678*Conceptualization Formal analysis Funding acquisition Methodology Project Administration Supervision Visualization Writing – original draft Writing – review and editing 1 Microbiology and Immunology, University of Texas Medical Branch, Galveston, TX, United States 2 Institute for Translational Sciences, University of Texas Medical Branch, Galveston, TX, United States 3 Pediatrics, University of Texas Medical Branch, Galveston, TX, United States 4 Biochemistry and Molecular Biology, University of Texas Medical Branch, Galveston, TX, United States 5 Pathology, University of Texas Medical Branch, Galveston, TX, United States 6 Institute for Human Infection and Immunity, University of Texas Medical Branch, Galveston, TX, United States 7 World Reference Center of Emerging Viruses and Arboviruses, University of Texas Medical Branch, Galveston, TX, United States 8 Center for Biodefense and Emerging Infectious Disease, University of Texas Medical Branch, Galveston, TX, United States * Corresponding Author: Vineet D. Menachery, [email protected] 10 7 2023 2023.04.17.536926https://creativecommons.org/licenses/by-nc/4.0/ This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License, which allows reusers to distribute, remix, adapt, and build upon the material in any medium or format for noncommercial purposes only, and only so long as attribution is given to the creator. nihpp-2023.04.17.536926.pdf SARS-CoV-2 Omicron variants emerged in 2022 with >30 novel amino acid mutations in the spike protein alone. While most studies focus on receptor binding domain changes, mutations in the C-terminus of S1 (CTS1), adjacent to the furin cleavage site, have largely been ignored. In this study, we examined three Omicron mutations in CTS1: H655Y, N679K, and P681H. Generating a SARS-CoV-2 triple mutant (YKH), we found that the mutant increased spike processing, consistent with prior reports for H655Y and P681H individually. Next, we generated a single N679K mutant, finding reduced viral replication in vitro and less disease in vivo. Mechanistically, the N679K mutant had reduced spike protein in purified virions compared to wild-type; spike protein decreases were further exacerbated in infected cell lysates. Importantly, exogenous spike expression also revealed that N679K reduced overall spike protein yield independent of infection. Although a loss-of-function mutation, transmission competition demonstrated that N679K had a replication advantage in the upper airway over wild-type SARS-CoV-2 in hamsters, potentially impacting transmissibility. Together, the data show that N679K reduces overall spike protein levels during Omicron infection, which has important implications for infection, immunity, and transmission. ==== Body pmcIntroduction Since its introduction, SARS-CoV-2 has continuously evolved giving rise to multiple Variants of Concern (VOCs) with diverse mutations in the spike protein 1 (Extended Fig. 1A). Present as a trimer on virions, spike is composed of S1 and S2 subunits, responsible for receptor binding and membrane fusion, respectively 2,3. The S1 subunit contains the N-terminal domain (NTD), receptor binding domain (RBD), and the C-terminus of the S1 subunit (CTS1), which harbors a furin cleavage site (FCS) in SARS-CoV-2. Following receptor binding, the spike is cleaved at the S1/S2 site by host proteases to expose the fusion machinery for entry. With the diverse mutations in the spike protein (Extended Fig. 1B), most Omicron studies have focused on the RBD and the impact on vaccine- or infection-induced immunity. However, mutations surrounding the FCS and S1/S2 cleavage site have been demonstrated to drive SARS-CoV-2 pathogenesis 4–9 and have been largely unstudied in the context of Omicron. With this in mind, we set out to evaluate the role of Omicron CTS1 mutations on infection and pathogenesis. Omicron maintains three mutations adjacent to the FCS and S1/S2 cleavage site: H655Y, N679K, and P681H (Extended Fig. 1B). Both H655Y and P681H have previously been observed in the Gamma and Alpha variants 10,11; in contrast, N679K is unique to and maintained by all Omicron subvariants 12. To evaluate the role of these mutations, we used reverse genetics to generate SARS-CoV-2 mutants with all three CTS1 mutations (YKH) or N679K alone in the original WA1 backbone from early 2020. While YKH modestly increases viral replication and spike processing in vitro, N679K results in a loss-of-function mutation that attenuates viral replication in vitro and disease in vivo while skewing replication toward the upper airways through reduced spike protein expression. Given the importance of spike protein for immunity, our finding may have major implications for vaccine efficacy and breakthrough infections. Results H655Y, N679K, and P681H together increase viral replication and spike processing. While the majority of the > 30 spike mutations Omicron acquired are localized to the RBD, three are harbored in the CTS1 adjacent to the furin cleavage site – H655Y, N679K, and P681H (Fig. 1A). Both H655Y and P681H have been observed individually in Gamma and Alpha variants and are associated with increased spike processing. In contrast, N679K is a mutation unique to Omicron and is maintained in all subsequent Omicron subvariants despite involving a single nucleotide change (T/C to A/G) in the wobble codon position 12. Importantly, N679K is adjacent to an important O-linked glycosylation site at T678 13,14; our group has previously shown this glycosylation is important for SARS-CoV-2 infection and protease usage 8. Several motifs within the CTS1 spike domain, including the furin cleavage site and the upstream QTQTN motif, are key to spike cleavage and host protease interactions, which drive SARS-CoV-2 infection and pathogenesis. All three Omicron mutations in the CTS1, H655Y, N679K, and P681H, are adjacent to or within these motifs and may impact their function (Fig. 1A and 1B). To evaluate this, we generated a mutant SARS-CoV-2 harboring H655Y, N679K, and P681H in the original WA1 backbone (YKH) (Fig. 1C) 15,16. Plaques produced by the YKH mutant were smaller compared to the parental WA1 (WT) (Fig. 1D). However, the YKH mutant did not attenuate stock titers nor replication kinetics in Vero E6 cells as compared to wild-type (WT) SARS-CoV-2 (Fig. 1E and 1F). Notably, while replication was slightly reduced at 24 hpi, end point titers for YKH were augmented at 48 hpi in Calu-3 2B4 cells compared to WT (Fig. 1G). The results suggest that the combination of the three mutations alters infection dynamics, which may offer some advantages to the Omicron variant in human respiratory cells (Fig. 1G). As H655Y and P681H have individually been shown to increase spike processing, we next evaluated spike processing on purified virions from YKH and WT infection. Similar to Delta and Omicron, YKH spike was more processed than WT (Fig. 1H and 1I). At 24 hpi, the S1/S2 cleavage ratio to full length spike ratio was ~2.4:1 for the YKH spike (55% S1/S2 product, 23% full-length); in contrast, WT had roughly equivalent amounts of S1/S2 product and full length. Overall, the combination of H655Y, N679K, and P681H in the YKH mutant resulted in increased viral endpoint yields in human respiratory cells and contributed to Omicron’s enhanced spike processing. N679K mutation attenuates SARS-CoV-2 infection. The increase in spike processing found in the YKH mutant is consistent with prior work examining H655Y and P681H mutations individually; however, the contribution of N679K had yet to be evaluated. Based on its location adjacent to a key O-linked glycosylation site 8, we hypothesized that N679K might impact SARS-CoV-2 infection (Fig. 2A). To evaluate potential changes, we generated a SARS-CoV-2 mutant with only N679K in the WA1 backbone (N679K) (Fig. 2B). Our initial characterization found that the N679K plaque sizes were distinctly smaller at days 2 and 3 post-infection (Fig. 2C), and stock titers were also slightly lower than WT (Fig. 2D). These differences in plaque size and stock titers are consistent with observations of most Omicron strains 17–20. Notably, unlike the minimal differences seen in YKH replication kinetics, the N679K mutant had attenuated replication in both Vero E6 and Calu-3 2B4 cells at 24 hpi (Fig. 2E and 2F). Although N679K viral titer recovered by 48 hpi, the results suggest that N679K is a loss-of-function mutation in terms of replication in both cell lines. We next evaluated N679K in vivo by infecting 3-to-4-week-old golden Syrian hamsters and monitored weight loss and disease over 7 days (Fig. 2G). Hamsters infected with N679K displayed significantly attenuated body weight loss compared to those infected with WT (Fig. 2H). Despite the stark attenuation seen in weight loss, N679K viral titers in the lungs were equivalent to WT at 2 dpi and 4 dpi (Fig. 2I). Similarly, N679K viral titers were comparable to WT at 2 dpi in nasal washes; however, the mutant virus resulted in reduced replication at 4 dpi (Fig. 2J). In addition, analysis of lung histopathology showed a modest, but not significant reduction in disease of the N679K infected hamsters as compared to control (Extended Fig. 2). Taken together, our results indicate that N679K has a distinct loss-of-function phenotype in vitro and in vivo. N679K mutation results in decreased spike protein expression. We next sought to determine the mechanism driving the loss-of-function observed with the N679K mutant. Given its location adjacent to the FCS, we first evaluated N679K effects on proteolytic spike processing. Virions were purified from WT, N679K or the Omicron variant BA.1 (Omicron) and blotted for spike processing. Nearly identical to YKH, the N679K mutant had increased spike processing with a ~2.5:1 ratio of S1/S2 cleavage product to full-length spike compared to 1:1 ratio for WT at 24 hpi (Fig. 3A and 3B). However, we noted distinct differences in total spike protein with N679K and Omicron compared to WT, despite similar levels of nucleocapsid protein. Densitometry analysis revealed that the total spike to nucleocapsid (S/N) ratio of N679K and Omicron virions was reduced 21% and 36%, respectively, as compared to WT (Fig. 3C). Overall, our results indicate that the N679K mutant and Omicron variant incorporate less spike protein into their virions. We then sought to determine if changes in the virion spike were due to changes to total protein expression in the cell or spike incorporation into the particle. To examine spike protein expression, we measured total spike relative to nucleocapsid from infected Vero E6 cell lysates 24 hpi (Fig. 3D and 3E). N679K resulted in a S/N ratio 66% less than WT, displaying an even further decrease in spike protein compared to the reduction in purified virions. Additionally, a similar decrease in S/N ratio was observed in Omicron, indicating that the phenotype is maintained in the context of all the Omicron mutations (Fig. 3D and 3E). Importantly, the RNA transcript ratio for both spike and N following infection of WT and N679K were nearly identical indicating no deficits in RNA expression of spike in the mutant (Fig. 3F). Together, the results indicate that the N679K mutation reduces the Omicron spike protein levels compared to WT following infection. Having established reduced spike protein in the context of N679K, we next wanted to determine if this reduction only occurs in the context of virus infection or is inherent to the protein. Therefore, we introduced the mutation into the Spike HexaPro plasmid to exogenously express spike protein and separate N679K driven changes from other aspects of viral infection 21. Vero E6 cells were transfected with the WT or N679K mutant spike HexaPro and harvested at 24 and 48 hours post transfection (hpt). Similar to what was observed in viral infection, N679K spike was reduced 43% at 24 hpt and 46% at 48 hpt (Fig. 3G and 3H). Overall, the results across virions, cell lysates, and overexpression systems demonstrate that the reduction in spike protein is governed by the N679K mutation in a manner independent of viral infection (Fig. 3I). N679K mutation results in preference for the upper airways. Recognizing that decreased spike expression impacts virus infection, we next evaluated the role of N679K on SARS-CoV-2 transmissibility. Using transmission competition, donor hamsters were infected with a 1:1 ratio of WT:N679K SARS-CoV-2 at a total of 105 pfu (Fig. 4A). At 24 hpi, donors were paired with naive recipients, cohoused for 8 hrs, separated, and donors nasal washed. Nasal wash, trachea, and lung were collected to measure viral RNA populations at 2 and 4 days post infection (dpi) for donors and post contact (dpc) for recipients. Surprisingly, while both viruses transmitted, WT and N679K demonstrated distinct replication sites along the respiratory tract (Fig. 4B). N679K dominated the nasal washes and upper airways while WT primarily seeded the lungs and lower airways. The trachea serves as a midpoint, with no clear delineation between the viruses (Fig. 4B). Having observed this gradation, we returned to the prior hamster study and examined antigen staining of the lung (Fig. 4C). While no significant differences in total antigen were noted, the localization of viral antigen in the N679K infection was distinct and concentrated in the large airways. In contrast, WT was more uniformly distributed in the parenchyma and airways (Fig. 4D). Overall, the results indicate that the N679K mutation shifts viral replication towards airway replication. Discussion Most Omicron studies have focused on determining the impact that the RBD mutations have on immune escape, largely overlooking mutations in other spike domains like the CTS1. Harboring the FCS and S1/S2 cleavage site, the CTS1 has been demonstrated as a hotspot for both attenuating and augmenting mutations 4–9. Focusing on Omicron’s three CTS1 mutations – H655Y, N679K, and P681H, we generated infectious clones with all three (YKH) or N679K alone in the SARS-CoV-2 WA1 background. The combination of YKH produced a modest increase in endpoint titers after infection of human respiratory cells, and augmented spike processing, consistent with prior studies that tested the effects of H655Y and P681H individually 10,11. However, the N679K mutant reduced viral replication in vitro and weight loss in vivo. Mechanistic studies determined that both N679K and Omicron have reduced spike protein incorporated into their virions, less spike protein in infected cell lysates, and inferior production using exogenous spike protein expression systems. Our results argue that reduced spike protein in the context the N679K mutation attenuates Omicron strains and may have implications for SARS-CoV-2 immunity by reducing spike antigen thus shifting immune recognition. Additionally, while the N679K mutation attenuated the virus in vivo, our studies indicate a shift toward the upper airways replication. Overall, the data argue that N679K acts as a loss-of-function mutation that has a significant impact on SARS-CoV-2 Omicron infection, pathogenesis, and transmission dynamics. N679K is likely attenuated because of its decrease in spike protein production. Starting with ~20-30% less spike in its virions, one possibility was a change in spike incorporation. However, an even greater decrease (66%) in spike protein was present in infected Vero E6 cell lysates, indicated that overall spike protein levels were affected. In addition, we found no change in the ratio of spike message relative to N transcript, suggesting the N679K mutation impacts the protein itself. To confirm that the reduction in spike was not a product of virus infection or host immune interactions, we exogenously expressed the spike protein to demonstrate that the N679K spike protein itself was less stable than the WT control. One possible mechanism is that the asparagine-to-lysine change introduces a ubiquitination site that could lead to spike degradation. Another possible mechanism is that the N679K mutation itself may destabilize the protein structurally. Additionally, the N679K substitution adds another basic amino acid to the stretch including the furin cleavage site; the positively charged lysine extends the polybasic cleavage motif and may facilitate cleavage by additional host proteases 22. Overall, while the exact mechanism is unclear, the N679K mutation results in a less stable spike protein that impacts infection and pathogenesis of SARS-CoV-2. Surprisingly, N679K is uniformly found in 100% of Omicron sequences in GSAID, despite being a single nucleotide change in the wobble position 12. Though attenuated in vitro, N679K does replicate to similar titers as WT in hamster lungs and at day 2 in nasal washes. Notably, N679K outcompetes WT in the upper airways when in direct transmission competition. These results suggest no deficits in transmission and may augment spread as virus replication in the upper airway is more likely to seed new infections. These results also potentially explain why N679K is maintained despite clear attenuation of SARS-CoV-2 infection. Notably, addition of H655Y and P681H in the YKH mutant rescues replication in Calu-3 cells, suggesting that other Omicron mutations may compensate for N679K. However, it is unclear if reverting N679K in the Omicron strains would result in a gain in terms of in vitro replication or in vivo pathogenesis. While N679K in SARS-CoV-2 WA1 produces a clear loss-of-function, the constellation of spike mutations and epistatic interactions may mitigate the deficit in Omicron strains. Importantly, the complete conservation of N679K in Omicron also implies some fitness advantage 23–28. From our data, the shift toward upper airway replication by N679K may explain how it is maintained despite lower overall spike protein expression. In addition to impacting primary infection, the reduction in spike protein may have important implications for SARS-CoV-2 and human immunity. Compared to WT, the N679K mutant produces less spike protein upon infection and can potentially skew the ratio of antibodies targeting spike and nucleocapsid. Prior work with SARS-CoV had shown that an altered spike/nucleocapsid antibody ratio contributed to vaccine failure in aged mice 29. Therefore, infection with Omicron could increase N targeting antibodies at the expense of spike antibodies. The result would be less protective neutralizing antibody, which may facilitate more breakthrough infections. Furthermore, SARS-CoV-2 vaccines based on the Omicron spike may produce less spike protein due to N679K mutation. In the context of the mRNA bivalent vaccines, the N679K mutation may alter the 1:1 ratio of WT to Omicron spike protein; N679K may bias immune responses towards WT spike protein instead of equally between both spike proteins. In addition, the total amount of spike protein produced may be less than previous vaccines formulations, thus diminishing the overall antibody response. These factors potentially contribute to the less than expected increase in immunity against Omicron strains despite the new bivalent vaccine formulations. Moving forward, reverting K679 back to N679 in vaccine may improve spike protein yields and subsequently improve vaccine response to the Omicron variants. Together, our results demonstrate that Omicron N679K is a loss-of-function mutation consistently maintained in subvariants. Mechanistically, the N679K mutation attenuates the virus in vitro and in vivo by increasing spike degradation. While the N679K mutation is attenuating in isolation, other Omicron mutations like H655Y and P681H may compensate for the N679K loss of function by amplifying spike processing and infection. However, the decreased spike protein expression by N679K may have implications for immunity induced by infection and vaccines. In addition, while N679K attenuated viral pathogenesis, the shift to the upper airway replication may have enhanced transmissibility and contribute to Omicron emergence. Overall, the data highlight that the Omicron CTS1 mutations have a significant impact on SARS-CoV-2 infection and are worthy of continued study and surveillance. Methods Cell Culture Vero E6 cells were grown in high glucose DMEM (Gibco #11965092) with 10% fetal bovine serum and 1x antibiotic-antimycotic. TMPRSS2-expressing Vero E6 cells were grown in low glucose DMEM (Gibco #11885084) with sodium pyruvate, 10% FBS, and 1 mg/mL Geneticin™ (Invitrogen #10131027). Calu-3 2B4 cells were grown in high glucose DMEM (Gibco #11965092) with 10% defined fetal bovine serum, 1 mM sodium pyruvate, and 1x antibiotic-antimycotic. Viruses The SARS-CoV-2 infectious clones were based on the USA-WA1/2020 sequence provided by the World Reference Center of Emerging Viruses and Arboviruses and the USA Centers for Disease Control and Prevention 30. Mutant viruses (YKH and N679K) were generated with restriction enzyme-based cloning using gBlocks encoding the mutations (Integrated DNA Technologies) and our reverse genetics system as previously described 15,16. Virus stock was generated in TMPRSS2-exrpressing Vero E6 cells to prevent mutations from occurring at the FCS. Viral RNA was extracted from virus stock and cDNA was generated to verify mutations by Sanger sequencing. Delta isolate (B.1.617.2) was obtained from the World Reference Center of Emerging Viruses and Arboviruses. Infectious clone of Omicron (BA.1) was obtained from Dr. Pei Yong Shi and Dr. Xuping Xie. In vitro Infection Vira infections in Vero E6 and Calu-3 2B4 were carried out as previously described 8. Briefly, growth media was removed, and cells were infected with WT or mutant SARS-CoV-2 at an MOI of 0.01 for 45 min at 37°C with 5% CO2. After absorption, cells were washed three times with PBS and fresh complete growth media was added. Three or more biological replicates were collected at each time point and each experiment was performed at least twice. Samples were titrated with plaque assay or focus forming assays. Plaque Assay Vero E6 cells were seeded in 6-well plates and grown to 80-100% confluency in complete growth media. Ten-fold serial dilutions in PBS were performed on virus samples. Growth media was removed from cells and 200 μl of inoculum was added to monolayers. Cells were incubated for 45 min at 37°C with 5% CO2. After absorption, 0.8% agarose overlay was added, and cells were incubated at 37°C with 5% CO2 for 2 days. Plaques were visualized with neutral red stain. Average plaque size was determined using ImageJ. Focus Forming Assay Focus forming assays (FFAs) were performed as previously described 31. Briefly, Vero E6 cells were seeded in 96-well plates to be 100% confluent. Samples were 10-fold serially diluted in serum-free media and 20 μl was to infect cells. Cells were incubated for 45 min at 37°C with 5% CO2 before 100 μl of 0.85% methylcellulose overlay was added. Cells were incubated for 24 h 45 min at 37°C with 5% CO2. After incubation, overlay was removed, and cells were washed three times with PBS before fixed and virus inactivated by 10% formalin for 30 min at room temperature. Cells were then permeabilized and blocked with 0.1% saponin/0.1% BSA in PBS before incubated with α-SARS-CoV-2 Nucleocapsid primary antibody (Cell Signaling Technology) at 1:1000 in permeabilization/blocking buffer overnight at 4°C. Cells are then washed three times with PBS before incubated with Alexa Fluor™ 555-conjugated α-mouse secondary antibody (Invitrogen #A28180) at 1:2000 in permeabilization/blocking buffer for 1 h at room temperature. Cells were washed three times with PBS. Fluorescent foci images were captured using a Cytation 7 cell imaging multi-mode reader (BioTek), and foci were counted manually. Hamster Infection Three- to four-week-old male golden Syrian hamsters (HsdHan:AURA strain) were purchased from Envigo. All studies were conducted under a protocol approved by the UTMB Institutional Animal Care and Use Committee and complied with USDA guidelines in a laboratory accredited by the Association for Assessment and Accreditation of Laboratory Animal Care. Procedures involving infectious SARS-CoV-2 were performed in the Galveston National Laboratory ABSL3 facility. Hamsters were intranasally infected with 105 pfu of WT or N679K SARS-CoV-2 in 100 μl. Infected hamsters were weighed and monitored for illness over 7 days. Hamsters were anesthetized with isoflurane and nasal washes were collected with 400 μl of PBS on endpoint days (2, 4, and 7 dpi). Hamsters were euthanized by CO2 for organ collection. Nasal wash and lung were collected to measure viral titer and RNA. Left lungs were collected for histopathology. Transmission Competition Three- to four-week-old male golden Syrian hamsters (HsdHan:AURA strain) were purchased from Envigo. Ten donor hamsters were intranasally infected with a 1:1 ratio of WT:N679K SARS-CoV-2 totaling 105 pfu in 100 μl and were subsequently singly housed. After 24 hrs post infection, individual donor hamsters were cohoused with a recipient hamster for 8 hrs for contact transmission. Following 8 hrs, hamster pairs were separated and housed singly, and nasal washes were collected from donors. At 2 and 4 days post infection for donors and post contact for recipients, hamsters were nasal washed with 400 μl of PBS and euthanized for trachea and lung collection. Nasal washes, tracheas, and lungs were processed in TRIzol and RNA was extracted to perform next generation sequencing. Virion Purification Vero E6 cells were grown in T175 flasks to be 100% confluent at time of infection. Cells were infected with 50 μl of virus stock in PBS for 45 min at 37°C with 5% CO2, and growth media with 5% FBS was added after absorption. Supernatant was harvested at 24 hpi and clarified by low-speed centrifugation. Virions were purified from supernatant by ultracentrifugation through a 20% sucrose cushion at 26,000 rpm for 3 hrs using a Beckman SW28 rotor. Pellets were resuspended with 2x Laemmli buffer to obtain protein samples for Western blot. Western Blot Protein levels were determined by SDS–PAGE followed by western blot analysis as previously described 8. In brief, sucrose-purified SARS-CoV-2 virions were inactivated by resuspending in 2x Laemmli buffer and boiling. SDS-PAGE gels were run with equal volumes of samples on Mini-PROTEAN TGX gels (Bio-Rad #4561094) followed by transfer onto PVDF membrane. Membranes were incubated with α-SARS-CoV S primary antibody (Novus Biologicals #NB100-56578) at 1:1000 dilution in 5% BSA in TBST to measure spike protein processing and expression. For loading control, α-SARS Nucleocapsid primary antibody (Novus Biologicals #NB100-56576) at 1:1000 in 5% BSA in TBST was used for viral loading control and α-GAPDH primary antibody (Invitrogen #AM4300) at 1:1000 in 5% BSA in TBST for cellular loading control. Primary antibody incubation was followed by HRP-conjugated α-rabbit secondary antibody (Cell Signaling Technology #7074) or HRP-conjugated α-mouse secondary antibody (Cell Signaling Technology #7076) at 1:3000 in 5% milk in TBST. Chemiluminescence signal was developed using Clarity Western ECL substrate (Bio-Rad #1705060) or Clarity Max Western ECL substrate (Bio-Rad #1705062) and imaged with a ChemiDoc MP System (Bio-Rad). Densitometry analysis was performed using ImageLab 6.0.1 (Bio-Rad). RT-qPCR Vero E6 cells were infected with an MOI of 1 as detailed above in in vitro infection. Cell lysate was collected at 8 hpi in TRIzol. RNA was extracted from TRIzol samples using Direct-zol RNA Miniprep Plus kit (Zymo #R2072) to be used in two-step RT-qPCR. cDNA was reverse transcribed from 1 μg of total RNA using LunaScript RT Supermix kit (NEB #E3010) according to manufacturer’s instructions. RT-qPCR was performed using Luna Universal qPCR Master mix (NEB #M3003) according to manufacturer’s instructions. RT-qPCR cycle was performed as follows: 95°C for 60 s (1 cycle), 95°C for 15 s and 51°C for 30 s then plate read (40 cycles), and melt curve from 65°C to 95°C for 5 s. For spike and nucleocapsid transcripts, a forward primer binding upstream of the transcription regulatory sequence (TRS) leader region (ACCAACCAACTTTCGATCTCT) was used with reverse primers for spike (TGCAGGGGGTAATTGAGTTCT) and nucleocapsid (CCCACTGCGTTCTCCATTCT). The 18S ribosomal RNA primers were forward (CCGGTACAGTGAAACTGCGAATG) and reverse ((GTTATCCAAGTAGGAGAGGAGCGAG). RNA transcript levels for spike and nucleocapsid were determined by ΔΔCt method with 18S as the internal control. Ratios of ΔΔCt spike over ΔΔCt nucleocapsid was reported for each sample. Spike HexaPro Cloning and Transfection SARS-CoV-2 S HexaPro was a gift from Jason McLellan (Addgene plasmid #154754) 21. The N679K mutation was cloned into spike HexaPro using a gBlock encoding the mutation (Integrated DNA Technologies) and restriction enzyme-based cloning. Sequences were verified by Sanger sequencing. Vero E6 cells were grown in 24-well plates to be 100% confluent at time of transfection. Cells were transfected with spike HexaPro WT or N679K plasmid and Lipofectamine 2000 following manufacturer’s instructions (Invitrogen). Briefly, 100 ng of spike HexaPro plasmid and 1.5 μl of Lipofectamine 2000 were separately diluted in 50 μl Opti-MEM (Gibco #31985070) before mixing together. After 20 min of room temperature incubation, 100 μl of the transfection mixture was added to cells, and cells were incubated at 37°C with 5% CO2. Cell lysate was harvested with 2x Laemmli buffer at 24 and 48 hours post transfection to be analyzed by Western blot. Structural Modeling Structural models previously generated were used as a base to visualize residues mutated in Omicron 8. Briefly, structural models were generated using SWISS-Model to generate homology models for WT and glycosylated SARS-CoV-2 spike protein on the basis of the SARS-CoV-1 trimer structure (Protein Data Bank code 6ACD). Homology models were visualized and manipulated in PyMOL (version 2.5.4) to visualize Omicron mutations. Next Generation Sequencing Next generation sequencing to determine viral RNA populations was performed as previously described 31. Briefly, RNA samples were extracted and prepared for Tiled-ClickSeq libraries 32. A modified pre-RT annealing protocol was applied as previously described31. The final libraries comprising of 300–700 bps fragments were pooled and sequenced on an Illumina NextSeq platform with paired-end sequencing. The raw Illumina data of the Tiled-ClickSeq libraries were processed with previously established bioinformatics pipelines32. One modification is the introduction of ten wild cards (“N”) covering the N679K mutation in the reference genome to allow bowtie233 to align reads to wild type or variant genomes without bias. PCR duplications were removed using UMI-tools34, and the number of unique reads representing WT and N679K variants were counted thereafter. Histology Left lung lobes were harvested from hamsters and fixed in 10% buffered formalin solution for at least 7 days. Fixed tissue was then embedded in paraffin, cut into 5 μM sections, and stained with hematoxylin and eosin (H&E) on a SAKURA VIP6 processor by the University of Texas Medical Branch Surgical Pathology Laboratory. Immunohistochemistry Fixed and paraffin-embedded left lung lobes from hamsters were cut into 5 μM sections and mounted onto slides by the University of Texas Medical Branch Surgical Pathology Laboratory. Paraffin-embedded sections were warmed at 56°C for 10 min, deparaffinized with xylene (3x 5-min washes) and graded ethanol (3x 100% 5-min washes, 1x 95% 5-min wash), and rehydrated in distilled water. After rehydration, antigen retrieval was performed by steaming slides in antigen retrieval solution (10 mM sodium citrate, 0.05% Tween-20, pH 6) for 40 min (boil antigen retrieval solution in microwave, add slides to boiling solution, and incubate in steamer). After cooling and rinsing in distilled water, endogenous peroxidases were quenched by incubating slides in TBS with 0.3% H2O2 for 15 min followed by 2x 5-min washes in 0.05% TBST. Sections were blocked with 10% normal goat serum in BSA diluent (1% BSA in 0.05% TBST) for 30 min at room temperature. Sections were incubated with primary anti-N antibody (Sino #40143-R001) at 1:1000 in BSA diluent overnight at 4°C. Following overnight primary antibody incubation, sections were washed 3x for 5 min in TBST. Sections were incubated in secondary HRP-conjugated anti-rabbit antibody (Cell Signaling Technology #7074) at 1:200 in BSA diluent for 1 hour at room temperature. Following secondary antibody incubation, sections were washed 3x for 5 min in TBST. To visualize antigen, sections were incubated in ImmPACT NovaRED (Vector Laboratories #SK-4805) for 3 min at room temperature before rinsed with TBST to stop the reaction followed by 1x 5-min wash in distilled water. Sections were incubated in hematoxylin for 5 min at room temperature to counterstain before rinsing in water to stop the reaction. Sections were dehydrated by incubating in the previous xylene and graded ethanol baths in reverse order before mounted with coverslips. Extended Data Extended Figure 1. Emergence of Omicron subvariants. (A) Timeline of SARS-CoV-2 variants emergence by earliest documented case reported by the WHO. (B) Spike mutations across Omicron subvariants with shared mutations across all subvariants (gray box) and mutations unique to the specific variant (bolded) indicated. Mutations key to this study indicated in bold red. Extended Figure 2. Histopathology of hamsters infected with WT or N679K SARS-CoV-2. (A) H&E staining of left lung of hamsters infected with 105 pfu of WT (top) or N679K (bottom) SARS-CoV-2 at 2 (left), 4 (middle), and 7 (right) dpi. Lungs for both WT and N679K show bronchiolitis and interstitial pneumonia at 2 dpi that become more severe at 4 and 7 dpi. (B) H&E staining of left lung of hamsters infected with 105 pfu of WT (black) or N679K (green) were scored for histopathological analysis. Acknowledgments Research was supported by grants from NIAID of the NIH to (R01-AI153602 and R21-AI145400 to VDM; R24-AI120942 (WRCEVA) to SCW). ALR was supported by a UTMB Institute for Human Infection and Immunity grant and the Sealy and Smith Foundation. Research was also supported by STARs Award provided by the University of Texas System to VDM and Data Acquisition award provided by the Institute for Human Infections and Immunity at UTMB to MNV. Trainee funding provided by NIAID of the NIH to MNV (T32-AI060549). Mass spectrometry experiments and analysis was provided by the Mass Spectrometry Facility at the University of Texas Medical Branch (https://www.utmb.edu/MSF). Figures were created with BioRender.com Figure 1. The combination of Omicron mutations H655Y, N679K, and P681H increases viral replication and spike processing. (A) Comparison of CTS1 region near the S1/S2 cleavage site between SARS-CoV-2 variants. (B) Structure of loop containing the S1/S2 cleavage site on SARS-CoV-2 spike protein. The residues that are mutated in Omicron are shown: H655 (magenta), N679 (green), and P681 (blue). The furin cleavage site RRAR (cyan) and QTQT motif (red) are also shown. (C) Schematic of WT and YKH SARS-CoV-2 mutant genomes. (D) WT and YKH SARS-CoV-2 plaques on Vero e6 cells at 2 dpi. (E) Viral titer from WT and YKH virus stock with the highest yield generated from TMPRSS2-expressing Vero E6 cells. (F-G) Growth kinetics of WT and YKH in Vero E6 (F) and Calu-3 2B4 (G) cells. Cells were infected at an MOI of 0.01 (n=3). Data are mean ± s.d. Statistical analysis measured by two-tailed Student’s t-test. (H) Purified WT, YKH, Delta isolate (B. 1.617.2), and Omicron (BA.1) virions from Vero E6 supernatant were probed with α-Spike and α-Nucleocapsid (N) antibodies in Western blots. Full-length spike (FL), S1/S2 cleavage product, and S2’ cleavage product are indicated. (I) Densitometry of FL and S1/S2 cleavage product was performed, and quantification of FL and S1/S2 cleavage product percentage of total spike is shown. Quantification was normalized to N for viral protein loading control. WT (black), YKH (blue), Delta isolate (purple), Omicron (orange). Results are representative of two experiments. Figure 2. N679K attenuates SARS-CoV-2 replication and disease when isolated. (A) Structural modeling of O-linked glycosylation of threonine 678 (yellow) of QTQTN motif (red) and the residues mutated in Omicron – H655 (magenta), N679 (green), and P681 (blue) – with N679 adjacent to the glycosylation. The furin cleavage site RRAR is also shown (cyan). (B) Schematic of WT and N679k sArS-CoV-2 genomes. (C) WT and N679K SARS-CoV-2 plaques on Vero E6 cells at 2 dpi (left) and 3 dpi (right). Average plaque size noted below. (D) Viral titer from WT and N679K virus stock with the highest yield generated form TMPRSS2-expressing Vero E6 cells. (E-F) Growth kinetics of WT and N679K in Vero E6 (E) and Calu-3 2B4 (F) cells. Cells were infected at an MOI of 0.01 (n=3). Data are mean ± s.d. Statistical analysis measured by two-tailed Student’s t-test. (G) Schematic of experiment design for golden Syrian hamster infection with WT (black) or N679K (green) SARS-CoV-2. Three- to four-week-old make hamsters were infected with 105 pfu and monitored for weight loss over 7 days. At 2, 4, and 7 dpi, nasal wash and lung was collected for viral titer, and lung was collected for histopathology. (H) Weight loss of hamsters infected with WT (black) or N679K (green) SARS-CoV-2 over 7 days. Data are mean ± s.e.m. Statistical analysis measured by two-tailed Student’s t-test. (I-J) Viral titer of lung (I) and nasal wash (J) collected at 2 and 4 dpi from hamsters infected with WT (black) or N679K (green) SARS-CoV-2. Data are mean ± s.d. Statistical analysis measured by two-tailed Student’s t-test. Figure 3. N679K results in decreased spike expression on virions and in cell lysate. (A) Purified WT, N679K, and Omicron (BA.1) virions from Vero E6 supernatant were probed with α-Spike and α-Nucleocapsid (N) antibodies in Western blots. Full-length spike (FL), S1/S2 cleavage product, and S2’ cleavage product are indicated. (B) Densitometry of spike processing from purified virion Western blot in (A) was performed, and quantification of FL and S1/S2 cleavage product percentage of total spike is shown. Quantification was normalized to N as viral protein loading control. WT (black), N679K (green), Omicron (orange). Results are representative of two experiments. (C) Densitometry of spike expression from purified virion Western blot in (A) was performed, and quantification of total spike protein to nucleocapsid ratio is shown. Spike/N ratio is relative to WT. WT (black), N679K (green), Omicron (orange). Results are representative of two experiments. (D) Vero E6 cells were infected with WT, N679K, or Omicron at an MOI of 0.01. Cell lysate was collected at 24 hpi and probed with α-Spike and α-Nucleocapsid (N) antibodies in Western blots. Full-length spike (FL), S1/S2 cleavage product, and S2’ cleavage product are indicated. (E) Densitometry of spike expression from infected cell lysate Western blot in (D) was performed, and quantification of total spike protein to nucleocapsid ratio is shown. Spike/N ratio is relative to WT. WT (black), N679K (green), Omicron (orange). Results are representative of three biological replicates. (F) Vero E6 cells were infected with WT or N679K at an MOI of 1 infectious units/cell. Cell lysate was collected at 8 hpi in Trizol to extract RNA. RNA transcripts for spike, nucleocapsid and 18S were measured using RT-qpCR. The ratios of ΔΔCt spike to ΔΔCt nucleocapsid are shown. Data are mean ± s.d. Statistical analysis measured by two-tailed Student’s t-test. (G) Vero E6 cells were transfected with Spike HexaPro WT and N679K and cell lysate was collected at 8, 24, and 48 hpt. Lysates were probed with α-Spike and α-GAPDH antibodies in Western blots. (H) Densitometry of spike expression from transfected cell lysates by Western blot in (G) was performed, and quantification of relative total spike protein is shown. Spike protein levels were normalized to GAPDH and are relative to WT. WT (black), N679K (green). Results are representative of three biological replicates. (I) While WT virus and exogenous spike plasmid produces abundant spike protein, the N679K mutation results in less spike protein expression in virions and intracellularly by infection and transfection of exogenous spike plasmid. Figure 4. N679K results in preference for upper airways. (A) Schematic of experimental design of transmission competition in golden Syrian hamsters. Donor three- to four-week-old male hamsters were intranasally infected with 105 pfu of WT:N679K SARS-CoV-2 in a 1:1 ratio and housed singly. Donors were paired with recipients 24 hpi and cohoused for 8 hrs before separating and nasal washing donors. Nasal washes, tracheas, and lungs were collected at 2 and 4 days post infection for donors (dpi) and post contact for recipients (dpc). (B) Next generation sequencing was performed on extracted RNA to measure the percentage of WT (black) and N679K (green) present in nasal wash (left), trachea (middle), and lung (right) of donors (top) and recipients (bottom). (C) Immunohistochemistry of left lung lobes at 2, 4 and 7 dpi staining for nucleocapsid. Hamsters were singly infected with 105 pfu of either WT or N679K SARS-CoV-2. (D) Immunohistochemistry staining of left lung lobes form hamsters infected with WT (black) or N679K (green) SARS-CoV-2 were scored by total section (left), airway (middle), or parenchyma (left) staining. Data are mean showing minimum and maximum (n=5). Statistical analysis measured by two-tailed Student’s t-test. Competing Interest Statement VDM has filed a patent on the reverse genetic system and reporter SARS-CoV-2. MNV and VDM have filed a provisional patent on a stabilized SARS-CoV-2 spike protein. Other authors declare no competing interests. ==== Refs References 1 Callaway E. Heavily mutated Omicron variant puts scientists on alert. Nature 600 , 21 (2021). 10.1038/d41586-021-03552-w 34824381 2 Belouzard S. , Millet J. K. , Licitra B. N. & Whittaker G. R. Mechanisms of Coronavirus Cell Entry Mediated by the Viral Spike Protein. Viruses 4 , 1011–1033 (2012).22816037 3 Vu M. N. & Menachery V. D. Binding and entering: COVID finds a new home. PLOS Pathogens 17 , e1009857 (2021). 10.1371/journal.ppat.1009857 34460863 4 Coutard B. The spike glycoprotein of the new coronavirus 2019-nCoV contains a furin-like cleavage site absent in CoV of the same clade. Antiviral Research 176 , 104742 (2020). 10.1016/j.antiviral.2020.104742 32057769 5 Johnson B. A. Loss of furin cleavage site attenuates SARS-CoV-2 pathogenesis. Nature (2021). 10.1038/s41586-021-03237-4 6 Liu Y. Delta spike P681R mutation enhances SARS-CoV-2 fitness over Alpha variant. bioRxiv, 2021.2008.2012.456173 (2021). 10.1101/2021.08.12.456173 7 Peacock T. P. The furin cleavage site in the SARS-CoV-2 spike protein is required for transmission in ferrets. Nat Microbiol 6 , 899–909 (2021). 10.1038/s41564-021-00908-w 33907312 8 Vu M. N. QTQTN motif upstream of the furin-cleavage site plays a key role in SARS-CoV-2 infection and pathogenesis. Proceedings of the National Academy of Sciences 119 , e2205690119 (2022). 10.1073/pnas.2205690119 9 Lau S.-Y. Attenuated SARS-CoV-2 variants with deletions at the S1/S2 junction. Emerging Microbes & Infections 9 , 837–842 (2020). 10.1080/22221751.2020.1756700 32301390 10 Escalera A. Mutations in SARS-CoV-2 variants of concern link to increased spike cleavage and virus transmission. Cell Host & Microbe 30 , 373–387.e377(2022). 10.1016/j.chom.2022.01.006 35150638 11 Liu Y. Delta spike P681R mutation enhances SARS-CoV-2 fitness over Alpha variant. Cell Reports 39 (2022). 10.1016/j.celrep.2022.110829 12 Hadfield J. Nextstrain: real-time tracking of pathogen evolution. Bioinformatics 34 , 4121–4123 (2018). 10.1093/bioinformatics/bty407 29790939 13 Gao C. SARS-CoV-2 Spike Protein Interacts with Multiple Innate Immune Receptors. bioRxiv, 2020.2007.2029.227462 (2020). 10.1101/2020.07.29.227462 14 Sanda M. , Morrison L. & Goldman R. N- and O-Glycosylation of the SARS-CoV-2 Spike Protein. Anal Chem 93 , 2003–2009 (2021). 10.1021/acs.analchem.0c03173 33406838 15 Xie X. An Infectious cDNA Clone of SARS-CoV-2. Cell Host Microbe 27 , 841–848 e843 (2020). 10.1016/j.chom.2020.04.004 32289263 16 Xie X. Engineering SARS-CoV-2 using a reverse genetic system. Nature Protocols 16 , 1761–1784 (2021). 10.1038/s41596-021-00491-8 33514944 17 Peacock T. P. The altered entry pathway and antigenic distance of the SARS-CoV-2 Omicron variant map to separate domains of spike protein. bioRxiv, 2021.2012.2031.474653 (2022). 10.1101/2021.12.31.474653 18 Meng B. Altered TMPRSS2 usage by SARS-CoV-2 Omicron impacts infectivity and fusogenicity. Nature 603 , 706–714 (2022). 10.1038/s41586-022-04474-x 35104837 19 Phadke K. S. , Higdon N. B. A. & Bellaire B. H. In vitro comparison of SARS-CoV-2 variants. bioRxiv, 2023.2003.2011.532212 (2023). 10.1101/2023.03.11.532212 20 Zhao H. SARS-CoV-2 Omicron variant shows less efficient replication and fusion activity when compared with Delta variant in TMPRSS2-expressed cells. Emerging Microbes & Infections 11 , 277–283 (2022). 10.1080/22221751.2021.2023329 34951565 21 Hsieh C.-L. Structure-based design of prefusion-stabilized SARS-CoV-2 spikes. Science 369 , 1501–1505 (2020). 10.1126/science.abd0826 32703906 22 Mustafa Z. , Kalbacher H. & Burster T. Occurrence of a novel cleavage site for cathepsin G adjacent to the polybasic sequence within the proteolytically sensitive activation loop of the SARS-CoV-2 Omicron variant: The amino acid substitution N679K and P681H of the spike protein. PLOS ONE 17 , e0264723 (2022). 10.1371/journal.pone.0264723 35436320 23 Yuan S. Pathogenicity, transmissibility, and fitness of SARS-CoV-2 Omicron in Syrian hamsters. Science 377 , 428–433 (2022). 10.1126/science.abn8939 35737809 24 McCormack C. P. Modelling the viral dynamics of the SARS-CoV-2 Delta and Omicron variants in different cell types. bioRxiv, 2023.2003.2015.529513 (2023). 10.1101/2023.03.15.529513 25 Hui K. P. Y. SARS-CoV-2 Omicron variant replication in human bronchus and lung ex vivo. Nature 603 , 715–720 (2022). 10.1038/s41586-022-04479-6 35104836 26 Carabelli A. M. SARS-CoV-2 variant biology: immune escape, transmission and fitness. Nature Reviews Microbiology 21 , 162–177 (2023). 10.1038/s41579-022-00841-7 36653446 27 Papanikolaou V. From delta to Omicron: S1-RBD/S2 mutation/deletion equilibrium in SARS-CoV-2 defined variants. Gene 814 , 146134 (2022). 10.1016/j.gene.2021.146134 34990799 28 Shrestha L. B. , Foster C. , Rawlinson W. , Tedla N. & Bull R. A. Evolution of the SARS-CoV-2 omicron variants BA.1 to BA.5: Implications for immune escape and transmission. Reviews in Medical Virology 32 , e2381 (2022). 10.1002/rmv.2381 35856385 29 Deming D. Vaccine Efficacy in Senescent Mice Challenged with Recombinant SARS-CoV Bearing Epidemic and Zoonotic Spike Variants. PLOS Medicine 3 , e525(2006). 10.1371/journal.pmed.0030525 17194199 30 Harcourt J. Severe Acute Respiratory Syndrome Coronavirus 2 from Patient with Coronavirus Disease, United States. Emerging infectious diseases 26 , 1266–1273 (2020). 10.3201/eid2606.200516 32160149 31 Johnson B. A. Nucleocapsid mutations in SARS-CoV-2 augment replication and pathogenesis. PLOS Pathogens 18 , e1010627(2022).10.1371/journal.ppat.1010627 35728038 32 Jaworski E. Tiled-ClickSeq for targeted sequencing of complete coronavirus genomes with simultaneous capture of RNA recombination and minority variants. eLife 10 , e68479 (2021). 10.7554/eLife.68479 34581669 33 Langmead B. & Salzberg S. L. Fast gapped-read alignment with Bowtie 2. Nature Methods 9 , 357–359 (2012). 10.1038/nmeth.1923 22388286 34 Smith T. , Heger A. & Sudbery I. UMI-tools: modeling sequencing errors in Unique Molecular Identifiers to improve quantification accuracy. Genome Res 27 , 491–499(2017).10.1101/gr.209601.116 28100584
PMC010xxxxxx/PMC10196684.txt
==== Front J Leadersh Organ Stud J Leadersh Organ Stud JLO spjlo Journal of Leadership & Organizational Studies 1548-0518 1939-7089 SAGE Publications Sage CA: Los Angeles, CA 10.1177/15480518231176231 10.1177_15480518231176231 Articles Charismatic Leadership Is Not One Size Fits All: The Moderation Effect of Intolerance to Uncertainty and Furlough Status During the COVID-19 Pandemic https://orcid.org/0000-0002-4062-4446 Klein Galit 1 https://orcid.org/0000-0002-7895-9233 Delegach Marianna 2 1 The Department of Economic and Business Administration, 42732 Ariel University , Ariel, Israel 2 The Human Resource Management Department, 42741 Sapir Academic College , D.N. Hof Ashkelon Israel Galit Klein, The Department of Economic and Business Administration, Ariel University, Ariel 40700, Israel. Email: [email protected] 18 5 2023 8 2023 18 5 2023 30 3 297313 12 10 2021 15 2 2023 22 4 2023 © The Authors 2023 2023 Baker College https://creativecommons.org/licenses/by-nc/4.0/ This article is distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 License (https://creativecommons.org/licenses/by-nc/4.0/) which permits non-commercial use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access page (https://us.sagepub.com/en-us/nam/open-access-at-sage). This study aims to examine the effect of charismatic leadership on followers’ attitudinal, emotional, and well-being outcomes in a crisis setting. Combining leadership literature with Conservation of Resources and leader-follower distance theories, we propose that the effect of charismatic leadership on follower outcomes depends on the interplay between the follower's furlough status during the lockdown period and their Intolerance to Uncertainty (IU) dispositional characteristic. A cross-sectional study was conducted at two points in time: during the first lockdown (March–April 2020) and four months after the lockdown (August 2020). The final sample included 336 employees with data for both points in time (n = 199 continued to work during the lockdown, n = 137 were on furlough). The findings confirmed the study's hypotheses and revealed that charismatic leadership significantly contributed to employee outcomes only in the case of furloughed employees with low levels of IU and of continuously-employed employees with high levels of IU. It did not make a similar contribution in the edge cases—employees with low IU levels who continued to work during the lockdown or those with high levels of IU who were furloughed. This study provides novel insights into the relationship between charismatic leadership effectiveness and follower outcomes, and informs managers how to better adjust their leadership style to their followers in a crisis setting. The findings extend our knowledge about charismatic leadership by suggesting the mutual contribution of the distance dimension and employee dispositional characteristics as a boundary condition to charismatic leadership effectiveness. charismatic leadership uncertainty avoidance furlough crisis COVID-19 intolerance to uncertainty distance theory Ariel University Grant number 7496 Heth Academic Center for Research of Competition and Regulation Grant number RA2000000666 typesetterts19 ==== Body pmcIntroduction The COVID-19 pandemic broke out unexpectedly and created health, economic, and psychological crises worldwide. While a crisis has an adverse impact on organizational functioning (Klein & Eckhaus, 2017), it is also proposed as a critical antecedent of charismatic leadership (House, 1977; Weber, 1947). Weber (1947) defines charismatic leaders as leaders with “divine gifts”, which are unique qualities that allow them to inspire their followers and motivate them to transcend the status quo in pursuit of striving for a new vision. These qualities are much needed during troublesome times such as the COVID-19 pandemic, when people are called on to act for the public good, sometimes at their own expense (Antonakis, 2021). While charismatic leadership is perceived as possessing unequaled capabilities and constructive forces during ambiguous and stressful situations (e.g., De Hoogh et al., 2004; House, 1977), Klein and House (1995) have suggested that charisma resides not only in the qualities of the leader, but also depends on follower readiness to accept it. In fact, scholars have criticized the ultimately unidirectional relationship and romanticization of the leader's role (Meindl, 1995), suggesting that follower attributions may attenuate or even neutralize the positive impact of charismatic leadership on follower attitudes and organizational outcomes (e.g., De Vries et al., 1998; Wegge et al., 2022). For example, Wegge et al. (2022) demonstrated that charismatic leadership reduced the performance of participants with a high level of self-direction in a crisis context; this is because such followers have a strong need for autonomy and, as a result, less need for charismatic guidance. This idea stands on a par with the growing effort to revive the contingency leadership theory (Day & Antonakis, 2012; Oc, 2018; Sharma & Kirkman, 2015) and the cumulative body of research on the paradoxical effect of positive leadership styles (e.g., Judge et al., 2009; Sharma & Kirkman, 2015), which suggest that positive leadership styles are not ultimately advantageous across all organizational settings and different employee characteristics. Following the above, we propose that follower dispositional characteristics can provide a more nuanced and balanced view of the impact of charismatic leadership, thus answering the call for adopting a more follower-centered perspective in the charismatic leadership literature (Uhl-Bien et al., 2014). Accordingly, the current study explores one of the followers’ critical characteristics that is highly relevant to crisis contexts: their level of intolerance to uncertainty (IU). This dispositional characteristic refers to the individual's negative beliefs about uncertainty and its consequences (Carleton et al., 2007). Previous research has shown it to be a strong predictor of a person's perception of and response to a crisis, and to contribute to the individual's well-being outcomes in the crisis context (Celik et al., 2021; Larsen et al., 2021; Maftei & Holman, 2022; Parlapani et al., 2020; Voitsidis et al., 2021). Moreover, leadership's influence on followers does not occur in a vacuum. Leaders are “tenants of time and context” (Bryman et al., 1996, p. 355). Therefore, we cannot analyze the leader-follower dynamic in isolation from the context where this relationship occurs (Shamir & Howell, 1999). Specifically, in this study we focus on the employee's work status during the COVID-19 pandemic as an indicator of the distance between a leader and a follower (Antonakis & Atwater, 2002). We examine the distance based on employee work status; that is, whether the employee continued to work during the lockdown (close followers) or was furloughed (distant followers) and thus detached from the organization during the lockdown period. We propose that employee work status can play a critical role in understanding charismatic leadership effectiveness in crisis conditions. Crises are relatively rare (Bass & Bass, 2008), hence the COVID-19 pandemic provides an opportunity for a real-time investigation of the effectiveness of a charismatic leader in such contexts. In the current study, based on the Conservation of Resources theory (COR; Hobfoll, 1988) and adopting the perception of charisma as a function of the interaction between three elements—leader qualities, follower characteristics, and contextual features (Klein & House, 1995)—we suggest that the interplay between follower characteristics (i.e., IU trait High: IU+; Low: IU-) and contextual features (i.e., being on furlough: FS+; or working regularly during the COVID-19 pandemic: FS-) serves as a boundary condition of the effect of charismatic leadership on employee attitudes, emotions, and well-being. Specifically, we propose that charismatic leadership's contribution to follower outcomes is more prominent in intermediate (i.e., FS-, IU+ or FS+, IU-) than in edge cases (i.e., FS-, IU- or FS+, IU+). This is because in the latter situation, followers either have enough resources to handle the crisis and thus are less susceptible to charismatic leadership (FS-, IU-), or their pool of resources is so drained that such leadership may not be sufficient to overcome their lack of means (FS+, IU+). To explore the proposed relationships, the current study focused on three dependent variables—psychological contract violation (i.e., the emotional distress and feelings of anger resulting from an unfulfilled psychological contract between employer and employee; Morrison & Robinson, 1997), emotional exhaustion (i.e., the draining of emotional resources and a feeling of being overloaded; Maslach et al., 2001), and job insecurity (i.e., employee perception of a potential threat to the continuity of their current job; Vander Elst et al., 2014)—as the emotional, well-being, and attitudinal indicators for charismatic leadership's effect. These specific variables were chosen based on their relevance to the COVID-19 pandemic (e.g., Ganson et al., 2021; Hwang et al., 2021; Wu et al., 2021) and downsizing (e.g., Arshad, 2016; Marques et al., 2014; Paulsen et al., 2005) contexts, and on leadership literature (i.e., Jiang & Lavaysse, 2018; Kaluza et al., 2020; Restubog et al., 2010). Moreover, previous studies demonstrated that these emotional, well-being, and attitudinal indicators are the key contributors to a variety of employees’ behaviors and organizational outcomes (e.g., Meyer et al., 2002; Morrison & Robinson, 1997; Swider & Zimmerman, 2010). For the study model, see Figure 1. Figure 1. The study model. Theoretical Framework Charismatic Leadership's Role During a Crisis The construct of charismatic leadership informs academic and practitioner attention in an attempt to understand leadership's effect on follower emotions, attitudes, and behaviors (Banks et al., 2017). This attention is required to remove the mystical aura from the charismatic leadership construct (Shamir, 1992) in order to provide a more rigorous and pragmatic definition. Accordingly, House (1977) presented a theoretical framework that outlined charismatic leadership's role in an organizational context. He suggested that charismatic leaders have the ability to motivate and influence their followers by inspiring a clear vision and radical behaviors, which in turn promote the belief that the leader is blessed with extraordinary capabilities, evoking strong emotional bonds that lead followers to high levels of compliance with a commitment to the leader's foresight. Later on, Shamir et al. (1993) added that charismatic leaders inspire their followers by implicating their self-concept and elevating their self-esteem through the communication of higher expectations and the promotion of followers’ self-confidence to achieve better outcomes. One of the critical antecedents for charismatic leadership acceleration is a crisis (e.g., House, 1977; Klein & House, 1995). Indeed, charismatic leaders positively contribute to financial performance and to subordinates’ positive work attitudes only under uncertain and changeable environments (e.g., De Hoogh et al., 2004; Waldman et al., 2001). The need for charismatic leadership can be explained as a coping mechanism that followers employ under extreme conditions and high-stress levels (Devereux, 1955; Madsen & Snow, 1991). In situations of uncertainty, charismatic leadership serves as a self-preservation strategy that restores the followers’ sense of coping capability by linking themselves to a dominant and effective model (Madsen & Snow, 1991). From a psychoanalytic perspective, the leader acts as the followers’ “significant other” by ensuring their better future and possibilities to cope with stressful events and offering security during times of uncertainty (Kets de Vries, 1988). Thus, a crisis creates environmental contingencies in which followers increase their need for charismatic leadership to reassure and decrease their anxieties and fears (Wegge et al., 2022). As research on the influence of charismatic leadership expands, the role of follower characteristics as a neutralizer of charismatic leadership's effect on follower outcomes receives more scholarly attention (e.g., Den Hartog & Belschak, 2012; Wegge et al., 2022). For example, Kets de Vries et al. (1998) suggested that a low need for leadership attenuated the contribution of charismatic leadership to employee attitudes such as job satisfaction and organizational commitment. Likewise, Wegge et al. (2022) found that although charismatic leadership encouraged followers to expend efforts in performing challenging assignments, this relationship was weakened by the employees’ self-direction trait. These two examples imply that charismatic leadership may have an affirmative effect, but only for those who need and desire that kind of leadership. However, despite some conceptual (e.g., Kets de Vries, 1988; Shamir & Howell, 1999) and empirical (e.g., De Hoogh et al., 2004; Waldman et al., 2001) evidence, it is not clear to what extent follower characteristics might affect the charismatic leader's contribution to follower outcomes in a crisis setting. The COVID-19 pandemic provides an opportunity to understand how and whether charismatic leadership contributes to followers’ attitudinal, emotional, and well-being outcomes in a crisis context (e.g., Antonakis, 2021). The current study argues that in a crisis context, charismatic leadership is more effective in an intermediate condition, that is, when employees need the support and have the capability to be supported. To examine this argument we focus on the employees’ dispositional trait of intolerance of uncertainty (IU), arguing that IU bounds the contribution of charismatic leaders to their followers’ attitudinal (i.e., job insecurity), emotional (i.e., contract violation), and well-being (i.e., emotional exhaustion) outcomes. Intolerance of Uncertainty IU is a dispositional characteristic that refers to experiencing fear concerning the unknown future (Carleton et al., 2007). High IU provokes fear, worry, and anxiety, thus increasing a person's vulnerability and negatively impacting the quality of their decision-making process (Hillen et al., 2017). In the course of unexpected events, intolerant individuals demonstrate high activation of stress responses (Barling & Frone, 2017), perceive a lack of control over the situation (Endler et al., 2000), and use maladaptive coping strategies more frequently (Rettie & Daniels, 2021). When faced with the unexpected COVID-19 pandemic, individuals who had high levels of IU showed emotional and mental distress, including higher levels of anxiety and depression (Reizer et al., 2021; Rettie & Daniels, 2021). The increase in stress among this group may have evolved from a drain on their beneficial resources, as explained by the Conservation Of Resources theory (COR; Hobfoll, 1988; Hobfoll et al., 2018). COR theory suggests that people are evolutionarily motivated to retain, foster, and protect necessary resources that are valuable for their survival. Resources can be tangible assets (e.g., a house, money), conditions (e.g., job security status), and also personal characteristics (Hobfoll et al., 2018; Baranik et al., 2019). According to COR theory, stress occurs when important resources are actually lost or perceived as having the potential to be lost, which forces individuals to “invest their available resources in order to protect against resource loss, recover from losses, and gain resources” (Hobfoll et al., 2018, p. 105). However, among vulnerable groups such as individuals with high levels of IU, the availability of resources is a priori relatively low because they perceive uncertain situations as highly stressful and upsetting, which leads to an inability to act (Buhr & Dugas, 2002). Moreover, IU is positively associated with worry (Dugas et al., 2001) and rumination (Satici et al., 2022), which demand an investment of additional resources to be dealt with (Baranik et al., 2019). Hence, the resources of people with high IU are depleted faster; furthermore, during unstable situations such as the COVID-19 pandemic, these individuals are more prone to a resource loss spiral, which increases their levels of anxiety (Reizer et al., 2021; Rettie & Daniels, 2021). One strategy to reduce employee anxiety is to rely on the leader as a prominent organizational factor that creates meaning for employees (Smircich & Morgan, 1982) and structures their reality (Delegach et al., 2017), thereby affecting employee attitudes, emotions, and well-being (for meta-analytical findings, see Schyns & Schilling, 2013). Given intolerant individuals’ higher levels of stress and ambiguity, their feeling of environmental control loss can provoke them into searching for a leader who can calm their anxiety by guarantying a safe and stable future; i.e., they become more “charisma hungry” (Bass & Bass, 2008, p. 593) and reach a higher level of readiness to accept the authority of a charismatic leader (Devereux, 1955), compared to individuals with low IU. Thus, although individuals with lower levels of IU may benefit from charismatic leadership, people with higher levels of IU are more susceptible to charismatic leaders in a crisis setting. Employee Work Status During the Pandemic Following Klein and House (1995), in addition to follower characteristics, we further highlight the pivotal role of context in charismatic leadership outcomes. In our study, context is represented by employee work status during the COVID-19 pandemic. The COVID-19 pandemic generated enormous disruption, forcing organizations to quickly alter their workforce arrangements (McKibbin & Fernando, 2021). Many employees were forced to work remotely; some were fired, while others were furloughed. Furlough is defined as a reduction in working hours, from a few to an entire postponement of work. Furloughed employees are placed in a non-work status and do not receive payment or any economic or organizational benefits for the period of the leave (Mandeville et al., 2019). On the other hand, they are still considered part of the workforce and are expected to return to work as soon as the crisis passes. Hence, they are in limbo: still considered part of the organization, and at the same time, detached from it. The furlough status represents a potential or actual threat to employee resources (Carnevale & Hatak, 2020) such as social conditions, organizational status, and financical means (Hobfoll et al., 2018). This potential or actual resource loss amplifies emotional exhaustion (e.g., Costa & Neves, 2017) and contributes to contract violation perception (Delegach et al., 2022). Moreover, while even expected organizational changes threaten employee resources and contribute positively to perceived job insecurity (e.g., Ito & Brotheridge, 2007), the lack of predictability and control during unforseable crises is associated with a heightened perception of job insecurity (Østhus, 2007; Probst & Lawler, 2006). This is especially so for furloughed employees, who face both a threat to and the actual loss of their jobs (Halbesleben et al., 2013). Accrodingly, most research refers to furlough as a disruptive strategy that impairs employee job security, accelerates work-family conflict, and breaches the psychological contract between employer and employee (e.g., Baranik et al., 2019; Mandeville et al., 2019). In addition, furlough creates a physical distance between the employees and their employers that disrupts interaction frequency (especially in the case of Israel 1 ) and corresponds with the concept of leader-follower distance. Napier and Ferris (1993) defined distance between the leader and followers as a “multidimensional construct that describes the psychological, structural, and functional separation, disparity, or discord between a supervisor and a subordinate” (p. 326). The psychological dimension of distance refers to the degree of intimacy in social interactions between leader and follower and is based on perceived differences in rank, status, power, and social standing (Antonakis & Atwater, 2002). The structural distance dimension refers to the physical distance between the leader and follower due to organizational structure, the leader's span of control, and physical placement (Napier & Ferris, 1993). Finally, the functional distance dimension refers to the quality of the working relationships between leader and follower (Napier & Ferris, 1993). Thus, employees who continued to work during COVID-19 lockdowns experienced mainly the expansion of structural distance due to remote or shifting work arrangements. Contrary to working employees, furloughed employees are in a more complex situation. During the furlough period, managers are prevented from requesting their employees to contribute to the organization or to work during the furlough period, and even from consulting with them about their organizational tasks and responsibilities. At the same time, since furloughed employees are still considered part of the workforce, their managers are expected to retain them by occasionally keeping them informed on their organizational status (Huffman et al., 2022) and by trying to minimize the loosening of the furloughed employees’ network ties with their peers in the organization (Ruiz-Palomino et al., 2022). Therefore, compared to employees who continued to work during COVID-19 lockdowns, furloughed employees experienced the expansion of distance in at least two dimensions—structural and functional—that resemble those of physical distance and perceived frequency of leader-follower interaction put forward by Antonakis and Atwater (2002). The influence of distance on charismatic leadership effectiveness is still undetermined. While some scholars have argued that distance is a “necessary requisite” for the charismatic leader's influence (Antonakis & Atwater, 2002; Katz & Kahn, 1978), others suggest that it may neutralize the leadership effect (Kerr & Jermier, 1978) or that intervening factors, such as employee characteristics, may moderate the association between leader distance and its influence on the followers (Antonakis & Atwater, 2002). Based on this inconclusiveness, we suggest that the relationships between charismatic leadership and employees’ attitudinal, emotional, and well-being outcomes, for employees who were furloughed or for those who continued to work during the lockdown period, depend on the employees’ receptiveness to this leadership style. The charismatic leader's positive emotional displays, confidence, optimism, and enthusiasm (Bono et al., 2007) bust followers’ emotional and motivational resources, which buffer the latter's negative reactions to stress (LePine et al., 2016). These resources are valuable since their availability enables individuals to overcome obstacles and thereby attenuates their emotional exhaustion (Wright & Hobfoll, 2004). Moreover, the charismatic leader primes the followers’ level of self-esteem and collective identity (Shamir et al., 1993) and reinforces their sense of belonging to the organization (Epitropaki, 2003). This enhances the followers’ social identification with their workgroup and the organization (Epitropaki, 2013) and thus leads to lower levels of contract violation and job insecurity. The charismatic leader as the channel of resources is particularly essential for furloughed employees since he/she is a key source of information, and sometimes even the only connection between the individual and the organization. Thus, furloughed employees stand to benefit more from a charismatic leader than employees who continue to work in their positions, since the reservoir of resources for those on furlough is more depleted. However, given that charismatic leadership helps elevate followers’ self-esteem, self-confidence, and collective identity (Shamir et al., 1993), we suggest that it is crucial that employees attune themselves to this kind of leadership style and possess enough psychological availability to use affirmative communication with their leader. In this respect, employees who retained their work during COVID-19 (i.e., closer followers) preserved a relatively stable level of employment and functioning compared to furloughed employees (i.e., more distant followers). Among those in the first group, employees who were also characterized by a low level of IU experienced less worry (Laugesen et al., 2003) and felt less anxiety and stress (Greco & Roger, 2003); in other words, their resources were a priori less drained, which in turn reduced their readiness for charismatic leadership (Madsen & Snow, 1991). Therefore, such employees may have been less susceptible to the positive influence of charismatic leadership than those characterized by high IU levels who retained their work during the lockdowns. On the other hand, for furloughed (i.e., more distant) employees, we propose inverse associations between charismatic leadership style and employee attitudinal, emotional, and well-being outcomes depending on the latter's IU level. The optimal interactional frequency between a leader and a follower is contingent on situational variables (Kerr & Jermier, 1978). In situations characterized by ambiguity and uncertainty, followers need more socioemotional interaction with their leader (Antonakis & Atwater, 2002). However, furloughs increase the extension of functional distance with the leader and consequently decrease the opportunity for leader support (Graen & Uhl-Bien, 1995) and the leader's ability to influence followers’ self-concept and motivation, which in turn may also contribute to the employees’ perceived loss of resources (i.e., developmental opportunities, status, and support; Hobfoll et al., 2018; Mandeville et al., 2019). In addition, tolerance of uncertainty is a valuable personal resource, an aspect of the self linked to an individual's ability to successfully control and impact their environment (Hobfoll et al., 2018). Thus, furloughed employees who are also characterized by high levels of IU may face a resource loss spiral, which increases stress levels to the point that the charismatic leader's affirmative contribution is not enough or not perceived to exist due to functional distance, which cannot overcome said depletion. This progressive loss spiral undermines the employees’ coping abilities and drains their resources, a situation that spills over and generalizes into emotional exhaustion and negative context-free outcomes (Hakanen & Schaufeli, 2012), such as ruminations about job insecurity (Richter et al., 2020). Moreover, since psychological contract maintenance and re-building often require resource investment (Bankins, 2015; Tomprou et al., 2015), the depletion of resources is likely to erode the foundation of the psychological contract and to deepen the feeling of contract violation. In this respect, we suggest that high-IU employees furloughed during a COVID-19 lockdown were less capable of using adaptive coping strategies, such as relying on charismatic leadership, and this amplified emotional exhaustion, contract violation, and job insecurity. To conclude, we posit that followers’ IU and employment status during lockdown jointly moderate the association between charismatic leadership and followers’ emotional, well-being, and attitudinal indicators after returning from the lockdown to regular work. Specifically, we suggest that charismatic leadership is more useful in intermediate conditions (FS+, IU- or FS-, IU+) compared to extreme conditions, either because the followers are less susceptible to charismatic leadership (FS-, IU-) or because they are overwhelmed and thus cannot embrace its “bright side” (FS+, IU+). Hypotheses 1–3:There will be a three-way interaction among charismatic leadership, furlough status, and IU on psychological contract violation (H1), emotional exhaustion (H2), and job insecurity (H3), such that: The associations between charismatic leadership style and psychological contract violation (H1a), emotional exhaustion (H2a), and job insecurity (H3a) will be negative for furloughed employees with low levels of IU (FS+, IU–), while for furloughed employees with high levels of IU (FS+, IU+) the negative contribution of charismatic leadership style to employees’ outcomes will be weak or nonexistent. The associations between charismatic leadership style and psychological contract violation (H1b), emotional exhaustion (H2b), and job insecurity (H3b) will be negative for employees who worked during the lockdown and had high levels of IU (FS-, IU+), while for employees who worked during the lockdown and had low levels of IU (FS-, IU-) the negative contribution of charismatic leadership style to employees outcomes will be weak or nonexistent. Methods Sample and Procedures Data for this research were part of a broader data collection effort implemented through a firm that provides online survey services. This firm has access to a comprehensive sample of employees in a variety of occupations and work roles. The criteria for participation in the study were: (1) over 21 years of age; (2) full-time employment prior to the COVID-19 pandemic; (3) had worked with their direct manager for at least six months, to allow the participant to become familiar with the leader's leadership style; and (4) participant tenure in the current organization of at least a year. The research protocol received the necessary approvals from the Institutional Review Board. Stage 1 (time 1; T1) of the data collection process was conducted in March–April 2020 (the beginning of the first lockdown in Israel). The participants were asked to complete the online survey measuring the independent study variables—the evaluation of their direct manager's charismatic leadership style and their own intolerance of uncertainty. In addition, we collected information about the participants’ current employment status and demographics. Stage 2 of the data collection process (time 2; T2) was conducted in mid-August 2020, when the economy started to recover after the lockdown. We returned to participants who took part in Stage 1 and asked them to complete an additional online survey. In this stage, we collected the dependent variables—contract violation, emotional exhaustion, and job insecurity—and asked the participants to report their current employment status. Four-hundred and ninety-nine individuals participated in Stage 1 of the data collection process, while 360 followed up in Stage 2 (i.e., attrition of 27.86%). Twenty-four participants remained on furlough also in Stage 2 or were fired from their organization. Thus, the final research sample was comprised of N = 336 who continued to work during the lockdown (n = 199) or returned to the organization after being furloughed during the first lockdown (n = 137). The final sample demographics are as follows: mean age was 43.49 years (SD = 11.60), 57.1% of the respondents were women, the mean seniority in the organization was 8.83 years (SD = 10.49), and the mean seniority with their current direct manager was 5.08 years (SD = 5.08). Sixty-three point one percent (63.1%) of participants worked in private sector organizations and 36.9% were employed in the public sector. Participants received a small honorarium for their participation. Measures The study scales were translated and back-translated into Hebrew to check the reliability of the translation. The study's independent variables were measured in Stage 1: Charismatic leadership style was measured using De Hoogh et al.'s (2005) 8-item Charismatic Leadership scale (sample item: “Has a vision and imagination of the future”; α = .89). A 7-point Likert scale was used to score responses ranging from (1): strongly disagree to (7): strongly agree. Intolerance of Uncertainty was measured using Carleton et al.'s (2007) 12-item Intolerance of Uncertainty short-form scale (sample item: “I should be able to organize everything in advance”; α = .91). A 5-point Likert scale was used to score responses ranging from (1): not at all characteristic of me to (5): entirely characteristic of me. The study's dependent variables were measured at Stage 2: Feeling of psychological contract violation was measured using Robinson and Wolfe Morrison's (2000) four-item scale (sample item: “I feel extremely frustrated by how I have been treated by my organization”; α = .92). A 5-point Likert scale was used to score responses ranging from (1): strongly disagree to (5): strongly agree. Emotional exhaustion was measured using Wilk and Moynihan's (2005) four-item scale (sample item: “I feel burned out from my work”; α = .91). Participants were asked to assess the frequency of experiencing certain emotions over the previous weeks using a scale that ranged from (1): never to (7): almost every day. Job insecurity was assessed using the four-item Job Insecurity Scale proposed by Vander Elst et al.(2014), which captures employees’ cognitive and emotional perceptions of job insecurity (sample item: “I feel insecure about the future of my job”; α = .85). A 5-point Likert scale was used to score responses in a range going from (1): strongly disagree to (5): strongly agree. Control variables. We controlled for employees’ organizational tenure and educational level since previous research has demonstrated negative associations between these constructs and job insecurity (e.g., Adkins et al., 2001; Näswall & De Witte, 2003) and emotional exhaustion (e.g., Dunford et al., 2012; Hwang et al., 2021; Qin et al., 2014). We also controlled for gender because prior research has suggested that it may be related to emotional exhaustion (e.g., Purvanova & Muros, 2010) and perceived contract violation (e.g., Stoner & Gallagher, 2010). Additionally, given that our research focuses on the contribution of leadership style to employees’ outcomes and the latter are likely to be influenced by how long the leader and the employee have worked together (Hu & Shi, 2015; Kark et al., 2015), we controlled for the length of the dyadic relationship between the leader and employee. Results First, we conducted the omnibus test of the hypothesized five-factor model. The results revealed the following fit indices: (χ2 = 924.45, df = 445, p < .001; CFI = .92; TLI = .92; SRMR = .06; RMSEA = .06), where CFI refers to the comparative fit index, TLI is a Tucker–Lewis index, SRMR refers to the standardized root mean squared residual, and RMSEA refers to the root mean square error of approximation. In addition, we examined three alternative models. The first model was a general model in which all items loaded on a single factor revealed a nonacceptable level of fit (χ2 = 3686.67, df = 455, p < .001; CFI = .49; TLI = .44; SRMR = .17; RMSEA = .15). The second two-correlated higher-order factor model examines the items’ loading according to the time of the questionnaire distribution. The items that were measured at the same time point were loaded on the same factor (i.e., the charismatic scale and IU items were loaded on one factor and the research's dependent variables were loaded on a different factor). The results of this model also demonstrated a nonacceptable level of fit (χ2 = 2699.10, df = 454, p < .001; CFI = .64; TLI = .61; SRMR = .15; RMSEA = .12). The third three-correlated higher-order factor model, where the charismatic leadership style items were loaded on the first factor, the IU items were loaded on the second factor, and the study's dependent variables were loaded on the third factor (χ2 = 1843,82, df = 452, p < .001; CFI = .78; TLI = .76; SRMR = .09; RMSEA = .10), also revealed a nonacceptable fit index. Table 1 presents the means, standard deviations, and correlations among demographic and research variables. Table 1. Means, Standard Deviations, and Correlations among the Study Variables. M (Sd) 1 2 3 4 5 6 7 8 9 1. Furlough statusa 1.41 (.49)   2. Charismatic leadership 5.18 (1.15) −.01   3. Intolerance of uncertainty 2.89 (.78) .17** .10   4. Job insecurity 2.35 (.95) .26** −.10 .28**   5. Contract violation 1.78 (.98) .20** −.12* .33** .43** 6. Emotional exhaustion 3.47 (1.44) .07 −.15** .28** .36** .51** 7. Genderb 1.57 (.50) .21** .11 .05 −.04 −.06 .06 8. Educational years 14.97 (3.09) −.16** .03 −.08 −.12* .02 .02 −.09 9. Organizational tenure 8.83 (10.49) −.06 .03 −.08 −.09 −.05 −.08 .07 .12* 10. Tenure with manager 5.08 (5.08) .06 −.03 .04 .04 −.06 −.07 .06 −.02 .44** Note. N = 336. *p < .05. **p < .01. a 1 = participants who worked during the lockdown, 2 = participants who were furloughed; b 1 = male, 2 = female. In order to test the research hypotheses, we applied the SPSS PROCESS macro to test the interactive effects (Hayes, 2013; Model 3). To estimate the hypothesized conditional relationships, we used a bootstrap procedure (Preacher & Hayes, 2004) with 95% bias-corrected confidence intervals (CIs) and 5,000 sampling replications. We mean-centered the study's independent variables to enhance the regression coefficients’ interpretability (Hayes, 2013). Table 2 presents the results of the conditional analyses. We controlled for participants’ gender, years of education, organizational tenure, and tenure with their direct manager in all our analyses. Table 2. Regression Results for Mediation and Conditional Indirect Effects. Effect Contract violation Emotion exhaustion Job Insecurity Model 1 Model 2 Model 3 B (SE) B (SE) B (SE) Constant 1.71** (.31) 2.87** (.46) 2.81** (.30) Gender −.17 (.10) .20 (.15) −.18+ (.10) Educational years −.02 (.02) .03 (.02) −.02 (.02) Organizational tenure .00 (.01) −.00 (.01) −.01 (.01) Tenure with manager −.02 (.01) −.02 (.02) .01 (.01) Furlough Statusa .32** (.11) .00 (.16) .37** (.10) Charismatic leadership −.10 (.06) −.18+ (.09) .00 (.06) Intolerance of uncertainty .40** (.08) .58** (.13) .16+ (.08) Furlough Status X Charismatic leadership −.06 (.09) −.13 (.13) −.19* (.09) Furlough Status X Intolerance of uncertainty .02 (.13) −.10 (.20) .31* (.13) Charismatic leadership X Intolerance of uncertainty −.09 (.07) −.08 (.11) −.07 (.07) Furlough Status X Charismatic leadership X Intolerance of uncertainty .24* (.10) .37* (.16) .22* (.10) Intolerance of uncertainty Furlough status B (SE) B (SE) B (SE) Low Work −.02 (.08) −.11 (.12) .06 (.08) Furloughed −.27** (.09) −.54** (.13) −.31** (.08) High Work −.17+ (.09) −.24+ (.13) −.05 (.09) Furloughed −.05 (.08) −.09 (.12) −.07 (.08) Note. a 1 = participants who worked during the lockdown, 2 = participants who were furloughed; N = 336 for contract Model 1 and Model 2, and N = 321 for Model 3; +p < .10, *p < .05, **p < .01. The results demonstrated that the three-way interactions of furlough status, charismatic leadership, and IU on contract violation (F(11,323) = 6.70, p < .001), emotional exhaustion (F(11,323) = 5.20, p < .001), and job insecurity (F(11,309) = 6.71, p < .001) were significant. The interactions are depicted in Figures 2, 3, and 4. A further inspection of the interaction results revealed that the conditional effects were significant for furloughed participants with low levels of IU (FS+, IU-; b = −.27, p = .002; b = −.54, p > .001; b = −.31, p > .001 for contract violation, emotional exhaustion, and job insecurity, respectively), but not for those with high levels of IU (FS+, IU-; b = −.05, p = .55; b = −.09, p = .45; b = −.07, p = .39 for contract violation, emotional exhaustion, and job insecurity, respectively). Moreover, the conditional effects were marginally significant for participants who continued to work during the lockdown and had high levels of IU (FS-, IU+; although not significant at the traditional p < .05, the interactions were significant at b = −.17, p = .058; b = −.24, p = .069 for contract violation and emotional exhaustion, respectively), but not for those of them who had low levels of IU (FS-, IU-; b = −.02, p = .81; b  =  -.11, p  =  .37 for contract violation and emotional exhaustion, respectively). We conducted simple slope analyses (Aiken & West, 1991) that replicated the findings and demonstrated that negative effects of charismatic leadership style on psychological contract violation, emotional exhaustion, and job insecurity were apparent for participants who had a low IU and were on furlough (FS+, IU-; t = −2.99, p = .003; t = −3.80, p < .001; t = −3.62, p < .001 for contract violation, emotional exhaustion, and job insecurity, respectively) and for participants with a high IU who continued working in their organization (FS-, IU+; although not significant at the traditional p < .05, the results of the t-tests were significant at t = −1.66, p = .097; t = −1.80, p = .073 for contract violation and emotional exhaustion, respectively). Figure 2. Interaction of charismatic leadership style, intolerance of uncertainty, and furlough status on psychological contract violation. Figure 3. Interaction of charismatic leadership style, intolerance of uncertainty, and furlough status on emotional exhaustion. Figure 4. Interaction of charismatic leadership style, intolerance of uncertainty, and furlough status on job insecurity. Discussion The purpose of the current study was to investigate the contribution of charismatic leadership to employees’ attitudinal, emotional, and well-being outcomes during a crisis. Based on Klein and House's (1995) notion of charisma as a combination of leader qualities (“the spark”), follower readiness to accept charismatic influence (“flammable material”), and a charisma-conducive environment (”oxygen”), we suggested that the interplay between two potential moderators (i.e., employee work status as a contextual factor and IU as a trait characteristic) may offer a more expansive and nuanced understanding of these relationships. Indeed, the study results revealed that the relationships between charismatic leadership style and employee outcomes depend on the combination of furlough status and employees’ levels of IU. Specifically, we found a negative contribution of charismatic leadership to employees’ levels of job insecurity, contract violation, and emotional exhaustion for furloughed employees who had low levels of IU, and a negative contribution of charismatic leadership to employees’ levels of contract violation and emotional exhaustion for working employees who showed high levels of IU. Thus, following the study hypotheses, our results demonstrated that charismatic leadership in a crisis context makes more of a contribution to follower outcomes in situations of intermediate levels of stress (i.e., FS+, IU- or FS-, IU+) and less at the edges (i.e., FS -, IU- or FS+, IU+; see Figure 1). Explicitly, we did not find any significant contribution of charismatic leadership style to employee outcomes for employees who continued to work during the lockdown and had low levels of IU and for furloughed employees characterized by high levels of IU. The results of the first group (FS-, IU-) corresponded with the previous literature and demonstrated that susceptibility to charismatic leadership depended on followers’ feelings of insecurity and ambiguity (Wegge et al., 2022). Thus, the employees who continued to work during the lockdown experienced less concern about an unpredictable future and felt less control loss and distress, which reduced their receptiveness to charismatic leadership. A more interesting finding refers to the opposite extreme and includes the furloughed employees with a high level of IU (FS+, IU+). While this group was the neediest in terms of support and reassurance from their leaders, the results demonstrated that charismatic leadership did not significantly contribute to their feeling of psychological contract violation, their level of emotional exhaustion, and the degree of insecurity regarding their job. The explanation for these results is driven by the COR theory (Hobfoll, 1989; Hobfoll et al., 2018) and suggests that a furlough status diminishes the employees’ work stability, reduces their economic, psychological, and mental resources (Baranik et al., 2019), and expands their functional distance from the leader. If the furloughed employees also suffered from high IU, their perception of uncertainty may have increased their anxiety (McEvoy & Mahoney, 2012) and avoidance behaviors (Shapiro et al., 2020), and thus depleted their personal resources, creating a resource loss spiral in which their levels of stress increased even more (Hobfoll, 1989). The stress level may have become so intense that it blocked their emotional availability to be attuned to leadership messages and behaviors, as charismatic as the latter may be. These study findings may not only extend the COR theory by providing unique opportunities to better understand resource dynamics in the context of a crisis and its behavioral outcomes, but also respond to Hobfoll et al.'s (2018) call to advance the understanding of the interplay of different resources and their combined effect on individuals. The current study also focused on the role of distance in leader-follower dynamics. Although the distance concept has been defined as a critical moderator of leadership's influencing process (Antonakis & Atwater, 2002), this concept is underexplored in the leadership literature (Antonakis & Jacquart, 2013). Even though several scholars believe that distance is essential to create an impact because intimacy between leader and followers may destroy the “aura of magic” (e.g., Katz & Kahn, 1978), some (e.g., Yagil, 1998) are less decisive and suggest that proximal charismatic leadership also has an affirmative influence on its followers, while others view distance as detrimental to leader-follower relationships (Kerr & Jermier, 1978). The functional distance created through the furlough experience may lead to adverse employee outcomes (Napier & Ferris, 1993) and undermine the leader's ability to motivate and inspire followers (Antonakis & Atwater, 2002). However, our study revealed that the effect of the distance dimension cannot be separated from the employees’ dispositional characteristics, which emphasizes the complicated connection between distance and charismatic leadership effectiveness. More studies are needed to explore follower traits as the basis for the leader effectiveness-distance dynamic. The study results also corresponded with the emerging body of research that highlights the paradoxical effect of positive leadership styles, such as the contingency approach (e.g., Fiedler, 1978; House, 1971), the too-much-of-a-good-thing effect (TMGT; e.g., Pierce & Aguinis, 2013), and the “dark side” possibility of the positive leadership styles (e.g., House & Howell, 1992; Sharma & Kirkman, 2015). For example, Sumanth (2011) found that upward communication between inclusive leaders and their followers is characterized by high quantity but lower quality. Another study demonstrated that transformational leaders’ behaviors contribute to an increase in their emotional exhaustion and turnover intentions (Lin et al., 2019). In the context of charismatic leadership, the idea that it has a similar effect across all settings and employees has been criticized before (e.g., Judge et al., 2009; Wegge et al., 2022) and several scholars even suggested that under certain circumstances, charismatic leadership can arouse a negative impact on their followers (e.g., Fragouli, 2018; Yukl, 1999). In line with this, House and Howell (1992) proposed that this leadership style may intensify organizational risk and make the decision-making process more uncertain. The current study's results revealed that in a crisis context, charismatic leadership did not diminish its negative impact on edge-case employees’ emotional, well-being, and attitudinal outcomes. Thus, the current study contributes to this line of thought and suggests that the common notion that charismatic leadership is essential during crises should be considered with added caution. Like many other leadership styles, this socially desirable leadership characteristic can have positive, neutral, or negative effects depending on both the situation and the followers’ dispositional traits. However, only a scarcity of studies empirically explores the charismatic leadership effect during a crisis (e.g., Crayne & Medeiros, 2021; Williams et al., 2021). This scarcity offers an opportunity for future research to further develop an understanding of when and why charismatic leadership might not always be the best “fit” in crisis situations. Managerial Implications Our research indicates that followers differ in their susceptibility to charismatic leadership influence in a crisis setting. This susceptibility depends on the interplay between employee status and dispositional characteristics. Previous studies have revealed the adverse effect of furlough on employees’ attitudes and behaviors (e.g., Baranik et al., 2019; Mandeville et al., 2019); thus, furloughed employees are a critical group with whom leaders have to re-establish their contribution to organizational outcomes. Our findings suggest that in order to implement charismatic leadership's positive contribution, leaders need to ensure that the furloughed employees are tolerant of uncertainty. As such, organizations need to invest in followership interventions, such as mentoring and coaching programs, that help build up employee resilience to uncertainty (Boswell et al., 2013). Additionally, the current crisis has given rise to the implementation of the furlough strategy. Furlough helps the organization save its economic resources and at the same time preserve its workforce. However, furlough also has a negative emotional impact on employees (Mandeville et al., 2019). Unfortunately, crises are inevitable in organizational life, and we can assume that the use of furlough will remain one of the accepted strategies for dealing with future crises. Thus, organizations that are forced to implement this strategy should inspect their employees’ IU dispositions to decide who will be furloughed. This follower characteristic may influence the organization's decision about workforce rearrangement in a crisis setting. Moreover, since furlough has garnered only limited attention in the literature, this study suggests novel insights into the contribution of leadership to furloughed employees. However, additional studies are needed to capture the full impact of this work arrangement on the employee-employer relationship. The study also highlighted the “bright side” (Judge et al., 2009) of charismatic leadership in a crisis context, but only for intermediate-cases employees. Since organizational crises occur from time to time (Klein & Eckhaus, 2017), organizations should encourage and promote employees who are high in charisma into managerial positions, and above all, into crisis management roles. However, as Osborn et al. (2002) suggest, in crisis situations a subtle dialog between leaders and their followers is essential. This dialog requires the leader to give an interpretation of the crisis, formulate the meaning of success, and clearly define the process to overcome the crisis. These patterns of behaviors are more “mundane” and “managerial” than the charismatic aspects of leadership commonly emphasized in crisis contexts. Thus, charismatic leaders must include in their toolbox also managerial tools that may help them tap into the edge case employees, since this clear dialog may reduce the latter's feelings of uncertainty and ambiguity. Limitations and Future Research This research is subject to some potential limitations. First, the study results may be susceptible to same-source bias because all variables were collected from the study participants via online surveys. However, the study's design minimizes the potential for this bias given that we used the objective data as one of the study's independent variables and implemented a five months’ temporal separation between measuring the independent and dependent variables (T2). Using time lag and objective data variables can reduce the threat of common method variance (Podsakoff et al., 2003). Second, in the current research we proposed that furlough status corresponds with the functional distance dimension. Although this assumption stemmed from the government's strict regulations regarding contact with furloughed employees, we did not explicitly examine the functional dimension construct. Future research may take this up to validate the correspondence. An additional weakness of the study is associated with the measurement of charismatic leadership. Although the charismatic leadership style continues to play a prominent role in leadership research (Antonakis et al., 2016), it evokes a bit of criticism due to its ambiguous conceptualization and overlapping structural composition (Antonakis et al., 2016; Van Knippenberg & Sitkin, 2013). Critics have suggested that scholars should go “back to the drawing board” and focus on measures of the constituents’ parts of charisma. However, these parts of charisma share a “meaningful core that makes them greater than the sum of their parts” (Sy et al., 2018, p. 68). Nevertheless, given this criticism, we recommend that future research makes use of objective measures of charisma (e.g., Fanelli et al., 2009, Jacquart & Antonakis, 2015, Mio et al., 2005) as well as the manipulation of charisma in experimental settings (e.g., Antonakis et al., 2011). Finally, this research was conducted in Israel, which allows us to examine the most extreme furlough case, which could not be downgraded to part-time due to existing government regulations. Israeli culture is characterized by low power distance, and its citizens are less willing to accept inequality between the less powerful and more powerful members of society (Hofstede, 1980). The furlough status may expand the differences in organizational power distribution, resulting in more acute employee reactions compared to those of employees from cultures with a higher level of power distance. Future research may examine the study model in other cultural contexts. The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding: The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was funded by Heth Academic Center for Research of Competition and Regulation (grant number RA2000000666), and Ariel University (grant number 7496). ORCID iDs: Galit Klein https://orcid.org/0000-0002-4062-4446 Marianna Delegach https://orcid.org/0000-0002-7895-9233 1. The Israeli model allows only a binary option—either keep the employees at work or fully furlough them. ==== Refs References Adkins C. L. Werbel J. D. Farh J. L. (2001). A field study of job insecurity during a financial crisis. Group & Organization Management, 26 (4), 463-483. 10.1177/1059601101264004 Aiken L. S. West S. G. (1991). Multiple regression: Testing and interpreting interactions. Sage Publications, Inc. Antonakis J. (2021). Leadership to defeat COVID-19. Group Processes & Intergroup Relations, 24 (2), 210-215. 10.1177/1368430220981418 Antonakis J. Atwater L. (2002). Leader distance: A review and a proposed theory. The Leadership Quarterly, 13 (6), 673-704. 10.1016/S1048-9843(02)00155-8 Antonakis J. Bastardoz N. Jacquart P. Antonakis J. Bastardoz N. Jacquart P. Shamir B. (2016). Charisma: An ill-defined and ill-measured gift. Annual Review of Organizational Psychology and Organizational Behavior, 3 (1), 293-319. 10.1146/annurev-orgpsych-041015-062305 Antonakis J. Fenley M. Liechti S. (2011). Can charisma be taught? Tests of two interventions. Academy of Management Learning & Education, 10 (3 ), 374-396. 10.5465/amle.2010.0012 Antonakis J. Jacquart P. (2013). The far side of leadership: Rather difficult to face”. In Bligh M. C. Riggio R. E. (Eds.), When near is far and far is near: Distance in leader follower relationships (pp. 155-187). Routledge. Arshad R. (2016). Psychological contract violation and turnover intention: Do cultural values matter? Journal of Managerial Psychology, 31 (1), 251-264. 10.1108/JMP-10-2013-0337 Bankins S. (2015). A process perspective on psychological contract change: Making sense of, and repairing, psychological contract breach and violation through employee coping actions. Journal of Organizational Behavior, 36 (8), 1071-1095. 10.1002/job.2007 Banks G. C. Engemann K. N. Williams C. E. Gooty J. McCauley K. D. Medaugh M. R. (2017). A meta-analytic review and future research agenda of charismatic leadership. The Leadership Quarterly, 28 (4 ), 508-529. 10.1016/j.leaqua.2016.12.003 Baranik L. E. Cheung J. H. Sinclair R. R. Lance C. E. (2019). What happens when employees are furloughed? A resource loss perspective. Journal of Career Development, 46 (4 ), 381-394. 10.1177/0894845318763880 Barling J. Frone M. R. (2017). If only my leader would just do something! passive leadership undermines employee well-being through role stressors and psychological resource depletion. Stress and Health, 33 (3 ), 211-222. 10.1002/smi.2697 27470980 Bass B. M. Bass R. (2008). Handbook of leadership: Theory, research, and application. Free Press. Bono J. E. Foldes H. J. Vinson G. Muros J. P. (2007). Workplace emotions: The role of supervision and leadership. Journal of Applied Psychology, 92 (5), 1357-1367. 10.1037/0021-9010.92.5.1357 17845090 Boswell J. F. Thompson-Hollands J. Farchione T. J. Barlow D. H. (2013). Intolerance of uncertainty: A common factor in the treatment of emotional disorders. Journal of Clinical Psychology, 69 (6 ), 630-645. 10.1002/jclp.21965 23381685 Bryman A. Stephens M. a Campo C. (1996). The importance of context: Qualitative research and the study of leadership. The Leadership Quarterly, 7 (3 ), 353-370. 10.1016/S1048-9843(96)90025-9 Buhr K. Dugas M. J. (2002). The intolerance of uncertainty scale: Psychometric properties of the English version. Behaviour Research and Therapy, 40 (8 ), 931-945. 10.1016/S0005-7967(01)00092-4 12186356 Carleton R. N. Norton M. P. J. Asmundson G. J. (2007). Fearing the unknown: A short version of the intolerance of uncertainty scale. Journal of Anxiety Disorders, 21 (1 ), 105-117. 10.1016/j.janxdis.2006.03.014 16647833 Carnevale J. B. Hatak I. (2020). Employee adjustment and well-being in the era of COVID-19: Implications for human resource management. Journal of Business Research, 116 , 183-187. 10.1016/j.jbusres.2020.05.037 32501303 Celik D. Alpay E. H. Celebi B. Turkeli A. (2021). Intolerance of uncertainty, rumination, post-traumatic stress symptoms and aggression during COVID-19: A serial mediation model. European Journal of Psychotraumatology, 12 (1 ), 1953790. 10.1080/20008198.2021.1953790 34408818 Costa S. P. Neves P. (2017). Forgiving is good for health and performance: How forgiveness helps individuals cope with the psychological contract breach. Journal of Vocational Behavior, 100 , 124-136. 10.1016/j.jvb.2017.03.005 Crayne M. P. Medeiros K. E. (2021). Making sense of crisis: Charismatic, ideological, and pragmatic leadership in response to COVID-19. American Psychologist, 76 (3 ), 462-474. 10.1037/amp0000715 32772539 Day D. V. Antonakis J. (2012). Leadership: Past, present, and future. In D. V. Day & J. Antonakis (Eds.), The Nature of Leadership (2nd ed.). Thousand Oaks: Sage. De Hoogh A. den Hartog D. Koopman P. Thierry H. van den Berg P. van der Weide J. Wilderom C. (2004). Charismatic leadership, environmental dynamism, and performance. European Journal of Work and Organizational Psychology, 13 (4 ), 447-471. 10.1080/13594320444000164 De Hoogh A. H. Den Hartog D. N. Koopman P. L. Thierry H. Van den Berg P. T. Van der Weide J. G. Wilderom C. P. (2005). Leader motives, charismatic leadership, and subordinates’ work attitude in the profit and voluntary sector. The Leadership Quarterly, 16 (1 ), 17-38. 10.1016/j.leaqua.2004.10.001 De Vries R. E. Roe R. A. Taillieu T. C. (1998). Need for supervision: Its impact on leadership effectiveness. The Journal of Applied Behavioral Science, 34 (4 ), 486-501. 10.1177/0021886398344015 Delegach M. Kark R. Katz-Navon T. Van Dijk D. (2017). A focus on commitment: The roles of transformational and transactional leadership and self-regulatory focus in fostering organizational and safety commitment. European Journal of Work and Organizational Psychology, 26 (5 ), 724-740. 10.1080/1359432X.2017.1345884 Delegach M. Klein G. Katz-Navon T. (2022). Furlough and its effects on employees after returning to work: The roles of psychological contract breach and violation, and perceived organizational support. Journal of Management & Organization, 1-18. 10.1017/jmo.2022.71 Den Hartog D. N. Belschak F. D. (2012). When does transformational leadership enhance employee proactive behavior? The role of autonomy and role breadth self-efficacy. Journal of Applied Psychology, 97 (1 ), 194-202. 10.1037/a0024903 21842977 Devereux G. (1955). Charismatic leadership and crisis. International Universities Press. Incorporated. Dugas M. J. Gosselin P. Ladouceur R. (2001). Intolerance of uncertainty and worry: Investigating specificity in a nonclinical sample. Cognitive Therapy and Research, 25 (5 ), 551-558. 10.1023/A:1005553414688 Dunford B. B. Shipp A. J. Boss R. W. Angermeier I. Boss A. D. (2012). Is burnout static or dynamic? A career transition perspective of employee burnout trajectories. Journal of Applied Psychology, 97 (3 ), 637-650. 10.1037/a0027060 22309410 Endler N. S. Macrodimitris S. D. Kocovski N. L. (2000). Controllability in cognitive and interpersonal tasks: Is control good for you? Personality and Individual Differences, 29 (5 ), 951-962. 10.1016/S0191-8869(99)00246-9 Epitropaki O. (2003). Transformational leadership, psychological contract breach and organizational identification. Academy of Management Best Paper Proceedings, OB, pp. M1-M6. Epitropaki O. (2013). A multi-level investigation of psychological contract breach and organizational identification through the lens of perceived organizational membership: Testing a moderated–mediated model. Journal of Organizational Behavior, 34 (1), 65-86. 10.1002/job.1793 Fanelli A. Misangyi V. F. Tosi H. L. (2009). In charisma we trust: The effects of CEO charismatic visions on securities analysts. Organization Science, 20 (6 ), 1011-1033. 10.1287/orsc.1080.0407 Fiedler F. E. (1978). The contingency model and the dynamics of the leadership process. In Advances in experimental social psychology (Vol. Vol. 11 , pp. 59-112). Academic Press. Fragouli E. (2018). The dark-side of charisma and charismatic leadership. Business & Management Review, 9 (4 ), 298-307. Ganson K. T. Tsai A. C. Weiser S. D. Benabou S. E. Nagata J. M. (2021). Job insecurity and symptoms of anxiety and depression among US young adults during COVID-19. Journal of Adolescent Health, 68 (1 ), 53-56. 10.1016/j.jadohealth.2020.10.008 Graen G. B. Uhl-Bien M. (1995). Relationship-based approach to leadership: Development of leader-member exchange (LMX) theory of leadership over 25 years: Applying a multi-level multi-domain perspective. The Leadership Quarterly, 6 (2 ), 219-247. 10.1016/1048-9843(95)90036-5 Greco V. Roger D. (2003). Uncertainty, stress, and health. Personality and Individual Differences, 34 (6 ), 1057-1068. 10.1016/S0191-8869(02)00091-0 Hakanen J. J. Schaufeli W. B. (2012). Do burnout and work engagement predict depressive symptoms and life satisfaction? A three-wave seven-year prospective study. Journal of Affective Disorders, 141 (2-3), 415-424. 10.1016/j.jad.2012.02.043 22445702 Halbesleben J. R. Wheeler A. R. Paustian-Underdahl S. C. (2013). The impact of furloughs on emotional exhaustion, self-rated performance, and recovery experiences. Journal of Applied Psychology, 98 (3), 492-503. 10.1037/a0032242 23506412 Hayes A. F. (2013). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach. Guilford Press. Hillen M. A. Gutheil C. M. Strout T. D. Smets E. M. Han P. K. (2017). Tolerance of uncertainty: Conceptual analysis, integrative model, and implications for healthcare. Social Science & Medicine, 180 , 62-75. 10.1016/j.socscimed.2017.03.024 28324792 Hobfoll S. E. (1988). The ecology of stress. New-York:Taylor & Francis. Hobfoll, S. E. (1989). Conservation of resources: A new attempt at conceptualizing stress. American Psychologist, 44(3), 513-524. 10.1037/0003-066X.44.3.513. Hobfoll S. E. Halbesleben J. Neveu J. P. Westman M. (2018). Conservation of resources in the organizational context: The reality of resources and their consequences. Annual Review of Organizational Psychology and Organizational Behavior, 5 , 103-128. 10.1146/annurev-orgpsych-032117-104640 Hofstede G. (1980). Culture and organizations. International Studies of Management & Organization, 10 (4 ), 15-41. 10.1080/00208825.1980.11656300 House R. J. (1971). A path goal theory of leader effectiveness. Administrative Science Quarterly, 16 (3 ), 321-339. 10.2307/2391905 House R. J. (1977). A theory of charismatic leadership. In Hunt J. G. Larson L. L. (Eds.), Leadership: The cutting edge (pp. 189-207). Southern Illinois University. House R. J. Howell J. M. (1992). Personality and charismatic leadership. The Leadership Quarterly, 3 (2 ), 81-108. 10.1016/1048-9843(92)90028-E Hu X. Shi J. (2015). Employees’ surface acting in interactions with leaders and peers. Journal of Organizational Behavior, 36 (8 ), 1132-1152. 10.1002/job.2015 Huffman A. H. Albritton M. D. Matthews R. A. Muse L. A. Howes S. S. (2022). Managing furloughs: How furlough policy and perceptions of fairness impact turnover intentions over time. The International Journal of Human Resource Management, 33 (14 ), 2801-2828. 10.1080/09585192.2021.1879207 Hwang H. Hur W. M. Shin Y. (2021). Emotional exhaustion among the south Korean workforce before and after COVID-19. Psychology and Psychotherapy: Theory, Research and Practice, 94 (2 ), 371-381. 10.1111/papt.12309 Ito J. K. Brotheridge C. M. (2007). Exploring the predictors and consequences of job insecurity’s components. Journal of Managerial Psychology, 22 (1), 40-64. 10.1108/02683940710721938 Jacquart P. Antonakis J. (2015). When does charisma matter for top-level leaders? Effect of attributional ambiguity. Academy of Management Journal, 58 (4 ), 1051-1074. 10.5465/amj.2012.0831 Jiang L. Lavaysse L. M. (2018). Cognitive and affective job insecurity: A meta-analysis and a primary study. Journal of Management, 44 (6 ), 2307-2342. 10.1177/0149206318773853 Judge T. A. Piccolo R. F. Kosalka T. (2009). The bright and dark sides of leader traits: A review and theoretical extension of the leader trait paradigm. The Leadership Quarterly, 20 (6 ), 855-875. 10.1016/j.leaqua.2009.09.004 Kaluza A. J. Boer D. Buengeler C. van Dick R. (2020). Leadership behaviour and leader self-reported well-being: A review, integration and meta-analytic examination. Work & Stress, 34 (1 ), 34-56. 10.1080/02678373.2019.1617369 Kark R. Katz-Navon T. Delegach M. (2015). The dual effects of leading for safety: The mediating role of employee regulatory focus. Journal of Applied Psychology, 100 (5 ), 1332-1348. 10.1037/a0038818 25664472 Katz D. Kahn R. L. (1978). The social psychology of organizations. Wiley. Kerr S. Jermier J. M. (1978). Substitutes for leadership: Their meaning and measurement. Organizational Behavior and Human Performance, 22 (3 ), 375-403. 10.1016/0030-5073(78)90023-5 Kets de Vries M. F. R. (1988). Origins of charisma: Ties that bind the leader and the led. In Conger J. A. Kanungo R. N. (Eds.), Charismatic leadership: The elusive factor in organizational effectiveness (pp. 237-252). Jossey-Bass. Klein G. Eckhaus E. (2017). Sensemaking and sensegiving as predicting organizational crisis. Risk Management, 19 (3 ), 225-244. 10.1057/s41283-017-0019-7 Klein K. A. House R. J. (1995). On fire—charismatic leadership and levels of analysis. Leadership Quarterly, 6 (2), 183-198. 10.1016/1048-9843(95)90034-9 Larsen E. M. Donaldson K. R. Liew M. Mohanty A. (2021). Conspiratorial thinking during COVID-19: The roles of paranoia, delusion-proneness, and intolerance of uncertainty. Frontiers in Psychiatry, 12 , 698147. 10.3389/fpsyt.2021.698147 34483993 Laugesen N. Dugas M. J. Bukowski W. M. (2003). Understanding adolescent worry: The application of a cognitive model. Journal of Abnormal Child Psychology, 31 (1 ), 55-64. 10.1023/A:1021721332181 12597699 LePine M. A. Zhang Y. Crawford E. R. Rich B. L. (2016). Turning their pain to gain: Charismatic leader influence on follower stress appraisal and job performance. Academy of Management Journal, 59 (3 ), 1036-1059. 10.5465/amj.2013.0778 Lin S. H. Scott B. A. Matta F. K. (2019). The dark side of transformational leader behaviors for leaders themselves: A conservation of resources perspective. Academy of Management Journal, 62 (5 ), 1556-1582. 10.5465/amj.2016.1255 Madsen D. Snow P. G. (1991). The charismatic bond: Political behavior in time of crisis. Harvard University Press. Maftei A. Holman A. C. (2022). Beliefs in conspiracy theories, intolerance of uncertainty, and moral disengagement during the coronavirus crisis. Ethics & Behavior, 32 (1 ), 1-11. 10.1080/10508422.2020.1843171 Mandeville A. Whitman M. Halbesleben J. (2019). The meaning of furloughs on family identification. Personnel Review, 48 (6), 1596-1610. 10.1108/PR-07-2018-0245 Marques T. Galende J. Cruz P. Ferreira M. P. (2014). Surviving downsizing and innovative behaviors: A matter of organizational commitment. International Journal of Manpower, 35 (7), 930-955. 10.1108/IJM-03-2012-0049 Maslach C. Schaufeli W. B. Leiter M. P. (2001). Job burnout. Annual Review of Psychology, 52 (1 ), 397-422. 10.1146/annurev.psych.52.1.397 McEvoy P. M. Mahoney A. E. (2012). To be sure, to be sure: Intolerance of uncertainty mediates symptoms of various anxiety disorders and depression. Behavior Therapy, 43 (3 ), 533-545. 10.1016/j.beth.2011.02.007 22697442 McKibbin W. Fernando R. (2021). The global macroeconomic impacts of COVID-19: Seven scenarios. Asian Economic Papers, 20 (2 ), 1-30. 10.1162/asep_a_00796 Meindl J. R. (1995). The romance of leadership as a follower-centric theory: A social constructionist approach. The Leadership Quarterly, 6 (3 ), 329-341. 10.1016/1048-9843(95)90012-8 Meyer J. P. Stanley D. J. Herscovitch L. Topolnytsky L. (2002). Affective, continuance, and normative commitment to the organization: A meta-analysis of antecedents, correlates, and consequences. Journal of Vocational Behavior, 61 (1 ), 20-52. 10.1006/jvbe.2001.1842 Mio J. S. Riggio R. E. Levin S. Reese R. (2005). Presidential leadership and charisma: The effects of metaphor. The Leadership Quarterly, 16 (2 ), 287-294. 10.1016/j.leaqua.2005.01.005 Morrison E. W. Robinson S. L. (1997). When employees feel betrayed: A model of how psychological contract violation develops. Academy of Management Review, 22 (1 ), 226-256. 10.2307/259230 Napier B. J. Ferris G. R. (1993). Distance in organizations. Human Resource Management Review, 3 (4 ), 321-357. 10.1016/1053-4822(93)90004-N Näswall K. De Witte H. (2003). Who feels insecure in Europe? Predicting job insecurity from background variables. Economic and Industrial Democracy, 24 (2 ), 189-215. 10.1177/0143831X03024002003 Oc B. (2018). Contextual leadership: A systematic review of how contextual factors shape leadership and its outcomes. The Leadership Quarterly, 29 (1 ), 218-235. 10.1016/j.leaqua.2017.12.004 Osborn R. N. Hunt J. G. Jauch L. R. (2002). Toward a contextual theory of leadership. The Leadership Quarterly, 13 (6 ), 797-837. 10.1016/S1048-9843(02)00154-6 Østhus S. (2007). For better or worse? Workplace changes and the health and well-being of Norwegian workers. Work, Employment and Society, 21 (4), 731-750. 10.1177/0950017007082881 Parlapani E. Holeva V. Nikopoulou V. A. Sereslis K. Athanasiadou M. Godosidis A. , … Diakogiannis I. (2020). Intolerance of uncertainty and loneliness in older adults during the COVID-19 pandemic. Frontiers in Psychiatry, 11 , 842. 10.3389/fpsyt.2020.00842 32973584 Paulsen N. Callan V. J. Grice T. A. Rooney D. Gallois C. Jones E. , … Bordia P. (2005). Job uncertainty and personal control during downsizing: A comparison of survivors and victims. Human Relations, 58 (4 ), 463-496. 10.1177/0018726705055033 Pierce J. R. Aguinis H. (2013). The too-much-of-a-good-thing effect in management. Journal of Management, 39 (2 ), 313-338. 10.1177/0149206311410060 Podsakoff P. M. MacKenzie S. B. Lee J. Y. Podsakoff N. P. (2003). Common method biases in behavioral research: A critical review of the literature and recommended remedies. Journal of Applied Psychology, 88 (5 ), 879-903. 10.1037/0021-9010.88.5.879 14516251 Preacher K. J. Hayes A. F. (2004). SPSS And SAS procedures for estimating indirect effects in simple mediation models. Behavior Research Methods, Instruments, & Computers, 36 (4 ), 717-731. 10.3758/BF03206553 Probst T. M. Lawler J. (2006). Cultural values as moderators of employee reactions to job insecurity: The role of individualism and collectivism. Applied Psychology: An International Review, 55 (2), 234-254. 10.1111/j.1464-0597.2006.00239.x Purvanova R. K. Muros J. P. (2010). Gender differences in burnout: A meta-analysis. Journal of Vocational Behavior, 77 (2 ), 168-185. 10.1016/j.jvb.2010.04.006 Qin X. DiRenzo M. S. Xu M. Duan Y. (2014). When do emotionally exhausted employees speak up? Exploring the potential curvilinear relationship between emotional exhaustion and voice. Journal of Organizational Behavior, 35 (7 ), 1018-1041. 10.1002/job.1948 Reizer A. Geffen L. Koslowsky M. (2021). Life under the COVID-19 lockdown: On the relationship between intolerance of uncertainty and psychological distress. Psychological Trauma: Theory, Research, Practice, and Policy, 13 (4 ), 432-437. 10.1037/tra0001012 33539162 Restubog S. L. D. Bordia P. Tang R. L. Krebs S. A. (2010). Investigating the moderating effects of leader–member exchange in the psychological contract breach–employee performance relationship: A test of two competing perspectives. British Journal of Management, 21 (2 ), 422-437. 10.1111/j.1467-8551.2009.00673.x Rettie H. Daniels J. (2021). Coping and tolerance of uncertainty: Predictors and mediators of mental health during the COVID-19 pandemic. American Psychologist, 76 (3 ), 427-437. 10.1037/amp0000710 32744841 Richter A. Vander Elst T. De Witte H. (2020). Job insecurity and subsequent actual turnover: Rumination as a valid explanation? Frontiers in Psychology, 11 , 712. 10.3389/fpsyg.2020.00712 32373033 Robinson S. L. Wolfe Morrison E. (2000). The development of psychological contract breach and violation: A longitudinal study. Journal of Organizational Behavior, 21 (5 ), 525-546. 10.1002/1099-1379(200008)21:5<525::AID-JOB40>3.0.CO;2-T Ruiz-Palomino P. Yáñez-Araque B. Jiménez-Estévez P. Gutiérrez-Broncano S. (2022). Can servant leadership prevent hotel employee depression during the COVID-19 pandemic? A mediating and multigroup analysis. Technological Forecasting and Social Change, 174 , 121192. 10.1016/j.techfore.2021.121192 34538969 Satici B. Saricali M. Satici S. A. Griffiths M. D. (2022). Intolerance of uncertainty and mental wellbeing: Serial mediation by rumination and fear of COVID-19. International Journal of Mental Health and Addiction, 20 , 2731-2742. 10.1007/s11469-020-00305-0 32427165 Schyns B. Schilling J. (2013). How bad are the effects of bad leaders? A meta-analysis of destructive leadership and its outcomes. The Leadership Quarterly, 24 (1 ), 138-158. 10.1016/j.leaqua.2012.09.001 Shamir B. (1992). Attribution of influence and charisma to the leader: The romance of leadership revisited 1. Journal of Applied Social Psychology, 22 (5 ), 386-407. 10.1111/j.1559-1816.1992.tb01546.x Shamir B. House R. J. Arthur M. B. (1993). The motivational effects of charismatic leadership: A self-concept based theory. Organization Science, 4 (4 ), 577-594. 10.1287/orsc.4.4.577 Shamir B. Howell J. M. (1999). Organizational and contextual influences on the emergence and effectiveness of charismatic leadership. The Leadership Quarterly, 10 (2), 257-283. 10.1016/S1048-9843(99)00014-4 Shapiro M. O. Short N. A. Morabito D. Schmidt N. B. (2020). Prospective associations between intolerance of uncertainty and psychopathology. Personality and Individual Differences, 166 , 110210. 10.1016/j.paid.2020.110210 Sharma P. N. Kirkman B. L. (2015). Leveraging leaders: A literature review and future lines of inquiry for empowering leadership research. Group & Organization Management, 40 (2 ), 193-237. 10.1177/1059601115574906 Smircich L. Morgan G. (1982). Leadership: The management of meaning. The Journal of Applied Behavioral Science, 18 (3 ), 257-273. 10.1177/002188638201800303 10260212 Stoner J. S. Gallagher V. C. (2010). Who cares? The role of job involvement in psychological contract violation. Journal of Applied Social Psychology, 40 (6 ), 1490-1514. 10.1111/j.1559-1816.2010.00626.x Sumanth J. J. (2011). Be careful what you ask for: How highly inclusive leaders diminish upward communication quality. University of North Carolina at Chapel Hill. Swider B. W. Zimmerman R. D. (2010). Born to burnout: A meta-analytic path model of personality, job burnout, and work outcomes. Journal of Vocational Behavior, 76 (3 ), 487-506. 10.1016/j.jvb.2010.01.003 Sy T. Horton C. Riggio R. (2018). Charismatic leadership: Eliciting and channeling follower emotions. The Leadership Quarterly, 29 (1 ), 58-69. 10.1016/j.leaqua.2017.12.008 Tomprou M. Rousseau D. M. Hansen S. D. (2015). The psychological contracts of violation victims: A post-violation model. Journal of Organizational Behavior, 36 (4), 561-581. 10.1002/job.1997 Uhl-Bien M. Riggio R. E. Lowe K. B. Carsten M. K. (2014). Followership theory: A review and research agenda. The Leadership Quarterly, 25 (1 ), 83-104. 10.1016/j.leaqua.2013.11.007 Van Knippenberg D. Sitkin S. B. (2013). A critical assessment of charismatic—transformational leadership research: Back to the drawing board? Academy of Management Annals, 7 (1 ), 1-60. 10.5465/19416520.2013.759433 Vander Elst T. De Witte H. De Cuyper N. (2014). The job insecurity scale: A psychometric evaluation across five European countries. European Journal of Work and Organizational Psychology, 23 (3 ), 364-380. 10.1080/1359432X.2012.745989 Voitsidis P. Nikopoulou V. A. Holeva V. Parlapani E. Sereslis K. Tsipropoulou V. , … Diakogiannis I. (2021). The mediating role of fear of COVID-19 in the relationship between intolerance of uncertainty and depression. Psychology and Psychotherapy: Theory, Research and Practice, 94 (3 ), 884-893. 10.1111/papt.12315 Waldman D. A. Ramirez G. G. House R. J. Puranam P. (2001). Does leadership matter? CEO leadership attributes and profitability under conditions of perceived environmental uncertainty. Academy of Management Journal, 44 (1 ), 134-143. 10.2307/3069341 Weber M. (1947). The theory of social and economic organization. New York:Free Press. Wegge J. Jungbauer K. L. Shemla M. (2022). When inspiration does not fit the bill: Charismatic leadership reduces performance in a team crisis for followers high in self-direction. Journal of Management & Organization, 28 (6), 1201-1218. 10.1017/jmo.2019.64 Wilk S. L. Moynihan L. M. (2005). Display rule” regulators": The relationship between supervisors and worker emotional exhaustion. Journal of Applied Psychology, 90 (5 ), 917-927. 10.1037/0021-9010.90.5.917 16162064 Williams E. A. Pillai R. McCombs K. M. Lowe K. B. (2021). The roles of character and charisma in evaluating leadership: Effects of dark triad and honesty-humility traits, attributed charisma, and COVID-19 crisis anxiety in rejecting a presidential leadership candidate. Personality and Individual Differences, 181 , 111035. 10.1016/j.paid.2021.111035 Wright T. A. Hobfoll S. E. (2004). Commitment, psychological well-being and job performance: An examination of conservation of resources (COR) theory and job burnout. Journal of Business & Management, 9 (4), 389-406. Wu X. Lin L. Wang J. (2021). When does breach not lead to violation? A dual perspective of psychological contract in hotels in times of crisis. International Journal of Hospitality Management, 95 , 102887. 10.1016/j.ijhm.2021.102887 Yagil D. (1998). Charismatic leadership and organizational hierarchy: Attribution of charisma to close and distant leaders. The Leadership Quarterly, 9 (2 ), 161-176. 10.1016/S1048-9843(98)90003-0 Yukl G. (1999). An evaluation of conceptual weaknesses in transformational and charismatic leadership theories. The Leadership Quarterly, 10 (2 ), 285-305. 10.1016/S1048-9843(99)00013-2
PMC010xxxxxx/PMC10197700.txt
==== Front bioRxiv BIORXIV bioRxiv Cold Spring Harbor Laboratory 37214830 10.1101/2023.05.10.540258 preprint 2 Article Influence of expression and purification protocols on Gα biochemical activity: kinetics of plant and mammalian G protein cycles http://orcid.org/0000-0002-0994-0790 Gookin Timothy E. 12 http://orcid.org/0000-0002-6591-4853 Chakravorty David 12* http://orcid.org/0000-0003-4541-1594 Assmann Sarah M. 1* 1 Biology Department, Pennsylvania State University, University Park, Pennsylvania 16802 2 These authors contributed equally to the article. * Authors for correspondence: Sarah M. Assmann, [email protected], 814-863-9579, David Chakravorty, [email protected], 814-863-9578 08 7 2023 2023.05.10.540258https://creativecommons.org/licenses/by-nc-nd/4.0/ This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which allows reusers to copy and distribute the material in any medium or format in unadapted form only, for noncommercial purposes only, and only so long as attribution is given to the creator. nihpp-2023.05.10.540258.pdf Heterotrimeric G proteins, composed of Gα, Gβ, and Gγ subunits, are a class of signal transduction complexes with broad roles in human health and agriculturally relevant plant physiological and developmental traits. In the classic paradigm, guanine nucleotide binding to the Gα subunit regulates the activation status of the complex. We sought to develop improved methods for heterologous expression and rapid purification of Gα subunits. Using GPA1, the sole canonical Gα subunit of the model plant species, Arabidopsis thaliana, we observed that, compared to conventional purification methods, rapid StrepII-tag mediated purification facilitates isolation of protein with increased GTP binding and hydrolysis activities. This allowed us to identify a potential discrepancy with the reported GTPase activity of GPA1. We also found that human GNAI1 displayed expected binding and hydrolysis activities when purified using our approach, indicating our protocol is applicable to mammalian Gα subunits, potentially including those for which purification of enzymatically active protein has been historically problematic. We then utilized domain swaps of GPA1 and human GNAO1 to demonstrate that the inherent instability of GPA1 is a function of the interaction between the Ras and helical domains. Additionally, we found that GPA1-GNAO1 domain swaps uncouple the instability from the rapid nucleotide binding kinetics displayed by GPA1. Heterotrimeric G protein GTPase Arabidopsis thaliana Signal transduction Recombinant protein expression GPA1 GNAI1 GNAO1 BODIPY GTP ==== Body pmcIntroduction The heterotrimeric G protein complex consists of an alpha (Gα), beta (Gβ), and gamma (Gγ) subunit, in which Gβ and Gγ exist as a non-dissociable dimer. Heterotrimeric G proteins (G proteins) are well-studied conserved eukaryotic signal transduction components. Mutations of G protein subunits in humans have been associated with diseases and developmental abnormalities including cancer (1–3), neurodevelopmental disorders (4), McCune-Albright Syndrome (5), diabetes (6, 7), hypertension (8) and ventricular tachycardia (9). In plants, null mutants of G protein subunits have been utilized to implicate G proteins in agronomically important traits such as morphological development (10, 11), grain shape and yield (12), hormone sensitivity (13, 14), stomatal responses (15–17), salinity tolerance (18), drought tolerance (19) and pathogen resistance (20, 21). The Gα subunit of the G protein heterotrimer binds guanine nucleotides in a binding pocket located within a cleft between the Ras-like and helical domains of the protein. The identity of the nucleotide, GDP or GTP, determines the activation state of the heterotrimer in the canonical signaling paradigm. In the inactive heterotrimer, Gα exists in the GDP-bound form, while stimulation of GTP-binding results in activation of the heterotrimer and dissociation of Gα from the Gβγ dimer. Gα and Gβγ are then able to signal to downstream effectors, until the intrinsic GTPase activity of the Gα subunit hydrolyzes GTP to GDP, thereby stimulating reassociation of the inactive heterotrimer. The activation status of the complex can be regulated by guanine nucleotide exchange factors (GEFs) including 7-transmembrane spanning G protein-coupled receptors (GPCRs) that stimulate GTP-binding, and GTPase activating proteins (GAPs) such as regulator of G protein signaling (RGS) proteins that stimulate GTP hydrolysis (22). In mammals the GPCR superfamily is large and perceives diverse ligands, with over 800 GPCRs encoded in the human genome (23). In contrast, only a few 7TM proteins have been identified as candidate GPCRs in plants (24, 25), including in the intensively studied model dicot Arabidopsis thaliana (26). Instead, receptor-like kinases (RLKs) may predominate in the GPCR role in plants (17, 27–31). Here, we sought to assess the kinetics of the sole canonical Arabidopsis Gα subunit, GPA1, in comparison with two closely related mammalian G proteins in the Gαi family, GNAI1 and GNAO1. GPA1 has previously been described to: i) display self-activating properties due to spontaneous nucleotide exchange and fast GTP-binding, and ii) exhibit slow GTPase activity, which skews the protein to the GTP-bound form, especially when compared to the human Gαi1 protein, GNAI1 (32, 33). However, in the course of our investigations, we found that specific purification protocols and protein storage impair the activity of both plant and mammalian Gα proteins. Purification of active recombinant heterotrimeric G protein Gα subunits is integral to understanding structure-function relationships. Jones et al. (32) determined the structure of GPA1 by x-ray crystallography, discovering that the GPA1 tertiary structure bears a strong resemblance to that of GNAI1. Yet, GPA1 and GNAI1 also display distinctly different enzymatic activities, indicating that differences arise on a finer scale, possibly due to high levels of intrinsic disorder within, and dynamic motion of, the GPA1 helical domain (32). Therefore, we developed a robust expression and rapid purification protocol utilizing dual StrepII-tags (34), which allowed for elution in a buffer directly compatible with downstream BODIPY-GTP/-GDP binding assays, thereby abrogating any need for protracted buffer component removal, e.g. by dialysis. GPA1 exhibits increased activity when rapidly purified using a specific purification protocol. Furthermore, comparison with the two human Gα subunits from the Gαi family demonstrated that mammalian Gα activity is also impacted by the choice of purification regime, and that our StrepII-tag approach to expression and purification is applicable to mammalian Gα subunits. Results Rapid purification of recombinant GPA1 yields higher activity in vitro The dual StrepII tag consists of tandem StrepII tags separated by a flexible linker. The original Strep tag was identified as a streptavidin binding tag that could be used to isolate recombinantly expressed antibodies (35). Both the Strep tag as well as streptavidin were further engineered to form a StrepII-Streptactin system with increased affinity, and the ability for N-terminal and C-terminal tagging (36, 37). As the resin conjugated Streptactin used to purify StrepII-tagged proteins exists in a tetrameric state, the use of two tandem StrepII tags separated by a linker was subsequently employed to further increase binding affinity via the avidity effect while still allowing efficient competitive elution (36) in a buffer compatible with many in vitro assays. We therefore utilized dual N-terminal StrepII tags separated by a linker sequence (SGGSGTSGGSA), similar to the linker used in the Twin-Strep-tag (GGGSGGGSGGSA) (36), to purify GPA1, the sole canonical Gα subunit from Arabidopsis. We adapted the base pGEX vector backbone to include N-terminal dual-StrepII tags, thrombin and TEV protease sites, a multiple cloning site, and the option of a C-terminal FLAG tag in a new vector we named pSTTa. GPA1 expressed from pSTTa was purified and observed to be highly pure (Fig. S1A). In a classic in vitro assay, fluorescent signal increases as Gα proteins bind BODIPY-conjugated GTP and decreases when GTP hydrolysis outpaces GTP binding, due to resultant partial fluorophore re-quenching (38–43). After numerous trials of multiple growth and induction protocols, we found that use of BL21 (DE3) cells cultured in high salt LB (HSLB) media resulted in the highest yield of StrepII-GPA1, so we utilized HSLB for growth and induction in all subsequent experiments. We found that freshly purified GPA1 exhibits a characteristic BODIPY-GTP kinetic curve indicative of rapid GTP binding and rapid GTP hydrolysis but GPA1 assayed after overnight storage at 4 °C, a proxy for standard buffer dialysis protocols, shows reduced GTP binding with slower kinetics (Fig. 1A). To circumvent the detrimental overnight dialysis step, we prepared His-GPA1 and GST-GPA1 fusion proteins fresh and as rapidly as possible, utilizing 10 kDa molecular weight cut-off centrifugal filter units for post-elution buffer exchange into “EB Base”, the buffer used for StrepII-GPA1 elution but lacking desthiobiotin. When purified side-by-side with StrepII-GPA1, the buffer exchange steps applied to the His-GPA1 and GST-GPA1 proteins added approximately 45 minutes of additional handling, which was performed on ice and in a 4 °C refrigerated centrifuge, while the StrepII-GPA1 sample was kept on ice. When separated by SDS-PAGE, the StrepII-GPA1 protein was obviously more pure than the His-GPA1 or GST-GPA1 preparations under these rapid purification conditions (Fig. S1B). Despite comparable affinity purification protocols and rapid handling at cold temperatures for the post-purification buffer exchange steps, the His-GPA1 and GST-GPA1 proteins displayed lower apparent binding and hydrolysis activities than the StrepII-GPA1 protein (Fig. 1B). As peak fluorescence of BODIPY-GTP is a function of net binding and hydrolysis, another interpretation for the lower activity observed for His- and GST- fusions in Figure 1B is that GTP binding rate is unaltered between the proteins, but GTP hydrolysis is faster in the His- and GST-fusions. We therefore assayed binding to the non-hydrolyzable BODIPY-GTPγS using rapid sampling in well mode for 30 seconds to assess the initial relative GTPγS-binding rates. Indeed the StrepII-GPA1 protein displayed appreciably faster BODIPY-GTPγS binding compared to the His-GPA and GST-GPA1 preparations (Fig. 1C), consistent with the interpretation that the StrepII-GPA1 preparation displays higher activity. One possible explanation for the increased apparent activity of the StrepII-GPA1 protein compared to His-GPA1 and GST-GPA1 (Fig. 1B) is that a co-purifying contaminant in the StrepII-GPA1 preparations displays GTP-binding and hydrolysis activities. Though the StrepII purifications commonly yield highly pure protein, we almost always observed a minor contaminating band slightly larger than 70 kDa in our GPA1 elutions (Fig. S1A). To verify the lack of GTP binding by any co-purifying contaminant, we expressed and purified a GPA1S52N mutant that, as observed previously for a GPA1S52C mutant (41), does not bind GTP: StrepII-GPA1S52N displayed no BODIPY-GTP or BODIPY-GTPγS-binding activity (Fig. S1C and D), confirming that the binding and hydrolysis observed for StrepII-GPA1 was solely a result of GPA1 activity. Mass spectrometric identification was performed on the 70+ kDa protein and it was found to correspond to E. coli DnaK, which is not expected to display GTP binding activity (44). GPA1 stability To investigate the underlying cause of in vitro GPA1 activity loss, we assessed GPA1 conformational stability utilizing SYPRO Orange fluorescence. SYPRO Orange fluorescence increases upon interaction with hydrophobic regions of the protein, which will have greater accessibility upon protein unfolding. We observed that at our standard BODIPY-GTP binding assay temperature of 25 °C, GPA1 protein displayed a steady increase in SYPRO Orange fluorescence over the course of 30 min, indicative of protein unfolding (Fig. 1, D and E). Therefore, for future assays we stored GPA1 and other Gα proteins for matched assays on ice in a 4 °C refrigerator during post-elution quantification steps, prepared reaction mixes on ice, and allowed no more than 2 minutes of temperature equilibration for samples immediately prior to initiating assays, to mitigate loss of activity. Incubation of GPA1 with excess unlabelled GDP (Fig. 1D) or GTPγS (Fig. 1E) restored protein stability while allowing for nucleotide exchange assessment, yet GDP inclusion also appears to compete with BODIPY-GTP/-GTPγS for binding under reaction conditions. For example, we found that inclusion of GDP in the binding buffer allowed for pre-loading of GPA1 with GDP, presumable short-term protein stabilization, and increased BODIPY-GTPγS binding activity (Fig. S1E). By contrast, inclusion of GDP in the binding and elution buffers resulted in protein with decreased BODIPY-GTPγS binding (Fig. S1E); presumably the lower apparent binding equilibrium is due to an excess of unlabeled GDP competing with BODIPY-GTPγS for binding to GPA1. Given the stability issues of GPA1 outlined above, we wondered if storage at −80 °C and freeze-thawing also has an impact on GPA1 activity. Therefore we purified StrepII-GPA1 and added either an additional 10% glycerol or 8.33% sucrose as a cryoprotectant (45), snap froze the protein with liquid N2 and stored the proteins for three weeks at −80 °C. Upon thawing and assaying the protein we observed that GPA1 frozen with sucrose as a cryoprotectant displayed more rapid BODIPY-GTP binding and faster hydrolysis than GPA1 frozen with glycerol (Fig. 1F). This difference suggests the standard use of glycerol as a cryoprotectant is suboptimal for storage of GPA1. In fact our data indicate the use of glycerol alone as a cryoprotectant could result in an underestimate of the peak net hydrolysis rate of GPA1 by ~55%. The BODIPY-GTP binding and hydrolysis curves of GPA1 frozen with sucrose (Fig. 1F) were similar to freshly prepared GPA1, e.g. as observed in Figure 1A. Therefore, storage supplemented with 8.33% sucrose is a viable alternative approach for GPA1 provided that all proteins to be compared are handled equivalently. The remaining data presented in this paper are results from freshly isolated GPA1, GNAI1 and GNAO1 proteins. Purification of the RGS1 cytosolic domain Arabidopsis RGS1 encodes a protein with seven transmembrane spanning domains at the N-terminus and a cytosolic RGS domain at the C-terminus. Previous assays of RGS1 activity have therefore utilized only the cytosolic RGS domain (46, 47). We recombinantly produced the RGS1 cytosolic domain utilizing the pSTTa vector to confirm: i) that RGS1 is amenable to rapid purification via dual StrepII-tags, and more importantly; ii) that the StrepII-tagging approach for GPA1 did not disrupt GPA1 interaction with the primary regulator, RGS1. The addition of StrepII-RGS1 to StrepII-GPA1 in a BODIPY-GTP assay strongly promoted the hydrolysis of GTP by GPA1 (Fig. S1F), indicating both i and ii are true. Comparison of GPA1 to human GNAI1/GNAO1 Arabidopsis GPA1 bears high structural similarity to human GNAI1, which has provided a rationale for previous biochemical comparisons of GPA1 with GNAI1 (32). We therefore sought to reexamine this comparison utilizing our newly optimized recombinant GPA1 purification protocol. We cloned GNAI1 into pSTTa using both codon harmonized (GNAI1ch) and wild-type (GNAI1wt) sequence (Fig. S2, A and B). We found that proteins derived from the two constructs were essentially interchangeable as side-by-side comparisons showed the GNAI1wt and GNAI1ch proteins did not differ in yield or purity (Fig. S2C), in BODIPY-GTP binding and hydrolysis (Fig. S2D), or in BODIPY-GTPγS binding (Fig. S2E). For a human Gα contrast with GNAI1, we prepared a construct for another human Gαi subfamily member, GNAO1, which has been shown to display considerably faster kinetics than GNAI1 (42). On a sequence level, the GNAI1 protein shares 38.2% identity and 56.8% similarity with GPA1, while GNAO1 displays 37.0% identity and 54.1% similarity with GPA1. Therefore, it is reasonable to compare GPA1 to both of these mammalian Gαi proteins. We purified StrepII-GNAI1 and StrepII-GNAO1, and compared their wild-type activities to those of their constitutively active mutants (StrepII-GNAI1Q204L/GNAO1Q205L (48)), and to activities of mutants corresponding to the plant nucleotide-free GPA1S52N mutant (StrepII-GNAI1S47N/StrepII-GNAO1S47N). Both wild-type GNAI1 and GNAO1 proteins displayed GTP-binding and hydrolysis activities, and as expected based on previous studies (42), the net GTP binding activity of GNAI1, reflected by the amplitude of peak fluorescence, was considerably lower than that of GNAO1 (Fig. 2, A and B). The constitutively active mutants (Q204L/Q205L) displayed slower binding than the wild-type proteins and no hydrolysis activity, as expected, while the S47N mutants displayed no BODIPY-GTP binding activity (Fig. 2, A and B). Surprisingly, the S47 mutants both displayed BODIPY-GTPγS binding activity that occurred faster than was observed for the wild-type GNAI1 and GNAO1 proteins (Fig. 2, C and D). The binding activity was, however, transient with a peak observed at 3–4 minutes, followed by a steady decline in signal. The decline is unlikely to be due to hydrolysis as GTPγS is considered non-hydrolysable, and no evidence of BODIPY-GTPγS hydrolysis was evident in any of our assays with wild-type GNAI1 or GNAO1. These results are in contrast to the analogous GPA1S52N mutant, which displayed no BODIPY-GTPγS binding (Fig. S1D). The S47 residue within the G1 motif is important for Mg2+ cofactor coordination (49) and since other metal ions are known to inhibit Gα nucleotide binding (50), we routinely utilized trace-metal-free (TMF) grade components to standardize our assays, which explicitly ruled out any effect of extraneously-present divalent ions, including Mg2+. Notably, we also show that TMF components are not necessary for basic assays, and our methodology can be performed using standard grade reagents (Fig. S2F). The GNAI1S47N and GNAO1S47N mutants retain some ability to bind BODIPY-GTPγS, unlike GPA1S52N; yet, without proper coordination of the Mg2+ cofactor, this binding is transient (Fig. 2, C and D). Given that the BODIPY fluorophore is covalently attached differently in BODIPY-GTP (ribose ring) and BODIPY-GTPγS (γ-phosphate), inconsistencies in binding results between the two BODIPY reagents could arise from a combination of steric differences of the binding pocket between mutants and the respective locations of the BODIPY fluorophore. To check if the S47N mutants do retain some residual Mg2+ binding, we performed BODIPY-GTPγS binding assays ±Mg2+. In our ±Mg2+ assay, both wild-type GNAI1 and GNAO1 displayed a clear requirement for Mg2+, with a very low level of BODIPY-GTPγS binding activity observed in the absence of Mg2+ (Fig. 3, A and B). Similarly, in the S47N mutants the more rapid but transient binding of BODIPY-GTPγS was only observed in the presence of Mg2+ (Fig. 3, A and B), confirming a requirement for Mg2+ for in guanine nucleotide coordination. We then investigated protein instability as the potential underlying cause for the transient BODIPY-GTPγS binding by both StrepII-GNAI1S47N and StrepII-GNAO1S47N. We found that in the presence of excess GTPγS, GNAI1 and GNAI1S47N exhibited similar protein stabilities (Fig. 3C) as determined by SYPRO Orange, a fluorescent indicator of protein unfolding, yet GNAI1S47N unfolding was singularly increased during the timecourse in the absence of GTPγS (Fig. 3C). These results indicate protein instability potentially contributes to the loss of activity by GNAI1S47N over time, yet the analogous assay comparing GNAO1 to GNAO1S47N did not directly support this hypothesis. GNAO1S47N did not exhibit appreciably more protein unfolding over the timecourse (Fig. 3D), with wild-type and mutant displaying similarly shaped curves. While protein instability could not explain the loss of activity for GNAO1S47N, the mutant protein did display a different basal level of SYPRO Orange fluorescence, even in the presence of additional GTPγS (Fig. 3D). This difference between the wild-type and mutant GNAO1 may indicate a difference in protein conformation, which could be reflected in the different abilities of GNAO1S47N to bind BODIPY-GTP vs. BODIPY-GTPγS (Fig. 2B vs. 2D). If the same phenomenon explains the differential binding of GNAI1S47N to BODIPY-GTP vs. BODIPY-GTPγS (Fig. 2A vs. 2C), the effect must be more local to the binding pocket and not reflected in the basal SYPRO Orange signal corresponding to the entire surface of the protein (Fig. 3C). Taken together, these data suggest GNAI1S47N and GNAO1S47N do retain some affinity for Mg2+ and a requirement for this cofactor in nucleotide coordination. The transient nature of the BODIPY-GTPγS binding may reflect transient Mg2+-binding rather than protein instability. Next we compared the binding activities of StrepII-GPA1 to StrepII-GNAI1 and StrepII-GNAO1. StrepII-GNAI1 displayed a much lower apparent BODIPY-GTP binding peak than StrepII-GPA1, while StrepII-GNAO1 displayed an intermediate activity (Fig. 4A). As peak fluorescence reflects a net activity of GTP binding and hydrolysis, these initial results were consistent with GPA1 displaying a faster binding and slower hydrolysis rate than GNAI1, however, interpretation of the comparison to GNAO1 was less clear. As the StrepII purification protocol was superior to His purification for GPA1, we characterized StrepII-tagged GNAI1 and GNAO1 in comparison to the commonly used His-tagged GNAI1 and GNAO1. We found optimal tags and purifications differed not just between human Gα subunits and Arabidopsis GPA1, but also between the human Gα subunits. Minimal differences in activity were observed between His-GNAI1 and StrepII-GNAI1 (Fig. 4B), Indicating that StrepII or His purification is suitable for GNAI1. By contrast, His-GNAO1 displayed higher net BODIPY-GTP binding and hydrolysis activities than StrepII-GNAO1 (Fig. 4C), and the difference was just as clear for binding of BODIPY-GTPγS (Fig. 4D), indicating the His purification protocol is the more suitable method to assay GNAO1 activity. Given the above results, we performed side-by-side purifications of StrepII-GPA1 and His-GNAO1, which demonstrated that GPA1 does indeed display a faster GTP binding rate than GNAO1, but the net hydrolysis rates appear to not to be as different between plant and human Gα subunits as previously thought (Fig. 4E). To isolate the observed binding rate of the Gα proteins, we performed assays with the non-hydrolyzable GTP analog BODIPY-GTPγS. Indeed the initial rate of BODIPY-GTPγS binding to GPA1 was more rapid than to GNAO1, yet the GPA1 maximal binding signal was unexpectedly much lower than GNAO1, and rather than plateau as with GNAO1, the GPA1 BODIPY-GTPγS signal decreased over time (Fig. 4F). Possible reasons for the difference in signal maxima include: i) steric differences of the Gα binding pockets resulting in differential levels of BODIPY fluorophore unquenching upon protein binding, and; ii) inherent instability of GPA1 resulting in a lower apparent binding activity in vitro. We believed hypothesis i) was unlikely as the empirically derived crystal structures of GPA1 and GNAO1 are highly similar (Fig. S3A), just as are the structures of GPA1 and GNAI1 (Fig. S3B). We therefore sought to assess the amount of each enzyme necessary to observe saturated and stable binding of 50 nM BODIPY-GTPγS, as a reflection of Gα activity retained in vitro. Despite being in excess, 100 nM, 200 nM and 400 nM concentrations of StrepII-GPA1 were unable to attain a maximal binding signal with 50 nM BODIPY-GTPγS. Only 800 nM or 1.2 μM GPA1 displayed a stable binding plateau at the maximal level (Fig. 5A). In comparison, all concentrations of GNAO1 either attained a maximal plateau, or neared maximal fluorescence in the case of 100 nM GNAO1, within the course of our assay (Fig. 5B). The necessity for higher GPA1 concentrations in reaching binding saturation reflects the established lower stability of GPA1 in vitro (Fig. 1, D and E), but also provides insight regarding the GNAO1>GPA1 signal maxima in Figure 4F. Notably, the maximal levels of BODIPY-GTPγS fluorescence were quite similar between high concentrations of GPA1 and GNAO1 when assayed side-by-side (Fig. 5, A and B), thereby refuting hypothesis i) above by indicating that steric differences in the binding pockets do not result in different levels of BODIPY fluorophore unquenching. Next we compared the ability of excess (10 μM) GDP to suppress binding of 50 nM BODIPY-GTP to 100 nM Gα proteins. BODIPY-GTP binding was partially suppressed by 10 μM GDP for GPA1, but almost completely abolished for GNAO1 (Fig. 5C). The striking difference in GDP suppression of GTP binding likely reflects a higher relative affinity for GTP than GDP and significantly faster nucleotide exchange rate of GPA1 than GNAO1. GPA1-GNAO1 helical domain swaps It has been proposed that the helical domain of GPA1 displays a marked level of intrinsic disorder and increased dynamic motion compared to that of GNAI1 (32). Jones et al. (32) confirmed that a helical domain swap between GPA1 and GNAI1 largely swapped the relative kinetics between the two Gα proteins. In those studies, the GPA1 helical domain conferred rapid spontaneous activation to GNAI1 while the GNAI1 helical domain conferred slower activation to GPA1. We wondered if a helical domain swap between GPA1 and GNAO1 would: i) display as strong a difference as the GPA1-GNAI1 domain swap, and ii) confirm that the helical domain of GPA1 is responsible for the poor stability of GPA1. Therefore, we created our reciprocal domain swap constructs GPA1GNAO1hel (GPA1 Ras domain fused to the GNAO1 helical domain) and GNAO1GPA1hel (GNAO1 Ras domain fused to the GPA1 helical domain). To not confound any tag/purification effects with the domain swap effects, we utilized our StrepII tagging and purification methods for all proteins. A comparison of BODIPY-GTPγS binding demonstrated that binding rates increased in the following order: GNAO1<GPA1GNAO1hel<GPA1<GNAO1GPA1hel. Beyond this initial binding rate, GPA1 displayed the lowest signal amplitude corresponding to peak binding, while GNAO1GPA1hel displayed the highest signal plateau (Fig. 6A). In BODIPY-GTP assays, which integrate GTP binding and hydrolysis, a similar initial trend was largely displayed during the binding phase, GNAO1<GPA1GNAO1hel=GNAO1GPA1hel<GPA1 (Fig. 6B). Once BODIPY-GTP hydrolysis exceeded the binding rate, we observed the following order of maximal net hydrolysis rates: GNAO1<GPA1GNAO1hel<GNAO1GPA1hel<GPA1 (Fig. 6B). Therefore, although the rate of GTPγS binding by GNAO1GPA1hel was the fastest of the four proteins assayed, peak BODIPY-GTP fluorescent signal was dampened by a rapid switch to net hydrolysis. We then assessed the relative conformational stability of the GPA1, GNAO1, GPA1GNAO1hel and the GNAO1GPA1hel proteins in a SYPRO Orange assay ±10 μM GTPγS (no BODIPY label). As suspected, the GNAO1 helical domain conferred a similar stability to GPA1 as did excess GTPγS (Fig. 6, C and D). As before, GPA1 samples without nucleotide supplementation displayed increased SYPRO Orange signal indicative of protein unfolding, and GPA1 was the only protein in the domain swap assays to display considerable divergence between the ±GTPγS samples (Fig. 6, C and D). Interestingly, at “T=0” of the SYPRO Orange assay the fluorescence of GPA1 in the absence of nucleotide supplementation was already much higher than that of most other samples. We note that all of these samples were prepared on ice in duplicate, pipetted into the assay plate and loaded into the plate reader; a process that took ~4 minutes for the number of samples in Figures 6C and 6D. To investigate the difference at “T=0” of the assays comparing multiple samples, we performed a 1 vs. 1 assay comparing single wells of GPA1 vs. GPA1 +10 μM GDP. This assay can be initiated in seconds and allowed us to monitor SYPRO Orange fluorescence almost immediately after removal from ice. Indeed, in this rapid assay, the initial fluorescence levels were similar between the samples before a steady rise in fluorescence signal was observed in the GPA1 alone reaction (Fig. S3C). The initial similarity of fluorescence between ±nucleotide samples was quite similar to the results shown in Figures 1D and 1E, which were assays run on an intermediate scale compared to the large assay in Figures 6C and 6D, and small assay in Figure S3C. GNAO1GPA1hel in the large scale assay also displayed a higher initial value of SYPRO Orange fluorescence, and noticeably more signal variation between timepoints, though it should be noted that the noise-like variation was not always observed for GNAO1GPA1hel (Fig. S3D). Unlike GPA1, the addition of GTPγS did not repress the T=0 high fluorescence values for GNAO1GPA1hel, yet the fluorescence signals for GNAO1GPA1hel did not rise as markedly through the assay as they did for GPA1 in the absence of GTPγS (Fig. 6, C and D). These traits appear consistent with GNAO1GPA1hel achieving a stable but different conformation than the other Gα proteins assayed in Figure 6; a conformation that is seemingly characterized by increased surface accessibility of hydrophobic residues for SYPRO Orange binding but not increased instability. In summary, for GPA1 in vitro, unfolding at room temperature and then at 25 °C began almost immediately and was evident on a scale of seconds to minutes, further underscoring the need to use a rapid purification protocol. Interestingly, our domain swap assays indicate that neither the Ras nor the helical domain alone accounts for the lack of stability. Discussion Purification of functional recombinant heterotrimeric G protein Gα subunits is integral to understanding their roles in both animals and plants. The former is of importance due to their well-described functions in human health (51), and the latter is important due to G protein involvement in controlling agriculturally important traits (12). We demonstrate for the Arabidopsis Gα subunit, GPA1, that protracted handling and/or storage using the standard protocol of glycerol as a cryoprotectant are detrimental to isolating optimally functional protein (Fig. 1). Therefore we developed a StrepII-tag purification protocol that allowed rapid on-column binding to isolate highly pure protein for immediate downstream analyses. The utilization of an elution buffer compatible with downstream assays, abolishing the need for buffer exchange steps, is a major advantage of the StrepII purification protocol. Even with the rapid StrepII purification protocol, our data were consistent with some loss of GPA1 activity during the purification and assay timeframe, based on the high concentrations of GPA1 required to saturate binding of 50 nM BODIPY-GTPγS, and the BODIPY signal rundown observed at lower concentrations of GPA1 (Fig. 5A, compared to GNAO1 in Fig. 5B). Nonetheless, matched purifications demonstrated that the loss of activity for StrepII-GPA1 was substantially less than that observed for commonly used tags: His-GPA1 or GST-GPA1 (Fig. 1B). It should be noted that inclusion of GDP in the binding buffer can lead to greater stabilization of GPA1 activity (Fig. S1E), presumably by preloading the protein with GDP during lysis and column-binding. Yet, inclusion of GDP in elution or storage fractions is not optimal when assaying intrinsic binding affinity as, at least in the case of GPA1, excess concentrations of GDP can compete with GTP for binding to Gα (Fig. S1E) and introduce a confounding nucleotide release step. As our method proved to be an improvement over existing protocols for GPA1 purification (Fig. 1B), we applied it to the purification of two closely related human Gα subunits, GNAI1 and GNAO1. We show our method is also applicable to human Gα subunit expression and purification such as for GNAI1 (Fig. 4B). We therefore establish StrepII-mediated purification as an addition to the toolkit of possibilities for recombinant investigation of G proteins. However, His-GNAO1 outperformed StrepII-GNAO1 in our hands (Fig. 4C), reinforcing that tag choice is not universal, and should be optimized for each protein of interest. We then utilized our newly improved purification protocol to address the following four questions of interest. 1) Does GPA1 indeed display self-activating properties? 2) Is the balance of GTP-loading of GPA1 further skewed to the active state by slow GTP hydrolysis? 3) What are the functional consequences of mutations to the serine residue important for Mg2+ ion coordination in the active site? 4) Given that GNAO1 displays rapid enzyme kinetics compared to GNAI1, but without the loss of stability observed in GPA1, can we employ a domain swap approach between GPA1 and GNAO1 to assess the relative contributions of the Ras vs. helical domains to enzyme function and stability? As GPA1 was sensitive to differences in handling, in all assays including GPA1 we only directly compared proteins prepared fresh side-by-side, and we recommend that as the best practice. Re-evaluation of GPA1 enzymatic kinetics Jones et al. (32) characterized GPA1 as a self-activating Gα protein due to rapid GDP release followed by rapid GTP binding. They also reported slow GTP hydrolysis kinetics. Urano et al. (33) followed this study with confirmation that Gα subunits from evolutionarily distant branches of the plant kingdom also exhibit these properties. However, both studies utilized His-tag purification protocols; Jones et al. purified Gα proteins using a 90 min batch binding step with post-elution processing steps and compared GPA1 to the slow GTP-binding Gα, GNAI1, and Urano et al. stored purified Gα subunits at −80 °C with glycerol as a cryoprotectant. Though these are standard protocols for Gα purification, with hindsight we suggest these steps are not optimal for isolation of active GPA1. It should also be noted that both studies included GDP in their elution buffers, which assists in GPA1 stabilization (Figs. 1D and S1E) but, depending on concentration, can slow GTP binding and therefore the maximal observable hydrolysis rate (Fig. 5C). As a result we sought to reassess the conclusions drawn from these studies by utilizing our newly developed rapid Gα purification protocol. All of our data are consistent with the original assertion that GPA1 displays rapid GTP binding (Figs. 1A–C and S1C–F). Even when compared to GNAO1, which displays much faster binding than GNAI1 (Fig. 2), GPA1 clearly displays a faster comparative rate of GTP binding (Fig. 4, E and F). Furthermore, our analyses of GPA1 stability in vitro (Figs. 1D–E, 6C–D and S3C), and the inability of moderately excess concentrations of GPA1 to saturate BODIPY-GTPγS-binding (Fig. 5A), suggests that this assessment of GTP binding is still an underestimate due to functional decline under assay conditions. We also found that storage of GPA1 with glycerol as the only cryoprotectant resulted in an underestimation of GTPase activity (Fig. 1F). Specifically, the peak net hydrolysis rate was 55% lower for samples stored with glycerol compared to samples stored with sucrose, which would certainly skew the extent to which GPA1 would be estimated to be GTP loaded. Additionally, the comparison to GNAO1 reveals GPA1 to be less of an outlier than previous (32) and current comparisons to GNAI1 would suggest. The apparent peak net hydrolysis rate of GPA1 is only 12.5% higher than that of GNAO1, as determined by BODIPY-GTP signal decreases across a 30 second moving window in Figure 4E, indicating relatively similar levels of activity. With regard to spontaneous nucleotide exchange, our data in Figure 5C are particularly compelling. In that GDP competition assay, GDP was provided at 10 μM, i.e. 100x in molar excess of the Gα proteins and 200x in molar excess of BODIPY-GTP. This massive overabundance of GDP was sufficient to completely outcompete BODIPY-GTP binding by GNAO1 (Fig. 5C), reflecting the crucial role of GPCR-mediated stimulation in nucleotide exchange for animal Gα subunits (52, 53). Contrastingly, 10 μM GDP was only partially able to suppress GPA1 BODIPY-GTP binding activity (Fig. 5C), consistent with GPA1 displaying a spontaneous nucleotide exchange activity and relatively much higher affinity for GTP than GDP, as previously reported (32, 33, 54). Overall, our data indicate that GPA1 does display rapid properties of both nucleotide exchange and GTP binding, but there likely has been underestimation of the GTP hydrolysis activity of GPA1 in the past due to choice of purification protocol. Side-by-side comparisons with two closely related mammalian Gα proteins, all isolated under optimal conditions (Fig. 4, A and E), reveals that the GTP hydrolysis rate of GPA1 falls within the range of that observed for these animal Gα subunits. GNAI1S47N and GNAO1S47N mutants display transient GTP binding With the advent of affordable mass patient genetic testing, a number of mutations of the equivalent sites to GNAI1S47/GNAO1S47 and GNAI1Q204/GNAO1Q205 of multiple Gα subunits have been associated with various medical conditions in ClinVar (55) and Catalogue Of Somatic Mutations In Cancer (COSMIC) (56) databases, as summarized in Tables S1 and S2, respectively. The Q204/Q205 site resides within the G3 motif (one of five G box motifs important for nucleotide binding) of Gα subunits, mutations of which are well-known to impart a constitutively active status upon Gα proteins (48). Mutations at this site specifically in GNAQ and GNA11 are strongly linked to uveal melanoma (1, 57). The S47 site is relatively less well-understood, though it is a crucial residue within the G1 motif involved in Mg2+ cofactor coordination (49). Mutants of S47 and equivalent sites in G proteins have been used as tools of functional investigation before disease associations were identified for the site. For example, an equivalent mutant to S47N in the small monomeric G protein Ras, S17N, was characterized as displaying a 23,000-fold reduction in affinity for GTP (58). Subsequently, a S47C mutation was identified in a random mutagenesis screen of GNAO1 as a protein with low to no GTPγS binding activity (59). In other examples, a GαT protein in which a region or subregions of amino acids 215–295 have been replaced with the equivalent GNAI1 residues to facilitate expression and purification, has been utilized and named GαT*. When assaying binding of radiolabeled GTPγS by GαT* chimeric proteins, there was an apparent discrepancy between the results of Natochin et al. (60) who reported the S43N and S43C mutants failed to bind GTP, and Ramachandran and Cerione (61) who reported a faster rate of spontaneous GDP-GTPγS exchange for the GαT*S43N mutant compared to GαT*. Our reassessment with real-time BODIPY-GTPγS binding suggests a mechanism by which the discrepancy may be understood. Figures 2C and 2D indicate the initial rate of BODIPY-GTPγS binding is faster for GNAI1S47N and GNAO1S47N than the respective wild-type proteins, while Figures 3A and 3B demonstrate this rapid binding is Mg2+-dependent. However, the binding is only transient, as shown by the observation that BODIPY-GTPγS signal initially increased, but then gradually decreased 3–4 minutes after binding initiation (Fig. 2, C and D), a phenomenon that was not caused by protein instability (Fig. 3, C and D). Thus, the binding signal could be missed and/or washed off if the protein is subjected to protracted handling in a radiolabeled GTPγS binding assay, which may account for the previously reported GTP-binding discrepancy. Our results provide additional insight into the mechanism by which S47 and equivalent position mutations of human Gα subunits manifest in disease states. Moreover, our results suggest an advantage of BODIPY assays in facile revelation of real-time kinetics. GPA1 instability is conferred by combined effects of the Ras and helical domains, and is not inherently linked to rapid nucleotide binding As mentioned above, studies from the Jones and Dohlman groups have indicated that the GPA1 helical domain displays high levels of intrinsic disorder based on comparisons of the electron density map and atomic displacement parameters of monomers determined by x-ray crystallography, and motion away from the Ras domain as predicted by molecular dynamics simulations (32, 62). Interdomain motion is a mechanism proposed to potentiate nucleotide exchange (52, 53, 63, 64) and therefore these observations for GPA1 are consistent with its status as a Gα subunit capable of spontaneous nucleotide exchange (32). As previously established, a domain substitution using the helical domain of GNAI1 conferred slower nucleotide exchange, faster GTP hydrolysis and increased stability to GPA1. Those stability experiments utilized circular dichroism over a temperature gradient of 15–80 °C, and proteins were assayed in the presence of excess GDP. We however observed using a SYPRO Orange fluorescence assay, that when incubated at 25 °C in the absence of additional nucleotides, GPA1 displayed reduced stability (Figs. 1, D and E and S3C). We also observed the enzymatic differences between GPA1 and GNAO1 were less than those between GPA1 and GNAI1 (Fig. 4), though GNAO1 was likely more stable than GPA1 based on the plateau in BODIPY-GTPγS binding signal observed in Figure 5B. It was intriguing to speculate that the helical domain of GNAO1 may confer stability to GPA1 while also allowing the fast GTP binding kinetics of GPA1 to be retained. Indeed this proved to be the case with GPA1GNAO1hel displaying almost as rapid BODIPY-GTPγS and BODIPY-GTP binding as GPA1 (Fig. 6 A and B). When protein stability was assayed, we observed that GPA1GNAO1hel displayed a similar resistance to unfolding at 25 °C as GNAO1, distinguishing it from the less stable GPA1 protein (Fig. 6C). When provided with a molar excess of GTPγS, GPA1 was as stable as GNAO1 and the chimeric Gα subunits (Fig. 6D). In the reciprocal domain swap, GNAO1GPA1hel displayed rapid BODIPY-GTPγS binding (Fig. 6A) and fast hydrolysis (Fig. 6B). Unexpectedly, GNAO1GPA1hel exhibited a higher basal level of SYPRO Orange interaction than the other Gα proteins, but unlike GPA1, this dye binding by GNAO1GPA1hel did not increase with time, indicating a relatively higher protein stability (Fig. 6, C and D). As the GNAO1GPA1hel protein also displays strong enzymatic activity (Fig 6, A and B), we conclude that the protein is not unfolded, but more likely resides in a stable but alternative conformation to the other Gα proteins assayed. Therefore, as the GPA1 helical domain did not confer instability to GNAO1, we conclude that GPA1 instability is a result of interdomain forces, and that rapid kinetics and instability can be uncoupled by the use of chimeric domain swaps. Future directions Our results suggest the need for further evaluation of the GTPase activity of GPA1 in comparison to mammalian Gα subunits. Here we used BODIPY-GTP/-GTPγS to test our newly developed purification approach for GPA1, and screen relative G protein activities. We report our purification method as a tool for the community and highlight important contrasts to data from established methods, as well as point to several general consistencies between our data and those of others. We also illustrate an advantage for BODIPY-GTP/GTPγS as it is a real-time method for measurement of direct binding with a sampling rate and processing speed that cannot be matched by traditional radiolabeled nucleotide approaches. These aspects are particularly useful for proteins with rapid kinetics and low stability in vitro. However, we also observed a drawback of the BODIPY labeling approach in the inconsistency observed between BODIPY-GTP and BODIPY-GTPγS binding for GNAI1S47N and GNAO1S47N (Fig. 2, A-D). Conjugation of a fluorophore such as BODIPY to GTP can result in differences in apparent binding compared to unlabelled GTP (65, 66), and therefore dictates caution in calculation of absolute rates. Therefore, in the results presented here we limited our interpretations to relative rates. While our study demonstrates greater stability in vitro for GNAO1 than GPA1, not all human Gα subunits have been as easy to produce recombinantly as GNAO1. For example, chimeric approaches have previously been required to express Gα proteins in the soluble state, including for mammalians Gα subunits such as GNAT1 (61, 67, 68), GNA12 and GNA13 (69). These chimeras integrate short regions of the kinetically particularly slow but easily purified GNAI1 enzyme. Our success with GPA1 purification indicates that our expression and rapid StrepII purification method is worth evaluating for purification of full length recombinant human Gα proteins that are enzymatically active, without the need to resort to chimeric sequence substitutions. Experimental procedures Cloning GPA1 was amplified from Arabidopsis cDNA with flanking NcoI and BspEI restriction sites. GNAI1 with the same flanking restriction sites was amplified from a wild type clone (Genscript, clone OHu13586) and from a designed codon harmonized (70) gBlock synthesized by Integrated DNA Technologies. These Gα subunits were cloned into the NcoI and BspEI sites of pSTTa, a vector we adapted from pGEX to include N-terminal dual-StrepII tags, thrombin and TEV protease sites, a multiple cloning site and an optional C-terminal FLAG tag. GNAO1 was amplified from a commercial clone (Genscript, clone OHu15183), adapting a 5’ BspHI restriction site (yields a sticky end compatible with NcoI) and a blunt 3’ end to clone into NcoI/PmlI sites of pSTTa. The C-terminal RGS box of RGS1 (corresponding to residues 247–459) was amplified from Arabidopsis cDNA with flanking NcoI and BspEI sites to clone into pSTTa in the same manner as GPA1 and GNAI1. All genes cloned into pSTTa included a stop codon, so the ORF did not read through to the C-terminal FLAG tag included in the vector. Mutants of GPA1, GNAI1 and GNAO1 were generated by REPLACR mutagenesis (71). GPA1-GNAO1 helical domain swaps were generated by overlap-extension PCR (72) and cloned into pSTTa as above, with the exception that the GPA1GNAO1hel construct was amplified with a 5’ BspHI site. Helical domains were defined as GPA1 residues E68-Y188 and GNAO1 residues G63-R177, with the remainder of the protein flanking these regions defined at the Ras domains, consistent with the regions used in the GPA1-GNAI1 domain swap performed by Jones et al. (32). His- and GST-tagged constructs were generated by amplifying the ORFs of GPA1, GNAI1 and GNAO1, which were A-tailed, TOPO cloned into pCR8 and mobilized by LR Gateway recombination into pDEST17 (for His-tagged expression), and in the case of GPA1, pDEST15 (for GST-tagged expression) (Thermo). Primers for ORF cloning, mutagenesis and overlap-extension PCRs are listed in Table S3. All sequences were verified as correct by Sanger sequencing. Protein expression Proteins were heterologously expressed in E. coli BL21 DE3 cells using 75 μg/ml carbenicillin for plasmid selection. Typically, fresh transformants were grown in 7.5 ml overnight cultures (LB media supplemented with 0.5% D-glucose (w/v) and 3 g/L MgCl−2), pelleted by centrifugation at 5000 g for 10 minutes, and resuspended in 5 ml fresh pre-warmed LB and grown at 37°C. Five ml of pre-warmed HSLB (LB media supplemented with 17 g/L NaCl and 3 g/L MgCl−2, pH 7.0) was added at T=20 and 40 min. At T=60 min the pre-culture was added to 600 ml prewarmed HSLB in a vigorously shaking (225 rpm) 2 L baffled flask (OD600 = 0.04–0.06). Cultures were grown to an OD600 of 0.7–0.8, transferred to a room temperature (20–21 °C) shaker and grown for 20 minutes before induction with 125 μM IPTG for 3–4 hours. Cells were pelleted by 6000 g centrifugation for 10 minutes at 4 °C. Cell pellets were promptly frozen and typically processed the following morning, though proteins retained activity when cell pellets were stored for multiple weeks at −80 °C. Protein purification All buffers were prepared with high purity premium grade reagents (e.g. Honeywell TraceSelect, Sigma BioXtra or EMD Millipore EmSure) to minimize introduction of extraneous metals, and supplemented with one tablet Complete EDTA-free protease inhibitor (Roche, 5056489001) or Pierce Protease Inhibitor Tablets, EDTA-free (Thermo, A32965) per 50 ml. Columns were pre-rinsed with 1 ml of 0.25% Tween-20. Frozen cell pellets containing expressed StrepII-tag fusion proteins were resuspended with a 10 ml Pasteur pipet in 10 ml buffer W1 (100 mM Tris-HCl, 500 mM NaCl, 2 mM MgCl2, 5 mM TCEP and 5% glycerol pH 8.0) supplemented with ~10 mg lysozyme (Sigma, L1667), 25 μl/ml BioLock biotin blocker (IBA) and 5 μl Pierce Universal Nuclease (Thermo), and kept on ice. Cells were lysed by three rounds of sonication on ice using a Fisher Sonic Dismembrator equipped with a 3 mm tip with 1 second on/off pulses set to 20% amplitude for 15 seconds (i.e. 15x one second pulses), and the cell debris were pelleted by centrifugation at 10000 g at 4 °C for 10–20 minutes. The supernatant was passed through a 0.2 μM PES filter directly into a 1 ml column, with a 6 ml total capacity, containing a 0.25 ml resin bed of Streptactin sepharose (IBA) pre-washed with buffer W1. Loaded columns were washed sequentially with 0.5 ml W1 (1x) and 0.3 ml W2 (3x) (50 mM Tris-HCl, 100 mM NaCl and 5% glycerol pH 7.7) before eluting with sequential fractions of 220, 350, and 165 μl of “EB base” (25 mM Tris-HCl, 50 mM NaCl and 5% glycerol pH 7.4) supplemented with freshly added 5 mM desthiobiotin (Sigma) to form “EB”. The identity of the minor contaminate DnaK was performed via gel band excision, NH4HCO3/CH3CN destaining, dehydration, and subsequent MS/MS sequencing by the P.S.U. College of Medicine Mass Spectrometry and Proteomics Facility. For GST-fusion proteins, cell pellets were resuspended in TBS-NoCa binding buffer (50 mM Tris-HCl, 150 mM NaCl, 1 mM MgOAc, 10 mM β-mercaptoethanol and 1 mM Imidazole, pH 8.0) and sonicated and centrifuged as above. The resultant supernatant was passed through a 0.2 μM PES filter into a 1 ml column with a 0.25 ml Pierce Glutathione Agarose (Pierce, 16100) resin bed, essentially mimicking the StrepII purification protocol. Sequential washes were performed with 2 ml (x1) and 1 ml (x2) TBS-NoCa before protein elution with sequential fractions of 220 μl (E1), 350 μl (E2), and 165 μl (E3) TBS-NoCa supplemented with 10 mM glutathione. His-fusion proteins were purified essentially as previously described for BODIPY reactions (41). Briefly, our purification protocol mimicked the StrepII protocol, with the following modifications: lysis/binding buffer was replaced with 15 ml of 50 mM Tris-HCl pH 8.0, 100 mM NaCl, 2 mM MgCl2, 0.2% C12E10, supplemented with 5 μl β-mercaptoethanol post sonication, cell debris was pelleted by centrifugation at 30000 g for 15 min, a 125 μl Talon (Takara) resin bed was used, the resin bed was washed with 1 ml of 50 mM Tris-HCl pH 8.0, 500 mM NaCl, 2 mM MgCl2, 5 mM β-mercaptoethanol, 0.2% C12E10 and 10 mM imidazole, and elution was performed with 20 mM Tris-HCl (pH 8.0), 250 mM NaCl, 5 mM β-mercaptoethanol, 10% glycerol and 250 mM imidazole. Peak elution fractions (second eluate fraction; E2) of GST and His tagged proteins were subjected to buffer exchange using Amicon Ultra 0.5 ml 10 kDa cutoff columns (Millipore Sigma) with five sequential rounds of concentration performed by centrifugation at 14000 g and 4 °C for approximately 10 min and dilution with “EB base” (5x, 5x, 5x, 5x, 2x) for a total dilution of 1250x. Protein quality and quantity were evaluated immediately after elution by SDS-PAGE of 10–20 μl fractions with a 3–4 lane mass ladder of Fraction V BSA (e.g. 0.5, 1.0, 1.5, 2.0 μg/lane) followed by Gel-Code Blue (Thermo) staining. Biochemical assays were initiated on the fraction displaying peak yield (almost always E2) immediately after PAGE analysis, generally 2–3 hours post-elution, during which time proteins had been stored on ice in a 4 °C refrigerator. We note that, under routine conditions and if pre-quantification of exact yield isn’t critical, the StrepII-tag E2 purity and concentration is consistent enough to allow for immediate biochemical analysis, within minutes of elution. BODIPY assays BODIPY-GTP (BODIPY FL GTP - product #G12411) and BODIPY-GTP-γS (BODIPY™ FL GTPγ-S - product #G22183) stocks were purchased from Thermo and diluted to 100 nM in Tris-HCl pH 7.4 immediately prior to use. BSA or buffer alone was used as a negative control as indicated in each assay. Proteins were diluted to twice the final assay concentration, generally 200 nM (GPA1, GNAO1 or BSA) or 400 nM (GNAI1 or BSA) in “EB base” and supplemented with 10 mM MgCl2 on ice, normally in a master mix sufficient to perform reactions in triplicate. 100 μl of each diluted protein was aliquoted to wells of a Costar 96 well plate (Corning #3631 - black with clear flat bottom non-treated plate) and loaded into a Synergy Neo2 multimode reader (Biotek), or in Figures S1E and 1F, an Flx800 plate reader (Biotek), set at 25 °C. Pre-injection background readings were taken with monochromators set to 486/18 nm excitation and 525/20 nm emission with a gain setting within the range 90–100 (Synergy Neo2), or 485/20 nm excitation and 528/20 nm emission filters with the sensitivity set to 90 (Flx800). Reactions were initiated utilizing plate reader injectors to dispense 100 μl of BODIPY-GTP or BODIPY-GTPγS to each well (at a rate of 250 μl/sec), yielding a final assay concentration of 50 nM BODIPY-GTP/-GTPγS, 100 or 200 nM protein and 5 mM Mg2+ cofactor. Kinetics were normally monitored in “plate mode” for 30 min with a kinetic interval of 3–6 seconds (Synergy Neo2) or 25–30 seconds (Flx800). In cases where rapid monitoring of initial BODIPY-GTPγS binding rates were assayed, samples were monitored in “well mode” for 30 seconds with an 80 msec kinetic interval (Synergy Neo2). SYPRO Orange assays We adapted the protein unfolding assay of Biggar et al. (73) to assess protein stability over time at 25 °C. Protein was diluted to 600 nM in “EB Base” supplemented with 5 mM MgCl2 and nucleotides as indicated with 5x SYPRO Orange dye (Thermo #S6650 – 5000X stock). Forty μl per reaction was aliquoted into wells of a FLUOTRAC 200 96 well half area plate (Greiner Bio-One #675076), loaded into a Synergy Neo2 multimode reader (Biotek) and fluorescence was monitored for the indicated length of time with monochromators set to 470/20 nm excitation and 570/20 nm emission with a gain setting of 100 and a kinetic interval of 5 or 6 seconds. Data analysis BODIPY assays represent the average of 3 technical replicates and were repeated 2–4 independent times (independent biological replicates) with the following exceptions; samples in Figure 1C were assayed in duplicate due to the time constraints of assaying an unstable protein in well-mode and GNAO1GPA1hel was assayed in duplicate in Figure 6B due to yield constraints. SYPRO Orange assays represent the average of 2 technical replicates and were repeated 2–3 independent times. Instrument-collected raw data were imported into GraphPad Prism (v9.5) for analysis and graphical presentation of the mean ± SEM for all timepoints. Supplementary Material Supplement 1 Figure S1. Comparison of GPA1 purification methods. A. Gel illustrating the purity of StrepII-GPA1 in our protein preparations, with the commonly co-purified 70+ kDa DnaK band. B. Proteins were purified in parallel for StrepII-GPA1, His-GPA1 and GST-GPA1 (marked by *) before His-GPA1 and GST-GPA1 proteins underwent buffer exchange into “EB base”. Proteins were separated by SDS-PAGE for quantification of yield and qualitative assessment of purity. C-D. StrepII-GPA1S52N does not display any binding activity when assayed with C. BODIPY-GTP or D. BODIPY-GTPγS. E. BODIPY-GTPγS binding curves of StrepII-GPA1 supplemented with no GDP, 10 μM GDP in the lysis/binding buffer, or 10 μM GDP in the lysis/binding buffer and elution buffers. F. BODIPY-GTP binding and hydrolysis data for 100 nM StrepII-GPA1 ±100 nM StrepII-RGS1 (cytosolic domain). All kinetic data in this manuscript were generated using a Synergy Neo2 multimode reader, with the exception of panels E and F in this figure, which were generated using an Flx800 plate reader. Figure S2. Control data for GNAI1 codon harmonization, and buffer reagent choices. A. DNA sequence of the codon harmonized GNAI1ch synthesized clone. B. Alignment of GNA1wt (native) and GNAI1ch protein sequences, generated with Clustal Omega. C. SDS-PAGE illustrates the relative yield and purity of Strep-tag purified GNAI1wt and GNAI1ch proteins. D-E. Assays comparing the activities of 250 nM StrepII-GNAI1wt vs. StrepII-GNAI1ch for D. BODIPY-GTP binding and hydrolysis, and E. BODIPY-GTPγS binding. F. Comparison of the BODIPY-GTP binding and hydrolysis activities of 100 nM StrepII-GPA1 purified and assayed in buffers prepared with standard grade reagents or trace metal free (TMF) grade reagents. Figure S3. A-B. Structural alignments of empirically derived A. GPA1-GNAO1 and B. GPA1-GNAI1 structures. PDB structure 2XTZ chain A (GPA1 – blue) was aligned in PyMol with 3C7K chain A (GNAO1 – green) or 1GIA chain A (GNAI1 – orange). The nucleotide (yellow for GPA1/light and dark blue for GNAO1 and GNAI1) is located in the binding pocket within the interdomain cleft, which is flanked by the Ras domain (upper domain) and helical domain (lower domain) in both panels. C. SYPRO Orange assay with rapid setup to demonstrate unfolding of GPA1 (600 nM) in vitro when not provided with excess nucleotide is almost immediate. D. An example of a GNAO1GPA1hel SYPRO Orange protein unfolding assay in which the rapid signal variation between timepoints displayed in Figures 6C and 6D was not observed. (Note the same y-axis scale was used in Figures 6C, 6D and S3D.) Table S1. Clinvar data associated with equivalent sites to GNAI1S47/GNAO1S47 and GNAI1Q204/GNAO1Q205 of Gα subunits. Table S2. COSMIC data associated with equivalent sites to GNAI1S47/GNAO1S47 and GNAI1Q204/GNAO1Q205 of Gα subunits. Table S3. Sequences of primers used in this study. Acknowledgments We thank Mr. David Arginteanu for technical assistance and the suggestion to evaluate sucrose as a cryoprotectant. Supported by NIGMS 5R01GM126079 to SMA. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Data availability All data supporting the findings of this manuscript are contained within the manuscript and supporting information. Raw data are available upon request from David Chakravorty. Abbreviations BSA bovine serum albumin BODIPY boron-dipyrromethene GAP GTPase activating protein GDP guanosine diphosphate GEF guanine nucleotide exchange factor GNAI1 Human guanine nucleotide-binding protein G(i) subunit alpha-1 GNAO1 Human guanine nucleotide-binding protein G(o) subunit alpha GPA1 Arabidopsis guanine nucleotide-binding protein alpha-1 subunit GPCR G protein-coupled receptor GTP Guanosine triphosphate GST glutathione S-transferase HSLB high salt Luria-Bertani RGS regulator of G protein signaling RLK receptor-like kinase Figure 1. Fresh StrepII-tagged protein displays higher enzymatic activity than GPA1 proteins purified using other methods or that have been stored. A. BODIPY-GTP binding and hydrolysis curves of GPA1 either freshly prepared or subjected to overnight storage at 4 °C to simulate the temperature and time effect of dialysis. B. Comparison of BODIPY-GTP binding and hydrolysis of GPA1 isolated by StrepII-, His- and GST-tag purification procedures. C. Comparison of the initial binding rates (note x-axis units) of the samples in panel B when assayed with BODIPY-GTPγS. D-E. SYPRO Orange protein unfolding assays conducted at 25 °C with GPA1 in the absence of additional nucleotides, compared to GPA1 supplemented with D. 125 μM GDP or E. 125 μM GTPγS. F. Comparison of BODIPY-GTP binding and hydrolysis activities of StrepII-GPA1 stored at −80 °C for 3 weeks with either 10% glycerol or 8.33% sucrose added to the elution fraction (which contains 5% glycerol) as a cryoprotectant. Note: For assays depicted in panels A and F the detector gain was set to 70, as opposed to 90–100 for other assays, hence the lower relative fluorescence values. 100 nM protein was used in BODIPY assays and 600 nM protein in SYPRO Orange assays. Bovine serum albumin (BSA) at equimolar concentration was used as a negative control as indicated. Figure 2. Comparison of StrepII-GNAI1 and StrepII-GNAO1 with dominant negative (S47N) and constitutively active (Q204L/Q205L) mutants. A-B. BODIPY-GTP binding and hydrolysis curves of A. GNAI1 or B. GNAO1, with corresponding mutants. C-D. BODIPY-GTPγS binding curves of C. GNAI1 or D. GNAO1, with corresponding mutants. Note: for all graphs, wild-type = blue, S47N = red, and Q204L/Q205L = orange. Figure 3. Influence of Mg2+ and GTPγS on StrepII-GNAI1, StrepII-GNAO1, and their associated S47N and Q204L/Q205L mutants. A-B. Mg2+ dependency assays for BODIPY-GTPγS binding by A. GNAI1 vs. GNAI1S47N or B. GNAO1 vs. GNAO1S47N. C-D. SYPRO Orange protein unfolding assay in the presence or absence of GTPγS for C. StrepII-GNAI1, -GNAI1S47N, and -GNAI1Q204L or D. StrepII-GNAO1, GNAO1S47N, and GNAO1Q205L. Note the reduced magnitude of RFU values compared to Figures 1D and 1E. The initial decrease in SYPRO Orange signal may result from temperature equilibration, as reactions were loaded immediately following sample preparation on ice. 200 nM protein was used in GNAI1 and 100 nM protein in GNAO1 BODIPY assays. 600 nM protein was used in the SYPRO Orange assays. Figure 4. Comparison of GPA1 activity to GNAI1 and GNAO1 activity. A. BODIPY-GTP binding and hydrolysis curves of StrepII-GPA1, StrepII-GNAI1 and StrepII-GNAO1. B-C. BODIPY-GTP binding and hydrolysis curves of B. StrepII-GNAI1 vs. His-GNAI1 and C. StrepII-GNAO1 vs. His-GNAO1. D. BODIPY-GTPγS binding curves of StrepII-GPA1 vs. His-GNAO1. E-F. Comparison of enzyme kinetics of StrepII-GPA1 vs. His-GNAO1. E. Binding and hydrolysis of BODIPY-GTP or F. binding of BODIPY-GTPγS. Gα proteins were used at 200 nM in panel B and 130 nM protein in panels C and D, while 100 nM protein was used in panels A, E and F. Figure 5. Saturation of BODIPY-GTPγS binding occurs at lower concentrations for GNAO1 than GPA1. A-B. Concentration-dependent kinetics and maximal binding of 50 nM BODIPY-GTPγS by A. StrepII-GPA1 or B. His-GNAO1. C. Comparison of binding and hydrolysis of BODIPY-GTP by StrepII-GPA1 vs. His-GNAO1 ±10 μM GDP. Gα proteins were used at 100 nM protein in panel C. Figure 6. Helical domain swap between GPA1 and GNAO1. Regions encoding the helical domains of GPA1 (residues 68–188) and GNAO1 (residues 63–177) were reciprocally swapped by overlap-extension PCR and the resultant constructs were all expressed with dual StrepII-tags, to eliminate the tag as a variable. A-B. Curves of GPA1, GNAO1 and helical domain swaps for A. BODIPY-GTPγS binding and B. BODIPY-GTP binding and hydrolysis. C-D. SYPRO Orange protein unfolding assays conducted at 25 °C with GPA1, GNAO1, GPA1GNAO1hel or GNAO1GPA1hel C. in the absence of supplementation with additional nucleotides, or D. in the presence of 10 μM GTPγS. 100 nM protein was used in BODIPY assays and 400 nM protein in SYPRO Orange assays. The authors declare that they have no conflicts of interest with the contents of this article. ==== Refs References 1. Van Raamsdonk C. D. , Bezrookove V. , Green G. , Bauer J. , Gaugler L. , O’Brien J. M. (2009) Frequent somatic mutations of GNAQ in uveal melanoma and blue naevi Nature 457 , 599–602 10.1038/nature07586 19078957 2. O’Hayre M. , Vazquez-Prado J. , Kufareva I. , Stawiski E. W. , Handel T. M. , Seshagiri S. (2013) The emerging mutational landscape of G proteins and G-protein-coupled receptors in cancer Nat. Rev. Cancer 13 , 412–424 10.1038/nrc3521 23640210 3. Hewitt N. , Ma N. , Arang N. , Martin S. A. , Prakash A. , DiBerto J. F. (2023) Catalytic site mutations confer multiple states of G protein activation Sci. Signal. 16 , eabq7842 10.1126/scisignal.abq7842 36787384 4. Muir A. M. , Gardner J. F. , van Jaarsveld R. H. , de Lange I. M. , van der Smagt J. J. , Wilson G. N. (2021) Variants in GNAI1 cause a syndrome associated with variable features including developmental delay, seizures, and hypotonia Genet. Med. 23 , 881–887 10.1038/s41436-020-01076-8 33473207 5. Weinstein L. S. , Shenker A. , Gejman P. V. , Merino M. J. , Friedman E. , and Spiegel A. M. (1991) Activating Mutations of the Stimulatory G Protein in the McCune-Albright Syndrome N. Engl. J. Med. 325 , 1688–1695 10.1056/nejm199112123252403 1944469 6. Gawler D. , Milligan G. , Spiegel A. M. , Unson C. G. , and Houslay M. D. (1987) Abolition of the expression of inhibitory guanine nucleotide regulatory protein Gi activity in diabetes Nature 327 , 229–232 10.1038/327229a0 3106832 7. Moxham C. M. , and Malbon C. C. (1996) Insulin action impaired by deficiency of the G-protein subunit Giα2 Nature 379 , 840–844 10.1038/379840a0 8587610 8. Siffert W. , Rosskopf D. , Siffert G. , Busch S. , Moritz A. , Erbel R. (1998) Association of a human G-protein β3 subunit variant with hypertension Nat. Genet. 18 , 45–48 10.1038/ng0198-45 9425898 9. Lerman B. B. , Dong B. , Stein K. M. , Markowitz S. M. , Linden J. , and Catanzaro D. F. (1998) Right Ventricular Outflow Tract Tachycardia Due To a Somatic Cell Mutation in G Protein subunit alphai2 J. Clin. Invest. 101 , 2862–2868 10.1172/jci1582 9637720 10. Ullah H. , Chen J. G. , Young J. C. , Im K. H. , Sussman M. R. , and Jones A. M. (2001) Modulation of Cell Proliferation by Heterotrimeric G Protein in Arabidopsis Science 292 , 2066–2069 10.1126/science.1059040 11408654 11. Ashikari M. , Wu J. Z. , Yano M. , Sasaki T. , and Yoshimura A. (1999) Rice gibberellin-insensitive dwarf mutant gene Dwarf 1 encodes the α-subunit of GTP-binding protein Proc. Natl. Acad. Sci. U. S. A. 96 , 10284–10289 10.1073/pnas.96.18.10284 10468600 12. Botella J. R. (2012) Can heterotrimeric G proteins help to feed the world? Trends Plant Sci. 17 , 563–568 10.1016/j.tplants.2012.06.002 22748359 13. Ullah H. , Chen J. G. , Wang S. C. , and Jones A. M. (2002) Role of a Heterotrimeric G Protein in Regulation of Arabidopsis Seed Germination Plant Physiol. 129 , 897–907 10.1104/pp.005017 12068128 14. Trusov Y. , Rookes J. E. , Tilbrook K. , Chakravorty D. , Mason M. G. , Anderson D. (2007) Heterotrimeric G Protein γ Subunits Provide Functional Selectivity in Gβγ Dimer Signaling in Arabidopsis Plant Cell 19 , 1235–1250 10.1105/tpc.107.050096 17468261 15. Wang X. Q. , Ullah H. , Jones A. M. , and Assmann S. M. (2001) G Protein Regulation of Ion Channels and Abscisic Acid Signaling in Arabidopsis Guard Cells Science 292 , 2070–2072 10.1126/science.1059046 11408655 16. Coursol S. , Fan L. M. , Le Stunff H. , Spiegel S. , Gilroy S. , and Assmann S. M. (2003) Sphingolipid signalling in Arabidopsis guard cells involves heterotrimeric G proteins Nature 423 , 651–654 10.1038/nature01643 12789341 17. Yu Y. Q. , Chakravorty D. , and Assmann S. M. (2018) The G Protein β-Subunit, AGB1, Interacts with FERONIA in RALF1-Regulated Stomatal Movement Plant Physiol. 176 , 2426–2440 10.1104/pp.17.01277 29301953 18. Colaneri A. C. , Tunc-Ozdemir M. , Huang J. P. , and Jones A. M. (2014) Growth attenuation under saline stress is mediated by the heterotrimeric G protein complex BMC Plant Biol. 14 , 129 10.1186/1471-2229-14-129 24884438 19. Ferrero-Serrano A. , and Assmann S. M. (2016) The α-subunit of the rice heterotrimeric G protein, RGA1, regulates drought tolerance during the vegetative phase in the dwarf rice mutant d1 J. Exp. Bot. 67 , 3433–3443 10.1093/jxb/erw183 27194741 20. Llorente F. , Alonso-Blanco C. , Sanchez-Rodriguez C. , Jorda L. , and Molina A. (2005) ERECTA receptor-like kinase and heterotrimeric G protein from Arabidopsis are required for resistance to the necrotrophic fungus Plectosphaerella cucumerina Plant J. 43 , 165–180 10.1111/j.1365-313X.2005.02440.x 15998304 21. Trusov Y. , Rookes J. E. , Chakravorty D. , Armour D. , Schenk P. M. , and Botella J. R. (2006) Heterotrimeric G Proteins Facilitate Arabidopsis Resistance to Necrotrophic Pathogens and Are Involved in Jasmonate Signaling Plant Physiol. 140 , 210–220 10.1104/pp.105.069625 16339801 22. McCudden C. R. , Hains M. D. , Kimple R. J. , Siderovski D. P. , and Willard F. S. (2005) G-protein signaling: back to the future Cell. Mol. Life. Sci. 62 , 551–577 10.1007/s00018-004-4462-3 15747061 23. Hauser A. S. , Attwood M. M. , Rask-Andersen M. , Schiöth H. B. , and Gloriam D. E. (2017) Trends in GPCR drug discovery: new agents, targets and indications Nat. Rev. Drug Discov. 16 , 829–842 10.1038/nrd.2017.178 29075003 24. Gookin T. E. , Kim J. , and Assmann S. M. (2008) Whole proteome identification of plant candidate G-protein coupled receptors in Arabidopsis, rice, and poplar: computational prediction and in-vivo protein coupling Genome Biol. 9 , R120 10.1186/gb-2008-9-7-r120 18671868 25. McFarlane H. E. , Mutwil-Anderwald D. , Verbancic J. , Picard K. L. , Gookin T. E. , Froehlich A. (2021) A G protein-coupled receptor-like module regulates cellulose synthase secretion from the endomembrane system in Arabidopsis Dev. Cell 56 , 1484–1497 10.1016/j.devcel.2021.03.031 26. Urano D. , and Jones A. M. (2013) “Round Up the Usual Suspects”: A Comment on Nonexistent Plant G Protein-Coupled Receptors Plant Physiol. 161 , 1097–1102 10.1104/pp.112.212324 23300167 27. Liu J. M. , Ding P. T. , Sun T. J. , Nitta Y. , Dong O. , Huang X. C. (2013) Heterotrimeric G Proteins Serve as a Converging Point in Plant Defense Signaling Activated by Multiple Receptor-Like Kinases Plant Physiol. 161 , 2146–2158 10.1104/pp.112.212431 23424249 28. Bommert P. , Il Je B. , Goldshmidt A. , and Jackson D. (2013) The maize Gα gene COMPACT PLANT2 functions in CLAVATA signalling to control shoot meristem size Nature 502 , 555–558 10.1038/nature12583 24025774 29. Ishida T. , Tabata R. , Yamada M. , Aida M. , Mitsumasu K. , Fujiwara M. (2014) Heterotrimeric G proteins control stem cell proliferation through CLAVATA signaling in Arabidopsis EMBO Rep. 15 , 1202–1209 10.15252/embr.201438660 25260844 30. Aranda-Sicilia M. N. , Trusov Y. , Maruta N. , Chakravorty D. , Zhang Y. L. , and Botella J. R. (2015) Heterotrimeric G proteins interact with defense-related receptor-like kinases in Arabidopsis J. Plant Physiol. 188 , 44–48 10.1016/j.jplph.2015.09.005 26414709 31. Chakravorty D. , and Assmann S. M. (2018) G protein subunit phosphorylation as a regulatory mechanism in heterotrimeric G protein signaling in mammals, yeast, and plants Biochem. J. 475 , 3331–3357 10.1042/bcj20160819 30413679 32. Jones J. C. , Duffy J. W. , Machius M. , Temple B. R. S. , Dohlman H. G. , and Jones A. M. (2011) The Crystal Structure of a Self-Activating G Protein α Subunit Reveals Its Distinct Mechanism of Signal Initiation Sci. Signal. 4 , ra8 10.1126/scisignal.2001446 21304159 33. Urano D. , Jones J. C. , Wang H. , Matthews M. , Bradford W. , Bennetzen J. L. (2012) G Protein Activation without a GEF in the Plant Kingdom PLoS Genet. 8 , e1002756 10.1371/journal.pgen.1002756 22761582 34. Schmidt T. G. M. , and Skerra A. (2007) The Strep-tag system for one-step purification and high-affinity detection or capturing of proteins Nat. Protoc. 2 , 1528–1535 10.1038/nprot.2007.209 17571060 35. Schmidt T. G. , and Skerra A. (1993) The random peptide library-assisted engineering of a C-terminal affinity peptide, useful for the detection and purification of a functional Ig Fv fragment Protein Eng. 6 , 109–122 10.1093/protein/6.1.109 8433964 36. Schmidt T. G. M. , Batz L. , Bonet L. , Carl U. , Holzapfel G. , Kiem K. (2013) Development of the Twin-Strep-tag® and its application for purification of recombinant proteins from cell culture supernatants Protein Expr. Purif. 92 , 54–61 10.1016/j.pep.2013.08.021 37. Schmidt T. G. M. , Koepke J. , Frank R. , and Skerra A. (1996) Molecular Interaction Between the Strep-tag Affinity Peptide and its Cognate Target, Streptavidin J. Mol. Biol. 255 , 753–766 10.1006/jmbi.1996.0061 8636976 38. McEwen D. P. , Gee K. R. , Kang H. C. , and Neubig R. R. (2001) Fluorescent BODIPY-GTP Analogs: Real-Time Measurement of Nucleotide Binding to G Proteins Anal. Biochem. 291 , 109–117 10.1006/abio.2001.5011 11262163 39. Willard F. S. , Kimple A. J. , Johnston C. A. , and Siderovski D. P. (2005) A direct fluorescence-based assay for RGS domain GTPase accelerating activity Anal. Biochem. 340 , 341–351 10.1016/j.ab.2005.02.015 15840508 40. Jameson E. E. , Roof R. A. , Whorton M. R. , Mosberg H. I. , Sunahara R. K. , Neubig R. R. (2005) Real-time Detection of Basal and Stimulated G Protein GTPase Activity Using Fluorescent GTP Analogues J. Biol. Chem. 280 , 7712–7719 10.1074/jbc.M413810200 15613467 41. Maruta N. , Trusov Y. , Chakravorty D. , Urano D. , Assmann S. M. , and Botella J. R. (2019) Nucleotide exchange-dependent and nucleotide exchange-independent functions of plant heterotrimeric GTP-binding proteins Sci. Signal. 12 , eaav9526 10.1126/scisignal.aav9526 31690635 42. Solis G. P. , Kozhanova T. V. , Koval A. , Zhilina S. S. , Mescheryakova T. I. , Abramov A. A. (2021) Pediatric Encephalopathy: Clinical, Biochemical and Cellular Insights into the Role of Gln52 of GNAO1 and GNAI1 for the Dominant Disease Cells 10 , 2749 10.3390/cells10102749 34685729 43. Jeong Y. , and Chung K. Y. (2023) Structural and Functional Implication of Natural Variants of Gαs Int. J. Mol. Sci. 24 , 4064 10.3390/ijms24044064 36835474 44. Barthel T. K. , and Walker G. C. (1999) Inferences Concerning the ATPase Properties of DnaK and Other HSP70s Are Affected by the ADP Kinase Activity of Copurifying Nucleoside-diphosphate Kinase J. Biol. Chem. 274 , 36670–36678 10.1074/jbc.274.51.36670 10593971 45. Hauptmann A. , Podgoršek K. , Kuzman D. , Srčič S. , Hoelzl G. , and Loerting T. (2018) Impact of Buffer, Protein Concentration and Sucrose Addition on the Aggregation and Particle Formation during Freezing and Thawing Pharm. Res. 35 , 101 10.1007/s11095-018-2378-5 29556730 46. Willard F. S. , and Siderovski D. P. (2004) Purification and In Vitro Functional Analysis of the Arabidopsis thaliana Regulator of G-Protein Signaling-1 In Methods in Enzymology: Regulators of G-Protein Signaling, Part A, 389 , 320–338 47. Jones J. C. , Temple B. R. S. , Jones A. M. , and Dohlman H. G. (2011) Functional Reconstitution of an Atypical G Protein Heterotrimer and Regulator of G Protein Signaling Protein (RGS1) from Arabidopsis thaliana J. Biol. Chem. 286 , 13143–13150 10.1074/jbc.M110.190355 21325279 48. Kleuss C. , Raw A. S. , Lee E. , Sprang S. R. , and Gilman A. G. (1994) Mechanism of GTP hydrolysis by G-protein α subunits Proc. Natl. Acad. Sci. U. S. A. 91 , 9828–9831 10.1073/pnas.91.21.9828 7937899 49. Wall M. A. , Posner B. A. , and Sprang S. R. (1998) Structural basis of activity and subunit recognition in G protein heterotrimers Structure 6 , 1169–1183 10.1016/s0969-2126(98)00117-8 9753695 50. Gao X. , Du Z. , and Patel T. B. (2005) Copper and Zinc Inhibit Gαs Function: A Nucleotide-free State of Gαs Induced by Cu2+ and Zn2+ J. Biol. Chem. 280 , 2579–2586 10.1074/jbc.M409791200 15546818 51. Weinstein L. S. , and Shenker A. (1993) G Protein Mutations in Human Disease Clin. Biochem. 26 , 333–338 10.1016/0009-9120(93)90109-j 8299203 52. Dror R. O. , Mildorf T. J. , Hilger D. , Manglik A. , Borhani D. W. , Arlow D. H. (2015) Structural basis for nucleotide exchange in heterotrimeric G proteins Science 348 , 1361–1365 10.1126/science.aaa5264 26089515 53. Mafi A. , Kim S.-K. , and Goddard W. A. (2022) The mechanism for ligand activation of the GPCR–G protein complex Proc. Natl. Acad. Sci. U. S. A. 119 , e2110085119 doi:10.1073/pnas.2110085119 35452328 54. Johnston C. A. , Taylor J. P. , Gao Y. , Kimple A. J. , Grigston J. C. , Chen J. G. (2007) GTPase acceleration as the rate-limiting step in Arabidopsis G protein-coupled sugar signaling Proc. Natl. Acad. Sci. U. S. A. 104 , 17317–17322 10.1073/pnas.0704751104 17951432 55. Landrum M. J. , Lee J. M. , Riley G. R. , Jang W. , Rubinstein W. S. , Church D. M. (2013) ClinVar: public archive of relationships among sequence variation and human phenotype Nucleic Acids Res. 42 , D980–D985 10.1093/nar/gkt1113 24234437 56. Tate J. G. , Bamford S. , Jubb H. C. , Sondka Z. , Beare D. M. , Bindal N. (2018) COSMIC: the Catalogue Of Somatic Mutations In Cancer Nucleic Acids Res. 47 , D941–D947 10.1093/nar/gky1015 57. Silva-Rodríguez P. , Fernández-Díaz D. , Bande M. , Pardo M. , Loidi L. , and Blanco-Teijeiro M. J. (2022) GNAQ and GNA11 Genes: A Comprehensive Review on Oncogenesis, Prognosis and Therapeutic Opportunities in Uveal Melanoma Cancers 14 , 3066 10.3390/cancers14133066 35804836 58. John J. , Rensland H. , Schlichting I. , Vetter I. , Borasio G. D. , Goody R. S. (1993) Kinetic and Structural Analysis of the Mg2+-binding Site of the Guanine Nucleotide-binding Protein p21H-ras J. Biol. Chem. 268 , 923–929 10.1016/S0021-9258(18)54022-9 8419371 59. Slepak V. Z. , Quick M. W. , Aragay A. M. , Davidson N. , Lester H. A. , and Simon M. I. (1993) Random Mutagenesis of G protein α Subunit Goα - Mutations Altering Nucleotide Binding J. Biol. Chem. 268 , 21889–21894 10.1016/S0021-9258(20)80624-3 8408043 60. Natochin M. , Barren B. , and Artemyev N. O. (2006) Dominant Negative Mutants of Transducin-α That Block Activated Receptor Biochemistry 45 , 6488–6494 10.1021/bi060381e 16700559 61. Ramachandran S. , and Cerione R. A. (2011) A Dominant-negative Gα Mutant That Traps a Stable Rhodopsin-Gα-GTP-βγ Complex J. Biol. Chem. 286 , 12702–12711 10.1074/jbc.M110.166538 21285355 62. Jones J. C. , Jones A. M. , Temple B. R. S. , and Dohlman H. G. (2012) Differences in intradomain and interdomain motion confer distinct activation properties to structurally similar Gα proteins Proc. Natl. Acad. Sci. U. S. A. 109 , 7275–7279 10.1073/pnas.1202943109 22529365 63. Rasmussen S. G. F. , DeVree B. T. , Zou Y. Z. , Kruse A. C. , Chung K. Y. , Kobilka T. S. (2011) Crystal structure of the β2 adrenergic receptor-Gs protein complex Nature 477 , 549–555 10.1038/nature10361 21772288 64. Westfield G. H. , Rasmussen S. G. F. , Su M. , Dutta S. , DeVree B. T. , Chung K. Y. (2011) Structural flexibility of the Gαs α-helical domain in the β2-adrenoceptor Gs complex Proc. Natl. Acad. Sci. U. S. A. 108 , 16086–16091 10.1073/pnas.1113645108 21914848 65. Dorn M. , Jurk M. , and Schmieder P. (2012) Blue News Update: BODIPY-GTP Binds to the Blue-Light Receptor YtvA While GTP Does Not PLoS One 7 , e29201 10.1371/journal.pone.0029201 22247770 66. Goody R. S. (2014) How not to do kinetics: examples involving GTPases and guanine nucleotide exchange factors Febs J. 281 , 593–600 10.1111/febs.12551 24112651 67. Skiba N. P. , Bae H. , and Hamm H. E. (1996) Mapping of Effector Binding Sites of Transducin α-Subunit Using Gαt/Gαi1 Chimeras J. Biol. Chem. 271 , 413–424 10.1074/jbc.271.1.413 8550597 68. Slep K. C. , Kercher M. A. , He W. , Cowan C. W. , Wensel T. G. , and Sigler P. B. (2001) Structural determinants for regulation of phosphodiesterase by a G protein at 2.0 Å Nature 409 , 1071–1077 10.1038/35059138 11234020 69. Kreutz B. , Yau D. M. , Nance M. R. , Tanabe S. , Tesmer J. J. G. , and Kozasa T. (2006) A New Approach to Producing Functional Gα Subunits Yields the Activated and Deactivated Structures of Gα12/13 Proteins Biochemistry 45 , 167–174 10.1021/bi051729t 16388592 70. Angov E. , Hillier C. J. , Kincaid R. L. , and Lyon J. A. (2008) Heterologous Protein Expression Is Enhanced by Harmonizing the Codon Usage Frequencies of the Target Gene with those of the Expression Host PLoS One 3 , e2189 10.1371/journal.pone.0002189 18478103 71. Trehan A. , Kiełbus M. , Czapinski J. , Stepulak A. , Huhtaniemi I. , and Rivero-Müller A. (2016) REPLACR-mutagenesis, a one-step method for site-directed mutagenesis by recombineering Sci. Rep. 6 , 19121 10.1038/srep19121 26750263 72. Ling M. M. , and Robinson B. H. (1997) Approaches to DNA Mutagenesis: An Overview Anal. Biochem. 254 , 157–178 10.1006/abio.1997.2428 9417773 73. Biggar K. K. , Dawson N. J. , and Storey K. B. (2012) Real-time protein unfolding: a method for determining the kinetics of native protein denaturation using a quantitative real-time thermocycler Biotechniques 53 , 231–238 10.2144/0000113922 23046506
PMC010xxxxxx/PMC10246017.txt
==== Front bioRxiv BIORXIV bioRxiv Cold Spring Harbor Laboratory 37292844 10.1101/2023.05.26.542547 preprint 2 Article Replication initiation in bacteria: precision control based on protein counting Fu Haochen †Department of Physics, University of California San Diego, 9500 Gilman Dr, La Jolla, CA 92093 Xiao Fangzhou †Department of Physics, University of California San Diego, 9500 Gilman Dr, La Jolla, CA 92093 Jun Suckjoon *Department of Physics and Department of Molecular Biology, University of California San Diego, 9500 Gilman Dr, La Jolla, CA 92093 † These authors contributed equally to this work. * Corresponding author: [email protected] 04 7 2023 2023.05.26.542547https://creativecommons.org/licenses/by-nc-nd/4.0/ This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which allows reusers to copy and distribute the material in any medium or format in unadapted form only, for noncommercial purposes only, and only so long as attribution is given to the creator. nihpp-2023.05.26.542547.pdf Balanced biosynthesis is the hallmark of bacterial cell physiology, where the concentrations of stable proteins remain steady. However, this poses a conceptual challenge to modeling the cell-cycle and cell-size controls in bacteria, as prevailing concentration-based eukaryote models are not directly applicable. In this study, we revisit and significantly extend the initiator-titration model, proposed thirty years ago, and explain how bacteria precisely and robustly control replication initiation based on the mechanism of protein copy-number sensing. Using a mean-field approach, we first derive an analytical expression of the cell size at initiation based on three biological mechanistic control parameters for an extended initiator-titration model. We also study the stability of our model analytically and show that initiation can become unstable in multifork replication conditions. Using simulations, we further show that the presence of the conversion between active and inactive initiator protein forms significantly represses initiation instability. Importantly, the two-step Poisson process set by the initiator titration step results in significantly improved initiation synchrony with CV~1/N scaling rather than the standard 1/N scaling in the Poisson process, where N is the total number of initiators required for initiation. Our results answer two long-standing questions in replication initiation: (1) Why do bacteria produce almost two orders of magnitude more DnaA, the master initiator proteins, than required for initiation? (2) Why does DnaA exist in active (DnaA-ATP) and inactive (DnaA-ADP) forms if only the active form is competent for initiation? The mechanism presented in this work provides a satisfying general solution to how the cell can achieve precision control without sensing protein concentrations, with broad implications from evolution to the design of synthetic cells. ==== Body pmcI. INTRODUCTION Most biology textbooks explain biological decision-making by emphasizing the control and sensing of key protein concentrations through programmed gene expression and protein degradation in eukaryotes. Protein concentration gradients can encode spatial or temporal information across different scales, such as morphogen gradients in the French flag model in developmental biology [1] or cyclin oscillations in eukaryotic cell-cycle controls [2] [Fig. 1(a)]. However, in bacterial cell physiology, balanced biosynthesis has been the hallmark since the 1950s at the population and single-cell levels [3–5]. Balanced biosynthesis means that the synthesis rate of all cellular components is the same as the cell’s growth rate in steady-state growth, wherein the concentrations of stable proteins are steady by the balance of their production and dilution [Fig. 1(b)]. However, balanced biosynthesis poses a fundamental conceptual challenge to modeling the cell-cycle and cell-size controls, as the prevailing concentration-based models are not directly applicable if the concentration of cell-cycle proteins remains constant (within stochasticity). Indeed, for the billion-year divergent model bacterial organisms Escherichia coli and Bacillus subtilis, their size control is based on (1) balanced biosynthesis of division initiator protein FtsZ and (2) its accumulation to a threshold number (not concentration) [6]. These two conditions lead to the adder phenotype [6]. Unfortunately, a mechanistic investigation of threshold FtsZ number sensing is a formidable challenge because division initiation involves multiple interacting proteins with unknown properties [7]. Replication initiation in bacteria, which is exclusively controlled by the widely-conserved master regulator protein, DnaA, is an attractive problem for mechanistic investigation because it exhibits the adder phenotype [8–11]. That is, the added cell size between two consecutive initiation events is independent of the cell size at initiation, as originally suggested by Sompayrac and Maaloe [12]. The adder phenotype implies that cells likely accumulate the DnaA molecules to a threshold number [6], and the synthesis of DnaA is balanced [13]. Furthermore, DnaA has been extensively studied, and most properties required for modeling are known or can be estimated [14–18]. Therefore, we view E. coli replication initiation as a tractable problem to understand the mechanism of protein copy-number sensing to control the cell cycle and cell size, and gain mechanistic insight into the general class of precision control in biology. In this work, we revisit and significantly extend the initiator-titration model proposed by Hansen, Christensen, and Atlung thirty years ago [19], the model closest to the protein-number-sensing idea (see Section A). In Section A, we summarize the original initiator-titration model and introduce our initiator-titration model v2. In Section B, we first introduce the “protocell” model, a minimal version of the initiator-titration model, and derive the first expression of the protocell size at initiation (known as the “initiation mass”). In Section C, we perform a dynamical stability analysis of the protocell model and show the existence of initiation instability. In Section D, we extend the protocell to our “initiator-titration model v2” and derive an analytical expression for the initiation mass in a special case (the Δ4 mutant [13]) based on three mechanistic biological control parameters: the expression level of DnaA, the ratio of the active vs. passive forms of DnaA, namely, [DnaA-ATP]/[DnaA-ADP], and the number of DnaA titration boxes on the chromosome. In the same section, we show that adding the replication-dependent, biologically-observed DnaA-ATP → DnaA-ADP conversion element (RIDA) restores initiation stability [20, 21]. In Section E, we discuss initiation asynchrony and cell-to-cell variability using the concept of intrinsic and extrinsic noise in the framework of initiator-titration model v2. Our model provides a quantitative and mechanistic explanation for several long-standing questions in bacterial replication initiation with the following findings: DnaA titration boxes are the protein-counting device that measures the threshold number of initiator proteins, and the two forms of DnaA (DnaA-ATP and DnaA-ADP), and especially the replication-dependent DnaA-ATP → DnaA-ADP, are needed to suppress initiation instability. Given the fundamental nature of replication initiation and its profound differences from eukaryotic cell-cycle control, we anticipate broad applications of our results, from the design of synthetic cells to the evolution of biological mechanisms in precision control. II. RESULTS AND DISCUSSION A. The “initiator-titration model v2” and intuition Consider engineering a synthetic cell capable of self-replication. For such a cell to be viable, it must meet a fundamental requirement for cell-cycle control: initiating replication only once during cell division. A possible “simple” strategy to implement this requirement could be as follows [Fig. 2(a)]: (1) The chromosome has one origin of replication. (2) The cell produces one initiator protein during the division cycle. (3) The initiator protein binds to ori (the replication origin) and immediately triggers initiation. (4) Upon initiation, the cell destroys the initiator protein. While this seemingly straightforward strategy could limit the replication origin to a single site and produce a single initiator protein during cell division, the underlying mechanisms required to achieve this are likely more complex. For instance, how would the cell “know” when to produce them and when to degrade them? While E. coli exhibits characteristics similar to the hypothetical strategy described above, there are notable differences. E. coli has one replication origin (ori), but replication initiation requires 10–20 master regulator DnaA molecules binding to the eleven DnaA boxes at ori [15–17, 22]. Furthermore, DnaA is stable and not degraded upon initiation [15, 16]. Strikingly, E. coli produces approximately 300 copies of DnaA per ori, or 30 times more than required at ori, with almost all being titrated by DnaA boxes encoded on the chromosome [15, 16]. In 1991, Hansen and colleagues proposed the initiator-titration model to explain these observations [Fig. 2(b)] [19]. Their model posits that DnaA is first titrated by high-affinity DnaA boxes on the chromosome, which allows it to bind ori with weak affinity and initiate replication only after the chromosomal DnaA boxes are nearly saturated. This highlights the importance of DnaA boxes on the chromosome as the timing device for replication initiation. Our model builds upon the initiator-titration model and incorporates the knowledge in DnaA accumulated in the past 30 years [15–17, 22]. Specifically, we have learned that DnaA exists in two forms, DnaA-ATP and DnaA-ADP, with different binding affinities to DNA [23]. DnaA-ATP is the active form that can trigger initiation, while DnaA-ADP is inactive as it cannot bind ori specifically [24, 25]. Further genetic, biochemical, and bioinformatic studies have revealed that approximately 300 high-affinity DnaA boxes are distributed across the circular chromosome [15, 26]. By contrast, ori contains a cluster of eleven DnaA binding sites, wherein only three have high affinities [25, 27]. Therefore, most DnaA, whether DnaA-ATP or DnaA-ADP, will first bind the high-affinity chromosomal DnaA boxes. Only after the titration step, do DnaA-ATP molecules bind the weak binding sites within ori and trigger initiation. We refer to this updated model as the initiator-titration model v2, in recognition of the pioneering work of Hansen et al. [19, 28]. Figure 2(c) illustrates how our initiator-titration model v2 works in more detail. To provide intuition without losing the generality of our ideas, let us consider a naked circular chromosome without bound DnaA. As DnaA binds to ATP or ADP tightly [23] and the cellular concentration of ATP is almost 10x higher than ADP [29, 30], newly synthesized DnaA molecules become DnaA-ATP. During steady-state growth, both DnaA-ATP and DanA-ADP exist in the cell due to multiple interconversion mechanisms [16]. (See Section D and Appendix D for detailed discussion). DnaA-ATP and DnaA-ADP will first bind to around 300 high binding-affinity chromosomal DnaA boxes KD≈1nM [26], whereas only DnaA-ATP can bind to around ten low-affinity boxes within ori KD≈100nM [26] [31]. When most chromosomal DnaA boxes are saturated, the probabilities for DnaA-ATP binding to ori vs. the remaining chromosomal DnaA boxes become comparable. Initiation is triggered once the low-affinity ori binding sites are saturated by DnaA-ATP. As we elaborate below, the initiator-titration model v2 answers two long-standing fundamental questions: Why does E. coli produce so many more DnaA proteins than required for initiation, only to be titrated? Why does E. coli maintain two forms of DnaA in the first place if they only need DnaA-ATP for initiation? B. The “protocell”: a minimal initiator-titration model To gain analytical insight, we first construct a minimal initiator-titration model, named “protocell” [Fig. 3(a)]. The protocell has the complexity between the two versions of the initiator-titration model [Figs. 2(b) & 2(c)]. The protocell has one ori, the active initiator protein (e.g., DnaA-ATP in E. coli), and the initiator binding sites on the chromosome. We assume the following based on the experimental data: The cell grows exponentially V(t)=V0eλt in steady-state [3], where V(t) is the total cell size at time t, and λ is the growth rate. The mass-doubling time τ is given by τ=ln2λ. Synthesis of the initiator protein is balanced, i.e., its concentration is constant during growth [3]. We denote the initiator protein copy number at time t as I(t) and its concentration as cI. The rate of DNA synthesis is constant [34, 35], with the duration of chromosome replication C, independent of the mass-doubling time τ [36]. The chromosome encodes specific DNA sequences for binding of the initiator proteins. NB high-affinity sites are evenly distributed on the chromosome [15], and nB low-affinity sites are localized at ori [16]. For the E. coli chromosome, we set NB=300 and nB=10, as explained in Section A. During replication, the total number of initiator binding sites increases as B(t). Initiators tightly bind to the binding sites rather than staying in the cytoplasm, and initiators preferentially bind the chromosomal binding sites before binding to the ones at ori. Therefore, replication initiates at t=tini when It=tini=Bt=tini, namely, all binding sites are saturated by the initiator proteins. For illustration purposes, we consider an intermediate growth condition, where two cell cycles slightly overlap without exhibiting multifork replication [36] [Fig. 3(b)]. In the Helmstetter-Cooper model [37], this corresponds to C<τ<C+D, where D is the duration between replication termination and cell division. As such, the cell can have two intact chromosomes between termination and the next initiation [Fig. 3(b)]. The steady-state curves of I(t) and B(t) are shown in Fig. 3(b) (in our model, a steady state means all derived quantities are periodic with a period of τ). In general, I(t) increases exponentially because of exponential growth and balanced biosynthesis (Assumptions 1&2 above), whereas B(t) increases piecewise linearly because of replication initiation and termination (Assumptions 3&4). Therefore, the number of initiators catches up with the total number of binding sites between replication termination and the new round of initiation at It=tini=Bt=tini=2NB+nB (Assumption 5). Here, the factor “2” refers to the fact that there are two entire chromosomes and two ori’s right before the initiation event in the specific growth condition depicted in Fig. 3(b). Upon initiation, the number of binding sites B(t) increases discontinuously by 2nB due to the duplication of both ori’s and the binding sites therein. After that, B(t) increases at the rate 2NB/C, steeper than the slope of I(t). Once the cell divides, I(t) and B(t) drop by half, and the cell repeats its cycle. From this picture, the initiation mass vi, defined by cell volume per ori at initiation [36], can be easily calculated by the number of initiators at initiation, (1) vi=Itini2cI=1cINB+nB, where cI is the initiator protein concentration, and “2” reflects the copy number of ori before initiation. The above result can be extended to different growth conditions. For example, in slow growth (τ>C+D), the replication cycles do not overlap, and all the factor “2” will vanish in the above analysis due to the single chromosome at initiation. This results in the same initiation mass vi as in the intermediate growth condition. In fast-growth conditions (τ<C), replication cycles overlap, exhibiting multifork replication. Since a new round of replication starts before the previous round of replication is completed, the initiation mass is given by (2) vi=1cIαNB+nB, with the cell-cycle dependent parameter α≤1 given as (3) α=12n+2−n+22nτC,n=⌊Cτ⌋, which applies to any growth conditions (see Appendix A for a derivation). α=1 when τ≥C (non-multifork replication), and 0<α<1 when τ<C (multifork replication) [Fig. 3(c)]. Thus, α refers to the degree of overlapping replication. Some of the most salient predictions of these results include (1) The initiation mass is inversely proportional to the initiator concentration cI (2) The initiation mass linearly depends on the number of chromosomal binding sites NB. The basis of the protocell’s behavior is that the initiator increases exponentially, whereas the number of binding sites increases piecewise linearly only during DNA replication. This allows the cell to reach the initiation point I(t)=B(t) from any initial conditions. Therefore, the protocell can always trigger initiation by protein number counting through titration. C. The protocell exhibits initiation instability. In the last section, we addressed if a solution exists in the minimal protocell model with a period of τ. We showed that this periodic solution always exists (Eq. 2). We defined it as the “steady-state” solution in the biological sense that the cell can grow in a steady state with the periodic cell cycle. However, since the model is dynamic, convergence to a steady state from a given initial condition, I(0) and B(0), is not guaranteed. Hence, in this section, we study how the replication cycle propagates in the lineage from an arbitrary initial condition at t=0, and under what conditions the cycle converges to the steady-state solution. Intuitively, if the two consecutive initiations are separated by τ, thus periodic, the system is in a steady state. Suppose an initiation event at t=0, and its initiation mass deviates from the steady-state solution Eq. 2. Typically, the next initiation occurs at t=t+≠τ. However, if this time interval between two consecutive initiations eventually converges to τ after generations, the steady-state solution is stable under perturbations on the initial conditions. Otherwise, the steady-state solution is unstable. In the rest of this section, we analyze a dynamical system based on Assumptions 1–5 in Section B on the protocell. 1. Setup We consider a protocell containing one chromosome with ongoing multifork replication [Fig. 4(a)]. We block the cell division so the protocell grows indefinitely as the chromosome replicates and multiplies starting from the initial condition. As the cell size approaches infinity, does the initiation mass have a fixed value (stable) or multiple values (unstable)? The analysis is non-trivial, as we need to accommodate arbitrary initial conditions. To this end, we start with the dynamics of I(t) and B(t). First, we have (4) It=I0eλt, as a consequence of exponential cell growth and balanced biosynthesis of the initiator proteins. Next, the dynamics of the number of binding sites B(t) is more subtle because it increases piecewise linearly depending on the replication state of the chromosome and the number of replication forks. To accommodate the possibility of arbitrary initial conditions, we define the “multifork tracker” vector variable, ρ(t), as follows. (5) ρ(t)≡ρ1(t),ρ2(t),⋯,ρd(t),ifd≥1,0,ifd=0. Here, the index d is the total number of generations (namely, the total rounds of replication cycles) since the initial chromosome, so d can grow indefinitely with time. That is, at every new round of the replication cycle, the size of the vector increases by one from d to d+1.d=0 is for the initial cell supposed to have an intact single chromosome without ongoing replications. We use the variable ρ to indicate the relative position of a replication fork of interest between ori and ter (the replication terminus), and therefore 0≤ρ(t)≤1 [Fig. 4(b)]. For example, ρ would be 0.5 if a pair of forks is exactly halfway between ori and ter [Figs 4(a) & 4(b)]. To track multifork replication, we use ρi(t) to represent the group of replication forks that are the i-th closest to the ori [Fig. 4(a)]. For example, i=1 always refers to the newest group of replication forks. To record the replication history, we set ρi(t)=1 for those replication forks that have already reached ter [Fig. 4(b)]. By these definitions, ρ(t) applies to both multifork replication and non-multifork replication. Based on the multifork tracker vector, the number of binding sites B(t) is completely determined by ρ as (6) B[ρ(t)]=NB1+∑i=1dρi(t)2d−i+2dnB. The dynamics of ρ(t) consists of two parts: First, between two initiation events, ρi(t) increases linearly with a slope of 1/C until it reaches 1, as replication forks travel from ori to ter [Fig. 4(b)]. Second, at initiation, the dimension of ρ increases by one, shifting its components to the right as S:Rd→Rd+1,ρ1,ρ2,⋯,ρd↦0,ρ1,ρ2,⋯,ρd to accommodate the new pair of replication forks at each ori [see also, Fig. 4(a)]. 2. Properties of the steady state The steady-state solution assumes periodicity of dynamics so that I(t) and B(t) double in each replication cycle. We consider the mapping between two consecutive initiation events to solve for the steady-state condition. We denote the first initiation event as ρ(t=0)=ρ at t=0, and the second initiation event as ρt=t+=ρ+ at t=t+. The mapping ℱ:Rd−1→Rd,ρ↦ρ+ requires a time-translation and a shift: (7) ρi+=ρi−1+t+C,ifρi−1+t+C<1,1,else, where the initiation time t+ is determined by the initiation criteria that I(t=0)=B(t=0) and It=t+=Bt=t+, Eq. 4, and Eq. 6, (8) eλt+2NB2−(d−1)+∑i=1d−1ρi2−i+nB=NB2−d+∑i=1dρi+2−i+nB. Equations 7 & 8 describe the dynamics of the system at initiation. We can now obtain the fixed point of the mapping ℱ by setting d→∞ and ρ+=ρ (Appendix B): (9) t+=τ,ρiss=iτC,ifi≤⌊Cτ⌋,1,else. The resulting expression for steady-state initiation mass is the same as Eq. 2, i.e., the fixed point of ℱ is the steady-state solution (see Appendix B for more details). Next, we study the stability of the fixed point of ℱ by calculating the Jacobian matrix of ℱ at the fixed point: (10) J=∂ρi+∂ρjss. This matrix can be reduced to an n×n matrix (n=⌊Cτ⌋), since all other matrix elements are zero. In E. coli, 0≤n≤2 in most growth conditions; here, we consider the range of 0≤n≤3 to accommodate cells theoretically doubling as frequently as at every τ=10 minutes, with a C period of 40 minutes. Therefore, we can calculate the eigenvalues of J for each n. Stability requires the largest eigenvalue of J to be smaller than 1. Eventually, we can obtain the stable and unstable regimes in the nB/NB vs. C/τ phase diagram, as shown in Fig. 4(c) (see Appendix B, also Appendix Fig. 2). Importantly, the phase diagram reveals both stable (n<1) and unstable (small nB/NB when n>1) steady states [Fig. 4(c)]. What happens when the system becomes unstable? As discussed earlier, in fast growth conditions, α<1 in the steady-state initiation mass expression (Eq. 2). Indeed, using numerical simulations, we found that the initiation mass oscillates between two values [Fig. 4(c)]. This indicates that the cell cycle can oscillate between multifork and non-multifork replication. Mathematically, this oscillatory behavior means that the fixed points of ℱo2=ℱ∘ℱ are stable, although the fixed point of ℱ is unstable. By fixing one of the fixed points of ℱo2 as ρ1=1, we can compute the other fixed point with ρ1<1 (see Appendix C). In extreme cases, ρ1 can be as small as 0.1. That is, the second round of replication starts only after 10% of the chromosome has been replicated by the replication forks from the previous initiation. When the replication forks from two consecutive rounds of initiation are too close to each other, they cannot be separated into two division cycles. This should result in two initiation events in one division cycle, and no initiation in the next division cycle. Therefore, although initiation triggering is guaranteed, the performance of the protocell is imperfect in terms of initiation instability in certain growth conditions. We show how the initiator-tiration model v2 resolves the instability issue in Section D. D. The initiator-titration model v2: replication-dependent DnaA-ATP → DnaA-ADP conversion stabilizes the cell cycle. In the previous section, we showed that the protocell can show initiation instability. In understanding why wild-type E. coli initiation is stable, we have to consider unique features of DnaA in E. coli, namely, its two distinct forms: the active DnaA-ATP and the inactive DnaA-ADP [23]. Several extrinsic elements, categorized into two main groups, interconvert between these DnaA forms [16, 20, 21, 38–41] [Fig. 5(a)]. The first group catalyzes the conversion of DnaA-ATP → DnaA-ADP. This includes the Regulatory Inactivation of DnaA (RIDA) [20, 21] and datA-dependent DnaA-ATP Hydrolysis (DDAH) [38, 39]. RIDA’s functionality requires active replication forks [42], thus rendering it replication-dependent, while DDAH’s datA, a DnaA binding chromosomal locus, fosters DnaA-ATP hydrolysis. By contrast, the second group, comprised of DARS1 and DARS2 (types of DnaA Reactivating Sequences), facilitates the ADP → ATP exchange for DnaA-ADP [40, 41]. Importantly, DnaA harbors intrinsic ATPase activity that facilitates its own conversion from DnaA-ATP → DnaA-ADP [23, 27], a feature not depicted in Fig. 5(a). Intriguingly, Δ4 cells — cells with a full deletion of all extrinsic DnaA-ATP ↔ DnaA-ADP interconversion path-ways — exhibit a nearly identical initiation phenotype as wild-type cells [13], solely relying on DnaA’s intrinsic ATPase activity. Figure 5(b) presents our numerical simulation results illuminating the alterations in the [DnaA-ATP]/[DnaA-ADP] ratio during the cell cycle in a DNA replication-dependent manner (see Appendix E for simulation details). Notably, this ratio should remain steady in Δ4 cells during cell elongation in a steady state [13]. Furthermore, re-initiation is prohibited within a certain timeframe post-initiation (approximately 10 minutes; the “eclipse period”), attributable to the sequestration of newly synthesized DNA by SeqA [43]. In this section, we integrate each of these features into our protocell model to formulate our initiator-titration model v2 and compute the initiation stability phase diagram. Our findings reveal that the replication-dependent DnaA-ATP → DnaA-ADP conversion by RIDA largely alleviates initiation instability, thus reinstating the stability characteristic of wild-type cells. 1. Analytical expression of the initiation mass in the initiator-titration model v2 with a constant DnaA-ATP/DnaA-ADP ratio (Δ4 cells) First, we incorporate the two forms of DnaA with the intrinsic DnaA-ATP → DnaA-ADP activity by DnaA into the protocell model to construct the Δ4 cells, a minimal version of the initiator-titration model v2 [Fig. 2(c)]. As noted earlier, both DnaA-ATP and DnaA-ADP can bind the chromosomal DnaA boxes because of their strong binding affinity (KD~1nM [15, 26, 44]), whereas only DnaA-ATP can bind the weak DnaA boxes at ori with KD~102nM [15, 26, 45]. With the same Assumptions 1–4 in Section B and this additional assumption, we can derive an analytical expression for steady-state initiation mass for Δ4 E. coli (see Appendix D for the derivation): (11) vi=αNB+(1+[DnaA-ADP][DnaA-ATP])nB[DnaA]−(1+[DnaA-ADP][DnaA-ATP])KeffnB, where Keff is the effective dissociation constant of DnaA at ori, and α is in Eq. 3. Therefore, this equation brings together the expression level of DnaA via [DnaA], the ratio [DnaA-ATP]/[DnaA-ADP], and the degree of overlapping replication (α). Note that, under physiological conditions, KeffnB≪[DnaA] (Appendix D). If [DnaA-ATP] ≫ [DnaA-ADP], all DnaA molecules are in their active form DnaA-ATP, and the Δ4 E. coli converges to the protocell (i.e., Eq. 11 converging to Eq. 2). 2. The Δ4 cells show initiation instability. We also investigated the initiation stability of the Δ4 cells using numerical simulations [Fig. 5(c)] (see Appendix E for simulation details). The initiation stability phase diagram is analogous to that of the protocells in Fig. 4(c), showing an island of instability regime. This occurs during the transition into multifork replication, wherein initiation mass alternates between two values. Importantly, changing the [DnaA-ATP]/[DnaA-ADP] ratio does not significantly impact the stability [Appendix Fig. 2(b)]. 3. Replication-dependent DnaA-ATP → DnaA-ADP by RIDA alone can restore initiation stability. Next, we implemented the extrinsic DnaA-ATP ↔ DnaA-ADP conversion elements in the Δ4 cells. In contrast to the constant [DnaA-ATP]/[DnaA-ADP] in Δ4, the extrinsic conversion elements induce temporal modulations in [DnaA-ATP]/[DnaA-ADP] during cell elongation [46]. This ratio reaches its maximum at initiation and its minimum at termination due to the activation/deactivation of the RIDA mechanism [Fig. 5(b)] [47]. We also investigated the initiation stability across growth conditions [Fig. 5(c) and Appendix Fig. 3]. Among all the known extrinsic conversion elements we tested, the replication-dependent DnaA-ATP → DnaA-ADP by RIDA alone was sufficient to restore initiation stability [Fig. 5(c), also Appendix Fig. 3]. Other elements only had mild effects on the stability. RIDA is replication-dependent; thus, it immediately decreases the level of DnaA-ATP upon initiation. This reduction in the initiation-competent DnaA-ATP level is likely the reason for suppressing premature re-initiation. Although we found RIDA to be the initiation stabilizer, it still significantly delays initiation due to the reduced level of DnaA-ATP. Our simulations show that the delayed initiation can be alleviated by the other DnaA-ADP → DnaA-ATP conversion elements without causing instability [Fig. 5(c) and Appendix Fig. 3]. Interestingly, the initiation mass becomes nearly invariant across a wide range of growth conditions in the presence of all four extrinsic conversion elements [Fig. 5(c)], as long as the concentration [DnaA] is growth-condition independent. We previously used this growth-conditionin-dependent [DnaA] hypothesis to explain the invariance of initiation mass [36], and the data so far supports the hypothesis [32, 33]. Based on these results, we conclude that the replication-dependent DnaA-ATP → DnaA-ADP by RIDA can significantly enhance the initiation stability, and the other DnaA-ADP → DnaA-ATP conversion elements keep the initiation mass nearly constant against physiological perturbations. 4. The eclipse period or origin sequestration does not improve stability. We also tested the effect of the eclipse period [43] in our simulations (Appendix Fig. 4). During the predefined eclipse period, we did not allow the binding of the initiator to ori. Surprisingly, the eclipse period did not improve stability significantly in the multifork replication regime. However, the amplitude of the initiation mass oscillation decreased slightly (Appendix Fig. 4). Therefore, we predict the effect of SeqA on steady-state stabilization to be modest. 5. Comparison with previous modeling by Berger and ten Wolde and recent experimental work In their recent study, Berger and ten Wolde [11] conducted a thorough investigation into E. coli DNA replication. They utilized extensive numerical simulations that factored in the known dynamics between DnaA-ATP and DnaA-ADP conversion, as well as the aspects of DnaA titration. To our knowledge, Berger and ten Wolde were the first to suggest possible instability during multifork replication. Under relatively fast growth conditions (with the doubling time 35 mins and the C period 40 mins), their observations noted oscillations in the initiation mass between two distinct values, which occurred in the absence of DnaA-ATP ↔ DnaA-ADP conversion. Our instability phase diagram [Fig. 4(c)] explains this observation. For example, in the case of Δ4 mutant cells, the initiation mass should oscillate between two values when 1<C/τ<1.8 [Fig. 5(c)]. However, the complexity of these instability regimes needs to be noted. Our phase diagrams show that multifork replication does not always lead to instability [Fig. 5(c) and Appendix Fig. 2]. Berger and ten Wolde propose the DnaA-ATP ↔ DnaA-ADP conversion as the key mechanisms in initiation control, as DnaA-ATP ↔ DnaA-ADP conversion could avoid initiation instability in the absence of titration boxes in their simulations. By contrast, we favor that titration plays a more fundamental role in initiation control, because it is the protein counting device in the protocell and also the Δ4 cells, where DnaA-ATP ↔ DnaA-ADP conversion is absent. Furthermore, titration boxes, which are prevalent in bacteria, ensure synchronous initiation (as explained in Section E), and explain as to why bacteria produce significantly more DnaA molecules than necessary for ori. Albeit titration is fundamental in our model, its performance is not perfect in terms of initiation instability, and we demonstrated that RIDA is the key conversion element required for initiation stability when titration is in place. While the details of molecular effects on initiation are beyond the scope of this theory work, we suggest recent work by Elf and colleagues [48] and by us on various deletion mutants, including Δ4 [13], for single-cell level experimental investigation as confirmations of some of our predictions.. E. Asynchrony and cell-to-cell variability of initiation in the initiator-titration framework Initiation stability raises a related issue of stochasticity in initiation. In the systems biology literature, “noise” is mainly discussed in the context of stochasticity in gene expression, decomposed into “intrinsic” vs. “extrinsic” components [49–52]. In our view, there are parallel observations in replication initiation: the initiation asynchrony among ori’s within the same cell [13, 53, 54], and the cell-to-cell variability of the initiation mass [6, 13, 55], as illustrated in Fig. 6(a). In this section, we discuss their origins and statistical properties within our initiator-titration model v2 framework. 1. Definition of the intrinsic and extrinsic noise During overlapping cell cycles, the cell contains multiple replication origins at initiation. These origins share the same biochemical environment within one cell, so their initiation events are correlated; the initiation timing in different cells can vary because of stochasticity in biological processes, such as gene expression [49, 52]. On the other hand, since these origins in the same cell do not interact with each other, they can initiate asynchronously due to the innate stochasticity of initiator accumulation at origins [28, 53]. To quantify initiation asynchrony and cell-to-cell variability, we consider two overlapping replication cycles. Suppose the two ori’s initiate at initiation mass vi(1) and vi(2), respectively. Similar to stochastic gene expression [49], we can define the intrinsic noise and the extrinsic noise of the initiation mass by the coefficient of variation as (12) CVint2=(vi(1)−vi(2))22vi(1)vi(2),CVext2=vi(1)vi(2)−vi(1)vi(2)vi(1)vi(2). Note that this definition fulfills the relation CVtot2=CVint2+CVext2, where CVtot is the coefficient of variation of the single variable vi(1) or vi(2) (Appendix F). We use CVint as a measure of asynchrony. Visually, CVint describes the width of the off-diagonal axis of the ellipsoid, while CVext describes the elongation extent of the diagonal axis compared to the short axis [Fig. 6(a)]. For example, if CVext=0,vi(1) are vi(2) fully uncorrelated, and the ellipsoid becomes a circle. In this case, the intrinsic noise is the sole source of cell-to-cell variability. Generally, while asynchrony is fully determined by the intrinsic noise, the cell-to-cell variability is a result of both the intrinsic noise and the extrinsic noise (see Appendix F for details). 2. A first-passage-time (FPT) model based on a one-step Poisson process To study the behavior of the extrinsic noise and the intrinsic noise, we convert the initiation mass variables, vi(1) and vi(2), into the first passage time (FPT) variables [56], T(1) and T(2), respectively. That is, the initiator proteins bind to binding sites at ori, increasing its occupancy O(t), and initiate replication as soon as ori is fully saturated O(t)=nB. Although the relation between vi and FPT is nonlinear, to the zeroth-order approximation, we have (13) CVint2≈⟨(T(1)−T(2))2⟩2⟨T(1)⟩⟨T(2)⟩,CVext2≈⟨T(1)T(2)⟩−⟨T(1)⟩⟨T(2)⟩⟨T(1)⟩⟨T(2)⟩. To obtain the scaling law of the noise of FPT, we assume the production of initiator proteins as a Poisson process with a constant production rate β [52]. We further assume that all cells are characterized by the same set of physiological parameters without noise. (By this assumption, we are considering the lower bound of the extrinsic noise, and we discuss the contribution of parameter noises in Section E.4.) Let us first consider a simple scenario of initiation without initiator-titration. In this scenario, there is no chromosomal binding site; all nB binding sites are localized at each ori, and the initiator protein has an equal probability of binding to either ori. That is, the two ori’s accumulate the initiator proteins independently. This results in uncorrelated T(1) and T(2) and hence CVext=0 based on Eq. 13. The intrinsic noise then becomes (14) CVint=σT(1)⟨T(1)⟩, where ⟨T(1)⟩ is the mean FPT at ori1 and σT(1) is the standard deviation. In this simplest scenario, the accumulation at ori1 is a Poission process with a rate of β followed by a binomial trial with equal probability, leading to a Gamma distribution of T(1), with the mean ⟨T(1)⟩=2nB/β and the standard deviation σT(1)=2nB/β (see Appendix G). Thus, the CVint is independent of β [56, 57], (15) CVint=1nB=2N, where N=2nB is the mean total number of initiator proteins needed for triggering initiation at both ori’s (Appendix G). Therefore, in this one-step Poisson process, the intrinsic noise of FPT scales with the square root of the required total number of initiators N. If the number of binding sites at ori is nB≈10, we have CVint≈30% [Fig. 6(b)]. If the cell localizes all NB≈300 DnaA boxes at each ori to increase the threshold, the noise will decrease to CVint=1/300≈6% [Fig. 6(c)]. The reason for the 1/N intrinsic noise scaling is that the stochasticity in gene expression fully propagates to the initiation timing, and T(1) and T(2) are uncorrelated. As we explain below, E. coli suppresses the intrinsic noise using an ingenious two-step Poisson process by compressing T(1) and T(2) into a narrow range during the cell cycle using titration. In other words, titration of the initiator proteins redirects most gene expression noise to the extrinsic noise, effectively synchronizing T(1) and T(2). 3. A two-step Poisson process in the initiator-titration framework predicts the 1/N scaling of the intrinsic noise, leading to initiation synchrony. Due to the significant differences in the binding affinity between the chromosomal binding sites KD≈1nM and ori KD≈100nM, E. coli titrates DnaA sequentially in two steps: (1) saturation of the ~NB chromosomal DnaA boxes by DnaA-ATP and DnaA-ADP, followed by (2) accumulation of DnaA-ATP at ori with nB≪NB binding sites. Thus, we modify the one-step Poisson process by adding the titration step, namely, a two-step Poisson process [Fig. 6(c)]. The first step delays the accumulation processes at ori1 and ori2 and they synchronize their initiations, and the intrinsic noise (asynchrony) is a result of stochasticity in the second step. To analyze the two-step Poisson process, we rewrite the two FPT variables T(1) and T(2) as, T(1)=T(0)+ΔT(1),T(2)=T(0)+ΔT(2). Here, T(0) is the time required to saturate the chromosomal binding sites, whereas ΔT(1) and ΔT(2) denote the additional respective times for the two ori’s to accumulate the initiator proteins to trigger initiation. We assume that T(0),ΔT(1), and ΔT(2) are three independent stochastic variables. Specifically, T(0) follows the original Poisson process with an accumulation rate of β, while ΔT(1) and ΔT(2) each independently follows the same Poisson process with an accumulation rate of β/2 (initiator proteins produced at the rate β bind the two ori’s), as derived in Appendix G. By this decomposition, Eq. 13 can be rewritten as (16) CVint=σΔT(1)⟨T(1)⟩,CVext=σT(0)⟨T(1)⟩. According to the corresponding Gamma distributions, the mean FPT reads ⟨T(1)⟩=N/β, where N is the mean total number of initiator proteins needed for triggering initiation at both ori’s; the standard deviation of the first-step FPT reads σT(0)=N−2nB/β, and the standard deviation of the second-step FPT reads σΔT(1)=2nB/β (see Appendix G). Therefore, based on Eq. 16, we obtain the CV’s scaling law as (17) CVint=2nBN,CVext=N−2nBN≈1N. This result indicates that CVint decays in ~1/N, much faster than the total noise CVtot~1/N, and CVext becomes the dominant noise component when N is large. For example, if nB=10,NB≈300, and N≈2NB+nB≈600 (two-overlapping cell cycle), the noise of the two-step processes decreases dramatically from ~30% to only 1%, while the extrinsic noise is around 4%. To test the predictions of the two-step Poisson process, we conducted simulation by considering a Poissonian protein production followed by a partitioning among three destinations: chromosomal binding sites, ori1, and ori2 (see Appendix G for model settings). As shown in Fig. 6(d), the scaling behavior of the intrinsic noise and the extrinsic noise is consistent with Eq. 17. In summary, the chromosomal titration boxes effectively synchronize the accumulation of DnaA-ATP at multiple ori’s by titration, compressing their initiation timing into a narrow temporal window during the cell cycle [53, 58]. This is consistent with long-standing experimental observations of synchronous initiation of minichromosomes [59, 60], and more recent observations of ectopic chromosomal origins [61]. This improvement of precision by two sequential binding processes is reminiscent of the ratchet-like kinetic proofreading model, and our results are generalizable. 4. Other noise sources not quantified in this work In the previous section, we have mainly discussed asynchrony and cell-to-cell variability in initiation resulting from stochastic protein production, which predicts CVint≈1%, and CVext≈4% in E. coli. However, experimentally measured CVint is about 3% – 4% [13] and CVtot is about 10% [6, 13, 55], both larger than the prediction. For mutants lacking DnaA-ATP ↔ DnaA-ADP conversion elements, the cell-to-cell variability can increase up to 20% [13]. The likely sources of additional asynchrony and cell-to-cell variability are likely as follows. For the intrinsic noise, the initiator accumulation at the two ori’s can be negatively correlated because of the new round of replication. Once the first initiation event is triggered at one ori, the newly produced DnaA boxes will titrate DnaA, and the newly activated RIDA decreases the DnaA-ATP pool [Fig. 5(b)], further delaying the initiation of the second ori. This anti-correlation between two asynchronous initiations should increase the intrinsic noise CVint. For the extrinsic noise, we suggest two extra main sources other than the 1/N for titration by Poisson process (Eq. 17): (1) cell-to-cell variability in the initiator concentration [52, 62, 63], DnaA-ATP/ADP ratio, doubling time [9, 64], and C period [36, 65], and (2) the growth-condition (C/τ) dependent initiation instability discussed in Section D [Figs. 4(c) & 5(c)]. In principle, even without noises in C period and doubling time, instability can cause a bimodal distribution of the initiation mass that significantly increases the extrinsic noise [11]. The noise caused by initiation instability should be significant in mutants without the RIDA mechanism, such as the Δ4 cells [13]. For the quantification of these noise contributions, we leave a more detailed analysis to future work. III. CONCLUSION AND PERSPECTIVE In this work, we have provided a comprehensive quantitative explanation of how bacteria control the cell cycle under balanced growth, particularly focusing on replication initiation as a tractable problem. Our analysis builds upon the original initiator-titration model proposed by Hansen and colleagues [19], which offered valuable insights into the two-step initiation processes that trigger initiation. Over the past three decades, significant progress has been made in understanding the conserved master replication initiator protein, DnaA. One perplexing aspect has been the coexistence of two forms of DnaA (DnaA-ATP and DnaA-ADP), with only DnaA-ATP being initiation competent. Expanding upon the original model by Hansen and colleagues, we developed the initiator-titration model v2, which incorporates the two-state DnaA model and accounts for DnaA box distribution. We have derived an analytical expression for the initiation mass in terms of three mechanistic parameters for DnaA: its concentration, the average ratio [DnaA-ATP]/[DnaA-ADP], and the number of DnaA titration boxes (Eq. 11). However, through our dynamical stability analysis, we have also revealed a previously unexplored instability in initiation within this model [Fig. 4(c)], thereby elucidating recent observations from numerical simulations by Berger and ten Wolde [11]. We have demonstrated that the replication-dependent DnaA-ATP → DnaA-ADP conversion (by RIDA) alone restores initiation stability [66]. Additionally, when considering all extrinsic DnaA-ATP ↔ DnaA-ADP elements, the initiation mass remains remarkably invariant across a wide range of growth conditions, in agreement with experimental observations [33, 36, 67]. Moreover, we have discovered that the titration process of the chromosomal DnaA boxes suppresses the intrinsic noise or asynchrony in initiation by CV~1/N scaling. This finding represents a significant improvement over the naively expected standard coefficient of variation scaling CV~1/N for a Poisson process. It underscores the extraordinary consequences of the two-step initiation processes in the initiator-titration models, highlighting the remarkable precision achieved by bacteria. In conclusion, we propose that titration may have been a pivotal evolutionary milestone, acting as a protein-counting mechanism that co-evolved with balanced biosynthesis. This system would not only enable bacteria to homeostatically control their size via the adder principle, but also lead to synchronous initiation by effectively separating titration and initiation in two steps. Our results thus illuminate how bacteria employ a seemingly straightforward yet efficient titration-based strategy to address fundamental biological challenges. This differentiates them from eukaryotes, which use programmed gene expression and protein degradation to sense and control protein concentrations. While our findings focus on a specific case of initiation control, they also trigger intriguing questions about the potential pervasiveness of titration-based precision control in diverse biological systems. Uncovering additional examples of such mechanisms will significantly advance our overall understanding of precision control and pave the way for practical applications, including the design of synthetic cells. Supplementary Material Supplement 1 ACKNOWLEDGMENTS We thank Flemming Hansen, Tove Atlung, Tsutomu Katayama, Anders Lobner-Olesen, Godefroid Charbon, Thias Boesen, Johan Elf, Dongyang Li, Fangwei Si, Guillaume Le Treut, Cara Jensen, Mareike Berger, Pieter-Rein ten Wolde, Alan Leonard, Julia Grimwade, Conrad Woldringh, Charles Helmstetter, and Willie Donachie for many invaluable discussions and exchange of ideas over the years that inspired and helped shape the ideas presented in this work. This work was supported by NSF MCB-2016090 and NIH MIRA (R35GM139622) to SJ. FIG. 1. Protein concentration in eukaryotes vs. bacteria. (a) (left) morphogen gradient in the French flag model in developmental biology. (right) Oscillation of cyclin concentration for eukaryotic cell-cycle control. (b) Balanced biosynthesis in bacteria. FIG. 2. Initiation control. (a) Hypothetical minimal cell. (b) Initiator-titration model v1 [19]. (c) Initiator-titration model v2 (this work). FIG. 3. Initiation control of the protocell by initiator protein counting. (a) Model sequence of titration and initiation. (b) Change in the copy numbers of initiators and initiator binding sites during the cell cycle under the condition of two overlapping cell cycle (C<τ<C+D). The initiation condition is It=tini=Bt=tini. (c) Predicted initiation mass in different growth conditions (C/τ) by assuming that cI is a constant [32, 33]. FIG. 4. Dynamical stability analysis for initiation of the protocell. (a) Multifork replication tracker. i=1 represents the group of replication forks closest to ori. (b) Linear (in time) progression of the replication forks from ori to ter on the circular chromosome. ter is on the opposite end of ori on the chromosome. (c) Stability phase diagram (nB/NB vs. C/τ space). Below: initiation mass vs. C/τ in the condition of nB/NB=1/30 and constant cI [32, 33]. Inset: stable initiation events vs. unstable (oscillatory) initiation events. FIG. 5. Initiator-titration model v2 predictions (see Section D for more details). (a) External DnaA-ATP ↔ DnaA-ADP conversion elements in E. coli. RIDA is the only component that depends on the active replication forks. (b) [DnaA-ATP]/[DnaA-ADP] varies during the cell cycle predicted by computer simulations in the wildtype cells. By contrast, the Δ4 mutant that lacks all extrinsic DnaA-ATP ↔ DnaA-ADP conversion elements in (a) show constant ratio. (c) Predicted initiation mass in different growth conditions (C/τ with fixed C) by assuming a constant [DnaA] [32, 33]. RIDA, the replication-dependent DnaA-ATP → DnaA-ADP mechanism alone can restore stability as long as titration is present. None of the other extrinsic DnaA-ATP ↔ DnaA-ADP conversion elements can restore the initiation stability. FIG. 6. Initiation precision and the reduction in asynchrony by 1/N scaling in the two-step Poisson process by titration. (a) Asynchrony (intrinsic noise), extrinsic noise, and cell-to-cell variability in initiation control. Grey dots are single-cell data of wildtype E. coli from [13]. (b) Initiation by the first-passage time model based on a simple Poisson process with nB the threshold at each ori. (left) nB=10vs (right) nB=300. (c) Synchronized initiation by titration in the two-step Poisson process. N is the mean total number of initiator proteins required for triggering initiation at both ori’s. (d) Simulation of the scaling behavior of the intrinsic noise CVint, the extrinsic noise CVext, and the total CV CVtot (see Appendix G). Top: varing N with fixed nB; bottom: varing nB with fixed N. The grey dashed lines are from Eq. 17. ==== Refs [1] Wolpert L. , Positional information and the spatial pattern of cellular differentiation, J. Theor. Biol. 25 , 1 (1969).4390734 [2] Evans T. , Rosenthal E. T. , Youngblom J. , Distel D. , and Hunt T. , Cyclin: a protein specified by maternal mRNA in sea urchin eggs that is destroyed at each cleavage division, Cell 33 , 389 (1983).6134587 [3] Jun S. , Si F. , Pugatch R. , and Scott M. , Fundamental principles in bacterial physiology—history, recent progress, and the future with focus on cell size control: a review, Rep. Prog. Phys. 81 , 056601 (2018).29313526 [4] Schaechter M. , Maaløe O. , and Kjeldgaard N. O. , Dependency on medium and temperature of cell size and chemical composition during balanced growth of salmonella typhimurium, Microbiology 19 , 592 (1958). [5] Kjeldgaard N. O. , Maaloe O. , and Schaechter M. , The transition between different physiological states during balanced growth of salmonella typhimurium, J. Gen. Microbiol. 19 , 607 (1958).13611203 [6] Si F. , Le Treut G. , Sauls J. T. , Vadia S. , Levin P. A. , and Jun S. , Mechanistic origin of Cell-Size control and homeostasis in bacteria, Curr. Biol. 29 , 1760 (2019).31104932 [7] den Blaauwen T. , Hamoen L. W. , and Levin P. A. , The divisome at 25: the road ahead, Curr. Opin. Microbiol. (2017). [8] Campos M. , Surovtsev I. V. , Kato S. , Paintdakhi A. , Beltran B. , Ebmeier S. E. , and Jacobs-Wagner C. , A constant size extension drives bacterial cell size homeostasis, Cell 159 , 1433 (2014).25480302 [9] Taheri-Araghi S. , Bradde S. , Sauls J. T. , Hill N. S. , Levin P. A. , Paulsson J. , Vergassola M. , and Jun S. , Cell-Size control and homeostasis in bacteria, Curr. Biol. 25 , 385 (2015).25544609 [10] Jun S. and Taheri-Araghi S. , Cell-size maintenance: universal strategy revealed, Trends Microbiol. 23 , 4 (2015).25497321 [11] Berger M. and Wolde P. R. T. , Robust replication initiation from coupled homeostatic mechanisms, Nat. Commun. 13 , 6556 (2022).36344507 [12] Sompayrac L. and Maaloe O. , Autorepressor model for control of DNA replication, Nat. New Biol. 241 , 133 (1973).4573387 [13] Boesen T. , Charbon G. , Fu H. , Jensen C. , Li D. , Jun S. , and others, Robust control of replication initiation in the absence of DnaA-ATP DnaA-ADP regulatory elements in escherichia coli, bioRxiv (2022). [14] Leonard A. C. , Rao P. , Kadam R. P. , and Grimwade J. E. , Changing perspectives on the role of DnaA-ATP in orisome function and timing regulation, Front. Microbiol. 10 , 2009 (2019).31555240 [15] Hansen F. G. and Atlung T. , The DnaA tale, Front. Microbiol. 9 , 319 (2018).29541066 [16] Katayama T. , Kasho K. , and Kawakami H. , The DnaA cycle in escherichia coli: Activation, function and inactivation of the initiator protein, Front. Microbiol. 8 , 2496 (2017).29312202 [17] Riber L. , Frimodt-Møller J. , Charbon G. , and Løbner-Olesen A. , Multiple DNA binding proteins contribute to timing of chromosome replication in e. coli, Front Mol Biosci 3 , 29 (2016).27446932 [18] Katayama T. , Ozaki S. , Keyamura K. , and Fujimitsu K. , Regulation of the replication cycle: conserved and diverse regulatory systems for DnaA and oric, Nat. Rev. Microbiol. 8 , 163 (2010).20157337 [19] Hansen F. G. , Christensen B. B. , and Atlung T. , The initiator titration model: computer simulation of chromosome and minichromosome control, Res. Microbiol. 142 , 161 (1991).1925015 [20] Katayama T. , Kubota T. , Kurokawa K. , Crooke E. , and Sekimizu K. , The initiator function of DnaA protein is negatively regulated by the sliding clamp of the e. coli chromosomal replicase, Cell 94 , 61 (1998).9674428 [21] Kato J. and Katayama T. , Hda, a novel DnaA-related protein, regulates the replication cycle in escherichia coli, EMBO J. 20 , 4253 (2001).11483528 [22] Leonard A. C. and Grimwade J. E. , The orisome: structure and function, Front. Microbiol. 6 , 545 (2015).26082765 [23] Sekimizu K. , Bramhill D. , and Kornberg A. , ATP activates dnaa protein in initiating replication of plasmids bearing the origin of the e. coli chromosome, Cell 50 , 259 (1987).3036372 [24] Nishida S. , Fujimitsu K. , Sekimizu K. , Ohmura T. , Ueda T. , and Katayama T. , A nucleotide switch in the escherichia coli DnaA protein initiates chromosomal replication: evidence from a mutant DnaA protein defective in regulatory ATP hydrolysis in vitro and in vivo, J. Biol. Chem. 277 , 14986 (2002).11839737 [25] McGarry K. C. , Ryan V. T. , Grimwade J. E. , and Leonard A. C. , Two discriminatory binding sites in the escherichia coli replication origin are required for DNA strand opening by initiator DnaA-ATP, Proc. Natl. Acad. Sci. U. S. A. 101 , 2811 (2004).14978287 [26] Schaper S. and Messer W. , Interaction of the initiator protein DnaA of escherichia coli with its DNA target, J. Biol. Chem. 270 , 17622 (1995).7615570 [27] Kawakami H. , Keyamura K. , and Katayama T. , Formation of an ATP-DnaA-specific initiation complex requires DnaA arginine 285, a conserved motif in the AAA+ protein family, J. Biol. Chem. 280 , 27420 (2005).15901724 [28] Herrick J. , Kohiyama M. , Atlung T. , and Hansen F. G. , The initiation mess?, Mol. Microbiol. 19 , 659 (1996).8820637 [29] Buckstein M. H. , He J. , and Rubin H. , Characterization of nucleotide pools as a function of physiological state in escherichia coli, J. Bacteriol. 190 , 718 (2008).17965154 [30] Hsieh L. S. , Rouviere-Yaniv J. , and Drlica K. , Bacterial DNA supercoiling and [ATP]/[ADP] ratio: changes associated with salt shock, J. Bacteriol. 173 , 3914 (1991).1646791 [31] More precisely, there are 11 DnaA boxes are clustered within ori, with three high-affinity sites and eight low-affinity sites. Only DnaA-ATP can bind to the eight low-affinity sites. For simplicity, we used “10” to denote the number of low-affinity binding sites at ori throughout this paper. [32] Hansen F. G. , Atlung T. , Braun R. E. , Wright A. , Hughes P. , and Kohiyama M. , Initiator (DnaA) protein concentration as a function of growth rate in escherichia coli and salmonella typhimurium, J. Bacteriol. 173 , 5194 (1991).1860829 [33] Zheng H. , Bai Y. , Jiang M. , Tokuyasu T. A. , Huang X. , Zhong F. , Wu Y. , Fu X. , Kleckner N. , Hwa T. , and Liu C. , General quantitative relations linking cell growth and the cell cycle in escherichia coli, Nat Microbiol 5 , 995 (2020).32424336 [34] Pham T. M. , Tan K. W. , Sakumura Y. , Okumura K. , Maki H. , and Akiyama M. T. , A single-molecule approach to DNA replication in escherichia coli cells demonstrated that DNA polymerase III is a major determinant of fork speed, Mol. Microbiol. 90 , 584 (2013).23998701 [35] Bhat D. , Hauf S. , Plessy C. , Yokobayashi Y. , and Pigolotti S. , enCorrection: Speed variations of bacterial replisomes, Elife 12 (2023). [36] Si F. , Li D. , Cox S. E. , Sauls J. T. , Azizi O. , Sou C. , Schwartz A. B. , Erickstad M. J. , Jun Y. , Li X. , and Jun S. , Invariance of initiation mass and predictability of cell size in escherichia coli, Curr. Biol. 27 , 1278 (2017).28416114 [37] Cooper S. and Helmstetter C. E. , Chromosome replication and the division cycle of escherichia colibr, J. Mol. Biol. (1968). [38] Kitagawa R. , Mitsuki H. , Okazaki T. , and Ogawa T. , A novel DnaA protein-binding site at 94.7 min on the escherichia coli chromosome, Mol. Microbiol. 19 , 1137 (1996).8830270 [39] Kasho K. and Katayama T. , DnaA binding locus datA promotes DnaA-ATP hydrolysis to enable cell cycle-coordinated replication initiation, Proceedings of the National Academy of Sciences 110 , 936 (2013). [40] Fujimitsu K. , Senriuchi T. , and Katayama T. , Specific genomic sequences of e. coli promote replicational initiation by directly reactivating ADP-DnaA, Genes Dev. 23 , 1221 (2009).19401329 [41] Kasho K. , Fujimitsu K. , Matoba T. , Oshima T. , and Katayama T. , Timely binding of IHF and fis to DARS2 regulates ATP-DnaA production and replication initiation, Nucleic Acids Res. 42 , 13134 (2014).25378325 [42] Moolman M. C. , Krishnan S. T. , Kerssemakers J. W. J. , van den Berg A. , Tulinski P. , Depken M. , Reyes-Lamothe R. , Sherratt D. J. , and Dekker N. H. , Slow unloading leads to DNA-bound β2-sliding clamp accumulation in live escherichia coli cells, Nat. Commun. 5 , 5820 (2014).25520215 [43] Lu M. , Campbell J. L. , Boye E. , and Kleckner N. , SeqA: a negative modulator of replication initiation in e. coli, Cell 77 , 413 (1994).8011018 [44] Hansen F. G. , Christensen B. B. , Nielsen C. B. , and Atlung T. , Insights into the quality of DnaA boxes and their cooperativity, J. Mol. Biol. 355 , 85 (2006).16298387 [45] Rozgaja T. A. , Grimwade J. E. , Iqbal M. , Czerwonka C. , Vora M. , and Leonard A. C. , Two oppositely oriented arrays of low-affinity recognition sites in oric guide progressive binding of DnaA during escherichia coli pre-RC assembly, Mol. Microbiol. 82 , 475 (2011).21895796 [46] Donachie W. D. and Blakely G. W. , Coupling the initiation of chromosome replication to cell size in escherichia coli, Curr. Opin. Microbiol. 6 , 146 (2003).12732304 [47] Kurokawa K. , Nishida S. , Emoto A. , Sekimizu K. , and Katayama T. , Replication cycle-coordinated change of the adenine nucleotide-bound forms of DnaA protein in escherichia coli, EMBO J. 18 , 6642 (1999).10581238 [48] Knöppel A. , Broström O. , Gras K. , Elf J. , and Fange D. , Regulatory elements coordinating initiation of chromosome replication to the Escherichia coli cell cycle, Proceedings of the National Academy of Sciences 120 , e2213795120 (2023). [49] Elowitz M. B. , Levine A. J. , Siggia E. D. , and Swain P. S. , Stochastic gene expression in a single cell, Science 297 , 1183 (2002).12183631 [50] Thattai M. and van Oudenaarden A. , Intrinsic noise in gene regulatory networks, Proc. Natl. Acad. Sci. U. S. A. 98 , 8614 (2001).11438714 [51] Paulsson J. , Summing up the noise in gene networks, Nature 427 , 415 (2004).14749823 [52] Taniguchi Y. , Choi P. J. , Li G.-W. , Chen H. , Babu M. , Hearn J. , Emili A. , and Sunney Xie X. , Quantifying e. coli proteome and transcriptome with Single-Molecule sensitivity in single cells, Science (2010). [53] Skarstad K. , Boye E. , and Steen H. B. , Timing of initiation of chromosome replication in individual escherichia coli cells, EMBO J. 5 , 1711 (1986).3527695 [54] Løbner-Olesen A. , Skarstad K. , Hansen F. G. , von Meyenburg K. , and Boye E. , The DnaA protein determines the initiation mass of escherichia coli K-12, Cell 57 , 881 (1989).2541928 [55] Sauls J. T. , Cox S. E. , Do Q. , Castillo V. , Ghulam-Jelani Z. , and Jun S. , Control of bacillus subtilis replication initiation during physiological transitions and perturbations, MBio 10 (2019). [56] Pandey P. P. , Singh H. , and Jain S. , Exponential trajectories, cell size fluctuations, and the adder property in bacteria follow from simple chemical dynamics and division control, Phys. Rev. E 101 , 062406 (2020).32688579 [57] Ghusinga K. R. , Dennehy J. J. , and Singh A. , First-passage time approach to controlling noise in the timing of intracellular events, Proceedings of the National Academy of Sciences 114 , 693 (2017). [58] Wallden M. , Fange D. , Lundius E. G. , Baltekin Ö. , and Elf J. , The synchronization of replication and division cycles in individual e. coli cells, Cell 166 , 729 (2016).27471967 [59] Leonard A. C. and Helmstetter C. E. , Cell cycle-specific replication of escherichia coli minichromosomes, Proc. Natl. Acad. Sci. U. S. A. 83 , 5101 (1986).3523483 [60] Helmstetter C. E. and Leonard A. C. , Coordinate initiation of chromosome and minichromosome replication in escherichia coli, J. Bacteriol. 169 , 3489 (1987).3301802 [61] Wang X. , Lesterlin C. , Reyes-Lamothe R. , Ball G. , and Sherratt D. J. , Replication and segregation of an escherichia coli chromosome with two replication origins, Proc. Natl. Acad. Sci. U. S. A. 108 , E243 (2011).21670292 [62] Salman H. , Brenner N. , Tung C.-K. , Elyahu N. , Stolovicki E. , Moore L. , Libchaber A. , and Braun E. , Universal protein fluctuations in populations of microorganisms, Phys. Rev. Lett. 108 , 238105 (2012).23003996 [63] Wang P. , Robert L. , Pelletier J. , Dang W. L. , Taddei F. , Wright A. , and Jun S. , Robust growth of escherichia coli, Curr. Biol. 20 , 1099 (2010).20537537 [64] Schaechter M. , Williamson J. P. , Hood J. R. , Jr, and A. L. Koch, Growth, cell and nuclear divisions in some bacteria, J. Gen. Microbiol. 29 , 421 (1962).13976593 [65] Le Treut G. , Si F. , Li D. , and Jun S. , Quantitative examination of five stochastic Cell-Cycle and Cell-Size control models for escherichia coli and bacillus subtilis, Front. Microbiol. 12 , 721899 (2021).34795646 [66] Riber L. , Olsson J. A. , Jensen R. B. , Skovgaard O. , Dasgupta S. , Marinus M. G. , and Løbner-Olesen A. , Hda-mediated inactivation of the DnaA protein and dnaa gene autoregulation act in concert to ensure homeostatic maintenance of the escherichia coli chromosome, Genes Dev. 20 , 2121 (2006).16882985 [67] Donachie W. D. , Relationship between cell size and time of initiation of DNA replication, Nature 219 , 1077 (1968).4876941
PMC010xxxxxx/PMC10251734.txt
==== Front iScience iScience iScience 2589-0042 Elsevier S2589-0042(23)01162-8 10.1016/j.isci.2023.107085 107085 Article A nanobody recognizes a unique conserved epitope and potently neutralizes SARS-CoV-2 omicron variants Modhiran Naphak 1217 Lauer Simon Malte 317 Amarilla Alberto A. 117 Hewins Peter 417 Lopes van den Broek Sara Irene 5 Low Yu Shang 1 Thakur Nazia 67 Liang Benjamin 1 Nieto Guillermo Valenzuela 8 Jung James 1 Paramitha Devina 1 Isaacs Ariel 1 Sng Julian D.J. 1 Song David 4 Jørgensen Jesper Tranekjær 910 Cheuquemilla Yorka 8 Bürger Jörg 311 Andersen Ida Vang 59 Himelreichs Johanna 8 Jara Ronald 8 MacLoughlin Ronan 12 Miranda-Chacon Zaray 7 Chana-Cuevas Pedro 13 Kramer Vasko 14 Spahn Christian 3 Mielke Thorsten 11 Khromykh Alexander A. 115 Munro Trent 2 Jones Martina L. 2 Young Paul R. 1215 Chappell Keith 1215 Bailey Dalan 6 Kjaer Andreas 910 Herth Matthias Manfred 59 Jurado Kellie Ann 4 Schwefel David [email protected] 3∗ Rojas-Fernandez Alejandro [email protected] 716∗∗ Watterson Daniel [email protected] 121518∗∗∗ 1 School of Chemistry and Molecular Bioscience, the University of Queensland, Brisbane, QLD, Australia 2 Australian Institute for Bioengineering and Nanotechnology, Brisbane, QLD, Australia 3 Institute of Medical Physics and Biophysics, Charité – Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin and Humboldt Universität zu Berlin, Berlin, Germany 4 Department of Microbiology, University of Pennsylvania, Philadelphia, PA, USA 5 Department of Drug Design and Pharmacology, Faculty of Health and Medical Sciences, University of Copenhagen, Jagtvej 160, 2100 Copenhagen, Denmark 6 The Pirbright Institute, Ash Road, Guildford, UK 7 Nuffield Department of Medicine, University of Oxford, Oxford, UK 8 Institute of Medicine, Faculty of Medicine & Center for Interdisciplinary Studies on the Nervous System, CISNE, Universidad Austral de Chile, Valdivia, Chile 9 Department of Clinical Physiology, Nuclear Medicine & PET, Copenhagen University Hospital, Rigshospitalet, Blegdamsvej 9, 2100 Copenhagen, Denmark 10 Cluster for Molecular Imaging, Department of Biomedical Sciences, University of Copenhagen, Blegdamsvej 3, 2200 Copenhagen, Denmark 11 Microscopy and Cryo-Electron Microscopy Service Group, Max-Planck-Institute for Molecular Genetics, Berlin, Germany 12 Research and Development, Science and Emerging Technologies, Aerogen Limited, Galway Business Park, H91 HE94 Galway, Ireland 13 CETRAM & Faculty of Medical Science Universidad de Santiago de Chile, Chile 14 PositronPharma SA, Rancagua 878, 7500921 Providencia, Santiago, Chile 15 Australian Infectious Diseases Research Centre, Global Virus Network Centre of Excellence, Brisbane, QLD, Australia 16 Berking Biotechnology, Valdivia, Chile ∗ Corresponding author [email protected] ∗∗ Corresponding author [email protected] ∗∗∗ Corresponding author [email protected] 17 These authors contributed equally 18 Lead contact 09 6 2023 21 7 2023 09 6 2023 26 7 1070859 2 2023 12 4 2023 6 6 2023 © 2023. 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Summary The severe acute respiratory syndrome coronavirus 2 (SARS-CoV2) Omicron variant sub-lineages spread rapidly worldwide, mostly due to their immune-evasive properties. This has put a significant part of the population at risk for severe disease and underscores the need for effective anti-SARS-CoV-2 agents against emergent strains in vulnerable patients. Camelid nanobodies are attractive therapeutic candidates due to their high stability, ease of large-scale production, and potential for delivery via inhalation. Here, we characterize the receptor binding domain (RBD)-specific nanobody W25 and show superior neutralization activity toward Omicron sub-lineages in comparison to all other SARS-CoV2 variants. Structure analysis of W25 in complex with the SARS-CoV2 spike glycoprotein shows that W25 engages an RBD epitope not covered by any of the antibodies previously approved for emergency use. In vivo evaluation of W25 prophylactic and therapeutic treatments across multiple SARS-CoV-2 variant infection models, together with W25 biodistribution analysis in mice, demonstrates favorable pre-clinical properties. Together, these data endorse W25 for further clinical development. Graphical abstract Highlights • The nanobody W25 efficiently neutralizes SARS-CoV2 omicron sub-lineages • A dimeric W25-Fc fusion shows enhanced neutralization activity • W25 binds a unique conserved SARS-CoV2 spike RBD epitope • W25 protects K18-huACE2 mice from Wuhan and Omicron SARS-CoV-2 infections Public health; Information system model; Decision science Subject areas Public health Information system model Decision science Published: June 9, 2023 ==== Body pmcIntroduction Since emergence in late 2019, the highly transmissible coronavirus—severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)—has infected more than 300 million individuals and claimed more than 6.5 million lives (WHO, updated September 2022). Despite the rapid development of vaccine strategies and preventive measures that help curb disease progression, viral transmission persists, and viral variants with altered virulence and antigenic sites have emerged in successive waves across the globe. Understanding the antigenic nature of emerging variants and the development of broad-spectrum therapeutics and vaccines remain an urgent priority. SARS-CoV-2 expresses a surface spike (S) glycoprotein, which consists of S1 and S2 subunits that form a homotrimeric viral spike which serves as both viral attachment protein and membrane fusogen.1 Cell receptor interaction is mediated by the S1 receptor-binding domain (RBD), which binds the peptidase domain of human angiotensin-converting enzyme 2 (hACE2). Structural studies have revealed different conformations of the spike from both in vitro and in situ.2,3,4 In the prefusion stage, the RBD switches between a closed and an open conformation to facilitate hACE2 interaction.5 Subsequently, the S2 undergoes a substantial conformational change which drives the fusion of viral and cellular membranes.1 Investigations of sera from COVID-19 convalescent patients enabled isolation of potent neutralizing antibodies (NAbs), the majority of which target the RBD.6,7,8,9,10,11,12,13 Some of these have received Emergency Use Authorization (EUA) from the U.S. Food & Drug Administration for SARS-CoV2 treatment or pre-exposure prophylaxis and have shown to be effective in treating certain patients with COVID-19 (https//ww.fda.gov). However, virus evolution causes ongoing emergence of novel SARS-CoV2 variants of concern (VOCs), concomitant with reduction of vaccine14,15 and therapeutic antibody efficacy.16,17,18 The latest SARS-CoV-2 VOC B.1.1.529, designated Omicron (O), was first reported to WHO from South Africa at the end of November 2021 and is rapidly spreading across many countries, thereby replacing the already highly transmissible SARS-CoV-2 Delta variant (B.1.617.2).19 SARS-CoV-2 Omicron accumulated a large number of mutations compared to its earlier pandemic variants, of which >30 substitutions, deletions, or insertions are located in the spike protein. Scientists all over the world are unanimously reporting significantly reduced efficacies of vaccine-elicited sera against Omicron.20,21,22 To the best of our knowledge, the majority of potent monoclonal NAbs, including EUA NAbs, showed strongly reduced or no detectable neutralization activity toward Omicron. A notable exception is EUA NAb sotrovimab (S309), which however still exhibits 2- to 3-fold reduced neutralizing activity.23 Accordingly, the identification of novel, potent anti-Omicron NAbs would be highly desirable to strengthen the therapeutic repertoire. Nanobodies are considered the smallest antigen-binding entity, representing the variable heavy-chain fragment (VHH) derived from heavy-chain-only antibodies found in camelids (llamas, alpacas, guanacos, vicuñas, dromedary, and camels). Because of their small size (2.5 nm by 4 nm; 12–15 kDa) and unique binding domains, nanobodies offer many advantages over conventional antibodies including the ability to bind cryptic epitopes not accessible to bulky conventional antibodies, high tissue permeability, ease of production, and thermostability. Due to their superior stability, nanobodies are highly suited for development as potential bio-inhaled therapies against respiratory diseases. ALX-0171, an inhaled anti-respiratory syncytial virus (RSV) nanobody, demonstrated robust antiviral effect, reduced symptoms of virus infection in animal models,24,25 and displayed promising result in reducing RSV viral load.26 We recently described the neutralizing RBD-specific nanobody,27 W25, derived from an alpaca immunized with the S protein from the ancestral SARS-CoV-2 strain (Wuhan, Wu). Here, we extend these early findings and show that unlike all hitherto described NAbs, W25 neutralizes the Omicron variant even more potently than the ancestral isolate. In order to understand the molecular basis of neutralization activity and breadth of W25, we determined the cryoelectron microscopy (cryo-EM) structures of W25 in complex with SARS-CoV-2 spike from both ancestral and Omicron SARS-CoV-2, revealing that the vast majority of mutations in the Omicron variant RBD are outside of the W25-RBD binding interface. In addition, W25-binding and live virus neutralization assays demonstrated activity against a broad range of SARS-CoV-2 VOCs. Functional assays reveal that W25 triggers fusion and that this mechanism is conserved across VOCs, including the less fusogenic Omicron variant. Furthermore, we demonstrate a protection and robust reduction of viral burden and prevention of lung pathology in the K18-hACE2 mouse model of SARS-CoV-2 challenged with VOC Beta from both prophylactic and therapeutic administration of human immunoglobulin G (IgG) Fc-nanobody fusion, W25-Fc. Importantly, W25-Fc treatment results in survival of K18-hACE2 mouse challenged with lethal dose of Beta SARS-CoV-2. Finally, we analyzed the pharmacodynamics of radiolabeled W25-Fc in mice to gain a better understanding of the pharmacological characteristics of W25-Fc to inform further drug development efforts. Results W25 ultra-potently neutralizes the SARS-CoV2 omicron variant We have previously reported the neutralization activity of W25-Fc against SARS-CoV-2 Wu and D614G.27 Given the number of substitutions in the Omicron BA.1 spike protein, including G339D, S371L, S373P, S375F, K417N, N440K, G446S, S477N, T478K, E484A, Q493R, G496S, Q498R, N501Y, and Y505H and BA.2 spike including G339D, S371F, S373P, S375F, T376A, D405N, R408S, K417N, N440K, S477N, T478K, E484A, Q493R, N501Y, and Y505H in the RBD, neutralization activity of W25-Fc was tested against SARS-CoV-2 Omicron BA.1 and BA.2 live virus on Vero E6 cells.28 Notably, W25-Fc completely and potently neutralized the Omicron BA.1 and BA.2 variant, with an IC50 (half maximal inhibitory concentration) of 1.45 nM ± 0.31 nM and 2.07 nM ± 0.66 nM, respectively, which is ∼7-fold more potent compared to the ancestral strain (Figures 1A–1C). In comparison, EUA monoclonal antibodies (mAbs) including REGN10933, REGN10987, LY-CoV555, and CT-P59 significantly or completely lost neutralizing activity against Omicron BA.1 and BA.2. Other RBD-specific mAbs including CB6 and DH1047 displayed a significant reduction of neutralizing activity against Omicron BA.1 variant, by ∼4- and 7-fold compared to the Wu strain, respectively (Figures 1A–1C), and DH1047 completely lost neutralizing activity against BA.2. Interestingly, the neutralization profiles demonstrated that while W25-Fc exhibited complete inhibition of Wu-1 and Omicron infection, S309, a broadly neutralizing mAb, did not reach 100% neutralization, as seen previously29 (Figure 1A).Figure 1 Authentic virus neutralization of SARS-CoV-2 variants by W25, EUA, and other mAbs (A) Neutralization curves comparing the sensitivity of SARS-CoV-2 strains Wu (gray line) and Omicron BA.1 (red line) and Omicron BA.2 (blue line) to antibody as indicated including W25-Fc, C05, CB6, REGN10933, CT-P59, LY-CoV555, S309, REGN10987, and DH1047. The data were analyzed and plotted using nonlinear regression (curve fit, three parameters). (B and C) Comparison of IC50 values across mAbs tested between Wu and Omicron. IC50 was calculated from the neutralization curve by Graphpad Prism 8 software. Color represents different mAbs as indicated. Data are generated from two independent experiments, each performed in technical triplicate. Data are represented as mean ± SEM. Structural analysis of W25 binding to SARS-CoV-2 spike To gain mechanistic understanding of how W25 binds SARS-CoV-2 Wu and Omicron spikes, we determined their cryo-EM structures in complex with W25 (Table S1). After extensive 2D and 3D classification, two major conformational states of the Wu spike/W25 complex were identified (Figure S1). Class 1 spike trimers displayed two RBDs in an upward-facing position, with the third one in “down” position, while class 2 trimers possessed one “up” RBD and two “down” RBDs (Figure S1). For both classes, cryo-EM density maps showed additional density adjacent to the “up” RBDs, not covered by the structural model of the spike trimer, strongly suggesting that these density portions correspond to W25 (Figure 2A, upper panel, and S2). For the Omicron spike/W25 complex, multiple rounds of 2D and 3D classification identified “3 down”, “2 up, 1 down”, and “1 up, 2 down” RBD classes (Figure S3). The latter state displayed clear extra density corresponding to W25 in similar position as for the Wu spike (Figure 2A, lower panel).Figure 2 Structure analysis of the W25/spike interaction (A) Cryo-EM densities of Wu (upper panel) and Omicron (lower panel) spike/W25 complexes (gray semi-transparent surfaces) (map 1, Figure S1 and map 4, Figure S3). Fitted trimeric SARS-COV2 spike protein structures (PDB: 6ZXN (Wu) and PBD: 7WG6 (Omicron)30,31 are shown in cartoon representation, with protomer A colored purple, protomer B red, and protomer C cyan. W25 is shown as green cartoon. Densities corresponding to W25 are colored green. (B) Three views of a focused refinement of the regions illustrated in A. Proteins are shown in cartoon representation, and N-linked glycans as sticks, in the same colors as in A. Cryo-EM densities (map 3, Figure S1, and map 5, Figure S4) are shown as semi-transparent light-gray surfaces. (C) Detailed “open book” view of the W25/RBD interaction interface between RBD and W25. Amino acids involved in intermolecular interaction are shown as sticks. (D) Visualization of the RBD surface in contact with W25 (yellow). The RBD is shown as red cartoon. Residues mutated in the Omicron RBD are shown as spheres, and colored light blue when located outside of the W25 contact surface, or dark blue when inside. (E) Two views of a superposition of selected RBD-bound NABs with the RBD/W25 complex, in cartoon representation (CB6 – PDB: 7C01, LY-CoV555 – PDB: 7l3N, S309 – PDB: 7R6W, CR3022 – PDB: 6YLA). W25 is shown as semi-transparent green molecular surface. For clarity, only the W25-bound RBD is shown as red cartoon. The inset shows an enlarged view of the unique portion of the W25 epitope as yellow surface. Focused refinement of the Wu and Omicron spike RBDs, the N-terminal domain (NTD) regions of the adjoining spike protomer, and W25 resulted in contiguous density maps of sufficient quality for flexible fitting and refinement of a W25 structure model generated by AlphaFold232,33 (Figure 2B, Table S1 and Video S1). W25 binds to the side and upper edge of the RBD, with the downward-facing side of W25 surrounded by N-linked glycans arising from spike protomer B NTD N122 and N165 (Figure 2B). Since W25 positions relative to Wu or Omicron RBDs were equivalent and the map quality and resolution of the Wu spike/W25 complex were slightly better (Figures 2B and S1 and S3), the Wu spike/W25 model was subjected to further structural analysis. Superposition of the RBD/W25 complex structure with the Wu spike class 1 “down” RBD indicates that W25 binding would lead to a slight overlap with the adjacent spike NTD, rationalizing why W25 preferentially binds to RBDs in the “up” conformation (Figure S4A). Video S1. The movie illustrates the cryo-EM structure of the Wu spike/W25 complex, including a close-up of the RBD/W25 molecular model, highlighting the interaction interface, related to Figure 2A Detailed inspection of the RBD/W25 contact surface reveals an interface area of 847 Å2 with mostly hydrophobic character, involving W25 β-sheet 3 CDR1 residues Y34 and A35 and downstream amino acids H37 and F39 (Figure 2C). In addition, there are interfacing residues in the loop region connecting β-sheets 3 and 4, which include CDR2 (M41, R47, F49, F53, G54, N60, A62, Y61, A63, K66), and ultimately residues in β-sheet 7 (CDR3): H98, L100, and L105-W111. From the RBD side, amino acid residues in an N-terminal loop region contribute to W25 binding (T345-S349, Y351, A352, N354), as well as from the loop downstream of β-sheet 3 (Y449-L452) and from a large loop preceding β-sheet 4 (R466, I468, T470-I472, G482-E484, F490, S494). In addition, there are hydrogen bonds between W25 side chains D109 and W111 and RBD side chain S349 and the backbone carbonyl group of N450, as well as between RBD side chains N354 and T470 and backbone carbonyls of W25 F53 and L105. Twenty-two out of 23 epitope residues are conserved between Wu and Omicron. Only one interfacing residue is non-conservatively exchanged between Wu and Omicron SARS-CoV2 (E484A) (Figure 2D, S4B and Video S1). Comparison with RBD binding modes of other nanobodies, NAbs, and ACE2 In order to put the present structure analysis in context, PDB entries containing camelid or synthetic SARS-CoV2-specific nanobodies were collated and structurally aligned on their RBDs, together with the RBD/W25 complex (Figure S4C and Table S2). The vast majority of nanobodies attach either to one side of the RBD (“side-1” or cluster 1 in34) or to the upper RBD surface (“top” or cluster 2). Two of the nanobodies bind to an RBD region opposite of cluster 1 (“side-2”). Interestingly, the W25 binding mode seems to incorporate characteristics of two of these clusters by occupying a surface area overlapping with epitopes of both “top” and “side-2” binders. In addition, structures of the variable fragments of selected NAbs, representing distinct RBD binding modes (classes 1–4),35 were superimposed on the W25/RBD complex structure (Figure 2E). W25 partially overlaps with the VH fragment of class 2 NAb LY-CoV555, resulting in partially shared epitopes involving RBD amino acid residues 351, 449–453, and 480–490. Furthermore, W25 minimally clashes with loop regions of class 3 NAb S309 VH (residues 102–109) and VL (residue 94); their respective epitopes are adjacent but do not overlap. Importantly, a significant portion of the W25 epitope, including surface-exposed RBD amino acid stretches 348–354 and 466–471, is unique and not covered by the CDRs of the representative class 1–4 NAbs analyzed here (Figure 2E, right panel). Subsequently, structural alignment of the RBD moieties of the RBD/W25 complex presented here with the structure of the RBD bound to its human cell surface receptor ACE2 was performed. The analysis demonstrates that there is a slight overlap of the molecular surfaces (Figure S4D). In this hypothetical model, W25 side chain E46 directly opposes ACE2 residue E35, which might induce electrostatic repulsion and lead to obstruction of the ACE2-binding site or ACE2 displacement from the RBD (Figure S4E). W25 binds multiple VOCs of SARS-CoV-2 and retains binding to BA.2 Omicron sublineage To evaluate W25 binding to previous SARS-CoV-2 VOCs and the rapidly expanding BA.2 Omicron sublineage, we expressed and purified Wu and VOC spikes (bearing mutations indicated in Figure S5A) and validated spike purity and trimer formation by SDS-PAGE (Figure S5B) and size exclusion chromatography (SEC) (Figure S5C). Binding kinetics of W25-Fc to the spike variant proteins were assessed by surface plasmon resonance (SPR) (Figure 3A). Subnanomolar dissociation constants (KD) were observed for Wu, Alpha, Beta, Gamma, and Omicron BA.1 using a 1:1 kinetic model (Figure 3B). The KD values of W25-Fc for Delta and Lambda variants were higher compared to other variants, at 11.4 and 0.7 nM, respectively (Figures 3A and 3B). As control, S309 and DH1047 NAbs also yielded subnanomolar KD values (Figures 3B and S6). ELISA was consistent with these findings (Figures 3C and S7A), showing that W25 binds all variants with similar KD (0.033–0.048 nM), except for the Delta variant (1.8 nM). W25 still retains binding affinity to XBB1.5 spike (Kd value = 0.1 nM) (Figure S7D). In agreement with recent reports,36,37 and our neutralization assays, EUA RBD-specific mAbs including REGN10933, LY-CoV555, and CT-P59 showed significantly reduced binding affinity (>10-fold reduction of KD) for Omicron BA.1, with the exception of S309 (Figure 3C). We observed no binding of REGN10987 to Omicron BA.1 variant. Structurally defined cross-reactive mAbs including DH1047 (RBD-specific NAb), CR3022, and 2M10B11 (non-neutralizing cryptic RBD epitopes) showed significantly reduced binding to Omicron BA.1 variant (∼5-fold).Figure 3 Broad reactivity of W25-Fc is conferred by conserved patch on SARS-CoV-2 RBD and potential inhibition mechanism of W25 (A) SPR sensograms of W25-Fc. W25-Fc was immobilized on SPR protein A chips. Various concentrations of SARS-CoV-2 spike variant proteins as indicated were injected for 30 s, at 30 μL/min followed by dissociation for 600 s. Dissociation constants (KD) were determined on the basis of fits, applying a 1:1 interaction model. Similar experiments were conducted for S309, DH1047, and C05 (as control, Figure S6). (B) Summary table showing KD, Ka,Kd of indicated Nb/mAbs. (C) Kd values from ELISA binding curves of W25-Fc, EUA mAbs, and other epitope-specific mAbs to SARS-CoV-2 spikes variants. (D) Neutralization curves comparing the sensitivity of live SARS-CoV-2 viruses including Alpha, Beta, Delta, Kappa, Lambda, and Gamma to W25-Fc, EUA, and other RBD-specific mAbs as indicated. The data were analyzed and plotted using nonlinear regression (curve fit, three parameters), and IC50 value was calculated from neutralization curves by Graphpad Prism 8 software. (E) Summary of IC50 values (nM) of neutralistion of SARS-CoV-2 variants performed in Vero E6 cells. Values that approached 50% neutralization were estimated from (D). Data are generated from two independent experiments, each performed in technical duplicate. (F) Molecular surface conservation of RBD VOCs. (G) W25 and its derivatives enhance spike-mediated cell fusion. Cell-cell fusion assay was performed. Stable cells expressing the rLuc-GFP components (effector cells) were transiently transfected SARS-CoV-2 spike proteins bearing spike mutations for D614G (top left), Beta (top right), Omicron (Bottom left), or control plasmid (no envelope control). W25 Nb, W25 Dimeric Fc (W25-Fc), or W25 monomeric Fc (W25-FcM) were incubated with effector cells for 1 h prior to co-culturing with target cells (huACE2-HEK293T cells). After 24–48 h, Renilla luciferase was read and analyzed by subtraction with control plasmid treatment. Data are generated from at least three biological replicates. Data are represented as mean ± SEM. Continuing surveillance of Omicron evolution revealed the newly emerged BA.2 Omicron sublineage, containing 8 unique spike mutations while lacking 13 spike mutations found in BA.1.29 Of the anti-RBD panel we tested, only W25-Fc retained sub-nanomolar affinity to BA.2 with a KD value of ∼0.4 nM. Conversely, all EUA mAbs, including S309, showed significantly impaired binding to BA.2 spike protein (Figures 3C and S7B). S2-specific mAbs S2P6 and 2G12 bound all variants with KD’s of approximately 0.2 nM and 10–12 nM, respectively (Figure 3C). Additionally, a direct comparison with other previously characterized engineered Nb-Fc fusions demonstrated that W25-Fc exhibits superior binding affinity for Omicron BA.1 (Figure S7C). W25 neutralizes multiple VOCs of SARS-CoV-2 As SARS-CoV-2 continues to evolve, the dominant variants are continuously changing. To explore and compare the efficacy of W25 for neutralization of SARS-CoV-2 VOCs, we performed live virus neutralization on Vero E6 cells against major VOCs including Alpha, Beta, Gamma, Delta, Kappa, and Lambda and found that W25-Fc neutralized Alpha, Beta, and Gamma variants (Figures 3D and 3E). W25-Fc showed complete and potent inhibition of SARS-CoV-2 Alpha, Beta, and Gamma variant infections with IC50 values ranging from 0.38 to 1.31 nM. The monomeric W25 also neutralized similar VOCs with IC50 values ranging from 0.42 to 2.12 nM (Figures S8A and S8B). Interestingly, both monovalent and divalent forms exhibited significantly weaker neutralization for Kappa, Lambda, and Delta variants. Of note, monomeric W25 was effective against Beta and Gamma, two of the most resistant variants leading to first-generation RBD-associated antibodies.15 Similar to other reports, we demonstrated that introducing bivalency through Fc fusion further improved efficacy of W25-Fc (∼10-fold for Gamma variant). Overall, the neutralization data are consistent with the W25 binding mode observed in the cryo-EM analysis since key residues within the W25 binding interface, with the exception of L452 and E484, were highly conserved across all VOCs examined in this work (Figure 3F and S9). W25 stimulates spike-dependent cell-cell fusion We furthermore explored the modulation of spike-mediated membrane fusion as a potential neutralization mechanism, using established cell-cell fusion assays.38 W25 and its derivatives (Fc fusion and monomeric Fc fusion [FcM]) caused extensive cell-cell fusion for SARS-CoV2 D614G-, Beta-, and Omicron-infected cells, demonstrated by luciferase activity (Figure 3G), when compared to a control antibody (anti-Nipah virus Fusion protein, 5B3). This was most striking for Omicron, which in the absence of W25 did not cause significant cell-cell fusion. Overall, W25-Fc induced less spike-mediated fusion, followed by W25-FcM and W25 (W25 Nb) in all variants tested. No enhancement was observed for control antibody treatment. Additionally, live cell GFP-reporter IncuCyte assays were performed in parallel and showed consistent results, where W25 addition greatly enhanced the GFP signal (Figures S10 and 11). Thus, W25 may additionally act through premature fusion activation, causing irreversible inactivation of coronavirus particles. W25 prophylactically and therapeutically reduces disease burden and protects from SARS-CoV-2 infection mortality in mice Since nanobody affinity and function have shown to be limited by temperature,39 thermostability of W25 was tested and showed W25 retains its antigen-binding function after heat treatment at 90°C (Figures S12A and S12B). In addition, full stability of the monomeric W25 after nebulization was observed, suggesting potential stability for airway administration (Figures S12C and S12D). To assess the use of W25-Fc for pre- and post-exposure therapy to counter SARS-CoV-2 infection, the K18-hACE2 transgenic murine model was employed. W25-Fc or GFP-Nb-Fc (control) were administered by intraperitoneal (i.p.) injection 4 h prior to or 24 h after intranasal (i.n.) challenge of K18-hACE2 mice with SARS-CoV-2 Beta variant (Figures 4A and 4F). It has been shown previously that SARS-CoV-2 Beta is significantly more lethal and causes more severe organ damage in K18-hACE2 mice than both the Wu ancestral strain and the D614G variant, mimicking severe SARS-CoV-2 infection in humans.40 Infected mice were monitored for 10 or 14 days after infection. All SARS-CoV-2-infected animals from the prophylactical W25-Fc treatment group survived, as opposed to control-treated group (Figure 4B). Symptoms, including ruffled fur/piloerection and accelerated shallow breathing, were observed in control-treated animals at 4–5 days after infection and progressed rapidly. Some non-treated, infected mice reached the humane endpoint state by day 6 post infection. Prophylactical W25-Fc treatment prevented all respiratory disease (Figure S13) and weight loss (Figure 4D) caused by SARS-CoV-2 Beta infection. In a separate experiment, lungs were collected after 4 days of infection. W25-Fc treatment reduced infectious virus level up to 103-fold when compared to the control group, and viral RNA also showed a significant reduction (Figures 4D and 4E). Therapeutic treatment (24 h post infection) of SARS-CoV-2 Beta-infected mice (Figure 4F) with W25-Fc significantly improved survival (60%) (Figure 4G) and weight loss (Figure 4H) and lead to lower viral load in nasal turbinate (Figure 4I) and lung (Figure 4J), compared to GFP-Nb-Fc-treated infected mice.Figure 4 W25-Fc protects mice from lethal Beta and Omicron SARS-CoV-2 infection and W25-Fc pharmacodynamic analysis (A) Experimental schematic: eight- to twelve-week-old male and female K18 transgenic mice were inoculated via the intranasal route with 1 × 103 PFU of SARS-CoV-2 (Isolate B.1.351). W25-Fc or GFP Nb-Fc were administered intraperitoneally 4 h prior to infection. (B) Survival and (C) weight change were monitored and scored. Two experiments were performed (n = 9–12, Log rank (Mantel-Cox) test ∗∗∗∗p < 0.0001). (D) Viral burdens were determined in lung tissues 4 day post infection via plaque forming assays for infectious virus and (E) RT-qPCR for viral genome copy number. Two experiments were performed (N = 10–11). (F) Experimental schematic: five- to six-week-old female K18 transgenic mice were inoculated via the intranasal route with 1 × 104 FFU of SARS-CoV-2 (Beta, Isolate B.1.351) N = 8, and N = 4–5 each group for survival and tissue harvest experiment, respectively. For therapeutic treatment, W25-Fc or GFP-Nb-Fc were administered intraperitoneally 24 h after SARS-CoV-2 Beta infection at 5 × 103 FFU/mouse. (G) Survival and (H) weight change were monitored. Viral loaded were determined in (I) nasal turbinate and (J) lung tissues 2 day post infection by immuoplaque assays for infectious virus. (K) Experimental schematic: for Omicron infection, five- to six-week-old female K18 transgenic mice were inoculated via the intranasal route with 1 × 105 FFU of SARS-CoV-2 (Isolate B.1.1.529). W25-Fc or GFP-Nb-Fc were administered intraperitoneally or intranasally 24 h after infection. (L and P) Survival and (M and Q) weight changes were monitored and scored. Viral loaded were determined in (N and R) nasal turbinate and (O and S) lung tissues 2 day post infection by immunoplaque assays for infectious virus. Bars represent medians, dots are individual animals, and dotted horizontal lines indicate the limit of detection. Data are represented as mean ± SEM. (T) The pharmacodynamics of W25-Fc in blood was determined by conjugation of W25-Fc to radioactive 111Indium. One mg/kg were injected intravenously through the tail to 6 groups of mice (group 1: 5 min, group 2: 20 min, group 3: 60 min, group 4: 3 h, group 5: 5 h and group 6: 24 h). The mice were dissected, and concentration of 111In W25-Fc was measured in the blood using an Auto-Gamma Counter. The efficacy of W25-Fc against Omicron BA.1 was also tested. Omicron virus appears less pathogenic in K18-ACE2 mice41 compared to earlier isolates. In this work, we inoculated mice with a lethal dose (105 FFU/mouse), and W25-Fc was administrated 24 h after infection via i.p. or i.n. delivery (Figure 4K). In parallel experiments, tissues were harvested 2 days post infection. For i.p. delivery, mice treated with W25-Fc had slight improvement of survival rate as compared to those which received control treatment (62.5% vs. 25%, p=0.23). Some mice treated with W25-Fc reached human endpoint by 2 days post-treatment (Figure 4L). Mice treated with W25 via i.p. displayed weight loss between day 2–8 post infection (Figure 4M). No significant reduction of viral load in nasal turbinate (Figure 4N) and lung was observed (Figure 4O). Notably, i.n. delivery of W25-Fc to Omicron-infected mice significantly improved survival (75%) compared to control-treated group (Figure 4P). Mice that received W25-Fc i.n. had lower weight compared to control-treated mice (Figure 4Q) but also lower viremia in nasal turbinate compared to controls (Figure 4R). No significant difference between W25-Fc- and control-treated mice was observed for viremia in lung (Figure 4S). Consistent with other studies, we observed low level of viremia in lung from Omicron-infected mice compared to SARS-CoV-2 Beta variant infection.41,42 Finally, we determined the pharmacokinetic profile of single dose intravenously injected W25-Fc radiolabeled with 1.2–1.5 MBq/mouse (1 mg/kg) of indium-111 in C57BL/6 mice (Figure S14A). Ex vivo studies were carried out measuring radioactivity in several tissues including blood, lungs, liver, kidneys, spleen, brain, muscle, and tail after 5 min, 20 min, 1 h, 3 h, 6 h, and 24 h (Figures S14B–S14H). We found that the half-life of W25-Fc in blood is 20.6 h, and after 24 h, the concentration of W25-Fc in blood was 3.8 ± 03 nM, well above the in vitro neutralization IC50 values for most variants (Figures 3D, 3E, and 4T). Discussion COVID-19 vaccines have proven to be highly effective in reducing the number of severe cases43,44 and are associated with lower transmissibility,45 but SARS-CoV-2 continues to evolve. The Delta variant has been dominant around Sep–Dec 2021, but more recently Omicron and its sub-lineages prevailed due to their unparalleled transmissibility. Certain therapeutic antibody cocktails have been approved for post-exposure treatment to reduce severe illness. However, the potency and breadth of protection by the majority of therapeutic antibodies are compromised by emerging SARS-CoV-2 immune-escape mutations, underlining an urgent need for developing, preparing, and improving antiviral avenues. The COVID-19 pandemic highlighted the importance of the rapid discovery of antiviral drugs. In this effort, a growing list of neutralizing antibodies, nanobodies, and synthetic antivirals have been developed using tools such as human/mouse B cells or immunized camelids.8,9,10,11,46 In contrast to conventional antibodies that require post-translation modifications from mammalian cells, nanobodies can be expressed in prokaryotic cells and, hence, are cheaper to produce. Benefiting from their small size, nanobodies may bind with higher affinity, target epitopes not accessible to conventional antibodies, and are amenable to nasal administration to directly reach infected mucosal tissues and lungs.47 However, due to their low molecular weight, nanobodies may have very short half-lives in the bloodstream. Our work and previous reports48 showed that dimerization via Fc fusion significantly improved neutralization potency, potentially through simultaneous engagement of two adjacent “up” RBDs (Figures 3D and 3E). The majority of NAbs raised against Wu SARS-CoV-2 are not effective against SARS-CoV-2 Omicron. In this work, we demonstrate that W25 and W25-Fc display potent inhibitory activity in vitro against the SARS-CoV-2 Omicron variant, as well as other VOCs including Alpha, Beta, and Gamma (Figures 1, 3D, and 3E) in live virus neutralization assays. This is consistent with structure-based sequence alignments (Figure S9), demonstrating that the majority of W25 residues involved in RBD binding are highly conserved across SARS-CoV-2 VOCs (Figure 3F), with the exception of L452, which is centrally located in the binding interface. Neutralization of Kappa, Lambda, and Delta variants by W25-Fc is strongly impeded (Figures 3D and 3E). These variants share the L452R/Q mutation, which may lead to steric clashes with W25 residues F39 and/or W111 (Figures 2D, 3F, and S9), thus rationalizing weaker binding affinity and neutralization activity of W25 toward these variants (Figures 3C–3E). This also suggests that W25 might not be effective against Omicron BA4/5. Unlike other antibodies,49,50,51,52,53 mutations at RBD position 484 (E484 K/A/Q), presented on Beta, Gamma, Kappa. and Omicron spike variants, do not impede the binding and neutralizing activity of W25. These mutations might be tolerated since E484 is only peripherally situated in the W25/RBD interface (Figures 2D and 3F). Intriguingly, W25, but not S309, retains a relatively high affinity to BA2.1 spike bearing mutation at amino acid 371 which is thought to be the key change due to introduction of a bulky phenylalanine side chain, causing loss of efficacy for several EUA mAbs and nanobodies.53,54 This is also in accordance with our structure analysis, showing no involvement of residue 371 in W25 binding. Importantly, BA.2 derivative strains recently emerged; among them is the XBB.1.5 lineage. XBB.1.5 contains RBD amino acid exchanges R346T and F490S, which are located in the W25/RBD binding interface. F490S is also found in the Lambda variant, which is not sensitive to W25. However, W25 retains significant higher binding affinity to the XBB1.5. spike (Kd = 0.1 nM, Figure S7) than to the Lambda spike (Kd = 0.765, Figures 3A and 3B), suggesting that the Lambda L452R substitution may be the main determinant of W25 inefficiency. Ultimately, however, how and if these affinity differences translate into neutralizing activity of W25 against these variants remains to be tested. Several SARS-CoV-2 NAbs have been reported targeting NTD and S2 epitopes; the most effective antibody classes however appear to target the RBD due to its natural role in binding the hACE2 receptor. Prior structural studies have classified these SARS-CoV2 NAbs according to their RBD binding modes35,55 (Figure 2E). Class 1 and 2 NAbs obstruct the ACE2-binding site on top of the RBD. However, they are highly susceptible to VOCs escape mutations including K417N, G446S, E484A, and Q493R. Classes 3 and 4 bind largely outside of the ACE2 interaction interface, to opposite surfaces on the side of the RBD. Our structure analysis showed that W25 targets a unique RBD epitope, partially overlapping with class 2 and 3 NAbs, not affected by Omicron mutations. Accordingly, W25 might be suitable as therapeutic agent in combination with class 1 and/or 4 NAbs to improve treatment efficacy against SARS-CoV2 Omicron strains or newly emerging VOCs. Most N-glycosylation sites are highly conserved across multiple SARS-CoV-2 variants, particularly in the NTD (N61, N122, N165, and N234). Our structural analysis found that W25 is surrounded by N-linked glycans at N122 and N165 from NTD of the neighboring spike protomer (Figure 2B). Interestingly, beyond its shielding function, N165 act as a molecular switch for RBD conformation by helping to maintain RBD in the “up” or ACE2-accessible state.56,57 Accordingly, RBD/W25-interaction might interfere with these glycan-mediated RBD opening processes. In addition, the binding of a class 2 NAb, BD-23, is also facilitated by this glycan.58 The mode of action of W25 in activating fusion mechanism may have clinical implications. Our structural analysis indicates that W25 only associates with RBD-up state and slightly overlaps with the ACE2-binding surface. This is in line with our fusion assay demonstrating that W25 triggers fusion activity of transfected spike protein in the presence of the cognate receptor ACE2 on the target cell population. Thus W25 may promote RBD-up conformation facilitating ACE2 engagement and enhance fusion. However, no infectivity enhancement was observed in authentic virus neutralization assay (Figures 3D and 3E); thus W25 may trigger premature fusogenic conformational changes on spike leading to viral inactivation prior to cellular engagement. A similar mechanism has been reported for RBD-specific mAbs including antibodies CR302259 and S23060 and nanobodies such as VHH E, W, and U.53 However, the mode of engagement of W25 to spike is unique to these previously reported binders, and it incorporates the outward face of RBD and is largely non-overlapping with the ACE2 site (Figure S15). Interestingly, induction of fusion was reduced using the dimeric W25-Fc relative to the monomeric form, although dimeric W25-Fc showed increased antiviral activity; thus multiple modes of action may underpin the potency of W25. In K18-hACE2 mouse model, W25-Fc administrated before infection provided full protection to all treated animals from a fatal SARS-CoV-2 challenge dose. Mice that received W25 prophylactic treatment exhibited no sickness and weight loss following the challenge with SARS-CoV-2 Beta variant (Figures 4A–4E). Single dose of therapeutic treatment of W25 after Beta and Omicron challenge provided improvement in survival (Figures 4F–4S). Interestingly, i.n. delivery of W25 conferred a better outcome compared to i.p. delivery This could be attributed to virus tropism where Omicron is reported to have higher replication in ex vivo bronchial epithelium as compared to other VOCs and Wu.61 This suggests that i.n. delivery of W25 may directly block the key infection site for Omicron. Nebulization, being more efficient at targeting the deep and local pulmonary structures, could be tested for W25 efficacy in preclinical setting. W25 in combination with other Nbs could be used as novel cocktail treatment that could be rapidly generated to block virus mutational escape. Of note, the efficiency of delivery methods for each VOC may need to be considered and assessed. In conclusion, our results show that the W25 nanobody, raised against the RBD of the ancestral SARS-CoV-2 strain, has highly potent neutralizing activity against multiple SARS-CoV-2 variants including Omicron in vitro. Our structural analyses reveal that the binding interface of W25 on the RBD is unique and highly conserved across multiple VOCs. In vivo mouse model SARS-CoV-2 infection studies demonstrated that W25 confers protection when administrated prophylactically or therapeutically, suggesting its therapeutic potential as a passive immunotherapy. The COVID-19 pandemic is still ongoing, with SARS-CoV-2 constantly evolving, prompting speculation that there will be new VOCs driving a resurgence of infections. Although vaccination remains the main measure against the pandemic, it is not effective in or possible for all patients. Accordingly, there is still an urgent need to provide additional means for infection prevention and/or treatment of high-risk patients and those in mid- to low-income countries. Because of its high efficiency, remarkable stability, and resilience to nebulization, W25 has high therapeutic potential in SARS-CoV-2-infected individuals. Limitations of the study We also recognize that there are some limitations in our present study. Firstly, our experiments analyzing the neutralization profile and cell-to-cell fusion were performed in Vero E6 cells and in HEK293 cells overexpressing ACE2. While these cells are widely used and validated in preclinical settings, future investigations using primary nasal and/or lung epithelial cells may offer insight into the observed in vivo therapeutic activity. Secondly, as nebulization was only performed in uninfected mice, nebulization in an infection model would need to be performed to access the efficacy. Relevant to this, however, i.n. delivery of W25-Fc was shown to be therapeutically beneficial for Omicron infection. STAR★Methods Key resources table REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies W25 Valenzuela Nieto et al.27 N/A W270 Valenzuela Nieto et al.27 N/A NB3 Schoof et al.52 N/A NB11 Schoof et al.52 N/A NB20 Schoof et al.52 N/A MR17 Li et al.62 N/A SB14 N/A SB42 N/A SB4 N/A Ty1 Hanke et al.31 N/A H11 Huo et al.63 N/A Nb against GFP Kubala et al.64 N/A S309 Pinto et al.9 N/A REGN10987 Hansen et al.10 N/A REGN10933 Hansen et al.10 N/A Ly-CoV555 Jones et al.11 N/A CT-P59 Kim et al.12 N/A CB6 Shi et al.8 N/A DH1047 Martinez et al.13 N/A 5B3 Dang et al.65 N/A C05 Ekiert et al.66 N/A Anti-human IgG conjugated with an IRDye 800CW Millennium Scientific Cat#LCR-926-32232 Goat Anti-Human IgG Fc Highly Cross-Adsorbed Secondary Antibody Thermofisher Cat#A18829 Bacterial and virus strains StellarTM Competent cells Takara Cat#636763 hCoV-19/Australia/QLD02/20200 This paper GISAID accession ID, EPI_ISL_407896 hCoV-19/Australia/QLD1517/2021 This paper GISAID accession ID EPI_ISL_944644 hCoV-19/Australia/QLD1520/2020 This paper GISAID accession ID EPI_ISL_968081 hCoV-19/Australia/QLD1893C/2021 This paper GISAID accession ID EPI_ISL_2433928 hCoV-19/Australia/NSW4318/2021 This paper GISAID accession ID EPI_ISL_1121976 hCoV-19/Australia/NSW4431/2021 This paper GISAID accession ID EPI_ISL_1494722 hCoV-19/Australia/NSW-RPAH-1933/2021 This paper GISAID accession ID EPI_ISL_6814922 Chemicals, peptides, and recombinant proteins Spike proteins (Wuhan, Alpha, Beta, Gamma, Delta Lambda, BA.1, BA.2) This paper N/A Critical commercial assays KPL milk diluent/blocking solution concentrate SeraCare Cat#5140-0011 3,3,5,5′-Tetramethylbenzidine Thermofisher Cat#002023 Experimental models: Cell lines ExpiCHO-STMcells Thermofisher Cat#29127 Flp-In™ T-REx™ 293 cells Thermofisher Cat#R78007 HEK293T ATCC CRL-3216 VeroE6 ATCC CRL-1586 VeroE6-TMPRSS2 This paper N/A VeroE6-ACE2-TMPRSS2 BEI resource NR-54970 Experimental models: Organisms/strains Transgenic hACE2 mice (2B6.Cg-Tg(K18-ACE2)2Prlmn/J) Jackson laboratory RRID:IMSR_JAX:034860 Oligonucleotides primers for SARS-CoV-2 N gene Forward 5′- TTACAAACATTGGCCGCAAA-3′ This paper N/A primers for SARS-CoV-2 N gene Reverse 5′-GCGCGACATTCCGAAGAA-3′ This paper N/A primers for HPRT1 Forward 5′- GTTGGATACAGGCCAGACTTTGTTG-3′ This paper N/A primers for HPRT1 Reverse 5′-GAGGGTAGGCTGGCCTATTGGCT-3′ This paper N/A Software and algorithms GraphPad Prism 8.01 GraphPad Software Inc. https://www.graphpad.com/scientific-software/prism/" https://www.graphpad.com/scientific-software/prism/ MotionCorr2 Zheng et al.67 N/A CTFFind4 Rohou et al.68 N/A Relion 3.0.7 Zivanov et al.69 N/A Relion 3.1.3 Zivanov et al.69 N/A cryoSPARC 3.3.1 Punjani et al.70 N/A AlphaFold2 Jumper et al.,71 Mirdita et al.33 N/A Coot 0.9.5 Emsley et al.72 N/A Phenix 1.19.2–4158 Liebschner et al.73 N/A ChimeraX Petterson et al.84 N/A Biacore Insight Evaluation Cytiva https://www.cytivalifesciences.com/en/us/shop/protein-analysis/spr-label-free-analysis/software/biacore-insight-evaluation-software-p-23528 Resource availability Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Daniel Watterson ([email protected]). Material availability Mabs plasmids and mAbs generated in this study are available from the authors upon request. Experimental model and study participant details Cell lines HEK293T (ATCC), Vero-E6 (ATCC) Vero-Ace2-Tmprss2 cells (BEI resources) were maintained in DMEM supplemented with 10% heat-inactivated (HI) Fetal Calf Serum (FCS, Bovogen), penicillin (100 IU/mL)/streptomycin (100 μg/mL) (P/S, Gibco) and L-glutamine (2 mM) (Life Technologies). Cells were cultured at 37°C and 5% CO2. ExpiCHO cells were maintained in ExpiCHO Expression media as per the manufacturer’s protocol. Flp-In T-REx 293 cells were cultured according to the manufacturer’s protocol. All cell lines used in this study were routinely tested for mycoplasma and found to be mycoplasma-free (MycoAlert Mycoplasma Detection Kit MycoAlert, Lonza). Evaluating the efficacy of W25 in K18-huACE2 mice Six to 8 weeks old female transgenic hACE2 mice (strain: 2B6.Cg-Tg(K18-ACE2)2Prlmn/J) were purchased from The Jackson Laboratory and Animal Resources Center and/or bred in-house. C57BL/6J female mice weighing 20–23 g were purchased from Janvier, France, La Rochelle. All mice were kept in a climate-controlled facility with a 12 h light/dark cycle. Study approval All work with infectious virus was performed in a biosafety level 3 laboratory. In vitro work and protocol were approved by the University of Queensland Biosafety committee (IBC/447B/SCMB/2021) and the Penn Institutional Biosafety Committee and Environmental Health and Safety. All the animals infected by SARS-CoV-2 were handled in a biosafety level 3 animal facilities in accordance with the recommendations for care and use of the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health and the Guidelines to promote the wellbeing of animals used for Scientific purposes. The protocols were approved by the Institutional Animal Care and Use Committee at the University of Pennsylvania (protocol #807017) and the University of Queensland animal ethics committee (2021/AE000929 and 2021/AE001119). All the authors declare their compliance with publishing ethics. C57BL/6J mice were conducted in accordance with the European Commission’s Directive 2010/63/EU, FELASA and ARRIVE guidelines for animal research and, with approval from The Danish Council for Animal Ethics (license numbers 2017-15-0201-01283). Method details Transient expression of S protein, mAbs and Nbs Expression and purification of SARS-CoV-2 Spike Soluble, trimeric spikes (residue 1–1204 amino acid) of SARS-CoV-2/human/China/Wuhan-Hu-1/2019 (referred to as Wu) (GenBank: MN908947), SARS-CoV-2 Beta variant (B.1.351) (L18F, D80A, D215G, Δ242–244, K417N, E484K, N501Y, D614G and A701V), Alpha variant (B.1.1.7) (Δ69-70, ΔY144–145, N501Y, A570D, D614G, P681H, T716I, S982A and D1118H), Gamma variant (P.1) L18F, T20N, P26S, D138Y, R190S, K417T, E484K, N501Y, D614G, D655Y, T1027I and V1176F), Lambda variant (C.37) (G75V, T76I, R246N, Δ247–253, L452Q, F490S, D614G and T859N), Delta variant (B.1.617) (T19R, E156G, Δ157–158, L452R, T478K, D614G, P681R and D950N), Omicron (BA.1) (A67V, Δ69–70, T95I, G142D, Δ143-145, Δ211, L212I, 214EPEinsert, G339D, S371L, S373P, S375F, K417N, N440K, G446S, S477N, T478K, E484A, Q493R, G496S, Q498R, N501Y, Y505H, T547K, D614G, H655Y, N679K, P681H, N796Y, N856K, Q954H, N969K, L981F) and Omicron (BA.2) (T19I, L24S, Δ25–26, A67V, Δ69-70, G142D, V213G, S371F, S373P, S375F, T376A, D405N, R408S, K417N, N440K, S477N, T478K, E484A, Q493R, Q498R, N501Y, D614G, H655Y, N679K, P681H, N764K, D796Y, Q954H and N969K) spike mutations were added in silico into the codon-optimised Wuhan reference stain and were cloned to pNBF plasmid. Spike proteins contain 6 proline mutations (F817P, A892P, A899P, A942P, K986P and V987P) and substituted at the furin cleavage site (residues 682–685).74 Cells were harvested at 5 days post transfection and purified using 0.1M NaAc pH3 before buffer exchanged to PBS pH7.4. Purified proteins were subjected to size exclusion chromatography (SEC) using Superose6 Increase 10/300 (Cytiva) in 10 mM Tris-HCl pH 7.8, 150 mM NaCl. A stable Flp-In T-REx 293 cell line (Thermo Fisher) expressing the prefusion-stabilized HexaPro variant of the Wu Spike75 was generated according to the manufacturer’s protocol (polyclonal selection). Cells were seeded in T-300 flasks (0.3 × 106 cells/mL), and after 24 h the medium was exchanged against fresh medium containing 1 μg/mL tetracycline to induce protein expression. After 4 days, cell culture supernatant was dialyzed against binding buffer (50 mM Tris-HCl pH 7.8, 500 mM NaCl), and spike was purified using Strep-Tactin XT beads (IBA Lifesciences), followed by SEC using a Superdex200 Increase 10/300 column (Cytiva), equilibrated in 10 mM Tris-HCl pH 7.8, 150 mM NaCl. SEC peak fractions containing spike protein were pooled, concentrated to 1 mg/mL, flash-frozen in liquid nitrogen and stored at −80°C. For antibodies and nanobodies, DNA encoding antibody variable domains including W25, W270,27 NB3, Nb11, Nb20,52 MR17,62 SB14, SB42, SB4,76 Ty131 and H1163 and Nb against GFP protein67 were cloned in-frame with an upstream IgG leader sequence and downstream human IgG1 Fc open reading frame into the mammalian expression vector pNBF. For monoclonal antibodies, heavy and light chains of S309,9 REGN10987, REGN10933,10 Ly-CoV555,11 CT-P59,12 CB6,8 DH1047,13 5B3,65 and C0566 were cloned into a human IgG1 expression vector as described previously.77 After validation of the cloning by Sanger sequencing, the plasmids were transfected into ExpiCHO cells (Thermo Scientific) according to the manufacturer’s protocol (1 μg DNA/mL of cells; 2:3 ratio of heavy chain to light chain). After 7 days, cell culture media were clarified by centrifugation and the IgG captured using Protein A resin (GE Healthcare). Proteins were eluted from the resin using citrate buffer pH 3, eluate was buffer exchanged to PBS pH7.4. For nanobody production, W25 and W270 were expressed as previously described.27 The pHen6-W25 vector inoculated in 20 mL of liquid LB-medium containing 100 μg/mL ampicillin and 1% glucose. The bacteria were cultured at 37°C with agitation for 16 h. The bacteria were then diluted into 1L Terrific Broth (TB) medium containing 100 μg/mL ampicillin, 2 mM MgCl2, 0.1% glucose and incubated at 37°C to an OD600 of 0.6–0.9. The expression of the nanobodies was induced by adding 1 mM of IPTG (isopropyl-β-d-1-thiogalactopyranoside) for 20 h at 28°C. Bacteria were collected by centrifugation at 8000 rpm for 8 min at 4°C. The harvested bacteria were resuspended in a 12 mL TES buffer (0.2 M Tris pH 8.0, 0.5 mM EDTA, 0.5 M sucrose) and incubated for 1 h on ice, then incubated for another hour on ice in a 18 mL TES buffer, diluted 4 times and centrifuged at 8000 rpm at 4°C to pellet down cell debris. The supernatant was loaded on 5 mL of HisPur Ni–NTA agarose resin which was pre-equilibrated with binding buffer (Tris 50 mM, NaCl 500 mM, imidazole 10 mM pH 7.5). The lysed cells containing His- and myc-tagged nanobodies were added to the column followed by adding the column’s volume in binding buffer for a total of eight times. The column was washed by adding 8-fold the columnś volume with wash buffer (Tris 50 mM, NaCl 500 mM, imidazole 30 mM pH a 7.5), and eluted with 15 mL of elution buffer (Tris 50 mM, NaCl 150 mM, 150 mM imidazole, 1 mM DTT pH 7.5). Purified antibodies, and nanobodies were verified by SDS-PAGE Coomassie staining analysis. Cryo-EM sample preparation, imaging and processing Wu spike (0.5 μM trimer) and W25 (1.7 μM) were incubated for 2 h on ice in a volume of 50 μL buffer containing 10 mM Tris-HCl pH 7.8 and 150 mM NaCl. 3.5 μL sample were adsorbed for 45 s on glow-discharged 400 mesh R1.2/R1.3 holey carbon grids (Quantifoil) and plunge-frozen in liquid ethane using a Vitrobot Mark IV (Thermo Fisher) with 1 s blotting time at 4°C and 99% humidity. Data were collected on a 300 kV Tecnai Polara cryo-EM (Thermo Fisher) equipped with a K2 Summit direct electron detector (Gatan) in super-resolution mode, at a nominal magnification of 31000×, with a pixel size of 0.63 Å/px and the following parameters: defocus range of −1 to −2.5 μm, 50 frames per movie, 200 ms exposure per frame, electron dose of 6.2 e/Å2/s, leading to a total dose of 62 e/Å2 per movie. Data were collected using Leginon.78 Movies were aligned and dose-weighted using MotionCor2.67 Initial estimation of the contrast transfer function (CTF) was performed with the CTFFind4 package via the Relion 3.0.7 GUI.68,69 Power spectra were manually inspected to remove ice contamination and astigmatic, weak, or poorly defined spectra. Subsequent image processing procedures are shown in Figure S1. Particles were picked initially using the Relion 3.0.7 LoG autopicker. After 2D and 3D classification, the best defined 3D class was used as template for a second round of autopicking in Relion 3.0.7. 2D and 3D classification yielded 141.257 clean particles, which were further processed using Relion 3.0.7 or cryoSPARC 3.3.170 as indicated in Figure S1. Particles were subjected to an additional ab-initio reconstruction (3 classes, class similarity = 0), followed by a consensus refinement, exhibiting a 2 RBD-up, 1 RBD-down conformation. Subsequently, particles were locally refined to one RBD in the “up” conformation using a stalk/RBD/W25 mask, followed by a focused classification without alignment using an RBD/W25 mask for the identification of the most defined WuRBD/W25 complex subset. This particle subset was subjected to another local refinement using the RBD/NTD mask. Locally refined maps were used as reference in a heterogeneous refinement. This revealed that particles were partially in a 1 RBD-up, 2 RBD-down conformation. Clean particles were re-assigned to both global conformations, followed by non-uniform refinement, resulting in 3.8 Å map for the 2 RBD-up, 1 RBD-down class (map 1, Figure S1) and a 3.81 Å map for the 1 RBD-up, 2 RBD-down class (map 2). Both particle subsets were pooled again and locally refined to the shared RBD in “up” conformation, resulting in a 5.92 Å map. Omicron spike and W25 were mixed in 1:6 M ratio and incubated on 37°C for 45 min. Omicron spike-W25 complex were adsorbed for 10 s onto glow discharged Quantifoil grids (Q1.2/1.3) and plunge frozen into liquid ethane using an EMGP2 system (Leica). Cryo-EM data collection was performed with Serial EM software (Version 3.1) using CRYO ARM 300 (JEOL) at the Center of Microscopy and microanalysis UQ. Movies were acquired in super resolution and CDS mode with a slit width of 10 eV using a K3 detector (Gatan) for 3.3 s during which 40 movie frames were collected with a 1 e−/Å2 per frame. Data were collected at a magnification of 60,000x, corresponding to a calibrated pixel size of 0.4 Å/pixel. Movies were binned 2x during motion Correction.67 A total of 9,258 micrographs were collected, 8,845 micrographs were selected based on an estimated resolution cutoff of 4.5 Å. Particles were extracted using Relion 3.0.7. All following processing steps were performed in Cryosparc 3.3.170 as indicated in Figure S3. Extracted particles were cleaned up from ice and false-positive picks using 2D classification. After 2D classification, most defined classes with high-resolution features were retained (164.806 particles) and used for an ab-initio 3D reconstruction (3 classes, class similarity = 0). Particles were further processed using heterogeneous refinement and previously generated maps, followed by a re-assignment to both global conformations from the Wuhan dataset (maps 1 and 2, Figure S1). A 3D variability analysis was performed for particles in 1 RBD-up, 2 RBD down conformation using a mask for the RBD in the “up” conformation. This revealed that particles were partially in a 3 RBD-down conformation. Using a less strict 2D class selection (222237 particles), the particle subset were then cleaned again using heterogeneous refinement, followed by re-assignment to the three identified global conformations. Particles subsets in the 2 RBD-up, 1 RBD-down and 1 RBD-up, 2 RBD-down conformation were cleaned in an additional round of heterogeneous refinement. The 1 RBD-up, 2 RBD-down particle subset showed the most defined RBD/W25 density and was therefore refined using non-uniform refinement (4.97 Å, map 4, Figure S3), followed by local refinement using an RBD-up/NTD mask, resulting in a 6.04 Å map (map 5). Model building and refinement The trimeric base, and the individual RBDs and NTDs of a Wu spike model (PDB 6ZXN,31) were fitted as rigid bodies in map 1 (Figure S1). One RBD and the adjacent NTD, as well as an AlphaFold233,71 model of the W25 nanobody were then fitted as rigid bodies in focus map 3 (Figure S1), trimmed manually using Coot 0.9.5,72 subjected to molecular dynamics flexible fitting using the Namdinator server79 with standard parameters, followed by real space and ADP refinement in Phenix 1.19.2–415873 and manual adjustment using Coot 0.9.5. For Omicron spike model building, a similar strategy was pursued, however using PDB 7WG630 as starting model, and map 4 and 5 for fitting (Figure S3). Molecular models and maps were visualized using the ChimeraX software.80 Surface plasmon resonance (SPR) SPR measurements were performed using a Biacore 8K+ instrument (Cytiva). Purified W25-Fc, S309, DH1047 or C05 antibody were captured on a Protein A Series S Sensor Chip (Cytiva). Binding of SARS-CoV-2 spike variants were tested using a Multi-Cycle High Performance Kinetics assay. The antibodies were all prepared at 1 μg/mL in running buffer (HBS-P+, pH 7.4, Cytiva) and injected for 30 s at 30 μL/min over Fc2 on all 8 channels. Multiple concentrations of SARS-CoV-2 spike variants were included: 0 nM, 1.40625 nM, 2.8125 nM, 5.625 nM, 11.25 nM, 22.5 nM, 45 nM and 90 nM. The proteins were injected over both Fc1 and Fc2 of Channels 1–8 in series for 180 s at 30 μL/min, followed by a dissociation period of 600 s. The chip surface was regenerated between each cycle using 10 mM glycine pH 1.5. The data was double reference subtracted; reference cell subtraction (Fc1) from the active cell (Fc2), and zero concentration subtraction for each analyte-antibody pair. Sensorgrams for the association and dissociation phases were fitted to a 1:1 binding model, using Biacore Insight Evaluation software (Cytiva). ELISA To test nanobody and antibodies, SARS-CoV-2 spike variant proteins in PBS pH7.4 were immobilised on Maxisorb ELISA (Nunc) plates at a concentration of 2 μg/mL overnight. Serial 5-fold dilutions of Fc-fusion nanobodies or antibodies in 1X KPL in blocking buffer (Seracare) were incubated with the immobilized antigen, followed by incubation with HRP-coupled anti-human IgG (MilleniumScience) and the chromogenic substrate TMB (ThermoFisher). Reactions were stopped with 2 M H2SO4 and absorption measured at 450 nm. Viruses The SARS-CoV-2 Ancestral (Wu) variant, hCoV-19/Australia/QLD02/2020 (GISAID accession ID, EPI_ISL_407896), Alpha variant, hCoV-19/Australia/QLD1517/2021, (GISAID accession ID EPI_ISL_944644), Beta variant hCoV-19/Australia/QLD1520/2020, (GISAID accession ID EPI_ISL_968081), Delta: hCoV-19/Australia/QLD1893C/2021 (GISAID accession ID EPI_ISL_2433928) were kindly provided by Dr Alyssa Pyke (Queensland Health Forensic & Scientific Services, Queensland Department of Health, Brisbane, Australia). For Omicron BA.1 variant (hCoV-19/Australia/NSW-RPAH-1933/2021), Gamma variants, hCoV-19/Australia/NSW4318/2021, Lambda variants hCoV-19/Australia/NSW4431/2021 were kindly provided by A/Prof. Stuart Turville (University of New South Wales, Australia). Omicron BA.2 variant was kindly provided by Prof. Andreas Suhrbier (QIMR Berghofer Medical Research Institute). The passage 2 of SARS-CoV-2 variants (except Gamma and Lambda) was received, and passage 3 were propagated in TMPRSS2-expressing VeroE6 and used as virus stock. Passage 3 of and passage 4 of Gamma, Lambda and BA.2 variants were obtained and passage 4 was propagated in TMPRSS2-expressing VeroE6 and used as virus stock, viruses were stored at −80°C. All working virus stocks were sequenced and validated using either Nanopore sequencing or Sanger sequencing on spike genes. For SARS-CoV-2 Beta variant (B.1.351) used in the animal experiment, was obtained from Andy Pekosz (Center for Emerging Viruses and Infectious Diseases (CEVID)). Infectious stocks were grown in Vero-Ace2-TMPRSS2 cells (BEI Resources) and stored at −80°C. All work with infectious virus was performed in a biosafety level 3 laboratory and approved by the University of Queensland Biosafety committee (IBC/447B/SCMB/2021) and the Penn Institutional Biosafety Committee and Environmental Health and Safety. Plaque reduction neutralisation test (PRNT) The levels of neutralising antibodies were assessed using our established PRNT protocol28 and supplementary materials and methods. Briefly, purified mAbs were 5-fold serially diluted in DMEM containing 2% HI-FCS and 1% P/S (Gibco). Subsequently, serial diluted antibodies were incubated with 50–100 immuno-plaques of SARS-CoV-2 and incubated at 37°C for 1 h. Then, 50 μL of mixture (virus/antibody complex) was added onto pre-seeded VeroE6 cells in 96-well plates at 6 × 104 cells/well and incubated at 37°C for 30 min. Following, overlay medium (containing 1% CMC, 1X M199, 2% HI-FCS and 1% (P/S)) was added on top of the inoculum and incubated at 37°C with 5% of CO2. Twenty-four hours after infection, the overlay was removed, and the monolayer was fixed with cold fixative solution (80% acetone and 20% phosphate buffered saline (PBS)) and incubated at −20°C for 30min. Fixative reagent was removed, and the monolayer was fully dried. Monolayer was then treated with 100 μL of 1X milk blocking solution (KPL, Seracare) diluted in 1X-PBS containing 0.05% Tween 20 (PBS-T) and incubated at 4°C for overnight. Next, the blocking buffer was removed, and immuno-plaques were stained using anti-spike antibody (mouse CR3022) as the primary antibody for the following variants: SARS-CoV-2 Wu, Alpha, Gamma, Beta, Lambda, Kappa and Delta variants. Anti-M SARS-CoV-2 antibody was used as a primary antibody to stain the immune-plaques for Omicron variant. The monolayer was incubated at 37°C for 1 h with the primary antibody and the unbounded antibody was removed by five consecutive washes with 5min incubation between each washed. To reveals the immuno-plaques, an infra-red dye conjugated secondary antibody (IRDye 800CW Goat anti-Mouse or Goat anti-streptavidin, MillenniumScience) was added and incubated at 37°C for 1 h, followed by five consecutive washes with 5min incubation in between. Finally, plate was fully dried avoiding long light exposure and scanned using the LI-COR Biosciences Odyssey Infrared Imaging System (Odessey CLx, Li-COR, USA). Immuno-plaques were analyzed and counted using an automated foci counter program, Viridot.81 A stock of primary and secondary antibodies at 1 mg/mL were diluted 1/1000 and 1/2500 in blocking buffer solution, respectively and 50 μL/well was used for the staining. Cell-cell fusion assay Cell-cell fusion assay was conducted as previously described.38 HEK293T Lenti rLuc-GFP 1–7 (effector cells) and HEK293T Lenti rLuc-GFP 8–11 (target cells) were seeded separately at 7.5×105 per well in a six-well dish in 3 mL of PRF-DMEM-10% and incubated overnight at 37°C, 5% CO2. Transfection mixes were set up in 200 μL Opti-MEM (Gibco) with the TransIT-X2 Dynamic Delivery System as per the manufacturer’s recommendations (Mirus). SARS-CoV-2 Spike glycoproteins of D614G, Beta and Omicron and human ACE2 plasmids were transfected into effector cells. A mock-transfected (pcDNA3.1 empty plasmid, - vGP) and positive transfection control (250 ng rLuc-GFP 8–11 plasmid) was also set up. W25, its derivatives and control antibody were diluted to specific concentration in sterile 1.5 mL tubes using serum-free PRF-DMEM and plated at 25 μL/well in a white-bottomed, sterile 96-well plate (Corning), including no antibody controls. The antibodies were incubated with 2×104 effector cells in 50 μL at 37°C, 5% CO2 for 1 h, after which target cells were co-cultured to corresponding wells and incubated for 18–24 h, after which GFP-positive syncytia and Renilla luciferase were quantified. Negative controls (effector cells only, target cells only) and positive transfection controls (HEK293T Lenti rLuc-GFP 1–7 cells transfected with rLuc-GFP 8–11 plasmid) were always included. The assay was conducted separately for Renilla luciferase and GFP readout. To quantify Renilla luciferase expression in fusion assays media were replaced with 100 μL of phosphate-buffered saline (PBS) followed by 60 μL of diluted substrate, Coelenterazine-H, 1 μM (Promega) 1: 400 with PBS. The plate was incubated in the dark for 2 min then read on the GloMax Multi+ Detection System (Promega). To quantify GFP expression, cells were plated in clear flat-bottomed 96-well plates (Nunc) and imaged every hour using the IncuCyte S3 live cell imaging system (Essen BioScience). Five fields of view were taken per well at 10× magnification, and GFP expression was determined using the total integrated intensity metric included in the IncuCyte S3 software (Essen BioScience). To analyze images generated on the IncyCyte S3, a collection of representative images is first taken to set fluorescence and cellular thresholds, which allows for the removal of background fluorescence, and selection of cell boundaries (‘objects’) by creating ‘masks’. Following this, the total integrated intensity metric can be accurately calculated by the software, which takes the total sum of objects’ fluorescent intensity in the image, expressed as green count units (GCU) μm. Mouse infections Animal studies were carried out in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health and the Guidelines to promote the wellbeing of animals used for Scientific purposes. The protocols were approved by the Institutional Animal Care and Use Committee at the University of Pennsylvania (protocol #807017) and the University of Queensland animal ethics committee (2021/AE000929 and 2021/AE001119). Heterozygous K18-hACE C57BL/6J mice (strain: 2B6.Cg-Tg(K18-ACE2)2Prlmn/J) were obtained from The Jackson Laboratory and/or bred in-house. Animals were housed in groups and fed standard chow diets. Virus inoculations were performed under anesthesia that was induced and maintained with ketamine hydrochloride and xylazine, and all efforts were made to minimize animal suffering. For prophylactic treatment study, mice of different ages and both sexes were infected with SARS-CoV-2 Beta variant at 1 × 103 plaque-forming units (PFU) via intranasal administration 4 h after intraperitoneal administration of 100 μg W25-Fc or GFP-Nb-Fc in 100 μl PBS. For therapeutic treatment study, 5–6 weeks old female mice were infected with SARS-CoV-2 Beta (5 × 103 FFU/mouse) or Omicron (1 × 105 FFU/mouse) variants via intranasal administration. Infected mice were administrated with 100 μg W25-Fc or GFP-Nb-Fc via intraperitoneal (100 μL) or intranasal administration (20 μL) 24 h post infection. The dose of W25 Fc was based on previous report.82 Mice were monitored daily for weights and clinical respiratory disease scores were obtained. Mice were sacrificed when exhibiting greater than 20% weight loss or after reaching a respiratory disease score of 5 for longer than 2 days or greater than 5. Respiratory disease was scored based on activity, ruffled fur/piloerection, accelerated shallow breathing, labored breathing, tremor/tension/paralysis and/or weight loss of greater than 7.5% with a point added for each identified. Lungs and Nasal turbinates were collected on 2 or 4 days post-infection and viral titers were determined using plaque-forming assay, immunoplaque-forming assay and RT-PCR. Plaque-forming assay and immunoplaque-forming assay Tissues were weighed and homogenized with zirconia beads in a FastPrep-24 instrument (MP Biomedicals) in 0.6 mL of DMEM media supplemented with 10% heat-inactivated FBS. Tissue homogenates were clarified by centrifugation at 10,000 r.p.m. for 5 min and stored at −80°C. Vero-Ace2-Tmprss2 cells (BEI Resources) were seeded at a density of 2.5 × 106 cells per plate in flat-bottom 6-well tissue culture plates. The following day, media was removed and replaced with 200 μL of 10-fold serial dilutions of the tissue homogenate to be titered, diluted in DMEM. One hour later, seaplaque/seakem agarose was added. Plates were incubated for 72 h, then fixed with 4% paraformaldehyde in phosphate-buffered saline for at least 1 h. Plaques were visualized with 0.05% (w/v) crystal violet in 20% methanol and washed with water prior to enumeration of plaques. Viral load via RT-qPCR or immuno-plaque assay (iPA) For RT-qPCR, tissue homogenate was added to Trizol at 1:3 ratio and extracted using Zymo Direct-zol RNA kit following manufactures protocol. RNA was reverse transcribed and amplified using iScript cDNA Synthesis Kit (Biorad). Gene-specific primers for SARS-CoV-2 N gene (F:TTACAAACATTGGCCGCAAA & R:GCGCGACATTCCGAAGAA) and mouse HPRT1 (F: GTTGGATACAGGCCAGACTTTGTTG & R: GAGGGTAGGCTGGCCTATTGGCT) with Power SYBR Green PCR Master Mix (Applied Biosystems) were used to amplify viral and cellular RNA by QuantStudio 3 Flex Real-Time PCR Systems (Applied Biosystems). The relative expression levels of target genes were calculated using the standard curve method using quantitative synthetic RNA (ATCC) and normalized to HPRT1 RNA as an internal control. For iPA, organs such as lung and nasal turbinates were collected in preweighted tubes and homogenised using a tissue homogenizer (TissueLyser LT, Qiagen). Homogenates were then clarified by centrifugation at 10000 × g, 4°C for 5 min. Supernatants were collected and stored at −80°C. Viral loads were determined by iPA on VeroE6 following our optimised protocol for viral titration. Viral titers were expressed in FFU per grams of tissue (FFU/g). Radiochemistry [111In]InCl3 was purchased from Curium. The analytical-HPLC system consists of a 170U UVD detector, a Scansys radiodetector and a Dionex system connected to a P680A pump. The system was run by Chromeleon software. The radiochemical conversion (RCC) of the radiolabeled compounds was determined by analyzing an aliquot of the crude reaction mixture by radio-HPLC analysis integrating the radioactive peaks of the chromatogram.83 Radiolabeled products were characterized by associating the UV-HPLC traces of the authentic cold compounds with the radio-HPLC chromatogram of the reaction mixtures. The radiochemical yield (RCY) was determined using the initial activity at the beginning of the synthesis and that of the formulated product at the end of thes, corrected for decomposition and corrected for decay. 111In-labeling DOTA-tetrazine 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA) -PEG11-tetrazine was dissolved (2 mg/mL) in metal-free water and stored at −80°C before use. In general, an aliquot of 50–100 μL (10–30 MBq) of [111In]indium chloride in 0.1M HCl was combined with 2 μL DOTA-PEG11-tetrazine and 1M NH4OAc buffer (pH 5.5) at a volume ratio of 1:10 was added. The mixture was shaken at 600 rpm for 5 min at 60°C in an Eppendorf ThermoMixer C. Then, 10 mM diethylenetriamine-pentaacetic acid (DTPA, volume ratio 1:10) and 2 μL 10 mg/mL gentisic acid in saline was added and the solution was shaken for an additional 5 min at 60°C in an Eppendorf ThermoMixer C. Quantitative labeling yield and a radiochemical purity reater than 95% were obtained with this method, as confirmed by radio-HPLC. Trans–cyclooctene (TCO)-modifications W25-Fc W25-Fc (2.0 mg/mL) in PBS (pH 7.4) was aliquoted in five vials of 100 μL. To each aliquot, 25 equiv TCO-PEG4-NHS (Broadpharm, BP-22418) and sodium carbonate buffer (1 M, 3.1 μL, pH 8.0) was added. The mixture was shaken at 600 rpm for 2 h at room temperature in the dark. Unreacted TCO-PEG4-NHS was removed by purification with Zeba spin desalting columns (7K MWCO, 0.5 mL, 89882, Thermofisher) and eluted in PBS (pH 7.4), with >95% protein recovery after purification. Final protein concentration was 1.7 mg/mL, measured with NanoDrop (NanoDrop 2000, ThermoScientific). Titration experiments were conducted to quantify the amount of reactive TCOs per protein-conjugate. 111In-labeled Tz stock was diluted accordingly to add 5 μL an excess of 111In-labeled Tz (2 equiv of Tz per protein-conjugate) to 5 μL of TCO-HSA and the mixture was shaken at 600 rpm for 1 h at 37°C. 3 μL NuPAGE LDS Sample Buffer (NP0007, Invitrogen) was added and the mixture was shaken for 10 min at 70°C. Samples were applied to NuPAGE 4 to 12%, Bis-Tris, 1.0 mm, Mini Protein Gel, 12-well (NP0322BOX, Invitrogen) SDS-PAGE gels. SDS-PAGE gels were exposed to phosphor storage screens and read by a Cyclone Storage Phosphor System (PerkinElmer Inc.). Quantification of plate readings was done with Optiquant software (version 5.00, PerkinElmer Inc.). Quantification by radioactive SDS-PAGE revealed presence of approximately 1.6 reactive TCO/protein. Radiolabeling of TCO-W25-Fc with 111In-tetrazine To a 5 mL Eppendorf vial was added TCO-W25-Fc (465 μL, 10.16 nmol protein, 16.26 nmol TCO) and 111In-Tz (68 MBq, 95 μL, 1.6 nmol, 0.1 TCO/Tz eq.). The mixture was shaken at 600 rpm for 1 h at 37°C, giving a radiochemical conversion of 68%. Ultrafiltration (Vivaspin 500, 5,000 MWCO PES, Sartorius) yielded in 40 MBq 111In-W25-Fc with a radiochemical purity of 96%, determined by radio-HPLC. Protein concentration after purification was measured with NanoDrop (NanoDrop One, ThermoScientific). The 111In-labeled protein was formulated in PBS to a final concentration of 15 MBq/mL. Biodistribution of 111In-labelled W25-Fc in vivo C57BL/6J female mice weighing 20–23 g (Janvier, France, La Rochelle) were housed in cages of 7–8 mice per cage. All mice were kept in a climate-controlled facility with a 12 h light/dark cycle. The cages contained a biting stick, fed with commercial breeding diet ad libitum (1310 FORTI- Avlsfoder, Brogaarden, Altromin International) and had access to water. All procedures were conducted in accordance with the European Commission’s Directive 2010/63/EU, FELASA and ARRIVE guidelines for animal research and, with approval from The Danish Council for Animal Ethics (license numbers 2017-15-0201-01283) as well as the Department of Experimental Medicine, University of Copenhagen. Mice were evenly divided into six different groups. All mice were weighted prior to injection of 111In-W25Fc (1.2–1.5 MBq/mouse) through a tail vein. The animals were sacrificed under anesthesia (3% sevoflurane) by decapitation at the designated time points after tracer administration (group 1: 5 min, group 2: 20 min, group 3: 60 min, group 4: 3 h, group 5: 5 h and group 6: 24 h). The mice were dissected and of each mouse the blood, lungs, liver, kidney, spleen, brain, muscle and tail were collected and placed into preweighed gamma counting tubes (Polypropylene, 5 mL 75 × 12 mm Ø, round base, Hounisens). The tubes with collected tissue were weighed and radioactivity measurements were conducted on a Cobra II Auto-Gamma Counter (Model D5005, Packard BioScience Company). The measurements were corrected for radioactivity decay and background. The ex vivo biodistribution results were expressed as a percentage of the injected radioactivity dose per gram of tissue (% ID/g) and nmol protein per tissue volume. For these calculations, tissue weight was converted to volume based on different tissue density for different organs. Radioactive counts in the tail were used as a quality control for the injections, which led to exclusion of one mouse in group 4 that had >20% ID in remaining in the tail. Quantification and statistical analysis All statistical analyses were performed using GraphPad Prism 8.01 software (GraphPad) ANOVA or Mann-Whitney test was performed for group comparisons. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 and ∗∗∗∗p < 0.0001 was considered as statistically significant with mean ± SEM. All of the statistical details of experiments can be found in the figure legends. Supplemental information Document S1. Figures S1–S15 and Tables S1 and S2 Data and code availability All data are available in the main text or the supplementary materials. Cryo-EM maps and molecular models have been deposited in the Electron Microscopy DataBank (EMDB) and the Protein DataBank (PDB), respectively, with accession codes EMD-15994 (map 1), EMD-15997 (map 2), EMD-16010 (map 3), EMD-16028 (map 4), EMD-16030 (map 5) and PDB: 8BEV (Wu NTD/RBD/W25) and PDB: 8BGG (Omicron NTD/RBD/W25). The accession numbers and more detailed information are also available in Supplementary information. This paper does not report original code. Any additional information required to reanalyse the data reported in this paper is available from the lead contact upon request. Acknowledgments We thank Drs Matthias Flotenmeyer and Lou Brillault from Center for Microscopy and Microanalysis (CMM), the University of Queensland, for facilitating cryo-EM work to be carried out and for scientific and technical assistance. We thank UQBR animal staff (Maya Patrick and Barb Arnts) and UQBR facility at the Australian Institute for Bioengineering and Nanotechnology (AIBN). We thank Research Computing Center (RCC), UQ, for partially providing computational resources. We thank Queensland Health Forensic and Scientific Services, Queensland, and the Kirby Institute University of New Souths Wales for providing SARS-CoV-2 virus isolates. We thank the National Biologics Facility, AIBN, for access to the Biacore 8K+ instrument for the SPR assay. We thank Prof. Andreas Suhrbier for providing Omicron BA.2 virus isolate. We thank Dr. Michael Landsberg for advice and helpful discussion. We thank Ian Mortimer (ITS Infrastructure Operations, UQ) for IT support. We thank the Max-Planck-Institut für molekulare Genetik (MPI-MG) for granting access to the TEM instrumentation of the microscopy and cryo-EM service group. N.M would like to thank Alpaca ‘Budda’ whose nanobody was isolated and Dzongsar Jamyang Khyentse Rinpoche for generous contribution. Funding. This work was supported by 10.13039/501100000925 NHMRC MRFF Coronavirus Research Response grant APP1202445 to D.W., K.C., P.R.Y., and T.M. Cryo-EM equipment at CMM is supported by ARC The Linkage Infrastructure, Equipment and Facilities (LIEF) scheme. N.M. was supported by UQ Research Stimulus. D. S. was supported by the 10.13039/501100001659 German Research Foundation (DFG) Emmy Noether Programme (SCHW1851/1-1) and by an 10.13039/501100003043 EMBO Advanced grant (aALTF-1650). The Chilean platform for the generation and characterization of Camelid Nanobodies to A.R.F is funded by ANID-FONDECYT No. 1200427; the regional Council of the “Los Rios region” projects FICR19-20, FICR21-01; FICR20-49 to R.J; the Bio & Medical Technology Development Program of the 10.13039/501100001321 National Research Foundation (NRF) of the Korean government (10.13039/501100014188 MSIT ) (NRF-2020M3A9H5112395); the ANID-MPG MPG190011 and ANID-STINT CS2018-7952 grants; and the EU-LAC T010047. DB is funded by 10.13039/100014013 UKRI Biotechnology and Biological Sciences Research Council (UKRI - 10.13039/501100000268 BBSRC , https://www.ukri.org/) Institute Strategic Program Grant (ISPG) to The 10.13039/501100000870 Pirbright Institute (BBS/E/I/00007034, BBS/E/I/00007030 and BBS/E/I/00007039). Author contributions N.M.: Experimental design; conceptualized; designed, generated, purified spike proteins, monoclonal antibodies, W25-Fc and control Fc; characterized spike proteins and nanobodies; generated W25/Omicron spike dataset including sample preparation, optimization and determination of optimal freezing conditions, image acquisition, performed single particle analysis (SPA); performed PRNTs; performed K18-huACE2 mouse challenge experiments; generated figures; drafted and edited manuscript. S.M.L.: Performed Wu spike Hexapro/W25 and Omicron spike Hexapro/W25 cryo-EM single particle and structure analysis. A.A.A.: Propagated live virus stock, performed PRNTs, performed K18-huACE2 mouse challenge experiments and data analysis. P.W.: Performed K18-huACE2 mouse challenge experiment. S.I.L.B.: Experimental design for the biodistribution studies, performed them with the help of J.T.J. and I.V.A. Y.S.L: Sample optimization for W25/Omicron spike cryo-EM, determination optimal freezing conditions, data acquisition, data analysis. N.T.: Performed cell-cell fusion experiments. B.L.: Assisted in cloning and purification of spike proteins and antibodies; assisted in K18-huACE2 mouse challenge experiments. G.V.N.: Identification of W25 cloning of W25 and performed thermostability and nebulization stability assays of W25. J.J.: Performed SPA analysis and focuses classification for W25/Omicron dataset. D.P.: Cloned, expressed, and purified published nanobodies. A.I.: Reformatted published nanobodies. J.D.S.: Generated live virus stock and generated VeroE6-TMPRSS6 cells for virus propagation. D.S.: Performed K18-huACE2 mouse challenge experiment. J.T.J.: Performed the biodistribution study with S.I.L.B and I.V.A. Y.C.: Performed optimization and purification of W25. J.B.: Assisted in Wu spike Hexapro/W25 cryo-EM single particle and structure analysis. I.V.A.: Performed the biodistribution study with S.I.L.B and J.T.J. J.H.: Performed optimization and purification of W25 for biodistribution assays. R.J.: Supervised vibrating mesh nebulizer stability assays. R.M.: Supervised vibrating mesh nebulizer stability assays. Z.M.C.: Performed optimization of W25. P.C.C.: supervised optimization of purification of W25 for cryo-EM and biodistribution. V.K.: Experimentally designed the biodistribution studies. T.M. and C.M.T.S.: Supervised Wu spike Hexapro/W25 cryo-EM single particle and structure analysis. A.A.K.: Obtained SARS-CoV-2 ancestral, Alpha, Beta isolates. Chief Investigator for BSL3 biocontainment facility at School of Chemistry and Molecular Biosciences. M.L.J.: Performed and analyzed SPR experiment. A.K.: Designed the biodistribution studies, supervised the work of J.T.J. M.M.H.: Designed the labeling and biodistribution studies, supervised the work of I.V.A. and S.I.L.B. K.A.J.: Supervised K18-huACE2 prophylactic mouse work. D. Sc.: Prepared and purified Wu Hexapro spike, prepared Wu spike Hexapro/W25 complex cryo-EM sample, performed and supervised Wu spike Hexapro/W25 cryo-EM and Omicron spike Hexapro/W25 single particle and structure analysis, generated figures, wrote initial manuscript draft. D.W.: Performed SPA of W25/Omicron dataset. The overall project was supervised, conceptualized, and edited by D.W., A.R.F., and D.S. Declaration of interests D.W., K.C., and P.R.Y are listed as inventors of ‘Molecular Clamp’ patent, US 2020/0040042. Supplemental information can be found online at https://doi.org/10.1016/j.isci.2023.107085. ==== Refs References 1 Lan J. Ge J. Yu J. Shan S. Zhou H. Fan S. Zhang Q. Shi X. Wang Q. Zhang L. Wang X. Structure of the SARS-CoV-2 spike receptor-binding domain bound to the ACE2 receptor Nature 581 2020 215 220 10.1038/s41586-020-2180-5 32225176 2 Cai Y. Zhang J. Xiao T. Peng H. Sterling S.M. Walsh R.M. Jr. Rawson S. Rits-Volloch S. Chen B. Distinct conformational states of SARS-CoV-2 spike protein Science 369 2020 1586 1592 10.1126/science.abd4251 32694201 3 Klein S. Cortese M. Winter S.L. Wachsmuth-Melm M. Neufeldt C.J. Cerikan B. Stanifer M.L. Boulant S. Bartenschlager R. Chlanda P. SARS-CoV-2 structure and replication characterized by in situ cryo-electron tomography Nat. Commun. 11 2020 5885 10.1038/s41467-020-19619-7 33208793 4 Ke Z. Oton J. Qu K. Cortese M. Zila V. McKeane L. Nakane T. Zivanov J. Neufeldt C.J. Cerikan B. Structures and distributions of SARS-CoV-2 spike proteins on intact virions Nature 588 2020 498 502 10.1038/s41586-020-2665-2 32805734 5 Zhou T. Tsybovsky Y. Gorman J. Rapp M. Cerutti G. Chuang G.Y. Katsamba P.S. Sampson J.M. Schön A. Bimela J. Cryo-EM structures of SARS-CoV-2 spike without and with ACE2 reveal a pH-dependent switch to mediate endosomal positioning of receptor-binding domains Cell Host Microbe 28 2020 867 879.e5 10.1016/j.chom.2020.11.004 33271067 6 Rogers T.F. Zhao F. Huang D. Beutler N. Burns A. He W.T. Limbo O. Smith C. Song G. Woehl J. Isolation of potent SARS-CoV-2 neutralizing antibodies and protection from disease in a small animal model Science 369 2020 956 963 10.1126/science.abc7520 32540903 7 Tortorici M.A. Beltramello M. Lempp F.A. Pinto D. Dang H.V. Rosen L.E. McCallum M. Bowen J. Minola A. Jaconi S. Ultrapotent human antibodies protect against SARS-CoV-2 challenge via multiple mechanisms Science 370 2020 950 957 10.1126/science.abe3354 32972994 8 Shi R. Shan C. Duan X. Chen Z. Liu P. Song J. Song T. Bi X. Han C. Wu L. A human neutralizing antibody targets the receptor-binding site of SARS-CoV-2 Nature 584 2020 120 124 10.1038/s41586-020-2381-y 32454512 9 Pinto D. Park Y.J. Beltramello M. Walls A.C. Tortorici M.A. Bianchi S. Jaconi S. Culap K. Zatta F. De Marco A. Cross-neutralization of SARS-CoV-2 by a human monoclonal SARS-CoV antibody Nature 583 2020 290 295 10.1038/s41586-020-2349-y 32422645 10 Hansen J. Baum A. Pascal K.E. Russo V. Giordano S. Wloga E. Fulton B.O. Yan Y. Koon K. Patel K. Studies in humanized mice and convalescent humans yield a SARS-CoV-2 antibody cocktail Science 369 2020 1010 1014 10.1126/science.abd0827 32540901 11 Jones B.E. Brown-Augsburger P.L. Corbett K.S. Westendorf K. Davies J. Cujec T.P. Wiethoff C.M. Blackbourne J.L. Heinz B.A. Foster D. The neutralizing antibody, LY-CoV555, protects against SARS-CoV-2 infection in nonhuman primates Sci. Transl. Med. 13 2021 eabf1906 10.1126/scitranslmed.abf1906 33820835 12 Kim C. Ryu D.K. Lee J. Kim Y.I. Seo J.M. Kim Y.G. Jeong J.H. Kim M. Kim J.I. Kim P. A therapeutic neutralizing antibody targeting receptor binding domain of SARS-CoV-2 spike protein Nat. Commun. 12 2021 288 10.1038/s41467-020-20602-5 33436577 13 Martinez D.R. Schaefer A. Gobeil S. Li D. De la Cruz G. Parks R. Lu X. Barr M. Manne K. Mansouri K. A broadly neutralizing antibody protects against SARS-CoV, pre-emergent bat CoVs, and SARS-CoV-2 variants in mice Preprint at bioRxiv 2021 10.1101/2021.04.27.441655 14 Garcia-Beltran W.F. Lam E.C. St Denis K. Nitido A.D. Garcia Z.H. Hauser B.M. Feldman J. Pavlovic M.N. Gregory D.J. Poznansky M.C. Multiple SARS-CoV-2 variants escape neutralization by vaccine-induced humoral immunity Cell 184 2021 2372 2383.e9 10.1016/j.cell.2021.03.013 33743213 15 Hoffmann M. Arora P. Groß R. Seidel A. Hörnich B.F. Hahn A.S. Krüger N. Graichen L. Hofmann-Winkler H. Kempf A. SARS-CoV-2 variants B.1.351 and P.1 escape from neutralizing antibodies Cell 184 2021 2384 2393.e12 10.1016/j.cell.2021.03.036 33794143 16 Chen R.E. Zhang X. Case J.B. Winkler E.S. Liu Y. VanBlargan L.A. Liu J. Errico J.M. Xie X. Suryadevara N. Resistance of SARS-CoV-2 variants to neutralization by monoclonal and serum-derived polyclonal antibodies Nat. Med. 27 2021 717 726 10.1038/s41591-021-01294-w 33664494 17 Wang P. Nair M.S. Liu L. Iketani S. Luo Y. Guo Y. Wang M. Yu J. Zhang B. Kwong P.D. Antibody resistance of SARS-CoV-2 variants B.1.351 and B.1.1.7 Nature 593 2021 130 135 10.1038/s41586-021-03398-2 33684923 18 Planas D. Veyer D. Baidaliuk A. Staropoli I. Guivel-Benhassine F. Rajah M.M. Planchais C. Porrot F. Robillard N. Puech J. Reduced sensitivity of SARS-CoV-2 variant Delta to antibody neutralization Nature 596 2021 276 280 10.1038/s41586-021-03777-9 34237773 19 Viana R. Moyo S. Amoako D.G. Tegally H. Scheepers C. Althaus C.L. Anyaneji U.J. Bester P.A. Boni M.F. Chand M. Rapid epidemic expansion of the SARS-CoV-2 Omicron variant in southern Africa Nature 603 2022 679 686 10.1038/s41586-022-04411-y 35042229 20 Liu L. Iketani S. Guo Y. Chan J.F.W. Wang M. Liu L. Luo Y. Chu H. Huang Y. Nair M.S. Striking antibody evasion manifested by the Omicron variant of SARS-CoV-2 Nature 602 2022 676 681 10.1038/s41586-021-04388-0 35016198 21 Yu J. Collier A.R.Y. Rowe M. Mardas F. Ventura J.D. Wan H. Miller J. Powers O. Chung B. Siamatu M. Neutralization of the SARS-CoV-2 omicron BA.1 and BA.2 variants N. Engl. J. Med. 386 2022 1579 1580 10.1056/NEJMc2201849 35294809 22 Sheward D.J. Kim C. Ehling R.A. Pankow A. Castro Dopico X. Dyrdak R. Martin D.P. Reddy S.T. Dillner J. Karlsson Hedestam G.B. Neutralisation sensitivity of the SARS-CoV-2 omicron (B.1.1.529) variant: a cross-sectional study Lancet Infect. Dis. 22 2022 813 820 10.1016/S1473-3099 35305699 23 Cameroni E. Bowen J.E. Rosen L.E. Saliba C. Zepeda S.K. Culap K. Pinto D. VanBlargan L.A. De Marco A. di Iulio J. Broadly neutralizing antibodies overcome SARS-CoV-2 Omicron antigenic shift Nature 602 2022 664 670 10.1038/s41586-021-04386-2 35016195 24 Larios Mora A. Detalle L. Gallup J.M. Van Geelen A. Stohr T. Duprez L. Ackermann M.R. Delivery of ALX-0171 by inhalation greatly reduces respiratory syncytial virus disease in newborn lambs mAbs 10 2018 778 795 10.1080/19420862.2018.1470727 29733750 25 Detalle L. Stohr T. Palomo C. Piedra P.A. Gilbert B.E. Mas V. Millar A. Power U.F. Stortelers C. Allosery K. Generation and characterization of ALX-0171, a potent novel therapeutic nanobody for the treatment of respiratory syncytial virus infection Antimicrob. Agents Chemother. 60 2016 6 13 10.1128/AAC.01802-15 26438495 26 Cunningham S. Piedra P.A. Martinon-Torres F. Szymanski H. Brackeva B. Dombrecht E. Detalle L. Fleurinck C. RESPIRE study group Nebulised ALX-0171 for respiratory syncytial virus lower respiratory tract infection in hospitalised children: a double-blind, randomised, placebo-controlled, phase 2b trial Lancet Respir. Med. 9 2021 21 32 10.1016/S2213-2600(20)30320-9 33002427 27 Valenzuela Nieto G. Jara R. Watterson D. Modhiran N. Amarilla A.A. Himelreichs J. Khromykh A.A. Salinas-Rebolledo C. Pinto T. Cheuquemilla Y. Potent neutralization of clinical isolates of SARS-CoV-2 D614 and G614 variants by a monomeric, sub-nanomolar affinity nanobody Sci. Rep. 11 2021 3318 10.1038/s41598-021-82833-w 33558635 28 Amarilla A.A. Modhiran N. Setoh Y.X. Peng N.Y.G. Sng J.D.J. Liang B. McMillan C.L.D. Freney M.E. Cheung S.T.M. Chappell K.J. An optimized high-throughput immuno-plaque assay for SARS-CoV-2 Front. Microbiol. 12 2021 625136 10.3389/fmicb.2021.625136 33643253 29 Yamasoba D. Kimura I. Nasser H. Morioka Y. Nao N. Ito J. Uriu K. Tsuda M. Zahradnik J. Shirakawa K. Virological characteristics of SARS-CoV-2 BA.2 variant Preprint at bioRxiv 2022 10.1101/2022.02.14.480335 30 Cui Z. Liu P. Wang N. Wang L. Fan K. Zhu Q. Wang K. Chen R. Feng R. Jia Z. Structural and functional characterizations of infectivity and immune evasion of SARS-CoV-2 Omicron Cell 185 2022 860 871.e13 10.1016/j.cell.2022.01.019 35120603 31 Hanke L. Vidakovics Perez L. Sheward D.J. Das H. Schulte T. Moliner-Morro A. Corcoran M. Achour A. Karlsson Hedestam G.B. Hällberg B.M. An alpaca nanobody neutralizes SARS-CoV-2 by blocking receptor interaction Nat. Commun. 11 2020 4420 10.1038/s41467-020-18174-5 32887876 32 Tunyasuvunakool K. Adler J. Wu Z. Green T. Zielinski M. Žídek A. Bridgland A. Cowie A. Meyer C. Laydon A. Highly accurate protein structure prediction for the human proteome Nature 596 2021 590 596 10.1038/s41586-021-03828-1 34293799 33 Mirdita M. Schütze K. Moriwaki Y. Heo L. Ovchinnikov S. Steinegger M. ColabFold - making protein folding accessible to all Preprint at bioRxiv 2022 10.1101/2021.08.15.456425 34 Tang Q. Owens R.J. Naismith J.H. Structural biology of nanobodies against the spike protein of SARS-CoV-2 Viruses 13 2021 2214 10.3390/v13112214 34835020 35 Barnes C.O. Jette C.A. Abernathy M.E. Dam K.M.A. Esswein S.R. Gristick H.B. Malyutin A.G. Sharaf N.G. Huey-Tubman K.E. Lee Y.E. SARS-CoV-2 neutralizing antibody structures inform therapeutic strategies Nature 588 2020 682 687 10.1038/s41586-020-2852-1 33045718 36 VanBlargan L.A. Errico J.M. Halfmann P.J. Zost S.J. Crowe J.E. Jr. Purcell L.A. Kawaoka Y. Corti D. Fremont D.H. Diamond M.S. An infectious SARS-CoV-2 B.1.1.529 Omicron virus escapes neutralization by therapeutic monoclonal antibodies Nat. Med. 28 2022 490 495 10.1038/s41591-021-01678-y 35046573 37 Cao Y. Wang J. Jian F. Xiao T. Song W. Yisimayi A. Huang W. Li Q. Wang P. An R. Omicron escapes the majority of existing SARS-CoV-2 neutralizing antibodies Nature 602 2022 657 663 10.1038/s41586-021-04385-3 35016194 38 Thakur N. Conceicao C. Isaacs A. Human S. Modhiran N. McLean R.K. Pedrera M. Tan T.K. Rijal P. Townsend A. Micro-fusion inhibition tests: quantifying antibody neutralization of virus-mediated cell-cell fusion J. Gen. Virol. 102 2021 jgv001506 10.1099/jgv.0.001506 33054904 39 Esparza T.J. Martin N.P. Anderson G.P. Goldman E.R. Brody D.L. High affinity nanobodies block SARS-CoV-2 spike receptor binding domain interaction with human angiotensin converting enzyme Sci. Rep. 10 2020 22370 10.1038/s41598-020-79036-0 33353972 40 Radvak P. Kwon H.J. Kosikova M. Ortega-Rodriguez U. Xiang R. Phue J.N. Shen R.F. Rozzelle J. Kapoor N. Rabara T. SARS-CoV-2 B.1.1.7 (alpha) and B.1.351 (beta) variants induce pathogenic patterns in K18-hACE2 transgenic mice distinct from early strains Nat. Commun. 12 2021 6559 10.1038/s41467-021-26803-w 34772941 41 Halfmann P.J. Iida S. Iwatsuki-Horimoto K. Maemura T. Kiso M. Scheaffer S.M. Darling T.L. Joshi A. Loeber S. Singh G. SARS-CoV-2 Omicron virus causes attenuated disease in mice and hamsters Nature 603 2022 687 692 10.1038/s41586-022-04441-6 35062015 42 Suryawanshi R.K. Chen I.P. Ma T. Syed A.M. Brazer N. Saldhi P. Simoneau C.R. Ciling A. Khalid M.M. Sreekumar B. Limited cross-variant immunity from SARS-CoV-2 Omicron without vaccination Nature 607 2022 351 355 10.1038/s41586-022-04865-0 35584773 43 Lopez Bernal J. Andrews N. Gower C. Gallagher E. Simmons R. Thelwall S. Stowe J. Tessier E. Groves N. Dabrera G. Effectiveness of Covid-19 vaccines against the B.1.617.2 (Delta) variant N. Engl. J. Med. 385 2021 585 594 10.1056/NEJMoa2108891 34289274 44 Dagan N. Barda N. Kepten E. Miron O. Perchik S. Katz M.A. Hernán M.A. Lipsitch M. Reis B. Balicer R.D. BNT162b2 mRNA Covid-19 vaccine in a nationwide mass vaccination setting N. Engl. J. Med. 384 2021 1412 1423 10.1056/NEJMoa2101765 33626250 45 Eyre D.W. Taylor D. Purver M. Chapman D. Fowler T. Pouwels K.B. Walker A.S. Peto T.E.A. Effect of Covid-19 vaccination on transmission of alpha and Delta variants N. Engl. J. Med. 386 2022 744 756 10.1056/NEJMoa2116597 34986294 46 Li D. Edwards R.J. Manne K. Martinez D.R. Schäfer A. Alam S.M. Wiehe K. Lu X. Parks R. Sutherland L.L. In vitro and in vivo functions of SARS-CoV-2 infection-enhancing and neutralizing antibodies Cell 184 2021 4203 4219.e32 10.1016/j.cell.2021.06.021 34242577 47 Van Heeke G. Allosery K. De Brabandere V. De Smedt T. Detalle L. de Fougerolles A. Nanobodies(R) as inhaled biotherapeutics for lung diseases Pharmacol. Ther. 169 2017 47 56 10.1016/j.pharmthera.2016.06.012 27373507 48 Wrapp D. De Vlieger D. Corbett K.S. Torres G.M. Wang N. Van Breedam W. Roose K. van Schie L. VIB-CMB COVID-19 Response TeamHoffmann M. Structural basis for potent neutralization of betacoronaviruses by single-domain camelid antibodies Cell 181 2020 1004 1015.e15 10.1016/j.cell.2020.04.031 32375025 49 Pymm P. Adair A. Chan L.J. Cooney J.P. Mordant F.L. Allison C.C. Lopez E. Haycroft E.R. O'Neill M.T. Tan L.L. Nanobody cocktails potently neutralize SARS-CoV-2 D614G N501Y variant and protect mice Proc. Natl. Acad. Sci. USA 118 2021 10.1073/pnas.2101918118 e2101918118 50 Huo J. Mikolajek H. Le Bas A. Clark J.J. Sharma P. Kipar A. Dormon J. Norman C. Weckener M. Clare D.K. A potent SARS-CoV-2 neutralising nanobody shows therapeutic efficacy in the Syrian golden hamster model of COVID-19 Nat. Commun. 12 2021 5469 10.1038/s41467-021-25480-z 34552091 51 Xiang Y. Nambulli S. Xiao Z. Liu H. Sang Z. Duprex W.P. Schneidman-Duhovny D. Zhang C. Shi Y. Versatile and multivalent nanobodies efficiently neutralize SARS-CoV-2 Science 370 2020 1479 1484 10.1126/science.abe4747 33154108 52 Schoof M. Faust B. Saunders R.A. Sangwan S. Rezelj V. Hoppe N. Boone M. Billesbølle C.B. Puchades C. Azumaya C.M. An ultrapotent synthetic nanobody neutralizes SARS-CoV-2 by stabilizing inactive Spike Science 370 2020 1473 1479 10.1126/science.abe3255 33154106 53 Koenig P.A. Das H. Liu H. Kümmerer B.M. Gohr F.N. Jenster L.M. Schiffelers L.D.J. Tesfamariam Y.M. Uchima M. Wuerth J.D. Structure-guided multivalent nanobodies block SARS-CoV-2 infection and suppress mutational escape Science 371 2021 eabe6230 10.1126/science.abe6230 33436526 54 Iketani S. Liu L. Guo Y. Liu L. Chan J.F.W. Huang Y. Wang M. Luo Y. Yu J. Chu H. Antibody evasion properties of SARS-CoV-2 Omicron sublineages Nature 604 2022 553 556 10.1038/s41586-022-04594-4 35240676 55 Shrestha L.B. Tedla N. Bull R.A. Broadly-neutralizing antibodies against emerging SARS-CoV-2 variants Front. Immunol. 12 2021 752003 10.3389/fimmu.2021.752003 34646276 56 Casalino L. Gaieb Z. Goldsmith J.A. Hjorth C.K. Dommer A.C. Harbison A.M. Fogarty C.A. Barros E.P. Taylor B.C. McLellan J.S. Beyond shielding: the roles of glycans in the SARS-CoV-2 spike protein ACS Cent. Sci. 6 2020 1722 1734 10.1021/acscentsci.0c01056 33140034 57 Henderson R. Edwards R.J. Mansouri K. Janowska K. Stalls V. Kopp M. Haynes B.F. Acharya P. Glycans on the SARS-CoV-2 spike control the receptor binding domain conformation Preprint at bioRxiv 2020 10.1101/2020.06.26.173765 58 Cao Y. Su B. Guo X. Sun W. Deng Y. Bao L. Zhu Q. Zhang X. Zheng Y. Geng C. Potent neutralizing antibodies against SARS-CoV-2 identified by high-throughput single-cell sequencing of convalescent patients' B cells Cell 182 2020 73 84.e16 10.1016/j.cell.2020.05.025 32425270 59 Huo J. Zhao Y. Ren J. Zhou D. Duyvesteyn H.M.E. Ginn H.M. Carrique L. Malinauskas T. Ruza R.R. Shah P.N.M. Neutralization of SARS-CoV-2 by destruction of the prefusion spike Cell Host Microbe 28 2020 445 454.e6 10.1016/j.chom.2020.06.010 32585135 60 Walls A.C. Xiong X. Park Y.J. Tortorici M.A. Snijder J. Quispe J. Cameroni E. Gopal R. Dai M. Lanzavecchia A. Unexpected receptor functional mimicry elucidates activation of coronavirus fusion Cell 176 2019 1026 1039.e15 10.1016/j.cell.2018.12.028 30712865 61 Hui K.P.Y. Ho J.C.W. Cheung M.C. Ng K.C. Ching R.H.H. Lai K.L. Kam T.T. Gu H. Sit K.Y. Hsin M.K.Y. SARS-CoV-2 Omicron variant replication in human bronchus and lung ex vivo Nature 603 2022 715 720 10.1038/s41586-022-04479-6 35104836 62 Li T. Cai H. Yao H. Zhou B. Zhang N. van Vlissingen M.F. Kuiken T. Han W. GeurtsvanKessel C.H. Gong Y. A synthetic nanobody targeting RBD protects hamsters from SARS-CoV-2 infection Nat. Commun. 12 2021 4635 10.1038/s41467-021-24905-z 34330908 63 Huo J. Le Bas A. Ruza R.R. Duyvesteyn H.M.E. Mikolajek H. Malinauskas T. Tan T.K. Rijal P. Dumoux M. Ward P.N. Neutralizing nanobodies bind SARS-CoV-2 spike RBD and block interaction with ACE2 Nat. Struct. Mol. Biol. 27 2020 846 854 10.1038/s41594-020-0469-6 32661423 64 Kubala M.H. Kovtun O. Alexandrov K. Collins B.M. Structural and thermodynamic analysis of the GFP:GFP-nanobody complex Protein Sci. 19 2010 2389 2401 10.1002/pro.519 20945358 65 Dang H.V. Chan Y.P. Park Y.J. Snijder J. Da Silva S.C. Vu B. Yan L. Feng Y.R. Rockx B. Geisbert T.W. An antibody against the F glycoprotein inhibits Nipah and Hendra virus infections Nat. Struct. Mol. Biol. 26 2019 980 987 10.1038/s41594-019-0308-9 31570878 66 Ekiert D.C. Kashyap A.K. Steel J. Rubrum A. Bhabha G. Khayat R. Lee J.H. Dillon M.A. O'Neil R.E. Faynboym A.M. Cross-neutralization of influenza A viruses mediated by a single antibody loop Nature 489 2012 526 532 10.1038/nature11414 22982990 67 Zheng S.Q. Palovcak E. Armache J.P. Verba K.A. Cheng Y. Agard D.A. MotionCor2: anisotropic correction of beam-induced motion for improved cryo-electron microscopy Nat. Methods 14 2017 331 332 10.1038/nmeth.4193 28250466 68 Rohou A. Grigorieff N. CTFFIND4: fast and accurate defocus estimation from electron micrographs J. Struct. Biol. 192 2015 216 221 10.1016/j.jsb.2015.08.008 26278980 69 Zivanov J. Nakane T. Forsberg B.O. Kimanius D. Hagen W.J. Lindahl E. Scheres S.H. New tools for automated high-resolution cryo-EM structure determination in RELION-3 Elife 7 2018 e42166 10.7554/eLife.42166 30412051 70 Punjani A. Rubinstein J.L. Fleet D.J. Brubaker M.A. cryoSPARC: algorithms for rapid unsupervised cryo-EM structure determination Nat. Methods 14 2017 290 296 10.1038/nmeth.4169 28165473 71 Jumper J. Evans R. Pritzel A. Green T. Figurnov M. Ronneberger O. Tunyasuvunakool K. Bates R. Žídek A. Potapenko A. Highly accurate protein structure prediction with AlphaFold Nature 596 2021 583 589 10.1038/s41586-021-03819-2 34265844 72 Emsley P. Lohkamp B. Scott W.G. Cowtan K. Features and development of Coot Acta Crystallogr. D Biol. Crystallogr. 66 2010 486 501 10.1107/S0907444910007493 20383002 73 Liebschner D. Afonine P.V. Baker M.L. Bunkóczi G. Chen V.B. Croll T.I. Hintze B. Hung L.W. Jain S. McCoy A.J. Macromolecular structure determination using X-rays, neutrons and electrons: recent developments in Phenix Acta Crystallogr. D Struct. Biol. 75 2019 861 877 10.1107/S2059798319011471 31588918 74 Watterson D. Wijesundara D.K. Modhiran N. Mordant F.L. Li Z. Avumegah M.S. McMillan C.L. Lackenby J. Guilfoyle K. van Amerongen G. Preclinical development of a molecular clamp-stabilised subunit vaccine for severe acute respiratory syndrome coronavirus 2 Clin. Transl. Immunol. 10 2021 e1269 10.1002/cti2.1269 75 Hsieh C.L. Goldsmith J.A. Schaub J.M. DiVenere A.M. Kuo H.C. Javanmardi K. Le K.C. Wrapp D. Lee A.G. Liu Y. Structure-based design of prefusion-stabilized SARS-CoV-2 spikes Science 369 2020 1501 1505 10.1126/science.abd0826 32703906 76 Ahmad J. Jiang J. Boyd L.F. Zeher A. Huang R. Xia D. Natarajan K. Margulies D.H. Structures of synthetic nanobody-SARS-CoV-2 receptor-binding domain complexes reveal distinct sites of interaction J. Biol. Chem. 297 2021 101202 10.1016/j.jbc.2021.101202 34537245 77 Modhiran N. Song H. Liu L. Bletchly C. Brillault L. Amarilla A.A. Xu X. Qi J. Chai Y. Cheung S.T.M. A broadly protective antibody that targets the flavivirus NS1 protein Science 371 2021 190 194 10.1126/science.abb9425 33414219 78 Suloway C. Pulokas J. Fellmann D. Cheng A. Guerra F. Quispe J. Stagg S. Potter C.S. Carragher B. Automated molecular microscopy: the new Leginon system J. Struct. Biol. 151 2005 41 60 10.1016/j.jsb.2005.03.010 15890530 79 Kidmose R.T. Juhl J. Nissen P. Boesen T. Karlsen J.L. Pedersen B.P. Namdinator - automatic molecular dynamics flexible fitting of structural models into cryo-EM and crystallography experimental maps IUCrJ 6 2019 526 531 10.1107/S2052252519007619 80 Pettersen E.F. Goddard T.D. Huang C.C. Meng E.C. Couch G.S. Croll T.I. Morris J.H. Ferrin T.E. UCSF ChimeraX: structure visualization for researchers, educators, and developers Protein Sci. 30 2021 70 82 10.1002/pro.3943 32881101 81 Katzelnick L.C. Coello Escoto A. McElvany B.D. Chávez C. Salje H. Luo W. Rodriguez-Barraquer I. Jarman R. Durbin A.P. Diehl S.A. Viridot: an automated virus plaque (immunofocus) counter for the measurement of serological neutralizing responses with application to dengue virus PLoS Negl. Trop. Dis. 12 2018 e0006862 10.1371/journal.pntd.0006862 30356267 82 Schepens B. van Schie L. Nerinckx W. Roose K. Van Breedam W. Fijalkowska D. Devos S. Weyts W. De Cae S. Vanmarcke S. An affinity-enhanced, broadly neutralizing heavy chain-only antibody protects against SARS-CoV-2 infection in animal models Sci. Transl. Med. 13 2021 eabi7826 10.1126/scitranslmed.abi7826 34609205 83 Herth M.M. Ametamey S. Antuganov D. Bauman A. Berndt M. Brooks A.F. Bormans G. Choe Y.S. Gillings N. Häfeli U.O. On the consensus nomenclature rules for radiopharmaceutical chemistry - reconsideration of radiochemical conversion Nucl. Med. Biol. 93 2021 19 21 10.1016/j.nucmedbio.2020.11.003 33232876 84 Pettersen E.F. Goddard T.D. Huang C.C. Meng E.C. Couch G.S. Croll T.I. Morris J.H. Ferrin T.E. UCSF ChimeraX: Structure visualization for researchers, educators, and developers Protein Sci. 30 2021 70 82 10.1002/pro.3943 32881101
PMC010xxxxxx/PMC10260200.txt
==== Front Neuro Oncol Neuro Oncol neuonc Neuro-Oncology 1522-8517 1523-5866 Oxford University Press US 10.1093/neuonc/noad073.130 noad073.130 Final Category: Genomics/Epigenomics/Metabolomics - METB AcademicSubjects/MED00300 AcademicSubjects/MED00310 METB-13. A SINGLE-CELL GENETIC IN VIVO LINEAGE-TRACING PLATFORM FOR MEDULLOBLASTOMA Jung Jangham University of California, San Francisco, San Francisco, USA Yu Bohyeon University of California, San Francisco, San Francisco, USA Hoare Owen University of California, San Francisco, San Francisco, USA Diaz Aaron University of California, San Francisco, San Francisco, USA 6 2023 12 6 2023 12 6 2023 25 Suppl 1 Abstracts from the 2023 Pediatric Neuro-Oncology Research Conference i33i33 © The Author(s) 2023. Published by Oxford University Press on behalf of the Society for Neuro-Oncology. 2023 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial License (https://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact [email protected] Abstract Comparative single-cell studies of group 3/4 medulloblastomas (MBs) and fetal brain have identified a common hierarchy of glutamatergic-lineage cell types and an apparent cell-of-origin in the rhombic lip. An increased understanding of the genetic regulators of stem-cell maintenance and differentiation in MB would clearly be of therapeutic benefit. There is a significant need for lineage-tracing systems to model MB malignant transformation, progression, and response to therapy. We developed in vivo and in organoid genetic lineage-tracing systems with a single-cell readout. A high-complexity lentiviral library expressing heritable polyadenylated molecular barcodes was transduced into D283 and D425 patient-derived cell lines. Barcoded cells were injected into the brains of immunocompromised mice. Both preimplantation barcoded cultures and the resulting mature tumors were profiled by single-cell RNA-sequencing (scRNA-seq). This captured both endogenous RNA and barcode transcripts. While only minimal barcode clash was detected in the preimplantation cultures, we observed a clonal expansion of barcodes in mature tumors which aligned with phylogenetics analysis of expressed mutations in endogenous RNA. Methods for assessing lineage coupling between transcriptional clusters and inferring their lineage relationship were developed. Ongoing efforts to recover barcodes from tumor sections via spatial transcriptomics will be presented. These studies fill a gap in status quo models of MB which are needed to leverage recent findings derived from single-cell analysis of human tumors. ==== Body pmc
PMC010xxxxxx/PMC10274881.txt
==== Front bioRxiv BIORXIV bioRxiv Cold Spring Harbor Laboratory 37333381 10.1101/2023.06.09.544399 preprint 3 Article A non-oscillatory, millisecond-scale embedding of brain state provides insight into behavior Parks David F. 1* Schneider Aidan M. 2* Xu Yifan 2 Brunwasser Samuel J. 2 Funderburk Samuel 2 Thurber Danilo 3 Blanche Tim 4 Dyer Eva L. 5 Haussler David 1 Hengen Keith B. 2† 1 Department of Biomolecular Engineering, University of California, Santa Cruz 2 Department of Biology, Washington University in Saint Louis 3 Exeter, NH 4 White Matter LLC, Seattle, WA 5 Department of Biomedical Engineering, Georgia Tech, Atlanta GA * These authors contributed equally to this work Author Contributions D.F.P. developed and ran the models and performed behavioral analyses, A.M.S. performed statistical analyses, wrote the paper, performed the single unit spiking analyses, and performed behavioral analyses, Y.X. provided sleep scoring expertise and animal care, S.J.B contributed substate analyses, S.F. provided sleep scoring expertise, D.T. performed flicker identification, T.B. provided intellectual and technical consultation, E.L.D. provided mentorship and consultation, D.H. provided mentorship and consultation, K.B.H. led, directed, and envisioned the project, edited figures, and wrote the paper. † Corresponding Author. [email protected] 27 6 2023 2023.06.09.544399https://creativecommons.org/licenses/by-nc-nd/4.0/ This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which allows reusers to copy and distribute the material in any medium or format in unadapted form only, for noncommercial purposes only, and only so long as attribution is given to the creator. nihpp-2023.06.09.544399.pdf Sleep and wake are understood to be slow, long-lasting processes that span the entire brain. Brain states correlate with many neurophysiological changes, yet the most robust and reliable signature of state is enriched in rhythms between 0.1 and 20 Hz. The possibility that the fundamental unit of brain state could be a reliable structure at the scale of milliseconds and microns has not been addressed due to the physical limits associated with oscillation-based definitions. Here, by analyzing high resolution neural activity recorded in 10 anatomically and functionally diverse regions of the murine brain over 24 h, we reveal a mechanistically distinct embedding of state in the brain. Sleep and wake states can be accurately classified from on the order of 100 to 101 ms of neuronal activity sampled from 100 μm of brain tissue. In contrast to canonical rhythms, this embedding persists above 1,000 Hz. This high frequency embedding is robust to substates and rapid events such as sharp wave ripples and cortical ON/OFF states. To ascertain whether such fast and local structure is meaningful, we leveraged our observation that individual circuits intermittently switch states independently of the rest of the brain. Brief state discontinuities in subsets of circuits correspond with brief behavioral discontinuities during both sleep and wake. Our results suggest that the fundamental unit of state in the brain is consistent with the spatial and temporal scale of neuronal computation, and that this resolution can contribute to an understanding of cognition and behavior. ==== Body pmcINTRODUCTION For nearly a century, stereotyped electrical waves traveling across the surface of the brain have been used to define neural activity patterns correlated with sleep/wake state (Berger, 1929). State-related waves are slow, travel across the isocortex, and are detectable in anatomically distributed structures (Gervasoni et al., 2004; Volgushev et al., 2006). These waves require multiple seconds of observation for identification and can be measured through the scalp, which displays filtered activity averaged across many millimeters of isocortex (Burle et al., 2015). In effect, state-related waves reflect a powerful mechanism of widely distributed electrical coordination, well poised to structure the global state of millions to billions of neurons over seconds to hours. As a result, the neural basis of sleep and wake states is generally understood to be a brain-wide phenomenon and orders of magnitude slower than the submillisecond precision of neural encoding of active behavior (Ding et al., 2016; Lee & Dan, 2012). However, contemporary studies of brain function and animal behavior are revealing complex state-related dynamics spanning multiple spatiotemporal scales, prompting calls for a reevaluation of traditional perspectives (Nir & de Lecea, 2023). Specifically, there is significant heterogeneity in oscillatory activity within both sleep and wake (Routtenberg et al., 1968; Sainsbury et al, 1987; Harris & Thiele, 2011; Engel et al., 2016; Lacroix et al., 2018; Simor et al., 2020). The low frequency waves that define sleep and wake travel across the isocortex but may be enriched or impoverished in different regions at any moment in time (Huber et al., 2006; Nir et al., 2011; Liu et al., 2022). There is evidence that sleep and wake states may intrude on one another to some extent, even in the course of normal behavior (Emrick et al., 2016; Jang et al., 2022; Soltani et al., 2019; Vyazovskiy et al., 2011; Harris & Thiele, 2011; Engel et al., 2016). The ability of the vertebrate brain to locally regulate states is exemplified by unihemispheric sleep in migratory birds (Rattenborg et al., 2016), marine mammals (Serafetinides & Brooks, 1971), and potentially even humans (Tamaki et al., 2016). The three broadest states, NREM sleep, REM sleep, and waking, are composed of a hierarchy of substates and nested neurophysiological events, such as sleep spindles and sharp wave ripples. Additionally, the notion of discrete states is somewhat misleading, as transitions between states are characterized by intermediate behavior and neurophysiology (Emrick et al., 2016; Jang et al. 2022; Soltani et al., 2019). Low frequency (<100 Hz) dynamics are the foundation of previous descriptions of sleep/wake as well as emerging evidence of brief and localized state-related phenomena. These dynamics influence many levels of neuronal activity. For example, delta waves (0.1 – 4 Hz) originating in the prefrontal isocortex drive high amplitude EEG waves that correlate with alternating periods of quiescence and bursting in isocortical neurons (Amzica and Steriade, 1998). Bursting and silence is, unsurprisingly, supported by hyperpolarization and depolarization in neuronal membranes (Volgushev et al., 2006). However, despite the fact that delta waves can be measured in a small patch of membrane, the fundamental unit of the wave is slow (250 ms to 10 s) and widely distributed, coordinating the activity of neurons over multiple centimeters (Volgushev et al., 2006). As a result, local measurements of low frequency waves are understood to reflect global, synchronizing forces (Buzsáki and Schomburg, 2015; Varela et al., 2001; Buzsáki and Draguhn, 2004; Molle et al., 2006). This is the basis of a widely accepted model of brain states in which neuronal activity (fast and local) in distinct circuits is systematically synchronized or desynchronized by oscillations (slow and distributed) (Girardeau & Lopes-Dos-Santos, 2021; Muñoz-Torres et al., 2023). Due to the physics of waves, it is impossible to consider a substrate of state shorter than the frequency of one cycle (although multiple cycles must be observed for practical purposes). As a result, frequency-based definitions of waves have a minimal resolution that is far slower and larger than the fundamental unit of neuronal activity: the action potential. A wave-based derivation that the fundamental unit of state is slow and global is, while perhaps true, tautological. We sought to learn the minimally resolvable structure of sleep and wake directly from raw data, independent of assumptions. Convolutional neural networks (CNNs) are well suited to extract the rules of sleep and wake at different spatiotemporal scales. Crucially, CNNs function well with noisy label data - in practice, if a CNN is trained using human labels (based on waves) but learns a more reliable latent signature of state, it can overrule the training label and disagree with high confidence (Rolnick et al., 2018) (Extended Data Fig.8). Using this bottom-up approach, we found that brain states are robustly resolvable in >= 40 ms of data from a single wire placed in any circuit in the brain. Removing oscillatory information below 750 Hz had no impact on accuracy. This suggests that the fundamental unit of brain state is at or below the spatiotemporal order of 101 ms and 102 μm. Fast and local embedding of state provides insights into brain function: individual circuits intermittently switch states independently of the rest of the brain, correlating with brief behaviors in both sleep and wake. RESULTS To empirically evaluate the minimally resolvable fingerprint of brain state in diverse circuitry throughout the mammalian brain, we analyzed a series of long-term, continuous multi-site recordings in freely behaving mice. Briefly, each of nine included animals was implanted with multiple 64-channel microelectrode arrays. To facilitate stable, high signal-to-noise recordings of single unit activity, arrays were composed of tetrodes. Arrays were attached to custom microelectronics in the headstage by flex cables, thus allowing an arbitrary geometry of stereotactic targets. After recovery, recordings (25 KHz) were conducted in the home cage continuously for between 4 and 16 weeks. Preamplification, analog-to-digital conversion, and multiplexing were achieved in the headstage, which was contained in a sparse frame of 3-D printed resin (Fig. 1A). The entire assembly weighed 4 g for a 512 ch (eight module) implant. High resolution video of behavior was collected in parallel. Even in the case of an eight-site implantation, animal movement was similar to unimplanted animals (Fig. S3). For inclusion in this study, we selected animals carrying arrays in a minimum of three and a maximum of eight separate brain regions (see Fig. S1 for a map of each animal). Given the diversity of implantation geometries between animals, we ensured that each brain region included in this study was recorded in at least two animals. As a result, we examined a grand total of 10 unique circuits across these recordings (Fig. 1A): CA1 hippocampus (CA1), primary visual cortex (VISp), nucleus accumbens (ACB), primary somatosensory cortex (SSp), primary motor cortex (MOp), caudate/putamen (CP), superior colliculus (SC), anterior cingulate cortex (ACA), retrosplenial cortex (RSP), and lateral geniculate nucleus (LGN). Within each animal’s dataset, we arbitrarily selected a 24 h period for analysis, ensuring only that there was high quality electrophysiological data in every brain region (i.e., evidence of single unit activity on some channels; Fig. S2; Supplemental Table 1). Note that while we used the presence of single unit activity to indicate high quality data, our analyses capitalize on the broadband signal (raw data) unless otherwise indicated (Fig. 1B). Three human experts independently sleep-scored the 24 h datasets, labeling waking, rapid eye movement (REM), and non-REM (NREM) sleep using polysomnography (Fig. S4). The three experts then met to address points of disagreement in each dataset, thus generating a consensus score that served as the basis of subsequent comparison. Substates within the three states, such as active/quiet wake and sleep spindles, were identified algorithmically and confirmed by human scorers. A CNN model can extract sleep and wake states from brief, local observations. Sleep scoring is traditionally conducted taking advantage of EEG, which reflects electrical waves on the dorsal isocortical surface, albeit with low spatiotemporal resolution. In addition, traditional sleep scoring requires some measure of animal motor output, such as EMG. While neural activity throughout the brain is influenced by sleep and wake (Emrick et al., 2016; Jang et al., 2022; Gent et al., 2018; Saper 2006), the degree to which sleep and wake are discernable solely on the basis of these dynamics is less explored. To empirically test this, and allow for circuit-by-circuit variability in state-related dynamics, we trained and tested a unique convolutional neural network (CNN) on the broadband raw data recorded in each circuit (Fig. 1C). Summarily, provided only locally-recorded raw neural data, the CNN attempts to learn rules that consistently predict human-generated labels, even though those labels were derived from EEG and behavior (Fig. 1D). Each CNN was shown 2.6 s of data at a time (2.6 s CNN) and was trained on 18 h of data and tested on a withheld, contiguous 6 h block of data that spanned a light/dark transition (Fig. S5). Despite the fact that CNNs only observed data from an individual circuit and had no information about animal movement, all three states (REM, NREM, and wake) were robustly separable in 2.6 s increments of data from every circuit. Accuracy was comparable to that of human experts (Fig. 1D–F). Notably, CNNs were effective at detecting brief states, such as microarousals, and in some cases identified examples missed by individual human scorers (Fig. S4F). CNNs can learn complex patterns from raw data (Rolnick et al., 2018), but understanding the rules learned by models is a widely recognized challenge (Ellis et al., 2022). One approach to this is ablation - by removing key components of a dataset, one can ascertain whether those components are integral to a model’s success. Because human experts rely on low frequency patterns (0.1–16 Hz; Hengen et al., 2016), we hypothesized that CNNs were using the same information. To test this, we applied a series of band-pass filters to data from each circuit in two animals (a total of 12 circuits) - after filtering, only a specific range of frequencies remained. In each circuit, a new series of 2.6 s CNNs was trained and tested on each of five band-passed datasets: A) 0.1 – 16 Hz, B) 100 – 200 Hz, C) 200 – 300 Hz, D) 300 – 400 Hz, and E) 400 – 500 Hz (Fig. 1G,H). Canonical state information is richly embedded in the 0.1 – 16 Hz band, but gamma power (30 – 100 Hz) changes by state as well. If CNNs required canonical oscillatory information, we expected accuracy to decline across these filters. Consistent with this, the 0.1 – 16 Hz model performed within 3% of broadband models (Fig. 1G). CNNs were progressively impaired by the remaining band-passes: all models reached chance levels by 300 – 400 Hz. Taken together, these data demonstrate that CNNs can robustly identify REM, NREM, and wake in nothing but 2.6 s of raw neural data from any circuit and that, consistent with nearly a century of observation (Berger, 1929), models rely on low frequency information to achieve this (Fig. 1H). Sleep/wake structures circuit activity at the kilohertz and 101 ms scale. The lowest band-pass suggests that complete information about states is available at frequencies below 16 Hz, and the progressive band-passes confirm that state information declines to 0 by ~ 200 Hz. However, these results do not rule out the possibility that a mechanistically distinct source of state embedding might emerge in ultra-fast frequencies. We sought to rule this out by training 1 s CNNs on datasets subjected to a series of increasing high-pass filters. In other words, we eliminated low frequency information progressively, allowing all faster information through at each step. Consistent with our prior results, high-passing at 0.3 Hz did not affect model accuracy. Similarly, high-passing above the delta band (4 Hz) had no impact on accuracy. Unexpectedly, this trend continued across almost 4 orders of magnitude (10−1 to 103): models maintained full accuracy at 750 Hz, and dropped to 0 only at > 3,000 Hz (Fig. 2A). Models failed in a stepwise fashion: accuracy generally remained above 70% prior to dropping to chance (Fig. S6A). The ~1 kHz range is understood to contain only fast neuronal events such as action potentials (Chung et al., 2017). Suggesting action potentials as an underlying mechanism, we observed that extracellular action potentials disappeared between 1,000 and 5,000 Hz (Fig. 2A). Across regions, 1 s CNNs trained on data high-passed at 750 Hz matched the performance of our baseline models (2.6 s broadband CNNs) as well as models restricted to low frequency information (1 s low-pass CNNs) (Fig 2B). Alongside the failure of band-pass models between 200 and 500 Hz, these data strongly suggest that there is a distinct source of brain state embedding in spike-band frequencies. Despite being separated from traditional metrics by multiple orders of magnitude, models restricted to kHz range data suffer no loss in accuracy. There is a simple explanation by which spike-band information could merely reflect low frequency rhythms present in EEG recordings: periods of action potential bursting and silence are shaped by low frequency field potentials (Vyazovskiy et al., 2011). Put simply, high-pass CNNs may be learning to reconstruct slow information in the timing of fast events. To investigate this possibility, we took advantage of the timescale of oscillatory information: for example, a 1 Hz wave cannot fit in windows smaller than 1 s. We trained another series of CNNs, this time systematically manipulating the size of input data. We progressively reduced the input size from 2.6 s down to a single sample point, or 1/25,000 s (Fig. 2C). Each model was provided with data from all channels within an individual circuit, but limited to a given input size. While median model performance declined monotonically as a function of input size, CNNs maintained accuracy significantly above chance down to 5 ms (0.43 +/−0.02, chance = 0.33). Given only 40 ms of input data, median CNN accuracy was nearly 70% (0.68, +/−0.02, chance = 0.33). Note that chance is precisely 0.33 - CNNs are provided with balanced datasets, and the mode of failure is characterized by universally guessing a single class (e.g., NREM). We excluded models which failed in this respect from statistical comparisons. Interestingly, we observed relatively balanced performance across the three states all the way down to 5 ms (Fig. 2D); in other words, CNNs did not struggle to identify one of the states relative to the other two. Taken with the high-pass results, these data suggest that all three states impose discriminable structure on neural activity at the millisecond timescale. Finally, we reasoned that CNNs may be learning to reconstruct slow waves in short intervals by virtue of spatial information gleaned by sampling 64 channels. Succinctly, because a wave travels in space, an instantaneous measurement of voltage on multiple electrodes could provide a measure of wavelength. To address this, we trained another series of CNNs exposed to progressively smaller and smaller input sizes. This time, however, we only passed in data from a single channel. While overall accuracy was slightly reduced and more variable in this heavily constrained situation, we observed qualitatively similar results in all single-channel models from all circuits of all animals (Fig. S6B). Many single-channel models performed significantly above chance down to 1 ms. Brain states impose millisecond-scale patterning in neuronal activity. We next sought to understand how sleep and wake state are embedded in neuronal activity at high frequencies and in small intervals. Logically, there are three possible mechanisms by which this phenomenon could be explained. First, models could learn that the average instantaneous voltage differs by state. This is the only possible explanation of performing above chance when testing on a single sample point. Second, variance might differ by state. In this case, two or more sample points could provide insight. Each of these explanations is decodable in fast/small samples, but does not necessitate a fast underlying process. Third, neural activity could be patterned at the millisecond timescale. In this case, the sequence of voltage measurements would carry state information and require a fast organizing process. To test these possibilities, we trained a series of single-channel CNNs (Fig. 3A) on the same datasets in two conditions. One was low-passed at 16 Hz and one was high-passed at 750 Hz (Fig. 3B). In effect, after high- and low- pass filtering, the same sample was passed into corresponding CNNs for labeling. In these parallel high- and low- pass datasets, we trained a series of CNNs on systematically decreased sample intervals. Finally, to test the hypothesis that high-frequency state embedding might reflect patterned information, we trained another parallel set of models but shuffled each sample prior to input (Fig. 3C). To summarize, progressively smaller chunks of data from a single wire were sleep scored by CNNs in four scenarios: intact low-pass, intact high-pass, shuffled-low pass, and shuffled-high pass. In the time intervals examined (2.6 s - 1/25,000 s), intact low-pass models performed only slightly better than their shuffled counterparts - a comparison of paired intact/shuffled low-pass models reveals a tendency to cluster near the unity line (Fig. 3D). The performance of shuffled and intact low-pass models was only separable down to 655 ms (p<.0001; linear mixed model: balanced_accuracy ~ sample size * high_low_pass * shuffle + (1|animal) + (1|circuit), where animal and circuit are random effects), after which they converged and then drop to near chance (<40%) by 40 ms (Fig. 3F). These data demonstrate that low frequencies carry some information about brain state at relatively short intervals, but that this information is distributional - the fact that shuffling does not impair performance reveals that models are learning from mean and/or variance. In contrast, intact 750 Hz high-pass models significantly outperformed their shuffled analogues at all sizes above 2.5 ms. A comparison of paired intact/shuffled models reveals a rightward shift toward intact models (Fig. 3E), consistent with the hypothesis that high frequency/short latency state information is temporally patterned (Fig. 3C). Not only did intact 750 Hz high-pass models outperform their shuffled variants, they also significantly outperformed all other models above 2.5 ms, including intact low-pass models at 2.6 s (p=0.0036). At 40 ms, the intact highpass model was the only model to perform above 40% (near-chance), yielding an accuracy of 51.3 +/− 0.9%. This is consistent with the results of 64 channel CNNs tested on broadband data (Fig. 2C), which suggested that 40 ms is the minimal sample size that maintains high accuracy. Taken together, these data reveal that high frequency temporal patterning carries state-related information on the timescale of 5 ms and above. Due to the effects of space filtering, it is likely that these patterns are generated within <100 μm of each recording site (Bédard et al., 2006). Surprisingly, the effect of shuffling was observed similarly on channels both with and without high amplitude spiking. That even non-spiking channels outperformed their shuffled counterparts reveals that the background noise in recordings contains temporally structured information about brain state. The most likely source of this information is low amplitude spiking of nearby neurons, i.e. multiunit-hash (Harris et al., 2016; Trautmann et al., 2019). This may account for the ability of high-pass CNNs to accurately identify state even when input samples contain no obvious single unit spiking. Substates and fast neurophysiological events. It has long been recognized that sleep and wake comprise many substates. In addition, stereotyped, intermittent neurophysiological events unfold in a state-dependent fashion. Sharp wave ripples are enriched in NREM sleep and quiet waking (Vanderwolf, 1969; Girardeau et al, 2009), while cortical ON- and OFF- states are primarily associated with NREM sleep (Vyazovskiy et al., 2011). These examples raise two questions about short and fast embedding of brain states. First, could it be that CNNs rely on fast substates and events (e.g., sharp wave ripples) to achieve their accuracy? This is unlikely, because fast events such as sharp wave ripples comprise a tiny fraction of total time in a state. Second, given that key neurophysiological events are defined by rapid and dramatic changes in activity, do such events lead to model errors and confusion? To address these questions, we used established algorithmic methods to detect sharp wave ripples (Karlsson & Frank, 2009; Kay et al, 2016), sleep spindles (Vallat & Walker, 2021), and cortical ON- and OFF- states (Vyazovskiy et al., 2011) in each of our datasets. In addition, we divided all of waking into two substates, active and quiet wake. We then evaluated the accuracy of 40 ms CNNs in and around these events. Cortical ON- and OFF- states during NREM did not confuse 40 ms models, despite the fact that individual ON- and OFF- states lasted longer than 40 ms (Fig. 4A,B). This suggests that, despite the fact that the CNN was only able to view data entirely within an ON- (or OFF-) state, there was a reliable latent signature of NREM. Cortical ON- and OFF- states can also occur in waking and REM sleep, albeit less frequently (Vyazovskiy et al., 2011; Funk et al., 2016). To a human observer, 40 ms of data within an OFF-state from REM, NREM and wake are identical, yet 40 ms CNNs correctly identified the superstate of these ON- and OFF- states (Fig. 4C,D). Likewise, the 40 ms CNN was robust to sharp wave ripples (Fig. 4E) and sleep spindles (Fig. 4F). Finally, network activity differences between quiet and active wake did not drive confusion (Fig. S8A,B). Microcircuits exhibit brief, independent sleep and wake-like states. We noted that CNNs of all sizes occasionally gave rise to high confidence errors and even disagreed with labels in training datasets (Fig. S7). Interestingly, these disagreements often spanned independent models trained in distinct circuits in the same brain. Given the confidence and independent reproducibility of these events, we reasoned that they may be evidence of two known phenomena. First, micro states have been described extensively (Halasz 1998). Micro arousals are brief (3–15 s) global arousals within NREM sleep (Halasz 1998; Ekstedt et al., 2004) into wake followed by a prompt return to NREM. Microsleeps are equivalent but involve a brief intrusion of sleep into waking. Second, slow waves characteristic of sleep may appear in portions of the cortex of awake animals in conditions of sleep deprivation (Vyzaovskiy et al., 2011; Alfonsa et al., 2023) or inactivity (Andrillon et al., 2021). This is referred to as local sleep. To account for these, we operationally defined microarousals as brief periods (< 20 s) occurring during otherwise consolidated sleep in which independent models in each circuit identified waking with high confidence (Fig. 5A, left). We applied an equivalent definition to microsleeps, identifying periods during waking when all models briefly identified a sleep state (Fig. 5A, middle). We eliminated these from our analyses. This excluded roughly 10% of nrem-to-wake disagreements and 5% of wake-to-nrem disagreements. After this, we were left with brief states that occurred in only a subset of circuits (i.e., not global), which we termed “flickers” (Fig. 5A, right). We initially hypothesized that flickering reflected local sleep (Vyazovskiy et al., 2011). To evaluate this, we deployed 1 s CNNs on 16 Hz low-passed data sets and identified flickers in this output. Approximately 60% of flickers identified in the broadband model were captured by low-pass models (Fig. 5B). These were consistent with human readable, low frequency local states, such as local slow waves during wake (Vyazovskiy et al., 2011) and REM (Funk et al., 2016), local changes in slow wave activity during NREM (Siclari and Tononi 2017), and local theta power during patterned behavior (Routtenberg et al., 1968; Sainsbury et al., 1986). We eliminated these from our analyses. We also directly evaluated whether flickers represent confusion driven by episodic oscillatory events. Neither ON/OFF states, sleep spindles, nor sharp wave ripples correlated positively with flickers. We reasoned that remaining events might represent transient, local shifts in the high frequency latent patterning of state. We also note that more than 99% of the excluded global events were also detected by low-pass models. This supports the efficacy of low-frequency monitoring tools (like EEG/LFP) for detection of microarousals and microsleeps. To test the sensitivity of CNNs to momentary switches of state, we created a positive control. We randomly transposed variable length snippets of data between stretches of REM, NREM and wake (Fig. 5C). 1 s CNNs recovered such synthetic flickers down to 10 ms, and recovered all examples > 66 ms. This suggests that flicker events identified by CNNs are consistent with brief intrusions of one state into another. Because such flickering appeared to be a plausible neurobiological event, we systematically quantified flicker rate and duration as a function of circuit. Flickers occurred at a rate of 10 – 50 per hour in each circuit (Fig. 5D. p=1e-22, main effect of circuit, linear mixed model: flicker_rate ~ circuit + (1 | animal), where animal is a random effect; see Fig. S9A for pairwise comparisons between circuits). Notably, there was a trend for isocortical regions to generate flickers at a higher rate than subcortical regions (p=0.071, Spearman Rank Correlation). Consistent with our synthetic control (Fig. 5C), flickers were observable down to 67 ms. Mean flicker duration varied significantly by circuit from 142 to 416 ms (Fig. 5D. p=1e-53, linear mixed model: flicker_duration ~ region + (1 | animal); see Fig. S9A for pairwise comparisons). Subcortical circuits produced significantly longer flickers than isocortical circuits (p=0.037, Spearman Rank Correlation). We observed all six possible flicker types, i.e., combinations of surrounding and flicker states. There was a significant effect of flicker type when comparing rate (Fig. 5E. p=3e-22, linear mixed model, flicker_rate ~ type + (1 | animal). See Fig. S9B for pairwise comparisons). The most frequent flicker type was NREM to REM, while the least was REM to NREM. Similarly, there was a significant effect of flicker type on duration (Fig. 5E. p=1e-306, linear mixed model, flicker_rate ~ type + (1 | animal). See Fig. S9B for pairwise comparisons). NREM to wake flickers were the longest, while REM to NREM flickers were the shortest. Interregional coordination is known to play a major role in the emergence and maintenance of sustained arousal states as well as the transient modulation of attentional states (Harris & Thiele, 2011; Poulet & Petersen 2008; Engel et al., 2016; Tan et al., 2014; Zagha et al., 2013; Mitra et al., 2016). We hypothesized that, if flickers represent neurobiologically meaningful substates, their timing might reflect functional connectivity between circuits. In other words, flickering in one circuit might strongly influence the state-related dynamics in downstream circuits. Consistent with this, subnetworks of circuits reliably exhibited coincident flickering far above chance (Fig. 5F). It is interesting to note that the LGN exhibited the highest rate of co-flickering, consistent with the broad connectivity of the thalamus. Flickering corresponds to transition-like spiking activity. Our results thus far demonstrate that CNNs identify unique state-related information in the 1 KHz range, which contains action potential traces (Fig. 6A). We hypothesized that single unit spiking might show evidence of state alterations during flickers. If so, this would suggest that flickering corresponds with a meaningful shift in neuronal spiking. Alternately, if unit activity were unaffected by flickering, this would suggest that these brief events detected in the broadband data are statistical anomalies. We spike-sorted all raw data and extracted ensembles of well-isolated single units from every circuit (Fig. 6B and S2). We then separated spiking into bins corresponding to each state, transitions between states, and flickers (Fig. 6C). Many units systematically changed their firing rates as a function of state - the proportion of units displaying alterations between states was maintained both in transitions and in flickers (wake to NREM example shown in Fig. 6D; all combinations shown in S10A). Likewise, population mean firing rates showed consistent shifts between states, in transitions, and in flickers (Fig. 6E, all combinations shown in S10B). Lastly, we sought to compare the effect of flickers and transitions on individual units whose activity varied substantially between states; it is these units in particular that might be expected to be modulated by flickering. We used principal components analysis (PCA) to examine correlations between the spiking patterns of single units (sets of 100 ISIs) on a circuit-by-circuit basis. By definition, the first principle component (PC1) represented the largest source of variance in unit activity - fortuitously, the axis defined by PC1 reliably separated each pair of states with near perfect accuracy, suggesting that ensemble spiking activity is sufficient to support arousal state classification. We then projected spiking activity during flickers and transitions onto these axes, allowing us to directly ask whether spiking during flickers showed evidence of a separation from the surrounding state. Ensemble spiking during both flickers and transitions was significantly shifted from the surrounding state towards the flicker state in all six flicker types and all four transition types (Fig. 6F). Transitions were significantly separable from the surrounding state in all regions, and flickers were separable in 9/10 regions (Fig. S10C, Extended Data Table 2–5). In 5/6 flicker types, flicker duration was uncorrelated with position along this axis (the exception was nrem-to-wake flickers, p<0.001). This suggests even very short flickers corresponded to spiking changes. The number of regions a flicker was detected in was also not significantly correlated with its position along this axis. Thus, time periods identified as flickers and transitions by the CNN correspond to significant alterations in single unit activity. Spiking activity during flickers and state transitions were generally not significantly different; this was the case in 3/4 transition types and 9/10 circuits (Extended Data Table 2–5). K-means clustering was applied along these axes to further explore the relationship between transitions and flickers on a recording-by-recording basis. On each PC1 axis, an optimal k was determined by the silhouette score k=2.79+/-0.10, and flickers co-clustered more frequently with transitions (68.53%) than the surrounding (63.13%) or predicted states (34.14%) (Rousseeuw 1987). This suggests, despite significant variability, the primary configuration along this axis is three clusters: one where the majority is the surrounding state, a second shared by transitions and flickers, and a third populated by the predicted state. Through the lens of single unit firing, flickers are consistent with circuit dynamics during transitions between the surrounding and predicted state. Flickering correlates with transient behavioral changes. As a whole, our data suggest that the minimally reliably resolvable unit of brain state is on the order of 101 ms and arises from local circuit activity. One possibility is that this is an epiphenomenon - in other words, resolving states at this scale provides no insight into brain function. Alternatively, understanding states at fast and local resolution could offer novel insight into the structure of natural behavior. Because flickers comprise brief changes in neuronal spiking, we hypothesized that they might correlate with animal behaviors that unfold on a similar timescale. We used optical flow - the difference between adjacent video frames - as a coarse measure of total activity. Optical flow differed significantly between all pairs of states (Fig. S4G). Interestingly, there was a slight but significant reduction in optical flow during REM relative to NREM, consistent with REM paralysis. Given that flickers represent momentary discontinuities in state, we next asked if similar discontinuities in behavior (Kramer & McLaughlin, 2001) corresponded with flickering. We algorithmically identified three forms of naturally occurring behavioral discontinuities: 1) motor twitches during extended NREM sleep (Fig. 7A), 2) brief pauses amidst extended locomotor sequences (Fig. 7B), and 3) momentary reductions in slight movements during NREM sleep, such as those associated with muscle tone and respiration (i.e., “freezing”, Fig. 7C). We examined the relationship of these discontinuities to flickering. Curious to know whether a potential relationship might depend on the number of circuits contributing to a flicker, we divided our analyses into single-region flickers and multi-region co-flickers. In all three examples of brief behavioral switching, we observed significantly increased flicker rates. Specifically, NREM-to-wake flickers were enriched during sleep twitches (Fig. 7A. p<0.001, linear mixed regression: flicker_rate ~ motor_state * flicker_type * n_circuits + (1 | animal/circuit), where circuit nested within animal is a random effect), wake-to-NREM flickers were enriched during brief pauses in high activity (Fig. 7B. p=0.0018), and NREM-to-REM flickers trended towards an increase during freezing in sleep (Fig. 7C. p=0.0722). In the first two cases (NREM-to-wake and wake-to-NREM), multi-fold increases in co-flickering appeared to drive these effects, suggesting that when more circuits are recruited during a flicker, there is more impact on behavioral structure. Paradoxically, co-flickering masked a significant enrichment of single-circuit NREM-to-REM flickers during freezing in sleep (Fig. 7I. p=0.006). These data suggest that short states identified in individual circuits correlate with short behavioral states within sleep and wake. By demonstrating a relationship between flickers and behavior, these data imply that consideration of brain states at the level <100 ms and kHz frequency has the potential to explain meaningful features of brain function and organization. DISCUSSION Sleep and wake states are widely assumed to arise from extended changes in global patterns of brain activity (Ding et al., 2016; Gervasoni et al., 2004; Lee & Dan, 2012; Volgushev et al., 2006). Here we report the unexpected finding that sleep and wake determine distinct, millisecond-scale patterns in microcircuits throughout the brain. We find that individual circuits regularly switch states (which we refer to as “flicker”) independently of other circuits. Flickers demonstrate that fast, state-related patterning of neuronal activity arises locally. Flickers, as brief neurophysiological discontinuities, are enriched in brief behavioral discontinuities in both sleep and wake. This implies that fast, state-related dynamics may contribute to cognition and behavior. Our data suggest a fundamental unit by which state determines brain activity that is distinct from low frequency oscillations. Our data also suggest that extremely short intervals of neural data contain latent structure that is determined by state. This information takes the form of distinct temporal patterns, although the mechanism behind such a phenomenon remains unclear. An immediate possibility is that of nested oscillations, similar to how 1 Hz slow waves in NREM coordinate sharp wave ripples (150 Hz). However, this is unlikely for three reasons. First, it appears that fast embedding can exist despite slow rhythms, a property illustrated by flickering. For example, within NREM slow waves, fast embedding can reveal a flicker (Fig. 5B, bottom two rows. Note that high frequency information is visibly altered during flickers despite ongoing slow wave activity). Second, there is a 500 Hz gap between low frequency rhythms and fast embedding (Fig. 1G,H, and Fig. 6A). Third, analysis of nested events and substates (Fig. 4) reveals that, amidst dramatic changes in low frequency activity, reliable fast embedding is trivially recoverable. As a result, it is not immediately clear how traditional wave information below 100 Hz could establish the kHz frequency fast embedding described here. However, extensive empirical evidence suggests that slow, broad signals are the foundation of sleep and wake states in the brain (Steriade et al., 1993; Gervasoni et al., 2004). Consistent with this, activation of broadly projecting nuclei in the midbrain and brainstem drives changes in brain state (Carter et al., 2010; Chen et al., 2018; Moruzzi & Magoun, 1949; Li et al., 2022). Our data thus appear paradoxical: sleep and wake can be experimentally controlled by slow and anatomically distributed mechanisms, yet cannot account for millisecond patterns nor local switches in state. This incongruity could be solved by simply shifting the mechanism of state from the global signal to the local circuit. In this framework, traveling waves and neuromodulatory tone function as coordinating signals, instructing the state of distributed circuits. From this perspective, the fundamental unit of sleep and wake states might comprise non-overlapping libraries of spike patterns available to each circuit. A slow and global signal would generally coordinate circuits (Buzsáki and Schomburg, 2015; Varela et a., 2001; Buzsáki and Draguhn, 2004; Molle et al., 2006), thus determining the behavioral macrostate of the organism. Future work will be required to understand how fast embedding interacts with global signals such as neuromodulators. Prior work in many species and brain regions describes shifts in neuronal activity that correlate with brain state (Noda and Adey, 1970; Abásolo et al., 2015; Watson et al., 2016; Levenstein et al., 2017; Brunwasser et al., 2022). However, such changes typically require extended observation, and while exhibiting statistically significant shifts across a large dataset, are too variable for use as a classification tool (Hengen et al., 2016; Xu et al., 2022). Here, small intervals of ensemble spiking activity are highly effective in separating each state (Fig. 6F). This is due to treating each neuron independently - it appears that, across many circuits, individual neurons are diverse in their state-dependent activity. While this is far more effective for classifying samples of sleep and wake than summary statistics, it requires a priori knowledge of each neuron’s profile, and the ability to track cells over extended periods of time. Despite this challenge, these data indicate that sleep and wake have complex effects on the activity of neighboring cells in the same microcircuit. Based on behavior and low frequency rhythms, short states, namely microarousals and microsleeps, have received attention in both humans (Torsvall 1987; Halász 2005; Carskadon and Rechtschaffen, 2011) and rodents (Franken et al., 1998; Soltani et al., 2019). While there is disagreement regarding precise definitions (Kjaerby et al., 2022; Lüthi et al., 2022), these are common experiences, such as momentarily slipping into sleep when drowsy. Local states have also been described previously. Utilizing sleep deprivation, Vyazovskiy et al., (2011) demonstrated the ability of the waking cortex to support patches of slow wave activity, i.e. “local sleep”. This influential work led to the identification of wake-like and sleep-like oscillatory activity during sleep and wake, respectively, in humans (Nobili et al., 2011; Nir et al., 2011; Hung et al., 2013), as well as rats (Emrick et al., 2016). In our data, we found robust evidence of these previously described phenomena - each of which was based on low frequency rhythms. By excluding these events from our subsequent analyses of flickers, we demonstrate that the contribution of fast embedding to brain function is mechanistically distinct from brief and local alterations in low frequency rhythms. We were initially surprised to record all six possible types of flickers, given that some pairings of states do not map onto transitions observed under normal conditions, in particular, wake-to-REM flickers. However, similar transitions in the vertebrate brain are not impossible: in narcolepsy, for example, REM-like neural activity arises during wake (Kroeger & de Lecea, 2009; Cao & Guillemiault, 2017). Given the robust ability of CNNs to recover synthetic wake-to-REM flickers and their presence in all animals and all circuits, there is clearly a latent, high frequency signature of REM that can arise during waking. How this relates to the content of behavior remains unclear. In this work, we take a neural-activity forward approach to discerning the minimal resolvable unit of brain state in anatomically and functionally diverse circuits. Two immediate approaches for future hold great promise. First, it will be fascinating to approach this problem in an unsupervised fashion; the number of reliably detectable substates may be far larger than appreciated. Second, a behavior-forward approach to state has great potential to further connect neuronal dynamics and brain function. Methods Mice All procedures involving mice were performed in accordance with protocols approved by the Washington University in Saint Louis Institutional Animal Care and Use Committee, following guidelines described in the US National Institutes of Health Guide for the Care and Use of Laboratory Animals. C57BL/6 mice from Charles River were used. Seven female and two male mice were used in this study. Mice were at least 100 d old at the beginning of recording (mean age 220 d). Mice were housed in an enriched environment and kept on a 12:12 h light:dark cycle. Mice had ad libitum access to food and water. Surgery All mice underwent multisite electrode array implantation surgery. Mice were anesthetized with isoflurane (1–2% in air) and administered slow release buprenorphine (ZooPharm, 0.1 mg kg−1). The mouse’s skull was secured in a robotic stereotaxic instrument (NeuroStar, Tubingen, Germany), and the skin and periosteum covering the dorsal surface of the skull was removed. Pitch, yaw, and roll were calculated to maximize alignment with stereotaxic atlases. Three to eight craniotomies (diameter 1–1.5 mm) were drilled using the automatic drilling function of the stereotaxic robot, and the dura was resected. In each animal, each of three to eight brain regions were implanted with a custom 64-channel tetrode-based array. Arrays were fixed (not drivable) and separated from headstage hardware by a flex cable, thus allowing an arbitrary geometry of multiple probes. Across nine mice, a total of 45 implants spanned 10 unique brain regions. Coordinates were as follows (AP/ML/DV relative to bregma and dura, in mm): CA1 (n=6, −2.54/−1.75/−1.5), VISp (n=5, −3.8/−2.8/−0.8), ACB (n=2, 1.25/−0.81/−3.94), SSp (n=6, −0.5/2.25/−0.8), MOp (n=7, 1.25/1.75/−1), CP (n=4, 0.5/1.52/−2.56), SC (n=3, −4.0/−1.0/−1.0), ACA (n=3, 0.6/0.8/−0.94), RSP (n=4, −1.5/0.3/−0.9), and LGN (n=2, −2.25/2.26/−2.40). Electrode bundles were lowered into brain tissue at a rate of 5mm/minute using a custom built stereotaxic vacuum holder. Anatomical location was confirmed post hoc via histological reconstruction (Extended Data Fig.1b). Arrays were secured with dental cement (C&B-Metabond Quick! Luting Cement, Parkell Products Inc; Flow-It ALC Flowable Dental Composite, Pentron), and headstage electronics (eCube, White Matter LLC) were bundled and secured in custom 3D-printed housing. Eight-module (512 channel) implants including arrays, cement, and headstage materials weighed approximately 4g. Mice were administered meloxicam (Pivetal, 5 mg kg−1 day−1 for three days) and dexamethasone (0.5 mg kg−1 day−1 for three days) and allowed to recover in the recording chamber for at least one week prior to recording. Recording Recordings were made using custom tetrode-based arrays. 16 tetrodes (64 channels) were soldered to a custom-designed PCB (5 mm × 5 mm × 200 μm) which stacked horizontally with a similarly sized amplifier chip. PCB/amplifier pairs further stacked with up to 7 additional pairs (total 8 modules, 512 channels). Recordings were conducted in an enriched home cage environment with social access to a litter mate through a perforated acrylic divider. Freely behaving mice were attached to a custom built cable with in-line commutation. Neuronal signals were amplified, digitized, and sampled at 25 kHz alongside synchronized 15–30 fps video using the eCube Server electrophysiology system (White Matter LLC). Recordings were conducted continuously for between two weeks and three months. Data and video were continuously monitored using Open Ephys (Siegle et al., 2017) and Watch Tower (White Matter LLC). 24 h blocks of data were identified for inclusion in these studies first by the absence of hardware and/or software problems, for example cable disconnects or dropped video frames, respectively, and second by the maximal yield of active channels. Beyond these criteria, selection of 24 h blocks was arbitrary. The same 24 h block was utilized for all recorded circuits in an individual animal. For experiments involving spike-sorted data, raw data were bandpass filtered between 350 and 7,500 Hz and spike waveforms were extracted and clustered using a modified version of SpikeInterface (Buccino et al., 2020) and MountainSort4 (Chung et al., 2017) with curation turned off. A custom XGBoost was used to identify those clusters constituting single units. Clusters identified as single units were manually inspected to confirm the presence of high amplitude spiking, stable spike amplitude over time, consistent waveform shapes, and little to no refractory period contamination. Probe Localization Following recording, mice were perfused with 4% formaldehyde (PFA), the brain was extracted and immersion fixed for 24h at 4° C in PFA. Brains were then transferred to a 30% sucrose solution in PBS and stored at 4° C until brains sank. Brains were then sectioned at 50 μm on a cryostat. Sections were rinsed in PBS prior to mounting on charged slides (SuperFrost Plus, Fisher) and stained with cresyl violet. Stained sections were aligned with the Allen Institute Mouse Brain Atlas (Allen Institute for Brain Science, 2012) and tetrode tracks were identified under a microscope (Extended Data Fig. 1B). Consensus Sleep Scoring Three experts independently scored the arousal state of each mouse using custom software (Extended Data Fig. 4D). Briefly, the LFP spectral power (0.1–60 Hz) was extracted from five channels selected from cortical implantations and averaged. Movement data was extracted from video recordings using DeepLabCut (Mathis et al., 2018) or EMG. LFP and movement data were preliminarily sleep-scored in 4 s epochs by a random forest. Human experts then evaluated LFP spectral density, movement, and random forest output in 4 s epochs. Scoring software provided immediate access to temporally aligned video for disambiguation. The three independently generated arrays of state assignments were then compared and all disagreements were identified. The three contributing human experts then, as a group, reevaluated each epoch of disagreement and generated a consensus state label. Convolutional Neural Network Construction and Experiments In this work, we used a single neural network model. We chose this architecture as it is particularly robust to label error, and is thus well suited to tasks in which there may be substantial variation in labels (Rolnick et al., 2018). Specifically, we coded a 1D eight-layer fully convolutional neural network (CNN) comprised of seven convolutional and one fully connected layer (size 150) in TensorFlow (Python). We used a stride of four to reduce input size four-fold in each layer. This size reduction dictated the number of layers in the network to support our largest model of 65,536 inputs (2.6s). We used a kernel size of 30, which indicates how many data points each layers sees in the convolution step. Features are built based on the kernel size at each layer. Layers contained the following number of filters: layer 1 had 320, layer 2 had 384, layer 3 had 448, layer 4–7 had 512. The output of the model is 3 values, a probability distribution over wake, non-rem and rem. This model architecture was the same for all input sizes from 1 sample to 2.6 s (65,536 data points). L2 weight regularization with 1e-6 was utilized. Learning rates were progressively reduced from 1e-4 to 5e-6 throughout training. Each layer utilized a standard ReLU activation function with the exception of the final convolutional layer which had no activation function applied. The loss function minimized by the CNN was softmax cross entropy (Good, 1952). CNN confidence was quantified as the entropy of the output distribution scaled to [0, 1]. Model parameters were chosen through a manual process of hyperparameter tuning. The manual hyperparameter tuning follows a coordinate descent approach in which each hyperparameter is varied until an optimal value is identified for that parameter, then the next parameter is varied, holding all others fixed. Hyperparameter selection was conducted early in the experimentation cycle and remained fixed throughout the experiments so as not to introduce added confounds. In all experiments, the CNN was presented with raw neural data. Two exceptions exist: 1) if data was low-passed below 16 Hz to examine the role of canonical oscillations, or 2) if data was high-passed above 750 Hz to examine the role of high-frequency (predominantly neuronal) activity. When these exceptions occur, it is for the purpose of interpretation and is directly specified in text, figure legends, and often the figure itself. Models were tasked with learning to predict sleep and wake states (REM, NREM, and wake) based on labels generated by a consensus amongst three expert human scorers. In some experiments, raw neural data were progressively ablated (see below) in preprocessing. To avoid common numerical problems, input data were linearly scaled by 10−3. The batch size during training was one, and the model was trained for 175,000 training steps. We selected a train/test split such that the training was composed of 18 consecutive hours of data, and the held-out test set comprised 6 h spanning a light/dark transition. Unless otherwise noted, step size was 1/15th of a second to match video fps. Model performance was evaluated using balanced accuracy (Brodersen et al., 2010; Kelleher et al., 2015) unless otherwise noted below. Model code will be available at github.com/hengenlab at the time of publication. CNN models functioned as follows. At each time step, a model was presented with an interval of data (the duration of this interval was experimentally varied, see below). The model’s task was to assign a probability of REM, NREM and wake to the central sample point. As a result, the task of each CNN was to learn the instantaneous state label at the center of a window of data, with only the information surrounding the center. Crucially, CNNs have no memory, and thus each observed segment of data is an independent decision. A more simple fully convolutional model was chosen over a more complex model such as ResNet with the following rationale: 1) to reduce computational burden, enabling the training of many thousands of models with varying input conditions, 2) it’s not clear that more complex models benchmarked against unrelated datasets would be a better fit for this data, 3) the existence of substantial label error dictates that there is more value in focusing on data than on model tuning. Basic Accuracy of Sleep and Wake States by Brain Region (Fig. 1D–F) To address the question of whether sleep and wake states are robustly embedded in the dynamics of the ten brain regions sampled in our recordings, we asked if CNNs could learn to decode sleep and wake from the broadband neural data within a single brain region (64 channels of data). Model input was 2.6 s (65,536 sample points) of 64 channels of raw neural data from each implant site (n=45 implants from N=9 animals, 45 total models). Extended Accuracy Evaluation in High Gamma Range (Fig. 1G–H) To examine the decodability of state information within the high gamma range, we implemented per-brain region models on neural data from 2 distinct animals (n=2 animals, with 12 probes in total derived from animals 2 and 7). The data had been manipulated via bandpass filtering to confine it to progressively amplified frequency information. Model input consisted of data that underwent the following bandpass filters: broadband, low-pass 16 Hz, 100–200 Hz bandpass, 200–300 Hz bandpass, 300–400 Hz bandpass, 400–500 Hz bandpass. Six models were executed per probe (72 total models). Note that EMG signal can be extracted from 300–500 Hz (Watson et al., 2016). The failure of models to learn in these bands suggests the signal is insufficient to support state classification in the recordings. We further confirmed that in a true EMG recording (750 Hz high-passed) models failed to train. The impact on the models’ capacity to learn from these frequency-specific datasets was closely monitored. Accuracy was evaluated using balanced accuracy directly on CNN test output. Progressive High Pass (Fig. 2A) To test whether sleep and wake states could be extracted from neural data absent the canonical waves that human scorers rely on, we progressively eliminated slow components of neural dynamics from all implants within an animal and measured the impact on the models’ ability to learn. Model input was ~1 s (24,576 sample points) of high-pass filtered neural data (3rd order Butterworth) from all channels within an animal (n=1 animal). 24 models were run. Specifically, models were trained and tested on neural data after the following high-pass filters were applied: 0 (raw), 0.5 Hz, 1 Hz, 2 Hz, 4 Hz, 6 Hz, 8 Hz, 12 Hz, 20 Hz, 50 Hz, 100 Hz, 250 Hz, 500 Hz, 750 Hz, 1,000 Hz, 1,500 Hz, 1,600 Hz, 1,700 Hz, 1,900 Hz, 2,000 Hz, 3,000 Hz, 5,000 Hz, 7000 Hz, 10,000 Hz. Accuracy was evaluated using balanced accuracy directly on CNN test output. Exploration of Inter-Regional Differences in Low Pass, High Pass, and Broadband Data (Fig. 2B) In order to assess the potential disparities between brain regions apparent in low pass, high pass, and unaltered broadband variations of the data, we implemented an extensive suite of models corresponding to each probe across all subjects (n=9 mice incorporating 45 probes in total). The identical model structure was applied in all scenarios, wherein the unmodified broadband data was subjected to preprocessing via a low pass filter at 16 Hz, a high pass filter at 750 Hz, or was left unfiltered. For each model, a single probe targeting a specific brain region (comprising 64 channels) was utilized, with a data input duration of 2.6 seconds (encompassing 65,536 data points). Accuracy was evaluated using balanced accuracy directly on CNN test output. Incremental Diminution of Input Size (Fig. 2C–D) In pursuit of ascertaining the minimal temporal duration required to accurately decode sleep and wake states, we orchestrated an expansive set of models corresponding to each probe in all subjects (n=9 mice comprising 45 probes in total), and trained them on an input size that was systematically reduced. Each model was informed by data from a single implant located within a particular brain region. A model with identical dimensions and hyperparameters was employed in each training phase, with the variation of increasingly truncated input sizes. The model’s task was to generate predictions solely based on the temporal window of data presented, devoid of any prior knowledge pertaining to the animal’s state. The progressive series of input sizes utilized were as follows: 2.6 s (65,536 data points), 1.3 s (32,768 data points), 327 ms (8,192 data points), 82 ms (2,048 data points), 41 ms (1,024 data points), 10 ms (256 data points), 5 ms (128 data points), 2.5 ms (64 data points), 1.3 ms (32 data points), 0.6 ms (16 data points), 0.3 ms (8 data points), 0.16 ms (4 data points), 0.08 ms (2 data points), and 0.04 ms (1 data point). Accuracy was evaluated using balanced accuracy directly on CNN test output. Model Training and Testing on Low-Pass Filtered Data (Fig. 3D) Data for this experiment were prepared using two manually selected channels, both spiking and non-spiking, with clear local field potentials. In some cases only one channel was included due to simple computational reasons. No exclusion criteria was applied beyond selecting only high quality channels. Each channel’s data underwent two transformations: in the first condition, it was left in its original, temporally intact form; in the second, the data was shuffled, reordering the data within each sampled segment. The shuffling process was executed by randomly drawing samples from the dataset and shuffling the data points within the sample, maintaining the same temporally intact sampling method as used for the unshuffled data. The models were then trained on the transformed data using a variety of input sizes from 2.6 seconds (65,536 data points) to 0.04 milliseconds (1 data point), generating a total of 2,028 single-channel models. Model performance was evaluated using balanced accuracy for each scenario. Model Training and Testing on High-Pass Filtered Data (Fig. 3E) The same channels and range of input sizes as in Fig. 3D were used for this part of the experiment. However, the preprocessing stage was different, employing a high-pass filter set at 750 Hz to remove low-frequency details from the raw data. The models were then trained on both the unaltered and shuffled high-pass data. Evaluation of Model Accuracy vs. Input Size (Fig. 3F) The data displayed in 3F is the same data from 3D and 3E, presented as a function of the input size. Flicker Experiments Flicker definition (Extended Data Fig. 7) We applied a series of criteria to CNN output to identify flickers for inclusion in our analyses. To start, we trained three CNNs to identify NREM, REM and wake based on raw broadband data (1 s input) from all 64 channels contained in each implantation site (n=45 implantation sites, 9 animals, 126 models). As a result, we had triplicate CNN-generated state scores for each recorded brain region. This was to avoid sporadic random error due to subtle inconsistencies in training. Often, we observed that CNN output would preemptively begin to increase confidence in an incoming state a few seconds prior to a global transition. To avoid transition-related ambiguity, we generally did not consider flickers within 30 s of a global transitions, but to account for the fact that some transitions were modified in time (e.g., a slow transition from quiet wake to NREM sleep). We manually evaluated CNN confidence surrounding each global state transition in the 24h of data from each implantation site in each animal in each replicate of the CNN (n=45 implantations from 9 animals, 3 replicates). We extended the window of exclusion around transitions in which evidence of the incoming state was present beyond the 30 s window. We then applied a series of confidence filters to the remaining data in each replicate. To avoid general periods of low confidence output, we identified and excluded any 35 s epochs with a mean confidence <75%. To restrict our analyses to high confidence flickering, we next eliminated 1s epochs with a mean confidence <75%. Together, these criteria excluded 20.91% of our recordings. In each replicate, we then assigned the high confidence state label to each time step (1/15th of a second), and collapsed the three model outputs into a single array by selecting the majority state at each interval. We then slid a 35 s rolling mode filter across the majority state array to create a label corresponding to the stable macro state surrounding every point in time. We defined flickers as disagreements between these two arrays. We had two further exclusion criteria to avoid overlap between flickers and previously described episodic arousal phenomena. First, flickers which co-occured across all probes in the animals were excluded. Specifically, this includes microstates, such as microsleeps and microarousals. Due to the 35 s modal filter of the majority state array, these had a maximum duration of ~20 seconds. Second, please note that flickers were detected in raw broadband data. However, to avoid overlap with events that could be visible in low-frequency data, we also detected flickers in models trained on data low-passed below 16 Hz. We excluded any flickers detected in the broadband which had any temporal overlap with a flicker detected in any region of the low-pass. This excludes low frequency phenomena previously described in EEG & LFP such as local slow waves in wake and REM. We also observed high-confidence, localized CNN errors (similar to flickers) amidst transitions between states (Fig. 6C). Consistent with progression along a transition’s time course, we found their rate of occurrence was inversely correlated with time to/from a human-labeled state change (p=0.016892). Because they have a similar confidence profile in the CNN’s predictions to flickers and they reflect transition dynamics, we used these as samples of transition spiking activity. To extract these events, we performed a comparable procedure to the previously mentioned flicker detection with two exceptions: 1) rather than excluding intervals of transitions, we excluded the complement (all non-transition intervals), 2) we did not exclude general periods of low confidence (based on 35s window) because the time course of a transition is often a low confidence period. Synthetic Flickers (Fig. 5C) To quantify the ability of the CNN to accurately detect brief intervals of a state B embedded within a containing state A, we constructed synthetic flickers. To do this, we identified all intervals of state A (NREM, REM, and wake) that were assigned the same label by all three human scorers, as well as the standard 2.6s CNN model (confidence >90%. See: Basic Accuracy of Sleep and Wake States by Brain Region). High confidence intervals were a minimum of 3 s in length. To simulate state B, we spliced segments of each state A into each other state (6 total combinations of REM, NREM, and wake into one another). Splices were 9.5, 19.0, 28.6, 47.6, 66.7, 133.3, and 333.3 ms. 100 splices of each duration were randomly selected from all high confidence intervals in the 24 h period and pasted into a randomly selected high confidence interval of another state. To illustrate with a specific example, consider a continuous segment of 47.6 ms of high confidence REM. This was randomly selected from thousands of >3 s intervals. A high confidence segment of wake was chosen at random from thousands of examples, and the 47.6 ms REM splice was pasted into a random location in the selected segment of wake. The insertion of the REM splice overwrote the corresponding section of wake data. We then asked whether splices were correctly identified by a CNN with a 655 ms window and a step size of 9.5 ms. We chose this step size to establish a functional lower limit of sensitivity while maintaining a reasonable computational load. We evaluated whether the CNN correctly identified any portion of the spliced state B. Coflicker Analysis (Fig. 5F) To understand whether flickers are a locally regulated (independent) or global phenomenon, we calculated the conditional probability of all pairs of regions flickering simultaneously according to: P(A∣B)=P(A∩B)P(B) Single Unit Analysis (Fig. 6D,E,F) Principal Component Analysis (PCA) was employed on each 12-hour spike-sorted block of CNN-classified regional recordings to evaluate whether single unit spiking could reliably discriminate arousal states. For each arousal state (wake, NREM, and REM) and flicker/transition type state (wake-to-nrem, wake-to-rem, nrem-to-wake, nrem-to-rem, rem-to-wake, and rem-to-nrem) we aggregated all interspike intervals (ISIs) for each neuron present during that state. We then randomly sampled 100 ISIs from these aggregated ISIs in contiguous runs of 1–10 ISIs (short snippets). We repeated this 100 times for each state. For each sample of a neuron’s spiking we calculated the mean ISI of the neurons (the inverse-mean instantaneous firing rate- gave similar results for this analysis). Prior to PCA, for each flicker type, we z-score normalized the firing rates for each neuron in the performed PCA. This z-score normalization was fit to the samples from the surrounding state and the predicted state (e.g. for a wake-to-nrem flicker, Wake and NREM appropriately). PC1 consistently separated these two classes with little overlap. We then transformed samples from the appropriate flicker-type and transition onto this axis. Transitions were available for four of the six state-pair combinations (not wake-to-REM or REM-to-NREM because they are not commonly observed). To enable us to evaluate trends across clustering blocks, we MinMax scaled the projections of samples along PC1. PCA projections were multiplied by −1 if the mean value of the projections of the predicted state were less than the mean value of the projections of the surrounding state. This aided visualization by ensuring samples of the surrounding state took on negative values and samples of the predicted state took on positive values to make visualization. To avoid the possibility of overestimating statistical effects, we took the mean projection for each surrounding state, predicted state, flicker state, and transition state in a clustering block. Error bars were challenging to formulate due to ISI resampling and possible dependency between spiking in two regions during co-flickering. A solution which is intuitively similar to standard deviation, with a focus on capturing animal-wise variance, was used. We first grouped all samples of spiking from a clustering block to create a mean for each. The SEM was then calculated with degrees of freedom multiplying the SEM by the square root of the number of animals (i.e. number of uncorrelated observations). This gave error bars that were roughly reflective of the trends seen in independent statistical significance tests. This is intuitively similar to the standard deviation. For each flicker type we fit a linear mixed effects model relating state to PC1 projection with animal as a random effect. We then performed an ANOVA and post-hoc EMMeans with Tukey correction. When grouping flickers by flicker-type we multiplied p-values by 6 for Bonferroni multiple hypothesis correction. When grouping flickers by region we multiplied p-values by 10. To evaluate the directional perturbation of portions of the population, we used the same sampled data used in PCA. For each flicker type, we iterated through each of the 100 samples for the surrounding state, flicker state, transition state, and predicted state in parallel. For each neuron, for each state we identified whether it or the surrounding state had the higher instantaneous firing rate. For each sample we evaluated the percent of neurons whose firing rate was increased (by any amount) relative to the sample of the surrounding state. As a negative control, we compared intervals of the surrounding state to eachother (sample 1 to sample 100, sample 2 to sample 99,...). For this negative control, the portion of units with increased firing should ideally be ~50% (representing chance), however in the case of some rare flicker types, it deviated from this due to sampling error. Flickers During Waking Behavior(Extended Data Fig. 4G, Fig. 7B,E,H) To correlate animal behavior with flickers we aligned video (15 fps) with neural data using a dedicated digital data channel synchronized with electrophysiological data acquisition, providing nanosecond-accurate timestamps per video frame (E3Vision, WhiteMatter, LLC, Seattle, WA). We use Farnebäck dense optical flow (Farnebäck, 2003) to measure movement. Specifically, motion was measured using dense optical flow which provides a per-pixel movement vector calculated between temporally adjacent video frames. Dictated by the video sampling rate, 15 per-pixel optical flow vectors were produced per second of recording. A video frame is segmented into three notable parts; 1) the subject animal, which is the core focus, 2) the background which should be excluded, and 3) the headstage tether cable which should be excluded, but produces the highest movement vectors due to its contrast with the background, immediate proximity to the camera, and quick jittery motion. To reduce the effect of cable movement, we excluded the top 3% of motion vectors. To focus analysis on the animal (as opposed to the background) we took the remaining top 10% of motion vectors. Explicitly, this retained the 87th percentile to 97th percentile of motion vectors produced by dense optical flow. Total animal movement was computed as the mean of these motion vectors for each frame. We next calculated two normalizations to allow comparisons across animals and over time. To accommodate drift and environmental changes, such as light/dark, we rescaled all motion vectors per each hour of video to a [0, 1] range. To normalize differences between mice we computed the 75th percentile of motion vectors per mouse and rescaled the movement values associated with each frame such that each animal’s 75th percentile movement values were aligned. Finally, we collapsed movement values above 1 to 1, resulting in a movement vector in the [0, 1] range. In other words, periods of high movement appeared as a sustained sequence of 1s. These normalization steps produced a movement value that is sufficiently invariant to light/dark cycles, differences in recording resolution, and variations in camera orientation and zoom to align locomotor states across animals. Three substates of waking were defined. Specifically, sustained high activity, low to intermediate activity, and brief pauses embedded within high activity. We computed two median filters over the normalized per-video frame movement values. Median filters are particularly useful in this context because they produce a smoothing with a sharp contour to the data that is cleanly delineable. The first was a rolling median using a 60 s window. At a threshold of 0.75, this median filter broadly segmented periods of sustained, high activity and periods of intermediate to low activity. Within periods of high activity, we then used a rolling median using a 0.66 s window to segment pauses from within sustained high activity. We manually evaluated a random subset of each locomotor state to confirm the accuracy of the algorithm parameters. Flickers During Sleeping Behavior(Fig. 7A,C,D,F,G,I) Using an analogous process to wake, three substates of sleep were defined. Specifically, sustained high activity (indicative of an extended microarousal), low activity, and brief high activity embedded within low activity (indicative of a twitch). To capture minor fluctuations in optical flow (indicative of brief stillness, or “freezing”) we inverted the trace of the magnitude of the optical flow and applied the scipy find_peaks function with default hyperparameters. We identified freezing as the 1 second following each of these peaks. Software and Statistical Analyses Data are reported as mean ± SEM unless otherwise noted. We used a mixed effects regression analysis with animal as a random effect followed by EMMeans with Tukey post hoc correction (LME4, R) (Bates et al., 2015) to determine statistical significance (p<0.05) unless otherwise noted. Mixed effects models comparing accuracy of models considered only models which trained successfully (accuracy > 33%). Supplementary Material Supplement 1 Acknowledgments This work is supported by NIH BRAIN Initiative 1R01NS118442-01 (KBH), and the Schmidt Futures Foundation SF 857 (D.H.). Through the Pacific Research Platform, this work was supported in part by NSF awards CNS-1730158, ACI-1540112, ACI-1541349, OAC-1826967, the University of California Office of the President, and the University of California San Diego’s California Institute for Telecommunications and Information Technology/Qualcomm Institute. Thanks to CENIC for the 100 Gpbs networks. Data Availability The datasets generated and/or analyzed in this study constitute tens of terabytes of raw neural broadband. The data are stored in a cost efficient manner not immediately accessible to the internet. We are excited to share data upon reasonable request, and as technical limitations make possible. Fig.1| CNNs learn robust signatures of sleep and wake from raw neural data in all brain regions. A, Recording protocol. (left) Image of a freely behaving mouse carrying a continuous multi-site recording device in our laboratory. (right) Brainrenders showing examples of implant/recording geometry from two of the nine animals in this study (Claudi et al., 2020). Colored regions indicate recorded circuits. CA1: hippocampus, VISp: primary visual cortex, ACB: nucleus accumbens, SSp: primary somatosensory cortex, MOp: primary motor cortex, CP: caudoputamen, SC: superior colliculus, ACA: anterior cingulate, RSP: retrosplenial cortex, LGN: lateral geniculate nucleus of the thalamus. B, 1 s of raw data from 8/64 channels in each of eight implanted brain regions in Animal 6. C, The architecture of the convolutional neural network (CNN) used to decode brain state (wake, REM, and NREM sleep) from raw data. Please see the methods section “Convolutional Neural Network Construction and Experiments” for more details. D, Human scoring of sleep state (top) versus eight CNNs (bottom) trained and tested independently on eight brain regions in the same animal (y axis of each row represents probability of each of three states at each timepoint). E, CNN accuracy relative to a consensus score is slightly but significantly better than individual human experts (left; gray bars, p=0.011, 1-way ANOVA). Colored bars: CNNs trained and tested in each of 10 brain regions show comparable accuracy, text in bars indicates n animals in each region (p=0.180, 1-way ANOVA). Upper inset shows 5 s of raw data (gray) and a 2.6 s sample (blue) that is used by the CNN for classification. F, Confusion matrices comparing human scorers (left) and CNNs (right) against consensus scores. Human scorers utilize full polysomnography. CNNs achieve balanced results across three states by observing only raw neural data. G, To test the source of state information learned by the CNN, models were trained and tested on filtered raw data from a subset of probes (12 circuits from two animals) : broadband (unaltered raw data), low-pass filtered at 16 Hz, and a series of progressively higher bandpass filters. Balanced accuracy of models is shown as a function of filter. H, Visual summary of the various filters applied (gray shows the same 5 s of data in each example). Blue is the 2.6 s window visible to the CNN for scoring. Fig.2| Brain state can be recovered in the kHz band and in only 100 – 101 ms of data. A, (top) 1 s of raw neural data subjected to increasing high-pass filters between up to 5,000 Hz. Note that spikes are eliminated between 1,000 and 5,000 Hz. (Bottom) CNN performance as a function of increasing high-pass filters in an example animal. High-pass filtering only decreased brain state information above 1,000 Hz (y=-9.892e-05*x+0.84,p<2e-16,r2=0.43). Lower inset: graphical depiction of classical oscillatory bands used to define states. B, Bar plot of summarizing model accuracy by circuit when trained and tested on 16 Hz low-passed data, 750 Hz high-passed data, or broadband (unfiltered) data from all animals. Broadband accuracy was slightly but significantly higher than low-pass accuracy (p=0.020; linear mixed effects: model_accuracy ~ filter * circuit + (1 | animal) where animal is a random effect) and high-pass accuracy (p=0.014). C, (top) 1 s of raw data (gray) overlaid with progressively shorter CNN input sizes (blue). CNNs could only observe an individual sample for each classification. (bottom) Balanced accuracy of CNNs trained and tested on progressively reduced input size. Each recorded circuit (n=45 implants, 9 animals, 10 circuits) is plotted individually (dashed lines), with the overall median performance illustrated (solid black line). D, Confusion matrices of mean performance of all models above chance at four input sizes: 1.3 s, 80 ms, 40 ms, and 5 ms. Values and colors represent class balanced accuracy for all models. Fig.3| Brain states pattern high frequency neuronal dynamics on the order of 100 to 101 ms. A, Experimental overview: model input. Depiction of a single wire (∅=12μm) placed in a local circuit. Curved lines represent the spatial effects of single neuron current dipoles that influence measured voltage at high frequencies. B, Experimental overview: paired low-pass and high-pass models. Parallel models are trained and tested on data from the same single channel (A), one model observing 16 Hz low-passed data, the other observing 750 Hz high-passed data. Blue boxes depict a 40 ms observation interval in each case. C, In each condition (low- and high- pass), another pair of models are made: intact and shuffled data (each sample is shuffled prior to training/testing). If shuffling does not reduce the ability of a CNN to decode state (left), state information must be recoverable from sample mean and variance. If temporal pattern is determined by state, shuffling will reduce accuracy (right). D,E, Two single channels were selected in each recorded circuit (both with and without high amplitude spiking) for examination in four conditions: high- / low- pass filtering and shuffle / intact comparison. In each condition, models were trained/tested at 13 input sizes (shown along the top) from 2.6 s (65,536 data points) to 0.04 ms (1 data point), yielding a total of 2,028 single-channel CNNs. D, 16 Hz low-passed single channel data. Each square shows a pair of intact/shuffled models, the square is colored by the circuit on which they are trained, and the size of the square indicates input size. E, The same as D, but for 750 Hz high-passed single channel data. F, Summary of models in D and E. Red indicates 750 Hz high-pass, blue indicates 16-Hz low-pass. Filled points are intact data, and open points are shuffled samples. 40 ms is indicated by dashed light-blue line for ease of comparison with other results. *** indicates p<0.001. Linear mixed effects balanced_accuracy ~ input size * filter * shuffle + (1|animal) + (1|circuit). Fig.4| Fast embedding of states is robust to diverse low frequency activity and neurophysiological events. A, Top- Broadband trace of several seconds of exemplary high-delta (0.1 – 4 Hz) activity during NREM sleep. Data are recorded in VISp. Red boxes indicate cortical ON and OFF states. Blue box shows the width of an individual input sample used by the 40 ms CNN to predict state. Middle- Raster of subset of VISp single units spiking. Bottom- Stacked barplot of 40 ms CNN prediction probabilities (the three colors in each bar show the probability that the corresponding sample came from each of the three states). To reduce computational burden, the CNN evaluates a 40 ms sample every 1/15 s (hence the slight gaps between samples). B, Zoomed 1 s view of ON/OFF-states in VISp. C, Example of a waking OFF-state in MOp. D, Example of a REM OFF-state in VISp. E, Example of a NREM sharp wave ripple (SWR) in CA1 hippocampus. Middle trace shows the same data as top trace but filtered to highlight SWR. F, Example of a NREM spindle in ACA. Middle trace shows the same data as top trace but filtered to highlight spindles. Fig.5| Individual circuits briefly switch states independently of the rest of the brain. A, Examples of three forms of disagreement between CNN classification and human consensus scoring of brain state. Top trace is neural broadband. Second row is human scoring of corresponding state. Bottom four rows are outputs of independent CNNs trained in each of four brain regions recorded in the same animal. The left column is a microsleep: all circuits (global) show a brief, high confidence intrusion of sleep into surrounding wake. The center column is a microarousal: all circuits show a brief, high confidence instruction of wake into surrounding sleep. The right column demonstrates a wake-to-NREM “flicker” in the anterior cingulate. Flickers are defined as high-confidence, non-global events that are not detected in low-pass models (B), and are distinct from transitions between states. B, Brief and local events were identified in models trained on 16 Hz low-passed data as well as models trained on broadband data. Events detected in both models were excluded from subsequent analyses. (top two rows) Examples of flickers detected in broadband data (gray trace) as well as low-passed data (teal). Red boxes denote the interval identified as a flicker in each model. Flicker type is shown on the left. (bottom four rows) Examples of flickers identified in the broadband but not low-passed data. C, (top) Schematic illustrating synthetic flicker positive control. Short segments of data from each state were transposed into segments of each other state. (bottom) Proportion of synthetic flickers captured by CNNs as a function of duration and flicker type. D, Flicker rate and duration vary significantly as a function of circuit (linear mixed effects model; Fig. S9A for post-hoc pairwise comparisons). There was a trend towards higher flicker rate in isocortical than subcortical regions (p=0.071, Spearman Rank Correlation). Subcortical regions exhibited significantly longer flickers than isocortical regions (p=0.037). E, (left) Mean rate of each flicker type per hour. (right) Mean duration by flicker type. See Fig. S9B for post-hoc pairwise comparisons. F, Coincident flickers (co-flickers) in two or more anatomically distinct circuits (top illustration, right brainrender) occurred significantly above chance (p<0.001, permutation test). Box plot of co-flickering probability by circuit. Shaded red area between dashed red lines indicates the range of chance levels (min to max) in all circuits. Solid red line is the mean chance level. Fig.6| Single neuron spiking shows evidence of flickers detected by CNNs. A, Schematic of state information as a function of frequency content. The shaded gray captures state information conveyed by canonical oscillations (δ-γ bar inset). The shaded teal illustrates state embedding in the kHz range, which contains action potential information (spike band). B, Example of a single unit. Top left is a broadband trace showing high signal-to-noise spiking. Top right shows the mean waveform (dark blue) of an extracted and spike-sorted single unit (individual traces shown in gray). Bottom left shows the mean waveform across the four channels of a tetrode. Bottom right is a histogram of the unit’s interspike intervals. Note the presence of a refractory period around 0–5 ms. C, Conceptual illustration of a flicker and a transition. D, The portion of units whose sampled instantaneous firing rate was different relative to a random sample of the surrounding state (wake): wake vs. wake (negative control), wake-to-NREM flicker, wake-to-NREM transition, and wake vs. NREM. See Fig. S10A for all state pairs. Error represents SEM. E, The mean single unit firing rate by circuit (color) during wake, a wake-to-NREM flicker, a wake-to-NREM transition, and NREM. See Fig. S10B for all state pairs. Box plots show the intermediate quartiles with outliers as swarm scatter colored by region. F. Mean scaled PC1 projections for the six flicker types- surrounding state (A: left square), predicted state (B: right square), A-to-B flicker (triangle), and A-to-B transition (circle) are shown. To incorporate the n animals into estimated variance, error bars are the SEM multiplied by the square-root of n animals. See Supplemental Tables 2–5 for significance of pairwise comparisons based on linear mixed effects model projection by sample type (i.e. flicker, transition, surrounding, predicted). Fig.7| Flickering predicts structure in free behavior. A, (top) Illustration of a twitch during inactive NREM sleep. (bottom) NREM-to-wake flicker rate before (dark blue), during (orange), and after (dark blue) twitches. Error shade is SEM. B, (top). Illustration of a brief pause during extended locomotion. (bottom) wake-to-NREM flicker rate before (light blue), during (gold), and after (light blue) pauses. Error shade is SEM. C, (top) Illustration of brief “freezing” during inactive NREM sleep (i.e., a slight but significant reduction in slight movements associated with muscle tone, respiration, etc.; Fig. S4G). (bottom) NREM-to-REM flicker rate before (green), during (slate), and after (green) freezing. Error shade is SEM. D, Rates (left) and durations (right) of NREM-to-wake single-region flickering (top brainrender) as a function of motion states shown in A. E, Same as D but for wake-to-NREM flickers during the states shown in B. F, Same as D but for NREM-to-REM flickers during the states shown in C. G, Rates (left) and durations (right) of NREM-to-wake co-flickers (multi-circuit: top brainrender) as a function of the motion states shown in A. H, Same as G but for wake-to-NREM flickers during the states shown in B. I, Same as G but for NREM-to-REM flickers during the states shown in C. Error bars for all plots are SEM. * p<0.05, ** p<0.01, *** p<0.001, linear mixed effects: flicker_rate ~ motor_state * flicker_type * n_circuits + (1 | animal/circuit) or flicker_duration ~ motor_state * flicker_type * n_circuits + (1 | animal/circuit), n=45 circuits, 9 mice. Competing Interests No competing interests disclosed Code Availability All relevant code from our lab, including software needed to run recordings or CNN models like ours is in python3 and will be publicly available at https://github.com/hengenlab at the time of publication. Other groups’ code including Open Ephys, SpikeInterface, and MountainSort4 is publicly available as specified in Methods. ==== Refs REFERENCES Abásolo D. , Simons S. , Morgado da Silva R. , Tononi G. , & Vyazovskiy V. V. (2015). Lempel-Ziv complexity of cortical activity during sleep and waking in rats. Journal of neurophysiology, 113 (7 ), 2742–2752.25717159 Alfonsa H. , Burman R. J. , Brodersen P. J. , Newey S. E. , Mahfooz K. , Yamagata T. , ... & Akerman C. J . (2023). Intracellular chloride regulation mediates local sleep pressure in the cortex. Nature neuroscience, 26 (1 ), 64–78.36510112 Amzica F. , & Steriade M. (1998). Electrophysiological correlates of sleep delta waves. Electroencephalography and clinical neurophysiology, 107 (2 ), 69–83.9751278 Andrillon T. , Burns A. , Mackay T. , Windt J. , & Tsuchiya N. (2021). Predicting lapses of attention with sleep-like slow waves. Nature Communications, 12 (1 ), 3657. 10.1038/s41467-021-23890-7 Bédard C. , Kröger H. , & Destexhe A. (2006). Model of low-pass filtering of local field potentials in brain tissue. Physical Review E, 73 (5 ), 051911. 10.1103/PhysRevE.73.051911 Berger H. (1929). Uber das Elektrenkephalogramm des Menschen. Archiv fur Psychiatrie, 87 , 527–570. Brunwasser S. J. , Farris C. , Elmore H. , Dyer E. L. , Nair K. B. , Whitesell J. D. , ... & Hengen K. B . (2022). Circuit-specific selective vulnerability in the DMN persists in the face of widespread amyloid burden. bioRxiv, 2022–11. Burle B. , Spieser L. , Roger C. , Casini L. , Hasbroucq T. , & Vidal F. (2015). Spatial and temporal resolutions of EEG: Is it really black and white? A scalp current density view. International Journal of Psychophysiology, 97 (3 ), 210–220. 10.1016/j.ijpsycho.2015.05.004 25979156 Buzsáki G. & Draguhn A. Neuronal oscillations in cortical networks. Science 304 , 1926–1929 (2004).15218136 Buzsáki G. , & Schomburg E. W. (2015). What does gamma coherence tell us about inter-regional neural communication?. Nature neuroscience, 18 (4 ), 484–489.25706474 Cao M. T. , & Guilleminault C. (2017). Narcolepsy: Diagnosis and management. In Principles and practice of sleep medicine (pp. 873–882). Elsevier. Carskadon M. A. , & Rechtschaffen A. (2011). Monitoring and staging human sleep. Principles and practice of sleep medicine, 5 , 16–26. Carter M. E. , Yizhar O. , Chikahisa S. , Nguyen H. , Adamantidis A. , Nishino S. , Deisseroth K. , & de Lecea L. (2010). Tuning arousal with optogenetic modulation of locus coeruleus neurons. Nature Neuroscience, 13 (12 ), 1526–1533. 10.1038/nn.2682 21037585 Chen K.-S. , Xu M. , Zhang Z. , Chang W.-C. , Gaj T. , Schaffer D. V. , & Dan Y. (2018). A Hypothalamic Switch for REM and Non-REM Sleep. Neuron, 97 (5 ), 1168–1176.e4. 10.1016/j.neuron.2018.02.005 29478915 Chung J. E. , Magland J. F. , Barnett A. H. , Tolosa V. M. , Tooker A. C. , Lee K. Y. , Shah K. G. , Felix S. H. , Frank L. M. , & Greengard L. F. (2017). A fully automated approach to spike sorting. Neuron, 95 (6 ), 1381–1394.e6. 10.1016/j.neuron.2017.08.030 28910621 Claudi F. , Tyson A. L. , & Branco T. (2020). Brainrender. A python based software for visualisation of neuroanatomical and morphological data (p. 2020.02.23.961748). bioRxiv. 10.1101/2020.02.23.961748 Ding F. , O’Donnell J. , Xu Q. , Kang N. , Goldman N. , & Nedergaard M. (2016). Changes in the composition of brain interstitial ions control the sleep-wake cycle. Science, 352 (6285 ), 550–555. 10.1126/science.aad4821 27126038 Ekstedt M. , Åkerstedt T. , & Söderström M. (2004). Microarousals during sleep are associated with increased levels of lipids, cortisol, and blood pressure. Psychosomatic medicine, 66 (6 ), 925–931.15564359 Ellis C. A. , Miller R. L. , & Calhoun V. D. (2022). A Systematic Approach for Explaining Time and Frequency Features Extracted by Convolutional Neural Networks From Raw Electroencephalography Data. Frontiers in Neuroinformatics, 16 . Emrick J. J. , Gross B. A. , Riley B. T. , & Poe G. R. (2016). Different Simultaneous Sleep States in the Hippocampus and Neocortex. Sleep, 39 (12 ), 2201–2209. 10.5665/sleep.6326 27748240 Engel T. A. , Steinmetz N. A. , Gieselmann M. A. , Thiele A. , Moore T. , & Boahen K. (2016). Selective modulation of cortical state during spatial attention. Science, 354 (6316 ), 1140–1144.27934763 Franken P. , Malafosse A. , & Tafti M. (1998). Genetic variation in EEG activity during sleep in inbred mice. American Journal of Physiology-Regulatory, Integrative and Comparative Physiology, 275 (4 ), R1127–R1137. Funk C. M. , Honjoh S. , Rodriguez A. V. , Cirelli C. , & Tononi G. (2016). Local slow waves in superficial layers of primary cortical areas during REM sleep. Current Biology, 26 (3 ), 396–403.26804554 Gent T. C. , Bassetti C. L. , & Adamantidis A. R. (2018). Sleep-wake control and the thalamus. Current opinion in neurobiology, 52 , 188–197.30144746 Gervasoni D. , Lin S.-C. , Ribeiro S. , Soares E. S. , Pantoja J. , & Nicolelis M. A. L. (2004). Global Forebrain Dynamics Predict Rat Behavioral States and Their Transitions. Journal of Neuroscience, 24 (49 ), 11137–11147. 10.1523/JNEUROSCI.3524-04.2004 15590930 Girardeau G. , Benchenane K. , Wiener S. I. , Buzsáki G. , & Zugaro M. B. (2009). Selective suppression of hippocampal ripples impairs spatial memory. Nature neuroscience, 12 (10 ), 1222–1223.19749750 Girardeau G. , & Lopes-Dos-Santos V. (2021). Brain neural patterns and the memory function of sleep. Science, 374 (6567 ), 560–564.34709916 Halász P. (2005). K-complex, a reactive EEG graphoelement of NREM sleep: an old chap in a new garment. Sleep medicine reviews, 9 (5 ), 391–412.16122950 Harris K. D. , Quiroga R. Q. , Freeman J. , & Smith S. L. (2016). Improving data quality in neuronal population recordings. Nature Neuroscience, 19 (9 ), 1165–1174. 10.1038/nn.4365 27571195 Harris K. D. , & Thiele A. (2011). Cortical state and attention. Nature reviews neuroscience, 12 (9 ), 509–523.21829219 Hengen K. B. , Torrado Pacheco A. , McGregor J. N. , Van Hooser S. D. , & Turrigiano G. G. (2016). Neuronal Firing Rate Homeostasis Is Inhibited by Sleep and Promoted by Wake. Cell, 165 (1 ), 180–191. 10.1016/j.cell.2016.01.046 26997481 Huber R. , Ghilardi M. F. , Massimini M. , Ferrarelli F. , Riedner B. A. , Peterson M. J. , & Tononi G. (2006). Arm immobilization causes cortical plastic changes and locally decreases sleep slow wave activity. Nature neuroscience, 9 (9 ), 1169–1176.16936722 Hung C. S. , Sarasso S. , Ferrarelli F. , Riedner B. , Ghilardi M. F. , Cirelli C. , & Tononi G. (2013). Local experience-dependent changes in the wake EEG after prolonged wakefulness. Sleep, 36 (1 ), 59–72.23288972 Jang R. S. , Ciliberti D. , Mankin E. A. , & Poe G. R. (2022). Recurrent Hippocampo-neocortical sleep-state divergence in humans. Proceedings of the National Academy of Sciences, 119 (44 ), e2123427119. Karlsson M.P. , and Frank L.M. (2009). Awake replay of remote experiences in the hippocampus. Nature Neuroscience 12 , 913–918.19525943 Kay K. , Sosa M. , Chung J.E. , Karlsson M.P. , Larkin M.C. , and Frank L.M. (2016). A hippocampal network for spatial coding during immobility and sleep. Nature 531 , 185–190.26934224 Kjaerby C. , Andersen M. , Hauglund N. , Untiet V. , Dall C. , Sigurdsson B. , ... & Nedergaard M . (2022). Memory-enhancing properties of sleep depend on the oscillatory amplitude of norepinephrine. Nature neuroscience, 25 (8 ), 1059–1070.35798980 Kramer D. L. , & McLaughlin R. L. (2001). The behavioral ecology of intermittent locomotion. American Zoologist, 41 (2 ), 137–153. Kroeger D. , & de Lecea L. (2009). The hypocretins and their role in narcolepsy. CNS & Neurological Disorders-Drug Targets (Formerly Current Drug Targets-CNS & Neurological Disorders), 8 (4 ), 271–280. Lacroix M. M. , de Lavilléon G. , Lefort J. , El Kanbi K. , Bagur S. , Laventure S. , ... & Benchenane K (2018). Improved sleep scoring in mice reveals human-like stages. BioRxiv, 489005. Lee A. M. , Hoy J. L. , Bonci A. , Wilbrecht L. , Stryker M. P. , & Niell C. M. (2014). Identification of a Brainstem Circuit Regulating Visual Cortical State in Parallel with Locomotion. Neuron, 83 (2 ), 455–466. 10.1016/j.neuron.2014.06.031 25033185 Lee S.-H. , & Dan Y. (2012). Neuromodulation of brain states. Neuron, 76 (1 ), 209–222. 10.1016/j.neuron.2012.09.012 23040816 Levenstein D. , Watson B. O. , Rinzel J. , & Buzsáki G. (2017). Sleep regulation of the distribution of cortical firing rates. Current Opinion in Neurobiology, 44 , 34–42.28288386 Li S. B. , Damonte V. M. , Chen C. , Wang G. X. , Kebschull J. M. , Yamaguchi H. , ... & de Lecea L . (2022). Hyperexcitable arousal circuits drive sleep instability during aging. Science, 375 (6583 ), eabh3021.35201886 Liu A. A. , Henin S. , Abbaspoor S. , Bragin A. , Buffalo E. A. , Farrell J. S. , ... & Buzsáki G . (2022). A consensus statement on detection of hippocampal sharp wave ripples and differentiation from other fast oscillations. Nature Communications, 13 (1 ), 6000. Loomis A. L. , Harvey E. N. , & Hobart G. (1935). Further observations on the potential rhythms of the cerebral cortex during sleep. Science, 82 (2122 ), 198–200.17844579 Lüthi A. , Franken P. , Fulda S. , Siclari F. , & Van Someren E. J. (2023). Do all norepinephrine surges disrupt sleep?. Nature neuroscience, 1–2. Mitra A. , Snyder A. Z. , Hacker C. D. , Pahwa M. , Tagliazucchi E. , Laufs H. , ... & Raichle M. E . (2016). Human cortical–hippocampal dialogue in wake and slow-wave sleep. Proceedings of the National Academy of Sciences, 113 (44 ), E6868–E6876. Molle M. , Yeshenko O. , Marshall L. , Sara S. J. , & Born J. (2006). Hippocampal sharp wave-ripples linked to slow oscillations in rat slow-wave sleep. Journal of neurophysiology, 96 (1 ), 62–70.16611848 Moruzzi G. , & Magoun H. W. (1949). Brain stem reticular formation and activation of the EEG. Electroencephalography and Clinical Neurophysiology, 1 (1 ), 455–473. 10.1016/0013-4694(49)90219-9 18421835 Muñoz-Torres Z. , Corsi-Cabrera M. , Velasco F. , & Velasco A. L. (2023). Amygdala and hippocampus dialogue with neocortex during human sleep and wakefulness. Sleep, 46 (1 ), zsac224.36124713 Nir Y. , & de Lecea L. (2023). Sleep and vigilance states: Embracing spatiotemporal dynamics. Neuron. Nir Y. , Staba R. J. , Andrillon T. , Vyazovskiy V. V. , Cirelli C. , Fried I. , & Tononi G. (2011). Regional slow waves and spindles in human sleep. Neuron, 70 (1 ), 153–169.21482364 Nobili L. , Ferrara M. , Moroni F. , De Gennaro L. , Russo G. L. , Campus C. , ... & De Carli F . (2011). Dissociated wake-like and sleep-like electro-cortical activity during sleep. Neuroimage, 58 (2 ), 612–619.21718789 Noda H. , & Adey W. R. (1970). Changes in neuronal activity in association cortex of the cat in relation to sleep and wakefulness. Brain Research, 19 (2 ), 263–275.4317333 Poulet J. F. , & Petersen C. C. (2008). Internal brain state regulates membrane potential synchrony in barrel cortex of behaving mice. Nature, 454 (7206 ), 881–885.18633351 Rattenborg N. C. , Voirin B. , Cruz S. M. , Tisdale R. , Dell’Omo G. , Lipp H.-P. , Wikelski M. , & Vyssotski A. L. (2016). Evidence that birds sleep in mid-flight. Nature Communications, 7 (1 ), 12468. 10.1038/ncomms12468 Rolnick D. , Veit A. , Belongie S. , & Shavit N. (2018). Deep Learning is Robust to Massive Label Noise. ArXiv:1705.10694 [Cs]. http://arxiv.org/abs/1705.10694 Rousseeuw P. J. (1987). Silhouettes: a graphical aid to the interpretation and validation of cluster analysis. Journal of computational and applied mathematics, 20 , 53–65. Routtenberg A. (1968). Hippocampal correlates of consummatory and observed behavior. Physiology & Behavior, 3 (4 ), 533–535. Sainsbury R. S. , Heynen A. , & Montoya C. P. (1987). Behavioral correlates of hippocampal type 2 theta in the rat. Physiology & behavior, 39 (4 ), 513–519.3575499 Saper C. B. (2006). Staying awake for dinner: hypothalamic integration of sleep, feeding, and circadian rhythms. Progress in brain research, 153 , 243–252.16876579 </>Serafetinides E. A. , & Brooks R. E. (1971). Electroencephalogram of the pilot whale globicephala scammoni, in wakefullness and sleep: Lateralization aspects. Siclari F. , & Tononi G. (2017). Local aspects of sleep and wakefulness. Current Opinion in Neurobiology, 44 , 222–227.28575720 Simor P. , van der Wijk G. , Nobili L. , & Peigneux P. (2020). The microstructure of REM sleep: why phasic and tonic?. Sleep medicine reviews, 52 , 101305.32259697 Soltani S. , Chauvette S. , Bukhtiyarova O. , Lina J. M. , Dubé J. , Seigneur J. , ... & Timofeev . (2019). Sleep–wake cycle in young and older mice. Frontiers in Systems Neuroscience, 13 , 51.31611779 Steriade M. , McCormick D. A. , & Sejnowski T. J. (1993). Thalamocortical oscillations in the sleeping and aroused brain. Science, 262 (5134 ), 679–685.8235588 Tamaki M. , Bang J. W. , Watanabe T. , & Sasaki Y. (2016). Night watch in one brain hemisphere during sleep associated with the first-night effect in humans. Current biology, 26 (9 ), 1190–1194.27112296 Tan A. Y. , Chen Y. , Scholl B. , Seidemann E. , & Priebe N. J. (2014). Sensory stimulation shifts visual cortex from synchronous to asynchronous states. Nature, 509 (7499 ), 226–229.24695217 Torsvall L. (1987). Sleepiness on the job: continuously measured EEG changes in train drivers. Electroencephalography and clinical Neurophysiology, 66 (6 ), 502–511.2438115 Trautmann E. M. , Stavisky S. D. , Lahiri S. , Ames K. C. , Kaufman M. T. , O’Shea D. J. , Vyas S. , Sun X. , Ryu S. I. , Ganguli S. , & Shenoy K. V. (2019). Accurate Estimation of Neural Population Dynamics without Spike Sorting. Neuron, 103 (2 ), 292–308.e4. 10.1016/j.neuron.2019.05.003 31171448 Vallat R. , & Walker M. P. (2021). An open-source, high-performance tool for automated sleep staging. Elife, 10 , e70092.34648426 Varela F. , Lachaux J.P. , Rodriguez E. & Martinerie J. The brainweb: phase synchronization and large-scale integration. Nat. Rev. Neurosci. 2 , 229–239 (2001).11283746 Vanderwolf C.H. (1969). Hippocampal electrical activity and voluntary movement in the rat. Electroencephalography and Clinical Neurophysiology 26 , 407–418.4183562 Volgushev M. , Chauvette S. , Mukovski M. , & Timofeev I. (2006). Precise Long-Range Synchronization of Activity and Silence in Neocortical Neurons during Slow-Wave Sleep. Journal of Neuroscience, 26 (21 ), 5665–5672. 10.1523/JNEUROSCI.0279-06.2006 16723523 Vyazovskiy V. V. , Olcese U. , Hanlon E. C. , Nir Y. , Cirelli C. , & Tononi G. (2011). Local sleep in awake rats. Nature, 472 (7344 ), 443–447. 10.1038/nature10009 21525926 Watson B. O. , Levenstein D. , Greene J. P. , Gelinas J. N. , & Buzsáki G. (2016). Network homeostasis and state dynamics of neocortical sleep. Neuron, 90 (4 ), 839–852.27133462 Xu Y. , Schneider A. , Wessel R. , & Hengen K. B. (2022). Sleep restores an optimal computational regime in cortical networks. bioRxiv, 2022–12. Zagha E. , Casale A. E. , Sachdev R. N. , McGinley M. J. , & McCormick D. A. (2013). Motor cortex feedback influences sensory processing by modulating network state. Neuron, 79 (3 ), 567–578.23850595
PMC010xxxxxx/PMC10312494.txt
==== Front bioRxiv BIORXIV bioRxiv Cold Spring Harbor Laboratory 37398315 10.1101/2023.05.31.542975 preprint 1 Article Learning structural heterogeneity from cryo-electron sub-tomograms with tomoDRGN Powell Barrett M. 1✉ Davis Joseph H. 12✉ 1 Department of Biology, Massachusetts Institute of Technology, Cambridge, MA 02139 2 Program in Computational and Systems Biology, Massachusetts Institute of Technology, Cambridge, MA 02139 AUTHOR CONTRIBUTIONS BMP and JHD conceived the work. BMP implemented the tomoDRGN method. BMP and JHD designed the experiments. BMP performed and analyzed the experiments. BMP and JHD wrote the manuscript. ✉ Correspondence: [email protected], [email protected] 02 6 2023 2023.05.31.542975https://creativecommons.org/licenses/by-nc-nd/4.0/ This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which allows reusers to copy and distribute the material in any medium or format in unadapted form only, for noncommercial purposes only, and only so long as attribution is given to the creator. nihpp-2023.05.31.542975.pdf Cryo-electron tomography (cryo-ET) allows one to observe macromolecular complexes in their native, spatially contextualized environment. Tools to visualize such complexes at nanometer resolution via iterative alignment and averaging are well-developed but rely on assumptions of structural homogeneity among the complexes under consideration. Recently developed downstream analysis tools allow for some assessment of macromolecular diversity but have limited capacity to represent highly heterogeneous macromolecules, including those undergoing continuous conformational changes. Here, we extend the highly expressive cryoDRGN deep learning architecture, originally created for cryo-electron microscopy single particle analysis, to sub-tomograms. Our new tool, tomoDRGN, learns a continuous low-dimensional representation of structural heterogeneity in cryo-ET datasets while also learning to reconstruct a large, heterogeneous ensemble of structures supported by the underlying data. Using simulated and experimental data, we describe and benchmark architectural choices within tomoDRGN that are uniquely necessitated and enabled by cryo-ET data. We additionally illustrate tomoDRGN’s efficacy in analyzing an exemplar dataset, using it to reveal extensive structural heterogeneity among ribosomes imaged in situ. ==== Body pmcINTRODUCTION Life relies on an array of large, dynamic macromolecular complexes to carry out essential cellular functions. The conformational flexibility and compositional variability in these complexes allow cells to mount targeted molecular responses to various stresses and stimuli. Structural biology has long aimed to visualize these diverse structures with the goals of gaining mechanistic insights into these responses and testing hypotheses related to macromolecular structure-function relationships. In pursuit of this goal, cryo-electron microscopy (cryo-EM) has proven to be a powerful tool for visualizing purified complexes with high resolution (Bai et al., 2015; Murata and Wolf, 2018). In cryoEM, ~104-107 individual particles are imaged by transmission electron microscopy (TEM), each from a single unknown projection angle. Single particle analysis (SPA) is then used to simultaneously estimate the most likely projection angle for each particle image and the k≥1 distinct 3-D volumes of the target complex, which, when projected to 2-D, are most likely to have produced the source dataset (Cheng et al., 2015). More recently, a number of tools have leveraged SPA datasets to deeply explore structural heterogeneity within these complexes (Chen and Ludtke, 2021; Dashti et al., 2020; Kinman et al., 2023; Punjani and Fleet, 2021; Sun et al., 2022; Zhong et al., 2021), dramatically expanding the range of insights and testable biological hypotheses that can be derived from cryo-EM. Cryo-electron tomography (cryo-ET) is a related imaging modality wherein a sample is repeatedly imaged from several known projection angles, enabling the reconstruction of a 3-D tomogram (Asano et al., 2016). As such, cryo-ET disentangles particles that overlap along a projection axis and enables the nanometer-scale 3-D visualization of highly complex samples, including subcellular volumes. Thus, cryo-ET affords the opportunity to inspect macromolecular structures in their native cellular context (Gemmer et al., 2023; Hoffmann et al., 2022; Lovatt et al., 2022; Xue et al., 2022), in contrast with cryo-EM’s typical requirement that particles be isolated from cells and purified. Sub-tomogram averaging (STA), a particle averaging approach analogous to SPA, is often employed in cryo-ET data processing. In STA, individual 3-D volumes, each a subtomogram corresponding to a unique particle, are extracted from the back-projected tilt series and are iteratively aligned to produce an average particle volume with increased SNR and resolution (Bharat and Scheres, 2016; Castano-Diez and Zanetti, 2019; Pyle and Zanetti, 2021; Zhang, 2019). Critically, STA can therefore offer insights to native protein complexes, enabling hypothesis generation in identifying unknown associated factors or novel complex ultrastructure. As with SPA, several tools have recently been developed to characterize heterogeneity among individual particles relative to the global average (Castano-Diez et al., 2012; Harastani et al., 2021; 2022; Himes and Zhang, 2018; Stolken et al., 2011), either during or after STA. Although these approaches have proven fruitful in answering specific biological questions such as nucleosome flexibility (Harastani et al., 2021; 2022), and ribosome heterogeneity (Himes and Zhang, 2018; Xue et al., 2022), each approach has specific constraints that limit their generality. For example, sub-tomogram PCA (Himes and Zhang, 2018) assumes heterogeneity can be modeled as a linear combination of voxel intensity, normal mode analysis (Harastani et al., 2021) requires a priori knowledge of an atomic model or density map to compute normal modes, and optical flow (Harastani et al., 2022) is inherently limited to conformational changes of the target particle in which the total voxel intensity across each sub-tomogram remains approximately constant. An unbiased and expressive tool to analyze heterogeneity is therefore highly desirable, particularly for in situ discovery of unexpected cofactors whose identity, binding site, and occupancy may be unknown. Here, we introduce tomoDRGN (Deep Reconstructing Generative Networks), a deep learning framework designed to learn a continuously generative model of per-particle conformational and compositional heterogeneity from cryo-ET datasets. TomoDRGN is related to our well-characterized cryoDRGN software (Kinman et al., 2023; Zhong et al., 2021), and therefore shares many overall design, processing, and analysis philosophies. As input, tomoDRGN uses particle images and corresponding metadata from upstream STA tools (Fig. 1a). It then learns to simultaneously embed each particle within a continuous low dimensional latent space and to reconstruct the corresponding unique 3-D volume (Fig. 1b). We have additionally developed and integrated software tools to visualize and interpret these outputs, and to integrate tomoDRGN outputs with external processing software for subsequent analyses, including contextualizing the tomoDRGN generated volumes within the in situ cellular tomography data. RESULTS A deep learning framework to reconstruct heterogeneous volumes from TEM tilt-series data. TomoDRGN was designed to efficiently train a neural network capable of: 1) embedding a collection of particles (each represented by multiple TEM images collected at different stage tilts) in a learned, continuous, low-dimensional latent space informed by structural heterogeneity; and 2) generating a 3-D volume for each individual particle using these embeddings. By design, cryoDRGN is unsuited for this task as it maps individual images to unique latent embeddings, which is expected for cryo-EM single particle datasets. Thus, cryoDRGN is not constrained to map multiple tilt images of the same particle to consistent regions of latent space, leading to uninterpretable learned latent spaces and generated volumes (see Discussion). To handle tilt-series data, we employed a variational autoencoder (VAE) framework (Kingma and Welling, 2013), featuring a purpose-built two-part encoder network feeding into a coordinate-based decoder network (Bepler et al., 2019; Zhong et al., 2019) (Fig. 1b). For each particle, the encoder network first uses encoder A (per tilt image) as a “feature extractor” to generate a unique intermediate embedding for each tilt image in a manner directly analogous to cryoDRGN’s encoder network. Encoder B then integrates these intermediate embeddings into a single latent embedding for the particle. The decoder network is supplied with this integrated latent embedding and a featurized voxel coordinate to reconstruct the signal at that coordinate. As in cryoDRGN, these operations are performed in reciprocal space. With this design, repeatedly evaluating the decoder network at multiple coordinates should allow for a rasterized reconstruction of the set of tilt images originally supplied to the encoder. Following a standard VAE (Kingma and Welling, 2013), we designed the network to be trained by minimizing a reconstruction loss between input and reconstructed images, and a latent loss quantified by the KL-divergence of the latent embedding from a standard normal distribution. Once trained, we expected a tomoDRGN network to enable detailed and systematic interrogation of structural heterogeneity within the input dataset. For example, similar to cryoDRGN, tomoDRGN’s learned latent space could be visualized either directly along any sets of latent dimensions or using a dimensionality reduction technique such as UMAP (Becht et al., 2018), where we have empirically found that distinct clusters often correspond to compositionally heterogeneous states, and diffuse, unfeatured distributions correspond to continuous structural variation. We reasoned that latent embeddings, sampled individually or following a well-populated path in latent space, could then be passed to the decoder to generate corresponding 3-D volumes for direct visualization. We predicted additional analysis could be performed in 3-D voxel space using standard cryoDRGN tools (Kinman et al., 2023; Sun et al., 2022). To complement tomoDRGN, we also constructed interactive tools to visualize and analyze heterogeneity in the spatial context of the original tomograms. Finally, we built tools to isolate particle subsets of interest for subsequent refinement with traditional STA software (Fig. 1c) as an iterative approach we speculated could maximize the value of a tomographic dataset. Sub-tomogram-specific image processing approaches. Having conceived the general tomoDRGN framework, we next considered additional image processing procedures that we hypothesized might improve model quality and computational performance. First, we noted that STA software tools commonly implement weighting schemes to model the signal-to-noise ratio (SNR) of each image as a function of the image tilt angle (i.e., electron pathlength through the sample) and cumulative dose (i.e., accumulated radiative damage) (Bharat et al., 2015; Grant and Grigorieff, 2015; Tegunov et al., 2021). Thus, we followed standard formulations for tilt weighting as the cosine of the stage tilt angle and dose weighting using fixed exposure curves, and we incorporated such weights into the reconstruction error calculated in tomoDRGN’s decoder network (Extended Data Fig. 1a). We expected such an approach would effectively downweigh the reconstruction loss of highly tilted and radiation damaged images, particularly at high spatial frequencies (Extended Data Fig. 1b–d). Second, tomoDRGN’s coordinate-based decoder is trained by evaluating a set of spatial frequencies per tilt image that, by default, is identical for all tilt images (i.e., independent of cumulative dose imparted at each tilt). However, prior work has shown that the SNR at a given spatial frequency can be maximized at an optimal electron dose (Hayward and Glaeser, 1979) and that during cryo-EM movie alignment, filtering spatial frequencies in each frame by their optimal dose can improve the aligned micrograph quality (Glaeser, 1979; Grant and Grigorieff, 2015). We therefore implemented a scheme applying optimal dose filtering to Fourier coordinates evaluated by the decoder during model training (Extended Data Fig. 1a). We expected that such filtering would restrict the set of spatial frequencies evaluated during decoder training without sacrificing 3-D reconstruction accuracy, thereby decreasing the computational burden of model training, particularly for high resolution datasets at large box sizes (Extended Data Fig. 1b–d). Finally, real-world datasets frequently contain particles missing some tilt images, often due to upstream micrograph filtering (Extended Data Fig. 2a). To flexibly handle such nonuniform input data, we implemented an approach that surveys the dataset for the fewest tilt images associated with a single particle (n), then randomly samples n tilt images from each particle during model training and evaluation (Extended Data Fig. 2b). Because this approach subsets and permutes tilt images at random, encoder B must learn a permutation-invariant function mapping from encoder A’s output (per tilt image) to the final latent space (per particle), and we hypothesized that this permutation-invariant learning goal might provide added regularization that could decrease overfitting. TomoDRGN robustly recovers simulated heterogeneity. To judge the efficacy of these architectural choices, we simulated (Baxter et al., 2009) cryo-ET particle stacks (see Methods) corresponding to four assembly states (B-E) of the ribosomal large subunit (LSU) from E. coli (Davis et al., 2016; Davis and Williamson, 2017) (Fig. 2a). We initially tested the ability of the isolated decoder network to perform a homogeneous reconstruction of the class E particles (i.e., no encoder was trained, and no latent space learned). We observed rapid convergence of the decoder network, with it reproducing the ground-truth density maps within 10 epochs (Fig. 2b). To assess tomoDRGN’s ability to faithfully embed and reconstruct structurally heterogeneous 3-D volumes, we next trained the full VAE network using particle stacks containing a labeled mixture of all four LSU structural classes. After training for 24 epochs, we observed four distinct clusters of latent embeddings by PCA and UMAP (Fig. 2c). Furthermore, the decoder network generated volumes from the center of each latent cluster that were consistent with the ground truth volumes (Fig. 2d). Finally, we quantified the fidelity of the embeddings to their corresponding ground truth volume classes on a per-particle basis. We observed a nearly one-to-one mapping between tomoDRGN particle embeddings and the correct ground truth class (Fig. 2e), indicating that the tomoDRGN network effectively learned discrete structural heterogeneity without supervision. We next assessed the benefits of our aforementioned reconstruction loss weighting, lattice coordinate filtering, and random tilt sampling approaches. Testing the weighting and filtering schemes on the homogeneous reconstruction of the LSU class E ribosomes, we observed modest improvements to final resolution with either or both schemes over using neither. Notably, however, the lattice coordinate filtering scheme led to large reductions in wall clock runtime and GPU memory utilization (Extended Data Fig. 1c–e, Supplementary Table 1). To assess the efficacy of the random sampling scheme, we compared heterogeneous networks trained on the 4-class LSU dataset with and without random tilt sampling. We observed higher average volume correlation coefficients (CC) for tomoDRGN volumes against ground truth volumes when using random sampling. Random sampling also provided our hypothesized robustness to model overfitting compared to sequential tilt sampling, as evidenced by the stable and elevated average CCs during further model training (Extended Data Fig. 2c). Finally, using the random sampling scheme, we observed an interpretable and wellfeatured latent space, even when using as few as 11 of the 41 available tilt images for each particle (Extended Data Fig. 2d–e). We additionally measured the accuracy and consistency of volumes generated from each such latent embedding to the corresponding ground truth volume, per particle per epoch, again observing robust performance with the random sampling scheme (Extended Data Fig. 2f). Notably, each of these metrics exhibited a dramatic drop in quality when only using a single tilt sampled per particle. This observation was consistent with our prediction that the mapping of one image to one latent embedding would be unsuitable for tilt series data. Combined, these strategies allowed efficient and flexible analysis of diverse input datasets, and we have benchmarked tomoDRGN performance for a range of network architectures (Extended Data Figs. 3–4, Supplemental Tables 2–4). Generally, we observe that performance is robust to network architecture hyperparameters, with slight improvements for deeper and narrower encoder A modules, and wider and shallower decoder modules. TomoDRGN uncovers structurally heterogeneous ribosomes imaged in situ. We next assessed tomoDRGN’s performance on the publicly available cryo-ET dataset EMPIAR-10499 (Tegunov et al., 2021), using it to analyze heterogeneity among chloramphenicol-treated ribosomes imaged in the bacterium Mycoplasma pneumoniae. Following published STA methods (Tegunov et al., 2021), we reproduced a Nyquist-limited ~3.5 Å resolution reconstruction of the 70S ribosome (Fig. 3a). We subsequently extracted corresponding ribosome images from the aligned tilt micrographs and used this particle stack to train a homogeneous tomoDRGN model. The tomoDRGN-reconstructed volume recapitulated high-resolution features observed in the STA map (Fig. 3a–c), highlighting the tomoDRGN decoder network’s ability to learn to accurately represent high-resolution structures. Encouraged by this result, we trained a heterogeneous tomoDRGN model on a down-sampled version of the particle stack and observed several distinct clusters in the resulting latent space (Fig. 4a, left). Generating volumes from these populated regions of latent space revealed that the majority of latent encodings corresponded to bonafide 70S ribosomes, as expected, whereas one subset corresponded to 50S ribosomal subunits, and another subset corresponded to apparent non-ribosomal particles (Fig. 4a, right). The non-ribosomal particles were further characterized by localizing them within each tomogram and providing them to RELION for ab initio reconstruction. Doing so suggested that most of these particles were false positive particle picks (Extended Data Fig. 5), highlighting tomoDRGN’s efficacy in sorting particles by structural heterogeneity generally, and in identifying errant particle picks specifically. We explored other approaches to separate 70S, 50S, and non-ribosomal particles, including using the trained tomoDRGN model to generate unique volumes corresponding to every particle’s latent embedding and either computing each volume’s similarity to the 70S STA map (Fig. 4b) or performing principal component analysis (PCA) in voxel space (Fig. 4c). Although these approaches produced results consistent with the clusters identified in latent space, for this dataset, the latent space clustering most clearly separated the 70S, 50S, and non-ribosomal particles. Guided by the latent embeddings, we next filtered out the non-ribosomal particles and used this ‘clean’ subset to train a new heterogeneous tomoDRGN model. The resulting latent space and generated volumes revealed an array of structurally heterogeneous ribosomes (Fig. 4d). Prior analyses of this dataset focusing on translation cycle heterogeneity (Xue et al., 2022) identified a major class with the A- and P-tRNA binding sites occupied by tRNAs and several minor classes featuring variable occupancy and positions of tRNAs in the A and P sites and EF-Tu in the A site. Consistently, we observed that these states are highly represented in our sampled volumes, and we further observed additional conformational and compositional heterogeneity throughout the ribosome (Supplemental Movie 1). For example, we found a set of volumes lacking EF-Tu and with helix 17 of the 16S bent towards the nowunoccupied EF-Tu binding site. In other volumes, we observed pronounced motions of the L1 stalk. We also observed volumes with clear density for r-proteins L7/L12 in the expected 1:4 ratio of L10CTD:L7NTD/L12NTD dimer of dimers. This observation was notable as this structural element is often unresolved in cryo-EM maps (Fromm et al., 2023; Stojkovic et al., 2020), likely due to this stalk’s dynamic nature and L7/L12’s ability to exchange off of the particle during purification (Chen et al., 2012). Observing this structure highlighted tomoDRGN’s ability to identify low abundance classes and emphasized the promise of the purification-free in situ structural analyses afforded by cryo-ET. We next applied MAVEn (Kinman et al., 2023; Sun et al., 2022), which has previously been used to systematically interrogate the structural heterogeneity of volume ensembles guided by atomic models. Here, we observed a broadly uniform distribution of occupancies for all queried structural elements (i.e., rRNA helices and r-proteins), with a notable exception of the 50S particle block, which lacks occupancy for any small subunit structural elements (Fig. 4e). We thus concluded that compositionally heterogeneous assembly intermediates were rare in this sample. TomoDRGN learns intermolecular heterogeneity. A grand promise of in situ cryo-ET is its potential to structurally characterize interactions between individual macromolecular complexes and their local environment (Tegunov et al., 2021; Turk and Baumeister, 2020). We hypothesized that tomoDRGN might perform well in this regard as its variational autoencoder architecture has a significant capacity to learn heterogeneity from the provided images, independent of the images being tightly or loosely cropped to the complexes under consideration. Indeed, our initial analysis revealed volume classes containing apparent intermolecular density truncated by the extracted box borders (Fig. 4d). To test tomoDRGN’s ability to analyze inter-complex structural heterogeneity, we extracted each ribosomal particle with a larger real-space box, effectively surveying the molecular neighborhood of each ribosome in the imaged cell. Training a new tomoDRGN model with these images revealed a similarly featured latent space with correspondingly diverse volumes (Fig. 5a). Many of the structures appeared to be disomes and trisomes, as previously reported (Tegunov et al., 2021), with measures of interparticle distance and the angular distribution to each ribosome’s nearest neighbor consistent with this interpretation (Fig. 5b). Intriguingly, the 50S population had an exceptionally broad distribution of nearest neighbor distances, and a subset of tomograms consisted almost exclusively of 50S ribosomes, whereas all other tomograms bore a more balanced distribution of all structural classes (Fig. 5b–c). Through this analysis, we observed a previously unreported ribosome structure with additional density corresponding to a lipid bilayer (Fig. 5a). To validate that this observed membrane density was not an artifact of the neural network, we mapped this set of apparently membrane-associated ribosomes to their original tomograms and observed that they exclusively corresponded to particles at the cell’s surface (Fig. 5d). To further identify residual heterogeneity within this group, we trained a new tomoDRGN model on this particle subset. We observed a relatively unfeatured latent space, with the majority (~80%, as quantified by MAVEn), of sampled volumes bearing a flexible periplasmic density protruding from the membrane (Fig. 5e). Notably, we observed significant relative motion between the ribosome and the adjacent membrane, indicating that the ribosome is not held in rigid alignment with the membrane and holotranslocon during translocation (Supplemental Movie 2). Traditional STA on this periplasmic-positive subpopulation of ribosomes further resolved the periplasmic density, as well as smaller arches of density connecting the ribosome to the membrane (Fig. 5f, Extended Data Fig. 6c). Rigid body docking using atomic models of likely transmembrane protein complexes into this density supported that we had identified ribosomes bound to SecDF, a subcomplex of the Sec holotranslocon with a relatively large extracellular globular domain encoded in the M. pneumoniae genome (Fig. 5f). This result highlighted the efficacy of tomoDRGN’s iterative particle curation and refinement approach in unveiling new structures buried in highly heterogeneous in situ datasets. DISCUSSION In this work, we introduce tomoDRGN, which, to our knowledge, is the first neural network framework capable of simultaneously modeling compositional and conformational heterogeneity from cryo-ET data on a per-particle basis. TomoDRGN achieves this using a bespoke deep learning architecture and numerous accelerations designed to exploit redundancies inherent to cryo-ET data collection. We note that the major heterogeneity analyses demonstrated in this manuscript were also tested with cryoDRGN (Zhong et al., 2021). However, cryoDRGN ultimately did not match tomoDRGN’s performance on cryo-ET data as it incorrectly classified simulated data, predominantly learned non-biological structural heterogeneity, and produced highly variable latent embeddings and volumes for different tilt images of the same particle (Extended Data Figs. 7–9). We note that an alternative approach of mapping single sub-tomogram volumes to single latent coordinates would theoretically function within the cryoDRGN framework but would: 1) be less computationally tractable due to cubic scaling of the number of voxel coordinates to be evaluated per particle; and 2) may be predisposed towards learning heterogeneity driven by missing wedge artifacts common to sub-tomogram volumes. Other tools to explore conformational heterogeneity from a cryo-ET dataset exist (Bharat and Scheres, 2016; Harastani et al., 2021; 2022; Himes and Zhang, 2018). However, they each rely on some degree of imposed prior knowledge, either in the form of “mass conservation” to describe continuous changes from a consensus structure, which is often derived from a provided atomic model (Harastani et al., 2021; 2022); assumptions of linear relationships between structures (Himes and Zhang, 2018); or the assertion that a small number of discrete structures exist (Bharat and Scheres, 2016). In contrast, the cryoDRGN/tomoDRGN approaches provide a greater degree of generality that we have found enables largely unsupervised learning of highly complex combinations of compositional and continuous conformational heterogeneity. Given the extent of structural heterogeneity observed with cryoDRGN in single particle datasets using purified samples (Sekne et al., 2022; Vasyliuk et al., 2022), we expect tomoDRGN to uncover similar structural variation within a rapidly expanding set of samples imaged in situ with cryo-ET. In addition to characterizing the structural heterogeneity of isolated particles, we expect that tomoDRGN’s ability to reanalyze particle stacks at different spatial scales (i.e., different real space box sizes) will prove widely useful in correlating intramolecular structural changes with structural variability in areas adjacent to the particle (Fig. 6). For example, when analyzing isolated ribosomal particles in EMPIAR-10499, we identified a set of particles decoding EF-Tu•tRNA and, when analyzing particles in a larger spatial context, we found a set of ribosomes associated with the cell membrane. A straightforward comparison enabled calculating the co-occurrence of these properties, revealing that approximately 13% of the membrane-associated ribosomes also appear to be decoding EF-Tu•tRNA. Of particular note, tomoDRGN allows us to generate a unique 3-D volume corresponding to each particle’s latent embedding. We anticipate this approach will allow researchers to populate low SNR tomograms with particle-specific density maps at approximately nanometer resolution and to explore the resultant spatial distributions of heterogeneous structures. Finally, the analyses enabled by tomoDRGN are inherently iterable. Our initial tomoDRGN analysis of EMPIAR-10499 revealed a population of non-ribosomal particles that we had failed to filter with traditional classification-based approaches. Excluding such particles and retraining at multiple spatial scales resolved intra- and inter-molecular structural heterogeneity, and retraining exclusively on a subset of membrane-associated ribosomes identified extracellular density that likely corresponded to the SecDF subcomplex. Given that tomoDRGN has the potential to identify many such distinct classes, we encourage users to embrace this iterative, branching approach. Some recently introduced software packages (Rice et al., 2022; Tegunov et al., 2021) explicitly support such “molecular sociology” where co-refinement of multiple distinct structures derived from a common data source enables global improvement of map quality. We anticipate tomoDRGN will form a virtuous cycle when interfacing with such software. MATERIALS AND METHODS TomoDRGN design and software implementation General architecture TomoDRGN is forked from cryoDRGN, and although we summarize the core aspects of the method here, readers are pointed to related cryoDRGN publications for further details (Bepler et al., 2019; Kinman et al., 2023; Zhong et al., 2021; Zhong et al., 2019). In brief, tomoDRGN is a variational autoencoder (VAE) with encoder and decoder networks comprised of multi-layer perceptrons (MLPs). TomoDRGN’s encoder learns a function (E) to map a set of j distinct tilt images (size × pixels) of particle i to a low dimensional latent encoding zi of dimension z; that is, E:ℝj*D*D→ℝZ The encoder MLP is comprised of two sub-networks that process j tilt images for each particle as follows. First, the 2-D Hartley transform of each tilt image is passed separately through encoder A to produce a set of j intermediate encodings. These j intermediate encodings are then pooled and passed together through encoder B to output the particle’s final latent embedding zi. The pooling step concatenates intermediate encodings along the tilt image axis by default, but also supports operations such as max and mean, which are inherently permutation-invariant. All experiments presented here concatenate the intermediate encodings. TomoDRGN’s decoder follows from that of cryoDRGN (Zhong et al., 2021), and uses a Gaussian featurization scheme for positional encoding in Fourier space (Tancik et al., 2020) as follows. Spatial coordinates are normalized to span [−0.5, 0.5] in each dimension, and a (fixed) positional encoder transforms each spatial coordinate to a basis set of D sinusoids with frequencies sampled from a scaled standard normal feat_sigma × 𝒩 (0,1) for each spatial coordinate axis, where D is the box size of an input image, and feat_sigma is set to 0.5. These positionally encoded coordinates, concatenated with the z-D latent coordinate, are then passed to the decoder; that is, in totality, D:ℝ3+z→ℝ Unless otherwise specified, models were trained for 50 epochs with batch size 1 (particle), AdamW optimizer with learning rate 0.0002, and weight decay 0. Training system Input images are modeled as linear 2-D projections of 3-D volumes, convolved by the contrast transfer function (CTF), with externally-derived rotation, translation, and CTF parameters. Heterogeneity among volumes is modeled via a continuous latent space sampled by a latent variable z per particle. The latent encoding for a given image X is taken as the maximum a posteriori of a Gaussian distribution parameterized by outputs from the encoder network, μZ|X and ΣZ|X, whereas the prior on the latent distribution is a standard normal distribution (0, I). Thus, the variational encoder qξ(z|X) produces a variational approximation of the true posterior p(z|X). The coordinate-based decoder models structures in reciprocal space: given a spatial frequency k∈ℝ3 and a latent variable z, the decoder predicts the corresponding voxel intensity as pθ(V|k, z). Applying the Fourier Slice Theorem (Bracewell, 1956), 3-D Fourier coordinates corresponding to 2-D projection image Xi are derived by rotating a 2-D lattice by the orientation of the volume Vi during imaging. Given a fixed latent coordinate sampled from qξ(zi|Xi) and the posed coordinate lattice, the reciprocal space image is reconstructed pixel-by-pixel via the decoder pθ(V|k, zi). The reconstructed image is then translated in-plane and multiplied by the CTF. The negative log-likelihood of the image is then computed as the mean squared error between the input and reconstructed image. The optimization function is the sum of the image reconstruction error and the KL divergence (KLD) of the latent encoding: ℒ(X;ξ,Θ)=Eqξ(z∣X)(log p(X∣z))−βKL(qξ(z∣X)‖p(z)) In this equation, the regularizing KLD term is weighted by β, which is set to 1|z|*t*D2 where D is the box size, t is the number of tilts, and |z| is the dimensionality of the latent space. Lattice masking and reconstruction weighting Critical dose is calculated for each spatial frequency using an empirical exposure-dependent amplitude attenuation curve derived for cryo-EM data (Grant and Grigorieff, 2015). The optimal dose is approximated to 2.51284 × critical dose as in the original study (Grant and Grigorieff, 2015; Hayward and Glaeser, 1979). Spatial frequencies (coordinates) of a tilt image exceeding the corresponding optimal doses are excluded from decoder network evaluation and loss calculation by a lattice mask during network training. Following error calculation of the input image against the reconstructed and CTF-weighted voxels, the squared differences are weighted (1) per-frequency by the exposure dependent amplitude attenuation curve (a function of tilt image index and spatial frequency), and (2) globally by the cosine of the stage tilt angle (a function of tilt image index). This weighted reconstruction error is backpropagated accordingly. Random tilt sampling During dataset initialization, the number of tilt images per particle is parsed via the rlnGroupName star file column using the syntax in Warp/M of tomogramID_particleID. The minimal number of tilt images present for any particle is then stored as the number of images to be sampled from each particle during network training and evaluation (this value also sets the input dimensionality of encoder B when using concatenation pooling). By default, sampling is performed randomly without replacement per-particle, and the subset and ordering of sampled tilts is updated each time a particle is retrieved during training or evaluation. Simulated dataset generation Cryo-ET data simulation was performed using scripts in the cryoSRPNT (cryo-EM Simulation of Realistic Particles via Noise Terms) GitHub repository. Density maps of four assembly states of the bacterial 50S ribosome (classes B - E) were obtained from EMD-8440, EMD-8441, EMD-8445, and EMD-8450, respectively (Davis et al., 2016). The project3d. py script was used to create noiseless projections of each volume as follows. First 5,000 random particle poses were sampled over SO(3). Each randomly posed particle was then rotated following a dose-symmetric tilt series scheme from 0° to ±60° with 3° steps in groups of 2 over 41 tilts, resulting in a total of 205,000 unique poses per volume. Each posed volume was projected along the z-axis to create noiseless images. The acn.py script was used to corrupt the noiseless projections using a standard cryo-EM image formation model (Baxter et al., 2009) augmented by tilt-series specific subroutines as follows. First, noiseless projections were Fourier-transformed, dose-weighted following an empirical exposure dependent amplitude attenuation curve at 3 e−/Å2/tilt to simulate SNR decrease due to radiation damage (Grant and Grigorieff, 2015), and inverse Fourier-transformed. Structural noise was added with an SNR of 1.4, and particles were then weighted by cosine(tilt) to simulate SNR decrease due to increased sample thickness. Projections were then convolved with the 2-D CTF with defocus values sampled from a mixture of Gaussian-distributed defoci with means between −1.5 μm to −3.5 μm in 0.5 μm steps and a standard deviation of 0.3 μm. Other CTF parameters included no astigmatism, 300 kV accelerating voltage, 2.7 mm spherical aberration, 0.1 amplitude contrast ratio, and 0° phase shift. Finally, shot noise was added with a SNR of 0.1, for a final SNR of 0.05. Particle stacks of each class were Fourier cropped to box sizes of 256px (bin1), 128px (bin2), and 64px (bin4). TomoDRGN network training on simulated data TomoDRGN homogeneous network training was performed on the 5,000 simulated class E particles. TomoDRGN heterogeneous network training was performed on all 20,000 simulated particles from classes B-E. Unless otherwise specified, figures illustrate results on the bin2 datasets, with network architectures summarized as nodes_per_layer × layers as follows: of 128×3 (encoder A), 128×3 (encoder B), and 256×3 (decoder). The dimensionality of the intermediate encoding was 32 and that of the final latent encoding was 128. Each model was trained utilizing dose and tilt loss weighting, dose frequency masking, and random tilt sampling, unless specified otherwise. Classification was performed directly on the latent embeddings with k=4 k-means clustering as implemented in scikit-learn. The dataset’s latent value nearest each k-means cluster center was used to generate a 3-D volume representative of that cluster. Sub-tomogram averaging of EMPIAR-10499 ribosomes Raw tilt movie data was downloaded from EMPIAR-10499. Movies were aligned and initial CTF estimation was performed in Warp (Tegunov and Cramer, 2019) as previously (Tegunov et al., 2021). Automated fiducial-based tilt series alignment was performed using dautoalign4warp (Burt et al., 2021) within the Dynamo package running in a Matlab environment (Castano-Diez et al., 2012). Alignment parameters were then used to generate tomograms at 10 Å/px in Warp. Template matching was performed in Warp using a 40 Å lowpass filtered ribosome volume generated from manually picked particles, keeping particles with a minimum separation of 80 Å (974,804 particles). The top 3% of particles by figure-of-merit across all tomograms were kept (29,245 particles). Sub-tomograms were extracted in Warp at 10 Å/px. Ab initio model generation and 3-D refinement were performed in RELION 3.1 (Bharat and Scheres, 2016) resulting in a density map with Nyquist-limited resolution. Sub-tomograms were re-extracted in Warp at 4 Å/px for further RELION 3-D refinement and 3-D classification with k=4 classes to remove false positive particle picks. The remaining 22,291 ribosomal particles were refined to a nominal resolution of 8.1 Å. Between each round of refinement and classification, particles were deduplicated in RELION with a cutoff distance of 80Å (removing a total of 360 particles throughout processing). The final 22,291 particles were imported to M and to produce a 3.5 Å resolution map as reported previously (Tegunov et al., 2021). Particles were then exported as image series sub-tomograms from M at several pixel and box sizes for tomoDRGN training, including three “single ribosome diameter” scales: 96 px at 3.71 Å/px, 210 px at 1.71 Å/px, 352 px at 1.71 Å; and one “multiple ribosome diameter” scale: 200 px at 3.71 Å/px. Particles were also exported as volume series sub-tomograms using M at 64 px 6 Å/px and 192 px 4 Å/px for validation of tomoDRGN heterogeneity analysis with traditional STA tools (see below) and for generation of requisite metadata for mapping particles to tomogram-contextualized locations in the tomoDRGN analysis Jupyter notebook. TomoDRGN network training on EMPIAR-10499 TomoDRGN homogeneous network training was performed on the 22,291 image series particles extracted at each of the “single ribosome diameter” image series sub-tomograms described above, or on select subsets at 96 px at 3.71 Å/px for interrogating heterogeneity in specific particle subsets. Unless specified otherwise, the network architecture was 512×3 (decoder). Each model was trained utilizing dose and tilt loss weighting, dose frequency masking, and random tilt sampling. TomoDRGN heterogeneous network training was performed on the same stack of 22,291 image series particles at box 96 px and 3.71 Å/px. Unless specified otherwise, the network architecture was 256×3 (encoder A), 256×3 (encoder B), and 256×3 (decoder) with the dimensionality of the intermediate encoding set to 128, and that of the final latent encoding set to 128. Each model was trained utilizing dose and tilt loss weighting, dose frequency masking, and random tilt sampling. Classification was performed directly on the latent embeddings with either k=20 (used for general visualization) or k=100 (used for detailed visualization and particle filtering) k-means clustering as above. The dataset’s latent value nearest each k-means cluster center was used to generate a 3-D volume representative of that cluster. Following exclusion of 1,310 non-ribosomal particles by separation of such volumes from k-100 classification, the remaining 20,981 particles were used to train new tomoDRGN models at box sizes of 96 and 200 px with 3.71 Å/px sampling. Membrane associated ribosomes (482) identified by k-100 classification of the 200 px trained dataset were further isolated to train a new tomoDRGN model with the parameters noted as above. Visualization and validation Python scripts A number of Python scripts were generated to quantify various properties of tomoDRGN outputs. Classification accuracy of tomoDRGN latent encodings learned for simulated datasets was evaluated by generating a confusion matrix (Fig. 2e). Classification reproducibility was evaluated for 100 randomly initialized classifications by calculating the Adjusted Rand Index (ARI) (Hubert and Arabie, 1985) (Extended Data Fig. 7f). The ARI measures a label-permutation-invariant similarity between two sets of clusterings and scales from 0 (random labeling) to 1 (identical labeling). Here, we used ARI to measure the similarity between the tomoDRGN or cryoDRGN latent clusters and the ground truth class labels. Volume consistency was assessed by calculating the map-map correlation coefficient (Real space) and map-map FSC scripts available within the tomoDRGN software. Before calculating map-map FSC curves, a soft mask was calculated and applied in Real space. Masks were defined by binarizing the map at ½ of the 99th voxel intensity percentile, dilating the mask by 3 px, and softening the mask using a falling cosine edge applied over 10 px. For computational efficiency, in some instances, the map-map correlation coefficient metric (CC) (Afonine et al., 2018) was used to quantify map-to-map similarity. Heterogeneity of a set of EMPIAR-10499 pre-filtered ribosome volumes generated by tomoDRGN was quantified by generating all volumes from the final epoch of training’s latent values and either (1) calculating the map-map CC to the STA 70S map for each tomoDRGN volume (Fig. 4b), or (2) performing principal component analysis on the array of all volume’s voxels (shape nvolumes × D3) followed by UMAP dimensionality reduction of the first 128 principal components (Fig. 4c). Finally, Python scripts were used to identify each particle’s nearest neighbor in each tomogram, calculate the distance to the nearest neighbor, and calculate the angle to the nearest neighbor after rotating to the STA consensus reference frame (Fig. 5c). Volume subset validation Subsets of the EMPIAR-10499 ribosomes were identified by tomoDRGN as non-ribosomal (n=1,310), 50S (n=852), 70S (n=20,129), or membrane-associated (n=482). Non-ribosomal particles were reprocessed in RELION 3.1 using ab initio volume generation with k=5 volume classes and all other parameters at their defaults. The 50S, 70S, and membrane-associated ribosome populations were reprocessed in RELION 3.1 using 3-D refinement against a corresponding real-space cropped 70S volume lowpass filtered to 60 Å. The same three particle subsets were also used to train tomoDRGN homogeneous networks as an additional validation, with identical training parameters to the full particle stack training detailed above. Visualization of tomoDRGN volumes in situ The subtomo2chimerax script (https://zenodo.org/record/6820119) was adapted to handle tomoDRGN’s unique sub-tomogram volumes per particle and is implemented in tomoDRGN. This script places each particle’s volume at its source location and orientation in the tomogram context using ChimeraX for visualization (Goddard et al., 2018; Pettersen et al., 2021). All volumes corresponding to EMPIAR-10499 tomogram 00256 were generated by tomoDRGN at box size 64 px and 5.55 Å/px using latent coordinates from tomoDRGN models in Fig. 4d and Fig. 5a, and placed in tomogram 00256 with coordinate and angle values extracted from the STA refinement in M. Atomic model-guided analyses To aid interpretation of tomoDRGN density maps, atomic models of the 70S ribosome (7PHA, 7PHB, and 4V89 which highlighted the L7/L12 dimers) were docked into density maps as rigid bodies using ChimeraX. The rRNA of 7PHB was segmented into distinct chains corresponding to rRNA helices (Petrov et al., 2014) following the MAVEn protocol (Kinman et al., 2023) for model-based analysis of volume ensembles (https://github.com/lkinman/MAVEn). The predicted atomic model for M.pneumoniae SecDF was downloaded from AlphaFold (ID: A0A0H3DPH3) and docked into the membrane-associated ribosome STA map in ChimeraX as a rigid body. Other components of the canonical Sec holotranslocon and oligosaccharyltransferases were either absent in the M. pneumoniae genome or lacked the observed extracellular domain. CryoDRGN network training CryoDRGN v0.3.4 was used to train models for both the simulated ribosome dataset (n=20,000) and the unfiltered EMPIAR-10499 dataset (n=22,291), using corresponding simulated or STA-derived poses and CTF parameters. Because cryoDRGN treats each input image independently, each dataset was reshaped to collapse the tilt axis dimension, resulting in particle stacks of size n=820,000 and n=913,931, respectively. Networks were trained with architecture 128×3 or 128×6 (encoder), latent dimensionality 8 or 128, and 256×3 (decoder), as annotated. All models were trained with hyperparameters intended to maximize similarity to the respective tomoDRGN analysis: batch size 40, gaussian positional featurization, 50 epochs of training, automatic mixed precision enabled, and all other parameters adopting default values. Latent space classification and volume sampling were performed as described for tomoDRGN above. Performance benchmarking All tomoDRGN and cryoDRGN models were trained on a cluster with nodes each equipped with 2x Intel Xeon Gold 6242R CPU (3.10 GHz, 512 GB RAM) and 2x Nvidia GeForce RTX 3090. Reported training times may in some cases be overestimates as up to two jobs were allowed to train or evaluate simultaneously on the same node. Extended Data Extended Data Figure 1: Training on a weighted subset of pixels improves reconstruction quality and compute performance. (a) Graphical overview of the dose filtering scheme (applied upstream of the decoder) and dose and tilt weighting scheme (applied during reconstruction error calculation) for a single representative tilt image. Filtering: the fixed optimal exposure curve is used to determine which spatial frequencies will be considered as a function of dose; the decoder processes only Fourier lattice coordinates within this mask (green lattice circle). Weighting: the squared error of the reconstructed Fourier slice is weighted per-frequency by the exposure-dependent amplitude attenuation curve and per-slice by the cosine of the corresponding stage tilt angle, before mean reduction and backpropagation (red arrows). (b) Relative weight of each tilt image assigned to a particle’s reconstruction error during model training as a function of spatial frequencies (x-axis), and tilt and dose, which are colored yellow to blue from low-to-high dose and tilt angle, assuming a dose symmetric tilt scheme (Hagen et al., 2017). (c) Map-map FSC of simulated class E large ribosomal subunit volumes (Davis et al., 2016) compared to tomoDRGN homogeneous network reconstructions in the presence or absence of the weighting or masking schemes at varying box and pixel sizes. (d) Spatial frequencies corresponding to FSC=0.5 map-map correlation with the ground truth volume plotted against wall time for model training. (e) Final tomoDRGN reconstructed volumes (left and center) and ground truth volumes (right) in the presence or absence of the weighting or masking schemes at box and pixel sizes assessed in panels (c) and (d). Extended Data Figure 2: Random per epoch tilt selection enables flexible and robust model training for datasets with non-uniform numbers of tilt-images per particle. (a) Graphical summary of a dataset with non-uniform numbers of tilt images per particle. Here, the minimum number of tilt images for any particle is 3. (b) Corresponding tomoDRGN network architecture for random sampling and ordering of 3 tilt images per particle. (c) Mean per-class volumetric correlation coefficient for identical tomoDRGN models trained on 41 sequentially sampled tilts (top) or 41 randomly sampled tilts (bottom). At 5 epoch intervals, 25 random volumes were generated from each class for correlation coefficient calculation to ground truth ribosome assembly intermediate volumes (classes B-E). Error bars denote standard error of the mean CC. (d) Nine tomoDRGN models with identical architectures were trained with the indicated number of tilts sampled per particle (total available tilts = 41). PCA (left) and UMAP (right) dimensionality reduction of each final epoch’s latent embeddings. Once trained, up to 10 randomly sampled and permuted tilt images for one representative particle from each volume class were embedded using the corresponding pretrained tomoDRGN model and are superimposed as colored points. Note increased dispersion of colored points as number of tilts sampled during training decreased. (e) For each ribosomal large subunit class (B-E), 25 particles were randomly selected and up to 10 subsets of their tilt images were randomly sampled and permuted as in (d). In the heatmap, row indices refer to models trained in (d) using different numbers of sampled tilts (1–41), and columns denote epochs of training with that model. For each particle, each tilt subset was evaluated with the corresponding tomoDRGN model and the ratio of standard deviations of each particle’s 10 latent embeddings to all particles’ latent embeddings was calculated. The mean ratio across all particles, which measures the dispersion of encoder embeddings, is plotted per ribosomal LSU class. Here, lower dispersion indicates better performance. (f) Particles and tilt subsets were selected as in (e). At each indicated epoch of training, the corresponding tomoDRGN model was used to generate volumes for each particle’s tilt subsets. For each such volume, the correlation coefficient was calculated between that volume and the corresponding ground truth volume. The mean across all particles at each epoch for each model is shown as a heatmap per ribosomal LSU class. Here, higher CC indicates improved performance. Extended Data Figure 3: TomoDRGN training statistics for homogeneous simulated datasets as a function of decoder architecture. (a) Map-map FSC of final volumes generated from tomoDRGN homogeneous network training on simulated class E ribosomes with indicated decoder architectures against the corresponding ground truth volume. Panels correspond to different box and pixel sizes; colors correspond to different tomoDRGN model architectures. (b) Volumes were generated at each epoch during training and spatial frequencies at which map-map FSC=0.5 are plotted against cumulative wall time for models of different architectures (colors) on images of different box and pixel sizes (panels). Circles note total wall time elapsed and resolution achieved after 50 epochs of training. Extended Data Figure 4: TomoDRGN model quality with heterogeneous simulated datasets as a function of encoder architectures. (a, b) Mean per-class volumetric correlation coefficient for tomoDRGN models trained with indicated encoder A architectures (panel A titles) or encoder B architectures (panel B titles). At 5 epoch intervals, 10 volumes from each volume class were generated and used to calculate volumetric correlation coefficients to the corresponding ground truth ribosome assembly intermediate volume. Error bars denote standard error of the mean in correlation coefficient among the tomoDRGN volumes at that epoch and the corresponding ground truth volume. Extended Data Figure 5: Using tomoDRGN to identify non-ribosomal particles picked from EMPIAR-10499 tomograms. (a) UMAP and corresponding sampled volumes from tomoDRGN heterogeneous network training from Fig. 4a. Eight representative non-ribosomal particles identified through manual inspection of k=100 k-means clustering of latent space are rendered at a constant isosurface and pose. (b) Two tomograms are shown in slice view using Cube (https://github.com/dtegunov/cube) with locations of particles labeled as non-ribosomal annotated within each tomogram. (c) RELION3-based multiclass (k=5) ab initio sub-tomogram volume generation using particles annotated as non-ribosomal (n=1,310). Extended Data Figure 6: Validation of tomoDRGN-generated volumes. Comparison of volumes generated by a full tomoDRGN network (row 1), an isolated decoder neural network (row 2), or traditional subtomogram averaging (row 3). A full tomoDRGN network was trained on the heterogeneous ribosomal particle stack (row 1, n=20,981, see Figs. 4d and 5a) and representative volumes are depicted. Separate tomoDRGN homogeneous decoder networks were trained on one of three homogeneous substacks corresponding to (a) 70S particles (n=20,129); (b) 50S particles (n=852); or (c) SecDF-positive ribosomes (n=380). Traditional STA was also performed on each of these three particles stacks. Extended Data Figure 7: CryoDRGN fails to consistently encode structural heterogeneity using a simulated tilt series dataset (a) Schematic of two cryoDRGN network architectures that were tested, and the tomoDRGN architecture used in Fig. 2c–e. Each model was trained using the same simulated dataset of ribosome large subunit assembly classes B-E (Davis et al., 2016) consisting of 41 tilt images for each of 5,000 particles for each of the four assembly states and thus the dataset was treated by cryoDRGN as n=820,000 images (see Methods). (b) UMAP of final epoch latent embeddings of each particle image, represented as a kernel density estimate (KDE) is plotted, with KDEs independently estimated and plotted for each of the four ground truth assembly states (bottom). (c) UMAP of final epoch latent embedding with k=4 k-means latent classification of the resulting latent space. KDEs were independently estimated and plotted for each of the four k-means classes. The predicted labels are annotated by both the k-means class index (0–3) and corresponding ground truth class label (B-E) of the central particle within each k-means class. (d) Confusion matrix of ground truth class labels versus k=4 k-means latent classification. (e) Volumes sampled at the k=4 k-means cluster centers illustrated in (c). Volumes are annotated by the k-means class index and ground truth class label and colored by the ground truth class label. (f) Violin plot of consistency of k=4 k-means clustering of each model by Adjusted Rand Index (Hubert and Arabie, 1985) (n = 100 randomly seeded initializations, higher values correspond to greater fidelity to ground truth classification). Extended Data Figure 8: CryoDRGN learns non-physical structural heterogeneity in an exemplar tomographic dataset. Two cryoDRGN models (a, b) were trained on the unfiltered particle stack of Mycoplasma pneumoniae ribosomes from Fig. 4a (n = 22,291 particles, treated as n = 913,931 images). The latent space is shown as a KDE plot following UMAP dimensionality reduction, with k=20 k-means class center particles annotated (left) and corresponding volumes visualized (right). Note that many putative 70S particles lack density in the particle core. A reference 70S volume sampled from tomoDRGN’s model in Fig. 4a is shown in the same pose for comparison. Extended Data Figure 9: CryoDRGN’s learned embeddings exhibit undesirable correlations with tilt image index. (a) Two cryoDRGN models were tested on the unfiltered particle stack of Mycoplasma pneumoniae ribosomes from Fig. 4a. The latent space is shown as a KDE plot following UMAP dimensionality reduction. The latent embeddings were binned by the tilt image index, and the median value across each bin is annotated. (b) KDEs from panel A replotted after binning by tilt image index quartiles. (c) KDEs from panel A with annotated positions corresponding to three representative particles evaluated using their 5th, 15th, 25th, or 35th tilt images. (d) Volumes generated from cryoDRGN using the latent embeddings highlighted in panel C. Supplementary Material Supplement 1 Supplemental Movie 1: Structural heterogeneity in the large ribosomal subunit. Volumes were sampled from the tomoDRGN model in Fig. 4d using k=100 k-means clustering of latent space. Density for the 30S was removed using the Volume Zone feature in ChimeraX, guided by atomic model 7PHB, to reveal distinct conformation and compositional states of the large subunit. Note conformational and compositional heterogeneity in tRNA and elongation factor binding sites, which are found along the midline of the particle. Supplement 2 Supplemental Movie 2: Membrane-associated ribosomes exhibit flexible attachment. Volumes were generated for all particles used to train the model in Fig. 5d. The tertile of volumes with highest SecDF occupancy are displayed, ordered by increasing occupancy (n = 162). Note significant dynamics in the orientation of the membrane relative to the associated ribosome. ACKNOWLEDGEMENTS We thank Laurel Kinman and Ellen Zhong for helpful discussion and feedback, and the MIT-IBM Satori team and the MIT SuperCloud Supercomputing Center for HPC computing resources and support. This work was supported by NIH grants R01-GM144542, 5T32-GM007287, and NSF-CAREER grant 2046778, the Sloan Foundation and the Whitehead Family. DATA AVAILABILITY Extracted particle sub-tomograms from EMPIAR-10499 will be deposited to EMPIAR. The membrane-associated ribosome from EMPIAR-10499 generated by RELION will be deposited to EMDB. The trained tomoDRGN and cryoDRGN models used to analyze EMPIAR-10499 will be deposited at zenodo.org. Simulated data and corresponding trained tomoDRGN and cryoDRGN models are available upon request. Figure 1: A neural network architecture to analyze structurally heterogeneous particles imaged by cryo-ET. (a) A typical sample and data processing workflow to produce tomoDRGN inputs. The sample (e.g., a bacterial cell) is applied to a grid, plunge frozen, and optionally thinned. A series of TEM images of a target region are collected at different stage tilt angles. A tomographic volume is reconstructed using weighted back-projection of all tilt images. Instances of the target particle are identified (blue boxes) and extracted as 3-D voxel arrays. Iterative sub-tomogram averaging (STA) is used to reconstruct a consensus density map. Per-particle 2-D tilt images are then re-extracted from the source tilt series images and parameters (e.g. pose, defocus, etc.) estimated from STA are associated with the images. (b) The tomoDRGN network architecture and training design. Each particle’s set of tilt images are independently passed through encoder A (Enc A), then jointly passed through encoder B (Enc B), thereby mapping all tilt images of a particle to one embedding (z) in a low dimensionality latent space. The decoder network (Dec) uses the latent embedding and a featurized voxel coordinate to decode a corresponding set of images pixel-by-pixel. Note that the decoder can learn a homogeneous structure by excluding the encoder module. The network is trained using a loss function (grey arrows) that depends on the input images, reconstructed images, and z (red arrows). (c) Graphical signposts for volumes generated or analyzed by distinct reconstruction tools. These signposts are used throughout this manuscript when volumes are displayed to clarify how they were generated. Figure 2: TomoDRGN recovers known heterogeneity in simulated datasets. (a) Illustration of the method used to simulate tilt series particle stacks corresponding to four assembly states (B-E) (Davis et al., 2016) of the bacterial large ribosomal subunit. (b) Left, a tomoDRGN homogeneous network reconstruction of the simulated class E dataset after 50 epochs of training at a resolution of 3.55 Å/px. Right, FSC between the tomoDRGN reconstruction and the ground truth volume at each of 50 epochs of training (purple to yellow). (c) First two principal components (left) and UMAP embeddings (right) of tomoDRGN latent space when trained on the simulated four class dataset, colored by k=4 k-means classification of latent space. (d) Ground truth ribosomal volumes (top) and corresponding tomoDRGN-reconstructed volumes (bottom) sampled from the median latent encoding of each of the k=4 k-means classes in (c). (e) Confusion matrix of k-means clustering class labels from (c) against ground truth class labels. Figure 3: TomoDRGN resolves high resolution features from sub-tomograms collected in situ. (a) M. pneumoniae in situ ribosomal volume obtained from traditional STA processing (n=22,291 particles) (top) and tomoDRGN homogeneous reconstruction of the same particles (bottom). (b) Density maps from the tomoDRGN homogeneous reconstruction around indicated ribosome components. (c) Map-to-map Fourier Shell Correlations (FSC) of three tomoDRGN reconstructions of the particle stack in (a) extracted at indicated box and pixel sizes against corresponding STA volumes. Circles denote the Nyquist limit for each particle stack. Figure 4: TomoDRGN uncovers structural heterogeneity in ribosomes imaged in situ (a) UMAP of tomoDRGN latent embeddings (n=22,291 particles) shown as gray kernel density estimate (KDE), overlaid with scatter plot depicting latent embedding locations of large-ribosomal-subunit-only (yellow) or non-ribosomal particles (blue) identified via k=100 k-means classification of latent space and manual inspection of the 100 related volumes. Representative volumes generated from latent embeddings annotated as 70S, 50S, or non-ribosomal (NR) also depicted. (b) Volumes (box=96 px) were generated from every particle’s latent embedding, and volumetric cross-correlation (CC) between the 70S STA map and these volumes was calculated. Histograms of CC are shown for volumes assigned as 70S (top), 50S (middle) and non-ribosomal (bottom) particles as in (a). (c) Volumes from panel (b) were subjected to principal component analysis. UMAP dimensionality reduction of the first 128 principal components is plotted as a KDE with scatterplot corresponding to assignments of 70S, 50S, or non-ribosomal from (a) superimposed. (d) UMAP of tomoDRGN latent embeddings (n=20,981; non-ribosomal particles excluded). Colored volumes sampled from correspondingly colored points on UMAP plot are shown with red asterisks and insets highlighting regions of notable structural variability. A transparent grey volume corresponding to a tomoDRGN reconstruction of a 70S•EF-Tu volume is provided for visual reference. (e) MAVEn analysis (Sun et al., 2022) of 500 volumes sampled from the tomoDRGN model in panel (d) plotted as a clustered heatmap with columns corresponding to proteins and rRNA structural elements (Ward-linkage, Euclidean-distance), and rows corresponding to the 500 sampled volumes (Ward-linkage, Correlation-distance). Distinct volume classes corresponding to 50S and 70S particles as identified by a row-wise threshold on this clustermap are also shown. Figure 5: TomoDRGN captures intermolecular heterogeneity in situ. (a) UMAP of tomoDRGN latent embeddings of particles (n=20,981) re-extracted with box size ~3x particle radius. Colored volumes sampled from correspondingly colored points in UMAP are shown. (b) Violin plot of the distance from each particle in the indicated classes from panel (a) to its nearest neighbor ribosome. Distribution colors are paired with those in (a). The right bound of the x-axis corresponds to the box diameter, and the red interval on the x-axis corresponds to typical inter-ribosome distances in a prokaryotic polysome. Mollweide projection histograms for each class highlighted in panel (a), depicts directions to each ribosome’s nearest neighbor ribosome, following rotation to the consensus pose. (c) Distribution of k=20 k-means classification of latent embeddings per tomogram. Column width is proportional to each tomogram’s fraction of the total particle count. Within a column, the height of each color is proportional to the population of that k-means class within that tomogram. Classes are colored as in (a). (d) Screenshot from tomoDRGN’s interactive tomogram viewer showing all ribosomes for a single tomogram (blue cones) with ribosomes corresponding to membrane-associated classes further annotated as red spheres. (e) UMAP of tomoDRGN latent embeddings (n=482) of membrane-associated ribosomes. Colored volumes are sampled from correspondingly colored points in latent space. Relative occupancy of globular periplasmic density (n=482) is plotted as a histogram with a red line noting manually assigned threshold defining particles bearing the periplasmic density (n=380). (f) STA reconstruction of membrane-associated ribosomes bearing periplasmic density identified by tomoDRGN with docked atomic model of Mycoplasma pneumoniae SecDF predicted using Alphafold (AF: A0A0H3DPH3). Figure 6: TomoDRGN visualizes structurally heterogeneous macromolecular complexes with spatial context. (a) Ribosomes from one EMPIAR-10499 tomogram rendered in ChimeraX. Volumes were generated for each ribosome using tomoDRGN and colored according to the intramolecular latent classification shown in Fig. 4d, and positioned correspondingly within the reconstructed cell. Transparent ribosomes correspond to k=20 k-means classes not highlighted in Fig. 4d. (b) Ribosomes from the same tomogram depicted in (a) rendered in ChimeraX. Volumes were generated for each ribosome and colored according to the intermolecular latent classification shown in Fig. 5a, and positioned correspondingly within the reconstructed cell. Transparent ribosomes correspond to k=20 k-means classes not highlighted in Fig. 5a. Table 1: Impact of weighting and masking on performance of tomoDRGN homogeneous reconstruction. Box size (px) Pixel size (Å) Weighting Masking Architecture # Trainable parameters # Data points per particle Training time per 1k particles (min) VRAM per particle (GB) Max resolution (1/px) Epochs to max resolution Wall clock to max resolution (min) 64 6.55 − − 256 × 3 247,298 131,528 0.26 1.95 0.48 1 1.30 64 6.55 + − 256 × 3 247,298 131,528 0.26 1.95 0.48 1 1.30 64 6.55 − + 256 × 3 247,298 96,776 0.22 1.76 0.48 1 1.11 64 6.55 + + 256 × 3 247,298 96,776 0.22 1.76 0.48 1 1.10 128 3.28 − − 256 × 3 296,450 526,932 0.88 4.59 0.49 16 70.58 128 3.28 + − 256 × 3 296,450 526,932 0.88 4.59 0.49 3 13.17 128 3.28 − + 256 × 3 296,450 194,064 0.43 2.47 0.49 3 6.38 128 3.28 + + 256 × 3 296,450 194,064 0.43 2.47 0.49 3 6.39 256 1.63 − − 256 × 3 394,754 2,108,712 4.42 18.88 0.25 30 662.49 256 1.63 + − 256 × 3 394,754 2,108,712 4.47 18.88 0.33 36 804.64 256 1.63 − + 256 × 3 394,754 378,516 1.21 4.42 0.33 42 253.25 256 1.63 + + 256 × 3 394,754 378,516 1.20 4.42 0.32 39 234.58 Summary statistics for tomoDRGN homogeneous network training using the simulated ribosome class E particles at different box and pixel sizes, in the presence or absence of reconstruction weighting and masking. Table 2: Impact of network architecture on tomoDRGN homogeneous network reconstruction. Box size (px) Architecture # Trainable parameters # Data points per particle Training time per 1k particles (min) VRAM per particle (GB) Max resolution (1/px) Epochs to max resolution Wall clock to max resolution (min) 64 64 × 3 24,962 96,776 0.15 1.42 0.48 1 0.75 64 128 × 3 74,498 96,776 0.17 1.54 0.48 1 0.85 64 256 × 3 247,298 96,776 0.22 1.76 0.48 1 1.11 64 512 × 3 887,810 96,776 0.37 2.15 0.48 1 1.84 64 768 × 3 1,921,538 96,776 0.58 2.79 0.48 1 2.88 128 64 × 3 37,250 194,064 0.28 1.85 0.38 43 60.74 128 128 × 3 99,074 194,064 0.33 2.04 0.49 15 24.94 128 256 × 3 296,450 194,064 0.43 2.47 0.49 4 8.69 128 512 × 3 986,114 194,064 0.75 3.33 0.49 2 7.48 128 768 × 3 2,068,994 194,064 1.22 4.53 0.49 1 6.12 256 64 × 3 61,826 378,516 0.88 3.48 0.20 42 185.81 256 128 × 3 148,226 378,516 0.94 3.67 0.26 47 221.67 256 256 × 3 394,754 378,516 1.15 4.42 0.32 33 190.30 256 512 × 3 1,182,722 378,516 1.81 6.10 0.34 22 199.42 256 768 × 3 2,363,906 378,516 3.61 11.92 0.35 20 360.84 Summary statistics for tomoDRGN homogeneous network training using the simulated ribosome class E particles at various box and pixel sizes, sweeping the number of nodes per layer in the decoder network. Table 3: Impact of encoder network A architecture on tomoDRGN heterogeneous network reconstruction. Box size (px) EncA architecture EncA-EncB intermediate dimensionality EncB architecture Dec architecture # Trainable parameters (encoder) # Data points per particle (encoder) # Trainable parameters (decoder) # Data points per particle (decoder) Training time per 1k particles (min) VRAM per particle (GB) 64 64 × 2 128 128 × 2 256 × 3 959,936 131,528 280,066 96,776 0.41 2.10 64 64 × 4 128 128 × 2 256 × 3 968,256 131,528 280,066 96,776 0.43 2.10 64 128 × 2 128 128 × 2 256 × 3 1,198,208 131,528 280,066 96,776 0.42 2.10 64 128 × 4 128 128 × 2 256 × 3 1,231,232 131,528 280,066 96,776 0.43 2.10 64 256 × 2 128 128 × 2 256 × 3 1,723,904 131,528 280,066 96,776 0.42 2.10 64 256 × 4 128 128 × 2 256 × 3 1,855,488 131,528 280,066 96,776 0.44 2.11 128 64 × 2 128 128 × 2 256 × 3 1,577,152 526,932 329,218 194,064 0.83 3.95 128 64 × 4 128 128 × 2 256 × 3 1,585,472 526,932 329,218 194,064 0.84 3.95 128 128 × 2 128 128 × 2 256 × 3 2,432,640 526,932 329,218 194,064 0.88 3.95 128 128 × 4 128 128 × 2 256 × 3 2,465,664 526,932 329,218 194,064 0.83 3.95 128 256 × 2 128 128 × 2 256 × 3 4,192,768 526,932 329,218 194,064 0.79 3.97 128 256 × 4 128 128 × 2 256 × 3 4,324,352 526,932 329,218 194,064 0.81 3.97 256 64 × 2 128 128 × 2 512 × 3 4,046,272 2,108,712 1,248,258 378,516 3.05 11.07 256 64 × 4 128 128 × 2 512 × 3 4,054,592 2,108,712 1,248,258 378,516 3.03 11.07 256 128 × 2 128 128 × 2 512 × 3 7,370,880 2,108,712 1,248,258 378,516 3.23 11.12 256 128 × 4 128 128 × 2 512 × 3 7,403,904 2,108,712 1,248,258 378,516 2.95 11.10 256 256 × 2 128 128 × 2 512 × 3 14,069,248 2,108,712 1,248,258 378,516 2.66 11.07 256 256 × 4 128 128 × 2 512 × 3 14,200,832 2,108,712 1,248,258 378,516 2.67 11.22 Summary statistics for tomoDRGN heterogeneous network training using the simulated ribosome 4-class particles at various box and pixel sizes, sweeping the encoder A architecture (number of nodes per layer and number of layers). Table 4: Impact of encoder network B architecture on tomoDRGN heterogeneous network reconstruction. Box size (px) EncA architecture EncA-EncB intermediate dimensionality EncB architecture Dec architecture # Trainable parameters (encoder) # Data points per particle (encoder) # Trainable parameters (decoder) # Data points per particle (decoder) Training time per 1k particles (min) VRAM per particle (GB) 64 128 × 3 32 64 × 3 256 × 3 577,568 131,528 280,066 96,776 0.41 2.10 64 128 × 3 32 128 × 3 256 × 3 715,040 131,528 280,066 96,776 0.41 2.11 64 128 × 3 128 64 × 3 256 × 3 841,856 131,528 280,066 96,776 0.41 2.10 64 128 × 3 128 128 × 3 256 × 3 1,231,232 131,528 280,066 96,776 0.41 2.10 128 128 × 3 32 64 × 3 256 × 3 1,812,000 526,932 329,218 194,064 0.79 3.95 128 128 × 3 32 128 × 3 256 × 3 1,949,472 526,932 329,218 194,064 0.85 3.96 128 128 × 3 128 64 × 3 256 × 3 2,076,288 526,932 329,218 194,064 0.87 3.95 128 128 × 3 128 128 × 3 256 × 3 2,465,664 526,932 329,218 194,064 0.86 3.95 256 128 × 3 32 64 × 3 256 × 3 6,750,240 2,108,712 427,522 378,516 2.23 9.43 256 128 × 3 32 128 × 3 256 × 3 6,887,712 2,108,712 427,522 378,516 2.22 9.43 256 128 × 3 128 64 × 3 256 × 3 7,014,528 2,108,712 427,522 378,516 2.18 9.43 256 128 × 3 128 128 × 3 256 × 3 7,403,904 2,108,712 427,522 378,516 2.16 9.43 Summary statistics for tomoDRGN heterogeneous network training using the simulated ribosome 4-class particles at various box and pixel sizes, sweeping the encoder A output layer size and the encoder B architecture (number of nodes per layer and number of layers). SOFTWARE AVAILABILITY TomoDRGN is distributed as free and open-source software under the GPL-3.0 license. Source code, installation instructions, and example usage are available at https://github.com/bpowell122/tomodrgn. Version 0.2.2 was used in this study. Scripts used to generate simulated data are available at https://github.com/bpowell122/cryoSRPNT. Version 0.1.0 was used in this study. ==== Refs REFERENCES Afonine P.V. , Klaholz B.P. , Moriarty N.W. , Poon B.K. , Sobolev O.V. , Terwilliger T.C. , Adams P.D. , and Urzhumtsev A. (2018). New tools for the analysis and validation of cryo-EM maps and atomic models. Acta Crystallogr D Struct Biol 74 , 814–840. 10.1107/S2059798318009324.30198894 Asano S. , Engel B.D. , and Baumeister W. (2016). In Situ Cryo-Electron Tomography: A Post-Reductionist Approach to Structural Biology. J Mol Biol 428 , 332–343. 10.1016/j.jmb.2015.09.030.26456135 Bai X.C. , McMullan G. , and Scheres S.H. (2015). How cryo-EM is revolutionizing structural biology. Trends Biochem Sci 40 , 49–57. 10.1016/j.tibs.2014.10.005.25544475 Baxter W.T. , Grassucci R.A. , Gao H. , and Frank J. (2009). Determination of signal-to-noise ratios and spectral SNRs in cryo-EM low-dose imaging of molecules. J Struct Biol 166 , 126–132. 10.1016/j.jsb.2009.02.012.19269332 Becht E. , McInnes L. , Healy J. , Dutertre C.A. , Kwok I.W.H. , Ng L.G. , Ginhoux F. , and Newell E.W. (2018). Dimensionality reduction for visualizing single-cell data using UMAP. Nat Biotechnol. 10.1038/nbt.4314. Bepler T. , Zhong E. , Kelley K. , Brignole E. , and Berger B. (2019). Explicitly disentangling image content from translation and rotation with spatial-VAE. Advances in Neural Information Processing Systems 32 . Bharat T.A. , and Scheres S.H. (2016). Resolving macromolecular structures from electron cryo-tomography data using subtomogram averaging in RELION. Nat Protoc 11 , 2054–2065. 10.1038/nprot.2016.124.27685097 Bharat T.A.M. , Russo C.J. , Lowe J. , Passmore L.A. , and Scheres S.H.W. (2015). Advances in Single-Particle Electron Cryomicroscopy Structure Determination applied to Subtomogram Averaging. Structure 23 , 1743–1753. 10.1016/j.str.2015.06.026.26256537 Bracewell R.N. (1956). Strip integration in radio astronomy. Australian Journal of Physics 9 , 198–217. Burt A. , Gaifas L. , Dendooven T. , and Gutsche I. (2021). A flexible framework for multi-particle refinement in cryo-electron tomography. PLoS Biol 19 , e3001319. 10.1371/journal.pbio.3001319.34437530 Castano-Diez D. , Kudryashev M. , Arheit M. , and Stahlberg H. (2012). Dynamo: a flexible, user-friendly development tool for subtomogram averaging of cryo-EM data in high-performance computing environments. J Struct Biol 178 , 139–151. 10.1016/j.jsb.2011.12.017.22245546 Castano-Diez D. , and Zanetti G. (2019). In situ structure determination by subtomogram averaging. Curr Opin Struct Biol 58 , 68–75. 10.1016/j.sbi.2019.05.011.31233977 Chen M. , and Ludtke S.J. (2021). Deep learning-based mixed-dimensional Gaussian mixture model for characterizing variability in cryo-EM. Nat Methods 18 , 930–936. 10.1038/s41592-021-01220-5.34326541 Chen S.S. , Sperling E. , Silverman J.M. , Davis J.H. , and Williamson J.R. (2012). Measuring the dynamics of E. coli ribosome biogenesis using pulse-labeling and quantitative mass spectrometry. Mol Biosyst 8 , 3325–3334. 10.1039/c2mb25310k.23090316 Cheng Y. , Grigorieff N. , Penczek P.A. , and Walz T. (2015). A primer to single-particle cryo-electron microscopy. Cell 161 , 438–449. 10.1016/j.cell.2015.03.050.25910204 Dashti A. , Mashayekhi G. , Shekhar M. , Ben Hail D. , Salah S. , Schwander P. , des Georges A. , Singharoy A. , Frank J. , and Ourmazd A. (2020). Retrieving functional pathways of biomolecules from single-particle snapshots. Nat Commun 11 , 4734. 10.1038/s41467-020-18403-x.32948759 Davis J.H. , Tan Y.Z. , Carragher B. , Potter C.S. , Lyumkis D. , and Williamson J.R. (2016). Modular Assembly of the Bacterial Large Ribosomal Subunit. Cell 167 , 1610–1622 e1615. 10.1016/j.cell.2016.11.020.27912064 Davis J.H. , and Williamson J.R. (2017). Structure and dynamics of bacterial ribosome biogenesis. Philosophical transactions of the Royal Society of London. Series B, Biological sciences 372 . 10.1098/rstb.2016.0181. Fromm S.A. , O’Connor K.M. , Purdy M. , Bhatt P.R. , Loughran G. , Atkins J.F. , Jomaa A. , and Mattei S. (2023). The translating bacterial ribosome at 1.55 A resolution generated by cryo-EM imaging services. Nat Commun 14 , 1095. 10.1038/s41467-023-36742-3.36841832 Gemmer M. , Chaillet M.L. , van Loenhout J. , Cuevas Arenas R. , Vismpas D. , Grollers-Mulderij M. , Koh F.A. , Albanese P. , Scheltema R.A. , Howes S.C. , (2023). Visualization of translation and protein biogenesis at the ER membrane. Nature 614 , 160–167. 10.1038/s41586-022-05638-5.36697828 Glaeser R.M. (1979). Prospects for extending the resolution limit of the electron microscope. J Microsc 117 , 77–91. 10.1111/j.1365-2818.1979.tb00232.x.490637 Goddard T.D. , Huang C.C. , Meng E.C. , Pettersen E.F. , Couch G.S. , Morris J.H. , and Ferrin T.E. (2018). UCSF ChimeraX: Meeting modern challenges in visualization and analysis. Protein Sci 27 , 14–25. 10.1002/pro.3235.28710774 Grant T. , and Grigorieff N. (2015). Measuring the optimal exposure for single particle cryo-EM using a 2.6 A reconstruction of rotavirus VP6. eLife 4 , e06980. 10.7554/eLife.06980.26023829 Hagen W.J.H. , Wan W. , and Briggs J.A.G. (2017). Implementation of a cryo-electron tomography tilt-scheme optimized for high resolution subtomogram averaging. J Struct Biol 197 , 191–198. 10.1016/j.jsb.2016.06.007.27313000 Harastani M. , Eltsov M. , Leforestier A. , and Jonic S. (2021). HEMNMA-3D: Cryo Electron Tomography Method Based on Normal Mode Analysis to Study Continuous Conformational Variability of Macromolecular Complexes. Front Mol Biosci 8 , 663121. 10.3389/fmolb.2021.663121.34095222 Harastani M. , Eltsov M. , Leforestier A. , and Jonic S. (2022). TomoFlow: Analysis of Continuous Conformational Variability of Macromolecules in Cryogenic Subtomograms based on 3D Dense Optical Flow. J Mol Biol 434 , 167381. 10.1016/j.jmb.2021.167381.34848215 Hayward S.B. , and Glaeser R.M. (1979). Radiation damage of purple membrane at low temperature. Ultramicroscopy 04 , 201–210. 10.1016/s0304-3991(79)90211-0.473421 Himes B.A. , and Zhang P. (2018). emClarity: software for high-resolution cryo-electron tomography and subtomogram averaging. Nat Methods 15 , 955–961. 10.1038/s41592-018-0167-z.30349041 Hoffmann P.C. , Kreysing J.P. , Khusainov I. , Tuijtel M.W. , Welsch S. , and Beck M. (2022). Structures of the eukaryotic ribosome and its translational states in situ. Nat Commun 13 , 7435. 10.1038/s41467-022-34997-w.36460643 Hubert L. , and Arabie P. (1985). Comparing partitions. Journal of Classification 2 , 193–218. 10.1007/BF01908075. Kingma D.P. , and Welling M. (2013). Auto-encoding variational bayes. arXiv preprint arXiv:1312.6114. Kinman L.F. , Powell B.M. , Zhong E.D. , Berger B. , and Davis J.H. (2023). Uncovering structural ensembles from single-particle cryo-EM data using cryoDRGN. Nat Protoc 18 , 319–339. 10.1038/s41596-022-00763-x.36376590 Lovatt M. , Leistner C. , and Frank R.A.W. (2022). Bridging length scales from molecules to the whole organism by cryoCLEM and cryoET. Faraday Discuss. 10.1039/d2fd00081d. Murata K. , and Wolf M. (2018). Cryo-electron microscopy for structural analysis of dynamic biological macromolecules. Biochim Biophys Acta Gen Subj 1862 , 324–334. 10.1016/j.bbagen.2017.07.020.28756276 Petrov A.S. , Bernier C.R. , Gulen B. , Waterbury C.C. , Hershkovits E. , Hsiao C. , Harvey S.C. , Hud N.V. , Fox G.E. , Wartell R.M. , and Williams L.D. (2014). Secondary structures of rRNAs from all three domains of life. PLoS One 9 , e88222. 10.1371/journal.pone.0088222.24505437 Pettersen E.F. , Goddard T.D. , Huang C.C. , Meng E.C. , Couch G.S. , Croll T.I. , Morris J.H. , and Ferrin T.E. (2021). UCSF ChimeraX: Structure visualization for researchers, educators, and developers. Protein Sci 30 , 70–82. 10.1002/pro.3943.32881101 Punjani A. , and Fleet D.J. (2021). 3D variability analysis: Resolving continuous flexibility and discrete heterogeneity from single particle cryo-EM. J Struct Biol 213 , 107702. 10.1016/j.jsb.2021.107702.33582281 Pyle E. , and Zanetti G. (2021). Current data processing strategies for cryo-electron tomography and subtomogram averaging. Biochem J 478 , 1827–1845. 10.1042/BCJ20200715.34003255 Rice G. , Wagner T. , Stabrin M. , and Raunser S. (2022). TomoTwin: Generalized 3D Localization of Macromolecules in Cryo-electron Tomograms with Structural Data Mining. bioRxiv, 2022.2006.2024.497279. 10.1101/2022.06.24.497279. Sekne Z. , Ghanim G.E. , van Roon A.M. , and Nguyen T.H.D. (2022). Structural basis of human telomerase recruitment by TPP1-POT1. Science 375 , 1173–1176. 10.1126/science.abn6840.35201900 Stojkovic V. , Myasnikov A.G. , Young I.D. , Frost A. , Fraser J.S. , and Fujimori D.G. (2020). Assessment of the nucleotide modifications in the high-resolution cryo-electron microscopy structure of the Escherichia coli 50S subunit. Nucleic Acids Res 48 , 2723–2732. 10.1093/nar/gkaa037.31989172 Stolken M. , Beck F. , Haller T. , Hegerl R. , Gutsche I. , Carazo J.M. , Baumeister W. , Scheres S.H. , and Nickell S. (2011). Maximum likelihood based classification of electron tomographic data. J Struct Biol 173 , 77–85. 10.1016/j.jsb.2010.08.005.20719249 Sun J. , Kinman L.F. , Jahagirdar D. , Ortega J. , and Davis J.H. (2022). KsgA facilitates ribosomal small subunit maturation by proofreading a key structural lesion. bioRxiv. Tancik M. , Srinivasan P. , Mildenhall B. , Fridovich-Keil S. , Raghavan N. , Singhal U. , Ramamoorthi R. , Barron J. , and Ng R. (2020). Fourier features let networks learn high frequency functions in low dimensional domains. Advances in Neural Information Processing Systems 33 , 7537–7547. Tegunov D. , and Cramer P. (2019). Real-time cryo-electron microscopy data preprocessing with Warp. Nat Methods 16 , 1146–1152. 10.1038/s41592-019-0580-y.31591575 Tegunov D. , Xue L. , Dienemann C. , Cramer P. , and Mahamid J. (2021). Multi-particle cryo-EM refinement with M visualizes ribosome-antibiotic complex at 3.5 A in cells. Nat Methods 18 , 186–193. 10.1038/s41592-020-01054-7.33542511 Turk M. , and Baumeister W. (2020). The promise and the challenges of cryo-electron tomography. FEBS Lett 594 , 3243–3261. 10.1002/1873-3468.13948.33020915 Vasyliuk D. , Felt J. , Zhong E.D. , Berger B. , Davis J.H. , and Yip C.K. (2022). Conformational landscape of the yeast SAGA complex as revealed by cryo-EM. Sci Rep 12 , 12306. 10.1038/s41598-022-16391-0.35853968 Xue L. , Lenz S. , Zimmermann-Kogadeeva M. , Tegunov D. , Cramer P. , Bork P. , Rappsilber J. , and Mahamid J. (2022). Visualizing translation dynamics at atomic detail inside a bacterial cell. Nature 610 , 205–211. 10.1038/s41586-022-05255-2.36171285 Zhang P. (2019). Advances in cryo-electron tomography and subtomogram averaging and classification. Curr Opin Struct Biol 58 , 249–258. 10.1016/j.sbi.2019.05.021.31280905 Zhong E.D. , Bepler T. , Berger B. , and Davis J.H. (2021). CryoDRGN: reconstruction of heterogeneous cryo-EM structures using neural networks. Nat Methods 18 , 176–185. 10.1038/s41592-020-01049-4.33542510 Zhong E.D. , Bepler T. , Davis J.H. , and Berger B. (2019). Reconstructing continuous distributions of 3D protein structure from cryo-EM images. arXiv preprint arXiv:1909.05215.
PMC010xxxxxx/PMC10316386.txt
==== Front BMJ BMJ BMJ-UK bmj The BMJ 0959-8138 1756-1833 BMJ Publishing Group Ltd. 37402509 10.1136/bmj-2022-073721 BMJ-2022-073721.R2 saga073721 Analysis 2474 Rethinking Health and Care SystemsAssessing resilience of a health system is difficult but necessary to prepare for the next crisis Sagan Anna technical officer 1 2 3 Thomas Steve Edward Kennedy chair of health policy and management research 4 Webb Erin research fellow 1 5 McKee Martin professor of European public health 1 3 1 European Observatory on Health Systems and Policies, London, UK 2 London School of Economics and Political Science, London, UK 3 London School of Hygiene and Tropical Medicine, London, UK 4 Centre for Health Policy and Management, School of Medicine, Trinity College Dublin, Dublin, Ireland 5 Department of Health Care Management, Technical University of Berlin, Berlin, Germany Correspondence to: A Sagan [email protected] 2023 4 7 2023 382 e073721© Author(s) (or their employer(s)) 2019. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ. 2023 BMJ https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/. Health systems responses to covid-19 can help to identify factors within and outside of the health system that affect its resilience to shocks, suggest Anna Sagan and colleagues ==== Body pmcHealth systems must constantly prepare for crises that threaten their operations. These include shocks that arise rapidly and are largely unforeseen, like pandemics or extreme weather events.1 Other types of shocks can arise more insidiously, such as the strains created by prolonged austerity or ageing populations. Yet as covid-19 showed, when a shock arrives, health systems are often unprepared. The concept of resilience has often been invoked in discussions of health system preparedness and response to crises. However, with this increased interest comes confusion about what health system resilience actually is and how it can be applied and measured to decide if a health system is resilient.2 While it is difficult to agree a comprehensive measure of resilience, valuable insights can be gained by looking at how well health systems performed during covid-19 to help prepare for future shocks.2 What is health system resilience? Put simply, something is resilient if it can, at least, recover rapidly after being exposed to a shock. However, in the case of a health system that can learn from experience we would hope that it does more than bounce back, especially if its original state could be improved. This learning is dynamic and often unpredictable, reflecting the multiplicity of actors and the complexity of power structures, processes, and feedback loops within the health system.3 Therefore a health system should not just react but also reflect and act, both during a shock and in its aftermath, to enhance its preparedness for any future shocks and ideally improve on how it was before. This points to a need to consider three phases in a health system response: absorption of a shock, adaptation to it, and transformation to strengthen the health system. Understood this way, resilience is a dynamic property of a health system that changes over time, sometimes rapidly, rather than being a constant state. Challenges to measuring the resilience of health systems Quantifying health system resilience to a shock is therefore challenging (box 1). To start with, deciding which indicators to measure resilience is not easy because there are different measures of success and what is considered important will vary over time.2 3 8 During the covid-19 pandemic the immediate priority was to interrupt transmission, and success was initially measured by numbers of covid-19 cases. Mortality during the pandemic was measured as case and infection fatality rates and excess mortality rates,9 each influenced by factors such as the extent of testing and diagnosis or, for excess mortality, the baseline chosen.10 Box 1 Challenges with quantifying a health system shock and its response Engineers test products to ensure that they can withstand shocks, but we cannot do the same with a health system. Modelling provides an alternative. However, although models can make good short term predictions, they face many challenges.4 Quantifying the scale and nature of both the shock and the predicted response in a health system is difficult. First, they may be difficult to measure—for example, because of the delay between infections occurring and data becoming available. In addition, infection may spread unnoticed within marginalised communities (which may be disproportionately affected because, for example, they live in precarious circumstances) making the size of an epidemic uncertain. Both the shock and the response are, using the word in its mathematical meaning, complex5 as they involve many diverse human responses.4 In a complex system, the spread of a shock and responses to it are influenced by where people and organisations start from (path dependency), reflecting factors such as morale, training, and reserve capacity; they are often non-linear and include feedback loops, with people adapting to the behaviour of those around them.6 Healthcare is especially constrained by how so many people and things must interact, especially when there are no easy substitutes. Scaling up hospital beds to cope with the surging numbers of people with covid-19, as was the aim of the Nightingale hospitals established in England, is relatively easy. Finding specialist staff is not. With so many factors at play, trying to model how health systems will respond to potential shocks is akin to informed guesswork. However, it does allow scenarios to be tested.7 Excess all-cause mortality has the benefit of also capturing deaths that are an indirect effect of the shock, such as those resulting from the decreased use of routine health services.11 12 However, more detailed and complex analysis is needed to disentangle the myriad factors that have contributed to excess mortality and, specifically, the functioning of the health system. This is inevitable given the blurred borders and complexity of health systems, and the importance of the socioeconomic context within which they are embedded. For example, indicators which focus on assessing health outcomes during covid-19, such as morbidity and mortality rates,13 14 15 16 17 miss some important factors, including morale of health workers and the effect on people who were socially isolated or dependent on essential services.11 Other indicators relevant to health system resilience relate to policy responses that lie beyond the health system, such as stringency of lockdowns, or other factors, such as the role of civil society in filling service gaps.18 Issues such as public trust and governance capacity must also be considered. For example, in the UK, public trust and confidence in the government’s ability to manage the pandemic was compromised at a critical time, which meant some people were less willing to follow rules and guidelines fundamental to controlling infection.19 Another challenge is knowing when to measure. Policies and outcomes change over time, and health systems often depend on decisions made beyond the health sector, the timing of which can be critical. When infections are increasing exponentially, for example, even a short delay in imposing restrictions on mixing will make an enormous difference in terms of the numbers of people infected and the subsequent effect on the health system.20 Some countries that initially responded rapidly and decisively to covid-19 had later surges of infections when they relaxed restrictions. This is exemplified by China, which having failed to achieve high vaccine uptake among its older population, saw high case numbers when it relaxed its successful control measures.21 22 These considerations apply equally to policies that contribute to societal resilience, such as social protection mechanisms that act over many years to reduce the number of people who might be more vulnerable to the effects of the pandemic and its countermeasures.23 Finally, given the complex environment within which health systems work, too narrow a focus on aspects of health system resilience, such as use of traditional measures of capacity (eg, workforce numbers, beds, and equipment), has been criticised for encouraging simplistic short term solutions to the detriment of other preparations such as prevention and control measures.24 Before the covid pandemic the US and UK were rated the top performing countries in pandemic preparedness based on traditional measures of health system capacity and technical measures of prevention (eg, biosafety) and detection (eg, laboratories). However, they had among the highest case and death rates, largely because of wider institutional weaknesses and poor political decision making.25 26 Operationalising a broader perspective of health system resilience requires understanding the challenges to comprehensively measuring it. Ideally, this would be a dynamic process, looking at various aspects of resilience, from patient care and frontline workers to political decision making, at different points in time. Barely coping workforce reflects lack of system resilience The pandemic was a stark reminder of the crucial importance of investment in a motivated, skilled, and engaged health workforce.27 Health workers in countries that acted rapidly to interrupt transmission, or which had previously invested in health system capacity, were relatively fortunate because they were better able to manage the shock.28 29 30 Others, whose governments had delayed acting and had underinvested over many years, struggled to respond, as did health systems where workers were already tired and demoralised.27 31 Yet, even in countries that were among the worst affected by rising numbers of cases, there were many examples of frontline health workers designing and implementing innovative responses in almost impossible circumstances. It is thus not surprising that in a 2022 systematic review of 68 studies measuring health system resilience across different health system shocks, workforce wellbeing was the second most commonly used indicator.32 Health system assessments of covid-19 responses often focused on the ability to ensure delivery of services both for patients with covid-19 and for those with other conditions by safeguarding and supporting the health workforce and, once vaccines became available, rolling them out at pace and scale, while reducing the accumulated backlogs of care. However, measures of the scale of disruption to care and subsequent recovery were fragmentary.33 34 One of the most detailed analyses, which included 31 services such as maternal and child health, HIV, and malaria, covered only 10 countries,35 and comparable data on important measures such as burnout in the health workforce are scarce. This is important because while a system may seem to be coping, it will be unsustainable if maintained by superhuman efforts of health workers. As with excess mortality, data on staff retention, sickness absence, or broader labour force participation can contribute useful insights because exhausted and sick health workers are a symptom of resilience being exhausted at the system level.36 What do we know about health system resilience during covid-19? Even though it is the best studied global shock ever, it is difficult, and perhaps impossible, to unequivocally say which health systems were the most resilient to covid-19 as the answer will change over time. Resilience of the health system is also influenced by decisions taken outside of it, but a focus on service delivery provides a useful starting point by looking at the ways health systems were able to absorb, adapt, transform, and learn during the covid-19 pandemic. Ability to absorb the shock The ability of a health system to absorb a shock is largely a function of its preparedness. Health systems need the resources to respond and a plan to deploy them rapidly and appropriately (box 2). Unfortunately, few countries were well prepared. Box 2 Response plans were inadequate A plan to guide the rapid and appropriate deployment of resources is critical to health system preparedness and response. Many countries did not have a pandemic preparedness and response plan in place, and even when one did exist, it had not been tested in realistic exercises. Even when plans had been tested, such as Exercise Alice in the UK, which sought to identify challenges to managing a Middle East respiratory syndrome (MERS) outbreak in 2016, recommendations were not always followed up.37 38 Many countries used plans designed for pandemic influenza rather than coronaviruses.39 This had serious consequences, not least because of the delayed recognition of the importance of airborne spread.40 The resources needed—facilities, materials (eg, pharmaceuticals, medical equipment, protective equipment), and people—must not only be available in the right quantities but also in the right combination, with the right characteristics (skill mix, function, technological specifications), and in the right place. While there were many examples of staff working flexibly—for example, by redeploying and repurposing facilities24—the absence of a single health worker with specialist skills, such as an anaesthetist, can prevent an entire intensive care team from functioning.41 Unfortunately, data systems enumerate individual items and not their performance in combination. The data we do have reveal stark differences in countries’ resource preparedness. Germany, with almost 34 intensive care beds per 100 000 population, was much better placed than England with 10.5 beds/100 000 and Ireland with 5/100 000 people.42 Low numbers of health workers in some countries compounded the challenges they faced. Within the European Economic Area the number of physicians differed by 60%, from fewer than 3 physicians/1000 population in Poland to 5/1000 people in Austria, while the variation in numbers of nurses was even greater, from fewer than 5/1000 population in Latvia and Bulgaria to 18/1000 in Norway.42 Having sufficient well motivated health workers is a core element of health system performance and resilience. Treatable mortality, which captures deaths that can be avoided through timely and effective care, can shed additional light on the combined performance of individual health system elements before and during the pandemic. European countries that were achieving lower treatable mortality, indicating stronger health system performance before the pandemic, also had lower excess mortality rates during covid-19 (fig 1). Investment in strengthening health systems is thus not only essential to reduce barriers to accessing health services and achieve the goal of universal health coverage43 but may improve resilience to shocks. Fig 1 Association between excess covid-19 mortality16 and pre-pandemic treatable mortality (Eurostat database, 2019) in European Economic Area countries. A negative excess mortality rate means that the mortality rate observed during the pandemic was lower than the mortality rate that would have been expected if covid-19 did not happen (ie, an improvement) Some of the resources needed can be stockpiled, repurposed, or scaled up rapidly, particularly non-specialist equipment such as beds. However, this will translate into meaningful healthcare capacity only if it is accompanied by a sufficient and capable workforce. In the short term, capacity can be partly boosted by mobilising and redeploying the existing workforce, although the risk of burnout is high. Medical students or retired health professionals can also be deployed, but this may require changes to regulatory systems and is not a panacea to acute shortages and chronic underinvestment in the workforce.24 There may, however, be potential to recruit staff for roles that require less training and supervision, as seen with volunteer vaccinators in the UK. The UK also recruited large numbers of contact tracers to work in private call centres, but at high cost and with questionable results.44 In addition, the appropriate design of health facilities is an overlooked aspect of preparedness. Countries in Asia benefited from having created physically separate pathways for infected and non-infected patients.45 This, coupled with attention to air quality, enabled them to maintain non-covid-19 services to a greater extent than in Europe.46 Other measures implemented in these countries in the aftermath of the severe acute respiratory syndrome (SARS) and Middle East respiratory syndrome (MERS epidemics, such as strengthening public health structures and response capacities, meant they were better able to respond to a new outbreak. How health systems adapted Adaptation involves reconfiguring the health system to respond to a threat. This is a reflective and proactive process, in contrast to the reactive aspects of absorption. It includes changes to the roles of health workers, adopting new technologies (such as lateral flow tests), and implementing new models of care. For example, many countries expanded the scope of work of nurses, pharmacists, and other health professionals.24 Countries where the roles of non-physicians had been most advanced before the pandemic—such as the UK, where nurses have long been able to prescribe within protocols—had an advantage.47 However, these changes brought substantial pressures, including high rates of burnout and moral injury (arising from health workers being unable to give lifesaving care because of inadequate resources).48 Advances in technology offer increasing possibilities for adaptation to a crisis. Teleconsultation greatly expanded in some countries, particularly in primary care.24 In Lithuania, for example, the number of remote primary care consultations was nearly 70 times higher in April 2020 than in April 2019.24 Although face-to-face consultations in primary care in England fell, online consultations more than compensated.49 Adaptation was also facilitated by earlier investment in data, research, and learning health systems. Having the right information and the capacity to learn was crucial to implementing adaptive actions. Initial capacity for surveillance and monitoring mattered, but many countries managed to put new systems in place.24 Yet, many gaps remain. For example, genomic sequencing, essential for tracking variants, is inadequate in many countries. Denmark and the UK were notable exceptions. Few countries collect data by ethnicity or migration status, essential to understand inequalities.49 Prior investment in research infrastructure made a difference. The UK’s Recovery trial, implemented at unprecedented speed and scale, provided crucial insights—for example on effectiveness of hydroxychloroquine and dexamethasone50—while the existence of learning networks, communities of practice that drove clinical and service innovation across England, was invaluable.51 Yet these UK successes were accompanied by some spectacular failures of governance, including possible corruption in public procurement of personal protective equipment, testing, and other services.52 53 Using pandemic response to transform the health system Sometimes, weaknesses revealed in the absorption and adaptation phases highlight a need to transform the system to enhance its preparedness for future shocks.2 A shock can thus be an opportunity to tackle longstanding problems by accelerating (or resurrecting) existing reforms or introducing new ones. For example, the Irish government took advantage of its pandemic response to unlock stalled progress towards universal health coverage (the Sláintecare reform), having made all covid related care free of charge,54 while Finland finalised its largest ever social and healthcare reform (Sote), transferring responsibilities for health, social, and rescue services from over 300 municipalities to about 20 larger entities.55 Thailand made similar advances while recuperating from the 1997 Asian financial crisis when it introduced ambitious universal health coverage reforms.56 Transformation can also be spurred by the absorptive and adaptive responses to the shock. For example, some short term absorptive measures, such as postponing elective surgery, are not sustainable in the longer term and may necessitate more profound changes such as establishing virtual hospitals or introducing dual patient pathways, as Asian countries did after the SARS epidemic. Increased use of teleconsultations and some of the skill mix adjustments, such as allowing certain health professionals to prescribe medicines or vaccinate, are examples of adaptive measures that could be become permanent after the pandemic.57 These changes can be either reactive, focusing on fixing problems uncovered during the pandemic, or proactive, drawing lessons from the traumas of the past three years to ensure that we are better prepared for the next pandemic or other shock.1 For example, a project initiated by the European Commission at the end of 2021 is testing the resilience of European health systems to different shocks, from natural disasters to the cost-of-living crisis, as well as antimicrobial resistance, climate change, and further pandemics, to identify potential weaknesses.58 At the same time, the EU’s Recovery and Resilience Facility offers member states funding to invest in the health sector for post-pandemic recovery.59 But with the political focus shifting to other problems,60 61 the urgency of investing in health system transformations may wane, even in the EU.62 Assessing health system resilience is worth the challenge A comprehensive measurement of resilience is a challenge, but efforts to assess it help us gain important insights into how health system resilience can be enhanced. In particular, looking at some of the actions health systems took to absorb, adapt, and transform in the midst of a crisis help us understand how various leverage points within and outside of the health system can enhance its preparedness and ability to maintain performance during a shock. Efforts to assess the resilience of health systems are justified not only to increase preparedness for future shocks but also to achieve universal health coverage and related sustainable development goals. Health systems that performed better before the pandemic were better placed when it arose, but a coordinated response within and among countries will be needed to prepare better for the next pandemic or shock. Key messages Health systems are often unprepared for shocks such as covid-19, but assessing resilience can help them prepare Health system resilience to a shock is difficult to measure comprehensively It can, however, be assessed by looking at what absorptive, adaptive, and transformative measures health systems adopted in response to a shock Responses to covid-19 show how health system strengthening and factors outside the health system affect resilience Contributors and sources: All contributors are engaged in research covering health systems, health systems resilience, and health systems responses to the covid-19 pandemic, and this article draws on their previous publications in these areas. AS, ST, and MM contributed equally to the conceptualisation, analysis, and implementing revisions. AS and MM led on the writing of the article and MM is the guarantor. EW supported the conceptualisation, analysis, and writing. Competing interests: We have read and understood BMJ policy on declaration of interests and have no interests to declare. Provenance and peer review: Commissioned; externally peer reviewed. This article is part of a collection proposed by the Health Foundation, which also provided funding for the collection, including open access fees. The BMJ commissioned, peer reviewed, edited, and made the decision to publish these articles. Rachael Hinton and Paul Simpson were the lead editors for The BMJ. ==== Refs 1 McKee M . Drawing light from the pandemic: a new strategy for health and sustainable development. World Health Organization. Regional Office for Europe for the European Observatory on Health Systems and Policies, 2021. 2 Thomas S, Sagan A, Larkin J, et al. Strengthening health systems resilience: Key concepts and strategies. European Observatory on Health Systems and Policies, 2020. 3 Topp SM . Power and politics: the case for linking resilience to health system governance. BMJ Glob Health 2020;5 :e002891. 10.1136/bmjgh-2020-002891.  32546590 4 Pagel C Yates CA . Role of mathematical modelling in future pandemic response policy. BMJ 2022;378 :e070615. 10.1136/bmj-2022-070615.  36109042 5 Nolte E McKee M Wait S . Describing and evaluating health systems. In: Bowling A Ebrahim S , eds. Handbook of health research methods: investigation, measurement and analysis. Open University Press, 2005: 12-43. 6 Reicher S Stott C . On order and disorder during the COVID-19 pandemic. Br J Soc Psychol 2020;59 :694-702. 10.1111/bjso.12398.  32609398 7 Rogers HL Barros PP Maeseneer J . Resilience testing of health systems: how can it be done? Int J Environ Res Public Health 2021;18 :4742. 10.3390/ijerph18094742.  33946804 8 Saulnier DD Blanchet K Canila C . A health systems resilience research agenda: moving from concept to practice. BMJ Glob Health 2021;6 :e006779. 10.1136/bmjgh-2021-006779.  34353820 9 Focacci CN Lam PH Bai Y . Choosing the right COVID-19 indicator: crude mortality, case fatality, and infection fatality rates influence policy preferences, behaviour, and understanding. Humanities and Social Sciences Communications 2022;9 :19. 10.1057/s41599-021-01032-0. 10 Shkolnikov VM Klimkin I McKee M . What should be the baseline when calculating excess mortality? New approaches suggest that we have underestimated the impact of the COVID-19 pandemic and previous winter peaks. SSM Popul Health 2022;18 :101118. 10.1016/j.ssmph.2022.101118.  35573866 11 Douglas M Katikireddi SV Taulbut M McKee M McCartney G . Mitigating the wider health effects of covid-19 pandemic response. BMJ 2020;369 :m1557. 10.1136/bmj.m1557.  32341002 12 Mansfield KE Mathur R Tazare J . Indirect acute effects of the COVID-19 pandemic on physical and mental health in the UK: a population-based study [published Online First: 20210218]. Lancet Digit Health 2021;3 :e217-30. 10.1016/S2589-7500(21)00017-0.  33612430 13 Oh J Lee H-Y Khuong QL . Mobility restrictions were associated with reductions in COVID-19 incidence early in the pandemic: evidence from a real-time evaluation in 34 countries. Sci Rep 2021;11 :13717. 10.1038/s41598-021-92766-z.  34215764 14 Kontis V Bennett JE Parks RM . Lessons learned and lessons missed: impact of the coronavirus disease 2019 (COVID-19) pandemic on all-cause mortality in 40 industrialised countries and US states prior to mass vaccination [version 2; peer review: 2 approved]. Wellcome Open Res 2022;6 :279. 10.12688/wellcomeopenres.17253.2.  35252592 15 Schöley J Aburto JM Kashnitsky I . Life expectancy changes since COVID-19. Nat Hum Behav 2022;6 :1649-59. 10.1038/s41562-022-01450-3.  36253520 16 Wang H Paulson KR Pease SA COVID-19 Excess Mortality Collaborators . Estimating excess mortality due to the COVID-19 pandemic: a systematic analysis of COVID-19-related mortality, 2020-21. Lancet 2022;399 :1513-36. 10.1016/S0140-6736(21)02796-3.  35279232 17 COVID-19 Forecasting Team . Variation in the COVID-19 infection-fatality ratio by age, time, and geography during the pre-vaccine era: a systematic analysis. Lancet 2022;399 :1469-88. 10.1016/S0140-6736(21)02867-1.  35219376 18 Blavatnik School of Government. Covid-19 government response tracker. 2023. https://www.bsg.ox.ac.uk/research/covid-19-government-response-tracker 19 Fancourt D Steptoe A Wright L . The Cummings effect: politics, trust, and behaviours during the COVID-19 pandemic. Lancet 2020;396 :464-5. 10.1016/S0140-6736(20)31690-1.  32771083 20 Arnold KF Gilthorpe MS Alwan NA . Estimating the effects of lockdown timing on COVID-19 cases and deaths in England: A counterfactual modelling study. PLoS One 2022;17 :e0263432. 10.1371/journal.pone.0263432.  35421094 21 Burki T . Moving away from zero COVID in China. Lancet Respir Med 2023;11 :132. 10.1016/S2213-2600(22)00508-2.  36535298 22 Mallapaty S . China is relaxing its zero-COVID policy - here’s what scientists think. Nature 2022;612 :383-4. 10.1038/d41586-022-04382-0.  36481953 23 McNamara CL McKee M Stuckler D . Precarious employment and health in the context of COVID-19: a rapid scoping umbrella review. Eur J Public Health 2021;31 (suppl_4 ):iv40-9. 10.1093/eurpub/ckab159.  34751369 24 Sagan A, Webb E, Azzopardi-Muscat N, et al. Health systems resilience during COVID-19: lessons for building back better. World Health Organization, Regional Office for Europe for the European Observatory on Health Systems and Policies, 2021. 25 Baum F Freeman T Musolino C . Explaining covid-19 performance: what factors might predict national responses? BMJ 2021;372 :n91. 10.1136/bmj.n91.  33509924 26 Nuclear Threat Initiative (NTI) and the Johns Hopkins Center for Health Security. Global Health Security Index. 2021. https://www.ghsindex.org/about/ 27 Fleming P Caffrey L Van Belle S How international health system austerity responses to the 2008 financial crisis impacted health system and workforce resilience – a realist review. Int J Health Policy Manage 2022. 10.34172/ijhpm.2022.7420 28 Hanefeld J Mayhew S Legido-Quigley H . Towards an understanding of resilience: responding to health systems shocks. Health Policy Plan 2018;33 :355-67. 10.1093/heapol/czx183.  29325025 29 Stuckler D Reeves A Loopstra R Karanikolos M McKee M . Austerity and health: the impact in the UK and Europe. Eur J Public Health 2017;27 (suppl_4 ):18-21. 10.1093/eurpub/ckx167.  29028245 30 Thomson S García-Ramírez JA Akkazieva B Habicht T Cylus J Evetovits T . How resilient is health financing policy in Europe to economic shocks? Evidence from the first year of the COVID-19 pandemic and the 2008 global financial crisis. Health Policy 2022;126 :7-15. 10.1016/j.healthpol.2021.11.002.  34857406 31 Baskin RG Bartlett R . Healthcare worker resilience during the COVID-19 pandemic: An integrative review. J Nurs Manag 2021;29 :2329-42. 10.1111/jonm.13395.  34182609 32 Fleming P O’Donoghue C Almirall-Sanchez A . Metrics and indicators used to assess health system resilience in response to shocks to health systems in high income countries-A systematic review. Health Policy 2022;126 :1195-205. 10.1016/j.healthpol.2022.10.001.  36257867 33 OECD, European Union. Dealing with backlogs: disruptions in non-COVID care during the pandemic. In: Health at a Glance: Europe 2022: State of Health in the EU Cycle. OECD, 2022. 34 van Ginneken E Reed S Siciliani L . Addressing backlogs and managing waiting lists during and beyond the COVID-19 pandemic. World Health Organization. Regional Office for Europe, 2022. 35 Arsenault C Gage A Kim MK . COVID-19 and resilience of healthcare systems in ten countries. Nat Med 2022;28 :1314-24. 10.1038/s41591-022-01750-1.  35288697 36 McKee M Hiam L . Britain’s not working. BMJ 2022;378 :o1883. 10.1136/bmj.o1883.  35905984 37 Dyer C . Pandemic preparedness: UK government kept coronavirus modelling secret. BMJ 2021;373 :n1501. 10.1136/bmj.n1501.  34117008 38 McKee M . The UK government tested the response to a coronavirus-why are we only discovering this now? BMJ 2021;375 :n2485. 10.1136/bmj.n2485.  34645602 39 An BY Tang S-Y . Lessons From covid-19 responses in east Asia: institutional infrastructure and enduring policy instruments. Am Rev Public Adm 2020;50 :790-800. 10.1177/0275074020943707. 40 Greenhalgh T Ozbilgin M Tomlinson D . How covid-19 spreads: narratives, counter narratives, and social dramas. BMJ 2022;378 :e069940. 10.1136/bmj-2022-069940. 41 van Klei WA Hollmann MW Sneyd JR . The value of anaesthesiologists in the COVID-19 pandemic: a model for our future practice? Br J Anaesth 2020;125 :652-5. 10.1016/j.bja.2020.08.014.  32896431 42 OECD, European Union. Chapter 7. Accessibility: affordability, availability and use of services. In: Health at a Glance: Europe 2020: State of Health in the EU Cycle. OECD, 2020. 43 Papanicolas I, Rajan D, Karanikolos M, et al. Health system performance assessment: a framework for policy analysis. World Health Organization, 2022. 44 Scally G Jacobson B Abbasi K . The UK’s public health response to covid-19. BMJ 2020;369 :m1932. 10.1136/bmj.m1932.  32414712 45 Cheng VCC Wong SC Chen JHK . Escalating infection control response to the rapidly evolving epidemiology of the coronavirus disease 2019 (COVID-19) due to SARS-CoV-2 in Hong Kong. Infect Control Hosp Epidemiol 2020;41 :493-8. 10.1017/ice.2020.58.  32131908 46 Han E Tan MMJ Turk E . Lessons learnt from easing COVID-19 restrictions: an analysis of countries and regions in Asia Pacific and Europe. Lancet 2020;396 :1525-34. 10.1016/S0140-6736(20)32007-9.  32979936 47 Anderson M O’Neill C Macleod Clark J . Securing a sustainable and fit-for-purpose UK health and care workforce. Lancet 2021;397 :1992-2011. 10.1016/S0140-6736(21)00231-2.  33965066 48 Expert Panel on Effective Ways of Investing in Health . Supporting mental health of health workforce and other essential workers. European Commission, 2022. 49 Routen A Akbari A Banerjee A . Strategies to record and use ethnicity information in routine health data. Nat Med 2022;28 :1338-42. 10.1038/s41591-022-01842-y.  35641824 50 Wise J Coombes R . Covid-19: the inside story of the Recovery trial. BMJ 2020;370 :m2670. 10.1136/bmj.m2670.  32641290 51 Sidhu MS Ford GA Fulop NJ Roberts CM . Learning networks in the pandemic: mobilising evidence for improvement. BMJ 2022;379 :e070215. 10.1136/bmj-2022-070215.  36207025 52 Iacobucci G . Covid-19: One in five government contracts had signs of possible corruption, report finds. BMJ 2021;373 :n1072. 10.1136/bmj.n1072.  33893123 53 McKee M . England’s PPE procurement failures must never happen again. BMJ 2020;370 :m2858. 10.1136/bmj.m2858.  32680840 54 Burke S Parker S Fleming P Barry S Thomas S . Building health system resilience through policy development in response to COVID-19 in Ireland: From shock to reform. Lancet Reg Health Eur 2021;9 :100223. 10.1016/j.lanepe.2021.100223.  34642676 55 European Commission. Finland finalises its largest-ever social and healthcare reform. European Commission, 2022. https://ec.europa.eu/social/BlobServlet?docId=25947&langId=en 56 Sumriddetchkajorn K Shimazaki K Ono T Kusaba T Sato K Kobayashi N . Universal health coverage and primary care, Thailand. Bull World Health Organ 2019;97 :415-22. 10.2471/BLT.18.223693.  31210679 57 van Schalkwyk MC Bourek A Kringos DS European Commission Expert Panel on Effective ways of Investing in Health . The best person (or machine) for the job: Rethinking task shifting in healthcare. Health Policy 2020;124 :1379-86. 10.1016/j.healthpol.2020.08.008.  32900551 58 Tille F Van Ginneken E Winkelmann J . Perspective: lessons from COVID-19 of countries in the European region in light of findings from the health system response monitor. Front Public Health 2023;10 :1058729. 10.3389/fpubh.2022.1058729.  36684940 59 Anderson M Forman R Mossialos E . Navigating the role of the EU Health Emergency Preparedness and Response Authority (HERA) in Europe and beyond. Lancet Regional Health . Europe 2021;9 :100203. 10.1016/j.lanepe.2021.100203 34661186 60 Gilson L Barasa E Nxumalo N . Everyday resilience in district health systems: emerging insights from the front lines in Kenya and South Africa. BMJ Glob Health 2017;2 :e000224. 10.1136/bmjgh-2016-000224.  29081995 61 Murphy A Fuhr D Roberts B Jarvis CI Tarasenko A McKee M . The health needs of refugees from Ukraine. BMJ 2022;377 :o864. 10.1136/bmj.o864.  35383103 62 World Bank. From double shock to double recovery: health financing in the time of covid-19. 2022. https://www.worldbank.org/en/topic/health/publication/from-double-shock-to-double-recovery-health-financing-in-the-time-of-covid-19
PMC010xxxxxx/PMC10317471.txt
==== Front Open Forum Infect Dis Open Forum Infect Dis ofid Open Forum Infectious Diseases 2328-8957 Oxford University Press US 37404953 10.1093/ofid/ofad234 ofad234 Major Article AcademicSubjects/MED00290 High-Sensitivity Troponins and Subclinical Coronary Atherosclerosis Evaluated by Coronary Calcium Score Among Older Asians Living With Well-Controlled Human Immunodeficiency Virus Chattranukulchai Pairoj Division of Cardiovascular Medicine, Department of Medicine, Faculty of Medicine, Chulalongkorn University, and Cardiac Center, King Chulalongkorn Memorial Hospital, Bangkok, Thailand Vassara Manasawee Division of Cardiovascular Medicine, Department of Medicine, Faculty of Medicine, Chulalongkorn University, and Cardiac Center, King Chulalongkorn Memorial Hospital, Bangkok, Thailand Siwamogsatham Sarawut Division of Hospital and Ambulatory Medicine, Department of Medicine, Faculty of Medicine, Chulalongkorn University, and Cardiac Center, King Chulalongkorn Memorial Hospital, Bangkok, Thailand Buddhari Wacin Division of Cardiovascular Medicine, Department of Medicine, Faculty of Medicine, Chulalongkorn University, and Cardiac Center, King Chulalongkorn Memorial Hospital, Bangkok, Thailand Tumkosit Monravee Department of Radiology, Faculty of Medicine, Chulalongkorn University, Bangkok, Thailand Ketloy Chutitorn Department of Laboratory Medicine, Faculty of Medicine, Chulalongkorn University, Bangkok, Thailand Shantavasinkul Prapimporn Division of Nutrition and Biochemical Medicine, Department of Medicine, Faculty of Medicine, Ramathibodi Hospital, Mahidol University, Bangkok, Thailand Apornpong Tanakorn HIV-NAT, Thai Red Cross AIDS Research Centre, Bangkok, Thailand Lwin Hay Mar Su HIV-NAT, Thai Red Cross AIDS Research Centre, Bangkok, Thailand Kerr Stephen J HIV-NAT, Thai Red Cross AIDS Research Centre, Bangkok, Thailand Biostatistics Excellence Centre, Faculty of Medicine, Chulalongkorn University, Bangkok, Thailand Boonyaratavej Smonporn Division of Cardiovascular Medicine, Department of Medicine, Faculty of Medicine, Chulalongkorn University, and Cardiac Center, King Chulalongkorn Memorial Hospital, Bangkok, Thailand Avihingsanon Anchalee HIV-NAT, Thai Red Cross AIDS Research Centre, Bangkok, Thailand Center of Excellence in Tuberculosis, Department of Medicine, Faculty of Medicine, Chulalongkorn University, Bangkok, Thailand HIV-NAT 006/207 study team Phanuphak Praphan Ruxrungtham Kiat Avihingsanon Anchalee Gatechompol Sivaporn Lwin Hay Mar Su Han Win Min Woratanarat Kobchoke Hiransuthikul Akarin Wichiansan Thanathip Boonrungsirisap Jedsadakorn Kerr Stephen J Apornpong Tanakorn Sophonphan Jiratchaya Phonphithak Supalak Wongvoranet Chuleeporn Chaiyahong Prachya Jirapasiri Jaravee Sarachat Paritaporn Setta Nattawadee Supakawee Khuanruan Duchchanutat Supaporn Ruengpanyathip Chavalun Phadungphon Chowalit Treepattanasuwan Orathai Boonmangum Theeradej Lertarrom Plengsri Uanithirat Anuntaya Chanthaburanun Sararut Anuchadbut Anongnart Tanjedrew Piyaporn Longcharaen Ratree Wongthai Niti Sattong Threepol Ubolyam Sasiwimol Mahanontharit Apicha Sopa Bunruan Chobkarching Umaporn Bouko Channuwat Phongam Nuchtida Iampornsin Thatri Dalodom Theera Khlaiphuengsin Apichaya Plakunmonthon Sasitorn Nanthapisal Kesdao Methanggool Umaporn Thangjitthanom Chornarin Sirichumpa Kanokon Chobkarjing Jutharos Jamrasrak Adisak Pitayanon Natthapa Phuengchangam Engon Chattranukulchai Pairoj Vassara Manasawee Buddhari Wacin Songmuang Smonporn Boonyaratavej Thimaporn Weerayut Siwamogsatham Sarawut Tumkosit Monravee Ketloy Chutitorn Shantavasinkul Prapimporn Sunthomyothin Sarat Wattanachanya Lalita Chaiwatanarat Tawachai Chutinet Aurauma Vongsayan Pongpat Samajarn Jitrada Putcharoen Opass Satitthummanid Sudarat Ariyachaipanich Aekarach Correspondence: Anchalee Avihingsanon, MD, PhD, HIV-NAT, Thai Red Cross AIDS Research Centre, 104 Ratchadamri Rd, Pathumwan, Bangkok 10330, Thailand ([email protected]); Pairoj Chattranukulchai, MD, Division of Cardiovascular Medicine, Department of Medicine, Faculty of Medicine, Chulalongkorn University, Bangkok 10330, Thailand ([email protected]). Potential conflicts of interest. All authors: No reported conflicts. 7 2023 03 5 2023 03 5 2023 10 7 ofad23406 12 2022 25 4 2023 01 5 2023 03 7 2023 © The Author(s) 2023. Published by Oxford University Press on behalf of Infectious Diseases Society of America. 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs licence (https://creativecommons.org/licenses/by-nc-nd/4.0/), which permits non-commercial reproduction and distribution of the work, in any medium, provided the original work is not altered or transformed in any way, and that the work is properly cited. For commercial re-use, please contact [email protected] Abstract Background Elevated levels of high-sensitivity cardiac troponin (hs-cTn) are suggestive of myocardial cell injury and coronary artery disease. We explored the association between hs-cTn and subclinical arteriosclerosis using coronary artery calcification (CAC) scoring among 337 virally suppressed patients with human immunodeficiency virus (HIV) who were ≥50 years old and without evidence of known coronary artery disease. Methods Noncontrast cardiac computed tomography and blood sampling for hs-cTn, both subunit I (hs-cTnI) and subunit T (hs-cTnT), were performed. The relationship between CAC (Agatston score) and serum hs-cTn levels was analyzed using Spearman correlation and logistic regression models. Results The patients, of whom 62% were male, had a median age of 54 years and had been on antiretroviral therapy for a median of 16 years; the CAC score was >0 in 50% of patients and ≥100 in 16%. Both hs-cTn concentrations were positively correlated with the Agatston score, with correlation coefficients of 0.28 and 0.27 (P < .001) for hs-cTnI and hs-cTnT, respectively. hs-cTnI and hs-cTnT concentrations of ≥4 and ≥5.3 pg/mL, respectively, provided the best performance for discriminating patients with Agatston scores ≥100, with a sensitivity and specificity of 76% and 60%, respectively, for hs-cTnI and 70% and 50% for hs-cTnT. In multivariable logistic regression analysis, each log unit increase in hs-cTnI level was independently associated with increased odds of having an Agatston score ≥100 (odds ratio, 2.83 [95% confidence interval, 1.69–4.75]; P <.001). Although not an independent predictor, hs-cTnT was also associated with an increased odds of having an Agatston score ≥100 (odds ratio, 1.58 [95% confidence interval, .92–2.73]; P = .10). Conclusions Among Asians aged ≥50 years with well-controlled HIV infection and without established cardiovascular disease, 50% had subclinical arteriosclerosis. Increasing hs-cTnI and hs-cTnT concentrations were associated with an increased risk of severe subclinical arteriosclerosis, and hs-cTn may be a potential biomarker to detect severe subclinical arteriosclerosis. Among 337 Asians ≥50 years old with well-controlled human immunodeficiency virus infection and without established cardiovascular disease, 50% had subclinical arteriosclerosis. High-sensitivity cardiac troponin (hs-cTnI) concentrations (subunits I and T) were positively correlated with Agatston score. Each log unit increase in hs-cTnI meant a 2.83-fold increase in the odds of severe subclinical atherosclerosis (Agatston score ≥100). PLWH coronary artery calcification high-sensitivity cardiac troponin subclinical atherosclerosis Higher Education Research Promotion 10.13039/100007469 Office of the Higher Education Commission 10.13039/100012527 Chulalongkorn University 10.13039/501100002873 NRU59-[014]-HR Ratchadapiseksompotch Fund Faculty of Medicine 10.13039/100019603 Chulalongkorn University 10.13039/501100002873 RA59/082 ==== Body pmcWith improvements in effective antiretroviral therapy (ART) and accessibility to treatment, people living with human immunodeficiency virus (HIV) (PLWH) are living longer with better quality of life. However, as PLWH age, the major causes of disease and death have shifted from those related to HIV toward comorbid conditions including atherosclerosis and cardiovascular disease (CVD) [1–3]. Several recent studies have demonstrated that HIV infection accelerates the process of atherosclerosis, by mechanisms including HIV-related immune dysregulation and inflammation, HIV therapies, host factors, and conventional CVD risk factors [4–7]. Several studies have assessed the prevalence of subclinical coronary atherosclerosis using different modalities [8]. Coronary artery calcification (CAC) detected with noncontrast cardiac computed tomography (CT) can predict future cardiovascular events and all-cause mortality rates in both PLWH and HIV-negative populations [4, 9–11]. Despite this, CAC screening in asymptomatic patients is still limited owing to accessibility, concerns about radiation exposure, and price [12]. Therefore, having an alternative technique for predicting CVD is desirable. Potential methods include the detection of high-sensitivity cardiac troponin (hs-cTn), both subunit I (hs-cTnI) and subunit T (hs-cTnT). hs-cTn has an important diagnostic role in acute cardiac care and provides prognostic value in predicting long-term cardiovascular events in asymptomatic, at-risk populations [13–15]. Several recent studies conducted in healthy volunteers without known CVDs showed that hs-cTn was associated with subclinical coronary atherosclerosis detected based on CAC [16–18]. However, its ability to predict subclinical coronary atherosclerosis in virologically suppressed, older PLWH has yet to be validated. We therefore sought to investigate the correlation between hs-cTn and subclinical atherosclerosis among Asian PLWH with well-controlled HIV infection and without established CVD, who had received ART for a median of 16 years. MATERIALS AND METHODS Study Design and Population PLWH aged ≥50 years were consecutively recruited from the prospective HIV-NAT 006 long-term cohort (clinicaltrials.gov NCT00411983) from March 2016 to May 2017 at the HIV-NAT, Thai Red Cross AIDS Research Centre, Bangkok, Thailand. Thorough medical history and HIV and non–HIV-related parameters were extracted from HIV-NAT's electronic health database. For the present study, the following were recorded on the same day noncontrast cardiac CT was performed: clinical examination findings, weight, height, waist circumference, body mass index (BMI), blood pressure, other medical conditions, including traditional cardiovascular risk factors, and fasting blood samples for lipid profile, glucose, and hs-cTn. The 10-year cardiovascular morbidity and mortality risks were calculated using the atherosclerotic CVD (ASCVD) risk score, with risk categorized as low (<7.5%), intermediate (7.5 to <20%), or high (≥20%) [19]. The exclusion criteria were established CVD and estimated glomerular filtration rate <50 mL/min/1.73 m2, calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation; the latter can affect hs-cTn levels [20]. All eligible participants underwent transthoracic echocardiography to exclude those with abnormal structural heart conditions that could increase hs-cTn levels—in particular, aortic stenosis, left or right ventricular systolic dysfunction, severe pulmonary hypertension, and severe ventricular hypertrophy [21]. Patient Consent Statement The protocol was approved by the Institutional Review Board of the Faculty of Medicine, Chulalongkorn University, Bangkok, Thailand. All participants gave their written informed consent. Multisection CT Study of CAC A multisection CT scanner (Somatom Sensation 64; Siemens Medical Systems) was used to quantify the coronary calcium using the Agatston score, which was calculated by multiplying the weighted density score by the pixel area of the calcification speck [22, 23]. After the initial scout images were acquired, the field of scan was placed on the chest area, covering the whole heart. The standard scan parameters included a 3-mm section thickness, 1.2 × 24-mm collimation, 0.37-second rotation, and spiral mode with 120 kVp at 80 mAs. The reconstruction was performed at 60% of the R-R interval. The Agatston scores were analyzed and confirmed by a single experienced observer who was blinded to the participant's medical history and biomarker levels. Biochemical Analyses Peripheral venous blood samples were taken on the same day as cardiac CT, processed, and stored at −80°C until analysis. The hs-cTnI levels were measured with the STAT high-sensitive cardiac troponin-I immunoassay on an Architect analyzer (Abbott Diagnostics), with a lower limit of detection (LLD) of 1.9 pg/mL. The 99th percentile reference limit for the overall population is 26.2 pg/mL, with a 4% intra-assay coefficient of variation. The hs-cTnT levels were quantified with the Cobas analyzer (Roche Diagnostics), with an LLD of 3 pg/mL and intra-assay coefficients of variation of 5% at 10 and 1% at 100 pg/mL. Statistical Analysis Categorical variables were expressed as frequencies and percentages, and continuous variables were summarized as mean (standard deviation) for normally distributed data and median (interquartile range [IQR]) for nonnormally distributed data. The hs-cTnT and hs-cTnI concentrations below the LLD were imputed as the LLD. The hs-cTnI and hs-cTnT levels were log-transformed to linearize their relationship with the logit function. The Agatston score was categorized into 4 groups—0, 1–99, 100–399, and ≥400—and the distribution of hs-cTn levels were assessed across the quartiles. A linear trend test was used to formally assess hs-cTn concentrations across ordered categories of Agatston score, and the relationship between hs-cTn and Agatston score as continuous variables were assessed using Spearman rank correlation coefficients. Multivariable logistic regression analysis was used to assess the association between CAC scores ≥100 and hs-cTn. Potential confounders included sex, age, waist circumference, hypertension, dyslipidemia, diabetes mellitus, absolute CD4 cell count, weight, smoking status, and hepatitis C virus infection. These were adjusted for in the multivariable model if the univariable P value was <.2. We assessed the relationship between log-transformed hs-cTnI and hs-cTnT levels and these confounders in multivariable linear regression models. Receiver operator characteristic (ROC) curves were constructed to assess the ability of hs-cTn to predict the Agatston score ≥100, for both hs-cTn alone and after adding it to the ASCVD score. Nested models were compared using the Aikake information criterion (AIC), and a likelihood ratio test was used compare the model fit. Differences were considered statistically significant at P < .05. Analyses were performed using Stata/SE software, version 15.0. RESULTS Population Characteristics A total of 337 PLWH were included, and their characteristics are shown in Table 1. Their median age (IQR) was 54 (52–59) years, and 210 (62%) were male. Nearly one-third had hypertension (n = 107 [32%]), and 60 (18%) had diabetes mellitus. Fourteen percent were current smokers. The median systolic blood pressure was 128 mm Hg, and the median diastolic blood pressure, 77 mm Hg. The median BMI (calculated as weight in kilograms divided by height in meters squared) was 23 kg/m2, and 11 participants (3%) had a BMI >30. The median waist circumference was 84 cm, and 34% of male and 58% of female participants had waist circumferences >90 and >80 cm, respectively. One-third of participants (33%) were on a statin regimen at the time of the study. Table 1. Demographics and Clinical Characteristics in Study Participants Baseline Characteristic Participants, No. (%)a (N = 337) Male sex 210 (62) Age, median (IQR), y 54 (52–59) Hypertension 107 (32) Diabetes mellitus 60 (18) BMI, median (IQR)b 23 (21–25) Waist circumference, median (IQR), cm 84 (78–90) Current smoker 48 (14) Blood pressure, median (IQR), mm Hg  Systolic 128 (118–137)  Diastolic 77 (71–84) Current statin therapy 111 (33) Laboratory values, median (IQR),  Total cholesterol, mg/dL 203 (179–236)  Triglycerides, mg/dL 160 (103–221)  LDL cholesterol, mg/dL 123 (99 -147)  HDL cholesterol, mg/dL 46 (39–57)  Absolute CD4 cell count, cells/µL 614 (483–803) Viral load <50 copies/mL 330 (98) ART duration, median (IQR), y 16 (13–19) NNRTI-based ART regimen 301 (89)  NRTI 197 (58)  PI 129 (38)  INSTI 9 (2.7)  Other 1 (0.3) HCV infection 30 (8.9) hs-cTnI, median (IQR), pg/mL 3.7 (2.7–5.2) hs-cTnI above LLD 316 (94) hs-cTnT, median (IQR), pg/mL 5.5 (3.8–8.7) hs-cTnT above LLD 288 (85) CAC (Agatston) score  0 170 (50)  1–99 113 (34)  100–399 31 (9)  ≥400 23 (7) ASCVD risk  Low risk (<7.5%) 191 (57)  Intermediate risk (≥7.5 to <20%) 107 (32)  High risk (≥20%) 39 (11) Abbreviations: ART, antiretroviral therapy; ASCVD, atherosclerotic cardiovascular disease; BMI, body mass index; CAC, coronary artery calcification; HCV, hepatitis C virus; HDL, high-density lipoprotein; hs-cTnI, high-sensitivity troponin I; hs-cTnT, high-sensitivity troponin T; INSTI, integrase strand transfer inhibitor; IQR, interquartile range; LDL, low-density lipoprotein; LLD, lower limit of detection; NNRTI, nonnucleoside reverse-transcriptase inhibitor; NRTI, nucleoside reverse-transcriptase inhibitor; PI, protease inhibitor. a Data represent no. (%) of participants unless otherwise specified. b BMI calculated as weight in kilograms divided by height in meters squared. The median (IQR) serum total cholesterol, low-density lipoprotein cholesterol, triglyceride and high-density lipoprotein cholesterol levels were 203 (179–236) mg/dL, 123 (99–147) mg/dL, 160 (103–221) mg/L, and 46 (39–57) mg/L, respectively. All participants were on ART, with a median duration (IQR) of 16 (13–19) years, and 98% had HIV-1 RNA levels <50 copies/mL. The median and nadir CD4 cell counts (IQR) were 614/µL (483–803/µL) and 176/µL (88–256/µL), respectively. Based on the 10-year ASCVD risk score, 56.8%, 31.7%, and 11.5% of participants were classified as low, moderate, and high risk, respectively. A total of 316 participants (94%) had hs-cTnI concentrations above the limit of detection (1.9 pg/mL); the median (IQR) was 3.7 (2.7–5.2) pg/mL. The overall 99th percentile for hs-cTnI was 39.7 pg/mL. Ten (3%) participants had hs-cTnI levels elevated to >26 pg/mL. A total of 288 participants (85%) had hs-cTnT concentrations above the limit of detection (3 pg/mL), with a median (IQR) of 5.5 (3.8–8.7) pg/mL. The 99th percentile for hs-cTnT was 38.3 pg/mL. Half of the participants had an Agatston score of 0, 34% had a score of 1–99, and 16% had an score of ≥100 (31 [9%] 100–399 and 23 [7%] ≥400; Table 1). In adjusted linear regression models, older age, higher serum creatinine levels, hypertension, and diabetes were associated with higher log hs-cTnT levels; older age, higher serum creatinine, and hypertension, with higher log hs-cTnI levels. Female sex was associated with lower levels of both troponin isoforms (Supplementary Table 1).  hs-cTn and Coronary Artery Calcium Across quartiles of hs-cTnI and hs-cTnT values, there were increasing proportions of participants with higher Agatston scores, particularly in the third and fourth quartiles (P < .001; Figure 1). Box-and-whisker plots (Figure 2) demonstrate increasing hs-cTn levels with increasing Agatston scores (P < .001 for both hs-cTnI and hs-cTnT). These levels differentiated clearly between participants with Agatston scores <100 and those with scores ≥100 (P < .001 for both subunits). The hs-cTn levels were slightly correlated with CAC, with Spearman correlation coefficients of 0.28 for hs-cTnI and 0.27 for hs-cTnT (both P < .001). Figure 1. Distribution of Agatston score expressed as percentages across high-sensitivity cardiac troponin (hs-cTn) quartiles, for hs-cTn subunit I (hs-cTnI) (left) and subunit T (hs-cTnT) (right). Figure 2. Box-and-whisker plots for high-sensitivity cardiac troponin subunit I (hs-cTnI) (left) and subunit T (hs-cTnT) (right) concentrations by Agatston score category. Lower edges of boxes represent the 25th percentile; upper edges, the 75th percentile; line within boxes, median values; whiskers, range of data, including outliers. P < .001 for both hs-cTnI and hs-cTnT, across score categories. ROC Curves to Detect Agatston Score ≥100 Figure 3 demonstrates that the 10-year ASCVD risk score and hs-cTn concentrations had similar abilities to discriminate Agatston scores ≥100. The area under the ROC curve (AUROC) was 0.757 (95% confidence interval [CI], .692–.822) for ASCVD risk score, 0.733 .659–.808) for hs-cTnI, and 0.672 (.594–.750) for hs-cTnT. The AIC for ASCVD risk score alone was 271.8436. Combining hs-cTn concentrations with the 10-year ASCVD risk resulted in an increase in AUROC after adding hs-cTnI (0.797 [95% CI, .734–.860]). The AIC for the combined model decreased to 257.2909 (likelihood ratio χ2P < .001). Figure 3. Receiver operating characteristic (ROC) curves to detect the presence of Agatston score ≥100 and atherosclerotic cardiovascular disease (ASCVD) risk score only or ASCVD risk score combined with high-sensitivity cardiac troponin subunit I (hs-cTnI) or subunit T (hs-cTnT) levels and multivariable (MV) models for hs-cTnI or hs-cTnT. Abbreviation: AUROC, area under the ROC curve. After adding hs-cTnT to the ASCVD risk score, there was a small increase in median AUROC to 0.760 (95% CI, .696–.826). The AIC of the combined model decreased to 266.2953 (likelihood ratio χ2P = .006). A hs-cTnI concentrations ≥4 pg/mL was the cutoff that maximized the sensitivity and specificity for an end point of Agatston score ≥100, with a sensitivity and specificity of 76% and 60%, respectively. For hs-cTnT, a cutoff value of ≥5.3 pg/mL provided the best performance to discriminating patients with Agatston scores ≥100, with a sensitivity and specificity of 70% and 50%, respectively. Multivariable Logistic Regression Evaluating Factors Associated With Agatston Scores ≥100 The results from the logistic regression are presented in Table 2. In the univariable model, both subunits of hs-cTn were significantly associated with Agatston score ≥100 (odds ratio [OR] per log unit increase [95% CI], 3.92 [95% CI 2.46–6.24] for hs-cTnI and 2.75 (1.74–4.33 for hs-cTnT; both P < .001) After adjustment for potential CVD risk factors in the multivariable model, hs-cTnI continued to be significantly associated with Agatston scores ≥100. Each log unit increase in hs-cTnI concentration increased the odds of Agatston scores ≥100 by 2.83 times (OR, 2.83 [95% CI, 1.69–4.75]; P < .001). In the multivariable model for hs-cTnT, male sex (OR, 2.44; P = .048), diabetes (2.34; P = .03), hypertension (2.27; P = .02), and waist circumference (1.04; P = .02) were also independently associated with Agatston scores ≥100. The AUROC was 0.830 (95% CI, .774–.885), and the AIC for this model was 243.6867, representing a significant improvement over the AUROC for hs-cTnI combined with ASCVD risk score (likelihood ratio χ2P < .001). Table 2. Univariable and Multivariable Logistic Regression Models for High-Sensitivity Troponin Subunits and Agatston Scores ≥100 Variable Univariable Model Multivariable Model for hs-cTnIa,b Adjusted Model for hs-cTnTa,c OR (95% CI) P Value OR (95% CI) P Value OR (95% CI) P Value Loge hs-cTnI 3.92 (2.46–6.24) <.001d 2.83 (1.69–4.75) <.001d … … Loge hs-cTnT 2.75 (1.74–4.33) <.001d … … 1.58 (.92–2.73) .10 Male sex 3.58 (1.68–7.60) .001d 2.44 (1.01–5.92) .048d  2.68 (1.14–6.31) .02d Age 1.09 (1.04–1.14) <.001d 1.05 (.99–1.11) .08  1.06 (1.001–1.19) .04d Hypertension 4.03 (2.21–7.38) <.001d 2.27 (1.14–4.52) .02d  2.36 (1.20–4.63) .01d Diabetes 4.11 (2.11–8.01) <.001d 2.34 (1.07–5.10) .03d  2.15 (1.004–4.60) .049d Dyslipidemia 1.34 (.48–3.74) .57 … … … … Waist circumference 1.07 (1.04–1.11) <.001d 1.04 (1.01–1.08) .02d  1.04 (1.01–1.08) .01d Current smoker 1.25 (0.57–2.76) .58 … … … … Absolute CD4 cell count 1.00 (0.99–1.00) .39 … … … … Serum creatinine 2.73 (1.00–7.45) .05 0.44 (.11–1.72) .24  0.48 (.13–1.82) .28 HCV infection 1.34 (.43–4.18) .62 … … … … Abbreviations: CI, confidence interval; HCV, hepatitis C virus; hs-cTnI, high-sensitivity troponin, subunit I; hs-cTnT, high-sensitivity troponin, subunit T; loge, natural logarithm transformed; OR, odds ratio. a Adjusted for variables with P < .1 in univariable models. b For the multivariable model for hs-cTnI, the Aikake information criterion (AIC) was 243.6867, and the area under the receiver operating curve (AUROC) was 0.8298. c For the adjusted model for hs-cTnT, the AIC was 257.4729, and the AUROC, 0.8020. d Significant at P < .05. In the multivariable hs-cTnT model, hs-cTnT was no longer an independent predictor, but the 95% CI remained consistent with an increased risk (OR, 1.58, [95% CI, .92–2.73]; P = .10). The AUROC for this model was 0.802 (95% CI, .745–.859), and the AIC was 257.429, representing a significant improvement over the AUROC for hs-cTnT combined with the ASCVD risk score (likelihood ratio χ2P = .002). Other potential confounders in the model had ORs and 95% CIs that were consistent in magnitude and precision with the multivariable model for hs-cTnI (Table 2). The multivariable models for both hs-cTnI and hs-cTnT showed adequate calibration (Hosmer and Lemeshow χ2P = .58 and P = .50, respectively). DISCUSSION This present study evaluated the association between hs-cTn and subclinical atherosclerosis in Asian PLWH ≥50 years old with well-controlled HIV infection and without documented coronary artery disease. We used CAC as determined by Agatston score as a surrogate for subclinical atherosclerosis burden because it has been shown to be superior to carotid intima-media thickness in predicting future cardiovascular events [8]. The overall prevalence of detectable CAC (score >0) in our cohort was 50%, which was close to the findings reported in a previous study conducted in PLWH without known CVDs in the Multicenter AIDS Cohort Study [24]. We found that both hs-cTn concentrations were associated with Agatston score ≥100. Multivariable analysis revealed that hs-cTnI concentration could discriminate well for PLWH with Agatston scores ≥100 or <100. In the present study, 94% and 85% of our participants had detectable hs-cTnI and hs-cTnT concentrations, respectively. Although hs-cTn can be detected in 74%–94% of the general population, hs-cTnI and hs-cTnT are detected at very low levels [15, 25, 26]. Korosoglou et al [27] suggested that repetitive microrupture of atherosclerotic plaques could result in embolization of the coronary microcirculation, which possibly accounts for microleakage of hs-cTnI and hs-cTnT into the bloodstream. Thus, elevated hs-cTnI and hs-cTnT levels could indirectly indicate an increased atherosclerotic burden [27]. The 99th percentile reference value of hs-cTnI in our cohort was 39.7 pg/mL, slightly higher than the 32.5 pg/mL found in a study using the same immunoassay in 4138 healthy individuals aged 55–64 years [15]. However, the overall distribution of hs-cTnI in the 2 cohorts was quite similar, with comparable median and IQR values. Accordingly, the difference in sample size between the studies could account for the difference in 99th percentile reference values. Because there is no established reference normal range of hs-cTnI specifically for the PLWH population, we were not able to compare our reference ranges with those in the prior study, conducted in an HIV-negative population. In our study, 3% of participants who were asymptomatic had hs-cTnI levels elevated above the recommended threshold for the diagnosis of acute myocardial infarction (26 pg/mL) [28]. Of these, 9 (90%) had detectable CAC. All of them had baseline characteristics similar to those in the whole HIV-NAT 006 cohort from which our study participants were drawn. Consistent with findings of a previous study in people living without HIV, our study demonstrates a weak, although significant, correlation between both subunits of hs-cTn and CAC. Olson et al [17] found that hs-cTnI levels increase along with increasing Agatston scores in the general population, with a correlation coefficient of 0.23 (P < .001), which is comparable to our finding of a correlation coefficient of 0.28 (P < .001), Notably, about one-third (37% for hs-cTnI and 38% for hs-cTnT) of our PLWH in the highest hs-cTn quartile had CAC scores of 0. Similar findings have been observed in the general population, with CAC scores of 0 in 45% of participants in the highest hs-cTnI quartile [17]. This indicates that the coronary plaque calcification, a result of the healing process after silent rupture, is not the only pathogenetic mechanism that accounts for hs-cTn elevation. Lee et al [29] reported that 7% of the plaques distributing along the coronary tree in healthy individuals are noncalcified. Several studies reported that asymptomatic PLWH had higher rates of noncalcified coronary plaque than persons without HIV [30, 31], Thus, if repetitive microrupture or erosion of such plaques occurred, it could be responsible for the elevation of hs-cTn in the absence of CAC. This has been highlighted in the study from Korosoglou et al [27], who reported a stronger correlation of hs-cTn with noncalcified plaques than with calcified plaques. The current study has some limitations. First, it was conducted in PLWH aged ≥50 years, almost all of whom were virologically suppressed without ventricular dysfunction. These findings may not be generalizable to PLWH with different rates of virological suppression or in those with ventricular dysfunction, although rates of ventricular dysfunction in PLWH ≥50 years old at our center are low and comparable to those in age and sex-matched HIV-negative controls [21]. Second, apart from subclinical atherosclerosis, some non–coronary atherosclerosis–related conditions that potentially influence on hs-cTn level—such as occult chronic opportunistic infection, myocardial inflammation, or structural heart diseases—may exist [32, 33]. However, almost all of our participants had good immune status, and abnormal structural heart conditions were excluded by echocardiogram. Third, because we collected only a single baseline blood sample, possible intraindividual variation in levels of both hs-cTn subunits was not examined. Fourth, in contrast to a previous study reporting that hs-cTnT concentrations were significantly associated with age [34] we found significant associations with age, diabetes, sex, and creatinine levels, which could potentially bias our study results. Elevated hs-cTn levels have demonstrated prognostic value in patients with chronic kidney disease from the general population with suspected acute coronary syndrome [35], and our multivariable logistic models adjusted for these significant associations. Fifth, studies in the general population have shown that while hs-cTnT levels are significantly increased in patients with coronary artery disease, with or without myocardial ischemia, only those with coronary artery disease and ischemia showed higher levels of another biomarker [N-terminal pro B-type natriuretic peptide (NT-pro-BNP)] [36]. We did not measure other biomarkers in the current study, so whether this finding is also applicable to PLWH is uncertain but is an area for future research. Finally, owing to the nature of cross-sectional studies, it remains unclear whether hs-cTn elevation provides prognostic benefit for long-term adverse clinical outcomes in this population. In conclusion, this study demonstrates a correlation between hs-cTn levels and subclinical atherosclerosis as indicated by CAC in Asians aged ≥50 years with well-controlled HIV-infection. hs-cTn could be a potential biomarker for early atherosclerotic risk stratification in this population. The clinical significance of elevated hs-cTnI levels and CAC score ≥100 with long-term adverse cardiovascular outcomes should be prospectively evaluated. Supplementary Material ofad234_Supplementary_Data Click here for additional data file. Acknowledgments Thanks to all of the participants for supporting this study. HIV-NAT 006 study team. Physician team: Praphan Phanuphak, Kiat Ruxrungtham, A. A., Sivaporn Gatechompol, H. M. S. L., Win Min Han, Kobchoke Woratanarat, Akarin Hiransuthikul, Thanathip Wichiansan, and Jedsadakorn Boonrungsirisap. Statistical team: S. J. K., T. A., and Jiratchaya Sophonphan. Nurse team: Supalak Phonphithak, Chuleeporn Wongvoranet, Prachya Chaiyahong, Jaravee Jirapasiri, Paritaporn Sarachat, and Nattawadee Setta. CRA team: Khuanruan Supakawee, and Supaporn Duchchanutat. Data team: Chavalun Ruengpanyathip, Chowalit Phadungphon, Orathai Treepattanasuwan, and Theeradej Boonmangum. Pharmacy team: Plengsri Lertarrom, Anuntaya Uanithirat, Sararut Chanthaburanun, Anongnart Anuchadbut, Piyaporn Tanjedrew, Ratree Longcharaen, Niti Wongthai, and Threepol Sattong. Laboratory team: Sasiwimol Ubolyam, Apicha Mahanontharit, Bunruan Sopa, Umaporn Chobkarching, Channuwat Bouko, Nuchtida Phongam, Thatri Iampornsin, Theera Dalodom, Apichaya Khlaiphuengsin, and Sasitorn Plakunmonthon. Finance team: Kesdao Nanthapisal, Umaporn Methanggool, Chornarin Thangjitthanom, Kanokon Sirichumpa, and Jutharos Chobkarjing. Administrative team: Adisak Jamrasrak, Natthapa Pitayanon, and Engon Phuengchangam HIV-NAT 207 aging study team. P. C., M. V., W. B., and S. B., and Weerayut Thimaporn, Division of Cardiovascular Medicine, Department of Medicine, Faculty of Medicine, Chulalongkorn University, and Cardiac Center, King Chulalongkorn Memorial Hospital, Bangkok, Thailand; S. S., Cardiac Center, King Chulalongkorn Memorial Hospital and Division of Hospital and Ambulatory Medicine, Department of Medicine, Faculty of Medicine, Chulalongkorn University; M. T., Department of Radiology, Faculty of Medicine, Chulalongkorn University; C. K., Department of Laboratory Medicine, Faculty of Medicine, Chulalongkorn University; P. S., Division of Nutrition and Biochemical Medicine, Department of Medicine, Faculty of Medicine, Ramathibodi Hospital, Mahidol University, Bangkok, Thailand; Sarat Sunthomyothin and Lalita Wattanachanya, Division of Endocrinology and Metabolism, Faculty of Medicine, Chulalongkorn University; Tawachai Chaiwatanarat, Division of Nuclear Medicine, Department of Radiology, Faculty of Medicine, Chulalongkorn University; Aurauma Chutinet, Pongpat Vongsayan, and Jitrada Samajarn, Division of Stroke, Department of Neurology, Faculty of Medicine, Chulalongkorn University; Opass Putcharoen, Division of Infectious Diseases, Faculty of Medicine, Chulalongkorn University; and Sudarat Satitthummanid and Aekarach Ariyachaipanich, Division of Cardiovascular Diseases, Faculty of Medicine, Chulalongkorn University. Financial support. This work was supported by the Higher Education Research Promotion and National Research University Project of Thailand, Office of the Higher Education Commission and the Ratchadapiseksompotch Fund, Faculty of Medicine, Chulalongkorn University (project code RA59/082). Supplementary Data Supplementary materials are available at Open Forum Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author. ==== Refs References 1 Sackoff JE , HannaDB, PfeifferMR, TorianLV. Causes of death among persons with AIDS in the era of highly active antiretroviral therapy: New York City. Ann Intern Med 2006; 145 :397–406.16983127 2 Triant VA . HIV infection and coronary heart disease: an intersection of epidemics. J Infect Dis 2012; 205 :S355–61.22577208 3 Burgess MJ , KastenMJ. Human immunodeficiency virus: what primary care clinicians need to know. Mayo Clin Proc 2013; 88 :1468–74.24290121 4 Vachiat A , McCutcheonK, TsabedzeN, ZachariahD, MangaP. HIV and ischemic heart disease. J Am Coll Cardiol 2017; 69 :73–82.28057253 5 Guaraldi G , ZonaS, OrlandoG, et al Human immunodeficiency virus infection is associated with accelerated atherosclerosis. J Antimicrob Chemother 2011; 66 :1857–60.21653601 6 Masia M , PadillaS, BernalE, et al Influence of antiretroviral therapy on oxidative stress and cardiovascular risk: a prospective cross-sectional study in HIV-infected patients. Clin Ther 2007; 29 :1448–55.17825696 7 Bloomfield GS , LeungC. Cardiac disease associated with human immunodeficiency virus infection. Cardiol Clin 2017; 35 :59–70.27886790 8 Folsom AR , KronmalRA, DetranoRC, et al Coronary artery calcification compared with carotid intima-media thickness in the prediction of cardiovascular disease incidence: the Multi-Ethnic Study of Atherosclerosis (MESA). Arch Intern Med 2008; 168 :1333–9.18574091 9 Blaha MJ , Cainzos-AchiricaM, GreenlandP, et al Role of coronary artery calcium score of zero and other negative risk markers for cardiovascular disease: the Multi-Ethnic Study of Atherosclerosis (MESA). Circulation 2016; 133 :849–58.26801055 10 Budoff MJ , HokansonJE, NasirK, et al Progression of coronary artery calcium predicts all-cause mortality. JACC Cardiovascular imaging 2010; 3 :1229–36.21163451 11 Budoff MJ , YoungR, BurkeG, et al Ten-year association of coronary artery calcium with atherosclerotic cardiovascular disease (ASCVD) events: the Multi-Ethnic Study of Atherosclerosis (MESA). Eur Heart J 2018; 39 :2401–8.29688297 12 Lin ZX , ZhouCS, SchoepfUJ, et al Coronary CT angiography radiation dose trends: a 10-year analysis to develop institutional diagnostic reference levels. Eur J Radiol 2019; 113 :140–7.30927938 13 Ford I , ShahAS, ZhangR, et al High-sensitivity cardiac troponin, statin therapy, and risk of coronary heart disease. J Am Coll Cardiol 2016; 68 :2719–28.28007133 14 Iribarren C , ChandraM, RanaJS, et al High-sensitivity cardiac troponin I and incident coronary heart disease among asymptomatic older adults. Heart 2016; 102 :1177–82.27030599 15 Zeller T , OjedaF, BrunnerFJ, et al High-sensitivity cardiac troponin I in the general population–defining reference populations for the determination of the 99th percentile in the Gutenberg Health Study. Clin Chem Lab Med 2015; 53 :699–706.25283141 16 Januzzi JLJ , SuchindranS, ColesA, et al High-sensitivity troponin I and coronary computed tomography in symptomatic outpatients with suspected coronary artery disease: insights from the PROMISE trial. JACC Cardiovasc Imaging 2018; 12 :1047–55.29550314 17 Olson F , EngborgJ, GronhojMH, et al Association between high-sensitive troponin I and coronary artery calcification in a Danish general population. Atherosclerosis 2016; 245 :88–93.26714045 18 Rusnak J , BehnesM, HenzlerT, et al Comparative analysis of high-sensitivity cardiac troponin I and T for their association with coronary computed tomography-assessed calcium scoring represented by the Agatston score. Eur J Med Res 2017; 22 :47.29145895 19 Goff DC Jr , Lloyd-JonesDM, BennettG, et al 2013 ACC/AHA guideline on the assessment of cardiovascular risk: a report of the American College of Cardiology/American Heart Association task force on practice guidelines. Circulation 2014; 129 :S49–73.24222018 20 deFilippi C , SeligerSL, KelleyW, et al Interpreting cardiac troponin results from high-sensitivity assays in chronic kidney disease without acute coronary syndrome. Clin Chem 2012; 58 :1342–51.22791885 21 Chattranukulchai P , ThimapornW, SiwamogsathamS, et al Echocardiographic findings among virally suppressed HIV-infected aging Asians compared with HIV-negative individuals. J Acquir Immune Defic Syndr 2020; 85 :379–86.32701821 22 Agatston AS , JanowitzWR, HildnerFJ, ZusmerNR, ViamonteMJr, DetranoR. Quantification of coronary artery calcium using ultrafast computed tomography. J Am Coll Cardiol 1990; 15 :827–32.2407762 23 McEvoy JW , BlahaMJ, DefilippisAP, et al Coronary artery calcium progression: an important clinical measurement? a review of published reports. J Am Coll Cardiol 2010; 56 :1613–22.21050970 24 Thomas GP , LiX, PostWS, et al Associations between antiretroviral use and subclinical coronary atherosclerosis. Aids 2016; 30 :2477–86.27490639 25 Zeller T , Tunstall-PedoeH, SaarelaO, et al High population prevalence of cardiac troponin I measured by a high-sensitivity assay and cardiovascular risk estimation: the MORGAM biomarker project Scottish cohort. Eur Heart J 2014; 35 :271–81.24104876 26 Blankenberg S , SalomaaV, MakarovaN, et al Troponin I and cardiovascular risk prediction in the general population: the BiomarCaRE consortium. Eur Heart J 2016; 37 :2428–37.27174290 27 Korosoglou G , LehrkeS, MuellerD, et al Determinants of troponin release in patients with stable coronary artery disease: insights from CT angiography characteristics of atherosclerotic plaque. Heart 2011; 97 :823–31.20884786 28 Thygesen K , AlpertJS, JaffeAS, et al Fourth universal definition of myocardial infarction (2018). Circulation 2018; 138 :e618–e51.30571511 29 Lee MS , ChunEJ, KimKJ, KimJA, YooJY, ChoiSI. Asymptomatic subjects with zero coronary calcium score: coronary CT angiographic features of plaques in event-prone patients. Int J Cardiovasc Imaging 2013; 29 :29–36. 30 D'Ascenzo F , CerratoE, CalcagnoA, et al High prevalence at computed coronary tomography of non-calcified plaques in asymptomatic HIV patients treated with HAART: a meta-analysis. Atherosclerosis 2015; 240 :197–204.25797313 31 Metkus TS , BrownT, BudoffM, et al HIV Infection is associated with an increased prevalence of coronary noncalcified plaque among participants with a coronary artery calcium score of zero: Multicenter AIDS Cohort Study (MACS). HIV Med 2015; 16 :635–9.25968104 32 Agzew Y . Elevated serum cardiac troponin in non-acute coronary syndrome. Clin Cardiol 2009; 32 :15–20.19143000 33 Sinning C , KellerT, ZellerT, et al Association of high-sensitivity assayed troponin I with cardiovascular phenotypes in the general population: the population-based Gutenberg health study. Clin Res Cardiol 2014; 103 :211–22.24271361 34 Oemrawsingh RM , ChengJM, Garcia-GarciaHM, et al High-sensitivity Troponin T in relation to coronary plaque characteristics in patients with stable coronary artery disease; results of the ATHEROREMO-IVUS study. Atherosclerosis 2016; 247 :135–41.26917225 35 Stacy SR , Suarez-CuervoC, BergerZ, et al Role of troponin in patients with chronic kidney disease and suspected acute coronary syndrome: a systematic review. Ann Intern Med 2014; 161 :502–12.25111593 36 Caselli C , PronteraC, LigaR, et al Effect of coronary atherosclerosis and myocardial ischemia on plasma levels of high-sensitivity troponin T and NT-proBNP in patients with stable angina. Arterioscler Thromb Vasc Biol 2016; 36 :757–64.26868212
PMC010xxxxxx/PMC10327009.txt
==== Front bioRxiv BIORXIV bioRxiv Cold Spring Harbor Laboratory 37425673 10.1101/2023.06.27.546733 preprint 1 Article The Salivary Microbiome and Predicted Metabolite Production are Associated with Progression from Barrett’s Esophagus to Esophageal Adenocarcinoma Solfisburg Quinn S 1 Baldini Federico 2 Baldwin-Hunter Brittany L 3 Lee Harry H 2 Park Heekuk 34 Freedberg Daniel E 35 Lightdale Charles J 3 Korem Tal 267* Abrams Julian A 358* 1 Department of Medicine, Boston University School of Medicine, Boston, MA, USA 2 Program for Mathematical Genomics, Department of Systems Biology, Columbia University Irving Medical Center, New York, NY, USA 3 Department of Medicine, Columbia University Irving Medical Center, New York, NY, USA 4 Microbiome and Pathogen Genomics Collaborative Center, Department of Medicine, Columbia University Irving Medical Center, New York, NY, USA 5 Digestive and Liver Disease Research Center, Columbia University Irving Medical Center, New York, NY, USA 6 Department of Obstetrics and Gynecology, Columbia University Irving Medical Center, New York, NY, USA 7 CIFAR Azrieli Global Scholars Program, CIFAR, Toronto, Canada 8 Herbert Irving Comprehensive Cancer Center, Columbia University Irving Medical Center, New York, NY USA Authors’ contributions: Quinn Solfisburg performed data analysis, data interpretation, and original drafting of the manuscript. Federico Baldini performed data analysis, data interpretation, and assisted with drafting the manuscript. Brittany Baldwin Hunter performed study conduct and provided critical input to the manuscript. Harry Lee performed data analysis and provided critical input to the manuscript. Daniel Freedberg performed data interpretation and provided critical input to the manuscript. Charles Lightdale performed study conduct and provided critical input to the manuscript. Tal Korem supervised data analysis, performed data interpretation, and assisted with drafting the manuscript. Julian Abrams designed the study, supervised data analysis, performed data interpretation, and assisted with drafting the manuscript. All authors approved the final version of the manuscript. * Co-corresponding authors: Tal Korem, PhD, 530 W 166th Street, Alianza Building, 3rd floor, New York, NY 10032, [email protected]; Julian Abrams, MD, MS, 630 W 168th Street, P&S 3-401, New York, NY 10032, [email protected] 28 6 2023 2023.06.27.546733https://creativecommons.org/licenses/by-nc/4.0/ This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License, which allows reusers to distribute, remix, adapt, and build upon the material in any medium or format for noncommercial purposes only, and only so long as attribution is given to the creator. nihpp-2023.06.27.546733.pdf Esophageal adenocarcinoma (EAC) is rising in incidence and associated with poor survival, and established risk factors do not explain this trend. Microbiome alterations have been associated with progression from the precursor Barrett’s esophagus (BE) to EAC, yet the oral microbiome, tightly linked to the esophageal microbiome and easier to sample, has not been extensively studied in this context. We aimed to assess the relationship between the salivary microbiome and neoplastic progression in BE to identify microbiome-related factors that may drive EAC development. We collected clinical data and oral health and hygiene history and characterized the salivary microbiome from 250 patients with and without BE, including 78 with advanced neoplasia (high grade dysplasia or early adenocarcinoma). We assessed differential relative abundance of taxa by 16S rRNA gene sequencing and associations between microbiome composition and clinical features and used microbiome metabolic modeling to predict metabolite production. We found significant shifts and increased dysbiosis associated with progression to advanced neoplasia, with these associations occurring independent of tooth loss, and the largest shifts were with the genus Streptococcus. Microbiome metabolic models predicted significant shifts in the metabolic capacities of the salivary microbiome in patients with advanced neoplasia, including increases in L-lactic acid and decreases in butyric acid and L-tryptophan production. Our results suggest both a mechanistic and predictive role for the oral microbiome in esophageal adenocarcinoma. Further work is warranted to identify the biological significance of these alterations, to validate metabolic shifts, and to determine whether they represent viable therapeutic targets for prevention of progression in BE. Columbia University Irving Scholar Award PR181960 Department of Defense Peer Reviewed Medical Research Program Clinical Trial AwardThis study was supported in part by the National Cancer Institute (U54 CA163004; R01 CA238433) and the Digestive Disease Research Foundation. T.K. is a CIFAR Azrieli Global Scholar in the Humans & the Microbiome Program. DEF was supported in part by a Department of Defense Peer Reviewed Medical Research Program Clinical Trial Award (PR181960) and by a Columbia University Irving Scholar Award. ==== Body pmcINTRODUCTION: Esophageal adenocarcinoma (EAC) has seen a dramatic rise in incidence in the past several decades, is often diagnosed at advanced stages, and is associated with poor survival.1,2 The factors that drive EAC remain incompletely understood. Barrett’s esophagus (BE) is the precursor lesion to EAC, but the overwhelming majority of BE patients do not progress to EAC. Established EAC risk factors, including gastroesophageal reflux disease (GERD) and obesity, do not fully explain the rise in its incidence.3,4 Increasing evidence suggests that the microbiome plays an important role in modifying the risk of a variety of epithelial cancers5–8 as well as in modulating the response to treatment.9–11 Changes in the esophageal microbiome have been observed in EAC and with progression from BE to EAC,12,13 raising the possibility that bacteria contribute to esophageal neoplasia. Reliable sampling of the esophageal microbiome, however, requires invasive procedures. A more accessible “window” to the esophageal ecosystem is the oral microbiome, which was shown to strongly influence it.14 A small study of the tumor-associated microbiome in EAC found a high prevalence of domination by oral flora such as Streptococcus,12 pointing to a link between the oral microbiome and EAC. Oral microbiome alterations have been associated with future risk of EAC15, and differences in the oral microbiome of BE patients were described previously in a small study of 49 patients.16 Alterations in the oral microbiome have also been associated with poor oral health17, which was in itself associated with increased risk of EAC in a recent analysis.18 It remains unclear how oral dysbiosis and poor oral health interact in their association with EAC. Finally, little is known with regard to oral microbiome alterations associated with neoplastic progression in BE patients. A clearer understanding of these oral microbiome changes could identify factors that may drive progression of neoplasia, representing novel therapeutic targets. Here, we profiled the salivary microbiome from 250 patients with various stages of BE and EAC who were undergoing upper endoscopy. We identify multiple characteristics of the oral microbiome associated with neoplastic progression in BE and show that they are independent of oral health. Using metabolic modeling, we predict metabolite profiles associated with alterations in BE, suggesting a mechanistic role for microbially produced metabolites. Finally, we show that the salivary microbiome offers a mild improvement in diagnostic accuracy compared to models based on clinical risk factors. Our results demonstrate the potential of studying the oral microbiome in the context of progression to EAC. RESULTS: Oral microbial composition from a large endoscopy cohort We recruited 250 adult patients undergoing upper endoscopy and characterized their oral microbiome using 16S rRNA gene sequencing. (Methods) A total of 244 patients were included in the analyses: 125 controls without Barrett’s esophagus (BE), and 119 BE patients (20 with non-dysplastic BE, 11 indefinite for dysplasia, 10 low grade dysplasia, 54 high grade dysplasia, and 24 intramucosal (T1a) adenocarcinoma). Patients with BE were more likely to be older (t-test p<0.001), male (Fisher exact p<0.001), white (p=0.001), or ever-smokers (defined as ≥100 lifetime cigarettes smoked) (p=0.003). They were also more likely to have GERD (p<0.001), to be treated with proton pump inhibitor (PPI; p<0.001), to take aspirin (p<0.001), and to have a higher BMI (p<0.001). (Table 1) There was no significant difference in the use of mouthwash between BE and non-BE patients (p=0.61), but non-BE patients were more likely to brush their teeth at least daily (98% vs 92%, p=0.03). Compared to non-BE, a significantly higher proportion of patients with BE had tooth loss (63% vs. 42%, p=0.001), largely due to an increase in tooth loss in patients with advanced neoplasia (defined as high grade dysplasia or adenocarcinoma, 66%; non-dysplastic BE, 40%; non-BE, 42%; Fisher’s exact p=0.001). (Figure 1) Older age (per year, adjusted OR 1.06, 95% CI 1.04–1.08) and a history of smoking (adjusted OR 2.12, 95% CI 1.07–4.19) were independently associated with tooth loss. (Supplementary Table 1) In multivariable analyses adjusting for EAC risk factors (age, male sex, white race, and GERD), tooth loss was associated with a non-significant increased risk of advanced neoplasia (vs. non-BE, adjusted OR 1.49, 95%CI 0.96–2.47). (Supplementary Table 2) These results are in line with a recent analyses of data from the Nurses’ Health Study, which found an association between both tooth loss and periodontal disease and risk of esophageal adenocarcinoma, but showed a decrease in association strength after adjusting for covariates.18 Established EAC risk factors were independently associated with advanced neoplasia even after controlling for daily tooth brushing, use of mouthwash, and presence of tooth loss, whereas these measures of oral health and hygiene were not independently associated with advanced neoplasia (p=0.21, 0.57, and 0.34, respectively). The oral microbiome of BE patients is progressively altered with dysplastic changes. To assess whether the salivary microbiome is associated with neoplastic progression, we focused our analyses on comparisons between three groups: non-BE (n=125), non-dysplastic BE (n=20), and advanced neoplasia (high grade dysplasia and adenocarcinoma [HGD/EAC]; n=78). We found that neoplastic progression was associated with significantly lower alpha diversity (Shannon: Kruskal-Wallis p=0.005, Figure 2A; Simpson: p=0.0029, Supplementary Figure 1). Compared to patients without BE, the alterations in alpha diversity were more pronounced in patients with advanced neoplasia than in those with nondysplastic BE (non-BE vs. non-dysplastic BE, Mann-Whitney p=0.11; non-BE vs advanced neoplasia, p=0.0006). There was no significant difference in alpha diversity comparing nondysplastic BE with advanced neoplasia (p=0.23). We further found that the oral microbiome from patients with advanced neoplasia tended to cluster separately than the rest of the cohort (weighted UniFrac, ANOSIM p<0.001; Figure 2B). Similar results were found when including all the subjects in their individual groups (non-BE, nondysplastic BE, IND, LGD, HGD, and EAC; Supplementary Figure 2). Our results indicate that the salivary microbiome alterations observed in advanced neoplasia are reflected both in the diversity of each individual’s microbiome (alpha diversity) and in compositional differences between individuals (beta diversity). We next checked whether specific microbes were associated with BE progression. We therefore compared relative abundance of different OTUs between all BE patients vs. non-BE using ALDEx2.19 (Methods) A total of 26 OTUs were identified as differentially abundant (p < 0.05, FDR corrected at 0.1; Figure 3). To assess whether the dysbiotic signature associated with BE is more pronounced with dysplastic changes, we checked whether these 26 OTUs were correlated with progression across the neoplastic spectrum, from no dysplasia to EAC (Methods). There was a significant association (p<0.05, FDR corrected at 0.1) for 23 of the 26 taxa, with a clear shift in composition with neoplastic progression, notably in the transition from low grade dysplasia (LGD) to high grade dysplasia (HGD). (Figure 3) This transition in composition from LGD to HGD is consistent with our prior observations of esophageal microbiome alterations with progression to EAC.20 The taxonomic alterations associated with progression were notable for increased relative abundance of several Streptococcus species. Streptococci form biofilms in the oral cavity21–23. The salivary microbiome is associated with advanced neoplasia even when controlling for tooth loss. Tooth loss is known to be strongly associated with oral microbiome composition17, and there was an increased proportion of advanced neoplasia patients with tooth loss. (Figure 1) Comparing patients who did and did not have all or most of their natural adult teeth, we found that those with tooth loss had lower alpha diversity (Shannon, Mann-Whitney p=0.001; Supplementary Figure 3A), and that oral microbiomes from both groups clustered separately (weighted UniFrac, ANOSIM p=0.003; Supplementary Figure 3B). We also identified 29 OTUs that had significantly different abundance between patients with and without tooth loss. (ALDEx2 p < 0.05, FDR corrected at 0.1; Figure 4) As poor oral health is associated with oral dysbiosis, this raised the question of whether the oral microbiome is associated with advanced neoplasia independent of tooth loss. We first examined whether salivary microbiome composition as a whole is associated with advanced neoplasia independently of tooth loss. We therefore calculated microbiome principal coordinates (PCos) using weighted UniFrac distances and used the top five PCos, which represented two thirds of the variance in microbiome composition. We then used multivariable logistic regression and found that PCo2 (explaining 22% of microbiome variation) and PCo4 (4.6%) were independently associated with advanced neoplasia. (p<0.001 and p=0.004, respectively) We then added the major EAC risk factors (age, sex, race, BMI, GERD history, and smoking history) to the model, and found that PCo2 remained independently associated with advanced neoplasia (p=0.004), suggesting that salivary microbiome composition represents a potential novel independent risk factor for EAC. Adding tooth loss to the model did not alter the association between PCo2 and advanced neoplasia (p=0.004), and in this model tooth loss was not independently associated with advanced neoplasia (p=0.12). Our results suggest that the association of tooth loss with advanced neoplasia is mediated through the oral microbiome. We next assessed whether associations between specific oral taxa and advanced neoplasia are independent of tooth loss. Of the 33 taxa associated with advanced neoplasia (N=78) vs. non-BE (N=125; ALDEx2 p < 0.05; FDR corrected at 0.1), 18 were also associated with tooth loss. (Figure 4) After adjusting for tooth loss in a generalized linear model, 20 of these taxa remained significantly associated with advanced neoplasia. (ALDEx2 p < 0.05, FDR corrected at 0.1) Notably, the four OTUs with the greatest increase in relative abundance in advanced neoplasia were all assigned to the genus Streptococcus, and the increased abundance of these Streptococcus OTUs in advanced neoplasia was independent of tooth loss. This corresponds with previous studies that found that the tumor-associated microbiome in EAC is often dominated by Streptococcus species.12 Metabolic modeling predicts distinct metabolic secretion capabilities in advanced neoplasia. Metabolite production by microbial communities is an important modality by which the microbiome affects the host. In order to assess if microbially produced metabolites might play a role as a driver or biomarker of neoplasia, we used microbiome community-scale metabolic models to predict metabolite secretion by the microbiome for every sample. (Methods) We found significant clustering of predicted metabolite profiles comparing advanced neoplasia cases with non-BE subjects (PERMANOVA p=0.001). Using principal components analysis, we found notable shift in the second component (15% explained variance; Mann-Whitney p=0.0003). (Figure 5A) Forty-four predicted metabolites had significantly altered abundance (p < 0.05, FDR corrected at 0.1) in advanced neoplasia. (Figure 5B) Notable alterations included increased predicted levels of L-lactic acid (p=0.023), a by-product of aerobic glycolysis, a hallmark of cancer which can contribute to neoplasticity;24 and 2-ketobutyric acid (p=0.033), previously reported to support mitochondrial respiration and cell proliferation.25 We also predicted that advanced neoplasia features a decrease in butyric acid (p=0.0089), a key promoter of gut homeostasis that was previously shown to be depleted in colon cancer and inflammatory bowel disease;26–28 and a decrease in L-tryptophan (p=0.0017). (Figure 5C) Circulating levels of tryptophan have been inversely associated with colon cancer risk29, and melatonin, a by-product of L-tryptophan metabolism, is under investigation for EAC prevention.30 The pattern of the shifts we predict is therefore consistent with the potential promotion of proliferation, inflammation, and cancer. Salivary microbiome data improves on clinical risk factors-based prediction of advanced neoplasia. We next performed an exploratory analysis to determine whether salivary microbiome features in this cohort could be used to distinguish advanced neoplasia from non-BE patients. As a baseline, we first trained a gradient boosted decision trees model which uses clinical risk factors for EAC and tooth loss to classify advanced neoplasia. The classifier was tested in cross-validation on patients not seen in the training of that model and achieved an area under the receiver operating characteristic curve (AUROC) of 0.84 (95%CI 0.79–0.89). The same process was then used to train a classifier using microbiome data, whereas, within each training fold, 10 OTUs were selected based on a Kruskal-Wallis test. This classifier had an AUROC of 0.72 (95%CI 0.65–0.79). Finally, a model trained on the combination of both microbiome data and EAC risk factors resulted in somewhat higher model accuracy, producing an AUROC of 0.88 (95%CI 0.83–0.91; vs. clinical risk factor model AUROC 0.84, 95%CI 0.79–0.89; DeLong p=0.053). (Supplementary Figure 4) A combined microbiome and clinical risk factor model with the outcome limited to HGD and excluding intramucosal EAC showed similar results with an AUROC 0.86 (95%CI 0.80–0.92), compared to an AUROC of 0.83 (95%CI 0.77–0.89) using only clinical risk factors for the same task. (Supplementary Figure 5) DISCUSSION: In this cross-sectional study of patients with and without BE, we detected marked shifts in the salivary microbiome with progression to EAC, with changes that appeared to be most pronounced in patients with advanced neoplasia. These changes included reduced diversity as well as significantly increased relative abundance of several taxa in the genus Streptococcus. As in previous studies, we found that tooth loss is more common in patients with advanced neoplasia. However, we show that many of the salivary microbiome associations observed in BE and advanced neoplasia persisted even when accounting for it. Further, we used metabolic modeling to identify distinct predicted metabolic secretion capabilities in advanced neoplasia. Our findings add to the growing body of evidence that the oral microbiome is linked to the esophageal microbiome and may contribute to esophageal neoplasia. In a case-control study of patients with EAC, BE, and controls, the EAC-associated microbiome had significantly reduced alpha diversity, similar to our observations in saliva.12 Interestingly, in that study 5/15 of the EAC tumors were dominated by Streptococcus spp. (relative abundance 69%-98%). In our salivary microbiome analyses, 4 of the 5 taxa most strongly associated with advanced neoplasia were also Streptococcus spp. Our group conducted a randomized controlled trial and found that an antimicrobial mouth rinse can produce esophageal microbiome and tissue gene expression changes, highlighting the relevance of oral bacteria to esophageal disease.31 In another cohort study analyzing mouth rinse samples from patients enrolled in two large cancer prevention studies, oral microbiome alterations were noted to precede an EAC diagnosis by several years.15 In a small study of 49 patients we previously noted marked salivary microbiome alterations associated with BE and also with advanced neoplasia.16 Our study features the use of microbiome metabolic models to identify broad shifts in metabolites predicted to be produced by the saliva microbiome. Many of the predicted changes to metabolite outputs correspond with existing knowledge. Lactic acid, for example, was predicted to be increased in advanced neoplasia. Lactic acid can serve as a major energy source for proliferative cancer cells, and is known to activate hypoxia inducible factors, which in turn contribute to proliferation, angiogenesis, and other neoplastic features.24 Our findings could therefore support the hypothesis that the oral and esophageal microbiota promotes EAC development and progression via production of metabolites.32 Our findings further correspond with a previous study detecting lactic-acid bacteria in many esophageal adenocarcinomas.12 However, the biological significance of predicted metabolite production is unclear, and future studies are needed to validate these predictions and to elucidate the biological effects of specific bacterial metabolites on esophageal neoplasia. Prior work has associated tooth loss and periodontal disease with increased risks of esophageal squamous cell cancer and gastric cancer, and a recent study found an association between both tooth loss and periodontal disease and risk of esophageal adenocarcinoma.18 This indicates a potential confounding effect, as tooth loss is also associated with major alterations in oral microbiome composition.17 Our study offers an explanation for these associations, demonstrating that salivary microbiome composition is independently associated with advanced neoplasia, even when adjusting for EAC risk factors and for tooth loss. These findings suggest that the association between tooth loss and esophageal neoplasia is mediated by changes in the salivary microbiome, and that the salivary microbiome may represent a novel independent risk factor for EAC. We performed exploratory analyses to assess whether the salivary microbiome could discriminate patients at highest EAC risk. The salivary microbiome is highly suitable for diagnostics, as it is stable over time33–35, especially compared to other body sites36, and is resistant to perturbations.37 Addition of a microbiome-based classifier to EAC risk factors resulted in modest improvement in discrimination. However, the current study was not specifically designed to address this question, and future studies should explore further the salivary microbiome as a potential biomarker for advanced neoplasia. Important strengths of the current study include the relatively large sample size and the inclusion of oral health and hygiene information from patients. The large sample size allowed for the detection of significant microbiome alterations, even when correcting for multiple comparisons. Previous studies of the oral microbiome in BE and EAC have not included oral health and hygiene data, key potential confounders. The patients were well characterized, with data collected on key EAC risk factors including GERD history, BMI, and smoking, which permitted microbiome analyses adjusting for these variables. The BE patients in the study were demographically similar to BE populations from other studies, enhancing the generalizability of the findings. Lastly, novel methods for predicted microbiome metabolic profiling allowed for insights into functional correlates of the salivary microbiome alterations. The study does have certain limitations. There were a relatively small number of non-dysplastic BE patients, limiting analyses in this subgroup. Analyses did not incorporate dietary intake; however, previous studies suggest that diet has minimal impact on salivary microbiome composition.38–40 No conclusions can be drawn with regard to temporality in this cross-sectional study. It is possible that the observed salivary microbiome alterations were caused by BE-associated advanced neoplasia, although we believe that this is unlikely. Tooth loss was self-reported rather than measured, and periodontal disease was not directly assessed. Community-scale metabolic models also have notable limitations. Our analysis was based on 16S rRNA gene sequencing, which does not allow us to tailor models to specific strains or genetic potential present in each sample. Additionally, while genome-scale models have been curated for common gut commensals, to our knowledge, such efforts have not been done for oral microbes. Consequently, some models may be missing, while existing ones may lack representation of niche-specific metabolic capacity. Despite these limitations, these models allow a systematic application of biochemical and genetic knowledge to our analysis and raise interesting hypotheses that could be experimentally validated. In conclusion, patients with BE-associated advanced neoplasia have a markedly altered salivary microbiome, and analyses of taxonomic alterations associated with stages of progression from BE to EAC appear to indicate that these changes are most notable at the transition from low- to high-grade dysplasia. Increased tooth loss was also observed with progression to EAC, although the salivary microbiome alterations were largely independent of tooth loss, suggesting that the association of tooth loss with advanced neoplasia is mediated through the oral microbiome. There were marked increases in various taxa in the genus Streptococcus in advanced neoplasia, possibly pointing to a biological contribution of these bacteria to neoplastic progression. In addition to the microbiome alterations, progression to EAC was associated with numerous changes to predicted bacterial metabolite production, with notable alterations that suggest possible proneoplastic effects related to these shifts. Further work is warranted to identify the biological significance of the microbiome alterations, to validate metabolic shifts, and to determine whether they represent viable therapeutic targets for prevention of progression in BE. METHODS: Study Design A total of 250 patients with and without BE undergoing upper endoscopy at Columbia University Irving Medical Center (New York, NY) were prospectively enrolled from February 2018 through February 2019. Patients were ≥18 years old and scheduled to undergo endoscopy for clinical indications. Patients were excluded if they had a concurrently scheduled colonoscopy, had a history of gastric or esophageal surgery, a history of esophageal squamous cell cancer, or use of antibiotics, steroids, or other immunosuppressants in the 3 months prior to the procedure. This study was approved by the Columbia University Institutional Review Board. All patients provided written informed consent. Data were collected on patient demographics and anthropometrics (to calculate BMI) as well as clinical information including medical history, history of gastroesophageal reflux disease (GERD; defined as experiencing frequent heartburn or fluid regurgitation), medication use at time of enrollment (with specific notation of daily use of proton pump inhibitors (PPIs), histamine-2 receptor antagonists, statins, and daily use of aspirin and non-steroidal anti-inflammatory drugs), alcohol history, and smoking history (ever smoking defined as having smoked >100 lifetime cigarettes). Data were collected on self-reported oral health and hygiene. Tooth loss was assessed using categories adapted from Borningen et al.17: all or most of natural adult teeth, partial plates or implants, full upper dentures or implants, full lower dentures or implants, full upper and lower dentures or implants. Data were also collected on tooth brushing and mouthwash use. Patients did not eat or drink after midnight prior to the endoscopy and saliva collection; saliva was collected prior to the endoscopy. Patients were categorized as BE if they had a history of endoscopically suspected BE with intestinal metaplasia on esophageal biopsies. BE patients were further categorized based on the highest degree of neoplasia ever (no dysplasia (NDBE), indefinite for dysplasia (IND), low grade dysplasia (LGD), high grade dysplasia (HGD), adenocarcinoma (EAC)). Microbiome Sequencing and Analysis The 16S rRNA V3-V4 region was amplified using Illumina adapter-ligated primers.41 The Illumina Nextera XT v2 index sets A-D were used to barcode sequencing libraries. Libraries were sequenced on an Illumina MiSeq using the v3 reagent kit (600 cycles) and a loading concentration of 12 pM with 10% PhiX spike-in. Sequences were assigned to operational taxonomic units (OTUs) using USEARCH42 with ≥ 97% sequence homology. Taxonomic assignments for the OTUs were based on the Human Oral Microbiome Database (HOMD).43 Any subsequently unassigned OTUs were assigned by referencing the Ribosomal Database Project (RDP).44 Samples were subsampled to 10,000 reads to compare across even sequencing depths while minimizing data loss. Five patients were excluded after sequencing because of relatively low sequencing depth with <10,000 total reads per sample. The median read count for the full cohort was >33,000. One patient was excluded because of a history of both EAC and esophageal squamous cell carcinoma. Microbiome metabolic modeling of oral microbial communities Microbiome metabolic modeling was performed using the Microbiome Modeling Toolbox (COBRA toolbox commit: 71c117305231f77a0292856e292b95ab32040711) 45,46 and the AGORA metabolic models (AGORA 1.02).47 All computations were performed in MATLAB version 2019a (Mathworks, Inc.), using the IBM CPLEX (IBM, Inc.) solver. We first matched species detected by our microbial sequencing analysis with the ones present in AGORA.47 Because AGORA metabolic models are available at the strain level, we generated species-level models using the createPanModels.m function of the Microbiome Modeling Toolbox (MMT)45 as previously described.48 To increase the number of species represented in our microbiome models we chose genus-level representative models for abundant microbes present in the oral cavity with >5% relative abundance in more than 10 samples. There were six species without a corresponding metabolic model, and these were either grouped with similar species or excluded from the analyses (See Supplementary Table 3 for details). We then used the mgPipe.m automated pipeline of the MMT to build and interrogate sample-specific microbiome metabolic models. Briefly, for each sample, personalized microbiome models are created by joining species-level metabolic models using the compartmentalization technique49; a lumen compartment enabling microbial metabolic interactions is added, as well as additional input and output compartments, allowing microbiome intake and secretion of metabolites. Altogether our microbiome models included 160 microbial species with an average of 50 species for each sample and a maximum of 69. As constraint-based metabolic modeling benefits from a specification of the metabolic environment such as media and carbon source availability50, we applied a “western diet”51 to each sample in the form of constraints on the metabolites uptake reactions.51 Finally, to obtain metabolic predictions, we used the Net Maximal Production Capabilities (NMPCs) through the mgPipe pipeline45 to provide predictions of the metabolite secretion profile of each sample. To detect significant changes in NMPCs distributions between cases and controls a Mann-Whitney U test was performed for each retained NMPCs. Only NMPCs which were present in at least 10% of the cases and had at least a value of 0.01 were retained for the significance analysis. FDR correction using the Benjamini–Hochberg procedure was applied. Statistical Analysis The primary groups of comparison were BE patients with advanced neoplasia (HGD or EAC), non-dysplastic BE (NDBE) and non-BE controls. Grouping high grade dysplasia and intramucosal adenocarcinoma together as advanced neoplasia reflects common practice as well as clinical guidelines for treatment.52 There is extremely low inter-observer agreement (even among expert gastrointestinal pathologists) for the diagnosis of LGD53–55, as inflammation-induced cytologic atypia mimics the findings of LGD. As a result, while estimates of cancer risk for LGD are relatively low on average,54 these estimates vary widely, thus making interpretations of findings for this group challenging. Patients with low grade dysplasia or indefinite for dysplasia were included in analyses assessing for alterations in the oral microbiome across the entire BE neoplastic spectrum. While there was no a priori reason to suspect that endoscopic therapy would have altered the salivary microbiome, comparisons were made between those patients with LGD or worse who had (n=78) and had not (n=10) received prior endoscopic therapy. There were no differences in alpha diversity (p=0.16), no evidence of clustering on beta diversity analyses (ANOSIM p=0.13), and no differentially abundant taxa. Thus, treated and untreated patients were grouped together for all analyses. Categorical variables were compared across groups using Fisher’s exact tests. Continuous variables were analyzed using t-tests or rank sum tests as appropriate, with ANOVA and Kruskal Wallis tests for ≥3 groups. For purposes of analyses, tooth loss was dichotomized as having all or most of natural adult teeth (yes/no). Multivariable logistic regression was performed to assess the association between tooth loss and advanced neoplasia, adjusted for known EAC risk factors (age, sex, GERD, body mass index (BMI), smoking). Alpha diversity was evaluated using the Shannon diversity index and beta diversity using weighted UniFrac56 distances. Groups were compared using both permutational multivariate analysis of variance (PERMANOVA) for predicted metabolite profiles and analysis of similarities (ANOSIM) for microbial compositions. To find differential abundances between study groups, the ALDEx219 R package was used. For differential abundance analyses, only OTUs present in at least 5% of all samples were included to allow for more meaningful comparisons. ALDEx2 was used to compare worst histological grades of BE as an ordinal variable in a generalized linear model and to assess correlation of BE-associated OTUs with neoplastic progression using aldex.corr to treat worst histological grade as a continuous variable. ALDEx2 was also used to find significance for differentially abundant taxa in a multivariate model with both advanced neoplasia and tooth loss. Generalized linear models were used to assess differential relative abundance of bacterial taxa in advanced neoplasia, adjusted for tooth loss. Multivariable logistic regression was performed to detect associations between advanced neoplasia and microbiome composition (represented by its top five principal coordinates), adjusted for EAC risk factors (age, sex, race, BMI, smoking, GERD). Supervised machine learning was used to classify patients with advanced neoplasia using the LightGBM package.57 Three models were created: 1) EAC risk factors alone (age, sex, race, BMI, smoking, GERD); 2) microbiome features alone; and 3) EAC risk factors and microbiome features together. Model parameters were optimized per fold in 10-fold cross-validation, with strict train-test sterility. The output of the models were predicted probabilities of whether a patient has advanced neoplasia or no BE, with the goal of identifying the patients at highest risk of mortality from EAC. All statistical analyses were performed in Python or R. Statistical significance was defined as p<0.05. Differential abundance analyses were corrected for multiple comparisons using the Benjamini-Hochberg procedure, and corrected statistical significance was defined as p<0.1. 95% confidence intervals for AUCs were calculated using the DeLong method using pROC.58 Supplementary Material Supplement 1 Figure 1. Tooth loss is significantly more common in advanced neoplasia. A significantly higher proportion of patients with advanced neoplasia (high grade dysplasia or esophageal adenocarcinoma) had tooth loss as compared to non-BE and non-dysplastic BE patients combined (Fisher’s exact p = 0.001). P – Fisher’s exact. Figure 2. A microbial signature of BE. (A) Patients with advanced neoplasia have significantly reduced alpha diversity compared to non-BE patients. Kruskal-Wallis overall p-value=0.005. p – Mann-Whitney U test. (B) Weighted UniFrac PCoA demonstrated significant clustering of patients with advanced neoplasia (ANOSIM p=0.001). NDBE, non-dysplastic BE; advanced neoplasia – high grade dysplasia or esophageal adenocarcinoma; ellipse, 2 standard deviation sigma ellipse. Figure 3. Increased oral dysbiosis with progressive dysplasia. Shifts in the oral microbiome compared to non-BE patients (shown as ALDEx2 effect sizes) were more pronounced with progression from no dysplasia to EAC, particularly notable in patients with high grade dysplasia and EAC. Bolded OTUs were significantly associated with neoplastic progression. (p < 0.05, FDR corrected at 0.1; Methods) NDBE, nondysplastic Barrett’s esophagus; Indef, indefinite for dysplasia; LGD, low grade dysplasia; HGD, high grade dysplasia; EAC, esophageal adenocarcinoma. Figure 4. Oral microbes are independently associated with advanced neoplasia and tooth loss. Differentially abundant taxa in patients with advanced neoplasia (high-grade dysplasia or esophageal adenocarcinoma) compared to non-BE (left) and in patients with and without tooth loss (right). Many taxa were associated with both neoplasia and tooth loss, yet most of these remained significantly differentially abundant after adjusting for tooth loss and advanced neoplasia, respectively (ALDEx2 p < 0.05, FDR corrected at 0.1; denoted by asterisks). Figure 5. The predicted metabolic profile is altered in patients with advanced neoplasia. (A) Significant clustering by advanced neoplasia status on principal components analysis (PERMANOVA p=0.001), with pronounced shifts in PC2 (p=0.0003). (B) Volcano plot demonstrating differentially abundant metabolites in advanced neoplasia. (C) Significantly altered predicted levels of L-lactic acid, 2-ketobutyric acid, butyric acid, and L-tryptophan in advanced neoplasia. Plot capped at −4 for butyrate. P, Mann-Whitney test Table 1. Patient Characteristics. P – t-test or Fisher exact p for difference between BE and non-BE. All (n=244) Non-BE (n=125) BE (n=119) P-value Age, years; mean (SD) 57.8 (18.7) 50.9 (18.7) 65.0 (15.8) <0.001 Sex, male; N (%) 140 (57%) 45 (36%) 95 (80%) <0.001 Ever smokers; N (%) 103 (42%) 41 (33%) 62 (52%) 0.002 BMI mean (SD) 27.5 (6.7) 27.0 (7.9) 28.1 (5.1) <0.001 Race, white; N (%) 222 (90%) 105 (84%) 117 (98%) <0.001 GERD; N (%) 167 (68%) 64 (51%) 103 (87%) <0.001 PPI use; N (%) 148 (60%) 41 (33%) 107 (90%) <0.001 Aspirin use; N (%) 77 (31%) 25 (20%) 52 (44%) <0.001 Oral health and hygiene  Tooth loss; N (%) 127 (52%) 52 (42%) 75 (63%) <0.001  Tooth brushing frequency, ≥ daily; N (%) 233 (95%) 123 (98%) 110 (92%) 0.03  Mouthwash frequency, ≥ daily; N (%) 139 (56%) 69 (55%) 70 (59%) 0.61 Data Availability: 16S rRNA gene sequencing files were uploaded to NCBI Sequence Read Archive (PRJNA785879). DECLARATIONS Ethical approval and consent to participate: This study was approved by the Columbia University Institutional Review Board. All patients provided written informed consent. Consent for publication: Not applicable Availability of data and materials: 16S rRNA gene sequencing files were uploaded to NCBI Sequence Read Archive (PRJNA785879). Competing interests: The authors have none to disclose. ==== Refs References: 1. Abrams JA , Sharaiha RZ , Gonsalves L , Lightdale CJ , Neugut AI . Dating the rise of esophageal adenocarcinoma: analysis of Connecticut Tumor Registry data, 1940–2007. Cancer Epidemiol Biomarkers Prev. Jan 2011;20 (1 ):183–6. doi:10.1158/1055-9965.EPI-10-0802 1055-9965.EPI-10-0802 [pii] 21127287 2. Hur C , Miller M , Kong CY , Trends in esophageal adenocarcinoma incidence and mortality. Cancer. Mar 15 2013;119 (6 ):1149–58. doi:10.1002/cncr.27834 23303625 3. Kong CY , Nattinger KJ , Hayeck TJ , The impact of obesity on the rise in esophageal adenocarcinoma incidence: estimates from a disease simulation model. Cancer Epidemiol Biomarkers Prev. Nov 2011;20 (11 ):2450–6. doi:10.1158/1055-9965.EPI-11-0547 21930957 4. Friedenberg FK , Hanlon A , Vanar V , Trends in gastroesophageal reflux disease as measured by the National Ambulatory Medical Care Survey. Dig Dis Sci. Jul 2010;55 (7 ):1911–7. doi:10.1007/s10620-009-1004-0 19830561 5. Rubinstein MR , Wang X , Liu W , Hao Y , Cai G , Han YW . Fusobacterium nucleatum promotes colorectal carcinogenesis by modulating E-cadherin/beta-catenin signaling via its FadA adhesin. Cell host & microbe. Aug 14 2013;14 (2 ):195–206. doi:10.1016/j.chom.2013.07.012 23954158 6. Tomkovich S , Dejea CM , Winglee K , Human colon mucosal biofilms from healthy or colon cancer hosts are carcinogenic. J Clin Invest. Mar 11 2019;130 :1699–1712. doi:10.1172/JCI124196 7. Kadosh E , Snir-Alkalay I , Venkatachalam A , The gut microbiome switches mutant p53 from tumour-suppressive to oncogenic. Nature. Oct 2020;586 (7827 ):133–138. doi:10.1038/s41586-020-2541-0 32728212 8. Pleguezuelos-Manzano C , Puschhof J , Rosendahl Huber A , Mutational signature in colorectal cancer caused by genotoxic pks(+) E. coli. Nature. Apr 2020;580 (7802 ):269–273. doi:10.1038/s41586-020-2080-8 32106218 9. Gopalakrishnan V , Spencer CN , Nezi L , Gut microbiome modulates response to anti-PD-1 immunotherapy in melanoma patients. Science. Jan 5 2018;359 (6371 ):97–103. doi:10.1126/science.aan4236 29097493 10. Matson V , Fessler J , Bao R , The commensal microbiome is associated with anti-PD-1 efficacy in metastatic melanoma patients. Science. Jan 5 2018;359 (6371 ):104–108. doi:10.1126/science.aao3290 29302014 11. Routy B , Le Chatelier E , Derosa L , Gut microbiome influences efficacy of PD-1-based immunotherapy against epithelial tumors. Science. Jan 5 2018;359 (6371 ):91–97. doi:10.1126/science.aan3706 29097494 12. Elliott DRF , Walker AW , O’Donovan M , Parkhill J , Fitzgerald RC . A non-endoscopic device to sample the oesophageal microbiota: a case-control study. Lancet Gastroenterol Hepatol. Jan 2017;2 (1 ):32–42. doi:10.1016/s2468-1253(16)30086-3 28404012 13. Snider EJ , Compres G , Freedberg DE , Alterations to the Esophageal Microbiome Associated with Progression from Barrett’s Esophagus to Esophageal Adenocarcinoma. Cancer Epidemiol Biomarkers Prev. Aug 29 2019;doi:10.1158/1055-9965.EPI-19-0008 14. Annavajhala MK , May M , Compres G , Relationship of the Esophageal Microbiome and Tissue Gene Expression and Links to the Oral Microbiome: A Randomized Clinical Trial. Clinical and Translational Gastroenterology. 2020;11 (12 ):e00235. doi:10.14309/ctg.0000000000000235 33512805 15. Peters BA , Wu J , Pei Z , Oral Microbiome Composition Reflects Prospective Risk for Esophageal Cancers. Cancer research. Dec 1 2017;77 (23 ):6777–6787. doi:10.1158/0008-5472.CAN-17-1296 29196415 16. Snider EJ , Compres G , Freedberg DE , Barrett’s esophagus is associated with a distinct oral microbiome. Clin Transl Gastroenterol. Feb 20 2018;9 (3 ):135. doi:10.1038/s41424-018-0005-8 29491399 17. Bornigen D , Ren B , Pickard R , Alterations in oral bacterial communities are associated with risk factors for oral and oropharyngeal cancer. Sci Rep. Dec 15 2017;7 (1 ):17686. doi:10.1038/s41598-017-17795-z 29247187 18. Lo CH , Kwon S , Wang L , Periodontal disease, tooth loss, and risk of oesophageal and gastric adenocarcinoma: a prospective study. Gut. Mar 2021;70 (3 ):620–621. doi:10.1136/gutjnl-2020-321949 32690603 19. Fernandes AD , Reid JNS , Macklaim JM , McMurrough TA , Edgell DR , Gloor GB . Unifying the analysis of high-throughput sequencing datasets: characterizing RNA-seq, 16S rRNA gene sequencing and selective growth experiments by compositional data analysis. Microbiome. 2014/05/05 2014;2 (1 ):15. doi:10.1186/2049-2618-2-15 24910773 20. Snider EJ , Compres G , Freedberg DE , Alterations to the Esophageal Microbiome Associated with Progression from Barrett’s Esophagus to Esophageal Adenocarcinoma. Cancer Epidemiol Biomarkers Prev. Oct 2019;28 (10 ):1687–1693. doi:10.1158/1055-9965.Epi-19-0008 31466948 21. Hall-Stoodley L , Costerton JW , Stoodley P . Bacterial biofilms: from the natural environment to infectious diseases. Nat Rev Microbiol. Feb 2004;2 (2 ):95–108. doi:10.1038/nrmicro821 15040259 22. Froeliger EH , Fives-Taylor P . Streptococcus parasanguis fimbria-associated adhesin fap1 is required for biofilm formation. Infect Immun. Apr 2001;69 (4 ):2512–9. doi:10.1128/IAI.69.4.2512-2519.2001 11254614 23. Lee SF , Li YH , Bowden GH . Detachment of Streptococcus mutans biofilm cells by an endogenous enzymatic activity. Infect Immun. Mar 1996;64 (3 ):1035–8. doi:10.1128/iai.64.3.1035-1038.1996 8641755 24. Certo M , Tsai CH , Pucino V , Ho PC , Mauro C . Lactate modulation of immune responses in inflammatory versus tumour microenvironments. Nat Rev Immunol. Mar 2021;21 (3 ):151–161. doi:10.1038/s41577-020-0406-2 32839570 25. Sullivan LB , Gui DY , Hosios AM , Bush LN , Freinkman E , Vander Heiden MG . Supporting Aspartate Biosynthesis Is an Essential Function of Respiration in Proliferating Cells. Cell. Jul 30 2015;162 (3 ):552–63. doi:10.1016/j.cell.2015.07.017 26232225 26. O’Keefe SJ . Diet, microorganisms and their metabolites, and colon cancer. Nature reviews Gastroenterology & hepatology. Dec 2016;13 (12 ):691–706. doi:10.1038/nrgastro.2016.165 27848961 27. Bultman SJ . Molecular pathways: gene-environment interactions regulating dietary fiber induction of proliferation and apoptosis via butyrate for cancer prevention. Clinical cancer research : an official journal of the American Association for Cancer Research. Feb 15 2014;20 (4 ):799–803. doi:10.1158/1078-0432.CCR-13-2483 24270685 28. Smith PM , Howitt MR , Panikov N , The microbial metabolites, short-chain fatty acids, regulate colonic Treg cell homeostasis. Science. Aug 2 2013;341 (6145 ):569–73. doi:10.1126/science.1241165 23828891 29. Papadimitriou N , Gunter MJ , Murphy N , Circulating tryptophan metabolites and risk of colon cancer: Results from case-control and prospective cohort studies. International journal of cancer Journal international du cancer. Jul 1 2021;doi:10.1002/ijc.33725 30. Majka J , Wierdak M , Brzozowska I , Melatonin in Prevention of the Sequence from Reflux Esophagitis to Barrett’s Esophagus and Esophageal Adenocarcinoma: Experimental and Clinical Perspectives. Int J Mol Sci. Jul 13 2018;19 (7 )doi:10.3390/ijms19072033 31. Annavajhala MK , May M , Compres G , Relationship of the Esophageal Microbiome and Tissue Gene Expression and Links to the Oral Microbiome: A Randomized Clinical Trial. Clin Transl Gastroenterol. Dec 2020;11 (12 ):e00235. doi:10.14309/ctg.0000000000000235 33512805 32. Krautkramer KA , Fan J , Backhed F . Gut microbial metabolites as multi-kingdom intermediates. Nat Rev Microbiol. Feb 2021;19 (2 ):77–94. doi:10.1038/s41579-020-0438-4 32968241 33. Vogtmann E , Hua X , Zhou L , Temporal Variability of Oral Microbiota Over 10 Months and the Implications for Future Epidemiologic Studies. Cancer Epidemiol Biomarkers Prev. Feb 23 2018;doi:10.1158/1055-9965.EPI-17-1004 34. Belstrom D , Holmstrup P , Bardow A , Kokaras A , Fiehn NE , Paster BJ . Temporal Stability of the Salivary Microbiota in Oral Health. PLoS One. 2016;11 (1 ):e0147472. doi:10.1371/journal.pone.0147472 26799067 35. Cameron SJ , Huws SA , Hegarty MJ , Smith DP , Mur LA . The human salivary microbiome exhibits temporal stability in bacterial diversity. FEMS Microbiol Ecol. Sep 2015;91 (9 ):fiv091. doi:10.1093/femsec/fiv091 26207044 36. Caporaso JG , Lauber CL , Costello EK , Moving pictures of the human microbiome. Genome biology. 2011;12 (5 ):R50. doi:10.1186/gb-2011-12-5-r50 21624126 37. Cabral DJ , Wurster JI , Flokas ME , The salivary microbiome is consistent between subjects and resistant to impacts of short-term hospitalization. Sci Rep. Sep 8 2017;7 (1 ):11040. doi:10.1038/s41598-017-11427-2 28887570 38. Belstrom D , Holmstrup P , Nielsen CH , Bacterial profiles of saliva in relation to diet, lifestyle factors, and socioeconomic status. J Oral Microbiol. 2014;6doi:10.3402/jom.v6.23609 39. Kato I , Vasquez A , Moyerbrailean G , Nutritional Correlates of Human Oral Microbiome. J Am Coll Nutr. Feb 2017;36 (2 ):88–98. doi:10.1080/07315724.2016.1185386 27797671 40. Nearing JT , DeClercq V , Van Limbergen J , Langille MGI . Assessing the Variation within the Oral Microbiome of Healthy Adults. mSphere. Sep 30 2020;5 (5 )doi:10.1128/mSphere.00451-20 41. Schirmer M , Ijaz UZ , D’Amore R , Hall N , Sloan WT , Quince C . Insight into biases and sequencing errors for amplicon sequencing with the Illumina MiSeq platform. Nucleic Acids Res. Mar 31 2015;43 (6 ):e37. doi:10.1093/nar/gku1341 25586220 42. Edgar RC . SINTAX: a simple non-Bayesian taxonomy classifier for 16S and ITS sequences. bioRxiv. 2016:074161. doi:10.1101/074161 43. F. Escapa I, Huang Y , Chen T , Construction of habitat-specific training sets to achieve species-level assignment in 16S rRNA gene datasets. Microbiome. 2020/05/15 2020;8 (1 ):65. doi:10.1186/s40168-020-00841-w 32414415 44. Wang Q , Garrity GM , Tiedje JM , Cole JR . Naive Bayesian classifier for rapid assignment of rRNA sequences into the new bacterial taxonomy. Applied and environmental microbiology. Aug 2007;73 (16 ):5261–7. doi:10.1128/aem.00062-07 17586664 45. Baldini F , Heinken A , Heirendt L , Magnusdottir S , Fleming RMT , Thiele I . The Microbiome Modeling Toolbox: from microbial interactions to personalized microbial communities. Bioinformatics. Jul 1 2019;35 (13 ):2332–2334. doi:10.1093/bioinformatics/bty941 30462168 46. Heirendt L , Arreckx S , Pfau T , Creation and analysis of biochemical constraint-based models using the COBRA Toolbox v.3.0. Nat Protoc. Mar 2019;14 (3 ):639–702. doi:10.1038/s41596-018-0098-2 30787451 47. Magnusdottir S , Heinken A , Kutt L , Generation of genome-scale metabolic reconstructions for 773 members of the human gut microbiota. Nature biotechnology. Jan 2017;35 (1 ):81–89. doi:10.1038/nbt.3703 48. Baldini F , Hertel J , Sandt E , Parkinson’s disease-associated alterations of the gut microbiome predict disease-relevant changes in metabolic functions. BMC biology. Jun 9 2020;18 (1 ):62. doi:10.1186/s12915-020-00775-7 32517799 49. Klitgord N , Segre D . Environments that induce synthetic microbial ecosystems. PLoS computational biology. Nov 18 2010;6 (11 ):e1001002. doi:10.1371/journal.pcbi.1001002 21124952 50. Orth JD , Thiele I , Palsson BO . What is flux balance analysis? Nature biotechnology. Mar 2010;28 (3 ):245–8. doi:10.1038/nbt.1614 51. Noronha A , Modamio J , Jarosz Y , The Virtual Metabolic Human database: integrating human and gut microbiome metabolism with nutrition and disease. Nucleic Acids Res. Jan 8 2019;47 (D1 ):D614–D624. doi:10.1093/nar/gky992 30371894 52. Shaheen NJ , Falk GW , Iyer PG , Gerson LB . ACG Clinical Guideline: Diagnosis and Management of Barrett’s Esophagus. Am J Gastroenterol. Jan 2016;111 (1 ):30–50; quiz 51. doi:10.1038/ajg.2015.322 26526079 53. Vennalaganti P , Kanakadandi V , Goldblum JR , Discordance Among Pathologists in the United States and Europe in Diagnosis of Low-Grade Dysplasia for Patients With Barrett’s Esophagus. Gastroenterology. Feb 2017;152 (3 ):564–570.e4. doi:10.1053/j.gastro.2016.10.041 27818167 54. Wani S , Falk GW , Post J , Risk factors for progression of low-grade dysplasia in patients with Barrett’s esophagus. Gastroenterology. Oct 2011;141 (4 ):1179–86, 1186.e1. doi:10.1053/j.gastro.2011.06.055 21723218 55. Wani S , Rubenstein JH , Vieth M , Bergman J . Diagnosis and Management of Low-Grade Dysplasia in Barrett’s Esophagus: Expert Review From the Clinical Practice Updates Committee of the American Gastroenterological Association. Gastroenterology. Nov 2016;151 (5 ):822–835. doi:10.1053/j.gastro.2016.09.040 27702561 56. Lozupone C , Lladser ME , Knights D , Stombaugh J , Knight R . UniFrac: an effective distance metric for microbial community comparison. Isme j. Feb 2011;5 (2 ):169–72. doi:10.1038/ismej.2010.133 20827291 57. Ke G , Meng Q , Finley T , LightGBM: A Highly Efficient Gradient Boosting Decision Tree. In: Guyon I , Luxburg UV , Bengio S , et al, eds. 31st Conference on Neural Information Processing Systems (NIPS 2017). 2017. 58. Robin X , Turck N , Hainard A , pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinformatics. Mar 17 2011;12 :77. doi:10.1186/1471-2105-12-77 21414208
PMC010xxxxxx/PMC10327077.txt
==== Front bioRxiv BIORXIV bioRxiv Cold Spring Harbor Laboratory 37425740 10.1101/2023.06.27.546738 preprint 1 Article Implicit reward-based motor learning http://orcid.org/0000-0002-8531-9568 van Mastrigt Nina M. Conceptualization Investigation Software Visualization Methodology Writing – original draft Writing – review & editing Vrije Universiteit Amsterdam, Department of Human Movement Sciences, Amsterdam, The Netherlands http://orcid.org/0000-0002-3992-9023 Tsay Jonathan S. Conceptualization Methodology Writing – review & editing UC Berkeley, CognAc lab, Berkeley, California, United States http://orcid.org/0000-0002-0131-850X Wang Tianhe Conceptualization Software Methodology Writing – review & editing UC Berkeley, CognAc lab, Berkeley, California, United States http://orcid.org/0000-0002-6170-1041 Avraham Guy Conceptualization Methodology Writing – review & editing UC Berkeley, CognAc lab, Berkeley, California, United States http://orcid.org/0000-0001-5740-1874 Abram Sabrina J. Methodology Writing – review & editing UC Berkeley, CognAc lab, Berkeley, California, United States http://orcid.org/0000-0002-4190-7827 van der Kooij Katinka Funding acquisition Supervision Methodology Writing – review & editing Vrije Universiteit Amsterdam, Department of Human Movement Sciences, Amsterdam, The Netherlands http://orcid.org/0000-0002-3794-0579 Smeets Jeroen B. J. Funding acquisition Supervision Methodology Writing – review & editing Vrije Universiteit Amsterdam, Department of Human Movement Sciences, Amsterdam, The Netherlands http://orcid.org/0000-0003-4728-5130 Ivry Richard B. Conceptualization Supervision Methodology Writing – review & editing UC Berkeley, CognAc lab, Berkeley, California, United States [email protected] 28 6 2023 2023.06.27.546738https://creativecommons.org/licenses/by-nc-nd/4.0/ This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which allows reusers to copy and distribute the material in any medium or format in unadapted form only, for noncommercial purposes only, and only so long as attribution is given to the creator. nihpp-2023.06.27.546738.pdf Binary feedback, providing information solely about task success or failure, can be sufficient to drive motor learning. While binary feedback can induce explicit adjustments in movement strategy, it remains unclear if this type of feedback also induce implicit learning. We examined this question in a center-out reaching task by gradually moving an invisible reward zone away from a visual target to a final rotation of 7.5° or 25° in a between-group design. Participants received binary feedback, indicating if the movement intersected the reward zone. By the end of the training, both groups modified their reach angle by about 95% of the rotation. We quantified implicit learning by measuring performance in a subsequent no-feedback aftereffect phase, in which participants were told to forgo any adopted movement strategies and reach directly to the visual target. The results showed a small, but robust (2–3°) aftereffect in both groups, highlighting that binary feedback elicits implicit learning. Notably, for both groups, reaches to two flanking generalization targets were biased in the same direction as the aftereffect. This pattern is at odds with the hypothesis that implicit learning is a form of use-dependent learning. Rather, the results suggest that binary feedback can be sufficient to recalibrate a sensorimotor map. Reward-based motor learning reinforcement learning implicit learning visuomotor rotation use-dependent learning reward ==== Body pmcIntroduction The execution of accurate movements relies on sensory feedback. Variants of sensorimotor adaptation experiments have been used to study the role of different forms of feedback on motor learning. In a typical visuomotor adaptation experiment, participants perform target-directed center-out reaching movements with feedback of the unseen hand limited to a visual cursor. To study learning, the position of the cursor is altered, resulting in a sensory prediction error, defined by the difference between the predicted and actual cursor position (Izawa & Shadmehr, 2011; Kim et al., 2018; Morehead et al., 2017; Shadmehr et al., 2010; Synofzik et al., 2008; Tseng et al., 2007). This directional error can drive different forms of learning. It can produce recalibration of a so-called sensorimotor map such that a subsequent movement to that target will be shifted in the direction opposite to the perturbed feedback, a process known as sensorimotor adaptation (Kim et al., 2021; Krakauer, 2009; Krakauer et al., 2019). It can also elicit explicit strategies to reduce the error; for example, the participant might aim away from the target (Bond & Taylor, 2015; Taylor et al., 2014). Feedback can also be limited to binary information conveying success or failure. In reaching tasks, success can be defined by the hand intersecting an invisible reward zone. To elicit learning, the reward zone is displaced from the target. This might be done in an abrupt manner. For example, success suddenly requires reaches into a reward zone that is centered 30° from the target. Alternatively, the reward zone can be shifted in a gradual manner, for example in 5° increments eventually reaching a maximum displacement of 30°. Following the introduction of the perturbation, success requires a movement that is off-target. While participants can find it challenging to learn when the shift is large or introduced abruptly (Brudner et al., 2016; Holland et al., 2018), many studies have shown that binary feedback is sufficient to produce learning when the shift is introduced in a gradual manner (Cashaback et al., 2019; Izawa & Shadmehr, 2011; Therrien et al., 2016, 2018; van der Kooij et al., 2021; van der Kooij & Smeets, 2018). While sensory prediction errors and binary reward feedback can produce similar adjustments in behavior, there are marked differences in the representational changes associated with these two forms of learning (Morehead & Orban de Xivry, 2021; Therrien & Wong, 2022). For example, adaptation from sensory prediction errors is greatly attenuated when any delay is introduced between the movement and feedback, whereas adaptation from binary reward feedback is minimally impacted by delays up to a few seconds (Brudner et al., 2016; Schween & Hegele, 2017). In addition, the acquired behavior is more persistent following reward-based feedback compared to error-based feedback (Bao & Lei, 2022; Galea et al., 2015; Shmuelof et al., 2012; Therrien et al., 2016). Learning processes can also be evaluated in terms of the degree to which they result in implicit and explicit changes in behaviour. A large body of literature has shown that adaptation from sensory prediction errors occurs in an automatic and implicit manner (Kim et al., 2018; Mazzoni & Krakauer, 2006; Morehead et al., 2017). Adaptation can also result from re-aiming, which is explicit and under volitional control. To date, less is known about implicit changes in behaviour in response to binary feedback. Following the convention in the adaptation literature, a strong probe of implicit learning is to focus on behavioral changes that persist when the feedback is eliminated and participants are reminded to reach directly to the target (Maresch, Mudrik, et al., 2021; Maresch, Werner, et al., 2021). When probed in this manner following reward feedback, a small aftereffect is observed. For example, following a shift of the reward zone of 25°, the average heading angle at the start of the aftereffect phase was around 5° (Holland et al., 2018). This suggests that reward-based learning is largely the result of a volitional change in strategy. Consistent with this hypothesis, disrupting explicit processes by introducing a secondary task attenuates learning from binary feedback (Codol et al., 2018; Holland et al., 2018). Nonetheless, the fact that there is an aftereffect, even if small, indicates binary feedback can induce implicit learning (Codol et al., 2018; Holland et al., 2018, 2019). What might be the source of this implicit component? We can consider two, non-mutually exclusive hypotheses. The first hypothesis centers on the idea that the behavioral change resulting from binary feedback includes a contribution from implicit, use-dependent learning. As implied by the name, use-dependent learning refers to a movement bias towards frequently repeated movements (Diedrichsen et al., 2010; Huang et al., 2011; Marinovic et al., 2017; Mawase et al., 2017; Tsay et al., 2022; Verstynen & Sabes, 2011). Tracking the reward zone will result in movements that are shifted in a consistent direction relative to the visual target. In an aftereffect phase, a use-dependent bias would produce a residual implicit bias in this direction. Interestingly, the 3–4° aftereffect following training with binary feedback is similar in magnitude to that observed in studies of use-dependent learning that exclude errors in action selection (Tsay et al., 2022). A second hypothesis is that binary feedback induces implicit recalibration of a sensorimotor map. Mechanistically, implicit recalibration could occur because the binary feedback alters the contingency between action plans and their associated movements. Feedback that indicates task success would strengthen the association between the goal to reach to a visual target and movements linked to that target, even if these are towards a reward zone that is displaced relative to the visual target. Feedback that indicates task failure would weaken this association. Compared to error-based learning, recalibration from reward feedback would appear to be much more limited given that the aftereffect following binary feedback is much smaller than that following cursor feedback for similar perturbation sizes (Bond & Taylor, 2015; Codol et al., 2018; Holland et al., 2018; Leow et al., 2018; Taylor & Ivry, 2014). Here, we report the results of an experiment designed to assess these use-dependent learning and implicit recalibration hypotheses. Providing binary feedback only, we examined how participants learned to respond to either a small (7.5°) or large perturbation (25°) of the reward zone. For both groups, the perturbation was introduced in a gradual manner. Assuming that participants in the Small perturbation condition will have little awareness of the perturbation, this condition provides a strong test of the role of implicit processes in reward-based learning. In contrast, we assumed that participants in the Large condition would eventually adopt a strategy. To assess implicit learning in both conditions, we measured reaching in an aftereffect phase in which all feedback was eliminated and participants were instructed to reach directly to the target. The implicit recalibration and use-dependent hypothesis both predict aftereffects in the Small and Large conditions. To compare the two hypotheses, we included two probe targets in the aftereffect phase, displaced by 15° from the training target location (Fig 1c). The inclusion of the probe targets allowed us to ask how implicit learning, if observed, generalized. By the implicit recalibration hypothesis, we would expect that reaches to the probe targets would be biased to a similar extent and in the same direction as reaches to the trained target. By the use-dependent hypothesis, we should observe that reaches to the probe targets would be attracted towards the trained movement. For the Small perturbation condition, the biases to the two probe targets should be in the opposite direction since the trained movement falls between the two probe locations. The predictions are less clear for the Large perturbation condition and will depend on the magnitude of learning. Biases to the two probe targets will be in the same direction if participants fully track the 25° shift of the reward zone. However, if the trained movement falls short of the reward zone, the biases will become less symmetric and even have opposite signs once the trained movement is less than 15°. Methods Participants 68 right-handed young adults were recruited from the research participant pool of the Department of Psychology at the University of California, Berkeley. 28 (22 females, 6 males; reported age: mean 20.5, SD 2.3 years) were assigned to the “Small” perturbation group and 40 (27 females, 13 males; reported age: mean 21.5, SD 5.7 years) were assigned to the “Large” perturbation group. Participants received either course credit or financial compensation for their participation, along with a $5 completion bonus paid to all participants. Based on self-reports, participants had normal or corrected-to-normal vision and hearing. The protocol was approved by the institutional review board at UC Berkeley. Of the original 68 participants, 20 were excluded from all analyses. 16 of these (8 per group) were excluded based on their responses to a post-experiment questionnaire (see “Experimental design”) that indicated they failed to follow the instructions. Four other participants in the Large group were excluded for idiosyncratic reasons: One fell asleep during the task, one reported, after the experiment having performed in a similar experiment, one did not use the apparatus correctly, and one experienced an equipment failure. Thus, the analyses reported below are based on data obtained from 20 participants (16 females; 10 for credit; reported age: mean 20.9, SD 2.4 years) in the Small perturbation group and of 28 participants (16 females; 16 for credit; reported age: mean 21.8, SD 6.0 years) in the Large perturbation group. Experimental set up The participant sat in front of a table in a small, darkened room. A horizontally-oriented computer screen (24”, ASUS, Taipei, Taiwan) constituted the upper surface of the table, with a 17” digitizing tablet (Wacom Co., Kazo, Japan) positioned 27 cm below the screen (Fig 1a). Stimuli were presented on the computer (refresh rate = 60 Hz) and the participant’s movements along the digitizing tablet were recorded from a digitizing pen (sampling rate = 200 Hz) that was embedded in a custom-made paddle, ensuring the pen maintained a vertical position. Vision of the hand was obscured by the screen. A computer (Dell OptiPlex 7040, Round Rock, Texas) with a Windows 7 operating system (Microsoft Co., Redmond, Washington) was used to run the custom experimental software in Matlab (The MathWorks, Natick, Massachusetts), using Psychtoolbox extensions (Brainard, 1997; Kleiner et al., 2007). Trial structure Each trial started with the appearance of a white “start” circle (radius = 0.42 cm), presented near the center of the screen. The participant was required to move the paddle to position the digitizing pen within the start circle. To guide the participant to the start location, a white ring was presented, with the radius of the ring indicating the distance from the pen to the start position. Movement towards the start position reduced the size of the ring. When the pen was within 0.84 cm of the start circle, the ring was replaced by a white circle (radius = 0.17 cm) that indicated the position of the pen, allowing the participant to move the pen into the start circle. When the paddle had been in the start circle for 300 ms, a visual target (circle with radius = 0.28 cm) appeared 7 cm from the start circle at either 45°, 60° or 75° (Figure 1b, c). The participant was instructed to move in rapid manner, attempting to slice through the target. Auditory feedback was presented when the movement amplitude exceeded 7 cm. On trials with performance feedback (see below), a pleasant “bing” indicated that the movement was successful (e.g., passed through the target when feedback was veridical) and an aversive “buzz” indicated that the movement was unsuccessful. On no feedback trials (in the baseline and aftereffect phases), a “knock” sound was played. This indicated that the required reach amplitude had been exceeded but it did not provide feedback on whether the movement was within the reward zone or not. To make participants move at similar, and relatively rapid speeds, 800 ms after the performance feedback an auditory message “Too slow” was played if the movement time was longer than 600 ms. This was the case on 3% of the trials. We did not exclude these slow trials from the analyses. The feedback ring appeared directly after the feedback was given. Note that by using a ring during the return movement, the participant received feedback indicating only the radial position of the hand. Angular position was only provided when the hand was very close to the start position: then, the ring turned into a cursor. This method was used so that any effect of adaptation to the rotated feedback (see below) would be minimally visible to the participant during the return movement. Experimental design The experimenter instructed the participant that the purpose of the experiment was to study how well people can control arm movements in the absence of visual feedback. The participant was told that they would control an invisible cursor, and they were asked to make reaching movements that would make the invisible cursor intersect a visual target (Fig 1a). The experimenter described how a “bing” and “buzz” would indicate if the reach had intersected or missed the target, respectively. The experimenter then completed 10 demonstration trials to demonstrate how the hand controlled the cursor movement. The target was always presented at the 60° location and during these trials, the auditory feedback was accompanied by veridical cursor feedback. After the ten demonstration trials, the participant was told that the cursor would no longer be visible during the reach, but that auditory feedback would be presented on most trials to indicate task outcome. However, on some trials, the participant would hear a “knock” sound, and this sound was uninformative concerning task outcome. To motivate the participant for all trials, the participant was informed that the computer kept track of all successful reaches and that a score in the top-third of high scores across participants would result in a $5 bonus (which was actually paid to all participants). The main experiment consisted of three phases: baseline, training, and aftereffect, with the experimenter providing instructions at the beginning of each phase. The baseline phase was composed of 150 trials with feedback limited to the uninformative “knock” sound. The target appeared at each of the three possible locations on 50 of the baseline trials, with the order determined randomly. These trials allowed the participant to become familiar with the apparatus, learn to move at the appropriate speed, and provided a measure of natural biases for each of the three target locations (Kuling et al., 2019; van der Kooij et al., 2013). The training phase was composed of 700 trials, with the target always appearing at the middle location (60°) and auditory feedback provided to indicate target hits or misses. For the first 100 trials, the reward zone was centered around the participant’s individual bias while reaching to the trained target and extended 2° in both directions; if, for example, the individual’s mean reach to the central target was rotated by 3° in the clockwise direction (at 57°), the initial reward zone spanned from 55° – 59°. Unbeknownst to the participant, the reward zone was gradually shifted over the next 500 trials. This was achieved by rotating the reward zone by 1.5° every 100 trials for the Small perturbation group and by 2.5° every 50 trials for the Large perturbation group. The rotation was either clockwise or counterclockwise, counterbalanced between participants. For the last 100 trials of the training phase, the reward zone remained fixed, displaced by 7.5° or 25° from its starting position for the Small and Large perturbation groups, respectively. A two-minute break was provided halfway through the 700-trial training phase. Note that we expected that the participants in the Small group would likely remain unaware of the perturbation since the shift was introduced gradually and the total displacement fell within 1–2 standard deviations of normal reach variability (Gaffin-Cahn et al., 2019). In contrast, we expected that participants in the Large group would likely become aware of the perturbation at some point during the training phase as the discrepancy between the visual target and hand movement would likely fall outside the individuals’ normal reach variability. Following the training phase, the participant completed an aftereffect phase of 150 trials. Prior to the start of the phase, the participant was instructed that the feedback might have been altered over the course of the training phase. To equally inform and instruct participants with different levels of awareness of the perturbation, the participant was informed that there were two groups of participants, an aligned group and a misaligned group. For the aligned group, the invisible cursor had always moved exactly with the position of the hand; for the misaligned group, the invisible cursor was slightly displaced from the position of the hand. To ensure that the participant understood the difference, they were asked to explain the difference between the two groups in their own words. If the explanation failed to capture the difference, the experimenter repeated the explanation. The experimenter then stated that for the final phase of the experiment, the cursor would be aligned with the hand for everyone, irrespective of initial group assignment and thus, they should reach straight to the target to make the cursor hit the target. As in the baseline phase, reaches during this phase were performed with only the uninformative feedback, with the phase composed of 50 reaches to each of the three targets. Participants were again instructed that accuracy would be recorded during this phase to determine a final performance bonus. At the end of the experiment, the participant completed a questionnaire consisting of five questions (Online resource 1). Question 1 asked if they believed the feedback had been veridical or perturbed and Question 2 asked for their confidence concerning their response to Question 1, using a 7-point rating scale (1= no confidence, 7= very confident). For Questions 3 and 4, the participants were asked to report (forced choice) where they had aimed during the training and aftereffect phases, respectively. Note that Question 4 was used to determine if the participant had followed the instructions. Those who answered that they had aimed to the left or right of the target during the aftereffect phase were excluded from all of the analyses (n=16). For Question 5, the participant was informed that they had been in the Misaligned feedback group and were asked to indicate (forced choice) if the feedback had been perturbed: to the left or to the right. As the answers to this question were below chance level in the Small perturbation group, for the Large perturbation group, the illustrations for the two choices were slightly changed (Online resource 1) to match the hand movements of the participants better. The total duration of the experiment was approximately one hour. Data analysis Reach angle was determined by the line from the start position to where the digitizing pen crossed the 7 cm radius around the start position. The mean reach angle during the baseline trials was used to characterize individual biases for each of the three target locations separately (50 reaches/target). All analyses were based on the reach angles during the training and aftereffect phases, with these angles expressed relative to that participant’s baseline bias for the corresponding target. Positive values correspond to reach angles shifted in the direction of the rotated reward zone. We calculated final learning as the mean reach angle in the last 100 trials of the training phase. To test for implicit learning, we calculated the mean reach angle to the training target in the aftereffect phase. For generalization, we calculated the mean reach angle for each of the two probe targets in the aftereffect phase. Statistics A preliminary analysis indicated that the final learning and aftereffect scores were not normally distributed (see Fig 2). Therefore, we employed non-parametric tests in the statistical evaluation of the results. To test whether the final learning and aftereffect were larger than zero, we performed a one-tailed Wilcoxon signed-rank test on these variables for each group (Small and Large). To test whether implicit learning was different for the two perturbation sizes, we performed a two-tailed Wilcoxon rank-sum test on the aftereffect scores for the two groups. For the generalization data, we defined the percentage generalization as the mean of the two probe target biases, divided by the aftereffect at the training target. We used a one-tailed Wilcoxon signed-rank test to test whether the percentage generalization values were significantly larger than zero. To evaluate the form of generalization, we defined generalization asymmetry as the difference between the reaching bias to the probe target opposite the reward zone and the probe target in the direction of the reward zone. The use-dependent learning hypothesis predicts that this value will be positive for the Small perturbation condition. The implicit recalibration hypothesis predicts that this value will be zero (if generalization is exactly the same for both targets, but see (Nikooyan & Ahmed, 2015)). To evaluate the two hypotheses, we used a Wilcoxon signed-rank test to test whether the generalization asymmetry values were significantly greater than zero. No statistics were performed on the questionnaire data. Results Learning To evaluate how people modified their movements given the gradual change in the reward zone, we analyzed the reach angle at the end of learning in both the Small (max shift of 7.5°) and Large (max shift of 25°) perturbation groups. Both groups learned to compensate for the feedback perturbation (Fig 2a,b). Participants in the Small perturbation group showed a median final learning of 7.1° (IQR [5.8°, 7.8°], p < 0.01, z = 3.9, Ws = 210, r = 0.20) (Fig 2c, horizontal axis). Participants in the Large perturbation group showed a median final learning of 23.7° (IQR [9.6°, 25.2°], p < 0.01, z = 4.2, Ws = 390, r = 0.16 (Fig 2d, horizontal axis). For both groups, this corresponds to a mean percentual change of 95% of the perturbation size (Small: IQR = 77% – 104%; Large: IQR = 38% – 101%). As can be seen in Fig 2c,d (horizontal axes), learning was more variable in the Large perturbation group than in the Small perturbation group. For the latter, all of the participants changed their reaches in the direction of the perturbation and 86% ended up with a mean heading angle over the final 100 trials that was within the final reward zone. In contrast, only 70% of the participants in the Large perturbation group reached the final reward zone (Online resource 3). Four participants in this group exhibited a mean hand angle over the final 100 trials that was in the opposite direction of the reward zone. Aftereffect The central aim of our study was to examine whether binary feedback regarding success or failure induces implicit motor learning. To this end, we focused on the reach direction during the aftereffect phase when the feedback was removed and participants were instructed to reach directly to the target. Both groups showed a significant aftereffect (Fig 2c, d vertical axes). Participants in the Small perturbation group had a median aftereffect of 3.4° (IQR [2.2°, 7.8°]; p < 0.01, z=3.90, Ws = 210, r = 0.20). Participants in the Large perturbation group had a median aftereffect of 2.2° (IQR [−3.1°, 10.7°], p < 0.05, z = 2.02, Ws = 292, r = 0.07). Importantly, we found no difference between the magnitude of the aftereffect for the Small and Large perturbation groups (p = 0.24, z=−1.2, U = 434). In summary, the aftereffect data indicate that there is an implicit component to learning that occurs in response to binary feedback. The magnitude of the aftereffect in both the Small and Large perturbation groups was of similar size and quite small. Generalization We included reaches to two probe targets in the aftereffect phase, asking how learning generalized to regions of the workspace neighboring the training target. Both groups exhibited generalization in that the reaches to the probe locations were significantly shifted from the baseline phase. In terms of the direction of the shift, the mean values were all positive, meaning that the change in reach direction for the probe targets was in the same direction as the change in reach direction to the training target (Fig 3a). Participants in the Small perturbation group had a median reaching bias of 3.5° to the probe target in the direction of the learning and of 3.6° to the other probe target (Online resource 4). The corresponding biases were 1.6° and 0.7° for the Large perturbation group. The generalization data are consistent with the implicit recalibration hypothesis. For each participant, generalization (Fig 3c) was calculated as the mean of the two probe target biases as a percentage of the aftereffect. We found significant generalization of 83% of the aftereffect in both the Small (IQR [53.3%, 100.5%], p < 0.01, z = 3.7, Ws = 205, r = 0.19) and Large (IQR [54.3%,100.6%], p < 0.01, z = 4.0, Ws = 379, r = 0.14) perturbation group. The generalization data are not consistent with the use-dependent learning hypothesis. The use-dependent learning hypothesis had predicted biases in opposite directions for the two probes in the Small perturbation group since the trained movement was between the two probe targets. This would predict positive generalization asymmetry scores. In the Large perturbation group, predictions were less clear since they depend on the location of the trained movement relative to the probe targets. For participants who fully followed the reward zone, the trained movement was beyond both probe targets; for others, the trained movement was either between or beyond both probe targets. For both groups, the analyses showed that the asymmetry scores were not significantly larger than zero (Small: median = −1.0°, IQR [−3.0°, 3.5°], p = 0.55, z = −0.06, Ws=89; Large: median = 0.0°, IQR [−2.7°, 2.0°], IQR [53.3%, 100.5%], p = 0.96, z = −0.05, Ws=201). These null results, coupled with the fact that we did observe significant generalization, are consistent with the implicit recalibration hypothesis. Awareness of the feedback perturbation As expected, participants in the Small perturbation group were generally unaware that the reward zone had shifted over the course of the experiment. When asked to judge if they had been in the group with veridical feedback or shifted feedback, 60% reported that the feedback was not perturbed with an average confidence of 3.3 on a 7-point scale (Online resource 1, 3). When forced to choose between saying if they aimed left, right, or straight to the target during the training phase, 50% reported having aimed straight to the target and 50% reported aiming away from the target. However, of the latter, half reported aiming in the direction of the shifted reward zone and the other half reported aiming in the opposite direction. These survey data, in combination with the fact that all participants in the Small perturbation group showed a shift in reaching in the direction of the perturbation, provide compelling evidence that there was little if any awareness of the experimental manipulation nor use of a compensatory strategy. A very different picture emerged from the survey data for the Large perturbation group. The majority (82%) reported that the feedback was perturbed with an average confidence of 4.8 on the 7-point scale. When asked whether they aimed left of, right of or straight to the target during the training phase, 75% of the participants reported having aimed off target in the direction of the shifted reward zone, whereas 21% reported having aimed straight to the target. In summary, the survey data indicate that the participants in the Large perturbation group were aware of the experimental manipulation and adopted a re-aiming strategy to compensate for the shift in the reward zone. There was no clear relation between the questionnaire reports and aftereffects (Online resource 2). Discussion In the present study, we examined whether binary feedback can induce implicit learning in response to shifts in a hidden reward zone. Based on previous work (Codol et al., 2018; Holland et al., 2018, 2019), we expected that the learning would include an implicit component. Participants performed a center-out reaching task and were only provided binary feedback to indicate if the movement ended in a reward zone that gradually shifted to be centered 7.5° or 25° from the visual target, with the expectation that awareness of the perturbation would be minimal in the former and that the latter would entail some explicit component. During training, participants in both groups learned to compensate for the rotated feedback. When the feedback was removed after training and participants were instructed to move to the target, their reaches were biased in the direction of learning, with an aftereffect of 2–3° in both groups. To test generalization, the no-feedback phase also included reaches to probe targets that flanked the training target. On these probe target trials, participants exhibited a shift in reach angle that was in the same direction as the shift associated with the training target. These results suggest that binary feedback can induce implicit reward-based motor learning and that this learning reflects implicit recalibration of a sensorimotor map. Small and saturated implicit learning in response to binary feedback Our study employed multiple approaches to prevent explicit processes from contaminating our assessment of implicit learning. First, we focused on the aftereffect in a phase without feedback and in which we provided explicit instructions to stop using any strategy that might have been used during the training. Second, we introduced the perturbation in a gradual manner, and most importantly, included a small perturbation group in which the displacement per step was within 1.5 standard deviations of baseline reach variability (Online resource 4) (Gaffin-Cahn et al., 2019). Thus, for this group, it is likely that behavioral changes during the training phase occurred implicitly. Third, we used questionnaires to directly assess awareness of the perturbation. The responses to the survey confirmed that, during the perturbation phase, awareness and strategy use were minimal in the Small perturbation group but high in the Large perturbation group. We observed a small, but consistent aftereffect of around 2–3° in both the Small and Large perturbation groups. The magnitude of this effect for the latter group is smaller than that previously reported in other studies using a perturbation of comparable size; for example, in Holland et al. (2018, see also, 2019), the aftereffects in response to a perturbation of 25° were around 5°. However, during their no-feedback aftereffect phase, Holland et al. first instructed participants to keep reaching as they had done during training. After this phase, they were instructed to stop using a strategy. This protocol may have contaminated the final aftereffect measure by adding extra strategy trials and the challenge to switch between tasks. The inclusion of the Small perturbation group not only provided a condition in which awareness should be minimized during the training phase, but also allowed us to directly compare how perturbation size impacted the magnitude of implicit learning from binary feedback. Interestingly, the size of the aftereffect did not scale with perturbation size. Indeed, in terms of mean value, the size was larger in the Small condition (3.4°) compared to the Large condition (2.2°), although this difference was not significant. While future testing is required to sample a broader range of perturbation sizes, the present results suggest that the magnitude of implicit learning from binary feedback is relatively small and saturates, at least for perturbations larger than 7.5°. Mechanisms of implicit learning in response to binary feedback In the following section, we will consider the mechanisms underlying implicit learning in response to binary feedback. Similar to what has been reported in studies of error-based learning (Bond & Taylor, 2015; Morehead et al., 2017) and use-dependent learning (Tsay et al., 2022), implicit learning in response to binary feedback seems to saturate. However, there are notable differences between these three implicit forms of learning. While the magnitude of implicit use-dependent biases is similar to the magnitude of the aftereffect observed in the present study, the generalization pattern did not show any evidence of attraction towards the training location. As such, the current results fail to support the idea that implicit learning from binary feedback is a manifestation of use-dependent learning. On the other hand, while the generalization pattern is similar for binary and cursor feedback, the magnitude of the binary feedback effect is much smaller than that observed in response to cursor feedback (Bond & Taylor, 2015; Morehead et al., 2017). This size discrepancy makes it unlikely that binary feedback operates on similar mechanisms in inducing implicit recalibration of the sensorimotor map. How, then, does binary feedback result in implicit learning? We outline three implicit recalibration hypotheses. First, implicit learning in response to binary feedback could be the result of motor recalibration, retuning the mapping between a visual target location and its associated movement. The contingency between action and reward outcome will lead to that action being associated with a new movement plan (Avraham et al., 2022). This hypothesis predicts that there is no sensory recalibration: training would not influence reports of where the visual target is perceived and perceived locations of the hand so that they are similar before and after training. Second, implicit learning could be the result of visual recalibration of the target, i.e., a bias in the perceived location of the visual target. This hypothesis predicts visual sensory remapping: for example, if asked to report the perceived target location by reaching with the non-trained hand, we would observe a bias towards the reward zone (Simani et al., 2007). Third, implicit learning could be the result of proprioceptive recalibration, i.e., a bias in perceived hand position. This hypothesis predicts proprioceptive sensory remapping. For example, static reports of perceived hand position would be biased in the opposite direction of the perturbation (Tsay & Ivry, 2022). Future studies employing fine-grained psychophysical tests can evaluate the merits of these different hypotheses, asking if implicit learning in response to binary feedback originates from implicit recalibration of a sensory and/or motor mapping, and how this evolves over the course of learning. Conclusion Our data add to a growing body of evidence indicating that motor learning encompasses multiple processes where both explicit and implicit processes drive behavioral changes (Kim et al., 2021; Morehead & Orban de Xivry, 2021; Therrien & Wong, 2022). The results provide compelling evidence of implicit learning in response to binary feedback and rule out that this effect is a form of use-dependent learning. Less clear is whether this implicit learning entails the same mechanisms, albeit in attenuated form, as occur during learning from sensory prediction errors, or reflects the operation of distinct, reward-based mechanisms. Supplementary Material Supplement 1 Supplement 2 Supplement 3 Supplement 4 Acknowledgments The research was funded by the Nederlandse Organisatie voor Wetenschappelijk Onderzoek, Toegepaste en Technische Wetenschappen Open Technologie Programma (NWO-TTW OTP grant 15989), and by the United States National Institutes of Health (NIH grants R35NS116883 and NS105839). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Fig. 1 Schematic outline of key hypotheses on implicit reward-based motor learning. a. Schematic of a participant in the experimental apparatus. b. Training phase. Participants made center-out reaching movements from a white starting circle to a black training target. A pleasant auditory “ding” was provided when the movement passed within the reward zone (green arch); otherwise, an unpleasant “buzz” was played. c. No-feedback phase. Participants were instructed to reach directly to a visual target. The target appeared at the training location or one of two probe locations (+−15°). Participants were instructed to forgo any strategy adopted during the training phase. Left panel shows implicit learning as measured by an aftereffect, defined as a change in hand angle for reaches to the training target from pre-training (translucent hand) to post-training (solid hand). Middle panel shows probe target reaching predictions for the implicit recalibration hypothesis. Reaches will be biased in same direction for both probe targets independent of size of the perturbation. Right panel shows probe target reaching predictions for the use-dependent learning hypothesis. For the Small perturbation condition, the biases will be in opposite directions since the reaches during training fall between the two probe locations. For the Large perturbation condition, the direction of the bias for the probe target nearest the reward zone will depend on the degree of learning (example here is for a participant who shows full learning) Fig. 2 The effect of binary reward feedback on reaching. a, b. The training phase. Gradually changing the rewarded hand angles (green zone) leads to learning, as indicated by the change in reach angle. We plot the median (solid thick lines) over all participants with the interquartile range (opaque lines) for the Small perturbation group (a) and Large perturbation group (b). Note that the vertical axes are scaled to the perturbation size. For display purposes, the curves are smoothed with a running average with a window size of 10 trials. c, d. Aftereffect as a function of the final learning for both groups. Each grey dot corresponds to a participant and error bars indicate the interquartile range per group. The green lines indicate perturbation size Fig. 3 Aftereffects and generalization of learning. Bars and error bars indicate medians and interquartile ranges. a. Reaching biases for the training target (black) and two probe targets (see Fig 1c). Thin lines indicate data from individual participants. b. Asymmetry in reaching biases to probe targets. Dots indicate the individual participants in the two groups. c. Generalization quantified as a percentage of the aftereffect. The participants with large positive and negative values are the ones with a small aftereffect Competing interests Richard B. Ivry is a co-founder with equity in Magnetic Tides, Inc. Data and code availability Data and code can be accessed on the Open Science Framework (https://osf.io/x7hp9/). ==== Refs References Avraham G. , Taylor J. A. , Breska A. , Ivry R. B. , & McDougle S. D. (2022). Contextual effects in sensorimotor adaptation adhere to associative learning rules. ELife, 11 , 2020.09.14.297143. 10.7554/elife.75801 Bao S. , & Lei Y. (2022). Memory decay and generalization following distinct motor learning mechanisms. J Neurophysiol, 128 , 1534–1545. 10.1152/jn.00105.2022 36321731 Bond K. M. , & Taylor J. A. (2015). Flexible explicit but rigid implicit learning in a visuomotor adaptation task. Journal of Neurophysiology, 113 (10 ), 3836–3849. 10.1152/jn.00009.2015 25855690 Brainard D. H. (1997). The Psychophyics Toolbox. Spatial Vision, 10 (4 ), 433–436.9176952 Brudner S. N. , Kethidi N. , Graeupner D. , Ivry R. B. , & Taylor J. A. (2016). Delayed feedback during sensorimotor learning selectively disrupts adaptation but not strategy use. Journal of Neurophysiology, 115 (3 ), 1499–1511. 10.1152/jn.00066.2015 26792878 Cashaback J. G. A. , Lao C. K. , Palidis D. J. , Coltman S. K. , McGregor H. R. , & Gribble P. L. (2019). The gradient of the reinforcement landscape influences sensorimotor learning. PLoS Computational Biology, 15 (3 ), e1006839. 10.1371/journal.pcbi.1006839 30830902 Codol O. , Holland P. J. , & Galea J. M. (2018). The relationship between reinforcement and explicit control during visuomotor adaptation. Scientific Reports, 8 (9121 ), 1–11. 10.1038/s41598-018-27378-1 29311619 Diedrichsen J. , White O. , Newman D. , & Lally N. (2010). Use-dependent and error-based learning of motor behaviors. Journal of Neuroscience, 30 (15 ), 5159–5166. 10.1523/JNEUROSCI.5406-09.2010 20392938 Gaffin-Cahn E. , Hudson T. E. , & Landy M. S. (2019). Did I do that? Detecting a perturbation to visual feedback in a reaching task. Journal of Vision, 19 (1 ), 5. 10.1167/19.1.5 Galea J. M. , Mallia E. , Rothwell J. , & Diedrichsen J. (2015). The effects of reward and punishment on motor skill learning. Nature Neuroscience, 18 (4 ), 597–604. 10.1016/j.cobeha.2017.11.011 25706473 Holland P. , Codol O. , & Galea J. M. (2018). Contribution of explicit processes to reinforcement-based motor learning. Journal of Neurophysiology, 119 (6 ), 2241–2255. 10.1152/jn.00901.2017 29537918 Holland P. , Codol O. , Oxley E. , Taylor M. , Hamshere E. , Joseph S. , Huffer L. , & Galea J. (2019). Domain-specific working memory, but not dopamine-related genetic variability, shapes reward-based motor learning. The Journal Of Neuroscience, 49 (47 ), 9383–9396. 10.1523/JNEUROSCI.0583-19.2019 Huang V. S. , Haith A. , Mazzoni P. , & Krakauer J. W. (2011). Rethinking Motor Learning and Savings in Adaptation Paradigms: Model-Free Memory for Successful Actions Combines with Internal Models. Neuron, 70 (4 ), 787–801. 10.1016/j.neuron.2011.04.012 21609832 Izawa J. , & Shadmehr R. (2011). Learning from sensory and reward prediction errors during motor adaptation. PLoS Computational Biology, 7 (3 ), e1002012. 10.1371/journal.pcbi.1002012 21423711 Kim H. E. , Avraham G. , & Ivry R. B. (2021). The Psychology of Reaching: Action Selection, Movement Implementation, and Sensorimotor Learning. Annual Review of Psychology, 72 , 61–95. 10.1146/annurev-psych-010419-051053 Kim H. E. , Morehead J. R. , Parvin D. E. , Moazzezi R. , & Ivry R. B. (2018). Invariant errors reveal limitations in motor correction rather than constraints on error sensitivity. Communications Biology, 1 (1 ). 10.1038/s42003-018-0021-y Kleiner M. , Brainard D. , Pelli D. , Ingling A. , Murray R. , & Broussard C. (2007). What’s new in psychtoolbox-3. Perception, 36 (14 ), 1–16. Krakauer J. W. (2009). Motor Learning and Consolidation: The Case of Visuomotor Rotation. Progress of Motor Control, 957 , 405–421. 10.1007/978-3-319-47313-0 Krakauer J. W. , Hadjiosif A. M. , Xu J. , Wong A. L. , & Haith A. M. (2019). Motor learning. Comprehensive Physiology, 9 (2 ), 613–663. 10.1002/cphy.c170043 30873583 Kuling I. A. , de Brouwer A. J. , Smeets J. B. J. , & Flanagan J. R. (2019). Correcting for natural visuo-proprioceptive matching errors based on reward as opposed to error feedback does not lead to higher retention. Experimental Brain Research, 237 (3 ), 735–741. 10.1007/s00221-018-5456-3 30560507 Leow L. A. , Marinovic W. , de Rugy A. , & Carroll T. J. (2018). Task errors contribute to implicit aftereffects in sensorimotor adaptation. European Journal of Neuroscience, 48 (11 ), 3397–3409. 10.1111/ejn.14213 30339299 Marinovic W. , Poh E. , De Rugy A. , & Carroll T. J. (2017). Action history influences subsequent movement via two distinct processes. ELife, 6 , 1–23. 10.7554/eLife.26713 Mawase F. , Uehara S. , Bastian A. J. , & Celnik P. (2017). Motor learning enhances use-dependent plasticity. Journal of Neuroscience, 37 (10 ), 2673–2685. 10.1523/JNEUROSCI.3303-16.2017 28143961 Mazzoni P. , & Krakauer J. W. (2006). An Implicit Plan Overrides an Explicit Strategy during Visuomotor Adaptation. Journal of Neuroscience, 26 (14 ), 3642–3645. 10.1523/JNEUROSCI.5317-05.2006 16597717 Morehead J. R. , & Orban de Xivry J.-J. (2021). A Synthesis of the Many Errors and Learning Processes of Visuomotor Adaptation. BioRxiv, James 1891 , 1–50. Morehead J. R. , Taylor J. A. , Parvin D. E. , & Ivry R. B. (2017). Characteristics of implicit sensorimotor adaptation revealed by task-irrelevant clamped feedback. Journal of Cognitive Neuroscience, 26 (6 ), 1–10. 10.1162/jocn Nikooyan A. A. , & Ahmed A. A. (2015). Reward feedback accelerates motor learning. Journal of Neurophysiology, 113 (2 ), 633–646. 10.1152/jn.00032.2014 25355957 Schween R. , & Hegele M. (2017). Feedback delay attenuates implicit but facilitates explicit adjustments to a visuomotor rotation. Neurobiology of Learning and Memory, 140 , 124–133. 10.1016/j.nlm.2017.02.015 28257877 Shadmehr R. , Smith M. A. , & Krakauer J. W. (2010). Error correction, sensory prediction, and adaptation in motor control. Annual Review of Neuroscience, 33 , 89–108. 10.1146/annurev-neuro-060909-153135 Shmuelof L. , Huang V. S. , Haith A. M. , Delnicki R. J. , Mazzoni P. , & Krakauer J. W. (2012). Overcoming Motor “Forgetting” Through Reinforcement Of Learned Actions. Journal of Neuroscience, 32 (42 ), 14617–14621a. 10.1523/JNEUROSCI.2184-12.2012 23077047 Simani M. C. , McGuire L. M. M. , & Sabes P. N. (2007). Visual-shift adaptation is composed of separable sensory and task-dependent effects. Journal of Neurophysiology, 98 (5 ), 2827–2841. 10.1152/jn.00290.2007 17728389 Synofzik M. , Lindner A. , & Thier P. (2008). The Cerebellum Updates Predictions about the Visual Consequences of One’s Behavior. Current Biology, 18 (11 ), 814–818. 10.1016/j.cub.2008.04.071 18514520 Taylor J. A. , & Ivry R. B. (2014). Cerebellar and Prefrontal Cortex Contributions to Adaptation, Strategies, and Reinforcement Learning. In Progress in Brain Research (1st ed., Vol. 210 ). Elsevier B.V. 10.1016/B978-0-444-63356-9.00009-1 Taylor J. A. , Krakauer J. W. , & Ivry R. B. (2014). Explicit and Implicit Contributions to Learning in a Sensorimotor Adaptation Task. Journal of Neuroscience, 34 (8 ), 3023–3032. 10.1523/JNEUROSCI.3619-13.2014 24553942 Therrien A. S. , Wolpert D. M. , & Bastian A. J. (2016). Effective Reinforcement learning following cerebellar damage requires a balance between exploration and motor noise. Brain, 139 (1 ), 101–114. 10.1093/brain/awv329 26626368 Therrien A. S. , Wolpert D. M. , & Bastian A. J. (2018). Increasing Motor Noise Impairs Reinforcement Learning in Healthy Individuals. Eneuro, 5 (June ), ENEURO.0050–18.2018. 10.1523/ENEURO.0050-18.2018 Therrien A. S. , & Wong A. L. (2022). Mechanisms of Human Motor Learning Do Not Function Independently. Frontiers in Human Neuroscience, 15 (January ), 1–9. 10.3389/fnhum.2021.785992 Tsay J. S. , & Ivry R. B. (2022). Understanding implicit sensorimotor adaptation as a process of proprioceptive re-alignment. ELife, 1–45. Tsay J. S. , Kim H. E. , Saxena A. , Parvin D. E. , Verstynen T. , & Ivry R. B. (2022). Dissociable use-dependent processes for volitional goal-directed reaching. Proc. R. Soc. B., 289 , 20220415. 10.21856/j-pep.2021.4.08 Tseng Y. W. , Diedrichsen J. , Krakauer J. W. , Shadmehr R. , & Bastian A. J. (2007). Sensory prediction errors drive cerebellum-dependent adaptation of reaching. Journal of Neurophysiology, 98 (1 ), 54–62. 10.1152/jn.00266.2007 17507504 van der Kooij K. , Brenner E. , Van Beers R. J. , Schot W. D. , & Smeets J. B. J. (2013). Alignment to natural and imposed mismatches between the senses. Journal of Neurophysiology, 109 (7 ), 1890–1899. 10.1152/jn.00845.2012 23343893 van der Kooij K. , & Smeets J. B. J. (2018). Reward-Based Motor Adaptation Can Generalize Across Actions. Journal of Experimental Psychology: Learning Memory and Cognition, 0 (999 ), 71–81. 10.1037/xlm0000573 van der Kooij K. , van Mastrigt N. M. , Crowe E. M. , & Smeets J. B. J. (2021). Learning a reach trajectory based on binary reward feedback. Scientific Reports, 11 (1 ), 1–15. 10.1038/s41598-020-80155-x 33414495 van Mastrigt N. M. , van der Kooij K. , & Smeets J. B. J. (2021). Pitfalls in quantifying exploration in reward-based motor learning and how to avoid them. Biological Cybernetics, 115 (4 ), 365–382. 10.1007/s00422-021-00884-8 34341885 Verstynen T. , & Sabes P. N. (2011). How each movement changes the next: An experimental and theoretical study of fast adaptive priors in reaching. Journal of Neuroscience, 31 (27 ), 10050–10059. 10.1523/JNEUROSCI.6525-10.2011 21734297
PMC010xxxxxx/PMC10327146.txt
==== Front bioRxiv BIORXIV bioRxiv Cold Spring Harbor Laboratory 37425711 10.1101/2023.06.27.546744 preprint 1 Article Self-extinguishing relay waves enable homeostatic control of human neutrophil swarming http://orcid.org/0000-0001-5131-5174 Strickland Jack 12 http://orcid.org/0000-0002-4597-6942 Pan Deng 3 http://orcid.org/0009-0000-5064-358X Godfrey Christian 45 http://orcid.org/0000-0001-6927-4649 Kim Julia S. 16 http://orcid.org/0000-0003-4860-4121 Hopke Alex 45 Degrange Maureen 7 Villavicencio Bryant 8 http://orcid.org/0000-0001-8892-8695 Mansour Michael K. 59 http://orcid.org/0000-0003-4219-0503 Zerbe Christa S. 10 http://orcid.org/0000-0001-7347-2082 Irimia Daniel 45 http://orcid.org/0000-0003-2611-0139 Amir Ariel 311 http://orcid.org/0000-0002-1778-6543 Weiner Orion D. 12 1 Department of Biochemistry and Biophysics, University of California San Francisco, San Francisco, CA, USA 2 Cardiovascular Research Institute, University of California San Francisco, San Francisco, CA, USA 3 John A. Paulson School of Engineering and Applied Sciences, Harvard University, Cambridge, MA, USA. 4 BioMEMS Resource Center and Center for Surgery, Innovation and Bioengineering, Department of Surgery, Massachusetts General Hospital, Boston, MA, USA. 5 Harvard Medical School, Boston, MA, USA. 6 Tetrad Graduate Program, UCSF, San Francisco, CA, USA. 7 Leidos Biomedical Research, Inc., Frederick National Laboratory for Cancer Research, Frederick, MD 21702, USA. 8 Kelly Government Services, Bethesda, MD, USA. 9 Division of Infectious Diseases, Massachusetts General Hospital, Boston, MA, USA. 10 Laboratory of Clinical Immunology and Microbiology (LCIM), National Institute of Allergy and Infectious Diseases (NIAID), National Institutes of Health (NIH), Bethesda, MD, USA. 11 Department of Complex Systems, Faculty of Physics, The Weizmann Institute of Science, Rehovot 7610001, Israel. Correspondence: [email protected] 28 6 2023 2023.06.27.546744https://creativecommons.org/licenses/by-nc-nd/4.0/ This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which allows reusers to copy and distribute the material in any medium or format in unadapted form only, for noncommercial purposes only, and only so long as attribution is given to the creator. nihpp-2023.06.27.546744.pdf Neutrophils exhibit self-amplified swarming to sites of injury and infection. How swarming is controlled to ensure the proper level of neutrophil recruitment is unknown. Using an ex vivo model of infection, we find that human neutrophils use active relay to generate multiple pulsatile waves of swarming signals. Unlike classic active relay systems such as action potentials, neutrophil swarming relay waves are self-extinguishing, limiting the spatial range of cell recruitment. We identify an NADPH-oxidase-based negative feedback loop that is needed for this self-extinguishing behavior. Through this circuit, neutrophils adjust the number and size of swarming waves for homeostatic levels of cell recruitment over a wide range of initial cell densities. We link a broken homeostat to neutrophil over-recruitment in the context of human chronic granulomatous disease. chemotaxis collective migration biological self-organization neutrophil swarming CGD self-extinguishing relay ==== Body pmcDuring collective migration, individual organisms coordinate their movement to solve critical tasks. Birds flock, fish school, and insects swarm to escape predation, find food, and move more efficiently Chandra et al. (2021); Gordon (2019); Parrish and Edelstein-Keshet (1999). These organismal collectives rely on individuals acting on local information, including distance to neighbors, alignment with neighbors, and local chemical cues, to generate complex emergent decisions. Collective migration is also organized at the cellular level, where groups of cells coordinate wound healing LaChance et al. (2022); Sun et al. (2023), cancer metastasis Muinonen-Martin et al. (2014), multicellular morphogenesis Donà et al. (2013), and the transition from single cell to multicellular existence Gregor et al. (2010). We are just beginning to understand the molecular logic of cell-cell communication that enables collective behaviors to arise. Here we probe the logic of collective migration in the context of the human immune response. Neutrophils are first responders of the innate immune system that are recruited to sites of injury and infection to neutralize invading pathogens and aid in tissue repair Ley et al. (2018); Kolaczkowska and Kubes (2013); Peiseler and Kubes (2019). Because the primary signals of injury/infection are relatively short ranged, activated neutrophils release chemoattractants such as Leukotriene B4 (LTB4) for positive feedback-based recruitment of additional neutrophils. This self-amplified ‘swarming’ process significantly enhances the speed and range of recruitment Kienle et al. (2021); Sun and Shi (2016); Poplimont et al. (2020); Reátegui et al. (2017); Lämmermann et al. (2013), but this process must be tightly regulated to limit collateral damage Kienle et al. (2021); Uderhardt et al. (2019); Jiwa et al. (2020); Mawhin et al. (2018). While some brakes on swarming are known Kienle et al. (2021); Reátegui et al. (2017); Uderhardt et al. (2019), how neutrophils control the spatiotemporal dynamics of cell-cell communication to recruit the appropriate number of cells is not well understood, and several fundamental questions remain unanswered. Are swarms primarily constrained by the neutrophils themselves Kienle et al. (2021), or are other cells required Uderhardt et al. (2019)? Do different molecular programs control the duration versus the range of swarming? And how robust is swarming to initial conditions? While many neutrophil swarming studies have been performed in living animals Kienle et al. (2021); Lämmermann et al. (2013); Uderhardt et al. (2019); Poplimont et al. (2020); Isles et al. (2021); Chtanova et al. (2008); Khazen et al. (2022); Ng et al. (2011), the in vivo context presents several challenges for mechanistic dissection of the swarming process. For example, multiple cell types potentially modulate swarm initiation and propagation Uderhardt et al. (2019), there are a multitude of diffusive signals following tissue damage Uderhardt et al. (2019); Lämmermann et al. (2013), and the in vivo migration environment is complex. These challenges make it difficult to mechanistically dissect the regulation of swarming. Furthermore, a focus on model organisms precludes the analysis of swarming for human neutrophils despite known differences in primary human neutrophil behaviors compared to model systems Nauseef (2023); Siwicki and Kubes (2023). To address these limitations, we are leveraging an ex vivo assay for studying human neutrophil swarming in response to controllable, reproducible, well-defined cues. This assay expands upon a previously-developed ex vivo swarming system Reátegui et al. (2017); Hopke et al. (2020) in which a defined grid of heat-killed Candida albicans “targets” are spotted on a slip of glass to act as an array of swarm initiation points. Healthy human primary neutrophils isolated from whole blood are added to the assay, and swarming responses are observed with live cell microscopy. Fast-moving multicellular Ca2+ waves define the zone of recruitment in human neutrophil swarming To monitor cell-cell communication in conjunction with more traditional swarming readouts like motility, we used a cytosolic calcium dye. Neutrophils increase cytosolic calcium following exposure to primary chemoattractants as well as swarming cues such at LTB4 Molski et al. (1981), and we envisioned that monitoring calcium influx would enable us to follow the propagation of swarming signals and correlate these to the regulation of directed cell movement (Figure 1a). Furthermore, calcium signals at the core of injury sites in vivo have been shown to be linked to key swarm signaling events Poplimont et al. (2020), and the simpler planar environment and higher signal-to-noise imaging of our assay should enable a more sensitive detection of swarm signal propagation to neutrophils far from the heat-killed Candida albicans target. Calcium signaling has been studied during mice and zebrafish swarming Poplimont et al. (2020); Khazen et al. (2022) but not during human neutrophil swarming. Following neutrophil introduction to the assay, we observe a calcium influx of the neutrophils that are in contact with the fungal target as well as multiple multicellular waves of calcium activity that radially propagate away from the target to surrounding neutrophils (Supplemental Movie SM1). An example time course of these waves is shown in Figure 1b with the overlayed tracked boundary of the calcium wave. Wave tracks are calculated by fitting a circle to a cloud of cells that are determined to be active via calcium signal binning and grouped using the ARCOS algorithm Gagliardi et al. (2022) (Supplemental Figure S1a,b; Supplemental Movie SM2). Wave movement is roughly isotropic, and the fitting of waves to a circle enables the tracking of wave propagation kinetics (Supplemental Figure S1c,d). Immediately following the passage of a calcium wave, cells rapidly polarize and radially migrate towards the wave origin site. Cells within the wave perimeter move towards the fungal target, while cells outside the wave perimeter lack coordinated movement (Figure 1c,d). By analyzing cell responses across many targets and donors, we observe that most, if not all, movement occurs within the boundaries of the tracked calcium wave, indicating that these wave fronts represent an effective boundary of the swarming guidance cue reception (Figure 1e). Though the propagating calcium waves delimit the zone of recruited neutrophils, they operate at very different timescales with respect to neutrophil movement. Calcium waves move an order of magnitude faster than the resulting neutrophil chemotaxis towards the wave origin (Figure 1f), and calcium waves propagate in an opposite direction to cell movement during swarming. Wave propagation during neutrophil swarming is consistent with active relay but not core diffusion. LTB4, an inflammatory lipid of the leukotriene family, is thought to be one of the key secreted molecules that regulates neutrophil swarming Lämmermann et al. (2013); Reátegui et al. (2017). Neutrophils release LTB4 in response to various damage-associated pattern molecules and pathogen-associated pattern molecules Reátegui et al. (2017); Afonso et al. (2012); McDonald et al. (1994), including Candida albicans Fischer et al. (2021). To test whether LTB4 reception is required for the rapid long-range calcium waves in our ex vivo swarming assay, we blocked LTB4 reception with the LTB4 receptor antagonist BIIL315 Birke et al. (2001). Blocking LTB4 receptors inhibited the rapidly propagating long-range calcium waves that accompany swarming. In the absence of LTB4 reception, a much smaller range, slow-moving Ca2+ wave was observed (Supplemental Figure S2b,c; Supplemental Movie SM3). Quantitative analysis of wave propagation in both settings indicate that the rapid, long-range calcium waves that accompany swarming are dependent on LTB4 reception (Figure 2a). We next investigated how the LTB4 ligands are propagated from the target site to the rest of the field during swarming. Two models have been proposed for the signal amplification observed during swarming. For the relay model of swarming, pioneer neutrophils at the site of infection secrete LTB4, which activates surrounding neutrophils to secrete additional LTB4, thereby continuing the relay Lämmermann et al. (2013); Afonso et al. (2012); Dieterle et al. (2019). This active relay mechanism could enable neutrophils to collectively signal across significant distances from the injury/infection site, analogous to cell-cell signal propagation in aggregating Dictyostelium Shaffer (1975); van Oss et al. (1996). In contrast, for the core production model of swarming, only neutrophils in direct contact with the site of injury/infection produce significant LTB4, which then passively diffuses into the tissue to attract more neutrophils Poplimont et al. (2020). These two swarming models give very different predictions for the dynamics of propagation of the LTB4 wavefront as it moves away from the site of infection. The relay model predicts a wave that travels at a fixed velocity because of continuous signal re-amplification at the traveling front. This produces an activation zone whose area (wave fit circle radius squared / R2) scales quadratically in time. In contrast, since the core diffusion model predicts LTB4 production restricted to the central source, this system is limited by diffusion, and therefore its activation area (R2) scales linearly in time (Figure 2b). We tracked the radius2 versus time for multiple waves across many healthy donors in control cells (Figure 2c) versus LTB4-blockaded cells (Figure 2d) and fit a power law to each individual wave trajectory. This analysis reveals that control cells exhibit an LTB4 reception-dependent wave propagation with an alpha close to 2, consistent with the active relay model of swarming and inconsistent with the core production model of swarming (Figure 2e,f). In the absence of LTB4 signal reception, cells exhibit a slower, smaller pattern of signal propagation with an alpha close to 1, which is consistent with core diffusion. These data suggest that the residual release of non-LTB4 ligands from cells on the target passively diffuse to the rest of the field (Figure 2e,f). These two examples establish that we can distinguish between diffusive waves and relay-mediated ones, and LTB4-mediated waves exhibit relay-like ballistic spreading in the early phase of wave propagation. Swarming relay waves self-extinguish through an NADPH Oxidase negative feedback loop Our observation that neutrophil swarming cues radiate from killed yeast targets via active relay might be expected to produce waves that continue to propagate as long as there are nearby receptive cells to continue the wave. This is the expected behavior of actively relayed systems including action potentials, mitotic waves, and Dictyostelium aggregation Gelens et al. (2014). We previously modeled an active relay model for swarming in which cells that pass a threshold amount of LTB4 themselves create more LTB4, as diagramed in Figure 3a Dieterle et al. (2019). Once initiated, these waves continue to propagate indefinitely given a field of responsive cells. In contrast with the predictions of a simple relay model, our experimentally-observed waves propagate and then abruptly stop (Figure 3b,c), indicating a more complex mode of regulation than positive feedback alone. While negative feedback loops could theoretically collaborate with positive feedback to generate self-extinguishing relay, we were only able to find one such example in the literature. In this work, a 1D model of two separately-generated waves of diffusing activator and inhibitor were able to generate a fixed radius of response in a cell-free system Ataullakhanov et al. (1998). We sought a simpler potential mechanism of wave stopping that does not depend on two independently propagated waves (see Supplementary Modeling Text, Supplemental Figure S3) and that is consistent with the dynamics of wave termination in our experimental context. This self-extinguishing relay model depends on two different responses initiated at two different concentrations of LTB4. Cells begin generating a non-diffusive internal inhibitor once they pass a low extracellular threshold of LTB4. After passing a higher threshold of extracellular LTB4, cells begin to release LTB4 themselves (Figure 3d). In this system, cells farther from an initiation event will have an increasing time delay between crossing the lower ‘inhibition circuit initiation’ threshold and the higher LTB4 production threshold. When a cell passes the threshold for inhibition, it begins accumulating the inhibitory cue that limits the relay strength of that cell. Close to the source of the activating cue at the target, only a small amount of inhibitor is generated prior to LTB4 production, and relay is still possible. Further from the source, the inhibitor accumulates for longer periods before the LTB4 production threshold is passed, thereby terminating further signal relay. This model can recapitulate the experimentally-observed wavefront dynamics (Figure 3d). What might serve as the activation-dependent inhibitor in this self-extinguishing relay mechanism? The inhibitor should be produced downstream of LTB4 reception, and the accumulation of this inhibitor should attenuate LTB4 production. NADPH oxidase activation satisfies both conditions. NADPH oxidase is used in ROS-dependent pathogen killing downstream of LTB4 and other chemoattractants Song et al. (2020); Hopke et al. (2020), and the inhibition of NADPH oxidase activation potentiates LTB4 production and neutrophil accumulation during swarming Song et al. (2020); Hopke et al. (2020); Henrickson et al. (2018); Roxo-Junior and Simão (2014); Hamasaki et al. (1989). If NADPH oxidase activity is a critical component in self-extinguishing relay, we would expect NADPH oxidase inhibition to prevent swarming wave termination, thereby reverting the swarming process to a simple relay system (as in Figure 3a). Indeed, the NADPH oxidase inhibitor DPI (Figure 3e, Supplemental Movie SM4) abolished the ability of swarming waves to self-extinguish, with swarming waves continuing to propagate from the targets out of the microscopic field of view. To ensure that this effect is not an off-target of DPI, we analyzed neutrophils from Chronic Granulomatosis Disease (CGD) patients with a genetic defect in neutrophil NADPH oxidase machinery. CGD neutrophils exhibit similar non-extinguishing wave dynamics as DPI-treated healthy donor cells (Supplemental Figure S4a). While both DPI based NADPH Oxidase-inhibited and a CGD donor wave activity led to strong recruitment of the cells within a calcium event, they only generated a single large wave compared to the multiple waves observed in unperturbed cells (Supplemental Figure S4b; Supplemental Movie SM4). Neutrophils tune the size and number of swarming waves for homeostatic recruitment to targets To investigate the physiological significance of multiple self-extinguishing waves (control neutrophils) versus a single uncontrolled relay wave (DPI-treated or CGD neutrophils), we next sought to investigate the relation between wave size/number and the resulting cell movement and recruitment. We hypothesized that these wave parameters may be actively adjusted to ensure robust homeostatic control of neutrophil recruitment across a wide range of initial conditions. We created quantitative metrics for both the propagating Ca2+ wave as well as neutrophil chemotaxis response that could be extended to multiple wave sizes and intensities. First, we aligned single-cell tracks for neutrophils entering the zone of recruitment and averaged their radial migration velocity profile for single, well-separated waves (Figure 4a, Supplemental Figure S5a). Upon entering a wave, neutrophils execute a discrete ‘run’ of movement towards the wave center but then revert to random, slower movement. Such a ‘step’ of movement inwards is consistent with the expected behavior from our self-extinguishing model where the chemotactic gradient decays behind the wavefront as the relay strength becomes more dominated by the local inhibitory mechanism. These data indicate that a single wave event does not recruit all cells to the center in one burst. To expand our analysis to multiple waves, we integrated the radial movement of all cells within a wave event (Figure 4b, Supplemental Figure S5b). Each wave independently recruited of a bolus of neutrophils toward a given target, with larger wave events more potently stimulating cell recruitment than smaller wave events (Supplemental Movie SM5). When looking across many experiments and donors, the total integrated wave area highly correlated with the Integrated Radial Movement of the cells within wave events (Figure 4c). We asked next whether cells tune the number or size of swarming waves for robust homeostatic control of swarming over a range of initial conditions. Towards this end, we seeded our swarming assay at a range of initial neutrophil densities. As seeding density increased, neutrophils generated smaller and fewer swarming waves, resulting in a decreased overall integrated calcium area (Figure 4d, Supplemental Figure S5d, e). This reduction in swarming waves with increasing cell density could potentially enable a homeostatic recruitment of the same number of neutrophils across a range of initial cell densities. To investigate this possibility, we monitored neutrophil recruitment to sites of heat-killed Candida albicans at a range of seeding densities over 60 min (Supplemental Figure S5f, g; Supplemental Movie SM6). Over a 6-fold change in initial cell density, there was a nearly constant accumulation of neutrophils at the target, with targets at the lowest densities attracting approximately 85% of the cells attracted at the highest cell densities (compared to 6x differences that would be expected for a non-homeostatic system) (Figure 4e, Supplemental Figure S5h). Because CGD cells are defective in the negative feedback loop that constrains wave size, we predicted that these cells should exhibit a defective homeostat and therefore recruit cells proportional to their surrounding density. Indeed, CGD cells exhibit a much stronger density dependence on recruitment than healthy donor cells (Figure 4f, Supplemental Figure S4e, f). Discussion By leveraging a cell-cell signaling readout in a highly controllable ex vivo swarming assay, our work reveals self-extinguishing waves of human neutrophil recruitment to sites of heat-killed Candida albicans. The active relay is based on an LTB4-positive feedback loop (Figure 2), and the ability to self-extinguish depends on LTB4-based NADPH-oxidase activation (Figure 3, 4). The active relay enables cells to transmit guidance cues much farther and faster than would be possible if only the cells directly on the target secreted swarming signals. Furthermore, the propagated waves may generate temporally-evolving guidance cues that enable more effective chemotaxis than static spatial gradients Geiger et al. (2003); Tweedy et al. (2016); Aranyosi et al. (2015); Skoge et al. (2014). Behind a wavefront, decaying gradients could further prevent cells from over accumulating at the core of swarms as cellular memory mechanisms lose strength Skoge et al. (2014) and as chemotaxis-inhibitory LTB4 metabolites accumulate Archambault et al. (2019); Pettipher et al. (1993); Tweedy et al. (2016). The self-extinguishing nature of the relay enables the system to adjust the number and size of swarming waves to achieve robust homeostatic control of neutrophil recruitment over a wide range of initial cell concentrations. Disruption of this homeostat has severe consequences for neutrophil recruitment in vivo. Previous work has shown that CGD patient neutrophils (that are defective in NADPH-oxidase activation) overproduce LTB4 and hyperaccumulate at sites of injury/infection in vitro and in vivo Song et al. (2020); Hamasaki et al. (1989); Henrickson et al. (2018); Hopke et al. (2020); Dinauer (2019). Our work shows that these cells are also defective in the negative feedback arm that limits the range of swarming signals and therefore lack the homeostat that normally constrains cell swarming. Our ex vivo system demonstrates that human neutrophils can modulate their cell-cell signaling to limit the range of cell swarming without feedback from other cell types or environmental cues. In future work, it will be interesting to probe how other cell types influence the initiation, termination, and resolving phases of swarming. We have focused on killed yeast-mediated neutrophil swarming (whose closest parallel is likely an infected lymph node Chtanova et al. (2008); Lämmermann et al. (2013)), but it will be interesting to compare how sterile injury Lämmermann et al. (2013); Ng et al. (2011); Poplimont et al. (2020); Park et al. (2018) and combined injury/infection Poplimont et al. (2020); Chtanova et al. (2008) change the dynamics of the swarming process. Other swarming terminators (such as Grk2 activation Kienle et al. (2021)) control the duration of swarming, whereas the NADPH oxidase negative feedback circuit studied in our work also controls the spatial range of swarming signal propagation. It will be interesting to probe how these swarming termination programs relate to one another. Finally, as there are a number of other human diseases with known swarming defects Knooihuizen et al. (2021); Alexander et al. (2021); Barros et al. (2021), it will be interesting to determine how these disease states interact with our model parameters and effect the relay system we study in this work. Methods For a detailed description of all the methods and code used in this work, please see Supplemental Methods. Supplementary Material Supplement 1 Supplement 2 Supplement 3 Supplement 4 Supplement 5 Supplement 6 Supplement 7 Supplement 8 Supplement 9 Supplement 10 Supplement 11 Acknowledgements We thank members of the Weiner, Amir, and Irimia lab for helpful discussions, particularly Nick Martin and Henry De Belly for a critical reading of the manuscript. We thank Paul Dieterle, Wencheng Ji, and Allyson Sgro for their detailed feedback and advice regarding the theoretical model. We also thank the blood donors and phlebotomists without whom this project would have been impossible. This work was supported by an AHA predoctoral fellowship (JDS), the National Institutes of Health grants (GM118167 to ODW, GM092804 to DI, R01 AI132638 to MKM), the National Science Foundation/Biotechnology and Biological Sciences Research Council grant (2019598 to ODW), the National Science Foundation Center for Cellular Construction Grant (DBI-1548297 to ODW), the Department of Defense (HU00012020011 to MKM) the Novo Nordisk Foundation grant for the Center for Geometrically Engineered Cellular Systems (NNF17OC0028176 to ODW), the Clore Center for Biological Physics (AA), the NSF CAREER Grant (1752024 to AA), and the Shriners Hospitals for Children (71010-BOS-22 to DI). CZ was supported by the Intramural Research Program of the National Institutes of Health Clinical Center and the National Institute of Allergy and Infectious Diseases. Finally, we would like to thank the Center for Advanced Light Microscopy-Nikon Imaging Center at UCSF for their support in using the CREST/C2 Confocal microscope. BIIL 315 was kindly provided by Boehringer Ingelheim via its open innovation platform opnMe, available at https://opnme.com. Figure 1. Fast-moving multicellular Ca2+ waves define the zone of recruitment in human neutrophil swarming. (A) Ex vivo assay for neutrophil swarming with self-amplified recruitment of cells in response to local site of heat-killed Candida albicans. Human neutrophils are isolated via immunomagnetic selection and dyed with CalBryte 520-AM (cytosolic Ca2+ reporter, used to monitor neutrophil detection of pathogen and swarming cues) and Hoechst 3334 (nuclear marker, used to track cell movement). Cells are placed on printed targets of heat-killed Candida albicans, and neutrophil swarming responses are monitored via confocal microscopy. (B) Time lapse sequence of a multicellular wave of neutrophil Ca2+ activity following detection of the fungal target. Dotted line represents wave boundary as tracked by our analysis software; scale bar = 200 µm. (C) Ca2+ wave defines the zone of neutrophil recruitment towards the fugal target. Final radius of Ca2+ wave from cells in 1b is shown in blue. Cell tracks for 3 minutes following wave initiation are indicated with color corresponding to radial velocity towards the wave center; scale bar = 100 µm. (D) Average radial velocity for cells inside versus outside the final wave boundary is plotted over 4 minutes for a population of cells during a single Ca2+ wave (Cell tracks inside the wave n = 3720, outside wave tracks n = 9499). Wave boundary predicts zone of recruitment during swarming. (E) Comparison of mean radial velocity for cells inside versus outside largest Ca2+ wave boundary in each target ROI. Tracks were averaged over the duration of the wave event, and lines connect mean radial velocities of outer versus inner cells for each target ROI. Dependent paired t-test p-value = 5.8 × 10−27 (Target ROIs n = 56) (F) Average velocity of Ca2+ waves propagated across the field of cells during swarming compared to the average cell velocity for the cells within these wave boundaries. Propagated Ca2+ waves are approximately one order of magnitude faster than cell movement during swarming. Figure 2. Wave propagation during human neutrophil swarming is consistent with active relay but not core diffusion. (A) Ca2+ wave propagation across a field of neutrophils during swarming for cells in the absence (50 waves) or presence (17 waves) of an LTB4 receptor inhibitor (1 µM BIIL315); averages shown with 95% confidence interval shaded. LTB4 reception is required for rapid, long-range Ca2+ wave propagation. (B) Two proposed models of LTB4 propagation from a site of infection to the rest of the field with the predicted kinetics of signal propagation shown for each. In a relay model of swarming Afonso et al. (2012); Lämmermann et al. (2013); Dieterle et al. (2019), each activated neutrophil releases LTB4, which stimulates the adjacent cell to release LTB4; this wave should spread with a constant velocity over time, giving an α of 2. For a core production model of swarming Poplimont et al. (2020), only neutrophils at the center of a swarm produce LTB4, which moves across the field through passive diffusion; this wave should spread with an α of 1 (linear on R2 plot). (C) Average early wave kinetics for control cells with mean and 95% confidence interval. Individual tracks are fit to a power law equation to determine the α of wave propagation for each experiment. (D) Wave propagation for LTB4R-inhibited cells averaged with 95% confidence interval in grey are individually fit to a power law equation for the first two minutes following swarm initiation. (E) Resulting curve fit alpha parameters for both unperturbed and LTB4R blockade conditions are plotted, with control cells showing wave propagation consistent with active relay. LTB4 reception-inhibited cells showed residual wave propagation consistent with core diffusion. Welch’s t-test P-value = 1.7 × 10−7 (F) Summary of this figure. LTB4-mediated waves propagate with kinetics consistent with active relay (and not core diffusion), while the zone of cell activation for LTB4-blockaded cells propagates away from the site of heat-killed Candida albicans with kinetics consistent with simple diffusion. Figure 3. Neutrophils employ a NADPH-oxidase-based negative feedback loop to generate a self-extinguishing relay during swarming. (A) Previously-proposed simple diffusive relay model of swarming Dieterle et al. (2019). As cells exceed a threshold amount of extracellular LTB4, they begin to release more LTB4 to their surroundings. This wave propagates indefinitely once initiated in a homogenous field of cells. (B) Comparing the simple diffusive relay prediction with an average of experimentally-observed swarming wavefronts shows how wave behavior diverges, as experimental relay waves come to a stop, whereas simple relay propagates indefinitely (Axes in µm. Wave n = 5. Volunteer N = 2). (C) While experimental waves initially propagate with a relay-like R2 relation, in the latter phase the wave decelerates then stops. (D) With the introduction of an activation-dependent inhibitor, it is possible to generate models of a self-extinguishing relay. Here different extracellular LTB4 thresholds for production of a local inhibitor (low threshold) versus release of more LTB4 (high threshold) generate an area of local inhibition that limits the wave propagation. Since the local inhibition requires accumulation over time to inhibit relay, wave termination is only achieved at further distances from the source. Our self-extinguishing relay model yields wavefront kinetics that closely resemble the experimentally observed Ca2+ swarming wavefront. The red X denotes where the model predicts the wave extinguishing (2D panel axes in µm.). (E) Here we test a candidate activation-dependent inhibitor for the self-extinguishing relay during swarming. In neutrophils, LTB4-stimulated activation of the NADPH Oxidase complex limits the Ca2+ influx that potentiates LTB4 synthesis Song et al. (2020). Inhibition of NADPH oxidase activity via the NADPH oxidase inhibitor DPI produce swarming waves that no longer stop in the observable microscopic field (lower panels, scale: 200 µm) and mimic simple relay dynamics throughout their propagation (right panel) (Volunteer N = 3, Wave n = 11). Figure 4. Neutrophils achieve homeostatic recruitment to sites of heat-killed Candida albicans by modulating the size and frequency of swarming waves. (A) Averaged radial velocities of neutrophil movement towards a target following time of exposure to swarming Ca2+ wave with 95% confidence interval plotted for 1758 cell tracks. A single wave produces a transient bolus of neutrophil recruitment. (B) All cellular tracks inside a wave are integrated for their radial movement (Cell track n = 5813); multiple Ca2+ wave events for the same target are plotted as shaded areas colored by maximum final area for an example ROI. Each wave induces a bolus of neutrophil movement towards the target, with larger waves inducing larger neutrophil responses. (C) Integrated radial movement of all cell tracks and integrated wave max area measurements calculated across 57 target ROIs demonstrate the strong positive correlation between Ca2+ wave signaling area and movement towards a target. Linear fit slope: 7.8 µm ± 0.8, 95% confidence interval. A control where all cell tracks are considered is in Supplemental Figure S5C. (D) Relation of seeding density to cell accumulation at target. Integrated wave max area measurements taken for each target ROI over 45min. Linear fit slope: −6.3 × 104 µm2 per 1000 cells/mm2 ± 1 × 104, 95% confidence interval. Lower cell densities elicit larger and more numerous swarming waves. (Scale bar: 200 µm). (E) Z-stack measurements taken across multiple ROIs per well 60 min after seeding swarming assays at different cell densities. Accumulation measured as integrated fluorescence at target site divided by median single-cell nuclear fluorescence. Low Density ROI Estimation: 402 cells, High Density ROI Estimation: 551 cells. Linear fit slope: 27 cells accumulate at target site per 1000 cells/mm2 ± 4 cells; 95% confidence interval. Only a small change in final neutrophil target accumulation is seen over a 6x range of cell swarming densities, compared to expectation for swarming directly proportional to cell density (linear scaling, orange line) (Scale bar: 100 µm). (F) Widefield measurements taken across multiple ROIs per well 60min after seeding swarming assays with different cell densities of either control donor cells or CGD (NADPH oxidase-defective) human donor cells. Accumulation measured similar to E, see Methods. Control ROI Estimation: 295 cells, CGD ROI Estimation: 4780 cells. CGD Linear fit slope: 297 cells accumulate at a target per 1000 cells/mm2 ± 19 cells, 95% confidence interval (Volunteer N=4, Target n = 253). Healthy control linear fit slope: 40 cells accumulate at a target per 1000 cells/mm2 ± 4 cells (Volunteer N = 3, Target n = 157)(Scale bar: 100 µm). ==== Refs Bibliography Afonso P. V. , Janka-Junttila M. , Lee Y. J. , McCann C. P. , Oliver C. M. , Aamer K. A. , Losert W. , Cicerone M. T. , and Parent C. A. LTB4 is a Signal-Relay molecule during neutrophil chemotaxis. Dev. Cell, 22 (5 ):1079–1091, May 2012.22542839 Alexander N. J. , Bozym D. J. , Farmer J. R. , Parris P. , Viens A. , Atallah N. , Hopke A. , Scherer A. , Dagher Z. , Barros N. , Knooihuizen S. A. I. , Saff R. R. , Pasternack M. S. , Thompson R. W. , Irimia D. , and Mansour M. K. Neutrophil functional profiling and cytokine augmentation for patients with multiple recurrent infections: A case study. J. Allergy Clin. Immunol. Pract., 9 (2 ):986–988, Feb. 2021.32853820 Aranyosi A. J. , Wong E. A. , and Irimia D. A neutrophil treadmill to decouple spatial and temporal signals during chemotaxis. Lab Chip, 15 (2 ):549–556, Jan. 2015.25412288 Archambault A.-S. , Poirier S. , Lefebvre J.-S. , Robichaud P.-P. , Larose M.-C. , Turcotte C. , Martin C. , Provost V. , Boudreau L. H. , McDonald P. P. , Laviolette M. , Surette M. E. , and Flamand N. 20-hydroxy- and 20-carboxy-leukotriene (LT) B4 downregulate LTB4-mediated responses of human neutrophils and eosinophils. J. Leukoc. Biol., 105 (6 ):1131–1142, 2019.30676680 Ataullakhanov F. I. , Guria G. T. , Sarbash V. I. , and Volkova R. I. Spatiotemporal dynamics of clotting and pattern formation in human blood. Biochim. Biophys. Acta, 1425 (3 ):453–468, Nov. 1998.9838209 Barros N. , Alexander N. , Viens A. , Timmer K. , Atallah N. , Knooihuizen S. A. I. , Hopke A. , Scherer A. , Dagher Z. , Irimia D. , and Mansour M. K. Cytokine augmentation reverses transplant recipient neutrophil dysfunction against the human fungal pathogen candida albicans. J. Infect. Dis., 224 (5 ):894–902, Sept. 2021.33417688 Birke F. W. , Meade C. J. , Anderskewitz R. , Speck G. A. , and Jennewein H. M. In vitro and in vivo pharmacological characterization of BIIL 284, a novel and potent leukotriene b(4) receptor antagonist. J. Pharmacol. Exp. Ther., 297 (1 ):458–466, Apr. 2001.11259574 Chandra V. , Gal A. , and Kronauer D. J. C. Colony expansions underlie the evolution of army ant mass raiding. Proc. Natl. Acad. Sci. U. S. A., 118 (22 ), June 2021. Chtanova T. , Schaeffer M. , Han S.-J. , van Dooren G. G. , Nollmann M. , Herzmark P. , Chan S. W. , Satija H. , Camfield K. , Aaron H. , Striepen B. , and Robey E. A. Dynamics of neutrophil migration in lymph nodes during infection. Immunity, 29 (3 ):487–496, Sept. 2008.18718768 Dieterle P. B. , Min J. , Irimia D. , and Amir A. Dynamics of diffusive cell signaling relays. Elife, 9 : e61771, Dec. 2019. Dinauer M. C. Inflammatory consequences of inherited disorders affecting neutrophil function. Blood, 133 (20 ):2130–2139, May 2019.30898864 Donà E. , Barry J. D. , Valentin G. , Quirin C. , Khmelinskii A. , Kunze A. , Durdu S. , Newton L. R. , Fernandez-Minan A. , Huber W. , Knop M. , and Gilmour D. Directional tissue migration through a self-generated chemokine gradient. Nature, 503 (7475 ):285–289, Nov. 2013.24067609 Fischer J. , Gresnigt M. S. , Werz O. , Hube B. , and Garscha U. Candida albicans-induced leukotriene biosynthesis in neutrophils is restricted to the hyphal morphology. FASEB J., 35 (10 ):e21820, Oct. 2021.34569657 Gagliardi P. A. , Grädel B. , Jacques M.-A. , Hinderling L. , Ender P. , Cohen A. R. , Kastberger G. , Pertz O. , and Dobrzyński M. Automatic detection of spatio-temporal signalling patterns in cell collectives. July 2022. Geiger J. , Wessels D. , and Soll D. R. Human polymorphonuclear leukocytes respond to waves of chemoattractant, like dictyostelium. Cell Motil. Cytoskeleton, 56 (1 ):27–44, Sept. 2003.12905529 Gelens L. , Anderson G. A. , and Ferrell J. E. Jr. Spatial trigger waves: positive feedback gets you a long way. Mol. Biol. Cell, 25 (22 ):3486–3493, Nov. 2014.25368427 Gordon D. M. The ecology of collective behavior in ants. Annu. Rev. Entomol., 64 :35–50, Jan. 2019.30256667 Gregor T. , Fujimoto K. , Masaki N. , and Sawai S. The onset of collective behavior in social amoebae. Science, 328 (5981 ):1021–1025, May 2010.20413456 Hamasaki T. , Sakano T. , Kobayashi M. , Sakura N. , Ueda K. , and Usui T. Leukotriene B4 metabolism in neutrophils of patients with chronic granulomatous disease: phorbol myristate acetate decreases endogenous leukotriene B4 via NADPH oxidase-dependent mechanism. Eur. J. Clin. Invest., 19 (4 ):404–411, Aug. 1989.2550242 Henrickson S. E. , Jongco A. M. , Thomsen K. F. , Garabedian E. K. , and Thomsen I. P. Non-infectious manifestations and complications of chronic granulomatous disease. J Pediatric Infect Dis Soc, 7 (suppl_1 ):S18–S24, May 2018.29746679 Hopke A. , Scherer A. , Kreuzburg S. , Abers M. S. , Zerbe C. S. , Dinauer M. C. , Mansour M. K. , and Irimia D. Neutrophil swarming delays the growth of clusters of pathogenic fungi. Nat. Commun., 11 (1 ):2031, Dec. 2020.32341348 Isles H. M. , Loynes C. A. , Alasmari S. , Kon F. C. , Henry K. M. , Kadochnikova A. , Hales J. , Muir C. F. , Keightley M.-C. , Kadirkamanathan V. , Hamilton N. , Lieschke G. J. , Renshaw S. A. , and Elks P. M. Pioneer neutrophils release chromatin within in vivo swarms. Elife, 10 : e68755, July 2021.34292151 Jiwa K. , Ward C. , Parry G. , Brodlie M. , Lordan J. , Fisher A. , and Garnett J. Immunohistochemical evidence of increased neutrophil swarms in cystic fibrosis lung removed at transplantation. Eur. Respir. J., 56 (suppl 64 ):698, Sept. 2020. Khazen R. , Corre B. , Garcia Z. , Lemaître F. , Bachellier-Bassi S. , d’Enfert C. , and Bousso P. Spatiotemporal dynamics of calcium signals during neutrophil cluster formation. Proceedings of the National Academy of Sciences, 119 (29 ):e2203855119, 2022. Kienle K. , Glaser K. M. , Eickhoff S. , Mihlan M. , Knöpper K. , Reátegui E. , Epple M. W. , Gunzer M. , Baumeister R. , Tarrant T. K. , Germain R. N. , Irimia D. , Kastenmüller W. , and Lämmermann T. Neutrophils self-limit swarming to contain bacterial growth in vivo. Science, 372 (6548 ), June 2021. Knooihuizen S. A. I. , Alexander N. J. , Hopke A. , Barros N. , Viens A. , Scherer A. , Atallah N. J. , Dagher Z. , Irimia D. , Chung R. T. , and Mansour M. K. Loss of coordinated neutrophil responses to the human fungal pathogen, candida albicans, in patients with cirrhosis. Hepatol Commun, 5 (3 ):502–515, Mar. 2021.33681682 Kolaczkowska E. and Kubes P. Neutrophil recruitment and function in health and inflammation. Nat. Rev. Immunol., 13 (3 ):159–175, Mar. 2013.23435331 LaChance J. , Suh K. , Clausen J. , and Cohen D. J. Learning the rules of collective cell migration using deep attention networks. PLoS Comput. Biol., 18 (4 ):e1009293, Apr. 2022.35476698 Lämmermann T. , Afonso P. V. , Angermann B. R. , Wang J. M. , Kastenmüller W. , Parent C. A. , and Germain R. N. Neutrophil swarms require LTB4 and integrins at sites of cell death in vivo. Nature, 498 (7454 ):371–375, June 2013.23708969 Ley K. , Hoffman H. M. , Kubes P. , Cassatella M. A. , Zychlinsky A. , Hedrick C. C. , and Catz S. D. Neutrophils: New insights and open questions. Sci Immunol, 3 (30 ), Dec. 2018. Mawhin M.-A. , Tilly P. , Zirka G. , Charles A.-L. , Slimani F. , Vonesch J.-L. , Michel J.-B. , Bäck M. , Norel X. , and Fabre J.-E. Neutrophils recruited by leukotriene B4 induce features of plaque destabilization during endotoxaemia. Cardiovasc. Res., 114 (12 ):1656–1666, Oct. 2018.29800147 McDonald P. P. , McColl S. R. , Braquet P. , and Borgeat P. Autocrine enhancement of leukotriene synthesis by endogenous leukotriene B4 and platelet-activating factor in human neutrophils. Br. J. Pharmacol., 111 (3 ):852–860, Mar. 1994.8019762 Molski T. F. , Naccache P. H. , Borgeat P. , and Sha’afi R. I. Similarities in the mechanisms by which formyl-methionyl-leucyl-phenylalanine, arachidonic acid and leukotriene B4 increase calcium and sodium influxes in rabbit neutrophils. Biochem. Biophys. Res. Commun., 103 (1 ): 227–232, Nov. 1981.6274342 Muinonen-Martin A. J. , Susanto O. , Zhang Q. , Smethurst E. , Faller W. J. , Veltman D. M. , Kalna G. , Lindsay C. , Bennett D. C. , Sansom O. J. , Herd R. , Jones R. , Machesky L. M. , Wakelam M. J. O. , Knecht D. A. , and Insall R. H. Melanoma cells break down LPA to establish local gradients that drive chemotactic dispersal. PLoS Biol., 12 (10 ):e1001966, Oct. 2014.25313567 Nauseef W. M. Human neutrophils ≠ murine neutrophils: Does it matter? Immunol. Rev., 314 (1 ):442–456, Mar. 2023.36380497 Ng L. G. , Qin J. S. , Roediger B. , Wang Y. , Jain R. , Cavanagh L. L. , Smith A. L. , Jones C. A. , Veer M. d. , Grimbaldeston M. A. , Meeusen E. N. , and Weninger W. Visualizing the neutrophil response to sterile tissue injury in mouse dermis reveals a Three-Phase cascade of events. J. Invest. Dermatol., 131 (10 ):2058–2068, Oct. 2011.21697893 Park S. A. , Choe Y. H. , Park E. , and Hyun Y.-M. Real-time dynamics of neutrophil clustering in response to phototoxicity-induced cell death and tissue damage in mouse ear dermis. Cell Adh. Migr., 12 (5 ):424–431, Sept. 2018.29733749 Parrish J. K. and Edelstein-Keshet L. Complexity, pattern, and evolutionary trade-offs in animal aggregation. Science, 284 (5411 ):99–101, Apr. 1999.10102827 Peiseler M. and Kubes P. More friend than foe: the emerging role of neutrophils in tissue repair. J. Clin. Invest., 129 (7 ):2629–2639, June 2019.31205028 Pettipher E. R. , Salter E. D. , Breslow R. , Raycroft L. , and Showell H. J. Specific inhibition of leukotriene B4 (LTB4)-induced neutrophil emigration by 20-hydroxy LTB4: implications for the regulation of inflammatory responses. Br. J. Pharmacol., 110 (1 ):423–427, Sept. 1993.8220903 Poplimont H. , Georgantzoglou A. , Boulch M. , Walker H. A. , Coombs C. , Papaleonidopoulou F. , and Sarris M. Neutrophil swarming in damaged tissue is orchestrated by connexins and cooperative calcium alarm signals. Curr. Biol., 30 (14 ):2761–2776.e7, July 2020.32502410 Reátegui E. , Jalali F. , Khankhel A. H. , Wong E. , Cho H. , Lee J. , Serhan C. N. , Dalli J. , Elliott H. , and Irimia D. Microscale arrays for the profiling of start and stop signals coordinating human-neutrophil swarming. Nature Biomedical Engineering, 1 (7 ):0094, July 2017. Roxo-Junior P. and Simão H. M. L. Chronic granulomatous disease: why an inflammatory disease? Braz. J. Med. Biol. Res., 47 (11 ):924–928, Nov. 2014.25296353 Shaffer B. M. Secretion of cyclic AMP induced by cyclic AMP in the cellular slime mould dictyostelium discoideum. Nature, 255 :549–552, June 1975.167286 Siwicki M. and Kubes P. Neutrophils in host defense, healing, and hypersensitivity: Dynamic cells within a dynamic host. J. Allergy Clin. Immunol., 151 (3 ):634–655, Mar. 2023.36642653 Skoge M. , Yue H. , Erickstad M. , Bae A. , Levine H. , Groisman A. , Loomis W. F. , and Rappel W.-J. Cellular memory in eukaryotic chemotaxis. Proc. Natl. Acad. Sci. U. S. A., 111 (40 ): 14448–14453, Oct. 2014.25249632 Song Z. , Huang G. , Chiquetto Paracatu L. , Grimes D. , Gu J. , Luke C. J. , Clemens R. A. , and Dinauer M. C. NADPH oxidase controls pulmonary neutrophil infiltration in the response to fungal cell walls by limiting LTB4. Blood, 135 (12 ):891–903, Mar. 2020.31951647 Sun D. and Shi M. Neutrophil swarming toward cryptococcus neoformans is mediated by complement and leukotriene B4. Biochem. Biophys. Res. Commun., 477 (4 ):945–951, Sept. 2016.27402276 Sun Y. , Reid B. , Zhang Y. , Zhu K. , Ferreira F. , Estrada A. , Sun Y. , Draper B. W. , Yue H. , Copos C. , Lin F. , Bernadskaya Y. , Zhao M. , and Mogilner A. Electric field-guided collective motility initiation of large epidermal cell groups. Mol. Biol. Cell, 34 (5 ):ar48, May 2023.36989037 Tweedy L. , Knecht D. A. , Mackay G. M. , and Insall R. H. Self-Generated chemoattractant gradients: Attractant depletion extends the range and robustness of chemotaxis. PLoS Biol., 14 (3 ):e1002404, Mar. 2016.26981861 Uderhardt S. , Martins A. J. , Tsang J. S. , Lämmermann T. , and Germain R. N. Resident macrophages cloak tissue microlesions to prevent Neutrophil-Driven inflammatory damage. Cell, 177 (3 ):541–555.e17, Apr. 2019.30955887 van Oss C. , Panfilov A. V. , Hogeweg P. , Siegert F. , and Weijer C. J. Spatial pattern formation during aggregation of the slime mould dictyostelium discoideum. J. Theor. Biol., 181 (3 ):203–213, Aug. 1996.8869123
PMC010xxxxxx/PMC10327220.txt
==== Front bioRxiv BIORXIV bioRxiv Cold Spring Harbor Laboratory 37425784 10.1101/2023.05.21.541618 preprint 1 Article Network-based approach elucidates critical genes in BRCA subtypes and chemotherapy response in Triple Negative Breast Cancer Agrawal Piyush 1* Jain Navami 2 Gopalan Vishaka 1 Timon Annan 3 Singh Arashdeep 1 Rajagopal Padma S 1 Hannenhalli Sridhar 1* 1 Cancer Data Science Lab, National Cancer Institute, NIH, Bethesda, MD, USA 2 Stanford University, Stanford, CA, USA 3 University of Pennsylvania, Philadelphia, PA, USA * Equal Correspondence Author contribution PA and NJ download and processed the data. VG downloaded and processed the single cell data. PA, AT, AS and SH perform the analysis. PA and SH perform the statistical analysis. PA, PSR and SH wrote the manuscript. PA and SH supervised the study. All authors read the article and approved the submitted version. Corresponding authors Email: Sridhar Hannenhalli: [email protected], Piyush Agrawal: [email protected] 23 5 2023 2023.05.21.541618https://creativecommons.org/licenses/by-nc-nd/4.0/ This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which allows reusers to copy and distribute the material in any medium or format in unadapted form only, for noncommercial purposes only, and only so long as attribution is given to the creator. nihpp-2023.05.21.541618.pdf Breast cancers exhibit substantial transcriptional heterogeneity, posing a significant challenge to the prediction of treatment response and prognostication of outcomes. Especially, translation of TNBC subtypes to the clinic remains a work in progress, in part because of a lack of clear transcriptional signatures distinguishing the subtypes. Our recent network-based approach, PathExt, demonstrates that global transcriptional changes in a disease context are likely mediated by a small number of key genes, and these mediators may better reflect functional or translationally relevant heterogeneity. We apply PathExt to 1059 BRCA tumors and 112 healthy control samples across 4 subtypes to identify frequent, key-mediator genes in each BRCA subtype. Compared to conventional differential expression analysis, PathExt-identified genes (1) exhibit greater concordance across tumors, revealing shared as well as BRCA subtype-specific biological processes, (2) better recapitulate BRCA-associated genes in multiple benchmarks, and (3) exhibit greater dependency scores in BRCA subtype-specific cancer cell lines. Single cell transcriptomes of BRCA subtype tumors reveal a subtype-specific distribution of PathExt-identified genes in multiple cell types from the tumor microenvironment. Application of PathExt to a TNBC chemotherapy response dataset identified TNBC subtype-specific key genes and biological processes associated with resistance. We described putative drugs that target top novel genes potentially mediating drug resistance. Overall, PathExt applied to breast cancer refines previous views of gene expression heterogeneity and identifies potential mediators of TNBC subtypes, including potential therapeutic targets. This work was supported by the Intramural Research Program of the National Cancer Institute. ==== Body pmcIntroduction Breast cancer (BRCA) is one of the leading cancers among women worldwide1. In 2023, 300,500 are expected to be diagnosed and 43,700 are expected to die from the disease [https://www.cancer.org/cancer/breast-cancer/about/how-common-is-breast-cancer.html]. Triple negative breast cancers (TNBC), defined clinically by the lack of estrogen receptor (ER), progesterone receptor (PR), or human epidermal growth factor receptor (HER2) on immunohistochemistry2 are especially aggressive with limited therapeutic options. Gene expression is used, primarily in the translational context, to characterize BRCA subtypes: Luminal A and Luminal B (which overlap with hormone-receptor positive breast cancers), HER2-enriched (which overlaps with HER2-positive tumors), and Basal-like (which overlaps with TNBC)3,4. To identify potential targets for BRCA treatment, previous studies have relied on differentially expressed genes (DEGs). However, individual genes are subject to stochastic variability in gene expression, limiting their reproducibility and consistency5. Furthermore, transcriptional changes are often due to perturbations of key mediators (such as transcription factors, kinases, and other regulatory proteins) in a complex gene regulatory network. Identifying these potential mediators is more likely to provide mechanistic insights and therapeutic targets. Network-based models have been shown to improve potential target identification in various cancers6–9. In contrast with DEGs, our previously developed network-based tool PathExt identifies differentially active pathways in a knowledge-based network and the corresponding central genes in the subnetwork (called TopNet) composed of the differentially active paths. Central genes identified by PathExt offer a more robust and mechanistic view of transcriptional changes across conditions compared to DEGs, explaining differential expression of downstream genes10. We applied PathExt framework to the TCGA BRCA transcriptional data (tumor and normal samples)11 to identify key genes mediating the global transcriptional changes in each BRCA subtype. Against multiple benchmark datasets, PathExt-identified genes substantially outperform DEGs in recapitulating potential driver genes and performs favorably compared to another network-based approach, MOMA12. PathExt-identified genes are further validated by their effect on cellular viability, based on CRISPR knock-out data from the DepMap database13,14, as well as their greater-than-expected mutation frequencies in TCGA samples, in a subtype-specific manner (note that the PathExt relies only on the transcriptome data and does not utilize the mutational data). We further applied PathExt in a clinical trial of neoadjuvant chemotherapy in TNBC to identify key genes associated with treatment-resistant tumors. Lastly, based on computational drug screening, we propose potential therapeutic strategies for targets identified by PathExt. Overall, PathExt refined prior characterization of inter-tumor heterogeneity and identified more consistent genes associated with BRCA subtypes, as well as candidate genes to study resistance to neoadjuvant chemotherapy in TNBC. Results Overview of the workflow The overall workflow of the study is shown in Figure 1. Given a BRCA transcriptomic profile and control samples, as well as a knowledge-based protein interaction network, PathExt10 identifies key genes likely to mediate the observed global transcriptomic changes in the specific sample relative to the control samples. We separately identify genes mediating up-regulation (based on Activated TopNet) as well as down-regulation (based on Repressed TopNet) of gene expression. We applied PathExt to 1059 TCGA BRCA transcriptomic samples across 4 BRCA subtypes using 112 healthy controls to identify top 100 key mediator genes in each sample (Methods). We further identified the top 200 genes most frequently identified across samples in each cancer subtype as a key mediator. As a control, we followed an analogous procedure to identify the top 200 most frequent up and downregulated DEGs in each subtype. Complete lists of genes and their frequencies are provided in the Supplementary Table S1–S4. We performed a series of downstream analyses to assess the relative merits of these identified genes in terms of their functional roles in BRCA. We additionally analyzed a TNBC dataset to identify central genes associated with nonresponse to neoadjuvant chemotherapy in various TNBC subtypes. PathExt reveals putative BRCA subtype-specific and shared mechanisms more consistently than DEGs The following analyses are based on the PathExt-identified top 200 most frequent central genes in Activated and Repressed TopNets for each subtype (Supplementary Table S5) and the analogously calculated top 200 most frequent upregulated and downregulated DEGs (Supplementary Table S6). PathExt genes are much more frequently represented within subtype-specific samples than DEGs, suggesting that PathExt may better reveal shared mechanisms across tumors. For example, in the Basal subtype, top Activated PathExt genes CDC20 and TTK were key genes in 175/177(~99%) tumors, while the most frequently upregulated DEG - CT83 was the top DEG in only 117/177(~66%) tumors. Likewise, in the HER2 subtype, out of 80 samples, PathExt identified ERBB2 (encoding HER2) as the central gene in 42 samples (>50%) whereas it was differentially upregulated in only 11 patients (~15%), underscoring the potential of PathExt in capturing functionally relevant genes. PathExt genes are also much more frequently represented than DEGs when comparing across all breast cancer samples, as shown in Figure 2A for Activated TopNets and Figure 2B for Repressed TopNets. Direct comparison of genes prioritized by PathExt and DEGs also revealed substantial differences (Supplementary Figure S1 A&B). We next identified the enriched biological processes associated with pan-subtype and subtype-specific genes. For the PathExt Activated TopNets, as expected, the common genes were associated mostly with cell cycle (Figure 3A, Supplementary Table S7). However, subtype-specific genes were enriched for distinct processes that are largely supported by literature. For instance, the Basal subtype was associated with cell fate determination, regulation of T-cell proliferation, neuron differentiation, interferon-gamma production, etc.15–18 [Figure 3B]; enriched processes for other subtypes are summarized in Supplementary Figure S2A–B (except LumB, where no enrichment was found), and the complete list of enriched processes is provided in the Supplementary Table S8–S10. Among top genes in Repressed TopNets, most subtypes share similar processes, including neuropeptide signaling pathway, cellular hormone metabolic pathway, muscle cell development, etc. (Figure 3C) whereas unique genes for the Basal subtype were enriched for the processes such as terpenoid metabolic process, liver development, etc. (Figure 3D). For remaining subtypes see Supplementary Figure S2C–E and the complete list of enriched processes are provided in the Supplementary Table S11–S15. In contrast, when analyzing pansubtype and subtype-specific DEGs, distinct, but fewer, sets of processes were enriched. Notably, for many subtypes we didn’t observe any significant enriched processes associated with up and downregulated DEGs; all results are provided in Supplementary Figure S3, Supplementary Table S16–S21. Enriched molecular functions for Activated TopNets of pan-subtype genes include chemokine binding, extracellular matrix structural constituent, and transcription corepressor binding (Figure 3E). Subtype-specific genes (except Her2, where no enrichment was found) also revealed subtype-specific molecular functions supported by literature19–22 (Supplementary Figure S4A–C, Supplementary Table S22–S25). Likewise for the Repressed TopNets, molecular functions enriched among pan-subtype genes include hormone activity, glycosyltransferase activity, and titin binding. (Figure 3F). Enriched functions among subtype-specific PathExt genes (except LumB, where no enrichment was found) included somewhat related functions, e.g., G protein-coupled peptide receptor activity23 (Supplementary Figure S4D–F, Supplementary Table S26–S29). Pan-subtype and subtype-specific DEGs revealed processes only partly overlapping with PathExt (Supplementary Figure S5, Supplementary Table S30–S34). Notably, for many subtypes we didn’t observe any significant enriched molecular functions associated with up and downregulated DEGs. Overall, PathExt reveals greater commonality across samples and across subtypes, and identifies many shared and subtype-specific genes and processes mediating the global transcriptome shifts that are missed by the conventional DEG analysis, underscoring the complementary value of PathExt approach. Next, we assessed the extent to which the subtype-specific genes identified by PathExt have any support for their subtype-specific functionality. We did this with respect to subtype-specific expression and mutational patterns of those genes. First, for every gene, in a given subtype, we computed the log-fold difference of that gene’s expression in the subtype relative to other subtypes. Unsurprisingly, subtype-specific central Activated genes show higher relative gene expression in the particular BRCA subtype, while central Repressed genes exhibit lower relative expression in the BRCA subtype (Supplementary Figure S6). To characterize subtype-specific mutational patterns (which was not used by PathExt), for every gene, we separately computed the frequency of activating (copy number amplification; CNA) and inactivating (copy number loss, nonsense, frameshift indel) mutations in a subtype-specific manner. The mutation frequency was normalized by average frequency across all genes in a given BRCA subtype and further normalized by the same quantity for the other three subtypes to get a subtype-specific relative log-normalized mutation frequency. While the signals were weak owing to the sparsity of mutations, we broadly observed a subtype-specific enrichment of activating mutations in subtype-specific genes (both from Activated and Repressed TopNets) (Figure 4A) and a depletion of inactivating mutations (Figure 4B). Supported by previously reported dominance of CNAs in BRCA genomic landscape24, our results link CNAs to key genes affecting global transcriptome in a BRCA subtype-specific fashion. Overall, our analyses support the role of key genes identified by PathExt in BRCA subtype-specific pathogenesis. PathExt-identified genes perform better than DEGs in cell line, functional, and benchmarking tests The key feature of PathExt is identification of genes that may not be overtly differentially expressed but may play a central role in mediating the tumor phenotype. First, we assess the extent to which subtype-specific PathExt genes exhibit subtype-specific function and clinical relevance. To do this, we utilized the publicly available DepMap dataset reporting cellular viability upon genome-wide CRISPR-Cas9 Knock Out (KO) across hundreds of cell lines, providing a dependency probability score for each gene in each cell line, where a higher score indicates greater essentiality or dependency25; by default, a score >0.5 indicates dependency. Critically, one can obtain dependency for each gene in BRCA subtype-specific cell lines. For each gene-subtype pair, we computed the gene’s average dependency probability scores across the cell lines derived from the specific BRCA subtype. We then compared the score distribution for PathExt genes with that for DEGs. As shown in Figure 5A, PathExt Activated genes exhibited significantly greater dependency in subtype-specific cell lines compared to upregulated DEGs. Interestingly, gene dependency of the Repressed TopNets and downregulated DEGs were very low suggesting that these genes do not affect cell survival. Our conclusions drawn from Figure 5A do not change when we compare the fraction of essential genes (dependency probability score > 0.5) between PathExt and DEGs based on Fisher’s test (Supplementary Figure S7). Depmap gene dependency probability value for the central genes (when available) of PathExt and DEGs in various subtypes is provided in the Supplementary Table S35–38. Next, we assessed how significantly PathExt and DEGs recapitulated previously compiled driver genes in three 3 publicly available datasets: (i) DriverDBv326 (including 313 BRCA driver genes) (ii) IntoGen27 (including 99 BRCA driver genes) and (iii) Huang et.al. dataset28 (including 500 BRCA driver genes). For each benchmark gene set, we assessed the significance of overlap (using Fisher Exact test) with the top 200 PathExt genes or DEGs in Activated or Repressed scenarios. As shown in Figure 5B, in all benchmarks, PathExt Activated TopNet central genes show significant overlap with the BRCA drivers, while upregulated DEGs do not. Integrating protein interaction networks and tumor mutation profiles, Zheng et al. have reported a map of 395 protein systems (NEST) which are recurrently mutated in one or more cancer types likely under somatic mutation selection29. PathExt genes are associated significantly with several NEST systems, including Cell cycle, Nucleosome and Ribosome, Cytoplasm and Extracellular space, and Signaling Systems. More specifically, PathExt Activated TopNet central genes from all four subtypes showed significant overlap with NEST genes while central genes from Repressed TopNets showed significant overlap for subtypes Her2 and LumA. In sharp contrast, none of the DEGs (up or downregulated) from any of the subtypes showed significant overlap with the NEST systems (Figure 5C). Next, we compiled an additional translational benchmark gene set comprising 474 unique target genes of 46 FDA-approved BRCA drugs from CancerDrugs_DB30. Again, we found that these genes exhibit significantly higher overlap with the top 200 Activated PathExt genes compared to DEGs (Figure 5D). Finally, we compared PathExt with MOMA (Multi-omics Master-Regulator Analysis)12, another network-based approach which integrated transcription and mutational data in the context of inferred regulatory networks, and previously reported 407 master regulator (MR) proteins across 20 TCGA cohorts. These MR proteins are further grouped into 24 pan-cancer master regulator blocks. Specifically, the study has reported 39, 92 and 23 MR proteins for Basal, LumA and LumB subtypes, respectively. We compared the overlap of MOMA MR proteins with the DriverDBv3 driver cancer gene dataset and NEST PPI dataset with that of PathExt identified TopNet genes. For a fair comparison, we selected the same number of genes (based on frequency) which MOMA had for each subtype. Based on the Fisher test, PathExt genes exhibit a comparable overlap with the benchmark sets relative to MOMA (Figure 5E). Overall, PathExt identified genes shows greater cell dependency, recapitulates previously identified cancer drivers, BRCA specific mutated modules, and BRCA FDA approved drug targets far more effectively than the conventional DEG approach and performs favorably to a recent network based multi-omics approach MOMA. PathExt leverages single-cell data to identify cell-specific mediators in tumor micro-environment Breast cancer heterogeneity is attributable to both transcriptional variation in cancer cells as well as variation in cellular composition in the tumor microenvironment. We therefore analyzed BRCA subtype-specific single cell transcriptomic data by Qian et. al.31 to chart the distribution of expression of the top 200 PathExt-identified genes across different cell types in the tumor microenvironment (Methods). Within each subtype-specific dataset we first identified in each cell type the genes that are expressed in that cell type with an above-mean expression; note that a gene can be expressed in multiple cell types, but not all. We then examined, in a subtype-specific manner, the fractions of PathExt genes expressed in each cell type relative to background expectation across all genes. In the Qian et. al. dataset, the top genes from the Activated TopNets are expressed most frequently in the malignant cells, except in LumA, where top genes are most enriched in myeloid cells and depleted in malignant cells (Figure 6A). Additional cell types show enriched expression for top activated genes as well -- T cells in TNBC, myeloid cells in Her2, and B cells in LumB. Repressed central genes show similar broad trends except that in three subtypes central repressed genes are also enriched in fibroblasts, consistent with the enrichment of extracellular matrix processes among these genes (Figure 6B). PathExt identifies potential mediators of resistance to Neoadjuvant Doxorubicin/Cyclophosphamide followed by Ixabepilone/Paclitaxel in TNBC TNBC is the most aggressive BRCA subtype and presents a major therapeutic challenge. Although TNBC is a clinically defined subtype, it substantially overlaps with the Basal subtype defined based on transcriptional profile. Gene expression has been used to further classify TNBC breast cancers32. Here, we aim to apply PathExt to identify key genes potentially mediating therapeutic resistance. We used publicly available data from a phase II trial of neoadjuvant doxorubicin/cyclophosphamide followed by ixabepilone/paclitaxel therapy, where pathologic complete response (PCR) was the measure of response33. We specifically focused on the 138 TNBC samples classified into 4 subtypes (BL1–47, BL2–27, LAR-21, and M-43), further classified as responders or non-responders. We applied PathExt to identify central genes (Activated & Repressed) associated with non-responders (in each sample independently), using samples from responders as the control, and ranked genes based on number of samples of a specific TNBC subtype in which they were detected among the top 100 central genes. Complete lists of PathExt identified TNBC subtype-specific genes and their frequencies in Activated and Repressed TopNets associated with non-responders are provided in the Supplementary Table S39 & S40 respectively. First, we selected the top 20 most frequent central genes in Activated TopNets from each TNBC subtype yielding 60 non-redundant genes, classified in 5 parts: Common - the genes present in at least 2 of the 4 TNBC subtypes among the top 20 genes, and 4 gene sets comprising genes uniquely found in a subtype among the top 20. Figure 7 shows the subtype-specific frequencies of these genes; note that a gene, say TGFB1, may be among the top 20 uniquely in BL1 and yet may have a high frequency (although not among top 20) in another subtype (M). As can be seen, most subtypes show relatively high specificity among the top genes. Supplementary Figure S8 shows a similar figure for top 20 central genes in Repressed TopNets in each subtype. Next, we examined the top 100 most frequent central genes from each subtype and identified genes common among all subtypes and unique to a given subtype. For the activated TopNets, 5 genes -- FOXA1, CTNNB1, JUN, FOS and ALB were found in all TNBC subtypes and 49, 55, 61 and 48 genes were unique to BL1, BL2, LAR and M subtypes respectively. Common genes were broadly associated with developmental processes (Figure 8A & Supplementary Table S41); BL1 subtype was enriched mainly for hemostasis and coagulation processes34–36; BL2 subtype for chemotaxis and signaling processes37,38; LAR subtype for metabolic process39,40 and lastly M subtype was associated with modification and nuclear division processes41–43 (Figure 8B–E). Detailed discussion of the processes associated with TNBC subtype-specific mediators of resistance, and their functional relevance is discussed in Supplementary Note 1. Complete lists of the enriched common and unique processes in each subtype are provided in the Supplementary Table S42–S45. Likewise, in the case of Repressed TopNets, 6 genes -- STAT3, TGFB1, TNF, EP300, TP53 and FOXA1 were common in all the subtypes and were mainly associated with the transcription associated processes. 61 genes were unique in the BL1 subtype and were associated with the immune system and function. BL2 with 45 unique genes were enriched for lipid related processes; LAR with 40 unique genes were associated with signaling processes and lastly M subtype with 55 unique genes were enriched with cell division processes. Complete lists of the enriched common and unique processes in each subtype for Repressed TopNets are provided in the Supplementary Table S46–S50 & Supplementary Figure S9A–E. Overall, these results reveal genes potentially mediating resistance across all TNBC subtypes and also distinct processes mediating resistance in each subtype. Next, we assessed the extent to which PathExt-identified genes can help differentiate responders from non-responders. For this, we identified 21 genes common among the top 200 Activated genes in all four TNBC subtypes; for comparison, we analogously selected 13 upregulated DEGs (Supplementary Table S51). We plotted these genes’ expression to see if they can discriminate responders and non-responders in an independent dataset by Stickeler et.al (GSE21974)44. As shown in Figure 9A, PathExtidentified genes’ expression discriminate responders and non-responders with significant p-value of 0.028, however, DEGs failed to discriminate between responders and non-responders (p-value = 0.61) (Figure 9B). These results underscore the generalizability of PathExt-identified genes across cohorts. As a community resource, we perform drug repurposing analysis and provide potential drugs targeting the genes potentially mediating resistance in TNBC as currently only poly (ADP-ribose) polymerase (PARP) inhibitors and immune checkpoint inhibitors have been approved for TNBC treatment45. We selected top100 most frequent central genes present in the Activating TopNets of TNBC subtypes and selected the 28 genes present in at least 3 TNBC subtypes, of which 3 were excluded due to unsuitability for docking due to their structural properties. Therefore, we proceeded with 25 genes for virtual drug screening. Based on the drug repurposing tool CLUE46 (https://clue.io/repurposing-app), 16 of the 25 genes are targeted by at least one approved drug. Table 1 shows the targets for which at least one drug is mapped in the CLUE data (we have provided up to 3 approved drugs mapped to each target). For the remaining 9 targets, we performed virtual screening using FDA approved drugs (Methods; Supplementary Table S52) and provided top 3 mapped drugs for each gene in Supplementary Table S53. Supplementary Figure S10 provides an example of docking results for one of the target genes. Discussion Tumor heterogeneity in breast cancer, particularly TNBC, remains an ongoing hurdle for identifying therapeutic targets with broad applicability. By identifying central mediators of gene expression changes, PathExt is designed to refine expression variance and can be used in multiple translational contexts. PathExt achieves its goal by first identifying critical pathways in a pre-defined knowledge-based gene network, i.e., those that exhibit significantly different activity in a specific transcriptomic sample compared to controls, then identifying key mediators of the critical pathways. In doing so PathExt relies less on differential expression of individual genes (which are highly variable47) and more on network relationships. Unlike differential expression, which relies on sufficient sampling, PathExt can be applied to a single sample, making it applicable for clinical scenarios where large cohorts are rarer. By identifying key genes in a sample-wise fashion (as opposed to a cohort for differential expression), PathExt directly accounts for inter-sample heterogeneity. By focusing on genes mediating the global transcriptome, PathExt shows far greater commonality across BRCA subtypes across multiple benchmark sets compared to DEGs. PathExt also performs favorably to a previous network-based approach - MOMA. We note that the central genes mediating the Activated TopNet (comprising paths with significantly higher activity in the condition of interest relative to control) are not necessarily differentially upregulated (Supplementary Figure S11). Likewise, central genes mediating the Repressed TopNet (comprising paths with significantly lower activity in the condition) are not necessarily differentially downregulated (Supplementary Figure S12); in particular, many of these genes may indeed have repressive effects on other genes while themselves being upregulated. This explains the somewhat counterintuitive observation that the key genes in both the Activated and the Repressed TopNets exhibit elevated inactivating mutations in cancer (Figure 4B). Furthermore, the cell types expressing the genes identified based on bulk sequencing is not immediately clear. Our single-cell analysis of PathExt-identified genes points to the role of tumor microenvironment in oncogenesis, where not only the malignant but various immune compartments may play a key role in mediating the global gene expression, as profiled by bulk sequencing. As shown above PathExt recapitulates the genes previously associated with BRCA more frequently compared to DEGs. For instance, ERBB2 gene (which encodes for HER2) was identified as a top central gene in over 50% (42/80) of the patients by PathExt compared to fewer than 15% patients (11/80) by DEG. Besides revealing an overall greater commonality in the key genes across samples and BRCA subtypes, PathExt also reveals shared genes and pathways between specific BRCA subtypes that are consistent with known BRCA subtype-specific biology recently reviewed by Nolan et.al.24. For instance, Basal and Her2 like-tumors show low and medium expression of LumA signatures respectively, and we observed that top10 most frequent LumA genes (except TTK, NEK2 and BIRC5) were among the top genes in very few Basal tumors and in nearly half of the Her2 tumors. Equally importantly, PathExt contributes to the currently limited knowledge of subtype-specific keys genes and their associated biological processes. For instance, Tan et.al.18 recently showed upregulation of neuronal genes preferentially in TNBC associated with neural crest and glial development. Encouragingly, PathExt identified cell-fate determination and regulation of T-cell proliferation among the top enriched processes uniquely in basal subtype (which substantially overlaps with TNBC). Likewise, Hartman et. al.48 have shown high expression of HER2 leads to secretion of interleukin-6 (IL6) which further activates STAT3 ultimately contributing to tumorigenesis. Uniquely in Her2 BRCA, PathExt reveals biological processes associated with IL6 response and inflammation. Consistent with reports linking aging with luminal breast cancer49–51, PathExt identified “aging” as the top enriched process specifically in LumA BRCA. Focusing on subtype-specific identified genes involved in subtype-specific enriched biological processes, and further filtering them based on gene co-expression network52 (https://coxpresdb.jp/), we identified core subtype-specific genes worthy of future follow up. These include MYC, MAX, E2F2 and IL2RA in Basal; JAK2, ERBB2, and MAPK9 in Her2, and MMP2, FGFR3, F2R, SHH and ADCY1 in LumA subtypes, most of which have known associations with specific BRCA subtypes. PathExt application to TNBC response revealed 5 genes -- FOXA1, CTNNB1, JUN, FOS and ALB associated with non-responsive Activated TopNets in all four TNBC subtypes. These genes were broadly enriched for development, signaling, and transcription. One of the top enriched processes we observed associated with resistance was ‘cell fate determination’ which was also reviewed by O’Reilly et.al. as one of the factors associated with chemoresistance in TNBC15. FOXA1 upregulation in basal-like cell line MDA-MB-231 leads to increased drug resistance53. Likewise, CTNNB1 is associated with Wnt signaling pathways, known to be associated with BRCA54. There is literature support for other genes as well55,56. For the Repressed TopNets, 6 genes were common among all the subtypes, viz. STAT3, TGFB1, TNF, EP300, TP53 and FOXA1. Interestingly, FOXA1 is revealed as a key gene in both Activated and Repressed TopNets across all subtypes. Dai et.al. have shown that downregulation of FOXA1 leads to increased malignancy and cancer stemness by suppressing SOD2 and IL657, underscoring a pleiotropic role of FOXA1. Another resistance-associated gene PathExt revealed is EP300, a known modulator of paclitaxel resistance and stemness58. Interestingly, the study we used for our analysis included Paclitaxel treatment after neoadjuvant chemotherapy. Overall, PathExt is complementary to the conventional DEG approach and reveals common and BRCA subtype-specific key genes and processes, as well as gene mediating chemotherapy response in TNBC. Lastly, as a community resource, we have provided potential drugs that are either approved or undergoing trials targeting novel key genes revealed by our analyses. Methods Data Collection and processing Transcriptomic profiles for 1059 primary BRCA tumors and 112 normal adjacent samples were downloaded from TCGA59. Gene expression was jointly quantile normalized and was used as an input for the PathExt60. We also analyzed the TNBC transcriptomic dataset by Horak et.al.33 where they look for the biomarkers associated with the response to the neoadjuvant doxorubicin/cyclophosphamide followed by Ixabepilone. In a follow up study61, Lehmann et.al. classified 138 of the above TNBC samples into 4 TNBC subtypes (see61 for classification details) - BL1 (47 samples), BL2 (27 samples), LAR (21 samples), and M (43 samples). We applied PathExt to each TNBC subtype separately to identify central genes mediating resistance (non-response). Identification of sample-specific central genes using PathExt PathExt method has been described in detail in our previous publications10,60. Here we provide a very brief sketch. Based on a set of control transcription profiles (e.g., normal adjacent breast samples for a BRCA sample), we first compute weight for each node in a given knowledge-based protein interaction network, such that the node weight represents the extent of upregulation (Activated) or downregulation (Repressed) of the gene in the foreground sample relative to the control. PathExt then identifies significantly perturbed paths in the network and finally in the subnetwork composed of such significantly perturbed paths (Activated or Repressed TopNet), it then identifies the top central genes based on betweenness centrality measure62. These central genes represent key mediators of global transcription changes (upregulation for Activated TopNet and downregulation for Repressed TopNet) in a particular sample. Having computed the top 100 genes for each BRCA sample we compute the number of sample (of a given BRCA subtype) in which a gene was among the central genes, and then selected the top 200 most frequent genes to represent subtype-specific key genes; this was done independently for Activated and Repressed TopNets. For comparison, based on log-fold change for each gene, we selected top 100 up and downregulated genes for each sample and further based on frequency we selected top 200 genes (up and downregulated) from each subtype. Analogous procedure was applied to identify key genes corresponding to non-responders relative to responders among the TNBC BRCA cohort. Gene Ontology Enrichment Analysis To identify enriched biological processes and molecular functions in various subtypes, we used the Clusterprofiler 4.0 package63, where the foreground was the top200 central genes (PathExt or DEGs) and the background was the default background used by the package. The database used was the human database and the minimum gene size and maximum gene size was set as 10 and 200 respectively to ensure specific terms. We then used the function “simplify” with cutoff value 0.8 to remove the redundant terms. Default parameters were used otherwise. In the case of TNBC responder and non-responders, we used similar parameters except foreground and background gene list. In this case, the foreground was top100 genes and the background was a customized list of 10994 genes, those that were detected in the cohort [Supplementary Table S54] Cell-line specific genetic dependency From the DepMap database13, we obtained the gene dependency probability scores for 16708 genes in 44 breast cancer cell lines, further categorized into BRCA subtypes. For each gene, we then obtained the average dependency probability score across the cell lines in a subtype-specific manner. As recommended, genes with average dependency probability score >= 0.5 were recorded as essential for comparing PathExt and DEG genes. Comparison of PathExt and DEGs genes with benchmarks gene sets We compiled various previously published datasets of potential BRCA driver genes, approved FDA drugs for BRCA and assessed their overlap with the PathExt central genes or DEGs. We used Fisher exact test to assess the statistical significance of overlap and reported Odds Ratio (>1 indicates greater than expected overlap). Breast cancer subtype single cell data analysis We used a single cell RNA-seq dataset for the 4 BRCA subtypes by Qian et.al.31. Read count matrices of scRNA-seq data (obtained using 10X v2 sequencing) breast cancer tumors were downloaded from http://blueprint.lambrechtslab.org. Author-supplied annotations were used to label each cell. The miQC package64 was used to remove dead cells, with a probability threshold of 0.5 (i.e., a probability of at least 0.5 that the cell is not dead) used to retain high-quality cells for downstream analyses. After filtering the dataset, we retained only those cell types having a minimum of 50 cells. This provided 7 cell type data for Basal and Her2 subtypes, and 4 cell type data for LumA & LumB subtypes. Next, we computed the z-score of gene expression for the PathExt top200 central genes in each cell (using all cells to estimate the mean and the standard deviation) and then took the average z-score value of the cells in a given cell type. Next, we selected the genes with z-score > 0 in each cell type and were considered as ‘expressed’ in that cell type. Next, we computed the fraction (%) of such expressed genes in every cell type and normalized the fraction across all cell types to get ‘Observed frequency’ in a cell type. We also computed the ‘Expected frequency’ in a similar manner by considering all genes. Finally, Log(Observed/Expected) value was computed for the central genes across cell types. Identifying key genes associated with neoadjuvant treatment in TNBC subtypes PathExt was applied to a previously published cohort of TNBC tumors having undergone neoadjuvant therapy with Doxorubicin/Cyclophosphamide followed by Ixabepilone or Paclitaxel drug, and further classified into responders and non-responders33 where responders include patients with complete or partial response, whereas non-responders include patients with stable or progressive disease. TNBC is further subdivided into 4 subtypes, BL1, BL2, LAR and M61, and hence, we applied PathExt in a subtype-specific manner to identify central genes among nonresponders relative to responders. Drug repurposing study We selected the top 100 most frequent central genes present in Activated TopNets from each TNBC subtype and then further selected genes present in at least 3 subtypes. This resulted in 29 genes out of which 3 genes were not suitable for analysis because of their structural properties. Based on the CMap database (using CLUE platform)65, we first identified the drugs targeting these genes. We consider only those drugs which are either approved or undergoing trials. For the remaining unmapped targets, we performed virtual screening using drugs which are either FDA approved or are under clinical trials. The 3D structures of the targets were downloaded from the RCSB-PDB66 and further refined using OpenBabel software67. Next, we used online server DrugRep68 to locate various binding pockets on the protein surface and selected the pocket with the largest volume covering most active site residues. Identification of active sites and residues facilitate the docking experiment and help in identifying drugs with improved binding. Virtual Screening was performed using Autodock Vina software69. The software requires a receptor and ligand file in a specific file format (“pdbqt” file format), which was prepared using MGLTools software70. Post virtual screening, based on docking free energy and root mean square deviation (RMSD), top 3 drugs for each target were proposed. Further, experimental work is required to 66prove the efficacy of these drugs. Supplementary Material Supplement 1 Supplement 2 Acknowledgement This work utilized the computational resources of the NIH HPC Biowulf cluster. Funding This work was supported by the Intramural Research Program of the National Cancer Institute. Data Availability Statement The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding authors. Figure 1: Architecture of the study. PathExt was implemented on the TCGA-BRCA dataset to identify subtype-specific most frequent central mediating genes as well as DEGs for comparison. Central genes were used for various analyses such as identifying enriched biological processes, genetic dependencies of the cell, identifying cell types mediating global transcriptomic response in single cell data, recapitulating breast cancer driver genes, FDA approved breast cancer targets, comparison with previous network-based methods and DEGs. PathExt was also implemented on the TNBC dataset to identify key genes associated with non-response to neoadjuvant chemotherapy treatment in subtype-specific manner. Figure 2: Venn diagram showing gene overlap among central genes in various subtypes identified in (A) PathExt Activated TopNets and Upregulated DEGs and for (B) PathExt Repressed TopNets and Downregulated DEGs. Figure 3: Top central PathExt central genes are associated with essential biological processes and molecular functions. Top20 enriched biological processes associated with activated common central genes among all the 4 BRCA subtypes (A); Top20 enriched biological processes associated with activated genes uniquely in Basal (B); Top20 enriched biological processes associated with repressed common central genes among all the 4 BRCA subtypes (C); and Top20 enriched biological processes associated with repressed genes uniquely in Basal (D); Top20 enriched molecular functions associated with activated common central genes among all the 4 BRCA subtypes (E); and, Top20 enriched molecular functions associated with repressed common central genes among all the 4 BRCA subtypes (F). Figure 4: PathExt reveals subtype specific mutational properties. Boxplot representation of subtype-specific PathExt Activated and Repressed TopNets unique gene CNV amplification distribution (A); and subtype-specific PathExt Activated and Repressed TopNets unique gene inactivating mutation distribution (B). Figure 6: Evaluation of PathExt approach with DEGs and MOMA. Boxplot representation of gene dependency probability distribution by PathExt Activated TopNet genes and Upregulated DEGs and PathExt Repressed TopNet genes and Downregulated DEGs in various BRCA subtypes (A); Fishers Odds Ratio (OR) showing that PathExt recapitulated BRCA specific driver genes from multiple datasets (B); NEST PPI mutated modules comprising a number of genes affecting various mechanisms (C); targets associated with FDA approved BRCA drugs compared to DEGs with high statistical significance (D). PathExt showed comparable performance with MOMA (another network-based method) on various datasets (E). Figure 6: Gene expression distribution of top200 PathExt (A) Activated central genes and (B) Repressed central genes in individual cell types in each BRCA subtype. Within each subtype-specific scRNA-seq dataset, we first identified in each cell type the genes that are expressed in that cell type with an above-mean expression. We then examined the fractions of PathExt genes expressed in each cell type (Observed Frequency) relative to background expectation across all genes (Expected Frequency). Finally, Log (Observed/Expected) value was computed for the central genes across cell types. Figure 7: Top PathExt Activated unique and common gene fraction distribution in various TNBC subtypes. Top20 most frequent genes were selected from each TNBC subtype, from which pan-subtype common genes (top20 in at least two subtypes) and subtype-specific unique genes were identified. The heat plot shows, in each subtype, the fraction of samples in which the gene was among the top 20 central genes. Figure 8: Top central PathExt Activated central genes are associated with essential biological processes in TNBC non-responders. Top20 enriched biological processes associated with common central genes among all the 4 TNBC subtypes (A). Top20 enriched biological processes associated with unique BL1 (B); BL2 (C); LAR (D); and M (E) subtype central genes. Figure 9: PathExt central genes discriminate responder-nonresponder significantly compared to DEGs. Average gene expression of 21 central genes from PathExt Activated TopNets and 13 upregulated DEGs were computed in each sample in an independent dataset. Boxplot show that PathExt discriminate responder and nonresponder significantly (Mann Whitney Test) and DEGs do not. Table 1: List of the targets from the Activated TopNets and the drugs mapped in CLUE database. This table comprises of those targets for which at least one drug was mapped in CLUE database. Maximum of 3 drugs are provided for each target gene. Target Top 3 potential drug ALB Erythromycin-estolate, Erythromycin-ethylsuccinate, Iodipamide CALM1 Chlorpromazine, Trifluoperazine, Loperamide EGFR Brigatinib, Olmutinib, Gefitinib ESR1 Hexestrol, Danazol, Gestrinone JUN Ephedrine-(racemic), Irbesartan, Vinblastine ADCY2 Forskolin APP Curcumin CCND1 Arsenic-trioxide CCR5 Maraviroc CREB1 Adenosine-phosphate, Naloxone CTNNB1 Urea EP300 Curcumin FOS Ephedrine-(racemic) IL6 Ibudilast STAT3 Acitretin, Niclosamide Competing Interests The authors declare no competing interests. ==== Refs References 1. Siegel R.L. , Miller K.D. , Wagle N.S. , and Jemal A. (2023). Cancer statistics, 2023. CA Cancer J Clin 73 , 17–48. 10.3322/CAAC.21763.36633525 2. Onitilo A.A. , Engel J.M. , Greenlee R.T. , and Mukesh B.N. Breast Cancer Subtypes Based on ER/PR and Her2 Expression: Comparison of Clinicopathologic Features and Survival. Original Research Clinical Medicine & Research 7 , 4–13. 10.3121/cmr.2009.825. 3. Bernard P.S. , Parker J.S. , Mullins M. , Cheung M.C.U. , Leung S. , Voduc D. , Vickery T. , Davies S. , Fauron C. , He X. , (2009). Supervised Risk Predictor of Breast Cancer Based on Intrinsic Subtypes. Journal of Clinical Oncology 27 , 1160. 10.1200/JCO.2008.18.1370.19204204 4. Bastien R.R. , Rodríguez-Lescure Á. , Ebbert M.T. , Prat A. , Munárriz B. , Rowe L. , Miller P. , Ruiz-Borrego M. , Anderson D. , Lyons B. , (2012). PAM50 breast cancer subtyping by RT-qPCR and concordance with standard clinical molecular markers. BMC Med Genomics 5 . 10.1186/1755-8794-5-44. 5. Cui W. , Xue H. , Wei L. , Jin J. , Tian X. , and Wang Q. (2021). High heterogeneity undermines generalization of differential expression results in RNA-Seq analysis. Hum Genomics 15 . 10.1186/S40246-021-00308-5. 6. Pe’er D. , and Hacohen N. (2011). Principles and strategies for developing network models in cancer. Cell 144 , 864–873. 10.1016/J.CELL.2011.03.001.21414479 7. Pai S. (2022). Network Approaches for Precision Oncology. Adv Exp Med Biol 1361 , 199–213. 10.1007/978-3-030-91836-1_11.35230690 8. Mundi P.S. , Dela Cruz F.S. , Grunn A. , Diolaiti D. , Mauguen A. , Rainey A.R. , Guillan K. , Siddiquee A. , You D. , Realubit R. , (2023). A transcriptome-based precision oncology platform for patient-therapy alignment in a diverse set of treatment resistant malignancies. Cancer Discov, OF1–OF22. 10.1158/2159-8290.CD-22-1020. 9. Hofree M. , Shen J.P. , Carter H. , Gross A. , and Ideker T. (2013). Network-based stratification of tumor mutations. Nat Methods 10 , 1108–1118. 10.1038/NMETH.2651.24037242 10. Sambaturu N. , Pusadkar V. , Hannenhalli S. , and Chandra N. (2021). PathExt: a general framework for path-based mining of omics-integrated biological networks. Bioinformatics 37 , 1254–1262. 10.1093/BIOINFORMATICS/BTAA941.33305329 11. Koboldt D.C. , Fulton R.S. , McLellan M.D. , Schmidt H. , Kalicki-Veizer J. , McMichael J.F. , Fulton L.L. , Dooling D.J. , Ding L. , Mardis E.R. , (2012). Comprehensive molecular portraits of human breast tumours. Nature 490 , 61–70. 10.1038/NATURE11412.23000897 12. Paull E.O. , Aytes A. , Jones S.J. , Subramaniam P.S. , Giorgi F.M. , Douglass E.F. , Tagore S. , Chu B. , Vasciaveo A. , Zheng S. , (2021). A modular master regulator landscape controls cancer transcriptional identity. Cell 184 , 334–351.e20. 10.1016/J.CELL.2020.11.045.33434495 13. Tsherniak A. , Vazquez F. , Montgomery P.G. , Weir B.A. , Kryukov G. , Cowley G.S. , Gill S. , Harrington W.F. , Pantel S. , Krill-Burger J.M. , (2017). Defining a Cancer Dependency Map. Cell 170 , 564–576.e16. 10.1016/J.CELL.2017.06.010.28753430 14. Chiu Y.C. , Zheng S. , Wang L.J. , Iskra B.S. , Rao M.K. , Houghton P.J. , Huang Y. , and Chen Y. (2021). Predicting and characterizing a cancer dependency map of tumors with deep learning. Sci Adv 7 . 10.1126/SCIADV.ABH1275. 15. O’Reilly E.A. , Gubbins L. , Sharma S. , Tully R. , Guang M.H.Z. , Weiner-Gorzel K. , McCaffrey J. , Harrison M. , Furlong F. , Kell M. , (2015). The fate of chemoresistance in triple negative breast cancer (TNBC). BBA Clin 3 , 257–275. 10.1016/J.BBACLI.2015.03.003.26676166 16. Oshi M. , Asaoka M. , Tokumaru Y. , Yan L. , Matsuyama R. , Ishikawa T. , Endo I. , and Takabe K. (2020). CD8 T Cell Score as a Prognostic Biomarker for Triple Negative Breast Cancer. Int J Mol Sci 21 , 1–16. 10.3390/IJMS21186968. 17. Singh S. , Kumar S. , Srivastava R.K. , Nandi A. , Thacker G. , Murali H. , Kim S. , Baldeon M. , Tobias J. , Blanco M.A. , (2020). Loss of ELF5-FBXW7 stabilizes IFNGR1 to promote the growth and metastasis of triple-negative breast cancer through interferon-γ signalling. Nat Cell Biol 22 , 591–602. 10.1038/S41556-020-0495-Y.32284542 18. Tan R. , Li H. , Huang Z. , Zhou Y. , Tao M. , Gao X. , and Xu Y. (2020). Neural Functions Play Different Roles in Triple Negative Breast Cancer (TNBC) and non-TNBC. Sci Rep 10 . 10.1038/S41598-020-60030-5. 19. Perreault A.A. , Sprunger D.M. , and Venters B.J. (2019). Epigenetic and transcriptional profiling of triple negative breast cancer. Sci Data 6 . 10.1038/SDATA.2019.33. 20. Fiscon G. , Pegoraro S. , Conte F. , Manfioletti G. , and Paci P. (2021). Gene network analysis using SWIM reveals interplay between the transcription factor-encoding genes HMGA1, FOXM1, and MYBL2 in triple-negative breast cancer. FEBS Lett 595 , 1569– 1586. 10.1002/1873-3468.14085.33835503 21. Gudjonsson T. , Rønnov-Jessen L. , Villadsen R. , Rank F. , Bissell M.J. , and Petersen O.W. (2002). Normal and tumor-derived myoepithelial cells differ in their ability to interact with luminal breast epithelial cells for polarity and basement membrane deposition. J Cell Sci 115 , 39–50. 10.1242/JCS.115.1.39.11801722 22. Hu Y. , Xie Q. , Wu X. , Liu W. , Li D.F. , Li C. , Zhao W.X. , Chen L.L. , Zheng Z. , Li G.M. , (2022). Tension of plus-end tracking protein Clip170 confers directionality and aggressiveness during breast cancer migration. Cell Death Dis 13 . 10.1038/S41419022-05306-6. 23. Asano S. , Ono A. , Sakamoto K. , Hayata-Takano A. , Nakazawa T. , Tanimoto K. , Hashimoto H. , and Ago Y. (2023). Vasoactive intestinal peptide receptor 2 signaling promotes breast cancer cell proliferation by enhancing the ERK pathway. Peptides (N.Y.) 161 . 10.1016/J.PEPTIDES.2023.170940. 24. Nolan E. , Lindeman G.J. , and Visvader J.E. (2023). Deciphering breast cancer: from biology to the clinic. Cell 186 . 10.1016/J.CELL.2023.01.040. 25. Chiu Y.C. , Zheng S. , Wang L.J. , Iskra B.S. , Rao M.K. , Houghton P.J. , Huang Y. , and Chen Y. (2021). Predicting and characterizing a cancer dependency map of tumors with deep learning. Sci Adv 7 . 10.1126/SCIADV.ABH1275. 26. Liu S.H. , Shen P.C. , Chen C.Y. , Hsu A.N. , Cho Y.C. , Lai Y.L. , Chen F.H. , Li C.Y. , Wang S.C. , Chen M. , (2020). DriverDBv3: a multi-omics database for cancer driver gene research. Nucleic Acids Res 48 , D863–D870. 10.1093/NAR/GKZ964.31701128 27. Martínez-Jiménez F. , Muiños F. , Sentís I. , Deu-Pons J. , Reyes-Salazar I. , Arnedo-Pac C. , Mularoni L. , Pich O. , Bonet J. , Kranas H. , (2020). A compendium of mutational cancer driver genes. Nat Rev Cancer 20 , 555–572. 10.1038/S41568-020-0290X.32778778 28. Xi J. , Yuan X. , Wang M. , Li A. , Li X. , and Huang Q. (2020). Inferring subgroupspecific driver genes from heterogeneous cancer samples via subspace learning with subgroup indication. Bioinformatics 36 , 1855–1863. 10.1093/BIOINFORMATICS/BTZ793.31626284 29. Zheng F. , Kelly M.R. , Ramms D.J. , Heintschel M.L. , Tao K. , Tutuncuoglu B. , Lee J.J. , Ono K. , Foussard H. , Chen M. , (2021). Interpretation of cancer mutations using a multiscale map of protein systems. Science 374 . 10.1126/SCIENCE.ABF3067. 30. Pantziarka P. , Capistrano I R. , De Potter A. , Vandeborne L. , and Bouche G. (2021). An Open Access Database of Licensed Cancer Drugs. Front Pharmacol 12 . 10.3389/FPHAR.2021.627574. 31. Qian J. , Olbrecht S. , Boeckx B. , Vos H. , Laoui D. , Etlioglu E. , Wauters E. , Pomella V. , Verbandt S. , Busschaert P. , (2020). A pan-cancer blueprint of the heterogeneous tumor microenvironment revealed by single-cell profiling. Cell Res 30 , 745–762. 10.1038/S41422-020-0355-0.32561858 32. Lehmann B.D. , Bauer J.A. , Chen X. , Sanders M.E. , Chakravarthy A.B. , Shyr Y. , and Pietenpol J.A. (2011). Identification of human triple-negative breast cancer subtypes and preclinical models for selection of targeted therapies. J Clin Invest 121 , 2750–2767. 10.1172/JCI45014.21633166 33. Horak C.E. , Pusztai L. , Xing G. , Trifan O.C. , Saura C. , Tseng L.M. , Chan S. , Welcher R. , and Liu D. (2013). Biomarker analysis of neoadjuvant doxorubicin/cyclophosphamide followed by ixabepilone or Paclitaxel in early-stage breast cancer. Clin Cancer Res 19 , 1587–1595. 10.1158/1078-0432.CCR-12-1359.23340299 34. SenGupta S. , Hein L.E. , Xu Y. , Zhang J. , Konwerski J.R. , Li Y. , Johnson C. , Cai D. , Smith J.L. , and Parent C.A. (2021). Triple-Negative Breast Cancer Cells Recruit Neutrophils by Secreting TGF-β and CXCR2 Ligands. Front Immunol 12 . 10.3389/FIMMU.2021.659996. 35. Chen J.Q. , and Russo J. (2009). ERalpha-negative and triple negative breast cancer: molecular features and potential therapeutic approaches. Biochim Biophys Acta 1796 , 162–175. 10.1016/J.BBCAN.2009.06.003.19527773 36. Fulda S. , and Debatin K.M. (2006). Extrinsic versus intrinsic apoptosis pathways in anticancer chemotherapy. Oncogene 25 , 4798–4811. 10.1038/SJ.ONC.1209608.16892092 37. Wu D. (2005). Signaling mechanisms for regulation of chemotaxis. Cell Res 15 , 52–56. 10.1038/SJ.CR.7290265.15686628 38. Garaud S. , Buisseret L. , Solinas C. , Gu-Trantien C. , De Wind A. , Van Den Eynden G. , Naveaux C. , Lodewyckx J.N. , Boisson A. , Duvillier H. , (2019). Tumor infiltrating B-cells signal functional humoral immune responses in breast cancer. JCI Insight 5 . 10.1172/JCI.INSIGHT.129641. 39. Sun X. , Wang M. , Wang M. , Yu X. , Guo J. , Sun T. , Li X. , Yao L. , Dong H. , and Xu Y. (2020). Metabolic Reprogramming in Triple-Negative Breast Cancer. Front Oncol 10 . 10.3389/FONC.2020.00428. 40. Blasiak J. , Pawlowska E. , Chojnacki J. , Szczepanska J. , Fila M. , and Chojnacki C. (2020). Vitamin D in Triple-Negative and BRCA1-Deficient Breast Cancer-Implications for Pathogenesis and Therapy. Int J Mol Sci 21 . 10.3390/IJMS21103670. 41. Kohale I.N. , Yu J. , Zhuang Y. , Fan X. , Reddy R.J. , Sinnwell J. , Kalari K.R. , Boughey J.C. , Carter J.M. , Goetz M.P. , (2022). Identification of Src Family Kinases as Potential Therapeutic Targets for Chemotherapy-Resistant Triple Negative Breast Cancer. Cancers (Basel) 14 . 10.3390/CANCERS14174220. 42. Singh A. , Nunes J.J. , and Ateeq B. (2015). Role and therapeutic potential of G-protein coupled receptors in breast cancer progression and metastases. Eur J Pharmacol 763 , 178– 183. 10.1016/J.EJPHAR.2015.05.011.25981295 43. Deng N. , Chen K. , Fan H. , and Jin F. (2022). The synergistic effect of CDKN2B-AS1 and SPC25 on triple-negative breast cancer. Ann Transl Med 10 , 783–783. 10.21037/ATM-22-2900.35965791 44. Stickeler E. , Pils D. , Klar M. , Orlowsk-Volk M. , Zur Hausen A. , Jäger M. , Watermann D. , Gitsch G. , Zeillinger R. , and Tempfer C.B. (2011). Basal-like molecular subtype and HER4 up-regulation and response to neoadjuvant chemotherapy in breast cancer. Oncol Rep 26 , 1037–1045. 10.3892/OR.2011.1392.21769435 45. Li Y. , Zhan Z. , Yin X. , Fu S. , and Deng X. (2021). Targeted Therapeutic Strategies for Triple-Negative Breast Cancer. Front Oncol 11 . 10.3389/FONC.2021.731535. 46. Corsello S.M. , Bittker J.A. , Liu Z. , Gould J. , McCarren P. , Hirschman J.E. , Johnston S.E. , Vrcic A. , Wong B. , Khan M. , (2017). The Drug Repurposing Hub: a nextgeneration drug library and information resource. Nat Med 23 , 405–408. 10.1038/NM.4306.28388612 47. Cui W. , Xue H. , Wei L. , Jin J. , Tian X. , and Wang Q. (2021). High heterogeneity undermines generalization of differential expression results in RNA-Seq analysis. Hum Genomics 15 . 10.1186/S40246-021-00308-5. 48. Hartman Z.C. , Yang X.Y. , Glass O. , Lei G. , Osada T. , Dave S.S. , Morse M.A. , Clay T.M. , and Lyerly H.K. (2011). HER2 overexpression elicits a proinflammatory IL-6 autocrine signaling loop that is critical for tumorigenesis. Cancer Res 71 , 4380–4391. 10.1158/0008-5472.CAN-11-0308.21518778 49. Hofstatter E.W. , Horvath S. , Dalela D. , Gupta P. , Chagpar A.B. , Wali V.B. , Bossuyt V. , Storniolo A.M. , Hatzis C. , Patwardhan G. , (2018). Increased epigenetic age in normal breast tissue from luminal breast cancer patients. Clin Epigenetics 10 . 10.1186/S13148-018-0534-8. 50. Yau C. , Fedele V. , Roydasgupta R. , Fridlyand J. , Hubbard A. , Gray J.W. , Chew K. , Dairkee S.H. , Moore D.H. , Schittulli F. , (2007). Aging impacts transcriptomes but not genomes of hormone-dependent breast cancers. Breast Cancer Res 9 . 10.1186/BCR1765. 51. Benz C.C. (2008). Impact of aging on the biology of breast cancer. Crit Rev Oncol Hematol 66 , 65–74. 10.1016/J.CRITREVONC.2007.09.001.17949989 52. Obayashi T. , Kodate S. , Hibara H. , Kagaya Y. , and Kinoshita K. (2023). COXPRESdb v8: an animal gene coexpression database navigating from a global view to detailed investigations. Nucleic Acids Res 51 , D80–D87. 10.1093/NAR/GKAC983.36350658 53. Kumar U. , Ardasheva A. , Mahmud Z. , Coombes R.C. , and Yagüe E. (2021). FOXA1 is a determinant of drug resistance in breast cancer cells. Breast Cancer Res Treat 186 , 317– 326. 10.1007/S10549-020-06068-5.33417085 54. van Schie E.H. , and van Amerongen R. (2020). Aberrant WNT/CTNNB1 Signaling as a Therapeutic Target in Human Breast Cancer: Weighing the Evidence. Front Cell Dev Biol 8 . 10.3389/FCELL.2020.00025. 55. Xie X. , Kaoud T.S. , Edupuganti R. , Zhang T. , Kogawa T. , Zhao Y. , Chauhan G.B. , Giannoukos D.N. , Qi Y. , Tripathy D. , (2017). c-Jun N-terminal kinase promotes stem cell phenotype in triple-negative breast cancer through upregulation of Notch1 via activation of c-Jun. Oncogene 36 , 2599–2608. 10.1038/ONC.2016.417.27941886 56. Tanaka S. , Ishii T. , Sato F. , Toi M. , and Itou J. (2020). Eribulin mesylate-induced c-Fos upregulation enhances cell survival in breast cancer cell lines. Biochem Biophys Res Commun 526 , 154–157. 10.1016/J.BBRC.2020.03.042.32201082 57. Dai X. , Cheng H. , Chen X. , Li T. , Zhang J. , Jin G. , Cai D. , and Huang Z. (2019). FOXA1 is Prognostic of Triple Negative Breast Cancers by Transcriptionally Suppressing SOD2 and IL6. Int J Biol Sci 15 , 1030–1041. 10.7150/IJBS.31009.31182923 58. Asaduzzaman M. , Constantinou S. , Min H. , Gallon J. , Lin M.L. , Singh P. , Raguz S. , Ali S. , Shousha S. , Coombes R.C. , (2017). Tumour suppressor EP300, a modulator of paclitaxel resistance and stemness, is downregulated in metaplastic breast cancer. Breast Cancer Res Treat 163 , 461–474. 10.1007/S10549-017-4202-Z.28341962 59. Koboldt D.C. , Fulton R.S. , McLellan M.D. , Schmidt H. , Kalicki-Veizer J. , McMichael J.F. , Fulton L.L. , Dooling D.J. , Ding L. , Mardis E.R. , (2012). Comprehensive molecular portraits of human breast tumours. Nature 490 , 61–70. 10.1038/NATURE11412.23000897 60. Agrawal P. , Sambaturu N. , Olgun G. , and Hannenhalli S. (2022). A Path-Based Analysis of Infected Cell Line and COVID-19 Patient Transcriptome Reveals Novel Potential Targets and Drugs Against SARS-CoV-2. Front Immunol 13 . 10.3389/FIMMU.2022.918817. 61. Lehmann B.D. , Jovanović B. , Chen X. , Estrada M. V. , Johnson K.N. , Shyr Y. , Moses H.L. , Sanders M.E. , and Pietenpol J.A. (2016). Refinement of Triple-Negative Breast Cancer Molecular Subtypes: Implications for Neoadjuvant Chemotherapy Selection. PLoS One 11 . 10.1371/JOURNAL.PONE.0157368. 62. Koschützki D. , and Schreiber F. (2008). Centrality analysis methods for biological networks and their application to gene regulatory networks. Gene Regul Syst Bio 2 , 193– 201. 10.4137/GRSB.S702. 63. Wu T. , Hu E. , Xu S. , Chen M. , Guo P. , Dai Z. , Feng T. , Zhou L. , Tang W. , Zhan L. , (2021). clusterProfiler 4.0: A universal enrichment tool for interpreting omics data. Innovation (Cambridge (Mass.)) 2 . 10.1016/J.XINN.2021.100141. 64. Hippen A.A. , Falco M.M. , Weber L.M. , Erkan E.P. , Zhang K. , Doherty J.A. , Vähärautio A. , Greene C.S. , and Hicks S.C. (2021). miQC: An adaptive probabilistic framework for quality control of single-cell RNA-sequencing data. PLoS Comput Biol 17 . 10.1371/JOURNAL.PCBI.1009290. 65. Lamb J. , Crawford E.D. , Peck D. , Modell J.W. , Blat I.C. , Wrobel M.J. , Lerner J. , Brunet J.P. , Subramanian A. , Ross K.N. , (2006). The Connectivity Map: using gene-expression signatures to connect small molecules, genes, and disease. Science 313 , 1929–1935. 10.1126/SCIENCE.1132939.17008526 66. Berman H.M. , Westbrook J. , Feng Z. , Gilliland G. , Bhat T.N. , Weissig H. , Shindyalov I.N. , and Bourne P.E. (2000). The Protein Data Bank. Nucleic Acids Res 28 , 235–242. 10.1093/NAR/28.1.235.10592235 67. O’Boyle N.M. , Banck M. , James C.A. , Morley C. , Vandermeersch T. , and Hutchison G.R. (2011). Open Babel: An open chemical toolbox. J Cheminform 3 . 10.1186/17582946-3-33. 68. Gan J. hong , Liu J. xiang , Liu Y. , Chen S. wen , Dai W. tao , Xiao Z.X. , and Cao Y. (2023). DrugRep: an automatic virtual screening server for drug repurposing. Acta Pharmacol Sin 44 . 10.1038/S41401-022-00996-2. 69. Eberhardt J. , Santos-Martins D. , Tillack A.F. , and Forli S. (2021). AutoDock Vina 1.2.0: New Docking Methods, Expanded Force Field, and Python Bindings. J Chem Inf Model 61 , 3891–3898. 10.1021/ACS.JCIM.1C00203.34278794 70. Morris G.M. , Ruth H. , Lindstrom W. , Sanner M.F. , Belew R.K. , Goodsell D.S. , and Olson A.J. (2009). AutoDock4 and AutoDockTools4: Automated docking with selective receptor flexibility. J Comput Chem 30 , 2785–2791. 10.1002/JCC.21256.19399780
PMC010xxxxxx/PMC10327268.txt
==== Front medRxiv MEDRXIV medRxiv Cold Spring Harbor Laboratory 37425815 10.1101/2023.06.28.23291950 preprint 1 Article Appropriate sampling and long follow-up are required to rigorously evaluate longevity of humoral memory after Vaccination Ganusov Vitaly V. 123 1 Department of Microbiology, University of Tennessee, Knoxville, TN, USA 2 Department of Mathematics, University of Tennessee, Knoxville, TN, USA 3 Texas Biomedical Research Institute, San Antonio, TX, USA 29 6 2023 2023.06.28.23291950https://creativecommons.org/licenses/by-nc-nd/4.0/ This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which allows reusers to copy and distribute the material in any medium or format in unadapted form only, for noncommercial purposes only, and only so long as attribution is given to the creator. nihpp-2023.06.28.23291950.pdf One of the goals of vaccination is to induce long-term immunity against the infection and/or disease. However, evaluating the duration of protection following vaccination often requires long-term follow-ups that can conflict with the desire to rapidly publish results. Arunachalam et al. JCI 2023 followed individuals receiving third or fourth dose of mRNA COVID19 vaccines for up to 6 months and in finding that the levels of SARS-CoV2-specific antibodies (Abs) declined with similar rates for the two groups came to the conclusion that additional boosting is unnecessary to prolong immunity to SARS-CoV-2. However, this may be premature conclusion to make. Accordingly, we demonstrate that measuring Ab levels at 3 time points and only for a short (up to 6 month) duration does not allow to accurately and rigorously evaluate the long-term half-life of vaccine-induced Abs. By using the data from a cohort of blood donors followed for several years, we show that after re-vaccination with vaccinia virus (VV), VV-specific Abs decay bi-phasically and even the late decay rate exceeds the true slow loss rate of humoral memory observed years prior to the boosting. We argue that mathematical modeling should be used to better optimize sampling schedules to provide more reliable advice about the duration of humoral immunity after repeated vaccinations. ==== Body pmc“To consult the statistician after an experiment is finished is often merely to ask him to conduct a post mortem examination. He can perhaps say what the experiment died of. ” Ronald Fisher. Results and discussion The holy grail for vaccinologists is to induce long term effective immunity against infections, such as against the SARS-CoV-2, the cause of the recent pandemic1. It is well known that specific antibodies (Abs) to the spike protein of SARS-CoV-2 account for disease protection2, and it is of great interest to discover how long such antibodies persist after infection or various forms and regimens of vaccinations3,4. In one such study, Arunachalam et al. 4, compared the durability of neutralizing antibody (nAb) levels in persons that received either 3 or 4 vaccinations with Pfizer or Moderna mRNA vaccines. By using a simple linear regression, Arunachalam et al. 4 showed that the decay of Abs between 1 and 6 months was similar in individuals receiving 3rd or 4th vaccination allowing them to conclude that “The durability of serum antibody responses improves only marginally following booster immunizations with the Pfizer-BioNTech or Moderna mRNA vaccines.” inferring that extra boosters are unnecessary to confer long-term immunity. From multiple studies in mice and humans it is clear that the kinetics of Ab decay are generally more complex than a simple exponentially decaying function would suggest5–7. Specifically, after the peak immune response, Ab titers initially decline rapidly and approach their long-term decay kinetics over months. To investigate how the dynamics of Ab titers after revaccination relates to their long-term maintenance, we re-analyzed unique data from a cohort of long-term blood donors that recorded Ab titers to multiple vaccines over a long time period8. For this analysis, we selected 4 individuals illustrated in Figure 2 of Amanna et al. 8 that were followed for 20+ years and subsequently been revaccinated with vaccinia virus (Figure 1 and Supplemental Figure S1). The kinetics of Ab decay prior to revaccination showed remarkable stability of VV-specific Abs with half-life times of 31–91 years (Figure 1A). By fitting a modification of the mathematical model of Ab response reported previously7 to the data on Ab expansion and contraction after revaccination, we found much smaller half-life times of VV-specific Abs 1.4–2.6 years as compared to the half-life times prior to revaccination (except of one volunteer (ID=514) showing no long-term decay after revaccination, Figure 1B). Given that these individuals were followed for nearly 2 years after VV revaccination their VV-specific Abs levels still decayed more rapidly than during years prior to boosting. This analysis indicates that long follow up studies are required before any conclusions can be made about the true longevity of vaccine-induced Abs. Additional analysis of the kinetics of Ab decay after the peak following revaccination with VV or tetanus vaccine suggested that these decays are not well described by a simple exponential curve indicating that different sub-populations of antibody-secreting cells (ASCs) with discordant lifespans may be present (Supplemental Figure S1). Interestingly, the long-term loss of tetanus-specific Abs occurred at a slower rate than the loss of VV-specific Abs, which may be surprising since VV is a live vaccine and tetanus vaccine is a protein-based vaccine. The difference here likely lies in the different times of follow up allowing more rigorous estimates of the half-life time of tetanus-specific Abs (Supplemental Figure S1). In our recent work, we formulated mathematical models aimed to describe kinetics of ASCs and Ab titer following vaccination of humans7. The example of the immune response following VV immunization showed that to capture Ab loss, one needs to measure Ab titers frequently around the peak of the immune response (Figure 1C; VV-specific response for Donor 97 in the model given in eqns. (S.2)–(S.3) are P0=0.022, Ton=9.5 day, ρ=1.02 day, Toff=14.79 day, δA=0.15/day, Tmem=28 day, A0=6.7, p=261.3/day)7. We then used the model that described well the Ab response to VV (eqn. (S.1)) and selected parameters that would allow matching the average Ab titers measured at three time points chosen in the report of Arunachalam et al. 4. Importantly, we could choose two extreme sets of parameters that predict either relatively short- lived Ab response with half-life T1/2=0.3 years (model parameters: A0=20000, ρ=0.091/day, Toff=30 day, fm=0, δ1=0.0058/day, δ2=0.13/day) or extremely long Ab response (T1/2>100 years, model parameters: A0=20000, ρ=0.48/day, Toff=7 day, fm=0.23, δ1=0.04/day, δ2=0.000013/day, Figure 1). These results further demonstrate that measuring Ab titers at only 3 time sparsely spaced points does not allow to rigorously evaluate the duration of humoral memory following revaccination. What are the possible solutions? Obviously, following revaccinated individuals for longer times may provide some useful information, but inapparent boostings of immunity following reinfection from currently circulating SARS-CoV-2 cannot be excluded. A better approach would be to improve the experimental design by using mathematical modeling-assisted power analyses that should indicate the most informative time points that samples should be collected at to provide the most reliable estimate of the parameter of immunity that is being advocated such as the Ab half-life times9,10. Following Ab response longitudinally in individual volunteers may also allow to use a more powerful approach of mixed effect modeling that may help to better define average and variability in Ab half-life times11. Ultimately, better collaborations between experimentalists and mathematical modelers may help to design more reliable experiments that provide more rigorous estimates of the longevity of humoral immunity afforded by vaccination, and (to paraphrase Ronald Fisher) “not let postmortem analysis to identify reasons experiment died of”. Supplementary Material Supplement 1 Supplement 2 Acknowledgement We would like to thank Mark Slifka and Ian Amanna for providing experimental data for this analysis. This was work supposed by the NIH grant (R01AI158963) to VVG. Abbreviations: VV vaccinia virus EU ELISA units NLS nonlinear least squares ASCs antibody-secreting cells NLME nonlinear mixed effects Figure 1: Stable levels of antibodies are reached only months after revaccination with vaccinia virus. A: We analyzed kinetics of Ab titers in four long-term blood donors from subjects 1–4 shown in Figure 2 of Amanna et al.8). These individuals were followed up for 20+ years during which they had been revaccinated with vaccinia virus. B: We fitted a mathematical model of humoral immune response (eqn. (S.1))) to subsets of the data that include revaccination and estimated the rate Ab expansion (ρ), the proportion of Ab conversion into long-lived population fm, and the half-life of the humoral immunity T1/2, panel B and see Supplemental Table S1 for estimated model parameters). C: Kinetics of Ab response following VV revaccination in one volunteer (Donor 97) suggests infinite half-life of the long-term memory (see Main text for best fit parameters)7. D: Sparse measurements of Ab titers after vaccination allow for alternative mathematical models (eqn. (S.1) with different parameter sets) with drastically different predicted longevities of Abs. Here the markers are average Ab titers from Figure 1Ciii of Arunachalam et al.4, and lines are predictions of two alternative mathematical models with different assumed sub-populations of ASCs. ==== Refs References 1. Plotkin S. A. 2008 Vaccines: correlates of vaccine-induced immunity. Clin Infect Dis, 47 (3 ), 401–9.18558875 2. Gilbert P. B. , Montefiori D. C. , McDermott A. B. , Fong Y. , Benkeser D. , Deng W. , Zhou H. , Houchens C. R. , Martins K. 2022 Immune correlates analysis of the mRNA-1273 COVID-19 vaccine efficacy clinical trial. Science (New York, N.Y.), 375 , 43–50. doi:10.1126/science.abm3425.34812653 3. Goel R. R. , Painter M. M. , Apostolidis S. A. , Mathew D. , Meng W. , Rosenfeld A. M. , Lund-green K. A. , Reynaldi A. , Khoury D. S. 2021 mRNA vaccines induce durable immune memory to SARS-CoV-2 and variants of concern. Science (New York, N.Y.), 374 , abm0829. doi:10.1126/science.abm0829.34648302 4. Arunachalam P. S. , Lai L. , Samaha H. , Feng Y. , Hu M. , Hui H. S.-Y. , Wali B. , Ellis M. , Davis-Gardner M. E. 2023 Durability of immune responses to mRNA booster vaccination against COVID-19. The Journal of clinical investigation, 133 . doi:10.1172/JCI167955. 5. Murali-Krishna K. , Altman J. , Suresh M. , Sourdive D. , Zajac A. , Miller J. , Slansky J. & Ahmed R. 1998 Counting antigen-specific CD8+ T cells: A re-evaluation of bystander actiation during viral infection. Immunity, 8 , 177–187.9491999 6. Miller J. , van der Most R. , Akondy R. , Glidewell J. , Albott S. , Masopust D. , Murali-Krishna K. , Mahar P. , Edupuganti S. 2008 Human effector and memory CD8+ T cell responses to smallpox and yellow fever vaccines. Immunity, 28 (5 ), 710–22.18468462 7. Le D. , Miller J. D. & Ganusov V. V. 2014 Mathematical modeling provides kinetic details of the human immune response to vaccination. Front Cell Infect Microbiol, 4 , 177. doi:10.3389/fcimb.2014.00177.25621280 8. Amanna I. J. , Carlson N. E. & Slifka M. K. 2007 Duration of humoral immunity to common viral and vaccine antigens. N Engl J Med, 357 (19 ), 1903–1915. doi:10.1056/NEJMoa066092.17989383 9. Rajakaruna H. & Ganusov V. V. 2022 Mathematical modeling to guide experimental design: T cell clustering as a case study. Bulletin of mathematical biology, 84 , 103. doi:10.1007/s11538-022-01063-x.35978047 10. Majumder B. , Budhu S. & Ganusov V. V. 2023 Mathematical modeling suggests cytotoxic T lymphocytes control growth of B16 tumor cells in collagin-fibrin gels by cytolytic and non-lytic mechanisms. bioRxiv. doi:https://doi.org/10.1101/2023.03.28.534600. Https://doi.org/10.1101/2023.03.28.534600. 11. Pinheiro J. C. & Bates D. M. 2000 Mixed Effects Models in S and S-Plus. Springer. 12. De Boer R. J. , Oprea M. , Antia R. , Murali-Krishna K. , Ahmed R. & Perelson A. S. 2001 Recruitment times, proliferation, and apoptosis rates during the CD8(+) T-cell response to lymphocytic choriomeningitis virus. J Virol, 75 (22 ), 10 663–10 669. 13. Ganusov V. V. , Milutinovic D. & De Boer R. J. 2007 IL-2 regulates expansion of CD4+ T cell populations by affecting cell death: insights from modeling CFSE data. J Immunol, 179 (2 ), 950–957.17617586 14. Slifka M. , Antia R. , Whitmire J. & Ahmed R. 1998 Humoral immunity due to long-lived plasma cells. Immunity, 8 (3 ), 363–72.9529153 15. Kindt T. J. , Osborne B. & Goldsby R. A. 2007 Kuby Immunology, 6th edition. New York: W. H. Freeman. 16. Bates D. M. & Watts D. G. 1988 Nonlinear regression analysis and its applications. Hoboken, NJ: John Wiles & Sons, Inc.
PMC010xxxxxx/PMC10327269.txt
==== Front medRxiv MEDRXIV medRxiv Cold Spring Harbor Laboratory 37425783 10.1101/2023.06.29.23292047 preprint 1 Article Inflammatory Biomarkers and Physiomarkers of Late-Onset Sepsis and Necrotizing Enterocolitis in Premature Infants Kumar Rupin 1 Kausch Sherry 2 Gummadi Angela K.S. 2 Fairchild Karen D. 2 Abhyankar Mayuresh 3 Petri William A. Jr. 3 Sullivan Brynne A. 2* 1 Department of Pediatrics, Division of Neonatology, University of Kentucky College of medicine 2 Department of Pediatrics, Division of Neonatology, University of Virginia School of Medicine, Charlottesville, VA 3 Department of Internal Medicine, Division of Infectious Diseases, University of Virginia School of Medicine, Charlottesville, VA Author Contributions: All authors made substantial contributions to the conception, performance, or writing of the work; RK, SK, and BS made substantial contributions to the acquisition, analysis, or interpretation of data; RK, BS, and AG drafted the work and all other authors have substantively revised it. All authors have approved the submitted version. All authors have agreed both to be personally accountable for the author’s own contributions and to ensure that questions related to the accuracy or integrity of any part of the work, even ones in which the author was not personally involved, are appropriately investigated, resolved, and the resolution documented in the literature. * Corresponding Author: Dr. Brynne Sullivan, University of Virginia School of Medicine, Department of Pediatrics, PO Box 800386, Charlottesville, Virginia, 22908. [email protected], Phone: Office: 434.924.5428 30 6 2023 2023.06.29.23292047https://creativecommons.org/licenses/by-nc-nd/4.0/ This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which allows reusers to copy and distribute the material in any medium or format in unadapted form only, for noncommercial purposes only, and only so long as attribution is given to the creator. nihpp-2023.06.29.23292047.pdf Background: Early diagnosis of late-onset sepsis (LOS) and necrotizing enterocolitis (NEC) in VLBW (<1500g) infants is challenging due to non-specific clinical signs. Inflammatory biomarkers increase in response to infection, but non-infectious conditions also cause inflammation in premature infants. Physiomarkers of sepsis exist in cardiorespiratory data and may be useful in combination with biomarkers for early diagnosis. Objectives: To determine whether inflammatory biomarkers at LOS or NEC diagnosis differ from times without infection, and whether biomarkers correlate with a cardiorespiratory physiomarker score. Methods: We collected remnant plasma samples and clinical data from VLBW infants. Sample collection occurred with blood draws for routine laboratory testing and blood draws for suspected sepsis. We analyzed 11 inflammatory biomarkers and a continuous cardiorespiratory monitoring (POWS) score. We compared biomarkers at gram-negative (GN) bacteremia or NEC, gram-positive (GP) bacteremia, negative blood cultures, and routine samples. Results: We analyzed 188 samples in 54 VLBW infants. Biomarker levels varied widely, even at routine laboratory testing. Several biomarkers were increased at the time of GN LOS or NEC diagnosis compared with all other samples. POWS was higher in patients with LOS and correlated with five biomarkers. IL-6 had 78% specificity at 100% sensitivity to detect GN LOS or NEC and added information to POWS (AUC POWS = 0.610, POWS + IL-6 = 0.680). Conclusion(s): Inflammatory biomarkers discriminate sepsis due to GN bacteremia or NEC and correlate with cardiorespiratory physiomarkers. Baseline biomarkers did not differ from times of GP bacteremia diagnosis or negative blood cultures. Category of Study: Clinical Research K23 HD097254 R01 AI124214 R01 AI152477 ==== Body pmcIntroduction Sepsis and necrotizing enterocolitis (NEC) lead to significant morbidity and mortality in premature infants1. Inflammation due to sepsis and NEC can cause end-organ damage2,3, including brain injury, which leads to long-term impairment.4–6 Early recognition and treatment may improve outcomes, but signs and symptoms often overlap with non-infectious conditions7. Therefore, a diagnosis is often made once the infection is advanced, and on the other hand, empiric antibiotics are given when there is no infection.8 Clinicians have limited tests and data to guide decisions on the likelihood of infection. Inflammatory molecules, upregulated as part of the immune response to infection, hold promise as biomarkers of neonatal sepsis. A biomarker with sufficient diagnostic accuracy could be a useful screening tool to aid in deciding whether to start antibiotics or continue monitoring for further signs of sepsis. Many studies have evaluated biomarkers for neonatal sepsis in premature infants.9–12 But, few have been translated into clinical care.13 Decisions to start or stop antibiotics in the NICU largely rely on the clinician’s assessment of clinical signs and non-specific laboratory tests, such as C-reactive protein14 and complete blood count (CBC) components.15 Physiologic markers of sepsis can be detected using predictive analysis of continuous vital sign data from standard bedside monitors.16 We used predictive analytics and a multicenter cohort of very low birth weight (VLBW, birth weight < 1500g) preterm infants to develop and externally validate a cardiorespiratory sepsis risk model. The model, called Pulse Oximetry Warning Score (POWS), detects abnormal heart rate and SpO2 patterns that occur near the clinical diagnosis of late-onset sepsis and calculates the relative risk of positive blood culture with sepsis in the subsequent 24 hours.17 We hypothesize that inflammatory biomarkers will correlate with physiologic markers of sepsis. Combining information from sepsis biomarkers and physiomarkers could add information on the likelihood of sepsis when vital sign-based predictive analytics detect a rising risk of sepsis. While many prior studies have analyzed inflammatory biomarkers in premature infants at the time of suspected sepsis,10,18–20 few have analyzed baseline levels or changes over time in patients who develop sepsis.21,22 Inflammatory biomarkers may be less useful in premature infants with non-infectious conditions that lead to inflammation, including respiratory distress syndrome and chronic lung disease. To understand the potential utility of biomarkers in this unique population, more data are needed. In this study, we aim to identify inflammatory biomarkers that discriminate sepsis from baseline levels and sepsis-like illnesses in VLBW infants. To do this, we measured multiple biomarkers in samples collected from remnant plasma, weekly when available and at the time of evaluation for late-onset sepsis or NEC. Methods Study Design This was a prospective, observational cohort study conducted at a single center Level IV academic NICU. The Institutional Review Board approved this study with verbal informed consent and a signed HIPAA data use authorization from a parent. We prospectively enrolled VLBW infants within seven days of birth and collected clinical data and remnant plasma from blood drawn for suspected sepsis and for routine laboratory testing. Clinical data and culture results were recorded from the electronic health record. We classified samples according to the indication for the laboratory tests ordered and the final diagnosis of a workup for LOS or NEC (see “clinical definitions” section below). Sample collection The study protocol did not require any extra blood draws and did not influence the clinical team’s decisions on ordering blood tests. Nurses collected blood samples in tubes according to the clinical laboratory guidelines for the laboratory test ordered. Blood for a complete blood count was collected in EDTA tubes, and blood for a basic metabolic panel or C- reactive protein was collected in sodium heparin tubes. These were the only clinical laboratory tests used to request remnant plasma for the study. The clinical lab stores samples at 4 degrees Celsius immediately after receipt, and the leftover blood from a clinical sample was requested by the study team. The clinical lab separated plasma from whole blood for samples collected in sodium heparin tubes, while the study team separated plasma from whole blood for samples collected in EDTA tubes by centrifugation (1800 g × 5 min). We stored all remnant plasma samples with at least 250 microliters at −80C until analysis. Biomarker analysis We selected subjects with at least two collected samples of adequate plasma volume (≥ 250 μL) for biomarker analysis. The multiplex assay measured a panel of inflammatory biomarkers: interleukin (IL)-6, IL-8, IL-10, IL-18, interferon gamma inducible protein (IP)-10, tumor necrosis factor alpha (TNFα), procalcitonin (PCT), human growth factor (HGF), endothelin growth factor (EGF), soluble suppression of tumorigenicity 2 (sST-2), and IL-1 receptor antigen (IL1-ra). We selected these biomarkers based on their role in the inflammatory response to bacterial infections and the existing literature. Analyte concentrations were quantified using a customized multiplex Luminex® magnetic bead-based antibody assay (R&D Systems, Minneapolis, USA). Fluorescence signals for each biomarker bead region were analyzed on a Luminex®200, a dual-laser flow-based detection instrument. Concentrations below the lowest standard were recorded as the value of the lower limit of detection for statistical analyses. Based on preliminary data, we used a 10-fold dilution to analyze PCT, sST-2, and IL-1ra, and the remaining biomarkers did not require dilution. Results of C- reactive protein (CRP) measurement obtained for clinical use were obtained from the electronic health record. Clinical definitions Samples were classified as “routine” if they were obtained from blood drawn for standard laboratory monitoring and not during a period of suspected infection. We classified samples obtained at the time of a blood culture using the diagnosis of the event according to the following definitions: Late-onset sepsis (LOS): a positive blood culture obtained after 72 hours of age and treated with at least five days of intravenous antibiotics. These were further categorized as sepsis due to Gram-positive (GP) or Gram-negative (GN) bacteremia according to the organism identified by blood culture. Necrotizing enterocolitis (NEC): radiographic evidence of necrotizing enterocolitis and clinical illness with or without a positive blood culture. Clinical sepsis (CS): negative blood and urine cultures treated with at least five days of antibiotics for presumed infection due to clinical illness Sepsis Ruled Out (SRO): a negative blood culture treated with fewer than five days of antibiotics. Cardiorespiratory sepsis risk prediction We previously developed a multivariable model to predict sepsis using continuous heart rate (HR) and oxygen saturation (SpO2) data called the Pulse Oximetry Warning Score, or POWS.17 The POWS model calculates the mean, standard deviation, skewness, kurtosis, and cross-correlation of HR and SpO2 every 10 minutes and uses logistic regression at each window to predict the relative risk of LOS with a positive blood culture in the next 24 hours. We calculated hourly POWS values from continuous bedside monitoring data during the 12 hours preceding blood cultures. POWS scores were calculated after discharge and thus did not influence decisions about clinical care. Statistical analyses We compared distributions of inflammatory biomarkers and POWS data using Kruskal-Wallis analysis across the four diagnosis groups, followed by a pairwise Wilcoxon test with corrections for multiple testing. The relationships between POWS and individual biomarker levels were assessed using univariate linear regression. The statistical significance of each model’s coefficient was adjusted for repeated measures using the Huber-White method. A p-value < 0.05 was considered statistically significant. The predictive performance of biomarkers and POWS were assessed individually or in combination using sensitivity, specificity at specific thresholds, and area under the receiver operator characteristics curve (AUC) using logistic regression to predict LOS or NEC. All statistical analysis was performed using RStudio (R Version 4.2.3, Vienna, Austria). Results Patients and samples We enrolled 118 VLBW infants with parental consent. Of those enrolled, we had sufficient samples from 54 infants for analysis. The final cohort had a mean gestational age of 25.9 ± 1.9 weeks and a mean birth weight of 803 ± 222 grams. Clinical characteristics overall and comparing those with and without LOS or NEC are shown in Table 1. In total, 188 plasma samples were analyzed, including 67 obtained near the time of blood cultures for suspected LOS or NEC and 120 at the time of routine blood sampling. Seven biomarkers (IP-10, IL-6, IL-10, IL-18, TNFa, IL-8, PCT) were analyzed at all times, while four (HGF, EGF, sST-2, IL-1ra) were successfully measured 80% of the time (152 samples in 45 patients). Of the 67 blood cultures with study samples analyzed, 28 were diagnosed as LOS or NEC with bacteremia. Of these, there were 22 cases of Gram-positive (GP) bacteremia, 5 Gram-negative (GN) bacteremia, and 1 with NEC and GN bacteremia. Organisms in the positive blood cultures were coagulase-negative Staphylococcus (CONS) species (n=19), Escherichia coli (n=2), Enterobacter species (n=1), Group B Streptococcus (n=1), Klebsiella species (n=5), and methicillin-susceptible Staphylococcus aureus (n=2). There were 2 cases of NEC with negative blood cultures and 17 negative blood cultures diagnosed as clinical sepsis (CS). The remaining 141 samples came from blood drawn at times with no sepsis (NS), including routine samples and blood cultures diagnosed as sepsis ruled out (SRO). Biomarkers Five of the eleven biomarkers (IL-6, TNF-α, IL-8, IL-10, and sST-2) were significantly higher in patients with GN sepsis or NEC than those with NS, CS, and GP sepsis. Overall, biomarkers at CS or GP diagnosis were not significantly different from NS samples (Figure 1). In a few cases, infants with serial biomarker measurements and LOS or NEC had a rise from baseline at the time of diagnosis. CRP was measured at the time of 92 samples (49%) and was highly variable. CRP was significantly higher in patients with GN sepsis or NEC (median CRP 4.4 g/dL, IQR 2.1 – 11.0) compared to patients with no sepsis (median CRP 0.1 g/dL, IQR 0.1 – 0.4) or clinical sepsis (median CRP 0.12 g/dL, IQR 0.1 – 1.2). There was no statistically significant difference in CRP for GN sepsis versus GP sepsis cases (median CRP 2.7 g/dL, IQR 0.1 – 3.3) or clinical sepsis (median CRP 0.12 g/dL, IQR 0.1 – 1.2). Hierarchical cluster analysis of cytokines from 30 LOS or NEC blood samples showed two distinct clusters of biomarker profiles (Figure 2). GN bacteremia or NEC was more prevalent in the cluster of samples with the highest cytokine levels. IL-6 levels had the best overall test accuracy for diagnosing gram-negative sepsis or NEC with a specificity of 78% and a negative predictive value of 100% at a threshold of 200 pg/mL with 100% sensitivity. Association with POWS, a cardiorespiratory sepsis risk prediction model Of the 188 samples, 187 had continuous heart rate and oxygen saturation data available around the time of sample collection to calculate POWS, a pulse oximetry warning score designed as a physiomarker for impending sepsis or sepsis-like illness. POWS was significantly associated with the levels of 5 biomarkers (IL-8, PCT, HGF, sST-2, and IL1-ra) irrespective of the associated diagnosis (Figure 3, p<0.05). The maximum POWS within the 12 hours preceding the sample had an AUC to predict LOS (GN or GP) or NEC (with or without bacteremia) of 0.610. The AUC increased to 0.680 when the IL-6 level of the sample was added to the model. Discussion We assayed inflammatory biomarkers at the time of blood culture for suspected sepsis and routine laboratory testing in VLBW infants and found differences that distinguish late-onset Gram-negative sepsis and NEC from other diagnoses, but similar biomarker distributions in samples collected at baseline and at the time of suspected sepsis or Gram-positive sepsis. We also found correlations between inflammatory biomarkers and POWS, an algorithm that detects abnormal patterns of heart rate and oxygen saturation that we previously developed as a physiomarker of sepsis. Our results confirm the findings in previous work that multiple biomarkers can discriminate between GN sepsis and GP sepsis at the time of positive blood culture.12 Since Gram-negative infections and NEC are known to carry higher morbidity and mortality, this finding might be a useful adjunct clinical decision support tool in helping pick the appropriate empiric antibiotic therapy and institute treatment early.23,24,25 The current study adds to our prior work because we analyzed samples remote from suspected sepsis. We found no significant difference in biomarker levels in these samples compared to those obtained at the time of clinical sepsis or Gram-positive sepsis compared to times when sepsis was not suspected or was ruled out. Some VLBW preterm infants may have a chronic or subacute systemic inflammatory response associated with lung disease or intracranial hemorrhage, which could confound sepsis prediction using blood biomarkers or cardiorespiratory physiomarkers.26,27,28 Plasma cytokines and chemokines are early markers of immune activation in response to infection, and have been shown to rise in septic premature infants.22,29,30 Several studies have evaluated the utility of biomarkers for diagnosing sepsis by comparing biomarker levels in cases versus controls11. Others have compared cytokine profiles obtained at the time of suspected infection in VLBW infants when the blood culture returns positive versus negative.31,32 Instead, we assessed longitudinal changes in inflammatory biomarkers measured at times when there was clinical suspicion for infection or sepsis and at times remote from infection. Kuster and colleagues also took the approach of measuring cytokines at baseline and at the time of suspected sepsis in VLBW infants and found IL-1ra and IL-6 to have high sensitivity and specificity in 21 cases of late-onset sepsis, even measured from samples taken the day prior to blood culture.30 A longitudinal study of extremely low birth weight (<1000g) measured cytokines at days 1, 3, 14, and 21 after birth and found that IFN-γ, IL-10, IL-18, TGF-β, and TNF-α levels differed among infants who developed fungal or bacterial LOS compared with those who never developed sepsis.22 This study enrolled a large cohort from the NICHD Neonatal Research Network centers but did not evaluate differences relative to the timing of blood cultures. They found that infants with higher levels of immune regulatory cytokines relative to pro-inflammatory cytokines were associated with an increased risk of LOS at any time during the NICU course.21 Of the biomarkers assayed in our study, IL-6 has been most consistently identified as a promising sepsis biomarker in previous studies, though with variable sensitivity and specificity.32 We found that IL-6 had a high specificity at a threshold set to detect 100% of Gram-negative sepsis and NEC cases, but low predictive accuracy for Gram-positive cases, which were mostly due to CONS bacteremia. We also found highly elevated levels of IL-6 measured from samples remote from sepsis with blood draws for routine laboratory tests. The inclusion of sST-2 as a sepsis biomarker was novel for this population as its utility has mainly been studied in adult populations. It is an IL-1 family receptor that binds IL-33 and has been implicated in diseases involving intestinal inflammation.33 Recent studies in adult patients demonstrate an association between elevated sST-2 and the severity of illness in Clostridium difficile colitis.34,35 Our results indicate that this may also be a useful biomarker for diagnosing NEC and Gram-negative sepsis, where gastrointestinal dysfunction and bacterial translocation cause systemic inflammation. Cytokines and chemokines such as IL-6, IL-1ra, and IL-8 have been demonstrated to have diagnostic utility as early sepsis markers 30,36, while acute phase reactants such as CRP and PCT rise during the later phases of systemic inflammation.37 CRP and PCT are also examples of the few inflammatory biomarkers available in U.S. clinical laboratories, which may drive their clinical use despite evidence of low clinical utility.14 A recent meta-analysis showed overall low sensitivity and specificity of CRP for late-onset sepsis diagnosis.38 We note several limitations of the study. First, the analysis was limited by the small sample size, where most late-onset sepsis events were due to CONS bacteremia, some of which may have represented contamination and not a true infection. Second, events diagnosed as clinical sepsis are heterogeneous in severity with subjective diagnostic criteria. We collected data on clinical characteristics, but not to the level of detail to account for concurrent inflammatory processes, such as lung disease, invasive mechanical ventilation, and minor procedures. Finally, the use of remnant plasma allowed us to enroll patients and collect samples without additional blood draws, but the volume of plasma available was small and therefore did not allow assays to be run in duplicate. Correlations with physiomarkers of sepsis using POWS, a cardiorespiratory sepsis risk score, resulted in promising correlations that warrant further confirmation in larger studies. Continuous cardiorespiratory predictive monitoring used in conjunction with biomarker testing could prove useful, both for early initiation of antibiotics when the likelihood of sepsis is high and for sparing antibiotics when biomarkers and physiomarkers indicate low sepsis risk. Conclusion In conclusion, in a prospective cohort of VLBW infants, inflammatory biomarkers discriminated between late-onset sepsis due to gram-negative bacteremia or NEC and all other samples collected at baseline or times of suspected sepsis with negative blood culture or gram-positive bacteremia. Several inflammatory biomarkers, measured at baseline and at the time of suspected or confirmed sepsis, correlated with cardiorespiratory physiomarkers of sepsis. Baseline inflammatory biomarker levels did not differ from levels obtained at the time of sepsis due to gram-positive bacteremia or negative blood cultures. Funding: We acknowledge the following grant for funding the work presented in this manuscript: K23 HD097254 (PI: B Sullivan); R01 AI124214, R01 AI152477 (PI: W. Petri) Figure 1. Plasma biomarker levels from remnant plasma collected at times with or without sepsis in VLBW infants. Samples were classified as clinical sepsis (CS, n = 17), Gram-positive bacteremia (GP, n = 22), Gram-negative bacteremia or NEC (GN, n = 8), or not sepsis (NS, n = 141). IL-6 (A), TNF-a (B), IL-8 (C), IL-10 (D), and sST-2 (E) were significantly higher in patients with GN sepsis as compared with NS, CS, and GP sepsis. Biomarker levels were not different when comparing NS, CS, and GP sepsis samples. Figure 2. Hierarchical cluster analysis of cytokine levels in 30 cases of blood culture-positive sepsis and NEC. The scaled natural log of each cytokine was taken prior to clustering. Higher than average cytokine levels are depicted in shades of red and lower than average levels are depicted in green. The resulting hierarchical clustering dendrogram is on the left-hand side of the heatmap. Figure 3. Relationship between POWS and inflammatory biomarkers. Plots of the scaled natural log of each biomarker level on the y-axis and the maximum sepsis risk within 12 hours before blood culture on the x-axis. Table 1. Cohort characteristics overall and grouped by infants with or with late-onset sepsis (LOS) or necrotizing enterocolitis (NEC) diagnosis. Values are presented as mean (standard deviation) or number (%). No LOS/NEC (N=30) LOS/NEC (N=24) Overall (N=54) Gestational age (weeks) 26.1 (1.8) 24.8 (2) 25.9 (1.9) Birth Weight (grams) 904.4 (208) 912.9 (206) 803 (222) Male 13 (24.0%) 15 (27.7%) 28 (51.9%) Race Black 2 (3.7%) 4 (7.4%) 6 (11.11%) Hispanic 3 (5.6%) 1 (1.9%) 4 (7.4%) White 23 (42.5%) 19 (35.1%) 42 (77.8%) Unknown 2 (3.7%) 0 2 (3.7%) Died 2 (3.7%) 2 (3.7%) 4 (7.4%) Impact: Late-onset sepsis and necrotizing enterocolitis (NEC) in very low birth weight (VLBW, <1500g) premature infants can result in severe morbidity and mortality. Diagnosis is challenging due to overlap with non-infectious conditions, leading to a delayed or unnecessary antibiotic use. In a single-center cohort of VLBW infants, inflammatory biomarkers were elevated at the time of sepsis due to Gram-negative sepsis or NEC, but not other sepsis; compared to times without sepsis or NEC. Physiomarkers of sepsis correlate with some biomarkers of sepsis, and combining their information could help in the early diagnosis of sepsis. Competing interests: The authors have no competing interests to disclose. Consent Statement: The Institutional Review Board of University of Virginia gave approval for this study with verbal informed consent and a signed HIPAA data use authorization from a parent. ==== Refs Bibliography 1. Shane A. L. , Sánchez P. J. , & Stoll B. J. (2017). Neonatal sepsis. The Lancet, 390 (10104 ), 1770–1780. 10.1016/S0140-6736(17)31002-4 2. Garg P. M. , Paschal J. L. , Ansari M. A. Y. , Block D. , Inagaki K. , & Weitkamp J.-H. (2022). Clinical impact of NEC-associated sepsis on outcomes in preterm infants. Pediatric Research, 92 (6 ), 1705–1715. 10.1038/s41390-022-02034-7 35352003 3. Wynn J. L. , Kelly M. S. , Benjamin D. K. , Clark R. H. , Greenberg R. , Benjamin D. K. , , (2017). Timing of Multiorgan Dysfunction among Hospitalized Infants with Fatal Fulminant Sepsis. American Journal of Perinatology, 34 (7 ), 633–639. 10.1055/s-0036-1597130 27923248 4. Adams-Chapman I. , & Stoll B. J. (2006). Neonatal infection and long-term neurodevelopmental outcome in the preterm infant. Current Opinion in Infectious Diseases, 19 (3 ), 290–297. 10.1097/01.qco.0000224825.57976.87 16645492 5. Mukhopadhyay S. , Puopolo K. M. , Hansen N. I. , Lorch S. A. , DeMauro S. B. , Greenberg R. G. , , & NICHD Neonatal Research Network. (2021). Neurodevelopmental outcomes following neonatal late-onset sepsis and blood culture-negative conditions. Archives of Disease in Childhood. Fetal and Neonatal Edition, 106 (5 ), 467–473. 10.1136/archdischild-2020-320664 33478957 6. Cai S. , Thompson D. K. , Anderson P. J. , & Yang J. Y.-M. (2019). Short- and Long-Term Neurodevelopmental Outcomes of Very Preterm Infants with Neonatal Sepsis: A Systematic Review and Meta-Analysis. Children (Basel, Switzerland), 6 (12 ). 10.3390/children6120131 7. Shane A. L. , & Stoll B. J. (2014). Neonatal sepsis: progress towards improved outcomes. The Journal of Infection, 68 Suppl 1 , S24–32. 10.1016/j.jinf.2013.09.011 24140138 8. Cantey J. B. (2019). The spartacus problem: diagnostic inefficiency of neonatal sepsis. Pediatrics, 144 (5 ). 10.1542/peds.2019-2576 9. Chauhan N. , Tiwari S. , & Jain U. (2017). Potential biomarkers for effective screening of neonatal sepsis infections: An overview. Microbial Pathogenesis, 107 , 234–242. 10.1016/j.micpath.2017.03.042 28377234 10. Ng S. , Strunk T. , Jiang P. , Muk T. , Sangild P. T. , & Currie A. (2018). Precision medicine for neonatal sepsis. Frontiers in Molecular Biosciences, 5 , 70. 10.3389/fmolb.2018.00070 30094238 11. Ye Q. , Du L.Z. , Shao W.X. , & Shang S.Q. (2017). Utility of cytokines to predict neonatal sepsis. Pediatric Research, 81 (4 ), 616–621. 10.1038/pr.2016.267 27997530 12. Raynor L. L. , Saucerman J. J. , Akinola M. O. , Lake D. E. , Moorman J. R. , & Fairchild K. D. (2012). Cytokine screening identifies NICU patients with Gram-negative bacteremia. Pediatric Research, 71 (3 ), 261–266. 10.1038/pr.2011.45 22278182 13. Cantey J. B. , & Lee J. H. (2021). Biomarkers for the diagnosis of neonatal sepsis. Clinics in Perinatology, 48 (2 ), 215–227. 10.1016/j.clp.2021.03.012 34030810 14. Cantey J. B. , & Bultmann C. R. (2020). C-Reactive Protein Testing in Late-Onset Neonatal Sepsis: Hazardous Waste. JAMA Pediatrics, 174 (3 ), 235–236. 10.1001/jamapediatrics.2019.5684 32011645 15. Mussap M. (2012). Laboratory medicine in neonatal sepsis and inflammation. The Journal of Maternal-Fetal & Neonatal Medicine, 25 Suppl 4 , 32–34. 10.3109/14767058.2012.715000 16. Kausch S. L. , Moorman J. R. , Lake D. E. , & Keim-Malpass J. (2021). Physiological machine learning models for prediction of sepsis in hospitalized adults: An integrative review. Intensive & Critical Care Nursing : The Official Journal of the British Association of Critical Care Nurses, 103035. 10.1016/j.iccn.2021.103035 17. Kausch S. L. , Brandberg J. G. , Qiu J. , Panda A. , Binai A. , Isler J. , , (2023). Cardiorespiratory signature of neonatal sepsis: development and validation of prediction models in 3 NICUs. Pediatric Research. 10.1038/s41390-022-02444-7 18. Hibbert J. , Strunk T. , Simmer K. , Richmond P. , Burgner D. , & Currie A. (2020). Plasma cytokine profiles in very preterm infants with late-onset sepsis. Plos One, 15 (5 ), e0232933. 10.1371/journal.pone.0232933 32407417 19. Ruoss J. L. , & Wynn J. L. (2019). Biomarkers in the Diagnosis of Neonatal Sepsis. In Benitz W. E. & Smith P. B. (Eds.), Infectious Disease and Pharmacology: Neonatology Questions and Controversies (pp. 103–112). Elsevier. https://www.sciencedirect.com/science/article/pii/B9780323543910000096 20. Lam H. S. , & Ng P. C. (2008). Biochemical markers of neonatal sepsis. Pathology., 40 (2 ), 141–148.18203036 21. Schelonka R. L. , Maheshwari A. , Carlo W. A. , Taylor S. , Hansen N. I. , Schendel D. E. , , & NICHD Neonatal Research Network. (2011). T cell cytokines and the risk of blood stream infection in extremely low birth weight infants. Cytokine, 53 (2 ), 249–255. 10.1016/j.cyto.2010.11.003 21145756 22. Sood B. G. , Shankaran S. , Schelonka R. L. , Saha S. , Benjamin D. K. , Sánchez P. J. , , & Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network. (2012). Cytokine profiles of preterm neonates with fungal and bacterial sepsis. Pediatric Research, 72 (2 ), 212–220. 10.1038/pr.2012.56 22562288 23. Prieto C. L. , Colomer B. F. , & Sastre J. B. L. (2013). Prognostic factors of mortality in very low-birth-weight infants with neonatal sepsis of nosocomial origin. American Journal of Perinatology, 30 (5 ), 353–358. 10.1055/s-0032-1324701 22893550 24. Goh G. L. , Lim C. S. E. , Sultana R. , De La Puerta R. , Rajadurai V. S. , , (2021). Risk Factors for Mortality From Late-Onset Sepsis Among Preterm Very-Low-Birthweight Infants: A Single-Center Cohort Study From Singapore. Frontiers in Pediatrics, 9 , 801955. 10.3389/fped.2021.801955 35174116 25. Makhoul I. R. , Sujov P. , Smolkin T. , Lusky A. , & Reichman B. (2002). Epidemiological, clinical, and microbiological characteristics of late-onset sepsis among very low birth weight infants in Israel: a national survey. Pediatrics, 109 (1 ), 34–39. 10.1542/peds.109.1.34 11773539 26. Fairchild K. D. , Sinkin R. A. , Davalian F. , Blackman A. E. , Swanson J. R. , Matsumoto J. A. , , (2014). Abnormal heart rate characteristics are associated with abnormal neuroimaging and outcomes in extremely low birth weight infants. Journal of Perinatology, 34 (5 ), 375–379. 10.1038/jp.2014.18 24556979 27. Collaco J. M. , McGrath-Morrow S. A. , Griffiths M. , Chavez-Valdez R. , Parkinson C. , Zhu J. , , (2022). Perinatal inflammatory biomarkers and respiratory disease in preterm infants. The Journal of Pediatrics, 246 , 34–39.e3. 10.1016/j.jpeds.2022.04.028 35460699 28. Bose C. L. , Dammann C. E. L. , & Laughon M. M. (2008). Bronchopulmonary dysplasia and inflammatory biomarkers in the premature neonate. Archives of Disease in Childhood. Fetal and Neonatal Edition, 93 (6 ), F455–61. 10.1136/adc.2007.121327 18676410 29. Arnon S. , & Litmanovitz I. (2008). Diagnostic tests in neonatal sepsis. Curr.Opin.Infect Dis., 21 (3 ), 223–227.18448965 30. Küster H. , Weiss M. , Willeitner A. E. , Detlefsen S. , Jeremias I. , Zbojan J. , , (1998). Interleukin-1 receptor antagonist and interleukin-6 for early diagnosis of neonatal sepsis 2 days before clinical manifestation. The Lancet, 352 (9136 ), 1271–1277. 10.1016/S0140-6736(98)08148-3 31. Gonzalez B. E. , Mercado C. K. , Johnson L. , Brodsky N. L. , & Bhandari V. (2003). Early markers of late-onset sepsis in premature neonates: clinical, hematological and cytokine profile. Journal of Perinatal Medicine, 31 (1 ), 60–68. 10.1515/JPM.2003.009 12661146 32. Ng P. C. , Li K. , Wong R. P. O. , Chui K. , Wong E. , Li G. , , (2003). Proinflammatory and anti-inflammatory cytokine responses in preterm infants with systemic infections. Archives of Disease in Childhood. Fetal and Neonatal Edition, 88 (3 ), F209–13. 10.1136/fn.88.3.f209 12719394 33. Nunes T. , Bernardazzi C. , & de Souza H. S. (2014). Interleukin-33 and inflammatory bowel diseases: lessons from human studies. Mediators of Inflammation, 2014 , 423957. 10.1155/2014/423957 24701033 34. Madden G. R. , Rigo I. , Boone R. , Abhyankar M. M. , Young M. K. , Basener W. , , (2023). Novel Biomarkers, Including tcdB PCR Cycle Threshold, for Predicting Recurrent Clostridioides difficile Infection. Infection and Immunity, 91 (4 ), e0009223. 10.1128/iai.00092-23 36975808 35. Abhyankar M. M. , Ma J. Z. , Scully K. W. , Nafziger A. J. , Frisbee A. L. , Saleh M. M. , , (2020). Immune Profiling To Predict Outcome of Clostridioides difficile Infection. MBio, 11 (3 ). 10.1128/mBio.00905-20 36. Ng P. C. , Ang I. L. , Chiu R. W. K. , Li K. , Lam H. S. , Wong R. P. O. , , (2010). Host-response biomarkers for diagnosis of late-onset septicemia and necrotizing enterocolitis in preterm infants. The Journal of Clinical Investigation, 120 (8 ), 2989–3000. 10.1172/JCI40196 20592468 37. Eschborn S. , & Weitkamp J.-H. (2019). Procalcitonin versus C-reactive protein: review of kinetics and performance for diagnosis of neonatal sepsis. Journal of Perinatology, 39 (7 ), 893–903. 10.1038/s41372-019-0363-4 30926891 38. Brown J. V. E. , Meader N. , Wright K. , Cleminson J. , & McGuire W. (2020). Assessment of C-Reactive Protein Diagnostic Test Accuracy for Late-Onset Infection in Newborn Infants: A Systematic Review and Meta-analysis. JAMA Pediatrics, 174 (3 ), 260–268. 10.1001/jamapediatrics.2019.5669 32011640
PMC010xxxxxx/PMC10327271.txt
==== Front medRxiv MEDRXIV medRxiv Cold Spring Harbor Laboratory 37425854 10.1101/2023.06.29.23292048 preprint 1 Article Expert-level pediatric brain tumor segmentation in a limited data scenario with stepwise transfer learning Boyd Aidan 12+ Ye Zezhong 12+ Prabhu Sanjay 3+ Tjong Michael C. 2 Zha Yining 12 Zapaischykova Anna 12 Vajapeyam Sridhar 3 Hayat Hasaan 12 Chopra Rishi 12 Liu Kevin X. 2 Nabavidazeh Ali 45 Resnick Adam 46 Mueller Sabine 789 Haas-Kogan Daphne 2 Aerts Hugo J.W.L. 121011 Poussaint Tina 3 Kann Benjamin H. 12* 1. Artificial Intelligence in Medicine (AIM) Program, Mass General Brigham, Harvard Medical School, Boston, MA 2. Department of Radiation Oncology, Dana-Farber Cancer Institute and Brigham and Women’s Hospital, Harvard Medical School, Boston, MA 3. Department of Radiology, Boston Children’s Hospital, Harvard Medical School, Boston, MA 4. Center for Data-Driven Discovery in Biomedicine (D3b), Children’s Hospital of Philadelphia, Philadelphia, PA 5. Department of Radiology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA 6. Department of Neurosurgery, Children’s Hospital of Philadelphia, Philadelphia, PA 7. Department of Neurology, University of California San Francisco, San Francisco, California 8. Department of Pediatrics, University of California San Francisco, San Francisco, California 9. Department of Neurological Surgery, University of California San Francisco, San Francisco, California 10. Department of Radiology, Brigham and Women’s Hospital, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, USA 11. Radiology and Nuclear Medicine, CARIM & GROW, Maastricht University, Maastricht, the Netherlands + These authors contributed equally to this manuscript. Author Contributions Study design: A.B., B.H.K.; code design, implementation and execution: A.B.; acquisition, analysis or interpretation of data: A.B., Z.Y., S.V., R.C., B.H.K.; image annotation: B.H.K., S.P., M.T.; writing of the manuscript: A.B., Z.Y., B.H.K.; critical revision of the manuscript for important intellectual content: all authors; statistical analysis: Z.Y., Y.Z.; study supervision: B.H.K., H.J.W.L.A., T.P., D.H.K. * Correspondence address to: Benjamin H. Kann, M.D., Department of Radiation Oncology, Dana-Farber Cancer Institute and Brigham and Women’s Hospital, Harvard Medical School, 75 Francis Street, Boston, MA 02115, MA, USA, Tel: +1 617-732-6310, [email protected] 30 6 2023 2023.06.29.23292048https://creativecommons.org/licenses/by-nc/4.0/ This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License, which allows reusers to distribute, remix, adapt, and build upon the material in any medium or format for noncommercial purposes only, and only so long as attribution is given to the creator. nihpp-2023.06.29.23292048.pdf Purpose: Artificial intelligence (AI)-automated tumor delineation for pediatric gliomas would enable real-time volumetric evaluation to support diagnosis, treatment response assessment, and clinical decision-making. Auto-segmentation algorithms for pediatric tumors are rare, due to limited data availability, and algorithms have yet to demonstrate clinical translation. Methods: We leveraged two datasets from a national brain tumor consortium (n=184) and a pediatric cancer center (n=100) to develop, externally validate, and clinically benchmark deep learning neural networks for pediatric low-grade glioma (pLGG) segmentation using a novel in-domain, stepwise transfer learning approach. The best model [via Dice similarity coefficient (DSC)] was externally validated and subject to randomized, blinded evaluation by three expert clinicians wherein clinicians assessed clinical acceptability of expert- and AI-generated segmentations via 10-point Likert scales and Turing tests. Results: The best AI model utilized in-domain, stepwise transfer learning (median DSC: 0.877 [IQR 0.715–0.914]) versus baseline model (median DSC 0.812 [IQR 0.559–0.888]; p<0.05). On external testing (n=60), the AI model yielded accuracy comparable to inter-expert agreement (median DSC: 0.834 [IQR 0.726–0.901] vs. 0.861 [IQR 0.795–0.905], p=0.13). On clinical benchmarking (n=100 scans, 300 segmentations from 3 experts), the experts rated the AI model higher on average compared to other experts (median Likert rating: 9 [IQR 7–9]) vs. 7 [IQR 7–9], p<0.05 for each). Additionally, the AI segmentations had significantly higher (p<0.05) overall acceptability compared to experts on average (80.2% vs. 65.4%). Experts correctly predicted the origins of AI segmentations in an average of 26.0% of cases. Conclusions: Stepwise transfer learning enabled expert-level, automated pediatric brain tumor auto-segmentation and volumetric measurement with a high level of clinical acceptability. This approach may enable development and translation of AI imaging segmentation algorithms in limited data scenarios. Summary Authors proposed and utilized a novel stepwise transfer learning approach to develop and externally validate a deep learning auto-segmentation model for pediatric low-grade glioma whose performance and clinical acceptability were on par with pediatric neuroradiologists and radiation oncologists. MRI Pediatric Low-Grade Glioma Automatic Tumor Segmentation nnUNet Deep Neural Network National Institutes of HealthU24CA194354 U01CA190234 U01CA209414 R35CA22052 K08DE030216 National Cancer Institute2P50CA165962 European Union – European Research Council866504 Radiological Society of North AmericaRSCH2017 ==== Body pmcINTRODUCTION Pediatric low-grade gliomas (pLGGs) are the most common type of brain tumors in children, accounting for approximately 30% of all pediatric brain tumors (1). pLGGs are heterogeneous in their molecular underpinnings, natural history, and aggressiveness, and therapies carry significant risks, thus making management decisions challenging (2–4). Optimal risk-stratification, response assessment, and surveillance for pLGG hinge on the ability to accurately localize and characterize brain tumors on magnetic resonance imaging (MRI) scans, which, in turn, relies on accurate tumor segmentation. Compared to adult gliomas, manual segmentation of pLGG carries distinct challenges, requiring expertise, resources, and time, thus limiting its clinical efficiency (5). Given the utility of accurate segmentation and the inherent limitations of manual segmentation, there has been interest in developing auto-segmentation tools for pediatric brain tumors (6–8). The progress in medical imaging techniques and computational methods have led to the development of various approaches for brain tumor segmentation, including machine learning-based methods, deep learning-based methods, and hybrid approaches (8–10). Recently, deep learning has emerged as a powerful tool in medical imaging, offering solutions to diverse clinical challenges (11–15). Auto-segmentation based on deep learning is thus a promissing approach for accurate and efficient brain tumor segmentation, including pLGGs (6,16,17), though distinct challenges remain. In particular, pLGGs are relatively rare tumors and there are no publicly available datasets for training models. Most brain tumor segmentation algorithms have been developed for adult glioma, which are much more common and have large volumes of public and institutional data for training (18,19). In contrast, there has been only limited study of dedicated pediatric glioma segmentation - with a paucity of pLGG-specific models that rely on small, single-institution datasets that have not been externally validated nor subjected to clinical testing (16,17). Human clinical evaluation of segmentation models is essential to benchmark performance to experts and determine their true level of performance and potential for clinical translation. There are numerous proposed ways to improve deep learning performance in limited-data settings (20). Aside from traditional techniques like data augmentation, ensembling, regularization, recently, advances have been made in knowledge transfer-learning (21) and self- (22) and semi-supervision (23). These have shown promise in improving medical image analysis algorithms, though can be challenging to implement. Pediatric brain tumors represent an ideally situated setting for which to apply these techniques, given the relative scarcity of data. Here, we aim to bridge the translational gap for pediatric brain tumor segmentation algorithms and achieve clinically acceptable performance despite a limited-data scenario with several innovations: 1) implementation of in-domain stepwise transfer learning; 2) aggregation of the largest pLGG imaging database to-date, and 3) performing blinded human acceptability testing. MATERIALS AND METHODS Study Design and Datasets This study was conducted in accordance with the Declaration of Helsinki guidelines and following the approval of the Dana-Farber/Boston Children’s/Harvard Cancer Center Institutional Review Board (IRB). Waiver of consent was obtained from IRB prior to research initiation due to public datasets or retrospective study. Data from one high-volume academic institution and one national consortium from 2001 through 2015 were included. The Children’s Brain Tumor Network (CBTN) dataset consists of one de-identified patient cohort (n=187). The Boston Children’s Hospital (BCH) dataset includes one patient cohort (n=100). These subsets represent all scans that passed initial quality control of DICOM metadata. Patient inclusion criteria were the following: 1) 0–18 years of age, 2) histopathologically confirmed pLGG, and 3) availability of preoperative brain MR imaging with a T2-weighted (T2W) imaging sequence. Spinal cord tumors were not considered for this study. Scans with spine tumors (n=12) and undetectable lesions (n=3) were excluded from analyses. Publicly available data from the Brain Tumor Segmentation (BraTS) 2021 competition was acquired from http://braintumorsegmentation.org (24–26). The BraTS competition data represents 1,251 adult glioma multiparametric MR scans along with expert-generated segmentation masks for T1-weighted (T1W), T2W, fluid attenuated inversion recovery (FLAIR), and contrast enhancement generated by radiologists (24,25). Full data description for BraTS is available here: http://braintumorsegmentation.org. MR Imaging Acquisition and Parameters Patients from the CBTN cohort underwent brain MR imaging at 1.5T or 3T Siemens MR scanners (Table S1). Sequences acquired included 2D axial T2-weighted turbo spin-echo (TR/TE, 1000–7300/80–530 ms; 0.5- to 5-mm section thickness), 3D axial or sagittal pre-contrast, and 3D axial gadolinium-based contrast agent–enhanced T1-weighted turbo or fast-field echo. Patients from the BCH cohort underwent brain MR imaging at 1.5T or 3T from various MRI vendors (Table S2). MRIs were performed using the brain tumor protocol of the institution, which included 2D axial T2-weighted fast spin-echo (TR/TE, 2600–12,000/11–120 ms; 2- to 5.5-mm section thickness), 2D axial or sagittal pre-contrast T1-weighted spin- echo, 2D axial T2 FLAIR and 2D axial gadolinium-based contrast agent–enhanced T1-weighted spin-echo sequences (Table S1–S2; Fig. S1–S4). All MR imaging data were extracted from the respective PACS and were de-identified for further analyses. Given that many pLGGs do not enhance with intravenous contrast, are hypointense on T1, and are hyperintense on T2-weighted sequences, we chose to develop a T2-weighted segmentation algorithm with a primary use case of volumetric tumor monitoring. MR Image Preprocessing MRI images were converted from DICOM format to NIFTI format via rasterization packages utilizing dcm2nii package (https://www.nitrc.org/projects/dcm2nii) in Python v3.8. N4 bias filed correction was adopted to correct the low-frequency intensity non-uniformity present in MRI images using SimpleITK (SITK) in Python v3.8. All scans were resampled to 1×1×3 mm3 voxel size using linear interpolation via SITK. After interpolation, the MR scans were co-registered using a rigid registration step with SITK. Lastly, a brain extraction step was performed for all the scans using HD-BET package in Python v3.8 (27). Preprocessing scripts are found here: https://github.com/BoydAidan/nnUNet_pLGG. MR image review and segmentation For model development and initial training, tumors in all T2w scans within the CBTN and 60 scans within the BCH cohort were initially segmented by a board-certified radiation oncologist (experience: 8 years) to serve as primary ground truth segmentations. Annotators were instructed to segment all areas of T2 signal abnormality concerning for tumor involvement, including areas of peritumoral T2 hyperintensity if suspicious for tumor involvement. Segmentations were performed and saved in NIFTI format using ITK-SNAP v4.0 (http://www.itksnap.org) in 3D utilizing axial, sagittal, and coronal views (Fig. S5). Deep learning approach: stepwise, in-domain transfer learning For this work, the nnUNet architecture (28) which is a deep learning-based segmentation approach that configures itself completely automatically, was chosen for the deep learning framework. nnUNet includes auto-configuration for preprocessing, network architecture selection, model training and post-processing and can be applied to any task. nnUNet performs well out of the box on many medical imaging tasks, and it has been specifically shown to produce strong results for brain tumor segmentation (28). The built-in ensembling of nnUNet was utilized for training and inference. Early stopping was implemented such that once the model does not improve on the validation set after 50 epochs, the training is stopped, up to a maximum of 1000 epochs. All other training parameters (learning rate, batch size, data augmentations, loss function) are the default nnUNet settings (28). All training settings described below are trained in the same manner. Adult brain tumor model performance [BraTS model] Competitions such as Brain Tumor Segmentation (BraTS) Challenge 2012–2021 have shown that deep learning-based solutions can effectively segment adult gliomas, mainly of the high-grade variety (18). Given the proven performance of nnUNet on adult tumors, we sought to determine the performance of these models when applied to the pediatric setting. We hypothesized that, given significant biological and morphological differences between adult and pediatric gliomas, adult tumor-trained model performance would degrade in the pediatric setting. To test this hypothesis, an nnUNet model was trained using the BraTS 2021 training dataset, which contains 1,251 MRI scans and associated expert annotations. To compare performance in the adult and pediatric settings, inference was performed first on an adult hold-out set from BraTS 2021. Inference was then completed on the internal CBTN pediatric test set (n=60). Training from scratch [Scratch model] Given the similarities in data distribution, we hypothesized that a model trained on pediatric brain tumors would perform better in a pediatric test set than a model trained on adult tumors. To test this hypothesis, using the CBTN training dataset (n=148), an nnUNet model was trained in the same way as the adult model. This model employed only the limited pediatric data available and, as such, relied on much less training data. In-domain transfer learning from adults [Transfer model] Transfer learning involves leveraging data from other sources, that is often out of distribution, domain, or modality, to initialize models with foundational knowledge that can be then fine-tuned to a particular task (21). This technique has been employed successfully in several medical imaging applications and is commonly performed by leveraging models trained on the 2D ImageNet (29). More recently, 3D medical imaging-specific pretrained models have become available and have demonstrated improved performance compared to ImageNet-trained models on medical tasks. Multiple studies suggest that the closer the data distribution between pretrained neural network and the task, the better (30). Thus, in the case of limited data scenario of pLGG segmentation, we sought to leverage in-domain transfer learning from the adult model combined with pediatric data. Specifically, the nnUNet model is initialized (i.e., pretrained) with the Adult model weights and then fine-tuned with additional training on the CBTN data under the same procedure as the Scratch model. Stepwise Transfer Learning Due to the limited quantity of pediatric data available for training, we hypothesized that reducing the number of parameters a network must optimize may help improve model convergence. This can be achieved by freezing specific model parameters during the fine-tuning process. This effectively reduces the complexity of the network for the optimizer. When the model is optimizing both the encoder and decoder simultaneously, it may not be able to find the optimal solution given the sparsity of data. Starting from the checkpoint of Transfer model, training is continued with all parameters frozen in either the nnUnet encoder block (Transfer-Encoder model) or decoder block (Transfer-Decoder model). Model evaluation and statistical analysis The primary performance endpoint of the study was the Dice similarity coefficient (DSC) (31), which measures the relative overlap of the predicted and ground-truth segmentations, with a DSC of >0.80 considered to be worthy of clinical testing (10). DSC was calculated based on Equation [1]: [1] DSC=2∑k yˆkyk∑k yˆk+yk where yk is the ground truth for voxel k of one image, and yˆk is the prediction for voxel k of that image. Median DSCs between models were compared with Wilcoxon Rank Sum tests within the CBTN test set. The highest performing model on the CBTN test set was then externally validated on the BCH dataset (n=60). Additionally, 3D volumetric measurement was calculated from each predicted and ground-truth segmentation, and median relative volume difference (RVD) and absolute volume difference (AVD) were evaluated. Other secondary endpoints included aggregated DSC, which calculates DSC over the entire test set population (rather than case-by-case), and intraclass coefficient (ICC) for calculated volumes. Aggregated DSC (DSCagg) was calculated based on Equation [2]: [2] DSCagg=2∑iN ∑k yˆi,kyi,k∑iN ∑k yˆi,k+yi,k where N is the number of test images, yi,k is the ground truth for voxel k and image i, and yˆi,k is the prediction for voxel k and image i. Means were compared with ANOVA. All tests were two-sided with a p<0.05 considered statistically significant. Statistical analyses were conducted using the Scikit-learn and SciPy packages in Python v3.8. Three inter-rater group DSCs were combined between three experts, between model 1 and experts, and between model 2 and experts. Mean inter-rater group DSCs between combined expert models and two AI models were compared with One-way ANOVA. Mean ratings of three individual experts, combined experts, and two individual segmentation models were compared by One-way ANOVA. Tukey’s ‘Honest Significant Difference’ method was used as post-hoc analysis to find specific groups with significantly different mean DSCs or mean ratings from other groups. All tests were two-sided with a p<0.05 considered statistically significant. Statistical analyses were conducted using R. Randomized, blinded clinical acceptability testing and inter-expert variability While DSC is an important quantitative measure of segmentation performance, positioning the algorithm for real-world use requires clinical validation and benchmarking (32,33). To determine segmentation variation across expert clinicians and diagnosticians of different specializations, a second radiation oncologist specializing in central nervous system tumors and a pediatric neuroradiologist were enlisted to annotate the entire BCH external dataset (n=100) for clinical acceptability testing (Fig. 1B). Pairwise inter-expert variability between the three annotators as well as two deep learning models (BraTS model and Transfer-Encoder model) was similarly evaluated with DSC to determine if model performance was within the range of inter-expert variability. To assess the clinical utility of the AI models, the three experts conducted a blinded, segmentation rating and acceptability study (Fig. 1C). For each of the 100 BCH cases, each expert was presented with three different segmentations overlaid with each other, with the ability to hide/unhide each individual segmentation as needed (Supplementary Protocol). The three segmentations consisted of at least one, and up to two expert segmentations (from the other two annotators) and at least one, and up to two AI-generated generated segmentations (selected from the BraTS model and/or the Transfer-Encoder model), selected at random (n=300 total segmentations per reviewer). The expert reviewers were blinded to the origin of the segmentations. The order and color of the segmentations displayed for each scan was randomized to reduce bias. The review was carried out using 3D Slicer (www.slicer.org). Experts were given written instructions and asked to rate each of the three segmentations on a scale from 1 (worst) to 10 (perfect), with a 7 being defined as a clinically acceptable segmentation for volumetric assessment (Fig. S6). For each segmentation, raters were also asked to choose whether the segmentation was AI-generated, also known as the Turing test. RESULTS Patient Characteristics The total pLGG patient cohort consisted of 284 pLGG patients from two cohorts, with 184 patients in the development set from CBTN cohort and 100 patients in the external test set from BCH (Table 1). Median age was 7 (range: 1–23) in the CBTN cohort and 8 (range: 1–19) in the BCH cohort. 84 patients (45.7%) were female in CBTN cohort, while 53 patients (53%) were female in BCH cohort. All patients had pathologically diagnosed grade I/II low-grade glioma, with a mixture of histologic subtypes of pilocytic astrocytoma (32.4%), optic pathway glioma (10.6%), juvenile pilocytic astrocytoma (4.6%), fibrillary astrocytoma (4.6%), ganglioglioma (4.2%), diffuse astrocytoma (4.2%), and other low-grade glioma/astrocytoma (44.0%). The primary tumor locations were posterior fossa (25.7%), temporal lobe (10.9%), Suprasellar (9.5%), frontal lobe (3.2%), and others (50.5%). In-domain Stepwise Transfer learning with fine-tuning improves performance The performance of the five model settings outlined in Fig. 1 is illustrated in Table 2. The BraTS model performed highly on held-out adult data, however, the performance of the model decreased significantly when applied to the pediatric data (Table 2; median DSC 0.926 [IQR 0.886–0.953] to 0.812 [IQR 0.559–0.888], p<0.05). Additionally, accuracy of volumetric assessment declined significantly when the BraTS model was applied to pediatric data (Table 2; median RVD: 0.052 [IQR 0.024–0.119] to 0.192 [IQR 0.109–0.682], p<0.05). This result highlighted the difference between adult gliomas and pediatric low-grade gliomas and emphasized the importance of developing a model capable of accurately segmenting pediatric cases. The Scratch model and all Transfer models performed significantly better (p<0.05 for each) than the BraTS model, with median DSC of 0.862 [IQR 0.672–0.91], 0.871 [IQR 0.724–0.914], 0,877 [IQR 0.708–0.914] and 0.877 [IQR 0.715–0.914] for Scratch, Transfer, Transfer-Decoder, and Transfer-Encoder models, respectively (Table 2 & Fig. 2). Of all approaches investigated, the highest performing was the Transfer-Encoder model (Table 2; median DSC: 0.877 [IQR 0.715–0.914]; median relative volume difference (RVD): 0.109 [IQR 0.032–0.31]). While both fine-tuned transfer learning models showed near equivalent performance for the primary endpoint, we chose Transfer-Encoder model for further testing giving its increased aggregated DSC (Table 2; aggregated DSC: 0.730 to 0.840), which indicated it had fewer incidences of empty segmentation masks. Overall, Transfer-Encoder model has the lowest empty mask percentage. External validation of Stepwise Transfer Learning On an external testing set with expert segmentations (n=60), the Transfer-Encoder model achieved median DSC 0.833 [IQR 0.743–0.900] and median RVD of 0.161 [IQR 0.058–0.393] as compared to manual segmentations. We performed failure analysis for cases with DSC<0.6 and found 6 cases in total. The failures were caused by: 1) tumor located in ventricle (Fig. 3E; n=1); 2) large cystic area in brain (Fig. 3A, C&D; n=3); empty segmentation from poor image quality due to respacing for large slice thickness (Fig. 3B; n=1); 4) under-segmentations for large heterogeneous tumor lesion (Fig. 3F; n=1). Stepwise transfer learning demonstrates clinical expert-level performance The distribution of inter-rater DSCs between experts and AI models is presented in Fig. 4A, while the heatmap in Fig. 4B displays the median inter-rater DSCs for each pair. We further compared the inter-rater DSCs from three experts, the Transfer-Encoder model, and the BraTS model. The Transfer-Encoder model (median: 0.834 [IQR 0.726–0.901]) did not show a significant difference (p=0.13) compared to the three experts (median: 0.861 [IQR 0.795–0.905]), but it exhibited a significantly higher DSC (p<0.01) than the BraTS model (median: 0.790 [IQR 0.662–0.870]) (Fig. 4C). For clinical acceptability testing, the segmentation ratings of the Transfer-Encoder model (median: 9 [IQR 7–9]) were significantly higher (p<0.01 for each) than those of Expert 1 (median: 7 [IQR 6–9]), Expert 3 (median: 7 [IQR 5–8]), the average expert (median: 7 [IQR 6–9]), and the BraTS model (median: 8 [IQR 6–9]). However, there was no significant difference between the Transfer-Encoder model and Expert 2 (median: 8 [IQR 7–9]) (Fig. 5A). Expert 2 had significantly higher ratings (p<0.01 for each) compared to Expert 1, Expert 3, and the average experts. There was no significant difference (p=0.54) between Expert 1 and Expert 3 in terms of the ratings (Fig. 5A). Furthermore, the Transfer-Encoder model demonstrated a significantly higher (p<0.05 for each) proportion of clinically acceptable segmentations (rating score⩾7) compared to two out of three experts (Expert 1 & Expert 3) and the experts average (Fig. 5B: Transfer-Encoder: 80.2%; BraTS: 72.1%; Expert 1: 68.3%; Expert 2: 78.7%; Expert 3: 49.3%; Experts average: 64.8%). Finally, results from the Turing test revealed consistently low accuracy in distinguishing AI-generated segmentations from those produced by experts. The Transfer-Encoder model segmentations were correctly identified by experts in only 26.0% of scans, which is lower than the correct identification rates for Expert 1 (35.0%), Expert 2 (66.5%), Expert 3 (26.9%), and the BraTS model (41.5%) (Fig. 5C). DISCUSSION In this study, we developed, validated, and clinically benchmarked a deep learning pipeline using stepwise transfer learning for automated, expert-level pLGG segmentation and volumetric measurement. Accurate tumor auto-segmentation models could be useful for risk-stratification, monitoring tumor progression, assessing treatment response, and surgical approach (6), though have had limited traction in pediatric tumors due to very sparse available training data. We leveraged a novel strategy of in-domain, stepwise transfer learning to demonstrate measurable gains in segmentation accuracy and clinical acceptability that was on par with clinician performance. To our knowledge, this is the first study to utilize stepwise transfer learning in this context and to evaluate clinical acceptability of auto-segmentation tools. The rigorous clinical benchmarking studies with three blinded experts suggest that this approach is nearing a performance ceiling for pLGG segmentation – i.e., output is comparable and indistinguishable to human experts. These findings position the model for prospective testing and clinical translation. The current state-of-the-art approaches for automated brain tumor segmentation rely on deep learning. However, most available auto-segmentation tools have been specifically developed and trained for adult brain cancers, particularly glioblastoma (GBM) (9,18,19). In this work, we find that tools such as these do not effectively generalize to pediatric brain tumors. Performance degradation may stem from the distinctive heterogeneous imaging appearance and types of pediatric brain tumors compared to adult brain tumors, as well as the anatomical differences resulting from the ongoing brain development in children. Several studies have proposed various DL solutions to address the segmentation of pediatric brain tumors, achieving Dice similarity coefficients (DSC) ranging between 0.68 and 0.88 (6,16,34,35), however the clinical acceptability of these approaches was not validated. To date, only one study has proposed an algorithm for pLGG segmentation, achieving a DSC of 0.77 (17). This study utilized FLAIR images from 311 patients from a single institution. The proposed model employed deep Multitask Learning, incorporating a tumor’s genetic alteration classifier as an auxiliary task to the main segmentation network (17). However, this model was trained on a limited number of MRI scans from a single institute and lacked external validation and clinical testing. In the present study, our stepwise transfer learning model achieved median DSC of 0.833 [IQR 0.743–0.900] with a median RVD of 0.161 [IQR 0.058–0.393] in the external test set, which represents a significant improvement over previous work (17). Improved performance may be due to sequential knowledge transfer - first from the adult setting, and then the pediatric setting. Additionally, freezing the encoder or decoder in the final fine-tuning step enabled optimization of a smaller parameter space, which may have mitigated overfitting given the limited amount of data. Training on a sufficient quantity of pediatric data from scratch may obviate the need for transfer learning, but what represents ”sufficient” is yet to be defined for pLGGs, and current datasets remain relatively small. While statistical metrics like the DSC and RVD offer valuable insights into a model’s overall segmentation performance, it is important to acknowledge their limitations in providing a comprehensive evaluation of a model’s utility (6). To ensure a thorough evaluation and facilitate the clinical translation of our model, we conducted a rigorous clinical-acceptability evaluation and validation process involving three expert clinicians. This multidimensional evaluation approach captures additional nuances and considerations that impact the model’s practical utility in a real-world clinical setting. The involvement of expert clinicians provides valuable feedback and insights, accelerating the translation of the model into clinical practice. In our study, we went a step further by conducting blinded, segmentation rating, acceptability, and Turing tests involving three expert clinicians. Notably, all experts performed worse than random chance (50%) in predicting the origin of the transfer learned model segmentations, suggesting that this model passes the Turing test, though this was not the case for the adult-trained model. To our knowledge, this is the first brain tumor segmentation study to incorporate such a clinical-acceptability test. The results highlight the importance of such clinical testing in positioning a model for potential clinical translation. There are several limitations of this study that should be taken into consideration. Firstly, the study is retrospective in nature and selection of scans for inclusion for this study, while performed a priori and based solely on availability, may introduce bias. Secondly, the model utilizes only T2W images, as these were the most commonly available for all the patients included in our analysis. Although T2 FLAIR images are commonly used for defining tumor regions, T2 FLAIR images were not available for many patients with pLGGs in our study, particularly in the CBTN cohort. Consequently, it becomes challenging to distinguish vasogenic edema from the tumor region on T2W images, making it difficult to accurately segment tumor areas based solely on T2W images. Despite these challenges, our study demonstrated that our DL model exhibited impressive segmentation performance compared to expert clinicians, suggesting that the model successfully captured the feature differences. However, we acknowledge that the incorporation of multiple MRI modalities, such as T2W, pre-contrast T1W, post-contrast T1W, and T2 FLAIR images, would enhance the granularity of tumor segmentation in pLGG patients. On the other hand, an advantage of a T2W-only model, is that it may be more widely applicable in situations where multiparametric and contrast-enhanced scans are unavailable, which can sometimes be the case for low-grade glioma studies. Furthermore, it is important to note that our study focused solely on whole tumor segmentation, and not the segmentation of specific tumor subregions. Consequently, the clinical utility of our findings may be restricted in certain cases, when change in cystic component is not relevant to the clinical response assessment. Finally, it is notable that the algorithm did fail on some cases, and while we identified certain factors that were associated with failures, it is difficult to predict with certainty why a failure occurred owing to the black-box nature of deep learning algorithms. Therefore, it is important for the model output to undergo a clinical review prior to use in clinical decision-making. CONCLUSION We developed, externally validated, and clinically benchmarked an automated deep learning pipeline using in-domain, stepwise transfer learning that enables expert-level MRI segmentation of pediatric low-grade gliomas. On blinded evaluation, the model demonstrated clinically acceptable performance that was higher on average than clinical experts. Prospective and longitudinal evaluation of the pipeline is planned to determine the algorithm’s potential for integration into the clinical care of children with low-grade glioma. Supplementary Material Supplement 1 Funding This study was supported in part by the National Institutes of Health (NIH) (U24CA194354, U01CA190234, U01CA209414, R35CA22052, and K08DE030216), the National Cancer Institute (NCI) Spore grant (2P50CA165962), the European Union – European Research Council (866504), the Radiological Society of North America (RSCH2017), the Pediatric Low-Grade Astrocytoma Program at Pediatric Brain Tumor Foundation, and the William M. Wood Foundation. Data availability BraTS data including raw MRI images may be requested from The Cancer Image Archive (https://www.cancerimagingarchive.net). Although raw MRI imaging data cannot be shared, all measured results to replicate the statistical analysis are shared at the GitHub webpage: https://github.com/BoydAidan/nnUNet_pLGG. Furthermore, we include test samples from a publicly available data set with deep learning and expert reader annotations. Abbreviations: AUC area under the curve CNN convolutional neural network DICOM Digital Imaging and Communications in Medicine pLGG pediatric low-grade glioma DSC dice similarity coefficient Figure 1. Schematic illustration of the study design. (A) An overview of the study workflow, including data preprocessing, expert segmentation on the tumors, model training/testing and well as the model clinical acceptability evolution. (B) A workflow showing the proposed in-domain stepwise transfer learning with detailed sequential steps involved in this approach. (C) A workflow detailing the 2-phase clinical acceptability evaluation. Figure 2. Comparative performance of deep learning training methodologies on internal validation set (n=60). Among the five different segmentation models assessed, the methods using stepwise transfer learning (Transfer-Decoder and Transfer-Encoder) had the highest segmentation accuracy. Additionally, the Transfer-Encoder model generated the fewest segmentations with a DSC: 0 (n=4, 6.7%) (indicating a complete segmentation miss). Conversely, the BraTS model exhibited the highest number of segmentations with a DSC: 0 (n=11, 18.3%). The Transfer-Encoder model demonstrated the highest median DSC (0.877 [IQR 0.715–0.914]), and was selected for further investigation. The BraTS model, trained only on adult glioma, demonstrated the lowest median DSC (0.812 [IQR 0.559–0.888]). Figure 3. Analysis of Transfer-Encoder model failures. Failure analysis was performed based on cases with DSC<0.6 and 6 cases were identified in total. The failures were caused by big cystic area in brain (A, C&D), empty segmentation from poor image quality due to respacing for large slice thickness (B), tumor located in ventricle (E), under-segmentations for large heterogeneous tumor lesion (F). Figure 4. Clinical acceptability testing. Three human experts were invited to rate randomized AI-generated (BraTS model or Transfer-Encoder model) or expert-generated segmentations blinded to segmentation origin (n=100 scans, 300 segmentations). (A) The distributions of inter-rater DSCs from three experts and two AI models for external test dataset (n=100). (B) The median inter-rater DSCs between each pair of the experts and AI models. (C) Boxplot shows inter-rater DSCs grouped by experts, Transfer-Encoder model and BraTS model. Both Transfer-Encoder model and average experts shows significantly higher (p<0.05) inter-rater DSCs than BraTS model. There was no significant difference on inter-rater DSCs between average experts and Transfer-Encoder model. DSC: Dice similarity coefficient; AI: artificial intelligence; E1: Expert 1; E2: Expert 2; E3: Expert 3; AI1: BraTS model; AI2: Transfer-Encoder model. Figure 5. (A) The mean segmentation rating scores for the three experts and two AI models, as evaluated by the experts, are presented in a heatmap. Violin plots are used to compare and group the rating scores for each expert and AI model, along with the corresponding p-values from statistical tests for group comparisons. (B) The percentage of segmentations that were considered acceptable, with a rating score of 7 or above, is summarized for each expert and model. (C) The accuracy of determining whether annotations were AI generated is shown for each expert and model, indicating how well the experts were able to distinguish between AI-generated and expert-generated segmentations. DSC: Dice similarity coefficient; AI: artificial intelligence; E1: Expert 1; E2: Expert 2; E3: Expert 3; AI1: BraTS model; AI2: Transfer-Encoder model. DSC: Dice similarity coefficient; AI: artificial intelligence; E1: Expert 1; E2: Expert 2; E3: Expert 3; Es: average experts; AI1: BraTS model; AI2: Transfer-Encoder model. Table 1. Patient demographics. Patient Cohorts CBTN (n = 184) BCH (n = 100) Age (years) median (range) 7 (1 – 23) 8 (1 – 19) Sex n (%) Female 84 (45.7%) 53 (53%) Male 94 (51.1%) 47 (47%) Unknown 6 (3.2%) 0 Race/Ethnicity n (%) Non-Hispanic Caucasian/white 124 (67.4%) 68 (68%) African American/Black 24 (13.0%) 5 (5%) Hispanic/Latinx 17 4 (4%) Asian American/Asian 3 (1.6%) 4 (4%) Other/Unknown 33 (17.9%) 19 (19%) Histologic diagnosis n (%) pilocytic astrocytoma 58 (31.5%) 34 (34%) Pilomyxoid Astrocytoma 10 (5.4%) 0 Juvenile Pilocytic Astrocytoma 0 13 (13%) Ganglioglioma 1 (0.5%) 11 (11%) Oligodendroglioma 1 Diffuse astrocytoma 9 (4.9%) 3 (3%) Fibrillary Astrocytoma 13 (7.1%) 0 Optic pathway glioma 27 (14.7%) 3 (3%) Other Low-grade glioma/astrocytoma 65 (35.3%) 36 (36%) Primary Tumor Location n (%) Posterior fossa 45 (24.5%) 28 (28%) Temporal lobe 13 (7.1%) 18 (18%) Frontal Lobe 7 (3.8%) 2 (2%) Cerebellum 0 18 (18%) Suprasellar 21 (11.4%) 6 (6%) Optic Pathway 27 (14.7%) 3 (3%) Brainstem 9 (4.9%) 3 (3%) Thalamus 5 (2.7%) 2 (2%) Others 47 (25.5%) 20 (20%) CBTN = Children Brain Tumor Consortium; BCH = Boston Children Hospital. Table 2. Model performance for internal test set (n=60). Models Median DSC (IQR) Aggregated DSC Median RVD (IQR) ICC Percentage of Cases with DSC 0 BraTS model 0.812 (0.559–0.888) 0.730 0.192 (0.109–0.682) 0.736 16.7% Scratch model 0.862 (0.672–0.91) 0.815 0.098 (0.04–0.407) 0.804 11.7% Transfer model 0.871 (0.724–0.914) 0.823 0.124 (0.036–0.334) 0.799 10% TransferDecoder model 0.877 (0.708–0.914) 0.832 0.126 (0.052–0.289) 0.832 8.3% TransferEncoder model 0.877 (0.715–0.914) 0.840 0.109 (0.032–0.31) 0.825 6.7% IQR = inter-quartile; DSC = Dice similarity coefficient; RVD: relative volume difference; ICC = intra-class correlation coefficient. Key Points There are limited imaging data available to train deep learning tumor segmentation for pediatric brain tumors, and adult-centric models generalize poorly in the pediatric setting. Stepwise transfer learning demonstrated gains in deep learning segmentation performance (Dice score: 0.877 [IQR 0.715–0.914]) compared to other methodologies and yielded segmentation accuracy comparable to human experts on external validation. On blinded clinical acceptability testing, the model received higher average Likert score rating and clinical acceptability compared to other experts (Transfer-Encoder model vs. average expert: 80.2% vs. 65.4%) Turing tests showed uniformly low ability of experts’ ability to correctly identify the origins of Transfer-Encoder model segmentations as AI-generated versus human-generated (mean accuracy: 26%). Competing Interests All the authors declare no competing interests. Code availability The code of the deep learning system, as well as the trained model and statistical analysis are publicly available at the GitHub webpage: https://github.com/BoydAidan/nnUNet_pLGG. ==== Refs REFERENCES 1. Ostrom QT , Price M , Ryan K , CBTRUS Statistical Report: Pediatric Brain Tumor Foundation Childhood and Adolescent Primary Brain and Other Central Nervous System Tumors Diagnosed in the United States in 2014–2018. Neuro Oncol. 2022;24 (3 ). doi: 10.1093/neuonc/noac161. 2. Ryall S , Tabori U , Hawkins C . Pediatric low-grade glioma in the era of molecular diagnostics. Acta Neuropathol Commun. 2020. doi: 10.1186/s40478-020-00902-z. 3. Manoharan N , Liu KX , Mueller S , Haas-Kogan DA , Bandopadhayay P . Pediatric low-grade glioma: Targeted therapeutics and clinical trials in the molecular era. Neoplasia (United States). 2023;36 . doi: 10.1016/j.neo.2022.100857. 4. Bandopadhayay P , Bergthold G , London WB , Long-term outcome of 4,040 children diagnosed with pediatric low-grade gliomas: An analysis of the Surveillance Epidemiology and End Results (SEER) database. Pediatr Blood Cancer. 2014;61 (7 ). doi: 10.1002/pbc.24958. 5. Deeley MA , Chen A , Datteri R , Comparison of manual and automatic segmentation methods for brain structures in the presence of space-occupying lesions: A multi-expert study. Phys Med Biol. 2011;56 (14 ). doi: 10.1088/0031-9155/56/14/021. 6. Fathi Kazerooni A , Arif S , Madhogarhia R , Automated tumor segmentation and brain tissue extraction from multiparametric MRI of pediatric brain tumors: A multi-institutional study. Neurooncol Adv. 2023;5 (1 ):vdad027. doi: 10.1093/noajnl/vdad027.37051331 7. Zhang S , Edwards A , Wang S , Patay Z , Bag AK , Scoggins MA . A Prior Knowledge Based Tumor and Tumoral Subregion Segmentation Tool for Pediatric Brain Tumors. ArXiv. 2021;abs/2109.14775. 8. Wels M , Carneiro G , Aplas A , Huber M , Hornegger J , Comaniciu D . A discriminative model-constrained graph cuts approach to fully automated pediatric brain tumor segmentation in 3-D MRI. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). 2008. doi: 10.1007/978-3-540-85988-8_9. 9. Jyothi P , Singh AR . Deep learning models and traditional automated techniques for brain tumor segmentation in MRI: a review. Artif Intell Rev. 2022; doi: 10.1007/s10462-022-10245-x. 10. Liu Z , Tong L , Chen L , Deep learning based brain tumor segmentation: a survey. Complex & Intelligent Systems. 2023;9 (1 ):1001–1026. doi: 10.1007/s40747-022-00815-5. 11. Kann BH , Likitlersuang J , Bontempi D , Screening for extranodal extension in HPV-associated oropharyngeal carcinoma: evaluation of a CT-based deep learning algorithm in patient data from a multicentre, randomised de-escalation trial. Lancet Digit Health. Elsevier; 2023;5 (6 ):e360–e369. doi: 10.1016/S2589-7500(23)00046-8.37087370 12. Kazmierski M , Welch M , Kim S , Multi-institutional prognostic modelling in head and neck cancer: evaluating impact and generalizability of deep learning and radiomics. Cancer Research Communications. 2023;CRC-22–0152 . doi: 10.1158/2767-9764.CRC-22-0152. 13. Ye Z , Qian JM , Hosny A , Deep Learning–based Detection of Intravenous Contrast Enhancement on CT Scans. Radiol Artif Intell. Radiological Society of North America; 2022;4 (3 ):e210285. doi: 10.1148/ryai.210285. 14. Hosny A , Bitterman DS , Guthier C V , Clinical validation of deep learning algorithms for radiotherapy targeting of non-small-cell lung cancer: an observational study. Lancet Digit Health. Elsevier; 2022;4 (9 ):e657–e666. doi: 10.1016/S2589-7500(22)00129-7.36028289 15. Jain A , Huang J , Ravipati Y , Head and Neck Primary Tumor and Lymph Node Auto-segmentation for PET/CT Scans. In: Andrearczyk V , Oreiller V , Hatt M , Depeursinge A , editors. Head and Neck Tumor Segmentation and Outcome Prediction. Cham: Springer Nature Switzerland; 2023. p. 61–69. 16. Nalepa J , Adamski S , Kotowski K , Segmenting pediatric optic pathway gliomas from MRI using deep learning. Comput Biol Med. 2022;142 . doi: 10.1016/j.compbiomed.2022.105237. 17. Vafaeikia P , Wagner MW , Hawkins C , Tabori U , Ertl-Wagner BB , Khalvati F . Improving the Segmentation of Pediatric Low-Grade Gliomas Through Multitask Learning. 2022 44th Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC). 2022. p. 2119–2122. doi: 10.1109/EMBC48229.2022.9871627. 18. Ghaffari M , Sowmya A , Oliver R . Automated Brain Tumor Segmentation Using Multimodal Brain Scans: A Survey Based on Models Submitted to the BraTS 2012–2018 Challenges. IEEE Rev Biomed Eng. 2020;13 . doi: 10.1109/RBME.2019.2946868. 19. Biratu ES , Schwenker F , Ayano YM , Debelee TG . A survey of brain tumor segmentation and classification algorithms. J Imaging. 2021;7 (9 ). doi: 10.3390/jimaging7090179. 20. Candemir S , Nguyen X V. , Folio LR , Prevedello LM . Training strategies for radiology deep learning models in data-limited scenarios. Radiol Artif Intell. 2021. doi: 10.1148/ryai.2021210014. 21. Zhuang F , Qi Z , Duan K , A Comprehensive Survey on Transfer Learning. Proceedings of the IEEE. 2021. doi: 10.1109/JPROC.2020.3004555. 22. Jaiswal A , Babu AR , Zadeh MZ , Banerjee D , Makedon F . A Survey on Contrastive Self-Supervised Learning. Technologies (Basel). 2021. doi: 10.3390/technologies9010002. 23. van Engelen JE , Hoos HH . A survey on semi-supervised learning. Mach Learn. 2020;109 (2 ). doi: 10.1007/s10994-019-05855-6. 24. Bakas S , Reyes M , Jakab A , Identifying the Best Machine Learning Algorithms for Brain Tumor Segmentation, Progression Assessment, and Overall Survival Prediction in the BRATS Challenge. CoRR. 2018;abs/1811.02629. http://arxiv.org/abs/1811.02629. 25. Menze BH , Jakab A , Bauer S , The Multimodal Brain Tumor Image Segmentation Benchmark (BRATS). IEEE Trans Med Imaging. 2015;34 (10 ):1993–2024. doi: 10.1109/TMI.2014.2377694.25494501 26. Baid U , Ghodasara S , Bilello M , The RSNA-ASNR-MICCAI BraTS 2021 Benchmark on Brain Tumor Segmentation and Radiogenomic Classification. CoRR. 2021;abs/2107.02314. https://arxiv.org/abs/2107.02314. 27. Isensee F , Schell M , Pflueger I , Automated brain extraction of multisequence MRI using artificial neural networks. Hum Brain Mapp. John Wiley & Sons, Ltd; 2019;40 (17 ):4952–4964. doi: 10.1002/hbm.24750.31403237 28. Isensee F , Jaeger PF , Kohl SAA , Petersen J , Maier-Hein KH . nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation. Nat Methods. 2021;18 (2 ). doi: 10.1038/s41592-020-01008-z. 29. Morid MA , Borjali A , Del Fiol G . A scoping review of transfer learning research on medical image analysis using ImageNet. Comput Biol Med. 2021. doi: 10.1016/j.compbiomed.2020.104115. 30. Mei X , Liu Z , Robson PM , RadImageNet: An Open Radiologic Deep Learning Research Dataset for Effective Transfer Learning. Radiol Artif Intell. 2022;4 (5 ). doi: 10.1148/ryai.210315. 31. Dice LR . Measures of the Amount of Ecologic Association Between Species. Ecology. 1945;26 (3 ). doi: 10.2307/1932409. 32. Kann BH , Hosny A , Aerts HJWL . Artificial intelligence for clinical oncology. Cancer Cell. 2021. doi: 10.1016/j.ccell.2021.04.002. 33. Hosny A , Bitterman DS , Guthier C v , Clinical validation of deep learning algorithms for radiotherapy targeting of non-small-cell lung cancer: an observational study. Lancet Digit Health. 2022;4 (9 ):e657–e666. doi: 10.1016/S2589-7500(22)00129-7.36028289 34. Artzi M , Gershov S , Ben-Sira L , Automatic segmentation, classification, and follow-up of optic pathway gliomas using deep learning and fuzzy c-means clustering based on MRI. Med Phys. 2020;47 (11 ). doi: 10.1002/mp.14489. 35. Peng J , Kim DD , Patel JB , Deep learning-based automatic tumor burden assessment of pediatric high-grade gliomas, medulloblastomas, and other leptomeningeal seeding tumors. Neuro Oncol. 2022;24 (2 ). doi: 10.1093/neuonc/noab151.
PMC010xxxxxx/PMC10327279.txt
==== Front medRxiv MEDRXIV medRxiv Cold Spring Harbor Laboratory 37425721 10.1101/2023.06.30.23291352 preprint 1 Article Online speech synthesis using a chronically implanted brain-computer interface in an individual with ALS Angrick Miguel 1+ Luo Shiyu 2 Rabbani Qinwan 3 Candrea Daniel N. 2 Shah Samyak 1 Milsap Griffin W. 4 Anderson William S. 5 Gordon Chad R. 56 Rosenblatt Kathryn R. 17 Clawson Lora 1 Maragakis Nicholas 1 Tenore Francesco V. 4 Fifer Matthew S. 4 Hermansky Hynek 89 Ramsey Nick F. 10 Crone Nathan E. 1+ 1 Department of Neurology, The Johns Hopkins University School of Medicine, Baltimore, MD, USA 2 Department of Biomedical Engineering, The Johns Hopkins University School of Medicine, Baltimore, MD, USA 3 Department of Electrical and Computer Engineering, The Johns Hopkins University, Baltimore, MD, USA 4 Research and Exploratory Development Department, Johns Hopkins Applied Physics Laboratory, Laurel, MD, USA 5 Department of Neurosurgery, The Johns Hopkins University School of Medicine, Baltimore, MD 6 Section of Neuroplastic and Reconstructive Surgery, Department of Plastic Surgery, The Johns Hopkins University School of Medicine, Baltimore, MD, USA 7 Department of Anesthesiology & Critical Care Medicine, The Johns Hopkins University School of Medicine, Baltimore, MD, USA 8 Center for Language and Speech Processing, The Johns Hopkins University, Baltimore, MD, USA 9 Human Language Technology Center of Excellence, The Johns Hopkins University, Baltimore, MD, USA 10 UMC Utrecht Brain Center, Department of Neurology and Neurosurgery, University Medical Center Utrecht, Utrecht, The Netherlands Author Contributions M.A. and N.C. wrote the manuscript. M.A., S.L., Q.R. and D.C. analyzed the data. M.A. and S.S. conducted the listening test. S.L. collected the data. M.A. and G.M. implemented the code for the online decoder and the underlying framework. M.A. made the visualizations. W.A., C.G. and K.R., L.C. and N.M. conducted the surgery / medical procedure. F.T. handled the regulatory aspects. H.H. supervised the speech processing methodology. M.F. N.R. and N.C. supervised the study and the conceptualization. All authors reviewed and revised the manuscript. + Corresponding authors 01 7 2023 2023.06.30.23291352https://creativecommons.org/licenses/by-nc-nd/4.0/ This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which allows reusers to copy and distribute the material in any medium or format in unadapted form only, for noncommercial purposes only, and only so long as attribution is given to the creator. nihpp-2023.06.30.23291352.pdf Recent studies have shown that speech can be reconstructed and synthesized using only brain activity recorded with intracranial electrodes, but until now this has only been done using retrospective analyses of recordings from able-bodied patients temporarily implanted with electrodes for epilepsy surgery. Here, we report online synthesis of intelligible words using a chronically implanted brain-computer interface (BCI) in a clinical trial participant (ClinicalTrials.gov, NCT03567213) with dysarthria due to amyotrophic lateral sclerosis (ALS). We demonstrate a reliable BCI that synthesizes commands freely chosen and spoken by the user from a vocabulary of 6 keywords originally designed to allow intuitive selection of items on a communication board. Our results show for the first time that a speech-impaired individual with ALS can use a chronically implanted BCI to reliably produce synthesized words that are intelligible to human listeners while preserving the participants voice profile. ==== Body pmcIntroduction A variety of neurological disorders, including amyotrophic lateral sclerosis (ALS), can severely affect speech production and other purposeful movements while sparing cognition. This can result in varying degrees of communication impairments, including Locked-In Syndrome (LIS) (Bauer, 1997; Smith, 2005), in which patients can only answer yes/no questions or select from sequentially presented options using eyeblinks, eye movements, or other residual movements. Individuals such as these may use augmentative and alternative technologies (AAT) to select among options on a communication board, but this communication can be slow, effortful, and may require caregiver intervention. Recent advances in implantable brain-computer interfaces (BCIs) have demonstrated the feasibility of establishing and maintaining communication using a variety of direct brain control strategies that bypass weak muscles, for example to control a switch scanner (Vansteensel, 2016; Chaudhary, 2022), a computer cursor (Pandarinath, 2017), to write letters (Willet, 2021) or to spell words using a hybrid approach of eye-tracking and attempted movement detection (Oxley, 2021). However, these communication modalities are still slower, more effortful, and less intuitive than speech-based BCI control (Chang, 2020). Recent studies have also explored the feasibility of decoding attempted speech from brain activity, outputting text or even acoustic speech, which could potentially carry more linguistic information such as intonation and prosody. Previous studies have reconstructed acoustic speech in offline analysis from linear regression models (Herff, 2016), convolutional (Angrick, 2019) and recurrent neural networks (Anumanchipalli, 2019; Wairagkar, 2023), and encoder-decoder architectures (Kohler, 2022). Concatenative approaches from the text-to-speech synthesis domain have also been explored (Herff, 2019; Wilson, 2020), and voice activity has been identified in electrocorticographic (ECoG) (Kanas, 2014) and stereotactic EEG recordings (Soroush, 2021). Moreover, speech decoding has been performed at the level of American English phonemes (Mugler, 2014), spoken vowels (Bouchard, 2013; Bouchard, 2014), spoken words (Kellis, 2010) and articulatory gestures (Mugler, 2015; Mugler, 2018). Until now, brain-to-speech decoding has primarily been reported in individuals with unimpaired speech, such as patients temporarily implanted with intracranial electrodes for epilepsy surgery. To date, it is unclear to what extent these findings will ultimately translate to individuals with motor speech impairments, as in ALS and other neurological disorders. Recent studies have demonstrated how neural activity acquired from an ECoG grid (Moses, 2021) or from microelectrodes (Willett, 2023) can be used to recover text from a patient with anarthria due to a brainstem stroke, or from a patient with dysarthria due to ALS, respectively. Prior to these studies, a landmark study allowed a locked-in volunteer to control a real-time synthesizer generating vowel sounds (Guenther, 2009). However, there have been no reports to date of direct closed-loop synthesis of intelligible spoken words. Here, we show that an individual with ALS participating in a clinical trial of an implantable BCI (ClinicalTrials.gov, NCT03567213) was able to produce audible, intelligible words that closely resembled his own voice, spoken at his own pace. Speech synthesis was accomplished through online decoding of ECoG signals generated during overt speech production from cortical regions previously shown to represent articulation and phonation, following similar previous work (Bouchard, 2013; Chartier, 2018; Anumanchipalli, 2019; Akbari, 2019). Our participant had considerable impairments in articulation and phonation. He was still able to produce some words that were intelligible when spoken in isolation, but his sentences were often unintelligible. Here, we focused on a closed vocabulary of 6 keywords, chosen for intuitive navigation of a communication board. Our participant was capable of producing these 6 keywords individually with a high degree of intelligibility. We acquired training data over a period of 6 weeks and deployed the speech synthesis BCI in several separate closed-loop sessions. Since the participant could still produce speech, we were able to time-align the individual’s neural and acoustic signals to enable a mapping between his cortical activity during overt speech production processes and his voice’s acoustic features. We chose to provide delayed rather than simultaneous auditory feedback in anticipation of ongoing deterioration in the patient’s speech due to ALS, with increasing discordance and interference between actual and BCI-synthesized speech. This design choice would be ideal for a neuroprosthetic device that remains capable of producing intelligible words as an individual’s speech becomes increasingly unintelligible. Here, we present the first closed-loop, self-paced BCI that translates brain activity to acoustic speech that resembles characteristics of the user’s voice profile, with most synthesized words of sufficient intelligibility to be correctly recognized by human listeners. This work makes an important step in translating recent results from speech synthesis from neural signals to their intended users with neurological speech impairments, by first focusing on a closed vocabulary that the participant can reliably generate at his own pace. Approach In order to synthesize acoustic speech from neural signals, we designed a pipeline that consisted of three recurrent neural networks (RNNs) to 1) identify and buffer speech-related neural activity, 2) transform sequences of speech-related neural activity into an intermediate acoustic representation, and 3) eventually recover the acoustic waveform using a vocoder. Figure 1 shows a schematic overview of our approach. We acquired ECoG signals from two electrode grids that covered cortical representations for speech production including ventral sensorimotor cortex and the dorsal laryngeal area (Figure 1.A). Here, we focused only on a subset of electrodes that had previously been identified as showing significant changes in high-gamma activity associated with overt speech production (see Supplementary Figure 2). From the raw ECoG signals, our closed-loop speech synthesizer extracted broadband high-gamma power features (70–170 Hz) that had previously been demonstrated to encode speech-related information useful for decoding speech (Figure 1.B) (Herff, 2019; Angrick, 2019). We used a unidirectional RNN to identify and buffer sequences of high-gamma activity frames and extract speech segments (Figure 1.C–D). This neural voice activity detection (nVAD) model internally employed a strategy to correct misclassified frames based on each frame’s temporal context, and additionally included a context window of 0.5 s to allow for smoother transitions between speech and non-speech frames. Each buffered sequence was forwarded to a bidirectional decoding model that mapped high-gamma features onto 18 Bark-scale cepstral coefficients (Moore, 2012) and 2 pitch parameters, henceforth referred to as LPC coefficients (Taylor, 2009; Valin, 2019) (Figure 1.E–F). We used a bidirectional architecture to include past and future information while making frame-wise predictions. Estimated LPC coefficients were transformed into an acoustic speech signal using the LPCNet vocoder (Valin, 2019) and played back as delayed auditory feedback (Figure 1.G). Results Synthesis Performance When deployed in sessions with the participant for online decoding, our speech-synthesis BCI was reliably capable of producing acoustic speech that captured many details and characteristics of the voice and pacing of the participant’s natural speech, often yielding a close resemblance to the words spoken in isolation from the participant. Figure 2.A provides examples of original and synthesized waveforms for a representative selection of words time-aligned by subtracting the duration of the extracted speech segment from the nVAD. Onset timings from the reconstructed waveforms indicate that the decoding model captured the flow of the spoken word while also synthesizing silence around utterances for smoother transitions. Figure 2.B shows the corresponding acoustic spectrograms for the spoken and synthesized words, respectively. The spectral structures of the original and synthesized speech shared many common characteristics and achieved average correlation scores of 0.67 (±0.18 standard deviation) suggesting that phoneme and formant-specific information were preserved. We conducted 3 sessions across 3 different days (approximately 5 and a half months after the training data was acquired, each session lasted 6 min) to repeat the experiment with acoustic feedback from the BCI to the participant (see Supplementary Video 1 for an excerpt). Other experiment parameters were not changed. All synthesized words were played back on loudspeakers while simultaneously recorded for evaluation. To assess the intelligibility of the synthesized words, we conducted listening tests in which human listeners played back individual samples of the synthesized words and selected the word that most closely resembled each sample. Additionally, we mixed in samples that contained the originally spoken words. This allowed us to assess the quality of the participant’s natural speech. We recruited a cohort of 21 native English speakers to listen to all samples that were produced during our 3 closed-loop sessions. Out of 180 samples, we excluded 2 words because the nVAD model did not detect speech activity and therefore no speech output was produced by the decoding model. We also excluded a few cases where speech activity was falsely detected by the nVAD model, which resulted in synthesized silence and remained unnoticed to the participant. Overall, human listeners achieved an accuracy score of 80%, indicating that the majority of synthesized words could be correctly and reliably recognized. Figure 2.C presents the confusion matrix regarding only the synthesized samples where the ground truth labels and human listener choices are displayed on the X- and Y-axes respectively. The confusion matrix shows that human listeners were able to recognize all but one word at very high rates. “Back” was recognized at low rates, albeit still above chance, and was most often mistaken for “Left”. This could have been due in part to the close proximity of the vowel formant frequencies for these two words. The participant’s weak tongue movements may have deemphasized the acoustic discriminability of these words, in turn resulting in the vocoder synthesizing a version of “back” that was often indistinct from “left”. In contrast, the confusion matrix also shows that human listeners were confident in distinguishing the words “Up” and “Left”. The decoder synthesized an intelligible but incorrect word in only 4% of the cases, and all listeners accurately recognized the incorrect word. Note that all keywords in the vocabulary were chosen for intuitive command and control of a computer interface, for example a communication board, and were not designed to be easily discriminable for BCI applications. Figure 2.D summarizes individual accuracy scores from all human listeners from the listening test in a histogram. All listeners recognized between 75% and 84% of the synthesized words. All human listeners achieved accuracy scores above chance (16.7%). In contrast, when tested on the participant’s natural speech, our human listeners correctly recognized almost all samples of the 6 keywords (99.8%). Anatomical and temporal contributions In order to understand which cortical areas contributed to identification of speech segments, we conducted a saliency analysis (Montavon, 2018) to reveal the underlying dynamics in high-gamma activity changes that explain the binary decisions made by our nVAD model. We utilized a method from the image processing domain (Simonyan, 2014) that queries spatial information indicating which pixels have contributed to a classification task. In our case, this method ranked individual high-gamma features over time by their influence on the predicted speech onsets (PSO). The absolute values of their gradients allowed interpretations of which contributions had the highest or lowest impact on the class scores from anatomical and temporal perspectives. The general idea is illustrated in Figure 3.B. In a forward pass, we first estimated for each trial the PSO by propagating through each time step until the nVAD model made a positive prediction. From here, we then applied backpropagation through time to compute all gradients with respect to the model’s input high-gamma features. Relevance scores |R| were computed by taking the absolute value of each partial derivative and the maximum value across time was used as the final score for each electrode (Simonyan, 2014). Note that we only performed backpropagation through time for each PSO, and not for whole speech segments. Results from the saliency analysis are shown in Figure 3.A. For each channel, we display the PSO-specific relevance scores by encoding the maximum magnitude of the influence in the size of the circles (bigger circles mean stronger influence on the predictions), and the temporal occurrence of that maximum in the respective color coding (lighter electrodes have their maximal influence on the PSO earlier). The color bar at the bottom limits the temporal influence to −400 ms prior to PSO, consistent with previous reports about speech planning (Indefrey, 2011) and articulatory representations (Bouchard, 2013). The saliency analysis showed that the nVAD model relied on a broad network of electrodes covering motor, premotor and somatosensory cortices whose collective changes in the high-gamma activity were relevant for identifying speech. Meanwhile, voice activity information encoded in the dorsal laryngeal area (highlighted electrodes in the upper grid in Figure 3.A) (Bouchard, 2013) only mildly contributed to the PSO. Figure 3.C shows relevance scores over a time period of 1 s prior to PSO for 3 selected electrodes that strongly contributed to predicting speech onsets. In conjunction with the color coding from Figure 3.A, the temporal associations were consistent with previous studies that examined phoneme decoding over fixed window sizes of 400 ms (Mugler, 2014) and 500 ms (Ramsey, 2018; Jiang, 2016) around speech onset times, suggesting that the nVAD model benefited from neural activity during speech planning and phonological processing (Indefrey, 2011) when identifying speech onset. We hypothesize that the decline in the relevance scores after −200 ms can be explained by the fact that voice activity information might have already been stored in the long short-term memory of the nVAD model and thus changes in neural activity beyond this time had less influence on the prediction. Discussion Here we demonstrate the feasibility of a closed-loop BCI that is capable of online synthesis of intelligible words using intracranial recordings from the speech cortex of an ALS participant. Recent studies (Anumanchipalli, 2019; Angrick, 2019; Kohler, 2022) suggest that deep learning techniques are a viable tool to reconstruct acoustic speech from ECoG signals. We found an approach consisting of three consecutive RNN architectures that identify and transform neural speech correlates into an acoustic waveform that can be streamed over the loudspeaker as neurofeedback, resulting in an 80% intelligibility score on a closed-vocabulary, keyword reading task. The majority of human listeners were able to correctly recognize most synthesized words – a significant advance to the field given that intelligible speech synthesis has so far been limited to offline analyses. All words from the closed vocabulary were chosen based on intuitive real-life applicability rather than being constructed to elicit discriminable neural activity that benefits decoder performance. The listening tests suggest that the words “Left” and “Back” were responsible for the majority of misclassified words. These words share very similar articulatory features, and our participant’s speech impairments likely made these words less discriminable in the synthesis process. Saliency analysis showed that our nVAD approach used information encoded in the high-gamma band across predominantly motor, premotor and somatosensory cortices, while electrodes covering the dorsal laryngeal area only marginally contributed to the identification of speech onsets. In particular, neural changes previously reported to be important for speech planning and phonological processing (Indefrey, 2011; Bouchard, 2013) appeared to have a profound impact. Here, the analysis indicates that our nVAD model learned a proper representation of spoken speech processes, providing a connection between neural patterns learned by the model and the spatio-temporal dynamics of speech production. Our participant was chronically implanted with 128 subdural ECoG electrodes, roughly half of which covered cortical areas where similar high-gamma responses have been reliably elicited during overt speech (Crone, 2001; Mugler, 2014; Ramsey, 2018; Bouchard, 2013) and have been used for offline decoding and reconstruction of speech (Angrick, 2019; Anumanchipalli, 2019). This study and others like it (Moses, 2021; Moses, 2019; Herff, 2015) explored the potential of ECoG-based BCIs to augment communication for individuals with motor speech impairments due to a variety of neurological disorders, including ALS and brainstem stroke. A potential advantage of ECoG for BCI is the stability of signal quality over long periods of time (Morrell, 2011). In a previous study of an individual with locked-in syndrome due to ALS, a fully implantable ECoG BCI with fewer electrodes provided a stable switch for a spelling application over a period of more than 3 years (Pels, 2019). Similarly, Rao et al. reported robust responses for ECoG recordings over the speech-auditory cortex for two drug-resistant epilepsy patients over a period of 1.5 years (Rao, 2017). The speech synthesis approach we demonstrated here used training data from 5 and a half months prior to testing and produced similar results over 3 separate days of testing, with recalibration but no retraining in each session. These findings suggest that the correspondence between neural activity in ventral sensorimotor cortex and speech acoustics were not significantly changed over this time period. Although longitudinal testing over longer time periods will be needed to explicitly test this, our findings provide additional support for the stability of ECoG as a BCI signal source for speech synthesis. Our approach used a speech synthesis model trained on neural data acquired during overt speech production. This constrains our current approach to patients with speech motor impairments in which vocalization is still possible and in which speech may still be intelligible. Given the increasing use of voice banking among people living with ALS, it may also be possible to improve the intelligibility of synthetic speech using an approach similar to ours, even in participants with unintelligible or absent speech. This speech could be utilized as a surrogate but would require careful alignment to speech attempts. Likewise, the same approach could be used with a generic voice, though this would not preserve the individual’s speech characteristics. Nevertheless, it remains to be seen how long our approach will continue to produce intelligible speech as our patient’s neural responses and articulatory impairments change over time due to ALS. Previous studies of long-term ECoG signal stability and BCI performance in patients with more severe motor impairments suggest that this may be possible (Vansteensel, 2016; Silversmith, 2021). Although our approach allowed for online, closed-loop production of synthetic speech that preserved our participant’s individual voice characteristics, the bidirectional LSTM imposed a delay in the audible feedback until after the patient spoke each word. We considered this delay to be not only acceptable, but potentially desirable, given our patient’s speech impairments and the likelihood of these impairments worsening in the future due to ALS. Although normal speakers use immediate acoustic feedback to tune their speech motor output (Denes, 1993), individuals with progressive motor speech impairments are likely to reach a point at which there is a significant, and distracting, mismatch between the subject’s speech and the synthetic speech produced by the BCI. In contrast, providing acoustic feedback immediately after each utterance gives the user clear and uninterrupted output that they can use to improve subsequent speech attempts, if necessary. While our results are promising, the approach used here did not allow for synthesis of unseen words. The bidirectional architecture of the decoding model learned variations of the neural dynamics of each word and was capable of recovering their acoustic representations from corresponding sequences of high-gamma frames. This approach does not capture more fine-grained and isolated part-of-speech units, such as syllables or phonemes. However, previous research (Anumanchipalli, 2019) has shown that speech synthesis approaches based on bidirectional architectures can generalize to unseen elements that were not part of the training set. Future research will be needed to expand the currently limited vocabulary for speech synthesis, and to explore to what extent similar or different approaches are able to extrapolate to words that are not in the vocabulary of the training set. Our demonstration here builds on previous seminal studies of the cortical representations for articulation and phonation (Chartier, 2018; Ramsey, 2018; Bouchard, 2013) in epilepsy patients implanted with similar subdural ECoG arrays for less than 30 days. These studies and others using intraoperative recordings have also supported the feasibility of producing synthetic speech from ECoG high-gamma responses (Anumanchipalli, 2019; Angrick, 2019; Akbari, 2019), but these demonstrations were based on offline analysis of ECoG signals that were previously recorded in subjects with normal speech. Here, a participant with impaired articulation and phonation was able to use a chronically implanted investigational device to produce acoustic speech that retained his unique voice characteristics. This was made possible through online decoding of ECoG high-gamma responses, using an algorithm trained on data collected months before. Notwithstanding the current limitations of our approach, our findings here provide a promising proof-of-concept that ECoG BCIs utilizing online speech synthesis can serve as alternative and augmentative communication devices for people living with ALS, as well as others with speech impairments due to paralysis. Moreover, our findings should motivate continued research on the feasibility of using BCIs to preserve or restore vocal communication in clinical populations where this is needed. Materials and Methods Participant Our participant was a male native English speaker in his 60s with ALS who was enrolled in a clinical trial (NCT03567213), approved by the Johns Hopkins University Institutional Review Board (IRB) and by the FDA (under an investigational device exemption) to test the safety and preliminary efficacy of a brain-computer interface composed of subdural electrodes and a percutaneous connection to external EEG amplifiers and computers. Diagnosed with ALS 8 years prior to implantation, our participant’s motor impairments had chiefly affected bulbar and upper extremity muscles and had resulted in motor impairments sufficient to render continuous speech mostly unintelligible (though individual words were intelligible), and to require assistance with most activities of daily living. The participant gave consent after being informed of the nature of the research and implant-related risks and was implanted with the study device in July 2022. Study device and implantation The study device was composed of two 8 × 8 subdural electrode grids (PMT Corporation, Chanhassen, MN) connected to a percutaneous 128-channel Neuroport pedestal (Blackrock Neurotech, Salt Lake City, UT). Both subdural grids contained platinum-iridium disc electrodes (0.76 mm thickness, 2-mm diameter exposed surface) with 4 mm center-to-center spacing and a total surface area of 12.11 cm2 (36.6 mm × 33.1 mm). The study device was surgically implanted during a craniotomy with the ECoG grids placed on the pial surface of sensorimotor representations for speech and upper extremity movements in the left hemisphere. Careful attention was made to assure that the scalp flap incision was well away from the external pedestal. Cortical representations were targeted using anatomical landmarks from pre-operative structural (MRI) and functional imaging (fMRI), in addition to somatosensory evoked potentials measured intraoperatively. Two reference wires attached to the Neuroport pedestal were implanted in the subdural space on the outward facing surface of the subdural grids. The participant was awoken during the craniotomy to confirm proper functioning of the study device and final placement of the two subdural grids. For this purpose, the participant was asked to repeatedly speak a single word as event-related ECoG spectral responses were noted. On the same day, the participant had a post-operative CT which was then co-registered to a pre-operative MRI to verify the anatomical locations of the two grids. Data Recording During all training and testing sessions, the Neuroport pedestal was connected to a 128-channel NeuroPlex-E headstage that was in turn connected by a mini-HDMI cable to a NeuroPort Biopotential Signal Processor (Blackrock Neurotech, Salt Lake City, UT, USA) and external computers. Acoustic speech was recorded through an external microphone (BETA® 58A, SHURE, Niles, IL) in a room isolated from external acoustic and electronic noise, then amplified and digitized by an external audio interface (H6-audio-recorder, Zoom Corporation, Tokyo, Japan). The acoustic speech signal was split and forwarded to: 1) an analog input of the NeuroPort Biopotential Signal Processor (NSP) to be recorded at the same frequency and in synchrony with the neural signals, and 2) the testing computer to capture high-quality (48 kHz) recordings. We applied cross-correlation to align the high-quality recordings with the synchronized audio signal from the NSP. Experiment Recordings & Task design Each recording day began with a syllable repetition task to acquire cortical activity to be used for baseline normalization. Each syllable was audibly presented through a loudspeaker, and the participant was instructed to recite the heard stimulus by repeating it aloud. Stimulus presentation lasted for 1 s, and trial duration was set randomly in the range of 2.5 s and 3.5 s with a step size of 80 ms. In the syllable repetition task, the participant was instructed to repeat 12 consonant-vowel syllables (Supplementary Table 1), in which each syllable was repeated 5 times. We extracted high-gamma frames from all trials to compute for each day the mean and standard deviation statistics for channel-specific normalization. To collect data for training our nVAD and speech decoding model, we recorded ECoG during multiple blocks of a speech production task over a period of 6 weeks. During the task, the participant read aloud single words that were prompted on a computer screen, interrupted occasionally by a silence trial in which the participant was instructed to say nothing. The words came from a closed vocabulary of 6 words (“Left”, “Right”, “Up”, “Down”, “Enter”, “Back”, and “...” for silence). In each block, there were five repetitions of each word (60 words in total) that appeared in a pseudo-randomized order by having a fixed set of seeds to control randomization orders. Each word was shown for 2 s per trial with an intertrial interval of 3 s. The participant was instructed to read the prompted word aloud as soon as it appeared. Because his speech was slow, effortful, and dysarthric, the participant may have sometimes used some of the intertrial interval to complete word production. However, offline analysis verified at least 1 s between the end of each spoken word and the beginning of the next trial, assuring that enough time had passed to avoid ECoG high-gamma responses leaking into subsequent trials. In each block, neural signals and audibly vocalized speech were acquired in parallel and stored to disc using BCI2000 (Schalk, 2004). We recorded training, validation, and test data for 10 days, and deployed our approach for synthesizing speech online 5 and a half months later. During the online task, the synthesized output was played to the participant while he performed the same keyword reading task as in the training sessions. The feedback from each synthesized word began after he spoke the same word, avoiding any interference with production from the acoustic feedback. The validation dataset was used for finding appropriate hyperparameters to train both nVAD and the decoding model. The test set was used to validate final model generalizability before online sessions. We also used the test set for the saliency analysis. In total, the training set was comprised of 1,570 trials that aggregated to approximately 80 min of data (21.8 min are pure speech), while the validation and test set contained 70 trials each with around 3 min of data (0.9 min pure speech), respectively. The data in each of these datasets were collected on different days, so that no baseline or other statistics in the training set leaked into the validation or test set. Signal Processing & Feature Extraction Neural signals were transformed into broadband high-gamma power features that have been previously reported to closely track the timing and location of cortical activation during speech and language processes (Crone, 2001; Leuthardt, 2012). In this feature extraction process, we first re-referenced all channels within each 64-contact grid to a common-average reference (CAR filtering), excluding channels with poor signal quality in any training session. Next, we selected all channels that had previously shown significant high-gamma responses during the syllable repetition task described above. This included 64 channels (Supplementary Figure 2, channels with blue outlines) across motor, premotor and somatosensory cortices, including the dorsal laryngeal area. From here, we applied two IIR Butterworth filters (both with filter order 8) to extract the high-gamma band in the range of 70 to 170 Hz while subsequently attenuating the first harmonic (118–122 Hz) of the line noise. For each channel, we computed logarithmic power features based on windows with a fixed length of 50 ms and a frameshift of 10 ms. To estimate speech-related increases in broadband high-gamma power, we normalized each feature by the day-specific statistics of the high-gamma power features accumulated from the syllable repetition task. For the acoustic recordings of the participant’s speech, we downsampled the time-aligned high-quality microphone recordings from 48 kHz to 16 kHz. From here, we padded the acoustic data by 16 ms to account for the shift introduced by the two filters on the neural data and estimated the boundaries of speech segments using an energy-based voice activity detection algorithm (Povey, 2011). Likewise, we computed acoustic features in the LPC coefficient space through the encoding functionality of the LPCNet vocoder. Both voice activity detection and LPC feature encoding were configured to operate on 10 ms frameshifts to match the number of samples from the broadband high-gamma feature extraction pipeline. Network Architectures Our proposed approach relied on three recurrent neural network architectures: 1) a unidirectional model that identified speech segments from the neural data, 2) a bidirectional model that translated sequences of speech-related high-gamma activity into corresponding sequences of LPC coefficients representing acoustic information, and 3) LPCNet (Valin, 2019), which converted those LPC coefficients into an acoustic speech signal. The network architecture of the unidirectional nVAD model was inspired by Zen et al. (Zen, 2015) in using a stack of two LSTM layers with 150 units each, followed by a linear fully connected output layer with two units representing speech or non-speech class target logits (Figure 4). We trained the unidirectional nVAD model using truncated backpropagation through time (BPTT) (Sutskever, 2013) to keep the costs of single parameter updates manageable. We initialized this algorithm’s hyperparameters k1 and k2 to 50 and 100 frames of high-gamma activity, respectively, such that the unfolding procedure of the backpropagation step was limited to 100 frames (1 s) and repeated every 50 frames (500 ms). Dropout was used as a regularization method with a probability of 50% to counter overfitting effects (Srivastava, 2014). Comparison between predicted and target labels was determined by the cross-entropy loss. We limited the network training using an early stopping mechanism that evaluated after each epoch the network performance on a held-out validation set and kept track of the best model weights by storing the model weights only when the frame-wise accuracy score was bigger than before. The learning rate of the stochastic gradient descent optimizer was dynamically adjusted in accordance with the RMSprop formula (Ruder, 2016) with an initial learning rate of 0.001. Using this procedure, the unidirectional nVAD model was trained for 27,975 update steps, achieving a frame-wise accuracy of 93.4% on held-out validation data. The architecture of the nVAD model had 311,102 trainable weights. The network architecture of the bidirectional decoding model had a very similar configuration to the unidirectional nVAD but employed a stack of bidirectional LSTM layers for sequence modelling (Anumanchipalli, 2019) to include past and future contexts. Since the acoustic space of the LPC components was continuous, we used a linear fully connected output layer for this regression task. Figure 4 contains an illustration of the network architecture of the decoding model. In contrast to the unidirectional nVAD model, we use standard BPTT to account for both past and future contexts within each extracted segment identified as spoken speech. The architecture of the decoding model had 378,420 trainable weights and was trained for 14,130 update steps using a stochastic gradient descent optimizer. The initial learning rate was set to 0.001 and dynamically updated in accordance with the RMSProp formula. Again, we used dropout with a 50% probability and employed an early stopping mechanism that only updates model weights when the loss on the held-out validation set is lower than before. Both the unidirectional nVAD and the bidirectional decoding model were implemented within the PyTorch framework. For LPCNet, we used the C-implementation and pretrained model weights by the original authors and communicated with the library via wrapper functions through the Cython programming language. Closed-Loop Architecture Our closed-loop architecture was built upon the ezmsg, a general-purpose framework which enables the implementation of streaming systems in the form a directed acyclic network of connected units, which communicate with each other through a publish/subscribe software engineering pattern using asynchronous coroutines. Here, each unit represents a self-contained operation which receives many inputs, and optionally propagates its output to all its subscribers. A unit consists of a settings and state class for enabling initial and updatable configurations and has multiple input and output connection streams to communicate with other nodes in the network. Figure 4 shows a schematic overview of the closed-loop architecture. ECoG signals were received by connecting to BCI2000 via a custom ZeroMQ (ZMQ) networking interface that sent packages of 40 ms over the TCP/IP protocol. From here, each unit interacted with other units through an asynchronous message system that was implemented on top of a shared-memory publish-subscribe multi-processing pattern. Figure 4 shows that the closed-loop architecture was comprised of 5 units for the synthesis pipeline, while employing several additional units that acted as loggers and wrote intermediate data to disc. In order to play back the synthesized speech during closed-loop sessions, we wrote the bytes of the raw PCM waveform to standard output (stdout) and reinterpreted them by piping them into SoX. We implemented our closed-loop architecture in Python 3.10. To keep the computational complexity manageable for this streamlined application, we implemented several functionalities, such as ringbuffers or specific calculations in the high-gamma feature extraction, in Cython. Contamination Analysis Overt speech production can cause acoustic artifacts in electrophysiological recordings, allowing learning machines such as neural networks to rely on information that is likely to fail once deployed – a phenomenon widely known as Clever Hans (Lapuschkin, 2019). We used the method proposed by Roussel et al. (Roussel, 2020) to assess the risk that our ECoG recordings had been contaminated. This method compares correlations between neural and acoustic spectrograms to determine a contamination index which describes the average correlation of matching frequencies. This contamination index is compared to the distribution of contamination indices resulting from randomly permuting the rows and columns of the contamination matrix – allowing statistical analysis of the risk when assuming that no acoustic contamination is present. For each recording day among the train, test and validation set, we analyzed acoustic contamination in the high-gamma frequency range. We identified 1 channel (Channel 46) in our recordings that was likely contaminated during 3 recording days (D5, D6, and D7), and we corrected this channel by taking the average of high-gamma power features from neighboring channels (8-neighbour configuration, excluding the bad channel 38). A detailed report can be found in Supplementary Figure 1, where each histogram corresponds to the distribution of permuted contamination matrices, and colored vertical bars indicate the actual contamination index, where green and red indicate the statistical criterion threshold (green: p > 0.05, red: p ≤ 0.05). After excluding the neural data from channel 46, Roussel’s method suggested that the null hypothesis could be rejected, and thus we concluded that no acoustic speech has interfered with neural recording. Listening Test We conducted a forced-choice listening test similar to Herff et al. (Herff, 2019) in which 21 native English speakers evaluated the intelligibility of the synthesized output and the originally spoken words. Listeners were asked to listen to one word at a time and select which word out of the six options most closely resembled it. Here, the listeners had the opportunity to listen to each sample many times before submitting a choice. We implemented the listening test on top of the BeaqleJS framework (Kraft, 2014). All words that were either spoken or synthesized during the 3 closed-loop sessions were included in the listening test, but were randomly sampled from a uniform distribution for unique randomized sequences across listeners. Supplementary Figure 3 provides a screenshot of the interface with which the listeners were working. All human listeners were only recruited through indirect means such as IRB-approved flyers placed on campus sites and had no direct connection to the PI. Anonymous demographic data was collected at the end of the listening test asking for age and preferred gender. Overall, recruited participants were 23.8% male and 61.9% female (14% other or preferred not to answer) ranging between 18 to 30 years old. Statistical Analysis Original and reconstructed speech spectrograms were compared using Pearson’s correlation coefficients for 80 mel-scaled spectral bins. For this, we transformed original and reconstructed waveforms into the spectral domain using the short-time Fourier transform (window size: 50 ms, frameshift: 10 ms, window function: Hanning), applied 80 triangular filters to focus only on perceptual differences for human listeners (Stevens, 1937), and Gaussianized the distribution of the acoustic space using the natural logarithm. Pearson correlation scores were calculated for each sample by averaging the correlation coefficients across frequency bins. The 95% confidence interval (two-sided) was used in the feature selection procedure while the z-criterion was Bonferroni corrected across time points. Lower and upper bounds for all channels and time points can be found in the supplementary data. Contamination analysis is based on permutation tests that use t-tests as their statistical criterion with a Bonferroni corrected significance level of α = 0.05 / N, where N represents the number of frequency bins multiplied by the number of selected channels. Overall, we used the SciPy stats package (version 1.10.1) for statistical evaluation, but the contamination analysis has been done in Matlab with the statistics and machine learning toolbox (version 12.4). Supplementary Video This study is accompanied by a video of the participant during one block of a closed-loop session, which demonstrates identification of speech segments from the participant through the unidirectional voice activity detection RNN and closed-loop reconstruction of the spoken speech signal. Note that we masked out the delayed feedback from the patient’s speech, which was recorded from the microphone while played back on the loudspeaker. Instead, we incorporated the same acoustic signal that was played back on the loudspeaker as a separate channel described as BCI (dark red). Supplementary Material Supplement 1 Acknowledgements This clinical trial is funded by a BRAIN Initiative UH3 NS114439 from NINDS (PI N.C., co-PI N.R). Data and Code Availability Neural data and anonymized speech audio will be made publicly available on www.osf.io upon acceptance of the manuscript. Corresponding source code for the closed-loop BCI and scripts for generating figures can be obtained from the official Crone Lab Github page at: https://github.com/cronelab/... and will be made publicly available upon acceptance of the manuscript. The ezmsg framework can be obtained from https://github.com/iscoe/ezmsg. Figure 1 | Overview of the closed-loop speech synthesizer. (A) Neural activity is acquired from a subset of 64 electrodes (highlighted in orange) from two 8 × 8 ECoG electrode arrays covering sensorimotor areas for face and tongue, and for upper limb regions. (B) The closed-loop speech synthesizer extracts high-gamma features to reveal speech-related neural correlates of attempted speech production and propagates each frame to a neural voice activity detection (nVAD) model (C) that identifies and extracts speech segments (D). When the participant finishes speaking a word, the nVAD model forwards the high-gamma activity of the whole extracted sequence to a bidirectional decoding model (E) which estimates acoustic features (F) that can be transformed into an acoustic speech signal. (G) The synthesized speech is played back as acoustic feedback. Figure 2 | Evaluation of the synthesized words. (A) Visual example of time-aligned original and reconstructed acoustic speech waveforms and their spectral representations (B) for 6 words that were recorded during one of the closed-loop sessions. Speech spectrograms are shown between 100 and 8000 Hz with a logarithmic frequency range to emphasize formant frequencies. (C) The confusion matrix between human listeners and ground truth. (D) Distribution of accuracy scores from all who performed the listening test for the synthesized speech samples. Dashed line shows chance performance (16.7%). Figure 3 | Changes in high-gamma activity across motor, premotor and somatosensory cortices trigger detection of speech output. (A) Saliency analysis shows that changes in high-gamma activity predominantly from 300 to 100 ms prior to predicted speech onset (PSO) strongly influenced the nVAD model’s decision. Electrodes covering motor, premotor and somatosensory cortices show the impact of model decisions, while electrodes covering the dorsal laryngeal area only modestly added information to the prediction. Grey electrodes were either not used, bad channels or had no notable contributions. (B) Illustration of the general procedure on how relevance scores were computed. For each time step t, relevance scores were computed by backpropagation through time across all previous high-gamma frames Xt. Predictions of 0 correspond to no-speech, while 1 represents speech frames. (C) Temporal progression of mean magnitudes of the absolute relevance score in 3 selected channels that strongly contributed to PSOs. Shaded areas reflect the standard error of the mean (N=60). Units of the relevance scores are in 10−3. Figure 4 | System overview of the closed-loop architecture. The computational graph is designed as a directed acyclic network. Solid shapes represent ezmsg units, dotted ones represent initialization parameters. Each unit is responsible for a self-contained task and distributes their output to all its subscribers. Logger units run in separate processes to not interrupt the main processing chain for synthesizing speech. Competing Interests The authors declare that they have no competing interests. ==== Refs References 1. Bauer Gerhard , Gerstenbrand Franz , and Rumpl Erik . “Varieties of the locked-in syndrome.” Journal of neurology 221 (1979): 77–91.92545 2. Smith E. , & Delargy M. (2005). Locked-in syndrome. Bmj, 330 (7488 ), 406–409.15718541 3. Vansteensel M. J. , Pels E. G. , Bleichner M. G. , Branco M. P. , Denison T. , Freudenburg Z. V. , ... & Ramsey N. F. (2016). Fully implanted brain–computer interface in a locked-in patient with ALS. New England Journal of Medicine, 375 (21 ), 2060–2066.27959736 4. Chaudhary U. , Vlachos I. , Zimmermann J. B. , Espinosa A. , Tonin A. , Jaramillo-Gonzalez A. , ... & Birbaumer N. (2022). Spelling interface using intracortical signals in a completely locked-in patient enabled via auditory neurofeedback training. Nature communications, 13 (1 ), 1236. 5. Pandarinath C. , Nuyujukian P. , Blabe C. H. , Sorice B. L. , Saab J. , Willett F. R. , ... & Henderson J. M. (2017). High performance communication by people with paralysis using an intracortical brain-computer interface. Elife, 6 , e18554.28220753 6. Willett F. R. , Avansino D. T. , Hochberg L. R. , Henderson J. M. , & Shenoy K. V. (2021). High-performance brain-to-text communication via handwriting. Nature, 593 (7858 ), 249–254.33981047 7. Oxley T. J. , Yoo P. E. , Rind G. S. , Ronayne S. M. , Lee C. S. , Bird C. , ... & Opie N. L. (2021). Motor neuroprosthesis implanted with neurointerventional surgery improves capacity for activities of daily living tasks in severe paralysis: first in-human experience. Journal of neurointerventional surgery, 13 (2 ), 102–108.33115813 8. Chang E. F. , & Anumanchipalli G. K. (2020). Toward a speech neuroprosthesis. Jama, 323 (5 ), 413–414.31880768 9. Herff C. , Johnson G. , Diener L. , Shih J. , Krusienski D. , & Schultz T. (2016). Towards direct speech synthesis from ECoG: A pilot study. In 2016 38th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) (pp. 1540–1543). IEEE. 10. Angrick M. , Herff C. , Mugler E. , Tate M. C. , Slutzky M. W. , Krusienski D. J. , & Schultz T. (2019). Speech synthesis from ECoG using densely connected 3D convolutional neural networks. Journal of neural engineering, 16 (3 ), 036019.30831567 11. Anumanchipalli G. K. , Chartier J. , & Chang E. F. (2019). Speech synthesis from neural decoding of spoken sentences. Nature, 568 (7753 ), 493–498.31019317 12. Wairagkar M. , Hochberg L. R. , Brandman D. M. , & Stavisky S. D. (2023). Synthesizing Speech by Decoding Intracortical Neural Activity from Dorsal Motor Cortex. In 2023 11th International IEEE/EMBS Conference on Neural Engineering (NER) (pp. 1–4). IEEE. 13. Kohler J. , Ottenhoff M. C. , Goulis S. , Angrick M. , Colon A. J. , Wagner L. , ... & Herff C. (2022). Synthesizing speech from intracranial depth electrodes using an encoder-decoder framework. Neurons, Behavior, Data Analysis and Theory (NBDT). 14. Herff C. , Diener L. , Angrick M. , Mugler E. , Tate M. C. , Goldrick M. A. , ... & Schultz T. (2019). Generating natural, intelligible speech from brain activity in motor, premotor, and inferior frontal cortices. Frontiers in neuroscience, 13 , 1267.31824257 15. Wilson G. H. , Stavisky S. D. , Willett F. R. , Avansino D. T. , Kelemen J. N. , Hochberg L. R. , ... & Shenoy K. V. (2020). Decoding spoken English from intracortical electrode arrays in dorsal precentral gyrus. Journal of neural engineering, 17 (6 ), 066007.33236720 16. Kanas V. G. , Mporas I. , Benz H. L. , Sgarbas K. N. , Bezerianos A. , & Crone N. E. (2014). Joint spatial-spectral feature space clustering for speech activity detection from ECoG signals. IEEE Transactions on Biomedical Engineering, 61 (4 ), 1241–1250.24658248 17. Soroush P. Z. , Angrick M. , Shih J. , Schultz T. , & Krusienski D. J. (2021). Speech activity detection from stereotactic eeg. In 2021 IEEE International Conference on Systems, Man, and Cybernetics (SMC) (pp. 3402–3407). IEEE. 18. Mugler E. M. , Patton J. L. , Flint R. D. , Wright Z. A. , Schuele S. U. , Rosenow J. , ... & Slutzky M. W. (2014). Direct classification of all American English phonemes using signals from functional speech motor cortex. Journal of neural engineering, 11 (3 ), 035015.24836588 19. Bouchard K. E. , Mesgarani N. , Johnson K. , & Chang E. F. (2013). Functional organization of human sensorimotor cortex for speech articulation. Nature, 495 (7441 ), 327–332.23426266 20. Bouchard K. E. , & Chang E. F. (2014). Neural decoding of spoken vowels from human sensory-motor cortex with high-density electrocorticography. In 2014 36th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (pp. 6782–6785). IEEE. 21. Kellis S. , Miller K. , Thomson K. , Brown R. , House P. , & Greger B. (2010). Decoding spoken words using local field potentials recorded from the cortical surface. Journal of neural engineering, 7 (5 ), 056007.20811093 22. Mugler E. M. , Goldrick M. , Rosenow J. M. , Tate M. C. , & Slutzky M. W. (2015). Decoding of articulatory gestures during word production using speech motor and premotor cortical activity. In 2015 37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) (pp. 5339–5342). IEEE. 23. Mugler E. M. , Tate M. C. , Livescu K. , Templer J. W. , Goldrick M. A. , & Slutzky M. W. (2018). Differential representation of articulatory gestures and phonemes in precentral and inferior frontal gyri. Journal of Neuroscience, 38 (46 ), 9803–9813.30257858 24. Moses D. A. , Metzger S. L. , Liu J. R. , Anumanchipalli G. K. , Makin J. G. , Sun P. F. , ... & Chang E. F. (2021). Neuroprosthesis for decoding speech in a paralyzed person with anarthria. New England Journal of Medicine, 385 (3 ), 217–227.34260835 25. Willett F. R. , Kunz E. , Fan C. , Avansino D. , Wilson G. , Choi E. Y. , ... & Henderson J. (2023). A high-performance speech neuroprosthesis. bioRxiv, 2023–01. 26. Guenther F. H. , Brumberg J. S. , Wright E. J. , Nieto-Castanon A. , Tourville J. A. , Panko M. , ... & Kennedy P. R. (2009). A wireless brain-machine interface for real-time speech synthesis. PloS one, 4 (12 ), e8218.20011034 27. Chartier J. , Anumanchipalli G. K. , Johnson K. , & Chang E. F. (2018). Encoding of articulatory kinematic trajectories in human speech sensorimotor cortex. Neuron, 98 (5 ), 1042–1054.29779940 28. Akbari H. , Khalighinejad B. , Herrero J. L. , Mehta A. D. , & Mesgarani N. (2019). Towards reconstructing intelligible speech from the human auditory cortex. Scientific reports, 9 (1 ), 874.30696881 29. Moore B. C. (2012). An introduction to the psychology of hearing. Brill. 30. Taylor P. (2009). Text-to-speech synthesis. Cambridge University Press. 31. Valin J. M. , & Skoglund J. (2019). LPCNet: Improving neural speech synthesis through linear prediction. In ICASSP 2019–2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (pp. 5891–5895). IEEE. 32. Montavon G. , Samek W. , & Müller K. R. (2018). Methods for interpreting and understanding deep neural networks. Digital signal processing, 73 , 1–15. 33. Simonyan K. , Vedaldi A. , & Zisserman A. (2014). Deep inside convolutional networks: Visualising image classification models and saliency maps. International Conference on Learning Representations (ICLR). 34. Indefrey P. (2011). The spatial and temporal signatures of word production components: a critical update. Frontiers in psychology, 2 , 255.22016740 35. Ramsey N. F. , Salari E. , Aarnoutse E. J. , Vansteensel M. J. , Bleichner M. G. , & Freudenburg Z. V. (2018). Decoding spoken phonemes from sensorimotor cortex with high-density ECoG grids. Neuroimage, 180 , 301–311.28993231 36. Jiang W. , Pailla T. , Dichter B. , Chang E. F. , & Gilja V. (2016). Decoding speech using the timing of neural signal modulation. In 2016 38th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) (pp. 1532–1535). IEEE. 37. Crone N. , Hao L. , Hart J. , Boatman D. , Lesser R. P. , Irizarry R. , & Gordon B. (2001). Electrocorticographic gamma activity during word production in spoken and sign language. Neurology, 57 (11 ), 2045–2053.11739824 38. Moses D. A. , Leonard M. K. , Makin J. G. , & Chang E. F. (2019). Real-time decoding of question- and-answer speech dialogue using human cortical activity. Nature communications, 10 (1 ), 3096. 39. Herff C. , Heger D. , De Pesters A. , Telaar D. , Brunner P. , Schalk G. , & Schultz T. (2015). Brain-to-text: decoding spoken phrases from phone representations in the brain. Frontiers in neuroscience, 9 , 217.26124702 40. Morrell M. J. (2011). Responsive cortical stimulation for the treatment of medically intractable partial epilepsy. Neurology, 77 (13 ), 1295–1304.21917777 41. Pels E. G. , Aarnoutse E. J. , Leinders S. , Freudenburg Z. V. , Branco M. P. , van der Vijgh B. H. , ... & Ramsey N. F. (2019). Stability of a chronic implanted brain-computer interface in late-stage amyotrophic lateral sclerosis. Clinical Neurophysiology, 130 (10 ), 1798–1803.31401488 42. Rao V. R. , Leonard M. K. , Kleen J. K. , Lucas B. A. , Mirro E. A. , & Chang E. F. (2017). Chronic ambulatory electrocorticography from human speech cortex. NeuroImage, 153 , 273–282.28396294 43. Silversmith D. B. , Abiri R. , Hardy N. F. , Natraj N. , Tu-Chan A. , Chang E. F. , & Ganguly K. (2021). Plug-and-play control of a brain–computer interface through neural map stabilization. Nature biotechnology, 39 (3 ), 326–335. 44. Denes P. B. , Denes P. , & Pinson E. (1993). The speech chain. Macmillan. 45. Schalk G. , McFarland D. J. , Hinterberger T. , Birbaumer N. , & Wolpaw J. R. (2004). BCI2000: a general-purpose brain-computer interface (BCI) system. IEEE Transactions on biomedical engineering, 51 (6 ), 1034–1043.15188875 46. Leuthardt E. C. , Pei X. M. , Breshears J. , Gaona C. , Sharma M. , Freudenberg Z. , ... & Schalk G. (2012). Temporal evolution of gamma activity in human cortex during an overt and covert word repetition task. Frontiers in human neuroscience, 6 , 99.22563311 47. Povey D. , Ghoshal A. , Boulianne G. , Burget L. , Glembek O. , Goel N. , ... & Vesely K. (2011). The Kaldi speech recognition toolkit. In IEEE 2011 workshop on automatic speech recognition and understanding (No. CONF). IEEE Signal Processing Society. 48. Zen H. , & Sak H. (2015, April). Unidirectional long short-term memory recurrent neural network with recurrent output layer for low-latency speech synthesis. In 2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (pp. 4470–4474). IEEE. 49. Sutskever I. (2013). Training recurrent neural networks (pp. 1–101). Toronto, ON, Canada: University of Toronto. 50. Srivastava N. , Hinton G. , Krizhevsky A. , Sutskever I. , & Salakhutdinov R. (2014). Dropout: a simple way to prevent neural networks from overfitting. The journal of machine learning research, 15 (1 ), 1929–1958. 51. Ruder S. (2016). An overview of gradient descent optimization algorithms. arXiv preprint arXiv:1609.04747. 52. Lapuschkin S. , Wäldchen S. , Binder A. , Montavon G. , Samek W. , & Müller K. R. (2019). Unmasking Clever Hans predictors and assessing what machines really learn. Nature communications, 10 (1 ), 1096. 53. Roussel P. , Le Godais G. , Bocquelet F. , Palma M. , Hongjie J. , Zhang S. , ... & Yvert B. (2020). Observation and assessment of acoustic contamination of electrophysiological brain signals during speech production and sound perception. Journal of Neural Engineering, 17 (5 ), 056028.33055383 54. Kraft S. , & Zölzer U. (2014, May). BeaqleJS: HTML5 and JavaScript based framework for the subjective evaluation of audio quality. In Linux Audio Conference, Karlsruhe, DE. 55. Stevens S. S. , Volkmann J. , & Newman E. B. (1937). A scale for the measurement of the psychological magnitude pitch. The journal of the acoustical society of america, 8 (3 ), 185–190.
PMC010xxxxxx/PMC10327285.txt
==== Front medRxiv MEDRXIV medRxiv Cold Spring Harbor Laboratory 37425750 10.1101/2023.06.28.23291995 preprint 1 Article Longitudinal multi-omics study reveals common etiology underlying association between plasma proteome and BMI trajectories in adolescent and young adult twins Drouard Gabin 1* Hagenbeek Fiona A. 123 Whipp Alyce 1 Pool René 23 Hottenga Jouke Jan 23 Jansen Rick 45 Hubers Nikki 26 Afonin Aleksei 7 BIOS Consortium8 BBMRI-NL Metabolomics Consortium9 Willemsen Gonneke 23 de Geus Eco J. C. 23 Ripatti Samuli 11011 Pirinen Matti 11012 Kanninen Katja M. 7 Boomsma Dorret I. 236 van Dongen Jenny 236 Kaprio Jaakko 1* 1 Institute for Molecular Medicine Finland (FIMM), HiLIFE, University of Helsinki, Helsinki, Finland 2 Department of Biological Psychology, Vrije Universiteit Amsterdam, Amsterdam, The Netherlands 3 Amsterdam Public Health Research Institute, Amsterdam, The Netherlands 4 Amsterdam UMC location Vrije Universiteit Amsterdam, Department of Psychiatry, Amsterdam, the Netherlands 5 Amsterdam Neuroscience, Mood, Anxiety, Psychosis, Sleep & Stress Program, Amsterdam, The Netherlands 6 Amsterdam Reproduction & Development (AR&D) Research Institute, Amsterdam, The Netherlands 7 A.I. Virtanen Institute for Molecular Sciences, University of Eastern Finland, Kuopio, Finland 8 Biobank-based Integrative Omics Study Consortium. Lists of authors and their affiliations appear in the supplementary material (see Additional file 1) 9 Lists of authors and their affiliations appear in the supplementary material (see Additional file 1) 10 Department of Public Health, Faculty of Medicine, University of Helsinki, Helsinki, Finland 11 Broad Institute of MIT and Harvard, Cambridge, MA, USA 12 Department of Mathematics and Statistics, University of Helsinki, Helsinki, Finland Authors’ contributions The study design was developed and discussed by GD, FAH, AW, NH, SR, MP, DIB, JD, and JK. The data were collected by GW, EJCG, KK, DIB, JD, and JK, and their availability for analysis was facilitated by RP and AF. Processing of FinnTwin12 data was performed by GD and AW. NTR data were processed by GD, FAH, RP, JH, and RJ. The bioinformatic and statistical analyses were performed by GD. GD and FAH wrote the original manuscript. All authors actively participated in the improvement of the manuscript by critically revising it. All of the authors read and approved the final version of the manuscript. * corresponding authors: [email protected] / [email protected] 01 7 2023 2023.06.28.23291995https://creativecommons.org/licenses/by-nc-nd/4.0/ This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which allows reusers to copy and distribute the material in any medium or format in unadapted form only, for noncommercial purposes only, and only so long as attribution is given to the creator. nihpp-2023.06.28.23291995.pdf Background: The influence of genetics and environment on the association of the plasma proteome with body mass index (BMI) and changes in BMI remain underexplored, and the links to other omics in these associations remain to be investigated. We characterized protein-BMI trajectory associations in adolescents and adults and how these connect to other omics layers. Methods: Our study included two cohorts of longitudinally followed twins: FinnTwin12 (N=651) and the Netherlands Twin Register (NTR) (N=665). Follow-up comprised four BMI measurements over approximately 6 (NTR: 23–27 years old) to 10 years (FinnTwin12: 12–22 years old), with omics data collected at the last BMI measurement. BMI changes were calculated using latent growth curve models. Mixed-effects models were used to quantify the associations between the abundance of 439 plasma proteins with BMI at blood sampling and changes in BMI. The sources of genetic and environmental variation underlying the protein abundances were quantified using twin models, as were the associations of proteins with BMI and BMI changes. In NTR, we investigated the association of gene expression of genes encoding proteins identified in FinnTwin12 with BMI and changes in BMI. We linked identified proteins and their coding genes to plasma metabolites and polygenic risk scores (PRS) using mixed-effect models and correlation networks. Results: We identified 66 and 14 proteins associated with BMI at blood sampling and changes in BMI, respectively. The average heritability of these proteins was 35%. Of the 66 BMI-protein associations, 43 and 12 showed genetic and environmental correlations, respectively, including 8 proteins showing both. Similarly, we observed 6 and 4 genetic and environmental correlations between changes in BMI and protein abundance, respectively. S100A8 gene expression was associated with BMI at blood sampling, and the PRG4 and CFI genes were associated with BMI changes. Proteins showed strong connections with many metabolites and PRSs, but we observed no multi-omics connections among gene expression and other omics layers. Conclusions: Associations between the proteome and BMI trajectories are characterized by shared genetic, environmental, and metabolic etiologies. We observed few gene-protein pairs associated with BMI or changes in BMI at the proteome and transcriptome levels. multi-omics proteome metabolome transcriptome polygenic risk scores weight gain body mass index longitudinal twin study Biobank-based Integrative Omics Study (BIOS) Consortium, BBMRI Metabolomics Consortium, BBMRI-NL, Research Infrastructure, NWO184.021.007 184033111 Wellcome Trust Sanger Institute, Broad Institute, ENGAGE – European Network for Genetic and Genomic EpidemiologyFP7-HEALTH-F4-2007 201413 National Institute of Alcohol Abuse and AlcoholismAA-12502 AA-00145 AA-09203 AA15416 K02AA018755 R01AA015416 Academy of Finland100499 205585 118555 141054 264146 308248 Centre of Excellence in Complex Disease Genetics312073 336823 352792 European Union’s Horizon 2020 research and innovation program874724 Netherlands Organization for Scientific Research (NWO), Netherlands Organisation for Health Research and Development (ZonMW)904-61-090 985-10-002 912-10-020 904-61-193 480-04-004 463-06-001 451-04-034 400-05-717 Addiction-31160008 016-115-035 481-08-011 400-07-080 056-32-010 Middelgroot-911-09-032 OCW_NWO Gravity program024.001.003 NWOGroot 480-15-001/674 Center for Medical Systems Biology (CSMB, NWO Genomics), NBIC/BioAssist/RK(2008.024), Biobanking and Biomolecular Resources Research Infrastructure, BBMRI – NL184.021.007 184.033.111 X-Omics184-034-019 Spinozapremie, NWO56-464-14192 KNAW Academy Professor AwardPAH/6635 University Research Fellow (URF)Amsterdam Public Health research institute (former EMGO+), Neuroscience Amsterdam research institute (former NCA); European Community’s Fifth and Seventh Framework ProgramFP5- LIFE QUALITY-CT-2002-2006 FP7- HEALTH-F4-2007-2013 01254 GenomEUtwin01413 ENGAGE602768 European Research Council284167 771057 230374 Rutgers University Cell and DNA RepositoryNIMH U24 MH068457-06 National Institutes of Health (NIH)R01D0042157-01A1 R01MH58799-03 MH081802 DA018673 R01 DK092127-04 Grand Opportunity1RC2 MH089951 1RC2 MH089995 Avera Institute for Human GeneticsGenetic Association Information Network (GAIN), Foundation for the National Institutes of Health, US National Institute of Mental HealthRC2 MH089951 NWO2018/EW/00408559 Doctoral Programme in Population Health (DOCPOP), University of Helsinki, FinlandAcademy of Finland265240 263278 Sigrid Juselius Foundation ==== Body pmcIntroduction In recent decades, the prevalence of obesity has been increasing [1], and it is predicted that almost one-fourth of the world’s population will be affected by obesity in 2035 [2]. The co-morbidities related to obesity include a wide range of high-prevalence diseases, including type 2 diabetes and cardiovascular disease [3–5], making it a major public health concern. Body mass index (BMI) is commonly used as a marker of obesity and body fat, the former being defined as BMI >30 kg.m−2. Both genetics and the environment influence BMI, with twin studies estimating ~75% of its variance attributable to genetic factors in young adults [6]. Multiple proteins and peptides, such as the hormones leptin, ghrelin, and resistin, take part in the complex processes of regulating energy balance, disturbances of which can induce obesity or anorexia [7,8]. High-throughput technologies have revolutionized obesity research, and omics data are key to an in-depth functional understanding of obesity [5]. An omics technique successful in investigating obesity-related co-morbidities is proteomics, which comprises the large-scale study of proteins [9]. For example, beyond the risk of general obesity in the development of diabetes, plasma protein markers of abdominal fat distribution are associated with an increased risk of diabetes [10]. Mendelian randomization identified unidirectional and bidirectional (i.e., protein-to-BMI and/or BMI-to-protein) causal associations of plasma proteins with BMI [11,12], signifying a direct imprint of the proteome on obesity, and vice versa. Mendelian randomization leverages genetic variants that influence protein levels and BMI, but little is known about how much of the association between protein levels and BMI is due to genetic and environmental influences. Multi-omics approaches that combine multiple omics layers in a single analysis better characterize the underlying biology of complex diseases than single-omics approaches [13,14]. The few multi-omics studies conducted for obesity demonstrated great potential in acquiring a better understanding of obesity, notably when coupling transcriptomic and metabolomic data [15,16]. Specifically, one study showed that some metabolite and gene expression associations with BMI were mediated by the epigenome [15], while another study observed associations of weight gain with changes in metabolite levels and blood cell function [16]. Among twin pairs discordant for BMI, weight differences were associated with lipidomics independent of genetic influences [17]. Overall, only a few multi-omics studies quantified biomolecules (i.e., metabolites or proteins) and omics stability in relation to participants’ post-intervention weight changes [18–20]. Weight development and obesity in adolescents and children from a multi-omics perspective is largely underexplored. The large inter- and intra-individual variability in the proteome, metabolome, and transcriptome in children [21] holds great promise for identifying obesity biomarkers across omics layers. Better identification of biomarkers for obesity in children and adolescents may allow both prevention of its onset in this age group and in adults, as childhood obesity increases the likelihood of adult obesity [22]. Longitudinal designs can help capture factors associated with weight change, and thus better predict individuals at risk for obesity. A few longitudinal single-omic studies have explored the associations between changes in BMI with proteomic [23–26] or metabolomic data [27]. While the studies involving proteomic data are mainly based on adult populations, one study to investigate the association between weight change and metabolites was conducted in children [28]. No one, to our knowledge, has conducted longitudinal multi-omics studies of weight change and BMI including proteomics data in children or adolescent populations yet. Plasma proteins are associated with BMI in adults, where some plasma proteins were shown to be causally influencing increases in BMI, while other plasma proteins were causally influenced by higher BMI [11,12]. As less is known about these associations in adolescence and for changes in BMI, the primary objective of the current study was to identify plasma proteins associated with BMI at blood sampling and changes in BMI during adolescence and in adulthood (Figure 1). Here, we refer to BMI at blood sampling and changes in BMI together as BMI trajectories. As changes in BMI in adolescents reflect substantial changes in lean body mass [29], observing an overlap of proteins associated with BMI changes during adolescence with those reported in the literature in adults could indicate whether proteins associated with BMI changes are due to changes in body fat or lean mass. We also elucidated the genetic and environmental sources underlying both the protein abundances and their correlations with BMI and changes in BMI. How genetic differences give rise to differences in plasma protein abundances and in BMI and changes in BMI is influenced by a complex interplay of multiple omics layers. To characterize the associations of multiple omics layers with BMI and changes in BMI, we next investigated whether gene expression of relevant protein-coding genes were associated with BMI trajectories in an external cohort. Finally, we linked identified proteins and their coding genes to plasma metabolites and polygenic risk scores (PRS) for obesity and coronary artery disease (CAD) using mixed-effect models and correlation networks. Methods Study population and longitudinal measurements FinnTwin12 FinnTwin12 is a longitudinal cohort based on a population of Finnish twins born between 1983 and 1987 aimed at investigating adolescent behavioral development and health habits [30,31]. Participants were identified from the Central Finnish Population Register and completed four questionnaires (response rate range: 85–90%) at approximate ages 12, 14, 17, and 22. A subset of twins, referred to as intensive subset, was studied more intensively, with additional psychiatric interviews and questionnaires starting at the wave corresponding to age 14. Besides health, lifestyle, and psychological indicators, questionnaires collected up to four self-reported weight and height measures for these participants across the four waves, from which BMI values (kg.m−2) were calculated. At the 22-year assessment wave, 786 twins from the intensive subset participated in in-person assessments and provided venous blood plasma samples after overnight fasting. We assessed the validity of self-reports of weight and height at the 22-year assessment wave versus in-person measurements in a sample of 756 of these participants (see Additional file 2). Self-reported and measured values correlated strongly for weight (Pearson correlation: r=0.98) and height (r=0.99). From the blood samples, extensive omic data were derived, including proteomics, metabolomics, and genotyping to generate the PRS data used in the current study. Of these participants, 651 had complete longitudinal anthropometric measurements over the 10 years of follow-up and constituted the final FinnTwin12 sample (Table 1). The sample included 250 monozygotic twins (106 complete pairs) and 401 dizygotic twins (164 complete pairs). Height and weight across the waves can be found in the supplementary material (see Additional file 3, Table S1). Netherlands Twin Register The Netherlands Twin Register (NTR) is a population-based cohort which includes twins and multiples from the Netherlands [32]. Adult participants of the NTR have completed a survey on health and behavior every 2–3 years since 1991. In contrast to the FinnTwin12 cohort, where data collection is based on age, the NTR data collection in adult twins is independent of age; adult participants are invited to fill out a survey every ~3 years. The blood samples for the omics experiments described in this study were collected in participants of the NTR biobank project (2004–2008)[33]. Venous blood samples were drawn between 7 and 10 AM after overnight fasting, and fertile women had their blood sample drawn in their pill-free week or on day 2–4 of their menstrual cycle. Information on BMI was obtained during the home visit for blood collection as part of the NTR Biobank project. Additionally, we calculated BMI from self-reported height and weight from NTR Survey 5 (2000), Survey 6 (2002), and Survey 7 (2004). Transcriptomics data was available for 3369 individuals, of which we selected 965 participants for whom blood sampling was performed before age 30 (i.e., age at blood sampling < 30 years old). Of these participants, we retained those for whom BMI was available at the time of blood sampling as well as at least one other BMI measure at Survey 5, Survey 6, or Survey 7. The final NTR sample comprised 665 participants with 2 to 4 BMI measurements over a mean follow-up period of ~6 years (Table 1). The sample included 359 monozygotic twins (141 complete pairs) and 306 dizygotic twins (103 complete pairs). Omics processing Proteomics Proteins from the plasma samples of FinnTwin12 participants were subjected to Liquid Chromatography-Electrospray Ionization-Mass Spectrometry (LC-ESI-MS/MS) as described previously [34]. First, proteins were precipitated with acetone and subjected to in-solution digestion according to the standard protocol of the Turku Proteomics Facility (Turku Proteomics Facility, Turku, Finland). After digestion, peptides were desalted using a 96-well Sep-Pak C18 plate (Waters), evaporated to dryness, and stored at −20°C. A commercial kit (High Select™ Top14 Abundant Protein Depletion Mini Spin Columns, cat. Number: A36370, ThermoScientific) was used to deplete the 14 most abundant proteins from plasma before the proteomic analysis. Samples were first analyzed by independent data acquisition LC-MS/MS using a Q Exactive HF mass spectrometer. Data were further analyzed using Spectronaut software and included local normalization of the data [35], as described elsewhere [34]. Raw matrix counts were log2-transformed and kit-depleted proteins were removed, including human serum albumin (HSA), albumin, IgG, IgA, IgM, IgD, IgE, kappa and lambda light chains, alpha-1 acid glycoprotein, alpha-1 antitrypsin, alpha-2 macroglobulin, apolipoprotein A1, fibrinogen, haptoglobin, and transferrin. None of the participants had scores greater or less than 5 standard deviations from the mean on the first two principal components (PCs) derived from principal component analysis (PCA), indicating none of the participants were identified as outliers. We excluded proteins with >10% missing values. Following the identification of 4 batches, imputation of missing values was performed (proportion of missing values in dataset: 0.86%) using the lowest observed value per batch. Corrections for batch effects were performed with Combat [36]. The final proteomic dataset comprised 439 proteins and protein abundances were scaled, such that one unit corresponded to one standard deviation (sd) with zero mean. Transcriptomics Details of pre-processing can be found in Jansen et al. [37]. In short, heparinized whole blood was transferred into PAXgene Blood RNA tubes (Qiagen, Valencia, Florida, USA) within 20 minutes of sampling, and stored at −30°C. Total RNA was extracted using the PAXgene Blood RNA MDx kit protocol in 96-well format with the BioRobot Universal System (Qiagen, Valencia, Florida, USA). RNA quality and quantity was assessed by Caliper AMS90 with HT DNA5K/RNA LabChips [37–39]. Samples were randomly assigned to plates and co-twins were randomized across plates to avoid bias in family correlation estimates. For cDNA synthesis, 50 ng of RNA was reverse transcribed and amplified in a plate format on a Biomek FX liquid handling robot (Beckman Coulter, Brea, California, USA) using Ovation Pico WTA reagents per the manufacturer’s protocol (NuGEN, San Carlos, California, USA). Products purified from single-primer isothermal amplification were then fragmented and labeled with biotin using Encore Biotin Module (NuGEN). Prior to hybridization, the labeled cDNA was analyzed using electrophoresis to verify the appropriate size distribution (Caliper AMS90, HT DNA 5K/RNA LabChip). Samples were hybridized to Affymetrix U219 array plates, and array hybridization, washing, staining and scanning were carried out per the manufacturer’s protocol (GeneTitan, Affymetrix, Santa Clara, California, USA). Gene expression data were required to pass standard Affymetrix quality control metrics (Affymetrix expression console). Probes were removed when their location was uncertain or if their location intersected a polymorphic single nucleotide polymorphism (SNP). Expression values were obtained using robust multi-array average normalization implemented in Affymetrix Power Tools (v 1.12.0). We excluded samples that displayed an average Pearson correlation below 0.8 with the probe set expression values of other samples and samples with incorrect sex chromosome expression. In the analyses, we retained only probes for genes encoding proteins associated with BMI or BMI changes in FinnTwin12. We attempted to match protein-coding gene names, however, we failed to retrieve three protein-coding genes (AMY2A, C4A, and CFHR1) identified in FinnTwin12 from the NTR transcriptomic data. We focused on log2-transformed gene-level expression in our analyses, where we obtained gene-level expression by averaging probe level expression per gene, for each gene associated with protein-coding genes. The same analyses were also performed for log2 probe expression as supplementary analyses. Metabolomics Metabolites were quantified from EDTA plasma samples using high-throughput proton nuclear magnetic resonance spectroscopy (1H-NMR) on a laboratory setup that combines a Bruker AVANCE III 500 MHz and a Bruker AVANCE III HD 600 MHz spectrometer (Nightingale Health Ltd, Helsinki, Finland)[40,41]. In short, this method provides simultaneous quantification of routine lipids, lipoprotein subclass profiling with lipid concentrations within 14 subclasses, fatty acid composition, and various low-molecular weight metabolites, including amino acids, ketone bodies, and glycolysis-related metabolites in molar concentration units. Of the 651 participants from the FinnTwin12 sample, 638 had metabolomic data available of which the preprocessing has been described elsewhere [42]. The data included 114 metabolites, of which 13 had missing values. None of the metabolites had more than 10% missing values, we therefore imputed all missing values with the lowest observed value of each metabolic biomarker. Besides LDL cholesterol as derived from the NMR platform, LDL cholesterol was quantified using the Friedewald equation [43]. Similar to the proteomic data, no outliers were detected on the first two PCs. The metabolite values were log2-transformed and scaled so that one unit corresponded to a change of one sd, with a mean of zero. In the NTR, with the exception of six non-overlapping metabolites (see Additional file 3, Table S2), we only retained metabolites that were significantly associated with BMI trajectories in FinnTwin12. Description for the NTR metabolomics (NMR platform) has been supplied in detail elsewhere [44]. Missing values represented less than 3% of the total data points, and no metabolite had more than 20% missing values, thus imputation was performed by the lowest observed value for each metabolite. Data were log2-transformed, and a technical variable indicating batch was used as a covariate in all relevant analyses. Polygenic risk scores PRSs were used to assess whether genetic susceptibility could explain the association of protein or metabolite levels with BMI trajectories. In FinnTwin12, genotype data for most participants (645 of 651) were available and genotyping was performed using the Illumina Human CNV370-Duo chip. Processing details are available elsewhere [45]. We calculated three PRSs describing the genetic susceptibility to BMI [46], Waist to Hip Ratio adjusted with BMI (WHR)[47], and CAD [48]. After correcting the PRSs for population stratification [49], by regressing out the top ten genetic PCs, the PRSs were scaled to a mean of zero and unit variance. Of the 665 NTR participants, 576 were genotyped using multiple platforms (see Additional file 4). For genotyped NTR participants, we calculated the PRS of BMI [46], using the infinitesimal prior (LDpred-inf) in Ldpred [50]. Detailed information on genotyping and PRS calculation in NTR participants is included in the supplementary material (see Additional file 4). The PRS was first corrected for population stratification [49] and for technical batch (i.e., platform) by regressing out the top ten genetic PCs and platform indicator, respectively, and then the PRS was scaled to have a mean of zero and unit variance. Statistical analysis The primary analyses comprised two sequential steps: 1) calculating growth factors summarizing BMI trajectories and 2) performing association analyses of omics layers with BMI trajectories using mixed effects models (Figure 1). Together, these analyses allowed for the joint investigation of longitudinal BMI measures and omics data, associating proteomic, transcriptomic, metabolomic, and PRS data with BMI trajectories. Genetic levers underlying protein abundances and the associations between proteins and BMI trajectories were further explored by univariate and bivariate twin modeling. A multi-omics correlation network was then constructed to put the omics variables associated with changes in BMI into perspective. Trajectories computation Latent Growth Curve Modeling Latent Growth Curve Modeling (LGCM) was used to summarize the anthropometric trajectories into intercepts and slopes, corresponding to baseline measurements and rates of change, respectively. LGCM analyses were run with the lavaan package version 0.6–12 [51] in R version 4.1.2. To accommodate the shape of the BMI trajectories, for which non-linearity in time is likely, we considered two modeling schemes, corresponding to linear and logarithmic dependence of time, respectively. When fitting the model, BMI measures were linearly adjusted for age differences within waves, with reference ages corresponding to the observed means per wave in the FinnTwin12 and NTR cohorts (Table 1). We tested for adjustment of the growth factors for sex, and further adjusted the model when the effect of sex was significant (p-value <0.05). Missing BMI values were allowed during model fitting by maximum likelihood. We calculated robust standard errors to correct for clustering in the data due to family relatedness. We assessed the goodness of fit of the linear and logarithmic schemes using the Comparative Fit Index (CFI), the Tucker-Lewis Index (TLI), and the Root Mean Square Error of Approximation (RMSEA). A model with RMSEA less than 0.06, CFI greater than 0.95, and TLI greater than 0.95 was considered a good fit [52,53]. Whenever the linear modeling did not enable good model fit, the logarithmic scheme was preferred when it did enable a good model fit. Sensitivity analyses We carried out two sensitivity analyses. The first sensitivity analysis was performed in FinnTwin12, which consisted of repeating the analyses of the association between protein levels and baseline BMI or changes in BMI with another marker of adiposity: the triponderal mass index (TMI). TMI is defined as mass divided by height cubed, in contrast to BMI dividing mass by the square of height [54]. BMI is a good indicator of adiposity for most age groups, but TMI is sometimes preferred as an indicator in children and adolescents [54–56]. We performed the second sensitivity analysis in NTR. In NTR, surveys were collected regardless of the age of the participants, thus, each wave comprised all participants that were 18 years or older at the time of the survey. Because LGCM corrected for within-survey age differences, we used a second approach that did not correct for wave-level age differences, which were larger in NTR than in FinnTwin12 (Table 1). Instead of considering reference ages by which ages are corrected, as in the main analyses, we took each individual’s actual age at baseline as the first time point and then modeled changes in BMI from this individual’s baseline age. We then quantified the associations between these changes in BMI with gene expression. Mixed-effect models To quantify associations between the plasma proteomics data and BMI trajectories, we used mixed-effects models. Mixed-effects modeling involves using two types of effects, fixed and random, the former being constant across individuals and the latter varying between individuals. We tested associations between proteins and BMI at blood sampling and slope of BMI (i.e., BMI changes) in FinnTwin12. We also tested associations between proteins and intercept of BMI (i.e., baseline BMI) as the intercept of BMI was that of young adolescents, whereas the BMI at blood sampling was that of young adults. We included age at blood sampling, age squared at blood sampling, sex, and interactions between sex and age and age squared as covariates. BMI intercept was also included as a covariate to correct for baseline BMI differences when testing associations between proteins and changes in BMI. The model included two random effects whose main function was to correct for clustering in the data due to family relatedness: a binary indicator for zygosity and the family identifier. We tested the association between the coefficients of the fixed effects and each protein using a t-test with Satterthwaite approximation [57] and Bonferroni correction. We considered an association significant if the Bonferroni corrected p-value was less than 0.05 (i.e., raw p-value < 0.05/439=1.1 ×10−4). The same modeling was used for identification of metabolites associated with BMI trajectories (Bonferroni: raw p-value < 0.05/114=4.4 ×10−4) as well as to quantify associations between PRSs and proteins (Bonferroni: raw p-value < 0.05/439=1.1 ×10−4). We repeated the same modeling in NTR, replacing proteomic variables with probe- or gene-level expression of protein-coding genes identified in FinnTwin12, or with metabolites. We added a technical covariate to correct for batches in the modeling between metabolites and BMI trajectories. Multiple correction was applied to the number of tests performed, with the associations of BMI and changes in BMI with omic variables (e.g., metabolite, gene) studied independently. We did not examine the associations between the omics layers and the BMI intercept, as BMI at baseline and BMI at blood sampling reflected both BMI at adulthood. The association between PRS of BMI and gene expression at the probe and gene level was also quantified. Classical twin modeling We used classical twin models (CTMs) to quantify the genetic and environmental contributions to protein abundances, and to decompose the associations between protein and BMI trajectories into genetic and environmental correlations. CTMs hinge on the comparison between MZ and DZ twins [58,59], since MZ twins share approximately 100% of the genomic sequence of their DNA, while DZ twins share, on average, only 50% of their segregating genes. By comparing the magnitude of correlations in DZ twin pairs (rDZ) to those in MZ twin pairs (rMZ), we can infer the presence of genetic factors. In univariate CTM, the variance (V) of a trait can be decomposed into several components: additive genetic (A), non-additive or dominant genetic (D), shared or common environmental (C), and nonshared or unique environmental (E). In classical twin modeling of pairs reared together, simultaneous quantification of the C and D components is not possible. A ratio rDZ/rMZ above or below 0.5 is therefore an indication that C or D, respectively, can be modeled. Since the quantification of D requires considerable statistical power, we have only estimated A and E, with C where appropriate. Heritability (h2) is defined as A/V, and the standardized coefficients e2 and c2 denote the part of the variance V explained by E and C, respectively [60,61]. We first fitted saturated models to obtain estimates of the twin correlations by maximum likelihood (ML), including sex and age at blood sampling as covariates. Analyses were conducted in R software using the OpenMx package, version 2.20.7 [62–65]. In univariate settings, proteins for which rMZ/rDZ<0.5 were fitted by AE models. Proteins for which rMZ/rDZ>0.5 were fitted with ACE models, which were compared with AE, CE, and E only. We used Akaike Information Criterion (AIC) to choose the best model for these proteins. Negative variance estimates were allowed [66]. Model comparison by AIC is available in the supplementary material (see Additional file 3, Table S3). Similarly, we decomposed phenotypically significant correlations between proteins and BMI trajectories. Since twin correlations indicated an AE model for BMI at blood sampling and changes in BMI in the univariate framework, we decomposed their phenotypic correlations with proteins into genetic (rA) and non-shared environmental (rE) correlations in the bivariate framework. The bivariate twin analysis between changes in BMI and protein abundance included the BMI intercept as a covariate in addition to sex and age. Proteins modeled by CE rather than AE (i.e., having h2=0) were excluded from the bivariate modeling. Cross-twin cross-trait correlations can be found in the Supplementary Document (see Additional file 3, Table S4, Table S5). Multi-omic network construction A correlation network was constructed to provide a multi-omics framework of the mechanisms underlying BMI change in FinnTwin12, using the significant findings from the prior analyses. The network included the 14 proteins that were associated with the change in BMI and the 3 PRSs. Metabolites associated with the change in BMI were also included, but the set of 48 highly correlated lipoproteins (out of 53 metabolites) were reduced by PCA into 3 principal components (LIPO.PC1, LIPO.PC2, and LIPO.PC3), explaining 87% of the variance. This allowed for better visual representation of the connections between omics. Each of the 25 final entries was linearly regressed on BMI slope, BMI intercept, and BMI at blood sampling, as well as simple and quadratic age at blood sampling, sex, age by sex interaction, and quadratic age by sex interaction. Correlations between these 25 residual variables were used to construct the multi-omics network. We colored the edges according to the significance of the correlation assessed by Pearson’s correlation (i.e., raw p-value < 0.05 or Bonferroni-corrected p-value < 0.05) and the direction of the correlation (i.e., positive or negative). In NTR, we used a similar methodology to correlate metabolomics residual variables (3 lipoprotein PCs and 5 low molecular weight molecules) with gene expression (N=14). Results Longitudinal development of BMI in the two cohorts In FinnTwin12, only the slope of BMI was associated with sex (intercept: p=0.70; slope: p < 0.001) and was therefore further adjusted by sex in the modeling. Logarithmic modeling achieved the best model fits (RMSEA: 0.05 [95% confidence interval (CI): 0.03,0.06], TLI=0.97, CTI=0.98) compared to linear modeling (RMSEA: 0.09 [95% Confidence Interval (IC): 0.07,0.11], TLI=0.93, CTI=0.94). We therefore kept logarithmic modeling for the calculation of growth factor values. The mean BMI intercept was 17.6 kg.m−2. Male and female participants gained an average of 6.2 and 5.0 BMI units, respectively, during the 10-year follow-up period (Table 1). In NTR, both the intercept and the slope were sex adjusted (intercept: p=0.01; slope: p=0.02) and linear modeling provided the best model fits (RMSEA: 0.04 [95% CI: 0.02,0.06], TLI=0.99; CTI=0.99). Table 1 describes the mean intercepts and slopes. Only twelve NTR participants had a decrease in BMI over the course of follow-up, and overall, the change in BMI in NTR showed substantial weight gain at the population level (z-value= 3.3, p=0.001). During the 6-year follow-up, male and female NTR participants gained an average of 1.5 and 1.3 BMI units, respectively, (Table 1). Growth factor variance was significantly non-zero in both FinnTwin12 and NTR datasets (z-values>3.0, p<0.01), indicating significant inter-individual differences longitudinally. Proteins and BMI trajectories in FinnTwin12 Phenotypic associations We examined associations between plasma protein abundance and BMI trajectories (i.e., BMI at blood sampling and slope of BMI) from mixed models using scaled BMI trajectory markers (Figure 1). A total of 66 proteins out of 439 proteins showed a significant association with BMI at blood sampling, of which 42 had a negative and 24 had a positive association with BMI (Table 2). Proteoglycan 4 had the strongest association with BMI (coefficient: 0.46; p=2.1E-34); an increase of one sd in the abundance of proteoglycan 4 was associated with an increase of almost a half of sd in BMI (Table 2). A total of 14 protein abundances were associated with changes in BMI, independent of baseline BMI levels (Table 3). These proteins included complement factors and proteoglycan 4, with proteoglycan 4 having the strongest association with adolescent BMI changes (estimate: 0.27; p=4.3E-10). Ten proteins had a positive association with changes in BMI and 4 had negative associations. Thirteen out of 14 proteins were also correlated with BMI at blood sampling. Similarly, we investigated associations between proteins and BMI intercept (i.e., baseline BMI; see Additional file 3, Table S6), of which 11 significant associations were detected. Of these, 9 were also associated with BMI at blood sampling. These findings suggest that the proteome in young adults contains proteins associated with both childhood and adulthood BMI. Proteoglycan 4, complement factor H, C-reactive protein, and secreted phosphoprotein 24 were associated with BMI at blood sample collection, BMI intercept, and BMI slope. Associations between proteins with triponderal mass index (TMI) at baseline (~12 years old) and slope of TMI can be found in the supplementary material (see Additional file 3, Table S7). In total, we found 10 and 22 proteins associated with baseline TMI and TMI changes, respectively. Seven of the ten proteins associated with baseline TMI were also associated with baseline BMI. All proteins associated with BMI changes were also associated with TMI changes. This shows that BMI and TMI associations with the plasma proteome tend to have similar proteins involved. Genetic and environmental sources underlying protein abundance Univariate twin modeling was used to quantify the genetic and environmental sources of variance in the 66 proteins identified as associated with either BMI at blood sampling and/or the slope of BMI (Figure 2A). According to the ratio between rMZ and rDZ (Figure 2B), we obtained 13 CE models, 1 ACE model, and 52 AE models. On average, genetic and nonshared environmental factors explained 35% (range: 0–78) and 59% (range: 22–89) of the variance in protein abundances, respectively (Figure 2C). Thirteen proteins showed no evidence for genetic effects (h2=0), and all of these had non-zero shared environmental components that accounted for an average of 28% (range: 19–41) of the variance, with nonshared environmental factors accounting for 72% (range: 59–81) of the variance in the abundances of these proteins (Figure 2A). Heritability of BMI at blood sampling was 72% [95% CI: 62,78], and the nonshared environment accounted for the remaining (28%) of the variance in BMI at blood sampling. Heritability of the slope of BMI was 63% [95% CI: 51,73] while the nonshared environment explained the remaining (37%) variance in BMI changes. All univariate estimates and their 95% CIs are available in the supplementary material (see Additional file 3, Table S3, Table S8). Genetic and environmental correlations between BMI trajectories and proteins The decomposition of significant phenotypic covariation between plasma protein abundances (h2>0) with BMI and BMI changes were performed with bivariate genetic twin modeling. For all associations between proteins and BMI at blood sampling or BMI change, we applied bivariate AE models. Of the 53 BMI-protein correlations tested, we observed 43 significant genetic correlations and 12 significant nonshared environmental correlations (Figure 3A). For 8 protein–BMI associations, we observed both significant genetic and nonshared environmental correlations. Overall, genetic correlations ranged from −0.50 to 0.50 (absolute mean: rA=0.29) and environmental correlations from −0.33 to 0.44 (absolute mean: rE=0.14). Seven proteins had genetic correlations with BMI greater than 0.4 in absolute value, and proteoglycan 4 had the highest nonshared environmental correlation with BMI among all proteins tested (rE= 0.44 [95% CI: 0.28,0.57]). The full set of genetic and nonshared environmental correlations is available in the supplemental material (see Additional file 3, Table S9). Similarly, the significant associations between proteins and changes in BMI were decomposed into genetic and nonshared environmental correlations for 11 of the 14 proteins associated with changes in BMI (Figure 3B). We observed 6 significant genetic correlations and 4 nonshared environmental correlations (see Additional file 3, Table S10). Because BMI change during adolescence predicted adult BMI independently of baseline BMI (linear regression: R2=45%), we examined whether the association between BMI at blood sampling and BMI change might confound the observed correlations between proteins and BMI. After adjusting the slope of BMI by BMI at blood sampling before bivariate twin modeling (Figure 3B), only apolipoprotein A-IV remained genetically correlated with the slope of BMI (rA= 0.20 [95% CI: 0.02, 0.38]), while all other correlations were no longer significant (see Additional file 3, Table S11). These results suggest that the genetic and environmental factors influencing the association between changes in BMI during adolescence and protein levels in adulthood may reflect associations with adult BMI. Gene expression and BMI trajectories in NTR Mixed-effects models were used to quantify the associations of BMI trajectories with the expression of genes encoding the proteins identified in FinnTwin12. Only S100A8 (protein encoded: S100 Calcium Binding Protein A8) was significantly associated with BMI at blood sampling (estimate=0.21; sd=0.05; p=1.5E-03). The PRG4 gene (protein encoded: proteoglycan 4) was significantly associated with the slope of BMI (estimate=0.09; sd=0.03; p=0.05). While the association of the CFI gene (protein encoded: complement factor I) with the slope of BMI did not pass Bonferroni correction (p=0.21), one of its probes (ID: 11718481_s_at; start: chr4:110661852; end: chr4:110663751) was significantly associated with changes in BMI (p=0.02) at the probe level. Summary statistics are available in the supplementary material (see Additional file 3, Table S12, Table S13, Table S14, Table S15), and estimates from the sensitivity analysis are also available in the supplementary material (see Additional file 3, Table S16, Table S17). Metabolites and BMI trajectories A total of 78 out of 114 metabolites were significantly associated with BMI at blood sampling in FinnTwin12. The mean absolute effect, which is 1.4 times greater than that observed for proteins, was 0.28 sd change in BMI per 1 sd change in metabolite level. Of the 78 metabolites, 68 were lipoproteins. The rest included Friedewald LDL cholesterol [43] and 9 low molecular weight molecules (LMWM): citrate, glycerol, glycoprotein acetyls, isoleucine, leucine, valine, CH2 groups of mobile lipids, phenylalanine, and tyrosine. In NTR, 44 out of the 78 metabolites were also significantly associated with BMI at blood sampling (Figure 4A; see Additional file 3, Table S18, Table S2). In FinnTwin12, 53 metabolites were associated with the slope of BMI, and all of these were also associated with BMI at the time of blood sampling. These metabolites comprised 48 lipoproteins and five LMWMs: glycoprotein acetyls, isoleucine, valine, phenylalanine, and tyrosine. In NTR, 14 of the 53 metabolites were also significantly associated with the BMI slope (Figure 4A), of which 13 were lipoproteins besides glycoprotein acetyls. In FinnTwin12, 19 metabolites were associated with the intercept of BMI (i.e., baseline BMI; baseline age: ~12 years). Summary statistics are available in the supplementary material (see Additional file 3, Table S19, Table S20, Table S21). Polygenic risk scores and BMI-associated omics traits We first examined the association of PRSs with proteins identified as associated with BMI and changes in BMI in FinnTwin12. Of the 66 proteins associated with BMI at blood sampling, none were associated with the PRS of BMI after Bonferroni correction (p > 0.05; see Additional file 3, Table S22). However, we observed consistent directions of effect between the genetic correlations of proteins and BMI and the associations between the PRS of BMI and proteins (Figure 3C). In NTR, we quantified the association of BMI protein-encoding gene expression with the PRS of BMI, but found no significant associations. We then compared the t-values of associations between gene expression and the PRS of BMI with the t-values of associations between gene expression and BMI, and found weak evidence for consistency in the direction of the compared t-values (linear regression coefficient nullity-test: p=0.015). In FinnTwin12, we also investigated the associations between proteins and the PRSs of CAD and WHR (see Additional file 3, Table S22). None of the associations between proteins and the PRS of WHR were significant, with P06702 (Bonferroni-corrected p-value=0.06) and Q15848 (p=0.10) having the lowest Bonferroni-corrected p-values. We found two significant associations between proteins and the PRS of CAD: C4b-binding protein alpha chain (protein ID: P04003; raw p-value=6.0E-04; Bonferroni p-value=4.0E-02) and Apolipoprotein B-100 (protein ID:P04114; raw p-value=5.2E-04; Bonferroni p-value=3.4E-02). We also conducted additional analyses between PRSs and proteins associated with the slope of BMI in the FinnTwin12 cohort (see Additional file 3, Table S25), between the PRS of BMI and genes encoding the proteins identified associated with the slope of BMI in NTR (for gene level results see Additional file 3, Table S26, and for probe level results see Additional file 3, Table S27), and between PRSs and metabolites in both cohorts (see Additional file 3, Table S28, Table S29). Multi-omic correlation network To examine the association of the proteome, metabolome, and PRSs with BMI trajectories, we constructed a multi-omics correlation network in FinnTwin12 (Figure 4B) with previously identified proteins and metabolites adjusted for covariates and BMI at blood sampling, slope of BMI, and BMI intercept. We observed 49 correlations between omic layers, of which 30 were positive and 19 negative (Figure 4B). Apart from associations between the PRS of CAD with P04003 and LIPO.PC3, the remainder were metabolome-proteome associations. Twelve of the fourteen proteins associated with changes in BMI shared at least one significant correlation with a residual metabolomic variable, the average number of connections being 3.9 out of 8 possible connections for these proteins. These correlations indicate multi-omic relationships between omics variables associated with a 10-year change in BMI, independent of BMI trajectories markers. We also examined the associations between the transcriptome and metabolome in NTR, but found no significant multi-omics relationships (see Additional file 3, Table S30, Table S31). Discussion Overall, a substantial number of plasma proteins were associated with adult BMI and BMI change during adolescence. The estimates for the heritability of plasma abundance of the identified proteins varied. For some proteins, familial resemblance was influenced by shared (i.e., familial) environmental factors rather than genetics. The etiology of the observed protein-BMI trajectory associations revealed genetic and/or nonshared environmental correlations between BMI and protein levels. Few associations between blood expression levels of protein-encoding genes and BMI trajectories were observed in NTR adults. This suggests a potential for observing gene-protein pairs associated with BMI trajectories at two different omic levels, among both adolescents and adults from two different countries. Finally, the use of metabolomic data and PRSs allowed both the observation of cross-omic associations and a better understanding of the connections that link BMI trajectories to different omic layers. The twin design allowed the quantification of the genetic and environmental sources of variation underlying the associations between BMI trajectories and the proteome from an epidemiologic perspective. Including two cohorts of different ages and populations also empowered the identification of biomolecules likely to be stable across generations and populations. This is in line with two longitudinal multi-omics studies that identified stable biomolecules in response to interventions and investigated proteome resistance to weight change [18,19]. Although BMI measurements at blood sampling were similar between the two cohorts (Table 1), both the length of follow-up and the intensity of BMI changes between the two cohorts differed. This may have limited our ability to compare biomolecule associations with changes in BMI between cohorts. Moreover, the relatively small sample sizes of the cohorts in the current study might have limited the statistical power to detect associations with PRSs or transcriptomic data. Using less conservative multiple testing corrections and cohorts with larger sample sizes may allow more associations to be detected. However, availability of larger multi-omics twin data is limited. An increase in the size and number of omics data in twin cohorts is therefore necessary to realize the full potential of twin designs. BMI changes in FinnTwin12 adolescents were substantial from early adolescence to young adulthood in both male and female participants. These changes reflect multiple processes such as increase in lean body mass, and increase in body fat as growth continued from prepuberty onwards. Longitudinal omics data would allow for a more in-depth investigation of the relationship that links changes in BMI to the proteome or metabolome. Investigations among adolescents need to evaluate changes in body composition, while among adults BMI weight gain is primarily change in fat depots. Beyond describing body composition, it is still needed to characterize the causal relationships between changes in BMI and proteins or metabolites. Twin designs combining MR and direction of causation (DOC) twin models, e.g. MR-DOC models, could be used for this purpose [67]. However, there is a lack of sufficiently large genome-wide association studies of BMI changes, which precludes the use of sufficiently robust instrumental variables. Complement factors associated with BMI at blood sampling in FinnTwin12, including complements I, B, and H, have been associated with BMI previously [11,12,68]. Other proteins with strong associations with BMI at blood sampling (|estimate|>0.3; Table 2) in FinnTwin12 replicate findings from the literature, such as C-reactive protein [11,12,18,68], sex hormone-binding globulin [11,12,68], and proteoglycan 4 [69]. Several proteins associated with BMI at blood sampling have also been reported in the literature to be causally associated with BMI. These include those encoded by the CRP, NCAM1, IGFBP2, or SHBG genes [11,12], the last four of which are causally influenced by BMI (i.e., BMI-to-protein association) [12]. We identified several of the proteins associated with changes in BMI (i.e., gains in BMI) during adolescence that were also identified in adult populations that experienced weight loss. For example, proteoglycan 4, which was most strongly associated with BMI changes in our study, was previously associated with weight loss in adults [25] with the same direction of effect, i.e., positive covariation between change in BMI and proteoglycan 4 level. C-reactive protein, although previously shown to rapidly respond to weight loss [25], was also associated with changes in BMI over a 10-year follow-up in our study. These observations suggest the proteins identified in adult populations are, at least in part, relevant to the study of adolescents. Moreover, we observed that the direction of the effect between protein levels and BMI change echoed those in the literature, i.e., positive associations with BMI gain were negative with weight loss and vice versa. This indicates a stability of certain biomolecules in the proteome to characterize changes in BMI across ages, both in the study of weight gain and weight loss. Weight loss involves changes in lean mass as well as in body fat [70], which would be analogously the reverse regarding growth in adolescence. Overall, we found few proteins for which the expression of their coding genes were associated with BMI or changes in BMI. Post-transcriptional modifications affecting proteins, e.g., the post-transcription addition of chemical groups or polypeptides [71,72], can explain this lack of association, as they increase the complexity of the transcriptome–proteome connection. However, we observed that the gene expression of S100A8 (protein: S100A8), PRG4 (protein: proteoglycan 4), and CFI (protein: complement factor I) were associated with BMI at blood sampling, BMI changes at the gene level, and BMI changes at the probe level, respectively. The S100A8 gene encodes the eponymous protein belonging to the S100 family corresponding to Ca2+-binding proteins widely known to be associated with BMI [73] in both adults [74,75] and children [76]. S100A8 expression levels have previously been associated with BMI [77], echoing the dual association with BMI at both the transcriptome and proteome levels in our study. In contrast, no previous study reported associations between the gene expression of CFI and PRG4 genes with changes in BMI to our knowledge. Heritability estimates of BMI trajectories in FinnTwin12 adolescents (h2=63%) were slightly larger than those observed in adult Finnish twins (h2= 52–57%) [78], which is consistent with the higher heritability of BMI reported in younger populations [6] and further motivates the study of factors genetically associated with BMI changes during adolescence. Of body mass, fat accounts for a smaller proportion in children and adolescents in general, while gaining fat is more typical in adult populations The previous estimates for the heritability of plasma protein abundance associated with BMI trajectories appear to be rare in the literature. Liu et al. reported that the heritability of 342 plasma proteins in a relatively small sample of older twins [79] varied widely, which is in line with our observations. Other studies have quantified the heritability of disease-associated plasma proteins [80], but none seem to focus on obesity. The heritability estimates of the proteins identified in our study therefore constitute a valuable resource for the further study of the proteome–obesity relationship. C4b-binding protein alpha chain (Bonferroni p-value=4.0E-02) and apolipoprotein B-100 (p=3.4E-02) were associated with the PRS of CAD, with apolipoprotein B-100 being genetically correlated to BMI (rA=0.32). While the links between C4b-binding protein alpha chain and CAD are new, apolipoprotein B-100 is genetically associated with cardiovascular disease [81,82], as well as chronic kidney disease, blood pressure, and various lipids [83,84]. A better understanding of the mechanisms induced or inferred by this protein could provide insights into both obesity and its co-morbidities. We identified many metabolites associated with BMI and BMI changes in FinnTwin12, of which we replicated a substantial proportion in NTR. These metabolites included mostly lipoproteins and branched-chain amino acids with known associations with BMI changes [27]. Despite the strong connections observed between the metabolome and proteome in the study of BMI trajectories, we found no significant connections between the metabolome and the transcriptome. Despite the lack of results, which could be explained by statistical power issues, it is still unclear how the different omics layers are connected biologically. Having omics from multiple tissues would be one approach, but perhaps feasible only in model organisms. The identification of mediation mechanisms between omics layers may help in acquiring a better understanding of the multi-omics mechanisms underlying obesity [85,86]. Conclusions In conclusion, the proteome is a promising resource for cross-sectional and longitudinal studies of obesity. Both environmental and genetic factors influence the proteome–obesity association, and integrating multiple omics may pave the way for a better understanding of the biological mechanisms underlying obesity. We observed proteins associated with weight gain in adolescents that are also associated with weight loss in adults in the literature, suggesting that these may have more to do with increases and decreases of lean mass, respectively, than changes in body fat. Future studies with different designs (i.e., different populations, ages, or follow-up lengths) could provide a holistic view of how the proteome-obesity relationship is expressed. Supplementary Material Supplement 1 Supplement 2 Supplement 3 Supplement 4 Acknowledgements We gratefully acknowledge the contribution of the Turku Proteomics Facility team supported by Biocenter Finland for mass spectrometry. We warmly thank the participants involved in this study. Funding The Biobank-based Integrative Omics Study (BIOS) Consortium and the BBMRI Metabolomics Consortium are funded by BBMRI-NL, a Research Infrastructure financed by NWO, project nos. 184.021.007 and 184033111. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Phenotype and genotype data collection in FinnTwin12 cohort has been supported by the Wellcome Trust Sanger Institute, the Broad Institute, ENGAGE – European Network for Genetic and Genomic Epidemiology, FP7-HEALTH-F4-2007, grant agreement number 201413, National Institute of Alcohol Abuse and Alcoholism (grants AA-12502, AA-00145, and AA-09203 to R J Rose; AA15416 and K02AA018755 to D M Dick; R01AA015416 to Jessica Salvatore) and the Academy of Finland (grants 100499, 205585, 118555, 141054, 264146, 308248 to JK, and the Centre of Excellence in Complex Disease Genetics (grants 312073, 336823, and 352792 to JKaprio). This research was partly funded by the European Union’s Horizon 2020 research and innovation program under grant agreement No 874724 (Equal-Life). Equal-Life is part of the European Human Exposome Network. For the Netherlands Twin Register, funding was obtained from the Netherlands Organization for Scientific Research (NWO) and The Netherlands Organisation for Health Research and Development (ZonMW) grants 904-61-090, 985-10-002, 912-10-020, 904-61-193,480-04-004, 463-06-001, 451-04-034, 400-05-717, Addiction-31160008, 016-115-035, 481-08-011, 400-07-080, 056-32-010, Middelgroot-911-09-032, OCW_NWO Gravity program – 024.001.003, NWO-Groot 480-15-001/674, Center for Medical Systems Biology (CSMB, NWO Genomics), NBIC/BioAssist/RK(2008.024), Biobanking and Biomolecular Resources Research Infrastructure (BBMRI –NL, 184.021.007 and 184.033.111), X-Omics 184-034-019; Spinozapremie (NWO- 56-464-14192), KNAW Academy Professor Award (PAH/6635) and University Research Fellow grant (URF) to DIB; Amsterdam Public Health research institute (former EMGO+), Neuroscience Amsterdam research institute (former NCA); the European Community’s Fifth and Seventh Framework Program (FP5- LIFE QUALITY-CT-2002-2006, FP7- HEALTH-F4-2007-2013, grant 01254: GenomEUtwin, grant 01413: ENGAGE and grant 602768: ACTION); the European Research Council (ERC Starting 284167, ERC Consolidator 771057, ERC Advanced 230374), Rutgers University Cell and DNA Repository (NIMH U24 MH068457-06), the National Institutes of Health (NIH, R01D0042157-01A1, R01MH58799-03, MH081802, DA018673, R01 DK092127-04, Grand Opportunity grants 1RC2 MH089951, and 1RC2 MH089995); the Avera Institute for Human Genetics, Sioux Falls, South Dakota (USA). Part of the genotyping and analyses were funded by the Genetic Association Information Network (GAIN) of the Foundation for the National Institutes of Health as well as the US National Institute of Mental Health (RC2 MH089951). Computing was supported by NWO through grant 2018/EW/00408559, BiG Grid, the Dutch e-Science Grid and SURFSARA. GD has received funding for his doctoral studies from the Doctoral Programme in Population Health (DOCPOP), University of Helsinki, Finland. JK acknowledges support by the Academy of Finland (grants 265240, 263278) and the Sigrid Juselius Foundation. Availability of data and materials FinnTwin12 data analyzed in this study is not publicly available due to the restrictions of informed consent. Requests to access these datasets should be directed to the Institute for Molecular Medicine Finland (FIMM) Data Access Committee (DAC) ([email protected]) for authorized researchers who have IRB/ethics approval and an institutionally approved study plan. To ensure the protection of privacy and compliance with national data protection legislation, a data use/transfer agreement is needed, the content and specific clauses of which will depend on the nature of the requested data. The data of the Netherlands Twin Register (NTR) may be accessed, upon approval of the data access committee, through the NTR (https://tweelingenregister.vu.nl/information_for_researchers/working-with-ntr-data). List of abbreviations BMI Body mass index PRSs polygenic risk scores CAD coronary artery disease NTR Netherlands Twin Register HSA human serum albumin PCs principal components PCA principal component analysis sd standard deviation SNP single nucleotide polymorphism 1H-NMR high-throughput proton nuclear magnetic resonance spectroscopy LGCM Latent Growth Curve Modeling CFI Comparative Fit Index TLI Tucker-Lewis Index RMSEA Root Mean Square Error of Approximation TMI triponderal mass index CTMs classical twin models rDZ correlations in DZ twins rMZ correlations in MZ twins V variance A additive genetic component D non-additive or dominant genetic component C shared or common environmental component E nonshared or unique environmental component h2 heritability ML maximum likelihood AIC Akaike Information Criterion rA genetic rE nonshared environmental correlations CI confidence interval LMWM low molecular weight molecules DOC direction of causation Figure 1: Study flowchart The study was divided into two main sequential steps. First, growth factors were calculated and then association studies were performed. In the FinnTwin12 sample, univariate and bivariate twin modeling was performed to quantify the genetic and environmental sources underlying both the protein abundances and the correlations between proteins and BMI trajectories. Cross-omics associations were performed to bring a holistic perspective to the analyses. BMI: body mass index. NTR: Netherlands Twin Register. Figure 2: Genetic and environmental sources underlying proteins for which plasma abundance was associated with BMI or BMI change in the FinnTwin12 sample (A) Univariate twin models were used to decompose the variance of proteins identified as associated with BMI or BMI changes in FinnTwin12. (B) The models were determined by the ratio of the correlations in MZ and the correlations in DZ twins. (C) Fractions attributed to non-shared environmental factors and to heritability. BMI: body mass index. Figure 3: Genetic and non-shared environmental correlations between BMI trajectories and protein levels in the FinnTwin12 sample (A) Among the 53 heritable proteins for which associations with BMI were significant, bivariate twin models were conducted to dissect the nature of the associations. (B) Genetic and non-shared environmental correlations were also quantified in the association between slope of BMI and protein levels; lines correspond to confidence intervals at the 95% threshold, arrows point to estimates obtained by adjusting protein levels by BMI at blood sampling prior to bivariate twin modeling rather than using BMI intercept as a covariate (pink). (C) The direction of the genetic correlations observed between BMI at blood sampling and protein levels were consistent with the t-values observed by crossing the PRS of BMI with protein levels. BMI: body mass index. PRS: polygenic risk score. Figure 4: Extended cross-omic analyses (A) The t-values of the associations between metabolites with BMI and BMI change have proportional directions between the two samples NTR and FinnTwin12. (B) Multi-omic correlations between PRSs with proteins and metabolites associated with BMI change in the FinnTwin12 sample, independent of BMI trajectory markers (BMI blood sampling, BMI slope and BMI intercept). PRS: polygenic risk score. BMI: body mass index. NTR: Netherlands Twin Register. Table 1: Descriptive Statistics of the FinnTwin12 and NTR Cohorts Body mass index (BMI) Male participants Female participants Age Timepoint Cohort N (% female) mean sd mean sd mean sd Wave 12 FinnTwin12 651 (59%) 17.7 2.8 17.6 2.4 11.4 0.3 Wave 14 FinnTwin12 651 (59%) 19.5 3 19.5 2.6 14.0 0.1 Wave 17 FinnTwin12 651 (59%) 21.8 2.9 21.2 2.8 17.6 0.2 O Wave 22 FinnTwin12 651 (59%) 23.9 3.4 22.6 3.6 22.4 0.7 Survey 5 NTR 412 (71%) 22.0 2.9 21.2 2.7 20.7 2.3 Survey 6 NTR 484 (67%) 22.5 2.8 21.6 2.7 23.2 2.4 Survey 7 NTR 501 (70%) 23.3 3.2 21.9 2.9 25.4 2.5 Obbi-biobanking NTR 665 (69%) 23.5 3.2 22.5 3.2 26.8 2.3 Growth factors Intercept FinnTwin12 651 (59%) 17.6 2.4 17.6 2.0 Slope (nonlinear) FinnTwin12 651 (59%) 2.3 0.6 2.0 0.6 Intercept NTR 665 (69%) 21.9 2.4 21.2 2.2 Slope (linear) NTR 665 (69%) 0.3 0.1 0.2 0.1 Four longitudinal measurements with ~10 and ~6 years of follow-up were collected in the FinnTwin12 and NTR cohorts, respectively. Time points marked with (*) included blood sampling from which proteomic, metabolomic, and polygenic risk score data were derived. Intercepts and slopes were obtained by Latent Growth Curve Models, in a logarithmic scheme for FinnTwin12 and linear scheme for NTR. Final samples included 651 participants (female: 59%) for FinnTwin12 and 665 (female: 69%) for NTR. In FinnTwin12, BMI at baseline and last BMI measurement (wave 22) ranged 11.0–30.1 and 17.2–42.0 respectively. In NTR, BMI at baseline and last BMI measurement (bb1) ranged 15.2–40.3 and 15.8–51.3 respectively. BMI is expressed as kg/m2 and age as years. NTR: Netherlands Twin Register. N: Number of participants without missing value. sd: standard deviation. Table 2: Mixed model-derived estimates of proteins significantly associated with body mass index at blood sampling in FinnTwin12 participants Protein Fixed effect coefficients P-value description UniProt ID estimate se t-value raw bonferroni Neural cell adhesion molecule L1-like protein 000533 −0.16 0.04 −4.1 4.9E-05 2.2E-02 ADAM DEC1 015204 −0.15 0.03 −4.2 3.0E-05 1.3E-02 Coagulation factor XIII A chain P00488 −0.18 0.04 −5.1 5.0E-07 2.2E-04 Complement factor B P00751 0.15 0.04 4.1 5.5E-05 2.4E-02 Antithrombin-III P01008 −0.25 0.04 −6.7 4.3E-11 1.9E-08 Complement C3 P01024 0.15 0.04 4.1 5.1E-05 2.3E-02 Apolipoprotein A-II P02652 −0.15 0.04 −4.1 4.7E-05 2.1E-02 C-reactive protein P02741 0.31 0.04 8.5 1.3E-16 5.6E-14 Serum amyloid P-component P02743 0.34 0.04 9.0 1.7E-18 7.6E-16 Transthyretin P02766 −0.19 0.04 −4.8 2.2E-06 9.5E-04 Vitamin D-binding protein P02774 −0.15 0.04 −4.0 7.8E-05 3.4E-02 C4b-binding protein alpha chain P04003 0.23 0.04 6.2 1.2E-09 5.4E-07 Apolipoprotein B-100 P04114 0.21 0.04 5.7 2.0E-08 8.6E-06 Sex hormone-binding globulin P04278 −0.33 0.05 −7.2 1.5E-12 6.6E-10 Pancreatic alpha-amylase P04746 −0.20 0.04 −5.6 3.4E-08 1.5E-05 Insulin-like growth factor I P05019 −0.15 0.04 −4.0 6.0E-05 2.6E-02 Fructose-bisphosphate aldolase B P05062 0.14 0.04 4.0 7.2E-05 3.2E-02 Apolipoprotein D P05090 −0.20 0.04 −5.3 1.3E-07 5.8E-05 Protein S100-A8 P05109 0.21 0.04 5.8 9.3E-09 4.1E-06 Complement factor I P05156 0.23 0.04 6.5 2.2E-10 9.8E-08 Cholinesterase P06276 0.21 0.04 5.0 9.0E-07 3.9E-04 Gelsolin P06396 −0.18 0.04 −4.5 6.9E-06 3.0E-03 Protein S100-A9 P06702 0.16 0.04 4.6 4.6E-06 2.0E-03 Apolipoprotein A-IV P06727 −0.17 0.04 −4.3 1.8E-05 8.0E-03 Complement component C8 gamma chain P07360 0.17 0.04 4.4 1.0E-05 4.6E-03 Corticosteroid-binding globulin P08185 −0.24 0.04 −5.9 6.2E-09 2.7E-06 4F2 cell-surface antigen heavy chain P08195 −0.16 0.04 −4.2 3.5E-05 1.5E-02 72 kDa type IV collagenase P08253 −0.20 0.04 −5.5 6.5E-08 2.8E-05 Complement factor H P08603 0.33 0.04 9.1 9.0E-19 4.0E-16 Complement C4-A P0C0L4 0.23 0.04 6.3 6.5E-10 2.9E-07 Serum amyloid A-1 protein P0DJI8 0.18 0.04 4.8 1.8E-06 8.0E-04 Serum amyloid A-2 protein P0DJI9 0.18 0.04 4.8 2.2E-06 9.7E-04 Complement component C7 P10643 −0.28 0.04 −7.5 2.3E-13 1.0E-10 Mast/stem cell growth factor receptor Kit P10721 −0.18 0.04 −4.8 1.6E-06 7.1E-04 Neural cell adhesion molecule 1 P13591 −0.15 0.04 −3.9 9.7E-05 4.3E-02 Bone marrow proteoglycan P13727 −0.15 0.04 −4.1 4.0E-05 1.8E-02 Carboxypeptidase N catalytic chain P15169 −0.16 0.04 −3.9 8.7E-05 3.8E-02 Metalloproteinase inhibitor 2 P16035 −0.14 0.04 −3.9 1.0E-04 4.6E-02 F umarylacetoacetase P16930 0.20 0.03 5.8 1.4E-08 6.1E-06 Endoglin P17813 −0.18 0.04 −5.0 9.4E-07 4.1E-04 Insulin-like growth factor-binding protein 2 P18065 −0.18 0.04 −4.9 1.5E-06 6.5E-04 Vascular cell adhesion protein 1 P19320 −0.19 0.04 −5.0 7.5E-07 3.3E-04 Inter-alpha-trypsin inhibitor heavy chain H2 P19823 −0.20 0.04 −5.6 2.8E-08 1.2E-05 Inter-alpha-trypsin inhibitor heavy chain HI P19827 −0.18 0.04 −4.8 2.1E-06 9.2E-04 C4b-binding protein beta chain P20851 0.18 0.04 4.7 3.9E-06 1.7E-03 Fibulin-1 P23142 −0.18 0.04 −4.9 1.2E-06 5.3E-04 Properdin P27918 0.20 0.04 5.3 1.7E-07 7.3E-05 Transketolase P29401 0.16 0.04 4.2 2.8E-05 1.2E-02 Kallistatin P29622 −0.19 0.04 −5.2 2.7E-07 1.2E-04 Serum amyloid A-4 protein P35542 0.19 0.04 5.2 2.2E-07 9.9E-05 Voltage-dependent calcium channel subunit alpha-2/delta-1 P54289 −0.16 0.04 −4.4 1.6E-05 6.8E-03 Cadherin-13 P55290 −0.17 0.04 −4.4 1.1E-05 4.6E-03 Basement membrane-specific heparan sulfate proteoglycan core protein P98160 −0.16 0.04 −4.3 2.2E-05 9.8E-03 Complement factor H-related protein 1 Q03591 0.15 0.04 4.1 4.0E-05 1.7E-02 Prolow-density lipoprotein receptor-related protein 1 Q07954 −0.15 0.04 −4.0 6.6E-05 2.9E-02 Contactin-1 Q12860 −0.20 0.04 −5.4 8.5E-08 3.7E-05 Secreted phosphoprotein 24 Q13103 −0.30 0.04 −8.1 3.4E-15 1.5E-12 Limbic system-associated membrane protein Q13449 −0.19 0.04 −5.2 3.0E-07 1.3E-04 SPARC-like protein 1 Q14515 −0.15 0.04 −4.2 3.6E-05 1.6E-02 Serum paraoxonase/lactonase 3 Q15166 −0.16 0.04 −4.3 2.2E-05 9.4E-03 Transforming growth factor-beta-induced proteinig-h3 Q15582 −0.17 0.04 −4.6 6.1E-06 2.7E-03 Adiponectin Q15848 −0.15 0.04 −4.1 4.7E-05 2.1E-02 ADAMTS-like protein 4 Q6UY14 −0.19 0.04 −5.2 2.7E-07 1.2E-04 Gamma-glutamyl hydrolase Q92820 0.15 0.04 4.2 3.1E-05 1.4E-02 Proteoglycan 4 Q92954 0.46 0.03 13.6 4.7E-37 2.1E-34 Interleukin-1 receptor accessory protein Q9NPH3 −0.20 0.04 −5.4 1.1E-07 4.7E-05 Protein estimates whose abundance were significantly associated with body mass index are shown if Bonferroni-corrected coefficient null-test p-values were less than 0.05. Mixed-effects models included as covariates: age at blood sampling, age squared, sex, and interactions of sex with age and age squared. Random effects included the zygosity indicator variable and family identifiers. Body mass index was z-scored; fixed effects estimates indicate effect on 1 sd change. se: standard error. Table 3: Mixed model-derived estimates of proteins significantly associated with the changes (slope) of body mass index over 10 years of follow-up in FinnTwin12 participants Protein Fixed effect coefficients P-value description UniProt ID estimate se t-value raw bonferroni Complement factor B P00751 0.16 0.04 4.3 1.9E-05 8.1E-03 C-reactive protein P02741 0.16 0.04 4.2 2.7E-05 1.2E-02 Serum amyloid P-component P02743 0.18 0.04 4.6 6.4E-06 2.8E-03 C4b-binding protein alpha chain P04003 0.18 0.04 4.8 1.7E-06 7.6E-04 Phosphatidylcholine-sterol acyltransferase P04180 0.16 0.04 4.3 2.3E-05 9.9E-03 Sex hormone-binding globulin P04278 −0.24 0.05 −5.1 4.5E-07 2.0E-04 Complement factor I P05156 0.16 0.04 4.5 1.0E-05 4.4E-03 Cholinesterase P06276 0.24 0.04 5.6 3.0E-08 1.3E-05 Complement factor H P08603 0.25 0.04 6.6 7.7E-11 3.4E-08 Complement component C7 P10643 −0.20 0.04 −5.2 2.4E-07 1.1E-04 Inter-alpha-trypsin inhibitor heavy chain H2 P19823 −0.15 0.04 −4.0 8.1E-05 3.6E-02 Properdin P27918 0.15 0.04 4.0 7.1E-05 3.1E-02 Secreted phosphoprotein 24 Q13103 −0.17 0.04 −4.3 2.2E-05 9.9E-03 Proteoglycan 4 Q92954 0.27 0.04 7.3 9.8E-13 4.3E-10 Protein estimates whose abundance were significantly associated with changes in body mass index (i.e., slope of body mass index) are shown if Bonferroni-corrected coefficient null-test p-values were less than 0.05. Mixed-effects models included as covariates: the intercept of body mass index, age at blood sampling, age squared, sex, and interactions of sex with age and age squared. Random effects included the zygosity indicator variable and family identifiers. The slope of body mass index was z-scored; fixed effects estimates indicate effect on 1 sd change. se: standard error. Competing interests The authors declare that they have no competing interests. Ethics approval and consent to participate In FinnTwin12, ethical approval for all data collection waves was obtained from the ethical committee of the Helsinki and Uusimaa University Hospital District and the Institutional Review Board of Indiana University. All data collection and sampling protocols were performed in compliance with the ethical guidelines. Parents provided consent for the twins aged 12 and 14 years old, while twins aged 17 and 22 years old provided written consent themselves for sample collection. In NTR, informed consent was obtained from all participants. Projects were approved by the Central Ethics Committee on Research Involving Human Subjects of the VU University Medical Centre, Amsterdam, an Institutional Review Board certified by the U.S. Office of Human Research Protections (IRB number IRB00002991 under Federal-wide Assurance- FWA00017598; IRB/institute code, NTR 03-180). Consent for publication Not applicable. ==== Refs References 1. Blüher M. Obesity: global epidemiology and pathogenesis. Nat Rev Endocrinol. 2019;15 (5 ):288–298.30814686 2. World Obesity Federation, World Obesity Atlas 2023. https://data.worldobesity.org/publications/?cat=19 (2023). Accessed February 2023. 3. Kivimäki M , Strandberg T , Pentti J , Body-mass index and risk of obesity-related complex multimorbidity: an observational multicohort study. Lancet Diabetes Endocrinol. 2022;10 (4 ):253–263.35248171 4. Huang J , Huffman JE , Huang Y , Genomics and phenomics of body mass index reveals a complex disease network. Nat Commun. 2022;13 (1 ):7973.36581621 5. Paczkowska-Abdulsalam M , Kretowski A . Obesity, metabolic health and omics: Current status and future directions. World J Diabetes. 2021;12 (4 ):420–436.33889288 6. Silventoinen K , Jelenkovic A , Sund R , Differences in genetic and environmental variation in adult BMI by sex, age, time period, and region: an individual-based pooled analysis of 40 twin cohorts. Am J Clin Nutr. 2017;106 (2 ):457–466.28679550 7. Klok MD , Jakobsdottir S , Drent ML . The role of leptin and ghrelin in the regulation of food intake and body weight in humans: a review. Obes Rev. 2007;8 (1 ):21–34.17212793 8. Atalayer D , Gibson C , Konopacka A , Geliebter A . Ghrelin and eating disorders. Prog Neuropsychopharmacol Biol Psychiatry. 2013;40 :70–82.22960103 9. Masood A , Benabdelkamel H , Alfadda AA . Obesity Proteomics: An Update on the Strategies and Tools Employed in the Study of Human Obesity. High Throughput. 2018;7 (3 ):27.30213114 10. Bao X , Xu B , Yin S , Proteomic Profiles of Body Mass Index and Waist-to-Hip Ratio and Their Role in Incidence of Diabetes. J Clin Endocrinol Metab. 2022;107 (7 ):e2982–e2990.35294966 11. Goudswaard LJ , Bell JA , Hughes DA , Effects of adiposity on the human plasma proteome: observational and Mendelian randomisation estimates. Int J Obes (Lond). 2021;45 (10 ):2221–2229.34226637 12. Zaghlool SB , Sharma S , Molnar M , Revealing the role of the human blood plasma proteome in obesity using genetic drivers. Nat Commun. 2021;12 (1 ):1279.33627659 13. Kreitmaier P , Katsoula G , Zeggini E . Insights from multi-omics integration in complex disease primary tissues. Trends Genet. 2023;39 (1 ):46–58.36137835 14. Hasin Y , Seldin M , Lusis A . Multi-omics approaches to disease. Genome Biol. 2017;18 (1 ):83.28476144 15. Zhang Q , Meng XH , Qiu C , Integrative analysis of multi-omics data to detect the underlying molecular mechanisms for obesity in vivo in humans. Hum Genomics. 2022;16 (1 ):15.35568907 16. Wahl S , Vogt S , Stückler F , Multi-omic signature of body weight change: results from a population-based cohort study. BMC Med. 2015;13 :48.25857605 17. Pietiläinen KH , Sysi-Aho M , Rissanen A , Acquired obesity is associated with changes in the serum lipidomic profile independent of genetic effects--a monozygotic twin study. PLoS One. 2007;2 (2 ):e218.17299598 18. Watanabe K , Wilmanski T , Diener C , Multiomic signatures of body mass index identify heterogeneous health phenotypes and responses to a lifestyle intervention [published online ahead of print, 2023 Mar 20]. Nat Med. 2023;10.1038/s41591-023-02248-0. 19. Piening BD , Zhou W , Contrepois K , Integrative Personal Omics Profiles during Periods of Weight Gain and Loss. Cell Syst. 2018;6 (2 ):157–170.e8.29361466 20. Lapatto HAK , Kuusela M , Heikkinen A , Nicotinamide riboside improves muscle mitochondrial biogenesis, satellite cell differentiation, and gut microbiota in a twin study. Sci Adv. 2023;9 (2 ):eadd5163.36638183 21. Gallego-Paüls M , Hernández-Ferrer C , Bustamante M , Variability of multi-omics profiles in a population-based child cohort. BMC Med. 2021;19 (1 ):166.34289836 22. Simmonds M , Llewellyn A , Owen CG , Woolacott N . Predicting adult obesity from childhood obesity: a systematic review and meta-analysis. Obes Rev. 2016;17 (2 ):95–107.26696565 23. Figarska SM , Rigdon J , Ganna A , Proteomic profiles before and during weight loss: Results from randomized trial of dietary intervention. Sci Rep. 2020;10 (1 ):7913.32404980 24. Lind L , Figarska S , Sundström J , Fall T , Ärnlöv J , Ingelsson E . Changes in Proteomic Profiles are Related to Changes in BMI and Fat Distribution During 10 Years of Aging. Obesity (Silver Spring). 2020;28 (1 ):178–186.31804015 25. Geyer PE , Wewer Albrechtsen NJ , Tyanova S , Proteomics reveals the effects of sustained weight loss on the human plasma proteome. Mol Syst Biol. 2016;12 (12 ):901.28007936 26. Carayol J , Chabert C , Di Cara A , Protein quantitative trait locus study in obesity during weight-loss identifies a leptin regulator. Nat Commun. 2017;8 (1 ):2084.29234017 27. Vijay A , Valdes AM . The Metabolomic Signatures of Weight Change. Metabolites. 2019;9 (4 ):67.30987392 28. Wahl S , Yu Z , Kleber M , Childhood obesity is associated with changes in the serum metabolite profile. Obes Facts. 2012;5 (5 ):660–670.23108202 29. Freedman DS , Wang J , Maynard LM , Relation of BMI to fat and fat-free mass among children and adolescents. Int J Obes (Lond). 2005;29 (1 ):1–8. doi:10.1038/sj.ijo.0802735 15278104 30. Kaprio J. Twin studies in Finland 2006. Twin Res Hum Genet. 2006; 9 (6 ):772–7.17254406 31. Rose RJ , Salvatore JE , Aaltonen S , Barr PB , Bogl LH , Byers HA , FinnTwin12 Cohort: An Updated Review. Twin Res Hum Genet. 2019;22 (5 ):302–11.31640839 32. Ligthart L , van Beijsterveldt CEM , Kevenaar ST , The Netherlands Twin Register: Longitudinal Research Based on Twin and Twin-Family Designs. Twin Res Hum Genet. 2019;22 (6 ): 623–636.31666148 33. Willemsen G , de Geus EJ , Bartels M , The Netherlands Twin Register biobank: a resource for genetic epidemiological studies. Twin Res Hum Genet. 2010;13 (3 ):231–245.20477721 34. Afonin Aleksei , Piironen Aino-Kaisa , de Sousa Maciel Izaque , Proteomic insights into mental health status: plasma markers in young adults. bioRxiv 2023.06.07.544039; doi:10.1101/2023.06.07.544039 35. Callister SJ , Barry RC , Adkins JN , Normalization approaches for removing systematic biases associated with mass spectrometry and label-free proteomics. J Proteome Res. 2006;5 (2 ):277–286.16457593 36. Leek JT , Johnson WE , Parker HS , Jaffe AE , Storey JD . The sva package for removing batch effects and other unwanted variation in high-throughput experiments. Bioinformatics. 2012;28 (6 ):882–883.22257669 37. Jansen R , Batista S , Brooks AI , Sex differences in the human peripheral blood transcriptome. BMC Genomics. 2014;15 :33.24438232 38. Wright FA , Sullivan PF , Brooks AI , Heritability and genomics of gene expression in peripheral blood. Nat Genet. 2014;46 (5 ):430–437.24728292 39. Vink JM , Jansen R , Brooks A , Differential gene expression patterns between smokers and non-smokers: cause or consequence?. Addict Biol. 2017;22 (2 ):550–560.26594007 40. Würtz P , Kangas AJ , Soininen P , Lawlor DA , Davey Smith G , Ala-Korpela M . Quantitative Serum Nuclear Magnetic Resonance Metabolomics in Large-Scale Epidemiology: A Primer on -Omic Technologies. Am J Epidemiol. 2017;186 (9 ):1084–1096.29106475 41. Soininen P , Kangas AJ , Würtz P , Suna T , Ala-Korpela M . Quantitative serum nuclear magnetic resonance metabolomics in cardiovascular epidemiology and genetics. Circ Cardiovasc Genet. 2015;8 (1 ):192–206.25691689 42. Whipp AM , Heinonen-Guzejev M , Pietiläinen KH , van Kamp I , Kaprio J . Branched-chain amino acids linked to depression in young adults. Front Neurosci. 2022;16 :935858.36248643 43. Friedewald WT , Levy RI , Fredrickson DS . Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem. 1972;18 (6 ):499–502.4337382 44. Hagenbeek FA , Pool R , van Dongen J , Heritability estimates for 361 blood metabolites across 40 genome-wide association studies. Nat Commun. 2020;11 (1 ):39.31911595 45. Kujala UM , Palviainen T , Pesonen P , Polygenic Risk Scores and Physical Activity. Med Sci Sports Exerc. 2020;52 (7 ):1518–1524.32049886 46. Yengo L , Sidorenko J , Kemper KE , Meta-analysis of genome-wide association studies for height and body mass index in ~700000 individuals of European ancestry. Hum Mol Genet. 2018;27 (20 ):3641–3649.30124842 47. Pulit SL , Stoneman C , Morris AP , Meta-analysis of genome-wide association studies for body fat distribution in 694 649 individuals of European ancestry. Hum Mol Genet. 2019;28 (1 ):166–174.30239722 48. Nikpay M , Goel A , Won HH , A comprehensive 1,000 Genomes-based genome-wide association meta-analysis of coronary artery disease. Nat Genet. 2015;47 (10 ):1121–1130.26343387 49. Price AL , Patterson NJ , Plenge RM , Weinblatt ME , Shadick NA , Reich D . Principal components analysis corrects for stratification in genome-wide association studies. Nat Genet. 2006;38 (8 ):904–909.16862161 50. Vilhjálmsson BJ , Yang J , Finucane HK , Modeling Linkage Disequilibrium Increases Accuracy of Polygenic Risk Scores. Am J Hum Genet. 2015;97 (4 ):576–592.26430803 51. Rosseel Y. lavaan: An R Package for Structural Equation Modeling. Journal of Statistical Software. 2012; 48 (2 ), 1–36. 52. Hu L , & Bentler PM . Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling. 1999; 6 , 1–55. 53. Xia Y , Yang Y . RMSEA, CFI, and TLI in structural equation modeling with ordered categorical data: The story they tell depends on the estimation methods. Behavior research methods. 2019; 51 (1 ): 409–428.29869222 54. Peterson CM , Su H , Thomas DM , Tri-Ponderal Mass Index vs Body Mass Index in Estimating Body Fat During Adolescence. JAMA Pediatr. 2017;171 (7 ):629–636.28505241 55. Gul Siraz U , Hatipoglu N , Mazicioglu MM , Ozturk A , Cicek B , Kurtoglu S . Triponderal mass index is as strong as body mass index in the determination of obesity and adiposity. Nutrition. 2023;105 :111846.36265325 56. Bramante CT , Palzer EF , Rudser KD , BMI metrics and their association with adiposity, cardiometabolic risk factors, and biomarkers in children and adolescents. Int J Obes (Lond). 2022;46 (2 ):359–365.34718333 57. Kuznetsova A , Brockhoff PB , Christensen RHB . lmerTest Package: Tests in Linear Mixed Effects Models. Journal of Statistical Software. 2017; 82 (13 ): 1–26. 58. Posthuma D , Beem AL , de Geus EJ , Theory and practice in quantitative genetics. Twin Res. 2003;6 (5 ):361–376.14624720 59. Boomsma D , Busjahn A , Peltonen L . Classical twin studies and beyond. Nat Rev Genet. 2002;3 (11 ):872–882.12415317 60. Rijsdijk FV , Sham PC . Analytic approaches to twin data using structural equation models. Brief Bioinform. 2002;3 (2 ):119–133. doi:10.1093/bib/3.2.119 12139432 61. Hagenbeek FA , van Dongen J , Pool R , Heritability of Urinary Amines, Organic Acids, and Steroid Hormones in Children. Metabolites. 2022;12 (6 ):474.35736407 62. Neale MC , Hunter MD , Pritikin JN , OpenMx 2.0: Extended structural equation and statistical modeling. Psychometrika. 2016; 81 (2 ): 535–549.25622929 63. Pritikin JN , Hunter MD , Boker SM . “Modular open-source software for Item Factor Analysis.” Educational and Psychological Measurement. 2015; 75 (3 ): 458–474.27065479 64. Hunter MD (2018). State Space Modeling in an Open Source, Modular, Structural Equation Modeling Environment. Structural Equation Modeling. 2018; 25 (2 ): 307–324. 65. Boker SM , Neale MC , Maes HH , OpenMx 2.21.1 User Guide. 2023. 66. Verhulst B , Prom-Wormley E , Keller M , Medland S , Neale MC . Type I Error Rates and Parameter Bias in Multivariate Behavioral Genetic Models. Behav Genet. 2019;49 (1 ):99–111. doi:10.1007/s10519-018-9942-y 30569348 67. Castro-de-Araujo LFS , Singh M , Zhou Y , MR-DoC2: Bidirectional Causal Modeling with Instrumental Variables and Data from Relatives. Behav Genet. 2023;53 (1 ):63–73.36322200 68. Cominetti O , Núñez Galindo A , Corthésy J , Obesity shows preserved plasma proteome in large independent clinical cohorts. Sci Rep. 2018;8 (1 ):16981.30451909 69. Oller Moreno S , Cominetti O , Núñez Galindo A , The differential plasma proteome of obese and overweight individuals undergoing a nutritional weight loss and maintenance intervention. Proteomics Clin Appl. 2018;12 (1 ):10.1002/prca.201600150. 70. Mesinovic J , Jansons P , Zengin A , Exercise attenuates bone mineral density loss during diet-induced weight loss in adults with overweight and obesity: A systematic review and meta-analysis. J Sport Health Sci. 2021;10 (5 ):550–559.34004388 71. Spoel SH . Orchestrating the proteome with post-translational modifications. J Exp Bot. 2018;69 (19 ):4499–4503.30169870 72. Wu L , Candille SI , Choi Y , Variation and genetic control of protein abundance in humans. Nature. 2013;499 (7456 ):79–82.23676674 73. Riuzzi F , Chiappalupi S , Arcuri C , Giambanco I , Sorci G , Donato R . S100 proteins in obesity: liaisons dangereuses. Cell Mol Life Sci. 2020;77 (1 ):129–147.31363816 74. Catalán V , Gómez-Ambrosi J , Rodríguez A , Increased levels of calprotectin in obesity are related to macrophage content: impact on inflammation and effect of weight loss. Mol Med. 2011;17 (11–12 ):1157–1167.21738950 75. Sekimoto R , Kishida K , Nakatsuji H , Nakagawa T , Funahashi T , Shimomura I . High circulating levels of S100A8/A9 complex (calprotectin) in male Japanese with abdominal adiposity and dysregulated expression of S100A8 and S100A9 in adipose tissues of obese mice. Biochem Biophys Res Commun. 2012;419 (4 ):782–789.22390934 76. Grand A , Rochette E , Dutheil F , Body Mass Index and Calprotectin Blood Level Correlation in Healthy Children: An Individual Patient Data Meta-Analysis. J Clin Med. 2020;9 (3 ):857.32245056 77. Chen N , Miao L , Lin W , Integrated DNA Methylation and Gene Expression Analysis Identified S100A8 and S100A9 in the Pathogenesis of Obesity. Front Cardiovasc Med. 2021;8 :631650.34055926 78. Drouard G , Silventoinen K , Latvala A , Kaprio J . Genetic and environmental factors underlying parallel changes in body mass index and alcohol consumption: a 36-year longitudinal study of adult twins [published online ahead of print, 2023 Mar 7]. Obes Facts. 2023;10.1159/000529835. 79. Liu Y , Buil A , Collins BC , Quantitative variability of 342 plasma proteins in a human twin population. Mol Syst Biol. 2015;11 (1 ):786.25652787 80. Haj AK , Hasan H , Raife TJ . Heritability of Protein and Metabolite Biomarkers Associated with COVID-19 Severity: A Metabolomics and Proteomics Analysis. Biomolecules. 2022;13 (1 ):46.36671431 81. Behbodikhah J , Ahmed S , Elyasi A , Apolipoprotein B and Cardiovascular Disease: Biomarker and Potential Therapeutic Target. Metabolites. 2021;11 (10 ):690.34677405 82. Sniderman AD , Thanassoulis G , Glavinovic T , Apolipoprotein B Particles and Cardiovascular Disease: A Narrative Review. JAMA Cardiol. 2019;4 (12 ):1287–1295.31642874 83. Gottesman O , Drill E , Lotay V , Bottinger E , Peter I . Can genetic pleiotropy replicate common clinical constellations of cardiovascular disease and risk?. PLoS One. 2012;7 (9 ):e46419.23029515 84. Rankinen T , Sarzynski MA , Ghosh S , Bouchard C . Are there genetic paths common to obesity, cardiovascular disease outcomes, and cardiovascular risk factors?. Circ Res. 2015;116 (5 ):909–922.25722444 85. Auwerx C , Sadler MC , Woh T , Reymond A , Kutalik Z , Porcu E . Exploiting the mediating role of the metabolome to unravel transcript-to-phenotype associations. Elife. 2023;12 :e81097.36891970 86. Sadler MC , Auwerx C , Lepik K , Porcu E , Kutalik Z . Quantifying the role of transcript levels in mediating DNA methylation effects on complex traits and diseases. Nat Commun. 2022;13 (1 ):7559.36477627
PMC010xxxxxx/PMC10352120.txt
==== Front Cancer Manag Res Cancer Manag Res cmar Cancer Management and Research 1179-1322 Dove 430213 10.2147/CMAR.S430213 Retraction Long Noncoding RNA VPS9D1-AS1 Sequesters microRNA-525-5p to Promote the Oncogenicity of Colorectal Cancer Cells by Upregulating HMGA1 [Retraction] Liu et al Liu et al 13 7 2023 2023 13 7 2023 15 685686 12 7 2023 12 7 2023 © 2023 Dove Medical Press. 2023 Dove Medical Press. https://creativecommons.org/licenses/by-nc/3.0/ This work is published and licensed by Dove Medical Press Limited. The full terms of this license are available at https://www.dovepress.com/terms.php and incorporate the Creative Commons Attribution – Non Commercial (unported, v3.0) License (http://creativecommons.org/licenses/by-nc/3.0/). By accessing the work you hereby accept the Terms. Non-commercial uses of the work are permitted without any further permission from Dove Medical Press Limited, provided the work is properly attributed. For permission for commercial use of this work, please see paragraphs 4.2 and 5 of our Terms (https://www.dovepress.com/terms.php). ==== Body pmcLiu H, Zhang X, Jin X, et al. Cancer Manag Res. 2020;12:9915–9928. We, the Editor and Publisher of Cancer Management and Therapy are retracting the published article. Since publication, concerns have been raised about the duplication of images in this article with those from other unrelated articles. Specifically, The image from Figure 1I, HCT116, si-NC, has been duplicated with the image from Figure 1H, Saos-2, si-NC, from He J, Guan J, Liao S, et al. Long Noncoding RNA CCDC144NL-AS1 Promotes the Oncogenicity of Osteosarcoma by Acting as a Molecular Sponge for microRNA-490-3p and Thereby Increasing HMGA2 Expression. Onco Targets Ther. 2021;14:1–13. https://doi.org/10.2147/OTT.S280912. When approached for an explanation, the authors have been unable to address the concerns raised and have not been able to provide sufficient original data from their study. As verifying the validity of published work is core to the integrity of the scholarly record, we are therefore retracting the article. The authors listed in this publication have been informed. We have been informed in our decision-making by our editorial policies and the COPE guidelines. The retracted article will remain online to maintain the scholarly record, but it will be digitally watermarked on each page as “Retracted”.
PMC010xxxxxx/PMC10352121.txt
==== Front Onco Targets Ther Onco Targets Ther ott OncoTargets and Therapy 1178-6930 Dove 430214 10.2147/OTT.S430214 Retraction Long Noncoding RNA ST7-AS1 Upregulates TRPM7 Expression by Sponging microRNA-543 to Promote Cervical Cancer Progression [Retraction] Qi et al Qi et al 13 7 2023 2023 13 7 2023 16 581582 12 7 2023 12 7 2023 © 2023 Dove Medical Press. 2023 Dove Medical Press. https://creativecommons.org/licenses/by-nc/3.0/ This work is published and licensed by Dove Medical Press Limited. The full terms of this license are available at https://www.dovepress.com/terms.php and incorporate the Creative Commons Attribution – Non Commercial (unported, v3.0) License (http://creativecommons.org/licenses/by-nc/3.0/). By accessing the work you hereby accept the Terms. Non-commercial uses of the work are permitted without any further permission from Dove Medical Press Limited, provided the work is properly attributed. For permission for commercial use of this work, please see paragraphs 4.2 and 5 of our Terms (https://www.dovepress.com/terms.php). ==== Body pmcQi H, Lu L, Wang L. Onco Targets Ther. 2020;13:7257–7269. We, the Editor and Publisher of OncoTargets and Therapy are retracting the published article. Since publication, concerns have been raised about the duplication of images in this article with those from other unrelated articles. Specifically, The image from Figure 4E, SiHa, si-NC, has been duplicated with the image from Figure 4E, T98, siNC, from Yang H, Song Z, Wu X, Wu Y, Liu C. MicroRNA-652 suppresses malignant phenotypes in glioblastoma multiforme via FOXK1-mediated AKT/mTOR signaling pathway. Onco Targets Ther. 2019;12:5563–5575. https://doi.org/10.2147/OTT.S204715 (RETRACTED). The image from Figure 5C, SiHa, si-ST7-AS1+miR-543 inhibitor, has been duplicated with the image from Figure 6C, MGC-803, si-NC, from Wang X, Chen X, Tian Y, Jiang D, Song Y. Long Noncoding RNA RGMB-AS1 Acts as a microRNA-574 Sponge Thereby Enhancing the Aggressiveness of Gastric Cancer via HDAC4 Upregulation. Onco Targets Ther. 2020;13:1691–1704. https://doi.org/10.2147/OTT.S234144 (RETRACTED). The image from Figure 5D, C-33A, si-ST7-AS1+miR-543 inhibitor, has been duplicated with the image from Figure 4F, U2OS, si-NC, from Zhao X, Li J, Yu D. MicroRNA-939-5p directly targets IGF-1R to inhibit the aggressive phenotypes of osteosarcoma through deactivating the PI3K/Akt pathway. Int J Mol Med. 2019;44:1833–1843. https://doi.org/10.3892/ijmm.2019.4333 (RETRACTED). When approached for an explanation, the authors have been unable to address the concerns raised and have not been able to provide sufficient original data from their study. As verifying the validity of published work is core to the integrity of the scholarly record, we are therefore retracting the article. The authors listed in this publication have been informed. We have been informed in our decision-making by our editorial policies and the COPE guidelines. The retracted article will remain online to maintain the scholarly record, but it will be digitally watermarked on each page as “Retracted”.
PMC010xxxxxx/PMC10352122.txt
==== Front Int J Womens Health Int J Womens Health ijwh International Journal of Women's Health 1179-1411 Dove 413670 10.2147/IJWH.S413670 Original Research A Retrospective Cohort Study on the Prevalence, Risk Factors, and Improvement of Overactive Bladder Symptoms in Women with Pelvic Organ Prolapse Aimjirakul et al Aimjirakul et al Aimjirakul Komkrit 1 Ng Jun Jiet 1 Saraluck Apisith 1 Wattanayingcharoenchai Rujira 1 Mangmeesri Peeranuch 1 http://orcid.org/0000-0001-5302-4370 Manonai Jittima 1 1 Department of Obstetrics & Gynaecology, Faculty of Medicine Ramathibodi Hospital, Mahidol University, Bangkok, Thailand Correspondence: Jun Jiet Ng; Jittima Manonai, Department of Obstetrics & Gynaecology, Faculty of Medicine Ramathibodi Hospital, Mahidol University, Bangkok, 10400, Thailand, Tel +66-2-2012167, Fax +66-2-2011416, Email [email protected]; [email protected] 13 7 2023 2023 15 10391046 12 4 2023 29 6 2023 © 2023 Aimjirakul et al. 2023 Aimjirakul et al. https://creativecommons.org/licenses/by-nc/3.0/ This work is published and licensed by Dove Medical Press Limited. The full terms of this license are available at https://www.dovepress.com/terms.php and incorporate the Creative Commons Attribution – Non Commercial (unported, v3.0) License (http://creativecommons.org/licenses/by-nc/3.0/). By accessing the work you hereby accept the Terms. Non-commercial uses of the work are permitted without any further permission from Dove Medical Press Limited, provided the work is properly attributed. For permission for commercial use of this work, please see paragraphs 4.2 and 5 of our Terms (https://www.dovepress.com/terms.php). Background Overactive bladder (OAB) symptoms are common in women with pelvic organ prolapse (POP), but the explanation is unclear. It is also uncertain whether OAB symptoms improve or persist after POP reduction. This study aimed to determine the prevalence and risk factors for OAB symptoms in women with POP, and to compare the improvement of OAB symptoms among women in three treatment groups: pelvic floor exercise, pessary, and surgery. Methods This retrospective cohort study included patients who visited our urogynecology clinic from January 2016 to December 2020. The Pelvic Floor Bother Questionnaire was used to evaluate selected pelvic floor symptoms (OAB and POP). Demographic characteristics and clinical findings, including Pelvic Organ Prolapse Quantification System and number of prolapsed compartments, were analyzed. Univariate and multivariate analyses were conducted to identify risk factors for OAB symptoms in women with POP. Subgroup analyses were performed in 533 patients to evaluate the improvement of OAB symptoms following POP treatment. Results A total of 754 patients were analyzed. The incidence of OAB symptoms was 70% (533/754) and two-thirds (65%) reported moderate to severe bother. The lowest points of the anterior wall (OR 0.60; 95% CI 0.41–0.87; p = 0.01), longer perineal body (OR 0.78; 95% CI 0.21–0.76; p = 0.02), and previous vaginal delivery (OR 2.10; 95% CI 1.14–3.89; p = 0.02) were identified as significant risk factors. In the subgroup analyses, improvement in OAB symptoms was observed in 36.6% (195/533) of women who underwent POP treatment. Compared with pelvic floor exercise, pessary (OR 1.40; 95% CI 0.94–2.07; p = 0.10) and surgery (OR 1.30; 95% CI 0.80–2.12; p = 0.28) had higher odd ratios but the effects were not significant. Conclusion The prevalence of OAB symptoms in women with POP was high at 70%. Improvement in OAB symptoms was observed in one-third of women who underwent POP treatment. However, there were no significant differences between the treatment methods. Keywords pelvic organ prolapse overactive bladder pelvic floor bother questionnaire prevalence risk factors pessary surgery funding There is no funding to report. ==== Body pmcIntroduction The female pelvic floor can be divided into three compartments: anterior, apical, and posterior.1 Pelvic floor disorders include pelvic organ prolapse (POP), urinary incontinence, voiding dysfunction, anal incontinence, defecatory dysfunction, pelvic pain syndrome, and sexual dysfunction.2 These conditions are usually related to one another and share common risk factors, such as aging, pregnancy, vaginal delivery, and heavy lifting.3 POP, referring to descent of the vaginal wall, uterus, or post-hysterectomy vaginal vault,4 is an extremely prevalent problem that occurs in approximately 50% of parous women.5 Overactive bladder (OAB) is a condition characterized by urinary urgency, typically with urinary frequency and nocturia, with or without urgency urinary incontinence (UUI).6 According to the Epidemiology Urinary Incontinence and Comorbidities (EPIC) study, OAB affects 13% of women regardless of age.7 However, the pathophysiology of OAB is not fully understood. POP and OAB symptoms often coexist. In a recent population-based study, 82% of women with POP reported at least one OAB symptom of any degree and 40% had at least one bothersome OAB symptom.8 However, evidence for correlations between specific anatomical defects and symptom severity is conflicting. Among many theories related to OAB, one theory that can explain the correlation of OAB symptoms with POP is the Integral Theory proposed by Petros and Ulstem.9 The Integral Theory views OAB pathogenesis as anatomical, resulting from defects and damage to the pelvic floor musculo-ligamentous system. Moreover, laxity of the suspensory ligaments inactivates the vector forces that contract against them.9 Observed improvement of OAB symptoms after repair of loose cardinal and uterosacral ligaments proved that the origins of OAB symptoms are not only within the bladder.8 Stretching of the bladder wall is likely to occur in vaginal prolapse, and this may trigger stretch receptors and result in detrusor contractions.10 Other theories such as bladder outflow obstruction causing bladder distention and release of detrusor stimulants, traction and kinking of the urethra, denervation of the bladder wall, detrusor muscle changes, and changes of the spinal micturition reflex have also been postulated.11 However, data for the relationships between stages of prolapse and OAB symptoms are sparse and conflicting. Bladder outlet obstruction from advanced anterior wall prolapse might be the most possible pathophysiology of OAB symptoms,11 though urgency and UUI were more frequently observed in women with less advanced prolapse overall.12,13 Treatment of POP can be divided into pelvic floor exercise, pessary, and surgery. Few studies have compared the improvement of OAB symptoms in women with POP after different modalities of treatment. Several studies attempted to predict the effects of POP treatment on OAB symptoms, but the results were inconsistent. While the majority of the studies indicated improvement in bothersome OAB symptoms after POP treatment,7,13–15 Johnson et al found that 40% of patients had persistent OAB symptoms and 6% of patients had de novo OAB symptoms after POP treatment.16 Thus, it remains unclear which patients will experience resolution or persistence of OAB symptoms after treatment to correct POP, and this may cause frustration and disappointment for patients. In this study, the prevalence and correlation between OAB symptoms and POP were reviewed and risk factors for OAB symptoms in women with POP were identified. The efficacies of the different treatment modalities for the improvement of OAB symptoms in women with POP were also compared. Materials and Methods Study Setting and Population This retrospective cohort study was conducted in Ramathibodi Hospital, Bangkok, Thailand. The inclusion criteria were consecutive female patients who visited the urogynecology clinic with POP symptoms from January 2016 to December 2020. The exclusion criteria were OAB treatment, pelvic reconstructive or anti-incontinence surgery, urinary tract infection, or neurological diseases. Patients with missing data, which were variables studied and 6-month follow-up, in their medical records were also excluded. Data Collection The electronic medical records of the eligible patients were reviewed. The Pelvic Floor Bother Questionnaire (PFBQ) was used to identify POP and OAB symptoms in the patients at their first visit. Data for demographic characteristics, including age, body mass index (BMI), menopausal status, parity, history of vaginal delivery, previous hysterectomy, and previous pelvic floor surgery, were collected for all patients. Physical examination findings, including Pelvic Organ Prolapse Quantification (POP-Q) and pelvic floor muscle strength (Brink score), were recorded and analyzed. The PFBQ is a nine-item self-administered questionnaire developed by The Cleveland Clinic that has been validated as a tool to evaluate symptoms and bother related to prolapse, bladder, and bowel symptoms.17 Symptom severity was assessed using a Likert scale bother score. The PFBQ has been translated, validated, and used in our center because of its clarity, simplicity, and ease of self-filling.18 Women with POP were identified through a positive response to Question 6 of the PFBQ (Do you experience the feeling of a bulge in the vagina?). Different stages of POP in the different compartments (anterior, apical, and posterior) were classified according to the IUGA and ICS terminology and grouped into four categories (Stages I, II, III, and IV). The study had two groups: women with OAB symptoms and women without OAB symptoms. Presence of OAB symptoms was determined using the PFBQ and defined as a positive response to Question 3 (Do you experience an abnormal strong feeling of urgency to urinate?) or Question 4 (Do you experience urine leakage associated with the feeling of urgency?). A urine analysis was conducted in all patients to rule out urinary tract infection. The POP treatments provided were pelvic floor exercise, pessary, and surgery. Data at 3 and 6 months post-treatment visits were evaluated for the improvement of OAB symptoms. Improvement was defined as either complete resolution of urgency and/or frequency and/or urgency incontinence, or a decrease in the degree of symptom bother. Statistical Analysis Data entry and statistical analyses were performed using the Statistical Package for Social Sciences (SPSS) version 24. The chi-square test was used to compare risk factors between women with and without OAB symptoms. Student’s t-test and the Mann–Whitney U-test were used to assess the differences in continuous variables. A multivariate logistic regression analysis was conducted to estimate the associations between other variables. The adjusted odds ratio (OR) and 95% confidence interval (CI) were estimated for each correlation. Values of p <0.05 were considered to indicate statistical significance. Data were presented as mean ± standard deviation, median (minimum, maximum), or percentage depending on the variable. The required sample size was calculated using a sample size formula with the following assumptions: confidence interval of 95%, precision of 5% and power of 80%. These assumptions were based on the reported prevalence of 37% for OAB symptoms among women with POP,19 and gave a sample size of 754 women. Results Patient Characteristics The medical records of 840 patients were evaluated. After exclusion of 86 patients with missing data and loss to follow-up, the final study population contained 754 patients with a mean age of 65.77±9.45 years. No differences were found between the included and the excluded groups in terms of age and OAB symptoms. The sociodemographic characteristics and physical examination findings were analyzed. Among the total 754 patients, 95% were postmenopausal, the mean BMI was 25.23±3.47 kg/m2, 94% of women had a history of vaginal delivery, and 5.8% were nulliparous. Regarding pelvic compartments, 27.7% had a single prolapsed compartment, 27.7% had two prolapsed compartments, and 38.2% had three prolapsed compartments. The POP-Q examination findings revealed a mean perineal body of 2.5±0.72 cm and a mean genital hiatus of 4.11±0.99 cm. The prevalence of OAB symptoms in women with POP was 70.4% (533/754). According to the PFBQ, approximately two-thirds of the patients (65%) were moderately and severely bothered by their OAB symptoms, such as the need for sanitary pad wearing, toilet mapping, and reduced participation in social activities. Risk Factors for OAB Symptoms The univariate analyses of associations between OAB symptoms and variables of interest are shown in Table 1. Age (p = 0.89), parity (p = 0.08), BMI (p = 0.08), menopausal status (p = 0.22), previous hysterectomy (p = 0.09), and previous pelvic floor surgery (p = 0.52) were not risk factors for OAB symptoms in women with POP. In contrast, history of vaginal delivery, point Ba beyond the hymen, and perineal body were associated with OAB symptoms. On physical examination, the apical compartment; point C (p = 0.37), the posterior compartment; point Bp (p = 0.19), GH (p = 0.71), TVL (p = 0.06), number of prolapsed compartments (p = 0.14), and Brink score (p = 0.7) were not related to OAB symptoms.Table 1 Women with POP with or Without Overactive Bladder (OAB) Symptoms (N = 754) Variables OAB Symptoms (N=533) No OAB Symptoms (N=221) p-value Age (years), mean ± SD 66.80 ± 9.62 65.70 ± 9.05 0.89 BMI (kg/m2), mean ± SD 25.37 ± 3.52 24.88 ± 3.33 0.08 Parity, n (%) 0.08  Nulliparity 26 (4.88) 18 (8.14)  Multiparity 507 (95.12) 203 (91.86) History of vaginal delivery, n (%) 0.04  Yes 506 (94.93) 201 (90.95)  No 27 (5.07) 20 (9.05) Menopausal status, n (%) 0.22  Premenopausal 31 (5.82) 8 (3.62)  Postmenopausal 502 (94.18) 213 (96.38) Previous hysterectomy, n (%) 0.09  Yes 72 (13.51) 20 (9.05)  No 461 (86.49) 201 (90.95) Previous pelvic floor surgery, n (%)  Yes 9 (1.69) 2 (0.90) 0.52  No 524 (98.31) 219 (99.10) POP-Q Examination Aa beyond hymen, n (%) 0.63  Yes 284 (53.28) 122 (55.20)  No 249 (46.72) 99 (44.80) Ba beyond hymen, n (%) 0.02  Yes 372 (69.79) 173 (78.28)  No 161 (30.21) 48 (21.72) Ap beyond hymen, n (%) 0.06  Yes 366 (68.67) 167 (31.33)  No 167 (31.33) 54 (24.43) Bp beyond hymen, n (%) 0.192  Yes 303 (56.85) 137 (61.99)  No 230 (43.15) 84 (38.01) C beyond hymen, n (%) 0.37  Yes 263 (49.34) 117 (52.94)  No 270 (50.66) 104 (47.06) D beyond hymen, n (%) 0.75  Yes 146 (27.39) 58 (26.24)  No 387 (72.61) 163 (73.76) GH (cm), mean ± SD 4.11 ± 1.02 4.09 ± 0.91 0.71 PB (cm), mean ± SD 2.47 ± 0.70 2.60 ± 0.77 0.03 TVL (cm), mean ± SD 7.02 ± 0.92 7.16 ± 0.87 0.06 Brink score (range 2–12), mean ± SD 7.47 ± 2.38 7.55 ± 2.46 0.70 Number of prolapsed compartment (POP stage more than stage I) 0.14 -Single compartment 148 (27.77) 61 (27.60) -Two compartments 144 (27.02) 65 (29.41) -Three compartments 200 (37.52) 88 (39.82) Note: chi-square, t-test or Mann–Whitney U-test. Abbreviations: BMI, body mass index; POP-Q, pelvic organ prolapse quantification system; GH, genital hiatus; PB, perineal body; TVL, total vaginal length. The results for the stepwise multivariate regression analysis are shown in Table 2. The analysis revealed that less advanced anterior vaginal wall prolapse with point Ba above the hymen (OR 0.60; 95% CI 0.41–0.87; p = 0.01) and longer perineal body (OR 0.78; 95% CI 0.21–0.76; p = 0.02) were protective factors against OAB symptoms in women with POP. Meanwhile, history of vaginal delivery (OR 2.10; 95% CI 1.14–3.89; p = 0.02) significantly increased the risk of OAB symptoms in women with POP.Table 2 Multivariate Associations Between Risk Factors and OAB Symptoms in Women with POP Variables Adjusted OR (95% CI) p-value Point Ba above hymen 0.60 (0.41–0.87) 0.01 Longer perineal body 0.78 (0.63–0.96) 0.02 History of vaginal delivery 2.10 (1.14–3.89) 0.02 Abbreviations: OAB, overactive bladder; POP, pelvic organ prolapse. Improvement of OAB Symptoms After POP Treatment Regarding POP treatment, the most common type of pessary used was ring with support (88.32%) and the three most common surgical procedures were apical with anterior compartment repair (32.69%), apical with anterior with posterior compartment repair (29.81%), and colpocleisis (22.11%). Table 3 shows the results for the three POP treatment groups: pelvic floor exercise, pessary, and surgery. Overall improvement in OAB symptoms after 6 months of treatment was seen in 36.6% (195/533) of women who underwent treatment for POP, comprising 40.9% (88/215) for pelvic floor exercise, 33.1% (71/214) for pessary, and 36% (36/104) for surgery (Figure 1). Compared with pelvic floor exercise, pessary (OR 1.40; 95% CI 0.94–2.07; p = 0.10) and surgery (OR 1.30; 95% CI 0.80–2.12; p = 0.28) had higher odd ratios, but the effects were not significant.Table 3 Comparison of Different Treatments Groups in Improvement of OAB Symptoms in Women with POP Treatment Adjusted Odds Ratio (95% CI) p-value Pelvic floor muscle exercise 1 Pessary 1.40 (0.94–2.07) 0.10 Surgery 1.30 (0.80–2.12) 0.28 Abbreviations: OAB, overactive bladder; POP, pelvic organ prolapse. Figure 1 Improvement of overactive bladder (OAB) symptoms in women with organ prolapse (POP) symptoms after 6 months of treatment. Discussion OAB symptoms and POP are two different pelvic floor disorders, but commonly coexist in patients. The findings of the present study confirmed the high correlation between OAB symptoms and POP. Moreover, the anatomical and structural factors for POP, namely advanced anterior wall prolapse and shorter perineal body, were clearly associated with OAB symptoms. Therefore, understanding the relationship between POP and OAB symptoms is important in clinical practice. Women who report prolapse symptoms should be asked about their OAB symptoms to achieve appropriate management of their symptoms. In addition to symptoms assessment and evaluation of their impact on quality of life, a focused physical examination is also needed, with special attention paid to the genitourinary examination and prolapse severity. In doing so, a complete clinical decision plan, reliable preoperative counseling, and appropriate goals of treatment would be developed. OAB symptoms should be examined in women with POP and the degree of bother caused by these symptoms should be quantified using validated patient-reported outcome measures such as the PFBQ. Knowledge of the risk factors for OAB symptoms in women with POP can assist in the prevention and treatment of OAB symptoms in women with POP. The identification of history of vaginal delivery as a risk factor for OAB symptoms in women with POP can be explained by damage to the musculoligamentous system during passage of the fetus through the birth canal, leading to loss of bladder support, more distal position of point Ba, and shorter perineal body, which were significant findings in the present study. A more distal position of point Ba indicated a more significant anterior vaginal wall prolapse, which predisposed women with POP to OAB symptoms. Advanced anterior wall prolapse, which causes stretching of the bladder wall and prolonged outlet obstruction, could have an influence through an increased bladder sensation and enhancement of the spinal micturition reflex.10,11 Effective POP treatment can help relief of bladder outlet obstruction and bladder wall overstretching. Consequently, the reduction of urinary urgency, frequency, and UUI has been shown after treatment in our study and other studies.13–16 For women with POP, the various treatment options should be explained and appropriate preoperative counseling should be provided to avoid disappointment. The risk of de novo OAB symptoms and persistance or even worsening of OAB symptoms should be highlighted to patients prior to POP surgery.20 Also prior to surgery, all patients should be offered a pessary trial for POP reduction to gauge the dynamics of their OAB symptoms, because a pessary has the benefits of being minimally invasive, inexpensive, and associated with a low complication rate.21 The present findings demonstrated that pelvic floor exercise, pessary, and surgery were equally effective for treatment of OAB symptoms in women with POP, with an overall treatment efficacy of 36.6%. These findings are consistent with those in a study by Zacharakis et al,14 in which women with coexisting POP and OAB symptoms experienced a significant improvement in urgency and frequency symptoms after successful pessary fitting. Urgency, UUI, and Overactive Bladder Symptom Score were also significantly improved in patients with POP and OAB symptoms at 3 and 6 months after pelvic reconstruction surgery in previous studies.16,20 These phenomena may be explained by anatomical correction of the lower urinary tract obstruction by pessary use or surgery that provides effective relief of OAB symptoms in women with POP. However, the predictors for OAB symptom relief in women with POP remain unclear. Based on previous studies, predictors of persistent OAB symptoms following surgery for advanced POP include overweight status, preoperative UUI and detrusor overactivity, lower preoperative maximal flow rate, and postoperative complications.22–25 Preoperative counseling should include a discussion about persistent OAB symptoms following prolapse repair. Our study did not demonstrate the difference between treatment in alleviating OAB symptoms. This could be explained by a low sample size with low statistical power to detect a true effect of treatment, which was considered as the secondary outcome. Addition of other treatment modalities, such as antimuscarinic drugs, beta-3 agonists, and neuromodulation, may be helpful for women with persistent OAB symptoms despite correction of POP. In medicine, prevention is better than cure. Because history of vaginal delivery was identified as a risk factor for OAB symptoms in women with POP, it is important to prevent maternal birth trauma. Counseling for vaginal delivery has always focused on obstetric anal sphincter injury, but the risk of OAB symptoms after vaginal delivery should also be included. Other variables such as fetal birth weight, instrumental delivery, and episiotomy should be investigated. Levator avulsion should be evaluated using digital palpation and ultrasound.26 Further examination for these variables will provide more information on the risk factors for OAB symptoms in women with POP. The strength of the present study was that its conduction in Asian women to reflect our local condition, given that much of the existing literature was extrapolated from studies conducted in Caucasian populations. The study was carried out in a urogynecology clinic where women were asked about and were comfortable with disclosing their pelvic floor symptoms. This was reflected in the high prevalence of OAB symptoms among women with POP. The study also had several limitations that should be considered. First, it had a retrospective design. Second, data on OAB symptoms were collected by a self-reporting questionnaire without confirmation by urodynamic examinations or bladder diaries. Third, a specific scale for OAB or urgency severity scale were not used, thus, we could not quantify the impact and bother of the urgency symptoms in detail. Further prospective studies in which OAB symptoms are verified and quantified using an OAB-specific scale should be considered. Comparative studies on other treatment modalities for OAB symptoms in women with POP, such as antimuscarinic drugs, beta-3 agonists, and neuromodulation, should also be taken into consideration. Conclusions In the present study, the prevalence of OAB symptoms in women with POP was high at 70%, and approximately two-thirds of the patients reported moderate to severe symptoms that affected their quality of life. Advanced anterior wall prolapse, shorter perineal body, and previous vaginal delivery were identified as significant risk factors for OAB symptoms in women with POP. There was a one-third chance of OAB symptom improvement in women who received POP treatment with pelvic floor exercise, pessary, or surgery. This information can be useful for patient counseling and to guide patient expectations. Other treatment modalities should be considered in women whose OAB symptoms persist despite POP correction. Acknowledgments The authors would like to thank Ms. Sasiporn Sitthisorn for assistance with the statistical analyses. The authors also thank Alison Sherwin, PhD, from Edanz (https://www.edanz.com/ac) for editing a draft of this manuscript. Data Sharing Statement The data that support the findings of this study are available from the corresponding authors upon reasonable request. Ethical Considerations This study was approved by the Medical and Research Ethics Committee of Ramathibodi Hospital Number MURA2022/773 (Expedited). Since this project was a retrospective chart review in which the data already exist, the informed consent was not required. However, the patient data were maintained with confidentiality and compliance with the Declaration of Helsinki. Author Contributions All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work. Disclosure The authors declare no conflicts of interest. ==== Refs References 1. Delancey JOL. Anatomy. In: Cardozo L, Staskin D, editors. Textbook of Female Urology and Urogynecology. 4th ed. Boca Raton: CRC Press; 2016. doi:10.1201/9781315378206 2. Bump RC, Norton PA. Epidemiology and natural history of pelvic floor dysfunction. Obstet Gynecol Clin North Am. 1998;25 (4 ):723–746. doi:10.1016/s0889-8545(05)70039-5 9921553 3. National Institute for Health and Care Excellence (NICE), London. Pelvic floor dysfunction: prevention and non-surgical management. NICE Guideline, No. 210; 2021. 4. Haylen BT, Maher CF, Barber MD, et al. An International Urogynecological Association (IUGA) / International Continence Society (ICS) joint report on the terminology for female pelvic organ prolapse (POP). Neurourol Urodynam. 2016;35 (2 ):137–168. doi:10.1002/nau.22922 5. Digesu GA, Chaliha C, Salvatore S, Hutchings A, Khullar V. The relationship of vaginal prolapse severity to symptoms and quality of life. BJOG. 2005;112 (7 ):971–976. doi:10.1111/j.1471-0528.2005.00568.x 15958002 6. Haylen BT, de Ridder D, Freeman RM, et al. An International Urogynecological Association (IUGA)/International Continence Society (ICS) joint report on the terminology for female pelvic floor dysfunction. Neurourol Urodyn. 2010;29 (1 ):4–20. doi:10.1002/nau.20798 19941278 7. Irwin DE, Milsom I, Hunskaar S, et al. Population-based survey of urinary incontinence, overactive bladder, and other lower urinary tract symptoms in five countries: results of the EPIC study. Eur Urol. 2006;50 (6 ):1306–1315. doi:10.1016/j.eururo.2006.09.019 17049716 8. Karjalainen PK, Tolppanen AM, Mattsson NK, et al. Pelvic organ prolapse surgery and overactive bladder symptoms—a population-based cohort (FINPOP). Int Urogynecol J. 2022;33 (1 ):95–105. doi:10.1007/s00192-021-04920-w 34245317 9. Petros PE, Ulmsten UI. An integral theory of female urinary incontinence. Experimental and clinical considerations. Acta Obstet Gynecol Scand Suppl. 1990;153 (Suppl ):7–31. doi:10.1111/j.1600-0412.1990.tb08027.x 2093278 10. Petros PEP. The Integral Theory System. A simplified clinical approach with illustrative case histories. Pelviperineology. 2010;29 (1 ):37–51. 11. Boer T, Salvatore S, Cardozo L, et al. Pelvic organ prolapse and overactive bladder. Neurourol Urodyn. 2010;29 (1 ):30–39. doi:10.1002/nau.20858 20025017 12. Burrows LJ, Meyn LA, Walters MD, et al. Pelvic symptoms in women with pelvic organ prolapse. Obstet Gynecol. 2004;104 (5 ):982–988. doi:10.1097/01.AOG.0000142708.61298.be 15516388 13. Dietz HP, Clarke B. Is the irritable bladder associated with anterior compartment relaxation? A critical look at the ‘integral theory of pelvic floor dysfunction’. Aust N Z J Obstet Gynaecol. 2001;41 (3 ):317–319. doi:10.1111/j.1479-828x.2001.tb01236.x 11592549 14. Zacharakis D, Grigoriadis T, Pitsouni E, Kypriotis K, Vogiatzis N, Athanasiou S. Assessment of overactive bladder symptoms among women with successful pessary placement. Int Urogynecol J. 2018;29 (4 ):571–577. doi:10.1007/s00192-017-3461-x 28871426 15. Coolen AWM, Troost S, Mol BWJ, et al. Primary treatment of pelvic organ prolapse: pessary use versus prolapse surgery. Int Urogynecol J. 2018;29 (1 ):99–107. doi:10.1007/s00192-017-3372-x 28600758 16. Johnson JR, High RA, Dziadek O, et al. Overactive bladder symptoms after pelvic organ prolapse repair. Female Pelvic Med Reconstr Surg. 2020;26 (12 ):742–745. doi:10.1097/SPV.0000000000000700 30681419 17. Peterson TV, Karp DR, Aguilar VC, et al. Validation of a global pelvic floor symptom bother questionnaire. Int Urogynecol J. 2010;21 (9 ):1129–1135. doi:10.1007/s00192-010-1148-7 20458467 18. Manonai J, Wattanayingcharoenchai R. Relationship between pelvic floor symptoms and POP-Q measurements. Neurourol Urodyn. 2016;35 (6 ):724–727. doi:10.1002/nau.22786 25919311 19. Kinno K, Sekido N, Takeuchi Y, Sawada Y, Watanabe S, Yoshimura Y. Association between overactive bladder and pelvic organ mobility as evaluated by dynamic magnetic resonance imaging. Sci Rep. 2021;11 (1 ):1–11. doi:10.1038/s41598-021-93143-6 33414495 20. Miranne JM, Lopes V, Carberry CL, Sung VW. The effect of pelvic organ prolapse severity on improvement in overactive bladder symptoms after pelvic reconstructive surgery. Int Urogynecol J. 2013;24 (8 ):1303–1308. doi:10.1007/s00192-012-2000-z 23229418 21. Yaranun S, Chiengthong K, Ruanphoo P, Bunyavejchevin S. Overactive bladder symptom score changes after pessary insertion in women with pelvic organ prolapse and overactive bladder. Thai J Obstet Gynaecol. 2022;28 :1–7. 22. Wu LY, Huang KH, Yang TH, et al. The surgical effect on overactive bladder symptoms in women with pelvic organ prolapse. Sci Rep. 2021;11 (1 ):20193. doi:10.1038/s41598-021-99537-w 34642384 23. Padoa A, Levy E, Fligelman T, Tomashev-Dinkovich R, Tsviban A, Serati M. Predictors of persistent overactive bladder following surgery for advanced pelvic organ prolapse. Int Urogynecol J. 2023;34 (3 ):759–767. doi:10.1007/s00192-022-05313-3 35907022 24. Liang CC, Hsieh WC, Lin YH, Tseng LH. Predictors of persistent detrusor overactivity in women with pelvic organ prolapse following transvaginal mesh repair. J Obstet Gynaecol Res. 2016;42 (4 ):427–433. doi:10.1111/jog.12927 26786248 25. Frigerio M, Manodoro S, Cola A, Palmieri S, Spelzini F, Milani R. Risk factors for persistent, de novo and overall overactive bladder syndrome after surgical prolapse repair. Eur J Obstet Gynecol Reprod Biol. 2019;233 :141–145. doi:10.1016/j.ejogrb.2018.12.024 30597338 26. Dietz HP, Moegni F, Shek KL. Diagnosis of levator avulsion injury: a comparison of three methods. Ultrasound Obstet Gynecol. 2012;40 (6 ):693–698. doi:10.1002/uog.11190 22605560
PMC010xxxxxx/PMC10352123.txt
==== Front Clin Cosmet Investig Dermatol Clin Cosmet Investig Dermatol ccid Clinical, Cosmetic and Investigational Dermatology 1178-7015 Dove 421152 10.2147/CCID.S421152 Case Report Granular Parakeratosis of the Eccrine Ostium: A Case Report Sriprachya-anunt et al Sriprachya-anunt et al http://orcid.org/0000-0003-3157-2549 Sriprachya-anunt Sittha 1 http://orcid.org/0000-0001-8268-8790 Rutnin Suthinee 1 http://orcid.org/0000-0001-9723-0563 Suchonwanit Poonkiat 1 1 Division of Dermatology, Department of Medicine, Faculty of Medicine, Ramathibodi Hospital, Mahidol University, Bangkok, Thailand Correspondence: Poonkiat Suchonwanit, Division of Dermatology, Department of Medicine, Faculty of Medicine, Ramathibodi Hospital, Mahidol University, 270 Rama VI Road, Ratchathewi, Bangkok, Thailand, 10400, Tel +66-2-2011141, Fax +66-2-201-1211 ext 4, Email [email protected] 13 7 2023 2023 16 18071810 12 5 2023 04 7 2023 © 2023 Sriprachya-anunt et al. 2023 Sriprachya-anunt et al. https://creativecommons.org/licenses/by-nc/3.0/ This work is published and licensed by Dove Medical Press Limited. The full terms of this license are available at https://www.dovepress.com/terms.php and incorporate the Creative Commons Attribution – Non Commercial (unported, v3.0) License (http://creativecommons.org/licenses/by-nc/3.0/). By accessing the work you hereby accept the Terms. Non-commercial uses of the work are permitted without any further permission from Dove Medical Press Limited, provided the work is properly attributed. For permission for commercial use of this work, please see paragraphs 4.2 and 5 of our Terms (https://www.dovepress.com/terms.php). Abstract Granular parakeratosis (GP) is a unique keratotic disorder that often affects the intertriginous areas. GP usually presents as erythematous or brownish hyperkeratotic papules or plaques and can be further classified into five types. GP of the eccrine ostium is a rare subtype; its pathological defects are mainly localized to the stratum corneum of the eccrine ostia. Due to its rarity, there is usually a delay in diagnosing GP, and these patients are often misdiagnosed with other dermatological conditions. In this report, we present the case of a 64-year-old Thai female who presented with recurrent pruritic erythematous rashes on her neck since approximately 40 years. She was previously diagnosed with eczema or folliculitis. Histopathological examination confirmed a final diagnosis of GP of the eccrine ostium. She was advised to avoid excessive heat and keep her intertriginous areas dry. Her condition improved significantly during the follow-up visit. Keywords flexural area granular parakeratosis hyperkeratosis keratinization disorder keratotic papules sweat gland funding No sources of funding were used to prepare this manuscript. ==== Body pmcIntroduction Granular parakeratosis (GP) is an uncommon dermatological disorder due to abnormal keratinization, with an incidence of 0.005%.1,2 The clinical presentations include brownish to erythematous papules that sometimes coalesce into plaques, mainly on the intertriginous areas.3 Because of its clinical characteristics and benign nature, GP is usually misdiagnosed, and the correct diagnosis may be delayed for up to 20 years.4 In this case report, we present a case of GP of the eccrine ostium, a rare type of GP, which was definitely diagnosed almost four decades after the initial presentation of the disease. Case Presentation A 64-year-old Thai female visited the dermatology clinic with a 4-day history of an itchy erythematous rash on her neck. She reported experiencing the same manifestations since her twenties, with a frequency of up to three to four episodes in some years. The lesions tended to be stimulated by hot climate and profuse sweating and resolved spontaneously or after the application of topical medications. Having been previously diagnosed with either eczema or folliculitis with similar presentations, she initially received topical and oral medications from a local pharmacy to manage her symptoms. However, the lesions did not improve despite a 3-day treatment with topical pimecrolimus and oral clindamycin 300 mg twice daily. She reported that her mother also suffered from the same condition but to a milder degree. The underlying conditions included hypertension, dyslipidemia, type-2 diabetes mellitus, allergic rhinitis, major depressive disorder, non-alcoholic steatohepatitis, and myasthenia gravis. Dermatological examination revealed multiple discrete brownish spiny keratotic papules and scaly erythematous papules on the neck (Figure 1A and B). No obvious skin lesions were observed on the axillae of the patient. Histological analysis revealed a mild epidermal hyperplasia with focal mound parakeratosis. Higher magnification showed that parakeratotic corneocytes contain multiple basophilic granules, similar to keratohyalin granules. These features were prominent in the stratum corneum of the eccrine ostia (Figure 2A and B).Figure 1 Clinical presentations: multiple erythematous to brownish spiny keratotic papules on the neck at the initial visit (A and B). Figure 2 Histopathology: (A) mild epidermal hyperplasia with focal mound parakeratosis (hematoxylin-eosin, x100). (B) Higher magnification showed parakeratotic corneocytes with retained keratohyalin granules, which were more prominent around the eccrine ostia (hematoxylin-eosin, x400). Based on the clinical manifestations and histopathological findings, the patient was diagnosed with GP of the eccrine ostium. She was advised to avoid excessive heat and keep her intertriginous areas dry. At the 2-week follow-up visit, the lesions significantly improved (Figure 3A and B).Figure 3 Significant improvement of the lesions at the 2-week follow-up visit (A and B). Discussion First described by Northcutt in 1991, GP is a rare acquired disorder of epidermal keratinization.2,3 It is usually found in the adult population, although children as young as 3 months of age have been reported to have such a condition.5,6 Females are more commonly affected than men.1 GP usually presents as erythematous to brownish hyperkeratotic papules or plaques, mainly affecting intertriginous areas. In a recent systematic review of 129 patients with GP from 60 studies, the axilla was found to be the most commonly affected area (56.5%), followed by the groins (31.8%), inter/submammary areas (10.9%), and anogenital areas (10.1%). The neck, as in our case, was affected in 7% of all cases.4 GP has been found to be associated with obesity, heat, sweating, friction, and topical agents such as zinc oxide, deodorants, and benzalkonium chloride.7–9 In terms of pathogenesis, the primary defect is in the processing of profilaggrin to filaggrin during keratinocyte cornification, resulting in failure to degrade keratohyalin granules.2 Altered proliferation and maturation of the epidermis are thought to be triggered and stimulated by the aforementioned factors.10,11 GP is also reported to be associated with other dermatological and systemic conditions such as ichthyosis, atopic dermatitis, dermatomyositis, molluscum contagiosum, fungal infection, and cancers.9,12–18 A recent report by Chirasuthat et al proposed the classification of GP into 5 types based on clinical presentations and histopathological findings. These include intertriginous GP, GP of the eccrine ostium, follicular GP, GP acanthoma, and incidental GP. As in our case, GP of the eccrine ostium usually presents with multiple brownish spiny keratotic papules on the neck with a tingling sensation when sweating. This is generally triggered by hot environments and excessive sweating. A histopathological study demonstrated parakeratotic corneocytes containing keratohyalin granules, which usually localize to the stratum corneum of the eccrine ostia.12 Owing to its diverse clinical presentations, GP may be misdiagnosed as other dermatological conditions, such as eczema, contact dermatitis, Darier’s disease, confluent and reticulated papillomatosis, epidermal nevus, folliculitis, or fungal infection. Moreover, routine biopsy may not be performed in such patients because of the benign nature of GP. As a result, GP is thought to be underrecognized. There is usually a delay in diagnosis, with a mean duration of 19.2 months and a range of up to 20 years from the first presentation of the condition.4,19 In our case, the patient reported experiencing itchy erythematous lesions similar to her current presentation since the age of 25 years. A review of her dermatologic outpatient visits revealed diagnoses of either eczema or folliculitis on multiple occasions, which were subsequently treated with topical corticosteroids or antifungals. Due to multiple recurrences of her condition, a skin biopsy was eventually performed, which revealed a diagnosis of GP. In terms of prognosis, GP is usually self-limiting and improves after discontinuation or avoidance of the triggering factors. A recent systematic review reported that the condition spontaneously resolves in 4.7% of cases.4 Several treatment options have been attempted with variable responses. These include topical corticosteroids, vitamin D analogues, retinoids, ammonium lactate, antifungals, oral isotretinoin, cryotherapy, and laser therapy.2,3,20–25 Our patient’s lesions resolved spontaneously. She was prescribed topical corticosteroids when she experienced itchiness, advised to avoid excessive sweating, and was kept dry. However, it is important to note that this case report has some limitations, including relying on anecdotal evidence from a single case and lacking generalizability. Conclusion We present a case of granular parakeratosis of the eccrine ostium, a rare keratotic disorder, in a patient who had suffered from the condition for almost 4 decades before the correct diagnosis was confirmed. Its recurrent and benign nature often results in GP being under-recognized and misdiagnosed as other conditions such as eczema or folliculitis. This report also highlights the importance of histopathological examination, which can help distinguish GP from other dermatological disorders with similar presentations and prevent delays in establishing a correct diagnosis. Ethics Approval and Consent to Participate This study was conducted in accordance with the principles of the Declaration of Helsinki. Ethical review and approval were not required to publish case details in accordance with local legislation and institutional requirements. Written informed consent was obtained from the patient for publication of this case report and any accompanying images according to our standard institutional rules. Disclosure The authors declare that this article was prepared in the absence of any commercial or financial relationships that could be construed as potential conflict of interest. ==== Refs References 1. Scheinfeld NS, Mones J. Granular parakeratosis: pathologic and clinical correlation of 18 cases of granular parakeratosis. J Am Acad Dermatol. 2005;52 (5 ):863–867. doi:10.1016/j.jaad.2004.12.031 15858479 2. Metze D, Rutten A. Granular parakeratosis: a unique acquired disorder of keratinization. J Cutan Pathol. 1999;26 (7 ):339–352. doi:10.1111/j.1600-0560.1999.tb01855.x 10487291 3. Northcutt AD, Nelson DM, Tschen JA. Axillary granular parakeratosis. J Am Acad Dermatol. 1991;24 (4 ):541–544. doi:10.1016/0190-9622(91)70078-G 2033126 4. Ip HK, Li A. Clinical features, histology, and treatment outcomes of granular parakeratosis: a systemic review. Int J Dermatol. 2022;61 (8 ):973–978. doi:10.1111/ijd.16107 35094385 5. Neri I, Patrizi A, Guerrini V, et al. Granular parakeratosis in a child. Dermatology. 2003;206 (2 ):177–178. doi:10.1159/000068454 12592091 6. Pimentel DRN, Michalany N, Morgado de Abreu MAM, et al. Granular parakeratosis in children: case report and review of literature. Pediatr Dermatol. 2003;20 (3 ):215–220. doi:10.1046/j.1525-1470.2003.20306.x 12787269 7. Ding CY, Liu H, Khachemoune A. Granular parakeratosis: a comprehensive review and a critical reappraisal. Am J Clin Dermatol. 2015;16 (6 ):495–500. doi:10.1007/s40257-015-0148-2 26242230 8. Kossard S, White A. Axillary granular parakeratosis. Australas J Dermatol. 1998;39 (3 ):186–187. doi:10.1111/j.1440-0960.1998.tb01280.x 9737049 9. Robinson AJ, Foster RS, Halbert AR, et al. Granular parakeratosis induced by benzalkonium chloride exposure from laundry rinse aids. Australas J Dermatol. 2017;58 (3 ):138–140. doi:10.1111/ajd.12551 10. Martorell A, Sanmartin O, Hueso-Gabriel L, et al. Granular parakeratosis: disease or reactive response? Actas Dermosifiliogr. 2011;1 (102 ):72–74. doi:10.1016/j.ad.2010.04.014 11. Wallace CA, Pichardo RO, Yosipovitch G, et al. Granular parakeratosis: a case report and literature review. J Cutan Pathol. 2003;30 (5 ):332–335. doi:10.1034/j.1600-0560.2003.00066.x 12753175 12. Chirasuthat P, Chirasuthat S, Suchonwanit P. Follicular granular parakeratosis: a case report, literature review, and proposed classification. Skin Appendage Disord. 2021;7 (2 ):144–148. doi:10.1159/000512950 33796563 13. Resnik KS, DiLeonardo M. Incidental granular parakeratosis associated with dermatomyositis. Am J Dermatopathol. 2007;29 (3 ):264–269. doi:10.1097/DAD.0b013e3180465860 17519624 14. Suchonwanit P, McMichael AJ. Alopecia in Association with Malignancy: a Review. Am J Clin Dermatol. 2018;19 (6 ):853–865. doi:10.1007/s40257-018-0378-1 30088232 15. Pock L, Cermakova A, Zipfelová J, et al. Incidental granular parakeratosis associated with molluscum contagiosum. Am J Dermatopathol. 2006;28 (1 ):45–47. doi:10.1097/01.dad.0000157448.54281.d9 16456325 16. Resnik KS, Kantor GR, DeLeonardo M. Dermatophyte-related granular parakeratosis. Am J Dermatopathol. 2004;26 (1 ):70–71. doi:10.1097/00000372-200402000-00011 14726826 17. Suchonwanit P, Kositkuljorn C, Pomsoong C. Alopecia areata: an autoimmune disease of multiple players. Immunotargets Ther. 2021;10 :299–312. doi:10.2147/itt.S266409 34350136 18. Resnik KS, DiLeonardo M. Incidental granular parakeratotic cornification in carcinomas. Am J Dermatopathol. 2007;29 (3 ):264–269.17519624 19. Reddy S, Swarnalata G, Mody T. Intertriginous granular parakeratosis persisting for 20 years. Indian J Dermatol Venereol Leprol. 2008;74 (4 ):405–407. doi:10.4103/0378-6323.42928 18797082 20. Chamberlain AJ, Tam MM. Intertriginous granular parakeratosis responsive to potent topical corticosteroids. Clin Exp Dermatol. 2003;28 (1 ):50–52. doi:10.1046/j.1365-2230.2003.01159.x 12558631 21. Suchonwanit P, Iamsumang W, Leerunyakul K. Topical finasteride for the treatment of male androgenetic alopecia and female pattern hair loss: a review of the current literature. J Dermatolog Treat. 2022;33 (2 ):643–648. doi:10.1080/09546634.2020.1782324 32538225 22. Contreras ME, Gottfried LC, Bang RH, et al. Axillary intertriginous granular parakeratosis responsive to topical calcipotriene and ammonium lactate. Int J Dermatol. 2003;42 (5 ):382–383. doi:10.1046/j.1365-4362.2003.01722.x 12755978 23. Brown SK, Heilman ER. Granular parakeratosis: resolution with topical tretinoin. J Am Acad Dermatol. 2002;47 (5 ):s279–280. doi:10.1067/mjd.2002.109252 12399751 24. Webster CG, Resnik KS, Webster GF. Axillary granular parakeratosis: response to isotretinoin. J Am Acad Dermatol. 1997;37 (5 ):789–790. doi:10.1016/S0190-9622(97)70119-1 9366832 25. Laimer M, Emberger M, Brunasso AM, et al. Laser for the treatment of granular parakeratosis. Dermatol Surg. 2009;35 (2 ):297–300. doi:10.1111/j.1524-4725.2008.01052.x 19215276
PMC010xxxxxx/PMC10352124.txt
==== Front Clin Ophthalmol Clin Ophthalmol opth Clinical Ophthalmology (Auckland, N.Z.) 1177-5467 1177-5483 Dove 411472 10.2147/OPTH.S411472 Original Research A Novel Procedure for Keratoconus/Corneal Ectasia Treating Epithelial Compensation of Higher-Order Aberrations, Topographic Guided Ablation, and Corneal Cross Linking – The CREATE+CXL Protocol Motwani Motwani http://orcid.org/0000-0001-8482-2017 Motwani Manoj 1 1 Cornea Revolution/Motwani LASIK Institute, San Diego, CA, 92121, USA Correspondence: Manoj Motwani, Cornea Revolution/Motwani LASIK Institute, 8710 Scranton Road, Ste 170, San Diego, CA, 92121, USA, Tel +1 858 554-0008, Email [email protected] 13 7 2023 2023 17 19811992 07 3 2023 22 6 2023 © 2023 Motwani. 2023 Motwani. https://creativecommons.org/licenses/by-nc/3.0/ This work is published and licensed by Dove Medical Press Limited. The full terms of this license are available at https://www.dovepress.com/terms.php and incorporate the Creative Commons Attribution – Non Commercial (unported, v3.0) License (http://creativecommons.org/licenses/by-nc/3.0/). By accessing the work you hereby accept the Terms. Non-commercial uses of the work are permitted without any further permission from Dove Medical Press Limited, provided the work is properly attributed. For permission for commercial use of this work, please see paragraphs 4.2 and 5 of our Terms (https://www.dovepress.com/terms.php). Purpose To present the outcomes of a retrospective study in keratoconus/corneal ectasia patients of treating the higher order aberrations compensated for the corneal epithelium in addition to topographic guided ablation followed by corneal cross linking. Methods Twenty-seven eyes of 14 patients were treated for keratoconus/corneal ectasia utilizing trans-epithelial topographic guided ablation photorefractive keratectomy (PRK) for treatment of corneal higher order aberrations and lower order astigmatism followed immediately by 15-minute cross linking were examined retrospectively. Six-month results were analyzed via measurement of vision, refraction, residual higher-order aberrations (HOAs), residual lower-order and higher-order aberrations, as well as for loss or gains of lines of best corrected visual acuity. Results All eyes save one had reduction in K1, K2, K Max, and K Mean. All eyes had reduction in manifest astigmatism, Contoura measured astigmatism, 57% reduction of higher-order aberrations (HOA), and 53% reduction of higher-order aberrations grouped with lower-order aberrations (Grouped). Nearly all (96.3%) eyes achieved 20/40 vision or better, 20 eyes had 1–7 lines gained of vision, and no eyes had any loss of lines of vision. Conclusion Use of the CREATE+CXL protocol combined with 15-minute corneal cross linking results in a significant increase in HOA reduction, as well as a significant improvement in corrected distance visual acuity over past procedures. Keywords astigmatism corneal cross linking higher-order aberrations keratoconus lower-order aberrations photorefractive keratectomy corneal ectasia ==== Body pmcIntroduction Corneal ectasia is a degenerative condition that causes progressive corneal protrusion, irregular astigmatism, and thinning. Keratoconus is a type of corneal ectasia, and the most common one.1 Historically in the United States treatment has utilized contact lenses (soft/rigid gas permeable (RGP)/scleral), spectacles, Intacs corneal implants to attempt to modify the shape, and in severe cases corneal transplant.2 Corneal cross linking has been available in the United States since 2016 to prevent progression of the ectasia, but does little to improve vision. Contact lenses do not modify disease progression, and become more difficult to fit in severe disease. Intacs have been shown to be useful, but are a gross attempt to modify the corneal shape rather than a specific normalization of the cornea. Corneal transplants have their own issues from newly induced irregular astigmatism, to rejection, and corneal failure over time.3,4 The goal has been to find a permanent solution that would allow the highest degree of normalization of the original cornea while preventing further progression. For well over a decade, the Athens Protocol has been utilized to decrease corneal irregularity via topographic guided ablation (WaveLight Contoura, WaveLight Laser Technologie AG, Erlangen, Germany) and then perform corneal cross linking to stiffen the cornea and “freeze” it into place.5 In a prior publication, the author described the United States experience with this procedure, with modifications from new scientific understanding from the LYRA Protocol/San Diego Protocol to attempt to achieve a better visual outcome.6–9 One of the fundamental problems with normalizing the cornea is that the topography system cannot measure corneal stromal irregularity that has been compensated for by changes in corneal epithelial thickness. This compensation has been described as significant (20–30 µm) in publications by Reinstein et al, Kanellopoulos et al, as well this author.10–15 Over the past several years commercial technology such as anterior segment optical coherence tomography has been available to map this epithelial compensation, but there is no existing technology that combines the topographic analysis with the epithelial compensation to create a map of the full irregularity. As such a device is unavailable, the authors used PTK (Phototherapeutic Keratectomy) to the depth of epithelial compensation provided by an Visionix/Optovue epithelial thickness mapping (ETM) device. This would ablate corneal stromal tissue compensated for by the corneal epithelium as corneal stroma and epithelium ablate similarly for each excimer laser pulse. This trans-epithelial PRK utilizing topographic guided ablation (WaveLight Contoura) was termed the CREATE Protocol, Corneal Repair Epithelium and Topography Enhanced, and when used in conjunction with corneal cross linking for corneal ectasia is termed the CREATE+CXL Protocol. Another major change from our prior publication has been the use of mitomycin-C. In Feb 2021, a publication demonstrated deep scarring and haze from the use of mitomycin-C with corneal cross linking.16 Since that time, we have eliminated the usage of mitomycin-C with CREATE+CXL procedures, and all patients in this manuscript’s series have not had mitomycin-c exposure. As stated in our prior publication,9 the goals of the procedure, in order of importance, are: Stop progression of keratoconus and strengthen the cornea to avoid cornea transplant. Normalize the cornea by decreasing the irregularity of the anterior cornea improving the optics of the cornea, as well as increasing the ability of further refractive correction via spectacles, soft contact lenses, and RGP/scleral lenses. In cases where sufficient tissue is available, also correct the sphere and cylinder to get as close to plano correction as possible. Materials and Methods We retrospectively examined patients who had keratoconus/corneal ectasia treatment the CREATE+CXL Protocol, and also had at least 6 months of follow-up results. As many of our patients come from out of town, we do not have full long-term results on many of our treatments, therefore we included all eyes in this study where at least 6-month results were obtained at our clinic. We were able to analyze data from 27 eyes of 14 patients. Results were tabulated for pre- and post-operative vision, average K and Kmax, and manifest cylinder. Pre-op Contoura (topography utilizing the Topolyzer Vario (WaveLight, Erlangen, Germany)) measured astigmatism was also averaged. Pre- and post-operative corrected distance visual acuity (CDVA) was tabulated, as were lines of vision gained or lost. Measurements up to 6th order Zernike polynomials to measure higher order aberrations (HOA) were tabulated, as were what we defined as Grouped. These are HOA plus lower order sphere and astigmatism, excluding piston and tilt. The first number was measured to demonstrate corneal irregularity reduction, the second to include lower order aberration reduction. Epithelial compensation was measured utilizing the Visionix/Optovue (Visionix/Optovue, Fremont, CA) epithelial mapping system on their OCT devices (Solix and Avanti), by measuring epithelial thickness of the thinnest and thickest areas in the central 5 mm of the cornea. The difference between these two numbers was the epithelial compensation for the corneal irregularity. Epithelium removal was performed utilizing either the PTK mode the Nidek EC-5000, or the WaveLight WFO myopic treatment to the depth of corneal epithelium removal determined by Optovue OCT mapping. This is due to the fact that there is no PTK mode available on WaveLight lasers in the United States. Residual epithelial removal was accomplished to a width of 9 mm via manual removal by cellulose sponge or corneal epithelial scraper. WaveLight Contoura was then used to treat the corneal higher-order aberrations and irregular astigmatism measurable by topography of the anterior cornea.13 Surgical planning utilized the measured astigmatism up to the maximum 3 diopters (D) allowed in the Contoura FDA approval. Most patients had no sphere treated due to tissue constraints. In milder keratoconus cases where the cone was not in the central visual axis spherical treatment was performed to a maximum of 70% of the manifest sphere after any adjustment for the spherical equivalent of astigmatism difference between manifest and measured astigmatism. Treatment parameters were designed to leave a corneal bed of 350 µm to maintain safety parameters for corneal cross linking. If the residual corneal bed after laser ablation was less than 350 µm (in this study between 320–350 µm) hypotonic saline was used to increase the thickness of the cornea to over 400 µm before cross-linking was performed. Measured astigmatism axis was always used for the topography guided ablation. It is notable that in many patients the highest amount of corneal tissue removal during topographic guided ablation was often not centrally, but in the superior periphery to normalize the superior flat area. It was essential during surgical planning to only use the central tissue removal depth, and not the overall ablation depth during calculations to leave a corneal bed of 350 µm to ensure safe corneal cross linking. We calculated this utilizing pachymetry from scheimpflug imaging, OCT pachymetry, OCT epithelial thickness maps, as well as the optical pachymetry on the WaveLight EX500. The amount of treatment was limited only by available tissue and the limit of 3 diopters of astigmatism treatment with WaveLight Contoura. No other excimer treatment was performed except for epithelial removal and the sole topographic guided ablation treatment. Corneal saturation with riboflavin was performed immediately after laser treatment for 2 minutes. As corneal cross linking (CXL) treatment was performed on the right eye first as a rule, the left eye would be re-saturated for another minute before UV treatment. An “off-label” 15-minute cross-linking procedure using a 10-mm spot size, homogenous UV light beam, with 6 mW/cm2 (ie a total UV light dose of 5.4 J/cm2), was performed. The riboflavin formulation used was 0.1% riboflavin w/Carboxymethyl Cellulose pH Balance (no dextran). All LASIK procedures were performed on the WaveLight EX500 excimer laser. All procedures were performed by one surgeon (MM) at one center in San Diego, California. All topographies were obtained utilizing the Topolyzer Vario (Alcon Surgical, Fort Worth, TX, USA). All epithelial thickness maps (ETM) were obtained with the Optovue Avanti or Solix devices (9-mm ETM) (Optovue, Fremont, CA, USA). Post-operative care consisted of fluorometholone 0.1% bid for 6 weeks, ofloxacin 0.3% qid x 1 week, and Prolensa qd as needed for pain during epithelial healing. Bandage contact lenses were removed with healing of the corneal abrasion which was between 4–5 days on average. As mentioned in the discussion, no Mitomycin-C was applied to any patients in this study. All patients signed written informed consent forms allowing their data to be used in this study and published including sample cases 1–4. This study falls under the exemption of the Health and Human Services (HHS) Policy for the Protection of Human Research Subjects 45 CFR 46.104 (d) for retrospective studies and 46.104 for exempt research, and thus, no Institutional Review Board approval was required. This study also conforms to the Declaration of Helsinki guidelines. There were no safety-related incidents that occurred or were reported to Alcon Inc. or WaveLight concerning patients involved in this study. Results Twenty seven (27) eyes of 14 patients are included in this study, 7 men and 7 women. The average patient age was 31.85 years, range 18–50 years. 25 eyes had keratoconus, and 2 post-LASIK ectasia. All results come from eyes that had at least 6 months of post-operative follow-up. Table 1 shows pre-op and post-op values for average K and Kmax, pre-op measured and manifest cylinder, and post-op manifest cylinder. Average post-op average K and Kmax all decreased, as did the post-operative manifest refraction. Pre-op average Contoura measured astigmatism was higher than the pre-op average manifest astigmatism.Table 1 Pre-Op and Post-Op Values for Average K and Kmax, Pre-Op Measured and Manifest Cylinder, and Post-Op Manifest Cylinder Avg. Kmax Avg. K Measured Cyl Manifest Cyl Pre-op 53.62 D (44.6–61.4) 45.49 D (32.7–52.10) 3.56 D (0.9–7.64) 2.37 D (0–6.5) Post-op 6 mos 49.41 D (43.6–71.5) 43.54 D (37.3–51.9) Not measured 1.07 D (0–6) Abbreviations: Cyl, cylinder; D, diopter. Table 2 presents the best corrected visual acuities pre-op and post-op by Snellen chart measurements, as well as the percentages of eyes in that category. It is notable that 96.3% of eyes were able to see 20/40 or better postoperatively, while only 70.35% could achieve 20/40 pre-op.Table 2 Best Corrected Visual Acuities Pre-Op and Post-Op by Snellen Chart Measurements BCVA Pre-Op Eyes Avg. Percentage Post-Op Eyes Avg. Percentage 20/15 0 0 1 3.7% 20/20 5 18.5% 13 48.15% 20/25 6 22.22% 8 29.63% 20/30 6 22.22% 3 11.11% 20/40 2 7.41% 1 3.7% 20/50 2 7.41% 0 0 20/60 3 11.11% 0 0 20/70 2 7.41% 0 0 20/80 1 3.7% 1 3.7% Abbreviations: Avg, average; BCVA, best-corrected visual acuity. Table 3 shows the lines of BCVA gained. 74% of eyes gained between 1–7 lines of vision, while the remaining 26% neither gained nor lost lines of vision. Notably no eyes lost any lines of BCVA.Table 3 Gains in Best Corrected Visual Acuity Lines of Vision Gained of BCVA Number of Eyes Percentage 0 7 25.92% 1 9 33.33% 2 4 14.81% 3 2 7.41% 4 1 3.7% 5 3 11.11% 6 0 0 7 1 3.7% Abbreviation: BCVA, best-corrected visual acuity. In Table 4 Zernike polynomials are presented for 6th order higher-order aberrations (HOA), and for Grouped polynomials which includes HOA and lower-order sphere and astigmatism (does not include tilt or constant height/piston). The average HOA were reduced by 57.3% and the Grouped values were reduced by 52.8%. This compares favorably with our prior corneal ectasia treatment study that did not treat epithelial compensation which had an average reduction of 31.8% for HOA and 25.5% for the Grouped values. Statistical testing with two tailed t-test for pre-and post-op HOA and Grouped values were statistically significant with P < 0.05, as was the difference in reduction values between our prior study and this study.Table 4 Zernike Polynomials for 6th Order Higher-Order Aberrations and for Grouped Polynomials HOA 5 Grouped Pre-op avg. microns 0.943 (0.26–2.1) 1.3 (0.46–2.5) Post-op avg. microns 0.403 (0.2–1.4) 0.61 (0.11–1.2) Avg. Reduction 57.27% (6.3–83.3%) 52.8% (6.9–64.1%) Abbreviations: Avg, average; HOA, higher-order aberrations. Table 5 shows the average amount of epithelial compensation of the stroma. This was determined for the entire cohort to be 18.9 µm, or 24% of the overall aberration treatment depth of 78 µm. The average number was calculated utilizing the depth of ablation treatment determined by WaveLight Contoura utilizing the 6mm treatment zone. This was limited by the maximum 3 diopters of astigmatism correction, as 15 out of the 27 eyes had measured astigmatism greater than 3 diopters. As treating lower-order aberrations (LOA) in combination with HOA creates a unique topography guided ablation pattern (in many cases treating the astigmatism does not increase the overall HOA reduction tissue removal as much as performing the astigmatism correction separately), it was not possible in these eyes to determine the total amount of aberration that topography system was measuring. In 12 eyes the measured astigmatism was below 3 diopters, and in these eyes the average epithelial compensation was 19.6 µm, or 28% of the total aberration of 69.2 µm. When statistical testing with two tailed t-test was performed comparing the two, it was not statistically significant with a P > 0.05.Table 5 Average Amount of Epithelial Compensation of the Stroma All Eyes Eyes with <3D of Contoura Cylinder Average Epithelial Compensation in microns (range) 18.92 (8–47) 48.58 (25–85) Total Contoura Measured HOA depth in microns (range) 59.51 (23–92) 19.58 (10–47) Total Aberration in microns 78.43 69.16 Epithelial compensation as % of HOA 24.1% 28% Abbreviation: HOA, higher-order aberrations. All eyes had minimal haze, trace to 1+, that was easily controlled and reduced with steroid treatment via either fluorometholone 0.1% or combined with a short burst of prednisolone 0.1%. It was noted through the entire patient group that haze was significantly less of a factor than when mitomycin-C was used. Two tailed t-tests were performed to look for statistical significance. Comparisons of pre-op to post op HOA values and HOA values were statistically significant, with P less than 0.05. This test was also performed comparing pre-op HOA and grouped values to the three different sub groups. Sample Cases Case 1: 49 year old male (Figure 1A–E).Figure 1 (A–E) Case 1: 49 year old male. (A) Pre-op topography, (B) Post-op topography, (C) Pre-op OCT pachymetry and epithelial thickness map, (D) Post-op OCT pachymetry and epithelial thickness map, (E) patient data. CASE 2: 32 year old male (Figure 2A–E).Figure 2 (A–E) Case 2: 32 year old male. (A) Pre-op topography, (B) Post-op topography, (C) Pre-op OCT pachymetry and epithelial thickness map, (D) Post-op OCT pachymetry and epithelial thickness map, (E) patient data. CASE 3: 20 year old female (Figure 3A–E).Figure 3 (A–E) Case 3: 20 year old female: (A) Pre-op topography, (B) Post-op topography, (C) Pre-op OCT pachymetry and epithelial thickness map, (D) Post-op OCT pachymetry and epithelial thickness map, (E) patient data. Discussion The amount of irregularity removed during corneal laser normalization increased significantly with use of the CREATE+CXL protocol in comparison to our prior publication results, from 31.87% for HOA and 25.5% for the Grouped values, to 57.3% HOA and 52.8% Grouped. This increase in the reduction of corneal irregularity was reflected in the increase in quantity of vision as well as significant improvements in the topographic shape of the corneas. Corresponding with that reduction in irregularity, 96.3% of eyes had post-operative vision of 20/40 or better (all but 2 eyes), and 74% eyes gained 1–7 lines of best corrected visual acuity (BCVA). The other 26% of eyes did not lose or gain lines of BCVA. No eyes lost lines of BCVA. It is important to note here that epithelial compensation of stromal irregularity can be seen with epithelial mapping devices such as the Optovue OCT machines, but cannot be measured by the Vario Topolyzer/WaveLight Contoura platforms, or for that matter any topography or wavefront measurement system. In other words, Contoura measures the topographic shape after epithelial compensation, and once that irregularity is removed the epithelium re-compensates to the smaller but not completely removed irregularity exposing residual irregularity. The only commercial epithelial thickness mapping device to measure epithelial compensation available in the United States are the Optovue OCT epithelial mapping devices, and as of yet no device exists to combine epithelial compensated HOA with topographic measured HOA. Disclaimer, such a device is described in United States patent no. 10857033 granted to the author in December 2020 but as of this writing has not yet been built. Furthermore, the author knows of no current device, software algorithm such as the Phorcides Analytic Engine, or any upcoming device that measures and includes epithelial compensation of stroma in its topographic guided laser ablation algorithm. The significantly increased reduction in HOA, combined with visual results and visual reduction of topographic abnormality as displayed in the case examples, make a strong case for including epithelial compensation as part of any algorithm to decrease corneal irregularity. It is notable that in a prior publication in 2019, the authors found that epithelial compensation negatively affected the accuracy of outcomes even in primary LASIK corrections utilizing topographic guided ablation and the LYRA Protocol.15 The outcomes in this study confirm that omitting Mitomycin-C in topographic guided ablation and cross-linking procedures actually produces significantly less haze. Haze has ceased to be an issue in our patients, with at most some +1 superficial corneal haze noticed on some patients that was easily treatable with steroid treatment. Notably, this lack of deep haze was part of the reason no eyes lost BCVA, along with the increased HOA reduction. The decrease in haze also allowed for a decrease in strength and frequency in the post-operative steroid regimen, which decreased post-operative epithelial healing time and aberrant healing issues. This increased reduction of HOA also seemed to lead to a faster visual rehabilitation time for patients. Many patients were achieving good, functional vision within 7–10 days post-operatively allowing for more rapid return to work and driving. Corneal cross linking was always performed after laser correction, and thus was always “epithelial off”. This has a deeper corneal penetration and depth of corneal stiffening than “epithelial on”. Past studies have shown that the depth of epithelial off cross linking is approximately double that of epithelial on,17,18 and it is our practice never to do epithelial on cross-linking on any patient due to the limited depth of structural strengthening. Reduction of average K, Kmax, and manifest astigmatism were all useful measures, but reduction of K is skewed by the large differences of K due to corneal irregularity, and reduction of manifest astigmatism is skewed by the limitation of WaveLight Contoura to treating astigmatism only up to 3 diopters as well as by the amount of tissue available for laser ablation. We believe that the most important measurements are the reduction of HOA and increase in BCVA, and secondarily the vision outcomes. It is reasonable to question why the amount of HOA reduction was not even greater in some patients, and we believe at this time the limitation is due to the maximum of 3 diopters of astigmatism treatable with Contoura. As we have shown in the original LYRA Protocol publications, the WaveLight Contoura laser ablation pattern for HOA and astigmatism correction corresponded closely with the anterior elevation irregularity as measured on scheimpflug devices. If the full amount of the measured astigmatism is not treated, the full amount of anterior elevation is not corrected, thereby leaving residual higher order aberration. In keratoconus patients, the main limiting factor is the tissue availability to ensure a 350 µm bed for corneal cross-linking, but in some eyes the limiting factor was the 3 diopter astigmatism treatment limit. Another limiting factor in treatment was the arbitrary limitation of Contoura treatment to 6.0 mm and 6.5 mm treatment zones as utilizing a 5.0 mm treatment zone would allow for greater irregularity reduction. It has long been confusing to us why treatment zones not included in the FDA approval, nor why they are not available to the surgeon with a precaution as is allowed for Wavefront Optimized corrections. Case 1 shows the treatment of a 49 y.o. male with keratoconus, more severe on the right eye which was chosen for this example (Figure 1A–E). Such a patient typically has keratoconus progress in their teens and 20’s, but stabilizes as they age. This patient had significant issues with scleral lens fit and complained of compromised vision even though the patient could achieve 20/30 BCVA. This patient had a 48.7% reduction in HOA, with a resultant 1 line improvement in BCVA to 20/25 (Figure 1E). Treatment included no spherical correction, but normalization of the cornea resulted in significant reduction in sphere by flattening the cone in the visual axis area. The resulting small amount of manifest astigmatism correlates with the small amount of cylinder left untreated due to the limits of Contoura of 3 diopters of cylinder treatment. The patient anecdotally reported increased vision and dramatically increased quality of vision which he described as the best since he was a teenager. Case 2 is a 32 year old male that had bilateral LASIK performed in 2012 (Figure 2A–E). After being diagnosed with lasik induced corneal ectasia in the right eye, he traveled to a surgeon experienced with Intacs for Keratoconus, and underwent Intacs placement with epithelial-on crosslinking in 2019 of his right eye. This procedure had a suboptimal visual outcome, so the patient had the Intacs removed several months later. He then had a surgeon perform refractive wavefront PRK with a VISX laser removing approximately 70 µm of tissue. Incidentally, since epithelial on cross-linking only strengthens the approximately 125 µm of the anterior cornea, a large portion of the strengthened tissue was removed by this procedure. The patient noted progressive worsening of his vision after this procedure, and presented to us with the LASIK induced ectasia in Figure 2 with multiple scars from the prior procedures. This patient had one of the largest amounts of epithelial compensation at 40 µm, but treatment demonstrates a cornea with a significant reduction of the irregularity on topography which correlates with the 64% reduction in HOA. The patient was limited in his post-operative BCVA by severe dry eye that was likely caused by the multiple procedures he had undergone. This was in the process of relatively successful treatment as of their last visit. The topographic outcome demonstrates a cornea that has had the majority of the cone normalized, which corresponds to the 65% reduction of HOA. The patient had epithelial off cross-linking performed subsequent to the transepithelial topography guided ablation, which would allow for the maximum depth of cross-linking. No progression has been noted since this procedure, and the patient anecdotally reports the best vision he has had since the ectasia was diagnosed. Case 3 is the classic young person with progressive keratoconus, a 20 year old female with bilateral disease and the case example is of the more severely affected of the two eyes (Figure 3A–E). Of interest is her younger brother also had progressive severe keratoconus, highlighting the genetic nature of this disease. This patient had a significant reduction of her cone on topography while having only a 37% reduction of HOA, a low number for this study, but yet had a dramatic improvement of her BCVA from 20/70 to 20/25 (Figure 3E). This correlates with what we observed in our earlier study, that HOA reduction does not correlate directly with BCVA improvement.9 It may well be that the cornea in the central visual axis greater improvement than the overall HOA reduction number, thus the significant improvement in vision. The right eye (OU) vision result was 20/20, allowing this young lady to have the vision to resume her education. No patient required or desired scleral lens fitting post-op. Some did wear glasses to correct post-operative refraction, but most chose to wear these glasses occasionally rather than regularly. This corresponds with the HOA reduction, as patient’s quality of vision would improve to where the LOA was the more prominent issue, so spectacles could correct the patient satisfactorily. Some patients did have slow epithelial healing issues. As haze became significantly less of an issue by terminating use of the mitomycin-C, steroid was used in lesser strengths and lower dosing which helped improved epithelial healing response. Slow healing epithelium responded well to temporary halting of the steroid, lubrication, and bandage lens replacement. Haze was not a significant problem in these patients even the steroid use was suspended. The enrollment of patients in this retrospective study was limited by the number of out-of-town patients, as comprehensive follow-up to obtain the requisite data was not available for many patients. Only limited visual and anecdotal data as reported by patients and their local ophthalmologists was available, and seemed to match the data measured in our center. We currently have no evidence or report of negative outcomes such as BCVA loss, significant haze, or other serious issues related to these procedures save one patient that ended up with a severe corneal ulcer during travel back home after the epithelial abrasion had already closed and bandage lens removed. Since we did not have the follow up information for this patient, we did not include them in this study. Keratoconus/corneal ectasia affects a small, but significant percentage of the population, but the potential visual debilitation can have a massive impact on the patient’s life, their career, and even their social interactions. Many patients report the inability to maintain a job, go to school, or drive. The goal for these patients is a permanent treatment that allows them to keep their own cornea but with significant visual rehabilitation, and the prevention of progression of the disease. The long desired ultimate goal for these patients is also to rehabilitate their vision to the point where they can lead “normal” lives with good vision that does not interfere with their lives or chosen careers. The results of the CREATE+CXL Protocol indicate that we are close to achieving this long desired goal. We realize this is a relatively small sample, but the overall results have been consistently good even with patients we do not have a full complement of data to include in this manuscript. We plan to continue to accumulate data and analyze a larger group of patients in a future manuscript. Acknowledgments The author would like to thank Julie Crider, PhD for editing contributions, and Gerardo Lozano and Charlene Lloyd for data collection and editing contributions. Abbreviations BCVA, best-corrected visual acuity; CXL, Corneal cross linking; D, diopters; ETM, epithelial thickness maps; FDA, Food and Drug Administration; HHS, Health and Human Services; HOAs, higher-order aberrations; MMC, mitomycin C; OD, right eye; OS, left eye; PRK, photorefractive keratectomy; qd, daily; qid, four times daily; RGP, rigid gas permeable; UCVA, uncorrected visual acuity; US, United States. Disclosure The author has been granted United States patent no. 10857032 concerning the creating of a more uniform cornea utilizing the topography measured astigmatism, and United States patent no. 10857033 for the treatment of epithelial compensation of corneal irregularity in conjunction with the use of topography guided ablation system. The author reports no other conflicts of interest in this work. ==== Refs References 1. Sorkin N, Varssano D. Corneal collagen crosslinking: a systematic review. Ophthalmologica. 2014;232 (1 ):10–27. doi:10.1159/000357979 24751584 2. Boyd K. Keratoconus diagnosis and treatment. American Academy of Ophthalmology; 2020. Available from: https://www.aao.org/eye-health/diseases/keratoconus-diagnosis. Accessed January 22, 2021. 3. Gomez-Benlloch A, Montesel A, Pareja-Arico L, et al. Causes of corneal transplant failure: a multicentric study. Acta Ophthalmol. 2021;99 (6 ):e922–e928. doi:10.1111/aos.14708 33421330 4. Lains I, Rosa AM, Guerra M, et al. Irregular astigmatism after corneal transplantation--efficacy and safety of topography-guided treatment. Cornea. 2016;35 (1 ):30–36. doi:10.1097/ICO.0000000000000647 26509761 5. Kanellopoulos AJ, Binder PS. Management of corneal ectasia after LASIK with combined, same-day, topography-guided partial transepithelial PRK and collagen cross-linking: the Athens protocol. J Refract Surg. 2011;27 (5 ):323–331. doi:10.3928/1081597X-20101105-01 21117539 6. Motwani M. The use of WaveLight® Contoura to create a uniform cornea: the LYRA Protocol. Part 3: the results of 50 treated eyes. Clin Ophthalmol. 2017;11 :915–921. doi:10.2147/OPTH.S133841 28553071 7. Motwani M. The use of WaveLight® Contoura to create a uniform cornea: the LYRA Protocol. Part 2: the consequences of treating astigmatism on an incorrect axis via excimer laser. Clin Ophthalmol. 2017;11 :907–913. doi:10.2147/OPTH.S133840 28553070 8. Motwani M. The use of WaveLight® Contoura to create a uniform cornea: the LYRA Protocol. Part 1: the effect of higher-order corneal aberrations on refractive astigmatism. Clin Ophthalmol. 2017;11 :897–905. doi:10.2147/OPTH.S133839 28553069 9. Motwani M. Treatment of keratoconus with WaveLight Contoura and corneal cross-linking combined. Clin Ophthalmol. 2021;15 :2455–2472. doi:10.2147/OPTH.S303559 34163129 10. Reinstein DZ, Archer TJ, Gobbe M. Refractive and topographic errors in topography-guided ablation produced by epithelial compensation predicted by 3D Artemis VHF digital ultrasound stromal and epithelial thickness mapping. J Refract Surg. 2012;28 (9 ):657–663. doi:10.3928/1081597X-20120815-02 22947295 11. Kanellopoulos AJ, Asimellis G. Novel Placido-derived topography-guided excimer corneal Normalization With cyclorotation adjustment: enhanced Athens protocol for keratoconus. J Refract Surg. 2015;31 (11 ):768–773. doi:10.3928/1081597X-20151021-06 26544565 12. Reinstein DZ, Archer TJ, Vida RS. Epithelial thickness mapping for corneal refractive surgery. Curr Opin Ophthalmol. 2022;33 (4 ):258–268. doi:10.1097/ICU.0000000000000867 35779050 13. Motwani M. A protocol for topographic-guided corneal repair utilizing the US Food and Drug Administration-approved Wavelight Contoura. Clin Ophthalmol. 2017;11 :573–581. doi:10.2147/OPTH.S127855 28356712 14. Motwani M. Epithelial Compensation of Corneal Higher-Order Aberrations and Its Effect on Topographic Guided LASIK Outcomes as Demonstrated by OCT Epithelial Thickness Mapping. ASCRS; 2020. 15. Motwani M. Analysis and causation of all inaccurate outcomes after WaveLight Contoura LASIK with LYRA protocol. Clin Ophthalmol. 2020;14 :3841–3854. doi:10.2147/OPTH.S267091 33223821 16. Awwad ST, Chacra LM, Helwe C, et al. Mitomycin C application after corneal cross-linking for keratoconus increases stromal haze. J Refract Surg. 2021;37 (2 ):83–90. doi:10.3928/1081597X-20201124-01 33577693 17. Yuksel E, Cubuk MO, Yalcin NG. Accelerated epithelium-on or accelerated epithelium-off corneal collagen cross-linking: contralateral comparison study. Taiwan J Ophthalmol. 2020;10 (1 ):37–44. doi:10.4103/tjo.tjo_11_19 32309123 18. Mazzotta C, Ramovecchi V. Customized epithelial debridement for thin ectatic corneas undergoing corneal cross-linking: epithelial island cross-linking technique. Clin Ophthalmol. 2014;8 :1337–1343. doi:10.2147/OPTH.S66372 25114495
PMC010xxxxxx/PMC10352125.txt
==== Front J Exp Pharmacol J Exp Pharmacol jep Journal of Experimental Pharmacology 1179-1454 Dove 416673 10.2147/JEP.S416673 Original Research OCE-205, a Selective V1a Partial Agonist, Reduces Portal Pressure in Rat Models of Portal Hypertension Bukofzer et al Bukofzer et al http://orcid.org/0009-0007-5545-8210 Bukofzer Stan 1 Harris Geoffrey 2 Song Susan 2 Cable Edward E 2 1 Ocelot Bio, Inc., San Diego, CA, USA 2 Ferring Research Institute Inc., San Diego, CA, USA Correspondence: Stan Bukofzer, Ocelot Bio, Inc., 12670 High Bluff Drive, San Diego, CA, 92130, USA, Tel +1 858-247-2272, Email [email protected] 13 7 2023 2023 15 279290 12 4 2023 04 7 2023 © 2023 Bukofzer et al. 2023 Bukofzer et al. https://creativecommons.org/licenses/by-nc/3.0/ This work is published and licensed by Dove Medical Press Limited. The full terms of this license are available at https://www.dovepress.com/terms.php and incorporate the Creative Commons Attribution – Non Commercial (unported, v3.0) License (http://creativecommons.org/licenses/by-nc/3.0/). By accessing the work you hereby accept the Terms. Non-commercial uses of the work are permitted without any further permission from Dove Medical Press Limited, provided the work is properly attributed. For permission for commercial use of this work, please see paragraphs 4.2 and 5 of our Terms (https://www.dovepress.com/terms.php). Purpose Management of decompensated cirrhosis may include the use of vasoconstrictors that can lead to serious adverse events. OCE-205 was designed as a highly selective V1a receptor partial agonist, intended to have a wider therapeutic window than full vasopressin agonists. Methods We aimed to characterize the activity of OCE-205 treatment in two rat models of portal hypertension (PHT). For both models, OCE-205 was administered as a subcutaneous bolus injection. Thirty male Wistar rats were fed a methionine/choline-deficient (MCD) diet to model PHT. Animals received OCE-205 (10, 25, 100, or 500 µg/kg) or intra-arterial terlipressin (100 µg/kg). In a more severe model of PHT, 11 male Sprague Dawley rats had the common bile duct surgically ligated (BDL) and received OCE-205. Portal pressure (PP) and mean arterial pressure (MAP) were measured. Results For PP in the MCD model, MAP increased while PP decreased in rats treated with OCE-205 or terlipressin; the peak changes to MAP were 14.7 and 33.5 mmHg, respectively. Changes in MAP began to plateau after 10 min in the OCE-205 groups, whereas in the terlipressin group, MAP rapidly increased and peaked after 20 min. Across all treatment groups in the BDL model, a dose-related decrease from baseline in PP was observed following OCE-205, plateauing as the dose increased. In all treatment groups, PP change remained negative throughout the 30-min testing period. In both PHT rat models, a reduction in PP was coupled to an increase in MAP, with both plateauing in dose–response curves. Conclusion Data support OCE-205 as a promising candidate for further development. Institutional Protocol Number Procedures were approved by the Ferring Research Institute (FRI) Institutional Animal Care and Use Committee on July 13, 2011, under protocol FRI-07-0002. Keywords mean arterial pressure ascites HRS-AKI vasoconstriction portal hypertension Ferring Pharmaceuticals Inc., and publication of the data was supported by Ocelot Bio, Inc The original study was supported by Ferring Pharmaceuticals Inc., and publication of the data was supported by Ocelot Bio, Inc. ==== Body pmcIntroduction Decompensated cirrhosis and its serious hemodynamic complications are a critical area unmet need for patients. Systemic hemodynamic complications are typical of portal hypertension (PHT) and cirrhosis.1–4 PHT produces alterations in cardiovascular function and tone, with splanchnic arterial vasodilation, reduced systemic vascular resistance, and lower effective arterial blood volume along with arterial pressure reductions.5 Effective hypovolemia, low perfusion pressure, and reduced glomerular filtration rate can lead to vasoconstriction within the kidney with subsequent sodium and water retention, resulting in hepatorenal syndrome–acute kidney injury (HRS-AKI).5,6 HRS-AKI treatment relies on increasing blood volume with albumin supplementation and discontinuing diuretics with the hope of increasing renal perfusion to restore renal function. Often, short-duration therapy with systemic vasoconstrictors is needed to increase mean arterial pressure (MAP; 10–15 mmHg, which correlates with reversal of HRS-AKI7,8) to counteract renal dysfunction. Ultimately, the therapeutic goal is liver transplantation.9 Vasopressin agonists are among the best-characterized vasoconstrictors for the management of HRS-AKI. Arginine vasopressin (AVP), a key hormonal regulator of osmotic balance, is typically synthesized in the hypothalamus and released by the posterior pituitary gland.10 The V2 receptor, expressed in the distal tubules and collecting ducts of the kidney, principally regulates vasopressin and, once activated, causes water retention.10–12 Vasopressin activates the V1a receptor at elevated physiologic and pharmacologic concentrations; it then results in systemic vasoconstriction due to V1a expression on smooth muscle cells in the vasculature walls,10 including those in the splanchnic circulation.6 Vasopressin also activates the V1b receptor, expressed in the anterior pituitary; once activated, this stimulates corticotropin secretion, further increasing water retention.13 Current vasopressin agonists have a narrow therapeutic window. Underdosing does not achieve the desired clinical effect, and overdosing can produce excess vasoconstriction and potentially lead to life-threatening ischemic adverse events (AEs).10,12,14,15 All clinically utilized vasopressin agonists target the V2 agonists as well as the V1a receptor.16 The agonist activity at the V2 receptor contributes to the known AE profile, involving fluid overload and respiratory complications.12 The combination of fluid retention and risk of excess vasoconstriction currently limits the utilization of existing vasopressin agonists to relatively late in the development of HRS-AKI and other cirrhotic complications where the risk/benefit can be justified. Historically, HRS-AKI treatment has utilized midodrine plus octreotide, norepinephrine,17,18 and, most recently, terlipressin. Norepinephrine and midodrine, as α-adrenergic agonists, can have limited therapeutic efficacy.19 Furthermore, norepinephrine administration requires a central venous line, typically given in an intensive care unit.9 Terlipressin was approved by the US Food and Drug Administration (FDA) in September 2022 to treat adults with HRS with rapid reduction in kidney function.20,21 Terlipressin (triglycyl-LVP) is a prodrug of lysine vasopressin (LVP, the porcine version of AVP) and shows activity at the V1a, V1b, and V2 receptors.22–25 Its systemic hemodynamic response has been associated with a decrease in portal pressure (PP) as well as lessening the hyperdynamic circulation without additional contraction of the arterial and central blood volume.26,27 The active metabolite of triglycyl-LVP is LVP, a full V1a agonist that is significantly more potent at the V2 receptor.22 Terlipressin is recommended for treatment of HRS-AKI,2,9 but can cause clinically significant AEs (eg, tissue hypoperfusion, ischemia) from its vasoconstrictive effects, likely due to LVP’s full agonism at the V1a receptor,14,28,29 as well as respiratory failure and fluid overload through V1a and V2 receptor activation, possibly related to sodium and water retention.23,30–33 An improved approach to targeting the V1a receptors with partial agonism (avoiding maximal stimulation), and no V2 receptor agonism, would lower the risks associated with vasopressin agonists14,28–30 and could offer therapeutic utility in conditions where a modest and capped increase in blood pressure is desirable. To improve the safety profile of vasoconstriction through the vasopressin system in patients with HRS-AKI, a molecule with V1a receptor selectivity and minimal V1a receptor agonism is needed. A single molecule was proposed in 1996 that, by possessing both agonist and antagonist properties, would perform as a partial agonist.34 OCE-205 is a novel peptide drug designed to target the vasopressin receptor system as a mixed agonist/antagonist for the V1a receptor. Unlike terlipressin,22–25 OCE-205 is not a prodrug and does not require a liver first pass or other modifications in vivo to produce the desired pharmacological effects. The activity at human V1a receptors plateaus at ~50% maximum possible effect, with no activity at human V2 receptors at clinically relevant concentrations.35 OCE-205 has similar functional properties at rat and human vasopressin receptors; by examining the use of OCE-205 to reduce PHT and increase MAP in two rat models, we hope to pave the way for improvements in clinical safety and efficacy in human patients with HRS-AKI. Materials and Methods Animal Use Housing conditions and animal care facilities were maintained in accordance with the Guide for the Care and Use of Laboratory Animals published by the National Research Council. Procedures were approved by the Ferring Research Institute (FRI) Institutional Animal Care and Use Committee (IACUC) on July 13, 2011, under protocol FRI-07-0002. Rat Models Methionine/Choline-Deficient Model The MCD model of PHT does not typically induce significant fibrosis nor development of cirrhosis, but leads to portal pressure (PP) elevation after 4 weeks of MCD.36–39 The MCD model of diet-induced PHT is clearly described elsewhere40 and is a considered a reliable model to induce PHT without being as severe and progressive as other interventions such as CCl4, or bile duct/portal vein ligation. The mechanism of inducing PHT is primarily from inducing a NAFLD/NASH-like phenotype with accumulation of intracellular lipid in hepatocytes. Presumably the increase in inflammation and congestion created by lipid accumulation increases hepatic resistance and therefore PP. Histological examination of the study animals were not performed as the livers were obviously steatotic upon macroscopic inspection. Thirty adult male Wistar rats (Harlan, Indianapolis) were studied over multiple testing days and weighed 200–250 g at the start of the study. Animals were housed two per cage in a controlled environment with free access to an MCD diet (Harlan Teklad) and water. After ≥8 weeks on the MCD diet, the adult male Wistar rats underwent surgical placement of catheters in the portal vein and/or femoral artery. Animals were allowed to recover for ≥5 days. A saline vehicle was administered 5 to 15 min prior to experimental compounds to control for nonspecific responses. OCE-205 was bolus administered subcutaneously at doses of 500, 100, and 25 µg/kg to determine the degree of systemic vasoconstriction achieved for a given degree of reduction in portal pressure and whether these occurred at similar, identical, or disparate drug concentrations. To avoid the need for a second surgery, terlipressin (100 µg/kg) was administered intra-arterially. Triglycyl-LVP (terlipressin) has been previously well characterized in both rats and dogs, at doses up to 50 µg/kg in combination with octreotide.41,42 Systolic blood pressure (SBP), diastolic blood pressure (DBP), and PP were measured by the fluidic transducers. Measurements were recorded continuously from time 0 (compound administration time) to 90 min following administration, unless technical difficulties (eg loss of catheter patency) resulted in cessation of the experiment prior to 90 min. To reduce the numbers of animals needed for analysis and as approved by the IACUC, 30 animals had repeat dosing with a minimum of 1 day of recovery between each administration. Altogether, 86 measurements with 1–8 repeat administrations per animal were used for this study. Bile Duct Surgically Ligated Model The BDL rat model typically causes larger increases in PHT than the MCD model, and significant renal dysfunction.36,43,44 Male Sprague Dawley rats had the common BDL to block the enterohepatic recirculation of bile acids. Eleven adult male Sprague Dawley rats (Harlan, Indianapolis) weighing 200–350 g were studied over multiple testing days. Animals were housed in a controlled environment with free access to food and water for ≥4 days after surgery and before experimentation. A laparotomy was performed, and the portal vein exposed, following the procedure described by Gervaz et al.45 The bifurcation of the portal vein and the confluence of the gastric vein were identified. Proximal clamping for ≤5 min was performed to minimize blood loss and was well tolerated. Puncture of the anterior wall of the portal vein was performed with a 30-gauge needle just proximal to its bifurcation. The tip of the catheter was cut at 45°, and its sharp edge was introduced tangentially into the portal vein. Care was taken to avoid inserting the tip of the catheter too far into the vein lumen. The catheter was exteriorized through the abdominal wall, tunneled subcutaneously, and exteriorized in the intrascapular region. The catheter was locked with 50% glycerol/heparin. A jacket was positioned to hold the exterior portion of the catheter in place. Animals were allowed to recover for ≥5 days before receiving OCE-205. To avoid the need for a second surgery, catheter placement and BDL were performed at the same time. The adult male Sprague Dawley rats had surgically placed portal vein catheters and BDL for an average of 25 days (range, 11–42 days). Vehicle (saline) was administered 5 to 15 min prior to experimental compound administration to control for nonspecific responses. No arterial measurements were taken due to bleeding issues in this more severe model of disease. OCE-205, administered as a subcutaneous bolus, was tested over a broader dose range (500, 100, 25, and 10 µg/kg) than in the MCD model, to establish a clearer dose–response relationship of changes in PP based on studies in healthy animals and evaluating changes in skin blood flow. The PP was measured via the fluidic transducers and typically recorded continuously from time 0 (compound administration time) to 30 min following administration by NOTOCORD-hem™ data acquisition software, unless technical difficulties (eg, loss of catheter patency) resulted in cessation of experiment prior to 30 min. To reduce the number of animals needed for analysis and as approved by the IACUC, 11 animals had repeat dosing with a minimum of 16 h of recovery between each administration. A total of 38 measurements with 1–11 repeat administrations per animal were used for this study. Test and Reference Compounds OCE-205–TFA salt was the test compound in both studies. Terlipressin (triglycyl-LVP) was the reference compound in the MCD study. Compound Formulation and Administration OCE-205 was administered in a dose volume of 1 mL/kg subcutaneously by injection (OCE-205) into the animal’s lower back. Terlipressin (triglycyl-LVP) was administered intra-arterially via the femoral catheter used to measure pressure. The OCE-205 doses were 500, 100, 25, and 10 µg/kg. The terlipressin (triglycyl-LVP) dose was 100 µg/kg. Subcutaneous Bolus Administration of Compounds Overview A minimum of four animals was used for testing each group to allow for the possibility of technical, and/or pharmacologic events. Most studied used each animal more than once. Animals were randomized to different testing stations (anesthesia stations and NOTOCORD™ software channels for automatic data acquisition). The investigators running the in-life portion of the studies were aware of the dosing solutions being administered. On the day of experimental measurements for each study, animals were anesthetized, and catheters were flushed to maintain or restore patency. Catheters were then connected to fluidic pressure transducers linked to data acquisition stations. Prior to administration of compound, a stable baseline was obtained for the pressure readings, saline was administered, and an additional 5–15 min of pressure readings were taken to solidify the baseline readings and provide a per-animal vehicle control. Animals were excluded from the study if the study cannular was not patent or if stable baseline values could not be obtained. Data Analysis Data recorded by NOTOCORD-hem™ were transferred to Microsoft Excel for analysis. Data were evaluated as change (Δ) from baseline. MAP and PP were reported as the average pressure value (mmHg) recorded over 10s beginning at the first systole (PP and SBP) or diastole (DBP) following the time point of interest at nominal times of 0 (compound administration), 1, 2, 3, 5, 7, 10, 15, 20, and 30 min after administration in both studies, along with 60 and 90 min after administration in the MCD study. Data collected over multiple test days were compiled. Mean, SEM, and N were reported for PP, MAP, ΔPP, and ΔMAP for each compound dose and time point. MAP was calculated from SBP and DBP: \documentclass[12pt]{minimal} \usepackage{wasysym} \usepackage[substack]{amsmath} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage[mathscr]{eucal} \usepackage{mathrsfs} \DeclareFontFamily{T1}{linotext}{} \DeclareFontShape{T1}{linotext}{m}{n} {linotext }{} \DeclareSymbolFont{linotext}{T1}{linotext}{m}{n} \DeclareSymbolFontAlphabet{\mathLINOTEXT}{linotext} \begin{document} $${\rm{MAP=}}\left[{\left({{\rm{2\times\,DBP}}}\right){\rm{+SBP}}}\right]{\rm{/3}}$$ \end{document} Change from baseline (delta Δ) for PP and MAP at each time point in each animal were calculated as follows: \documentclass[12pt]{minimal} \usepackage{wasysym} \usepackage[substack]{amsmath} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage[mathscr]{eucal} \usepackage{mathrsfs} \DeclareFontFamily{T1}{linotext}{} \DeclareFontShape{T1}{linotext}{m}{n} {linotext }{} \DeclareSymbolFont{linotext}{T1}{linotext}{m}{n} \DeclareSymbolFontAlphabet{\mathLINOTEXT}{linotext} \begin{document} $${\rm{\Delta\,PP=PP\ at\ each\ time-PP\ at\ baseline}}\ \left({{\rm{time\ 0}}} \right)$$ \end{document} \documentclass[12pt]{minimal} \usepackage{wasysym} \usepackage[substack]{amsmath} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage[mathscr]{eucal} \usepackage{mathrsfs} \DeclareFontFamily{T1}{linotext}{} \DeclareFontShape{T1}{linotext}{m}{n} {linotext }{} \DeclareSymbolFont{linotext}{T1}{linotext}{m}{n} \DeclareSymbolFontAlphabet{\mathLINOTEXT}{linotext} \begin{document} $${\rm{\Delta\,MAP=MAP\ at\ each\ time-MAP\ at\ baseline\ }}\left({{\rm{time\ 0}}} \right)$$ \end{document} Statistical Analysis The observed PP and MAP data were statistically analyzed using JMP software. Data from each animal at times 0 to 90 min (MCD study) and 0 to 30 min (BDL study) in the four treatment groups were compared using MANOVA with repeated measures. Each time course measurement was considered independent; animals were allowed to recover before any repeat administration. No corrections or extrapolations were performed if data were not collected for every time point. The MANOVA outcome for treatment between subjects was not considered significant if p ≥ 0.05 (Prob>F); in the MCD study, subsequent contrasts were performed to compare treatment groups. Data from time 0 were statistically determined using one-way ANOVA to compare starting values in each treatment group. In cases when this ANOVA was significant (p < 0.05), Tukey–Kramer HSD post hoc analysis was used to compare treatment groups. Results OCE-205 Modulates Disease Physiology in an MCD Model of PHT Mean Arterial Pressure OCE-205 and terlipressin, when given to MCD-fed rats, resulted in increased MAP. For the 25 μg/kg (n = 13), 100 μg/kg (n = 23), and 500 μg/kg (n = 17) OCE-205 treatment groups, the highest observed absolute MAP values (mean ± SEM) were 79.2 ± 1.4, 86.8 ± 2.5, and 90.4 ± 3.7 mmHg, respectively (Table 1). There were greater increases in MAP resulting from the intra-arterial administration of terlipressin (100 μg/kg; n = 13) compared with the OCE-205 treatment groups, with the highest observed MAP of 116.6 ± 4.1 mmHg.Table 1 Change in MAP from Baseline After OCE-205 or Terlipressin Treatment in the MCD Model Time (Min) OCE-205 Terlipressin MAP (mmHg) 25 μg/kg, SC 100 μg/kg, SC 500 μg/kg, SC 100 μg/kg, IA Mean SEM N Mean SEM N Mean SEM N Mean SEM N 0 73.9 1.7 13 75.5 1.6 23 75.6 1.9 17 83.0 2.3 13 1 75.0 1.7 13 77.9 1.7 22 77.9 2.1 17 89.1 2.8 13 2 75.8 1.7 13 79.4 1.8 22 80.1 2.1 17 93.8 3.1 13 3 76.5 1.6 13 80.3 1.7 22 82.0 2.1 17 101.1 3.9 13 5 77.4 1.6 13 83.0 1.8 22 84.7 2.3 17 109.3 4.5 13 7 77.5 1.6 12 84.6 2.1 21 87.7 2.8 17 113.1 4.5 13 10 78.7 1.5 13 85.8 2.1 22 90.4 3.7 17 115.0 4.3 13 15 79.2 1.4 13 86.5 2.3 23 89.3 3.8 17 116.0 4.2 13 20 79.1 1.2 13 86.8 2.5 23 90.0 3.8 16 116.6 4.1 13 30 78.3 1.3 13 86.3 2.5 20 87.0 3.3 17 114.7 3.9 13 60 76.9 1.2 13 86.8 3.6 17 84.8 2.5 17 96.3 3.2 13 90 76.8 1.3 12 85.2 3.2 17 86.1 2.4 16 88.3 2.6 13 Abbreviations: IA, intra-arterial; MAP, mean arterial pressure; SC, subcutaneous; SEM, standard error of the mean. Observed MAP using MANOVA with repeated measures of the OCE-205 treatment groups showed a significant difference between the 25 and 100 μg/kg OCE-205 treatment groups (p = 0.03). There was no significant difference between the 100 and 500 μg/kg OCE-205 treatment groups (p = 0.63). MAP was significantly different between intra-arterial administration of terlipressin and subcutaneous administration of OCE-205 at 500 μg/kg (p < 0.01). The maximum ΔMAP values (mean ± SEM) for OCE-205 were 5.3 ± 1.1, 11.3 ± 1.8, and 14.7 ± 2.6 mmHg in the 25, 100, and 500 μg/kg treatment groups, respectively (Figure 1). Following intra-arterial administration of terlipressin, the maximum ΔMAP (33.5 ± 2.6 mmHg) was greater than in any of the OCE-205 treatment groups. Figure 1 Change from baseline in mean arterial pressure following OCE-205 or terlipressin administration in a rat model of portal hypertension (methionine/choline-deficient diet). Abbreviations: IA, intra-arterial; SC, subcutaneous. Of note, the observed MAP baseline values (time 0) were statistically different for the terlipressin group versus each of the OCE-205 treatment groups. The values were determined by one-way ANOVA with Tukey–Kramer HSD post hoc analysis (p = 0.02 vs 25 μg/kg, p = 0.03 vs 100 µg/kg, p = 0.46 vs 500 μg/kg OCE-205). Portal Pressure There was a decrease in PP following treatment with OCE-205 or terlipressin. The observed PP values were not significantly different between any of the four treatment groups, as determined by MANOVA with repeated measures (Table 2). The maximum ΔPP (mean ± SEM) were −2.3 ± 0.2, −2.5 ± 0.3, and −3.9 ± 0.8 mmHg in the OCE-205 25, 100, and 500 μg/kg treatment groups, respectively (Figure 2). The change following intra-arterial administration of terlipressin was −2.8 ± 0.8 mmHg.Table 2 Change in PP from Baseline After OCE-205 or Terlipressin Treatment in the MCD Model Time (Min) OCE-205 Terlipressin PP (mmHg) 25 μg/kg, SC 100 μg/kg, SC 500 μg/kg, SC 100 μg/kg, IA Mean SEM N Mean SEM N Mean SEM N Mean SEM N 0 10.7 0.9 19 10.9 0.8 24 12.6 1.5 13 10.7 1.6 6 1 10.4 0.8 19 10.3 0.7 24 11.4 1.4 13 9.2 1.3 6 2 9.9 0.8 19 9.7 0.6 24 10.7 1.3 13 8.6 1.2 6 3 9.5 0.7 19 9.2 0.6 24 10.1 1.2 13 8.4 1.2 6 5 8.9 0.7 19 8.7 0.6 24 9.7 1.1 13 8.3 1.2 6 7 8.8 0.7 19 8.4 0.5 23 9.3 1.2 13 8.2 1.2 6 10 8.7 0.7 19 8.4 0.6 24 9.7 1.1 13 8.3 1.1 6 15 8.5 0.7 19 8.3 0.6 23 9.8 0.9 12 8.2 1.1 6 20 8.4 0.7 19 8.3 0.6 23 9.5 0.8 12 8.2 1.1 6 30 8.3 0.7 18 8.1 0.6 20 9.4 0.8 12 8.2 1.0 6 60 8.6 0.8 16 8.0 0.5 17 9.1 0.6 11 8.0 0.9 6 90 8.9 0.9 14 8.1 0.6 16 9.6 0.8 8 10.3 1.9 5 Abbreviations: IA, intra-arterial; PP, portal pressure; SC, subcutaneous; SEM, standard error of the mean. Figure 2 Change from baseline in portal pressure following OCE-205 or terlipressin administration in a rat model of portal hypertension (methionine/choline-deficient diet). Abbreviations: IA, intra-arterial; SC, subcutaneous. The observed PP values at baseline (time 0) were not statistically different between the treatment groups, as determined by one-way ANOVA. Tolerability No treatment-related deaths were observed after OCE-205 or terlipressin administration. OCE-205 Modulates Pathophysiology in a BDL Rat Model of More Severe Disease In the rats whose PHT was induced by BDL, OCE-205 treatment resulted in a decrease in PP from baseline. The observed PP values were not significantly different between any of the four treatment groups based on MANOVA with repeated measures (Table 3). Across all treatment groups, dose-related decreases from baseline in portal pressure (ΔPP) occurred following OCE-205 administration, with a plateau as the dose increased further (Figure 3).Table 3 Change in PP from Baseline After OCE-205 Treatment in the BDL Model Time (Min) OCE-205 Terlipressin PP (mmHg) 10 μg/kg, SC 25 μg/kg, SC 100 μg/kg, SC 500 μg/kg, SC Mean SEM N Mean SEM N Mean SEM N Mean SEM N 0 16.9 0.7 9 17.5 0.7 7 18.9 0.8 14 20.1 0.7 8 1 16.7 0.7 9 16.9 0.9 7 16.8 0.7 14 16.6 1.0 8 2 16.3 0.7 9 15.8 0.8 7 14.7 0.7 14 15.7 0.8 8 3 15.8 0.7 9 14.7 0.6 7 14.5 0.7 14 15.5 0.7 8 5 15.2 0.6 9 14.4 0.6 7 14.8 0.8 14 15.3 0.7 8 7 15.1 0.6 9 14.3 0.6 7 14.9 0.8 14 15.2 0.7 8 10 15.2 0.6 9 14.2 0.6 7 14.9 0.8 14 15.3 0.8 8 15 15.3 0.6 9 14.2 0.6 7 15.3 0.9 13 16.4 0.8 8 20 15.4 0.6 9 14.4 0.7 7 15.4 0.9 13 16.3 0.7 8 30 15.7 0.6 9 14.6 0.7 7 15.4 0.8 13 16.4 0.8 7 Abbreviations: IA, intra-arterial; PP, portal pressure; SC, subcutaneous; SEM, standard error of the mean. Figure 3 Change from baseline in portal pressure following subcutaneous (SC) OCE-205 administration in a rat model of bile duct ligation and portal hypertension. In the OCE-205 10 μg/kg (n = 9), 25 μg/kg (n = 7), 100 μg/kg (n = 14), and 500 μg/kg (n = 8) treatment groups, the maximum ΔPP (mean ± SEM) were −1.8 ± 0.2 mmHg at 7 min, −3.3 ± 0.2 mmHg at 10 min, −4.4 ± 0.2 mmHg at 3 min, and −4.9 ± 0.6 mmHg at 7 min, respectively. The decrease in PP remained present throughout the testing period (at 30 min) in all treatment groups: −1.2 ± 0.2, −2.9 ± 0.5, −3.4 ± 0.2, and −3.6 ± 0.7 mmHg, respectively. When comparing results for ΔPP, the observed PP values at baseline (time 0) were statistically different between the 10 μg/kg (16.9 ± 0.7 mmHg) and 500 μg/kg (20.1 ± 0.7 mmHg) OCE-205 treatment groups. This was determined by one-way ANOVA with Tukey–Kramer HSD post hoc analysis (p = 0.045). Tolerability No treatment-related deaths were observed after OCE-205 administration. Discussion In two rat models of portal hypertension, OCE-205 achieved the therapeutic goal of decreasing PP without causing excessive vasoconstriction over a broad dose range. The study clearly shows that modest but significant decreases in PP can be obtained with OCE-205. Studies in human blood vessels have demonstrated in vitro V1a selective activity and behavior akin to a partial agonist.35 Generating data on portal blood flow is much more challenging to accomplish in rodent models. Only two phenomena reduce PP: either splanchnic vasoconstriction restricts flow, or a change in hepatic resistance allows the same flow at decreased pressure. Vasopressin agonists exert their biological activity by increasing splanchnic resistance and reducing splanchnic blood flow. Here, we explored the potential therapeutic benefit of OCE-205 by demonstrating that it reduces PP in animals with PHT, and its safety by confirming a plateau effect where no further reductions in PP occur and there are no further increases in MAP compared with other known vasopressin analogs. In the MCD model, which does not induce significant fibrosis nor development of cirrhosis but does lead to PP elevation,36,38,39,46 MAP increased within the desired treatment window, and plateaued over time and with increasing dose. This systemic increase was coupled with a significant reduction in PP. In contrast, administration of the full V1a receptor agonist terlipressin (triglycyl-LVP) resulted in increases in MAP that were beyond the desired treatment window of 10–15 mmHg. PP reductions in this group were similar to those seen with OCE-205. Although the recorded MAP baseline values were statistically different between the terlipressin (triglycyl-LVP) versus OCE-205 treatment groups, this observation could be attributed to inter-animal variability and smaller sample size for the terlipressin group (n = 6). The differences observed in baseline PP values in the 10 µg/kg versus 500 µg/kg groups could have been similarly affected. In the BDL rat model, which causes more profound PHT, and significant renal dysfunction,43,44 OCE-205 reduced PP with capped maximal vasoconstrictive activity that is consistent with a partial agonist-like effect. The OCE-205 molecule was designed to widen the therapeutic window through an innovative combination of both a selective V1a agonist domain and a selective V1a antagonist domain that together achieve effective partial agonism. While this unique design allows the molecule to bind to the receptor in either orientation, any one molecule binds to one receptor at a time. Binding of the agonist domain to V1a receptors drives the desired vasoconstrictive effect, while binding of the antagonist domain in a competitive manner prevents maximal activation of the V1a system. Cell-based functional assays of OCE-205 support both its function as an effective partial agonist and its selectivity for the V1a receptor, with no activity at the human or rat V2 receptor at therapeutic concentrations.35 In healthy animals, this partial agonist mechanism was demonstrated, with a plateau effect achieving submaximal increases in MAP that were driven by similar increases in SBP and DBP (Bukofzer et al, “OCE-205 in Rats and Non-Human Primates: Pharmacokinetic and Pharmacodynamic Analysis”, submitted manuscript). Without treatment, patients with HRS-AKI have a mortality of ~80% at 3 months, with a median survival of only 4 weeks.47 Furthermore, almost one-third of patients with HRS-AKI are readmitted to hospital within 30 days after initial hospitalization.48 While the current EASL guidelines recommend the use of terlipressin plus albumin,9 and the US FDA recently approved terlipressin for patients with HRS with a rapid reduction in kidney function,49 the observed side effects limit its use. Despite significantly improving renal function, terlipressin use is associated with serious AEs, including gastrointestinal disorders, and respiratory failure.50 LVP, the active metabolite of terlipressin (triglycyl-LVP), has activity at V1a, V1b, and V2 receptors22,23 and is both a full V1a and V2 receptor agonist.22 The known AE profile of terlipressin is likely related to the full V1a14,28,29 agonist activity of the primary metabolite LVP (which is responsible for the known pharmacology of terlipressin), leading to excessive vasoconstriction and significant V2 receptor activity which could contribute to further water retention.23,30 Because of its selective V1a functional partial agonist activity, OCE-205 could further benefit patients with HRS-AKI through increasing MAP by ~10–15 mmHg, within the desired treatment window, without the development of ischemic or pulmonary complications. OCE-205 may have potential for more generalizable utility for other complications of decompensated cirrhosis, such as for resistant and refractory ascites and post-paracentesis–induced circulatory dysfunction, because of similar underlying mechanisms of renal dysfunction. Because of its innovative agonist/antagonist design, the selective binding of OCE-205 to V1a receptors should result in effective partial receptor agonism and limit maximal vasoconstriction. These properties could eliminate the excessive vasoconstriction and associated adverse effects observed with a full V1a receptor. Further studies utilizing OCE-205 in patients with refractory ascites are being planned. Neither of these rat disease models fully replicates human liver disease, but rather reproduces the hemodynamic and fluid imbalances observed in patients with cirrhosis. Moreover, chronic liver diseases progress over many years in humans, whereas in rats the disease progression occurs over weeks or months. Rodent metabolism is much faster than human metabolism, which could lead to issues in replicating liver diseases related to metabolism (eg, NAFLD).51 Because animal models cannot completely predict the response in humans or replicate all the features of human liver disease, clinical trials in humans are needed to confirm the potential of OCE-205. Conclusion In our study, OCE-205, a novel peptide drug, led to predictable increases in MAP within the desired range for efficacy and safety while decreasing PP in both rat models. Its wide dose range, selective V1a receptor activation, and vasoconstrictive effects that elicit the target MAP increase suggest that OCE-205 is a promising candidate for further development to treat HRS-AKI. The first Phase 2 study is ongoing in patients with cirrhosis and ascites who develop HRS-AKI (NCT05309200). Acknowledgments Medical writing and editorial assistance was provided by Innovative Strategic Communications (Milford, PA) and by Richard Perry, PharmD, both funded by Ocelot Bio, Inc. Disclosure SB is the founding Chief Scientific and Medical Officer of Ocelot Bio, Inc. GH is a founder of and consultant to Ocelot Bio, Inc., and has multiple draft patents pending for V1a agonists. GH, SS, and EC were employees of Ferring Research Institute Inc., at the time of the study. The authors report no other conflicts of interest in this work. ==== Refs References 1. Tellez L, Guerrero A. Management of liver decompensation in advanced liver disease (renal impairment, liver failure, adrenal insufficiency, cardiopulmonary complications). Clin Drug Investig. 2022;42 (Suppl 1 ):15–23. doi:10.1007/s40261-022-01149-3 2. Biggins SW, Angeli P, Garcia-Tsao G, et al. Diagnosis, evaluation, and management of ascites, spontaneous bacterial peritonitis and hepatorenal syndrome: 2021 practice guidance by the American association for the study of liver diseases. Hepatology. 2021;74 (2 ):1014–1048. doi:10.1002/hep.31884 33942342 3. Smith A, Baumgartner K, Bositis C. Cirrhosis: diagnosis and management. Am Fam Physician. 2019;100 (12 ):759–770.31845776 4. Moller S, Bendtsen F. The pathophysiology of arterial vasodilatation and hyperdynamic circulation in cirrhosis. Liver Int. 2018;38 (4 ):570–580. doi:10.1111/liv.13589 28921803 5. Ginès P, Solà E, Angeli P, Wong F, Nadim MK, Kamath PS. Hepatorenal syndrome. Nat Rev Dis Primers. 2018;4 (1 ):23. doi:10.1038/s41572-018-0022-7 30213943 6. Harper D, Chandler B. Splanchnic circulation. BJA Educ. 2015;16 (2 ):66–71. doi:10.1093/bjaceaccp/mkv017 7. Velez JCQ. Hepatorenal syndrome type 1: from diagnosis ascertainment to goal-oriented pharmacologic therapy. Kidney360. 2022;3 (2 ):382–395. doi:10.34067/KID.0006722021 35373127 8. Velez JC, Nietert PJ. Therapeutic response to vasoconstrictors in hepatorenal syndrome parallels increase in mean arterial pressure: a pooled analysis of clinical trials. Am J Kidney Dis. 2011;58 (6 ):928–938. doi:10.1053/j.ajkd.2011.07.017 21962618 9. European Association for The Study of the Liver. EASL clinical practice guidelines for the management of patients with decompensated cirrhosis. J Hepatol. 2018;69 (2 ):406–460. doi:10.1016/j.jhep.2018.03.024 29653741 10. Treschan TA, Peters J, Warltier D. The vasopressin system: physiology and clinical strategies. Anesthesiology. 2006;105 (3 ):599. doi:10.1097/00000542-200609000-00026 16931995 11. Barberis C, Mouillac B, Durroux T. Structural bases of vasopressin/oxytocin receptor function. J Endocrinol. 1998;156 (2 ):223–229. doi:10.1677/joe.0.1560223 9518866 12. Koshimizu TA, Nakamura K, Egashira N, Hiroyama M, Nonoguchi H, Tanoue A. Vasopressin V1a and V1b receptors: from molecules to physiological systems. Physiol Rev. 2012;92 (4 ):1813–1864. doi:10.1152/physrev.00035.2011 23073632 13. Albert RE, Smith WW, Eichna LW. Hemodynamic changes associated with fluid retention induced in noncardiac subjects by corticotropin (ACTH) and cortisone; comparison with the hemodynamic changes of congestive heart failure. Circulation. 1955;12 (6 ):1047–1056. doi:10.1161/01.cir.12.6.1047 13270361 14. Donnellan F, Cullen G, Hegarty JE, McCormick PA. Ischaemic complications of glypressin in liver disease: a case series. Br J Clin Pharmacol. 2007;64 (4 ):550–552. doi:10.1111/j.1365-2125.2007.02921.x 17880357 15. Gulberg V, Bilzer M, Gerbes AL. Long-term therapy and retreatment of hepatorenal syndrome type 1 with ornipressin and dopamine. Hepatology. 1999;30 (4 ):870–875. doi:10.1002/hep.510300430 10498636 16. Glavas M, Gitlin-Domagalska A, Debowski D, Ptaszynska N, Legowska A, Rolka K. Vasopressin and its analogues: from natural hormones to multitasking peptides. Int J Mol Sci. 2022;23 (6 ):3068. doi:10.3390/ijms23063068 35328489 17. Karwa R, Woodis CB. Midodrine and octreotide in treatment of cirrhosis-related hemodynamic complications. Ann Pharmacother. 2009;43 (4 ):692–699. doi:10.1345/aph.1L373 19299324 18. Israelsen M, Krag A, Allegretti AS, et al. Terlipressin versus other vasoactive drugs for hepatorenal syndrome. Cochrane Database Syst Rev. 2017;9 :CD011532. doi:10.1002/14651858.CD011532.pub2 28953318 19. Smythe CM, Nickel JF, Bradley SE. The effect of epinephrine (USP), L-epinephrine, and L -norepinephrine on glomerular filtration rate, renal plasma flow, and the urinary excretion of sodium, potassium, and water in normal man. J Clin Investig. 1952;31 (5 ):499–506. doi:10.1172/JCI102634 14927740 20. Mallinckrodt Pharmaceuticals. TERLIVAZ [terlipressin] for injection. 2022; Available from: https://www.terlivaz.com/PI/. 21. U.S. Food & Drug Administration. FDA approves treatment to improve kidney function in adults with hepatorenal syndrome; 2022. Available from: https://www.fda.gov/drugs/news-events-human-drugs/fda-approves-treatment-improve-kidney-function-adults-hepatorenal-syndrome. Accessed July 5, 2023. 22. Jamil K, Pappas SC, Devarakonda KR. In vitro binding and receptor-mediated activity of terlipressin at vasopressin receptors V1 and V2. J Exp Pharmacol. 2018;10 :1–7. doi:10.2147/JEP.S146034 29302194 23. Colson PH, Virsolvy A, Gaudard P, et al. Terlipressin, a vasoactive prodrug recommended in hepatorenal syndrome, is an agonist of human V1, V2 and V1B receptors: implications for its safety profile. Pharmacol Res. 2016;113 (Pt A ):257–264. doi:10.1016/j.phrs.2016.08.027 27586252 24. Ouattara A, Landi M, Le Manach Y, et al. Comparative cardiac effects of terlipressin, vasopressin, and norepinephrine on an isolated perfused rabbit heart. Anesthesiology. 2005;102 (1 ):85–92. doi:10.1097/00000542-200501000-00016 15618791 25. Henderson KK, Byron KL. Vasopressin-induced vasoconstriction: two concentration-dependent signaling pathways. J Appl Physiol. 2007;102 (4 ):1402–1409. doi:10.1152/japplphysiol.00825.2006 17204577 26. Moller S, Hansen EF, Becker U, Brinch K, Henriksen JH, Bendtsen F. Central and systemic haemodynamic effects of terlipressin in portal hypertensive patients. Liver. 2000;20 (1 ):51–59. doi:10.1034/j.1600-0676.2000.020001051.x 10726961 27. Freeman JG, Barton JR, Record CO. Haemodynamic responses to 1.25 and 2 mg of terlipressin intravenously in man. Aliment Pharmacol Ther. 1988;2 (4 ):361–367. doi:10.1111/j.1365-2036.1988.tb00709.x 2979260 28. Sarma P, Muktesh G, Singh RS, et al. Terlipressin-induced ischemic complications: a systematic review of published case reports. J Pharmacol Pharmacother. 2018;9 (2 ):76–85. doi:10.4103/jpp.JPP_23_18 29. Boyer TD, Sanyal AJ, Wong F, et al. Terlipressin plus albumin is more effective than albumin alone in improving renal function in patients with cirrhosis and hepatorenal syndrome type 1. Gastroenterology. 2016;150 (7 ):1579–1589 e2. doi:10.1053/j.gastro.2016.02.026 26896734 30. Krag A, Bendtsen F, Pedersen EB, Holstein-Rathlou NH, Moller S. Effects of terlipressin on the aquaretic system: evidence of antidiuretic effects. Am J Physiol Renal Physiol. 2008;295 (5 ):F1295–F1300. doi:10.1152/ajprenal.90407.2008 18753292 31. Messmer AS, Zingg C, Müller M, Gerber JL, Schefold JC, Pfortmueller CA. Fluid overload and mortality in adult critical care patients-a systematic review and meta-analysis of observational studies. Crit Care Med. 2020;48 (12 ):1862–1870. doi:10.1097/ccm.0000000000004617 33009098 32. Schrier RW, Wang W. Acute renal failure and sepsis. N Engl J Med. 2004;351 (2 ):159–169. doi:10.1056/NEJMra032401 15247356 33. Griffin BR, Liu KD, Teixeira JP. Critical care nephrology: core curriculum 2020. Am J Kidney Dis. 2020;75 (3 ):435–452. doi:10.1053/j.ajkd.2019.10.010 31982214 34. Zhu BT. Rational design of receptor partial agonists and possible mechanisms of receptor partial activation: a theory. J Theor Biol. 1996;181 (3 ):273–291. doi:10.1006/jtbi.1996.0130 8869127 35. Croston G, Cable E, Toy J, et al. Selective partial agonism of vasopressin 1a receptors in vitro by OCE-205. J Pharmacol Exp Ther. 2023;2023 :1. 36. Konigshofer P, Brusilovskaya K, Schwabl P, Reiberger T. Animal models of portal hypertension. Biochim Biophys Acta Mol Basis Dis. 2019;1865 (5 ):1019–1030. doi:10.1016/j.bbadis.2018.07.018 30055295 37. Bai Z, An Y, Guo X, et al. Role of terlipressin in cirrhotic patients with ascites and without hepatorenal syndrome: a systematic review of current evidence. Can J Gastroenterol Hepatol. 2020;2020 :5106958. doi:10.1155/2020/5106958 32676484 38. Francque S, Wamutu S, Chatterjee S, et al. Non-alcoholic steatohepatitis induces non-fibrosis-related portal hypertension associated with splanchnic vasodilation and signs of a hyperdynamic circulation in vitro and in vivo in a rat model. Liver Int. 2010;30 (3 ):365–375. doi:10.1111/j.1478-3231.2009.02136.x 19840249 39. Ryou M, Stylopoulos N, Baffy G. Nonalcoholic fatty liver disease and portal hypertension. Explor Med. 2020;1 :149–169. doi:10.37349/emed.2020.00011 32685936 40. Rogers AE, Newberne PM. Animal model: fatty liver and cirrhosis in lipotrope-deficient male rats. Am J Pathol. 1973;73 (3 ):817–820.4129061 41. Oberti F, Veal N, Kaassis M, et al. Hemodynamic effects of terlipressin and octreotide administration alone or in combination in portal hypertensive rats. J Hepatol. 1998;29 (1 ):103–111. doi:10.1016/s0168-8278(98)80184-0 9696498 42. Blei AT, Groszmann RJ, Gusberg R, Conn HO. Comparison of vasopressin and triglycyl-lysine vasopressin on splanchnic and systemic hemodynamics in dogs. Dig Dis Sci. 1980;25 (9 ):688–694. doi:10.1007/BF01308328 7418592 43. Pereira RM, Dos Santos RA, Oliveira EA, et al. Development of hepatorenal syndrome in bile duct ligated rats. World J Gastroenterol. 2008;14 (28 ):4505–4511. doi:10.3748/wjg.14.4505 18680230 44. Martinez-Prieto C, Ortiz MC, Fortepiani LA, Ruiz-Macia J, Atucha NM, Garcia-Estan J. Haemodynamic and renal evolution of the bile duct-ligated rat. Clin Sci. 2000;98 (5 ):611–617. doi:10.1042/cs0980611 45. Gervaz P, Scholl B, Gillet M. Permanent access to the portal vein in rats: an experimental model. Eur Surg Res. 2000;32 (3 ):203–206. doi:10.1159/000008764 10878463 46. Van der Graaff D, Kwanten WJ, Couturier FJ, et al. Severe steatosis induces portal hypertension by systemic arterial hyporeactivity and hepatic vasoconstrictor hyperreactivity in rats. Lab Invest. 2018;98 (10 ):1263–1275. doi:10.1038/s41374-017-0018-z 29326427 47. Alessandria C, Ozdogan O, Guevara M, et al. MELD score and clinical type predict prognosis in hepatorenal syndrome: relevance to liver transplantation. Hepatology. 2005;41 (6 ):1282–1289. doi:10.1002/hep.20687 15834937 48. Gilbertson EL, Krishnasamy R, Foote C, Kennard AL, Jardine MJ, Gray NA. Burden of care and quality of life among caregivers for adults receiving maintenance dialysis: a systematic review. Am J Kidney Dis. 2019;73 (3 ):332–343. doi:10.1053/j.ajkd.2018.09.006 30454885 49. Terlivaz. US FDA prescribing information (terlipressin); 2022. Available from: https://www.accessdata.fda.gov/drugsatfda_docs/label/2022/022231s000lbl.pdf. Accessed October 17, 2022. 50. Wong F, Pappas SC, Curry MP, et al. Terlipressin plus albumin for the treatment of type 1 hepatorenal syndrome. N Engl J Med. 2021;384 (9 ):818–828. doi:10.1056/NEJMoa2008290 33657294 51. Liu Y, Meyer C, Xu C, et al. Animal models of chronic liver diseases. Am J Physiol Gastrointest Liver Physiol. 2013;304 (5 ):G449–G468. doi:10.1152/ajpgi.00199.2012 23275613
PMC010xxxxxx/PMC10352140.txt
==== Front J Pain Res J Pain Res jpr Journal of Pain Research 1178-7090 Dove 409721 10.2147/JPR.S409721 Original Research Comparison of Edge of Lamina Block with Thoracic Paravertebral Block and Retrolaminar Block for Analgesic Efficacy in Adult Patients Undergoing Video-Assisted Thoracic Surgery: A Prospective Randomized Study Gao et al Gao et al Gao Xiaoyun 1* Chen Moxi 1* Liu Penghao 1 Zhou Shenyuan 1 Kong Sai 1 Zhang Junfeng 1 Cao Jun 1 1 Department of Anesthesiology, Shanghai Sixth People’s Hospital, Shanghai, 200233, People’s Republic of China Correspondence: Jun Cao, Department of Anesthesiology, Shanghai Sixth People’s Hospital, No. 600, Yishan Road, Shanghai, 200233, People’s Republic of China, Tel +86 18930173661, Email [email protected] * These authors contributed equally to this work 13 7 2023 2023 16 23752382 09 3 2023 09 6 2023 © 2023 Gao et al. 2023 Gao et al. https://creativecommons.org/licenses/by-nc/3.0/ This work is published and licensed by Dove Medical Press Limited. The full terms of this license are available at https://www.dovepress.com/terms.php and incorporate the Creative Commons Attribution – Non Commercial (unported, v3.0) License (http://creativecommons.org/licenses/by-nc/3.0/). By accessing the work you hereby accept the Terms. Non-commercial uses of the work are permitted without any further permission from Dove Medical Press Limited, provided the work is properly attributed. For permission for commercial use of this work, please see paragraphs 4.2 and 5 of our Terms (https://www.dovepress.com/terms.php). Background A novel ultrasound-guided paravertebral block, the edge laminar block (ELB) was reported recently. However, it was unclear how effective ELB was in comparison with traditional blocking methods. We conducted a trial to compare the analgesic efficacy of ELB with the thoracic paravertebral block (TPVB) and the retrolaminar block (RLB) in patients undergoing video-assisted thoracic surgery (VATS). Methods We identified 90 patients who were scheduled for VATS and randomly assigned them to three groups: ELB group (Group E), TPVB group (Group T), and RLB group (Group R). Each group underwent ELB, TPVB, and RLB, respectively, under ultrasound guidance before general anesthesia induction. All patients received post-operative routine analgesia protocol. Our primary outcome was the extent of dermatomal sensory loss on the midclavicular, midaxillary, and scapular lines, measured using a pinprick 15 minutes after the nerve block. Secondary outcomes included the intraoperative dose of sufentanil, the numerical rating scale (NRS) scores assessed in the post-anesthesia care unit (PACU) and at 6, 12, and 24 hours post-operatively, and pethidine administrated as analgesic rescue dose. Results The percentages of nerve block range reaching the midclavicular line, midaxillary line, and scapular line in Group E were 96.7%, 93.3%, 93.3%, and 60% in Group T and 30%, 56.7%, and 96.7% in Group R, respectively. Group E had wider dermatomal sensory loss on the midclavicular line and midaxillary line compared to Group R (P < 0.001) and had a wider range compared to Group T on the scapular line (P < 0.001). There was no significant difference in the intraoperative use of sufentanil in the three groups. Post-operative NRS scores at each time point were significantly lower in Group E than those in the other two groups (P < 0.01). Conclusion ELB had a wider nerve block range and applied better post-operative analgesia in comparison with TPVB and RLB. Keywords analgesia retrolaminar block thoracic paravertebral block ultrasound guidance video-assisted thoracic surgery grant from the Scientific research fund of Shanghai Sixth People’s Hospital to M.X.C This work was supported by grant from the Scientific research fund of Shanghai Sixth People’s Hospital to M.X.C. (YNTS202005). ==== Body pmcBackground Thoracic paravertebral block (TPVB) can provide effective perioperative analgesia and is currently used extensively in thoracic surgery as part of the enhanced recovery after surgery (ERAS) protocol.1,2 However, complications such as pneumothorax and nerve or vessel injury cannot be completely avoided.3 As an alternative to TPVB, interfascial plane block, including erector spinae plane block (ESPB) and RLB, has no risk of injection-induced injury and has also become commonly used analgesic method in chest surgery,4–7 but its analgesic effect is inferior to that of TPVB.4 ELB had sufficient analgesic effect in the internal fixation of rib fractures.8 As we previously reported, in the ultrasound-guided laminar block the puncture needle was advanced by in-plane approach until the needle tip touched the lateral edge of “lamina’s cliff” with some resistance. The hypoechoic local anesthetics diffused to both sides (up and down to the lamina’s cliff) around the injection point. This is the reason why this new approach is called ELB. ESPB, which is an interfascial block procedure, has the similar puncture sites with RLB. However, several studies have reported that the blocking effect of ESPB is not ideal for RLB, which has been concluded in a study that ESPB is equivalent, and not superior, to RLB for postoperative analgesia.9,10 Hence, ESPB was not considered in this trial. To evaluate the effectiveness of ELB in video-assisted thoracic surgery, more information is needed. Besides, as a new approach to therapeutic lamina block, there is no comparative study between it and traditional blocking methods. Therefore, this study was designed to fill the gaps by comparing the ELB with TPVB and RLB for analgesic efficacy in adult patients undergoing video-assisted thoracic surgery. Materials and Methods Study Participants This double-blinded, prospective, randomized controlled trial was approved by the Ethics Committee of Shanghai Sixth People’s Hospital (2021–005) and registered at www.chictr.org.cn (ChiCTR2000040053). Written informed consent was obtained from each participant. Meanwhile, this study had been conducted in compliance with the current version of the Declaration of Helsinki. The trial began on February 5, 2021, and ended on October 29, 2021. We enrolled 90 patients who underwent VATS, aged 18–80 years, with American Society of Anesthesiologists (ASA) physical status I and II. The exclusion criteria were as follows: 1. Presence of local infection, 2. Neuropathy of spinal cord or trunk nerve, 3. Coagulation dysfunction, 4. Allergy to anesthetic drugs. Patients shifted to open thoracic surgery were also excluded. Procedures We randomly assigned each individual to Group E, Group T, or Group R in a 1:1:1 ratio, and the assigned group was concealed in an opaque, sealed envelope. Each envelope was opened only by the medical personnel who performed the nerve block procedure. All patients, operating room anesthetists, staff who did the postoperative NRS follow-up, and statisticians were blinded to the group allocation. After the patients were admitted into an operating room, upper-limb blood pressure, pulse oxygen saturation, and electrocardiography were regularly monitored. All three nerve blockades were performed using the S-Nerve™ Ultrasound System (Fujifilm SonoSite Inc. Bothell, WA, USA) with a convex-array probe (5–2 MHz; C60x; Fujifilm SonoSite Inc. Bothell, WA, USA). The patients were asked to lie in a lateral decubitus position (with the affected side up). Each patient was injected with 20 mL of 0.375% ropivacaine after negative aspiration using in-plane approach. In Group E, the low-frequency convex transducer was placed in a transverse position at the sixth vertebral level, ensuring that the spinous process, vertebral lamina (VL), and transverse process (TP) were visible on one screen. We rotated the probe 30–45 degrees, keeping VL and TP of the sixth thoracic vertebral in view, and moved the probe caudally until the hyperechoic line of TP diminished. Thus, the lateral edge of the laminar arcus was silhouetted.8 Local analgesic (LA) was injected at the edge of the lamina. In Group T, we injected the LA into the sixth thoracic paravertebral space (TPVS) through the paralaminar in-plane approach described by Taketa.11 In Group R, the LA was injected on the surface of the sixth thoracic lamina through the RLB approach. The ultrasound images and demonstration figure were showed in Figure 1. Figure 1 Ultrasound-guided images and demonstration figure. (A) The image of TPVB. (B) The image of RLB. (C) The image of ELB. (D) The demonstration figure. Triangles indicate the needle; Arrow (black and white) and cross signs indicate the injection sites. Abbreviations: PP, Parietal Pleura; VL, Vertebral Lamina; TP, Transverse Process; SP, Spinae Process; IIM-SCTL, Internal Intercostal Membrane, and Superior Costotransverse Ligament. The patients remained resting for 15 min after the nerve block procedure and were sent into the operating room. Each patient’s loss of pinprick sensation was evaluated by an anesthesiologist in the room. The dermatomal sensory loss on the midclavicular line, midaxillary line, and scapular line were recorded. Anesthesia was induced with intravenous injection of 2 mg/kg propofol, 0.4 μg/kg sufentanil, 0.6 mg/kg rocuronium, and maintained with sevoflurane at 0.7–1.2 age-adjusted minimum alveolar concentration (MAC). Sufentanil administered intermittently maintained the blood pressure within 20% of the baseline value. At the end of the surgery, sevoflurane inhalation was withdrawn and 50 mg of flurbiprofen was administered. All patients received a controlled intravenous analgesia pump (PCA, 100 mL, infusion rate 2 mL/h) consisting of tramadol 500 mg + lornoxicam 16 mg with no initial infusion dose. A dosage of 50 mg of flurbiprofen was infused twice a day. If the patient’s NRS was more than 4, an additional intramuscular injection of 50 mg pethidine was given as an analgesic rescue. Outcomes and Data Collection Primary outcomes The dermatomal sensory loss on the midclavicular line, midaxillary line, and scapular line were measured with a pinprick. Secondary outcomes included: Dosages of sufentanil. NRS pain scores assessed at PACU (T0), 6 (T1), 12 (T2), and 24 (T3) hours after surgery. Pethidine dose administrated as analgesic rescue. Complications related to the nerve block procedure such as nerve or vessel injuries. Sample Size The sample size was calculated based on the ratio of blockade range at the midclavicular line with PASS 11.0 (NCSS, LLC, Kaysville, USA). According to our pilot study, the ratios were 80%, 80%, and 40% in ELB, TPVB, and RLB, respectively. A total of 23 cases in each group were required to achieve 80% power with an α-value of 0.05. Taking into consideration a potential dropout rate of 15%, we aimed to enroll 30 patients in each group in this trial. Statistical Analysis We used the Shapiro–Wilk test to check for data normality and Levene’s test to verify the homogeneity of variance. Normally distributed data were expressed as mean ± standard deviation. Student’s t-test was used for intergroup comparisons, and data in multiple groups were compared with analysis of variance (ANOVA). Non-normally distributed data were represented as median (interquartile range), and comparisons between groups were assessed by the Mann-Whitney U-test or Kruskal–Wallis H-test. Categorical variables were described as numbers (%) and analyzed using the Chi-square test or Fisher’s exact test. P < 0.05 was considered as the significance level. All statistical analyses were performed using SPSS 26.0 software. Results A total of 90 patients were enrolled in this study. As no one was shifted to open thoracic surgery, all patients were included in the final study and there were 30 patients in each of the groups E, T, and R. A flow chart of patient recruitment is shown in Figure 2. Patient characteristics are listed in Table 1. No significant differences were found between the three groups.Table 1 Demographic and Clinical Characteristics of the Patients Group E (n=30) Group T (n=30) Group R (n=30) p. Overall Age (y) 58.5±11.7 59.2±15.3 56.0±13.1 0.377 Gender (Male/Female) 11/19 12/18 13/17 x2=0.278, p=0.963 Body mass index (kg/m2) 23.1±2.6 22.5±2.3 24.4±3.7 0.074 Height (cm) 162.6±5.4 165.5±7.5 165.0±7.3 0.254 Operative duration 70.8 ±33.8 66.8±34.3 65.2±30.4 0.839 Notes: ELB group; Group T: TPVB group; Group R: RLB group. Figure 2 Flow diagram of the study. The dermatomal sensory loss on the midclavicular line, midaxillary line, and scapular line of all groups are presented in Table 2. In order to better reflect the effects of the three blocking methods on the skin segment, we presented the case results of the three observation lines with blocking effects separately. We also measured the effective ratio of patients who achieved a blocking effect on one observation line in a group. Group E had wider dermatomal sensory loss on the midclavicular and midaxillary line compared to Group R and had a wider range on the scapular line compared to Group T (P < 0.001), regardless of whether non-effective dermatomes were included (Table 2). Meanwhile, in Group E, T, and R, the percentages of analgesic effect on the three observation lines were 93.3% (28/30 cases), 60% (18/30 cases) and 26.7% (8/30 cases), respectively. This suggested that ELB was able to block the region of the three lines simultaneously when compared with the other two nerve block techniques (x2 = 27.778, P < 0.001). Compared with the values in Group R, Group T had wider dermatomal sensory loss on the midclavicular line and midaxillary line (P < 0.001), but the range on the scapular line was narrower (P < 0.01). In addition, bilateral blockade occurred in two patients in Group E. They both had 9 hypoesthesia dermatomes (T4–T12).Table 2 Dermatomes and Ratios of Sensory Loss on Midclavicular Line, Midaxillary Line, and Scapular Line of Three Groups Group E (n=30) Group T (n=30) Group R (n=30) p value E vs T E vs R T vs R All patients’ data calculated (noneffective dermatomes also included) Midclavicular line Dermatomes 4.6±2.4 3.9±2.5 0.0(0, 1.0) 0.274 <0.001 <0.001 Midaxillary line Dermatomes 4.5±2.4 3.5±1.9 2.0(0, 3.0) 0.087 <0.001 <0.001 Scapular line Dermatomes 5.3±2.7 1.0(0, 4.3) 4.5±3.2 <0.001 0.277 0.001 Only effective dermatomes involved Midclavicular line Dermatomes 4.8±2.3 4.3±2.3 2.6±2.0 0.441 0.012 0.051 Effective ratio 96.7%(29/30) 93.3%(28/30) 30%(9/30) 0.554 <0.001 <0.001 Midaxillary line Dermatomes 4.7±2.3 3.8±1.7 3.0±1.5 0.108 0.010 0.128 Effective ratio 96.7%(29/30) 93.3%(28/30) 56.7%(17/30) 0.554 <0.001 0.001 Scapular line Dermatomes 5.5±2.6 3.0(1, 6.3) 4.0(3, 5.5) 0.036 0.076 0.216 Effective ratio 96.7%(29/30) 60%(18/30) 96.7%(29/30) 0.001 1.000 0.001 Effective ratio on all lines in the same person 93.3%(28/30) 60%(18/30) 26.7%(8/30) x2=27.778, p<0.001 Notes: ELB group; Group T: TPVB group; Group R: RLB group. There was no significant difference in the intraoperative sufentanil dosage in the three groups. Post-operative NRS scores at each time point were significantly lower in Group E than in the other two groups (P < 0.05). There was no statistical difference in NRS scores between Group T and Group R at each time point (Table 3). Patients in Group E received less pethidine rescue compared to the other two groups (P = 0.006).Table 3 Intraoperative Consumption of Sufentanil, Postoperative Pethidine Rescue, and NRS Scores in Three Groups Group E (n=30) Group T (n=30) Group R (n=30) p value E vs T E vs R T vs R Dose of sufentanil (μg) 35(30, 40) 35(30, 40) 37.5(30, 45) 0.737 0.090 0.059 Unit dose of sufentanil (μg/kg/h) Avrange (95% CI) 0.5(0.38, 0.68) 0.59(0.34, 0.85) 0.49(0.40, 0.84) 0.663 0.574 0.918 Pethidine rescue Doses 3 15 15 0.006 Cases 3/30 13/30 13/30 x2=10.175, p=0.006 NRS PACU 1.3±1.1 2.6±1.1 2.5±1.2 <0.001 <0.001 0.741 6h 2.7±1.1 4.0(3.0, 4.3) 3.5±1.5 <0.001 0.014 0.804 12h 2.4±1.1 3.4±1.5 3.3±1.5 0.002 0.005 0.792 24h 1.0(0.8, 2.0) 2.2±1.2 2.3±1.3 0.001 0.001 0.612 Notes: Group E: ELB group; Group T: TPVB group; Group R: RLB group. Abbreviations: NRS, the numerical rating scale; PACU, post-anesthesia care unit. Perioperative nerve block-related complications were not observed in any of them. Discussion Although the safety and effectiveness of TPVB have been acknowledged,12 the outcome depends to some extent on the skill of the operator on account of the anatomy of TPVS. Previous studies reported that the continuous RLB shows effective surgical analgesia and satisfactory postoperative pain control in PCNL surgery,13 breast cancer surgery by landmark technique14 and rib fracture.15 These studies indicated that RLB could be developed as an alternative to TPVB due to its easier and safer properties. However, its analgesic effect is inferior to that of TPVB.4 We developed the ELB based on RLB. There were no major vessels, nerves, or pleura in the needle pathway, which was confirmed with ultrasonography. The operator could easily advance the needle to the point of injection without worrying about the risk of TPVB. In this study, we found that group T had wider dermatomal sensory loss on the midclavicular line and midaxillary line than Group R, but the blocking effect of Group R on the scapular line was superior to that of Group T. This is consistent with the conclusion of Sabouri.16 These results indicated that the injection sites of the two groups could be related to these results. Although TPVS is connected with the intervertebral foramen, the injection point is at a certain distance from the intervertebral foramen. Therefore, the TPVB local anesthetics mainly infiltrate the anterior thoracic nerve. However, the anesthetics for RLB were injected in the space between the erector spinae and the lamina that was close to the posterior branch. Although previous cadaver studies have demonstrated that LA of the RLB may penetrate the superior costal transverse ligament and the posterior ramus of the spinal nerve to the TPVS, its blocking effect is still inferior to that of the TPVB.16 The injection site of ELB which Shu8 introduced is at the edge of the lamina near the inferior articular process of the vertebra and is closer to the zygapophyseal than RLB. In this study, the analgesic dermatomal along the three lines in Group E were more than those of the other two groups, irrespective of whether the patients who did not experience the blocking effect were excluded or not. In addition, we found that ELB blocked anterior branches more effectively than RLB and blocked posterior branches more effectively than TPVB. A relatively wider range of blocks enabled Group E to experience better post-operative analgesia. Also, in Group E, the ratio of sensory loss on the midclavicular line, midaxillary line, and scapular line was 96.7% (29/30 cases). The percentage of patients who acquired anterior and posterior branch blockade simultaneously in Group E was 93.3% (28/30 cases), which was significantly higher than that in groups T and R (60% (18/30 cases), 26.7% (8/30 cases)), respectively. The above findings may be due to more anesthetics being diffused into the TPVS near the foraminal outlet as it spreads along the lamina, so the anterior and posterior branches of the thoracic nerve could be effectively blocked. TPVB and RLB have been widely used for post-operative analgesia in VATS.17–20 Wang19 and Sandeep20 noted that TPVB was superior to RLB, while in our study, there was no significant difference in NRS scores between our T and R groups. This may be related to the routine daily intravenous infusion of flurbiprofen in our trial. It is worth noting that the post-operative NRS scores in our Group E were significantly lower than those in the other two groups at each time point (P < 0.05). There was no statistical difference in the intraoperative usage of sufentanil among the three groups, whether in the analysis of total amount or unit dosage. Meanwhile, the frequency of pethidine rescue in Group E was also significantly lower than those in the other two groups. It is assumed that the decrease in post-operative NRS might be attributed to ELB. Satisfactory post-operative analgesia depends on effective analgesia in the anterior, middle, and posterior areas of the chest wall, namely successful blockade at the anterior and posterior branches. There were several limitations in this study. Firstly, there was no cadaver study, and we could not confirm whether the LA of the ELB was mainly diffused to the TPVS adjacent to the foramina. Even if this was the case, it is still not known whether the solution similarly infiltrates into the TPVS as it does in RLB. In addition, there were two patients with bilateral blocks in Group E. How the solution enters the vertebral canal can perhaps only be understood in a cadaver study. Secondly, only NRS scale was used to find the efficacy, other pain score assessment scales should be used to conclude this result. Finally, all the three nerve blocks in the study belong to the paravertebral block. Theoretically, LA spreads in different ways and ranges due to different doses. In a study with pig carcasses, Damjanovska21 found that LA of 10 mL could spread to the posterior branch of the spinal nerve, but could not spread to the TPVS when RLB was done. Nonetheless, LA was found in the TPVS when 30 mL dose was given. We do not know whether using higher doses in our study would have resulted in different results, and this needs to be addressed in further clinical studies. Conclusion In conclusion, ELB was effective in achieving block of the anterior and posterior branches of the thoracic nerve simultaneously, and had a wider nerve block range and better post-operative analgesia in comparison with TPVB and RLB. It has the potential to be a better alternative to TPVB in thoracic surgery. Acknowledgments We appreciate the supports from the cardiothoracic surgeons and nursing teams of Shanghai Sixth People’s Hospital. Data Sharing Statement The datasets used and analyzed during the current study had been submitted to publicly available repository https://pan.baidu.com/s/1FXkjstyAJ0XZuU_MpsyV_A, and the accession code is available from the corresponding author on reasonable request. Ethics Approval This trial had been approved by the Ethics Committee of Shanghai Sixth People’s Hospital (2021–005). This study was conducted in accordance with the declaration of Helsinki. Consent to Participate Written informed consents were obtained from all patients. Consent for Publication All authors read and approved the final manuscript and were in agreement with the content of the manuscript. Code Availability The sample size was calculated with PASS 11.0, and the statistical analyses were performed using SPSS 26.0 software. Disclosure The authors report no conflicts of interest in this work. ==== Refs References 1. El Mourad MB, Amer AF. Effects of adding dexamethasone or ketamine to bupivacaine for ultrasound-guided thoracic paravertebral block in patients undergoing modified radical mastectomy: a prospective randomized controlled study. Indian J Anaesth. 2018;62 (4 ):285–291. PMID: 29720754; PMCID: PMC5907434. doi:10.4103/ija.IJA_791_17 29720754 2. Ding W, Chen Y, Li D, et al. Investigation of single-dose thoracic paravertebral analgesia for postoperative pain control after thoracoscopic lobectomy - a randomized controlled trial. Int J Surg. 2018;57 :8–14. PMID: 30056127. doi:10.1016/j.ijsu.2018.07.006 30056127 3. Pace MM, Sharma B, Anderson-Dam J, Fleischmann K, Warren L, Stefanovich P. Ultrasound-guided thoracic paravertebral blockade: a retrospective study of the incidence of complications. Anesth Analg. 2016;122 (4 ):1186–1191. PMID: 26756911. doi:10.1213/ANE.0000000000001117 26756911 4. Murouchi T, Yamakage M. Retrolaminar block: analgesic efficacy and safety evaluation. J Anesth. 2016;30 (6 ):1003–1007. PMID: 27518726. doi:10.1007/s00540-016-2230-1 27518726 5. Forero M, Adhikary SD, Lopez H, Tsui C, Chin KJ. The erector spinae plane block: a novel analgesic technique in thoracic neuropathic pain. Reg Anesth Pain Med. 2016;41 :621–627. doi:10.1097/AAP.0000000000000451 27501016 6. Gurkan Y, Aksu C, Kus A, Yorukoglu UH, Kilic CT. Ultrasound guided erector spinae plane block reduces postoperative opioid consumption following breast surgery: a randomized controlled study. J Clin Anesth. 2018;50 :65–68. doi:10.1016/j.jclinane.2018.06.033 29980005 7. Yao Y, Li H, He Q, Chen T, Wang Y, Zheng X. Efficacy of ultrasound-guided erector spinae plane block on postoperative quality of recovery and analgesia after modified radical mastectomy: randomized controlled trial. Reg Anesth Pain Med. 2020;45 :5–9. doi:10.1136/rapm-2019-100983 8. Shu Z, Zhang J, Cao J. A novel approach of ultrasound-guided laminar block for rib fracture surgery. J Clin Anesth. 2021;70 :110191. PMID: 33561706. doi:10.1016/j.jclinane.2021.110191 33561706 9. Sotome S, Sawada A, Wada A, Shima H, Kutomi G, Yamakage M. Erector spinae plane block versus retrolaminar block for postoperative analgesia after breast surgery: a randomized controlled trial. J Anesth. 2021;35 (1 ):27–34. doi:10.1007/s00540-020-02855-y 32915300 10. Tao T, Zhou Q. 竖脊肌平面阻滞与后路椎板阻滞对后路腰椎手术术后镇痛效果的比较 [Efficacy of erector spinae block versus retrolaminar block for postoperative analgesia following posterior lumbar surgery]. Nan Fang Yi Ke Da Xue Xue Bao. 2019;39 (6 ):736–739. Chinese. doi:10.12122/j.issn.1673-4254.2019.06.17 31270055 11. Taketa Y, Fujitani T. A novel paralaminar in-plane approach for ultrasound-guided continuous thoracic paravertebral block using microconvex array transducer. Reg Anesth Pain Med. 2015;40 (4 ):390. PMID: 26079356. doi:10.1097/AAP.0000000000000259 12. Schnabel A, Reichl SU, Kranke P, Pogatzki-Zahn EM, Zahn PK. Efficacy and safety of paravertebral blocks in breast surgery: a meta-analysis of randomized controlled trials. Br J Anaesth. 2010;105 (6 ):842–852. Erratum in: Br J Anaesth. 2013;111(3):522. PMID: 20947592. doi:10.1093/bja/aeq265 20947592 13. Kumari P, Kumar A, Sinha C, Kumar A, Kumari A. Continuous retrolaminar block in percutaneous nephrolithotomy surgery. Saudi J Anaesth. 2023;17 (1 ):132–133. doi:10.4103/sja.sja_486_22 37032680 14. Jüttner T, Werdehausen R, Hermanns H, et al. The paravertebral lamina technique: a new regional anesthesia approach for breast surgery. J Clin Anesth. 2011;23 :443–450. doi:10.1016/j.jclinane.2010.12.015 21752620 15. Voscopoulos C, Palaniappan D, Zeballos J, Ko H, Janfaza D, Vlassakov K. The ultrasound-guided retrolaminar block. Can J Anesth. 2013;60 :888–895. doi:10.1007/s12630-013-9983-x 23797663 16. Sabouri AS, Crawford L, Bick SK, Nozari A, Anderson TA. Is a retrolaminar approach to the thoracic paravertebral space possible?: A human cadaveric study. Reg Anesth Pain Med. 2018;43 (8 ):864–868. PMID: 29923954. doi:10.1097/AAP.0000000000000828 29923954 17. Taketa Y, Irisawa Y, Fujitani T. Comparison of analgesic efficacy between two approaches of paravertebral block for thoracotomy: a randomised trial. Acta Anaesthesiol Scand. 2018;62 (9 ):1274–1279. PMID: 30047132. doi:10.1111/aas.13216 30047132 18. Nobukuni K, Hatta M, Nakagaki T, Yoshino J, Obuchi T, Fujimura N. Retrolaminar versus epidural block for postoperative analgesia after minor video-assisted thoracic surgery: a retrospective, matched, non-inferiority study. J Thorac Dis. 2021;13 (5 ):2758–2767. PMID: 34164168; PMCID: PMC8182553. doi:10.21037/jtd-21-238 34164168 19. Wang Q, Wei S, Li S, et al. Comparison of the analgesic effect of ultrasound-guided paravertebral block and ultrasound-guided retrolaminar block in uniportal video-assisted thoracoscopic surgery: a prospective, randomized study. BMC Cancer. 2021;21 (1 ):1229. PMID: 34784889; PMCID: PMC8594110. doi:10.1186/s12885-021-08938-7 34784889 20. Sandeep B, Huang X, Li Y, Xiong D, Zhu B, Xiao Z. A comparison of regional anesthesia techniques in patients undergoing video-assisted thoracic surgery: a network meta-analysis. Int J Surg. 2022;105 :106840. PMID: 36030040. doi:10.1016/j.ijsu.2022.106840 36030040 21. Damjanovska M, Stopar Pintaric T, Cvetko E, Vlassakov K. The ultrasound-guided retrolaminar block: volume-dependent injectate distribution. J Pain Res. 2018;11 :293–299. PMID: 29445296; PMCID: PMC5808708. doi:10.2147/JPR.S153660 29445296
PMC010xxxxxx/PMC10352141.txt
==== Front Int J Nanomedicine Int J Nanomedicine ijn International Journal of Nanomedicine 1176-9114 1178-2013 Dove 402678 10.2147/IJN.S402678 Review Recent Advances for Dynamic-Based Therapy of Atherosclerosis Wu et al Wu et al Wu Guanghao 1 * Yu Guanye 2 * Zheng Meiling 3 * Peng Wenhui 2 Li Lei 4 1 School of Materials Science and Engineering, Beijing Institute of Technology, Beijing, 100081, People’s Republic of China 2 Department of Cardiology, Shanghai Tenth People’s Hospital, Tongji University, School of Medicine, Shanghai, 200072, People’s Republic of China 3 Dongzhimen Hospital Beijing University of Chinese Medicine, Beijing, 101121, People’s Republic of China 4 National Clinical Research Center for Obstetric & Gynecologic Diseases, Department of Obstetrics and Gynecology, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing, 100730, People’s Republic of China Correspondence: Lei Li, Department of Obstetrics and Gynecology, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College, No. 1 Shuai Fu Yuan, Eastern District, Beijing, 100730, People’s Republic of China, Tel +8613426328126, Fax +86 10 69158100, Email [email protected] Guanghao Wu, School of Materials Science and Engineering, Beijing Institute of Technology, Beijing, 100081, People’s Republic of China, Email [email protected] * These authors contributed equally to this work 13 7 2023 2023 18 38513878 26 12 2022 06 5 2023 © 2023 Wu et al. 2023 Wu et al. https://creativecommons.org/licenses/by-nc/3.0/ This work is published and licensed by Dove Medical Press Limited. The full terms of this license are available at https://www.dovepress.com/terms.php and incorporate the Creative Commons Attribution – Non Commercial (unported, v3.0) License (http://creativecommons.org/licenses/by-nc/3.0/). By accessing the work you hereby accept the Terms. Non-commercial uses of the work are permitted without any further permission from Dove Medical Press Limited, provided the work is properly attributed. For permission for commercial use of this work, please see paragraphs 4.2 and 5 of our Terms (https://www.dovepress.com/terms.php). Abstract Atherosclerosis (AS) is a chronic inflammatory disease, which may lead to high morbidity and mortality. Currently, the clinical treatment strategy for AS is administering drugs and performing surgery. However, advanced therapy strategies are urgently required because of the deficient therapeutic effects of current managements. Increased number of energy conversion-based organic or inorganic materials has been used in cancer and other major disease treatments, bringing hope to patients with the development of nanomedicine and materials. These treatment strategies employ specific nanomaterials with specific own physiochemical properties (external stimuli: light or ultrasound) to promote foam cell apoptosis and cholesterol efflux. Based on the pathological characteristics of vulnerable plaques, energy conversion-based nano-therapy has attracted increasing attention in the field of anti-atherosclerosis. Therefore, this review focuses on recent advances in energy conversion-based treatments. In addition to summarizing the therapeutic effects of various techniques, the regulated pathological processes are highlighted. Finally, the challenges and prospects for further development of dynamic treatment for AS are discussed. Keywords atherosclerosis nanomaterial physiochemical reactive oxygen species energy conversion therapy ==== Body pmcIntroduction Cardiovascular disorders are the leading causes of morbidity and mortality worldwide. According to the latest report by the World Health Organization (WHO), approximately 32% of deaths are due to cardiovascular diseases (CVDs).1 CVDs are a group of disorders of the heart and blood vessels. Approximately 45.1% of these cardiovascular disease-related deaths were caused by coronary artery disease. Atherosclerosis (AS) is a chronic inflammatory vascular disease. Smoking, alcoholism, high-diet, lack of exercise, and genetic predisposition are important factors in the development of AS. These factors increase blood lipid concentration and lipid deposits in vessel wall would lead to formation of atherosclerotic plaque and gradually induce stenosis and obstruction of blood flow. Rupture of atherosclerotic plaque produces lesions that result in acute myocardial infarction, sudden cardiac death, or other severe outcomes.2,3 Until now, clinical treatments for AS focused on drugs and surgery; angioplasty and bypass grafting are the two main surgical interventions.4,5 However, surgery is invasive and may induce thrombosis in the endarterectomized segment, which may increase the risk of thrombosis, bleeding, and restenosis. Statins, which are an important drug intervention, are thought to be important in the prevention of AS. These drugs could not only reduce cholesterol but also alleviate pro-inflammatory cytokine effects. Besides statins, some other drugs, such as proprotein convertase subtilisin/kexin type 9 inhibitors (Alirocumab and Evolocumab), could enhance low-density lipoprotein (LDL) degradation and promote lower cholesterol levels.6 In addition, certain immunosuppressants (methotrexate and rapamycin) show a clear and promising impact in reducing atherosclerotic plaque sizes. However, there are still some limitations in terms of stability, targeting efficacy, toxicity, and production. Currently, the combination of nanotechnology and AS offers hope for the treatment of these diseases. Compared with traditional nano-therapies, energy conversion-based nano-therapy can not only leverage external energy for proapoptotic effects on the foam cells of vulnerable plaques but also allow for minimal dosing to achieve therapeutic effects and migrated side-effects. Previous studies have summarized the use of nanomedicine in the treatment and imaging of atherosclerotic plaques. To the best of our knowledge, there is no comprehensive and systematic review of nano-dynamic treatment strategies for AS. In this review, we explored and illustrated the mechanism of atherosclerotic plaque formation and highlighted the current characteristics of phototherapy—both photothermal therapy (PTT) and photodynamic therapy (PDT—and sonodynamic therapy (SDT)) for AS treatment (Figure 1). Figure 1 Schematics of nanomaterials and conventional agents for energy conversion-based therapies for atherosclerosis. The Pathogenesis of Atherosclerotic Plaques The formation of atherosclerotic plaque is a continuous and complicated process with early and advanced stages. A schematic diagram of the stages of atherosclerosis formation is shown in Figure 2. At early stage, toxic substances such as nicotine, high level of cholesterol and glucose in blood, which are brought by maladaptive lifestyle, would damage the balance of microenvironment, and hemodynamic disorder like hypertension further aggravates the situation, further damages endothelium.7–9 Maladaptive living and eating habits could lead to endothelial dysfunction. Endothelial dysfunction is a critical pathophysiological element in AS, inducing greater penetration of macromolecules such as lipoproteins, and increasing production of chemotactic molecules (ICAM-1, VCAM-1, and P-selectin). Next, monocytes differentiate into macrophages and transform into foam cells by ingesting oxidized low-density lipoprotein (ox-LDL) via scavenger receptor type A (SR-A), CD36, and lectin-like oxidized low-density lipoprotein receptor-1 (LOX-1). Foam cells and other immune cells secrete several cytokines (eg, IL-6, TNF-α, IL-1β, etc.). These cytokines can induce vascular smooth muscle cells (VSMCs) to migrate from tunica media into the tunica intima and proliferate. These differentiated VSMCs can phagocytize ox-LDL, form VSMC-derived foam cells and may produce extracellular traps which contributed to plaque progression by influencing the microenvironment of advanced plaques.10–12 Other VSMCs secrete collagen or fibronectin to form fibrous cap. Subsequently, the macrophage-derived and VSMC-derived foam cells gradually form lipid core. Plaque rupture will eventually take place in lesions with large lipid core and thin fibrous cap, leading to platelet adherence, aggregation and thrombosis, causing severe clinical events. Figure 2 The development of atherosclerosis plaques. The Underlying Mechanism-Reactive Oxygen Species (ROS) The basic processes of ROS generation have been widely studied in recent years. It is commonly recognized that ROS generation is the predominant mechanism by which PDT exerted its effects. It is a plausible theory that ROS production is the major mechanism by which PDT-derived SDT functions. However, the difference between the two is that for PDT, the mechanisms of ROS generation due to photodynamic reaction are well identified, whereas for SDT, there are relevant hypotheses but none with certainty. There are two photochemical interaction pathways to generate ROS—type I and type II.13 In the type-I mechanism, the photosensitizer (PS) is excited by a specific wavelength of light before reacting with the substrate to produce free radicals and anionic radicals through a hydrogen atom or electron transfer procedure. Since most of the previously formed free radicals would react with oxygen, ROS in cells is generated through a chain of intricate free radical reactions. The main component of the type II mechanism is the interaction of PS excitation to the triplet state with oxygen molecules, and the formation of singlet oxygen. Both reactions could occur simultaneously, and the proportion of the procedures is determined by the type of PS employed, the substrate, and the concentration of oxygen.14 Although the mechanism underlying the generation of ROS is currently not well defined for SDT, there are two potential hypotheses that dominate the mainstream perspective—sonoluminescence (SL) and pyrolysis. Considering the application of sound-sensitizers that are simultaneously photosensitizers, such as protoporphyrin IX (PpIX) and 5-aminolevulinic acid (ALA),15–18 SL is considered to be a key trigger for sonosensitizer-mediated ROS production in SDT. SL is a form of chemiluminescence, which is the phenomenon of light emission excited by the energy released by the rapid collapse of a bubble under ultrasonic excitation, and is based on the ability of ultrasound to generate cavitation.19,20 Cavitation includes both inertial cavitation and stable cavitation.21 SL occurs during the acoustic cavitation process, with inertial cavitation as the dominant process. Following the collapse, SL, as well as chemicals in the surrounding medium, such as free hydroxyl radicals, are generated.22 This is accompanied by destruction of cells and tissues to exert therapeutic effects. Therefore, it is essential to clarify the relationship between the cavitation dose and the generation of hydroxyl radicals. Peter et al assessed the creation of free radicals using the chemodose of hydrogen peroxide and showed that the higher the cavitation dose, the greater the generation of radicals.23 Similar findings were reported by Villeneuve et al24 In addition, Xu et al discovered that the reactive oxygen radicals were characterized by aqueous PpIX, and that the decomposition was inhibited by radical scavengers such as NAC, but not sodium azide, which showed that PpIX decomposition by ultrasound was determined by the hydroxyl radical.25 It is well known that inertial cavitation dominated the SL; however, whether stable cavitation could cause SL deserves further investigation. Generally, stable cavitation indicates that the bubbles are underactive during chemical and SL processes. The presence of some bubbles is called “high-energy stable cavitation”, and it has two properties—stable shape and an active state for SL or chemical reactions.26 Furthermore, the mechanism of multi-bubble sonoluminescence (MBSL) has been widely accepted as a chemiluminescence phenomenon caused by excited OH radicals.27 This has also been demonstrated in single-bubble acoustic chemistry studies. Yasui and his team demonstrated that O radicals were the main oxidants produced by gaseous bubbles; therefore, the generation of OH radicals subsequently triggered the same mechanism of ROS as PDT—Type I and II pathways. A contrasting hypothesis is that the pyrolysis that occurs at the gas–liquid interfacial regions after sonoluminescence is the mechanism for the production of free radicals. There are three regions,28,29 and using the ESR-spin trapping technique, the mechanisms underlying the assignment of free radical structures have been elucidated by numerous studies.29,30 The first region is the gas phase that is generated by a collapsed cavitation bubble. When the intense bubble lapse occurs, the temperature within the bubble increases to 5000 Kelvin or even 15,000 Kelvin, accompanied by high pressures. This causes water vapor and oxygen to be dissociated into oxidants such as OH radicals and O radicals, which are generated inside the bubble.20,30,31 For free radicals, a higher temperature inside the bubble is not always beneficial; there exists an optimal bubble temperature of approximately 5500 Kelvin for better production of free radicals. This is because the oxidants are consumed by nitrogen inside the bubble when the temperature is too high.32 On the other hand, the temperature inside the bubble is dependent on a variety of factors, such as the thermal conductivity and heat capacity ratio in the bubbles, and a reduction in the heat capacity ratio would significantly decrease the temperature.33 The gas–liquid interface area is the second region, which includes methyl radicals, phenyl, and tert-butyl radicals. The third region is the original solvent at room temperature, where the solute reacts with OH radicals and H atoms to produce radiation-chemistry-like byproducts. Redox homeostasis is the system that maintains the balance between oxidation and antioxidation in the body.34 Disruption of the balance is one of the pathological mechanisms of many diseases, such as AS, cancer, and cognitive impairment.35–37 ROS plays an important role in oxidative stress, which is a collective term for a series of free radicals, including superoxide, hydroxyl, peroxide, and hydroperoxide.38 ROS causes cellular damage, including oxidation of proteins, lipids, and carbohydrates, altering the genome and cellular structure, and acting as a signaling molecule to induce intracellular signaling pathway transduction.39,40 In addition, the equilibrium between ROS generation and the capacity of cells to achieve antitonicity or rehabilitate damage could lead to oxidative stress and ultimately trigger different cell death mechanisms, including apoptosis, necrosis, and autophagy.41,42 Multiple cell types, including endothelial cells, macrophages, foam cells, and VSMCs participate in the pathogenesis of AS,43 and ROS contributes to its pathophysiological process by affecting these cell functions.35 This is the mechanism that PDT and SDT exploit to develop great therapeutic potential for AS. Photo-Based Energy-Conversion Nano-Therapy Photothermal Therapy (PTT) PTT can be beneficial for eradicating or reversing AS plaque, which is beyond the effect of modern therapies for AS, such as lipid-lowering drugs or recanalization operations. PTT involves two key elements—photothermal agents and laser radiation, which are typically near-infrared radiation (NIR). Different strategies have been applied to ensure that photothermal agents can be precisely guided and absorbed by the plaque deposits. The agents then convert light energy into heat, inducing intracellular hyperthermia. This process induces cell necrosis, apoptosis, and autophagy depending on the thermal energy or ROS produced in the process, thus preventing the rapid progression of AS. Various photothermal agents have been used in AS, which can generally be categorized into two types—inorganic and hybrid nanoparticles (NPs). Inorganic NPs have small-sized effects and optical/magnetic/ultrasonic signal-response performance and may serve as better theranostic agents. Organic nanomaterials can be degraded and metabolized in vivo and have better biocompatibility, greater drug loading capability, and longer circulation time. Hybrid NPs usually combine the advantages of both organic and inorganic materials; thus, they have attracted great interest in recent years.44 Inorganic Photothermal Agents Initially, Lukianova-Hleb et al proved that gold NPs generated photothermal microbubbles around the lesion in vitro with a short laser pulse and were able to disrupt plaque tissue mechanically, without damaging the arterial wall.45 Gold NPs can be phagocytosed by macrophages at sites of vascular inflammation, and their vitality will not be affected unless exposed to NIR laser.46 In apolipoprotein E knockout (ApoE-/-) mice (an animal model used to mimic AS), gold nanorods (Au NRs) were applicable as both contrast agents for computed tomography (CT) imaging and photothermal agents for inflammatory macrophages irradiated by an 808 nm laser.47 However, noble metal nanostructures are expensive and experience obvious deformation under continuous irradiation, which eventually impairs their photothermal ability. Materials such as graphene and carbon nanotubes are much more reliable. Single-walled carbon nanotubes (SWNTs) are engulfed by 94 ± 6% of macrophages and induced 93 ± 3% in vitro cell death under irradiation; however, the nonspecific uptake of SWNTs would damage other organs, thus requiring more researches to improve their accuracy.48 Semiconductor photothermal nanomaterials have a number of advantages, including lower cost, simpler synthetic procedures, and better photothermal effect. Ultrasmall CuCo2S4 nanocrystals exhibit high photothermal conversion rate and can be rapidly cleared by the renal system.49 In addition, MoO2 nanoclusters can also eliminate macrophages with no serious side-effects in vivo under NIR.50 The photothermal conversion efficiency of Cu3BiS3 nanocrystals is up to 58.6% with a power density of 0.4 W·cm−2, and they can also be applied as a CT contrast agent to evaluate carotid inflammation.51 Janus heterodimers consisting of noble metal and semiconductor, such as the Janus Ag/Ag2S beads synthesized by Peng et al, possess better photothermal conversion efficiency, lower cytotoxicity and better biocompatibility than the individual components. They can be uptake by macrophages and remarkably eliminate these cells from arterial inflammation without further damage to the normal arterial wall and major organs.52 Hybrid NPs With the development of nanotheranostics, the construction of multifunctional NPs has become increasingly feasible. To make photothermal agents target different cells more precisely, they are coated with receptor-specific antibodies or signal molecules. Drugs that play a protective role in AS have also been loaded to strengthen their therapeutic effects on restoring the endothelial function. Some materials are combined with nanomotors that can help NPs penetrate deeper into the lesion. Last but not least, combination of nanomaterials that can strengthen imaging resolution enables image-guided PTT. SR-A is expressed by activated macrophages in AS plaque, which can be competitively blocked by dextran sulfate (DS) and inhibit the cellular endocytosis of ox-LDL. A multifunctional therapeutic nanoplatform, CS-CNCs@Ce6/DS, possesses rapid photothermal conversion ability of chitosan-coated carbon nanocages (CS-CNCs) and enables efficient PTT. The nanoplatform further induces the PDT effect on the activated macrophages, as it gradually releases chlorin e6 (Ce6) in the lesion, reducing secretion of pro-inflammatory cytokine and alleviating proliferation and migration of VSMCs.53 Oh et al synthesized mannosylated-reduced graphene oxide (Man-rGO) to enhance the therapeutic effect of graphene oxide, as it could bind to the mannose receptor, a marker of M2 macrophages that mediated phagocytosis.54 To realize synergistic treatment effect of drugs and PTT, Liu et al used dendritic mesoporous silica NPs as nanocarrier to cap anticoagulant drug heparin (Hep) and copper sulfide (CuS), which provided a photothermal effect. Meanwhile, the hyaluronic acid (HA) bounding to the nanocarrier enabled it to release drugs according to pH and specifically target CD44-positive macrophages.55 Transient receptor potential vanilloid subfamily 1 (TRPV1) is a thermo-sensitive channel in VSMCs, which induces a Ca2+ influx when the local temperature increases, further activating autophagy (Figure 3A(a)). Gao et al used copper sulfide (CuS) NPs coupled with antibodies of TRPV1 (Figure 3A(b)) as a photothermal switch to activate TRPV1 signaling (Figure 3A(c)), induced autophagosome formation (Figure 3A(d)), reduced lipid uptake, and thus prevented VSMCs from turning into foam cells. In vivo experiment exhibited significant reduction of lesion areas, indicating great therapeutic effect of AS (Figure 3A(e)). 56 Figure 3 (A) (a) Illustration of CuS-TRPV1 switch for photothermal activation of TRPV1 signaling to attenuate atherosclerosis. (b) Transmission electron microscopy (TEM) image of CuS-TRPV1. (c) Representative TEM images of targeting of CuS-TRPV1 to VSMCs membrane rather than endosome or lysosome. Red arrows: localization of nanoparticles. (d) Representative TEM image of autophagosomes in VSMCs after photothermal activation of TRPV1 by CuS-TRPV1. Black circles: double-membrane structures of autophagosomes. (e) NIR laser treatment and representative images of Oil Red O-stained aortic root sections and complete aorta. Reproduced from Gao W, Sun Y, Cai M, et al. Copper sulfide nanoparticles as a photothermal switch for TRPV1 signaling to attenuate atherosclerosis. Nat Commun. 2018;9(1):231, with permission under the terms of the CC-BY license.56 Copyright 2018, Springer Nature. (B) Schematic illustration of the synthesis process of MJAMS/PTX/aV and the mechanism of treatment of atherosclerosis using the MJAMS/PTX/ aV coated balloon. Reproduced with permission from Huang Y, Li T, Gao W, et al. Platelet-derived nanomotor coated balloon for atherosclerosis combination therapy. J Mater Chem B. 2020;8(26):5765–5775. Copyright 2020, Royal Society of Chemistry.57 (C) Schematic illustration of lipid metabolism in foam cells after phototherapy with bTiO2-based nanoprobes irradiated by 808 nm NIR laser. Reprinted from Bioact Mater, 17, Dai T, He W, Tu S, et al. Black TiO(2) nanoprobe-mediated mild phototherapy reduces intracellular lipid levels in atherosclerotic foam cells via cholesterol regulation pathways instead of apoptosis. 18–28. With permission under the terms of the CC-BY license.58 Using different imaging contrast, NIR radiation under guidance benefits precise treatment toward criminal lesions. Chen et al developed a polymeric nanosystem (V-PPZ-NPs) that encapsulated phthalocyanine zinc (ZnPc) and perfluorohexane (PFH) and conjugated it to an anti-VEGFR-2 antibody so that it could target the vascular endothelial growth factor receptor (VEGFR-2) on vascular endothelial cells (VECs), since inhibition of neovascularization could help stabilize atherosclerotic plaque and prevent cardiovascular events. The NPs can be observed in vivo by ultrasound and photoacoustic (PA) imaging. Thus, laser irradiation can be guided to the optimal location to kill the surrounding cells.59 Another nanosystem targeting VECs in the same way utilizes 3 nm MnFe2O4, PFH and polylactic acid-glycolic acid (PLGA) shells to realize magnetic resonance imaging (MRI) T1 and T2 imaging except for PA imaging and ultrasound molecular imaging.60 Several studies have shown that photothermal agents made of nanomotors can deeply penetrate target lesions owing to their self-propelled motion ability and combination with biomolecules such as anti-vascular cell adhesion molecule-1 (anti-VCAM-1) antibodies (aV), which target overexpressed VCAM-1 in AS plaque. Huang et al constructed a nanomotor of mesoporous silica (MS) and platinum (Pt), which was loaded with the anti-proliferative drug paclitaxel (PTX), and coated with a platelet membrane to prevent drug leakage (Figure 3B). Under NIR irradiation, Pt provided immediate elimination of inflammatory macrophages and drove the nanomotor to penetrate the lesion for drug retention, achieving long-term anti-proliferation effects.57 A similar micromotor MMS/Au/PTX/VEGF/aV designed by Li et al promoted extra short-term endothelialization by releasing vascular endothelial growth factor (VEGF).61 Another type of nanomotor was mainly composed of L-arginine (LA), β-cyclodextrin (β-CD) and Au which had dual-power source. LA induced the generation of NO which could not only work as a driving force but also restore endothelial function and decrease oxidative stress, while Au could provide additional thermophoresis power under irradiation. In vivo, the dual-mode nanomotor showed better enrichment in plaque area than the single-mode one and led to fewer lesions after two months of treatment.62 Considering the irreversible damage to tissues caused by high thermal exposure and the risk of plaque destabilization raised by cell apoptosis itself, Dai et al designed a nanoprobe (bTiO2-HA-p) that allowed simultaneous PTT and PDT with a single NIR laser excitation, which resulted in a relatively low temperature (44.5 °C). This temperature could prevent cell apoptosis through activation of HSP27, promote ABCA1-dependent cholesterol efflux, and reduce SREBP2-induced biosynthesis of LDLR, thus decreasing uptake of lipid (Figure 3C). The nanoprobes attenuated as much as 12.8% of intracellular lipid accumulation compared to 5% and 3% of single PTT and PDT functional nanoprobes, respectively.58 Most photothermal agents are stimulated under laser irradiation with the first near-infrared window (NIR-I 700−950 nm), while the NIR-II (1000–1700 nm) laser has better tissue penetration ability and higher maximum permissible exposure than NIR-I. Copper sulfide (CuS) can be excited in the NIR-II region and deeply penetrate the lesions, thus Cao et al developed HA- and PEG-modified CuS/TiO2 heterostructure nanosheets (HA-HNSs), which combined SDT and PTT for early atherosclerotic treatment and inhibited atherosclerotic plaque progression, with no obvious toxicity.63 A Clinical Trial of PTT So far, there has only been one clinical trial of PTT on AS. Based on the promising outcomes of a preclinical trial on 101 Yucatan miniature swine,64 Kharlamov et al conducted the NANOM-FIM trial to evaluate the safety and feasibility of PTT mediated by NPs in patients with coronary artery disease with flow-limiting plaque. In the two experimental groups, silica-gold NPs were delivered in a bioengineered on-artery patch and the silica-gold iron-bearing NPs were transferred with targeted microbubbles and stem cells using a magnetic navigation system, while patients in the control group were treated with standard stent implantation. The ongoing clinical follow-up analyzed the event-free survival and exhibited a significantly lower risk of cardiovascular death in the first group than in the others (91.7% vs 81.7% and 80%, respectively; p < 0.05), and no complications associated with the target lesion occurred.65 A subsequent 5-year observational prospective cohort analysis of the NANOM-FIM trial illustrated that PTT with silica-gold NPs was safe with lower mortality, fewer major adverse cardiovascular events, and better revascularization of the target plaque compared to that with stent implantation. Efficient regression of coronary AS suggested a high potential to introduce PTT with silica-gold NPs in clinical practice. However, the trial also raised questions like nanotoxicity of the iron-bearing NPs and undesirable distribution in other tissue and organs, which were the major limitations for the development and implementation of nanotechnique in real clinical practice.66 Photodynamic Therapy (PDT) Although the feasibility of PTT in macrophage elimination has been proven, a variety of obstacles, such as accuracy, stability, and pharmacokinetics, still need to be overcome before clinical practice. PDT is a promising alternative to PTT and is composed of photosensitizers which can accumulate selectively in lesions and laser irradiation to induce local therapy. Under irradiation at the maximum absorption wavelength of the photosensitizer, energy is transformed and leads to ROS generation within the lesion, which induces cell death through different pathways. Thus, PDT can help stabilize and reduce plaque size, enlarge the stenosis vessel, and prevent occlusion, further prevent serious situations such as acute coronary syndrome.67 To date, there have been three generations of photosensitizers. The first generation includes hematoporphyrin derivative (HpD) and photofrin II (a purified form of HpD). The majority of second-generation compounds are derived from porphyrin structures, such as benzoporphyrins, phthalocyanines, and PpIX, whereas others, including mono-aspartyl chlorin e6 (NPe6), temoporfin, and pheophorbide derivatives, are related to the chlorin structure. These agents possess a clear composition and better light-absorption capability. The third generation uses nanotechnology to combine traditional photosensitizers with nanomaterials and different drugs for AS, so that they have higher efficacy of energy transformation, better accuracy, and various functions.68 The First Generation In 1983, Spears et al suggested that hematoporphyrin derivative (HpD) could not only be used clinically to localize malignant neoplasms but also be concentrated selectively within atherosclerotic plaque in both rabbits and Patas monkeys.69 Since its introduction into this field, HpD has been shown to induce quantitative changes in atheroma in thoracic aortic segments under 636 nm light irradiation.70 Intravascular irradiations in vivo after HpD injection can inhibit the growth of smooth muscle cells (SMCs) and the intimal hyperplasia response, improving restenosis after angioplasty.71 In addition, the rapid increase in HpD concentration upon local application makes it possible to use PDT to prevent restenosis after angioplasty without serious side-effects.72 Neave et al proved that in comparison to the normal arterial wall, AS plaque took up more dihematoporphyrin ether (photofrin II), especially 48 hours after the injection, and the plaque showed significant reduction 6 weeks after PDT.73 It is proved that morbid SMCs are more sensitive to PDT than healthy cells both in vitro and in vivo, confirming the feasibility of photofrin II–induced PDT.74,75 Several experiments have been conducted to determine the most appropriate dose and light intensity that can avoid extensive vascular injury and fully ablate plaque.76–78 Amemiya et al found that photofrin-mediated PDT using the pulse wave yttrium aluminum garnet-optical parametric oscillator (YAG-OPO) laser had better outcomes than a continuous-wave argon dye (Ar-dye) laser, penetrating deeper into the media of the vessel wall.79 However, the complex composition of the first generation was not good for the specificity and stability of the damage intensity, which limited the clinical exploitation and necessitated the development of the second generation. The Second Generation Verteporfin, a benzoporphyrin derivative monoacid ring A (BPD-MA), may have a better treatment effect than HpD, because its maximal absorption is at 692 nm, which is beyond the absorption spectrum of blood at 630 nm. BPD-MA can be preferentially taken up by atherosclerotic plaque both in vitro and in vivo,80 and this process can be enhanced when it is combined with plasma lipoproteins.81 Jain et al performed intra-arterial administration of liposomal verteporfin and intra-arterial light irradiation in ApoE-/- mice, inducing dramatic macrophage apoptosis in the plaque.82 Another porphyrin-like chemical, hematoporphyrin monomethyl ether (HMME), induces apoptosis of macrophages through the caspase-9 and caspase-3 activation pathways, and may become feasible for unstable AS plaque.83 5-Amino-levulinic acid-induced protoporphyrin IX (ALA-PpIX) is a naturally occurring porphyrin precursor, which produces PpIX to sensitize the target cells under irradiation. Based on various animal models84–87 and clinical studies,88 its efficacy in depleting the VSMC population and inhibiting intimal hyperplasia or constrictive remodeling has been demonstrated. However, ALA can cause hemodynamic changes (reductions in systemic and pulmonary pressure and lung resistance), which may limit its use in the cardiovascular field.89 Eldar et al first demonstrated that porphyrin-based copper phthalocyanine tetrasulfonate, a water-soluble phthalocyanine dye, preferentially accumulated in atheromatous plaque in rabbits.90 In vitro, when fluoride is added before exposure to light, chloroaluminium phthalocyanine (AIPc) and its derivatives can selectively destroy SMCs without affecting endothelial cells and skin fibroblasts, without causing cutaneous phototoxicity, which is the most common side-effect of photosensitizers.91 Later, de Vries et al found that ox-LDL could deliver AlPc specifically to macrophages in the plaque, mediated by scavenger receptors.92 Indocyanine green (ICG) has also been promising for the treatment of restenosis of carotid arteries under extracorporeal light irradiation because of the better tissue penetration of NIR light.93 The photosensitizer, mono-L-aspartyl chlorin e6 (NPe6), specifically deposits in the AS plaque94 and can be eradicated soon after intravenous injection, so it is less likely to photosensitize the skin, which is a common side effect for photosensitizers. Hayashi et al proved that NPe6 excited by irradiation could destroy the construction of the elastic fiber network and dissociate ester bonds of cholesterol esters in the plaque.95 Chlorin (Ce6) covalently attaches to bovine serum albumin can be recognized and internalized by class A Type-I scavenger receptors on macrophages.96 Ce6 can be released by cathepsin-B-activatable theranostic agent (L-SR15) and can not only provide diagnostic visualization but also eliminate macrophages selectively under laser irradiation, attenuating cathepsin B activity to stabilize the plaque.97 However, cardiovascular drugs such as atorvastatin and clopidogrel, which are often prescribed for patients with AS, have undesirable effects on photosensitizers, significantly affecting photosensitization in vitro without interfering with the cellular uptake of L-SR15.98 Weiss-Sadan et al also designed a photosensitizing quenched activity-based probe to target cysteine protease cathepsin that was typically expressed in arterial lesions and decreased the number of macrophages without affecting SMC and collagen contents.99 Another chlorin-based second-generation photosensitizer, motexafin lutetium, is also effective for AS. Irradiated by 732 nm light (2 J/cm2), motexafin lutetium produces cytotoxic singlet oxygen and leads to changes in mitochondrial membrane potential (MMP), release of cytochrome c, and activation of the caspase family. This process leads to phosphatidylserine externalization and induces apoptotic DNA fragmentation.100 Motexafin lutetium has been proved to target and significantly reduce atherosclerotic lesions of both vein graft intimal hyperplasia and graft coronary artery disease.101,102 With pheophorbide derivative PH-1126, which is selectively deposited in AS plaque, foam cells can be expelled from the endothelium to the intimal surface of the plaque.103 Pyropheophorbide-alpha methyl ester (MPPa)-mediated PDT generates ROS and induces apoptosis of RAW264.7 cells, while the damage level is related to the dose of light irradiation.104 Synthesized CTSB-PPP that can be cleaved by cathepsin B in plaque, releases pheophorbide a, and selectively destroys cells with high cathepsin activity upon irradiation.105 Wang et al discovered that curcumin (CUR)-mediated PDT could mediate autophagy in VSMCs treated with ox-LDL, further inhibiting the phenotypic transformation, migration, and formation of foam cells in vitro.106 The Third Generation Dendrimers exhibit relatively low cytotoxicity and good biocompatibility with nanomaterials as drug carriers. Polyamidoamine (PAMAM) dendrite delivery of a ZnPc photosensitizer is found to be considerably aggregated on AS plaque against healthy tissue.107 ALA dendrimers enhance porphyrin synthesis and have higher affinity with macrophages than endothelial cells, which is mainly mediated by caveolae-mediated endocytosis.108 Cespedes et al later proved that ALA dendrimers exhibited selective PDT response on foam cells rather than endothelial cells as well.109 Wennink et al encapsulated temoporfin in polymeric micelles to realize the selective elimination of macrophages and alleviate side-effects, though the instability of the system might be great obstacles for clinical use.110 Chlorin-based photosensitizers are one of the most popular study objects in third-generation photosensitizers because they can result in fluorescence emission and provide NIR fluorescence imaging with a high target-to-background ratio. Ce6 is conjugated to HA to form macrophage-targeted theranostic nanoparticles (MacTNP), which releases Ce6 and becomes phototoxic upon illumination inside activated macrophages in vitro, as excess ROS degrades the NPs.111 As shown in Figure 4A(a), Song et al combined dextran sulfate (DS) to Ce6, which could be engulfed by activated macrophages and foam cells via SR-A-mediated endocytosis. Optical imaging-guided DS-Ce6 photoactivation can detect inflammatory activity in plaques in vivo and simultaneously reduce plaque burden and inflammation by inducing apoptosis, autophagy (Figure 4A(b)), and efferocytosis.112 With the triple-helix structure of beta-glucan (Glu), Glu/Ce6 nanocomplexes can be recognized by the dectin-1 receptor of macrophages to mediate efficient membrane destruction and apoptosis of foam cells under irradiation.113 Figure 4 (A) (a) Schematic overview of photoactivation activated by DS-Ce6 targeting macrophage SR-A for autophagy induction and efferocytosis enhancement to regress atherosclerosis. (b) Confocal laser scanning microscopy images of the double IF of autophagy marker LC3 (green) and p62 (red) at 1 day and 1 week after treatment. Blue: nucleus stained with DAPI. **: P < 0.01, ***: P < 0.001. Reproduced with permission from Song JW, Ahn JW, Lee MW, et al. Targeted theranostic photoactivation on atherosclerosis. J Nanobiotechnology. 2021;19(1):338. under the terms of CC-BY license. Copyright 2021, Springer Nature.112 (B) Schematic diagram of UCNPs-Ce6-mediated PDT and the mechanism of autophagy. Reproduced with permission from Han XB, Li HX, Jiang YQ, et al. Upconversion nanoparticle-mediated photodynamic therapy induces autophagy and cholesterol efflux of macrophage-derived foam cells via ROS generation. Cell Death Dis. 2017;8(6):e2864. under the terms of CC-BY license. Copyright 2017, Springer Nature.114 (C) Schematically illustrate construction of theranostic TPZ/IR780@HSAeOPN NPs, which could precisely regress the vulnerable atherosclerotic plaques through a cascade of synergistic events triggered by careful lasers irradiation under the guidance of NIR fluorescence/MR imaging. Red “up” arrow refers to the decreased content of oxygen, and the red “down” arrow indicates the increase of ROS. Reprinted from Acta Pharm Sin B, 12(4), Xu M, Mao C, Chen H, et al. Osteopontin targeted theranostic nanoprobes for laser-induced synergistic regression of vulnerable atherosclerotic plaques. 2014–2028. Copyright 2022, with permission from Elsevier B.V.115 Upconversion nanoparticles (UCNPs) are helpful for the shallow penetration depth of Ce6 owing to their ability to convert NIR light into visible light and can be used for noninvasive imaging of deep tissues, drug delivery, and PDT. UCNPs-Ce6 can induce translocation of the proapoptotic factor Bax, release of cytochrome C, and upregulation of other apoptotic factors such as cleaved caspase-3/9 under laser irradiation, which leads to the apoptosis of THP-1 cells.116 ROS generated during UCNPs-Ce6-mediated PDT also activates the PI3K/Akt/mTOR signaling pathway which enhances autophagy, promotes cholesterol efflux through ABCA1 (Figure 4B), 114 and inhibits the expression of pro-inflammatory factors in M1 peritoneal macrophages,117 suggesting that it may help relieve the lipid burden of plaque in vivo. Ma et al designed a platelet membrane-coated nanostructure (PAAO-UCNPs) containing UCNPs and Ce6 for accurate localization of plaque and noninvasive PDT of AS. Platelet membrane coating facilitates the specific targeting of the therapeutic system to macrophage-derived foam cells.118 Xu et al constructed theranostic nanoprobes, TPZ/IR780@HSA-OPN, in which HAS-OPN referred to human serum albumin (HAS) decorated with a high-affinity peptide targeting osteopontin (OPN), an overexpressed marker of foam cells in atherosclerotic plaque. The nanoprobes encapsulated the photosensitizer IR780 and hypoxia-activatable tirapazamine (TPZ), implementing biological suppression of foam cells under less oxygen content induced by PDT. Under the guidance of fluorescence/MR imaging, the nanoprobes can greatly regress the vulnerable plaque and decrease the degree of carotid artery stenosis in vivo115 (Figure 4C). Despite the extensive clinical use of interventional laser irradiation, the application of PDT without external light irradiation presents an exciting field. Mu et al designed a chemiexcited system (FeCNPs) that also possessed T1-weighted contrast ability. In vivo MRI and in vitro experiments have shown that FeCNPs can accumulate in plaque and effectively eliminate macrophages, reducing plaque size and thickness.119 Clinical Research Adjuvant photofrin-mediated PDT was performed in five patients undergoing coronary stent deployment. Photofrin was administered using a local delivery balloon catheter to the stent-implanted lesions. After follow-up for eighteen months, no major adverse effects (myocardial infarction, coronary artery spasm, thrombosis, dissection, or aneurysmal dilatation) or in-stent restenosis were detected.120 In 1999, Jenkins et al investigated the effect of adjuvant ALA-mediated PDT after femoral percutaneous transluminal angioplasty (PTA). Seven patients with symptomatic restenosis or occlusion at the same site after the first conventional angioplasty were enrolled in the study. After oral administration of 5-ALA, adjuvant PDT was performed using a laser fiber within the angioplasty balloon, following a second femoral PTA. No adverse complications, death, or no restenosis occurred, suggesting the safety and the potential for restenosis prevention of endovascular PDT following PTA.88 Rockson et al demonstrated the safety and underlying therapeutic effects of motexafin lutetium in 47 patients with peripheral arterial AS. At follow-up, there were no deleterious vascular effects, and the standardized classification of clinical outcomes showed improvement in 29 patients (62%), no change in 17 patients (36%), and moderate worsening in 1 patient (2%).121 Another study was conducted in 75 patients undergoing percutaneous coronary intervention (PCI) and stent implantation, by following them for 6 months and confirmed that motexafin lutetium-mediated PDT was safe as an adjunct to PCI and established a well-tolerated maximum dosage and range of light doses.122 All the clinical trials mentioned have been conducted in a small sample size which leads to a less convincing result; thus, we need more large cohort studies to provide precise and comprehensive information of treatment effect and side effect. Ultrasound-Based Energy-Conversion Nano-Therapy The application of PDT and PTT raises the question of their effectiveness being limited by their relatively shallow penetration depth of light through tissue, and concentration on superficial lesions.123 Sonodynamic therapy (SDT), which originates as a proxy for PDT, is a perfect solution to the problem as ultrasound provides deep tissue penetration. SDT uses low-frequency and low-intensity ultrasound to activate sonosensitizers, which yields localized cytotoxicity mainly through the generation of ROS. It is interesting that many photosensitizers can also be applied as sonosensitizers. SDT was first used in 1989, proposed by Umemura et al124 and has been widely investigated for the treatment of tumors. Because light can be absorbed in the blood to some extent, SDT has become a prospective alternative to PDT for stabilizing and reducing the size of atherosclerotic plaque. In 2002, Arakawa et al reported that SDT could prevent neointimal hyperplasia after iliac artery stent implantation in rabbits, suggesting that SDT might have great clinical value in the treatment of cardiovascular disease.125 Sonosensitizers that have been investigated for AS include porphyrin derivatives, herbal-derived agents, and agents combined with nano materials. Porphyrin Derivatives Many photosensitizers, especially porphyrin derivatives, can also be applied as sonosensitizers. PpIX-mediated SDT can produce intracellular singlet oxygen and lead to cytoskeleton disruption and apoptosis of THP-1 macrophages in vitro,15 as well as significantly induce membrane permeabilization and facilitate the entry of drug into macrophages. This process promotes the anti-atherosclerotic effect of atorvastatin on THP-1-derived foam cells, including reduction of intracellular lipid droplets and increasing in cholesterol efflux by upregulation of PPARγ and ABCG1 expression.126 A large number of cytological experiments have already proved that 5-ALA-mediated SDT (ALA-SDT) can alleviate the burden of AS plaque, mainly by inducing apoptosis and autophagy of macrophages. For example, ALA-SDT induces ROS generation and a significant loss of MMP in vitro.127 Chen et al found that, in ALA-SDT, voltage-dependent anion channel 1 (VDAC1) led to Ca²+-mediated oxidative stress128 and mitochondria-caspase pathway and further accelerated the apoptosis of macrophages.129 Under the irritation of ALA-SDT, other potential mechanisms include the caspase-3/8 pathway17 and mitochondrial 18 kDa translocator protein (TSPO)130 that induces apoptosis, PPARγ-LXRα-ABCA1/ABCG1 pathway that promotes efferocytosis,131 ROS-AMPK-mTORC1 pathway that promotes autophagy132 and the increased heme oxygenase-1 (HO-1) expression that reduces ox-LDL-mediated impairment.133 Apart from macrophages, ALA-SDT also influences the equilibrium between different T lymphocyte subtypes and facilitates a switch toward Th2 polarization, which helps stabilize the plaque.18 In vivo ALA-SDT experiment performed both in rabbits and ApoE-/- mice suggested that the treatment eliminated macrophage and inhibited matrix degradation, which helped inhibit plaque progression and decrease the occurrence of plaque disruption.133,134 Increased mean lumen area and reduced artery stenosis also suggested that ALA-SDT was superior to treatment with atorvastatin alone, the standard of care for AS.135 Sinoporphyrin sodium (DVDMS) is also a porphyrin derivative which has been validated in SDT for AS. In plaque with high inclination to intraplaque hemorrhage, DVDMS-SDT can reduce iron retention by increasing ferroportin 1 (Fpn1) expression and activate the ROS-Nrf2-FPN1 pathway to reduce levels of the unstable iron pool and ferritin expression.136 Meanwhile, the process inactivates the expression of CD47 to promote efferocytosis, thereby reducing inflammation.137 In both rabbits and mice experiments, arterial inflammation and angiogenesis were conspicuously alleviated after DVDMS-mediated SDT, which was similar to the treatment outcome of intensive statin treatment for 3 months.138 Herbal Derived Agents Emodin, an anti-inflammatory agent, is found to be an effective sonosensitizer that induces the dissociation of intracellular filaments, aggregation of cytoskeleton proteins, and subsequent destruction of the cytoskeleton after SDT.139 Curcumin, extracted from the traditional Chinese herb Curcuma longa, has been shown to have the same effect on THP-1 cells under pulsed ultrasound irradiation (2 W/cm2 with 0.86 MHz).140 Hydroxyl-acetylated curcumin (HAC) removes the unstable hydroxy groups of curcumin141 and induces autophagy in THP-1 macrophages through the ROS-dependent PI3K/AKT/mTOR pathway with decreased lipid accumulation.142 Hydroxysafflor yellow A-mediated SDT causes autophagy in the same way.143 Berberine-mediated SDT works in the same way and increases cholesterol efflux in either macrophages or foam cells.144 Li et al found that hypericin (HY)-mediated SDT produced ROS, promoted translocation of BAX from the cytosol to the mitochondria, and released cytochrome C, thereby inducing macrophage apoptosis.145 Pseudohypericin (P-HY)-SDT has a similar function146 and triggers the translocation of transcription factor EB (TFEB) from lysosome to nucleus, which promotes autophagy and lysosome regeneration. This process inhibits lipid uptake by reducing CD36 and SR-A expression, and enhances ABCA1 expression, which increases the release of free fatty acids, further decreasing the lipid burden of macrophages.147 Agents Combined with Nano Materials Gonçalvez et al synthesized ALA:AuNPs, which could be used as both photosensitizers and sonosensitizers. AuNPs could help increase ROS generation and improve ROS function. SDT with ALA:AuNPs showed better results than PDT alone, as it decreased the viability of macrophages by 87% in 2 min.148 SDT using the methyl ester of aminolevulinic acid (MALA) with AuNPs showed similar results in the reduction of macrophages.149 Curcumin nanosuspensions using polyvinylpyrrolidone (PVPK30) and sodium dodecyl sulfate (SDS) as stabilizers possess better water solubility and bioavailability than curcumin and can decrease total cholesterol (TC) and LDL, leading to transdifferentiation from M1 to M2 macrophages.150 Yao et al fabricated PFP-HMME@PLGA/MnFe2O4-ramucirumab nanoparticles (PHPMR NPs), which could also serve as a contrast agent for MRI/photoacoustic/ultrasound imaging and were able to inhibit neovascularization (Figure 5A). Under low-intensity focused ultrasound irradiation and multimodal imaging guidance, PHPMR NPs mediated SDT targeted to the mitochondria of VECs in the plaque, promoted apoptosis by ROS-induced caspase activation, and suppressed proliferation and migration, further reduced neovascularization and stabilized vulnerable plaques (Figure 5B). 151 Figure 5 (A) Schematic illustration of the synthetic process for PFP–HMME@PLGA/MnFe2O4–Ram nanoplatform and the corresponding theranostic functionality for targeted MRI/PA/US multimodal imaging-guided sonodynamic plaque neovascularization therapy. (B) Representative histopathological staining of plaque sections showing less plaque neovascularization and stable atherosclerotic plaque on day 28 after treatment of the advanced plaques in rabbits. Red arrows indicate abnormal adventitial neovessels. White arrows indicate intraplaque hemorrhage. (A and B) were reproduced with permission under the terms of the CC-BY license. Yao J, Yang Z, Huang L, et al. Low-Intensity focused ultrasound-responsive ferrite-encapsulated nanoparticles for atherosclerotic plaque neovascularization theranostics. Adv Sci. 2021;8(19):e2100850.151 Clinical Research Among the sonosensitizers mentioned above, the comprehensive mechanism and efficacy of ALA have made it the best choice for clinical trials. ALA-SDT has been performed mainly for peripheral AS diseases with great therapeutic results. A pilot clinical trial recruited 16 participants with peripheral AS and divided them into two groups. Atorvastatin was administered alone in one group, and a combination of atorvastatin and ALA-SDT was implemented in another. The combination therapy successfully decreased the ratio of vessel diameter to stenosis at 4 weeks, and the decrease was maintained until 40 weeks, indicating efficient regression of atherosclerotic lesions.135 Nonetheless, strict RCT with larger sample size has to be conducted to eliminate the side effects of ALA on hemodynamics, while prolonged follow-up is necessary to state better long-term outcomes of SDT than modern treatment. DVDMS-SDT had also been in a pilot clinical study, in which the researchers used contrast-enhanced ultrasonography analysis and fludeoxyglucose F18-positron emission tomography-computed tomography (PET-CT) to evaluate the reduced neovascularization and arterial inflammation, respectively. The beneficial effect after 1-month treatment was almost equivalent to the therapeutic outcome after 3-month intensive statin treatment, with no side effects in follow-up screening, indicating DVDMS-SDT alone or in combination with a reduced statin dosage might have lesion-specific therapeutic effects in patients with vulnerable atherosclerotic plaques.136 Jiang et al conducted a phase-2, randomized, sham-controlled, double-blind clinical trial (NCT03457662) that enrolled 32 patients with symptomatic femoropopliteal AS. PET/CT confirmed that SDT prominently decreased the target-to-background ratio in the worst lesion by 0.53 on day 30 compared with control group. It also reduced the plaque area by 7.2%, improved local stenosis by 9.6%, and increased the score evaluating walking speed and physical function. Moreover, these improvements were maintained for up to 6 months, suggesting the great potential of SDT in clinical use to regress and stabilize plaque and improve the quality of life.152 Another trial of DVDMS-SDT (NCT03382249) on carotid AS is still underway, and we are looking forward to see its treatment effect and safety problems. The Challenge and Resolutions for Clinical Transition Nanomaterial-based therapies have emerged as a promising approach for the treatment of AS, as they can be designed to target specific cells or tissues and deliver therapeutic agents directly to the site of AS plaques. However, the clinical translation of nanomaterial-based therapies for AS presents several challenges that should be overcome to ensure their safety and efficacy. One of the main challenges of clinical transition for nanomaterial-based therapy for AS is to ensure their targeted delivery to the site of plaques. The design of nanomaterials can be optimized to achieve targeted delivery, but this requires a thorough understanding of the underlying mechanisms of AS and the specific cells and tissues involved in plaque formation. In this review, we have included many nanomaterials that are combined with specific antibody to target certain cells, such as macrophages or VSMCs more precisely, which is very promising in clinical use, especially precision medicine. Additionally, the pharmacokinetics and biodistribution of the nanomaterials must be carefully optimized to ensure that the therapeutic agents are delivered to the site of plaque formation while minimizing off-target effects. During design and synthesis of the materials, detrimental materials should be avoided or neutralized to minimize the risks, while more in vivo experiments have to be conducted to verify the good biocompatibility of materials in blood and tissues. Whether the materials will cause side effects is still the main concern before clinical application and should be clearly mentioned in the research. Another challenge of clinical transition for nanomaterial-based therapy for AS is the evaluation of their safety and efficacy. Preclinical studies have shown promising results for nanomaterial-based therapies for AS, but it is important to evaluate their safety and efficacy in clinical trials (Table 1). The clinical trials should be carefully designed to ensure that they meet all safety and efficacy standards and that the outcomes are clinically meaningful. Table 1 Safety of Nanomaterials for AS Therapy Treatment Materials Target Animal Dosage Biocompatibility Reference PTT Gold nanoparticles – ApoE-/- mice 0.4 μmol Au per g body weight Large amounts of Au elements were found in reticuloendothelial system (RES) of spleen and liver. No significant organ damage and inflammatory lesions were found. [47] Single-walled carbon nanotubes (SWNTs) - Mice 0.6 nmol/mouse No observed significant acute or chronic toxicity with regard to clinical and laboratory parameters, histology, or survival in mice followed 3 to 5 months after injection. [48] CuCo2S4 - ApoE-/- mice 80 μg mL-1, 100 μL No obvious toxicity to the major organs. No significant side effect on liver and kidney function. [49] Cu3BiS3 nanocrystals - ApoE-/- mice 10 mg kg-1 No significant change in the shape and size of the cells. Cu3BiS3 nanocrystals mainly accumulate in the kidney and spleen after the intravenous administration. The content in these two organs gradually decreased over time, indicating that Cu3BiS3 nanocrystals were mainly degraded through these two organs. [51] MoO2 nanoclusters - ApoE-/- mice 80 ppm The material mainly accumulate in the liver and spleen, which indicates that it was mainly degraded in these organs. No particle accumulation, obvious organ damage, or abnormal blood test results were noted due to local administration. [50] CS-CNCs@Ce6/DS Activated macrophages that express the type A scavenger receptor (SR-A) ApoE-/- mice - No obvious tissue abnormalities. No clear toxicity to the major organs. No obvious hepatotoxicity or nephrotoxicity in mice. NPs accumulating in vivo are excreted in the urine. The nanoplatform caused no evident side effects on atherosclerotic mice and could be metabolized in vivo. [53] CuS-TRPV1 Vascular smooth muscle cells (VSMCs) ApoE-/- mice 10 mg kg−1 No noticeable organ damage. [56] V-PPZ-NPs Vascular endothelial cells (VECs) Nude mice 5 mg mL−1 No significant weight change Major organs (heart, liver, spleen, lung and kidney) showed relatively normal histological structures. [59] PFH@PLGA/MnFe(2)O(4)-Ram nanoparticles Endothelial cells New Zealand rabbits 50 mg In SD rats, the levels of blood indices and histopathological aspects of the organs were almost unchanged. 30.33% of the UMFNPs (3 nm MnFe2O4) was excreted via the renal route and approximately 67.40% via the hepatobiliary route within 60 h. In plaque-bearing rabbits, on day 28 after PTT treatment, blood biochemical indices and weight were found unchanged compared with the baseline. There was negligible toxicity toward the main organs compared with the other groups. [60] H-CuS@DMSN-N and H-CuS@ DMSN-N=C-HA CD44-positive inflammatory macrophages New Zealand rabbits 3 mg mL-1 PT and FIB indexes showed no significant difference. APTT and TT values increased significantly. No obvious toxicity to liver or kidney. No significant histological toxicity to the major organs (heart, liver, spleen, lung, and kidney). [55] Ag/Ag2S Janus beads (AAS JBs) - ApoE-/- mice 100 μL 250 μg mL-1 The material cause little production of pro-inflammatory cytokines. The blood indexes obtained from blood biochemistry and complete blood panel analysis show no abnormal change. No inflammation or damage is observed in major organs. The material mainly accumulates in the kidney and liver, thus, it was assumed to be excreted through these two organs. [52] MJAMS/PTX/aV nanomotors - New Zealand rabbits - The material causes low hemolytic rates (<5%), which means no obvious damage to red blood cells (RBCs). No significant change in weight and major organs (heart, liver, spleen, lungs and kidneys). [57] MMS/Au/PTX/VEGF/aV VCAM-1 adhesion molecules on the surface of vascular endothelial cells ApoE-/- mice 50 mg kg−1 During the entire treatment, no mice showed any abnormalities, and all mice maintained a slight weight gain. The materials were mainly distributed in the kidneys, indicating that the micromotors may be excluded from the metabolic organs. At the beginning (24 h) and the end (2 months) of treatment, the micromotors would not cause obvious main organ toxicity and blood toxicity in the mice. [61] CD-LA-Au-aV No ApoE-/- mice 50 mg kg−1 The materials are almost completely cleared from the blood after 48 h. There is no abnormality or death in the mice during the whole treatment process, and the mice maintained slight weight gain. The materials did not cause toxicity in major organs and blood of mice at the starting (1d) and ending (60d) stages of treatment. Au element is found in the kidney, suggesting that the materials may be cleared in the metabolic organs. [62] HA-HNSs CD44+ cells ApoE-/- mice 10 mg/kg The half life was determined to be 2.71 h. HA-HNSs were partially metabolized in the liver and bound to monocytes and macrophages which were accumulated in the spleen. The hemolysis rate was less than 5%, suggesting the good hemocompatibility. The curves of mouse body weight showed no difference within 14 days after treatment. No systemic monocytes/macrophages decrease. There was no significant change in the liver and kidney function markers, as well as in the myocardial injury marker, echocardiogram, electrocardiogram and SpO2. No substantial tissue damage or inflammatory lesions in major organs were observed [63] Silica–gold nanoparticles - Yucatan swine with inherited hyper-low-density lipoprotein (LDL) cholesterolemia and hypercholesterolemia and bearing mutant alleles for apolipoprotein B The highest level of accumulation with NPs was directly in the studied plaques (> 900–1000 NPs/cm3; > 10 lg/mL), in the surrounding and neighboring healthy vascular tissues (400–600 units per cm3; 5.4 lg/mL), and in the liver and spleen, without signs of fibrosis or allergic responses. without any detrimental morphological dynamics. [64–66] PDT PM-PAAO-UCNPs - ApoE-/- mice 15 mg/kg The elimination half-life was 7.45 h for PM-PAAO-UCNPs. The reticuloendothelial system (RES) in liver and spleen, contained the most nanoparticles. No distinguishable injuries in major organs, and no obvious body weight change was observed. Levels of red blood cells, white blood cells, platelets, and hemoglobin were in normal ranges. There is little influence on liver and kidney functions. [118] mTHPC-loaded Ben-PCL-mPEG micelles - Balb/c nude mice 5.0 and 0.6% mTHPC loading Half-life times (t1/2 values): 1.5 hours. The materials was present in liver, spleen, kidney and lung in varying degrees after 4 hours of injection. [110] TPZ/IR780@HSA-OPN Overexpression of OPN in activated foamy macrophages ApoE-/- mice 1mg/kg IR780 The materials had no distinct hemolytic activity in comparison with the control. No significant alteration on the levels of biochemical blood biomarkers, indicating no obvious damage to liver, renal, and cardiac function and little risk of diabetes mellitus. No overt tissue damage. [115] FeCNPs - ApoE-/- mice Low-dose (0.5 mg/kg CPPO), and high-dose (2 mg/kg CPPO) FeCNPs The materials accumulated mainly in spleens and kidneys. The content of the materials in hearts, aortae, and livers at 48 hours decreased obviously from 24 hours. [119] SDT PFP-HMME@PLGA/MnFe2O4-ramucirumab nanoparticles (PHPMR NPs) Mitochondria of rabbit aortic endothelial cells (RAECs) New Zealand rabbits 1 mg mL−1, 50 mL, Fe + Mn: 269.7 mg L−1 No obvious changes were observed in the blood indexes or histopathological lesions in the organs. The rapid excretion of MnFe2O4 within 60 h is beneficial to minimize systematic toxicity. On day 28 after treatment, the changes in blood biochemical indices and weight fluctuations were negligible. [151] To overcome these challenges, several resolution suggestions have been proposed. Thorough understanding of the underlying mechanisms of AS and the specific cells and tissues involved in plaque formation have to be developed to inform the design of nanomaterials for targeted delivery. The pharmacokinetics and biodistribution of nanomaterials should be optimized to achieve targeted delivery to the site of plaque formation. More well-designed clinical trials are necessary to evaluate the safety and efficacy of nanomaterial-based therapies for AS, especially random clinical trials (RCT), which should include larger sample size than those we mentioned in the articles. Researchers need to engage with regulatory agencies to ensure that the therapies meet all safety and efficacy standards. For instance, most photosensitizers will result in skin photosensitivity, and this side effect should be clearly investigated. Meanwhile, collaboration with stakeholders including patients and healthcare providers is recommended to ensure that the benefits and risks of nanomaterial-based therapies for AS are clearly communicated. In summary, the clinical transition of nanomaterial-based therapies for AS presents several challenges that must be overcome to ensure their safety and efficacy. However, with careful design and optimization, rigorous preclinical and clinical testing, and close collaboration with regulatory agencies and stakeholders, nanomaterial-based therapies have the potential to transform the treatment of AS and improve patient outcomes. Conclusion and Prospects In this review, we systematically introduced the basic physical principles and biological mechanisms associated with ROS-derived therapeutic strategy, and summarized the recent advances in PDT, PTT, and SDT treatment strategies based on AS (Table 2). The creation of sensitizers for PTT, PDT, and SDT requires the design and development of more organic and inorganic sensitizer small molecules, a reliance on good nanomaterial science platforms, and the establishment of more efficient and rigorous treatment assessment methods. Table 2 Summary of Nanomaterials for Dynamic-Based AS Therapy Field Materials Constitution Radiation Dose for PTT and PTT; Ultrasound Frequency, Intensity/ Pressure, Time for SDT Treatment Effect Reference PTT Gold nanoparticles – 808 nm; 2 W cm−2; 10 min Decreased the number of macrophages and the thickness of intima media. [47] CuCo2S4 – 808 nm; 0.56 W cm−2 Less macrophage infiltration. Significant reduction of the intima and media thickness. [49] Cu3BiS3 nanocrystals – 808 nm; 0.4 W cm−2; 5 min Less macrophage infiltration. Lower thickness of the intima/media. [51] MoO2 nanoclusters – 808 nm; 0.69 W cm−2; 10 min Inhibited carotid wall hyperplasia caused by inflammation. [50] Silica–gold nanoparticles Stable silica-gold NPs with 1,4,7,10,13,16,21, 24-octaazabicyclo[8.8.8]hexacosane (azacryptand) 821 nm; 35–44 W cm−2; 7 min; for human Destruction of a plaque with regression of atheroma.High safety with better rate of mortality, major adverse cardiovascular events and target lesion revascularization at the long-term follow-up if compare with stent [64–66] CS-CNCs@Ce6/DS Chitosan (CS) and Carbon nanocages (CNCs) formed chitosan-coated carbon nanocage (CS-CNC), chlorin e6 (Ce6) and dextran sulfate (DS) 808 nm; 1 W cm−2; 10 min and 633 nm; 80 mW cm−2; 10 min Less SR-A expression, lower mRNA levels of pro-inflammatory markers in lesions, reduced proliferation and migration of SMCs and reduced lesion areas. [53] CuS-TRPV1 CuS NP conjugated with a TRPV1 monoclonal antibody 980 nm; 5 W cm−2; at the cardiac region for 30 cycles Activated TRPV1. Reduced lesion areas. [56] PFH@PLGA/MnFe(2)O(4)-Ram nanoparticles (NPs) 3 nm manganese ferrite (MnFe2O4) Perfluorohexane (PFH) Polylactic acid-glycolic acid (PLGA) Anti-VEGFR-2 antibody 808nm; 2 W cm−2; 15 min Reduced the number of macrophages, the percentage ratio of lipids, the HIF-1α score and increased collagen content. Reduced the inflammation level in the plaque. Resulted in more apoptotic neovessels with reduced hypoxic areas. Lumen area was increased and lesion area was decreased. [60] H-CuS@DMSN-N and H-CuS@ DMSN-N=C-HA Dendritic mesoporous silica nanoparticles (DMSN) Copper sulfide (CuS) Anticoagulant drug heparin (Hep) Oxidized hyaluronic acid (oxi-HA) 980 nm; 2.0 W cm−2; 10 min Ablated macrophages and thrombus. [55] Ag/Ag2S Janus beads (AAS JBs) Ag, Ag2S and polyvinylpyrrolidone (PVP) 808 nm; 0.7 W cm−2; 5 min Lower numbers of macrophages and lower thickness of intima. [52] MJAMS/PTX/aV nanomotors Aminated mesoporous silica (AMS) modified with platinum (Pt) was on one-side to form Janus AMS (JAMS). Modified with anti-proliferative drug PTX and anti-vascular cell adhesion molecule-1 (anti-VCAM-1) antibody. Wrapped with platelet membrane. NIR laser; 2.0 W cm−2; 10 min Reduced hyperplasia area. [57] MMS/Au/PTX/VEGF/aV mesoporous-macroporous silica (MMS) Gold nanoparticles (Au NPs) Vascular endothelial growth factor (VEGF) Antiproliferative drug paclitaxel (PTX) anti-VCAM-1 polyclonal antibody (aV) NIR laser; 2.5 W cm−2; 10 min Reduced formation of lipid plaques and lesion area. [61] CD-LA-Au-aV L-arginine (LA) β-cyclodextrin (β-CD), Au nanoparticles (Au NPs) 6-(mercaptohexyl) ferrocene (FcH) Anti-VCAM-1 monoclonal antibody (aV) 808 nm; 2.5 W cm−2; 10 min Reduced plaque area. Decreased index of LDL and increased index of HDL in blood. [62] bTiO2-HA-p Black TiO2 (bTiO2) nanoparticles Hyaluronan (HA) Porphine 808 nm; 1 W cm−2; 10 min; in vitro Reduced the intracellular lipid burden without inducing evidently apoptosis or necrosis. [58] HA-HNSs Hyaluronic acid (HA)- and PEG-modified CuS/TiO2 heterostructure nanosheets 1064 nm; 0.8 W cm-2; 10 min Ultrasound intensity at 0.5 W cm−2; 10 min Lower ratio of the lesion area and decreased content of intraplaque macrophage and proinflammatory cytokines. [63] PDT Hematoporphyrin derivative (HpD) - 636 nm; total fluence of 27 mJ cm−2; 10min Induced cell necrosis in intimal and media and inhibited initial hyperplasia. [69–72] dihematoporphyrin ether (photofrin II) - 632 nm; 150 mW cm−2; 30 J cm−2 (total output); for human Decreased intimal thickness and enlarged luminal diameter. Reduced lipid content in plaque. In patients no adverse events such as photodermatosis, or myocardial ischemia had occurred. No in-stent restenosis was observed. [73–79,120] ALA-PPIX - 635 nm; a total dose of 50 J cm−2; 500 and 1500 s, and 200 and 390 s for the iliacs and coronaries, respectively in human Decreased number of VSMCs in the media and inhibited intimal hyperplasia. Increased maximal lumen diameter. Patients tolerated the procedure well without adverse complications or death. All were rendered asymptomatic throughout the study interval. All vessels remained patent and no lesion attained the duplex definition of restenosis. [84–89] Indocyanine green (ICG) - 780 nm; 4 J cm−2 Reduced arterial wall thickness. [93] e6 - 670 nm; 72.6 mW cm−2 (dose rate); 4.4 J cm−2 (irradiation dose) Damaged elastic fiber network in plaque. Dissociated ester bonds of cholesterol esters. Attenuated cathepsin B-related signal.Decreased macrophage infiltration by inducing apoptosis without affecting plaque size or number of VSMCs. [94–98] Activity-Based Probed (PS-qABP) Photosensitizer (bChlo) and quencher (QC-1) attached to GB111-NH2 scaffold 760nm, 50mW, 20min Reduced immune cell content without affecting VSMC and collagen contents. [99] Motexafin lutetium - 730±6 nm; 100 to 400 J/cm-fiber; 12min; in human Increased number of apoptotic cells. Therapy was well tolerated. No deleterious vascular effects. Several secondary end points suggested a favorable therapeutic effect. [100–102,121,122] PH-1126 9-desoxo-9 (R,S) -hydroxy-10 (R) -N,N-dimethylaminoethyl-pheophorbide-a 647 nm; 100 J cm−2 (irradiation dose) Endothelial layer of the lesion was ruptured and numerous teardrop-shaped cells resembled foam cells were observed. [103] Pyropheophorbide-alpha methyl ester (MPPa) - 630 nm; 30 mW cm−2; for the indicated time; in vitro Induced RAW264.7 cell apoptosis. Decreased secretion of inflammatory cytokines. [104] Curcumin (CUR) - 425 nm; 40 mW cm−2; 6 hours Promoted autophagy and inhibited differentiation of VSMCs. Reduced cell migration and lipid droplet numbers in cells. [106] ALA dendrimers - 400–700 nm; 0.5 mW cm−2 (power density); 0.15 or 0.6 J cm−2 (irradiation dose); in vitro High selectivity on macrophages as compared to endothelial cells. Induced cell death. [108,109] DS-Ce6 Dextran sulfate(DS) and Ce6 670 nm; 1 W cm−2 (fluence rate); 150 seconds Reduced plaque burden and inflammation. Increased apoptotic macrophages and induced autophagy and efferocytosis. [112] Glu/Ce6 nanocomplexes Glucan and Chlorin e6 670 nm; 50 mW cm−2; 1 min Induced significant membrane damage and apoptosis of foam cells. [113] UCNPs-Ce6 Chlorin e6 (Ce6), onto silica-coated upconversion nanoparticles 980nm; 1.0 W cm−2; 60s; in vitro Resulted in apoptosis of THP-1 macrophages and enhanced the cholesterol efflux and the induction of autophagy. [114,116,117] PM-PAAO-UCNPs Incorporating UCNPs and Ce6 into polyacrylic acid-n-octylamine (PAAO) micelles, followed by PM coating 980 nm; 10 mW cm−2; 30 min Alleviated plaque progression. Induced foam cell apoptosis and ameliorated inflammation. Triggered lipid efflux from foam cells. Increased the expression of cell marker of VSMCs. [118] TPZ/IR780@HSA-OPN Human serum albumin (HSA) conjugated with a high affinity-peptide targeting osteopontin (OPN) and encapsulated with photosensitizer IR780 and hypoxia-activatable tirapazamine (TPZ) 808 nm; 1.5 W cm−2; 3 min Resulted in plaque ablation and amplified hypoxia. Suppressed foamy macrophages. Reduced plaque area and degree of carotid artery stenosis. [115] FeCNPs Monomethoxypolyethylene glycol-block-poly(L-lysine) modified with Ce6 and 3,4-DA (mPEG-Plys -[DA-Ce6]) was synthesized and self-assembled with CPPO. To improve stability, Fe3+ was introduced to coordinate with catechol groups - Eliminated macrophages and prevented progression of plaque. Reduced plaque size and thickness [119] SDT PpIX - 1.0 MHz; 0.5 W cm−2; 5 min; in vitro Induced both apoptosis and necrosis of THP-1 macrophages [15,126] 5-ALA 1.0 MHz; 1.5W cm−2 for rabbits and 0.8 W cm−2 for mice;15min Decreased size of the atherosclerotic plaque and enlarged lumen. Reduction in lesional macrophages and lipids. [17,127–135] Sinoporphyrin sodium (DVDM) - 1.0 MHz; 1.5 W cm−2 for rabbits, 0.8 W cm−2 for mice; 15 min For human: 2.1 W cm−2 in the common femoral artery and superficial femoral artery, and 1.8 W cm−2 in popliteal artery Reduced arterial inflammation and angiogenesis. Alleviated iron retention in hemorrhagic plaques. Enhanced macrophage efferocytosis. inhibited the progression of atherosclerosis, reduced the macrophage content, and increased the smooth muscle cell content. [136–138,152] Emodin - 0.86 MHz; 0.44 W cm−2; 15 min; in vitro Decreased cell viability and increased apoptotic and necrotic cells. [139] Hydroxyl acetylated curcumin (HAC) - 1.0 MHz; 0.5 W cm−2; for indicated time; in vitro Induced apoptosis of macrophages and autophagy with decrease in the lipid uptake. [1,41,142] Curcumin nanosuspensions (Cur-ns) Polyvinylpyrrolidone (PVPK30), sodium dodecyl sulfate (SDS) and curcumin 1.0 MHz; 0.4 W cm−2; 15 minutes Reduced the level of total cholesterol (TC) and low density lipoprotein (LDL). Promoted the transformation from M1 to M2 macrophages. Relieved atherosclerosis syndrome. [150] ALA:AuNPs Methyl ester of aminolevulinic acid (MALA), gold nanoparticles (MALA:AuNPs) and polyethylene glycol (PEG) 1 MHz; 1 W cm−2; 2 minutes Culminated with total macrophage reduction [1,48,149] PFP-HMME@PLGA/MnFe2O4-ramucirumab nanoparticles (PHPMR NPs) Manganese ferrite (MnFe2O4), hematoporphyrin monomethyl ether (HMME), perfluoropentane (PFP), polylactic acid-glycolic acid (PLGA) shells and anti-VEGFR-2 antibody. 1 MHz; 1.5 Wcm−2; 15 min Induced apoptosis in neovessel endothelial cells and improved hypoxia. Reduced the density of neovessels, subsequently inhibiting intraplaque hemorrhage and inflammation and stabilizing the plaque. [151] With the development of nanotechnology and the demand for effective treatments for AS, a variety of novel nanomaterials combined with conversion-based characteristics have been discovered and applied in animal models. However, the clinical transition of these nanomaterials poses some challenges. The first challenge is that the inherent AS microenvironment (such as its acidic nature) may weaken PTT, PDT, and SDT treatment effects. The development of atherosclerotic plaque microenvironment-responsive multifunctional sonosensitizers is highly effective. In addition, biomolecules such as proteins, lipids, and nutrient molecules are abundantly distributed in the AS microenvironment. These biomolecules can absorb onto the surface of the NPs to form a layer called the “biomolecule corona”. For example, the biomolecule corona may alter their biodistribution in the body and modulate their interaction with the immune system.153 Therefore, it is necessary to consider the interactions between NPs and biological molecules to understand the behaviors in AS microenvironments. The second challenge is, as the recent research has revealed, the targeting abilities of sonosensitizers or photosensitizers are very low. To improve the AS targeting ability, some excellent targeting ligands, such as cell membrane coating, specific antibodies and protein modification, and extracellular vesicle loading, have been developed. Targeting therapy based on nanotechnology has brought new possibilities for the individualized precision treatment of atherosclerotic plaques. Finally, owing to the complexity of AS, nano-dynamic therapy methods usually kill plaque foam cells. A synergistic combination of multiple treatments, such as immunotherapy, will greatly improve the therapeutic efficacy of AS. With regard to practical applications, the development of a complete AS therapy (including PDT and other therapies) will lead to better clinical outcomes and give patients hope. Acknowledgments This work was funded by the National Natural Science Foundation of China (No.81972444) and the National High Level Hospital Clinical Research Funding (No.2022-PUMCH-A-231). Disclosure The authors declare no conflicts of interest in this work. ==== Refs References 1. Cardiovascular diseases (CVDs). Available from: https://www.who.int/en/news-room/fact-sheets/detail/cardiovascular-diseases-(cvds). Accessed November, 2022. 2. Bäck M, Yurdagul A, Tabas I, Öörni K, Kovanen PT. Inflammation and its resolution in atherosclerosis: mediators and therapeutic opportunities. Nat Rev Cardiol. 2019;16 (7 ):389–406. doi:10.1038/s41569-019-0169-2 30846875 3. Bentzon JF, Otsuka F, Virmani R, Falk E. Mechanisms of plaque formation and rupture. Circ Res. 2014;114 (12 ):1852–1866. doi:10.1161/CIRCRESAHA.114.302721 24902970 4. Tian J, Ren X, Vergallo R, et al. Distinct morphological features of ruptured culprit plaque for acute coronary events compared to those with silent rupture and thin-cap fibroatheroma: a combined optical coherence tomography and intravascular ultrasound study. J Am Coll Cardiol. 2014;63 (21 ):2209–2216. doi:10.1016/j.jacc.2014.01.061 24632266 5. T J, D H, T C, et al. Prevalence and characteristics of TCFA and degree of coronary artery stenosis: an OCT, IVUS, and angiographic study. J Am Coll Cardiol. 2014;64 (7 ):672–680. doi:10.1016/j.jacc.2014.05.052 25125298 6. Ascer E, Bertolami MC, Venturinelli ML, et al. Atorvastatin reduces proinflammatory markers in hypercholesterolemic patients. Atherosclerosis. 2004;177 (1 ):161–166. doi:10.1016/j.atherosclerosis.2004.07.003 15488879 7. Lobatto ME, Fuster V, Fayad ZA, Mulder WJM. Perspectives and opportunities for nanomedicine in the management of atherosclerosis. Nat Rev Drug Discov. 2011;10 (11 ):835–852. doi:10.1038/nrd3578 22015921 8. Gisterå A, Hansson GK. The immunology of atherosclerosis. Nat Rev Nephrol. 2017;13 (6 ):368–380. doi:10.1038/nrneph.2017.51 28392564 9. X-H Y, Zhang D-W, Zheng X-L, Tang C-K. Cholesterol transport system: an integrated cholesterol transport model involved in atherosclerosis. Prog Lipid Res. 2019;73 :65–91. doi:10.1016/j.plipres.2018.12.002 30528667 10. Zhai M, Gong S, Luan P, et al. Extracellular traps from activated vascular smooth muscle cells drive the progression of atherosclerosis. Nat Commun. 2022;13 (1 ):7500. doi:10.1038/s41467-022-35330-1 36473863 11. Lin X, Ouyang S, Zhi C, et al. Focus on ferroptosis, pyroptosis, apoptosis and autophagy of vascular endothelial cells to the strategic targets for the treatment of atherosclerosis. Arch Biochem Biophys. 2022;715 :109098. doi:10.1016/j.abb.2021.109098 34856194 12. Luo Y, Li J, Huang C, Wang X, Long D, Cao Y. Graphene oxide links alterations of anti-viral signaling pathways with lipid metabolism via suppressing TLR3 in vascular smooth muscle cells. Mol Omics. 2022;18 (8 ):779–790. doi:10.1039/d2mo00086e 35912640 13. Ochsner M. Photophysical and photobiological processes in the photodynamic therapy of tumours. J Photochem Photobiol B. 1997;39 (1 ):1–18. doi:10.1016/S1011-1344(96)07428-3 9210318 14. Castano AP, Demidova TN, Hamblin MR. Mechanisms in photodynamic therapy: part one-photosensitizers, photochemistry and cellular localization. Photodiagnosis Photodyn Ther. 2004;1 (4 ):279–293. doi:10.1016/S1572-1000(05)00007-4 25048432 15. Guo S, Sun X, Cheng J, et al. Apoptosis of THP-1 macrophages induced by protoporphyrin IX-mediated sonodynamic therapy. Int J Nanomedicine. 2013;8 :2239–2246. doi:10.2147/IJN.S43717 23818780 16. Dan J, Sun X, Li W, et al. 5-aminolevulinic acid-mediated sonodynamic therapy promotes phenotypic switching from dedifferentiated to differentiated phenotype via reactive oxygen species and p38 mitogen-activated protein kinase in vascular smooth muscle cells. Ultrasound Med Biol. 2015;41 (6 ):1681–1689. doi:10.1016/j.ultrasmedbio.2014.12.664 25796412 17. Tian F, Yao J, Yan M, et al. 5-Aminolevulinic Acid-Mediated Sonodynamic Therapy Inhibits RIPK1/RIPK3-Dependent Necroptosis in THP-1-Derived Foam Cells. Sci Rep. 2016;6 :21992. doi:10.1038/srep21992 26911899 18. Yang Y, Liu Y, Chen X, et al. 5-aminolevulinic acid-mediated sonodynamic therapy alleviates atherosclerosis via enhancing efferocytosis and facilitating a shift in the Th1/Th2 balance toward Th2 polarization. Cell Physiol Biochem. 2018;47 (1 ):83–96. doi:10.1159/000489751 29763901 19. Crum LA, Roy RA. Sonoluminescence. Science. 1994;266 (5183 ):233–234. doi:10.1126/science.266.5183.233 17771441 20. Lohse D. Sonoluminescence: cavitation hots up. Nature. 2005;434 (7029 ):33–34. doi:10.1038/434033a 15744286 21. Yasui K, Tuziuti T, Lee J, Kozuka T, Towata A, Iida Y. Numerical simulations of acoustic cavitation noise with the temporal fluctuation in the number of bubbles. Ultrason Sonochem. 2010;17 (2 ):460–472. doi:10.1016/j.ultsonch.2009.08.014 19751988 22. Suslick KS. Sonochemistry. Science. 1990;247 (4949 ):1439–1445. doi:10.1126/science.247.4949.1439 17791211 23. Huber PE, Debus J. Tumor cytotoxicity in vivo and radical formation in vitro depend on the shock wave-induced cavitation dose. Radiat Res. 2001;156 (3 ):301–309. doi:10.1667/0033-7587(2001)156[0301:TCIVAR]2.0.CO;2 11500139 24. Villeneuve L, Alberti L, Steghens JP, Lancelin JM, Mestas JL. Assay of hydroxyl radicals generated by focused ultrasound. Ultrason Sonochem. 2009;16 (3 ):339–344. doi:10.1016/j.ultsonch.2008.09.007 19010709 25. Xu H, Sun X, Yao J, et al. The decomposition of protoporphyrin IX by ultrasound is dependent on the generation of hydroxyl radicals. Ultrason Sonochem. 2015;27 :623–630. doi:10.1016/j.ultsonch.2015.04.024 25934126 26. Pankaj. Aqueous Inorganic Sonochemistry. In: Ashokkumar M, editor. Theoretical and Experimental Sonochemistry Involving Inorganic Systems. Berlin: Springer-Verlag; 2011:213–271. 27. Yasui K. Multibubble sonoluminescence from a theoretical perspective. Molecules. 2021;26 (15 ):4624. doi:10.3390/molecules26154624 34361777 28. Kondo T, Krishna CM, Riesz P. Pyrolysis radicals formed by ultrasound in aqueous solutions of nucleotides: a spin trapping study. Int J Radiat Biol. 1990;57 (1 ):23–33. doi:10.1080/09553009014550311 1967291 29. Riesz P, Kondo T. Free radical formation induced by ultrasound and its biological implications. Free Radic Biol Med. 1992;13 (3 ):247–270. doi:10.1016/0891-5849(92)90021-8 1324205 30. Torii T, Yasui K, Yasuda K, et al. Generation and consumption rates of OH radicals in sonochemical reactions. Res Chem Intermed. 2004;30 (7 ):713–721. doi:10.1163/1568567041856918 31. Yasui K, Tuziuti T, Kozuka T, Towata A, Iida Y. Relationship between the bubble temperature and main oxidant created inside an air bubble under ultrasound. J Chem Phys. 2007;127 (15 ):154502. doi:10.1063/1.2790420 17949168 32. Yasui K, Tuziuti T, Iida Y, Mitome H. Theoretical study of the ambient-pressure dependence of sonochemical reactions. J Chem Phys. 2003;119 (1 ):346–356. doi:10.1063/1.1576375 33. Yusof NSM, Anandan S, Sivashanmugam P, Flores EMM, Ashokkumar M. A correlation between cavitation bubble temperature, sonoluminescence and interfacial chemistry - A mini review. Ultrason Sonochem. 2022;85 :105988. doi:10.1016/j.ultsonch.2022.105988 35344863 34. Winyard PG, Moody CJ, Jacob C. Oxidative activation of antioxidant defence. Trends Biochem Sci. 2005;30 (8 ):453–461. doi:10.1016/j.tibs.2005.06.001 15996871 35. Nowak WN, Deng J, Ruan XZ, Xu Q. Reactive oxygen species generation and atherosclerosis. Arterioscler Thromb Vasc Biol. 2017;37 (5 ):e41–e52. doi:10.1161/ATVBAHA.117.309228 28446473 36. Moloney JN, Cotter TG. ROS signalling in the biology of cancer. Semin Cell Dev Biol. 2018;80 :50–64.28587975 37. Zhao Y, Zhao B. Oxidative stress and the pathogenesis of Alzheimer’s disease. Oxid Med Cell Longev. 2013;2013 :316523. doi:10.1155/2013/316523 23983897 38. Bayir S, Barras A, Boukherroub R, et al. Mesoporous silica nanoparticles in recent photodynamic therapy applications. Photochem Photobiol Sci. 2018;17 (11 ):1651–1674. doi:10.1039/c8pp00143j 30022180 39. Yu BP. Cellular defenses against damage from reactive oxygen species. Physiol Rev. 1994;74 (1 ):139–162. doi:10.1152/physrev.1994.74.1.139 8295932 40. Forrester SJ, Kikuchi DS, Hernandes MS, Xu Q, Griendling KK. Reactive oxygen species in metabolic and inflammatory signaling. Circ Res. 2018;122 (6 ):877–902. doi:10.1161/CIRCRESAHA.117.311401 29700084 41. Fiers W, Beyaert R, Declercq W, Vandenabeele P. More than one way to die: apoptosis, necrosis and reactive oxygen damage. Oncogene. 1999;18 (54 ):7719–7730. doi:10.1038/sj.onc.1203249 10618712 42. Scherz-Shouval R, Elazar Z. Regulation of autophagy by ROS: physiology and pathology. Trends Biochem Sci. 2011;36 (1 ):30–38. doi:10.1016/j.tibs.2010.07.007 20728362 43. Aprotosoaie AC, Costache A-D, Costache -I-I. Therapeutic strategies and chemoprevention of atherosclerosis: what do we know and where do we go? Pharmaceutics. 2022;14 (4 ):722. doi:10.3390/pharmaceutics14040722 35456556 44. Dai T, He W, Yao C, et al. Applications of inorganic nanoparticles in the diagnosis and therapy of atherosclerosis. Biomater Sci. 2020;8 (14 ):3784–3799. doi:10.1039/D0BM00196A 32469010 45. Lukianova-Hleb EY, Mrochek AG, Lapotko DO. Method for disruption and re-canalization of atherosclerotic plaques in coronary vessels with photothermal bubbles generated around gold nanoparticles. Lasers Surg Med. 2009;41 (3 ):240–247. doi:10.1002/lsm.20749 19291755 46. Kosuge H, Nakamura M, Oyane A, et al. Potential of gold nanoparticles for noninvasive imaging and therapy for vascular inflammation. Mol Imaging Biol. 2021;24 :692–699. doi:10.1007/s11307-021-01654-5 34580810 47. Qin J, Peng Z, Li B, et al. Gold nanorods as a theranostic platform for in vitro and in vivo imaging and photothermal therapy of inflammatory macrophages. Nanoscale. 2015;7 (33 ):13991–14001. doi:10.1039/C5NR02521D 26228112 48. Kosuge H, Sherlock SP, Kitagawa T, et al. Near infrared imaging and photothermal ablation of vascular inflammation using single-walled carbon nanotubes. J Am Heart Assoc. 2012;1 (6 ):e002568. doi:10.1161/jaha.112.002568 23316318 49. Zhang X, Liu J, Yang X, et al. CuCo2S4 nanocrystals as a nanoplatform for photothermal therapy of arterial inflammation. Nanoscale. 2019;11 (19 ):9733–9742. doi:10.1039/C9NR00772E 31066405 50. Wang X, Wu X, Qin J, et al. Differential phagocytosis-based photothermal ablation of inflammatory macrophages in atherosclerotic disease. ACS Appl Mater Interfaces. 2019;11 (44 ):41009–41018. doi:10.1021/acsami.9b12258 31599564 51. Lu R, Zhu J, Yu C, Nie Z, Gao Y. Cu3BiS3 nanocrystals as efficient nanoplatforms for CT imaging guided photothermal therapy of arterial inflammation. Front Bioeng Biotechnol. 2020;8 :981. doi:10.3389/fbioe.2020.00981 32923437 52. Peng X, Liu J, Li B, et al. Janus Ag/Ag(2)S beads as efficient photothermal agents for the eradication of inflammation and artery stenosis. Nanoscale. 2019;11 (42 ):20324–20332. doi:10.1039/c9nr04804a 31641707 53. Liu J, Zhou B, Guo Y, et al. SR-A-targeted nanoplatform for sequential photothermal/photodynamic ablation of activated macrophages to alleviate atherosclerosis. ACS Appl Mater Interfaces. 2021;2021 :1. 54. Oh B, Lee CH. Development of Man-rGO for targeted eradication of macrophage ablation. Mol Pharm. 2015;12 (9 ):3226–3236. doi:10.1021/acs.molpharmaceut.5b00181 26161461 55. Liu S, Zhao Y, Shen M, et al. Hyaluronic acid targeted and pH-responsive multifunctional nanoparticles for chemo-photothermal synergistic therapy of atherosclerosis. J Mater Chem B. 2022;10 (4 ):562–570.34982089 56. Gao W, Sun Y, Cai M, et al. Copper sulfide nanoparticles as a photothermal switch for TRPV1 signaling to attenuate atherosclerosis. Nat Commun. 2018;9 (1 ):231. doi:10.1038/s41467-017-02657-z 29335450 57. Huang Y, Li T, Gao W, et al. Platelet-derived nanomotor coated balloon for atherosclerosis combination therapy. J Mater Chem B. 2020;8 (26 ):5765–5775. doi:10.1039/D0TB00789G 32519722 58. Dai T, He W, Tu S, et al. Black TiO(2) nanoprobe-mediated mild phototherapy reduces intracellular lipid levels in atherosclerotic foam cells via cholesterol regulation pathways instead of apoptosis. Bioact Mater. 2022;17 :18–28. doi:10.1016/j.bioactmat.2022.01.013 35386468 59. Chen YL, Liu FQ, Guo Y, et al. PA/US dual-modality imaging to guide VEGFR-2 targeted photothermal therapy using ZnPc-/PFH-loaded polymeric nanoparticles. Biomater Sci. 2018;6 (8 ):2130–2143. doi:10.1039/c8bm00213d 29916500 60. Yang Z, Yao J, Wang J, et al. Ferrite-encapsulated nanoparticles with stable photothermal performance for multimodal imaging-guided atherosclerotic plaque neovascularization therapy. Biomater Sci. 2021;9 (16 ):5652–5664. doi:10.1039/d1bm00343g 34259244 61. Li X, Wu R, Chen H, et al. Near-infrared light-driven multifunctional tubular micromotors for treatment of atherosclerosis. ACS Appl Mater Interfaces. 2021;13 (26 ):30930–30940. doi:10.1021/acsami.1c03600 34156244 62. Wu Z, Wu R, Li X, et al. Multi-pathway microenvironment regulation for atherosclerosis therapy based on beta-cyclodextrin/L-arginine/au nanomotors with dual-mode propulsion. Small. 2022;18 (9 ):e2104120. doi:10.1002/smll.202104120 34918450 63. Cao Z, Yuan G, Zeng L, et al. Macrophage-targeted sonodynamic/photothermal synergistic therapy for preventing atherosclerotic plaque progression using CuS/TiO heterostructured nanosheets. ACS Nano. 2022;16 (7 ):10608–10622. doi:10.1021/acsnano.2c02177 35759554 64. Kharlamov AN, Gabinsky JL. Plasmonic photothermic and stem cell therapy of atherosclerotic plaque as a novel nanotool for angioplasty and artery remodeling. Rejuvenation Res. 2012;15 (2 ):222–230. doi:10.1089/rej.2011.1305 22533437 65. Kharlamov AN, Tyurnina AE, Veselova VS, Kovtun OP, Shur VY, Gabinsky JL. Silica-gold nanoparticles for atheroprotective management of plaques: results of the NANOM-FIM trial. Nanoscale. 2015;7 (17 ):8003–8015. doi:10.1039/c5nr01050k 25864858 66. Kharlamov AN, Feinstein JA, Cramer JA, Boothroyd JA, Shishkina EV, Shur V. Plasmonic photothermal therapy of atherosclerosis with nanoparticles: long-term outcomes and safety in NANOM-FIM trial. Future Cardiol. 2017;13 (4 ):345–363. doi:10.2217/fca-2017-0009 28644056 67. Rockson SG, Lorenz DP, Cheong WF, Woodburn KW. Photoangioplasty: an emerging clinical cardiovascular role for photodynamic therapy. Circulation. 2000;102 (5 ):591–596. doi:10.1161/01.cir.102.5.591 10920074 68. Kou J, Dou D, Yang L. Porphyrin photosensitizers in photodynamic therapy and its applications. Oncotarget. 2017;8 (46 ):81591–81603. doi:10.18632/oncotarget.20189 29113417 69. Spears JR, Serur J, Shropshire D, Paulin S. Fluorescence of experimental atheromatous plaques with hematoporphyrin derivative. J Clin Invest. 1983;71 (2 ):395–399. doi:10.1172/JCI110782 6822671 70. Litvack F, Grundfest WS, Forrester JS, et al. Effects of hematoporphyrin derivative and photodynamic therapy on atherosclerotic rabbits. Am J Cardiol. 1985;56 (10 ):667–671. doi:10.1016/0002-9149(85)91032-x 2931972 71. Usui M, Asahara T, Naitoh Y, Katoh T, Ibukiyama C. Photodynamic therapy for the prevention of intimal hyperplasia in balloon-injured rabbit arteries. Jpn Circ J. 1999;63 (5 ):387–393. doi:10.1253/jcj.63.387 10943619 72. Gonschior P, Erdemci A, Gerheuser F, et al. [Selective hematoporphyrin derivative (HMD) application in arterial vessels using a porous balloon catheter results in equivalent levels as compared to high-dose systemic administration]. Z Kardiol. 1991;80 (12 ):738–745. German.1837968 73. Neave V, Giannotta S, Hyman S, Schneider J. Hematoporphyrin uptake in atherosclerotic plaques: therapeutic potentials. Neurosurgery. 1988;23 (3 ):307–312. doi:10.1227/00006123-198809000-00004 2976125 74. Dartsch PC, Ischinger T, Betz E. Responses of cultured smooth muscle cells from human nonatherosclerotic arteries and primary stenosing lesions after photoradiation: implications for photodynamic therapy of vascular stenoses. J Am Coll Cardiol. 1990;15 (7 ):1545–1550. doi:10.1016/0735-1097(90)92824-l 2345234 75. Dartsch PC, Coppenrath E, Coppenrath K, Ischinger T. Photodynamic therapy of vascular stenosis: results from cell culture studies on human endothelial cells. Coron Artery Dis. 1993;4 (2 ):207–213. doi:10.1097/00019501-199302000-00012 8269213 76. Tang G, Hyman S, Schneider JH Jr, Giannotta SL. Application of photodynamic therapy to the treatment of atherosclerotic plaques. Neurosurgery. 1993;32 (3 ):438–443. doi:10.1227/00006123-199303000-00016 8455769 77. Hsiang YN, Crespo MT, Machan LS, Bower RD, Todd ME. Photodynamic therapy for atherosclerotic stenoses in Yucatan miniswine. Can J Surg. 1994;37 (2 ):148–152.8156469 78. Vincent GM, Mackie RW, Orme E, Fox J, Johnson M. In vivo photosensitizer-enhanced laser angioplasty in atherosclerotic Yucatan miniswine. J Clin Laser Med Surg. 1990;8 (3 ):59–61. doi:10.1089/clm.1990.8.59 10148962 79. Amemiya T, Nakajima H, Katoh T, Rakue H, Miyagi M, Ibukiyama C. Photodynamic therapy of atherosclerosis using YAG-OPO laser and Porfimer sodium, and comparison with using argon-dye laser. Jpn Circ J. 1999;63 (4 ):288–295. doi:10.1253/jcj.63.288 10475777 80. Hsiang YN, Crespo MT, Richter AM, Jain AK, Fragoso M, Levy JG. In vitro and in vivo uptake of benzoporphyrin derivative into human and miniswine atherosclerotic plaque. Photochem Photobiol. 1993;57 (4 ):670–674. doi:10.1111/j.1751-1097.1993.tb02935.x 8506397 81. Allison BA, Crespo MT, Jain AK, Richter AM, Hsiang YN, Levy JG. Delivery of benzoporphyrin derivative, a photosensitizer, into atherosclerotic plaque of Watanabe heritable hyperlipidemic rabbits and balloon-injured New Zealand rabbits. Photochem Photobiol. 1997;65 (5 ):877–883. doi:10.1111/j.1751-1097.1997.tb01938.x 9155261 82. Jain M, Zellweger M, Frobert A, et al. Intra-arterial drug and light delivery for photodynamic therapy using visudyne®: implication for atherosclerotic plaque treatment. Front Physiol. 2016;7 :400. doi:10.3389/fphys.2016.00400 27672369 83. Cheng J, Liang H, Li Q, et al. Hematoporphyrin monomethyl ether-mediated photodynamic effects on THP-1 cell-derived macrophages. J Photochem Photobiol B. 2010;101 (1 ):9–15. doi:10.1016/j.jphotobiol.2010.06.005 20630770 84. Pai M, Jamal W, Mosse A, Bishop C, Bown S, McEwan J. Inhibition of in-stent restenosis in rabbit iliac arteries with photodynamic therapy. Eur J Vasc Endovasc Surg. 2005;30 (6 ):573–581. doi:10.1016/j.ejvs.2005.07.003 16125418 85. Gabeler EEE, van Hillegersberg R, Statius van Eps RG, Sluiter W, Mulder P, van Urk H. Endovascular photodynamic therapy with aminolaevulinic acid prevents balloon induced intimal hyperplasia and constrictive remodelling. Eur J Vasc Endovasc Surg. 2002;24 (4 ):322–331. doi:10.1053/ejvs.2002.1723 12323175 86. Jenkins MP, Buonaccorsi G, MacRobert A, Bishop CC, Bown SG, McEwan JR. Intra-arterial photodynamic therapy using 5-ALA in a swine model. Eur J Vasc Endovasc Surg. 1998;16 (4 ):284–291. doi:10.1016/S1078-5884(98)80047-6 9818005 87. Jenkins MP, Buonaccorsi GA, Mansfield R, Bishop CC, Bown SG, McEwan JR. Reduction in the response to coronary and iliac artery injury with photodynamic therapy using 5-aminolaevulinic acid. Cardiovasc Res. 2000;45 (2 ):478–485. doi:10.1016/S0008-6363(99)00352-1 10728369 88. Jenkins GAB MP, Raphael M, Nyamekye I, McEwan JR, Bown SG, Bishop CCR. Clinical study of adjuvant photodynamic therapy to reduce restenosis following femoral angioplasty. Br J Surg. 1999;86 (10 ):1258–1263. doi:10.1046/j.1365-2168.1999.01247.x 10540128 89. Herman MA, Webber J, Fromm D, Kessel D. Hemodynamic effects of 5-aminolevulinic acid in humans. Biology. 1998;43 (1 ):61–65. doi:10.1016/S1011-1344(98)00086-4 9639916 90. Eldar M, Yerushalmi Y, Kessler E, et al. Preferential uptake of a water-soluble phthalocyanine by atherosclerotic plaques in rabbits. Atherosclerosis. 1990;84 (2–3 ):135–139. doi:10.1016/0021-9150(90)90083-u 2282094 91. Dartsch PC, Wunderlich K, Ben-Hur E. Aluminium phthalocyanines-induced photolysis of human vascular wall cells in culture and the effect of fluoride on photodynamic action. Coron Artery Dis. 1994;5 (10 ):851–855.7866605 92. de Vries HE, Moor AC, Dubbelman TM, van Berkel TJ, Kuiper J. Oxidized low-density lipoprotein as a delivery system for photosensitizers: implications for photodynamic therapy of atherosclerosis. J Pharmacol Exp Ther. 1999;289 (1 ):528–534.10087046 93. Lin JS, Wang CJ, Li WT. Photodynamic therapy of balloon-injured rat carotid arteries using indocyanine green. Lasers Med Sci. 2018;33 (5 ):1123–1130. doi:10.1007/s10103-018-2488-7 29594740 94. Yasunaka Y, Aizawa K, Asahara T, Kato H, Hayata Y, Ishikura IA. In vivo accumulation of photosensitizers in atherosclerotic lesions and blood in atherosclerotic rabbit. Lasers Life Sci. 1991;1991 :53–65. 95. Hayashi J, Saito T, Aizawa K. Photodynamic diagnosis and treatment for atherosclerosis by an endoscopic approach. Diagn Ther Endosc. 1999;5 (3 ):191–195. doi:10.1155/DTE.5.191 18493502 96. Hamblin MR, Miller JL, Ortel B. Scavenger-receptor targeted photodynamic therapy. Photochem Photobiol. 2000;72 (4 ):533–540. doi:10.1562/0031-8655(2000)072<0533:SRTPT>2.0.CO;2 11045726 97. Shon S-M, Choi Y, Kim J-Y, et al. Photodynamic therapy using a protease-mediated theranostic agent reduces cathepsin-B activity in mouse atheromata in vivo. Arterioscler Thromb Vasc Biol. 2013;33 (6 ):1360–1365. doi:10.1161/ATVBAHA.113.301290 23539220 98. Lee DK, Choi Y, Shon SM, Schellingerhout D, Park JE, Kim DE. Atorvastatin and clopidogrel interfere with photosensitization in vitro. Photochem Photobiol Sci. 2011;10 (10 ):1587–1592. doi:10.1039/c0pp00363h 21748161 99. Weiss-Sadan T, Ben-Nun Y, Maimoun D, et al. A theranostic cathepsin activity-based probe for noninvasive intervention in cardiovascular diseases. Theranostics. 2019;9 (20 ):5731–5738. doi:10.7150/thno.34402 31534515 100. Chen Z, Woodburn KW, Shi C, Adelman DC, Rogers C, Simon DI. Photodynamic therapy with motexafin lutetium induces redox-sensitive apoptosis of vascular cells. Arterioscler Thromb Vasc Biol. 2001;21 (5 ):759–764. doi:10.1161/01.atv.21.5.759 11348871 101. Yamaguchi A, Woodburn KW, Hayase M, Robbins RC. Reduction of vein graft disease using photodynamic therapy with motexafin lutetium in a rodent isograft model. Circulation. 2000;102 (19 Suppl 3 ):III275–III280. doi:10.1161/01.cir.102.suppl_3.iii-275 11082401 102. Yamaguchi A, Woodburn KW, Hayase M, Hoyt G, Robbins RC. Photodynamic therapy with motexafin lutetium (Lu-Tex) reduces experimental graft coronary artery disease. Transplantation. 2001;71 (11 ):1526–1532. doi:10.1097/00007890-200106150-00008 11435960 103. Saito T, Hayashi J, Sato H, Kawabe H, Aizawa K. Scanning electron microscopic analysis of acute photodynamic therapy for atherosclerotic plaques of rabbit aorta by using a pheophorbide derivative. J Clin Laser Med Surg. 1996;14 (1 ):1–6. doi:10.1089/clm.1996.14.1 9484092 104. Huang L, Chen Q, Yu L, Bai D. Pyropheophorbide-alpha methyl ester-mediated photodynamic therapy induces apoptosis and inhibits LPS-induced inflammation in RAW264.7 macrophages. Photodiagnosis Photodyn Ther. 2019;25 :148–156. doi:10.1016/j.pdpdt.2018.12.002 30562579 105. Jain M, Bouilloux J, Borrego I, et al. Cathepsin B-cleavable polymeric photosensitizer prodrug for selective photodynamic therapy: in vitro studies. Pharmaceuticals. 2022;15 (5 ):564. doi:10.3390/ph15050564 35631388 106. Wang G, Zhu Y, Li K, et al. Curcumin-mediated photodynamic therapy inhibits the phenotypic transformation, migration, and foaming of oxidized low-density lipoprotein-treated vascular smooth muscle cells by promoting autophagy. J Cardiovasc Pharmacol. 2021;78 (2 ):308–318. doi:10.1097/FJC.0000000000001069 34091481 107. Spyropoulos-Antonakakis N, Sarantopoulou E, Trohopoulos PN, et al. Selective aggregation of PAMAM dendrimer nanocarriers and PAMAM/ZnPc nanodrugs on human atheromatous carotid tissues: a photodynamic therapy for atherosclerosis. Nanoscale Res Lett. 2015;10 :210.25991914 108. Rodriguez L, Vallecorsa P, Battah S, et al. Aminolevulinic acid dendrimers in photodynamic treatment of cancer and atheromatous disease. Photochem Photobiol Sci. 2015;14 (9 ):1617–1627. doi:10.1039/c5pp00126a 26066768 109. Cespedes MA, Saenz DA, Calvo GH, et al. Apoptotic cell death induced by dendritic derivatives of aminolevulinic acid in endothelial and foam cells co-cultures. Photochem Photobiol Sci. 2021;20 (4 ):489–499. doi:10.1007/s43630-021-00025-x 33825180 110. Wennink JWH, Liu Y, Mäkinen PI, et al. Macrophage selective photodynamic therapy by meta-tetra(hydroxyphenyl)chlorin loaded polymeric micelles: a possible treatment for cardiovascular diseases. Eur J Pharm Sci. 2017;107 :112–125. doi:10.1016/j.ejps.2017.06.038 28679107 111. Kim H, Kim Y, Kim IH, Kim K, Choi Y. ROS-responsive activatable photosensitizing agent for imaging and photodynamic therapy of activated macrophages. Theranostics. 2013;4 (1 ):1–11. doi:10.7150/thno.7101 24396511 112. Song JW, Ahn JW, Lee MW, et al. Targeted theranostic photoactivation on atherosclerosis. J Nanobiotechnology. 2021;19 (1 ):338. doi:10.1186/s12951-021-01084-z 34689768 113. Ahn JW, Kim JH, Park K. In vitro photodynamic effects of the inclusion nanocomplexes of glucan and chlorin e6 on atherogenic foam cells. Int J Mol Sci. 2020;22 (1 ):177. doi:10.3390/ijms22010177 33375356 114. Han XB, Li HX, Jiang YQ, et al. Upconversion nanoparticle-mediated photodynamic therapy induces autophagy and cholesterol efflux of macrophage-derived foam cells via ROS generation. Cell Death Dis. 2017;8 (6 ):e2864. doi:10.1038/cddis.2017.242 28594401 115. Xu M, Mao C, Chen H, et al. Osteopontin targeted theranostic nanoprobes for laser-induced synergistic regression of vulnerable atherosclerotic plaques. Acta Pharm Sin B. 2022;12 (4 ):2014–2028. doi:10.1016/j.apsb.2021.12.020 35847489 116. Zhu X, Wang H, Zheng L, et al. Upconversion nanoparticle-mediated photodynamic therapy induces THP-1 macrophage apoptosis via ROS bursts and activation of the mitochondrial caspase pathway. Int J Nanomedicine. 2015;10 :3719–3736. doi:10.2147/IJN.S82162 26045663 117. Han X, Kou J, Zheng Y, et al. ROS generated by upconversion nanoparticle-mediated photodynamic therapy induces autophagy via PI3K/AKT/ mTOR signaling pathway in M1 peritoneal macrophage. Cell Physiol Biochem. 2019;52 (6 ):1325–1338.31050281 118. Ma Y, Ma Y, Gao M, et al. Platelet-Mimicking therapeutic system for noninvasive mitigation of the progression of atherosclerotic plaques. Adv Sci. 2021;8 (8 ):2004128. doi:10.1002/advs.202004128 119. Mu D, Wang X, Wang H, et al. Chemiexcited photodynamic therapy integrated in polymeric nanoparticles capable of MRI against atherosclerosis. Int J Nanomedicine. 2022;17 :2353–2366. doi:10.2147/IJN.S355790 35645560 120. Usui M, Miyagi M, Fukasawa S, et al. A first trial in the clinical application of photodynamic therapy for the prevention of restenosis after coronary-stent placement. Lasers Surg Med. 2004;34 (3 ):235–241. doi:10.1002/lsm.20018 15022250 121. Rockson SG, Kramer P, Razavi M, et al. Photoangioplasty for human peripheral atherosclerosis: results of a Phase I trial of photodynamic therapy with motexafin lutetium (Antrin). Circulation. 2000;102 (19 ):2322–2324. doi:10.1161/01.cir.102.19.2322 11067782 122. Kereiakes DJ, Szyniszewski AM, Wahr D, et al. Phase I drug and light dose-escalation trial of motexafin lutetium and far red light activation (phototherapy) in subjects with coronary artery disease undergoing percutaneous coronary intervention and stent deployment: procedural and long-term results. Circulation. 2003;108 (11 ):1310–1315. doi:10.1161/01.CIR.0000087602.91755.19 12939212 123. Choi V, Rajora MA, Zheng G. Activating drugs with sound: mechanisms behind sonodynamic therapy and the role of nanomedicine. Bioconjug Chem. 2020;31 (4 ):967–989. doi:10.1021/acs.bioconjchem.0c00029 32129984 124. Yumita N, Nishigaki R, Umemura K, Umemura S-I. Hematoporphyrin as a sensitizer of cell-damaging effect of ultrasound. Jpn J Cancer Chemother. 1989;80 (3 ):219–222. doi:10.1111/j.1349-7006.1989.tb02295.x 125. Arakawa K, Hagisawa K, Kusano H, et al. Sonodynamic therapy decreased neointimal hyperplasia after stenting in the rabbit iliac artery. Circulation. 2002;105 (2 ):149–151. doi:10.1161/hc0202.102921 11790692 126. Cao Z, Zhang T, Sun X, et al. Membrane-permeabilized sonodynamic therapy enhances drug delivery into macrophages. PLoS One. 2019;14 (6 ):e0217511. doi:10.1371/journal.pone.0217511 31181129 127. Cheng J, Sun X, Guo S, et al. Effects of 5-aminolevulinic acid-mediated sonodynamic therapy on macrophages. Int J Nanomedicine. 2013;8 :669–676. doi:10.2147/IJN.S39844 23426386 128. Chen H, Gao W, Yang Y, et al. Inhibition of VDAC1 prevents Ca²+-mediated oxidative stress and apoptosis induced by 5-aminolevulinic acid mediated sonodynamic therapy in THP-1 macrophages. Apoptosis. 2014;19 (12 ):1712–1726. doi:10.1007/s10495-014-1045-5 25342393 129. Wang H, Yang Y, Chen H, et al. The predominant pathway of apoptosis in THP-1 macrophage-derived foam cells induced by 5-aminolevulinic acid-mediated sonodynamic therapy is the mitochondria-caspase pathway despite the participation of endoplasmic reticulum stress. Cell Physiol Biochem. 2014;33 (6 ):1789–1801. doi:10.1159/000362958 24923653 130. Sun X, Guo S, Wang W, et al. Potential involvement of the 18 kDa translocator protein and reactive oxygen species in apoptosis of THP-1 macrophages induced by sonodynamic therapy. PLoS One. 2018;13 (5 ):e0196541. doi:10.1371/journal.pone.0196541 29746502 131. Wang H, Yang Y, Sun X, et al. Sonodynamic therapy-induced foam cells apoptosis activates the phagocytic PPARγ-LXRα-ABCA1/ABCG1 pathway and promotes cholesterol efflux in advanced plaque. Theranostics. 2018;8 (18 ):4969–4984. doi:10.7150/thno.26193 30429880 132. Yang Y, Wang J, Guo S, et al. Non-lethal sonodynamic therapy facilitates the M1-to-M2 transition in advanced atherosclerotic plaques via activating the ROS-AMPK-mTORC1-autophagy pathway. Redox Biol. 2020;32 :101501. doi:10.1016/j.redox.2020.101501 32179242 133. Wang Y, Wang W, Xu H, et al. Non-lethal sonodynamic therapy inhibits atherosclerotic plaque progression in ApoE-/- mice and attenuates ox-LDL-mediated macrophage impairment by inducing heme oxygenase-1. Cell Physiol Biochem. 2017;41 (6 ):2432–2446. doi:10.1159/000475913 28468003 134. Li Z, Sun X, Guo S, et al. Rapid stabilisation of atherosclerotic plaque with 5-aminolevulinic acid-mediated sonodynamic therapy. Thromb Haemost. 2015;114 (4 ):793–803. doi:10.1160/TH14-12-1030 26179778 135. Sun X, Guo S, Yao J, et al. Rapid inhibition of atherosclerotic plaque progression by sonodynamic therapy. Cardiovasc Res. 2019;115 (1 ):190–203. doi:10.1093/cvr/cvy139 29878150 136. Li B, Gong J, Sheng S, et al. Sonodynamic therapy reduces iron retention of hemorrhagic plaque. Bioeng Transl Med. 2021;6 (1 ):e10193. doi:10.1002/btm2.10193 33532592 137. Cao Y, Yao J, Gao W, et al. Sonodynamic therapy promotes efferocytosis via CD47 down-regulation in advanced atherosclerotic plaque. Int Heart J. 2022;63 (1 ):131–140. doi:10.1536/ihj.21-233 35034916 138. Yao J, Gao W, Wang Y, et al. Sonodynamic therapy suppresses neovascularization in atherosclerotic plaques via macrophage apoptosis-induced endothelial cell apoptosis. JACC. 2020;5 (1 ):53–65. doi:10.1016/j.jacbts.2019.10.007 32043020 139. Gao Q, Wang F, Guo S, et al. Sonodynamic effect of an anti-inflammatory agent--emodin on macrophages. Ultrasound Med Biol. 2011;37 (9 ):1478–1485. doi:10.1016/j.ultrasmedbio.2011.05.846 21767904 140. Wang F, Gao Q, Guo S, et al. The sonodynamic effect of curcumin on THP-1 cell-derived macrophages. Biomed Res Int. 2013;2013 :737264. doi:10.1155/2013/737264 23509769 141. Zheng L, Sun X, Zhu X, et al. Apoptosis of THP-1 derived macrophages induced by sonodynamic therapy using a new sonosensitizer hydroxyl acetylated curcumin. PLoS One. 2014;9 (3 ):e93133. doi:10.1371/journal.pone.0093133 24676101 142. Zheng L, Li Y, Li X, et al. Combination of Hydroxyl Acetylated Curcumin and Ultrasound Induces Macrophage Autophagy with Anti-Apoptotic and Anti-Lipid Aggregation Effects. Cell Physiol Biochem. 2016;39 (5 ):1746–1760. doi:10.1159/000447875 27744450 143. Jiang Y, Kou J, Han X, et al. ROS-Dependent Activation of Autophagy through the PI3K/Akt/mTOR Pathway Is Induced by Hydroxysafflor Yellow A-Sonodynamic Therapy in THP-1 Macrophages. Oxid Med Cell Longev. 2017;2017 :8519169. doi:10.1155/2017/8519169 28191279 144. Kou JY, Li Y, Zhong ZY, et al. Berberine-sonodynamic therapy induces autophagy and lipid unloading in macrophage. Cell Death Dis. 2017;8 (1 ):e2558. doi:10.1038/cddis.2016.354 28102849 145. Li X, Gao L, Zheng L, et al. The efficacy and mechanism of apoptosis induction by hypericin-mediated sonodynamic therapy in THP-1 macrophages. Int J Nanomedicine. 2015;10 :821–838. doi:10.2147/IJN.S75398 25653524 146. Zheng X, Wu J, Shao Q, et al. Apoptosis of THP-1 macrophages induced by pseudohypericin-mediated sonodynamic therapy through the mitochondria-caspase pathway. Cell Physiol Biochem. 2016;38 (2 ):545–557. doi:10.1159/000438649 26828938 147. Li X, Zhang X, Zheng L, et al. Hypericin-mediated sonodynamic therapy induces autophagy and decreases lipids in THP-1 macrophage by promoting ROS-dependent nuclear translocation of TFEB. Cell Death Dis. 2016;7 (12 ):e2527. doi:10.1038/cddis.2016.433 28005078 148. Goncalves KD, Vieira DP, Courrol LC. Synthesis and characterization of aminolevulinic acid gold nanoparticles: photo and sonosensitizer agent for atherosclerosis. J Lumines. 2018;197 :317–323. doi:10.1016/j.jlumin.2018.01.057 149. Gonçalvez KDO, Vieira DP, Courrol LC. Study of THP-1 macrophage viability after sonodynamic therapy using methyl ester of 5-aminolevulinic acid gold nanoparticles. Ultrasound Med Biol. 2018;44 (9 ):2009–2017. doi:10.1016/j.ultrasmedbio.2018.05.012 29936026 150. Jiang L, Wang J, Jiang J, et al. Sonodynamic therapy in atherosclerosis by curcumin nanosuspensions: preparation design, efficacy evaluation, and mechanisms analysis. Eur J Pharm Biopharm. 2020;146 :101–110. doi:10.1016/j.ejpb.2019.12.005 31841689 151. Yao J, Yang Z, Huang L, et al. Low-Intensity focused ultrasound-responsive ferrite-encapsulated nanoparticles for atherosclerotic plaque neovascularization theranostics. Adv Sci. 2021;8 (19 ):e2100850. doi:10.1002/advs.202100850 152. Jiang Y, Fan J, Li Y, et al. Rapid reduction in plaque inflammation by sonodynamic therapy inpatients with symptomatic femoropopliteal peripheral artery disease:A randomized controlled trial. Int J Cardiol. 2021;325 :132–139. doi:10.1016/j.ijcard.2020.09.035 32966832 153. Cao Y. Nutrient molecule Corona: an update for nanomaterial-food component interactions. Toxicology. 2022;476 :153253. doi:10.1016/j.tox.2022.153253 35811011
PMC010xxxxxx/PMC10352160.txt
==== Front Dig Dis Sci Dig Dis Sci Digestive Diseases and Sciences 0163-2116 1573-2568 Springer US New York 36401140 7727 10.1007/s10620-022-07727-x Original Article High Prevalence of Functional Gastrointestinal Disorders in Celiac Patients with Persistent Symptoms on a Gluten-Free Diet: A 20-Year Follow-Up Study http://orcid.org/0000-0002-8493-7698 Schiepatti Annalisa [email protected] 12 Maimaris Stiliano 12 Lusetti Francesca 1 Scalvini Davide 1 Minerba Paolo 1 Cincotta Marta 1 Fazzino Erica 1 Biagi Federico [email protected] 12 1 grid.8982.b 0000 0004 1762 5736 Dipartimento di Medicina Interna e Terapia Medica, University of Pavia, 27100 Pavia, Italy 2 grid.511455.1 Istituti Clinici Scientifici Maugeri, IRCCS, Gastroenterology Unit of Pavia Institute, Via Salvatore Maugeri 10, 27100 Pavia, Italy 18 11 2022 18 11 2022 2023 68 8 33743382 1 4 2022 20 7 2022 © The Author(s) 2022, corrected publication 2023 https://creativecommons.org/licenses/by-nc/4.0/ Open AccessThis article is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License, which permits any non-commercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc/4.0/. Background Ongoing symptoms in treated celiac disease (CD) are frequent and are commonly thought of as being due to infractions to a gluten-free diet (GFD) or complications. Aims To study the etiology and natural history of clinically relevant events (CREs) throughout follow-up and identify predictors thereof to guide follow-up. Methods CREs (symptoms/signs requiring diagnostic/therapeutic interventions) occurring in celiac patients between January-2000 and May-2021 were retrospectively collected between June and September 2021 and analysed. Results One-hundred-and-eighty-nine adult patients (133 F, age at diagnosis 36 ± 13 years, median follow-up 103 months, IQR 54–156) were enrolled. CREs were very common (88/189, 47%), but hardly due to poor GFD adherence (4%) or complications (2%). Interestingly, leading etiologies were functional gastrointestinal disorders (30%), reflux disease (18%) and micronutrient deficiencies (10%). Age at diagnosis ≥ 45 years (HR 1.68, 95%CI 1.05–2.69, p = 0.03) and classical pattern of CD (HR 1.63, 95%CI 1.04–2.54, p = 0.03) were predictors of CREs on a multivariable Cox model. At 5 years, 46% of classical patients ≥ 45 years old at diagnosis were event-free, while this was 62% for non-classical/silent ≥ 45 years, 60% for classical < 45 years, and 80% for non-classical/silent < 45 years. Conclusions CREs occurred in almost half of CD patients during follow-up, with functional disorders being very common. New follow-up strategies for adult CD may be developed based on age and clinical pattern at diagnosis. Keywords Gluten-free diet Celiac disease Follow-up Persistent symptoms Università degli Studi di PaviaOpen access funding provided by Università degli Studi di Pavia within the CRUI-CARE Agreement. issue-copyright-statement© Springer Science+Business Media, LLC, part of Springer Nature 2023 ==== Body pmcIntroduction Celiac disease (CD) is a chronic immune-mediated enteropathy triggered by dietary gluten in genetically predisposed individuals [1–3] with a prevalence of around 1% in the general population [1–4]. The clinical presentation of CD is very heterogeneous, ranging from an overt malabsorption syndrome, to mild intestinal/extra-intestinal symptoms, or to asymptomatic patients [1–3]. Positive endomysial (EmA)/tissue transglutaminase (tTA) antibodies and villous atrophy (VA) on correctly oriented duodenal biopsies are still recommended for the diagnosis of CD in adults in most countries [1–3], although great interest has been dedicated to the possibility of a biopsy-sparing approach also in adults [5, 6]. A strict lifelong gluten-free diet (GFD) is the cornerstone for treatment of CD, leading to complete resolution of symptoms/histological lesions in the vast majority of patients, and preventing long-term morbidity and mortality associated to CD [1–4, 7–11]. However, it has been reported that in up to 30% of patients symptoms and/or histological lesions may not resolve completely despite being on a GFD [12–18]. These clinical scenarios have a wide spectrum of etiologies, which may be related or unrelated to CD. Ongoing gluten ingestion (voluntary or inadvertent) has been reported to be the prevalent cause of persistent symptoms in CD. However, other etiologies may also be responsible, ranging from purely functional gastrointestinal disorders, to life-threatening malignant complications of CD [12–18]. Despite their wide clinical variability, these scenarios are very often grouped together under the umbrella term non-responsive CD (NRCD). The literature provides data on the underlying etiology of NRCD, with particular interest on refractory CD and malignant complications of CD, for which risk factors and natural history have also been delineated [11, 12, 19–21]. Conversely, very little is known on the natural history of symptoms and disorders unrelated to complicated CD occurring in celiac patients on a long-term GFD and how this may influence modalities for organizing the follow-up of these patients. Although major international guidelines suggest maintaining regular follow-up consultations for all adult celiac patients, there is no widespread consensus on timing and modalities for organizing it cost-effectively [1–3]. Therefore, the present study aims to: (1) evaluate occurrence and etiologies of symptoms leading to clinically relevant events (CREs), both related and unrelated to CD, in adult celiac patients on long-term follow-up; (2) identify any relevant predictors of CREs during follow-up; 3) suggest possible follow-up strategies in adult celiac patients. Patients and Methods Study Population and Setting This is a single-centre retrospective study of adult patients (≥ 18 years old) with biopsy-proven CD on a long-term GFD and follow-up, which aims to describe the natural history and identify the etiologies and predictors of persistent/recurrent symptoms despite a GFD. Patients enrolled in this study all underwent follow-up duodenal biopsy, as part of our standard of care in the last 20 years [22]. Enrolment and Exclusion Criteria Patients who were directly diagnosed with CD at our center between January-2000 and November-2019, and followed-up in clinic until May-2021 were the focus of the present study. All patients underwent follow-up duodenal biopsy and regular dietary assessment of GFD adherence, either via interview by expert personnel or using a validated questionnaire we previously developed [23]. In the last 20 years our center has provided care to over 800 patients, as previously described [24]. To avoid biases, for the purpose of the present study, we included only patients directly diagnosed by our center who had at least one follow-up biopsy. In other words we excluded all referred patients and any patients who did not undergo follow-up duodenal biopsy for various reasons. CD was diagnosed based on a certain degree of VA on duodenal biopsies from the second part of the duodenum and positive EmA and/or tTA while on a gluten-containing diet [1–3]. Data Collection Patients’ medical records were retrospectively reviewed between June-2021 and September-2021, and the following data were collected: age at diagnosis of CD, presenting symptoms of CD according to Oslo classification [25], first-degree family history of CD, results and time of follow-up duodenal biopsies and EmA testing, number of follow-up medical consultations, duration of follow-up, GFD adherence throughout follow-up, and occurrence of clinical events, both related and unrelated to CD during follow-up. Evaluation of Symptoms and Clinical Events During Follow-Up Symptoms/signs (persistent, recurrent or new symptoms/signs) at any time during follow-up requiring any diagnostic testing, treatment, emergency room access, or hospitalization were considered CREs for the purpose of our study. Their etiologies and outcomes were also evaluated. Persistent symptoms were defined as those already present at diagnosis of CD, which did not improve significantly or resolve during follow-up despite a GFD [22]; recurrent symptoms were those that relapsed despite initial resolution/improvement on a GFD; symptoms which developed for the first time while on a GFD, were considered ‘new onset symptoms’. Data on symptoms was collected regardless of whether a relationship with CD was suspected or not. Diagnostic tests performed for scheduled follow-up of chronic conditions in the absence of symptoms, including follow-up of persistent VA, or in the context of a screening program were not considered as clinical events. This strategy was adopted in order to provide the most objective evaluation of CREs throughout follow-up in a retrospective study. Clinical characteristics of patients with at least one event were compared to those of patients without events to identify predictors of developing events. Patients who had at least one clinically relevant event during follow-up as defined above were considered patients who developed events while those without any events were considered event-free. All events observed throughout follow-up were analysed to determine their aetiology and to estimate overall incidence of events. Criteria to Assess GFD Adherence All patients were instructed immediately after diagnosis of CD by expert personnel on how to follow a strict GFD correctly. Throughout follow-up, GFD adherence was assessed, until 2008, by dietary interview by expert personnel, after 2008 using a five-point validated score we previously developed [23, 24]. Patients scoring 0–2 were considered poorly adherent, while those scoring 3–4 were considered adherent. Statistical Analysis Statistical analysis was performed using R version 4.1.2 (R Core Team (2020). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/). Categorical variables were summarized as total counts and percentages. Univariate analysis of categorical variables was performed with Fisher’s exact test. The Cochran-Armitage test was used to evaluate trends in occurrence of events according to patient age groups. Continuous variables were summarized as mean and standard deviation or median and interquartile range. Testing for normality of data was performed using the Shapiro–Wilk test. For univariate analysis of continuous variables among groups the Mann–Whitney U test was used. Spearman rank correlation was used to evaluate for correlation among variables. Incidence rate of events and exact binomial 95% confidence intervals (95% CI) were calculated. Two-sided p-values < 0.05 were considered statistically significant. A multivariable Cox model was developed to identify predictors of event-free follow-up and to estimate hazard ratios and 95% CI. Harrel’s c statistic and 95% CI were calculated to evaluate model discriminatory ability. Variables for multivariable Cox model analysis were selected based on results of univariate analysis and on the basis of clinical relevance. Kaplan–Meier curves for event-free follow-up were generated and the logrank test was used to compare event-free follow-up rates among groups. Mortality among patients with and without events during follow-up was compared using the logrank test. Results One-hundred-and-eighty-nine celiac patients (133 F, mean age at diagnosis 36 ± 13 years) who underwent routine follow-up duodenal biopsy and were followed-up for a median of 112 months (IQR 73–164) were enrolled. At follow-up duodenal biopsy (after a median of 16 months since diagnosis, IQR 13–20 months), 29 patients (15.4%) still showed a certain degree of VA. Among these 29 patients, 2 (6.9%) had complicated CD, 9 (31.0%) had poor GFD adherence, while the remaining 18 (62.0%) showed initial but incomplete histological improvement in the context of good GFD adherence. GFD adherence at time of follow-up biopsy was good in the vast majority of patients (160/176, 90.9%, 13 data missing). Among patients without complications who underwent a further follow-up duodenal biopsy, histological recovery occurred in 75% of the patients while 25% still had atrophy due to poor GFD adherence. At time of follow-up duodenal biopsy 95 patients overall (50.3%) had ongoing symptoms, of which 64/95 (37.0%) had persistence of symptoms present at diagnosis despite a GFD and 42/95 (24.6%) developed new symptoms which were not present at diagnosis. Only 11 of the 29 patients with VA at follow-up biopsy (37.9%) had persistent symptoms at time of follow-up biopsy. Demographic and Clinical Characteristics of Patients With and Without Clinical Events Eighty-eight out of 189 patients (47%) had at least one clinically significant event (symptoms requiring additional diagnostic testing, medical treatment, emergency room access or hospitalization) during follow-up, while the remaining 101 patients (53%) had no events. As shown in Table 1, patients who had at least one clinical event during follow-up were older at diagnosis (p = 0.02) and more likely to have a classical pattern of CD at presentation (p = 0.01). They were also more likely to have persistent VA at follow-up duodenal biopsy (p = 0.03). As shown in Fig. 1, there was a trend for increasing prevalence of events with increasing age at diagnosis (p < 0.01). No relationship was found between GFD adherence at time of follow-up duodenal biopsy and occurrence of events during follow-up (p = 0.44). In total, 5 patients died over follow-up, after a median of 146 months from diagnosis of CD (IQR 122–161). All patients that died had at least one clinical event during follow-up, while none of the patients who had no clinical events died during follow-up (mortality 5.7% vs 0%, p = 0.03). As would be expected, patients with clinical events had a significantly higher number of follow-up medical consultations in our Unit (median 4 vs 2, p < 0.01) and a slightly, although not statistically significant, longer duration of clinical follow-up than those with no events (median 125 vs 100 months, p = 0.06). Likewise, total number of events correlated with number of follow-up medical consultations (Spearman rank correlation, rho = 0.49, p < 0.001).Table 1 Demographic and clinical characteristics of patients with and without clinical events Pts without clinical events (101) Pts with clinical events (88) p-value Odds ratio (95% CI) At diagnosis  Age at diagnosis median (IQR) 33 years (27–39) 36 years (30–47) 0.02 –  Gender (female) 69 (68.3%) 64 (72.7%) 0.53 –  Classical pattern of CD [25] 42 (42.0%) 53 (60.9%) 0.01 2.14 (1.17–3.91)  Anemia 43 (43.0%) 49 (44.8%) 0.88 –  Associated autoimmune conditions 14 (14.0%) 13 (14.9%) 1.00 –  Dermatitis herpetiformis 11 (11.0%) 10 (11.5%) 1.00 –  Dyspepsia 19 (19.0%) 13 (14.9%) 0.56 –  1st-degree family history of CD 21 (21.0%) 14 (16.1%) 045 –  GERD 14 (14.0%) 17 (19.5%) 0.33 – At follow-up  Persistence of VA at duodenal biopsy 10 (9.9%) 19 (21.8%) 0.03 2.53 (1.09–5.88)  Good GFD adherence 88 (92.6%) 72 (88.9%) 0.44 –  Duration of follow-up median (IQR) 100 months (68–163) 125 months (84–167) 0.06 –  Consultations at our unit -median (IQR) 2 (1–3) 4 (3–8)  < 0.001 –  Development of complications of CD 0 (0.0%) 3 (3.4%) 0.10 –  Mortality 0 (0.0%) 5 (5.7%) 0.03 – CD celiac disease, GERD gastroesophageal reflux disease, GFD gluten-free diet, VA villous atrophy, pts patients Fig. 1 Prevalence of patients with and without clinical events according to age group at diagnosis of celiac disease. *Cochran-Armitage trend test, p < 0.01 Clinical Events Symptoms and Etiologies In total 157 clinically significant events occurred throughout follow-up in all patients, resulting in an incidence of events of 83.5/1000 person-years (95% CI 70.9–97.6). Median time to event from diagnosis of CD was 50 months (IQR 26–107). Events, as shown in Fig. 2, were most prevalent in the first 5 years from diagnosis of CD and decreased thereafter (overall event-free follow-up rates were 65% at 5 years and 51% at 10 years). In managing the 157 events, which occurred, a total of 92 outpatient medical treatments, 63 diagnostic investigations and 13 emergency room or hospital admissions were required. The most common symptoms leading to clinical events were gastroesophageal reflux disease symptoms (30 cases, 19%), diarrhea (24, 15%), dyspepsia (21, 13%), and constipation (15, 10%). Figure 3 shows the underlying etiologies of clinical events the most common being gastroesophageal reflux disease (29, 18%) and functional gastrointestinal disorders (47, 30%) such as functional constipation (16, 10%), irritable bowel syndrome (15, 10%) and functional dyspepsia (14, 9%). Poor adherence to a GFD accounted for 4% of total events and complications of CD for only 2%.Fig. 2 Kaplan–Meier curve showing distribution of events over time from diagnosis of celiac disease Fig. 3 Bar chart showing etiology of clinical events. GERD gastroesophageal reflux disease, GI gastrointestinal, GFD gluten-free diet CD coeliac disease, IBD inflammatory bowel disease, COPD chronic obstructive pulmonary disease. Functional GI disorders include: functional constipation, functional dyspepsia, irritable bowel syndrome, functional diarrhea. Two events which occurred in the context of poor GFD adherence (due to Helicobacter Pylori infection and dyslipidemia) were not included under poor GFD adherence as they were unrelated in nature to GFD adherence. Other etiologies included: colonic diverticular disease, colon polyps, dyslipidemia, dermatitis herpetiformis relapse, small intestinal bacterial overgrowth, COPD exacerbation, diabetic ketoacidosis in type 1 diabetes mellitus, acute gastroenteritis, ANCA-associated vasculitis, multifactorial anemia, hypercalcaemic syndrome, spondyloarthritis, osteoporosis, NSAID abuse, erosive gastritis, recurrent aphthous stomatitis Diagnostic Tests Of the 63 diagnostic tests performed, the most common ones were upper GI endoscopy with biopsies (28 cases, 44%), colonoscopy (13 cases, 21%) and abdominal ultrasonography (11 cases, 17%). The most common symptoms leading to diagnostic testing were diarrhoea (16 cases, 25%), dyspepsia (13 cases, 21%), abdominal pain (12 cases, 19%) and hematochezia (6 cases, 10%). Irritable bowel syndrome (12 cases, 19%), functional dyspepsia (10 cases, 16%), hemorrhoids (6 cases, 10%), gastroesophageal reflux disease/oesophagitis (5 cases, 8%) and lactose intolerance (4 cases, 6%) were the most common etiologies for symptoms requiring diagnostic investigations. Need for Medical Treatment Medical treatment was required in 92 events. The most common reasons for medical treatment were gastroesophageal reflux disease/oesophagitis (29 cases, 32%), micronutrient deficiencies (16 cases, 17%) including iron, folate and vitamin B12 deficiency with or without consequent anemia, functional constipation (15 cases, 16%), Helicobacter pylori infection (11 cases, 12%). Symptoms resolved after treatment in 75% of cases while they persisted in the remaining 25%, with functional constipation (43%), iron-deficiency anemia (21%), and H pylori infection (14%) being the most common to persist. Emergency Room Access/Hospital Admission Eight patients required emergency room/hospital admission during follow-up, with a total of 13 such events. Three patients were admitted to hospital due to persistence or relapse of severe malabsorption, all due to development of complications of CD (enteropathy-associated T-cell lymphoma, B-cell lymphoma, type 1 refractory CD). Two of these patients had primary unresponsiveness to a GFD (VA at follow-up biopsy and persistent malabsorption) while the last patient initially responded clinically and histologically to a GFD and after several years developed a complication [20]. Other reasons for emergency room access/hospital admission were not directly related to CD and included biliary colic, diabetic ketoacidosis and subsequent diagnosis of type 1 diabetes mellitus, acute gastroenteritis, ANCA-associated vasculitis, small intestinal bacterial overgrowth and exacerbation of chronic obstructive pulmonary disease (one patient each). Finally, one patient attended the emergency room and was subsequently hospitalized 4 times for reasons unrelated to CD (including hypercalcaemic syndrome, spondyloarthritis flare, multifactorial anemia). GFD adherence was good in all patients who attended the emergency room or were hospitalized. Predictors of Clinical Events Multivariable Cox analysis identified age at diagnosis ≥ 45 years (HR 1.68, 95%CI 1.05–2.69, p = 0.03) and classical pattern of CD at diagnosis (HR 1.63, 95% CI 1.04–2.54, p = 0.03) as independent predictors of developing clinical events during follow-up. However, persistence of VA at follow-up duodenal biopsy did not reach statistical significance (HR 1.55, 95%CI 0.91–2.63, p = 0.11). Harrell’s c was 0.60 (95%CI 0.54–0.66). Distribution of Clinical Events During Follow-Up According to Age and Clinical Pattern at Diagnosis of CD As shown in Fig. 4 occurrence of events during follow-up differs significantly (p < 0.01) according to age group (< 45 years, ≥ 45 years) and clinical pattern of CD at diagnosis (classical vs non-classical/silent). Overall, at 5 years follow-up, 46% of classical patients diagnosed at age ≥ 45 years were event-free, while event-free rates were 62% for non-classical/silent patients ≥ 45 years at diagnosis, 60% for classical patients < 45 years old at diagnosis, and 80% for non-classical/silent patients < 45 years old at diagnosis. At 10 years, event-free rates were 25%, 47%, 51% and 60% for these four groups, respectively.Fig. 4 Event-free follow-up rates according to age at diagnosis and clinical pattern of celiac disease at diagnosis (Oslo classification [25]). Group 1: Patients diagnosed at age ≥ 45 years with classical CD. Group 2: Patients diagnosed at age ≥ 45 years with non-classical/silent CD. Group 3: Patients diagnosed at age < 45 years with classical CD. Group 4: Patients diagnosed at age < 45 years with non-classical/silent CD Discussion This retrospective study, conducted on data collected over the last two decades in a referral centre, provides a real-world overview on the etiologies, natural history, and clinical predictors of persistent, recurrent or newly developing symptoms both related and unrelated to CD in adult celiac patients on a long-term GFD. Major findings include the identification of celiac patients at risk of having CREs over follow-up, potentially providing new clinical perspectives on follow-up modalities for adult CD. Firstly, we have shown that functional gastrointestinal disorders were the leading cause (30% of cases) of clinical events during follow-up, whereas etiologies strictly related to CD, which included inadequate adherence to a GFD and malignant complications of CD were rare, accounting, respectively, for only 4% and 2% of all events. Although functional gastrointestinal disorders are not life-threatening conditions, their impact on patients’ quality of life is burdensome [26, 27]. Their association with CD has been known for a long time [18, 28, 29]; however, mechanisms underlying functional symptoms in celiac patients on a long-term GFD are still unclear [12–18, 30–33]. Based on our results, the high prevalence of functional gastrointestinal disorders, gastroesophageal reflux disease and micronutrient deficiencies while on a GFD suggest a possible role for dietary quality rebalancing, as crucial intervention, in addition to maintaining a strict lifelong GFD adherence. Recently, it has been reported that gene expression in the small intestine of treated celiac patients differed from healthy controls, particularly for genes involved in the transport of micronutrients [34, 35], therefore this may have implications for nutritional supplementation, especially after diagnosis. Secondly, one third of patients had at least one clinically relevant event in the first five years after diagnosis. Just under 15% of patients without previous events subsequently developed events between five and ten years from diagnosis. Very few developed events after 10 years. This suggests that strict follow-up immediately after diagnosis is necessary, and then it can be organized on a case-by case basis. We have shown that age (> 45 years old) and classical pattern at diagnosis of CD were independent predictors of CREs. These parameters may be used to stratify patients into three subgroups according to their risk of developing CREs during follow-up, regardless of severity or etiological relationship with CD. These results complete our previous findings that age at diagnosis and clinical pattern of CD were risk factors for the development of malignant complications of CD [11]. Currently, international guidelines recommend regular follow-up in adult CD, but timing and modalities of follow-up are not standardised [1–3]. Implications for clinical practice of our results include the development of personalized and cost-effective modalities for the follow-up of adult celiac patients, including timing of follow-up, role of healthcare practitioners, and decentralization of care for those at low risk. We believe a possible cost-effective proposal may be as follows: (1) for patients diagnosed at age ≥ 45 years with classical symptoms maintenance of regular annual follow-up specialist medical consultation; (2) for patients diagnosed at age < 45 years with non-classical/silent presentation annual specialist medical consultation for the first two years since diagnosis and subsequent discharge to general practice or dietitians with referrals as necessary; (3) annual follow-up for the first five years and then on a case-by-case basis in the remaining patients. These last two scenarios account for the vast majority of celiac patients in our cohort. This study has some limitations, which are predominantly related to the retrospective and single-centre design in a referral center, limited sample size and lack of standardized methods to retrospectively assess and categorize the heterogeneous symptoms, which develop in celiac patients. The referral centre setting may limit generalizability of our results, given the lack of data on patient outcomes in other settings such as general practice and other secondary care settings. However, it is extremely difficult to obtain data on celiac patients followed-up in these settings. We did not find adherence to a GFD or persistence of VA on follow-up duodenal biopsy as predictors of CREs. In our cohort, only a minority of patients had clinical events attributable to poor adherence to a GFD. This is in contrast with the current literature, which reports voluntary/involuntary dietary lapses as the most common cause of NRCD (persistent/recurrent symptoms and/or villous atrophy) [12–18]. Reasons behind this may include discrepancies in methods for assessing GFD adherence, cultural differences due to the culinary and social background and strict instruction received by our patients on how to maintain lifelong rigorous dietary adherence [23, 24]. It should be noted that although persistence of VA at follow-up duodenal biopsy did not reach significance on multivariate analysis, this may be due to our limited sample size. Nonetheless, the literature provides discordant data on the relationship between persistent VA during follow-up and risk of poor long-term outcomes in celiac patients [16, 36, 37]. In conclusion, this study has delineated the natural history, etiologies and predictors of CREs in patients with CD on a long-term GFD and provided a proposal for a cost-effective optimization of follow-up care of these patients. Although requiring a confirmation on larger sample sizes, and integration with emerging methods for implementing adherence to a GFD such as gluten peptides [38], we hope these results can help clinicians to deliver the best quality of care to celiac patients. Finally, our results may potentially be relevant in selecting patients for no-biopsy diagnostic strategies for adult CD, considering the growing interest in this approach [5, 6]. Author contributions AS and FB planned the study. AS, SM, FL, DS, PM, MC, EF collected the data. SM performed the statistical analysis. AS, SM, FB interpreted the data and drafted the manuscript. All the Authors revised and approved the final version of the manuscript. Funding Open access funding provided by Università degli Studi di Pavia within the CRUI-CARE Agreement. This study received no funding. Declarations Conflict of interest The authors declare that they have no conflict of interest. Ethical approval The study was approved by the ethical review board of IRCCS Pavia, ICS Maugeri, Pavia, Italy (protocol number 2381 CE, approved on 14th January 2020). The study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki (6th revision, 2008) as reflected in a priori approval by the institution's human research committee. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Change history 5/27/2023 The co-author "Federico Biagi" e-mail address has been corrected. ==== Refs References 1. Ludvigsson JF Bai JC Biagi F Diagnosis and management of adult coeliac disease: guidelines from the British Society of Gastroenterology Gut 2014 63 1210 1228 10.1136/gutjnl-2013-306578 24917550 2. Rubio-Tapia A Hill ID Kelly CP ACG clinical guidelines: diagnosis and management of celiac disease Am. J. Gastroenterol. 2013 108 656 676 10.1038/ajg.2013.79 23609613 3. Al-Toma A Volta U Auricchio R European Society for the Study of Coeliac Disease (ESsCD) guideline for coeliac disease and other gluten-related disorders United Eur. Gastroenterol. J. 2019 7 583 613 10.1177/2050640619844125 4. Singh P Arora A Strand TA Leffler DA Catassi C Green PH Kelly CP Ahuja V Makharia GK Global prevalence of celiac disease: systematic Review and Meta-analysis Clin. Gastroenterol. Hepatol. 2018 16 6 823 836 10.1016/j.cgh.2017.06.037 29551598 5. Fuchs V Kurppa K Huhtala H Serology-based criteria for adult coeliac disease have excellent accuracy across the range of pre-test probabilities Aliment. Pharmacol. Ther. 2019 49 277 284 10.1111/apt.15109 30592070 6. Penny HA Raju SA Lau MS Accuracy of a no-biopsy approach for the diagnosis of coeliac disease across different adult cohorts Gut 2021 70 876 883 10.1136/gutjnl-2020-320913 33139268 7. Holmes GK Prior P Lane MR Malignancy in celiac disease–effect of a gluten free diet Gut 1989 30 333 338 10.1136/gut.30.3.333 2707633 8. Hall NJ Rubin G Charnock A Systematic review: adherence to a gluten-free diet in adult patients with celiac disease Aliment. Pharmacol. Ther. 2009 30 315 330 10.1111/j.1365-2036.2009.04053.x 19485977 9. Zanchetta MB Longobardi V Bai JC Bone and celiac disease Curr. Osteoporos. Rep. 2016 14 43 48 10.1007/s11914-016-0304-5 26875096 10. Schiepatti A Sprio E Sanders DS Coeliac disease and obstetric and gynaecological disorders: where are we now? Eur. J. Gastroenterol. Hepatol. 2019 31 425 433 10.1097/MEG.0000000000001361 30676472 11. Biagi F Schiepatti A Maiorano G Risk of complications in celiac patients depends on age at diagnosis and type of clinical presentation Dig. Liver Dis. 2018 50 549 552 10.1016/j.dld.2017.12.001 29277481 12. Penny HA Baggus EMR Rej A Non-responsive coeliac disease: a comprehensive review from the NHS England National Centre for Refractory Coeliac Disease Nutrients 2020 12 216 10.3390/nu12010216 31947666 13. Leffler DA Dennis M Hyett B Etiologies and predictors of diagnosis in nonresponsive celiac disease Clin. Gastroenterol. Hepatol. 2007 5 445 450 10.1016/j.cgh.2006.12.006 17382600 14. O'Mahony S Howdle PD Losowsky MS Review article: management of patients with non-responsive coeliac disease Aliment. Pharmacol. Ther. 1996 10 671 680 10.1046/j.1365-2036.1996.66237000.x 8899074 15. Abdulkarim AS Burgart LJ See J Murray JA Etiology of nonresponsive celiac disease: results of a systematic approach Am. J. Gastroenterol. 2002 97 2016 2021 10.1111/j.1572-0241.2002.05917.x 12190170 16. Kaukinen K Peräaho M Lindfors K Persistent small bowel mucosal villous atrophy without symptoms in coeliac disease Aliment. Pharmacol. Ther. 2007 15 1237 1245 10.1111/j.1365-2036.2007.03311.x 17. Stasi E Marafini I Caruso R Frequency and cause of persistent symptoms in celiac disease patients on a long-term gluten-free diet J. Clin. Gastroenterol. 2016 50 239 243 10.1097/MCG.0000000000000392 26280705 18. Fine KD Meyer RL Lee EL The prevalence and causes of chronic diarrhea in patients with celiac sprue treated with a gluten-free diet Gastroenterology 1997 112 1830 1838 10.1053/gast.1997.v112.pm9178673 9178673 19. Al-Toma A Verbeek WH Hadithi M Survival in refractory coeliac disease and enteropathy-associated T-cell lymphoma: retrospective evaluation of single-centre experience Gut 2007 56 1373 1378 10.1136/gut.2006.114512 17470479 20. Rubio-Tapia A Kelly DG Lahr BD Clinical staging and survival in refractory celiac disease: a single center experience Gastroenterology 2009 136 99 353 10.1053/j.gastro.2008.10.013 18996383 21. Biagi F Marchese A Ferretti F A multicentre case control study on complicated coeliac disease: two different patterns of natural history, two different prognoses BMC Gastroenterol. 2014 14 139 10.1186/1471-230X-14-139 25103857 22. Biagi F Vattiato C Agazzi S A second duodenal biopsy is necessary in the follow-up of adult coeliac patients Ann. Med. 2014 46 430 433 10.3109/07853890.2014.913378 24857202 23. Biagi F Andrealli A Bianchi PI A gluten-free diet score to evaluate dietary compliance in patients with coeliac disease Br. J. Nutr. 2009 102 882 887 10.1017/S0007114509301579 19331704 24. Schiepatti A Maimaris S Nicolardi ML Determinants and trends of adherence to a gluten-free diet in adult celiac patients on a long-term follow-up (2000–2020) Clin. Gastroenterol. Hepatol. 2022 20 e741 e749 10.1016/j.cgh.2020.12.015 33338656 25. Ludvigsson JF Leffler DA Bai JC The Oslo definitions for coeliac disease and related terms Gut 2013 62 43 52 10.1136/gutjnl-2011-301346 22345659 26. Parker S Palsson O Sanders DS functional gastrointestinal disorders and associated health impairment in individuals with celiac disease Clin. Gastroenterol. Hepatol. 2021 S1542–3565 00758 768 27. Zingone F Swift GL Card TR Psychological morbidity of celiac disease: a review of the literature United Eur. Gastroenterol. J. 2015 3 136 145 10.1177/2050640614560786 28. Sanders DS Carter MJ Hurlstone DP Association of adult coeliac disease with irritable bowel syndrome: a case-control study in patients fulfilling ROME II criteria referred to secondary care Lancet 2001 358 1504 1508 10.1016/S0140-6736(01)06581-3 11705563 29. Sainsbury A Sanders DS Ford AC Prevalence of irritable bowel syndrome-type symptoms in patients with celiac disease: a meta-analysis Clin. Gastroenterol. Hepatol. 2013 11 359 365 10.1016/j.cgh.2012.11.033 23246645 30. Ludvigsson JF Lebwohl B Chen Q Anxiety after coeliac disease diagnosis predicts mucosal healing: a population-based study Aliment. Pharmacol. Ther. 2018 48 1091 1098 10.1111/apt.14991 30288774 31. Schiepatti A Bellani V Perlato M Inadvertent and minimal gluten intake has a negligible role in the onset of symptoms in patients with coeliac disease on a gluten-free diet Br. J. Nutr. 2019 121 576 581 10.1017/S0007114518003616 30526695 32. Wacklin P Laurikka P Lindfors K Altered duodenal microbiota composition in celiac disease patients suffering from persistent symptoms on a long-term gluten-free diet Am. J. Gastroenterol. 2014 109 1933 1941 10.1038/ajg.2014.355 25403367 33. Schiepatti A Bacchi S Biagi F Relationship between duodenal microbiota composition, clinical features at diagnosis, and persistent symptoms in adult coeliac disease Dig. Liver Dis. 2021 53 972 979 10.1016/j.dld.2021.02.019 33741248 34. Dotsenko V Oittinen M Taavela J Genome-wide transcriptomic analysis of intestinal mucosa in celiac disease patients on a gluten-free diet and postgluten challenge Cell. Mol. Gastroenterol. Hepatol. 2021 11 13 32 10.1016/j.jcmgh.2020.07.010 32745639 35. Roncoroni L Bascuñán KA Doneda L A low FODMAP gluten-free diet improves functional gastrointestinal disorders and overall mental health of celiac disease patients: a randomized controlled trial Nutrients 2018 10 1023 10.3390/nu10081023 30081576 36. Lebwohl B Granath F Ekbom A Mucosal healing and mortality in coeliac disease Aliment. Pharmacol. Ther. 2013 37 332 339 10.1111/apt.12164 23190299 37. Rubio-Tapia A Rahim MW See JA Mucosal recovery and mortality in adults with celiac disease after treatment with a gluten-free diet Am. J. Gastroenterol. 2010 105 1412 1420 10.1038/ajg.2010.10 20145607 38. Stefanolo JP Tálamo M Dodds S Real-world gluten exposure in patients with celiac disease on gluten-free diets, determined from gliadin immunogenic peptides in urine and fecal samples Clin. Gastroenterol. Hepatol. 2021 19 484 491 10.1016/j.cgh.2020.03.038 32217152
PMC010xxxxxx/PMC10352172.txt
==== Front Dig Dis Sci Dig Dis Sci Digestive Diseases and Sciences 0163-2116 1573-2568 Springer US New York 37258979 7948 10.1007/s10620-023-07948-8 Original Article Characterization of Duodenal Microbiota in Patients with Acute Pancreatitis and Healthy Controls Zhao Meng-Qi 12 Cui Meng-Yan 12 Jiang Qiao-Li 3 Wang Jing-Jing 1 Fan Miao-Yan 12 http://orcid.org/0000-0002-6366-4969 Lu Ying-Ying [email protected] 3 1 grid.16821.3c 0000 0004 0368 8293 Shanghai Key Laboratory of Pancreatic Diseases, Institute of Translational Medicine, Shanghai General Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China 2 grid.16821.3c 0000 0004 0368 8293 Department of Gastroenterology, Shanghai General Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China 3 grid.16821.3c 0000 0004 0368 8293 Department of Gastroenterology, Jiading Branch of Shanghai General Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, 201620 China 31 5 2023 31 5 2023 2023 68 8 33413353 8 12 2022 4 4 2023 © The Author(s) 2023 https://creativecommons.org/licenses/by-nc/4.0/ Open Access This article is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License, which permits any non-commercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc/4.0/. Objective The small intestinal bacterial overgrowth (SIBO) in acute pancreatitis correlates with the severity of the disease. However, corresponding studies on the microbial composition of the duodenal mucosa of patients are uncommon. Methods Duodenal mucosal biopsies were collected by gastroscopy from 16 patients with mild acute pancreatitis (the Ap group) and 16 healthy individuals (the control group) and subjected to histological studies as well as bacterial 16S rRNA gene sequencing. Caerulein and l-arginine were used to induce mild acute pancreatitis (MAP) and severe acute pancreatitis (SAP) in mice, respectively, and their pancreas and duodenum were collected for histological studies. Results H&E analysis displayed no significant pathological damage in the descending duodenum of patients with acute pancreatitis compared with that of the controls. Immunofluorescence and Real-time PCR revealed that the expressions of tight junction proteins (TJPs) in duodenal mucosa were decreased in acute pancreatitis. The results of the alpha diversity analysis revealed no significant difference between the two groups, while LEfSe and the random forest revealed a few differences, indicating that the descending duodenum mucosal microbiota changed slightly in patients with mild acute pancreatitis. We observed the pathological changes and the expression of TJPs in the duodenum in the three groups of mice and found that SAP mice had more severe pathological damage in the duodenum. Furthermore, the expression of TJPs in the duodenum was lower in the MAP and SAP groups of mice compared to control mice, but it was similar in both groups. Conclusion Patients with mild acute pancreatitis had mild duodenal barrier dysfunction and slight changes in duodenal mucosal microbiota. Supplementary Information The online version contains supplementary material available at 10.1007/s10620-023-07948-8. Keywords Acute pancreatitis Duodenal mucosa Microbiota Occludin ZO-1 http://dx.doi.org/10.13039/100007219 Natural Science Foundation of Shanghai 22ZR1453500 Science and Technology Project of Jiading Hospital, Shanghai General Hospital202135A Key Discipline Project of Shanghai Jiading District, Shanghai2020-jdyxzdxk-15 issue-copyright-statement© Springer Science+Business Media, LLC, part of Springer Nature 2023 ==== Body pmcIntroduction Acute pancreatitis (AP) often has a moderate and self-limiting clinical course, but in 20–30% of cases, it can escalate to a severe illness with systemic inflammatory response syndrome SIRS and multiple organ dysfunction syndromes (MODS). Necrotic pancreatic tissue infection with enteric-derived bacteria happens in 30–70% of these patients [1]. These infectious complications are typically the cause of death in patients suffering from acute pancreatitis, with a mortality rate as high as 50% [2–4]. It has been demonstrated that intestinal barrier dysfunction allows bacteria of intestinal origin to transfer to extraintestinal organs, leading to sepsis and subsequent infectious diseases with a high mortality rate [5]. The significance of gut microbiota in the emergence and transmission of critical illnesses is becoming more widely acknowledged. According to studies, the presence of small intestinal bacterial overgrowth (SIBO) in acute pancreatitis correlates with the severity of the disease [6–8]. SIBO is characterized as bacterial overgrowth in the small intestine caused by an abnormally high number of bacteria and/or changes in the types of bacteria. The authors deem the existence of more than 105 bacteria per milliliter of proximal jejunal aspiration to be indicative of SIBO [9]. Although the gold standard for the diagnosis of SIBO is the use of small intestinal fluid bacterial cultures, clinical studies have mostly used portable breath hydrogen concentration meters to measure patients' breath hydrogen concentrations for the diagnosis of intraluminal SIBO [10, 11]. SIBO was found to be more common in MSAP or SAP patients than in MAP patients, and SIBO was linked to organ failure and acute respiratory distress syndrome (ARDS), a complication of acute pancreatitis. Malnutrition, severe malabsorption, and deficiency syndromes can be caused by SIBO. However, it is still unknown how the bacteria in the duodenum change during AP. The response of acute pancreatitis to chemicals in the duodenum is facilitated by neuronal cross-talk between the duodenum and pancreas [12]. Additionally, when the pancreas is dysfunctional, it can cause a decrease in bicarbonate secretion, resulting in the inability of the proximal duodenum to neutralize gastric acid, thus inducing duodenal inflammation [13]. Numerous studies indicate that acute pancreatitis commonly alters the intestinal microbiota and that intestinal microbiota dysbiosis might impair immunological function, increase intestinal permeability, and encourage intestinal bacterial metastasis [14, 15]. It was found through animal experiments that the small intestine may be the main source of intestinal bacteria in infected pancreatic necrosis, rather than the colon [16]. Another study in rats, in which fluorescent bacteria were observed to migrate from the lumen of the small intestine into the pancreas by Intravital microscopy of fluorescent bacteria, provided substantial experimental evidence for the hypothesis of an intestinal origin-pancreatitis infection complication [17]. Similarly, duodenal ecological dysregulation has been found in patients as a potential risk for infection in patients with severe acute pancreatitis [18]. In this study, the examination of the generated metadata by the SourceTracker tool revealed that the duodenum was the most likely source of bacteria that later invaded pancreatic tissue in this model [19]. About 90% of the energy in the meal is absorbed by the small intestine mucosa, which has a large surface area [20]. A sufficient capacity for absorption depends on mucosal integrity, which is continually threatened by the intraluminal environment, including bacteria [21]. Several studies have used animal models to investigate how the gut microbiota changes during acute pancreatitis [22, 23]. Similarly, significant differences in the structure of the gut microbiota between acute pancreatitis and healthy control have been observed in adult participants, and a disordered microbiota is strongly associated with systemic inflammation and intestinal barrier dysfunction. However, these studies only focused on the microbiota in the intestine [15]. There aren't many studies that correspond to the microbial makeup of patients' duodenums. The big duodenal papilla, where the common bile duct and pancreatic duct exit, brings the duodenum and pancreas closer anatomically. Therefore, we hypothesize that microbiota can pass retrogradely through the pancreatic duct and that duodenal microbiota and infection in pancreatitis may be related. The structural alterations in the duodenal mucosal microbiota in people with acute pancreatitis are the main focus of this investigation. Materials and Methods Clinical Trial Design and Sampling This study examines the differences in intestinal microbiota among acute pancreatitis patients and healthy individuals. It involved 32 patients who underwent a gastroscopy at the Jiading Branch of Shanghai General Hospital (Shanghai) between October 2021 and February 2022. After undergoing a routine physical examination, 16 patients with acute pancreatitis (11 males and 6 women) were assigned to the Ap group. The Ap group was mostly mild to moderate. 16 healthy volunteers (11 males, 6 females, ages 32–58) were assigned to the control group. Abdominal pain was the most prevalent symptom among the 16 individuals with acute pancreatitis, followed by abdominal distension, nausea, and vomiting. Patients and healthy controls were excluded if they took antibiotics, probiotics, Chinese herbs, steroids, and other substances that may influence the structure of the microbiota up to 8 weeks before sampling. Other exclusion criteria involved patients with a history of gastroscopy intolerance, immunodeficiency, allergy, asthma, celiac disease, colon cancer, diabetes, HIV, inflammatory bowel disease, irritable bowel syndrome, gastroenteritis, narcotic small bowel colitis, and arthritis. Before collecting any samples, all patients provided written informed consent, which was authorized by the research ethics committees of Shanghai General Hospital. Every biochemical analysis was handled by the Shanghai General Hospital Laboratory. After an overnight fast, blood samples were taken on the second day of hospitalization. We collected mucosal biopsy samples that were obtained via gastroscopy from the two groups around the 5th day after diagnosis. Each patient had two duodenal biopsy samples obtained during the upper endoscopy. One tissue sample was frozen immediately and stored at − 80 °C, while the other was fixed in 10% formaldehyde in phosphate-buffered saline (PBS; pH 7.4) and embedded in paraffin. Mice Experimental Design Shanghai Leagene Biotechnology Co.Ltd (Shanghai, China) provided male C57BL/6 mice weighing 20–22 g. All mice were kept in special pathogen-free (SPF) environments that had a temperature of 23 ± 2 °C, a 12-h cycle of light and dark, and free access to water and mouse chow. The guidelines of the Animal Care and Use Committee of Shanghai Jiao Tong University were followed in all animal experimentation. The mice were randomly divided into three groups: healthy control (the CON group), caerulein-induced pancreatitis (the MAP group), and l-arginine-induced pancreatitis (the SAP group) and five mice in each group. Mice in the MAP group were given 100 μg/kg caerulein intraperitoneally every hour for 10 times and mice in the SAP group received two intraperitoneal injections of l-arginine hydrochloride (8%, pH = 7.0, 4.5 g/kg) given every 2 h. Histopathology Duodenal biopsy tissue samples were immobilized in 4% paraformaldehyde, dehydrated, embedded in paraffin, cut into 4 μm sections, and stained using hematoxylin and eosin (H&E). The morphologic evaluation was performed by a light microscope (Leica, Germany) at a magnification of × 200 according to a modified Chiu's score: Grade 0, normal mucosal villi, no ulceration or tissue damage, and no inflammatory reactions; Grade 1, Limited infiltration of inflammatory cells in the lamina propria and the development of a subepithelial Gruenhagen space, typically at the apex of the villus; Grade 2: diffuse infiltration of inflammatory cells into the lamina propria, extending into the subepithelial space, with moderate epithelial elevation; Grade 3: Subendothelium localized infiltration of inflammatory cells and severe epithelial raising down the sides of the villi; Grade 4, exposed lamina propria and dilated capillaries, as well as a diffuse infiltration of inflammatory cells in the subendothelium; Grade 5, digestion, disintegration, bleeding, and ulceration of the lamina propria along with severe inflammatory reactions [24]. The pancreatic and duodenal specimens were fixed with 4% paraformaldehyde and dehydrated. Then they were embedded in paraffin, cut into 4 μm sections and stained with H&E. A light microscope (Leica, Germany) was used to evaluate the morphology at magnifications of × 100 and × 200. According to Schmidt's criteria, the histological abnormalities in the pancreas, including pancreatic edema, acinar cell necrosis, bleeding, and inflammation, were assessed [25]. Immunofluorescence Duodenal mucosa sections were heated for 1 h at 60 °C. Next, sections were put in Leica Autostainer XL (Leica, USA) (xylene for 40 min, 100% ethanol for 10 min, 95% ethanol for 10 min, 80% ethanol for 5 min, 70% ethanol for 5 min, and doubly distilled water for 3 min) to deparaffinized and rehydrate the samples. A citrate antigen retrieval solution was used to extract antigens (Sangon Biotech, China). After washing the tissue several times in phosphate-buffered saline (PBS), a super pap pen (Sangon Biotech, China) was used to draw a circle around it. Slides were incubated with primary antibodies against Occludin and ZO-1 diluted with primary antibody dilution solution (Servicebio, China) for overnight incubation at 4 °C after being blocked for an hour at room temperature with immunostaining blocking buffer (Sangon Biotech, China). The slides were cleaned with PBS and then incubated with Alexa Fluor 488 AffiniPure donkey antirabbit IgG (Yeason, China) for 1 h at room temperature. Then, the slides were rinsed with PBS and dyed with dihydrochloride (Yeason, China) for 10 min. Images were captured with a fluorescence microscope (Leica, USA). Quantitative Reverse Transcription PCR Total RNA was extracted from tissues using TRIzol (Invitrogen, USA) and a tissue RNA purification kit plus (EZBioscience, USA). cDNA was synthesized using HyperScript III RT SuperMix for quantitative PCR (qPCR) with genomic DNA (gDNA) remover (EnzyArtisan, China). A NanoDrop2000 equipment (Thermo Scientific, USA) was used to measure the concentration of RNA or DNA. With QuantStudio 6 Flex real-time PCR systems, the real-time PCR was carried out according to the procedure listed below (Thermo Scientific, USA): predenaturation (95 °C for 30 s), 40 amplification cycles of denaturation (95 °C for 10 s), and annealing and extension (60 °C for 30 s). Gene expression was measured by the 2−ΔΔCT method. The detection primers are listed in Table S1. DNA Extraction PCR Amplification and Sequencing of Bacteria in Duodenal Mucosa Duodenal biopsy tissues were soaked in 20 μl of 20 mg/ml protease solution and 300 μl of PBS solution at 60 °C overnight. The samples were suspended in 790 μl of sterile lysis buffer (4 M guanidine thiocyanate; 10% N-lauroyl sarcosine; 5% N-lauroyl sarcosine-0.1 M phosphate buffer, pH 8.0) in 2 ml screw-cap tube containing 1 g glass beads (0.1 mm BioSpec Products, Inc., USA). After vigorous vortexing, this mixture was incubated for an hour at 70 °C while being forcefully bead beaten for 10 min. The E.Z.N.A.® Stool DNA Kit (Omega Bio-tek, Inc., GA), which skips the lysis steps, was used to extract microbial genomic DNA from the duodenal biopsies in accordance with the manufacturer's instructions and was stored at − 20 °C for further analysis. The universal bacterial primers F1 and R2 (5'- CCTACGGGNGGCWGCAG-3' and 5'-GACTACHVGGGTATCTAATCC-3'), which map to locations 341 to 805 in the Escherichia coli 16S rRNA gene, were used to amplify the V3–V4 region of the bacterial 16S ribosomal RNA gene from each sample. The following program was used to run PCR reactions in an EasyCycler 96 PCR system (Analytik Jena Corp., AG): 3 min of denaturation at 95 °C, followed by 21 cycles of 0.5 min of denaturation at 94 °C, 0.5 min of annealing at 58 °C, and 0.5 min elongation at 72 °C, with a final 5 min extension at 72 °C. Shanghai Mobio Biomedical Technology Co. Ltd. sequenced the products from various samples using the Miseq platform (Illumina Inc., USA) in accordance with the manufacturer's instructions. Bioinformatics Analysis of Sequencing Data USEARCH (version 11.0.667) was used to extract clean data from raw data using the following criteria: (i) Sequences of each sample were extracted with zero mismatches using each index. (ii) Sequences with less than 16 bp overlap were discarded. (iii) Any overlap error rate greater than 0.1 was discarded. (iv) Sequences with less than 400 bp after the merge was removed. Quality-filtered sequences were grouped into distinct sequences and sorted in decreasing abundance using UPARSE in accordance with the UPARSE OTU analysis pipeline, excluding singletons, in order to find representative sequences. Operational Taxonomic Units (OTUs) were classified using UPARSE (version 7.1 http://drive5.com/uparse/) and annotated using the SILVA reference database (SSU138) based on 97 percent similarity after chimeric sequences were removed. Mothur v1.42.1 was used to evaluate the Alpha diversity metrics (ACE estimator, Chao 1 estimator, Shannon–Wiener diversity index, and Simpson diversity index). We employed principal coordinates analysis (PCoA) and nonmetric multidimensional scaling (NMDS) plots based on the unweighted UniFrac dissimilarity to depict the structural diversity of the gut microbiome in the discovery group. The statistical significance of beta diversity was assessed using an Adonis analysis. To ascertain whether there were notable variations in microbial composition between groups, a Wilcoxon rank-sum test was utilized. To detect taxa with differential abundance among groups, the linear discriminant analysis (LDA) effect size (LEfSe) method was used (LEfSe version 1.1, https://github.com/SegataLab/Lefse). As expected by a random forests model, the microbiome heatmap analysis showed distinct gut microbiomes in both groups. PICRUSt2 v2.4.1 was employed to forecast functional abundances based on 16S rRNA gene sequences. Statistical Analysis The data were displayed as the mean ± standard deviation (SD). The Student's t test, Mann–Whitney test, and Spearman correlation test were carried out using SPSS 19.0 software, and a p value of < 0.05 was considered statistically significant. Data Availability The Sequence Read Archive database, accession number PRJNA877061, contains the 16S rRNA gene V3-V4 regions' raw sequencing data as well as the information that goes with it. Results Clinical Characteristics This duodenal mucosal microbial profiling study involved 32 participants. Table 1 summarizes the characteristics of the 16 healthy controls and the 16 patients with acute pancreatitis, including demographics, fasting blood glucose level, blood fat (TC, TG), and plasma inflammatory markers such as CRP, IL-6, WBC, and PCT. Some baseline characteristics, such as age, BMI, PT and blood glucose levels, did not differ significantly between the two groups. However, TC and TG levels were higher in the group of patients with acute pancreatitis. The patient group also had significantly higher levels of CRP, IL-6, WBC, and PCT, indicating an acute inflammatory disease.Table 1 Baseline characteristics of the individuals enrolled Variable Healthy controls (n = 16) Acute pancreatitis patients (n = 16) p Value Age (mean ± SD) 42.50 ± 7.16 46.06 ± 11.18 0.056 Gender  Female 6 (37.5%) 6 (37.5%)  Male 10 (62.5%) 10 (62.5%) BMI (kg/m2) 24.69 ± 2.91 25.67 ± 3.60 0.618 Glucose(mmol/L) 5.59 ± 0.81 6.10 ± 0.92 0.531 WBC (× 109) 5.98 ± 1.15 9.33 ± 3.43  < 0.05 PLT (× 109) 235.69 ± 45.93 206.31 ± 45.49 0.833 CRP (mg/L) 6.42 ± 1.70 65.68 ± 76.21  < 0.001 IL-6 (pg/mL) 4.96 ± 2.21 32.57 ± 33.75  < 0.001 PCT (ng /ml) 0.040 ± 0.01 0.12 ± 0.04  < 0.05 PT (s) 12.10 ± 0.72 12.78 ± 0.99 0.489 TC (mmol/L) 3.90 ± 0.80 4.81 ± 1.30  < 0.05 TG (mmol/L) 1.90 ± 0.98 2.49 ± 2.34  < 0.05 Histopathological and Expression Levels of TJPs Most healthy controls had normal duodenal epithelium and villi, but there were mucosal specimens in the lamina propria that showed mild chronic inflammation. (Fig. 1A). In the duodenum of patients with acute pancreatitis, we found no mucosal damage, villi atrophy, or rupture, but there was more severe hemorrhage and inflammatory cell infiltration. However, when compared to the control group, there were no appreciable variations in the duodenal pathology scores. To determine whether duodenal intestinal permeability had changed, immunofluorescence was used to examine the expression of intestinal TJPs Occludin and ZO-1 in the duodenum of patients with acute pancreatitis. Compared with healthy individuals, the expression of Occludin and ZO-1 decreased in AP individuals slightly, indicating some degradation of the duodenal mucosal barrier in acute pancreatitis (Fig. 1B). By using real-time qPCR, we discovered that the expression of TJPs (Occludin, ZO-1, and Claudin1) was higher in healthy controls than in patients with acute pancreatitis, with Occludin being statistically significant (Fig. 1C). To verify the duodenal mucosal barrier function at different disease severity, we observed the pathology and TJPs changes in the duodenum of three groups of mice (the CON, MAP and SAP groups). When compared to controls, mice with MAP had some pathological damage to the duodenum, whereas mice with SAP had more severe pathological damage to the duodenum (Fig. S4AB). Furthermore, the expression of TJPs in the mouse duodenum was found to be lower in the MAP and SAP groups compared to control mice, but similar in both groups (Fig. S4C).Fig. 1 H&E-stained tissue sections of duodenum and protein and mRNA expression levels of TJPs. A The evaluation of intestinal injury with modified Chiu's scores in the two groups. Microscopic images were taken at 200 × magnification. Scale bar = 50 μm. B Representative images of fluorescence staining of TJPs (red) of the duodenum. C mRNA expression levels of TJPs (Claudin1, Occludin and ZO-1). Symbol *means p < 0.05, **means p < 0.01, ***means p < 0.001, ns means p > 0.05, Student’s t test MiSeq Sequencing Data Summary and Alpha Diversity Analysis of the Duodenal Bacterial Microbiota In total, 877,091 usable sequences were obtained from 32 samples using Illumina MiSeq sequencing, with an average of 27,409 sequences per sample. From these, 732 OTUs were identified at a 97% similar level. With the current sequencing, the rarefaction curves achieved a plateau, and the Shannon diversity estimates of all the samples were constant (Fig. S1A, B), suggesting that most diversity had already been captured. According to Venn diagrams, there were 626 OTUs in both groups, with 54 specific to the group of patients with acute pancreatitis and 52 specific to the healthy control group (Fig. 2A). Based on the OTUs observed in each sample and the ACE and Chao indices, which reflect the richness of species diversity, it was discovered that the acute pancreatitis group had fewer OTUs, but the two groups’ microbiota diversity and richness did not significantly differ from one another. (Fig. 2B–F).Fig. 2 Operational Taxonomic Units clustering and alpha diversity analysis of the microbiota of the duodenal mucosa. A Venn diagram demonstrates the shared and unique Operational Taxonomic Units (OTUs) in both groups. B The OTUs in the single sample from each group. C–F The estimators of ACE, Chao, Shannon and Simpson of duodenal bacterial microbiota in each group Analysis of the Beta Diversity Based on OTU Levels Beta diversity was used to ascertain how samples differed from one another. To compare whether samples had significant microbial community differences, we used the widely used unweighted UniFrac analysis, which was calculated based on evolutionary information between individual sample sequences. Adonis analysis revealed a significant difference in the two groups (Adonis: p = 0.0169). Based on the unweighted UniFrac distances, a PCoA also showed that the microbial composition of the Ap group deviated from the control group according to the PC1, which accounts for 17.64%, demonstrating that the overall gut microbiota in the two groups had different bacterial microbiota compositions (Fig. 3A).Fig. 3 Analysis of beta diversity of the duodenal mucosa in both groups. A Principal coordinates analysis plots based on unweighted UniFrac distances between the two groups. B Nonmetric multidimensional scaling analysis plots based on unweighted UniFrac distances between the two groups. C Cluster tree between the two groups Comparison of the Bacterial Microbiota in Healthy Volunteers and Acute Pancreatitis To further elucidate the differences in microbial structure, we calculated the relative abundance of bacteria. The average relative abundance of microbiomes at the phylum level is represented in Fig. 4A. In the control group, the duodenal mucosal microbiota was dominated by Proteobacteria, followed by Actinobacteriota, Firmicutes, and Bacteroidota with proportions of 48.2%, 25.4%, 13.5%, and 9.1%, respectively. Firmicutes proliferated while Proteobacteria and Bacteroidetes decreased when compared with the CON group, but the Wilcoxon rank sum test did not find any significant differences in these bacterial phyla between the two groups. (Proteobacteria 43.8%, Actinobacteriota 24.6%, Firmicutes 22.5% and Bacteroidetes 7.1%).Fig. 4 Alterations in the microbiota of the duodenal mucosa in both groups. A Microbiota composition at the phylum level in the duodenal mucosa. B Microbiota composition at the genus level in the duodenal mucosa. C Differences in bacterial composition in the duodenal mucosa between the acute pancreatitis patients and healthy controls. Symbol *means p < 0.05, **means p < 0.01, Mann–Whitney U tests Figure S2A shows the average relative abundance of microbiomes at the class level, the duodenal mucosal microbiota was dominated by Gammaproteobacteria, Actinobacteria, Bacteroidia and Bacilli in the control group. The microbiota in the Ap group had a comparable makeup to that in the control group. The average relative abundance of microbiomes at the order level is depicted in Fig. S2B, the composition of both groups was dominated by Oceanospirillales, Corynebacteriales, Burkholderiales and Propionibacteriales. In comparison to the control group, the Ap group had a considerably higher concentration of Lactobacillales. At the family level, the duodenal mucosal microbiota was dominated by Halomonadaceae, Dietziaceae, Idiomarinaceae and Nocardioidaceae (Fig. S2C). Streptococcaceae abundance was noticeably higher in the Ap group compared to the control group. At the genus level, the duodenal mucosal microbiota was dominated by Halomonas in the Ap group, followed by Dietzia, Aliidiomarina and Aeromicrobium with proportions of 15.5%, 11.1%, 8.0%, 4.5%, respectively. Similarly, these four species were the dominant genera in the control group, and their proportions were 18.3%, 11.4%, 8.7%, and 8.1% (Fig. 4B). The duodenal mucosal microbiota in the acute pancreatitis group had a higher abundance of Streptococcus and Neisseria. Halomonas and Achromobacter were more prevalent in the healthy group than in the Ap group. The Wilcoxon rank sum test indicated that Actinobacillus and Oribacterium were significantly different in the two groups, and both were more abundant in the healthy tissue (Fig. 4C). Furthermore, we discovered that the abundance of Actinomycetes and Oribacterium was inversely correlated with clinical inflammatory indices in the control and Ap groups, but not statistically different. (p > 0.05; Fig. 3S). Differences in Microbiome Compositions Between the Groups We performed LEfSe, showing the microbial structure that differed most between patients with acute pancreatitis and healthy controls (Fig. 5A). According to LEfSe analysis at the genus level, Microtrichales, Pseudomonas, Ruminococcaceae, and Pediococcus were more abundant in the duodenal mucosa of acute pancreatitis patients. Conversely, the abundances of various genera, including Bacteria, Prevotellaceae, Bacilli_RF39, Bilophila, Candidatus Saccharimonas, Treponema, and Stomatobaculum, were significantly higher in the control group than in the Ap group.Fig. 5 Comparison of microbiome composition and Predictions of the functional alteration to the microbiota in both groups. A microbiota differences at the genus level, as assessed by Linear discriminant analysis effect size. B A heatmap analysis of the microbiomes using a random forest model. C Predictions of the functional alteration to the microbiota of the duodenal mucosa in patients with acute pancreatitis It was found that the gut microbiomes of both groups differed using a random forest model (Fig. 5B). There were discovered to be 21 OTUs in all that was different between the two groups. Among these OTUs, 6 OTUs were more abundant in the Ap group than in the control group, belonging to the genera of Pseudomonas, Halomonas, Lachnospiraceae_NK4A136_group, Microtrichales, Bacteria, and Nesterenkonia. Correspondingly, OTUs belonging to the genera of Streptococcus, Actinobacillus, Haemophilus, Prevotellaceae, Saccharimonadaceae, Candidatus_Saccharimonas, etc., were more abundant in the control group than in the Ap group. Functional Alterations of Duodenal Mucosal Microbiota Microbiomes in Both Groups Based on the KEGG module and KEGG pathway databases, 16S sequencing data was used for functional prediction, and LEfSe was then used to sort the different metabolic modules and pathways between the groups. The results showed that Endocrine_system, Glycerolipid_metabolism, and Dioxin_degradation were significantly higher in acute pancreatitis individuals compared to healthy individuals. However, ABC_transporters, Glycine_serine_and_threonine_metabolism, Selenocompound_metabolism, Apoptosis, Plant_pathogen_interaction, and Environmental_adaptation were all significantly higher in the CON group than in the Ap group (Fig. 5C). Discussion AP is distinguished by an inflammatory cascade response that typically results in an overgrowth of intestinal bacteria, which exacerbates the disease by activating the innate immune system and bacterial translocation. The ileum is thought to have the most altered intestinal microbiota in acute pancreatitis [22]. Although research has shown that the displaced microbiota is from the small intestine, no research has been conducted to determine whether or not the duodenal microbiota changes and how these changes affect disease [16]. In this study, we examined the mucosal microbiota of the descending duodenum and discovered that it changed slightly in acute pancreatitis. The baseline characteristics of participants showed significantly elevated inflammatory factors in blood in the acute pancreatitis group. However, when we examined the morphological changes of the descending duodenum by pathology, the findings revealed no significant pathological damage to the descending duodenum in patients with acute pancreatitis compared to controls, including mucosal damage, inflammation, and hemorrhage/congestion. The intestinal mucosa normally serves as a barrier, we further tested the mucosal tight junction protein (Occludin and ZO-1) of the descending duodenum and it was discovered that acute pancreatitis impairs the duodenal barrier function. Since intestinal damage can enhance mucosal permeability and bacterial translocation, we used 16S rRNA gene sequencing to examine the composition of the descending duodenum's mucosal microbiota in acute pancreatitis. The alpha diversity was not significantly different between the groups in this study; differences in microbiota between the two groups were detected by beta diversity and found to be different by the Adonis test based on unweighted UniFrac distances. It showed that the abundance of the mucosal microbiota of the descending duodenum had some slight changes in mild acute pancreatitis. This is consistent with a study that found no obvious change in the composition of the duodenal microbiota despite acute pancreatitis-induced duodenal bacterial overgrowth. Further examination of the bacteriophage structure revealed that both groups primarily belonged to the Proteobacteria and Actinobacteriota phyla. Other researchers have reported similar results [26]. The dominant genera were roughly the same in both groups at the genus level, but the descending duodenal mucosa had higher levels of Streptococcus and Neisseria than the control group, and the healthy controls had higher levels of Halothrix and Azithromycin. Streptococcus and Neisseria are well-known pathogenic bacteria that have been linked to diseases such as meningitis and pneumonia. The Wilcoxon rank sum test revealed that Actinobacillus and Oribacterium were significantly different between the two groups, with both being more prevalent in the healthy control group. There is no significant correlation between different genera and clinical inflammatory indicators using Spearman correlation analysis, which we think is because the duodenal mucosa in this investigation was collected from individuals with MAP. We speculate that changes in inflammatory markers and microbiota are more significant in SAP patients. We hypothesized that changes in inflammatory markers and microbiota are more significant in SAP patients, so we performed mouse experiments to confirm this. As a result, we examined the duodenum of healthy, MAP, and SAP mice and discovered that SAP mice had more serious duodenal mucosal damage than MAP and healthy mice. We hypothesized that changes in duodenal mucosal microbiota were related to disease severity. In addition, the fact that the number of bacteria in the digestive tract gradually increases from top to bottom as the digestive tract progresses could explain this result to some extent. Because gastric juice contains a high concentration of digestive enzymes, the pH in the stomach is low, preventing intestinal bacteria from reproducing. The microbiota of the duodenum and stomach were essentially identical, but they contained coliform and anaerobic bacteria, whereas the abundance and diversity of microbiota in the jejunum, ileum, and colon increased gradually. This is also supported by the results of the current investigation, which found no significant inflammatory changes in the mucosal pathology of the duodenum after observing the mucosal pathology of the descending duodenum in acute pancreatitis. This speculation was also confirmed by this study, which found a gradual increase in the number of different genera along the gastrointestinal tract, based on alpha and beta diversity measurements. Moreover, this study found that the SAP duodenal mucosal microbiota showed a decrease in the genus Alistipes belonging to the phylum Bacteroidetes and an increase in the phylum Firmicutes including the genera Romboutsia and Turicibacter compared to the control group [19]. According to one study that investigated the microbiota of MAP and SAP mice, alterations in the microbiota were particularly noticeable in duodenal aspirates. However, the changes in gut microbiota composition were less pronounced in MAP compared to SAP. Furthermore, this study clarified that facultatively pathogenic bacteria such as Escherichia/Shigella, Enterobacteriaceae diversa, Enterococcus, and Staphylococcus were significantly enriched during SAP, which is also a sign of duodenal bacterial overgrowth [27]. There are several limitations to this study. For instance, the sample size was initially somewhat tiny. The possibility of statistically significant results could be achieved by increasing the sample size. Second, the duodenal mucosal microbiota was not tested in patients with severe acute pancreatitis, and the effect of disease severity on the microbiota was not validated in mice and human. In conclusion, we demonstrate that minor damage to the duodenal mucosa occurs in MAP patients, as well as changes in the microbiota. We also found that the duodenal mucosal barrier was affected by disease severity by examining the duodenum of mice with varying disease severity. Although the relationship and mechanism between duodenal microbiota and AP remain to be confirmed, our research on duodenal mucosal microbiota extends the alteration of intestinal microbiota in acute pancreatitis and could be used as therapeutic targets in the future. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (DOCX 653 kb) The authors would like to thank Chen H for her kind assistance and skilled technical assistance. Thanks to the Shanghai Natural Science Foundation Project (Grant No.22ZR1453500), the Science and Technology Project of Jiading Hospital, Shanghai General Hospital (Grant No. 202135A) and the Key Discipline Project of Shanghai Jiading District, Shanghai (2020-jdyxzdxk-15) for funding. Author’s contribution YL designed the research work. MC and MF performed the research activities. QJ conducted a statistical analysis. JW and MZ prepared figures and wrote the manuscript. MZ edited the manuscript submitted. All authors have read and given their approval for publication. Funding This work was supported by the Shanghai Natural Science Foundation Project (Grant No.22ZR1453500), the Science and Technology Project of Jiading Hospital, Shanghai General Hospital (Grant No. 202135A) and the Key Discipline Project of Shanghai Jiading District, Shanghai (2020-jdyxzdxk-15). Data availability The 16S rRNA sequence data generated in this study have been deposited in the Sequence Read Archive database under accession number PRJNA877061. Declarations Conflict of interest We confirm that there are no known conflicts of interest associated with this publication and there has been no significant financial support for this work that could have influenced its outcome. The authors declare no competing financial interests. Ethical approval The studies involving humans and animals were reviewed and approved by the Clinical Center Laboratory Animal Welfare and Ethics committee of Shanghai First People’s Hospital. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Beger HG Rau BM Severe acute pancreatitis: Clinical course and management World J Gastroenterol 2007 13 5043 5051 10.3748/wjg.v13.i38.5043 17876868 2. Ammori BJ Role of the gut in the course of severe acute pancreatitis Pancreas 2003 26 122 129 10.1097/00006676-200303000-00006 12604908 3. Beger HG Rau B Isenmann R Natural history of necrotizing pancreatitis Pancreatology 2003 3 93 101 10.1159/000070076 12774801 4. Schmid SW Uhl W Friess H The role of infection in acute pancreatitis Gut 1999 45 311 316 10.1136/gut.45.2.311 10403749 5. Clark JA Coopersmith CM Intestinal crosstalk: a new paradigm for understanding the gut as the "motor" of critical illness Shock 2007 28 384 393 10.1097/shk.0b013e31805569df 17577136 6. Pimentel M Saad RJ Long MD ACG Clinical Guideline: Small Intestinal Bacterial Overgrowth Am J Gastroenterol 2020 115 165 178 10.14309/ajg.0000000000000501 32023228 7. Gerritsen J Timmerman HM Fuentes S Correlation between protection against sepsis by probiotic therapy and stimulation of a novel bacterial phylotype Appl Environ Microbiol 2011 77 7749 7756 10.1128/AEM.05428-11 21926217 8. Van Felius ID Akkermans LM Bosscha K Interdigestive small bowel motility and duodenal bacterial overgrowth in experimental acute pancreatitis Neurogastroenterol Motil 2003 15 267 276 10.1046/j.1365-2982.2003.00410.x 12787336 9. Bouhnik Y Alain S Attar A Bacterial populations contaminating the upper gut in patients with small intestinal bacterial overgrowth syndrome Am J Gastroenterol 1999 94 1327 1331 10.1111/j.1572-0241.1999.01016.x 10235214 10. Zhang M Zhu HM He F Association between acute pancreatitis and small intestinal bacterial overgrowth assessed by hydrogen breath test World J Gastroenterol 2017 23 8591 8596 10.3748/wjg.v23.i48.8591 29358867 11. Liang XY Jia TX Zhang M Intestinal bacterial overgrowth in the early stage of severe acute pancreatitis is associated with acute respiratory distress syndrome World J Gastroenterol 2021 27 1643 1654 10.3748/wjg.v27.i15.1643 33958849 12. Li C Zhu Y Shenoy M Anatomical and functional characterization of a duodeno-pancreatic neural reflex that can induce acute pancreatitis Am J Physiol Gastrointest Liver Physiol 2013 304 G490 500 10.1152/ajpgi.00012.2012 23306082 13. Futagami S Wakabayashi M Pancreatic Dysfunction and Duodenal Inflammatory Responses Coordinate with Refractory Epigastric Pain Including Functional Dyspepsia: A Narrative Review J Nippon Med Sch 2022 89 255 262 10.1272/jnms.JNMS.2022_89-311 35082205 14. Tan C Ling Z Huang Y Dysbiosis of Intestinal Microbiota Associated With Inflammation Involved in the Progression of Acute Pancreatitis Pancreas 2015 44 868 875 10.1097/MPA.0000000000000355 25931253 15. Zhu Y He C Li X Gut microbiota dysbiosis worsens the severity of acute pancreatitis in patients and mice J Gastroenterol 2019 54 347 358 10.1007/s00535-018-1529-0 30519748 16. Fritz S Hackert T Hartwig W Bacterial translocation and infected pancreatic necrosis in acute necrotizing pancreatitis derives from small bowel rather than from colon Am J Surg 2010 200 111 117 10.1016/j.amjsurg.2009.08.019 20637344 17. Samel S Lanig S Lux A The gut origin of bacterial pancreatic infection during acute experimental pancreatitis in rats Pancreatology 2002 2 449 455 10.1159/000064714 12378112 18. Ma X Huang L Huang Z The impacts of acid suppression on duodenal microbiota during the early phase of severe acute pancreatitis Sci Rep 2020 10 20063 10.1038/s41598-020-77245-1 33208878 19. van den Berg FF Hugenholtz F Boermeester MA Spatioregional assessment of the gut microbiota in experimental necrotizing pancreatitis BJS Open 2021 5 5 zrab061 10.1093/bjsopen/zrab061 34518874 20. Leser TD Molbak L Better living through microbial action: the benefits of the mammalian gastrointestinal microbiota on the host Environ Microbiol 2009 11 2194 2206 10.1111/j.1462-2920.2009.01941.x 19737302 21. Sekirov I Russell SL Antunes LC Gut microbiota in health and disease Physiol Rev 2010 90 859 904 10.1152/physrev.00045.2009 20664075 22. Chen J Huang C Wang J Dysbiosis of intestinal microbiota and decrease in paneth cell antimicrobial peptide level during acute necrotizing pancreatitis in rats PLoS One 2017 12 e0176583 10.1371/journal.pone.0176583 28441432 23. Huang C Chen J Wang J Dysbiosis of intestinal microbiota and decreased antimicrobial peptide level in paneth cells during hypertriglyceridemia-related acute necrotizing pancreatitis in rats Front Microbiol 2017 8 776 10.3389/fmicb.2017.00776 28522995 24 Chiu CJ McArdle AH Brown R Intestinal mucosal lesion in low-flow states. I. A morphological, hemodynamic, and metabolic reappraisal Arch Surg 1970 101 478 483 10.1001/archsurg.1970.01340280030009 5457245 25. Schmidt J Rattner DW Lewandrowski K A better model of acute pancreatitis for evaluating therapy Ann Surg 1992 215 44 56 10.1097/00000658-199201000-00007 1731649 26. Fan HN Zhu P Zhang J Mucosal microbiome dysbiosis associated with duodenum bulb inflammation Microb Pathog 2021 150 104711 10.1016/j.micpath.2020.104711 33385493 27. Glaubitz J, Wilden A, Frost F et al. Activated regulatory T-cells promote duodenal bacterial translocation into necrotic areas in severe acute pancreatitis [published online ahead of print, 2023 Jan 11]. Gut. 2023;gutjnl-2022–327448.
PMC010xxxxxx/PMC10352255.txt
==== Front Eur Heart J Case Rep Eur Heart J Case Rep ehjcr European Heart Journal: Case Reports 2514-2119 Oxford University Press US 10.1093/ehjcr/ytad297 ytad297 Images in Cardiology Valvular Heart Disease AcademicSubjects/MED00200 Prosthetic valve endocarditis: a combination of clinical findings and advanced imaging Storz Matthias Department of Internal Medicine, Stadtspital Zurich Triemli, Birmensdorferstrasse 497, 8063 Zurich, Switzerland Wissmeyer Michael Institute of Radiology and Nuclear Medicine, Stadtspital Zurich, Birmensdorferstrasse 497, 8063 Zurich, Switzerland https://orcid.org/0000-0003-4028-2869 Arrigo Mattia Department of Internal Medicine, Stadtspital Zurich Triemli, Birmensdorferstrasse 497, 8063 Zurich, Switzerland https://orcid.org/0000-0001-5378-4716 Huber Lars C Department of Internal Medicine, Stadtspital Zurich Triemli, Birmensdorferstrasse 497, 8063 Zurich, Switzerland Idris Amr Handling Editor Sharrack Noor Editor Basyal Binaya Editor Corresponding author. Tel: +41 44 416 30 01, Email: [email protected] Conflict of interest: None declared. 7 2023 06 7 2023 06 7 2023 7 7 ytad29721 4 2023 01 6 2023 04 7 2023 18 7 2023 © The Author(s) 2023. Published by Oxford University Press on behalf of the European Society of Cardiology. 2023 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial License (https://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact [email protected] ==== Body pmcCase description A 52-year-old woman presented to our hospital with fever (40.3°C) and shivering. Triple valve replacement surgery [mechanical aortic (Carbomedics 23 mm) and mitral (Carbomedics 27 mm) prostheses, and biological tricuspid prosthesis (Carpentier-Edwards Perimount Magna 27 mm)] was performed more than 20 years earlier because of severe rheumatic heart disease. Physical examination at admission was unremarkable (blood pressure 105/54 mmHg, heart rate 74/min, and SpO2 96%) with no hints for a specific focus. Cardiac auscultation revealed normal prosthesis closing sounds and no murmurs. Inflammation parameters were slightly elevated (leucocytes 11.3 g/L and C-reactive protein 30 mg/L), blood cultures were collected, and empiric antibiotic treatment started. Two days later, the patient developed Osler’s nodes, Janeway’s lesions, and splinter haemorrhages on both hands (Figure 1A). Blood cultures (six of six sets) turned positive at Day 3 for Streptococcus dysgalactiae. Repetitive transoesophageal echocardiography did not show typical signs of PVE. Cardiac 18F-fluorodeoxyglucose (FDG)–PET/CT scan on Day 8 revealed increased uptake in the paravalvular space confirming PVE of the three prostheses (Figure 1B). Figure 1 (A) Osler’s nodes (white arrow), Janeway’s lesions (black arrows), and splinter haemorrhages (arrowheads). (B) Cardiac positron emission tomography/computed tomography scan showing increased 18F-fluorodeoxyglucose uptake in the paravalvular space. Our case highlights the importance of careful skin examination in patients with fever and valve prosthesis. Osler’s nodes (white arrow) and Janeway’s lesions (black arrows) are separate minor Duke criteria since they have been historically considered as distinct immunologic and vascular phenomena, respectively.1 To date, it has been recognized that both entities result from septic emboli causing dermal micro-abscesses. Osler’s nodes are tender when localized on fingers and toes, and Janeway’s lesions involve palms and soles and have haemorrhagic appearance.2 Splinter haemorrhages (arrowheads) are not part of the Duke criteria. The modified Duke criteria are the mainstay of diagnosis for infective endocarditis but their diagnostic accuracy for PVE is limited by frequent inconclusive echocardiographic results, in particular at early stages. Advanced imaging modalities such as FDG–PET/CT have recently gained importance providing excellent diagnostic performance in PVE.3 Abnormal paravalvular FDG uptake is recognized as a major Duke criterion in the European Society of Cardiology guidelines. As illustrated in this case, the diagnostic process in suspected PVE still follows the integration of physical findings, laboratory data, and modern imaging technology. The presence of typical skin findings—despite low prevalence <10%—should increase the suspicion of PVE and act as gatekeeper for more sophisticated investigations such as FDG–PET/CT.   Consent: We confirm that informed written consent for submission and publication of this case report has been obtained from the patient in compliance with the COPE guidelines. Funding: None declared. Data availability No new data were generated or analysed in support of this research. ==== Refs References 1 Alpert JS . Osler’s nodes and Janeway lesions are not the result of small-vessel vasculitis. Am J Med 2013;126 :843–844.24054951 2 Marrie TJ . Osler’s nodes and Janeway lesions. Am J Med 2008;121 :105–106.18261495 3 Lancellotti P , HabibG, OuryC, NchimiA. Positron emission tomography/computed tomography imaging in device infective endocarditis: ready for prime time. Circulation 2015;132 :1076–1080.26276889
PMC010xxxxxx/PMC10352388.txt
==== Front Dig Dis Sci Dig Dis Sci Digestive Diseases and Sciences 0163-2116 1573-2568 Springer US New York 37338617 8001 10.1007/s10620-023-08001-4 Original Article Association of Antibody Responses to Fusobacterium nucleatum and Streptococcus gallolyticus Proteins with Colorectal Adenoma and Colorectal Cancer Genua Flavia 1 Butt Julia 2 Waterboer Tim 2 http://orcid.org/0000-0003-3724-7122 Hughes David J. [email protected] 1 1 grid.7886.1 0000 0001 0768 2743 Cancer Biology and Therapeutics Laboratory, UCD School of Biomolecular and Biomedical Science, UCD Conway Institute, University College Dublin, Dublin, D04 V1W8 Ireland 2 grid.7497.d 0000 0004 0492 0584 Infections and Cancer Epidemiology, German Cancer Research Center (Deutsches Krebsforschungszentrum, DKFZ), 69120 Heidelberg, Germany 20 6 2023 20 6 2023 2023 68 8 33003311 6 3 2023 2 6 2023 © The Author(s) 2023 https://creativecommons.org/licenses/by-nc/4.0/ Open Access This article is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License, which permits any non-commercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc/4.0/. Background Streptococcus gallolyticus subspecies gallolyticus (SGG) and Fusobacterium (F.) nucleatum have been implicated in colorectal carcinogenesis. Here, the association of immune responses to bacterial exposure with advancing stages of colorectal neoplasia was assessed by multiplex serology. Methods Immunoglobulin (Ig) A and G antibody responses to eleven proteins each of F. nucleatum and SGG were measured in plasma of controls (n = 100) and patients with colorectal cancer (CRC, n = 25), advanced adenoma (n = 82), or small polyps (n = 85). Multivariable logistic regression was used to evaluate the association of bacterial sero-positivity with colorectal neoplasia. In a cohort subset with matched data (n = 45), F. nucleatum sero-positivity was correlated with bacterial abundance in both neoplastic and matched normal tissue. Results IgG sero-positivity to Fn1426 of F. nucleatum was associated with an increased CRC risk (OR = 4.84; 95% CI 1.46–16.0), while IgA sero-positivity to any SGG protein or specifically Gallo0272 and Gallo1675 alone was associated with increased advanced adenoma occurrence (OR = 2.02, 95% CI 1.10–3.71; OR = 2.67, 95% CI 1.10–6.46; and OR = 6.17, 95% CI 1.61–23.5, respectively). Only F. nucleatum abundance in the normal mucosa positively correlated with the IgA response to the Fn1426 antigen (Correlation coefficient (r) = 0.38, p < 0.01). Conclusion Antibody responses to SGG and F. nucleatum were associated with occurrence of colorectal adenomas and CRC, respectively. Further studies are needed to clarify the role these microbes or the immune response to their antigens may have in colorectal carcinogenesis stages. Graphical Abstract Supplementary Information The online version contains supplementary material available at 10.1007/s10620-023-08001-4. Keywords Serology Colorectal cancer Colorectal neoplasms Fusobacterium nucleatum Streptococcus gallolyticus http://dx.doi.org/10.13039/501100001590 Health Research Board HRA_PHS/2013/397 Hughes David J. University College DublinOpen Access funding provided by the IReL Consortium issue-copyright-statement© Springer Science+Business Media, LLC, part of Springer Nature 2023 ==== Body pmcIntroduction Colorectal cancer (CRC) is the second leading cause of cancer-related death and the third most commonly diagnosed cancer in the world, accounting for 1.8 million new cases in 2018 [1]. Accumulating evidence suggests that genetic susceptibility, environmental exposure, metabolic dysfunction, microbiome composition, and breakdown of gut barrier integrity play major roles in CRC etiology [2, 3]. Furthermore, chronic inflammation caused by infections may promote CRC development [4]. Experimental data suggest that microbes damage the gut barrier lining and colonocyte DNA and increase proinflammatory cytokines and oxidative factors [5, 6, 7]. F. nucleatum may contribute to CRC development by the invasion of the colonic mucosa, the recruitment of immune cells, and the generation of an oncogenic microenvironment that facilitates the development of colorectal neoplasia [8]. Increased permeability of the gut barrier leads to extensive translocation of microbes into the lamina propria, such as SGG, which may then contribute to CRC development [9]. Antibodies against SGG and F. nucleatum may serve as markers for microbial invasion due to the presence of colorectal adenoma (CRA) and CRC. Prospective and patient cohort studies across several European populations, using the same multiplex serology employed here, observed a significant association between SGG antibody response and CRC development [10, 11, 12], while the latter study also found an association with CRA presence. While equivalent prospective studies in both Europe and North America based on multiplex serology did not show any association between F. nucleatum antibody response and CRC [13, 14], similar studies have not yet been conducted in precancerous lesions. Thus, due to limited knowledge of the immune response to these bacterial antigens in colorectal neoplastic progression, we assessed whether antibody responses to SGG and F. nucleatum proteins are associated with stages of neoplasia development from small polyps to more advanced adenomas and cancers in a patient case–control study conducted within a CRC screening population. Materials and Methods Clinical Characteristics This study included 292 individuals from Ireland who donated blood samples prior to bowel preparation for colonoscopy following a positive result for the immunochemical Fecal Occult Blood Test (FIT) or prior to surgical resection of a colorectal tumor. Patients with CRC (n = 25), advanced adenoma (n = 82), and polyps (n = 85) were diagnosed at the Departments of Gastroenterology and Surgery, the Adelaide & Meath Hospital, in Dublin, Ireland (most of the patients with CRC were recruited through the surgery department). Controls (n = 100) were individuals with no colorectal neoplasia detected upon colonoscopy (‘colonoscopy-negative’ controls). All CRCs were classified according to the tenth revision of the International Classification of Diseases (ICD-10). Advanced adenoma include adenomas with high-grade dysplasia (HGD), adenomas with at least 20% tubular villous or villous features, all adenomas greater than 10 mm, and the presence of three or more adenomas [15, 16]. The clinical data, including age at diagnosis, sex, pTNM (tumor stage, regional lymph node involvement, and distant metastasis) staging, and primary tumor localization were taken from patient medical records (see Table 1 for the summary of the clinical characteristics of our study cohorts). All patients gave informed consent in accordance with the Helsinki Declaration and all patient samples were pseudonymized to protect participant identity. The study was approved by the Ethical Committee of the St. James’s Hospital and Federated Dublin Voluntary Hospitals Joint Research Ethics Committee (Ireland, reference 2007-37-17).Table 1 Clinical characteristics of the studied cohort of patients Advanced adenoma (n = 82)b Controls (n = 100) Polyp (n = 85)a Adenoma (n = 60) HGD (n = 22) CRC (n = 25) Sex n(%)  Female 53 (53) 34 (40) 28 (47) 8 (36) 13 (52)  Male 47 (47) 51 (60) 32 (53) 14 (64) 12 (48) Age (years)  Mean (range) 61(42–75) 62(44–75) 64(50–109) 62(44–84) 66(36–89) F. nucleatum  (n positive/n total) Na Na 22/60 11/22 12/25 Localization  Colon/rectum/Na Na 58/25/2 38/20/2 13/9/0 20/4/1 Staging  T staging n (T0/T1/T2/T3/T4/Tx/Na) Na Na Na Na 1/2/3/11/4/1/3  N staging n (N0/N1/N2/Nx/Na) Na Na Na Na 16/2/3/1/3  M staging n (M0/M1/Mx/Na) Na Na Na Na 6/3/14/2 HGD high-grade dysplasia; CRC colorectal cancer. F. nucleatum Fusobacterium nucleatum aPolyps were generally hyperplastic and less than 2 mm bAdvanced adenoma include adenomas with high‐grade dysplasia (HGD), and/or at least 20% tubular villous or villous features, greater than 10 mm, or the presence of three or more adenomas Sample Collection The blood samples were collected within one day prior to surgery or colonoscopy in 6-mL VACUTAINER® tubes (Cruinn Diagnostics, Dublin, Ireland) with EDTA. Within 4 h of collection, bloods were centrifuged at 2000×g for 10 min to separate the top plasma layer, which was then stored at − 80 °C in cryovials. Disease and matched normal mucosal tissue samples were collected during resection of primary tumor or by biopsy, before treatment, while all adenoma biopsies were obtained at colonoscopy during a pilot CRC screening program as described previously [17]. Multiplex Serology Plasma samples were analyzed for antibody responses against each 11 F. nucleatum and SGG proteins in a final serum sample dilution of 1:100 using multiplex serology in a fluorescent bead-based suspension array, as described previously [18]. Briefly, antigens were recombinantly expressed as Glutathione S-transferase (GST)-tagged fusion proteins and affinity purified on glutathione casein-coupled polystyrene beads (Luminex Corp, Austin, TX, USA) with distinct internal fluorescence [19]. After the pre-incubation step, sera were incubated with the antigen-loaded bead mixture and bound IgG or IgA serum antibodies were labeled separately by biotinylated secondary antibodies (goat anti-Human IgG-Biotin #109-065-098 and goat anti-Human IgA-Biotin #109-065-011, Jackson ImmunoResearch, Westgrove, PA, USA) and a subsequent incubation with Streptavidin-R-Phycoerythrin (MossBio, Pasadena, MD, USA). A Luminex 200 Analyzer (Luminex Corp., Austin, TX, USA) was then used to distinguish the bead sets and their respective antigens as well as to quantify the amount of serum IgG or IgA bound to the antigen. The level of antibody response was given as the median fluorescence intensity (MFI) of at least 100 beads per type measured. Background values against the GST-tag, as well as the bead surface and secondary reagents, were subtracted to generate net MFI values. No gold standard serological assay was available to validate this multiplex serology. Therefore, antigen-specific cut-offs were defined at the approximate inflection point of frequency distribution curves under the assumption that a sudden rise in the distribution of antibody response over the percentile of sera indicates a cut-off for sero-positivity as we previously described [20]. Thus, for our current study data, MFI values were plotted against the percentage of samples that had at least that MFI value and the cut-off was then set where a higher cut-off would not significantly change the sero-positivity rate. The multiplex serology assay included 11 proteins from SGG (strain UCN34) and 11 proteins from F. nucleatum (strain ATCC25586) as previously described [11, 13, 21]. Antigen-specific cut-offs with putative protein function are listed in Supplementary Table S1. DNA Extraction from Colorectal Tissue Biopsies and Quantitative Real-Time Polymerase Chain (qPCR) To address whether the observed antibody responses to F. nucleatum reflect its presence in the colorectal tract rather than from other potential infection sites, we also correlated the immune responses to the bacterium with existing matched data for 45 subjects on the relative abundance of F. nucleatum in colorectal neoplasia tissue and in the respective normal adjacent mucosa. For the DNA extraction, 20–30 mg of tissue were lysed on ice in 400 μL of lysis buffer (50-mmol/L HEPES pH 7.5, 150-mmol/ L NaCl, 5-mmol/L EDTA) and protease inhibitor (Calbiochem, Hampshire, UK), followed by sonication on ice for 3 × 30 s. Lysates were centrifuged at 10,000g for 10 min at 4 °C. DNA was then extracted using the Norgen All-in-One Purification Kit (cat. no. 24210). DNA was quantified using a NanoDrop 2000c Spectrophotometer (Thermo Scientific, Asheville, NC, USA). DNA extractions were stored at − 80 °C. Quantitative real-time polymerase chain reaction (qPCR) to quantify the relative abundance of F. nucleatum in both disease and matched normal tissue from patients with CRA or CRC was performed on the Applied Biosystems 7500 Real-Time PCR System (Thermo Fisher Scientific, Dublin, Ireland). Relative quantification (RQ) of F. nucleatum was calculated by 2−ΔCT, where ΔCT is the difference in the copy number threshold (CT) for the test gene (NusG) and reference gene (human prostaglandin transporter, PGT), as described in [17]. Assessment by qPCR of SGG 16 s rRNA relative abundance in the tissue samples provided too few positives (n = 2) to conduct a similar analysis for SGG. Statistical Analysis We estimated the association of colorectal neoplasia with respective IgA and IgG sero-positivity to individual F. nucleatum and SGG proteins and combined sero-positivity to both bacteria, using conditional logistic regression models to compute odds ratios (ORs) and 95% confidence intervals (95% CI). Combined sero-positivity was arbitrarily defined as being simultaneously sero-positive to any F. nucleatum and any SGG protein for either IgA or IgG. Furthermore, we assessed sero-positivity for at least two proteins from a six marker panel subset of SGG antigens (Gallo0272, Gallo0748, Gallo1675, Gallo2018, Gallo2178, and Gallo2179) that have been previously shown to be more strongly associated with CRC risk compared to positivity toward any one of the eleven SGG proteins [11, 12]. To address whether minor inflammatory-related conditions could act as confounders for observed associations, we conducted a sensitivity analysis restricting the control group to those subjects with “no abnormalities detected after colonoscopy” (NAD, n = 37), including hemorrhoids, mild colitis and diverticulosis, or other minor inflammatory conditions. Analyses were adjusted by age and sex and are presented in the text, except where noted, and in main data tables. The results of the unadjusted analysis are included in the supplementary materials (Supplementary Tables S2 to S13). Point-biserial test was used to evaluate the correlation between F. nucleatum abundance in both colorectal neoplastic and matched normal tissue and antibody response to the bacterium in plasma (in a smaller cohort of patients with available matched data, n = 45). Multiple-testing adjustment was conducted using the False Discovery Rate (FDR). Given that the p-values are derived from a clear hypothesis-driven approach with a small number of comparisons across two bacteria, we base our interpretation on the observed p-values but, to be cautious, also present the q-values for the multiple-testing correction in Supplementary Tables S2 to S13. P- and q-values < 0.05 were considered statistically significant. All statistical analyses were performed with IBM SPSS Statistic for Windows, version 27.0 (SPSS Inc., Chicago, Ill., USA) and Rstudio, version 4.0.0 (RStudio Team (2020). RStudio: Integrated Development for R. RStudio, PBC, Boston, MA URL http://www.rstudio.com/). Results Association of Sero-positivity to F. nucleatum and SGG Proteins with Colorectal Neoplasia Risk IgG sero-positivity to F. nucleatum protein Fn1426 was associated with CRC (OR = 4.84, 95% CI 1.46–16.0, Table 3). The corresponding IgA sero-positivity to this antigen was also associated with CRC in the crude model (OR = 4.75, 95% CI 1.25–17.9, Supplementary Table S2), which did not retain statistical significance in the adjusted model although there was a similar disease risk point estimate (OR = 3.84, 95% CI 0.96–15.3, Table 2). We found no association of F. nucleatum sero-positivity with detection of polyps or advanced adenomas (Tables 2 and 3).Table 2 IgA sero-positivity to individual Fusobacterium nucleatum proteins and association with development of polyps, advanced adenoma, and CRC Secondary Ab Antigen Controls (n = 100) Polyps (n = 85) ORa 95% CI p-value AA (n = 82) ORa 95% CI p-value CRC (n-25) ORa 95% CI p-value n(%) n(%) n(%) n(%) α-IgA Fn0131 Negative 86(86) 78(91) 74(90) 23(92) Positive 14(14) 7(9) 0.48 0.18–1.29 0.148 8(10) 0.67 0.26–1.72 0.409 2(8) 0.50 0.10–2.50 0.403 Fn0253 Negative 94(94) 81(95) 79(96) 24(96) Positive 6(6) 4(5) 0.73 0.19–2.72 0.642 3(4) 0.52 0.12–2.20 0.377 1(4) 0.55 0.06–5.22 0.610 Fn0264 Negative 97(97) 79(93) 75(91) 24(96) Positive 3(3) 6(7) 2.35 0.56–9.85 0.241 7(9) 2.55 0.63–10.3 0.188 1(4) 0.83 0.07–9.63 0.885 Fn0387 Negative 97(97) 82(96) 80(97) 25(100) Positive 3(3) 3(7) 1.13 0.22–5.82 0.880 2(3) 0.69 0.11–4.32 0.698 0(0) Fn1426 Negative 95(95) 78(92) 79(96) 20(80) Positive 5(5) 7(8) 1.34 0.39–4.55 0.632 3(4) 0.64 0.14–2.86 0.647 5(20) 3.84 0.96–15.3 0.057 Fn1449 Negative 89(89) 75(88) 77(93) 23(92) Positive 11(11) 10(12) 1.05 0.41–2.65 0.914 5(7) 0.52 0.17–1.61 0.262 2(8) 0.53 0.09–2.99 0.478 Fn1526 Negative 98(98) 85(100) 81(98) 23(92) Positive 2(2) 0(0) 1(2) 0.56 0.05–6.42 0.648 2(8) 2.54 0.29–21.9 0.395 Fn1817_1 Negative 72(72) 78(91) 72(88) 21(84) Positive 13(13) 7(9) 0.61 0.23–1.62 0.326 10(12) 0.93 0.38–2.27 0.874 4(16) 1.41 0.40–4.96 0.592 Fn1817_2 Negative 90(90) 80(94) 73(89) 22(88) Positive 10(10) 5(6) 0.56 0.18–1.74 0.319 9(11) 1.09 0.40–2.91 0.862 3(12) 1.19 0.28–4.92 0.810 Fn1859 Negative 97(97) 82(96) 75(91) 25(100) Positive 3(3) 3(4) 1.05 0.19–5.54 0.953 7(9) 4.11 0.99–17.0 0.052 0(0) Fn1893 Negative 91(91) 73(86) 76(92) 25(100) Positive 9(9) 12(14) 1.46 0.57–3.72 0.427 6(8) 0.59 0.19–1.79 0.354 0(0) Any proteins Negative Positive 52(52) 48(48) 47(55) 38(45) 0.83 0.46–1.51 0.561 43(52) 39(48) 0.97 0.53–1.77 0.930 14(56) 11(44) 0.67 0.26–1.74 0.418 Fn and SGG Negative Positive 77(77) 23(23) 64(75) 21(25) 1.00 0.50–2.00 0.989 54(66) 28(34) 1.73 0.89–3.37 0.105 18(72) 7(28) 1.06 0.37–3.02 0.900 aLogistic regression analysis adjusted by age and sex Ab antibody; OR odds ratio; CI confidence interval; AA advanced adenoma; CRC colorectal cancer; Fn Fusobacterium nucleatum; SGG Streptococcus gallolyticus subspecies gallolyticus Table 3 IgG sero-positivity to individual Fusobacterium nucleatum proteins and association with development of polyps, advanced adenoma, and CRC Secondary Ab Antigen Controls (n = 100) Polyps (n = 85) ORa 95%CI p-value AA (n = 82) ORa 95% CI p-value CRC (n = 25) ORa 95% CI p-value n(%) n(%) n(%) n(%) Fn0131 Negative 81(81) 71(83) 68(82) 22(88) α-IgG Positive 19(19) 14(17) 0.80 0.36–1.73 0.800 14(18) 0.91 0.42–1.98 0.830 3(12) 0.48 0.12–1.94 0.310 Fn0253 Negative 94(94) 80(94) 76(92) 25(100) Positive 6(6) 5(6) 0.90 0.26–3.11 0.873 6(8) 1.03 0.31–3.42 0.958 0(0) Fn0264 Negative 94(94) 79(93) 78(95) 24(96) Positive 6(6) 6(7) 1.12 0.34–3.69 0.851 4(5) 0.99 0.26–3.74 0.996 1(4) 0.90 0.09–8.30 0.929 Fn0387 Negative 87(87) 80(94) 68(82) 24(96) Positive 13(13) 5(6) 0.42 0.14–1.26 0.125 14(18) 1.47 0.63–3.43 0.372 1(4) 0.35 0.04–2.89 0.330 Fn1426 Negative 91(91) 70(82) 75(91) 18(72) Positive 9(9) 15(18) 1.82 0.73–4.51 0.197 7(9) 0.82 0.27–2.41 0.721 7(28) 4.84 1.46–16.0 0.010 Fn1449 Negative 86(86) 77(90) 66(80) 24(96) Positive 14(14) 8(10) 0.63 0.25–1.60 0.336 16(20) 1.55 0.68–3.50 0.290 1(4) 0.26 0.03–2.13 0.211 Fn1526 Negative 97(97) 78(92) 77(94) 25(100) Positive 3(3) 7(8) 2.53 0.61–10.4 0.196 5(6) 2.17 0.48–9.83 0.312 0(0) Fn1817_1 Negative 91(91) 83(97) 75(91) 21(84) Positive 9(9) 2(3) 0.24 0.05–1.16 0.077 9(9) 1.37 0.50–3.72 0.533 4(16) 2.72 0.70–10.4 0.145 Fn1817_2 Negative 90(90) 72(84) 64(78) 22(88) Positive 10(10) 13(16) 1.15 0.49–2.70 0.739 18(22) 2.20 0.98–4.96 0.055 3(12) 1.07 0.25–4046 0.924 Fn1859 Negative 95(95) 78(91) 77(93) 100(0) Positive 5(5) 7(9) 1.43 0.42–4.88 0.566 5(7) 1.15 0.30–4.28 0.834 0(0) Fn1893 Negative 86(86) 79(93) 69(84) 24(96) Positive 14(14) 6(7) 0.47 0.17–1.29 0.145 13(16) 1.14 0.49–2.64 0.760 1(4) 0.28 0.03–2.27 0.234 Any proteins Negative Positive 40(40) 60(60) 42(49) 43(51) 0.59 0.32–1.10 0.098 26(31) 56(69) 1.47 0.78–2.77 0.228 14(56) 11(44) 0.60 0.24–1.50 0.278 Fn and SGG Negative Positive 47(47) 53(53) 52(61) 33(39) 0.46 0.24–0.87 0.017 43(52) 39(48) 0.80 0.43–1.47 0.476 15(60) 10(40) 0.50 0.19–1.31 0.162 aLogistic regression analysis adjusted by age and sex Ab antibody; OR odds ratio; CI confidence interval; AA advanced adenoma; CRC colorectal cancer; Fn Fusobacterium nucleatum; SGG Streptococcus gallolyticus subspecies gallolyticus For SGG, IgA sero-positivity to Gallo2179 was associated with CRC in the crude model (OR = 3.50, 95% CI 1.17–10.4, Supplementary Table S5). However, this association was not statistically significant in the adjusted model but with little change in the magnitude of the risk estimate (OR = 2.90, 95% CI 0.91–9.19, Table 4). IgA sero-positivity to any of the SGG antigens was associated with advanced adenoma occurrence (OR = 2.02, 95% CI 1.10–3.71, Table 4). IgA sero-positivity to Gallo0272 and Gallo1675 was associated with occurrence of advanced adenoma (OR = 2.67, 95% CI 1.10–6.46 and OR = 6.17, 95% CI 1.61–23.5, respectively, Table 4). Conversely, IgG sero-positivity to Gallo0112A was inversely associated with the presence of polyps (OR = 0.38; 95% CI 0.15–0.92, Table 5). Positivity to two or more proteins of the 6-marker panel of SGG was not associated with colorectal neoplasia (Tables 4 and 5).Table 4 IgA sero-positivity to individual SGG proteins and association with development of polyps, advanced adenoma, and CRC Secondary Ab Antigen Controls (n = 100) Polyps (n = 85) ORa 95%CI p-value AA (n = 82) ORa 95% CI p-value CRC (n-25) ORa 95% CI p-value n(%) n(%) n(%) n(%) Gallo0112A Negative 90(90) 73(86) 72(88) 22(88) α-IgA Positive 10(10) 12(14) 1.33 0.53–3.34 0.538 10(12) 1.20 0.46–3.12 0.703 3(12) 1.37 0.33–5.66 0.658 Gallo0112B Negative 94(94) 81(95) 78(95) 25(100) Positive 6(6) 4(5) 0.92 0.23–3.59 0.905 4(5) 1.07 0.27–4.23 0.913 0(0) Gallo0272 Negative 91(91) 76(89) 65(79) 20(80) Positive 9(9) 9(11) 1.14 0.42–3.08 0.783 17(21) 2.67 1.10–6.46 0.029 5(20) 2.38 0.65–8.31 0.193 Gallo0577 Negative 87(87) 72(85) 73(89) 20(80) Positive 13(13) 13(15) 1.15 0.49–2.68 0.746 9(11) 0.75 0.29–1.90 0.548 5(20) 1.43 0.43–4.75 0.551 Gallo0748 Negative 98(98) 78(92) 76(92) 25(100) Positive 2(2) 7(8) 4.10 0.81–20.7 0.087 6(8) 4.29 0.80–22.9 0.088 0(0) Gallo0933 Negative 89(89) 79(92) 79(95) 24(96) Positive 11(11) 6(8) 0.61 0.21–1.76 0.363 4(5) 0.46 0.14–1.53 0.208 1(4) 0.36 0.04–3.03 0.350 Gallo1570 Negative 91(9) 75(88) 72(87) 23(92) Positive 9(9) 10(12) 1.21 0.46–3.16 0.698 10(13) 1.39 0.53–3.66 0.499 2(8) 0.87 0.17–4.35 0.463 Gallo1675 Negative 96(96) 78(92) 70(85) 24(96) Positive 4(4) 7(8) 2.86 0.68–11.9 0.149 12(15) 6.17 1.61–23.5 0.008 1(4) 1.41 0.13–15.2 0.777 Gallo2018 Negative 97(97) 76(89) 75(91) 22(88) Positive 3(3) 9(11) 1.11 0.41–2.99 0.831 7(9) 2.86 0.68–11.8 0.148 3(12) 1.18 0.28–4.95 0.822 Gallo2178 Negative 91(91) 79(93) 74(90) 24(96) Positive 9(9) 6(7) 0.66 0.22–1.92 0.447 8(10) 0.91 0.33–2.48 0.865 1(4) 0.34 0.04–2.88 0.324 Gallo2179 Negative 90(90) 80(94) 68(82) 18(72) Positive 10(10) 5(6) 0.51 0.16–1.58 0.245 14(18) 1.77 0.72–4.31 0.208 7(28) 2.90 0.91–9.19 0.070 Any proteins Negative Positive 53(53) 47(47) 39(45) 46(55) 1.40 0.77–2.52 0.259 31(38) 51(62) 2.02 1.10–3.71 0.023 11(40) 15(60) 1.56 0.62–3.92 0.342 Fn and SGG Negative Positive 77(77) 23(23) 64(75) 21(25) 1.00 0.50–2.00 0.989 54(66) 28(34) 1.73 0.89–3.37 0.105 18(72) 7(28) 1.06 0.37–3.02 0.900  > 2 of 6-marker panel* Negative Positive 89(89) 11(11) 77(91) 8(9) 0.79 0.29–2.09 0.638 71(83) 14(27) 1.64 0.69–3.91 0.260 23(92) 2(8) 0.72 0.14–3.63 0.699 aLogistic regression analysis adjusted by age and sex Ab antibody; OR odds ratio; CI confidence interval; AA advanced adenoma; CRC colorectal cancer; Fn Fusobacterium nucleatum; SGG Streptococcus gallolyticus subspecies gallolyticus *Includes Gallo0272, Gallo0748, Gallo1675, Gallo2018, Gallo2178, and Gallo2179 Table 5 IgG sero-positivity to individual SGG proteins and association with development of polyps, advanced adenoma, and CRC Secondary Ab Antigen Controls (n = 100) Polyps (n = 85) ORa 95%CI p-value AA (n = 82) ORa 95% CI p-value CRC (n = 25) ORa 95% CI p-value n(%) n(%) n(%) n(%) Gallo0112A Negative 79(79) 77(91) 66(80) 22(88) α-IgG Positive 21(21) 8(9) 0.38 0.15–0.92 0.032 16(20) 0.88 0.41–1.88 0.756 3(12) 0.68 0.18–2.62 0.585 Gallo0112B Negative 91(91) 71(84) 71(86) 23(92) Positive 9(9) 14(16) 1.95 0.79–4.79 0.145 11(14) 1.58 0.61–4.11 0.342 2(8) 0.76 0.14–4.09 0.755 Gallo0272 Negative 88(88) 68(80) 65(79) 24(96) Positive 12(12) 17(20) 1.92 0.84–4.37 0.119 17(21) 1.90 0.83–4.33 0.126 1(4) 0.36 0.04–3.03 0.351 Gallo0577 Negative 79(79) 69(81) 68(83) 17(68) Positive 21(21) 16(19) 0.73 0.34–1.54 0.412 14(17) 0.74 0.34–1.60 0.449 8(32) 1.44 0.51–4.04 0.487 Gallo0748 Negative 87(87) 69(81) 66(80) 21(84) Positive 13(13) 16(19) 1.44 0.63–3.28 0.377 16(20) 1.47 0.64–3.32 0.356 4(16) 0.96 0.26–3.44` 0.950 Gallo0933 Negative 71(71) 56(66) 66(80) 18(72) Positive 29(29) 29(34) 1.20 0.64–2.27 0.560 16(20) 0.63 0.31–1.29 0.213 7(28) 0.96 0.35–2.63 0.941 Gallo1570 Negative 83(83) 73(85) 70(85) 24(96) Positive 17(17) 12(15) 0.75 0.33–1.71 0.507 12(15) 0.70 0.30–1.64 0.422 1(4) 0.20 0.02–1.67 0.140 Gallo1675 Negative 90(90) 72(84) 78(95) 23(92) Positive 10(10) 13(16) 1.54 0.62–3.79 0.344 4(5) 0.46 0.13–1.58 0.221 2(8) 0.78 0.15–3.91 0.765 Gallo2018 Negative 81(82) 70(82) 72(87) 21(84) Positive 18(18) 15(18) 0.92 0.42–2.00 0.835 10(13) 0.61 0.25–1.45 0.266 4(16) 1.05 0.31–3.59 0.928 Gallo2178 Negative 90(90) 79(93) 78(95) 23(92) Positive 10(10) 6(7) 0.67 0.23–1.96 0.470 4(5) 0.33 0.08–1.27 0.108 2(8) 0.61 0.12–3.14 0.559 Gallo2179 Negative 85(85) 67(78) 68(83) 19(76) Positive 15(15) 18(22) 1.32 0.61–2.86 0.476 14(17) 1.18 0.51–2.70 0.688 6(24) 1.74 0.57–5.29 0.325 Any proteins Negative Positive 22(22) 78(78) 21(24) 64(76) 0.75 0.37–1.53 0.442 25(30) 57(70) 0.60 0.30–1.20 0.15 5(20) 20 (80) 0.98 0.31–3.01 0.980 Fn and SGG Negative Positive 47(47) 53(53) 52(61) 33(39) 0.46 0.24–0.87 0.017 43(52) 39(48) 0.80 0.43–1.47 0.476 15(60) 10(40) 0.50 0.19–1.31 0.162  > 2 of 6-marker panel* Negative Positive 83(83) 17(17) 63(74) 22(26) 1.55 0.75–3.21 0.237 67(83) 15(17) 1.05 0.47–2.34 0.904 20(80) 5(20) 1.31 0.40–4.27 0.644 Ab antibody; OR odds ratio; CI confidence interval; AA advanced adenoma; CRC colorectal cancer; Fn: Fusobacterium nucleatum; SGG Streptococcus gallolyticus subspecies gallolyticus *Includes Gallo0272, Gallo0748, Gallo1675, Gallo2018, Gallo2178, and Gallo2179 aLogistic regression analysis adjusted by age and sex Finally, dual positivity to both F. nucleatum and SGG was not associated with any of the colorectal neoplasia stages assessed, except for IgG dual positivity to these microbes and a decreased risk of polyp development (OR = 0.46, 95% CI 0.24–0.87, Tables 3 and 5). Sensitivity Analysis Based on the Control Group A few differences were observed for the analyses conducted in the ‘NAD’ control group compared to the full control group. Firstly, IgG sero-positivity to Fn0387 was inversely associated with the occurrence of polyps (OR = 0.23, 95% CI 0.06–0.79, Supplementary Table S10). Secondly, we did not find any statistically significant association between IgA sero-positivity to Gallo1675 and advanced adenomas (Supplementary Table S12). Lastly, we no longer observed the inverse association between IgG dual positivity to F. nucleatum and SGG with polyps (Supplementary Tables S10 and S13). We confirmed all the other findings for the analysis conducted in the full control group (Supplementary Tables S8, S11, S12, and S13). Correlation Between F. nucleatum Tissue Levels and the Antibody Response in Plasma The relative abundance of F. nucleatum in CRC and CRA disease tissue and the respective normal adjacent mucosa, as previously ascertained by qPCR for 45 patients, showed little correlation with the immune responses. The only exception was a significant positive correlation between F. nucleatum abundance in the normal mucosa and levels of the IgA antibody response to the Fn1426 antigen (Correlation coefficient (r) = 0.38, p < 0.01, Supplementary Table S14). Discussion In this study we attempted to detect mucosal and systemic antibody responses to SGG and F. nucleatum by separate detection of IgA and IgG. We observed that sero-prevalence to some of the bacterial antigens varied significantly between cases with colorectal neoplasia and control groups and that some of these differences were associated with colorectal neoplasia across the major developmental stages from polyps to tumors. It remains to be clarified whether these microbes infect healthy colon tissue prior to the carcinogenic process or if they only infect a developing neoplasia due to bacterial translocation across the impaired gut barrier. In both hypotheses, the bacteria may induce mucosal and systemic antibody responses and potentially cause pro-carcinogenic effects, as previously hypothesized for genotoxic Escherichia coli species and Enterotoxigenic Bacteroides fragilis (ETBF) [20]. Regarding F. nucleatum, IgG and IgA sero-posivitiy to Fn1426 were associated with CRC, although caution is strongly advised in interpreting the large point estimates (4.84 and 4.75, respectively) due to the wide CIs and modest samples numbers for CRC. However, IgA sero-positivity to Fn1426 did not retain significance upon adjustment. Two large prospective studies—one from the European Prospective Investigation into Cancer and Nutrition (EPIC) and another from a cohort consortium in the USA suggested that pre-diagnostic antibody responses to F. nucleatum proteins were not associated with CRC risk [13, 14]. Conversely, a study conducted by Wu et al. showed that patients with CRC infected with F. nucleatum produced higher levels of IgA and IgG than the control groups [22]. Taken together with the results presented here, these studies support the hypothesis that F. nucleatum might act as a “passenger bacterium” increasing in abundance due to favorable growth conditions with dysplastic progression. A similar conclusion on a non-pathogenic reverse causality rationale for the presence of this bacteria was reached by a large German CRC screening study—conducted in stool samples and based on 16S rRNA analysis—where F. nucleatum abundance was strongly associated with CRC but not with adenomas (either advanced or or non-advanced) [23]. However, it remains possible from these observations that the bacterium is involved in the cancer transition from advanced adenomas without having a causative role at the early stage of carcinogenesis. Concerning SGG, IgA sero-positivity to Gallo2179 was associated with a 3.5-fold odds of CRC development compared to controls, although the statistical significance was not retained upon adjustment. A similar result was found in a Spanish case–control study, based on the same methodology employed here, where the antibody response to Gallo2179, alone and in combination with Gallo2178, was significantly associated with CRC risk [24]. Gallo2178 and Gallo2179 are pilus proteins that are assumed to be virulence factors in SGG-induced infective endocarditis but also in invasion of CRC tissue [9, 25]. The pattern of antibody responses to SGG proteins with precancerous lesions differed. Firstly, positivity for the IgA responses to any of the SGG antigens or separately to Gallo0272 and Gallo1675 were associated with the presence of advanced adenoma. Finally, IgA sero-positivity to Gallo0112A was negatively associated with the presence of polyps. So far, functions of Gallo0272, Gallo1675, and Gallo2018 were only predicted by amino acid similarities to proteins of other bacterial species: Gallo0272 is homologous to one of the agglutinin receptors in the oral bacterium Streptococcus gordonii, which mediates binding to host cell and bacterial receptors representing an important virulence factor [26]. Gallo1675 is a cell wall protein with unknown function [27], while Gallo2018 is putatively involved in bacteriocin synthesis [27]. Several studies have indicated the presence of SGG already in early colorectal lesions, including polyps and adenoma, although antibody responses to SGG were assessed by ELISA and Western blot, respectively [28, 29]. Both studies suggest that SGG may drive the transition from the normal epithelium to the early stages of colorectal neoplasia and thus, CRC. Our results align with these findings as we observed that sero-positivity to SGG proteins were more associated with colorectal adenomas rather than CRC, although, this may reflect the lower power with the smaller numbers of cancers analyzed. The findings that antibody responses to SGG appear before cancer diagnosis or in precancerous lesions suggests that SGG infection may be a potential etiological factor in the transition of a polyp to malignant disease and its detection could help to identify precursors that may more likely progress to cancer [24]. Dual positivity to both F. nucleatum and SGG was not associated with any of the colorectal neoplasia stages assessed except for an IgG dual positivity to these microbes and a decreased risk of polyp development. However, as there were limited group numbers dual positive to these bacterial species (33/85 patients with polyps, 39/85 with advanced adenoma and 10/25 with CRC), larger patient cohort studies are needed to confirm or refute a protective role of the combined sero-positivity to both bacteria in the early stages of carcinogenesis. We did not find any association between positivity to two or more proteins of the six markers SGG panel and colorectal neoplasia, in contrast with the findings in a German case–control study, where positivity to at least two proteins from this panel was significantly associated with a 1.81- and 2.98-fold higher risk of CRC and non-advanced adenoma, respectively [12]. However, this result may be due to the small number of positives for at least two proteins of the panel for both IgA (CRC = 2 and polyps = 8) and IgG antibody responses. In our sensitivity analysis, there were two notable deviations from the adjusted analyses using the full control group. These observations, while limited due to small number of NAD subjects (n = 37), warrants consideration of the control group characteristics in future studies, where a strength here was employing colonoscopy findings to dichotomise the controls into subjects that presented with minor inflammatory-related conditions and those with no pathologies. Since serology is an indirect systemic of past and/or current infection, it is unknown whether the observed antibody responses result from other infection sites than the colorectum. To partially address this, we correlated the relative quantification of F. nucleatum, ascertained by qPCR of its DNA in disease and matched normal mucosa tissues of 45 study participants, with levels of the IgA and IgG response to F. nucleatum. We observed that levels of IgA antibody response to Fn1426 were positively correlated with F. nucleatum abundance in matched normal mucosa tissue of patients with CRA and CRC. Sero-positivity to this antigen was associated with occurrence of CRC in this study, which indicates that the observed association might truly result from an infection in the gut with F. nucleatum. Our study has several notable strengths. Firstly, we analyzed the antibody response to F. nucleatum and SGG in different stages of colorectal neoplasia within  a similar demographic cohort and we also conducted an exploratory analysis to assess whether the nature of the control groups may act as a potential confounder. Another strength was the application of multiplex serology to detect different Ig classes and to assess a more detailed immune response to the bacterium beyond just overall sero-positivity. As numerous antigens and two Ig classes were assessed, correcting our findings for multiple comparisons may conceivably be warranted. While we acknowledge that the application of FDR correction would remove significance of our results, this may be considered over-stringent due to the correlated nature of the tested antigens. A main limitation in this study is the modest sample size across the neoplasia groups, especially for the CRC cases (n = 25). While it was an added strength for this study to have matched immune responses and measured colorectal disease and surrounding tissue abundances of F. nucleatum, this novel analysis was limited by the modest number of matched samples and the lack of measurements of SGG abundance in the tissue. Future prospective studies, specifically detecting F. nucleatum and SGG in stool or tissue biopsies in larger cohorts within different study settings, are needed to help clarify whether the antibody response originates from infections in the colorectal tissue or other sites of the body. Reverse causality is a highly possible factor underlying our results particularly considering a lengthy immune response during the generally long period of neoplastic development. Finally, residual confounding cannot be ruled out, as we could only control for age and sex among relevant covariates (as there were no data on, for example, BMI, smoking, diet, or antibiotic use). Conclusion In this study we found that sero-positivity to certain SGG and F. nucleatum proteins were associated with the presence of advanced stages of colorectal neoplasia, including CRC. Thus, the evaluation of antibody response to bacteria may be a useful resource to identify individuals at increased risk for developing CRC or to detect the presence of CRC at the early stages. These findings need to be validated in other settings with increased samples sizes to assess F. nucleatum and SGG serology as a potential biomarker of the immune response to bacterial agents in the developing colorectal neoplasia. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (DOCX 222 KB) We thank Dr Niall Swan for the pathology designation of the colorectal neoplasia samples (Departments of Gastroenterology and Surgery, The Adelaide & Meath Hospital, Dublin, Ireland). Author’s contribution Conceptualization: DJH and TW. Samples collection: DJH. Experiments: JB, TW, and FG. Data analysis: FG, JB, and DJH. Funding acquisition: DJH. Writing of the original draft: FG and DJH. Writing, reviewing, and editing of the manuscript: JB and TW. Funding Open Access funding provided by the IReL Consortium. This work was funded by the Health Research Board of Ireland, award  HRA-POR-2013-397 to DJH, and a PhD Research Scholarship award to FG from the School of Biomolecular and Biomedical Science, UCD. Support for this work was also provided by the COST Action CA17118 supported by COST (European Cooperation in Science and Technology, www.cost.eu) to FG and DJH. Data availability The datasets generated during and/or analyzed during the current study are not publicly available due to patients’ data confidentiality but are available from the corresponding author on reasonable request. Declarations Competing interest The authors have no relevant financial or non-financial interests to disclose. Ethical approval The study was approved by the Ethical committee of the St. James’s Hospital and Federated Dublin Voluntary Hospitals Joint Research Ethics Committee (Ireland, reference 2007-37-17). Consent to participate All patients gave informed consent in accord with the 1964 Helsinki Declaration and all patient samples were coded to protect participant identity. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Bray F Ferlay J Soerjomataram I Siegel RL Torre LA Jemal A Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries CA Cancer J Clin 2018 68 394 424 10.3322/caac.21492 30207593 2. Murphy N Moreno V Hughes DJ Vodicka L Vodicka P Aglago EK Lifestyle and dietary environmental factors in colorectal cancer susceptibility Mol Aspects Med 2019 69 2 9 10.1016/j.mam.2019.06.005 31233770 3. Genua F Raghunathan V Jenab M Gallagher WM Hughes DJ The role of gut barrier dysfunction and microbiome dysbiosis in colorectal cancer development Front Oncol 2021 10.3389/FONC.2021.626349 33937029 4. Schmitt M Greten FR The inflammatory pathogenesis of colorectal cancer Nat Rev Immunol 2021 21 653 667 10.1038/s41577-021-00534-x 33911231 5. Tilg H Adolph TE Gerner RR Moschen AR The intestinal microbiota in colorectal cancer Cancer Cell 2018 33 954 964 10.1016/j.ccell.2018.03.004 29657127 6. Kostic AD Chun E Robertson L Glickman JN Gallini CA Michaud M Fusobacterium nucleatum potentiates intestinal tumorigenesis and modulates the tumor-immune microenvironment Cell Host Microbe 2013 14 207 215 10.1016/j.chom.2013.07.007 23954159 7. Jans C Boleij A The road to infection: host-microbe interactions defining the pathogenicity of Streptococcus bovis/Streptococcus equinus complex members Front Microbiol 2018 10.3389/fmicb.2018.00603 29692760 8. Rubinstein MR Wang X Liu W Hao Y Cai G Han YW Fusobacterium nucleatum promotes colorectal carcinogenesis by modulating E-Cadherin/β-catenin signaling via its FadA adhesin Cell Host Microbe 2013 14 195 206 10.1016/j.chom.2013.07.012 23954158 9. Boleij A Muytjens CMJ Bukhari SI Cayet N Glaser P Hermans PWM Novel clues on the specific association of Streptococcus gallolyticus subsp gallolyticus with colorectal cancer J Infect Dis 2011 203 1101 1109 10.1093/INFDIS/JIQ169 21451000 10. Butt J Blot WJ Teras LR Visvanathan K Le Marchand L Haiman CA Antibody responses to Streptococcus Gallolyticus subspecies Gallolyticus proteins in a large prospective colorectal cancer cohort consortium Cancer Epidemiol Biomarkers Prev 2018 27 1186 1194 10.1158/1055-9965.EPI-18-0249 30038049 11. Butt J Jenab M Willhauck-Fleckenstein M Michel A Pawlita M Kyrø C Prospective evaluation of antibody response to Streptococcus gallolyticus and risk of colorectal cancer Int J Cancer 2018 143 245 252 10.1002/ijc.31283 29377173 12. Butt J Werner S Willhauck-Fleckenstein M Michel A Waterboer T Zörnig I Serology of Streptococcus gallolyticus subspecies gallolyticus and its association with colorectal cancer and precursors Int J Cancer 2017 141 897 904 10.1002/IJC.30765 28477334 13. Butt J Jenab M Pawlita M Overvad K Tjonneland A Olsen A Antibody responses to Fusobacterium nucleatum proteins in prediagnostic blood samples are not associated with risk of developing colorectal cancer Cancer Epidemiol Prev Biomarkers 2019 28 1552 1555 10.1158/1055-9965.EPI-19-0313 14. Lo C-H, Blot WJ, Teras LR, Visvanathan K, Marchand L, Haiman CA, et al. Prediagnostic Antibody Responses to Fusobacterium nucleatum Proteins Are Not Associated with Risk of Colorectal Cancer in a Large U.S. Consortium. Anne Zel n.d.;15:19. 10.1158/1055-9965.EPI-20-1471. 15. Kim DH Pickhardt PJ Taylor AJ Characteristics of advanced adenomas detected at CT colonographic screening: implications for appropriate polyp size thresholds for polypectomy versus surveillance AJR Am J Roentgenol 2007 188 940 944 10.2214/AJR.06.0764 17377027 16. Corley DA Jensen CD Marks AR Zhao WK Lee JK Doubeni CA Adenoma detection rate and risk of colorectal cancer and death N Engl J Med 2014 370 1298 1306 10.1056/NEJMOA1309086 24693890 17. Flanagan L Schmid J Ebert M Soucek P Kunicka T Liska V Fusobacterium nucleatum associates with stages of colorectal neoplasia development, colorectal cancer and disease outcome Eur J Clin Microbiol Infect Dis 2014 33 1381 1390 10.1007/s10096-014-2081-3 24599709 18. Waterboer T Sehr P Michael KM Franceschi S Nieland JD Joos TO Multiplex human papillomavirus serology based on in situ-purified glutathione s-transferase fusion proteins Clin Chem 2005 51 1845 1853 10.1373/CLINCHEM.2005.052381 16099939 19. Sehr P Zumbach K Pawlita M A generic capture ELISA for recombinant proteins fused to glutathione S-transferase: validation for HPV serology J Immunol Methods 2001 253 153 162 10.1016/S0022-1759(01)00376-3 11384677 20. Butt J Jenab M Werner J Fedirko V Weiderpass E Dahm CC Association of pre-diagnostic antibody responses to Escherichia coli and Bacteroides fragilis toxin proteins with colorectal cancer in a european cohort Gut Microbes 2021 13 1 14 10.1080/19490976.2021.1903825 33874856 21. Butt J Varga MG Blot WJ Teras L Visvanathan K Le Marchand L Serologic response to helicobacter pylori proteins associated with risk of colorectal cancer among diverse populations in the United States Gastroenterology 2019 156 175 186.e2 10.1053/j.gastro.2018.09.054 30296434 22. Wang HF Li LF Guo SH Zeng QY Ning F Liu WL Evaluation of antibody level against Fusobacterium nucleatum in the serological diagnosis of colorectal cancer Sci Rep 2016 10.1038/SREP33440 28442790 23. Amitay EL Werner S Vital M Pieper DH Höfler D Gierse IJ Fusobacterium and colorectal cancer: causal factor or passenger? Results from a large colorectal cancer screening study Carcinogenesis 2017 38 781 788 10.1093/carcin/bgx053 28582482 24. Butt J Romero-Hernández B Pérez-Gómez B Willhauck-Fleckenstein M Holzinger D Martin V Association of Streptococcus gallolyticus subspecies gallolyticus with colorectal cancer: serological evidence Int J Cancer 2016 138 1670 1679 10.1002/IJC.29914 26537841 25. Danne C Entenza JM Mallet A Briandet R Débarbouillé M Nato F Molecular characterization of a Streptococcus gallolyticus genomic island encoding a pilus involved in endocarditis J Infect Dis 2011 204 1960 1970 10.1093/INFDIS/JIR666 22043018 26. Demuth DR Duan Y Brooks W Holmes AR McNab R Jenkinson HF Tandem genes encode cell-surface polypeptides SspA and SspB which mediate adhesion of the oral bacterium Streptococcus gordonii to human and bacterial receptors Undefined 1996 20 403 413 10.1111/J.1365-2958.1996.TB02627.X 27. Hinse D Vollmer T Rückert C Blom J Kalinowski J Knabbe C Complete genome and comparative analysis of Streptococcus gallolyticus subsp. gallolyticus, an emerging pathogen of infective endocarditis Undefined 2011 10.1186/1471-2164-12-400 28. Abdulamir AS Hafidh RR Mahdi LK Al-jeboori T Abubaker F Investigation into the controversial association of Streptococcus gallolyticus with colorectal cancer and adenoma BMC Cancer 2009 10.1186/1471-2407-9-403 19925668 29. Garza-González E Ríos M Bosques-Padilla FJ Francois F Cho I González GM Immune response against Streptococcus gallolyticus in patients with adenomatous polyps in colon Int J Cancer 2012 131 2294 2299 10.1002/ijc.27511 22377818
PMC010xxxxxx/PMC10352424.txt
==== Front Dig Dis Sci Dig Dis Sci Digestive Diseases and Sciences 0163-2116 1573-2568 Springer US New York 37029308 7921 10.1007/s10620-023-07921-5 Original Article Cluster-Analytic Identification of Clinically Meaningful Subtypes in MCAS: The Relevance of Heat and Cold Häder Tinus [email protected] 12 Molderings Gerhard J. [email protected] 3 Klawonn Frank [email protected] 45 Conrad Rupert [email protected] 6 Mücke Martin [email protected] 17 http://orcid.org/0000-0001-5354-2872 Sellin Julia [email protected] 17 1 grid.412301.5 0000 0000 8653 1507 Institute for Digitalization and General Medicine, University Hospital RWTH Aachen, Aachen, Germany 2 grid.15090.3d 0000 0000 8786 803X Center for Rare Diseases Bonn (ZSEB), University Hospital Bonn, Bonn, Germany 3 grid.15090.3d 0000 0000 8786 803X Institute for Human Genetics, University Hospital Bonn, Bonn, Germany 4 grid.7490.a 0000 0001 2238 295X Biostatistics Group, Helmholtz Centre for Infection Research, Braunschweig, Germany 5 Department of Computer Science, Ostfalia University, Wolfenbuettel, Germany 6 grid.16149.3b 0000 0004 0551 4246 Department of Psychosomatic Medicine and Psychotherapy, University Hospital Muenster, Muenster, Germany 7 grid.412301.5 0000 0000 8653 1507 Center for Rare Diseases Aachen (ZSEA), University Hospital RWTH Aachen, Aachen, Germany 8 4 2023 8 4 2023 2023 68 8 34003412 15 7 2022 10 3 2023 © The Author(s) 2023 https://creativecommons.org/licenses/by-nc/4.0/ Open AccessThis article is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License, which permits any non-commercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc/4.0/. Background Mast cell activation syndrome (MCAS) is a clinically heterogeneous disease with allergy-like symptoms and abdominal complaints. Its etiology is only partially understood and it is often overlooked. Aims The aim of this study was to identify subgroups of MCAS patients to facilitate diagnosis and allow a personalized therapy. Methods Based on data from 250 MCAS patients, hierarchical and two-step cluster analyses as well as association analyses were performed. The data used included data from a MCAS checklist asking about symptoms and triggers and a set of diagnostically relevant laboratory parameters. Results Using a two-step cluster analysis, MCAS patients could be divided into three clusters. Physical trigger factors were particularly decisive for the classification as they showed remarkable differences between the three clusters. Cluster 1, labeled high responders, showed high values for the triggers heat and cold, whereas cluster 2, labeled intermediate responders, presented with high values for the trigger heat and low values for cold. The third cluster, labeled low responders, did not react to thermal triggers. The first two clusters showed more divers clinical symptoms especially with regard to dermatological and cardiological complaints. Subsequent association analyses revealed relationships between triggers and clinical complaints: Abdominal discomfort is mainly triggered by histamine consumption, dermatological discomfort by exercise, and neurological symptoms are related to physical exertion and periods of starvation. The reasons for the occurrence of cardiological complaints are manifold and triggers for respiratory complaints still need better identification. Conclusion Our study identified three distinct clusters on the basis of physical triggers, which also differ significantly in their clinical symptoms. A trigger-related classification can be helpful in clinical practice for diagnosis and therapy. Longitudinal studies should be conducted to further understand the relationship between triggers and symptoms. Supplementary Information The online version contains supplementary material available at 10.1007/s10620-023-07921-5. Keywords Mast cell activation syndrome MCAS Cluster analysis Systemic mastocytosis SM Food intolerance Histamine Multisystemic complaints IBS RWTH Aachen University (3131)Open Access funding enabled and organized by Projekt DEAL. issue-copyright-statement© Springer Science+Business Media, LLC, part of Springer Nature 2023 ==== Body pmcIntroduction Mast cell activation syndrome (MCAS) is characterized by allergy-like symptoms as a result of pathologically increased activation of mast cells and is sometimes associated with accumulation of mast cells of abnormal morphology [1]. MCAS-associated symptoms are caused by downstream effects of mast cell mediators, which mast cells release upon activation and which normally play a role in the defense against harmful substances or microorganisms, such as bacteria, parasites, and animal toxins [2, 3]. One important mast cell mediator is histamine, which increases permeability and dilatation of vessels, resulting in swelling and redness [4]. Mast cells are most known for their involvement in type I allergy, in which IgE binds tightly to the mast cell receptor FcεRI and, therefore, induces mast cell activation upon antigen contact [5]. MCAS is clinically indistinguishable from systemic mastocytosis (SM), a rare genetic disease typically caused by mutations in the KIT gene coding for the KIT tyrosine-protein kinase. Both diseases are characterized by increased mast cell activation and accumulation of mast cells in a variety of tissues [6]. The symptoms of MCAS and SM patients are very diverse and include abdominal, neurological, cardiological, respiratory, and dermatological complaints [7–11]. The severity of symptoms ranges from mild over severe to life-threatening [12, 13]. Characteristically, at the beginning of the manifestation of the disease, symptoms occur episodically, and a progression of symptoms is observed during the course of the disease [14]. For a lot of patients, abdominal complaints are the major problem in their disease. These complaints have many similarities with those in irritable bowel syndrome (IBS). IBS is classically characterized by chronic diffuse abdominal pain, constipation, and diarrhea. Studies have shown that refractory IBS patients present with symptoms typical of MCAS and have elevated mast cell-specific laboratory parameters indicative of mast cell activation [15]. It was also found that IBS patients on ketotifen therapy, an anti-histamine, had a significant reduction in symptoms [16]. Based on these similarities, it appears likely that there is an overlap of the two disorders, highlighting the importance of clear diagnostic criteria and more research about etiology and pathology of both disorders. MCAS is in real life severely underdiagnosed despite an estimated prevalence of up to 17% [17], with negative consequences for affected patients and the healthcare system. In an effort toward the definition of clear diagnostic criteria for MCAS, two main approaches (termed Consensus 1 and Consensus 2) have emerged [6, 12, 18–24]. The most recent versions of Consensus 1 [20] and 2 [24] assign different criteria as a basis for the diagnosis of MCAS (Table 1). As has been discussed recently [6], the criteria for diagnosis of Consensus 1 are slightly stricter than those of Consensus 2, whereas, at this point in time, it is debatable which approach is more accurate.Table 1 Overview of the MCAS diagnostic criteria according to Consensus 1 and Consensus 2 as discussed in [6] Consensus 1 [20] Consensus 2 [24] 1. Typical clinical symptoms of severe systemic (i.e., involving at least 2 organ systems) mast cell activation include urticaria, flushing, pruritus, angioedema, nasal congestion, nasal pruritus, wheezing, throat swelling, hoarseness, headache, hypotensive syncope, tachycardia, abdominal cramping diarrhea, and anaphylaxis 2. Involvement of mast cells is documented by biochemical changes; preferred marker: increase in serum tryptase level from the individual’s baseline to plus 20% + 2 ng/mL; other mast cell-derived markers of mast cell activation (histamine and histamine metabolites, PGD2 metabolites, and heparin) have also been proposed, but are less specific than tryptase 3. Response of symptoms to therapy with mast cell-stabilizing agents, drugs directed against mast cell mediator production, or drugs blocking mediator release or effects of mast cell-derived mediators Major Criterion Constellation of clinical complaints attributable to pathologically increased mast cell activity (mast cell mediator release syndrome) Minor Criteria 1. Genetic alterations of different MC genes shown to increase mast cell activity 2. Evidence (typically from body fluids such as whole blood, serum, plasma, or urine) of above-normal levels of mast cell mediators including • tryptase • histamine or its metabolites (e.g., N-methylhistamine) • heparin • chromogranin A (note potential confounders of cardiac or renal failure, neuroendocrine tumors, recent proton pump inhibitor use, or chronic atrophic gastritis) • other relatively mast cell-specific mediators (e.g., eicosanoids including prostaglandin (PG) D2, its metabolite 11-β-PGF2α, or leukotriene E4) 3. Symptomatic response to inhibitors of mast cell activation or mast cell mediator production or action 4. Multifocal order-disseminated infiltrates of mast cells in marrow and/or extracutaneous organ(s) (e.g., gastrointestinal or genitourinary tract; > 19 mast cells/high power field) 5. Abnormal spindle-shaped morphology in > 25% of mast cells in marrow other extracutaneous organs 6. Abnormal mast cell expression of CD2 and/or CD25 (i.e., co-expression of CD117/CD25 or CD117/CD2) Diagnosis made by demonstrating all three of the above-noted criteria; the diagnosed individual should then be assessed for primary, secondary, or idiopathic MCAS as outlined previously Diagnosis established upon demonstration of the major criterion combined with at least one minor criterion (and in the unstated but inferred absence of any other disease better accounting for the patient’s problems) Regardless of stricter or more lenient criteria, both approaches have the goal to improve MCAS diagnosis. Many MCAS patients have not yet been identified as such, are unaware of their disease, and therefore do not benefit from mast cell-specific therapy or even receive inappropriate therapies. Various medications are used to treat MCAS. These include anti-histamines [25], the mast cell stabilizer cromoglycic acid [26], and slow-release vitamin C [27], which all reduce mast cell activity and can reduce symptoms. In cases of treatment failure, immunosuppressants or omalizumab [28] can be tried [1]. A facultative symptomatic treatment is necessary to break the vicious cycle of continuously mutually activating mast cells. Prerequisite for a response to every drug therapy is the avoidance of mast cell trigger factors [10, 12]. As mentioned above, MCAS is a disorder with variable clinical presentation [6]. Accordingly, symptoms, triggers, and laboratory values vary widely among patients. Against this background and the estimated high prevalence [14], it is evident that a more precise characterization of specific subgroups could facilitate diagnosis and therapy. Cluster analysis is an explorative statistical procedure in which subgroups, i.e., clusters, are formed on the basis of various characteristics. Cluster analyses have already proven to be beneficial in the characterization of other clinical diseases, such as fibromyalgia [29] and irritable bowel syndrome [30]. We therefore conducted a cluster analysis of MCAS patients based on symptoms and triggers, with the aim of simplifying the classification of patients on the basis of their medical history and open the road toward a more personalized therapy. Methods Participants Data of 250 MCAS patients were included in the study either from the specialized private practice of one of the authors (MM) or from the Center for Rare Diseases at Bonn University Hospital, between January 2019 and June 2020. Requirements for inclusion in the study were an age of at least 18 years and a proven mast cell activation syndrome according to consensus 2 criteria [6]. Patients completed the checklist as part of the diagnostic procedure. In this process, a conspicuous score in the checklist represented one criterion for the diagnosis of MCAS. Only patients in whom the diagnosis was subsequently confirmed by the minor criteria abnormal biopsy result and/or abnormalities in mast cell-specific laboratory parameters were included in this study. Instruments and Collected Data The previously published checklist [14, 21] for the detection of mast cell mediator release syndrome [14] served as the data basis. In this checklist, patients were asked in a binary fashion for various abdominal, neurological, cardiological, dermatological, and respiratory symptoms as well as certain factors leading to elicitation or exacerbation of symptoms [14]. The initial trigger of the complaints could be named in a free text column. In addition, data on age, sex, weight, and height were collected. Mast cell-specific laboratory parameters such as tryptase in blood and N-methylhistamine in urine had been assessed for more than 180 patients by MVZ Labor Quade, Cologne, Germany. Chromogranin A, neuron-specific enolase, and immunoglobulin E levels had also been determined for the vast majority of MCAS patients [31–34]. Statistical Analysis Cluster Analysis Two-step [35] and hierarchical [36] cluster analyses were performed using SPSS version 27.0 for macOS from IBM SPSS, Chicago IL, USA. Factors that trigger discomfort or lead to worsening of discomfort turned out to lead to separable clusters and were therefore considered for cluster analysis. These included physical exertion, heat, cold, stress, alcohol consumption, sleep deprivation, periods of starvation, and consumption of histamine-containing foods. Bayesian information criterion (BIC) was used to determine the optimal number of clusters. Cluster solutions were compared with the silhouette measure for cohesion and separation. In the second step, the three identified clusters were compared in terms of baseline data collected, symptoms, and laboratory values using cross-tabulations. Significant differences in the subgroups were detected using the Chi-square test for categorical and the Kruskal–Wallis test for continuous variables. p-values of ≤ 0.01 were considered significant. Clinical Associations Possible associations between sex, triggers, symptoms, and laboratory values were calculated. For two binary variables, the p-value was calculated by Fisher’s exact test. When comparing a binary variable with a numerical one, the p-value for the Wilcoxon–Mann–Whitney test was calculated. In this study, due to a wide variety of questions and the exploratory nature of the study, many tests for significance were performed. It should therefore be noted that with a significance level of p ≤ 0.01, there is an average of 1 × erroneous rejection of the null hypothesis in 100 tests performed. Results Description of the Collective In the MCAS patient cohort, about 86% were female. The average patient age was 45 years (range 18–84 years). The mean weight of the patients was within the normal range with a BMI of 23.2 kg/m2 (range 12.4–46.2 kg/m2). Triggers of symptoms varied widely, with stress and consumption of histamine-containing foods reported most frequently (Fig. 1A). Heat was named as a trigger factor more often than cold. The initial trigger could be identified by almost half of the patients and was mainly related to stress, surgery or hospitalization, and infections (Fig. 1B).Fig. 1 A The prevalence of the triggers in the study cohort is given as a percentage. B The prevalence of initial triggers is given as a percentage. C Result of two-step cluster analysis. A value of 1 means that 100 percent of the patients in the cluster reported this trigger, whereas a value of 0 indicates that none of the patients reported it. D Prevalence of the symptoms in the three clusters given as a percentage. p-values were calculated with chi-square test The symptoms reported by the MCAS patients were very variable and affected different organ systems, including various abdominal, neurological, respiratory, cardiological, and dermatological symptoms. Neurological complaints, in particular word-finding disorder and weakness/exhaustion, affected almost every patient in the collective (Fig. 2A). On the other hand, other symptoms, in particular dermatological complaints, affected smaller subgroups (Fig. 2B–E). In the majority of the collective, symptoms occurred episodically and many observed a shortening of symptom-free intervals (Fig. 2F).Fig. 2 The respective prevalence of Abdominal (A), Respiratory (B), Neurological (C), Cardiological (D), and Dermatological (E) are given as percentage. In F, the prevalence are given for progressive course of the complaints Laboratory chemistry showed an increase of the mean and median value only for N-methylhistamine. Tryptase, immunoglobulin E, chromogranin A, and neuron-specific enolase showed abnormal values only in a few patients (Fig. 4A–E). Two-Step Cluster Analysis Worked Better than Hierarchical Cluster Analysis To identify clusters, hierarchical cluster analyses were performed first. Although the resulting clusters showed already great similarity with the final results, they were not completely stable with regard to cluster size and values for the respective triggers in response to altered data sorting. Therefore, a two-step cluster analysis for binary data was conducted, since it is typically more robust with regard to sorting dependency [37]. However, using the two-step cluster analysis and taking all factors into account also shows a slightly varying result with different sorting of the data. For this reason, a predictor importance table supplied with the two-step cluster analysis was generated. This showed with descending influence physical exertion, heat, and cold as the three most important provoking factors (Fig. 3A). The other triggers showed a significant decrease in importance and therefore were not chosen as cluster variables, but were presented via cross-tabulations. Setting physical exertion, heat, and cold as variables and log-likelihood as a measure of quality, a robust clustering result was shown, which remained unchanged with different sorting of the data.Fig. 3 A Predictor importance table created with SPSS two-step cluster analysis. The formation of the clusters should be limited to the most important factors [47]. In this example, these could be clearly identified as physical exertion, heat, and cold. B Chart created with SPSS two-step cluster analysis, BIC values against number of clusters. The largest change in BIC values marks the optimal number of clusters. The largest change in BIC values was seen in the step from a 1 cluster to a 2 cluster and in the change from a 2-cluster to a 3-cluster solution. C Silhouette measure of cohesion and separation for a 2-cluster solution, the larger the silhouette measure for cohesion and separation becomes, the better the cluster result. D Silhouette measure of cohesion and separation for a 3-cluster solution Bayesian information criterion (BIC) was used to determine the optimal number of clusters. The largest changes in the values were seen when jumping from a 1- to a 2-cluster solution and from a 2- to a 3-cluster solution. At 4 clusters and beyond, the curve flattened and the BIC values changed only slightly (Fig. 3B). The decision for a 3-cluster solution was made due to the better silhouette measure for cohesion and separation compared to the 2-cluster solution (Fig. 3C–D). In addition, the 3-cluster solution provided the best content interpretability (not shown). The results showed that MCAS patients could be classified into three groups based on the factors that provoke symptoms or cause an exacerbation of existing symptoms (Fig. 1C). These factors include physical exertion, heat, cold, stress, alcohol consumption, insomnia, periods of starvation, and histamine consumption. Cluster 1 Responded to All Triggers The largest cluster with a total of 105 patients showed high scores on all triggers (Fig. 1C). Discomfort was particularly triggered by physical exertion, heat, and cold; the three triggers were reported by all patients of cluster 1. Less important was sleep deprivation and alcohol consumption, although these triggers were still mentioned by at least 70 percent of the patients of cluster 1 (Fig. 1C). Cluster 2 Did Not React to Cold In the smallest cluster (54 patients), cold was not a trigger of discomfort (Fig. 1C). Main triggers were instead physical exertion and heat which were reported by all patients of the cluster (Fig. 1C). The response to the other provoking factors was similar to that of the third cluster, but slightly more pronounced. Alcohol consumption and starvation periods were the least mentioned provoking factors (Fig. 1C). Cluster 3 Showed Little Response to Physical Factors The third cluster, consisting of 91 patients, showed the least response to the mentioned triggers (Fig. 1C). In contrast to the first two clusters, physical exertion and temperature extremes were not significant factors (Fig. 1C). Except for stress, the other triggers mentioned above were reported less by this group of patients than by those in clusters 2 and especially 1 (Fig. 1C). Clinically, Clusters 1 and 2 Were More Severely Affected than Cluster 3 In the second step, the prevalence of the symptoms in the three clusters was compared (Fig. 1D). Patients from clusters 1 and 2 reported abdominal, cardiac, dermatologic, respiratory, and neurologic symptoms significantly more frequently than those of the third cluster (Fig. 1D). This was most pronounced for cardiological and dermatological symptoms. Increases in blood pressure, for example, were reported approximately twice as often by the first two clusters than by the third, and acne-like skin changes also occurred more than twice as frequently in the first two clusters. The first and the second cluster, on the other hand, did not differ significantly with regard to reported symptoms (Fig. 1D). Differences between these two patient groups (however, not reaching the level of significance) may exist with regard to neurological complaints and may reach statistical significance in future studies with larger patient numbers. Laboratory Chemistry Results Between the Clusters Laboratory chemistry revealed no evidence for significant differences between the three clusters with respect to the laboratory parameters immunoglobulin E, N-methylhistamine, tryptase, neuron-specific enolase, and chromogranin A (results not shown). Association Analysis of Symptoms and Triggers As our analysis revealed clusters that differed with respect to symptom triggers, we next asked if different triggers are typically associated with specific symptoms or if certain symptoms typically occur together. Indeed, closer inspection showed that abdominal symptoms were primarily triggered by histamine consumption. In addition, they were associated with episodic symptom onset and symptom progression during the course of the disease (Table 2).Table 2 Associations between symptoms and triggers Symptom section Most commonly associated triggers Associations Abdominal complaints Histamine consumption Episodic course of symptoms; symptom progression Dermatological complaints Physical exertion Respiratory symptoms (especially runny nose and irritable cough) Cardiological complaints Manifold Hot flushes; weakness and fatigue Respiratory complaints No clear association Mostly neurological symptoms Neurological complaints Physical exertion; starvation periods Many associations (especially cardiological and respiratory symptoms) For respiratory symptoms, no trigger showed a clear significant association with multiple respiratory symptoms. However, respiratory complaints co-occurred with symptoms in several other systems of the body, particularly neurological symptoms, like fatigue attacks, headache, and word-finding disorders (Table 2). Dermatological complaints were most often associated with the trigger physical exertion. In particular, the occurrence of flushing, itching, and acne-like skin lesions showed a significant relationship to physical activity. Skin and mucous membrane complaints occurred together with respiratory complaints, such as runny nose and irritable cough (Table 2). The triggers of the cardiological complaints were diverse. These symptoms were associated with the occurrence of physical weakness and exhaustibility as well as fatigue attacks (Table 2). Neurological complaints were mainly triggered by physical exertion and periods of hunger. Cold led to symptoms of weakness and fatigue as well as headaches. Heat played only a minor role with regard to this symptom group. Neurological complaints showed many associations with other symptoms, especially cardiac and respiratory symptoms (Table 2). There was a significant association between gender and renal N-methylhistamine excretion. Women had significantly higher mean and median values than men. In both sexes, median and mean values were above the reference value of 6.5 μg/mmol/Cr/m2 Body Surface Area (Fig. 4F).Fig. 4 A to E Boxplots for the respective laboratory parameters were in the reference range. F Boxplots of NMH created separately for men and women. The difference in urine NMH concentration between men and women was statistically significant as indicated by the p-value calculated with Wilcoxon–Mann–Whitney Test. Laboratory chemistry revealed normal mean values and medians for tryptase, chromogranin A, neuron-specific enolase, and immunoglobulin E (A–E). N-methylhistamine (NMH) in urine was elevated (average of 9.7 μg/mmol/Cr/m2 Body Surface Area versus the reference value of < 6.5). 75% of MCAS patients had median NMH levels above the reference value (F). The median NMH excretion was significantly higher in women than in men (F) All calculated p-values of associations between symptoms and triggers can be found in Supplemental Table 1. Discussion Cluster Analysis The central result of the present study is the possibility to divide MCAS patients into clinical subgroups. Taken together, the MCAS patient collective presents with a uniform reaction to stress, consumption of histamine, alcohol consumption, insomnia, and periods of hunger. The crucial difference appears in the evaluation of heat and cold as trigger factors, which resulted in the emergence of three clusters:Cluster 1 can be considered as the group of high responders; complaints were caused by many triggers, including temperature changes in either direction. The latter were reported by all patients of the cluster. Cluster 2 comprises the intermediate responders, who stated heat as a trigger but not cold, in contrast to the first cluster. Cluster 3 covers the low responders, who were particularly notable for their low response to the main triggers of the first two clusters. Classification of patients into three clusters allows the division of MCAS patients on the basis of their medical history. Physical triggers such as heat and cold are well known from previous studies [38, 39]. Similar findings can be observed in SM patients [40]. From a methodological perspective, a cluster analysis on the basis of trigger factors has not yet been applied to MCAS patients. Our study confirms the usefulness of this approach in MCAS patients as it allows for the subtyping of patients based on the anamnesis regarding the two physical triggers heat and cold, which is not only clinically meaningful but also can serve as an easily accessible and therefore an economic way to gain an estimate of the patient’s needs. Patients who report heat and/or cold will usually have more clinical symptoms (Fig. 1D) and require closer medical care than patients who do not report either trigger. Clusters 1 and 2 typically show more symptoms overall, but especially many dermatological and cardiological symptoms, to which special attention should therefore be paid. The commonly used laboratory parameters immunoglobulin E, N-methylhistamine, tryptase, neuron-specific enolase, and chromogranin A did not show any significant differences in the three clusters and therefore cannot serve as indicators for one of the clusters. This is in accordance with previous reports showing that laboratory values are highly diverse among patients and do not correlate with diseases severity or symptoms ([6] and references therein). Our results highlight the relevance of trigger factors and underline once more that patients should be advised to observe and subsequently avoid their specific triggers, in addition to drug therapy [1]. While three distinct clusters of patients emerged from our analysis, it should however be noted that individual complaints vary greatly among patients overall. This type of heterogeneity among patients poses particular challenges not only to diagnosis, but also therapy. The relatively new field of precision medicine attempts to extract information, often from large amounts of data, aiming at providing personalized therapies based on the precise understanding of individual or stratified differences among patients, in order to be able to deliver to each patient the best type of therapy at the optimal time point and dosage to maximize efficiency [41]. At the same time, it enables the identification of specific biomarkers for identification of the optimal therapy for an individual patient. Our study could be seen as a first step toward tackling this heterogeneity in MCAS, although much more data combined with machine or deep learning strategies would be needed to identify precise differences between patients. For such an approach, several omics data would ideally be needed, e.g., patient genomes, single-cell omics of mast cells, (gut) microbiomics, as well as longitudinal, high-resolution data on diet, laboratory parameters and symptoms, and potentially even health monitoring data from wearable or mobile sensors. Knowledge about, e.g., the exact changes in mast cell genetics or signaling could have direct implications on the best symptomatic therapy, like, for example, the choice between mast cell stabilizers, anti-histamines, anti-IgE antibodies, or suppression of mast cell development [42] or even suggest new types of interventions, including diets to optimize gut microbiota for MCAS patients. Indeed, a specialized sub-area of precision medicine is precision nutrition, which tries to optimize individualized nutritional advice based on large amounts of data and machine learning or deep learning approaches [43], which could prove particularly beneficial for MCAS patients in the future, as many suffer from food intolerances and are often looking for the right diet to manage their symptoms. Association Analysis With the help of association analyses, previously unknown relationships between triggers and symptoms could be established. To be more specific, this means for clinical practice that patients with certain complaints can be given recommendations for action to improve their quality of life. This can be illustrated by specific examples: A patient with neurological complaints could be informed about his symptoms being most likely associated with starvation periods and physical exertion. If abdominal complaints occur, a progression of complaints and an episodic course of symptoms can be expected. It should be noted that this method is rarely used, but offers great potential. In psychiatric research, association analyses and networks are already used to work out which symptoms are central to a disorder and how strongly they are related [44, 45]. With this in mind, it also seems possible that inferences can be made on a mechanistic basis. Thus, certain symptoms could occur together because of the same activation pathway or mediator being responsible. This also requires a better link between clinical and basic research to connect clinical findings with research results at the cellular level [46]. Limitations The applied symptom checklist currently has a binary scale level. A more differentiated assessment that gives estimates of symptom severity may facilitate the understanding of differences in clinical symptomatology. Furthermore, patients’ symptoms and triggers may change over time, which cannot be assessed by our study design. Therefore, it might be more appropriate to see clusters as disease stages, which evolve over time. Longitudinal, circadian, and environmental changes are also potential confounders for the interpretation of laboratory parameters, which were determined at only one time point as part of the diagnostic process. In follow-up studies, longitudinal data, including the laboratory parameters, should therefore be collected under controlled conditions (e.g., all patients on a specific diet) and at specific time points during the study, which would furthermore allow the inclusion of more objective data to complement subjective patient-reported experience measures, like the symptom checklist. Conclusion This is the first cluster analysis performed in MCAS patients and thus the first approach to dissect the genetically highly heterogeneous disease into less clinically heterogeneous subgroups. Clustering can have an immediate impact on daily clinical practice, as different clusters of a disease could have different needs with regard to therapy and supportive care. In this case, our results show that clinicians should especially take note of the triggers reported by patients, as they point toward different symptom loads that should be expected, which could prompt them to adjust symptomatic therapy accordingly. This is important in order to avoid exacerbation of symptoms by a vicious cycle of constantly mutually activating hyperactive mast cells. The information could be helpful to identify at-risk patients more easily. The association analyses are new in the sense that they establish connections between triggers and symptoms that have not been described before. This allows an even more differentiated approach to the patient. Depending on which complaints the patient describes, the trigger can be named, a statement can be made about accompanying complaints, and recommendations about measures to reduce complaints can be given. In summary, our study confirms the utility of a cluster analytic approach and the potential of association analysis to improve the understanding of MCAS and to personalize the therapy. To validate the results of our study, prospective longitudinal studies should be performed in future. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (XLSX 21 kb). In the first column and the top line, the gender, laboratory values, and all triggers and symptoms queried are listed in each case. The respective p-values from Fisher’s exact test for two binary variables, and Wilcoxon–Mann–Whitey test for a binary and a numerical one is shown. p-values ≤ 0.05 are not shown individually and are indicated with NA. Author's contribution Conceptualization: TH, RC, MM, and JS. Methodology: TH, FK, RC, and JS. Formal analysis: TH, FK, RC, and JS. Investigation: TH. Resources: TH and MM. Writing—original draft preparation: TH, GJM, RC, MM, and JS. Writing, reviewing, and editing of the manuscript: TH, GJM, RC, MM, and JS. Visualization: TH, FK, and JS. Supervision: FK, RC, MM, and JS. Project administration: RC, MM, and JS. Funding Open Access funding enabled and organized by Projekt DEAL. There was no funding for this study. Data Availability The dataset analyzed and discussed is available from the corresponding author upon request. Declarations Competing interests The authors declare that they have no financial or non-financial competing interests. Ethical approval The study was approved by the Ethics Committee of the Rheinische Friedrich-Wilhelms-Universität Bonn, Medical Faculty, on April 30th to May 2022. Consent for publication Not applicable. An editorial commenting on this article is available at 10.1007/s10620-023-07923-3. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Rupert Conrad, Martin Mücke, and Julia Sellin are senior authors. ==== Refs References 1. Molderings GJ Haenisch B Brettner S Homann J Menzen M Dumoulin FL Pharmacological treatment options for mast cell activation disease Naunyn Schmiedebergs Arch Pharmacol. 2016 389 671 694 10.1007/s00210-016-1247-1 27132234 2. Piliponsky AM Romani L The contribution of mast cells to bacterial and fungal infection immunity Immunol Rev. 2018 282 188 197 10.1111/imr.12623 29431211 3. Kim HS Kawakami Y Kasakura K Kawakami T Recent advances in mast cell activation and regulation F1000Res 2020 9 F1000 10.12688/f1000research.22037.1 32226609 4. da Silva EZM Jamur MC Oliver C Mast cell function: a new vision of an old cell J Histochem Cytochem. 2014 62 698 738 10.1369/0022155414545334 25062998 5. Galli SJ Tsai M IgE and mast cells in allergic disease Nat Med. 2012 18 693 704 10.1038/nm.2755 22561833 6. Afrin LB Ackerley MB Bluestein LS Brewer JH Brook JB Buchanan AD Diagnosis of mast cell activation syndrome: a global “consensus-2” Diagnosis (Berl). 2020 8 137 152 10.1515/dx-2020-0005 32324159 7. Afrin LB Pöhlau D Raithel M Haenisch B Dumoulin FL Homann J Mast cell activation disease: an underappreciated cause of neurologic and psychiatric symptoms and diseases Brain Behav Immun. 2015 50 314 321 10.1016/j.bbi.2015.07.002 26162709 8. Kolck UW Haenisch B Molderings GJ Cardiovascular symptoms in patients with systemic mast cell activation disease Transl Res. 2016 174 23 32.e1 10.1016/j.trsl.2015.12.012 26775802 9. Weinstock LB Pace LA Rezaie A Afrin LB Molderings GJ Mast cell activation syndrome: a primer for the gastroenterologist Dig Dis Sci. 2020 66 965 982 10.1007/s10620-020-06264-9 32328892 10. Picard M Giavina-Bianchi P Mezzano V Castells M Expanding spectrum of mast cell activation disorders: monoclonal and idiopathic mast cell activation syndromes Clin Ther. 2013 35 548 562 10.1016/j.clinthera.2013.04.001 23642289 11. Qureshi AA Friedman AJ A review of the dermatologic symptoms of idiopathic mast cell activation syndrome J Drugs Dermatol. 2019 18 162 168 30794367 12. Molderings GJ Brettner S Homann J Afrin LB Mast cell activation disease: a concise practical guide for diagnostic workup and therapeutic options J Hematol Oncol. 2011 4 10 10.1186/1756-8722-4-10 21418662 13. Bonadonna P Bonifacio M Lombardo C Zanotti R Hymenoptera allergy and mast cell activation syndromes Curr Allergy Asthma Rep. 2015 16 5 10.1007/s11882-015-0582-5 14. Molderings G Kolck U Scheurlen C Brüss M Frieling T Raithel M Die systemische Mastzellerkrankung mit gastrointestinal betonter Symptomatik - eine Checkliste als Diagnoseinstrument Dtsch med Wochenschr. 2006 131 2095 2100 10.1055/s-2006-951337 16981082 15. Frieling T Meis K Kolck U Homann J Hülsdonk A Haars U Evidence for mast cell activation in patients with therapy-resistant irritable bowel syndrome Z Gastroenterol. 2011 49 191 194 10.1055/s-0029-1245707 21298604 16. Klooker TK Braak B Koopman KE Welting O Wouters MM van der Heide S The mast cell stabiliser ketotifen decreases visceral hypersensitivity and improves intestinal symptoms in patients with irritable bowel syndrome Gut. 2010 59 1213 1221 10.1136/gut.2010.213108 20650926 17. Molderings GJ Haenisch B Bogdanow M Fimmers R Nöthen MM Familial occurrence of systemic mast cell activation disease PLoS ONE. 2013 8 e76241 10.1371/journal.pone.0076241 24098785 18. Valent P Akin C Doctor, I think I am suffering from MCAS: Differential diagnosis and separating facts from fiction J Allergy Clin Immunol Pract. 2019 7 1109 1114 10.1016/j.jaip.2018.11.045 30961836 19. Valent P Akin C Arock M Brockow K Butterfield JH Carter MC Definitions, criteria and global classification of mast cell disorders with special reference to mast cell activation syndromes: a consensus proposal Int Arch Allergy Immunol. 2012 157 215 225 10.1159/000328760 22041891 20. Valent P Akin C Bonadonna P Hartmann K Brockow K Niedoszytko M Proposed diagnostic algorithm for patients with suspected mast cell activation syndrome (MCAS) J Allergy Clin Immunol Pract. 2019 7 1125 1133.e1 10.1016/j.jaip.2019.01.006 30737190 21. Afrin LB Molderings GJ A concise, practical guide to diagnostic assessment for mast cell activation disease WJH. 2014 3 1 10.5315/wjh.v3.i1 22. Afrin LB The presentation, diagnosis and treatment of mast cell activation syndrome : review article Curr Allergy Clin Immunol. 2014 27 146 160 23. Afrin LB Self S Menk J Lazarchick J Characterization of mast cell activation syndrome Am J Med Sci. 2017 353 207 215 10.1016/j.amjms.2016.12.013 28262205 24. Molderings GJ Zienkiewicz T Homann J Menzen M Afrin LB Risk of solid cancer in patients with mast cell activation syndrome: results from Germany and USA F1000Res. 2017 6 1889 10.12688/f1000research.12730.1 29225779 25. Nurmatov UB Rhatigan E Simons FER Sheikh A H1-antihistamines for primary mast cell activation syndromes: a systematic review Allergy. 2015 70 1052 1061 10.1111/all.12672 26095756 26. Advenier C Ruff F [Cromoglycic acid (disodium cromoglycate) and inhibitors of mast cell degranulation] Sem Hop 1984 60 659 664 6324375 27. Hagel AF Layritz CM Hagel WH Hagel HJ Hagel E Dauth W Intravenous infusion of ascorbic acid decreases serum histamine concentrations in patients with allergic and non-allergic diseases Naunyn Schmiedebergs Arch Pharmacol. 2013 386 789 793 10.1007/s00210-013-0880-1 23666445 28. Kibsgaard L Skjold T Deleuran M Vestergaard C Omalizumab induced remission of idiopathic anaphylaxis in a patient suffering from indolent systemic mastocytosis Acta Derm Venereol. 2014 94 363 364 10.2340/00015555-1687 24162824 29. de Souza JB Goffaux P Julien N Potvin S Charest J Marchand S Fibromyalgia subgroups: profiling distinct subgroups using the Fibromyalgia Impact Questionnaire. A preliminary study Rheumatol Int. 2009 29 509 515 10.1007/s00296-008-0722-5 18820930 30. Windgassen S Moss-Morris R Goldsmith K Chalder T The importance of cluster analysis for enhancing clinical practice: an example from irritable bowel syndrome J Ment Health. 2018 27 94 96 10.1080/09638237.2018.1437615 29447026 31. Schwartz LB Metcalfe DD Miller JS Earl H Sullivan T Tryptase levels as an indicator of mast-cell activation in systemic anaphylaxis and mastocytosis N Engl J Med. 1987 316 1622 1626 10.1056/NEJM198706253162603 3295549 32. Valent P Bonadonna P Hartmann K Broesby-Olsen S Brockow K Butterfield JH Why the 20% + 2 tryptase formula is a diagnostic gold standard for severe systemic mast cell activation and mast cell activation syndrome Int Arch Allergy Immunol. 2019 180 44 51 10.1159/000501079 31256161 33. Payne V Kam PCA Mast cell tryptase: a review of its physiology and clinical significance Anaesthesia. 2004 59 695 703 10.1111/j.1365-2044.2004.03757.x 15200544 34. Valent P Horny HP Triggiani M Arock M Clinical and laboratory parameters of mast cell activation as basis for the formulation of diagnostic criteria Int Arch Allergy Immunol. 2011 156 119 127 10.1159/000323763 21576982 35. Chiu T, Fang D, Chen J, Wang Y, Jeris C. A robust and scalable clustering algorithm for mixed type attributes in large database environment. In: Proceedings of the seventh ACM SIGKDD international conference on Knowledge discovery and data mining [Internet]. New York, NY, USA: Association for Computing Machinery; 2001 [cited 2022 Apr 9]. p. 263–8. (KDD ’01). Available from: 10.1145/502512.502549 36. Li T. A general model for clustering binary data. In: Proceedings of the eleventh ACM SIGKDD international conference on Knowledge discovery in data mining [Internet]. New York, NY, USA: Association for Computing Machinery; 2005 [cited 2022 Apr 20]. p. 188–97. (KDD ’05). Available from: 10.1145/1081870.1081894 37. Clustering Binary Data (should be avoided) [Internet]. 2020 [cited 2021 Oct 12]. Available from: https://www.ibm.com/support/pages/clustering-binary-data-should-be-avoided 38. Hamilton MJ Hornick JL Akin C Castells MC Greenberger NJ Mast cell activation syndrome: a newly recognized disorder with systemic clinical manifestations J Allergy Clin Immunol. 2011 128 147 152.e2 10.1016/j.jaci.2011.04.037 21621255 39. Theoharides TC Tsilioni I Ren H Recent advances in our understanding of mast cell activation – or should it be mast cell mediator disorders? Expert Rev Clin Immunol. 2019 15 639 656 10.1080/1744666X.2019.1596800 30884251 40. Castells M Austen KF Mastocytosis: mediator-related signs and symptoms Int Arch Allergy Immunol. 2002 127 147 152 10.1159/000048188 11919427 41. Hartl D de Luca V Kostikova A Laramie J Kennedy S Ferrero E Translational precision medicine: an industry perspective J Transl Med. 2021 19 245 10.1186/s12967-021-02910-6 34090480 42. Valent P Akin C Hartmann K Nilsson G Reiter A Hermine O Mast cells as a unique hematopoietic lineage and cell system: from Paul Ehrlich’s visions to precision medicine concepts Theranostics. 2020 10 10743 10768 10.7150/thno.46719 32929378 43. Kirk D Catal C Tekinerdogan B Precision nutrition: a systematic literature review Comput Biol Med. 2021 133 104365 10.1016/j.compbiomed.2021.104365 33866251 44. Airaksinen J Gluschkoff K Kivimäki M Jokela M Connectivity of depression symptoms before and after diagnosis of a chronic disease: a network analysis in the U.S. Health and Retirement Study J Affect Disord. 2020 266 230 234 10.1016/j.jad.2020.01.170 32056882 45. Mullarkey MC Marchetti I Beevers CG Using network analysis to identify central symptoms of adolescent depression J Clin Child Adolesc Psychol. 2019 48 656 668 10.1080/15374416.2018.1437735 29533089 46. Kahn K Ryan G Beckett M Taylor S Berrebi C Cho M Bridging the gap between basic science and clinical practice: a role for community clinicians Implement Sci. 2011 6 34 10.1186/1748-5908-6-34 21463516 47. IBM Corporation. IBM: SPSS Modeler/ Predictor Importance [Internet]. Predictor Importance. 2021 [cited 2022 Mar 30]. Available from: https://prod.ibmdocs-production-dal-6099123ce774e592a519d7c33db8265e-0000.us-south.containers.appdomain.cloud/docs/en/spss-modeler/18.1.0?topic=SS3RA7_18.1.0/modeler_mainhelp_client_ddita/clementine/idh_common_predictor_importance.html
PMC010xxxxxx/PMC10352442.txt
==== Front Dig Dis Sci Dig Dis Sci Digestive Diseases and Sciences 0163-2116 1573-2568 Springer US New York 37338618 8005 10.1007/s10620-023-08005-0 Original Article sST2 Levels Show No Association with Helicobacter pylori Infection in Asymptomatic Patients: Implications for Biomarker Research Wernly Sarah 1 Paar Vera 2 Völkerer Andreas 1 Semmler Georg 3 Datz Christian 1 Lichtenauer Michael 2 http://orcid.org/0000-0003-4024-0220 Wernly Bernhard [email protected] 14 1 grid.21604.31 0000 0004 0523 5263 Department of Internal Medicine, General Hospital Oberndorf, Teaching Hospital of the Paracelsus Medical University, Salzburg, Austria 2 grid.21604.31 0000 0004 0523 5263 Clinic II for Internal Medicine, University Hospital Salzburg, Paracelsus Medical University, Salzburg, Austria 3 grid.22937.3d 0000 0000 9259 8492 Division of Gastroenterology and Hepatology, Department of Medicine III, Medical University of Vienna, Vienna, Austria 4 grid.21604.31 0000 0004 0523 5263 Institute of General Practice, Family Medicine and Preventive Medicine, Paracelsus Medical University, Salzburg, Austria 20 6 2023 20 6 2023 2023 68 8 32933299 28 3 2023 13 6 2023 © The Author(s) 2023 https://creativecommons.org/licenses/by-nc/4.0/ Open Access This article is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License, which permits any non-commercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc/4.0/. Introduction Helicobacter pylori (H. pylori) is a prevalent stomach bacterium that can cause a range of clinical outcomes, including gastric cancer. In recent years, soluble suppression of tumorigenicity-2 (sST2) has gained attention as a biomarker associated with various diseases, such as gastric cancer. The purpose of this study was to explore the possible connection between H. pylori infection and sST2 levels in patients who do not exhibit symptoms. Methods A total of 694 patients from the Salzburg Colon Cancer Prevention Initiative (Sakkopi) were included in the study. The prevalence of H. pylori infection was determined by histology, and sST2 levels were measured in serum samples. Clinical and laboratory parameters, such as age, sex, BMI, smoking status, hypertension, and metabolic syndrome, were also collected. Results The median sST2 concentration was similar between patients with (9.62; 7.18–13.44 ng/mL; p = 0.66) and without (9.67; 7.08–13.06 ng/mL) H. pylori. Logistic regression analysis did not show any association (OR 1.00; 95%CI 0.97–1.04; p = 0.93) between sST2 levels and H. pylori infection, which remained so (aOR 0.99; 95%CI 0.95–1.03; p = 0.60) after adjustment for age, sex, educational status, and metabolic syndrome. In addition, sensitivity analyses stratified by age, sex, BMI, smoking status, educational status, and the concomitant diagnosis of metabolic syndrome could not show any association between sST2 levels and H. pylori infection. Conclusion The results indicate that sST2 may not serve as a valuable biomarker in the diagnosis and treatment of H. pylori infection. Our findings are of relevance for further research investigating sST2, as we could not find an influence of asymptomatic H. pylori infection on sST2 concentration. What Is Already Known? Soluble suppression of tumorigenicity-2 (sST2) has gained attention as a biomarker associated with various diseases, such as gastric cancer. What Is New in This Study? The median sST2 concentration was similar between patients with (9.62; 7.18–13.44 ng/mL; p = 0.66) and without (9.67; 7.08–13.06 ng/mL) H. pylori. What Are the Future Clinical and Research Implications of the Study Findings? The results indicate that sST2 may not serve as a valuable biomarker in the diagnosis and treatment of H. pylori infection. Keywords Biomarker H. pylori Cancer sST2 Soluble suppression of tumorigenicity-2 Paracelsus Medical UniversityOpen access funding provided by Paracelsus Medical University. issue-copyright-statement© Springer Science+Business Media, LLC, part of Springer Nature 2023 ==== Body pmcIntroduction Helicobacter pylori (H. pylori) is a gram-negative bacterium that colonizes the human stomach, with an estimated prevalence of 50% worldwide [1–3]. Although H.pylori infection is usually asymptomatic, it can cause a range of clinical manifestations, including gastritis, peptic ulcer disease, gastric adenocarcinoma, and mucosa-associated lymphoid tissue lymphoma [1, 2]. The impact of H.pylori on individual and public health is significant, as it is estimated to be responsible for 90% of duodenal ulcers, 80% of gastric ulcers, and 60% of gastric cancers [4]. Further, H.pylori infection is associated with extraintestinal systemic diseases such as immune thrombocytopenia[5] and iron deficiency anemia of unknown origin [6]. Therefore, H. pylori infection should be treated once diagnosed [4]. Soluble suppression of tumorigenicity-2 (sST2) is a biomarker that has been extensively studied in recent years [7–9]. It is a member of the interleukin-1 receptor family [10] and is secreted in response to various forms of cellular stress, including inflammation, tissue injury, and mechanical strain [11]. sST2 was proposed as a marker for heart failure [12, 13], but emerging evidence suggests that it may also play a role in autoimmune diseases [14], infectious diseases [9], and cancer [15]. Of note, sST2 has been shown to be elevated in patients with gastric cancer [16]. Further, IL-33 and ST2 have been shown to be involved in exacerbated H. pylori infection in vitro [17]. Our study is predicated on the hypothesis that sST2 levels may be altered in asymptomatic patients with H. pylori infection, which holds significant implications for the diagnosis and management of H. pylori infection. While H.pylori infection can be diagnosed through invasive procedures like endoscopy and biopsy, less invasive methods such as stool antigen testing, breath tests, or serology are also available, albeit with potentially lower accuracy [2]. As a non-invasive biomarker, sST2 could offer a promising addition in combination to current diagnostic methods. Moreover, if H. pylori infection does influence sST2 levels, this may have implications for other disease contexts, such as autoimmune diseases and heart failure, where sST2 has been implicated. To the best of our knowledge, this is the first study to investigate the potential association between H. pylori infection and sST2 levels in asymptomatic patients. We aim to compare sST2 levels in patients with and without H. pylori infection, while assessing the relationship between sST2 and other clinical and laboratory parameters. It is worth noting that both sST2 and H. pylori infection have been linked to metabolic risk factors like obesity, insulin resistance, and metabolic syndrome, which may also play a role in our findings [18]. Another objective was to investigate whether there is a correlation between H. pylori infection and changes in sST2 levels, while also examining whether this correlation remains significant regardless of the presence of common metabolic risk factors. In summary, this paper investigates the potential association between H. pylori infection and sST2 levels in asymptomatic patients. H. pylori infection is a significant public health issue, and sST2 is a biomarker that has been proposed to play a role in various disease settings. Our study aims to shed light on the potential relationship between these two factors and may have implications for the diagnosis and management of H. pylori infection as well as other disease settings. Methods Subjects Our study recruited participants from the Salzburg Colon Cancer Prevention Initiative (Sakkopi), a cohort of asymptomatic patients screened for colorectal cancer at a single center in Austria who were offered a screening esophagogastroduodenoscopy (EGD). We assessed all patients (n = 1047) enrolled between 2018 and 2020. In 718 patients we had serum samples available for sST2 measurement. We further excluded 24 patients who refused EGD. Ultimately, our analysis consisted of 694 patients who had both serum samples available for sST2 measurement as well as undergone EGD with biopsy. All enrolled subjects had clinical and laboratory parameters collected as part of the study [19, 20]. In addition, patients were asked to complete a medical history questionnaire. We assessed and calculated various parameters, including body mass index (BMI), smoking status, arterial hypertension, and metabolic syndrome, based on current guidelines [21]. We assessed the presence of H. pylori through histology, using biopsies collected during EGD. We performed the study and all procedures according to the principles of the Declaration of Helsinki. The local ethics committee for the province Salzburg approved the study protocol (Approval No. 415-E/1262). Written informed consent was obtained from every participant. Measurement of sST2 Serum levels of sST2 were measured using a commercially available enzyme-linked immunosorbent assay (ELISA) kit (Duoset DY206; R&D Systems, USA). The assay was performed according to the manufacturer’s protocol. Serum samples and standard protein were added to the wells on a 96-well plate (Maxisorp plate, Nunc-Immuno, Sigma Aldrich, Austria) and incubated for 2 h. After this incubation period, 96-well plates were washed for three times with washing buffer containing Tween-20. Then, a biotin-labelled secondary antibody was added to the wells and plates were incubated for further 2 h. Later, plates were washed again and Streptavidin-horseradish-peroxidase was added. A colour reaction was achieved using tetramethylbenzidine (TMB; Sigma Aldrich, USA). The reaction was stopped using diluted sulfuric acid. Optical density values were measured at 450 nm on an ELISA plate-reader (Bio-Rad Laboratories, Austria). Serum levels of sST2 were then calculated for each sample using a Excel spreadsheet. Statistical Analysis Continuous variables were presented as median with interquartile range (IQR) and compared using Mann–Whitney U Test. Categorical variables were expressed as numbers with percentages and analyzed using the chi-square test. We visualized the distribution of sST2 (Fig. 1) in patients with and without H. pylori infection using the vioplot command [22]. All statistical tests were two-tailed, and a p value less than 0.05 was considered statistically significant. The primary endpoint in our study was the histological diagnosis of H. pylori, and the primary exposure was the concentration of sST2. We fitted logistic regression models to examine the relationship between sST2 concentration and H. pylori status. In Model-1, sST2 concentration was the independent variable, and the occurrence of H. pylori was the dependent variable. In Model-2, sST2 concentration, age, sex, educational status and concomitant metabolic syndrome diagnosis were fixed effect-dependent variables. Sensitivity analyses (Fig. 2) were also performed to explore the relationship between H. pylori status and sST2 concentration, with patients stratified by sex (male versus female), age (patients aged ≤ 55 years and patients aged > 55 years), BMI (BMI < 30 kg/m2 versus BMI ≥ 30 kg/m2), concomitant metabolic syndrome diagnosis, and current smoking status. We obtained adjusted odds ratios (ORs) and corresponding 95% confidence intervals (CIs). All statistical analyses were conducted using Stata/IC 17.Fig. 1 We visualized the distribution of sST2 in ng/mL (this figure) using vioplot command [22] in patients with and without H. pylori Fig. 2 To investigate the association between sST2 concentration and H. pylori infection, we performed sensitivity analyses using logistic regression models. The occurrence of H. pylori was set as the dependent variable, and sST2 concentration was the independent variable. The patients were stratified by sex, age, BMI, concomitant metabolic syndrome diagnosis, smoking, and educational status. However, we did not observe any significant relationship between sST2 concentration and H. pylori infection in any of the subgroups analyzed Results We compared various clinical and laboratory parameters between patients with (n = 131) and without (n = 563) H. pylori infection. There was no significant difference in age, with both groups having a median age of 57 years (p = 0.75). Similarly, there was no difference in the proportion of patients aged 55 years or younger versus those over 55 years of age (p = 0.95). The proportion of female patients was similar in both groups, with 48% of patients without H. pylori infection and 39% of patients with H. pylori infection being female (p = 0.73). The median BMI was slightly higher in patients with H. pylori infection compared to those without infection (27 kg/m2 vs. 26 kg/m2, p = 0.02). However, there was no significant difference in the proportion of patients with a BMI less than 30 kg/m2 versus those with a BMI of 30 kg/m2 or greater (p = 0.11). The frequency of hypertension, diabetes, and metabolic syndrome did not show any significant difference between the two groups (Table 1). Additionally, smoking status was comparable in both groups. We also found no significant variations between the two groups in the laboratory parameters analyzed, including HbA1c, cholesterol, LDL, HDL, creatinine, hemoglobin, white blood cell count, CRP, and platelet count. There was no significant difference in the median sST2 concentration between patients without (9.67; 7.08–13.06 ng/mL) and those with (9.62; 7.18–13.44 ng/mL; p = 0.66) H. pylori infection (Fig. 1). In the logistic regression, sST2 was not associated with the diagnosis of H. pylori (OR 1.00; 95%CI 0.97–1.04; p = 0.93) and remained so after multivariable adjustment for age, sex, educational status and the diagnosis of metabolic syndrome (aOR 0.99; 95%CI 0.95–1.03; p = 0.60). In the sensitivity analyses stratified by age, sex, BMI, smoking and educational status as well as the concomitant diagnosis of metabolic syndrome, we found no association between sST2 and H. pylori infection in any of the subgroups.Table 1 Baseline characteristics of patients with (n = 131) and without (n = 563) H. pylori infection No H. pylori H. pylori p value N = 563 N = 131 Age (years) 57 (52–63) 57 (53–64) 0.75 Age categories 0.95  Age ≤ 55 40% (226) 40% (53)  Age > 55 60% (337) 60% (78)  Female sex 48% (268) 39% (51) 0.73  BMI (kg/m2) 26 (24–29) 27 (25–30) 0.02 BMI categories 0.11  BMI < 30 kg/m2 80% (452) 74% (97)  BMI ≥ 30 kg/m2 20% (111) 26% (34)  Hypertension (yes/no) 61% (342) 62% (81) 0.82 Metabolic syndrome (yes/no) 83% (469) 80% (105) 0.39 Diabetes (yes/no) 31% (174) 32% (42) 0.80 Hba1c (%) 5.4 (5.2–5.6) 5.4 (5.2–5.7) 0.44 Cholesterol (mg/dL) 224 (194–253) 220 (195–244) 0.31 LDL (mg/dL) 147 (118–173) 143 (121–165) 0.49 HDL (mg/dL) 57 (48–68) 56 (47–66) 0.59 Creatinine (mg/dL) 0.9 (0.8–1.0) 0.9 (0.8–1.0) 0.24 Hemoglobine (mg/dL) 14.6 (13.8–15.4) 14.9 (13.9–15.6) 0.12 White blood count (G/L 5.6 (4.7–6.7) 6.0 (5.0–7.2) 0.35 CRP (mg/dL) 0.1 (0.1–0.3) 0.2 (0.1–0.3) 0.68 Platelets (G/L) 240 (208–276) 238 (208–274) 0.89 Smoking status 0.14  Non-smoker 83% (433) 78% (94)  Current smoker 17% (86) 22% (27)  ST2 (pg/mL) 9.57 (7.08–13.06) 9.62 (7.18–13.55) 0.66 Continuous data are given as median ± inter-quartile range (IQR) and compared using Mann’s Whitney U Test. Categorical data are given as numbers (percentage) and compared using the chi-square test. All tests were two-sided, and a p value of < 0.05 was considered statistically significant Discussion The previous literature on sST2 and H. pylori infection is scarce. Bergis et al. investigated the role of Interleukin-33 (IL-33), its membrane bound cellular receptor ST2L, and its soluble receptor sST2 in gastric cancer [16]. Thirty gastric cancer patients, 51 gastritis patients, and 40 healthy volunteers were enrolled, and the levels of IL-33 and sST2 were determined by ELISA. The results showed that sST2 levels in gastric cancer were significantly higher than in gastritis or healthy controls, and higher levels of sST2 were seen in patients with lower degree of tumor differentiation. Soluble ST2 was significantly associated with a more advanced tumor stage, metastatic disease, and significantly correlated with the duration of the disease [16]. Another study investigated the interaction between H. pylori and the proinflammatory cytokine IL-33 and its receptor ST-2 [17]. The results showed that H. pylori infection elevated expression levels of IL-33 and ST-2 and caused ST-2 mobilization into membrane lipid rafts. Depletion of membrane cholesterol reduced H. pylori-induced IL-33 and IL-8 production. In vivo studies revealed that H. pylori infection led to upregulation of IL-33/ST-2 and severe leukocyte infiltration in gastric tissues, indicating that ST-2 recruitment into the lipid rafts serves as a platform for IL-33-dependent H. pylori infection, which aggravates inflammation in the stomach. Two other studies showed that IL-33, for which sST2 is the soluble decoy receptor, is involved in inflammation mediated by H.pylori infection [23, 24]. Based on these preliminary studies, we speculated about a role of sST2 in asymptomatic H. pylori infection. However, our study did not find any significant association between sST2 levels and H. pylori infection in asymptomatic patients. This suggests that sST2 may not serve as a valuable biomarker for the diagnosis and management of H. pylori infection. All of our patients were asymptomatic and underwent gastroscopy because it was offered to them as part of an opportunistic colon cancer screening program using colonoscopy. However, some of our patients may have had more severe gastritis than others, or even subtle symptoms. Unfortunately, we do not have this data.sST2 is a biomarker that has been extensively studied in recent years [7–9]. An association between sST2 and mortality has already been demonstrated in several populations [25, 26]. sST2 was proposed as a marker for heart failure [27], but emerging evidence suggests that it may also play a role in autoimmune diseases [28], infectious diseases [9], and cancer [14]. In autoimmune diseases, sST2 has been shown to be associated with disease activity in conditions such as systemic lupus erythematosus [29] and rheumatoid arthritis [30]. In infectious diseases, sST2 levels have been reported to be elevated in severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection [31] and sepsis [9, 32]. In cancer, sST2 levels have been linked to tumor progression and metastasis [15, 16]. sST2 has also been investigated in cardiovascular diseases, such as acute myocardial infarction [33], heart failure [12], and atherosclerosis [34]. Furthermore, sST2 has been studied in the context of chronic obstructive pulmonary disease [35], chronic kidney disease [11], and type 2 diabetes mellitus [36]. Therefore, it is actually surprising that there has been no data on the association between sST2 and H. pylori infection so far. Based on our data, we can exclude such an association for asymptomatic patients. Nevertheless, we consider it important to document and publish our unequivocal negative results. First, to avoid reporting bias. Reporting bias refers to the systematic error that can arise in research studies when there is selective reporting or publication of certain results or outcomes over others [37]. It can occur when the researchers choose to report only the results that support their hypothesis or research question, or if journals are more likely to accept and publish positive results rather than negative or inconclusive ones [38, 39]. This bias can lead to an incomplete or misleading understanding of the research question being studied, as important information may be left out or overemphasized. It can also lead to a skewed perception of the efficacy or safety of a particular intervention or treatment. Reporting bias can be minimized by ensuring that all study outcomes are reported, regardless of whether they are statistically significant or not. Second, to state that sST2 is definitely not a suitable biomarker for the detection of H. pylori. And third, to show scientists who are investigating other diseases (diabetes, cancer, cardiovascular diseases) and sST2 that any concomitant asymptomatic H. pylori infection does not bias their results. Conclusion In conclusion, this study found no significant association between H. pylori infection and sST2 levels in asymptomatic patients. The median sST2 concentration was similar in both infected and non-infected patients. Logistic regression analysis and sensitivity analyses also did not show any association between sST2 levels and H. pylori infection. These results suggest that sST2 may not serve as a useful biomarker for the diagnosis and management of H. pylori infection. These findings have relevance for researchers investigating sST2 in various diseases and indicate that asymptomatic H. pylori infection does not bias the results of sST2 studies. Author's contribution SW, BW, ML and CD conceived the presented idea. All authors contributed to the final version of the manuscript and approved the final version. Funding Open access funding provided by Paracelsus Medical University. There is no funding to report for this submission. Data availability Data are available upon reasonable request. Declarations Competing interests CD is part of the scientific advisory board of SPAR Austria. Ethical approval We performed the study and all procedures according to the principles of the Declaration of Helsinki. The local ethics committee for the province Salzburg approved the study protocol (Approval No. 415-E/1262). Written informed consent was obtained from every participant. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Chey WD Leontiadis GI Howden CW ACG clinical guideline: treatment of Helicobacter pylori infection Am J Gastroenterol 2017 112 212 239 10.1038/ajg.2016.563 28071659 2. Braden B Diagnosis of Helicobacter pylori infection BMJ 2012 344 e828 10.1136/bmj.e828 22368293 3. Kavitt RT Cifu AS Management of Helicobacter pylori infection JAMA 2017 317 1572 1573 10.1001/jama.2017.1949 28418469 4. Malfertheiner P Megraud F O’Morain CA Management of Helicobacter pylori infection-the Maastricht V/Florence Consensus Report Gut 2017 66 6 30 10.1136/gutjnl-2016-312288 27707777 5. Lee A Hong J Chung H Author Correction: Helicobacter pylori eradication affects platelet count recovery in immune thrombocytopenia Sci Rep 2020 10 18198 10.1038/s41598-020-74316-1 33082421 6. Monzón H Forné M Esteve M Helicobacter pylori infection as a cause of iron deficiency anaemia of unknown origin World J Gastroenterol 2013 19 4166 4171 10.3748/wjg.v19.i26.4166 23864779 7. Zhang J Chen Z Ma M Soluble ST2 in coronary artery disease: clinical biomarkers and treatment guidance Front Cardiovasc Med 2022 9 924461 10.3389/fcvm.2022.924461 36225958 8. Boga S Alkim H Koksal AR Serum ST2 in inflammatory bowel disease: a potential biomarker for disease activity J Investig Med 2016 64 1016 1024 10.1136/jim-2016-000062 27001944 9. Hur M Kim H Kim HJ Soluble ST2 has a prognostic role in patients with suspected sepsis Ann Lab Med 2015 35 570 577 10.3343/alm.2015.35.6.570 26354344 10. Iwahana H Yanagisawa K Ito-Kosaka A Different promoter usage and multiple transcription initiation sites of the interleukin-1 receptor-related human ST2 gene in UT-7 and TM12 cells Eur J Biochem 1999 264 397 406 10.1046/j.1432-1327.1999.00615.x 10491084 11. Homsak E Gruson D Soluble ST2: a complex and diverse role in several diseases Clin Chim Acta 2020 507 75 87 10.1016/j.cca.2020.04.011 32305537 12. Savarimuthu S Goel P Harky A Soluble ST2: a valuable prognostic marker in heart failure Heart Fail Rev 2022 27 2155 2164 10.1007/s10741-022-10258-2 35759079 13. Pascual-Figal DA Januzzi JL The biology of ST2: the International ST2 Consensus Panel Am J Cardiol 2015 115 3B 7B 10.1016/j.amjcard.2015.01.034 25665766 14. De la Fuente M MacDonald TT Hermoso MA The IL-33/ST2 axis: role in health and disease Cytokine Growth Factor Rev 2015 26 615 623 10.1016/j.cytogfr.2015.07.017 26271893 15. Lu D-P Zhou X-Y Yao L-T Serum soluble ST2 is associated with ER-positive breast cancer BMC Cancer 2014 14 198 10.1186/1471-2407-14-198 24636276 16. Bergis D Kassis V Radeke HH High plasma sST2 levels in gastric cancer and their association with metastatic disease Cancer Biomark 2016 16 117 125 10.3233/CBM-150547 26835712 17. Kuo C-J Chen C-Y Lo H-R Helicobacter pylori induces IL-33 production and recruits ST-2 to lipid rafts to exacerbate inflammation Cells 2019 10.3390/cells8101290 31817208 18. Wernly S Wernly B Semmler G Non-alcoholic fatty liver disease is not independently associated with Helicobacter pylori in a Central European screening cohort Minerva Med. 2022 10.23736/S0026-4806.22.07928-9 35384436 19. Wernly S Semmler G Völkerer A Cardiovascular risk assessment by SCORE2 predicts risk for colorectal neoplasia and tumor-related mortality J Pers Med 2022 10.3390/jpm12050848 35629269 20. Semmler G Wernly S Wernly B Machine learning models cannot replace screening colonoscopy for the prediction of advanced colorectal adenoma J Pers Med 2021 10.3390/jpm11100981 34945736 21. Huang PL A comprehensive definition for metabolic syndrome Dis Model Mech 2009 2 231 237 10.1242/dmm.001180 19407331 22. Winter N, Nichols A. VIOPLOT: Stata module to produce violin plots with current graphics. Published Online First: 17 June 2012. https://EconPapers.repec.org/RePEc:boc:bocode:s456902 (Accessed 28 Feb 2023). 23. Gonciarz W Krupa A Moran AP Interference of LPS H pylori with IL-33-driven regeneration of caviae porcellus primary gastric epithelial cells and fibroblasts Cells 2021 10.3390/cells10061385 34199843 24. Buzzelli JN Chalinor HV Pavlic DI IL33 is a stomach alarmin that initiates a skewed Th2 response to injury and infection Cell Mol Gastroenterol Hepatol 2015 1 203 21.e3 10.1016/j.jcmgh.2014.12.003 28210674 25. Pfetsch V Sanin V Jaensch A Increased plasma concentrations of soluble ST2 independently predict mortality but not cardiovascular events in stable coronary heart disease patients: 13-year follow-up of the KAROLA study Cardiovasc Drugs Ther 2017 31 167 177 10.1007/s10557-017-6718-1 28283847 26. Hammer F Genser B Dieplinger B Soluble suppression of tumorigenesis-2 is a strong predictor of all-cause, cardiovascular and infection-related mortality risk in haemodialysis patients with diabetes mellitus Clinical Kidney J. 2022 15 1915 1923 10.1093/ckj/sfac142 36158148 27. Gürgöze MT van Vark LC Baart SJ Multimarker analysis of serially measured GDF-15, NT-proBNP, ST2, GAL-3, cTnI, creatinine, and prognosis in acute heart failure Circ Heart Fail 2023 16 e009526 10.1161/CIRCHEARTFAILURE.122.009526 36408685 28. Shakerian L Kolahdooz H Garousi M IL-33/ST2 axis in autoimmune disease Cytokine 2022 158 156015 10.1016/j.cyto.2022.156015 36041312 29. Song Y Wei F Liu Y IL-33/ST2 activation is involved in Ro60-regulated photosensitivity in cutaneous lupus erythematosus Mediators Inflamm 2022 2022 4955761 10.1155/2022/4955761 35909659 30. Shi L-J Liu C Li J-H Elevated levels of soluble ST2 were associated with rheumatoid arthritis disease activity and ameliorated inflammation in synovial fibroblasts Chin Med J 2018 131 316 322 10.4103/0366-6999.223847 29363647 31. Park M Hur M Kim H Soluble ST2 as a useful biomarker for predicting clinical outcomes in hospitalized COVID-19 patients Diagnostics (Basel) 2023 10.3390/diagnostics13020259 37443569 32. Van Nynatten LR Slessarev M Martin CM Novel plasma protein biomarkers from critically ill sepsis patients Clin Proteomics 2022 19 50 10.1186/s12014-022-09389-3 36572854 33. Søndergaard FT Beske RP Frydland M Soluble ST2 in plasma is associated with post-procedural no-or-slow reflow after primary percutaneous coronary intervention in ST-elevation myocardial infarction Eur Heart J Acute Cardiovasc Care 2023 12 48 52 10.1093/ehjacc/zuac146 36355574 34. Pfetsch V Sanin V Jaensch A Increased concentrations of soluble ST2 independently predict cardiac and total mortality but not non-fatal cardiovascular events in stable coronary heart disease patients Atherosclerosis. 2017 263 e46 e47 10.1016/j.atherosclerosis.2017.06.159 35. Urban MH Stojkovic S Demyanets S Soluble ST2 and all-cause mortality in patients with chronic obstructive pulmonary disease—a 10-year cohort study J Clin Med Res 2021 11 56 36. Zeyda M Wernly B Demyanets S Severe obesity increases adipose tissue expression of interleukin-33 and its receptor ST2, both predominantly detectable in endothelial cells of human adipose tissue Int J Obes 2013 37 658 665 10.1038/ijo.2012.118 37. Easterbrook PJ Berlin JA Gopalan R Publication bias in clinical research Lancet 1991 337 867 872 10.1016/0140-6736(91)90201-Y 1672966 38. Chan A-W Krleza-Jerić K Schmid I Outcome reporting bias in randomized trials funded by the Canadian Institutes of Health Research CMAJ 2004 171 735 740 10.1503/cmaj.1041086 15451835 39. Sahni P Aggarwal R Reporting and Publishing Research in the Biomedical Sciences 2018 Singapore Springer
PMC010xxxxxx/PMC10352469.txt
==== Front MedEdPORTAL MedEdPORTAL mep MedEdPORTAL : the Journal of Teaching and Learning Resources 2374-8265 Association of American Medical Colleges 11322 10.15766/mep_2374-8265.11322 Original Publication Improving Procedural Skill Confidence in Pediatric Residents: A Longitudinal Simulation-Based Workshop Curriculum https://orcid.org/0000-0002-7437-3622 Maleknia Lydia MD 1 * Boshuizen Vanessa MD, MPH 2 Caputo Heather MD 3 Shah Rina MD 4 1 Fellow, Department of Pediatric Hospital Medicine, Kaiser Permanente Oakland Medical Center 2 Faculty Pediatrician, Community Health Connection 3 Faculty Hospitalist, Department of Pediatric Hospital Medicine, Kaiser Permanente Oakland Medical Center 4 Associate Program Director, Kaiser Permanente Northern California Pediatric Residency Program; Assistant Chief and Faculty Hospitalist, Department of Pediatric Hospital Medicine, Kaiser Permanente Oakland Medical Center ∗Corresponding author: [email protected] 2023 18 7 2023 19 1132216 8 2022 3 4 2023 © 2023 Maleknia et al. 2023 Maleknia et al. https://creativecommons.org/licenses/by-nc/4.0/ This is an open-access publication distributed under the terms of the Creative Commons Attribution-NonCommercial license. Introduction Exit surveys among our pediatric residency graduates found 50% were not confident performing required procedures. While procedural competency poses many curricular challenges, simulation is an effective educational modality many programs have adopted, though often only through onetime workshops limited to single procedures, clinical settings, or levels of training. We sought to develop a comprehensive, recurring, yearlong, simulation-based curriculum covering many important pediatric procedures. Methods We created a longitudinal curriculum of recurring monthly workshops using both low- and high-fidelity simulators, highlighting 17 pediatric procedures. Comprehensive facilitator guides contained equipment lists, instructions, competency checklists, and quizzes for each workshop. Correlation between attendance and confidence was assessed for skills in which residents attended two or more workshops on the same skill. ACGME exit surveys compared graduates’ confidence regarding procedural skills before and after curriculum implementation. Results On exit surveys, graduates who agreed or strongly agreed to feeling comfortable with the procedures in our curriculum improved from 50% to 66% after 2 years, and those who disagreed or strongly disagreed decreased from 40% to 22%. A positive correlation existed between repeated workshop attendance and confidence in many procedures (R2 range, .60–.99). Discussion Longitudinal simulation is an effective educational modality that increases learner confidence in performing procedures. Our curriculum addresses adult learners’ need for repetition and can be adopted by other programs to improve graduates’ confidence. The curriculum's sustainability is underscored by use of cost-reducing low-fidelity simulators and comprehensive guides that allow any instructor to conduct the workshop. Keywords Clinical/Procedural Skills Training Pediatrics Simulation ==== Body pmcEducational Objectives By the end of this activity, learners will be able to: 1. Perform procedural skills using simulators with direct feedback from a facilitator. 2. Demonstrate proper use of basic equipment needed for each procedure. 3. Improve confidence in performing common pediatric procedural skills. Introduction All pediatric residency training programs governed by the Accreditation Council for Graduate Medical Education (ACGME) are to provide clinical training in procedural skills for pediatric resident physicians.1 Program directors face challenges in ensuring residents receive adequate opportunities for procedure practice, feedback, and assessment.2,3 Consistent with national attitudes on pediatric procedural skill readiness,4 an ACGME exit survey among trainees in our pediatric residency program indicated that 50% were not confident in performing ACGME-required procedures. Historically, studies have shown that residents gain procedural skills at a patient's bedside.5 However, there are often limitations in the number of procedures residents are exposed to, and patient safety sometimes limits resident opportunities to practice.6–8 This may be compounded by the frequency with which patients undergo procedures prior to their transfer to our tertiary care center. Because effective adult learning includes engagement in repetitive practice,9 this lack of practice poses a challenge to mastering procedural skills learned at the bedside.6 Simulation-based education can increase opportunities for trainees to practice their skills. Furthermore, standardized checklists and feedback in the setting of simulation-based education have been shown to improve residents’ confidence in procedural skills.10 In fact, residents trained on simulators perform better and are more likely to adhere to protocol compared to those receiving standard training.8,11 The literature has shown success in improving learner confidence after onetime simulation workshops,12 and many of the published studies available are limited to single clinical settings10,13–16 or a small number of procedures10,13,14,16–21 or include only senior residents10,13 or interns.22 Several MedEdPORTAL resources available pertain to pediatric procedural skills. Sagalowsky et al. provided a longitudinal curriculum for four simulated pediatric resuscitation scenarios in which a small number of basic procedures were embedded (including bag mask ventilation, intubation, and intravenous access).21 Good et al. published a simulation-based training for ultrasound-guided central venous catheter placement for pediatric trainees demonstrating improved confidence and knowledge, although they did not implement or assess this long-term.23 Auerbach et al. provided a comprehensive training package for formative feedback on simulated infant lumbar puncture, but like Good et al.'s curriculum, focused on only one procedural skill.20 Sawyer et al. created skills checklists for seven common pediatric procedures for use during a skills training day, but these were only available for incoming interns at an annual workshop.22 Our curriculum aims to address some of the limitations of prior work by offering a curriculum with repetitive, longitudinal practice for many procedures found across multiple clinical settings and all training levels in order to improve access to and confidence in performing required procedures. This was done by implementing recurring monthly procedure workshops using both low- and high-fidelity simulators accompanied by didactic sessions highlighting the ACGME-required procedures. Building off prior work by Sawyer et al.,22 facilitator workshop guides for 17 procedural skills were developed for implementation in a comprehensive yearlong curriculum in our pediatric residency program. Our facilitator guides offer a unique contribution to the literature by including goals and objectives, equipment lists, risks/benefits, indications/contraindications, workshop setup, skills checklists, and knowledge pretests, allowing other residencies to implement the curriculum within their respective programs. Methods In response to our ACGME exit survey results, we aimed to develop a longitudinal curriculum in which pediatric residents could practice core pediatric procedures on simulators. The workshops were added to our standing daily noontime conferences and were generally 1.5 hours in duration. The workshopped skills included venipuncture, intravenous line placement, arterial puncture, chest compressions, bag valve mask ventilation, defibrillation, intubation, lumbar puncture, intraosseous access, bladder catheterization, neonatal intubation, umbilical arterial and venous catheterization, incision/drainage, suturing, and joint reduction and splinting. The skills covered and those repeated in a given year (Figure) were determined by the program to be important high-value ones (i.e., deemed highest yield for residents and graduated pediatricians, including both ACGME-required and other important resuscitative skills). All the skills repeated once in an academic year, except for incision/drainage, suturing, joint reduction, and splinting, to prioritize repetition for the skills thought most likely to be encountered within our specialty. In the case of our residency, the unrepeated skills tended to be consistently encountered in senior emergency department rotations and had relatively higher confidence on the baseline exit survey. Prior to implementing the procedure curriculum, we obtained feedback regarding the format and content of the workshop and fidelity of the simulators following one pilot workshop. Feedback included having multiple stations available, limiting the number of learners per station, and increasing the time dedicated to learner engagement with the simulator. The curriculum was fully implemented during the 2018–2019 academic year. Figure. Timeline of procedure curriculum for an academic year color-coded by workshop. BLS skills workshops included cardiopulmonary resuscitation, defibrillator training, and bag mask ventilation. Airway workshops included endotracheal intubation and laryngeal mask airways. Abbreviations: BLS, basic life support; IV, intravenous line; UAC, umbilical artery catheterization; UVC, umbilical vein catheterization. Facilitators for each workshop were intended to be pediatric hospitalists without prerequisite knowledge outside of their normal scope of practice. Each workshop included two to three facilitators, often one or more pediatric hospitalists with or without expert faculty. For example, intensivists and anesthesiologists were invited to airway workshops, intensive care nurses to intraosseous line workshops, pediatric nurses to intravenous line workshops, and neonatologists to neonatal intubation and umbilical catheterization workshops. The use of expert faculty was optional but added the benefit of experience in discussing risks, benefits, indications, and equipment troubleshooting. The equipment necessary to run each workshop varied based on the skills included in that workshop, and the amount of supplies required was based on our 30-resident program (Appendix A). There was a mix of high-fidelity simulators and low-cost simulators. For example, instead of an abscess task trainer for incision/drainage, we made our own low-cost abscesses, like other simulation workshops in the literature.24 The workshops were carried out in either conference rooms or empty hospital patient rooms (depending on availability) and had computer access for the facilitators. At the start of each workshop, residents completed a short pretest on paper to assess knowledge of each skill (Appendices B–H), then split into groups to attend each station. Facilitators used checklists for each skill to review indications, risks, and benefits for each procedure as well as how to troubleshoot the equipment. Prior checklists published by Sawyer et al. were utilized alongside checklists for additional skills adapted for our curriculum (Appendices B–H).22 This gave the facilitators a guide by which to check off each resident's skill proficiency in real time and offer feedback. Facilitators also used answer sheets to review the correct responses to the pretests with the participants (Appendices B–H). Facilitator guides included a list of optional supplemental information facilitators could find via web search to help them prepare for each session; computers were available in the rooms if they wanted to display any supplemental information at their discretion, but specific resources were not provided. The ideal setup and timing for each workshop were provided in each facilitator guide. To assess our curriculum's main purpose of improving graduates’ procedural confidence, the ACGME exit survey was used to evaluate the effectiveness of the overall curriculum, comparing mean confidence for procedural skills in residency classes prior to and after implementation of the curriculum. Changes were evaluated using paired t tests to test for statistical significance. To evaluate the quality of each workshop in improving trainee confidence, as well as the effect of longitudinal repetition, trainees completed anonymous surveys on paper that ranked their self-reported confidence in each skill in a given workshop on a 10-point Likert scale (1 = least confident, 10 = most confident) at the beginning of the workshop (Appendix I). Trainees also indicated the number of times they had participated in that particular workshop. To limit cognitive bias, we did not reassess confidence immediately after the workshop but rather in future surveys of the same workshop. For example, a resident who attended three workshops on the same skill provided survey responses prior to their first workshop, prior to their second workshop, and prior to their third workshop. The latter two responses would represent delayed confidence responses after exposure to the first and second workshops, respectively. Mean confidence scores were plotted against number of workshops attended for each skill, and best-fit linear trendlines were obtained. The correlation coefficient (R2) was determined for each skill that had residents attending two or more of its workshops. Level of training (PGY year) was also obtained to assess for relationships between confidence and year. To protect anonymity, residents were not compared to their own selves. Instead, aggregate data were obtained, and confidence was assessed for correlation with number of workshops attended. Data were collected during the 2018–2019 and 2019–2020 academic years. The Research Determination Committee for the Kaiser Permanente Northern California region determined the project did not meet the regulatory definition of research involving human subjects per 45 CFR 46.102(d). Results The procedure curriculum was implemented successfully and continues as a regular monthly workshop series within the residency noon conferences. For our program of 30 residents, attendance varied from 10 to 25 residents for each session, consisting of two to three stations with facilitators at each station. Each workshop's ideal flow and setup are included in its respective facilitator guide. Residents split up into groups to attend each station, taking turns between observing and performing each skill, with the facilitator supervising. Faculty facilitators consisted of pediatric hospitalists (including the authors) but, depending on the workshop, also included expert faculty. To prepare for the workshops ahead of time, facilitators received with the facilitator guides with the goals of the workshops, checklists for procedural competence, and suggested types of optional resources they could find via web search should they need additional information in preparation for the session. While each of the workshops was not individually assessed by the residents after the initial pilot feedback, the pediatric hospitalists at each workshop included the authors, who could assure the sessions ran smoothly. Debriefs and informal feedback indicated that facilitators found the workshop timing and flow to be appropriate and had enough time to provide valuable feedback on performance, observe that residents used the equipment properly, and review the session quiz answers. Facilitators oversaw the residents verbalizing indications, contraindications, risks, and benefits of each procedure as they practiced at each station and guided them with real-time coaching as necessary. The performance checklists were not collected but rather were used for direct feedback to the residents. When evaluating the attitudes of graduating residents (n = 10 for each graduating class) on the ACGME exit survey, the percentage who agreed or strongly agreed they felt comfortable with the procedures included in our curriculum improved from 50% to 66% after 2 years of the curriculum ( p = .10). The proportion of graduating residents disagreeing or strongly disagreeing they felt comfortable across procedures included in our curriculum decreased from 40% to 22% after 2 years of the curriculum ( p = .002). Residents who graduated 3 years from the start of the curriculum were not assessed due to temporary cancellation of in-person didactics secondary to the COVID-19 pandemic, which limited our ability to perform these workshops for over 10 months. The residents were also evaluated on their confidence in performing each procedure, and correlation between repeated attendance and confidence was assessed for skills for which residents attended two or more workshops on the same procedural skill. For several skills, there was a positive correlation between the number of workshops of a particular skill attended and confidence in that skill, with R2 ranging between .60 and .99 (intubation R2 = .99, intraosseous access R2 = .99, venipuncture R2 = .96, bladder catheterization R2 = .84, peripheral IV placement R2 = .82, lumbar puncture R2 = .79, and umbilical catheterization R2 = .60). However, there was no change in confidence for bag mask ventilation and chest compressions, which remained stably high, or neonatal intubation and defibrillation, which remained stably low. Data collection on arterial puncture was interrupted due to lack of access to the arterial task trainer during the sampling period, and not enough workshops were attended to make assessments on incision/drainage, suturing, splinting, and reduction. The results were not controlled for potential confounding variables, such as number of procedures a resident may have completed on real patients between each workshop, or by PGY year, which we expand on in the Discussion section. Discussion This project adds to the existing body of literature indicating that longitudinal simulation is an effective educational modality that can increase learner confidence in performing procedures in a safe environment. After implementation of this curriculum, our residents demonstrated improved procedural skill confidence at the time of graduation, as evidenced by the ACGME exit surveys. To our knowledge, this is the first procedure-based publication that includes comprehensive facilitator guides with equipment lists, performance checklists, and competency quizzes for an entire year's worth of curriculum for a pediatric residency. Our curriculum addresses adult learners’ need for repetitive practice by building redundancy into the yearlong monthly curriculum and allowing pediatric trainees from three to six opportunities to attend the simulation workshop on a particular skill throughout the course of their residency. To avoid cognitive biases such as recency/availability bias, whereby resident confidence may be inflated immediately after a workshop, we chose to administer confidence surveys the subsequent time residents attended the same workshop, rather than immediately afterward, and to assess for long-term improvements in confidence. One trade-off of this decision is that it did not control for any real-life procedure exposures that could have affected confidence scores between workshops. Given the limited amount of procedure opportunities our pediatric residents faced, we felt that the increases in self-reported confidence with increased workshop attendance over time suggested a positive dose-related effect and justified continuing this simulation-based procedural curriculum in our residency program. Potential limitations of the curriculum could include the frequent need for facilitators, though this has been alleviated by the use of facilitator guides that include all the materials required for each workshop, as well as the need for sometimes costly simulator equipment. We were able to secure access to simulators through partnership with our institution's simulation lab and other departments, including the pediatric and neonatal intensive care units. However, despite this partnership, we were still limited by temporary lack of access to an arterial task trainer. Nonetheless, not all the workshops required high-fidelity simulators. Some required only bananas for suturing practice or bandages filled with toothpaste for incision/drainage practice, whereas others, such as for splinting practice, used residents as participants. Assessment of the curriculum also had limitations, including participants’ recall bias regarding how many workshops had been attended or potential errors in confusing which workshops had previously been attended (i.e., airway management vs. neonatal intubation), as well as the confounding factor of time, whereby residents may have had exposure to procedures on real patients in the time between workshops. Due to attendance variability, a workshop conducted at the end of an academic year could capture one senior resident's first time at the workshop and another's third time, yet both should have had comparable real-patient exposures. While the correlation we observed (between number of workshops attended and stated confidence) need not imply causation, we are nonetheless encouraged by the improvements in the ACGME exit survey since implementing the curriculum and will therefore continue the curriculum in our residency. Only self-reported confidence was assessed as an outcome of the curriculum, rather than competence, which could be a future aim to study. We attempted to use the pretests as a way to track improved competence/knowledge for each skill with repeated workshop exposure but were limited by resident engagement and inability to collect enough pretests to make a valuable assessment. We were not able to gather data on every skill for which residents had attended at least two workshops because not enough residents had that repeated exposure during our sampling period and the COVID-19 pandemic resulted in a pause of all in-person gatherings during the third year of curriculum implementation. Similarly, we were not able to meaningfully stratify the results by PGY year given our sampling period and relatively small residency size. However, this only underscores the need for a longitudinal repetitive curriculum to address the challenges of capturing all our residents at a given time (due to residents missing the workshops while on night shifts, away rotations, off-site rotations, vacation, etc.). Overall, this comprehensive curriculum can be adopted immediately by other programs to improve both procedure exposure and graduate confidence thanks to the instructor guides that allow any instructor to lead the workshop. Programs can further tailor the curriculum by conducting a needs assessment to understand the needs of their residents, evaluating their exit survey results, and deciding which skills to include and which to repeat. The curriculum's sustainability is underscored by its use of cost-reducing low-fidelity simulators. The workshops can also be utilized to formally evaluate struggling learners’ competence in particular procedural skills through the use of procedure checklists or checking off residents on skills they may not have had exposure to in clinical practice. Our residency program continues to use these recurring monthly procedure workshops in its didactic schedule given the impact they have had on trainee confidence both during and after residency. Furthermore, the small-group setup of the workshops made them accessible during the COVID-19 pandemic when social distancing was a priority. After an initial hiatus from in-person gatherings mandated by our graduate medical education office, we were able to safely restart the curriculum with appropriate personal protective equipment and cleaning procedures. Appendices Equipment Checklist for Procedure Workshops.docx Facilitator Guide - Venipuncture, IV, Arterial Puncture.docx Facilitator Guide - BLS Skills.docx Facilitator Guide - LP, IO, Bladder Cath.docx Facilitator Guide - UVC, UAC, Neonatal Intubation.docx Facilitator Guide - Airway Management.docx Facilitator Guide - I&D, Suturing.docx Facilitator Guide - Splinting and Reduction.docx Resident Confidence Surveys.docx All appendices are peer reviewed as integral parts of the Original Publication. Disclosures Funding/Support Prior Presentations Ethical Approval None to report. None to report. Maleknia L, Boshuizen V, Caputo H, Shah R. Development of a longitudinal procedure-based curriculum in a pediatric residency program. Presented at: Academic Pediatric Association Regional Meeting (Region 9 and 10); January 2019; Monterey, CA. The Kaiser Permanente Northern California Research Determination Committee approved this project. ==== Refs References 1. Accreditation Council for Graduate Medical Education. ACGME Program Requirements for Graduate Medical Education in Pediatrics. Accreditation Council for Graduate Medical Education; 2022. 2. Gaies MG, Landrigan CP, Hafler JP, Sandora TJ. Assessing procedural skills training in pediatric residency programs. Pediatrics. 2007;120 (4 ):715–722. 10.1542/peds.2007-0325 17908757 3. Auerbach MA, White ML, Bhargava S, et al. Are graduating pediatric residents prepared to perform infant lumbar punctures? A multi-institutional descriptive study. Pediatr Emerg Care. 2018;34 (2 ):116–120. 10.1097/PEC.0000000000000914 27741067 4. Schumacher DJ, Frintner MP, Cull W. Graduating pediatric resident reports on procedural training and preparation. Acad Pediatr. 2018;18 (1 ):73–78. 10.1016/j.acap.2017.08.001 28826730 5. Mills DM, Williams DC, Dobson JV. Simulation training as a mechanism for procedural and resuscitation education for pediatric residents: a systematic review. Hosp Pediatr. 2013;3 (2 ):167–176. 10.1542/hpeds.2012-0041 24340419 6. Weinberg ER, Auerbach MA, Shah NB. The use of simulation for pediatric training and assessment. Curr Opin Pediatr. 2009;21 (3 ):282–287. 10.1097/MOP.0b013e32832b32dc 19381090 7. Issenberg SB, McGaghie WC, Petrusa ER, Gordon DL, Scalese RJ. Features and uses of high-fidelity medical simulations that lead to effective learning: a BEME systematic review. Med Teach. 2005;27 (1 ):10–28. 10.1080/01421590500046924 16147767 8. Okuda Y, Bryson EO, DeMaria S Jr, et al. The utility of simulation in medical education: what is the evidence? Mt Sinai J Med. 2009;76 (4 ):330–343. 10.1002/msj.20127 19642147 9. Knowles MS. The Adult Learner: A Neglected Species. Gulf Publishing Company; 1973. 10. Goldman MP, Rudd AV, Baum SC, et al. Formative assessments promote procedural learning and engagement for senior pediatric residents on rotation in the pediatric emergency department. MedEdPORTAL. 2022;18 :11265. 10.15766/mep_2374-8265.11265 35892044 11. Velmahos GC, Toutouzas KG, Sillin LF, et al. Cognitive task analysis for teaching technical skills in an inanimate surgical skills laboratory. Am J Surg. 2004;187 (1 ):114–119. 10.1016/j.amjsurg.2002.12.005 14706600 12. Augustine EM, Kahana M. Effect of procedure simulation workshops on resident procedural confidence and competence. J Grad Med Educ. 2012;4 (4 ):479–485. 10.4300/JGME-D-12-00019.1 24294425 13. Tofil NM, Benner KW, Zinkan L, Alten J, Varisco BM, White ML. Pediatric intensive care simulation course: a new paradigm in teaching. J Grad Med Educ. 2011;3 (1 ):81–87. 10.4300/JGME-D-10-00070.1 22379527 14. Barsuk JH, McGaghie WC, Cohen ER, Balachandran JS, Wayne DB. Use of simulation-based mastery learning to improve the quality of central venous catheter placement in a medical intensive care unit. J Hosp Med. 2009;4 (7 ):397–403. 10.1002/jhm.468 19753568 15. Skertich NJ, Schimpke SW, Lee T, et al. Pediatric surgery simulation-based training for the general surgery resident. J Surg Res. 2021;258 :339–344. 10.1016/j.jss.2020.05.038 32561030 16. Rosen O, Campbell D, Bruno C, Gabelman L, Goffman D, Angert R. Low-cost, easy-to-assemble neonatal procedural trainers: chest tube, pericardiocentesis, and exchange transfusion. MedEdPORTAL. 2014;10 :9787. 10.15766/mep_2374-8265.9787 17. Sall D, Wigger GW, Kinnear B, Kelleher M, Warm E, O'Toole JK. Paracentesis simulation: a comprehensive approach to procedural education. MedEdPORTAL. 2018;14 :10747. 10.15766/mep_2374-8265.10747 30800947 18. Goldman MP, Bhatnagar A, Nagler J, Auerbach MA. Advanced pediatric emergency airway management: a multimodality curriculum addressing a rare but critical procedure. MedEdPORTAL. 2020;16 :10962. 10.15766/mep_2374-8265.10962 32908951 19. Gorelik N, Patil K, Chen SJ, Bhatnagar S, Faingold R. Impact of simulation training on radiology resident performance in neonatal head ultrasound. Acad Radiol. 2021;28 (6 ):859–867. 10.1016/j.acra.2020.06.040 32768353 20. Auerbach M, Chang T, Fein D, et al. A comprehensive infant lumber puncture novice procedural skills training package: an INSPIRE simulation-based procedural skills training package. MedEdPORTAL. 2014;10 :9724. 10.15766/mep_2374-8265.9724 21. Sagalowsky S, Boyle T, Winn A, et al. Four core cases: a simulation curriculum for pediatrics residents. MedEdPORTAL. 2014;10 :9943. 10.15766/mep_2374-8265.9943 22. Sawyer T, Creamer K, Puntel R, et al. Pediatric procedural skills training curriculum. MedEdPORTAL. 2010;6 :8094. 10.15766/mep_2374-8265.8094 23. Good RJ, Mashburn D, Jekich E, et al. Simulation-based training for ultrasound-guided central venous catheter placement in pediatric patients. MedEdPORTAL. 2022;18 :11276. 10.15766/mep_2374-8265.11276 36249594 24. Freeman M, Wathen P, Williams J, Zhang M. Teaching incision and drainage of abscess. MedEdPORTAL. 2014;10 :9736. 10.15766/mep_2374-8265.9736
PMC010xxxxxx/PMC10352541.txt
==== Front Acute Med Surg Acute Med Surg 10.1002/(ISSN)2052-8817 AMS2 Acute Medicine & Surgery 2052-8817 John Wiley and Sons Inc. Hoboken 10.1002/ams2.872 AMS2872 AMS-2023-0090.R1 Case Report Case Report Bilateral vocal cord palsy induced by long‐term use of small‐bore nasogastric tube Vocal cord palsy due to nasogastric tube Nihira et al. Nihira Takashi https://orcid.org/0000-0001-8375-7129 1 [email protected] Fukaguchi Kiyomitsu https://orcid.org/0000-0003-2262-1898 1 Taguchi Azusa 1 Fukui Hiroyuki 1 Sekine Ichiro https://orcid.org/0000-0002-0868-8431 1 Yamamoto Daisuke 2 Moriya Hidekazu 3 Yamagami Hiroshi 1 1 Department of Emergency Medicine Shonan Kamakura General Hospital Kamakura Japan 2 Department of Neurology Shonan Kamakura General Hospital Kamakura Japan 3 Department of General Internal Medicine Shonan Kamakura General Hospital Kamakura Japan * Correspondence Takashi Nihira, Department of Emergency Medicine, Shonan Kamakura General Hospital, 1370‐1 Okamoto, Kamakura, Kanagawa 247‐8533, Japan. Email: [email protected] 17 7 2023 Jan-Dec 2023 10 1 10.1002/ams2.v10.1 e87203 6 2023 18 4 2023 25 6 2023 © 2023 The Authors. Acute Medicine & Surgery published by John Wiley & Sons Australia, Ltd on behalf of Japanese Association for Acute Medicine. https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an open access article under the terms of the http://creativecommons.org/licenses/by-nc-nd/4.0/ License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non‐commercial and no modifications or adaptations are made. Abstract Background Nasogastric tube syndrome is a rare but life‐threatening complication of nasogastric tube placement due to acute upper airway obstruction caused by bilateral vocal cord paresis. Case Presentation An 86‐year‐old woman was brought to the emergency department with acute stridor. She had been diagnosed with stroke 106 days prior, and an 8F nasogastric tube was placed on the day following the diagnosis. A laryngeal fiberscopy revealed bilateral laryngeal edema and bilateral vocal cord palsy. Nasogastric tube removal and intubation were carried out, and the stridor disappeared. Two days later, a tracheostomy was performed. Unfortunately, the patient's vocal cord function had not improved at the 1 month follow‐up upon assessment with a laryngeal fiberscope. Conclusion Long‐term small‐bore nasogastric tube placement can cause upper airway obstruction due to bilateral vocal cord palsy. Nasogastric tube syndrome (NGTS) is a rare but lethal complication of nasogastric tube placement. Nasogastric tube syndrome can be fatal due to upper airway obstruction caused by vocal cord paresis. We report a woman who was diagnosed with NGTS caused by long‐term use of small‐bore nasogastric tube. airway obstruction endoscopy intubation tracheostomy vocal cord paralysis source-schema-version-number2.0 cover-dateJanuary/December 2023 details-of-publishers-convertorConverter:WILEY_ML3GV2_TO_JATSPMC version:6.3.2 mode:remove_FC converted:18.07.2023 Nihira T , Fukaguchi K , Taguchi A , Fukui H , Sekine I , Yamamoto D , et al. Bilateral vocal cord palsy induced by long‐term use of small‐bore nasogastric tube. Acute Med Surg. 2023;10 :e872. 10.1002/ams2.872 ==== Body pmcINTRODUCTION Nasogastric tube syndrome (NGTS) is a rare but life‐threatening complication of nasogastric tube placement. Nasogastric tube syndrome was first described by Sofferman et al. in 1990 as the triad of nasogastric tube placement with throat pain and vocal cord paresis. 1 Although its incidence is unknown, NGTS is potentially fatal due to acute upper airway obstruction caused by bilateral vocal cord paresis. The clinical risk of developing NGTS other than nasogastric tube insertion is not clear, and it remains unclear which patients we should be concerned about regarding the development of the disease. In this report, we present a case of NGTS that developed after prolonged placement, despite the nasogastric tube diameter being as small as 8F. This case represents a rare occurrence of NGTS in an adult who underwent placement of a nasogastric tube with a very small diameter. CASE REPORT An 86‐year‐old woman was brought to the emergency department with acute stridor. She had previously been diagnosed with stroke, right‐sided hemiparesis, and aphagia due to left middle cerebral artery occlusion 106 days earlier. Her National Institutes of Health Stroke Scale score had been 16/42. Emergency cerebrovascular mechanical thrombectomy was carried out. Scattered cerebral infarcts were observed in the left middle cerebral artery region. Evaluation after treatment revealed no dysarthria but dysphagia, and an 8F nasogastric tube was inserted for nutrition on the day following the diagnosis. At the time of discharge, her neurological condition was assessed as Glasgow Coma Scale E4V4M6, dysphagia present, no dysarthria, right upper extremity Manual Muscle Test score 1, and right lower extremity Manual Muscle Test score 2. No observations were made regarding the vocal cords. She was transferred to the hospital for rehabilitation for dysphagia and right‐sided paralysis and continued rehabilitation. On presentation, there was no sign of trauma, and her vital signs were as follows: heart rate, 96 b.p.m.; blood pressure, 152/123 mmHg; oxygen saturation, 97% on 2 L oxygen through a nasal cannula; respiratory rate, 18 breaths/min; body temperature, 36.5°C; and Glasgow Coma Scale score, E4V2M6. She did not have throat pain. Laryngeal fiberscopy revealed the nasogastric tube passing through the right side of the hypopharynx, bilateral laryngeal edema without ulcer, and bilateral vocal cord palsy (Figure 1). Nasogastric tube removal and intubation were carried out, and the stridor disappeared. There were no significant changes on neurological examination or head computed tomography; hence, we diagnosed NGTS. Two days later, a tracheostomy was undertaken. Unfortunately, the patient's vocal cord function had not improved at the 1 month follow‐up upon assessment with a laryngeal fiberscope. She was transferred to a rehabilitation hospital on day 58. FIGURE 1 Laryngeal fiberscopy in an 86‐year‐old woman with long‐term small‐bore nasogastric tube placement. (A) Bilateral laryngeal edema. (B) Bilateral vocal cord palsy at inspiratory phase. DISCUSSION Considering the clinical findings of our patient, we discovered the following: (i) long‐term small‐bore nasogastric tube placement can induce NGTS; (ii) NGTS is life‐threatening due to upper airway obstruction with bilateral vocal cord palsy; and (iii) bilateral vocal cord palsy might not improve in all cases. Four mechanisms have been proposed for the pathogenesis of NGTS: (i) the rubbing force between the fixed tube and movable structures of the larynx; (ii) gravitational compression on the tube by the cricoid cartilage against the vertebral column in the supine position; and (iii) the pressure of tonic contraction of the cricopharyngeal muscle on the tube. 1 Ulcers formed through these mechanisms are assumed to cause inflammation around the posterior cricoarytenoid muscle, resulting in vocal cord palsy. The fourth mechanism proposed by Isozaki et al. concerning cases without ulcer formation is that posterior cricoarytenoid muscle ischemia results from compression. 2 Considering these hypotheses, it has been speculated that tube diameter influences the pathogenesis of NGTS. 3 The nasogastric tube sizes reported previously were 16–18F, and it is not clear how large a tube should be to induce NGTS. In addition to the size, the material of the nasogastric tube is suspected to affect NGTS because of plasticizer elution, which alters tube flexibility. 3 Therefore, it is possible that the onset of NGTS cannot be avoided even with a small diameter of 8F. In addition, given that the involvement of size is relative, the fact that the patient was small and thin (150 cm, 45 kg) might have contributed to NGTS development. In this case, a polyurethane feeding tube without a plasticizer was used. Although NGTS commonly occurs between 12 h and 2 weeks after nasogastric tube insertion, 4 only three cases were diagnosed between 2 weeks and 2 years after nasogastric tube insertion. 2 Similarly, our patient had a prolonged NGTS development time of 105 days. The mechanism by which NGTS occurred in this case is assumed to be as follows. As the posterior cricoarytenoid muscle is a thin tissue that exists behind the cricoid cartilage, even with a small tube diameter, the repeated rubbing force of the tube over a long period of time, and the pressure exerted by the cricoid cartilage, vertebral body, and nasogastric tube in the supine position could have disrupted the perfusion of the posterior cricoarytenoid muscle, and induced muscle ischemia, as proposed by Isozaki et al. 2 Our case is unique in that it developed after long‐term implantation, despite using the smallest‐bore soft tube in an adult patient. 1 , 3 , 5 These facts indicate that NGTS is an unavoidable complication of nasogastric tube insertion. This case suggests that NGTS might not be avoidable, even with small‐diameter tubes. As a mechanism has been proposed in which tube stimulation is the trigger, shortening the duration of nasogastric indwelling might be effective in reducing the chances of stimulation and preventing NGTS. Clinicians should be aware of the risk of NGTS development in all patients requiring nasogastric tube placement and should undertake appropriate observations. Nasogastric tube syndrome could be fatal and is an acute complication due to bilateral vocal cord palsy, 77% of which requires tracheostomy. Given that only diagnosed cases have been reported, there could be undiagnosed cases that were treated as sudden deaths. Although most NGTS patients have good vocal cord function recovery within 2 months, some cases in which there was no improvement have been reported. 4 , 6 , 7 This difference might be due to pathological variations. Differences in prognosis according to pathophysiology remain unelucidated. Nasogastric tube placement is one of the most commonly performed procedures. Nasogastric tube syndrome is caused by nasogastric tube placement and can lead to upper airway obstruction due to vocal cord palsy. Airway obstruction can be fatal within minutes; thus, patients with suspected NGTS should be treated by an experienced medical team for airway management. Considering that palsy might be irreversible, physicians should be aware of the risks of NGTS and carry out appropriate observations for patients who undergo nasogastric tube placement. CONCLUSION Long‐term small‐bore nasogastric tube placement can cause upper airway obstruction due to bilateral vocal cord palsy. CONFLICT OF INTEREST STATEMENT The authors declare no conflicts of interest. ETHICS STATEMENT Approval of the research protocol: This report has been approved by our institutional Ethical Committee. The approval number is SKEC‐23‐6. Informed consent: Written informed consent was obtained from the patient to publish this case report and accompanying images. Registry and registration no. of the study/trial: N/A. Animal studies: N/A. DATA AVAILABILITY STATEMENT Data sharing is not applicable to this article as no new data were created or analyzed in this study. ==== Refs REFERENCES 1 Sofferman RA , Haisch CE , Kirchner JA , Hardin NJ . The nasogastric tube syndrome. Laryngoscope. 1990;100 (9 ):962–8.2395406 2 Isozaki E , Tobisawa S , Naito R , Mizutani T , Hayashi H . A variant form of nasogastric tube syndrome. Intern Med. 2005;44 (12 ):1286–90.16415551 3 Kanbayashi T , Tanaka S , Uchida Y , Hatanaka Y , Sonoo M . Nasogastric tube syndrome: the size and type of the nasogastric tube may contribute to the development of nasogastric tube syndrome. Intern Med. 2021;60 (12 ):1977–9.33518566 4 Brousseau VJ , Kost KM . A rare but serious entity: nasogastric tube syndrome. Otolaryngol Head Neck Surg. 2006;135 (5 ):677–9.17071292 5 Nehru VI , Al Shammari HJ , Jaffer AM . Nasogastric tube syndrome: the unilateral variant. Med Princ Pract. 2003;12 (1 ):44–6.12566968 6 Nayak G , Virk RS , Singh M , Singh M . Nasogastric tube syndrome: a diagnostic dilemma. J Bronchology Interv Pulmonol. 2018;25 (4 ):343–5.29771772 7 Apostolakis LW , Funk GF , Urdaneta LF , McCulloch TM , Jeyapalan MM . The nasogastric tube syndrome: two case reports and review of the literature. Head Neck. 2001;23 (1 ):59–63.11190859
PMC010xxxxxx/PMC10352544.txt
==== Front Clin Case Rep Clin Case Rep 10.1002/(ISSN)2050-0904 CCR3 Clinical Case Reports 2050-0904 John Wiley and Sons Inc. Hoboken 10.1002/ccr3.7134 CCR37134 CCR3-2022-12-2547.R1 Case Report Case Report A case of high‐risk AML in a patient with advanced systemic mastocytosis Fazio et al. Fazio Manlio https://orcid.org/0000-0001-8804-2640 1 Vetro Calogero 2 Markovic Uroš https://orcid.org/0000-0003-0868-5676 2 Duminuco Andrea 1 Parisi Marina 2 Maugeri Cinzia 2 Mauro Elisa 2 Parrinello Nunziatina Laura 2 Stagno Fabio 2 Villari Loredana 3 Triolo Anna Maria 2 Stella Stefania 4 Palumbo Giuseppe A. 1 5 6 Di Raimondo Francesco 2 5 Romano Alessandra https://orcid.org/0000-0002-6333-4433 5 [email protected] Zanotti Roberta https://orcid.org/0000-0003-2871-9730 7 1 Post Graduation School of Hematology University of Catania, A.O.U. Policlinico “G.Rodolico‐San Marco” Catania Italy 2 Division of Hematology A.O.U. Policlinico “G.Rodolico‐San Marco” Catania Italy 3 Department of Pathology A.O.U. Policlinico di Catania Catania Italy 4 Center of Experimental Oncology and Hematology A.O.U. Policlinico “G. Rodolico‐San Marco” Catania Italy 5 Dipartimento di specialità Medico‐Chirurgiche, CHIRMED, sezione di Ematologia Università degli Studi di Catania Catania Italy 6 Dipartimento di Scienze Mediche, Chirurgiche e Tecnologie Avanzate “G.F. Ingrassia” University of Catania Catania Italy 7 Department of Medicine, Hematology Unit Azienda Ospedaliera Universitaria Integrata di Verona Verona Italy * Correspondence Alessandra Romano, Dipartimento di specialità Medico‐Chirurgiche, CHIRMED, sezione di Ematologia, Università degli Studi di Catania, Catania, Italy. Email: [email protected] 17 7 2023 7 2023 11 7 10.1002/ccr3.v11.7 e713401 3 2023 22 12 2022 13 3 2023 © 2023 The Authors. Clinical Case Reports published by John Wiley & Sons Ltd. https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an open access article under the terms of the http://creativecommons.org/licenses/by-nc-nd/4.0/ License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non‐commercial and no modifications or adaptations are made. Abstract Aggressive SM + AML has limited therapeutic options. Even a strong combination of decitabine–venetoclax–midostaurin has a transient effect on AML and a mitigated effect on SM. Larger series are required to identify the best therapeutic strategy. Aggressive SM + AML has limited therapeutic options. Even a strong combination of decitabine–venetoclax–midostaurin has a transient effect on AML and a mitigated effect on SM. Larger series are required to identify the best therapeutic strategy. advanced systemic mastocytosis (AdvSM) cladribine concurrent hematologic disease decitabine–venetoclax (DEC‐VEN) HAM regimen refractory acute myeloid leukemia (R‐AML) source-schema-version-number2.0 cover-dateJuly 2023 details-of-publishers-convertorConverter:WILEY_ML3GV2_TO_JATSPMC version:6.3.2 mode:remove_FC converted:18.07.2023 Fazio M , Vetro C , Markovic U , et al. A case of high‐risk AML in a patient with advanced systemic mastocytosis. Clin Case Rep. 2023;11 :e7134. doi:10.1002/ccr3.7134 ==== Body pmc1 INTRODUCTION Systemic mastocytosis (SM) can be associated with another hematologic neoplasia including acute myeloid leukemia (AML). Adopting triplet combination for both diseases with decitabine–venetoclax–midostaurin could be a valid option. However, even if leukemia responds, mastocytosis can persist with a lower mutational burden, eventually prompting AML relapse with a very dismal prognosis. Systemic mastocytosis (SM) is a rare disease that derives from an abnormal proliferation of clonal mast cells (MCs) in extracutaneous organs such as bone marrow (BM), lymph nodes, spleen, liver, bones, and gastrointestinal tract. The affected patients are at high risk of suffering from life‐threatening anaphylaxis and may develop even a MC activation syndrome (MCAS). According to the WHO classification, the diagnosis of SM can be established when either one major and one minor or at least three minor criteria are fulfilled. 1 SM is an umbrella term which encompasses a wide spectrum of clinical entities, from indolent (ISM) and smoldering (SSM) to advanced SM (AdvSM). 2 SSM is characterized by the presence of two or more B‐findings which are indicative of high MCs burden but without organ damage (>30% MCs in the BM and/or serum tryptase >200 μg/dL and/or cKITD816V mutation with a variant allele frequency [VAF] ≥10%; hypercellular and dysplastic BM without criteria for myelodysplastic syndrome [MDS]; hepatomegaly or splenomegaly in the absence of functional impairment). AdvSM includes: aggressive SM (ASM) which has a median overall survival (OS) of 41 months, SM with associated non‐mast cell hematologic neoplasm (SM‐AHN) which has a median OS of 24 months, 3 and mast cell leukemia (MCL) which has a median OS of 2 months. ASM is characterized by the presence of one or more C‐findings which are indicative of organ damage (cytopenia, hepatomegaly and splenomegaly with functional impairment, intestinal malabsorption with hypoalbuminemia and weight loss, osteolytic lesions with pathological fractures). 1 AHNs derive from the myeloid lineage in 80%–90% of cases and chronic myelomonocytic leukemia (CMML) is the most frequently associated one. 4 The most common mutations detected in SM are somatic activating mutations in cKIT, most notably cKITD816V. 5 These mutations can also be detected in the concomitant myeloid (but not in lymphoid) AHNs. Moreover, most patients with AdvSM show additional somatic mutations both on SM and AHN cells (most frequently SRSF2, ASXL1, RUNX1 [S/A/R], TET2, and JAK2). 6 This mutational multilineage involvement strengthens the hypothesis of a clonal relationship between these two diseases. 7 Midostaurin is a multikinase inhibitor that targets both wild‐type and D816V‐mutated c‐KIT and has been largely approved for the treatment of both FLT3‐mutated acute myeloid leukemia (AML) and AdvSM. 8 2 CASE PRESENTATION A 69‐year‐old Caucasian man presented at our center in March 2017 due to neutrophilic leukocytosis (white blood cell [WBC] count 16.380 cells/μL) and normal Hb and platelet values, along with maculo‐papular cutaneous lesions dating from 2010. A skin biopsy performed shortly after confirmed the diagnosis of cutaneous mastocytosis (CM). In past medical history, the patient referred hypertension, glucose intolerance. No allergies or syncopal or other mediator‐related symptoms were reported. BM histological analysis revealed the presence of aggregates with >15 atypical mast cell CD117+, CD34−, CD25+, tryptase+, consensual to reticulin fibrosis and focally collagen (Grade 2–3). cKITD816V mutation was demonstrated in peripheral blood (PB) (VAF: 12%). 9 BM karyotype was normal. Baseline serum tryptase level was 184 μg/dL (n.r. 0.1–11.4 μg/dL); normal alkaline phosphatase (ALP); increased β2 microglobulin serum levels. The further presence of more than 90% neutrophils (N) in PB together with predominant maturing granulopoiesis in the BM histological sample was suggestive of a concomitant chronic neutrophilic leukemia (CNL). Twenty‐seven different myeloid genes were analyzed in PB, using next‐generation sequencing (NGS), including BCR‐ABL1, FIPIL1‐PDGFRα, JAK2V617F, MPL, CARL, with no evidence of mutations including that of the CSF3R gene, characteristic of CNL. Lastly, the absence of ≥10% peripheral monocytes excluded the hypothesis of a CMML. 4 Osteopenia was documented by DEXA (T score −1.8 at lumbar spine) and diffuse osteosclerosis with micro osteolysis at skull was documented by low‐dose CT scans. Abdominal ultrasound showed only mild hepatomegaly. Thus, the final diagnosis was ISM associated with MPN/MDS overlap syndrome (ISM‐AHN) 3 (Figure 1A–F). The patient was started on H1/H2 antagonist treatment and vitamin D supplementation together with hydroxyurea (HU, 500 mg bis in die) and maintained good general conditions for 4 years with moderate leukocytosis and decrease in serum tryptase level (from initial 184 to 73 μg/dL). FIGURE 1 Immunohistochemical features of SM‐AHN (A–F). Photomicrograph of the bone marrow histology and immunohistochemistry supporting the diagnosis of SM‐AHN. SM and MDS/MPN: hypercellularity, granulocytic lineage expansion and paratrabecular infiltration of mast cells (MCs) (arrow) with the characteristic delicate fibrosis (arrowed), hematoxylin and eosin (H&E) stain (A); paratrabecular infiltration of MCs. The mast cell granules are well highlighted by the metachromatic staining with toluidine blue (B); cellularity predominantly determined by granulopoiesis expressing CD15 (C); positive immunostaining (brown) of spindle‐shaped MCs with CD 117 (D), CD25 (E), and tryptase (F). Morphology (G–I). AML: blasts and neutrophils (leukemic hiatus) (G, H). This PB smear was made after the first cycle of venetoclax and shows blast clearance with the persistence of pathologic spindle‐shaped MCs (in the picture) (I). Follow‐up histological BM analysis performed in 2020 showed an increase in MC infiltration (30%–35%) with neoplastic immunophenotype and <5% of CD34+ blasts. Routine low‐dose CT scans revealed osteosclerosis, and the abdomen ultrasound revealed liver steatosis and mild splenomegaly, in absence of cytopenia. These results documented the progression from ISM to SSM. 3 In March 2021 the patient presented at our center with weakness, loss of sense of well‐being and weight loss (3 kg in the last 2 months). The complete blood count (CBC) showed 218,000 WBC/μL (N 43.5%, lymphocytes 11.9%, monocytes 24.6% and blasts 20% on PB differential count) without severe anemia or thrombocytopenia. The evolution of MPN/MDS syndrome to AML was confirmed at PB flow cytometry: presence of 23% blasts with a CD33+, HLA‐DR+, CD34+, and CD117+ immunophenotype (Figure 1G,H). BM biopsy was not performed due to patient refusal. Physical exam revealed the increase in both splenomegaly and hepatomegaly (palpable from costal margin at 6 and 3 cm, respectively). Laboratory investigations showed serum tryptase >200 μg/dL; NGS study on PB showed an increase in cKITD816V VAF (38%) along with an additional high risk RUNX‐1 I342 K mutation (VAF: 5%). 10 The presence of Grade 2 hypoalbuminemia (2.22 g/dL—n.r. 3.50–5.20) were considered related to SM and described as C‐findings, hence confirming the evolution from SSM to ASM with a final diagnosis of ASM associated with secondary AML. 3 We decided to treat the patient with the decitabine–venetoclax combination (DEC‐VEN) as a first‐line therapy for AML 11 and midostaurin as a first‐line therapy for ASM. We did not choose chemotherapy because of age and patient refusal to be hospitalized. In the absence of a sibling donor, we activated unrelated donor research. Due to significant leukocytosis and the high risk of tumor lysis syndrome, venetoclax was initially omitted, and decitabine was administered (20 mg/m2 i.v. for 5 days/cycle) associated with HU (1000 mg/day). After two cycles of decitabine and HU, the WBC was 9220 cells/μL: thereafter, the patient was treated with a third cycle of therapy with venetoclax (400 mg daily) for 28 days/cycle and decitabine (20 mg/m2) for 5 days/cycle. Midostaurin was initially imbricated with decitabine and HU and administered at a daily dose of 200 mg but was progressively reduced due to hematologic toxicity and suspended before venetoclax imbrication because of progressive symptomatic thrombocytopenia. After the third cycle of therapy including venetoclax, a BM biopsy was performed that revealed a total blast clearance (morphologic leukemia‐free state MLFS), although pathological CD25+ MC infiltration was stable (30%) (Figure 1I). Despite this, the cKITD816V mutation burden appeared reduced (10%), albuminemia returned within normal ranges, tryptase levels dropped to 84.3 μg/dL, and at physical examination spleen and liver appeared completely reseized. Thus, SM was deemed in clinical improvement by combined consensus IWG‐MRT‐ECNM response criteria. 12 He subsequently received a fourth cycle of therapy with DEC‐VEN combination but venetoclax was temporarily suspended throughout the cycle in order to allow neutrophil recovery. Unfortunately, blood examinations repeated 1 month later documented AML relapse with peripheral leukocytosis (WBC 67,880/μL) and 30% blasts on PB differential count; serum tryptase level was 91.3 μg/dL; spleen and liver appeared enlarged at manual palpation (4 and 2 cm from costal margin, respectively); albumin level was at the lower limit. Once the patient was again debulked with HU, another cycle of DEC‐VEN was administered, associated with midostaurin at half dosage (100 mg/die). Despite an initial benefit, the patient developed refractoriness to this regimen. Abdominal examination revealed a massive splenomegaly of 8 cm below the left costal margin, and a PB smear showed 60% blasts. As a consequence, the patient was hospitalized in order to undergo salvage chemotherapy as a possible bridge to allogeneic stem cell transplantation (allo‐HSCT). The patient was treated with HAM regimen 13 in association with cladribine. 14 Our intention was to treat both pathologies considering that cladribine is effective both on R/R‐AML 15 and AdvSM. 16 The scheme we administered consisted of: Ara‐C 2000 mg i.v. bis in die (from Day +1 to Day +4), mitoxantrone 10 mg/die i.v. (from Day +3 to Day +5), and cladribine 7 mg/die subcutaneously (from Day +1 to Day +5). Dose was calculated on ideal weight and reduced due to age and performance status. For the prophylaxis of HD‐Ara‐C‐induced photophobia and conjunctivitis, he received glucocorticoid eye drops every 6 h starting before the first dose and continuing for 24 h after the last dose of HD‐Ara‐C. Despite achieving a decreased serum tryptase level (from 69.9 at the admission to 54.8 μg/dL) and organomegaly reduction, the patient experienced prolonged Grade 4 pancytopenia (Hb <7 g/dL, PLT <2000/ μL, WBC <100/ μL), which required constant red blood cell and platelet transfusions. Furthermore, he developed a septic state that led to exitus on Day +18 from the HAM treatment start (Figure 2). FIGURE 2 Graphs representing the trend of white blood cell count (A), platelet count (B), and tryptase level (C) from SM‐AHN diagnosis in 2017 until exitus in 2021. The graphs reflect the patient's measured platelet count (PLT), white blood cell count (WBC), and tryptase level after treatment with HU and H1/H2 antagonists, MDS/MPN evolution to AML and SM‐AML I and II line therapies. AML, acute myeloid leukemia; C, cycle; DEC, decitabine; D, day; GCSF, granulocyte colony‐stimulating factor; HU, hydroxyurea; MID, midostaurine; MLFS, morphologic leukemia‐free state; SM‐AHN, systemic mastocytosis with associated hematologic neoplasm; VEN, venetoclax. 3 DISCUSSION The association between AML, especially cKIT D816 mut /CBF neg AML, and SM determines a particularly dismal prognosis. A retrospective analysis published in 2019 reported 40 KIT D816 mut /CBF neg SM‐AML patients (29 with secondary AML evolved from SM ± associated myeloid neoplasm). The median OS of the 40 SM‐AML patients was 5.4 months and thus even worse as compared to patients with MCL, which is defined by the presence of ≥20% MCs in a BM smear. SM‐AML patients treated with intensive chemotherapy (including HAM regimen) ± allo‐HSCT achieved a higher OS (16.7 months) than those who received non‐intensive treatment as hypomethylating agents (HMAs) ± cladribine (2.7 months). HMAs do not induce any complete remission (CR) in SM‐AML and none of the patients was treated with midostaurin. 17 The association of HMAs with venetoclax could achieve a CR/sustained response (SR) in both diseases, even with different response time, as reported in some case reports. 18 , 19 In none of them midostaurin was imbricated to the scheme. Venetoclax is a Bcl‐2 inhibitor with confirmed efficacy in combination with HMAs in patients with previously untreated AML who were ineligible for intensive induction therapy. The HMAs‐VEN regimen had significant improvement in terms of OS, CR, and duration of response. 20 High expression of Bcl‐2 is also typical in AdvSM, including SM‐AML. Considering that neoplastic MCs and AML derive from the same malignant clone and that leukemia stem cells are Bcl‐2+, venetoclax can prevent MC differentiation by targeting Bcl‐2 in the neoplastic progenitors. 18 The expression of Bcl‐2 determines partial resistance to midostaurin monotherapy in AdvSM. According to the study D2201, midostaurin is the only approved agent in first‐line therapy for AdvSM with KITD816V mutation, achieving mainly partial (PR) or major responses (MR) that are not sustained. 21 A recent preliminary study demonstrated that midostaurin restores apoptotic dependency to Bcl‐2 in MCL‐like cells and justified a possible employment of midostaurin associated with venetoclax front‐line therapy in mast cell tumors. 22 Our patient obtained MLFS and SM clinical improvement after two cycles of decitabine associated with midostaurin and one cycle of DEC‐VEN. Unfortunately, midostaurin administration necessitated interruption before the introduction of venetoclax due to hematologic toxicity. Nevertheless, we decided to reintroduce midostaurin at half dosage during the fifth cycle of DEC‐VEN, after AML relapse, motivated by the persistence of pathological MCs in the BM. Similar cases of SM‐AML have been reported in the literature describing the permanence of a residual mast cell infiltrate in the BM despite AML remission with an induction regimen or following stem cell allograft. 23 Most importantly, MCs release numerous factors (histamine, heparin, tryptase, cytokines) in the microenvironment, thus enhancing tumor‐associated angiogenesis. 24 As a consequence, the residual MC infiltrates potentially create a favorable substrate for further progression or the recurrence of the associated hematologic neoplasms. 25 Within a short time the patient became refractory to the triplet (DEC‐VEN‐MIDO) and was immediately hospitalized in order to undergo salvage therapy with HAM regimen (for R/R‐AML) associated with cladribine (both for AdvSM and R/R‐AML) as a bridge to transplant. 13 Relapsed/refractory (R/R) AML remains an impervious hematologic challenge with generally poor outcomes and despite the existence of a wide armamentarium of second‐line regimens, none of them is significantly beneficial compared to the others. 26 A German clinical phase I/II study investigated the tolerability and efficacy of high‐dose cytosine arabinoside and mitoxantrone (HAM) in heavily pretreated patients with advanced refractory disease. A total of 40 patients was recruited in the study (median age 45 years); 21 patients (53%) achieved CR (19 of them after one HAM course and two of them after two HAM courses); one additional patient obtained a PR (normalization of PB counts but with residual 15% blasts in the BM). Most common non‐hematologic toxicity was represented by nausea, vomiting, mucositis, and diarrhea. These results demonstrated that the HAM regimen could achieve a high ORR in a very unfavorable category of patients and suggested it should be employed in earlier stages of the disease. 13 Cladribine is a polyvalent antimetabolite which has been used for both R/R‐AML and SM. It is an adenosine‐deaminase resistant analog of adenosine which is activated inside the cells and induces apoptosis in hematopoietic cells, leukemic cells, and lymphatic malignancies. Several reports describe how cladribine induces CRs (100% regression) in cutaneous forms and major (complete resolution of at least one C‐findings and no progression of other C‐findings) or PRs in indolent and AdvSM. 16 On the other hand, a meta‐analysis of 10 prospective studies evaluated the role of cladribine in R/R‐AML showing a significant prognostic improvement, especially when combined with cytarabine in cladribine/cytarabine/G‐CSF/mitoxantrone regimen (CLAG‐M). 15 The reason for this synergistic effect is due to the fact that cladribine increases cytarabine uptake in leukemia cells and the accumulation of its active cytotoxic metabolite (Ara‐CTP). 27 The most common toxicities caused by cladribine are myelosuppression (severe neutropenia and thrombocytopenia), immunosuppression (low TCD4+ and TCD8+ levels) and subsequent opportunistic infections (Grades 3 and 4), the latter being the main cause of early death as in this case. 15 4 CONCLUSION The employment of an HMA‐venetoclax regimen in combination with midostaurin (or other KIT inhibitors) 5 , 28 could be an option for patients with AML associated with SM, but data about this combination, dosage of midostaurin and their toxicity are lacking. However, in our case, MLFS was achieved, but pathologic mastocyte infiltration appeared stable and only clinical improvement of SM was obtained. The efficacy and safety of the HAM regimen combined with cladribine in R/R‐AML have been documented, but there are no cases in the literature about the employment of this therapeutic scheme in SM associated with R/R‐AML. Unfortunately, our attempt was not valuable. More specifically, even though treatment allowed to decrease serum tryptase level, the patient developed Grade 4 neutropenia and died because of septic shock during aplasia. For these reasons, further studies are necessary to understand the validity of the abovementioned therapeutic combinations in this particular set of patients. AUTHOR CONTRIBUTIONS Manlio Fazio: Conceptualization; writing – original draft. Calogero Vetro: Conceptualization; data curation. Uros Markovic: Writing – review and editing. Andrea Duminuco: Writing – review and editing. Marina Silvia Parisi: Investigation. Cinzia Maugeri: Investigation. Elisa Mauro: Investigation. Nunziatina Laura Parrinello: Investigation. Fabio Stagno: Investigation. Loredana Villari: Investigation. Anna Maria Triolo: Investigation. Stefania Stella: Investigation. Giuseppe Palumbo: Supervision. Francesco Di Raimondo: Supervision. Alessandra Romano: Conceptualization; data curation; investigation; project administration; writing – review and editing. Roberta Zanotti: Data curation; supervision; writing – review and editing. ACKNOWLEDGMENTS Authors thank AIL (Associazione Italiana Leucemia e Linfomi) Sezione di Catania and ASIMAS (Associazione Italiana Mastocitosi) for their commitment to research and support to people affected by these diseases. CONFLICT OF INTEREST STATEMENT The authors declare that they had no sources of funding for this study and no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. ETHICS STATEMENT This study was conducted in accordance with the Declaration of Helsinki. The authors obtained written consent to use personal data of the patient for the publication of this case report. DATA AVAILABILITY STATEMENT All data analyzed during this study are included in this article. ==== Refs REFERENCES 1 Arber DA , Orazi A , Hasserjian RP , et al. International consensus classification of myeloid neoplasms and acute leukemias: integrating morphologic, clinical, and genomic data. Blood. 2022;140 (11 ):1200‐1228.35767897 2 Valent P , Akin C , Sperr WR , et al. New insights into the pathogenesis of mastocytosis: emerging concepts in diagnosis and therapy. Annu Rev Pathol. 2023;18 (1 ):361‐386.36270293 3 Pardanani A . Systemic mastocytosis in adults: 2021 update on diagnosis, risk stratification and management. Am J Hematol. 2021;96 (4 ):508‐525.33524167 4 Valent P , Orazi A , Savona MR , et al. Proposed diagnostic criteria for classical chronic myelomonocytic leukemia (CMML). CMML Variants and Pre‐CMML Conditions. Vol 104 . Haematologica. Ferrata Storti Foundation; 2019:1935‐1949. 5 Verstovsek S . Advanced systemic mastocytosis: the impact of KIT mutations in diagnosis, treatment, and progression. Eur J Haematol. 2013;90 :89‐98.23181448 6 Reiter A , George TI , Gotlib J . New Developments in Diagnosis, Prognostication, and Treatment of Advanced Systemic Mastocytosis [Internet]. 2020. Available from:http://ashpublications.org/blood/article‐pdf/135/16/1365/1724162/bloodbld2019000932c.pdf 7 Sotlar K , Colak S , Bache A , et al. Variable presence of KIT D816V in clonal haematological non‐mast cell lineage diseases associated with systemic mastocytosis (SM‐AHNMD). J Pathol. 2010;220 (5 ):586‐595.20112369 8 Gallogly MM , Lazarus HM , Cooper BW . Midostaurin: a novel therapeutic agent for patients with FLT3‐mutated acute myeloid leukemia and systemic mastocytosis. Ther Adv Hematol. 2017;8 (9 ):245‐261.29051803 9 Khoury JD , Solary E , Abla O , et al. The 5th edition of the World Health Organization Classification of Haematolymphoid Tumours: Myeloid and Histiocytic/Dendritic Neoplasms. Leukemia. 2022;36 (7 ):1703‐1719.35732831 10 Ni ZF , Ma LJ , Shi LL , Shen PL , Zhao JQ . Clinical characteristics of acute myeloid leukemia patients with RUNX1 gene mutation. Zhongguo Shi Yan Xue Ye Xue Za Zhi. 2021;29 (5 ):1411‐1416.34627418 11 Dinardo CD , Pratz K , Pullarkat V , et al. Venetoclax Combined with Decitabine or Azacitidine in Treatment‐Naive, Elderly Patients with Acute Myeloid Leukemia [Internet]. 2019 Available from: www.abbvie.com/our‐science/clinical‐trials/clinical‐trials‐data‐and 12 Gotlib J , Pardanani A , Akin C , et al. Perspectives International Working Group‐Myeloproliferative Neoplasms Research and Treatment (IWG‐MRT) & European Competence Network on Mastocytosis (ECNM) Consensus Response Criteria in Advanced Systemic Mastocytosis. 2013 Available from: http://ashpublications.org/blood/article‐pdf/121/13/2393/1364345/2393.pdf 13 Hiddemann W , Kreutzmann H , Straif K , et al. High‐dose cytosine arabinoside and mitoxantrone: a highly effective regimen in refractory acute myeloid leukemia. Blood. 1987 Mar;69 (3 ):744‐749.3469002 14 Jaglal MV , Duong VH , Bello CM , et al. Cladribine, cytarabine, filgrastim, and mitoxantrone (CLAG‐M) compared to standard induction in acute myeloid leukemia from myelodysplastic syndrome after azanucleoside failure. Leuk Res. 2014;38 (4 ):443‐446.24439565 15 Zhou A , Han Q , Song H , Zi J , Ma J , Ge Z . Efficacy and toxicity of cladribine for the treatment of refractory acute myeloid leukemia: a meta‐analysis. Drug Des Devel Ther. 2019;13 :1867‐1878. 16 Marlies E , Van H . Cladribine in the treatment of systemic mastocytosis, a review of the literature. Int J Rare Dis Disord. 2019;2 (1 ):1‐5. 17 Jawhar M , Döhner K , Kreil S , et al. KIT D816 mutated/CBF‐negative acute myeloid leukemia: a poor‐risk subtype associated with systemic mastocytosis. Leukemia. 2019;33 (5 ):1124‐1134.30635631 18 Pourhassan H , Kim Y , Pullarkat V . Clearance of bone marrow mast cells after hypomethylating agent and venetoclax for systemic mastocytosis associated with myeloid neoplasia. Eur J Haematol [Internet]. 2022;https://onlinelibrary.wiley.com/doi/10.1111/ejh.13894;110 :213‐216.36335580 19 Atieh T , Woodroof J , Yacoub A . Venetoclax and decitabine show robust activity in advanced systemic mastocytosis. Blood. 2020;136 (Supplement 1 ):25‐26. 20 DiNardo CD , Jonas BA , Pullarkat V , et al. Azacitidine and venetoclax in previously untreated acute myeloid leukemia. N Engl J Med. 2020;383 (7 ):617‐629.32786187 21 DeAngelo DJ , George TI , Linder A , et al. Efficacy and safety of midostaurin in patients with advanced systemic mastocytosis: 10‐year median follow‐up of a phase II trial. Leukemia. 2018;32 (2 ):470‐478.28744009 22 Canioni D , Moraly J , Lhermitte L , et al. BCL‐2 is expressed in advanced mastocytosis and midaustorin induces venetoclax sensitivity of mast leukemia cell lines. Blood. 2019;13 (134 Supplement_1 ):1683. 23 Zhrebker L , Cooper B , Krause JR . Systemic mastocytosis with associated acute myelogenous leukemia. Bayl Univ Med Cent Proc. 2014;27 (1 ):22‐24. 24 Xu P , Zhang C , Wang Y , et al. Increased number of mast cells in the bone marrow of chronic myeloid leukemia may herald the pending myeloid transformation—the mast cell is an indicator of myeloid transformation. Transl Cancer Res. 2019;8 (5 ):2121‐2129.35116962 25 Mehtani D , Puri N . Steering mast cells or their mediators as a prospective novel therapeutic approach for the treatment of hematological malignancies. Front Oncol. 2021;24 :11. 26 Megías‐Vericat JE , Martínez‐Cuadrón D , Sanz MÁ , Montesinos P . Salvage regimens using conventional chemotherapy agents for relapsed/refractory adult AML patients: a systematic literature review. Ann Hematol. 2018;97 (7 ):1115‐1153.29680875 27 Kornblau SM , Gandhi V , Andreeff HM , et al. Clinical and laboratory studies of 2‐chlorodeoxyadenosine +/− cytosine arabinoside for relapsed or refractory acute myelogenous leukemia in adults. Leukemia. 1996 Oct;10 (10 ):1563‐1569.8847890 28 Dhillon S . Avapritinib: first approval. Drugs. 2020;80 (4 ):433‐439.32100250
PMC010xxxxxx/PMC10352545.txt
==== Front Acute Med Surg Acute Med Surg 10.1002/(ISSN)2052-8817 AMS2 Acute Medicine & Surgery 2052-8817 John Wiley and Sons Inc. Hoboken 10.1002/ams2.871 AMS2871 AMS-2022-0239.R3 Original Article Original Articles Association of obesity paradox with prognosis of veno‐venous‐extracorporeal membrane oxygenation in patients with coronavirus disease 2019 Obesity paradox in patients with COVID‐19 Honzawa et al. Honzawa Hiroshi https://orcid.org/0000-0003-3337-3711 1 [email protected] Taniguchi Hayato https://orcid.org/0000-0001-6995-3529 2 Ogawa Fumihiro 1 Oi Yasufumi https://orcid.org/0000-0003-1956-9662 1 Abe Takeru https://orcid.org/0000-0003-3496-1953 2 Takeuchi Ichiro 1 2 1 Emergency Care Department Yokohama City University Hospital Yokohama Japan 2 Advanced Critical Care and Emergency Center Yokohama City University Medical Center Yokohama Japan * Correspondence Hiroshi Honzawa, Emergency Care Department, Yokohama City University Hospital, 3‐9 Fukuura, Kanazawa‐ku, Yokohama City, Kanagawa 236‐0004, Japan. Email: [email protected] 17 7 2023 Jan-Dec 2023 10 1 10.1002/ams2.v10.1 e87115 6 2023 05 1 2023 17 6 2023 © 2023 The Authors. Acute Medicine & Surgery published by John Wiley & Sons Australia, Ltd on behalf of Japanese Association for Acute Medicine. https://creativecommons.org/licenses/by-nc/4.0/ This is an open access article under the terms of the http://creativecommons.org/licenses/by-nc/4.0/ License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes. Abstract Aim Although the obesity paradox is known for various diseases, including cancer and acute respiratory distress syndrome, little is known about veno‐venous extracorporeal membrane oxygenation (VV‐ECMO) in patients with coronavirus disease 2019 (COVID‐19). In this study, we aimed to investigate the association between body mass index (BMI) and prognosis in critical patients with COVID‐19 requiring VV‐ECMO. Methods We conducted a retrospective observational single‐center study at Yokohama City University Civic General Medical Center between March 2020 and October 2021. Participants were patients with COVID‐19 who required VV‐ECMO. They were classified into two groups: BMI ≤30 kg/m2 and >30 kg/m2. Results In total, 23 patients were included in the analysis, with a median BMI of 28.7 kg/m2. Overall, 22 patients were successfully weaned from the ECMO. When comparing the two groups, there was a trend toward fewer days from onset to ECMO induction in the BMI >30 kg/m2 group. Moreover, the two groups had a similar prognosis. There were no statistically significant differences in the number of days from onset to hospitalization or the duration of ECMO induction between the groups. Conclusion VV‐ECMO induction for patients with COVID‐19 may lead to earlier indications in patients with BMI >30 kg/m2 than in those with BMI ≤30 kg/m2. In this study, we aimed to investigate the association between body mass index (BMI) and prognosis in critical patients with coronavirus disease 2019 (COVID‐19) requiring veno‐venous extracorporeal membrane oxygenation (VV‐ECMO). We conducted a retrospective observational single‐center study. Participants were patients with COVID‐19 who required VV‐ECMO. VV‐ECMO induction for patients with COVID‐19 may lead to earlier indications in patients with BMI >30 kg/m2 than in those with BMI ≤30 kg/m2. acute respiratory distress syndrome body mass index coronavirus disease 2019 obesity paradox veno‐venous‐extracorporeal membrane oxygenation JSPS KAKENHI21K09026 source-schema-version-number2.0 cover-dateJanuary/December 2023 details-of-publishers-convertorConverter:WILEY_ML3GV2_TO_JATSPMC version:6.3.2 mode:remove_FC converted:18.07.2023 Honzawa H , Taniguchi H , Ogawa F , Oi Y , Abe T , Takeuchi I . Association of obesity paradox with prognosis of veno‐venous‐extracorporeal membrane oxygenation in patients with coronavirus disease 2019. Acute Med Surg. 2023;10 :e871. 10.1002/ams2.871 ==== Body pmcINTRODUCTION According to the World Health Organization (WHO), overweight and obesity are defined as the accumulation of abnormal or excessive fat that poses health risks. Thus, the risk of obesity‐related diseases, such as diabetes and chronic renal failure, increases with the progression of obesity. 1 Patients with hypertension, diabetes, and obesity have been reported to be susceptible to severe respiratory failure due to coronavirus disease 2019 (COVID‐19). 2 , 3 By contrast, a paradoxical phenomenon called the obesity paradox has been reported in diseases such as acute respiratory distress syndrome (ARDS) 4 and cancer, 5 in which patients with obesity have a better prognosis than patients with a normal body mass index (BMI). In diseases with obesity paradox, a J‐shaped correlation between BMI and prognosis is often observed, suggesting that moderate obesity itself may be protective compared with patients who are normal or severely obese; however, the mechanism remains not fully understood. 6 Reports showed that obesity has no effect on the mortality of patients undergoing veno‐venous extracorporeal membrane oxygenation (VV‐ECMO) for severe respiratory failure, 7 and in Europe, patients with obesity with a BMI >30 kg/m2 may have a better prognosis than patients with a BMI <30 kg/m2 undergoing ECMO for COVID‐19; 8 however, it is unclear whether a similar trend is observed in Japan. In this study, we aimed to investigate the relationship between BMI and prognosis in patients receiving VV‐ECMO for COVID‐19. METHODS Study setting We retrospectively examined the association between ECMO withdrawal and BMI in patients who underwent VV‐ECMO for COVID‐19 at Yokohama City University Civic General Medical Center between March 2020 and October 2021. We collected medical information, including age, sex, BMI, medical history, Sequential Organ Failure Assessment (SOFA) score, Acute Physiology and Chronic Health Evaluation II (APACHE II) score, Respiratory ECMO Survival Prediction (RESP) score, Murray score, and whether or not patients received remdesivir (200 mg/day on day 1 only, 100 mg/day on days 2–10), dexamethasone (6.6 mg/day), unfractionated heparin (10,000 units/day on days 2–10), and supine therapy (at least 16 h). Data on the fraction of inspiratory oxygen (FiO2), positive end‐expiratory pressure (PEEP), maximum airway pressure, and arterial blood gas findings were also collected within 6 h before ECMO induction. Primary and secondary outcomes The primary outcome was death during the induction of ECMO. The secondary outcomes were the duration of ECMO and the number of days from onset to hospitalization, hospitalization to endotracheal intubation, onset to endotracheal intubation, endotracheal intubation to ECMO induction, and onset to ECMO induction. The participants were divided into two groups based on the WHO definition of obesity 9 : BMI ≤30 kg/m2 and BMI >30 kg/m2 (Figure 1). FIGURE 1 Patient flow diagram. BMI, body mass index. Clinical workflow and disease staging Ventilator management was limited to a maximum PEEP of 15 cmH2O, and airway pressure did not exceed 30 cmH2O. The lung protection strategy aimed at a tidal volume of 6–8 mL/kg, and deep sedation and muscle relaxants were used if the patient presented with large excess breaths. If computed tomography showed a strong image of pneumonia on the dorsal side and PaO2/FiO2 (P/F) was below 200, the patient was placed in the prone position. VV‐ECMO was introduced when oxygenation could not be maintained despite the aforementioned respiratory management. The indications for VV‐ECMO were as follows: patients with hypoxemia with FiO2 ≥0.8 and P/F <100, respiratory acidosis with pH ≤7.2 and plateau pressure >32 cmH2O, Murray score >3, or in the prone position despite treatment intervention for the original disease, lung protection strategy + high PEEP strategy and prone therapy, and poor response to therapy (Figure 2). FIGURE 2 ECMO induction criteria. ADL, activities of daily living; CT, computed tomography; ECMO, extracorporeal membrane oxygenation; P/F ratio, PaO2/FiO2 ratio; PEEP, positive end‐expiratory pressure. Statistical analysis The distribution of each variable in the two groups, classified according to BMI, was compared. The Mann–Whitney U test was used for continuous variables, and Fisher exact test was used for categorical variables. All statistical analyses were performed using JMP version 16 (SAS Institute Inc., Cary, NC, USA). Statistical significance was set at p < 0.05. RESULTS In total, 23 patients were included in the analysis (Appendix S1). The included patients were 57 [51–61] (median [interquartile range]) years old, 17 (74%) were male, BMI was 28.7 [26.5–36.2] kg/m2, SOFA score was 10 [8–12], APACHE score was 22 [19–26], Murray score was 3.3 [3.0–3.5], and RESP score was 2 [1–4]. Remdesivir, dexamethasone, unfractionated heparin, and supine therapy were administered to 14 (61%), 18 (78%), 23 (10%), and 10 (41.6%) patients, respectively. Onset to hospitalization was 5 [4–9] days, hospitalization to endotracheal intubation was 1 [0–3] days, onset to endotracheal intubation was 7 [6–10] days, endotracheal intubation to ECMO induction was 3 [0–5] days, onset to ECMO induction was 12.5 [7.3–15.0] days, and ECMO duration was 10.0; 2 (8.3%) patients died during ECMO induction (Table 1). TABLE 1 Patients’ characteristics. Characteristics All patients (N = 23), median (interquartile range)/frequency (%) Patient background Age (years) 57.0 (51.0–61.0) Male 17 (74) BMI (kg/m2) 28.7 (26.5–36.2) Comorbidities Hypertension 14 (60.8) Diabetes 12 (52.2) Dyslipidemia 8 (34.8) Coronary artery disease 2 (8.7) Chronic lung disease 1 (4.3) Chronic kidney disease 2 (8.7) Chronic liver disease 0 (0) Immunosuppression 1 (4.3) Metastatic solid tumor 0 (0) Scores SOFA score 10 (8–12) APACHE score 22 (19–26) Murray score 3.3 (3.0–3.5) RESP score 2 (1–4) Within 6 h before ECMO FiO2 1.0 (0.8–1.0) PEEP 15 (13–15) Maximum airway pressure 30 (26–30) P/F ratio 69.5 (48.9–75.4) pH 7.38 (7.27–7.42) pCO2 58.4 (48.9–72.8) Treatment Remdesivir 14 (61) Dexamethasone 18 (78) Unfractionated heparin 23 (100) Prone position 11 (42) Disposition Onset to admission (days) 5 (4–9) Admission to intubation (days) 1 (0–3) Onset to intubation (days) 7 (6–10) Intubation to ECMO (days) 3 (0–5) Onset to ECMO (days) 12 (7–15) ECMO duration (days) 10 (8–18) Death 2 (8) Abbreviations: APACHE score, Acute Physiology and Chronic Health Evaluation Score; BMI, body mass index; ECMO, extracorporeal membrane oxygenation; FiO2, fraction of inspiratory oxygen; P/F ratio, PaO2/FiO2 ratio; pCO2, partial pressure of carbon dioxide; PEEP, positive end‐expiratory pressure; RESP score, Respiratory ECMO Survival Prediction score; SOFA score, Sequential Organ Failure Assessment score. Comparison of two groups classified according to BMI The included patients were divided into two groups: BMI ≦30 kg/m2 (n = 13) and BMI >30 kg/m2 (n = 10). As for the patients’ background, age was 57 versus 56.5 years (p = 0.83) in the BMI ≦30 kg/m2 and BMI >30 kg/m2 groups, BMI was 26.5 kg/m2 versus 36.5 kg/m2 (p < 0.001), and sex (male) was 69% versus 80% (p = 0.66). As for the prognostic score, SOFA score was 8 versus 10 (p = 0.43), APACHE II score was 20 versus 23 (p = 0.13), Murray score was 3.3 versus 3.2 (p = 0.68), RESP score was 2 versus 2.5 (p = 0.90), with no statistical difference between the two groups. The time from onset to admission, 5 versus 5.5 days (p = 0.87); from admission to intubation, 3 versus 0.5 days (p = 0.23); from onset to intubation, 8 versus 6.5 days (p = 0.30); from intubation to ECMO, 4 versus 2 days (p = 0.17); from onset to ECMO, 14 versus 8 days (p = 0.07); and ECMO duration, 12 versus 8 days (p = 0.14). As a treatment, 53.9% versus 70% (p = 0.67) received remdesivir, 100% versus 100% received unfractionated heparin, 76.9% versus 80% (p = 1.00) received dexamethasone, and 53.9% versus 40% (p = 0.68) were received prone position therapy.  For ventilator or respiratory status within 6 h before ECMO, FiO2 was 0.8 versus 1.0 (p = 0.06), PEEP was 15 versus 15 cmH2O (p = 0.95), maximum airway pressure 26 versus 30 cmH2O (p = 0.22), P/F ratio 70.8 versus 60 (p = 0.39), pH 7.40 versus 7.32 (p = 0.19), and partial pressure of carbon dioxide (pCO2) 55.1 versus 67.9 mmHg (p = 0.14). There was no statistical difference in patients’ history of preexisting medical conditions: 54% versus 70% had hypertension (p = 0.66), 38% versus 70% had diabetes (p = 0.21), 31% versus 40% had dyslipidemia (p = 0.69), 15% versus 0% had coronary artery disease (p = 0.48), 0% versus 10% had chronic lung disease (p = 0.43), 8% versus 10% had chronic kidney disease (p = 1.00), and 0% versus 10% had immunosuppression (p = 0.43). Two patients died in the group with BMI <30 kg/m2 because of pulmonary fibrosis (ECMO duration was 56 days); another reason was sepsis (ECMO duration was 16 days; Table 2). TABLE 2 Comparison of patients’ information classified according to BMI. Patient information BMI ≦30 kg/m2 (n = 13) BMI >30 kg/m2 (n = 10) p value Patient background Age 57 (51.5–60) 56.5 (49–62.25) 0.83 Male 9 (69) 8 (80) 0.66 BMI (kg/m2) 26.5 (24.3–28.7) 36.5 (33.2–40.5) <0.001 Comorbidities Hypertension 7 (54) 7 (70) 0.66 Diabetes 5 (38) 7 (70) 0.21 Dyslipidemia 4 (31) 4 (40) 0.69 Coronary artery disease 2 (15) 0 (0) 0.48 Chronic lung disease 0 (0) 1 (10) 0.43 Chronic kidney disease 1 (8) 1 (10) >0.99 Chronic liver disease 0 (0) 0 (0) — Immunosuppression 0 (0) 1 (10) 0.43 Metastatic solid tumor 0 (0) 0 (0) — Scores SOFA score 8 (8–12) 10 (7.75–13) 0.43 APACHE score 20 (18–23) 23 (19.75–28.75) 0.13 Murray score 3.3 (2.8–3.6) 3.2 (3.0–3.6) 0.68 RESP score 2 (1.5–3.5) 2.5 (1–4.5) 0.90 Within 6 h before ECMO FiO2 0.8 (0.75–1.0) 1.0 (0.975–1.0) 0.06 PEEP 15 (12.5–15.5) 15 (14.25–15) 0.95 Maximum airway pressure 26 (25–30) 30 (26.5–32.5) 0.22 P/F ratio 70.8 (48.9–94.3) 60 (51.5–73.3) 0.39 pH 7.40 (7.33–7.49) 7.32 (7.25–7.41) 0.19 pCO2 55.1 (49.2–82.1) 67.9 (56.7–98.9) 0.14 Treatment Remdesivir 7 (53.9) 7 (70) 0.67 Dexamethasone 10 (76.9) 8 (80) 1.00 Unfractionated heparin 13 (100) 10 (100) — Prone position 7 (53.9) 4 (40) 0.68 Disposition Onset to admission (days) 5 (4–9.5) 5.5 (3–8.5) 0.87 Admission to intubation (days) 3 (0–4) 0.5 (0–2.25) 0.23 Onset to intubation (days) 8 (6–12) 6.5 (5.75–10) 0.30 Intubation to ECMO (days) 4 (2–5) 2 (0–4.5) 0.17 Onset to ECMO (days) 14 (9.5–16.5) 8 (6.5–13) 0.07 ECMO duration (days) 12 (8–25) 8 (7.5–12) 0.14 Death 2 (15.4) 0 (0) 0.48 Note: Data presented as median (interquartile range) or frequency (%). Abbreviations: APACHE score, Acute Physiology and Chronic Health Evaluation Score; BMI, body mass index; ECMO, extracorporeal membrane oxygenation; FiO2, fraction of inspiratory oxygen; P/F ratio, PaO2/FiO2 ratio; pCO2, partial pressure of carbon dioxide; PEEP, positive end‐expiratory pressure; RESP score, Respiratory ECMO Survival Prediction score; SOFA score, Sequential Organ Failure Assessment score. There were no statistically significant differences between the two groups in terms of age, sex, SOFA, APACHE, Murray, and RESP scores, whether patients received remdesivir, dexamethasone, unfractionated heparin, or the percentage of patients receiving supine therapy. Moreover, there were no statistically significant differences in PEEP, maximum airway pressure, P/F ratio, pH, or pCO2 within 6 h before ECMO induction. There was no statistically significant difference in ECMO survival and withdrawal rates between the two groups. However, 100% survival and withdrawal rates were achieved in the group with BMI >30 kg/m2. The number of days from disease onset to ECMO induction tended to be shorter in the BMI >30 kg/m2 group. There were no statistically significant differences between the two groups in terms of the number of days from onset to hospitalization, hospitalization to endotracheal intubation, onset to endotracheal intubation, intubation to ECMO induction, and ECMO duration (Table 2). DISCUSSION In this study, patients with severe respiratory failure due to COVID‐19 and in the BMI >30 kg/m2 group required earlier ECMO induction compared with those in the BMI ≤30 kg/m2 group. In addition, inhaled oxygen concentrations before ECMO tended to be higher in the BMI >30 kg/m2 group than in the BMI ≤30 kg/m2 group. A prospective observational cohort study by Daviet et al. 8 consisting of 76 patients with COVID‐19 requiring ECMO showed that a higher BMI was a positive independent factor for 90‐day survival, with better outcomes in the BMI >30 kg/m2 than in the BMI ≤30 kg/m2 group. Moreover, the study reported that the time from intensive care unit admission to endotracheal intubation and the time from intensive care unit admission to ECMO were significantly shorter in the BMI >30 kg/m2 group than in the BMI ≤30 kg/m2 group. 8 Mongero et al. 9 reported a shorter time from diagnosis to ECMO in the good prognosis group than in the poor prognosis group. This study cites the involvement of specific respiratory mechanics in patients with obesity, such as decreased chest wall compliance, increased intra‐abdominal pressure, and decreased lung volume, as reasons for the earlier induction of ECMO. In our study, the number of days from onset to ECMO induction tended to be fewer in the BMI >30 kg/m2 than in BMI ≤30 kg/m2 group, corroborating the results of previous studies. In the present study, there was no statistically significant difference in ECMO duration or mortality between the BMI >30 kg/m2 and BMI ≤30 kg/m2 groups. Obesity has been reported to be a risk factor for severe COVID‐19, 3 while patients with obesity have been reported to have a similar or better prognosis in ARDS cases with COVID‐19 using ECMO compared with patients with normal weight, 8 , 10 which is similar to the results of the present study. The mechanism of the obesity paradox in ARDS remains unclear; however, several possibilities have been reported. One possibility is due to decreased inflammatory cytokines in patients with obesity. A study of peripheral blood from 1409 patients with acute lung injury reported an inverse relationship between high BMI and levels of inflammatory cytokines, such as interleukin‐6 and interleukin‐8, and surfactant protein D. 11 Another possibility is the hypothesis that ventilatory management difficulties in patients with obesity lead to early introduction of ECMO and early implementation of a thorough lung protection strategy. Patients with ARDS with BMI >25 kg/m2 have been shown to have lower pulmonary compliance due to decreased thoracic compliance compared with patients with BMI ≤25 kg/m2. 12 Patients with obesity may have difficulty with ventilatory management as a result of thoracic restrictive mechanics even with less severe lung parenchyma, resulting in an earlier introduction of ECMO. Lowering drive pressure during ventilatory management of patients with ARDS has been reported to reduce mortality, 13 and the use of higher drive pressure during ECMO induction for ARDS has been associated with increased mortality. 14 In a report comparing patients on ventilatory management for COVID‐19 ARDS in three groups (BMI <25 kg/m2, 25 ≤ BMI ≤ 30 kg/m2, and BMI >30 kg/m2), the higher BMI group required higher drive pressures and showed decreased pulmonary compliance. 15 These findings suggest that early introduction of ECMO results in lower driving pressure and less ventilator‐related lung injury, which may explain the favorable prognosis of patients with obesity with early introduction of ECMO. This hypothesis has been discussed in several studies. 16 , 17 , 18 Thus, we believe that the introduction of ECMO for respiratory failure in patients with obesity should not be withheld because of obesity itself. By contrast, the use of ECMO in patients with respiratory failure requires significant medical resources 19 and prognosis may vary greatly depending on the balance between medical supply and demand, 20 and depression and decreased sexual activity have been reported 21 as 1‐year outcomes for patients with COVID‐19 on ECMO, so its use should be carefully considered. This study has certain limitations. This was a single‐center, retrospective study; therefore, selection bias may have been present, and caution should be exercised when generalizing these findings. Moreover, the sample size was limited and the statistical power may have been inadequate. Future studies should be conducted in a multicenter setting, with a large number of participants. CONCLUSION In the case of VV‐ECMO induction for COVID‐19, patients with BMI >30 kg/m2 may have received earlier induction than those with BMI ≤30 kg/m2. Failure of early ventilatory management leads to the early introduction of ECMO, reduction of ventilator‐related lung injury, and early implementation of a thorough lung protection strategy. CONFLICT OF INTEREST STATEMENT The authors declare no conflicts of interest for this article. ETHICS STATEMENT Approval of the research protocol: The protocol for this research project has been approved by a suitably constituted Ethics Committee of the institution, and it conforms to the provisions of the Declaration of Helsinki (Committee of the Yokohama City University Medical Center, Approval Number B200200049). Informed consent: All informed consent was obtained from the subject(s) and/or guardian(s). Registry and the registration no. of the study/trial: N/A. Animal studies: N/A. CONSENT FOR PUBLICATION Written informed consent was obtained from patients for the publication of this case report and the relevant images. A copy of the written consent is available for review by the Editor‐in‐Chief of Acute Medicine and Surgery. Supporting information Appendix S1 Click here for additional data file. ACKNOWLEDGMENTS This work was supported by JSPS KAKENHI (grant number 21 K09026). We thank Editage (www.editage.com) for English language editing. DATA AVAILABILITY STATEMENT Data requests should be made to the corresponding author. ==== Refs REFERENCES 1 Schetz M , De Jong A , Deane AM , Druml W , Hemelaar P , Pelosi P , et al. Obesity in the critically ill: a narrative review. Intensive Care Med. 2019;45 :757–69.30888440 2 Zhou F , Yu T , Du R , Fan G , Liu Y , Liu Z , et al. Clinical course and risk factors for mortality of adult inpatients with COVID‐19 in Wuhan, China: a retrospective cohort study. Lancet. 2020;395 :1054–62.32171076 3 Gao YD , Ding M , Dong X , Zhang JJ , Kursat Azkur A , Azkur D , et al. Risk factors for severe and critically ill COVID‐19 patients: a review. Allergy. 2021;76 :428–55.33185910 4 Zhi G , Xin W , Ying W , Guohong X , Shuying L . Obesity paradox in acute respiratory distress syndrome: a systematic review and meta‐analysis. PLoS One. 2016;11 :e0163677.27684705 5 Horwich TB , Fonarow GC , Clark AL . Obesity and the obesity paradox in heart failure. Prog Cardiovasc Dis. 2018;61 :151–6.29852198 6 Zhou D , Wang C , Lin Q , Li T . The obesity paradox in survivors for critically ill patients. Crit Care. 2022;26 :198.35781349 7 Cho WH , Oh JY , Yeo HJ , Han J , Kim J , Hong SB , et al. Obesity survival paradox in pneumonia supported with extracorporeal membrane oxygenation: analysis of the national registry. J Crit Care. 2018;48 :453–7.30409352 8 Daviet F , Guilloux P , Hraiech S , Tonon D , Velly L , Bourenne J , et al. Impact of obesity on survival in COVID‐19 ARDS patients receiving ECMO: results from an ambispective observational cohort. Ann Intensive Care. 2021;11 :157.34779944 9 Mongero LB , Stammers AH , Tesdahl EA , Petersen C , Patel K , Jacobs JP . The use of extracorporeal membrane oxygenation in COVID‐19 patients with severe cardiorespiratory failure: the influence of obesity on outcomes. J Extra Corpor Technol. 2021;53 :293–8.34992320 10 Shaefi S , Brenner SK , Gupta S , O'Gara BP , Krajewski ML , Charytan DM , et al. Extracorporeal membrane oxygenation in patients with severe respiratory failure from COVID‐19. Intensive Care Med. 2021;47 :208–21.33528595 11 Stapleton RD , Dixon AE , Parsons PE , Ware LB , Suratt BT . NHLBI acute respiratory distress syndrome network. The association between BMI and plasma cytokine levels in patients with acute lung injury. Chest. 2010;138 :568–77.20435656 12 Bein T . Driving pressure in obese ventilated patients: another brick in the (chest) wall. Intensive Care Med. 2018;44 :1349–51.29959454 13 Amato MB , Meade MO , Slutsky AS , Brochard L , Costa EL , Schoenfeld DA , et al. Driving pressure and survival in the acute respiratory distress syndrome. N Engl J Med. 2015;372 :747–55.25693014 14 Wang R , Sun B , Li X , Tang X , He H , Li Y , et al. Mechanical ventilation strategy guided by transpulmonary pressure in severe acute respiratory distress syndrome treated with venovenous extracorporeal membrane oxygenation. Crit Care Med. 2020;48 :1280–8.32590385 15 Schavemaker R , Schultz MJ , Lagrand WK , van Slobbe‐Bijlsma E , Serpa Neto A , Paulus F , et al. Associations of body mass index with ventilation management and clinical outcomes in invasively ventilated patients with ARDS related to COVID‐19—insights from the PRoVENT‐COVID study. J Clin Med. 2021;10 :1176.33799735 16 Ladosky W , Botelho MA , Albuquerque JP Jr . Chest mechanics in morbidly obese non‐hypoventilated patients. Respir Med. 2001;95 :281–6.11316110 17 Kon ZN , Dahi S , Evans CF , Byrnes KA , Bittle GJ , Wehman B , et al. Class III obesity is not a contraindication to venovenous extracorporeal membrane oxygenation support. Ann Thorac Surg. 2015;100 :1855–60.26296270 18 Littleton SW . Impact of obesity on respiratory function. Respirology. 2012;17 :43–9.22040049 19 Peek GJ , Mugford M , Tiruvoipati R , Wilson A , Allen E , Thalanany MM , et al. Efficacy and economic assessment of conventional ventilatory support versus extracorporeal membrane oxygenation for severe adult respiratory failure (Cesar): a multicentre randomised controlled trial. Lancet. 2009;374 :1351–63.19762075 20 Barbaro RP , MacLaren G , Boonstra PS , Combes A , Agerstrand C , Annich G , et al. Extracorporeal membrane oxygenation for COVID‐19: evolving outcomes from the international extracorporeal life support organization registry. Lancet. 2021;398 :1230–8.34599878 21 Chommeloux J , Valentin S , Winiszewski H , Adda M , Pineton de Chambrun M , Moyon Q , et al. One‐year mental and physical health assessment in survivors after ECMO for COVID‐19‐related ARDS. Am J Respir Crit Care Med. 2023;207 :150–9.36150112
PMC010xxxxxx/PMC10352564.txt
==== Front Acute Med Surg Acute Med Surg 10.1002/(ISSN)2052-8817 AMS2 Acute Medicine & Surgery 2052-8817 John Wiley and Sons Inc. Hoboken 10.1002/ams2.873 AMS2873 AMS-2023-0077.R2 Mini Review Mini Reviews Narrative minireview of the current status of hyperbaric oxygen therapy for pregnant women HBO for Pregnant Women Yanagawa et al. Yanagawa Youichi https://orcid.org/0000-0002-4270-6402 1 [email protected] Nunotani Marika 1 Abe Keiki 1 Hamada Michika 1 Ota Soichiro 1 Muramatsu Ken‐ichi 1 Takeuchi Ikuto 1 Nagasawa Hiroki https://orcid.org/0000-0001-7825-4244 1 Ohsaka Hiromichi https://orcid.org/0000-0002-0528-2743 1 Ishikawa Kouhei https://orcid.org/0000-0001-7911-4825 1 1 Department of Acute Critical Care Medicine Shizuoka Hospital, Juntendo University Izunokuni Japan * Correspondence Youichi Yanagawa, Department of Acute Critical Care Medicine, Shizuoka Hospital, Juntendo University, 1129 Nagaoka, Izunokuni 410‐2295, Shizuoka, Japan. Email: [email protected] 17 7 2023 Jan-Dec 2023 10 1 10.1002/ams2.v10.1 e87321 6 2023 08 4 2023 28 6 2023 © 2023 The Authors. Acute Medicine & Surgery published by John Wiley & Sons Australia, Ltd on behalf of Japanese Association for Acute Medicine. https://creativecommons.org/licenses/by-nc/4.0/ This is an open access article under the terms of the http://creativecommons.org/licenses/by-nc/4.0/ License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes. Abstract We performed a narrative minireview for a PubMed search on March 31, 2023, using the keywords “pregnant” and “hyperbaric oxygen” to identify any related articles. Most reports have described pregnant women with carbon monoxide (CO) poisoning being treated by hyperbaric oxygen therapy (HBOT). HBOT helped improve the maternal condition and ensure normal fetal development. Some pregnant women with CO poisoning treated by HBOT suffered abortions or gave premature birth to low‐weight babies or with congenital malformations. However, these results were considered sequelae of CO poisoning, not HBOT. We hypothesized that for pregnant women facing a life‐threatening situation, for which the effectiveness of HBOT has previously been suggested, prioritizing the stabilization of the mother may also be beneficial for normal fetal development. Most reports described pregnant women with carbon monoxide poisoning being treated by hyperbaric oxygen therapy (HBOT). We hypothesized that for pregnant women facing a life‐threatening situation for which HBOT has proven effective, prioritizing stabilizing the mother would also be beneficial for normal fetus development. hyperbaric oxygen therapy neonate pregnant women Promotion and Mutual Aid Corporation for Private Schools of Japan 10.13039/501100012359 source-schema-version-number2.0 cover-dateJanuary/December 2023 details-of-publishers-convertorConverter:WILEY_ML3GV2_TO_JATSPMC version:6.3.2 mode:remove_FC converted:18.07.2023 Yanagawa Y , Nunotani M , Abe K , Hamada M , Ota S , Muramatsu K‐i , et al. Narrative minireview of the current status of hyperbaric oxygen therapy for pregnant women. Acute Med Surg. 2023;10 :e873. 10.1002/ams2.873 ==== Body pmcINTRODUCTION Hyperbaric oxygen therapy (HBOT) consists of using pure oxygen at increased pressure (in general, 2–3 atmosphere absolute [ATA]) to augment oxygen levels in the blood (hyperoxemia) and tissue (hyperoxia). This results in a major supply of reactive oxygen species and reactive nitrite species, with a consequent increased expression of growth factors, promotion of neovascularization, and enhanced immunomodulatory properties. 1 HBOT is typically delivered in hyperbaric chambers to a single patient at a time (monoplaced) or in a room accommodating multiple patients at the same pressure (multiplace chambers). The commonly accepted indications for HBOT are air or gas arterial or venous emboli, decompression sickness (DCS), carbon monoxide (CO) poisoning, gas gangrene, selected crush injuries, selected arterial insufficiencies, severe otherwise untreatable anemias, certain intracranial abscesses, necrotizing soft tissue infections, chronic refractory osteomyelitis, delayed radiation–induced injury, compromised skin grafts, acute thermal burn injuries, and sudden sensorineural hearing loss. 2 However, a number of major adverse effects induced by HBOT have been reported. The two most common complications during HBOT are claustrophobia and barotrauma. In addition, as a result of the hyperoxic and hyperbaric environment, seizures might be induced due to the toxic properties of oxygen at high concentrations. In addition, retinopathy due to premature delivery in pregnant women might develop. 2 , 3 HBOT may also be teratogenic for fetuses when pregnant women receive HBOT. 4 , 5 We previously experienced a case wherein a pregnant woman suffering from nonobstructive ileus consulted our facility for HBOT. HBOT has been reported to be indicated for ileus. 6 , 7 However, the indications of HBOT for pregnant women remain unclear, except for cases of CO poisoning, and we therefore canceled this consultation. Given the above, we reviewed reports concerning the clinical usefulness, safety, and adverse effects of HBOT for pregnant women and summarized the essential data. METHODS We performed a narrative minireview for a PubMed search on March 31, 2023, to identify any related articles using the keywords “pregnant” and “hyperbaric oxygen”. The inclusion criterion was pregnant women who had actually received HBOT before delivery, abortion, or caesarean section. The exclusion criterion was reports written in languages other than English. RESULTS We found 288 articles after the PubMed search. Of these, 35 reports described pregnant women actually treated by HBOT before delivery, abortion, or caesarean section. After excluding 17 reports written in languages other than English and 7 that were review articles and did not include original clinical data, 8 , 9 , 10 , 11 , 12 , 13 , 14 the remaining 11 reports were summarized (Table 1). 15 , 16 , 17 , 18 , 19 , 20 , 21 , 22 , 23 , 24 , 25 There were eight case reports 15 , 16 , 17 , 18 , 19 , 20 , 21 , 22 and three original reports. 23 , 24 , 25 The number of pregnant women who were treated by HBOT before delivery, abortion, or cesarean section was 113 in total. The average age of pregnant women, whose ages were described in the reports, was 26.4 years. The average gestational week, when the weeks were described in the reports, was 21.5 weeks. Ten of the eleven reports involved pregnant women with CO poisoning treated by HBOT. 15 , 16 , 17 , 18 , 19 , 20 , 21 , 23 , 24 , 25 All 10 reports showed the efficacy of HBOT for improving the maternal condition and enabling normal fetal development. Some pregnant women with CO poisoning treated by HBOT had abortions or gave birth to low‐weight babies prematurely or with congenital malformations. 25 However, these results were considered sequelae of CO poisoning rather than HBOT. 23 , 24 , 25 Based on this clinical evidence, standard HBOT in pregnancy appears to be safe and is considered beneficial, reducing the severity of maternal and fetal injuries and fetal deformities. 14 In addition, HBOT may improve long‐term fetal outcomes even after birth. 23 TABLE 1 Summary of previous reports concerning HBOT for pregnant women before delivery, abortion, or cesarean section. Ref. no. First author Type N Maternal age (years) Fetal age (weeks of gestation) Disease Cause Maternal CO‐Hb HBOT ATA Duration (min) Number of times Side effects Indication or subject Delivery outcome Neonate follow‐up Key message 15 Kosaki Case 1 30 31 CO Accident (camp) 28.6 2 60 3 No High CO Hb At 41 weeks, she delivered a healthy infant weighing 2862 g After 3 months’ follow‐up, both the infant and mother are doing well During pregnancy, CO negatively affects both the fetus and mother, so it is critical for emergency physicians to promptly recognize and manage such poisoning cases 16 Gabrielli Case 1 14 37 CO Accident (heater) 5.3 3 180 1 No Headache, vomiting At 24 h after HBOT, she vaginally delivered a healthy infant weighing 3035 g Normal after 4 weeks CO intoxication in late pregnancy was treated with HBOT without adverse consequences 17 Greingor Case 1 31 21 CO Accident (heater) 24.6 2.5 90 1 No High CO Hb She was discharged in good health 1 day after exposure and delivered a healthy male infant at term weighing 3800 g No description Despite maternal well‐being, fetal morbidity or mortality can still occur. HBOT seems to be the treatment of choice, and all pregnant women suffering from CO intoxication should be referred for HBOT 18 Brown Case 1 27 32 CO Accident (fire) 25.1 3 + 2 46 + 50 No High CO Hb, syncope attack A healthy 3200 g male by vaginal delivery No description Both mother and fetus are doing fine. No negative effects from the HBOT were observed 19 Van Hoesen Case 1 17 37 CO Accident (car) 47.2 2.4 90 1 No High CO Hb and consciousness disturbance of mother with fetal distress syndrome A healthy 3600 g female by vaginal delivery Follow‐up examinations at 2 and 6 months old revealed normal growth and development with normal findings from a funduscopic examination Acute CO poisoning during pregnancy was successfully treated with HBOT. Recommendations are suggested for the use of HBOT during pregnancy 20 Mandal Case 1 31 35 CO Accident (heater) 25.3 2.8 100 2 No Fetal distress A healthy male by vaginal delivery Normal development after 1 year HBOT appears to be safe and effective for maternal body and fetus with CO intoxication 21 Santos Case series 2/37 ? ? CO Wildfire 3.7 2.5 90 ? No description Two of 37 were pregnant. No detailed information was available. ? No description Wildfires can cause clinically relevant CO poisoning, with typical signs and symptoms 22 Tchirikov Case 1 2 22 30 Severe placental insufficiency N/A N/A 1.4 50 7 No Severe placental insufficiency At 31 + 4 weeks’ gestation, the patient gave birth spontaneously to a preterm newborn weighing 1378 g. At 5 years old, the boy is doing well, but the speech development was delayed without any neurological disturbance Fetal nutrition combined with HBOT is technically possible and may allow the prolongation of the pregnancy 22 Tchirikov Case 2 2 30 25 Severe placental insufficiency N/A N/A 1.4 50 7 No Severe placental insufficiency During HBOT, the patient complained of uterine contractions. One day later, her child was delivered by cesarean section because of fetal late decelerations. The newborn weighed 420 g, Apgar 3/8/8 The newborn unfortunately developed pulmonary bleeding on day 4 of life with subsequent hypotonia, anuria, leucopenia, and thrombocytopenia, and died 6 days later Fetal nutrition combined with HBOT is technically possible and may allow the prolongation of the pregnancy 23 Ozgok‐Kangal Original 27 26.8 18 CO No description 27.9 2 or 2.5 30, 75 or 90 1 (3 in 30 min) No description Vaginal (66.6%), caesarean (29.6%), abortion (3.7%); Infant weight ≥2500 g (95.7%), <2500 g (4.3%); 3 of 28 had abortion, premature birth, or limb malformation The median infant age was 34 (8–44) months at the last interview. Only 1 child (43 months old) could not speak There was no definite evidence of fetal morbidity or mortality after HBOT. HBOT may improve long‐term fetal outcomes after in utero CO poisoning without complications 24 Arslan Original 32 25 23 CO Three patients were poisoned at work, and 29 were poisoned by stoves at home or in tents in which they were living 24.9 2.4 120 1 (81.8%) 2 (15.2%) 3 (3.0%) No description Four patients had premature births (32–34 weeks). Two neonates died after birth: one from congenital cyanotic heart disease, and the other from a twin pregnancy at 33rd week with a birth weight of 1800 g and respiratory distress. The others were healthy. No description HBOT is not advisable for pregnant patients except in cases of CO poisoning. HBOT under 2.4 ATA pressure for 120 min has no harmful effects on the mother or fetus 25 Elkharrat Original 44 27.5 21 CO No description 19 2 120 1 No description Ten patients sustained a loss of consciousness for a few seconds, two were in a coma, and the remainder had headaches, gastrointestinal dysfunction, and dizziness Obstetric follow‐up data were available for 38 women. Normal delivery of normal infants in 32 (85%), spontaneous abortion in 12 (5%), medical abortion (10 weeks’ gestation) in 1 (2%), premature delivery of a normal baby in 1 (2%), induction of labor (36 weeks’ gestation) in 1 (2%), and baby with Down syndrome in 1 (2%) The infant was born with major cardiopulmonary defects and died 6 days later There is no evidence that HBOT was involved with either abortion in our study. HBOT may be performed in pregnant women acutely intoxicated with CO Abbreviations: ATA, atmosphere absolute; CO, carbon monoxide; Hb, hemoglobin; HBOT, hyperbaric oxygen therapy; N/A, not applicable; ?, no description. The remaining article described two pregnant women with severe placental insufficiency treated by HBOT. 22 The first patient gave birth spontaneously to a newborn weighing 1378 g at 31 weeks’ gestation. In a follow‐up examination at 5 years of age, the boy was doing well without any developmental delay. In the second case, the patient gave birth to a hypotrophic newborn weighing 420 g at 25 weeks’ gestation; unfortunately, the extremely preterm newborn died 6 days later. The authors concluded that adequate fetal nutrition combined with HBOT was technically possible and facilitated the prolongation of the pregnancy. 22 One article described a 2‐hour‐old neonate with CO poisoning who underwent HBOT. 26 His 41‐year‐old mother had accidentally suffered CO poisoning at 38 weeks’ gestation by a heater. Given the decreased fetal movement and biophysical profile score, the obstetrics team decided to perform emergency caesarean section. After delivery, the patient's carboxyhemoglobin value was 11.9%, and the maternal value was 7.4%. Both mother and neonate underwent HBOT at 2.4 atmosphere absolutes (ATA) for 90 min starting 2.5 h after delivery. The infant was discharged in good condition 3 days after delivery but later lost to follow‐up. Another article described a case of residual severe neurological deficits and stillbirth due to delayed HBOT for a pregnant woman suffering from air embolism. 27 A 36‐year‐old woman in her 30th week of pregnancy developed convulsion and fell unconscious due to air embolism induced by orogenital sex. She underwent tracheal intubation and 100% oxygenation and magnesium sulfate administration. She spontaneously suffered stillbirth 3 h after the incident. She received delayed HBOT at 39 h after the incident but sustained severe sequalae from it. The author insisted that emergency HBOT was still required to obtain favorable outcomes for both the mother and fetus. DISCUSSION This review indicated that standard HBOT for pregnant women suffering from CO poisoning is safe and considered beneficial, reducing the severity of maternal and fetal injury. Concerning the safety of HBOT for pregnant women, HBOT ranging from 1.4 to 3 ATA with a duration of 30 to 180 min was found to be safe based on the results from Table 1. Typical side effects of HBOT are barotrauma for the middle ear, sinus, or oxygen toxicity. 28 , 29 However, such problems for pregnant women have not been documented in the previous reports as noted in Table 1. Strict operational protocols and in‐chamber monitoring might result in an improvement of maternal safety. 28 Fetal hemoglobin binds to CO as maternal hemoglobin does but with 2.5–3 times greater affinity. 14 A shift to the left of the dissociation curve is also expected in the fetal compartment. Furthermore, fetal carboxyhemoglobin has a half‐life approximately 4 times longer than maternal carboxyhemoglobin. 14 As such, even in cases where the mother is doing well, fetal morbidity or mortality can still occur. Accordingly, even in asymptomatic cases of maternal CO exposure, HBOT might be necessary to minimize fetal injuries and ensure normal fetal development. 14 Normal fetal development requires the stabilization of vital signs of the maternal body. 30 , 31 , 32 Accordingly, for pregnant women facing a life‐threatening situation for which HBOT has proven effective, prioritizing stabilizing the mother would also be beneficial for normal fetal development. At least a single episode of hypoxia from CO poisoning can be teratogenic for a fetus; however, there has so far been no report of a single HBOT being teratogenic for a fetus based on experimental studies. 33 Accordingly, fetal teratogenic complications are likely to be related to CO poisoning, but they are not related to HBOT. 23 No previous report has described the benefits and adverse effects of HBOT for healthy pregnant women and, in particular, their fetuses. Sapunar et al. 34 reported the effects of HBOT on rat embryos. They found that HBOT did not induce malformations at either 3.3 or 4.3 ATA and that the largest embryotoxic effects involved decreasing the fetal body weight and increasing the placental weight. Gilman et al. 35 reported the effects of DCS and treatment for fetal development using pregnant hamsters. They induced DCS by exposing hamsters to 7.1 ATA of compressed air breathing for 40 min. Comparisons were then made between three groups of pregnant hamsters: (1) those that developed DCS; (2) those that did not; and (3) a control (nondivided) group. Maternal DCS left untreated resulted in frequent and severe teratogenic effects. Furthermore, fetuses from females that did not develop DCS were significantly smaller at term than those from control animals. However, fetuses from females with DCS that were treated with HBOT did not differ markedly from controls. Accordingly, these experimental results support our hypothesis that prioritizing the stabilization of critically ill mothers with HBOT would also be beneficial for normal fetal development. One limitation associated with the present study was that no report investigated the effects of HBOT on all stages of pregnancy, and none evaluated the condition and development of the neonate after birth. In addition, experimental animals are not equal to humans, especially with regard to high cortical and visual functions. Moreover, only successful cases might have been previously reported. Furthermore, there have been no clinical reports of HBOT being applied to pregnant woman in a life‐threatening situation except for CO poisoning. Accordingly, further accumulation of clinical evidence concerning the effects of HBOT on pregnant women, fetuses, and development after birth is needed. CONCLUSION Standard HBOT for pregnant women suffering from CO poisoning has been proven to be safe and is considered beneficial, reducing the severity of both maternal and fetal injury. For pregnant women facing a life‐threatening situation, for which the effectiveness of HBOT has previously been suggested, prioritizing the stabilization of the mother may also be beneficial for normal fetal development. CONFLICT OF INTEREST STATEMENT We do not have conflict interest to declare. ETHICS STATEMENT Approval of the research protocol: Juntendo University review board (298). Informed consent: N/A. Registry and the registration no. of the study/trial: N/A. Animal studies: N/A. ACKNOWLEDGMENTS This work was supported in part by a Grant‐in‐Aid for Special Research in Subsidies for ordinary expenses of private schools from The Promotion and Mutual Aid Corporation for Private Schools of Japan. DATA AVAILABILITY STATEMENT We do not have data are available. ==== Refs REFERENCES 1 Ortega MA , Fraile‐Martinez O , García‐Montero C , Callejón‐Peláez E , Sáez MA , Álvarez‐Mon MA , et al. A general overview on the hyperbaric oxygen therapy: applications, mechanisms and translational opportunities. Medicina (Kaunas). 2021;57 :864.34577787 2 Kirby JP , Snyder J , Schuerer DJE , Peters JS , Bochicchio GV . Essentials of hyperbaric oxygen therapy: 2019 review. Mo Med. 2019;116 :176–179.31527935 3 Sola A , Chow L , Rogido M . Retinopathy of prematurity and oxygen therapy: a changing relationship. An Pediatr. 2005;62 :48–63. 4 Lane MKM , Garedew M , Deary EC , Coleman CN , Ahrens‐Víquez MM , Erythropel HC , et al. What to expect when expecting in lab: a review of unique risks and resources for pregnant researchers in the chemical laboratory. Chem Res Toxicol. 2022;35 :163–198.35130693 5 Wells PG , McCallum GP , Chen CS , et al. Oxidative stress in developmental origins of disease: teratogenesis, neurodevelopmental deficits, and cancer. Toxicol Sci. 2009;108 :4–18.19126598 6 Ambiru S , Furuyama N , Kimura F , Shimizu H , Yoshidome H , Miyazaki M , et al. Hyperbaric oxygen therapy as a prophylactic and treatment against ileus and recurrent intestinal obstruction soon after surgery to relieve adhesive intestinal obstruction. J Gastroenterol Hepatol. 2008;23 :e379–e383.17593223 7 Fukami Y , Kobayashi S , Sekoguchi E , Kurumiya Y . Randomized controlled trial of hyperbaric oxygen therapy in adhesive postoperative small bowel obstruction. Langenbecks Arch Surg. 2018;403 :555–559.29808324 8 Reid RL , Lorenzo M . SCUBA diving in pregnancy. J Obstet Gynaecol Can. 2018;40 :1490–1496.29503252 9 Culnan DM , Craft‐Coffman B , Bitz GH , Capek KD , Tu Y , Lineaweaver WC , et al. Carbon monoxide and cyanide poisoning in the burned pregnant patient: an indication for hyperbaric oxygen therapy. Ann Plast Surg. 2018;80 :S106–S112.29461288 10 Friedman P , Guo XM , Stiller RJ , Laifer SA . Carbon monoxide exposure during pregnancy. Obstet Gynecol Surv. 2015;70 :705–712.26584719 11 Kao LW , Nañagas KA . Toxicity associated with carbon monoxide. Clin Lab Med. 2006;26 :99–125.16567227 12 Gozubuyuk AA , Dag H , Kacar A , Karakurt Y , Arica V . Epidemiology, pathophysiology, clinical evaluation, and treatment of carbon monoxide poisoning in child, infant, and fetus. North Clin Istanb. 2017;4 :100–107.28752154 13 Jennings RT . Women and the hazardous environment: when the pregnant patient requires hyperbaric oxygen therapy. Aviat Space Environ Med. 1987;58 :370–374.3579829 14 Eleftheriou G , Butera R , Lonati D , Ferruzzi M , Costa M , Ferani R , et al. Open issues in management of carbon monoxide poisoning in pregnancy: practical suggestions. J Obstet Gynaecol. 2022;42 :2540–2541.35648870 15 Kosaki Y , Maeyama H , Nojima T , Obara T , Nakao A , Naito H . Carbon monoxide poisoning during pregnancy treated with hyperbaric oxygen. Clin Case Rep. 2021;9 :e04138.34026172 16 Gabrielli A , Layon AJ . Carbon monoxide intoxication during pregnancy: a case presentation and pathophysiologic discussion, with emphasis on molecular mechanisms. J Clin Anesth. 1995;7 :82–87.7772366 17 Greingor JL , Tosi JM , Ruhlmann S , Aussedat M . Acute carbon monoxide intoxication during pregnancy. One case report and review of the literature. Emerg Med J. 2001;18 :399–401.11559621 18 Brown DB , Mueller GL , Golich FC . Hyperbaric oxygen treatment for carbon monoxide poisoning in pregnancy: a case report. Aviat Space Environ Med. 1992;63 :1011–1014.1445151 19 Van Hoesen KB , Camporesi EM , Moon RE , Hage ML , Piantadosi CA . Should hyperbaric oxygen be used to treat the pregnant patient for acute carbon monoxide poisoning? A case report and literature review. JAMA. 1989;261 :1039–1043.2644457 20 Mandal NG , White N , Wee MY . Carbon monoxide poisoning in a parturient and the use of hyperbaric oxygen for treatment. Int J Obstet Anesth. 2001;10 :71–74.15321656 21 Santos LR , Alves‐Correia M , Câmara M , et al. Multiple victims of carbon monoxide poisoning in the aftermath of a wildfire: a case series. Acta Med Port. 2018;31 :146–151.29790465 22 Tchirikov M , Saling E , Bapayeva G , Bucher M , Thews O , Seliger G . Hyperbaric oxygenation and glucose/amino acids substitution in human severe placental insufficiency. Physiol Rep. 2018;6 :e13589.29536649 23 Ozgok‐Kangal K . Long‐term infant outcomes after hyperbaric oxygen treatment for acute carbon monoxide poisoning during pregnancy. Diving Hyperb Med. 2021;51 :248–255.34547775 24 Arslan A . Hyperbaric oxygen therapy in carbon monoxide poisoning in pregnancy: maternal and fetal outcome. Am J Emerg Med. 2021;43 :41–45.33497969 25 Elkharrat D , Raphael JC , Korach JM , Jars‐Guincestre MC , Chastang C , Harboun C , et al. Acute carbon monoxide intoxication and hyperbaric oxygen in pregnancy. Intensive Care Med. 1991;17 :289–292.1939875 26 Kreshak AA , Lawrence SM , Ontiveros ST , Castellano T , VanHoesen KB . Perinatal carbon monoxide poisoning: treatment of a 2‐hour‐old neonate with hyperbaric oxygen. AJP Rep. 2022;12 :e113–e116.35280718 27 Bray P , Myers RA , Cowley RA . Orogenital sex as a cause of nonfatal air embolism in pregnancy. Obstet Gynecol. 1983;61 :653–657.6835620 28 Hadanny A , Meir O , Bechor Y , Fishlev G , Bergan J , Efrati S . The safety of hyperbaric oxygen treatment–retrospective analysis in 2,334 patients. Undersea Hyperb Med. 2016;43 :113–122.27265988 29 Camporesi EM . Side effects of hyperbaric oxygen therapy. Undersea Hyperb Med. 2014;41 :253–257.24984321 30 Lai T , Xiang L , Liu Z , Mu Y , Li X , Li N , et al. Association of maternal disease and medication use with the risk of congenital heart defects in offspring: a case‐control study using logistic regression with a random‐effects model. J Perinat Med. 2019;47 :455–463.30794526 31 Martín‐Del‐Campo R , Sifuentes‐Romero I , García‐Gasca A . Hox genes in reptile development, epigenetic regulation, and Teratogenesis. Cytogenet Genome Res. 2019;157 :34–45.30566937 32 Calado AM , Dos Anjos Pires M . An overview of teratology. Methods Mol Biol. 2018;1797 :3–32.29896685 33 Palmer J , Von Rueden K . Carbon monoxide poisoning and pregnancy: critical nursing interventions. J Emerg Nurs. 2015;41 :479–483.26409658 34 Sapunar D , Saraga‐Babić M , Peruzović M , Marusić M . Effects of hyperbaric oxygen on rat embryos. Biol Neonate. 1993;63 :360–369.8373909 35 Gilman SC , Bradley ME , Greene KM , Fischer GJ . Fetal development: effects of decompression sickness and treatment. Aviat Space Environ Med. 1983;54 :1040–1042.6651733
PMC010xxxxxx/PMC10352571.txt
==== Front Acute Med Surg Acute Med Surg 10.1002/(ISSN)2052-8817 AMS2 Acute Medicine & Surgery 2052-8817 John Wiley and Sons Inc. Hoboken 10.1002/ams2.874 AMS2874 AMS-2023-0008.R2 Case Report Case Report Severe acute respiratory distress syndrome caused by Otsujito ARDS caused by Otsujito Hirasawa et al. Hirasawa Nobuhisa https://orcid.org/0000-0001-9329-0534 1 [email protected] Nakae Hajime https://orcid.org/0000-0003-3733-3530 1 Satoh Kasumi 1 Yoshida Kenji 1 Okuyama Manabu 1 1 Department of Emergency and Critical Care Medicine Akita University Graduate School of Medicine Akita Japan * Correspondence Nobuhisa Hirasawa, Department of Emergency and Critical Care Medicine, Akita University Graduate School of Medicine, 1‐1‐1 Hondo, Akita 010‐8543, Japan. Email: [email protected] 17 7 2023 Jan-Dec 2023 10 1 10.1002/ams2.v10.1 e87412 6 2023 15 1 2023 28 6 2023 © 2023 The Authors. Acute Medicine & Surgery published by John Wiley & Sons Australia, Ltd on behalf of Japanese Association for Acute Medicine. https://creativecommons.org/licenses/by-nc/4.0/ This is an open access article under the terms of the http://creativecommons.org/licenses/by-nc/4.0/ License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes. Abstract Background Kampo prescriptions can cause drug‐induced lung injury (DLI) and acute respiratory distress syndrome (ARDS). However, severe respiratory failure induced by Otsujito (OJT) is extremely rare. High‐dose steroid pulse therapy is generally given to patients with severe DLI. Case Presentation A 63‐year‐old man with respiratory distress was admitted to our hospital. The patient was diagnosed with severe ARDS caused by OJT, which had been prescribed 4 weeks prior to admission. Thus, OJT was discontinued, and intensive care for ARDS, including ventilation and prone positioning, was implemented. His respiratory condition rapidly improved after treatment with an initial methylprednisolone dose (1.5 mg/kg/day). He was extubated on day 4 and discharged on day 16. The steroid dose was gradually reduced and discontinued by day 116. Conclusion A severe case of ARDS caused by OJT was successfully treated with low‐dose steroids and specialized intensive care. A patient with severe acute respiratory distress syndrome caused by Otsujito was successfully treated with low‐dose steroids and specialized intensive care. Otsujito is widely prescribed in Japan, attention should be paid to its severe adverse effects. This case suggests that high‐dose steroids are not necessary for treating drug‐induced acute respiratory distress syndrome, even though it is recommended in Japanese guidelines for drug‐induced lung injury. abnormality drug‐induced Kampo Otsujito steroid source-schema-version-number2.0 cover-dateJanuary/December 2023 details-of-publishers-convertorConverter:WILEY_ML3GV2_TO_JATSPMC version:6.3.2 mode:remove_FC converted:18.07.2023 Hirasawa N , Nakae H , Satoh K , Yoshida K , Okuyama M . Severe acute respiratory distress syndrome caused by Otsujito. Acute Med Surg. 2023;10 :e874. 10.1002/ams2.874 ==== Body pmcBACKGROUND Kampo prescriptions are one of the causes of drug‐induced lung injury (DLI). In particular, Scutellaria root is a Kampo prescription known to cause DLI. Reports have shown that Sho‐saiko‐to and Saireito, which also contain Scutellaria root, caused severe DLI and acute respiratory distress syndrome (ARDS), leading to death. 1 Otsujito (OJT), used to relieve the symptoms of anal fissures and hemorrhoids, is made up of Japanese Angelica, Bupleurum, Scutellaria roots, Cimicifuga and Rhubarb rhizomes, and Glycyrrhiza. Cases of DLI caused by OJT have been reported, although there have been no reports of severe ARDS caused by OJT. Although high‐dose steroid pulse therapy is recommended for severe DLI in Japan, 2 the appropriate steroid dosage is still unknown. 3 Herein, we report a case of ARDS caused by OJT successfully treated with low‐dose steroid therapy and specialized intensive care, including prone positioning. CASE PRESENTATION A 63‐year‐old man presented to a local hospital with fever and hypoxia. The patient experienced respiratory distress and dizziness without cough for 1 day, prior to the first visit. He had a history of hypertension and subarachnoid hemorrhage, and was allergic to penicillin. He had been prescribed OJT 6.0 g/day (Kracie Holdings, Ltd) for the past 4 weeks. He was an ex‐smoker and had no episodes of resumption of smoking. His bodyweight was 55 kg. His initial vital signs were as follows: Glasgow Coma Scale score, 15 (E4V5M6); blood pressure, 114/86 mmHg; heart rate, 100 b.p.m.; respiratory rate, 42 breaths/min; oxygen saturation, 65% while breathing ambient air; and body temperature, 38.7°C. He was transferred to our hospital on the same day, intubated, and placed on a ventilator. The initial arterial blood gas showed pH 7.40, PaCO2 52 mmHg, PaO2 63 mmHg, HCO3 − 32.2 mmol/L, and lactate 10 mg/dL at FIO2 0.75. Blood tests showed an elevated inflammatory response, with a white blood cell count of 9700/mm3, eosinophil count of 0.0/mm3, and serum C‐reactive protein level of 18.5 mg/L. The following autoantibody screening was negative; antinuclear antibody, rheumatoid factor, myeloperoxidase‐antineutrophil cytoplasmic antibody, and proteinase‐3‐antineutrophil cytoplasmic antibody tests were negative. Bacteriological and virological tests did not detect any infectious pathogens, including Legionella pneumoniae, Mycoplasma pneumoniae, Chlamydia pneumoniae, cytomegalovirus, and severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2). Echocardiography revealed a normal ejection fraction of 60%. Chest radiography and computed tomography revealed bilateral ground‐glass opacity with air bronchograms and pleural effusion (Figure 1A,B). In view of the clinical presentation, mild hydrostatic pulmonary edema might be possible, however, the most plausible diagnoses were drug‐induced interstitial lung disease and severe ARDS due to OJT. He was admitted to the intensive care unit and started on meropenem and azithromycin, intravenous methylprednisolone (80 mg/day), and prone positioning (16 h/day). He was deeply sedated to suppress breathing efforts without neuromuscular blockade. In the ventilator setting, the tidal volume was limited to ≤8 mL/kg, and positive end‐expiratory pressure was maintained above 10 cmH2O. All oral medications were discontinued. FIGURE 1 Chest images of a 63‐year‐old man with severe acute respiratory distress syndrome caused by Otsujito. (A) Chest contrast‐enhanced computed tomography on admission, showing bilateral ground‐grass opacity with air bronchograms. (B) Supine chest radiograph on admission. (C) Erect chest radiograph on day 7 after acute treatment in the intensive care unit. On day 3, the patient's respiratory condition improved. The PaO2/FIO2 ratio was 238 at FIO2 0.40 and brain natriuretic peptide level was 35.2 pg/mL. Prone positioning and antimicrobial treatment were terminated, ventilator settings were weaned, and deep sedation was switched to light sedation. The patient was extubated on day 4. On day 7, his oxygen saturation was 97% on 1 L/min oxygen by nasal cannula, and the ground‐glass opacity was reduced (Figure 1C). Methylprednisolone was switched to oral prednisolone and the dose was reduced to 40 mg/day. The clinical course in the intensive care unit is described in Figure 2. On day 12, the Krebs von den Lungen‐6 (KL‐6) level was 798 U/mL. The patient was discharged on day 16. The dose of prednisolone was reduced based on clinical course indicators of interstitial lung disease such as KL‐6 and chest radiographs: 40 mg on day 7, 20 mg on day 10, 15 mg on day 20, 10 mg on day 26, 7.5 mg on day 40, 5 mg on day 61, 2.5 mg on day 89, and discontinued on day 116. FIGURE 2 Clinical course in the intensive care unit of a 63‐year‐old man with severe acute respiratory distress syndrome caused by Otsujito. DISCUSSION Two points are noteworthy: first, ARDS caused by OJT can be lethal, and second, Kampo‐induced ARDS was successfully treated with low‐dose steroids and intensive care including prone positioning. Kampo medicines can cause DLI and ARDS; however, this case is rare in that it was severe. In previous reports, the causative Kampo contained Scutellaria root in approximately 90% of the cases. The typical latent period of Kampo‐induced DLI is generally 1–12 weeks. 1 In this case, both the diagnostic criteria for DLI 2 and the Berlin definition for severe ARDS 4 were met. Otsujito was discontinued on day 1; on day 3, infection, autoimmune disease, and other causes were ruled out, and antimicrobial therapy was discontinued. Nevertheless, the patient's condition improved. We diagnosed DLI caused by OJT based on the clinical course. Eosinophilic pneumonia cannot be ruled out completely, however, there were no triggering episodes, such as the start of smoking. The resumption of OJT should have contributed to the diagnosis of DLI; however, it was not done due to the detriment of the patient. There are several types of DLI, with interstitial pneumonia being the most common. Three cases of DLI with interstitial pneumonia, due to OJT, have been reported previously, two of which were suspected to be ARDS. All cases showed good prognosis and ventilators were not required (Table 1). 5 , 6 , 7 In our case, severe ARDS occurred rapidly within 1 day of the onset of subjective symptoms of dyspnea. Although OJT is widely prescribed in Japan, attention should be paid to its adverse effects, such as severe DLI and its rapid progression. TABLE 1 Previous reports on adverse events with respiratory involvement related to Otsujito. No. Age (years) Computed tomography Latent period (weeks) Mechanical ventilation Initial steroid Prognosis 1 (ref. 5) 80 Bilateral GGO 8 No High‐dose pulse Survived Pleural effusion 2 (ref. 6) 53 Bilateral GGO 4 No No Survived 3 (ref. 7) 53 Bilateral GGO 2 No High‐dose pulse Survived Pleural effusion 4 (2022) 69 Bilateral GGO 4 Yes 1.5 mg/kg Survived Pleural effusion Abbreviation: GGO, ground grass opacity. Steroids are commonly used in treating DLI. The appropriate steroid dosage and dose‐reduction timing remain controversial. Japanese guidelines for DLI recommend that initial treatment of drug‐induced ARDS should include 500–1000 mg/day of methylprednisolone for 3 days, continued at a prednisolone equivalent of 0.5–1.0 mg/kg/day, and tapered off. 2 However, the benefits of high‐dose steroid pulse therapy have not been established yet. However, guidelines for ARDS recommend a methylprednisolone dose of 1–2 mg/kg/day initially. 4 Recent clinical studies suggest that early and prolonged use of low‐dose steroids can improve outcomes for patients with ARDS. 8 Conversely, high‐dose steroids could be harmful and increase mortality and ventilator‐free days. 9 It is unclear how the steroid dose can be reduced, and there is variability among studies. Short‐term reduction or discontinuation of steroids could elevate reintubation; therefore, even if ARDS improves and the patient can be weaned from the ventilator, it is better to taper the dose off gradually. 8 In our case, the patient's condition improved substantially after discontinuation of the causative agent and initial methylprednisolone treatment (1.5 mg/kg) with intensive care, including prone positioning. Compared to previous reports, 8 the initial steroid dose period was short and the time to completion was long. It is possible that the initial steroid dose period should have been prolonged. The reason it took 116 days to complete steroids was due to residual findings of drug‐induced interstitial pneumonia. This case suggests that high‐dose steroids are not necessary for treating Kampo‐induced ARDS, even in severe cases. However, it is difficult to determine which treatment was successful and the extent of success. The appropriate steroid dose for DLI types other than ARDS is not clear yet. Therefore, further studies are needed to determine the appropriate steroid dose for DLI. CONCLUSION Otsujito can cause severe ARDS. A patient with severe ARDS caused by OJT was successfully treated with low‐dose steroid and intensive care, including prone positioning. CONFLICT OF INTEREST STATEMENT The authors declare no conflicts of interest. ETHICS STATEMENT Approval of the research protocol: N/A. Informed consent: Written informed consent was obtained from the patient. Registry and the registration no. of the study/trial: N/A. Animal studies: N/A. DATA AVAILABILITY STATEMENT Data sharing is not applicable to this article as no new data were created or analyzed in this paper. ==== Refs REFERENCES 1 Enomoto Y , Nakamura Y , Enomoto N , Fujisawa T , Inui N , Suda T . Japanese herbal medicine‐induced pneumonitis: a review of 73 patients. Respir Investig. 2017;64 :138–144. 2 Kubo K , Azuma A , Kanazawa M , Kameda H , Kusumoto M , Genma A , et al. Consensus statement for the diagnosis and treatment of drug‐induced lung injuries. Respir Investig. 2013;51 :260–277. 3 Skeoch S , Weatherley N , Swift AJ , Oldroyd A , Johns C , Hayton C , et al. Drug‐induced interstitial lung disease: a systematic review. J Clin Med. 2018;7 :356.30326612 4 Tasaka S , Ohshimo S , Takeuchi M , Yasuda H , Ichikado K , Tsushima K , et al. ARDS clinical practice guideline 2021. J Intensive Care. 2022;10 :32.35799288 5 Tsuji T , Ihata A , Azuma C , Igarashi S. A case of Otsuji‐to‐induced pneumonitis. Jpn J Chest Dis. 1999;11 :852–856. (in Japanese). 6 Takeshita K , Saisho Y , Kitamura K , Kaburagi N , Funabiki T , Inamura T , et al. Pneumonitis induced by Scutellaria root (Scullcap). Intern Med. 2001;40 :764–768.11518120 7 Hiraya D , Kagohashi K , Satoh H . Pneumonitis due to an herbal medicine. Otsu‐Ji‐to Eur J Intern Med. 2010;3 :244. 8 Meduri GU , Bridges L , Shih MC , Marik PE , Siemieniuk RAC , Kocak M . Prolonged glucocorticoid treatment is associated with improved ARDS outcomes: analysis of individual patients' data from four randomized trials and trial‐level meta‐analysis of the updated literature. Intensive Care Med. 2016;42 :829–840.26508525 9 Kido T , Muramatsu K , Asakawa T , Otsubo H , Ogoshi T , Oda K , et al. The relationship between high‐dose corticosteroid treatment and mortality in acute respiratory distress syndrome: a retrospective and observational study using a nationwide administrative database in Japan. BMC Pulm Med. 2018;18 :28.29415701
PMC010xxxxxx/PMC10352592.txt
==== Front Respirol Case Rep Respirol Case Rep 10.1002/(ISSN)2051-3380 RCR2 Respirology Case Reports 2051-3380 John Wiley & Sons, Ltd Chichester, UK 10.1002/rcr2.1190 RCR21190 Case Report Case Reports Rifampicin‐induced acute tubulointerstitial nephritis during pulmonary tuberculosis treatment: A case report RIFAMPICIN‐INDUCED ACUTE RENAL INJURY Moussa et al. Moussa Chirine https://orcid.org/0000-0002-8123-9843 1 2 [email protected] Esbaa Samia 1 2 Rouis Houda 1 2 Sellami Nada 2 3 Hajji Meriam 2 3 Houcine Yoldez 2 4 Khattab Amel 1 Khouaja Ibtihel 1 Zendah Ines 1 2 Maâlej Sonia 1 2 1 Pneumology Department 1 Abderrahmen Mami Hospital Ariana Tunisia 2 Faculty of Medicine of Tunis El Manar University Tunis Tunisia 3 Internal Medicine A Charles Nicolle Hospital Tunis Tunisia 4 Pathology Department Abderrahmen Mami Hospital Ariana Tunisia * Correspondence Chirine Moussa, Pneumology Department 1, Abderrahmen Mami Hospital, 2080 Ariana, Tunisia. Email: [email protected] 17 7 2023 8 2023 11 8 10.1002/rcr2.v11.8 e0119006 6 2023 02 7 2023 © 2023 The Authors. Respirology Case Reports published by John Wiley & Sons Australia, Ltd on behalf of The Asian Pacific Society of Respirology. https://creativecommons.org/licenses/by-nc/4.0/ This is an open access article under the terms of the http://creativecommons.org/licenses/by-nc/4.0/ License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes. Abstract Drug‐induced tubulointerstitial nephritis is an uncommon complication in patients on anti‐tuberculosis therapy that can lead to permanent kidney damage. Rifampicin is the most offending drug. We report a case of a 41‐years old man being treated for pulmonary tuberculosis and presenting with tubulointerstitial nephritis associated with rifampicin. We focus on diagnosis features and therapeutic challenges. We present a case of rifampicin‐induced acute renal failure due to acute tubulointerstitial nephritis (ATIN), successfully managed with second‐line anti‐TB treatment. acute kidney injury acute renal failure acute tubulointerstitial nephritis rifampicin tuberculosis source-schema-version-number2.0 cover-dateAugust 2023 details-of-publishers-convertorConverter:WILEY_ML3GV2_TO_JATSPMC version:6.3.2 mode:remove_FC converted:18.07.2023 Moussa C , Esbaa S , Rouis H , Sellami N , Hajji M , Houcine Y , et al. Rifampicin‐induced acute tubulointerstitial nephritis during pulmonary tuberculosis treatment: A case report. Respirology Case Reports. 2023;11 :e01190. 10.1002/rcr2.1190 Associate Editor: Andrea Ban Yu‐Lin ==== Body pmcINTRODUCTION Tuberculosis (TB) is a global health problem that accounted for 1.5 million deaths in 2020. 1 Prompt treatment remains the most effective intervention to control the spread of the disease. Adverse reactions to first‐line anti‐TB medications compromise drug adherence and worsen anti‐TB treatment outcomes. Rifampicin (RIF) is the key drug in the treatment of TB. It has often been associated with hepatotoxic response and gastro enteropathy. 2 Only a few studies have investigated RIF‐induced nephrotoxicity. We present a case of RIF‐induced acute renal failure due to acute tubulointerstitial nephritis (ATIN), successfully managed with second‐line anti‐TB treatment. CASE REPORT A 41‐years old patient was admitted for acute renal failure. He has been diagnosed with a drug susceptible active pulmonary TB affecting both lungs. The initial diagnostic evaluation included a sputum smear examination for acid‐fast bacilli (AFB) using Ziehl‐Neelsen staining. The sputum smear results were positive for AFB, confirming the presence of active pulmonary tuberculosis. The patient was receiving first‐line anti‐TB drugs daily for 40 days according to the four‐drug fixed‐dose combination regimen: RIF (450 mg), isoniazid (INH) (225 mg), pyrazinamide (PZA) (1200 mg), and ethambutol (EMB) (825 mg). The patient reported taking the medication regularly at a dose of 4 tablets per day. No exposure to other medication was reported. The patient's renal function was normal before TB treatment. On admission, the patient presented with asthenia and fatigue. A physical examination revealed a body weight of 43 kg, a body temperature of 36.7°C, a heart rate of 110 beats per minute, a blood pressure of 113/68 mmHg, and an oxygen saturation of 96% on room air. Crackles in the left lung were noted. The sputum smear results were still positive for AFB. Chest x‐ray (Figure 1) screening showed bilateral pulmonary infiltrates and cavitary lesions on the upper lung and no significant improvement compared to the initial presentation. Creatinine levels rose from 60 μmol/L before treatment to 220 μmol/L, with a creatinine clearance of 23 mL/min. The urea level was 6 μmol/L. Other laboratory test analyses showed; elevated inflammatory biomarkers (white blood cell count: 20600 cells/μL, C‐reactive protein: 70 mg/dL mg/dL). Liver function tests revealed no abnormalities, indicating that there were no signs of liver dysfunction associated with the antituberculous treatment. HIV screening was negative. The etiological workup for renal insufficiency was initiated. Urine output was normal; eliminating obstructive causes of renal insufficiency. Autoimmune disorders such as systemic lupus erythematosus or granulomatosis with polyangiitis were ruled out based on negative immunological testing, and the absence of proteins and blood in the urine analysis. Renal sizes and cortical echogenicity were normal on kidney ultrasound. FIGURE 1 Chest x‐ray, at 40 days of treatment: Bilateral pulmonary infiltrates and cavitary lesions on the upper lobes. Renal tuberculosis or a drug‐induced non‐oliguric acute kidney injury (AKI) were suspected. Medications were held. A kidney biopsy‐performed 11 days later—confirmed ATIN. The appearance is consistent with acute tubulointerstitial nephritis (ATIN). This ATIN occurred on a background of non‐proliferative IgA mesangial deposition nephritis, classified as E0 M0 S0 T0 C0 according to the Oxford classification. Neither tuberculous granuloma nor acid‐fast bacilli were observed (Figure 2). FIGURE 2 Histology of kidney biopsy: Tubular necrosis, Masson's Trichrome ×40. The patient remained hospitalized for 12 weeks with close supervision. The antituberculous treatment was withheld until the normalization of creatinine levels. Serum creatinine levels progressively decreased to near baseline within 11 weeks (Figure 3). The suspicion of rifampicin as the causative agent was based on a comprehensive review of the literature and consultation with pharmacovigilance experts. FIGURE 3 Clinical course of creatinine level. The decision to initiate a treatment based on second‐line anti‐TB drugs was made following a multidisciplinary decision and considering that rifampicin is a key antitubercular agent. The new therapeutic regimen proposed consists of INH (225 mg), EMB (800 mg), PZA (1000 mg), levofloxacin (500 mg), ethionamide (500 mg) and cycloserine (500 mg) daily, along with pyridoxine supplementation. The culture‐negative conversion was achieved at 12 months. He completed 18 months of treatment without a relapse of renal failure. The chest x‐ray at the end of treatment showed radiological improvement with persistent left‐sided retractive sequelae (Figure 4). FIGURE 4 The chest x‐ray at the end of treatment showed radiological improvement with persistent left‐sided retractive sequelae. DISCUSSION Tubulointerstitial nephritis is an uncommon complication in patients on anti‐TB therapy. It can lead to permanent kidney damage. 1 Although RIF is the most frequently involved drug in this complication, other first‐line anti‐TB drugs, such as INH, ETB, and PYZ were also associated with AKI. 1 , 2 , 3 , 4 , 5 In a prospective study, the incidence of AKI during anti‐TB treatment was 10.3%. This percentage is higher than reported in previous retrospective studies (0.05%–7.1%). 1 RIF‐induced AKI usually occurs in patients who have previously taken this drug or who are undergoing intermittent treatment. 2 , 3 It is assumed that previous or intermittent exposure to RIF triggers an immune response. 3 Upon re‐exposure, anti‐RIF antibodies form immune complexes that are deposited in the renal vessels, the interstitial area, and the glomerular endothelium, leading to acute tubular necrosis, and acute interstitial nephritis. 2 , 6 The diagnosis of RIF‐induced AKI is based on the typical disease course and by excluding other potential etiologies. 6 It is important to identify the aetiology, as management may differ based on biopsy results. 4 When a biopsy is performed, ATIN and acute tubular necrosis are the most common histopathological findings during RIF‐induced AKI. 1 The kidney biopsy is the gold standard for the diagnosis of ATIN. 5 In cases where renal biopsy is contraindicated or not feasible, The diagnosis can be based on a compatible clinical course and laboratory findings. such as eosinophiluria and accumulation of 67Ga in the bilateral kidneys. 4 In the present case, acute renal failure occurred during anti‐TB treatment and improved after the medications were discontinued. The pathological findings on the kidney biopsy specimen were consistent with drug‐induced nephrotoxicity. Features of a non‐proliferative IgA nephropathy (IgAN) were also found. Confirming the aetiology of the IgAN remains challenging. In our case, rifampicin was the most implicated drug. Although isoniazid and ethambutol can also potentially cause acute interstitial nephritis, the available evidence and clinical experience strongly implicated rifampicin in this particular case. The assessment of causality (imputability) in pharmacovigilance involves considering various factors, including temporal relationship, known drug‐related adverse effects, alternative explanations, dechallenge/rechallenge, and published case reports. In this case, the temporal relationship between rifampicin administration and the onset of renal symptoms, along with the absence of other identifiable causes, led to a high probability of rifampicin's involvement in the development of acute interstitial nephritis. According to Chang et al., more than 50% of AKI cases occurred within 2 months of starting TB treatment. The onset of AKI was more commonly seen in older patients with a higher baseline estimated glomerular filtration rate and blood eosinophil count (>350 (109/L)). 1 Although chronic kidney disease has not been associated with AKI occurrence, it is, along with hypoalbuminemia, a possible risk factor for severe and permanent kidney damage. 4 , 6 Dehydration and hypoperfusion in patients with gastrointestinal disturbance can participate in renal function impairment. Therefore, fluid management and correct regimen modification are crucial to prevent further injuries. 6 Renal recovery is better in acute interstitial nephritis than in acute tubular necrosis. 6 The prognosis in acute interstitial nephritis is good, with a 1.6% mortality and a recovery rate between 73% and 100%. 1 , 3 But it can lead to serious complications such as Fanconi syndrome, resulting in bone pain and fracture, fatigue, and muscular weakness. 3 Tuberculosis treatment in the setting of ATIN is challenging. The offending medication should be discontinued as soon as possible. 5 In our case, the collegial decision was an alternative regimen based on second‐line anti‐TB treatment usually used for drug‐resistant TB, without RIF. Given the importance of rifampicin in standard anti‐tuberculous regimens, the exclusion of rifampicin necessitated an alternative treatment approach. Considering this situation, the management was approached as a possible case of drug‐resistant tuberculosis. Hence, a longer duration of therapy, involving a combination of multiple drugs such as isoniazid, ethambutol, pyrazinamide, levofloxacin, ethionamide and cycloserine, was initiated to ensure effective treatment and minimize the risk of relapse. Kizilbash et al. also reported using a drug‐resistant regimen based on linezolid, moxifloxacin, ethambutol and high doses of INH in such circumstances. 5 There are no clear recommendations for the management of RIF‐induced AKI. However, several therapeutic approaches have been reported in the literature. In some studies, authors have utilized steroid therapy in patients with a pathological or clinical diagnosis of acute interstitial nephritis (AIN) to expedite the recovery of renal function. However, it is important to note that the use of steroids for AIN remains a topic of controversy. 7 , 8 We debated the use of steroids however, given the potential for corticosteroids to exacerbate the underlying tuberculosis infection, it was decided to refrain from their use in order to avoid potential worsening of the initial disease. Several studies support RIF desensitization in patients with RIF‐induced AKI. Although the rifampicin desensitization protocol varies, success rates are high (80%–82%). 6 According to other studies, patients with RIF‐induced ATIN may experience more severe kidney damage if they restart RIF, even at lower doses. Desensitization therapy should be avoided. 4 Levofloxacin may be an alternative to rifampicin thanks to its safety and potency. A culture‐negative conversion in drug‐susceptible TB was observed using levofloxacin instead of RIF for at least 18 months with no major side effects and no AKI relapse. 4 In our case, we saw that adding levofloxacin was not sufficient. According to the national tuberculosis guidelines in Tunisia, the rationale for selecting a second‐line treatment was based on the risk of developing drug‐resistant tuberculosis due to the interruption of an effective treatment for 11 weeks. 9 The patient did not receive an adequate duration of the intensive phase of treatment, as evidenced by the lack of radiological improvement and positive direct examination results. The decision was made by a panel of experts. The selection of an unconventional treatment approach can be considered a limitation of this study. However, the favourable progression of the patient's condition and complete recovery serve as evidence that this therapeutic choice is a viable option in countries with a high tuberculosis endemicity. The key learning points of this case are that rifampicin‐induced AKI typically occurs during the discontinuation of rifampicin treatment. However, this case demonstrates that even with regular administration of the treatment, rifampicin can still lead to renal failure. A definitive diagnosis of ATIN requires renal biopsy as noninvasive laboratory tests and imaging studies lack sensitivity and specificity. The choice of treatment to replace rifampicin is challenging as it involves selecting a second‐line antitubercular drug, and there are multiple treatment regimens available, with or without corticosteroids. The decision should be made within the context of a multidisciplinary approach, involving expert opinions and a collegial decision‐making process. In conclusion, drug‐induced ATIN is a rare and serious adverse reaction of anti‐TB treatment. Early diagnosis and discontinuation of the offending medication are important to prevent further kidney damage and to promote renal function recovery. Conducting large‐scale studies to establish clear management protocols is necessary. AUTHOR CONTRIBUTIONS Chirine Moussa conceived the project. Chirine Moussa and Samia Esbaa worte the manuscript. Nada Sellami and Meriam Hajji gave the data. Yoldez Houcine, Houda Rouis, Amel Khattab, and Ibtihel Khouaja revised the manuscript critically for important intellectual content. Sonia Maâlej and Ines Zendah gave final approval for the version to be published. CONFLICT OF INTEREST STATEMENT None declared. ETHICS STATEMENT The authors declare that appropriate written informed consent was obtained for the publication of this manuscript and accompanying images. DATA AVAILABILITY STATEMENT Data sharing not applicable to this article as no datasets were generated or analysed during the current study. ==== Refs REFERENCES 1 Chang CH , Chang LY , Ko JC , Wen YF , Chang CJ , Keng LT , et al. Incidence of and risk factors for acute kidney injury during Antituberculosis treatment: a prospective cohort study and literature review. Infect Dis Ther [Internet]. 2023 Feb 11 [cited 2023 Feb 26]; Available from;12 :919–931. 10.1007/s40121-023-00761-w 36773200 2 Chiba S , Tsuchiya K , Sakashita H , Ito E , Inase N . Rifampicin‐induced acute kidney injury during the initial treatment for pulmonary tuberculosis: a case report and literature review. Intern Med [Internet]. 2013;52 (21 ):2457–2460. [cited 2021 Oct 9] Available from: https://www.jstage.jst.go.jp/article/internalmedicine/52/21/52_52.0634/_article 24190152 3 Beebe A , Seaworth B , Patil N . Rifampicin‐induced nephrotoxicity in a tuberculosis patient. J Clin Tuberc Other Mycobact Dis [Internet]. 2015;1 :13–15. [cited 2023 Feb 26] Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6850238/ 31723676 4 Sakashita K , Murata K , Takahashi Y , Yamamoto M , Oohashi K , Sato Y , et al. A case series of acute kidney injury during anti‐tuberculosis treatment. Intern Med [Internet]. 2019;58 (4 ):521–527. [cited 2021 Oct 9] Available from: https://www.jstage.jst.go.jp/article/internalmedicine/58/4/58_0813-18/_article 30333388 5 Kizilbash Q . Successful management of acute interstitial nephritis in two cases of disseminated tuberculosis. Tuberculosis (Edinb). 2016;101S :S135–S136.27729256 6 Chang CH , Chen YF , Wu VC , Shu CC , Lee CH , Wang JY , et al. Acute kidney injury due to anti‐tuberculosis drugs: a five‐year experience in an aging population. BMC Infect Dis [Internet]. 2014;14 (1 ):23. [cited 2021 Oct 9] Available from: 10.1186/1471-2334-14-23 24410958 7 Perazella MA , Markowitz GS . Drug‐induced acute interstitial nephritis. Nat Rev Nephrol. 2010;6 (8 ):461–470.20517290 8 Muriithi AK , Leung N , Valeri AM , Cornell LD , Sethi S , Fidler ME , et al. Biopsy‐proven acute interstitial nephritis, 1993‐2011: a case series. Am J Kidney Dis. 2014;64 (4 ):558–566.24927897 9 30102018Guide‐PNLT‐2018.pdf [Internet]. [cited 2023 Jun 23]. Available from: http://www.santetunisie.rns.tn/images/docs/anis/actualite/2018/octobre/30102018Guide-PNLT-2018.pdf
PMC010xxxxxx/PMC10352596.txt
==== Front J Am Coll Emerg Physicians Open J Am Coll Emerg Physicians Open 10.1002/(ISSN)2688-1152 EMP2 Journal of the American College of Emergency Physicians Open 2688-1152 John Wiley and Sons Inc. Hoboken 10.1002/emp2.13001 EMP213001 Original Research The Practice of Emergency Medicine Factors that influence interprofessional implementation of trauma‐informed care in the emergency department LEWIS‐O'CONNOR et al. Lewis‐O'Connor Annie NP, PhD https://orcid.org/0000-0003-1952-8027 1 2 [email protected] * Olson Rose MD 3 Grossman Samara LICSW 4 Nelson Derek BS 1 Levy‐Carrick Nomi MD, MPhil 4 5 Stoklosa Hanni MD, MPH 6 7 Banning Stephanie MD 3 Rittenberg Eve MD 1 2 1 Division of Women's Health, Department of Medicine Brigham and Women's Hospital Boston Massachusetts USA 2 Harvard Medical School Boston Massachusetts USA 3 Department of Medicine, Brigham and Women's Hospital Boston Massachusetts USA 4 Department of Psychiatry Brigham and Women's Hospital Boston Massachusetts USA 5 Department of Psychiatry Harvard Medical School Boston Massachusetts USA 6 Department of International Emergency Medicine and Humanitarian Programs Brigham and Women's Hospital Boston Massachusetts USA 7 Department of Medicine Harvard Medical School Boston Massachusetts USA * Correspondence Annie Lewis‐O'Connor, NP, PhD, Division of Women's Health, Department of Medicine, Brigham and Women's Hospital, Boston, MA 02116, USA. Email: [email protected] 17 7 2023 8 2023 4 4 10.1002/emp2.v4.4 e1300130 5 2023 05 4 2023 09 6 2023 © 2023 The Authors. JACEP Open published by Wiley Periodicals LLC on behalf of American College of Emergency Physicians. https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an open access article under the terms of the http://creativecommons.org/licenses/by-nc-nd/4.0/ License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non‐commercial and no modifications or adaptations are made. Abstract Background To describe factors that influence interprofessional staff decisions and ability to implement trauma‐informed care (TIC) in a level‐one emergency department (ED) trauma center. Methods This qualitative research study consisted of semi‐structured interviews and quantitative surveys that were conducted between March and December 2020 at an urban trauma center. Eligible participants were staff working in the ED. Interview questions were developed using the Theoretical Domains Framework (TDF), which is designed to identify influences on health professional behavior related to implementation of evidence‐based recommendations. Interview responses were transcribed, coded using Atlas software, and analyzed using thematic analysis. Results Key themes identified included awareness of TIC principles, impact of TIC on staff and patients, and experiences of bias. Participants identified opportunities to improve care for patients with a trauma history, including staff training, more time with patients, and efforts to decrease bias toward patients. Most participants (85.7%) felt that a TIC plan, tiered trauma inquiry, and warm handovers would be easy or very easy to implement. Conclusion We identified key interprofessional staff beliefs and attitudes that influence implementation of TIC in the ED. These factors represent potential individual, team‐based, and organizational targets for behavior change interventions to improve staff response to patient trauma and to address secondary trauma experienced by ED staff. emergency department staff interdisciplinary teams re‐traumatization trauma trauma‐informed care Robert Wood Johnson Foundation 10.13039/100000867 75720 source-schema-version-number2.0 cover-dateAugust 2023 details-of-publishers-convertorConverter:WILEY_ML3GV2_TO_JATSPMC version:6.3.2 mode:remove_FC converted:18.07.2023 Lewis‐O'Connor A , Olson R , Grossman S , et al. Factors that influence interprofessional implementation of trauma‐informed care in the emergency department. JACEP Open. 2023;4 :e13001. 10.1002/emp2.13001 By JACEP Open policy, all authors are required to disclose any and all commercial, financial, and other relationships in any way related to the subject of this article as per ICMJE conflict of interest guidelines (see www.icmje.org). The authors have stated that no such relationships exist. Supervising Editor: Christian Tomaszewski, MD, MS. ==== Body pmc1 INTRODUCTION 1.1 Background Traumatic experiences present a crucial challenge to emergency medicine because trauma has been shown to negatively impact health, influence how patients engage in health care, and affect the well‐being and safety of emergency department (ED) staff. Trauma is defined as experiences, on levels ranging from individual to collective, that cause physical, emotional, or life‐threatening harm and have long‐lasting impacts on health and well‐being. 1 In childhood, exposure to trauma has a dose‐response association with chronic conditions including cardiovascular disease, type 2 diabetes, and psychiatric disorders. 2 , 3 , 4 Trauma disproportionately impacts people of color and those with low‐socioeconomic status (SES) and contributes to health inequities. 5 , 6 , 7 , 8 , 9 , 10 In the ED, trauma is often understood to consist of physical injuries, such as stabbings or gunshot wounds, whereas emotional, social, and structural traumas may be overlooked. 1 , 11 The ED setting, with its sensory overload and perceived loss of privacy and control, can be a difficult environment for patients with a history of trauma. The resulting retraumatization can lead to behaviors rooted in a trauma response, such as patients being “combative” or labeled as “difficult.” 12 A trauma‐informed approach can allow ED staff to recognize these trauma responses and effectively help patients to de‐escalate and engage in care. In addition to providing effective tools to care for patients, trauma‐informed approaches are crucial for the well‐being of ED staff members, who frequently experience both direct and secondary (vicarious) trauma. 13 , 14 , 15 Trauma‐informed care (TIC) offers an approach and guiding principles that recognizes and mitigates the impact of trauma on health and promotes engagement in health care, and strengthens the well‐being of health care staff (Figure 1). The 4R's of TIC include: realizing how widespread trauma is and understanding paths for recovery; recognizing the signs and symptoms of trauma in patients and staff; responding using the guiding principles of TIC; and awareness of policies and procedures that may result in retraumatizing an individual. FIGURE 1 The 6 principles of trauma‐informed care. 1.2 Importance TIC strategies and their benefits have been well described 10 ; however, operationalization in health care settings has lagged. 16 , 17 , 18 , 19 , 20 The Theoretical Domains Framework (TDF) was developed to investigate implementation challenges in health care settings and provides a framework to identify potential behavior change interventions that could overcome these implementation barriers. 21 The COM‐B model is a simpler understanding that capability, opportunity, and motivation act together to influence behavior that can be used in combination with the TDF (Figure 2). FIGURE 2 Theoretical Domains Framework (TDF) and behavioral change wheel. Reproduced from Atkins et al,21 under the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0). No changes to the original figure were made. COM‐B sources of behavior are also shown, demonstrating how the TDF can be linked with this more simplified model. 1.3 Goals of this investigation Using the TDF and COM‐B frameworks, we conducted this study to gain an in‐depth understanding of the dynamic, complex interpersonal and environmental factors that impact staff's acceptance and understanding of TIC. Through this understanding, we aimed to describe the factors that influence interprofessional ED staff's understanding and acceptance of TIC, including 3 potential trauma‐informed strategies: TIC plan, tiered trauma inquiry, and warm handovers. 2 METHODS 2.1 Study design To explore the range of factors that could influence staff feasibility and acceptability of TIC implementation, we used a qualitative study design consisting of likert scaled survey questions and semi‐structured interviews. This approach followed the Consolidated Criteria for Reporting Qualitative Research (COREQ) reporting guideline as well as the Standards for Reporting Qualitative Research (SRQR) guideline. 22 2.2 Setting This study was conducted at an academic level I trauma center in Boston, Massachusetts. This study was approved by the institutional review board. 2.3 Selection of participants Eligibility criteria included English‐speaking staff who primarily work in patient‐facing roles in the ED, including physicians, nurses, social workers, security officers, medical assistants, domestic violence advocates, community health workers, and ultrasound technicians (Table 1). We actively recruited participants from December 2019 through March 2020 via purposive sampling with input from the ED leadership team to obtain a diverse range of interprofessional ED staff. Investigators emailed potential participants and assessed eligibility through a phone screening. Participants were recruited until thematic saturation was reached, resulting in 22 interviews. 23 TABLE 1 Characteristics of study participants. Characteristic No. (% of total responses) Gender a Women 17 (77.3) Men 4 (18.2) Prefer not to say 1 (4.5) Race Asian/Pacific Islander 2 (9.1) Black/African American 6 (27.3) Latin X/Hispanic 2 (9.1) Native American/Indigenous 1 (4.5) White 13 (59.1) Age, years 21–29 5 (22.7) 30–39 9 (40.9) 40–49 4 (18.2) 50–59 2 (9.1) 60 or older 2 (9.1) Role/position Physician 7(31.8) Nurse 4 (18.1) Social worker 3 (13.6) Emergency services assistant 2 (9.1) Security officer 2 (9.1) Ultrasound technician 1 (4.5) Domestic violence advocate 1 (4.5) Community health worker 1 (4.5) Physician assistant 1(4.5) a Participants self‐identified their gender. John Wiley & Sons, Ltd. 2.4 Study procedures Surveys and semi‐structured interviews were conducted between March and December 2020. All participants provided verbal informed consent. There was no financial compensation provided to participants. Surveys and interviews were conducted using video‐conferencing due to the onset of the COVID‐19 pandemic. Participants were given a list of potential risks of the study process and informed they could skip any question or end the interview at any point. All interviews were de‐identified, recorded via an encrypted digital audio recorder, and stored on password protected computers. Participants were asked to complete a short survey before the interview (Supporting Information Appendix S1). The survey answers were briefly referenced during the semi‐structured interview (Supporting Information Appendix S2) assessed factors that would influence staff's ability and willingness to implement TIC in the ED, including topics of prior knowledge and experience with TICs. The more granular questions aligned with our TDF Framework (Supporting Information Appendix S3) and explored specific aspects of behavior and implementation, whereas the COM‐B, a model that understands behavior as a sum of opportunity, motivation, and capability, allowed simplified understanding of the results. The semi‐structured interview guide (Supporting Information Appendix S2) was developed by the research team (A.L.O., E.R., N.L.C., S.G., H.S., J.L., S.B.) using the Theoretical Domains Framework (Supporting Information Appendix S3). 21 To explore potential TIC interventions that could be implemented the ED, the research team used a participatory approach, meeting with ED patient advisors, Coordinated Approach to Resilience and Empowerment Clinic patient advisors, and ED leadership. Based on these discussions and review of the literature, the research team selected 3 potential trauma‐informed interventions that could be further explored in the semi‐structured interviews. Participants were asked to rank their acceptance of these 3 interventions using a Likert scale. The interventions were as follows: warm handovers, broad trauma‐tiered inquiry, and TIC plans. A warm handover is a transfer of care between members of the health care team that occurs in front of the patient and/or their support individuals; it allows the patient to hear what is said, clarify or correct information, and ask questions about their care. 24 Broad trauma‐tiered inquiry is a discussion between patients and providers that elicits a trauma history using a humanistic approach. 25 A TIC plan is a document in the electronic health record (EHR) that is, developed with patients; it documents the patient's strengths, ways of coping, triggers, prior health care experiences, and any relevant trauma history that the patient feels is important to share with the care team. The Bottom Line Trauma informed care (TIC) can aid in recognition and engagement of vulnerable patients. A survey indicated that 86% of emergency department (ED) staff felt that TIC could be incorporated into daily ED care to improve both patient and staff experience. Five members of the research team developed the interview questions through an iterative process until consensus was achieved. Questions were categorized by TDF domains within a broader behavioral science framework of COM‐B influences on behavior by opportunity, motivation, and capability. The interviews began with an open‐ended question within each TDF domain, followed by a series of questions to probe more deeply into the target behavior. Participants were provided definitions and examples of each of the three potential TIC interventions. 2.5 Data analysis We summarized participants’ characteristics with proportions and percentages for categorical variables. Each interview was considered one unit of analysis and was transcribed verbatim using TranscribeMe software. Coding and analysis were conducted using qualitative analysis software, ATLAS.ti. To mitigate bias and enhance reflexivity, initial coding of transcripts was completed by two independent team members using an inductive approach. These inductive open codes were then combined into preliminary themes for the larger team (A.L.O., E.R., S.G., N.L.C., H.S.), which through regular meetings and discussion, we verified codes and discussed resolution of discrepancies and confirmation of final themes 26 , 27 Data saturation was determined when no new additional data were being identified. The survey questions regarding working in the ED and current TIC practices were analyzed, and percentages of responses were reported (Supporting Information Appendix S4). Last, we analyzed the survey responses about proposed TIC interventions and ease of implementation (Supporting Information Appendix S5). Data analysis was performed from March to December 2021. 3 RESULTS Of the 22 participants, 17 identified as women and 13 as White (Table 1). Most participants were 30–39 years old. Staff roles included 7 physicians, 4 nurses, 3 social workers, 2 emergency service assistants, 2 security officers, 1 physician assistant, 1 ultrasound technician, 1 community health worker, and 1 domestic violence advocate. 3.1 Survey results 3.1.1 Participant survey responses: perceived frequency of factors that influence TIC implementation Figure 3 demonstrates survey responses on perceived frequency of factors that influence staff implementation of TIC in the ED. Respondents reported frequently (40.9%) or usually (27.3%) feeling supported in the ED by other staff, and frequently (22.7%) or usually (27.3%) enjoying their work in the ED over the past week. Most participants (59.1%) stated that patients who have experienced trauma frequently or usually receive high quality care in the ED. Respondents reported feeling confident that they usually (22.7%), frequently (40.9%), or sometimes (31.8%) can positively affect the experience of a patient in the ED who has experienced trauma. Most participants reported addressing the emotional or behavioral distress that patients may experience, either every time (22.7%), usually (18.2%), or frequently (36.4%). FIGURE 3 Participant survey responses on perceived frequency of factors that influence decisions to implement trauma‐informed care in the emergency department. 3.1.2 Agreement with factors that influence TIC implementation Figure 4 demonstrates survey responses evaluating participant agreement with factors that influence staff implementation of TIC in the ED. Almost all participants agreed or strongly agreed (95.5%) they were optimistic that further TIC education and protocols would improve care for patients who have experienced trauma. Overall participants agreed or strongly agreed (100%) they were open to using TIC protocols, but with less certainty about other team members openness to TIC (31% somewhat agreeing). FIGURE 4 Participant survey responses on perceived agreement with factors that influence implementation of trauma‐informed care in the emergency department. 3.1.3 Perceived difficulty of TIC interventions Respondents were asked to rank the difficulty of implementing three potential TIC strategies (Figure 5). Eighteen participants (85.7%) felt a trauma‐informed acute care plan would be easy or very easy to implement in the ED. Fewer participants felt that trauma tiered inquiry (11 [52.3%]) and warm handovers (10 [47.6%]) would be easy or very easy to implement in the ED. FIGURE 5 Participant survey responses on perceived difficulty with implementation of trauma‐informed care interventions in the emergency department. 3.1.4 Perspectives on TIC training Ten participants reported minimal (7, 31.8%) to no (3, 13.6%) prior training on TIC, 6 (27.3%) reported some prior training, and 6 participants reported significant (4, 18.2%) to very significant (2, 9.1) prior training on TIC. Nineteen participants reported that ED staff training on TIC is either very important (13, 59.1%) or extremely important (6, 27.3%). 3.1.5 Thematic analysis Key themes emerging from participant interviews were categorized according to the COM‐B behavioral change framework and TDF domains (Figure 2). Six key themes emerged describing factors that influence staff decisions to accept/use TIC in the ED: self‐reported capability, environmental facilitators of change, environmental barriers to change, opportunities for change, and experiences of and witnessing implicit and explicit bias (Table 2). TABLE 2 Thematic analysis of factors that influence staff implementation of trauma‐informed care in the emergency department. COM‐B TDF domain Theme Subtheme Select quotations Capability Knowledge Awareness of TIC principles Familiarity with TIC “Yes, I have heard of it [laughter] and my understanding is it's really providing care that is centered around being able to understand a patient's experience with trauma and also providing appropriate sensitive care as well as understanding how medical care is affected by the experience of trauma that the patient has had.” Opportunity Environ‐mental context and resources Facilitators of staff emotional and psychological safety Debriefing difficult situations “They're doing nightly debriefing so that after each shift we can kind of talk about the hard things that happened and also some operational—what we could do better from an operational standpoint.” Team camaraderie “I feel like there's a really good sense of camaraderie and teamwork, and that everyone kind of lifts each other up or picks each other up when you've had a bad case…I mean, you will hear some of the sickest jokes that you'll ever hear. But in that moment, sometimes, it's what you need to just get you through to that next point.” Supportive leadership “We've got great leadership both at the physician and nursing level. An also an incredible social work department who have really kind of championed helping staff to deal with trauma. Hence, like I said, this trauma‐informed care initiative is coming at such a great time because I think the environment is right for it” Opportunity Environ‐mental context and resources Barriers to staff emotional and psychological safety Lack of debriefs “Most cases, we do not debrief. It's not talked about. And the nurse doesn't get that break that they need…I haven't really been able to talk to anybody about how I feel after a patient—or debrief it. And then most of the times, we don't even know what happens to our patients after they leave our department, whether dead or alive.” Opportunity Environ‐mental context and resources Opportunities to improve trauma patient experiences Staff training “For me, it would really be an education piece because I think that's the thing that's really lacking and that is foundational to any changes moving forward. Because if people don't understand what trauma‐informed care is, we're not going to be able to provide it” More time with patients “Having more time to explore some of these experiences that happened to the patient…And I want to be able to ask those questions. But again, I think there are many limitations. And one of them is usually the time.” Decreased bias toward patients “I'd also really like to see some sort of behind the scenes shift in the way patients are spoken about, including those who have dealt with trauma, and it's not kind of conducive to trauma‐informed care. If you don't give respect to a patient behind their back, I don't think you can necessarily give them the proper respect or care that they need to their face as well.” Opportunity Social influences Experiences of bias Racism Explicit bias: I think the example that I would most cite is patients with sickle cell disease who tend to be Black, although not always, are often perceived as being addicts rather than having a chronic health condition that requires chronic use of pain medications. And I have overheard racism in the tones of the way sometimes patients with sickle cell disease are treated.” Implicit bias: “There were times just walking through the hallway that you would happen to notice that the majority of the patients that are out in a hallway bed instead of in a room, with at the very least a curtain or a closed door, the majority of the people who were in the hallways were where people of color. And that's pretty disheartening to see.” Homophobia and transphobia Explicit bias: “’Don't be like, ‘I'm going to call it guy.’ No. That's just going to make it worse. And honestly, it has because even some people's like, ‘Look, it's clearly a dude. Why do I have to say ‘she’?’ And it's not your choice. How does it affect you?” Implicit bias: “But kind of how I mentioned before, homophobia, some people just aren't necessarily aware I guess of how to react to things.” Implicit bias: “It's one of those things that if we have a transgender person patient, sometimes there's a lot of discussion about it and I don't think they're served very well, very smoothly…it depends on whether or not someone else has talked to us about what they want their pronouns to be.” Sexism Implicit bias: “Agitation, aggression, they're treated differently if they're women versus men. I think for women that it's often kind of labeled as a kind of voluntary behavior and that women may have control over it, but they're not composing themselves. Whereas for men, a different kind of medical etiology is looked at under more scrutiny, if a male kind of exhibits the same type of behavior.” Mental health and substance use disorders Implicit bias: “Another group that gets that received a lot of bias are people who have a diagnosis of anxiety…because their presentation in the moment has an anxiety component, they, I think, are at higher risk, especially in an emergency department for being sort of written off.” Implicit bias: “I hear a lot of providers in the emergency room just like—and I can understand frustration like, ‘Well, here's this person again looking for detox.’ And yes, you can be frustrated but we all have to find a way.” Motivation Optimism, consequences. intentions Can improve care Desire to learn and apply TIC “I think we can and should try to improve the care we give to patients that live with violence and that have long history of traumatic events.” “Some days I leave feeling the care I gave was great, other days I don't feel so good. I think I don't always know how to interact with patients that have trauma‐like a crash or fall or some accident I am good.” Abbreviations: TDF, Theoretical Domains Framework; TIC, trauma‐informed care. John Wiley & Sons, Ltd. 3.1.6 Awareness of TIC principles Within the knowledge TDF domain, awareness of TIC principles emerged as a key theme. Participants reported familiarity with TIC principles and awareness of TIC strategies. 3.1.7 Facilitators of staff emotional and psychological safety Within the TDF domain environmental context and resources, facilitators of emotional and psychological safety emerged as a key theme. Participants described several facilitators to managing traumatic and other psychologically and emotionally challenging situations in the ED. For example, participants reported they debrief difficult cases with the care team, which promotes an opportunity to process together. Some also reported that team camaraderie and supportive leadership facilitated a safer and healthier work environment. Several participants were unable to name facilitators of emotional safety in their workplace. 3.1.8 Barriers to staff emotional and psychological safety Barriers to staff emotional and psychological safety emerged as a theme under the TDF domain of environmental context and resources. Key subthemes included the desire for more support sessions and debriefs to improve employee physical, psychological, and emotional safety. Other barriers mentioned that did not reach saturation included poor sleep patterns due to shift work and lack of staff diversity (race and ethnicity). 3.1.9 Opportunities to improve trauma patient experiences Within the TDF domain environmental context and resources, participants identified staff training, increased time with patients, and efforts to decrease discrimination toward patients as key opportunities to enhance experiences for patients with histories of trauma. Most shared that although they provided excellent care to patients with acute physical traumas, they could do better with patients who have experienced past traumas including patients who are often stigmatized, that is, obesity, substance use disorder, mental health, and non‐English speaking patients. 3.1.10 Experiences of bias Experiences of bias were categorized under the TDF domain of social influences. Participants reported witnessing or experiencing both implicit and explicit bias in multiple scenarios, including staff‐to‐staff, patient‐toward‐staff, and staff‐toward‐patient interactions. Staff reported witnessing episodes of homophobia, transphobia, sexism, and bias against people with mental health or substance use disorders. They discussed both the impact of these experiences on patient care, and their desire to deliver universally equitable patient‐centered care. Some participants also reported experiencing microaggressions themselves in the course of their work. 3.1.11 Motivation to learn Within the TDF domain of motivation participants shared a desire to do more for patients experiencing emotional distress, including patients with many complex health care needs. They reported “good” days that positively affected their well‐being and “bad” days that negatively affected their physical and mental well‐being. There was much curiosity if TIC could not only improve patient care but could improve the well‐being of staff. 4 LIMITATIONS Limitations include the small sample size and the single institution analysis that limits generalizability. The study relied on participants’ self‐report, which is subject to recall and social desirability. Different results might be found in different EDs, thus additional studies are needed. The study population was gathered using purposive sampling, and thus responses may have been skewed to support TIC. The initial protocol specified one‐to‐one in‐person interviews; however, in the context of COVID‐19, we pivoted to an audio‐recorded virtual platform. Because interviews were conducted during the COVID‐19 pandemic, our results may have influenced responses. 5 DISCUSSION TIC promotes a culture of safety, empowerment, resilience, and healing. TIC offers an opportunity to improve patient care and staff wellness by applying the 4Rs: realizing the prevalence of trauma, recognizing the impact of trauma, responding both individually and on an organizational level to trauma, and resisting re‐traumatization. 1 Trauma‐informed approaches are recommended by ED professional societies and in clinical guidelines, yet there has been limited operationalization of TIC interventions in ED settings. 16 , 17 , 18 , 19 , 20 This study has several salient findings. First, interdisciplinary ED staff frequently care for patients who have experienced trauma, witness human suffering and loss, and notice the impact of bias on patients and staff. Second, ED staff overwhelmingly felt capable and motivated to provide TIC in the ED and described a passionate commitment to providing equitable and unbiased care. They reported high acceptability of a novel EHR‐integrated TIC plan. Third, ED staff often reported that they lack the specific protocols, training, time, and support to implement TIC interventions and yet they endorsed the importance of trauma‐informed approaches. Participants described the personal toll of secondary trauma, which occurs through exposure to the traumatic experiences of other people, whether by listening to their stories or witnessing their suffering. Secondary trauma is especially common in “caring” professionals, occurs in up to 30% of physicians, and has been associated with multiple negative physical and psychological impacts, including depression, anxiety, post‐traumatic stress disorder, and suicide. 13 , 14 , 28 , 29 , 30 Secondary trauma also can contribute to staff attrition and turnover. 28 , 31 , 32 , 33 The majority of interprofessional staff in our study reported witnessing instances of bias during their work in the ED. Participants recognized that race, stigma, and bias exist in the ED and endorsed TIC approaches that may impact care provided to patients who experience bias. 34 In addition to affecting patient care, these experiences can contribute to secondary trauma of staff. 35 TIC can enhance clinicians’ sense of self‐efficacy and resilience by equipping clinicians with the skills, resources, and support to sensitively care for those who have experienced trauma. 1 Thus, TIC offers a strategy to reduce burnout in a high‐intensity work environment such as the ED, in which both direct and secondary trauma are common. 33 Notably, this study found high staff acceptability of an EHR‐integrated TIC plan that is tailored to the specific needs and preferences of an individual patient. To our knowledge, this is the first EHR‐integrated TIC care plan that has been reported in the literature. The TIC care plan provides an actionable strategy for health care organizations to better care for patients who experience trauma. Developed jointly by a patient and their health care team, the EHR TIC care plan enhances patient‐centeredness by improving interdisciplinary team communication and reducing re‐traumatization of patients (eg, by preventing the need for a patient to repeatedly describe traumatic events or by alerting health care providers to specific behaviors or physical examination maneuvers that might trigger a trauma response for the patient). The TIC care plan is currently being implemented in our hospital system. Based on our experience, in systems already using EHRs, the TIC care plan is likely to be acceptable and relatively easy to implement. More research is needed to evaluate the effectiveness of EHR‐based TIC interventions in improving patient experiences. Last, individual‐level, team‐level, and systems‐level factors have been found to influence whether TIC policies and procedures are implemented routinely and effectively. 35 , 36 Our study found that individual‐level factors such as knowledge and beliefs largely supported the implementation of TIC in the ED. Participants felt that trauma and TIC were relevant to their work setting, believed that TIC was beneficial, and felt able to positively influence the care of patients who experienced trauma. Participants identified positive team‐based factors such as camaraderie and the opportunity for frequent debriefs of difficult cases as important facilitators of TIC implementation. However, participants also identified systems‐level challenges, such as lack of training, time, and private spaces, as well as some aspects of organizational culture, as chief barriers to implementation. Our study is concordant with previous nonemergency medicine literature that has found that resource barriers and organizational factors strongly influence staff implementation of TIC interventions. 33 , 35 This finding suggests that individuals, teams, and systems need to align to effectively provide TIC, and additional organizational support and resources are needed to enhance TIC implementation in the ED. Our study results are not generalizable because a single study site was used. In summary, we identified key factors that influence interprofessional staff implementation of TIC in the ED. These findings reveal potential individual and organizational targets for behavior change interventions to improve TIC and patient and staff experiences in the ED. Future research should evaluate the impact of the trauma‐informed interventions on patient and staff outcomes in the ED setting. AUTHOR CONTRIBUTIONS AL‐OC, ER, NL‐C, SG, and HS participated in the study design, analysis, and preparation of the manuscript. DN participated in quantitative and qualitative analysis and contributed to the manuscript. RO assisted with manuscript preparation. SB assisted with the theoretical framework and methods. AL‐OC and ER assumed overall responsibility of the manuscript. All authors attest to meeting the 4 ICMJE.org authorship criteria: (1) substantial contributions to the conception or design of the work; or the acquisition, analysis, or interpretation of data for the work; and (2) drafting the work or revising it critically for important intellectual content; (3) final approval of the version to be published; and (4) agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. CONFLICT OF INTEREST STATEMENT The authors declare no conflicts of interest. Supporting information Supporting Information, Additional supporting information may be found online in the Supporting Information section at the end of the article. Click here for additional data file. ACKNOWLEDGMENTS Annie Lewis‐O'Connor, Eve Rittenberg, Nomi Levy‐Carrick, Samara Grossman, and Hanni Stoklosa are alumnae of Clinical Scholars, supported by the Robert Wood Johnson Foundation. The project was supported by the program. The views expressed here do not necessarily reflect the views of the Foundation. We appreciate the support of Mary Lynn Cala, Jeannie V. Lee, and Mara Hampson. The project was supported by the Robert Wood Johnson Foundation. Annie Lewis‐O'Connor, PhD, NP‐BC, is a dually board certified (pediatrics and women's health) nurse practitioner. She is the founder and director of the C.A.R.E. (Coordinated Approach to Resiliency & Empowerment) Clinic at Brigham and Women's Hospital, Boston, MA; an associate scientist in the division of women's health/department of medicine at Brigham Women's Hospital; and an instructor at Harvard Medical School in Boston, MA. ==== Refs REFERENCES 1 SAMHSA's Concept of Trauma and Guidance for a Trauma‐Informed Approach. In: Administration SAaMHS, editor. Substance Abuse and Mental Health Services Administration; 2014. 2 Moffitt TE . Childhood exposure to violence and lifelong health: clinical intervention science and stress‐biology research join forces. Dev Psychopathol. 2013;25 (4 ):1619‐1634. doi: 10.1017/s0954579413000801. Pt 2. Epub 2013/12/18. PubMed PMID: 24342859; PubMed Central PMCID: PMCPMC3869039.24342859 3 Felitti VJ . Adverse childhood experiences and adult health. Acad Pediatr. 2009;9 (3 ):131‐132. doi: 10.1016/j.acap.2009.03.001. Epub 2009/05/20. PubMed PMID: 19450768.19450768 4 Felitti VJ , Anda RF , Nordenberg D , et al. Relationship of childhood abuse and household dysfunction to many of the leading causes of death in adults. The Adverse Childhood Experiences (ACE) Study. Am J Prev Med. 1998;14 (4 ):245‐258. doi: 10.1016/s0749-3797(98)00017-8. Epub 1998/06/23. PubMed PMID: 9635069.9635069 5 Sacks V , Murphey D , The prevalence of adverse childhood experiences, nationally, by state, and by race/ethnicity. 2018. 6 Roberts AL , Gilman SE , Breslau J , Breslau N , Koenen KC . Race/ethnic differences in exposure to traumatic events, development of post‐traumatic stress disorder, and treatment‐seeking for post‐traumatic stress disorder in the United States. Psychol Med. 2011;41 (1 ):71‐83. doi: 10.1017/s0033291710000401. Epub 2010/03/30. PubMed PMID: 20346193; PubMed Central PMCID: PMCPMC3097040.20346193 7 Crouch JL , Hanson RF , Saunders BE , Kilpatrick DG , Resnick HS . Income, race/ethnicity, and exposure to violence in youth: results from the national survey of adolescents. J Community Psychol. 2000;28 :625‐641. doi: 10.1002/1520-6629(200011)28:6<625::AID‐JCOP6>3.0.CO;2‐R 8 Slopen N , Shonkoff JP , Albert MA , et al. Racial Disparities in Child Adversity in the U.S.: interactions With Family Immigration History and Income. Am J Prev Med. 2016;50 (1 ):47‐56. doi: 10.1016/j.amepre.2015.06.013. Epub 2015/09/08. PubMed PMID: 26342634.26342634 9 Mersky JP , Janczewski CE . Racial and ethnic differences in the prevalence of adverse childhood experiences: findings from a low‐income sample of U.S. women. Child Abuse Negl. 2018;76 :480‐487. doi: 10.1016/j.chiabu.2017.12.012. Epub 2017/12/23. PubMed PMID: 29272816.29272816 10 Gee GC , Ford CL . STRUCTURAL RACISM AND HEALTH INEQUITIES: old Issues, New Directions. Du Bois review : social science research on race. 2011;8 (1 ):115‐132. doi: 10.1017/s1742058x11000130. Epub 2011/04/01. PubMed PMID: 25632292; PubMed Central PMCID: PMCPMC4306458.25632292 11 Fischer KR , Bakes KM , Corbin TJ , et al. Trauma‐Informed Care for Violently Injured Patients in the Emergency Department. Ann Emerg Med. 2019;73 (2 ):193‐202. doi: 10.1016/j.annemergmed.2018.10.018. Epub 2018/12/07. PubMed PMID: 30503381.30503381 12 Molloy L , Fields L , Trostian B , Kinghorn G . Trauma‐informed care for people presenting to the emergency department with mental health issues. Emerg nurse: the journal of the RCN Accident and Emergency Nursing Association. 2020;28 (2 ):30‐35. doi: 10.7748/en.2020.e1990. Epub 2020/02/26. PubMed PMID: 32096370. 13 Stehman CR , Testo Z , Gershaw RS , Kellogg AR , Burnout . Drop Out, Suicide: physician Loss in Emergency Medicine, Part I. West J Emerg Med. 2019;20 (3 ):485‐494. doi: 10.5811/westjem.2019.4.40970. Epub 2019/05/28. PubMed PMID: 31123550; PubMed Central PMCID: PMCPMC6526882.31123550 14 Scott SD , Hirschinger LE , Cox KR , et al. Caring for our own: deploying a systemwide second victim rapid response team. Jt Comm J Qual Patient Saf. 2010;36 (5 ):233‐240. doi: 10.1016/s1553-7250(10)36038-7. Epub 2010/05/20. PubMed PMID: 20480757.20480757 15 Querin LB , Beck Dallaghan GL , Shenvi C . A Qualitative Study of Resident Physician and Health Care Worker Experiences of Verbal and Physical Abuse in the Emergency Department. Ann Emerg Med. 2022;79 (4 ):391‐396. doi: 10.1016/j.annemergmed.2021.04.019 34172300 16 Brown T , Ashworth H , Bass M , et al. Trauma‐informed Care Interventions in Emergency Medicine: a Systematic Review. West J Emerg Med. 2022;23 (3 ):334‐344. doi: 10.5811/westjem.2022.1.53674. Epub 2022/06/10. PubMed PMID: 35679503; PubMed Central PMCID: PMCPMC9183774.35679503 17 Caring for Patients Who Have Experienced Trauma: aCOG Committee Opinion, Number 825. Obstet Gynecol. 2021;137 (4 ):e94‐e9. Epub 2021/03/25. doi: 10.1097/aog.0000000000004326 PubMed PMID: 3375983033759830 18 Gerace A , Muir‐Cochrane E . Perceptions of nurses working with psychiatric consumers regarding the elimination of seclusion and restraint in psychiatric inpatient settings and emergency departments: an Australian survey. Int J Ment Health Nurs. 2019;28 (1 ):209‐225. doi: 10.1111/inm.12522. Epub 2018/07/19. PubMed PMID: 30019798; PubMed Central PMCID: PMCPMC7818138.30019798 19 Hoysted C , Babl FE , Kassam‐Adams N , et al. Perspectives of hospital emergency department staff on trauma‐informed care for injured children: an Australian and New Zealand analysis. J Paediatr Child Health. 2017;53 (9 ):862‐869. doi: 10.1111/jpc.13644. Epub 2017/08/07. PubMed PMID: 28782226.28782226 20 Bruce MM , Kassam‐Adams N , Rogers M , et al. Trauma Providers' Knowledge, Views, and Practice of Trauma‐Informed Care. J Trauma Nurs: the official journal of the Society of Trauma Nurses. 2018;25 (2 ):131‐138. doi: 10.1097/jtn.0000000000000356. Epub 2018/03/10. PubMed PMID: 29521782; PubMed Central PMCID: PMCPMC5968451. 21 Atkins L , Francis J , Islam R , et al. A guide to using the Theoretical Domains Framework of behaviour change to investigate implementation problems. Implementation science : IS. 2017;12 (1 ):77. doi: 10.1186/s13012-017-0605-9. Epub 2017/06/24. PubMed PMID: 28637486; PubMed Central PMCID: PMCPMC5480145.28637486 22 Tong A , Sainsbury P , Craig J . Consolidated criteria for reporting qualitative research (COREQ): a 32‐item checklist for interviews and focus groups. International journal for quality in health care : journal of the International Society for Quality in Health Care. 2007;19 (6 ):349‐357. doi: 10.1093/intqhc/mzm042. Epub 2007/09/18. PubMed PMID: 17872937.17872937 23 Guest G , Namey E , Chen M . A simple method to assess and report thematic saturation in qualitative research. PLoS One. 2020;15 (5 ):e0232076. doi: 10.1371/journal.pone.0232076. Epub 2020/05/06. PubMed PMID: 32369511; PubMed Central PMCID: PMCPMC7200005.32369511 24 Warm Handoff: Intervention Agency for Healthcare Research and Quality; 2017 [updated Content last reviewed December 2017; cited 2022 2022]. Accessed September 9, 2022. Available from: https://www.ahrq.gov/patient‐safety/reports/engage/interventions/warmhandoff.html 25 Lewis‐O'Connor A , Warren A , Lee JV , et al. The state of the science on trauma inquiry. Women's health (London, England). 2019;15 :1745506519861234. doi: 10.1177/1745506519861234. Epub 2019/08/29. PubMed PMID: 31456510; PubMed Central PMCID: PMCPMC6713964. 26 Braun V , Clarke V . Using thematic analysis in psychology. Qual Res Psychol. 2006;3 (2 ):77‐101. doi: 10.1191/1478088706qp063oa 27 Lincoln YS , Guba EG . Naturalistic inquiry. Newbury Park Cal:Sage. 1985. 28 Hamdan M , Hamra AA . Burnout among workers in emergency Departments in Palestinian hospitals: prevalence and associated factors. BMC Health Serv Res. 2017;17 (1 ):407. doi: 10.1186/s12913-017-2356-3. Epub 2017/06/18. PubMed PMID: 28619081; PubMed Central PMCID: PMCPMC5472878.28619081 29 Ma H , Huang SQ , We B , Zhong Y . Compassion fatigue, burnout, compassion satisfaction and depression among emergency department physicians and nurses: a cross‐sectional study. BMJ open. 2022;12 (4 ):e055941. doi: 10.1136/bmjopen-2021-055941. Epub 2022/04/30. PubMed PMID: 35487521; PubMed Central PMCID: PMCPMC9052053. 30 Chen Y , Shen X , Feng J , et al. Prevalence and predictors of depression among emergency physicians: a national cross‐sectional study. BMC psychiatry. 2022;22 (1 ):69. doi: 10.1186/s12888-022-03687-8. Epub 2022/01/30. PubMed PMID: 35090424; PubMed Central PMCID: PMCPMC8795725.35090424 31 Zhang Y , Feng X . The relationship between job satisfaction, burnout, and turnover intention among physicians from urban state‐owned medical institutions in Hubei, China: a cross‐sectional study. BMC Health Serv Res. 2011;11 :235. doi: 10.1186/1472-6963-11-235. Epub 2011/09/29. PubMed PMID: 21943042; PubMed Central PMCID: PMCPMC3197494.21943042 32 Estryn‐Behar M , Doppia MA , Guetarni K , et al. Emergency physicians accumulate more stress factors than other physicians‐results from the French SESMAT study. Emergency medicine journal : EMJ. 2011;28 (5 ):397‐410. doi: 10.1136/emj.2009.082594. Epub 2010/12/03. PubMed PMID: 21123828.21123828 33 Damian AJ , Gallo J , Leaf P , Mendelson T . Organizational and provider level factors in implementation of trauma‐informed care after a city‐wide training: an explanatory mixed methods assessment. BMC Health Serv Res. 2017;17 (1 ):750. doi: 10.1186/s12913-017-2695-0. Epub 2017/11/22. PubMed PMID: 29157241; PubMed Central PMCID: PMCPMC5696731.29157241 34 Agboola IK , Coupet E Jr , Wong AH . “The Coats That We Can Take Off and the Ones We Can't”: the Role of Trauma‐Informed Care on Race and Bias During Agitation in the Emergency Department. Ann Emerg Med. 2021;77 (5 ):493‐498. doi: 10.1016/j.annemergmed.2020.11.021 PMID: 33579587; PMCID: PMC808505433579587 35 Aarons GA , Hurlburt M , Horwitz SM . Advancing a conceptual model of evidence‐based practice implementation in public service sectors. Adm Policy Ment Health. 2011;38 (1 ):4‐23. doi: 10.1007/s10488-010-0327-7. Epub 2011/01/05. PubMed PMID: 21197565; PubMed Central PMCID: PMCPMC3025110.21197565 36 Greenhalgh T , Robert G , Macfarlane F , Bate P , Kyriakidou O . Diffusion of innovations in service organizations: systematic review and recommendations. TheMilbank quarterly. 2004;82 (4 ):581‐629. doi: 10.1111/j.0887-378X.2004.00325.x. Epub 2004/12/15. PubMed PMID: 15595944; PubMed Central PMCID: PMCPMC2690184.
PMC010xxxxxx/PMC10352597.txt
==== Front Ultrason Sonochem Ultrason Sonochem Ultrasonics Sonochemistry 1350-4177 1873-2828 Elsevier S1350-4177(23)00088-3 10.1016/j.ultsonch.2023.106376 106376 Preface Sonoprocessing of materials (Special issue) Eskin Dmitry Komarov Sergey Tzanakis Iakovos 17 3 2023 6 2023 17 3 2023 96 106376© 2023 Published by Elsevier B.V. 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). ==== Body pmcUltrasound-assisted technologies are experiencing a surge in research and development. This is clearly seen in the example of sonochemistry that has been originally developed to exploit cavitation-related phenomena aiming at enhancing chemical reactions mainly due to formation of highly active radicals like hydroxyl (HO•) and hydroperoxyl (HOO•) in aqueous media. Over the years, this field has undergone significant growth and development, and now includes a wide range of areas such as preparation of emulsions, synthesis and dispersion of nanoparticles and nanospheres, treatment of contaminated surfaces, etc. The expansion and significance of the field can be well illustrated by the impact factor growth of the flagship journal Ultrasonics Sonochemistry that has increased from around 0.95 in 2000 to 9.34 in 2021. The annual citations to this journal have increased from 125 in 2000 to 12,400 in 2021. There is, however, another group of ultrasound-assisted processes, which have received considerable attention recently, that deals with physical effects of ultrasound rather than ultrasonically stimulated chemical reactions. Examples include, but are not limited to, ultrasonic degassing and solidification of molten metals, synthesis of composite materials, metal joining, agglomeration and de-agglomeration of particulates, fragmentation of crystals, atomization, exfoliation of nanomaterials, additive manufacturing, high-precision machining and welding. It is to be noted that many of these processes have been proposed and examined for a very long time. Indeed, as early as the 1920 s, R. Wood and A. Loomis have conducted the first experiments on the ultrasonic atomization of liquids, the emulsification of immiscible liquids, and the structural changes in crystallized organic substances. Then, several years later, in 1935, S. J. Sokolov has conducted a pioneering experimental research on the crystallization of low-melting metals under conditions of ultrasonic irradiation. Since that time, a great number of studies, both fundamental and applied, have been performed to shed light on the ultrasound-related effects and the underlying mechanisms. For example, when ultrasound waves propagate through molten metals, the acoustic cavitation and streaming play a primary role in achieving the desirable ultrasonic effects, similar to those of aqueous media. However, in applications where ultrasound energy is delivered into gas or solid phases, it is not cavitation but quite different physical phenomena that are responsible for the ultrasonic effects. Another promising area where ultrasound waves can be beneficial is controlling interface phenomena. It is well known that when waves are incident upon an interface, the reflecting or scattering of the waves from the interface is responsible for a number of nonlinear phenomena that occur, affecting surface energy. These provide a unique tool for controlling the rates of the interfacial heat and mass transfer. In line is the cavitation-assisted production of 2D nanomaterials that have attracted a great deal of attention from the scientific community with efforts focused on harnessing dynamic interaction of cavitation with 2Ds and optimizing their production at a scale. New technical means of studying the physical phenomena upon ultrasonic processing, such as direct acoustic pressure measurements in a range of temperatures, particle-image velocimetry and ultra-high speed imaging in optical and X-rays spectra, revealed intricate mechanisms of interaction between cavitation bubbles, shockwaves, and acoustic streaming with solid and liquid phases. These mechanisms helped to inform advanced physics-based numerical models that now cover a range of spatial and temporal scales. Currently, the above mentioned and other ultrasonics related topics tend to be dispersed through a range of different journals and, hence, appear unconnected even though they all contain ultrasonic processing as their core subject. It is worth noting that papers on some of such topics do appear occasionally in regular issues of Ultrasonics Sonochemistry but, not as contributions to the main scope of the journal, as they go beyond the chemistry (core subject) related phenomena. This motivated us to propose and launch this Special Issue aiming at attracting papers on a wider range of topics related to Sonoprocessing of Materials with a view of becoming an established trend in the Journal and promote links and collaborations between scientists with complementary interests. This issue attracted a lot of interest and a healthy number of accepted papers on a wide range of topics. We hope that these topics will continue to be covered by Ultrasonic Sonochemistry and, in some not so distant future, will be incorporated in the scope.
PMC010xxxxxx/PMC10352598.txt
==== Front Ultrason Sonochem Ultrason Sonochem Ultrasonics Sonochemistry 1350-4177 1873-2828 Elsevier S1350-4177(23)00143-8 10.1016/j.ultsonch.2023.106431 106431 Ultrasound in Colloids & Polymer Dual effects of ultrasound on fabrication of anodic aluminum oxide Wu Zhicheng [email protected] a⁎ Zhao Yuxiao a1 Fan Jiasheng a1 Gao Chao b Yuan Xieyu a Wang Guoli b Zhang Qiaogen a a State Key Laboratory of Electric Insulation and Power Equipment, Xi’an Jiaotong University, Xi’an 710049, China b Electric Power Research Institute, China Southern Power Grid Co. Ltd, Guangzhou 510623, China ⁎ Corresponding author. [email protected] 1 Y. Zhao and J. Fan contributed equally to this work. 05 5 2023 6 2023 05 5 2023 96 1064317 12 2022 18 4 2023 1 5 2023 © 2023 The Author(s) 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Highlights • Decoupling of the effects of bubble desorption and mass transfer enhancement. • Nanopore-expansion and promoted AAO growth efficiency through bubble desorption and ion migration. • Inhibition of mass transfer and decrease in efficiency caused by jet cavitation-induced nanopore contraction. Ultrasound has been proven to enhance the mass transfer process and impact the fabrication of anodic aluminum oxide (AAO). However, the different effects of ultrasound propagating in different media make the specific target and process of ultrasound in AAO remain unclear, and the effects of ultrasound on AAO reported in previous studies are contradictory. These uncertainties have greatly limited the application of ultrasonic-assisted anodization (UAA) in practice. In this study, the bubble desorption and mass transfer enhancement effects were decoupled based on an anodizing system with focused ultrasound, such that the dual effects of ultrasound on different targets were distinguished. The results showed that ultrasound has the dual effects on AAO fabrication. Specifically, ultrasound focused on the anode has a nanopore-expansion effect on AAO, leading to a 12.24 % improvement in fabrication efficiency. This was attributed to the promotion of interfacial ion migration through ultrasonic-induced high-frequency vibrational bubble desorption. However, AAO nanopores were observed to shrink when ultrasound was focused on the electrolyte, accompanied by a 25.85 % reduction in fabrication efficiency. The effects of ultrasound on mass transfer through jet cavitation appeared to be the reason for this phenomenon. This study resolved the paradoxical phenomena of UAA in previous studies and is expected to guide AAO application in electrochemistry and surface treatments. Keywords Ultrasound Anodic aluminum oxide Dual effects Ultrasonic vibration Jet cavitation ==== Body pmc1 Introduction Anodic aluminum oxide (AAO), with low fabrication cost, tunable pore size, and unique electrochemical properties, has been made as a preferred choice for various applications such as optoelectronic components [1], surface modification [2], and marine antifouling [3]. Nevertheless, the AAO fabrication process still faces significant challenges like poor surface uniformity and slow fabrication efficiency despite extensive research in the 20th century. It is urgent to find new technological strategies for improving the AAO fabrication process. The AAO fabrication process belongs to the multiphase electrochemical processes, including solid (electrode), liquid (electrolyte), and gas (bubble) phases, and any affected phase can influence the anodizing process [4]. Based on this, ultrasound has been used to improve the anodizing process thanks to its modulating effect on multiphase fluids [5]. The advantages of ultrasound in the AAO fabrication have been detected, such as in cleaning oxide film [6], increasing anodization current density [7], and promoting film growth [8]. Recently, negative effects of ultrasound on AAO have also been found, in which ultrasound can reduce the anodization current and fabrication efficiency [9]. These paradoxical findings have been inferred to be attributable to the various ways in which ultrasound impacts the solid, liquid, and gas phases. Several hypotheses have been postulated to explain the mechanisms by which ultrasound affects electrochemical processes, including mass transfer [10], ultrasonic cavitation [11], microjet [12], and shock waves [13]. However, most experiments of ultrasonic-assisted anodization (UAA) have involved an “ultrasonic bath” as the ultrasonic application method, which cannot distinguish the ultrasound target. Clarifying the promotion or inhibition mechanisms in the UAA process is very important and thus the topic of this study. Two ultrasonic-focusing modes were introduced here to AAO fabrication, as ultrasound focused on the anode or electrolyte, which achieved focusing ultrasound on the different phases (solid, liquid, or gas phase) in anodization separately. Benefiting from separate ultrasound targets, this study achieved a reliable analysis of the different roles played by ultrasound in AAO. Here, based on an anodizing system with focused ultrasound, the effects of UAA on AAO and its mechanisms were investigated, with AAO fabrication in phosphoric acid as an example. The UAA system construction with two ultrasonic-focusing modes was introduced and the sound and flow fields of the two modes were calculated in Section 2. The influence of UAA on AAO morphology and current density was investigated with different ultrasonic-focusing modes in Section 3. The dual effects of ultrasound in AAO fabrication were analyzed in Section 4. This study clarified the reason for the dual effects of ultrasound and contributed guidance for applying ultrasound in electrochemistry and surface treatments. 2 Experimental setups 2.1 Anodizing system with focused ultrasound A UAA system which can focus ultrasound on different targets was constructed (Fig. 1). A 35 kHz sinusoidal voltage was generated by an arbitrary waveform generator and amplified by a power amplifier with a maximum output power of 500 W. The conversion of electricity to ultrasound relied on an ultrasound transducer with a rated power of 500 W. Ultrasound was introduced into the anodizing process through a horn and two different ultrasonic-focusing modes were proposed. The horn was connected to the aluminum sheet (Mode I), or the electrolyte (Mode II).Fig. 1 Anodizing system with focused ultrasound. To avoid the effect of size ratio on the results, aluminum (Al, purity 99.999 %) and platinum sheets of the same size (Pt, 15 m × 15 m × 0.2 m) were selected as anode and cathode, respectively. A water-cooling system was applied to ensure a constant temperature during anodizing and a 10- wt% phosphoric acid solution was selected as the electrolyte. 2.2 Sound fields and flow fields of two modes To understand the effects of ultrasound on different targets, a two-dimensional numerical model combining sound and flow fields was established to explore differences between the two modes. The computational domains included liquid, electrode, and boundaries. All boundaries were non-slip walls except for the horn boundary. A normal displacement with a harmonic oscillation was added to the horn boundary. The interfaces of liquid-electrode and electrode-air satisfied the impedance boundary condition, expressed as(1) n·(-1ρc∇p)=iωpZe where Ze=ρece is the impedance of the outer region,ρe and ce the air density and sound speed in air, respectively, n the unit vector perpendicular to the wall, p the sound pressure, ω the angular frequency, and ρc the liquid complex density [14]. The distributions of sound pressure and liquid flow velocity showed that, ultrasound on the anode (Fig. 2a) or electrolyte (Fig. 2c) presented an elliptical radial shape. When the horn was connected to the anode, the electrolyte did not flow significantly (Fig. 2b). When the horn was placed in the electrolyte, ultrasonic radiation force was applied to the fluid at the horn bottom, resulting in changing liquid flow velocity and jet cavitation (Fig. 2d). In short, the distributions of the flow field were significantly different in the two different ultrasonic-focusing modes.Fig. 2 Distributions of sound pressure and liquid flow velocity in different modes. Sound pressure in Mode I (a), liquid flow velocity in Mode I (b), sound pressure in Mode II (c), and liquid flow velocity in Mode II (d). 2.3 Experiment process and characterization The Al sheet was cleaned in a γ-butyrolactone solution with a concentration of 3.2- wt% at 50 °C for 5 min and rinsed with deionized water. Then, the sheet was fixed with a clamp and UAA applied in Mode I or Mode II at 25 °C under a 40 V constant voltage for 40 min. AAO samples were bent to an angle of 90° to expose cross-sections [15]. AAO morphologies and synthetic structure were assessed with field emission scanning electron microscope (FESEM; Gemini 500, Carl Zeiss AG, Oberkochen, Germany), which includes the energy dispersive X-ray spectrometer (EDS; Ultimax 100, Oxford, Britain). The AAO surface and cross-section were observed by tilting the specimen mount to an angle of 45° between the surface and microscope detector. The working distance between sample and probe was 5–8 mm and the accelerating voltage 5 kV. The nanopore size of AAO was analyzed by the open-source software ImageJ [16], and the detailed procedure is provided in the Supplementary Information. The anodization current was recorded every 2 s using an ammeter (FLUK298C, Fluke Corporation, Everett, USA). The ammeter is connected to the anodized current loop in a positive-in and negative-out way. Current density is defined as the magnitude of current per unit cm2. The surface element composition was analyzed by EDS (Fig. S4), which was determined that the substance formed on the surface of the Al sheet was AAO. X-ray diffraction (XRD), x-ray photoelectron spectra (XPS), and linear sweep voltammetry (LSV) were used to determine the crystal form, surface chemical bonds, and electrochemical properties of the film of AAO, respectively. The results are shown in Supplementary Information. 3 Results 3.1 Effects of ultrasound on AAO morphology The surface morphologies of AAO in No-ultrasound, Mode I, and Mode II showed that there were hundreds of round-like nanopores ∼ 100 nm diameter per square μm of AAO (Fig. 3).Fig. 3 AAO surface morphologies in different modes. No-ultrasound, Mode I, and Mode II (a–c, respectively). The statistical results of nanopore sizes in different modes were depicted in half-violin diagrams (Fig. 4). In each half-violin, a left histogram expresses the nanopore size distribution and a kernel smooth function on the right reflected the histogram trend. The significant difference in nanopore size between either No-ultrasound and Mode I (p < 0.0001) or No-ultrasound and Mode II (p < 0.0001) was shown through the Mann-Whitney U test (significance level α = 0.05) [17]. Considering their means (No-ultrasound, 64.7095 nm; Mode I, 76.1355 nm; and Mode II, 62.1023 nm), the nanopore size of Mode I was found to be significantly larger than that of No-ultrasound, while nanopore size of Mode II was significantly smaller than that of No-ultrasound.Fig. 4 AAO nanopore sizes in different modes. These results indicated that ultrasound focused on different targets had different effects on nanopore size: a nanopore-expansion effect reflected in Mode I, while Mode II showed a nanopore-contraction effect. The nanopore-expansion effect and nanopore-contraction effect are due to that ultrasound changed the dissolution rate of Al2O3 near the nucleation point and the inner wall of the pore, which in turn changed the sizes of nanopore size in different modes. 5 repeated anodization experiments were conducted in different modes, and the statistical results of the average equivalent nanopore size (Table S2) verified the nanopore-expansion and nanopore-contraction effects of ultrasound on AAO. 3.2 Effects of ultrasound on AAO growth rate The current density is an important factor affecting the AAO growth rate. To understand the effects of ultrasound on AAO growth rate, the time dependence of current density in No-ultrasound, Mode I, and Mode II were recorded (Fig. 5). During anodizing, the current density decreased exponentially at first, then increased slightly, and stabilized after 1500 s. Due to the anodic oxidation current development was not stable during the growth phase of the nanopore, the current densities in different modes were averaged from 1500 to 2400 s, as the stable current density Javg. Compared with Javg of No-ultrasound, the current density of Mode I increased by 13.66 %, while that of Mode II decreased by 19.65 %.Fig. 5 Time dependence of current density in different modes. The AAO growth rate is characterized as the oxide film growth rate v, which was calculated by measuring oxide film thicknesses (Fig. 6). Compared with the v of No-ultrasound (73.5 nm/min), the v of Mode I (82.5 nm/min) was increased by 12.24 %, while the v of Mode II (54.5 nm/min) was reduced by 25.85 %.Fig. 6 Thickness and growth rate of AAO in different modes with 40 min duration. No ultrasound, Mode I, and Mode II (a–c, respectively). The results in Section 3 showed the dual effects of ultrasound on AAO fabrication, which implied that ultrasound focused on different targets had different mechanism effects during anodizing. 4 Discussion Changes in ultrasonic-focusing modes caused different anodization results, in nanopore size and AAO growth rate. These indicated that ultrasound focused on different targets had different UAA mechanisms. Thus, the mechanisms of ultrasound on the anode and electrolyte were discussed below. 4.1 Ultrasound focused on the anode According to unchanged flow field distribution (Fig. 2b) and vibrating anode, ultrasonic vibration was dominant in anodization when the ultrasound was focused on the anode. To clarify the anodizing process with ultrasonic vibration, a two-stage anodization was designed, with the anodization carried out first in No-ultrasound for 526 s and subsequently in Mode I for 474 s. The current density was recorded in the same manner as in Section 2.3 and the anode surface photographed in each stage. Ultrasonic vibration had a significant effect on the current density and bubbles on the anode surface (Fig. 7a). In the first stage, current density stabilized at 1.5 × 10−2 A/cm2 and there were many bubbles attached to the anode surface. In the second stage, the current density increased, stabilized at 2.0 × 10−2 A/cm2, and surface bubbles disappeared.Fig. 7 Current density and images of anode surface (a) and schematic diagram of interface ion migration (b). The above phenomena were analyzed by the interfacial ion migration process (Fig. 7b). In the first stage, O2 oxidized by O2– stayed at the electrolyte/alumina interface [18], which inhibited Al3+ diffusion and AAO growth. In the second stage, by applying ultrasound to the anode surface, the anode high-frequency vibration promoted interfacial bubble desorption. Subsequently, bubbles were no longer attached to the anode surface. Therefore, Al3+ near the electrolyte/alumina interface moved freely, being free from the influence of bubbles which increased the anodization current density and improved AAO fabrication efficiency. The chemical reaction near the nucleation point and the inner wall of the nanopore channel is carried out at a faster rate because the ions in this area can move more freely in the nanopore growth stage. The development degree of nanopore size is more mature than that without ultrasound when the stable pore structure is stabilized, and ultrasound has a nanopore-expansion effect on AAO. 4.2 Ultrasound focused on the electrolyte The ion movement process in the electrolyte was a type of mass transfer [19], [20]. Due to the electrolyte being a bridge that conducted charged particles, the mass transfer in anodization was affected when ultrasound was focused on the electrolyte. To discuss the mechanisms of ultrasound on the electrolyte, it was necessary to study mass transfer in the electrolyte. The ion mass transfer in anodization can be described by the Nernst-Planck mass transfer equation [21], expressed as(2) J(x)=-D∂C(x)∂x-zFRTDC∂ϕ(x)∂x+Cν(x) where J(x) is the diffusion flux at distance × from the electrode surface, D the electrolyte diffusion coefficient, which characterizes substance mobilities, ∂C(x)/∂x the concentration gradient of ions, ∂φ(x)/∂x the potential gradient, v(x) the electrolyte convection velocity along the direction of ∂φ(x)/∂x, F the Faraday constant, z the valence state of matter (dimensionless), R the universal gas constant, and T the temperature. The three terms on the right side of Eq. (2) corresponded to diffusion, migration, and convection, respectively. According to Eq. (2) and Fig. 8, the effects of ultrasound on mass transfer were analyzed. When ultrasound was focused on the electrolyte, jet cavitation caused electrolyte agitation, which reduced ∂C(x)/∂x in ion diffusion. Ion migration was not affected because of the constant electric field. Jet cavitation makes ions gain a velocity perpendicular to v(x), which does not change v(x). Therefore, ion convection was not affected by jet cavitation. Combined with the above analysis, jet cavitation only inhibited ion diffusion, which reduces the J(x) and current density. Meanwhile, the chemical reaction rate near the nucleation point and the inner wall of the nanopore channel slows down due to the increasing concentration of the ion that cannot be well diffused. The development degree of nanopore size is more immature than that without ultrasound, and ultrasound has a nanopore-contraction effect on AAO.Fig. 8 Effects of ultrasound on mass transfer. In short, the UAA dual effects were attributed to different mechanisms of ultrasound on the anode and electrolyte. When ultrasound was focused on the anode, UAA showed a promotion effect on AAO fabrication efficiency. In this case, ultrasonic vibration promoted bubble desorption and ion migration. However, UAA also showed the inhibition effect when the ultrasound was focused on the electrolyte, which was attributed to the effect of ultrasonic cavitation, which reduced the concentration gradient of charged particles and inhibited mass transfer, resulting in decreased current density. Although the UAA mechanisms have been discussed in this study, the dual effects of UAA are not only caused by ultrasonic vibration and cavitation. Other factors, such as changes in the number of interfacial active sites [9] or cavitation bubble collapse [22] might affect AAO fabrication. Unlike the above studies, the focus here was on distinguishing the effects of ultrasound on targets in different phases. In addition, other factors were not discussed here because bubble desorption and jet cavitation were much more pronounced. 5 Conclusion Based on a UAA system which realized ultrasound focused on different targets, the dual effects of ultrasound in AAO fabrication were examined. The results were summarized as follows.(1) Ultrasound on the anode had a nanopore-expansion effect, while ultrasound on the electrolyte had a nanopore-contraction effect. (2) The current density was increased and AAO fabrication efficiency increased by 12.24 % when ultrasound was applied on the anode, while the current density was reduced and AAO fabrication efficiency decreased by 25.85 % when ultrasound was applied on the electrolyte. (3) The promotion effect of ultrasound on AAO fabrication was attributed to bubble desorption caused by anode vibration. Meanwhile, ultrasound inhibited AAO fabrication because the ultrasound cavitation weakened ion diffusion. To achieve efficient AAO fabrication, it would be necessary to focus as much ultrasonic energy as possible on the anode, while avoiding the generation of jet cavitation. This study provided new guidance for applying ultrasound to electrochemistry and surface treatment. CRediT authorship contribution statement Zhicheng Wu: Conceptualization, Formal analysis, Writing – review & editing, Project administration, Funding acquisition. Yuxiao Zhao: Methodology, Investigation, Writing – original draft, Visualization. Jiasheng Fan: Software, Investigation, Writing – original draft. Chao Gao: Validation, Data curation, Writing – review & editing, Project administration. Xieyu Yuan: Methodology. Guoli Wang: Validation, Resources. Qiaogen Zhang: Resources, Supervision. Declaration of Competing Interest The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Zhicheng Wu reports financial support was provided by National Natural Science Foundation of China. Appendix A Supplementary data The following are the Supplementary data to this article:Supplementary data 1 Data availability No data was used for the research described in the article. Acknowledgements This work was financially supported by National Natural Science Foundation of China (No. 52207173). Zhicheng Wu reports financial support was provided by National Natural Science Foundation of China. Appendix A Supplementary data to this article can be found online at https://doi.org/10.1016/j.ultsonch.2023.106431. ==== Refs References 1 Richard G. Akkermans R.P. Coles B.A. Marken F. Ultrasound in photoelectrochemistry: a new approach to the enhancement of the efficiency of semiconductor electrode processes1 Ultrason. Sonochem. 4 1997 223 228 10.1016/S1350-4177(96)00041-7 11232778 2 Kaseem M. Ko Y.G. A novel composite system composed of zirconia and LDHs film grown on plasma electrolysis coating: Toward a stable smart coating Ultrason. Sonochem. 49 2018 316 324 10.1016/j.ultsonch.2018.08.023 30150026 3 Olkowicz K. Buczko Z. Nasiłowska B. Kowalczyk K. Czwartos J. Superhydrophobic Coating Based on Porous Aluminum Oxide Modified by Polydimethylsiloxane (PDMS) Mater. 15 2022 1042 10.3390/ma15031042 4 Wu Y. Liu K. Su B. Jiang L. Superhydrophobicity-Mediated Electrochemical Reaction Along the Solid–Liquid–Gas Triphase Interface: Edge-Growth of Gold Architectures Adv. Mater. 26 2014 1124 1128 10.1002/adma.201304062 24243745 5 Qin L. Porfyrakis K. Tzanakis I. Grobert N. Eskin D.G. Fezzaa K. Mi J. Multiscale interactions of liquid, bubbles and solid phases in ultrasonic fields revealed by multiphysics modelling and ultrafast X-ray imaging Ultrason. Sonochem. 89 2022 106158 10.1016/j.ultsonch.2022.106158 6 Zhang R. Jiang K. Zhu Y. Qi H. Ding G. Ultrasound-assisted anodization of aluminum in oxalic acid Appl. Surf. Sci. 258 2011 586 589 10.1016/j.apsusc.2011.08.041 7 Zarei M. Nourouzi S. Jamaati R. Cano I.G. Dosta S. Sarret M. Formation of highly uniform tin oxide nanochannels by electrochemical anodization on cold sprayed tin coatings Surf. Coat. Technol. 410 2021 126978 10.1016/j.surfcoat.2021.126978 8 Paulo V. Neves-Araujoa J. Padrón-Hernándeza E. Fast and Room-temperature Synthesis of Porous Alumina Films in Ultrasonic Assisted Bath Inducing Superficial Cavitations Port. Electrochim. Acta 37 2019 123 129 10.4152/pea.201902123 9 Gao Z. Cao J. Muzammal H.M. Wang C. Sun H. Dong C. Ma H. Wang Y. Ultrasound assisted large scale fabrication of superhydrophilic anodized SnOx films with highly uniformed nanoporous arrays Mater. Chem. Phys. 242 2020 122540 10 Yang Y. Feng Y. Li K. Ajmal S. Cheng H. Gong K. Zhang L. Ultrasound-boosted selectivity of CO in CO2 electrochemical reduction Ultrason. Sonochem. 76 2021 105623 10.1016/j.ultsonch.2021.105623 11 Hou G. Ren Y. Zhang X. Dong F. An Y. Zhao X. Zhou H. Chen J. Cavitation erosion mechanisms in Co-based coatings exposed to seawater Ultrason. Sonochem. 60 2020 104799 10.1016/j.ultsonch.2019.104799 12 Low L.E. Tey B.T. Ong B.H. Tang S.Y. A facile and rapid sonochemical synthesis of monodispersed Fe3O4@cellulose nanocrystal nanocomposites without inert gas protection Asia Pac. J. Chem. Eng. 13 4 2018 e2209 13 Sundaresan P. Yamuna A. Chen S. Sonochemical synthesis of samarium tungstate nanoparticles for the electrochemical detection of nilutamide Ultrason. Sonochem. 67 2020 105146 10.1016/j.ultsonch.2020.105146 14 Wu J. Li Z. Density-functional theory for complex fluids Annu. Rev. Phys. Chem. 58 2007 85 112 10.1146/annurev.physchem.58.032806.104650 17052165 15 Wu Z. Fan J. Yuan X. Zhao Y. Wang B. Zhang Q. Li L. Key role of longitudinal disarranged growth in porous anodic alumina: Crystallographic orientation Mater. Lett. 330 2023 133294 10.1016/j.matlet.2022.133294 inpress 16 Trujillo C. Piedrahita-Quintero P. Garcia-Sucerquia J. Digital lensless holographic microscopy: numerical simulation and reconstruction with ImageJ Appl. Opt. 59 19 2020 5788 5795 10.1364/AO.395672 32609706 17 P.E. McKnight, J. Najab, Mann‐Whitney U Test, The Corsini encyclopedia psychology, (2010) 1-1. https://doi.org/10.1002/9780470479216.corpsy0524. 18 Crossland A. Habazaki H. Shimizu K. Skeldon P. Thompson G. Wood G. Zhou X. Smith C. Residual flaws due to formation of oxygen bubbles in anodic alumina Corro. Sci. 41 1999 1945 1954 10.1016/S0010-938X(99)00035-9 19 Kashid M.N. Renken A. Kiwi-Minsker L. Gas–liquid and liquid–liquid mass transfer in microstructured reactors Chem. Eng. Sci. 66 2011 3876 3897 10.1016/j.ces.2011.05.015 20 Aikens D. Electrochemical methods, fundamentals and applications ACS Publ. 60 1983 A25 10.1021/ed060pA25.1 21 Newman J. Thomas-Alyea K.E. Electrochemical systems 2012 John Wiley & Sons NJ 22 Pollet B. Power ultrasound in electrochemistry: from versatile laboratory tool to engineering solution 2012 John Wiley & Sons NJ
PMC010xxxxxx/PMC10352601.txt
==== Front Blood Blood Blood 0006-4971 1528-0020 The American Society of Hematology S0006-4971(23)01191-6 10.1182/blood.2023020181 Thrombosis and Hemostasis Structure of coagulation factor VIII bound to a patient-derived anti–C1 domain antibody inhibitor Childers Kenneth C. 1 Avery Nathan G. 1 Estrada Alamo Kevin A. 1 Davulcu Omar 23 Haynes Rose Marie 23 Lollar Pete 4 Doering Christopher B. 45 Coxon Carmen H. 6 Spiegel P. Clint Jr. [email protected] 1∗ 1 Chemistry Department, Western Washington University, Bellingham, WA 2 Pacific Northwest Center for Cryo-EM, Oregon Health & Science University, Portland, OR 3 Pacific Northwest National Laboratory, Environmental Molecular Sciences Laboratory, Richland, WA 4 Department of Pediatrics, Aflac Cancer and Blood Disorders Center, Children’s Healthcare of Atlanta, Emory University, Atlanta, GA 5 Expression Therapeutics Inc, Tucker, GA 6 National Institute for Biological Standards and Control, Potters Bar, Hertfordshire, United Kingdom ∗ Correspondence: P. Clint Spiegel Jr, Chemistry Department, Western Washington University, 516 High St, MS 9150 Bellingham, WA; [email protected] 18 5 2023 13 7 2023 18 5 2023 142 2 197201 21 2 2023 25 4 2023 © 2023 by The American Society of Hematology. Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0), permitting only noncommercial, nonderivative use with attribution. All other rights reserved. 2023 The American Society of Hematology https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Key Points • The structure of FVIII bound to an anti–C1 domain antibody inhibitor reveals a novel epitope. • Antibody binding blocks multiple lysine and arginine residues implicated in FVIII endocytosis by dendritic cells. Visual Abstract Abstract The development of pathogenic antibody inhibitors against coagulation factor VIII (FVIII) occurs in ∼30% of patients with congenital hemophilia A receiving FVIII replacement therapy, as well as in all cases of acquired hemophilia A. KM33 is an anti–C1 domain antibody inhibitor previously isolated from a patient with severe hemophilia A. In addition to potently blocking FVIII binding to von Willebrand factor and phospholipid surfaces, KM33 disrupts FVIII binding to lipoprotein receptor-related protein 1 (LRP1), which drives FVIII hepatic clearance and antigen presentation in dendritic cells. Here, we report on the structure of FVIII bound to NB33, a recombinant derivative of KM33, via single-particle cryo-electron microscopy. Structural analysis revealed that the NB33 epitope localizes to the FVIII residues R2090-S2094 and I2158-R2159, which constitute membrane-binding loops in the C1 domain. Further analysis revealed that multiple FVIII lysine and arginine residues, previously shown to mediate binding to LRP1, dock onto an acidic cleft at the NB33 variable domain interface, thus blocking a putative LRP1 binding site. Together, these results demonstrate a novel mechanism of FVIII inhibition by a patient-derived antibody inhibitor and provide structural evidence for engineering FVIII with reduced LRP1–mediated clearance. Using single-particle cryo-electron microscopy, Childers and colleagues report on the first structure of factor VIII (FVIII) bound to a recombinant derivative of an anti-C1 domain antibody inhibitor previously isolated from a patient with severe hemophilia A. The inhibitor targets positively charged amino acids on the C1 domain to block FVIII binding to von Willebrand factor, phospholipid membranes, and lipoprotein receptor–related protein 1, which drives FVIII hepatic clearance and antigen presentation in dendritic cells. These data help explain the in vivo consequences of this inhibitor and may assist future engineering of FVIII with a longer half-life due to reduced clearance. ==== Body pmcIntroduction Hemophilia A is an X-linked recessive bleeding disorder afflicting 1 in 4065 male births worldwide and is characterized by defective or deficient coagulation factor VIII (FVIII), leading to uncontrolled bleeding events.1 The formation of pathogenic antibody inhibitors against FVIII occurs in 30% of patients with congenital hemophilia A receiving FVIII replacement therapy as well as in all cases of acquired hemophilia A.2 Most of the characterized antibody inhibitors target the A2, C1, and C2 domains of FVIII and disrupt coagulation mechanistically. Anti–C1 domain inhibitor antibodies can block FVIII binding to von Willebrand factor (VWF) and phospholipid surfaces, resulting in premature proteolytic degradation, clearance of FVIII, and/or impeding access to membrane surfaces, in which activated FVIII serves to nucleate the intrinsic tenase complex. KM33, a group AB anti–C1 domain antibody inhibitor was previously isolated from a patient with severe hemophilia A who presented with multiple inhibitor antibodies.3 In addition to potently inhibiting FVIII binding to VWF and phospholipid membranes, KM33 has the novel property of blocking the FVIII endocytosis by dendritic cells, which regulates the hepatic clearance of FVIII and antigen presentation.4,5 Studies incorporating site-directed mutagenesis with surface plasmon resonance (SPR) and live cell microscopy as well as hydrogen-deuterium exchange mass spectrometry (HDX-MS) have localized the KM33 epitope to several membrane-binding spikes in the C1 domain.5, 6, 7, 8, 9, 10, 11 To identify the amino acids that make up the KM33 epitope, this study reports on the structure of ET3i, a bioengineered human/porcine FVIII chimera, bound to a Fab fragment of NB33, a recombinant immunoglobulin G derivative of KM33, by single-particle cryo-electron microscopy (cryo-EM). Study design ET3i and NB33 were expressed, purified, and analyzed as previously described (see supplemental Methods, available on the Blood website).12, 13, 14 Cryo-EM sample preparation, data collection, image processing, and structure determination and validation are detailed in the supplemental Methods. Results and discussion Characterization of ET3i inhibition by NB33 Neutralization of ET3i by NB33 was assessed using a factor X activation chromogenic assay and Bethesda protocol. We determined an IC50 value of 3.64 nM in the chromogenic assay (supplemental Figure 1A) and a specific inhibitory activity of 7451 Bethesda units per mL NB33 in the Bethesda assay (supplemental Figure 1B), consistent the reports from previous studies.5,14 These results demonstrate that the inhibition of ET3i by NB33 is analogous to the inhibition of human FVIII by KM33. Structure of ET3i:NB33 complex The structure of ET3i bound to NB33 was determined by single-particle cryo-EM at a nominal resolution of 4.23 Å (Figure 1A-B). The initial 2-dimensional classification showed intact particles with unambiguous densities in the NB33 Fab constant (heavy and light) and variable (heavy [VH] and light [VL]) domains (supplemental Figure 2A). The final sharpened map, excluding the flexible constant domains, displayed unequivocal densities for the A1-A2/A3-C1-C2 domains of ET3i and the variable domains of NB33. The ET3i A domains vary in local resolution, whereas the C2 domain has suboptimal resolution owing to its flexibility but sufficient density for complete model building. No large-scale conformational rearrangements of the C domains were observed as previously described.15,16Figure 1. Cryo-EM structure of the ET3i:NB33 complex. Cryo-EM map (A) and structure (B) of ET3i bound to NB33 Fab fragment. (dark blue, porcine A1 domain; slate, human A2 domain; cyan, porcine A3 domain; orange, human C1 domain; yellow, human C2 domain; dark purple, NB33 heavy chain; and light pink, NB33 light chain). (C-E) Intermolecular contacts between ET3i residues (C-D) R2090-F2093 and (E) I2158-R2159 (orange) and NB33 heavy (dark purple) and light (light pink) chains. The dashed lines represent noncovalent interactions ≤5 Å. The ET3i:NB33 interface, which exhibits a local resolution of ∼3.5 to 4 Å (supplemental Figure 2C), is stabilized by a combination of electrostatic and hydrophobic interactions with a buried surface area of 822 Å2. The NB33 epitope centers on membrane-binding spikes in the C1 domain, composed FVIII residues R2090-S2094 and I2158-R2159.17 Residue K2092 anchors the C1 domain to NB33 by binding to an acidic cleft at the center of the paratope, forming multiple salt bridges and hydrogen bonds with the NB33 VH domain (Figure 1C). NB33 residue Y32 in the VH domain forms a hydrogen bond with FVIII residue R2090, in addition to hydrophobic interactions with residue F2093 (Figure 1D). These results are consistent with prior SPR and enzyme-linked immunosorbent assay data, demonstrating that the FVIII R2090A/K2092A/F2093A triple variant abrogated binding to KM33.8 The cryo-EM structure of BIVV001, a bioengineered extended half-life therapeutic FVIII fused to the VWF D′D3 domains, helped identify interactions between FVIII residues K2092 and F2093 and the D3 domain, supporting the obstruction of VWF binding as an inhibitory mechanism by KM33.5,18 Additional contacts were present between FVIII residues I2158-R2159 and the CDR3 loop of the NB33 VH domain (Figure 1E). Although previous HDX-MS data demonstrated interactions between KM33 and FVIII residues from 2077 to 2085, the ET3i:NB33 structure reveals no interactions with this region.5 Most of the intermolecular contacts are with the NB33 VH domain, consistent with the isolation of KM33 through genetic recombination from the VH gene segment of a patient with hemophilia A and combined with a noninhibitory immunoglobulin G4 VL.3 Recent pull-down assays illustrated that the KM33 VH domain, but not the VL domain, retained its affinity toward FVIII.19 However, several interactions are observed in the ET3i:NB33 cryo-EM structure at lower map contour levels between FVIII residues Y2043-Q2045 and the NB33 VL domain, as previously suggested by HDX-MS experiments.5 Together, the structure that we report is in agreement with previous studies demonstrating that the KM33 epitope overlaps with amino acids in the C1 domain, previously shown to bind VWF and/or phospholipid membranes.5,10,11 A complete list of intermolecular contacts is provided in supplemental Table 2. NB33 and LRP1 bind to the FVIII lysine and arginine residues in the C1 domain The ET3i:NB33 structure revealed a patch of positively charged residues in the FVIII C1 domain docked onto an acidic cleft formed by the VH/VL domain interface of NB33 (Figure 2A), providing the first structural insight into how KM33 blocks the uptake of FVIII by dendritic cells through low-density lipoprotein receptor-related protein 1 (LRP1)-mediated endocytosis. LRP1 uses clusters of extracellular complement-type repeat domains to bind and endocytose a wide variety of ligands, including FVIII, for processing and antigen presentation to T cells.20 Ligand binding to LRP1 occurs through a conserved acidic patch and an aromatic residue in the complement-type repeat domains, which target ligands carrying surface-exposed positively charged residues.21,22 The binding of LRP1 to FVIII is predicted to rely on an array of surface-exposed lysine residues on the FVIII light chain in a charge-dependent manner.10 Further studies have suggested a partial overlap between the LRP1 binding region and KM33 epitope.5,7,8,23 Our structural analysis of ET3i:NB33 was consistent with these experiments, revealing that residues K2065, R2090, K2092, and R2159 docked onto a patch of acidic NB33 residues, thus blocking the putative LRP1 binding region (Figure 2B-C). These positively charged FVIII residues have previously been shown to bind to VWF and/or phospholipid membranes, supporting the inhibitory mechanism of KM33 disruption of VWF and phospholipid interactions (supplemental Table 2).6,11,17,18 Intriguingly, NB33 appears to mimic the predicted LRP1 binding site using an aliphatic paratope to target positively charged FVIII residues and neighboring hydrophobic residues. These results provide structural evidence for the mechanism of FVIII clearance by LRP1 and suggest that mutating positively charged FVIII residues in the C1 domain can reduce hepatic clearance rates.Figure 2. Electrostatic interactions between the C1 domain and NB33. (A) Electrostatic surface potential (blue, positive charge; red, negative charge) of the ET3i C1 domain (top) and NB33 Fab fragment (bottom) at ±10 kcal/(mol·e–). (B) FVIII residue K2065/R2090/K2092/R2159 (sticks) docked onto an acidic patch formed by the NB33 VH domain (surface). NB33 is colored based on the electrostatic surface potential, as in panel A. (C) The structure of the C1 domain bound to NB33. Charged residues in the epitope (C1 domain, orange) and paratope (NB33 VL, light pink; NB33 VH, dark purple) are depicted as spheres. (D) The proposed pipeline for the development of FVIII replacement therapeutics with reduced LRP1-mediated clearance and immunogenicity by targeting surface-exposed arginine and lysine residues. The structure of the C1 domain bound to NB33 bears resemblance to the crystal structure of the isolated C2 domain bound to BO2C11, a patient-derived anti–C2 domain antibody inhibitor that forms multiple salt bridges with FVIII residues R2215 and R2220 (supplemental Figure 4).24 Both antibodies target several β-hairpin loops and disrupt FVIII binding to VWF and phospholipid membranes, although the BO2C11 epitope was significantly more aliphatic than the electropositive KM33 epitope. In addition, the FVIII C1 domain forms extensive interactions with the VWF D′D3 domains in contrast to the C2 domain.18 Uptake of FVIII by monocyte-derived dendritic cells was blocked in the presence of BO2C11 or KM33 antibodies.25 Furthermore, the exposure of FVIII–/– mice to recombinant FVIII preincubated with BO2C11 or KM33 diminished the host immune response.7,25 Similar results were observed using a recombinant FVIII R2090A/K2092A/F2093A triple variant in the absence of antibodies, presumably because of the impaired binding to LRP1; however, the FVIII R2215A/R2220A double variant on the C2 domain showed no measurable effect on diminishing the immune response.8,25 Indeed, previous SPR studies have demonstrated that the C2 domain has no affinity for LRP1, suggesting that the BO2C11 antibody may indirectly block LRP1 binding through steric interference or overlap with the binding region for a non-LRP1 endocytic receptor.11 Although inhibitor development is a rare occurrence in cases of mild/moderate hemophilia A, our structural analysis may also indicate that certain missense mutations in the C1 domain may promote FVIII binding to LRP1 and induce an immune response.26 Together, these results suggest differential roles for positively charged residues in the C domains in driving LRP1-mediated FVIII endocytosis and immunogenicity. In summary, the ET3i:NB33 structure represents the first structural analysis of a complex of a therapeutically active FVIII construct bound to a patient-derived pathogenic inhibitory antibody. Our structural analysis delineates the role of surface-exposed, positively charged residues on the C1 domain in binding to KM33, which disrupts the endocytosis of FVIII by dendritic cells. Fundamental questions concerning FVIII clearance require further investigation, including the role of non-LRP1 endocytic receptors, such as macrophage mannose receptors, which are unaffected by KM33-bound FVIII.7 These findings are critical for therapeutic strategies for designing a recombinant FVIII molecule with an extended half-life, reduced immunogenicity, and decreased clearance by dendritic cells (Figure 2D). Conflict-of-interest disclosure: P.L. is listed as an inventor on a patent application describing ET3i and on patents owned by Emory University claiming compositions of matter that include modified FVIII proteins with reduced reactivity with anti-FVIII antibodies. C.B.D. and P.L. are cofounders of Expression Therapeutics and own equity in the company. Expression Therapeutics owns the intellectual property associated with ET3i. The terms of this arrangement have been reviewed and approved by Emory University in accordance with its conflict-of-interest policies. The remaining authors declare no competing financial interests. Supplementary Material Supplemental Methods, Tables, Figures, and References Acknowledgments This work was supported by the 10.13039/100000019 National Hemophilia Foundation Judith Graham Pool Postdoctoral Research Fellowship (K.C.C.) and the National Institutes of Health (NIH), 10.13039/100000050 National Heart, Lung, and Blood Institute (award numbers R15HL135658 and U54HL141981 [P.C.S.] and award numbers R44HL117511, R44HL110448, U54HL112309, and U54HL141981 [C.B.D., P.L.]). A portion of this research was supported by the NIH National Institute of General Medical Sciences grant U24GM129547, performed at the 10.13039/100008043 Pacific Northwest Center for Cryo-EM at 10.13039/100006668 Oregon Health & Science University , and accessed through the 10.13039/100019109 Environmental Molecular Sciences Laboratory (grid.436923.9), a 10.13039/100006132 DOE Office of Science User Facility sponsored by the 10.13039/100006206 Office of Biological and Environmental Research . Authorship Contribution: K.C.C. planned and performed the experiments, analyzed the data, and assisted with writing the manuscript; N.G.A. performed the experiments and analyzed the data; K.A.E.A. assisted with data processing and interpretation; O.D. and R.M.H. performed the experiments and assisted with data processing and interpretation; C.B.D. and P.L. developed the expression and purification procedures for ET3i and assisted with writing the manuscript; C.H.C. developed the expression and purification procedures for NB33 and assisted with writing the manuscript; and P.C.S. planned the experiments, analyzed the data, and wrote the manuscript. Atomic coordinates for the factor VIII/NB33 structure are deposited in the Protein Data Bank (accession number 8G6I), and the cryo-electron microscopy maps are deposited in the Electron Microscopy Data Base (accession number EMD-29770). Data are available on request from the corresponding author, P. Clint Spiegel Jr ([email protected]). The online version of this article contains a data supplement. The publication costs of this article were defrayed in part by page charge payment. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734. ==== Refs References 1 Iorio A Stonebraker JS Chambost H Establishing the prevalence and prevalence at birth of hemophilia in males a meta-analytic approach using national registries Ann Intern Med 171 8 2019 540 546 31499529 2 Lai JD Lillicrap D Factor VIII inhibitors: advances in basic and translational science Int J Lab Hematol 39 suppl 1 2017 6 13 28447409 3 van den Brink EN Turenhout EAM Bovenschen N Multiple VH genes are used to assemble human antibodies directed toward the A3-C1 domains of factor VIII Blood 97 4 2001 966 972 11159524 4 Meems H Van Den Biggelaar M Rondaij M van der Zwaan C Mertens K Meijer AB C1 domain residues Lys 2092 and Phe 2093 are of major importance for the endocytic uptake of coagulation factor VIII Int J Biochem Cell Biol 43 8 2011 1114 1121 21497669 5 Batsuli G Deng W Healey JF High-affinity, noninhibitory pathogenic C1 domain antibodies are present in patients with hemophilia A and inhibitors Blood 128 16 2016 2055 2067 27381905 6 Meems H Meijer AB Cullinan DB Mertens K Gilbert GE Factor VIII C1 domain residues Lys 2092 and Phe 2093 contribute to membrane binding and cofactor activity Blood 114 18 2009 3938 3946 19687511 7 Herczenik E Van Haren SD Wroblewska A Uptake of blood coagulation factor VIII by dendritic cells is mediated via its C1 domain J Allergy Clin Immunol 129 2 2012 501-9, 509.e1-5 8 Wroblewska A Van Haren SD Herczenik E Modification of an exposed loop in the C1 domain reduces immune responses to factor VIII in hemophilia A mice Blood 119 22 2012 5294 5300 22498747 9 Batsuli G Ito J Mercer R Anti-C1 domain antibodies that accelerate factor VIII clearance contribute to antibody pathogenicity in a murine hemophilia A model J Thromb Haemostasis 16 9 2018 1779 1788 29981270 10 van den Biggelaar M Madsen JJ Faber JH Factor VIII interacts with the endocytic receptor low-density lipoprotein receptor-related Protein 1 via an extended surface comprising “hot-spot” lysine residues J Biol Chem 290 27 2015 16463 16476 25903134 11 Bloem E Van Den Biggelaar M Wroblewska A Factor VIII C1 domain spikes 2092-2093 and 2158-2159 comprise regions that modulate cofactor function and cellular uptake J Biol Chem 288 41 2013 29670 29679 24009077 12 Doering CB Healey JF Parker ET Barrow RT Lollar P Identification of porcine coagulation factor VIIi domains responsible for high level expression via enhanced secretion J Biol Chem 279 8 2004 6546 6552 14660593 13 Spencer HT Denning G Gautney RE Lentiviral vector platform for production of bioengineered recombinant coagulation factor VIII Mol Ther 19 2 2011 302 309 21081907 14 Coxon CH Yu X Beavis J Characterisation and application of recombinant FVIII-neutralising antibodies from haemophilia A inhibitor patients Br J Haematol 193 5 2021 976 987 33973229 15 Gish JS Jarvis L Childers KC Structure of blood coagulation factor VIII in complex with an anti–C1 domain pathogenic antibody inhibitor Blood 137 21 2021 2981 2986 33529335 16 Stoilova-Mcphie S Lynch GC Ludtke S Pettitt BM Domain organization of membrane-bound factor VIII Biopolymers 99 7 2013 448 459 23616213 17 Lü J Pipe SW Miao H Jacquemin M Gilbert GE A membrane-interactive surface on the factor VIII C1 domain cooperates with the C2 domain for cofactor function Blood 117 11 2011 3181 3189 21156843 18 Fuller JR Knockenhauer KE Leksa NC Peters RT Batchelor JD Molecular determinants of the factor VIII/von Willebrand factor complex revealed by BIVV001 cryo-electron microscopy Blood 137 21 2021 2970 2980 33569592 19 Shestopal SA Parunov LA Olivares P Isolated variable domains of an antibody can assemble on blood coagulation factor VIII into a functional Fv-like complex Int J Mol Sci 23 15 2022 8134-15 20 Young PA Migliorini M Strickland DK Evidence that factor VIII forms a bivalent complex with the low density lipoprotein (LDL) receptor-related protein 1 (LRP1): identification of cluster IV on LRP1 as the major binding site J Biol Chem 291 50 2016 26035 26044 27794518 21 Gettins PGW Dolmer K A proximal pair of positive charges provides the dominant ligand-binding contribution to complement-like domains from the LRP (low-density lipoprotein receptor-related protein) Biochem J 443 1 2012 65 73 22181833 22 Fisher C Beglova N Blacklow SC Structure of an LDLR-RAP complex reveals a general mode for ligand recognition by lipoprotein receptors Mol Cell 22 2 2006 277 283 16630895 23 Meems H Meijer AB Mertens K Factor VIII Variants Having a Decreased Cellular Uptake https://patentimages.storage.googleapis.com/fe/c4/42/1bcbe7893cbbae/AU2011303916A1.pdf 2019 24 Spiegel PC Jacquemin M Saint-Remy JM Stoddard BL Pratt KP Structure of a factor VIII C2 domain–immunoglobulin G4k Fab complex: identification of an inhibitory antibody epitope on the surface of factor VIII Blood 98 1 2001 13 19 11418455 25 Gangadharan B Ing M Delignat S The C1 and C2 domains of blood coagulation factor VIII mediate its endocytosis by dendritic cells Haematologica 102 2 2017 271 281 27758819 26 Eckhardt CL Van Velzen AS Peters M Factor VIII gene (F8) mutation and risk of inhibitor development in nonsevere hemophilia A Blood 122 11 2013 1954 1962 23926300
PMC010xxxxxx/PMC10352604.txt
==== Front Ultrason Sonochem Ultrason Sonochem Ultrasonics Sonochemistry 1350-4177 1873-2828 Elsevier S1350-4177(23)00148-7 10.1016/j.ultsonch.2023.106436 106436 Ultrasonic Degradation of Pollutant Ultrasound-assisted alkali removal of proteins from wastewater generated during oil bodies extraction Song Hanyu a Zhong Mingming a Sun Yufan a Yue Qiang [email protected] b⁎ Qi Baokun [email protected] a⁎ a College of Food Science, Northeast Agricultural University, Harbin 150030, China b Heilongjiang Open University, Harbin, Heilongjiang 150030, China ⁎ Corresponding authors. [email protected]@163.com 08 5 2023 6 2023 08 5 2023 96 1064369 12 2022 24 4 2023 4 5 2023 © 2023 The Authors. Published by Elsevier B.V. 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). In this study, an ultrasonic-assisted alkaline method was used to remove proteins from wastewater generated during oil-body extraction, and the effects of different ultrasonic power settings (0, 150, 300, and 450 W) on protein recovery were investigated. The recoveries of the ultrasonically treated samples were higher than those of the samples without ultrasonic treatment, and the protein recoveries increased with increasing power, with a protein recovery of 50.10 % ± 0.19 % when the ultrasonic power was 450 W. Amino acid analysis showed that the amino acids comprising the recovered samples were consistent, regardless of the ultrasonic power used, but significant differences in the contents of amino acids were observed. No significant changes were observed in the protein electrophoretic profile using dodecyl polyacrylamide gel, indicating that sonication did not change the primary structures of the recovered samples. Fourier transform infrared and fluorescence spectroscopy revealed that the molecular structures of the samples changed after sonication, and the fluorescence intensity increased gradually with increasing sonication power. The contents of α-helices and random coils obtained at an ultrasonic power of 450 W decreased to 13.44 % and 14.31 %, respectively, whereas the β-sheet content generally increased. The denaturation temperatures of the proteins were determined using differential scanning calorimetry, and ultrasound treatment reduced the denaturation temperatures of the samples, which was associated with the structural and conformational changes caused by their chemical bonding. The solubility of the recovered protein increased with increasing ultrasound power, and a high solubility was essential in good emulsification. The emulsification of the samples was improved well. In conclusion, ultrasound treatment changed the structure and thus improved the functional properties of the protein. Keywords Wastewater Protein Ultrasound-assisted alkali method Structural property Functional property Abbreviations DI deionized DSC differential scanning calorimetry EAI emulsification activity ESI emulsion stability FC foaming capacity FS foaming stability FT-IR Fourier-transform infrared OAC oil-holding capacity OB oil body PAGE polyacrylamide gel electrophoresis SDS sodium dodecyl sulfate SEM scanning electron microscopy WAC water-holding capacity ==== Body pmc1 Introduction Soybean, which is a critical oilseed crop that provides edible vegetable oil and protein, is used as a functional nutritional component for humans. Soybean seeds store triacylglycerol in the form of spherical droplets denoted oil bodies (OBs). The outer membrane of an OB comprises a monolayer of phospholipids and membrane proteins. Recently, OBs attracted considerable attention. As a natural oil-in-water emulsion [1], the special structure provides a unique stability that renders it outstanding for use in the fields of food, cosmetics, and pharmaceuticals [2]. OB extraction from the slurry obtained by grinding and filtering soybeans requires the addition of large amounts of sucrose and alkali. After centrifugation, the slurry is divided into three layers: the upper OB layer, the middle layer containing soluble protein in a clear liquid, and the lower precipitated solid layer. After collecting the OBs, the middle and lower layers are discarded as waste. However, the middle layer contains numerous carbohydrates, proteins, and inorganic and organic salts, often with high chemical and biological oxygen demands, which, if discharged directly into the environment, may contaminate water bodies, thereby negatively affecting aquatic and terrestrial ecosystems [3]. Therefore, this study was conducted using the discarded intermediate layer, as few studies regarding the crucial proper treatment of this wastewater are reported. The current increase in the global population and shift in individual dietary habits towards protein-rich foods are increasing concerns regarding protein shortages [4]. Therefore, this study extracts protein from wastewater to address the negative environmental impact of excess nutrients in the wastewater and recover protein resources from the generated wastewater. Alkali extraction-isoelectric point and micellar precipitation and salt extraction-dialysis are three common methods of extracting plant proteins [5]. Alkali extraction-isoelectric point precipitation, which is the most commonly used method, is employed to extract purified protein products from various plants in high yields [6]. However, the extraction pH used in alkali extraction-isoelectric point precipitation may affect protein functionality: a higher alkalinity may increase protein denaturation and aggregation [7], which may adversely affect protein solubility and induce the oxidation of polyphenols to quinones, resulting in extracted proteins with darker colors [8]. Due to the structural binding of proteins to other biomolecules, such as polyphenols and cellulose, conventional alkali extraction-isoelectric point precipitation exhibits limitations. To address this challenge, pretreatment methods, including various enzymatic and physical techniques, are used to maximize structural disruption and improve protein extraction [9]. Due to the low installation and maintenance costs of ultrasound, ultrasonic pretreatment in combination with the alkaline method was used in this study to recover proteins from the wastewater generated by OB extraction. Ultrasound-assisted extraction methods are employed to extract proteins rapidly and efficiently from materials, using ultrasound to rupture cells. Ultrasound is classified as low (20–100 kHz), medium (100 kHz–1 MHz), or high frequency (1–10 MHz). The ultrasound-induced nucleation, growth, and collapse of microbubbles are known as acoustic cavitation. At a sufficiently high intensity and power, the negative pressure of the rarefaction cycle may exceed the threshold intermolecular force of attraction of the liquid medium, resulting in the formation of cavitation bubbles. The bubbles grow via the rectified diffusion of dissolved gas and solvent molecules, and over numerous cycles, the bubbles gradually grow until they reach critical sizes. When unable to withstand the internal vapor pressure, these bubbles burst violently, releasing a considerable amount of energy [10]. In liquid media, ultrasound propagates various types of physical forces, such as acoustic flow, cavitation, shear, microjets, and shockwaves. These contribute to the rupturing of cell walls and enhance mass transfer between protein molecules and the solvent, thus improving the efficiency of protein extraction [11]. Ultrasonic power is related to the pressure amplitudes of ultrasound waves and provides the threshold for cavitation. Different ultrasonic powers display different effects on the number and sizes of cavitation bubbles, maximum bubble collapse temperature, and ultrasonic chemical yield. Increasing the ultrasonic power may yield an optimal level of cavitation, beyond which marginal effects or reduced performance are observed. Therefore, determining the optimal power for a given application and experimental condition is a key step in the study [10]. Recently, this method improved the yields of various plant proteins. Chittapalo and Noomhorm [12] reported that sonication combined with the alkali method in extracting rice bran protein resulted in a 1.65-fold higher yield than that obtained via the conventional alkali method. The protein yield of ultrasonically treated watermelon seeds increased by 87 % compared to that obtained using conventional methods [13]. Wang et al. [14] reported that ultrasonication-assisted alkali extraction with a shorter extraction time afforded a higher yield of pea proteins (82.6 %) with enhanced functionalities compared to those of the proteins obtained via conventional alkali extraction. The propagation of ultrasound waves in the extraction medium results in acoustic cavitation, which promotes protein extraction by increasing the mass transfer and internal diffusion of the matrix [15]. This study used a combination of ultrasonic treatment and alkali extraction to remove proteins from the effluent produced during OB extraction. Alkali extraction was performed at pH 9 to investigate the effects of different ultrasound powers (0, 150, 300, and 450 W) on the recovery and basic composition of the protein recovered from wastewater to yield insight into the changes in protein structure and function. The process aids in protecting the environment and ensuring the sustainability of food systems, and the recovered proteins may be used to develop renewable protein resources for use in protein applications. 2 Material and methods 2.1 Materials Soybeans (Dong-Nong 42), which contained 41.2 % crude protein, 23.6 % oil, 11.7 % moisture, and 4.1 % ash, were harvested in Harbin, China [16], and soybean oil was purchased from local supermarkets (Harbin, China). All reagents used were of analytical grade. 2.2 Extraction The washed soybeans were placed in water in a ratio of 1:5 (w/v), and the solution was stored in a refrigerator at 4 °C for 18 h. Deionized (DI) water was added to the soaked soybeans (1:8, w/v), which were then ground using a tissue crusher at maximum speed for 90 s to yield a slurry. The obtained slurry was filtered through four layers of coarse cotton cloth to remove solid residues, and sucrose (20 %) was then added to the filtrate. The pH was adjusted to 9 (NaOH), followed by 1 h of stirring in an ice-water bath to dissolve the proteins. The filtrate was then centrifuged at 15 000 × g for 30 min (L535R-1, Changsha Xiangyi Centrifuge Instrument, Changsha, China) and transferred to partition funnels to separate the semisolid cream layers from the waste solutions, which were collected. The collected wastewater was treated using an ultrasonic processor (Scientz, Ningbo, China) at 20 kHz for 10 min with an output of 0, 150, 300, or 450 W (0 W: protein samples were recovered using conventional alkali extraction, without ultrasonic treatment, by directly adding HCl to the collected wastewater for precipitation). The wastewater was sonicated by applying a 4 s pulse, with two consecutive cycles separated by a 2 s rest period. After sonication, the pH of the wastewater was adjusted to 4.5 using HCl (1 M), and the wastewater was allowed to stand for 2 h and then centrifuged (15 000 × g, 30 min, 4 °C). The precipitate was collected and redissolved in DI water (1:5, w/v) via stirring for 1 h, and centrifugation was repeated a total of thrice (washing steps). Finally, the precipitates were solubilized in DI water, neutralized with 2 M NaOH, and centrifuged again (15 000 × g, 30 min, 4 °C). The supernatant was lyophilized to afford the resulting protein, which was denoted 0, 150, 300, or 450 W based on the employed output power of 0, 150, 300, or 450 W, respectively. 2.3 Basic composition 2.3.1 Chemical composition The crude protein contents recovered using the ultrasound-assisted alkaline method with different ultrasonic power settings were determined via the Dumas method (N × 6.25). Petroleum ether was used as the extraction solvent, the oil contents were determined using Soxhlet extraction, and the extracted proteins were determined via the method proposed by Jiang et al. [17]. The protein contents and masses after lyophilization of the samples recovered via the ultrasonic-assisted alkali method with different power settings were recorded and denoted Cr and Wr, respectively. The volume and protein content of the wastewater generated by OB extraction were recorded as Ve and Ce, respectively.(1) Recoveryyield%=Cr×WrCe×Ve×100 2.3.2 Amino acid analysis Samples were placed in sealed test tubes containing 6 mol/L HCl, hydrolyzed for 24 h at 110 °C, and then cooled to 20 ± 2 °C, as reported by Zheng et al. [18]. The sample was transferred to a 25 mL volumetric flask and filtered. A 1 mL aliquot of the treated sample solution was dried under vacuum and redissolved in 1 mL HCl, and 30 μL was introduced into the amino acid analyzer (L-8900, Hitachi, Tokyo, Japan). 2.4 Structural analysis 2.4.1 Sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) The samples were used to prepare protein suspensions with concentrations of 2 mg/mL by dissolving them in DI water, adding 2 × protein loading buffer, mixing well, and boiling for 5 min. The gels comprised separating (12 %) and concentrating gels (5 %), and 5 μL of the samples were added to the gels. The potential was set to 80 or 120 mV for the concentrating or separating gel, respectively, and the power was cut after the dye reached the bottom of the gel. The gels were dyed, decolorized for 24 h, and photographed. 2.4.2 Fluorescence spectroscopy Protein suspensions (0.25 mg/mL) were prepared using the samples recovered via the ultrasound-assisted alkali method with different ultrasonic power settings, and the fluorescence spectra were measured at excitation and emission wavelengths and a potential of 280 and 300–400 nm and 700 mV, respectively. 2.4.3 Fourier-transform infrared (FT-IR) spectroscopy The FT-IR spectra of the samples recovered via the ultrasound-assisted alkali method with different ultrasonic power settings were obtained using an FTIR-8400S spectrometer (Shimadzu, Kyoto, Japan) in the spectral range 4000–400 cm−1. 2.4.4 Free sulfhydryl (–SH) groups Tris-glycine buffer was prepared using 0.086 mol/L tris, 0.09 mol/L glycine, and 4 mM ethylenediaminetetraacetic acid at pH 8.0. Subsequently, 4 mg of 5,5′-dithiobis (2-nitrobenzoic acid) was added to 1 mL Tris-glycine buffer. Different samples were dissolved in 4.5 mL Tris-glycine buffer, added to 50 μL Ellman reagent and mixed well. Then, the solution was protected from light, and its absorbance was recorded using a UV spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). The extinction coefficient was 13 600 M−1 cm−1, which was used to calculate the free SH content. 2.4.5 Differential scanning calorimetry (DSC) DSC was conducted using a differential scanning calorimeter (Mettler Toledo, Columbus, OH, USA) to determine the thermal properties. First, 5–15 mg of soy protein was placed in a sealed sterile (Al) container and then scanned over 20–140 °C at a ramping rate of 10 °C/min. 2.4.6 Surface tension Different sample powders were weighed to prepare solutions with concentrations of 1 mg/mL, and the surface tension between the protein solution and air was measured for 300 s at 25 °C using a tensiometer equipped with a charge-coupled-device camera (Theta Lite, Biolin Scientific, Gothenburg, Sweden). 2.5 Microstructural characterization 2.5.1 Particle size and ζ-potential The particle sizes and ζ-potentials of the samples were measured using a Zetasizer Nano ZS90 analyzer (Malvern Panalytical, Malvern, UK). The samples were diluted to 0.001 % with phosphate-buffered saline, and the refractive indices of the dispersed and continuous phases were set to 1.456 and 1.333, respectively [19]. 2.5.2 Scanning electron microscopy (SEM) The lyophilized sample powder was evenly applied to double-sided carbon tape. A scanning electron microscope (S-3400 N, Hitachi) was used at an operating voltage of 30 kV to study the microstructures of the samples, which were coated with layers of Au prior to observation. 2.6 Functional characteristics 2.6.1 Protein solubility Different sample powders were weighed and dissolved in DI water to prepare solutions with concentrations of 1 mg/mL and the pH values were set to 3–9. The solutions were then centrifuged (10 000 × g, 30 min, 4 °C), and the protein contents in the supernatants were determined using the bicinchoninic acid protein assay.(2) Solubility%=proteincontentofsupernatanttotalproteincontent×100% 2.6.2 Surface hydrophobicity (H0) Protein samples recovered at different powers were dissolved in DI water at concentrations of 0.01–0.5 mg/mL. The protein solution (4 mL) was mixed with 20 μL ANS solution, and the fluorescence intensity was measured at 370 (excitation) and 465 nm (emission). The curve was plotted with the protein concentration as the horizontal axis and the measured fluorescence intensity as the vertical axis, and the slope of the curve was H0. 2.6.3 Emulsifying properties 8 mL of protein solution (5 mg/mL) was homogenized with 2 mL of soybean oil at 10 000 rpm for 2 min. After emulsion preparation, samples were immediately removed from the bottom of the emulsion, and their absorbances were measured at 500 nm via dilution (dissolve SDS in DI water,0.1 % (w/v)) [20].(3) EAIemulsionactivityindexm2g=2×2.303×A0×N10,000θLC (4) ESIemulsionstabilityindex(min)=A0A0-A30×T30-T0 where A0 and A30 represent absorbance at 0 and 30 min, respectively. The dilution factor, volume fraction of the oil phase, thickness of the cuvette, and concentration of the protein are represented by N (2 0 0), θ (0.20), L (1 cm), and C (g/mL), respectively. T0 and T30 represent 0 and 30 min. 2.6.4 Foaming properties The foaming capacity (FC) and stability (FSs) of the protein samples recovered using the ultrasound-assisted alkaline method were measured. Protein suspensions of 1 % (w/v) were homogenized at 15 000 rpm for 2 min. V1 is the volume of the foam after homogenization, and V2 is the volume of the foam after 10 min [15].(5) FC%=V110×100 (6) FS%=V2V1×100 2.6.5 Water/oil absorption capacity (WAC/OAC) DI water/oil (10 mL) and proteins (1 g) recovered at different ultrasonic power settings were mixed in centrifuge tubes and centrifuged (8000 × g, 10 min, 4 °C) after standing for 30 min. The precipitate was obtained and placed at a 45° for 20 min and the WAC/OAC is the number of grams of water/oil per gram of sample. 2.6.6 Rheology The proteins recovered at different ultrasonic power settings were used to prepare 100 g/L suspensions, and silicone oil, which was necessary to prevent evaporation, was prepared prior to the study. Samples were heated from 25 to 90 °C at a heating rate of 2 °C/min, maintained at 90 °C for 10 min, and cooled to 25 °C. The strain and frequency were fixed at 1 % and 0.1 Hz, respectively. 2.7 Statistical analysis Each experiment was performed in triplicate under the same conditions, and the experimental results are expressed as the mean ± standard deviation. Multiple comparisons were performed using Duncan's test (P < 0.05), and statistical analyses were conducted using SPSS version 26.0 (IBM, Armonk, NY, USA). 3 Results and discussion 3.1 Basic composition 3.1.1 Chemical composition The protein content of the wastewater generated by OB extraction at pH 9.0 was 1.95 % ± 0.03 %. The protein and oil contents and protein recoveries of the sonicated and untreated samples are shown in Table 1. All protein samples recovered in this study display similar protein contents (approximately 80 %). The oil contents of these samples range from 3.55 % to 3.91 %, which are much lower than those of raw soybeans because most oil in the plant is stored in the OBs as triglycerides. Protein recovery using the conventional alkali extraction method is 42.48 %. The recovery of protein from the ultrasound-treated samples increases markedly (P < 0.05) with ultrasonic power, reaching 50.10 % at 450 W. Notably, the process of protein extraction using ultrasound does not exhibit a single mechanism of action but the cumulative effect of shear, microjets, shockwaves, and the formation of hydroxyl radicals via the collapse of cavitation bubbles. These driving forces contribute to the collapse of the cell walls and enhance mass transfer between the protein molecules and solvent, resulting in increased protein recovery [11]. Therefore, ultrasound-assisted alkaline extraction is a suitable method for realizing high recovery rates of proteins from wastewater.Table 1 Compositions of protein samples recovered using the ultrasonic-assisted alkali method with different power settings and the protein recovery rates. Sample Protein (%) Oil (%) Protein yield (%) 0 W 80.59 ± 0.07a 3.55 ± 0.52a 42.48 ± 0.34d 150 W 78.99 ± 0.10c 3.91 ± 0.59a 45.31 ± 0.14c 300 W 78.49 ± 0.11d 3.88 ± 0.19a 47.75 ± 0.08b 450 W 79.65 ± 0.13b 3.75 ± 0.20a 50.10 ± 0.19a Different letters in the same column indicate significant differences (P < 0.05). 3.1.2 Amino acid analysis Table 2 shows the amino acid compositions and contents of the proteins recovered from wastewater using the ultrasound-assisted alkaline method with different ultrasonic power settings. Glu and Asp are the predominant amino acids, and Met is the limiting amino acid. The amino acids comprising the recovered protein remain identical, irrespective of the ultrasonic power setting, whereas the amino acid contents vary between the samples. The essential amino acid (EAA) contents of the samples increase in the order 26.09 % (0 W) < 26.45 % (150 W) < 26.72 % 300 W) < 27.08 % (450 W). When the ultrasonic power is increased, the content of Pro decreases, whereas those of the other amino acids gradually increase. Amino acids such as Pro and Arg serve as precursors of Asp and numerous enzymes, whereas Glu serves as a substrate for Arg and Pro. Therefore, changes in Pro are associated with Asp [21]. The contents of hydrophilic and -phobic amino acids increase with increasing ultrasonic power. All four samples contain higher contents of hydrophobic amino acids compared to those of hydrophilic amino acids. The compositions of amino acids correlate closely with the functionalities of proteins [22], and thus, the effect of the ultrasound-assisted alkali method on protein functionality was further investigated.Table 2 Amino acid analyses of proteins recovered using the ultrasonic-assisted alkaline method with different power settings. Total amino acid 0 W 150 W 300 W 450 W Asp 8.37 ± 0.02a 9.09 ± 0.13b 9.11 ± 0.08b 9.37 ± 0.16a Thr 2.75 ± 0.04b 2.93 ± 0.06a 2.94 ± 0.01a 2.95 ± 0.12a Ser 3.77 ± 0.06b 4.14 ± 0.09a 4.19 ± 0.07a 4.22 ± 0.07a Glu 14.73 ± 0.04c 17.06 ± 0.05b 17.14 ± 0.15b 17.55 ± 0.11a Gly 3.02 ± 0.07b 3.41 ± 0.14a 3.44 ± 0.03a 3.46 ± 0.13a Ala 2.95 ± 0.01b 3.28 ± 0.08a 3.30 ± 0.10a 3.33 ± 0.05a Cys 0.70 ± 0.08b 0.92 ± 0.11a 0.84 ± 0.04ab 0.91 ± 0.08a Val 3.26 ± 0.02b 3.37 ± 0.02a 3.38 ± 0.09a 3.38 ± 0.06a Met 0.85 ± 0.04b 0.96 ± 0.07ab 1.03 ± 0.14a 1.04 ± 0.09a Trp – – – – Ile 3.30 ± 0.12b 3.44 ± 0.09ab 3.47 ± 0.06a 3.54 ± 0.03a Leu 5.83 ± 0.07b 6.06 ± 0.16ab 6.10 ± 0.08a 6.20 ± 0.19a Tyr 2.59 ± 0.05b 2.79 ± 0.09a 2.84 ± 0.04a 2.88 ± 0.01a Phe 3.88 ± 0.05a 3.59 ± 0.07b 3.63 ± 0.12b 3.67 ± 0.13b Lys 4.39 ± 0.07a 4.17 ± 0.04c 4.22 ± 0.07bc 4.30 ± 0.05ab His 1.82 ± 0.03b 1.93 ± 0.07ab 1.95 ± 0.12ab 2.00 ± 0.08a Arg 5.70 ± 0.08b 5.69 ± 0.08b 5.77 ± 0.06ab 5.87 ± 0.06a Pro EAA Hydrophobic Hydrophilic 3.38 ± 0.05a 26.09 26.47 9.81 3.00 ± 0.15b 26.45 27.12 10.77 2.93 ± 0.06b 26.72 27.28 10.81 2.99 ± 0.16b 27.08 27.61 10.95 Different letters in the same column indicate significant differences (P < 0.05). 3.2 Structural analysis 3.2.1 Sds-page The results of SDS-PAGE of the proteins recovered using the ultrasonic-assisted alkali method with different power settings are shown in Fig. 2. Soy protein contains two major proteins, i.e., 7S and 11S, and their individual subunits are labeled in the figure. As shown in Fig. 2, the ultrasonicated samples exhibit roughly similar bands in the range of approximately 15–95 kDa compared to those of the control, which indicates that the primary structures of the proteins are not altered by sonication. Similar results were reported by Xiong et al. [23], with high-intensity ultrasound used to treat pea protein isolates.Fig. 1 Flow chart of protein recovery from wastewater. Fig. 2 SDS-PAGE profiles of proteins recovered at ultrasonic power of 0, 150, 300, and 450 W. 3.2.2 Fluorescence spectroscopy As shown in Fig. 3A, the fluorescence intensity increases with increasing ultrasonic power, suggesting that the cavitation forces of ultrasonic treatment unfold the protein structure, leading to an increase in the number of chromophores in the solvent [24]. The maximum emission wavelength (λm) of the initial sample is 328 nm, which increases to 331 nm when the ultrasonic power settings are 150 and 300 W and 330 nm at 450 W. Ultrasonic treatment loosens the protein structure, exposing the internal Trp residues and resulting in the overall redshift of λm [25]. This redshift may be explained by the disruption of the hydrophobic interactions during sonication, which exposes the hydrophobic groups to the molecular surface [26].Fig. 3 Fluorescence (A) and Fourier transform infrared spectra (B) of the proteins recovered using the ultrasound-assisted alkali method with different ultrasonic power settings. 3.2.3 FT-IR spectroscopy Fig. 3B shows the FT-IR spectra of the different protein samples. The FT-IR spectra display three main sets of characteristic transmission bands, among which the amide I band at 1600–1700 cm−1 mainly reflects the stretching vibration of C = O [27]. The secondary structures of the protein components may be analyzed based on the second-order derivatives of the amide I regions. Based on the data shown in Table 3, the respective α-helix and β-sheet and -turn contents are 13.44–15.25 %, 34.53–49.53 %, and 14.36–31.08 %. The content of β-sheets, which are the main secondary structures in the protein samples, is the highest. The α-helix contents of the ultrasonically treated samples are lower than that of the untreated sample. The α-helix content of the untreated sample is 15.25 %, and it is only 13.44 % when treated using an ultrasonic power of 450 W. The content of the random coils exhibits the same trend as that of the α-helices, with the smallest content (14.31 %) observed when an ultrasonic power of 450 W is applied. By contrast, the β-turns content increased to 20.73% at 450 W of ultrasonic power. Therefore, the power setting of ultrasonic treatment changes the secondary structure of the protein to varying degrees [28]. Hydrogen bonding is critical in maintaining the stabilities of the α-helices [29]. Therefore, decreased contents of α-helices and random coils indicate cleavage of the intramolecular hydrogen bonds, a decrease in the disordered protein content, and an increase in protein flexibility, yielding a protein with improved functional properties [29]. Although numerous studies report that ultrasonic treatment changes the secondary structure of a protein [30], [31], these changes should differ according to the protein, modification technology used, and protein components analyzed.Table 3 Contents (%) of the secondary structures of the proteins recovered using the ultrasonic-assisted alkali method with different power settings. Sample α-helices β-sheets β-turns Random coils 0 W 15.25 ± 0.02a 49.53 ± 0.06a 14.36 ± 0.03d 15.89 ± 0.05a 150 W 13.59 ± 0.07b 45.76 ± 0.09c 20.40 ± 0.09c 14.55 ± 0.14b 300 W 13.61 ± 0.10b 34.53 ± 0.02d 31.08 ± 0.05a 14.67 ± 0.13b 450 W 13.44 ± 0.06c 45.97 ± 0.14b 20.73 ± 0.13b 14.31 ± 0.04c Different letters in the same column indicate significant differences (P < 0.05). 3.2.4 Free –SH groups As shown in Fig. 4, the free -SH content was measured to further understand the variations in the molecular structures of the proteins. The untreated recovered samples contain 6.05 μmol/g of free –SH. With increasing ultrasonic power, the content of free SH gradually increases and reaches a maximum at 450 W (7.21 μmol/g). This increase is possibly due to the cavitation effect that cleaves the disulfide bonds of the protein, resulting in the formation of sulfhydryl groups [23]. However, the electrophoretic profile showed no change in molecular weight for all samples. Another explanation is that ultrasonic treatment promotes the partial unfolding of the protein, exposing the free –SH groups originally present within the protein molecule [30]. Sun et al. [29] treated OB proteins with ultrasound and the free sulfhydryl contents of the proteins increased with increasing ultrasound power, which is consistent with our experimental results. In contrast, Gulseren et al. [32] reported that sonication reduced the free sulfhydryl contents of the studied proteins. The varied findings suggest that differences in the type of protein used and solution and sonication conditions may lead to different results [33].Fig. 4 Free –SH groups of the proteins recovered using the ultrasound-assisted alkaline method with different ultrasonic power settings. 3.2.5 Dsc DSC provides insight into the conformational and structural stabilities of the ultrasonically treated recovered proteins [34]. As shown in Fig. 5, the thermograms of all samples exhibit single major heat absorption peaks and different denaturation temperatures. Notably, sonication reduces the denaturation temperatures of the proteins, with denaturation temperatures of 91.444, 89.846, 84.894, and 88.195 °C observed at 0, 150, 300, and 450 W of ultrasonic power, respectively. A decreased denaturation temperature may be related to several structural and conformational changes caused by breakage of the chemical bonds of proteins by ultrasound [35].Fig. 5 Differential scanning calorimetry thermograms of the proteins recovered using the ultrasound-assisted alkali method with different ultrasonic power settings. 3.3 Microstructures 3.3.1 Particle size and ζ-potential As shown in Fig. 6A, the average particle size of the proteins recovered using the ultrasound-assisted alkali method with different ultrasonic power settings decrease from 264.07 (0 W) to 237.17 nm (450 W) with increasing ultrasonic power. This is likely caused by the cavitation and turbulence effects generated by the ultrasonic probe [36]. Particle size reduction gives the protein a larger surface area, which increases the interactions between the protein and water molecules and leads to increased solubilities. Small particles also increase the surface hydrophobicity because cavitation exposes more groups to a polar environment [35].Fig. 6 Particle size (A) and ζ-potential (B) of the proteins recovered using the ultrasound-assisted alkali method with different ultrasonic power settings. As shown in Fig. 6B, the ζ-potential of all samples is negative, and its absolute value increases with increasing ultrasonic power. This is due to the unfolding of the protein caused by ultrasonic treatment, which results in the exposure of more anionic groups on the surface. The increased exposure of these groups leads to an increased electrostatic repulsive force, which enhances the dispersion of the protein and increases its stability [37]. 3.3.2 Sem As shown in Fig. 7, the image of the untreated recovered protein differs significantly from those of the ultrasonically treated recovered proteins. Compared to those of the untreated protein, the fragments of the ultrasonicated proteins become smaller and more regular with increasing ultrasonic power, which is consistent with the average particle size. This suggests that higher ultrasonic powers may lead to smaller structures and sonication changes the surface structure of the sample, which may affect its function.Fig. 7 Surface morphologies of the proteins recovered using the ultrasound-assisted alkali method with different ultrasonic power settings. 3.4 Functional characteristics 3.4.1 Protein solubility As shown in Fig. 8, the solubilities of the proteins recovered using the ultrasound-assisted alkali method display pH-dependent U-shaped curves. The lowest values (3–7 %) are observed at pH values of 4 and 5, which are related to the isoelectric points of the protein. Protein solubility also increases with increasing ultrasonic power, mainly because of the formation of large cavitation bubbles during sonication, which results in a significant increase in local temperature and pressure. Consequently, protein unfolding and conformational changes occur, leading to an increase in the number of hydrophilic amino acid residues and causing insoluble protein aggregates to become soluble aggregates [38].Fig. 8 Solubilities (A) and surface hydrophobicity(B) of the proteins recovered using the ultrasound-assisted alkali method with different ultrasonic power settings. 3.4.2 h0 As shown in Fig. 8, H0 increases with increasing ultrasonic power, which may be attributed to the unfolding of the protein structure and exposure of the internal hydrophobic groups to the polar environment. Several studies report similar conclusions, i.e., ultrasound treatment increases the H0 values of soy [30] and myofibrillar proteins [39]. The increase in H0 may lead to a decrease in the adsorption energy barrier at the air–water interface, thereby enhancing the adsorption kinetics. Consequently, the protein-water interactions increase, resulting in an improved protein FC and FS [35]. Therefore, the foaming properties and FS should be consistent with the H0 data. 3.4.3 Emulsifying properties The emulsification activity (EAI) of the sonicated samples increase with increasing sonication power (Table 4). The improved solubility and increased net surface charge may increase the protein-solvent interactions [40], and the protein may be adsorbed faster at the oil–water interface. Hence, the emulsification of the protein solution is enhanced [41]. Emulsion stability (ESI) increases from 365.43 (0 W) to 673.44 min (450 W), and thus, ESI is also enhanced by sonication. This may be due to the formation of smaller droplets or changes in the surfaces of the lipid droplets, thus altering the attraction or repulsion between the droplets [33]. Previous studies indicate that the emulsification properties of various proteins, such as pea [42] and walnut proteins [33], may be improved by sonication.Table 4 Effects of ultrasonic power on emulsification, foaming, and water/oil-holding capacities. Ultrasonic power EAI (m2/g) ESI (min) FC (%) FS (%) WAC (g/g) OAC(g/g) 0 W 10.05 ± 0.21c 365.43 ± 46.77c 69.79 ± 4.77b 56.67 ± 1.94c 2.16 ± 0.02d 4.49 ± 0.01c 150 W 12.10 ± 0.61b 424.12 ± 30.07bc 70.83 ± 3.61b 67.68 ± 4.63b 2.87 ± 0.04c 5.18 ± 0.08b 300 W 14.27 ± 1.10a 478.66 ± 24.16b 79.17 ± 9.55b 73.69 ± 2.87b 3.60 ± 0.03b 5.27 ± 0.01b 450 W 15.01 ± 0.79a 673.44 ± 64.94a 93.75 ± 6.25a 84.54 ± 2.89a 3.67 ± 0.01a 5.39 ± 0.08a Different letters in the same column indicate significant differences (P < 0.05). 3.4.4 Foaming properties and surface tension The FC and FS of the proteins recovered via ultrasound-assisted alkaline extraction at the same concentrations are shown in Table 4. The FC of the sonicated protein (450 W) is 1.34-fold higher than that of the untreated protein due to the effect of ultrasonic homogenization [43]. Moreover, the weaker intermolecular interactions between protein molecules may be due to the smaller particle size and higher solubility [30]. The FS increases from 56.67 % (0 W) to 84.54 % (450 W) with increasing ultrasonic power, which depends mainly on the unfolding of the protein structure. The proteins rapidly transfer and adsorb at the air–water interface, forming a thick viscous film around the bubble and generating a stable foam [23]. The surface tensions at the air–water interfaces of the proteins recovered using different ultrasonic power settings as functions of time are shown in Fig. 9. The surface tensions of all samples decrease rapidly in the first 50 s and the decrease gradually, indicating that the proteins diffuse rapidly to the interfaces and are adsorbed in the first 50 s, followed by the formation of energy barriers during adsorption and rearrangement. The surface tension of the recovered protein is significantly lower after sonication compared to that of the untreated protein. This may be attributed to the increased flexibility of the protein due to sonication and the large amount of protein adsorbed at the interface. Surface tension exhibits a good negative correlation with FC [40], which further supports the conclusion regarding the foaming properties of the proteins.Fig. 9 Surface tensions of the proteins recovered using the ultrasound-assisted alkali method with different ultrasonic power settings. 3.4.5 WAC and OAC The polar amino groups of proteins are the main sites of water-protein interactions, which affect their water-binding properties. Differences in the purities and conformational characteristics of proteins may lead to variations in WAC [44]. The WAC of the recovered proteins after sonication were higher than that of the untreated protein, slowly increasing with a gradual increase in sonication power and reaching a maximum of 3.67 g/g when the power was 450 W. This may be explained by the increased exposure of polar amino acids due to conformational changes in the protein during extraction [45]. The OAC of proteins is associated with enhanced flavor retention and improved mouthfeel, and the maximum OAC (5.39 g/g) was reached when the sonication power was 450 W. The OAC is associated with the amino acids in proteins, particularly the hydrophobic residues, interacting with the hydrocarbon chains of the lipid molecules [46]. The high OACs of the ultrasonically recovered proteins renders them suitable as ingredients for use in various food products, such as sauces and sausages [47]. 3.4.6 Gelation behavior As shown in Fig. 10, G′ > G′′ for all samples during temperature variation, indicating the dominant elastic behaviors of the protein gels. The initial G′ of the untreated sample exceeds those of the ultrasonically treated samples, while that of the untreated samples showed a decreasing trend during the first 1000 s because of the hydrophobic interactions of the proteins at low temperatures. G′ then begins to display an increasing trend, indicating the initiation of the formation of a gel network [48]. With time, the G′ values of the ultrasonically treated samples significantly exceed that of the untreated sample, suggesting that sonication enhances the strengths of the protein gels. G′ increases significantly with increasing sonication power because the ultrasound broke the noncovalent bond interactions, which facilitated the intermolecular interactions of the proteins, resulting in stronger gels [37]. The increase in G′ observed during the cooling phase is due to the unfolding of the protein structure and exposure of hydrophobic groups during the high-temperature phase. Consequently, enhanced interactions between protein molecules are more conducive to gel formation [49]. Previous studies have reported that the random coil content of gel samples may be related to gel properties, with higher random coil content associated with poorer gel properties[18]. The stronger gels formed after sonication may be associated with random coil content, with a decrease in random coil content of sonicated proteins compared to untreated proteins.Fig. 10 Thermal gelation processes of the proteins recovered using the ultrasound-assisted alkali method with different ultrasonic power settings. 4 Conclusions The ultrasound-assisted alkali method with different ultrasonication power settings was applied to successfully remove and recover proteins from wastewater generated by OB extraction and study the effects of the employed ultrasonication power on protein structure and function. The protein recovery increased with ultrasonic power, and the relative contents of α-helices and random coils of the sonicated samples gradually decreased, whereas those of the β-turns increased. The improvements in functionality of the sonicated samples could be attributed to the unfolding of the protein structure and exposure of the internal hydrophobic groups due to ultrasound treatment. This method reduces the nutrient content of wastewater and enables the effective recovery of proteins with specific properties and functions, which not only benefits the environment, but also aids in maximizing the functional value of soybean processing. However, this method still exhibits three problems during treatment: 1) a large amount of sucrose is added during OB extraction to enhance collection; 2) the wastewater still contains large amounts of other nutrients; and 3) the pH of the wastewater is alkaline. These limitations require further investigation in future studies. In terms of subsequent research, we may identify methods of processing them into functional beverages or extracting the sugars from the waste stream generated by oil-body extraction to realize a green recycling method. Currently, our studies are still in the laboratory stage, but the insight gained from our research should aid in application in industry as an alternative to existing techniques. CRediT authorship contribution statement Hanyu Song: Methodology, Formal analysis, Writing – original draft. Mingming Zhong: Methodology, Data curation. Yufan Sun: Methodology. Qiang Yue: . Baokun Qi: Conceptualization, Funding acquisition, Supervision. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Data availability The authors do not have permission to share data. Acknowledgements We gratefully acknowledge financial support received from the National Key Research and Development Program of China (No. 2021YFD2100301-01). ==== Refs References 1 Sun Y. Zhong M. Wu L. Wang Q.i. Li Y. Qi B. Loading natural emulsions with nutraceuticals by ultrasonication: Formation and digestion properties of curcumin-loaded soybean oil bodies Food Hydrocoll. 124 2022 107292 2 Liu C. Wang R. He S. Cheng C. Ma Y. The stability and gastro-intestinal digestion of curcumin emulsion stabilized with soybean oil bodies LWT 131 2020 109663 10.1016/j.lwt.2020.109663 3 Lee S.Y. Stuckey D.C. Separation and biosynthesis of value-added compounds from food-processing wastewater: Towards sustainable wastewater resource recovery J. Clean. Prod. 357 2022 131975 4 Momen S. Alavi F. Aider M. Alkali-mediated treatments for extraction and functional modification of proteins: Critical and application review Trends Food Sci. Technol. 110 2021 778 797 10.1016/j.tifs.2021.02.052 5 Tanger C. Engel J. Kulozik U. Influence of extraction conditions on the conformational alteration of pea protein extracted from pea flour Food Hydrocoll. 107 2020 105949 6 Cui L. Bandillo N. Wang Y. Ohm J.-B. Chen B. Rao J. Functionality and structure of yellow pea protein isolate as affected by cultivars and extraction pH Food Hydrocoll. 108 2020 106008 7 Gao Z. Shen P. Lan Y. Cui L. Ohm J.B. Chen B. Rao J. Effect of alkaline extraction pH on structure properties, solubility, and beany flavor of yellow pea protein isolate Food Res Int 131 2020 109045 10.1016/j.foodres.2020.109045 8 Hadidi M. Khaksar F.B. Pagan J. Ibarz A. Application of Ultrasound-Ultrafiltration-Assisted alkaline isoelectric precipitation (UUAAIP) technique for producing alfalfa protein isolate for human consumption: Optimization, comparison, physicochemical, and functional properties Food Res Int 130 2020 108907 10.1016/j.foodres.2019.108907 9 Ampofo J. Ngadi M. Ultrasound-assisted processing: Science, technology and challenges for the plant-based protein industry Ultrason Sonochem 84 2022 105955 10.1016/j.ultsonch.2022.105955 10 Meroni D. Djellabi R. Ashokkumar M. Bianchi C.L. Boffito D.C. Sonoprocessing: From Concepts to Large-Scale Reactors Chem Rev 122 2022 3219 3258 10.1021/acs.chemrev.1c00438 34818504 11 Sengar A.S. Thirunavookarasu N. Choudhary P. Naik M. Surekha A. Sunil C.K. Rawson A. Application of power ultrasound for plant protein extraction, modification and allergen reduction – A review Applied Food Research 2 2 2022 100219 12 Chittapalo T. Noomhorm A. Ultrasonic assisted alkali extraction of protein from defatted rice bran and properties of the protein concentrates Int. J. Food Sci. Technol. 44 2009 1843 1849 10.1111/j.1365-2621.2009.02009.x 13 Gadalkar S.M. Rathod V.K. Extraction of watermelon seed proteins with enhanced functional properties using ultrasound Prep Biochem Biotechnol 50 2020 133 140 10.1080/10826068.2019.1679173 31702437 14 Wang R. Li L. Feng W. Wang T. Fabrication of hydrophilic composites by bridging the secondary structures between rice proteins and pea proteins toward enhanced nutritional properties Food Funct 11 2020 7446 7455 10.1039/d0fo01182g 32808004 15 Pezeshk S. Rezaei M. Hosseini H. Abdollahi M. Impact of pH-shift processing combined with ultrasonication on structural and functional properties of proteins isolated from rainbow trout by-products Food Hydrocoll. 118 2021 106768 16 Liu G. Hu M. Du X. Liao Y.i. Yan S. Zhang S. Qi B. Li Y. Correlating structure and emulsification of soybean protein isolate: Synergism between low-pH-shifting treatment and ultrasonication improves emulsifying properties Colloids Surf A Physicochem Eng Asp 646 2022 128963 17 Jiang Y. Zhang C. Zhao X. Xu X. Recovery of emulsifying and gelling protein from waste chicken exudate by using a sustainable pH-shifting treatment Food Chem 387 2022 132886 10.1016/j.foodchem.2022.132886 18 Zheng L.i. Wang ZhongJiang Kong Y. Ma ZhaoLei Wu ChangLing Regenstein J.M. Teng F. Li Y. Different commercial soy protein isolates and the characteristics of Chiba tofu Food Hydrocoll. 110 2021 106115 19 Sun Y. Zhang S. Xie F. Zhong M. Jiang L. Qi B. Li Y.J.F.C. Effects of covalent modification with epigallocatechin-3-gallate on oleosin structure and ability to stabilize artificial oil body emulsions 341 2021 128272 20 Li Y. Zhong M. Xie F. Sun Y. Zhang S. Qi B. The effect of pH on the stabilization and digestive characteristics of soybean lipophilic protein oil-in-water emulsions with hypromellose Food Chem 309 2020 125579 10.1016/j.foodchem.2019.125579 21 Mir N.A. Riar C.S. Singh S. Effect of pH and holding time on the characteristics of protein isolates from Chenopodium seeds and study of their amino acid profile and scoring Food Chem 272 2019 165 173 10.1016/j.foodchem.2018.08.048 30309527 22 López-Monterrubio D.I. Lobato-Calleros C. Alvarez-Ramirez J. Vernon-Carter E.J. Huauzontle (Chenopodium nuttalliae Saff.) protein: Composition, structure, physicochemical and functional properties Food Hydrocoll. 108 2020 106043 23 Xiong T. Xiong W. Ge M. Xia J. Li B. Chen Y. Effect of high intensity ultrasound on structure and foaming properties of pea protein isolate Food Res Int 109 2018 260 267 10.1016/j.foodres.2018.04.044 29803449 24 Yan S. Xu J. Zhang S. Li Y. Effects of flexibility and surface hydrophobicity on emulsifying properties: Ultrasound-treated soybean protein isolate Lwt 142 2021 110881 25 Li J. Wu M. Wang Y. Li K. Du J. Bai Y. Effect of pH-shifting treatment on structural and heat induced gel properties of peanut protein isolate Food Chem 325 2020 126921 10.1016/j.foodchem.2020.126921 26 Jiang L. Wang J. Li Y. Wang Z. Liang J. Wang R. Chen Y. Ma W. Qi B. Zhang M. Effects of ultrasound on the structure and physical properties of black bean protein isolates Food Res. Int. 62 2014 595 601 10.1016/j.foodres.2014.04.022 27 Zhong M. Sun Y. Sun Y. Fang L. Wang Q. Qi B. Li Y. Soy lipophilic protein self-assembled by pH-shift combined with heat treatment: Structure, hydrophobic resveratrol encapsulation, emulsification, and digestion Food Chem 394 2022 133514 10.1016/j.foodchem.2022.133514 28 Ding Y. Ma H. Wang K.e. Azam S.M.R. Wang Y. Zhou J. Qu W. Ultrasound frequency effect on soybean protein: Acoustic field simulation, extraction rate and structure Lwt 145 2021 111320 29 Sun Y. Zhong M. Wu L. Huang Y. Li Y. Qi B. Effects of ultrasound-assisted salt (NaCl) extraction method on the structural and functional properties of Oleosin Food Chem 372 2022 131238 10.1016/j.foodchem.2021.131238 30 Hu H. Wu J. Li-Chan E.C.Y. Zhu L. Zhang F. Xu X. Fan G. Wang L. Huang X. Pan S. Effects of ultrasound on structural and physical properties of soy protein isolate (SPI) dispersions Food Hydrocoll. 30 2013 647 655 10.1016/j.foodhyd.2012.08.001 31 Eze O.F. Chatzifragkou A. Charalampopoulos D. Properties of protein isolates extracted by ultrasonication from soybean residue (okara) Food Chem 368 2022 130837 10.1016/j.foodchem.2021.130837 32 Gulseren I. Guzey D. Bruce B.D. Weiss J. Structural and functional changes in ultrasonicated bovine serum albumin solutions Ultrason Sonochem 14 2007 173 183 10.1016/j.ultsonch.2005.07.006 16959528 33 Zhu Z. Zhu W. Yi J. Liu N. Cao Y. Lu J. Decker E.A. McClements D.J. Effects of sonication on the physicochemical and functional properties of walnut protein isolate Food Res Int 106 2018 853 861 10.1016/j.foodres.2018.01.060 29579996 34 Sha L. Xiong Y.L. Comparative structural and emulsifying properties of ultrasound-treated pea (Pisum sativum L.) protein isolate and the legumin and vicilin fractions Food Res Int 156 2022 111179 10.1016/j.foodres.2022.111179 35 Mir N.A. Riar C.S. Singh S. Physicochemical, molecular and thermal properties of high-intensity ultrasound (HIUS) treated protein isolates from album (Chenopodium album) seed Food Hydrocoll. 96 2019 433 441 10.1016/j.foodhyd.2019.05.052 36 Amiri A. Sharifian P. Soltanizadeh N. Application of ultrasound treatment for improving the physicochemical, functional and rheological properties of myofibrillar proteins Int J Biol Macromol 111 2018 139 147 10.1016/j.ijbiomac.2017.12.167 29307807 37 Bi C.-H. Chi S.-y. Zhou T. Zhang J.-y. Wang X.-Y. Li J. Shi W.-T. Tian B. Huang Z.-G. Liu Y.i. Effect of low-frequency high-intensity ultrasound (HIU) on the physicochemical properties of chickpea protein Food Res. Int. 159 2022 111474 35940749 38 Huang L. Ding X. Li Y. Ma H. The aggregation, structures and emulsifying properties of soybean protein isolate induced by ultrasound and acid Food Chem 279 2019 114 119 10.1016/j.foodchem.2018.11.147 30611469 39 Zhang Z. Regenstein J.M. Zhou P. Yang Y. Effects of high intensity ultrasound modification on physicochemical property and water in myofibrillar protein gel Ultrason Sonochem 34 2017 960 967 10.1016/j.ultsonch.2016.08.008 27773327 40 Chang L. Lan Y. Bandillo N. Ohm J.-B. Chen B. Rao J. Plant proteins from green pea and chickpea: Extraction, fractionation, structural characterization and functional properties Food Hydrocoll. 123 2022 107165 41 Cai Y. Huang L. Chen B. Zhao X. Zhao M. Zhao Q. Van der Meeren P. Effect of alkaline pH on the physicochemical properties of insoluble soybean fiber (ISF), formation and stability of ISF-emulsions Food Hydrocoll. 111 2021 106188 42 Wang F. Zhang Y. Xu L. Ma H. An efficient ultrasound-assisted extraction method of pea protein and its effect on protein functional properties and biological activities Lwt 127 2020 109348 43 A. Stefanović, J. Jovanović, M. Dojčinović, S. Lević, M. Žuža, V. Nedović, Z.J.J.o.H.E. Knežević-Jugović, Design, Impact of high-intensity ultrasound probe on the functionality of egg white proteins, J. Hyg. Eng. Des. 6 (2014) 215-224. 44 López D.N. Ingrassia R. Busti P. Wagner J. Boeris V. Spelzini D. Effects of extraction pH of chia protein isolates on functional properties Lwt 97 2018 523 529 10.1016/j.lwt.2018.07.036 45 López D.N. Ingrassia R. Busti P. Bonino J. Delgado J.F. Wagner J. Boeris V. Spelzini D. Structural characterization of protein isolates obtained from chia (Salvia hispanica L.) seeds Lwt 90 2018 396 402 10.1016/j.lwt.2017.12.060 46 Das D. Mir N.A. Chandla N.K. Singh S. Combined effect of pH treatment and the extraction pH on the physicochemical, functional and rheological characteristics of amaranth (Amaranthus hypochondriacus) seed protein isolates Food Chem 353 2021 129466 10.1016/j.foodchem.2021.129466 47 Yu M. Zeng M. Qin F. He Z. Chen J. Physicochemical and functional properties of protein extracts from Torreya grandis seeds Food Chem 227 2017 453 460 10.1016/j.foodchem.2017.01.114 28274456 48 Hu H. Fan X. Zhou Z. Xu X. Fan G. Wang L. Huang X. Pan S. Zhu L. Acid-induced gelation behavior of soybean protein isolate with high intensity ultrasonic pre-treatments Ultrason Sonochem 20 2013 187 195 10.1016/j.ultsonch.2012.07.011 22925550 49 Malik M.A. Saini C.S. Rheological and structural properties of protein isolates extracted from dephenolized sunflower meal: Effect of high intensity ultrasound Food Hydrocoll. 81 2018 229 241 10.1016/j.foodhyd.2018.02.052
PMC010xxxxxx/PMC10352633.txt
==== Front J Caring Sci J Caring Sci J Caring Sci TBZMED Journal of Caring Sciences 2251-9920 2251-9920 Tabriz University of Medical Sciences 10.34172/jcs.2023.31856 Original Article Psychometric Properties of the Greek Version of Demoralization Scale-II (DS-II) in Patients with Cancer https://orcid.org/0000-0001-6117-5128 Elmasian Tania-Flora 1 Conceptualization Investigation Methodology Resources Writing – original draft https://orcid.org/0000-0001-6180-7092 Nikoloudi Maria 2 Data curation https://orcid.org/0009-0005-4215-1537 Tsilika Eleni 3 Formal analysis Supervision Validation Visualization https://orcid.org/0000-0001-6106-3927 Kostopoulou Sotiria 2 Resources Writing – original draft Writing – review & editing https://orcid.org/0000-0001-9698-6852 Zygogianni Anna 4 Supervision https://orcid.org/0000-0002-2354-9958 Katsaragakis Stylianos 5 Supervision https://orcid.org/0000-0002-5236-1014 Mystakidou Kyriaki 2 *Project administration 1Social Policy and Social Anthropology, Ministry of Migration and Asylum, Asylum Service Case Officer, Greece 2Pain Relief and Palliative Care Unit, Aretaieion Hospital, National & Kapodistrian University of Athens, School of Medicine, Athens, Greece 3Health Psychologist, Pain Relief and Palliative Care Unit, Aretaieion Hospital, National & Kapodistrian University of Athens, School of Medicine, Athens, Greece 4Department of Radiology, Aretaieion Hospital, School of Medicine, National & Kapodistrian University of Athens, Athens, Greece 5Department of Nursing, National and Kapodistrian University of Athens, Athens, Greece * Kyriaki Mystakidou [email protected] 6 2023 20 5 2023 12 2 103109 02 11 2022 17 3 2023 © 2023 The Author(s). 2023 https://creativecommons.org/licenses/by-nc/4.0/ This work is published by Journal of Caring Sciences as an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by-nc/4.0/). Non-commercial uses of the work are permitted, provided the original work is properly cited. http://jcs.tbzmed.ac.ir Introduction: The concept of demoralization is used to describe situations of existential distress and self-perceived inability to effectively deal with stressors. The Demoralization Scale-II (DS-II) is a short and modified version of the original DS that measures the level of demoralization in patients. The purpose of this study is to evaluate the psychometric properties of the Greek version of the Greek Demoralisation Scale-II (DS-II GR) in the population of patients with cancer. Methods: The main tool used in this cross-sectional study is the DS-II GR translated and evaluated for its psychometric properties in a sample of 150 Greek patients with cancer. Exploratory factor analysis (EFA), confirmatory factor analysis (CFA), convergent validity, known groups’ validity, cut-off points, internal consistency reliability and test-retest reliability were done. Results: According to the CFA, a two-factor model emerged with a different conceptual content and grouping than the original. The correlation coefficients between DS-II GR and Hospital Anxiety and Depression Scale-Greek (HADS-GR) The internal consistency of DS-II GR for factor 1, factor 2, and total score were measured with Cronbach’s alpha and calculated to be 0.906, 0.810, and 0.913. Conclusion: The Greek version of the demoralization scale is reliable and valid for assessing demoralization in Greek patients with cancer. Cancer Palliative care Anxiety Depression Demoralization ==== Body pmcIntroduction Demoralization refers to a persistent failure of coping with stress as defined by Jerome Frank thirty years ago. Feelings of despair, isolation, hopelessness, loss of meaning and existential distress are the core features of the definition of demoralization.1,2 The necessity of examining it in patients with cancer and its complications: Physical symptoms in cancer patients may affect the increment of demoralisation.3 In addition, when patients cannot effectively manage stressful situations they feel a sense of helplessness and incompetence, which results in the feeling of lost significance and purpose in life. This psychological reaction is common among patients with advanced cancer. Demoralization in patients with cancer is closely correlated to depression. Moreover, demoralization and depression increase the risk of suicide among patients with cancer. Compared with depressed patients, demoralized patients require additional suicide-risk assessments. These results suggest that health care professionals must be sensitive to depression and demoralization syndromes in patients with cancer. The signs of demoralization are loss of purpose, meaning, hopelessness-helplessness.1,2 Demoralization is classified as a serious and potentially treatable condition.4There is a variety of psychometric tools for measuring and evaluating demoralization.4-10 The novelty of the present study is that this is the only study regarding the psychometric properties of the Demoralisation Scale-II (DS-II) in Greek patients with cancer. The available tools to measure demoralization are: The DS,11The Short Demoralization Scale (SDS)12and a survey regarding demoralization as a diagnostic specifier for adjustment disorder and major depression.11 The aim of this study was to evaluate the psychometric properties of the Greek version of the DS II scale in oncology patients. The demoralization tool was chosen as it is the most commonly used instrument for the assessment of demoralization scale. It is of interest the history of demoralization asthe prominent existential psychotherapist Irvin Yalom is central to the discussion on demoralization as a large focus of his work was related to existential distress. Yalom defined existential psychotherapy as a therapy that is dynamic and focused on the distress that is grounded in the individual’s existence. Demoralization can be understood to result from this existential conflict when a person lacks the resources to cope with such conflict of life.13Hopelessness, loss of meaning, and existential distress are proposed as the core features of the diagnostic category of demoralization syndrome. This syndrome can be differentiated from depression and is recognizable in palliative care settings. It is associated with chronic medical illness, disability, bodily disfigurement, fear of loss of dignity, social isolation, and where there is a subjective sense of incompetence feelings of greater dependency on others or the perception of being a burden. The impact of cancer on adverse emotional conditions such as despair can instigate the progression of suicidal ideation which, in turn, can contribute into actual suicidal behaviour. Additionally, because of the sense of impotence or helplessness, those with the syndrome predictably progress to a desire to die or to commit suicide.14 A treatment approach is described which the potential to alleviate the distress has caused by this syndrome. Overall, demoralization syndrome has satisfactory face, descriptive, predictive, construct, and divergent validity, suggesting its utility as a diagnostic category in palliative care.4 In patients with cancer, demoralization has been associated with stressors that reflect a state of existential discomfort and disintegration, destroying one’s sense of self-worth to effectively manage internal and external stimuli throughout the duration-stages of the disease. Disease-related conditions range between existential integrity, spiritual well-being, a sense of serenity and severe existential despair.7 This syndrome includes diverse emotional states of despair and loss of meaning and purpose, along with cognitive perceptions of subjective impotence and personal failure in life that emerge from a sense of being trapped in a circumstance. The absence of a promising future due to the loss of values, roles, and goals, the lack of self-confidence and inner strength to achieve them, deprives the patient to establish effective and appropriate treatment methods.8 Jerome Frank coined the term demoralization in the 1970s.9 Clarke and Kissane hypothesized that demoralization syndrome is a distinct clinical entity characterized by symptoms: existential discomfort including despair, loss of meaning and purpose and cognitive-behavioural attitudes including pessimism, weakness, a sense of entrapment, personal failure, and lack of motivation. When combined with the absence of depressive symptoms, these effects should last for at least two weeks.2,4 The inclusion criteria included patients’ ≥ 18 years old, with advanced cancer diagnosis, considered to be as locally advanced, recurrent, or metastatic disease for solid cancers, and relapsed or refractory disease for hematologic tumours. A convenient sample of 150 from a total of 250 patients treated in the unit during that period participated in the study. All participants provided a written informed consent and underwent symptom evaluation by a specialist palliative care physician. There has been much discussion about the syndrome of demoralization in palliative care.1Cancer can be demoralizing to patients because disrupts their biopsychosocial status and equilibrium threating their physical and mental integrity.1-4 Demoralization is a mental condition that has negative consequences for patients, making them vulnerable and jeopardizing the success of therapeutic interventions, a factor that may be associated with the desire for premature death.5,6 Kissane et al differentiated demoralization from depression and found that 7–14% of patients with cancer were demoralized but not depressed.11 In a recent systematic review depression was significantly associated with high levels of demoralization.13 Untreated demoralization results in the development of depression at a later stage.15 The signs of depression are: lethargy and a loss of interest in enjoyable activities, while the signs of demoralization are: loss of purpose, meaning, hopelessness-helplessness.16,17 Demoralization is classified as a serious and potentially treatable condition.4 There is a variety of psychometric tools for measuring and evaluating demoralization.10,11,18-21The aim of this study was to evaluate the psychometric properties of the Greek version of the DS II scale in oncology patients. Measuring demoralization with valid instruments, such as the Greek version of the DS-II questionnaire, is important on reducing error in the measurement process. Measuring demoralization with valid instruments, such as the Greek version of the DS-II questionnaire, is important for the recognition of emotional and cognitive states of existential distress and demoralization in Greek patients with cancer. The DS-II has demonstrated convergent validity with measures of psychological distress, quality of life, and attitudes toward the end of life. It also demonstrated discriminant validity, as the DS-II differentiated patients who had different functional performance levels and high/low symptoms, with a difference of 2 points between groups on the DS-II considered clinically meaningful. Furthermore, discriminant validity was demonstrated, as comorbidity with depression was not observed at moderate levels of demoralization.22 A study by Boxley et al23 examined the internal consistency and factor structure of the Hospital Anxiety and Depression Scale (HADS( in a polytrauma/traumatic brain injury clinic. A confirmatory factor analysis (CFA) of the depression and anxiety subscales showed that the two factors were highly correlated (r=0.70). Goodness of fit statistics for the two-factor model were acceptable. The HADS demonstrated very good reliability overall (alpha=0.89) and for the individual subscales (alpha=0.84). This study supports the use of the HADS as a screen for depression and anxiety. Materials and Methods In this prospective cross-sectional observational study, data were collected from 150 advanced patients with cancer in the First Radiology Laboratory and Palliative Care Unit “Jenny Karezi” of Aretaieion Hospital. Approval No: 255/02-10-2020 was granted by Aretaieion Ethics committee. The study was conducted in accordance with the Declaration of Helsinki and data was collected between 03/04/2019 and 18/09/2019. The first time 150 patients completed the DS-II GR questionnaire, while the second time 30 patients completed the questionnaire after 1-3 days for the reliability analysis. A translation of the DS-II into Greek was performed using the “forward-backword” method: two independent bilingual health care professionals translated it to Greek and then two additional bilingual independent health-care professionals translated it back into English. A matching of these translations was then performed. Finally, a review of the translation of the English version and the reverse translation were performed with excellent results. The forward-back translation method was used to translate the original version of the questionnaire into Greek. The two versions were then compared, and minor changes were made to arrive at the final agreed-upon version. The main tool for the current study was the DS II which is a short form of the DS. The DS-II is a 16-item, 2-factor scale that has demonstrated item fit, uni-dimensionality, and reliability as a measure of demoralization in patients receiving palliative care.21 The HADS was originally developed by Zigmond and Snaith and is commonly used to determine the levels of anxiety and depression that a person is experiencing.24 The HADS is a fourteen-item scale that generates: Seven of the items relate to anxiety and seven relate to depression. Zigmond and Snaith created this outcome measure specifically to avoid reliance on aspects of these conditions that are also common somatic symptoms of illness, for example fatigue and insomnia or hypersomnia. This, it was hoped, would create a tool for the detection of anxiety and depression in people with physical health problems. The “receiver operating curve” (ROC) function curve analysis was used to find the cut-off points of the “Meaning and Purpose” factor, the “Discomfort” factor, and the overall score of the demoralization scale for the differentiation of subgroups of patients based on their level of stress ROC, calculating the corresponding areas under the curve (AUC). The maximum probability estimation method was used to calculate areas below the ROC curve with standard error and 95% confidence interval (CI), while the HADS score was used to assess the sensitivity and specificity of the different cut-off points of the “Meaning and Purpose” factor of “Discomfort” and the overall score. The HADS determined convergent validity, particularly the Greek version of the scale.24,25 The data were statistically processed using SPSS version 21. The indicators calculated and analysed performed were: exploratory factor analysis (EFA), CFA, convergent or criterion validity, known groups’ validity, cut-off points, internal consistency reliability and test-retest reliability. Results A two-factor model of the authentic DS-II scale emerged, marginally characterized by unacceptable global adjustment indicators. Subsequently, two factors also emerged of different conceptual content and grouping than the original scale. The correlation coefficients between DS-II GR and HAD anxiety were: factor 1 (r=0.60, P< 0.001), factor 2 (r=0.51, P<0.001), and total score (r=-0.62, P<0.001). The correlation coefficients between DS-II GR and HAD depression were: factor 1 (r=0.70, P<0.001), factor 2 (r=0.45, P<0.001), and total score (r=-0.66, P<0.001). The above results indicate a high correlation satisfying the convergent validity. The internal consistency of DS-II GR for factors 1, 2 and total score were measured with Cronbach’s alpha and calculated to be 0.90, 0.81, and 0.91 respectively. These values indicate excellent internal consistency. The qualitative-quantitative demographic and clinical data of the study sample are found in Table 1. Table 1 Demographic and disease-related patient’s characteristics Variables No. (%) Gender Men 61) 40.7) Women 89 (59.3) Educational level Primary studies 28 (18.7) Secondary studies 73 (48.7) University studies 49 (32.7) Marital status Married 108(72.0) Single 11(7.3) Divorced 15(10.0) Widowed 16(10.7) Cancer location Breast 50 (33.3) Lung 22 (14.7) Urogenital 28 (18.7) Gastrointestinal tract 36 (24.0) Other 14 (9.3) Eastern cooperative oncology group score 0–1 121 (80.7) 2–3 29 (19.3) Metastasis No 68 (45.3) Yes 82 (54.7) Radiotherapy No 53 (35.3) Yes 97 (64.7) Surgery No 32 (21.3) Yes 118 (78.7) Caregiver Children 40 (26.7) Relative 104 (69.3) Friends – Other 6 (4.0) Therapy Curative 146 (97.3) Palliative 4 (2.7) Age Mean (SD) 61.51 Disease duration Median 9.0 CFA: A two-factor model of the original DS-II was conducted by CFA giving unacceptable global fit indices. The resulting global fit indices χ2=256.5, chi-square-degrees of freedom ratio=2.49, root mean square error of approximation (RMSEA)=0.100, comparative fit index (CFI)=0.870, normed fit index (NFI)=0.803, goodness-of-fit index (GFI)=0.822, adjusted goodness-of-fit (AGFI)=0.766 showed that a two-factor solution proposed by the author should be rejected but marginally. EFA: 16 items were analysed using an Oblique rotation. Two factors, with eigenvalue>1 and items factor loadings were≥0.40 were identified (Tables 2 and 3). Table 2 Eigenvalues and explained variance of DS-II questionnaire Items Eigenvalues % of Variance Cumulative % 1 7.33 45.83 45.83 2 1.54 9.67 55.50 3 0.92 5.81 4 0.84 5.01 5 0.73 4.61 6 0. 65 4.06 7 0.56 3.50 8 0.49 3.08 9 0.48 3.02 10 0.44 2.76 11 0.42 2.65 12 0.39 2.45 13 0.27 1.72 14 0.26 1.64 15 0.24 1.54 16 0.17 1.07 Table 3 Factor loadings of DS-IIsubscales Items Factor 1 Factor 2 1 0.61 2 0.84 3 0.83 4 0.64 5 0.74 6 0.68 7 0.76 8 0.69 9 0.64 10 0.51 11 0.67 12 0.59 13 0.58 14 0.72 15 0.67 16 0.69 Extraction method: maximum likelihood; Rotation: oblique; only loadings with values>0.4 are presented. CFA new structure: A two-factor model of DS-II was examined by CFA (Figure 1). The resulting global fit indices χ2 =193.64, chi-square-degrees of freedom (df) ratio=1.88, RMSEA=.071, CFI=0.905, NFI=0.900, GFI=0.860, AGFI=0.810 showed that the new two factor solution could be retained. Figure 1 Confirmatory factor analysis Convergent or criterion validity: The correlation coefficients between Hospital Anxiety and Depression Scale anxiety and DS-II factors were: factor 1 (r=0.60, P< 0.001), factor 2 (r=0.51, P<0.001), total score (r=-0.62, P<0.001). The correlation coefficients between HAD depression and DS-II factors were: factor 1 (r=0.70, P<0.001), factor 2 (r=0.45, P<0.001), total score (r=-0.66, P<0.001), indicating high correlation between DS-II subscales and total score with HAD anxiety- depression scales which satisfied the criterion validity. Known-groups validity: DS-II factor 1, 2 and total score were higher for patients with anxiety score>11 compared with those with score<11 (P<0.001; Table 4) Table 4 Known-group’s validity HAD anxiety N Mean (SD) P value * DS-II factor1 <11 125 2.23 (3.06) <0.001 ≥11 25 7.44 (6.00) DS-II factor2 <11 125 3.38 (2.62) <0.001 ≥11 25 5.68 (2.59) DS-II total <11 125 5.61 (5.14) <0.001 ≥11 25 13.12 (7.63) *Significant statistically. The cut-off points of DS-II total score: The area under the curve (AUC) of DS-II total=0.818 (P<0.001). The AUC of DS-II factor 1 was 0.778 (P<0.001). The AUC of DS-II factor 2 was 0.752 (P<0.001) (Figure 2). Figure 2 Sensitivity and specificity of DS Internal consistency reliability: The internal consistency of the DS-II factor 1, 2 and total score was measured with Cronbach’s alpha and estimated as 0.906, 0.810 and 0.913 indicate excellent internal consistency. Test-retest reliability: The paired samples t-test between initial assessment and reassessment of DS-II subscales and total score found no statistically significant difference. The results of stability indicated that DS-II factor 1, 2 and total score were consistent between the two occasions (Table 5). Table 5 Test-retest reliability (N=30) ICC (95% CI) Paired samples t test P value Mean (SD) Initial Reassessment DS-II factor 1 0.89 (0.77-0.95) 2.90(3.65) 3.16(3.50) 0.52 DS-II factor 2 0.90 (0.79-0.95) 4.2(2.85) 4.1(2.92) 0.75 DS-II Total 0.92 (0.84-0.96) 7.10(6.06) 7.26(6.20) 0.78 ICC, intr aclass correlation coefficient Discussion The purpose of this study was to translate and investigate the psychometric properties of the demoralization scale by assessing its reliability and validity in Greek patients with cancer. According to the CFA a two-factor model of the authentic DS II index emerged, characterized by unacceptable global adjustment indicators. Thus, an EFA of the index followed, and two factors emerged. In the Greek version elements 4 and 9 were transferred to factor 1 resulting in two factors consisting of elements 10 and 6. In the original validation of the DS 5 factors had been emerged.11 Robinson et al suggested a 2-factor solution with 1 item deleted from each component: “I am not in good spirits” and “I am ashamed of what little I have accomplished”. Component 1 was labeled “Meaning and Purpose” while component 2 was labelled “Distress and Coping Ability”.26 The Spanish version the questionnaire contained the 2 factors of Robinson showing that item 10 “I have a lot of regret about my life” was not significant.27In the DS-II version Española-Colombia factor analysis had shown 3 factors: factor 1 “Sense of life” with 7 instruments, factor 2 “Lack of emotional control” with 7 instruments and factor 3 “Depressive symptomatology” with 2 instruments.22 A high correlation was found between the DS-II GR sub-scales and the overall HADS score, demonstrating criterion validity. There was a very strong relationship between demoralization and emotional distress-combined anxiety and depression.28The results of Española-Colombian study indicate that demoralization has within its structure components given by some depressive symptoms, but demoralization and depression are two distinct components In Robinson et al comorbidity between depression and demoralization existed at high levels of demoralization.1 Cronbach’s alpha was used indicating that the overall score and sub-scales had excellent internal consistency. In study by Robinson et al the Cronbach’s alpha index was equal to 0.89 for the overall score, 0.84 for factor 1, and 0.82 for factor 2.26Similar findings were found in the Spanish version of the DS II index, where Cronbach’s alpha was 0.88 for the overall score, 0.83 for factor 1, and 0.79 for factor 228 while Cronbach alpha index in DS-II Española-Colombian was 0.87 for total score, 0.75 for factor 1 and 0.78 for factor 2, 3. 22The test re-test reliability revealed no statistically significant differences. Accordingly, in Robinson et al the Intraclass Correlation Coefficient values for DS-II factor 1,2 and overall score were 0.68, 0.82, and 0.80 between the initial assessment and the reassessment.27 The coefficient values in Española-Colombian between the two repeated measurements for the total score were 0.56 and for factor 1: 0.55, factor 2: 0.46 and factor 3: 0.57.22 In the German version, DS-II related significantly with depression, anxiety, mental distress, and body image disturbance.29 Patients with a HAD anxiety score > 11 had higher score in “Meaning and Purpose”, “Discomfort” and overall scores, compared to patients with a score < 11. In contrast Robinson et al total score had moderate-strong positive correlations with burden, depression and desire to die. Psychological symptom burden had a higher correlation with “Distress and Coping Ability” than “Meaning and Purpose”.26 In a Spanish population there was a close relationship between demoralization and emotional distress while between depression and anxiety was less.28Thus, the factors that explain the demoralization scale and the overall score appear to be highly correlated with anxiety and depression scales assessing patients’ emotional states. Conclusion The Greek version of DS-II has shown to be valid, reliable and feasible with adequate psychometric properties in patients with cancer. Acknowledgments The authors did not receive any financial support in relation to the research, authorship, or publication of this article. Authors’ Contribution Conceptualization: Tania-Flora Elmasian. Data curation: Maria Nikoloudi. Formal analysis: Eleni Tsilika. Investigation: Tania-Flora Elmasian. Methodology: Tania-Flora Elmasian. Project administration: Kyriaki Mystakidou. Resources: Tania-Flora Elmasian, Sotiria Kostopoulou. Supervision: Eleni Tsilika, Anna Zygogianni, Stylianos Katsaragakis. Validation: Eleni Tsilika. Visualization: Eleni Tsilika. Writing–original draft: Tania-Flora Elmasian, Sotiria Kostopoulou. Writing–review & editing: Sotiria Kostopoulou. COI-statement The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article. Data Availability All the data generated or analyzed during this study have been incorporated within this published article. Ethical Approval Ethical approval and permission for this study was granted by the Aretaieion Ethics committee. Funding We do not have any funding. Research Highlights What is the current knowledge? Demoralization has negative consequences for patients, jeopardizing the outcome of interventions and may be linked with wishes for hastened death. What is new here? This study demonstrates psychometric properties of DS-II GR in grease context which maintains internal consistency. ==== Refs References 1 Robinson S Kissane DW Brooker J Burney S A systematic review of the demoralization syndrome in individuals with progressive disease and cancer: a decade of research J Pain Symptom Manage 2015 49 3 595 610 10.1016/j.jpainsymman.2014.07.008 25131888 2 Clarke DM Kissane DW Demoralization: its phenomenology and importance Aust N Z J Psychiatry 2002 36 6 733 42 10.1046/j.1440-1614.2002.01086.x 12406115 3 Hong YT Lin YA Pan YX Lin JL Lin XJ Zhang J Understanding factors influencing demoralization among cancer patients based on the bio-psycho-social model: a systematic review Psychooncology 2022 31 12 2036 49 10.1002/pon.6023 36016470 4 Kissane DW Clarke DM Street AF Demoralization syndrome--a relevant psychiatric diagnosis for palliative care J Palliat Care 2001 17 1 12 21 10.1177/082585970101700103 11324179 5 Breitbart W Rosenfeld B Pessin H Kaim M Funesti-Esch J Galietta M Depression, hopelessness, and desire for hastened death in terminally ill patients with cancer JAMA 2000 284 22 2907 11 10.1001/jama.284.22.2907 11147988 6 Chochinov HM Dying, dignity, and new horizons in palliative end-of-life care CA Cancer J Clin 2006 56 2 84 103 10.3322/canjclin.56.2.84 16514136 7 Kissane DW Kissane DWPsychospiritual and existential distressThe challenge for palliative care Aust Fam Physician 2000 29 11 1022 5 11127057 8 Kissane DW Demoralization: a life-preserving diagnosis to make for the severely medically ill J Palliat Care 2014 30 4 255 8 10.1177/082585971403000402 25962256 9 Frank J The role of hope in psychotherapy Int J Psychiatry 1968 5 5 383 95 5659469 10 Bobevski I Kissane D McKenzie D Murphy G Perera C Payne I The Demoralization Interview: reliability and validity of a new brief diagnostic measure among medically ill patients Gen Hosp Psychiatry 2022 79 50 9 10.1016/j.genhosppsych.2022.10.002 36274426 11 Kissane DW Wein S Love A Lee XQ Kee PL Clarke DM The Demoralization Scale: a report of its development and preliminary validation J Palliat Care 2004 20 4 269 76 10.1177/082585970402000402 15690829 12 Galiana L Rudilla D Oliver A Barreto P The Short Demoralization Scale (SDS): a new tool to appraise demoralization in palliative care patients Palliat Support Care 2017 15 5 516 23 10.1017/s1478951516000973 28065203 13 Tang PL Wang HH Chou FH A systematic review and meta-analysis of demoralization and depression in patients with cancer Psychosomatics 2015 56 6 634 43 10.1016/j.psym.2015.06.005 26411374 14 Lai Q Huang H Zhu Y Shu S Chen Y Luo Y Incidence and risk factors for suicidal ideation in a sample of Chinese patients with mixed cancer types Support Care Cancer 2022 30 12 9811 21 10.1007/s00520-022-07386-8 36269433 15 Jacobsen JC Vanderwerker LC Block SD Friedlander RJ Maciejewski PK Prigerson HG Depression and demoralization as distinct syndromes: preliminary data from a cohort of advanced cancer patients Indian J Palliat Care 2006 12 1 8 16 10.4103/0973-1075.25913 16 Rudilla D Oliver A Galiana L Barreto P A new measure of home care patients’ dignity at the end of life: the Palliative Patients’ Dignity Scale (PPDS) Palliat Support Care 2016 14 2 99 108 10.1017/s1478951515000747 26062752 17 Angelino AF Treisman GJ Major depression and demoralization in cancer patients: diagnostic and treatment considerations Support Care Cancer 2001 9 5 344 9 10.1007/s005200000195 11497387 18 de Figueiredo JM de Figueiredo JMDemoralization and psychotherapy: a tribute to Jerome DFrank, MD, PhD (1909-2005) Psychother Psychosom 2007 76 3 129 33 10.1159/000099839 17426411 19 Dohrenwend BP Dohrenwend BS Levav I Shrout P The psychiatric epidemiology research interview Harefuah 1981 100 6 274 6 20 Levitt EE A structural analysis of the impact of MMPI-2 on MMPI-1 J Pers Assess 1990 55 3-4 562 77 10.1207/s15327752jpa5503&4_13 21 Cockram CA Doros G de Figueiredo JM Diagnosis and measurement of subjective incompetence: the clinical hallmark of demoralization Psychother Psychosom 2009 78 6 342 5 10.1159/000235737 19713728 22 Palacios-Espinosa X Sánchez-Pedraza R Rodríguez C Psychometric properties of Demoralization Scale (DS-II Spanish version-Colombia) for oncologic patients in palliative care Av Psicol Latinoam 2020 38 3 84 101 10.12804/revistas.urosario.edu.co/apl/a.8408 23 Boxley L Flaherty JM Spencer RJ Drag LL Pangilinan PH Bieliauskas LA Reliability and factor structure of the Hospital Anxiety and Depression Scale in a polytrauma clinic J Rehabil Res Dev 2016 53 6 873 80 10.1682/jrrd.2015.05.0088 28273327 24 Zigmond AS Snaith RP The hospital anxiety and depression scale Acta Psychiatr Scand 1983 67 6 361 70 10.1111/j.1600-0447.1983.tb09716.x 6880820 25 Mystakidou K Tsilika E Parpa E Katsouda E Galanos A Vlahos L The Hospital Anxiety and Depression Scale in Greek cancer patients: psychometric analyses and applicability Support Care Cancer 2004 12 12 821 5 10.1007/s00520-004-0698-y 15480813 26 Robinson S Kissane DW Brooker J Hempton C Michael N Fischer J Refinement and revalidation of the demoralization scale: the DS-II-external validity Cancer 2016 122 14 2260 7 10.1002/cncr.30012 27171544 27 Robinson S Kissane DW Brooker J Michael N Fischer J Franco M Refinement and revalidation of the demoralization scale: the DS-II-internal validity Cancer 2016 122 14 2251 9 10.1002/cncr.30015 27171617 28 Belar A Arantzamendi M Rodríguez-Núñez A Santesteban Y Martinez M López-Saca M Multicenter study of the psychometric properties of the new Demoralization Scale (DS-II) in Spanish-speaking advanced cancer patients J Pain Symptom Manage 2019 57 3 627 34 10.1016/j.jpainsymman.2018.11.016 30472315 29 Koranyi S Hinz A Hufeld JM Hartung TJ Quintero Garzón L Fendel U Psychometric evaluation of the German version of the Demoralization Scale-II and the association between demoralization, sociodemographic, disease- and treatment-related factors in patients with cancer Front Psychol 2021 12 789793 10.3389/fpsyg.2021.789793 34899543
PMC010xxxxxx/PMC10352634.txt
==== Front J Caring Sci J Caring Sci J Caring Sci TBZMED Journal of Caring Sciences 2251-9920 2251-9920 Tabriz University of Medical Sciences 10.34172/jcs.2023.30726 Review Article Psychosocial Interventions by Nurses for Patients with HIV/ AIDS: A Systematic Review https://orcid.org/0000-0002-6860-3794 Davoudi Malihe 1 Conceptualization Data curation Formal analysis Methodology Writing – original draft https://orcid.org/0000-0002-1082-7488 Heydari Abbas 2 Conceptualization Funding acquisition Methodology Project administration Supervision Writing – review & editing https://orcid.org/0000-0001-8270-7357 Manzari Zahra Sadat 2 *Conceptualization Formal analysis Methodology Supervision Writing – review & editing 1Student Research Committee, School of Nursing and Midwifery, Mashhad University of Medical Sciences, Mashhad, Iran 2Department of Medical-Surgical Nursing, School of Nursing and Midwifery, Mashhad University of Medical Sciences, Mashhad, Iran * Zahra Sadat Marzari [email protected] 6 2023 26 4 2023 12 2 94102 11 5 2022 05 7 2022 © 2023 The Author(s). 2023 https://creativecommons.org/licenses/by-nc/4.0/ This work is published by Journal of Caring Sciences as an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by-nc/4.0/). Non-commercial uses of the work are permitted, provided the original work is properly cited. http://jcs.tbzmed.ac.ir Introduction: Providing psychological support is one of the traditional roles of nurses for patients with HIV/AIDS. Searching the literature showed that various psychological interventions have been performed by nurses to support HIV/AIDS patients; however, no summary of these interventions is available. We aimed to systematically review the interventional studies which investigated the effectiveness of psychosocial interventions delivered by nurses to HIV/AIDS patients. Methods: This systematic review was performed based on Cochrane’s handbook of systematic reviews of interventional studies. Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement were used in this study. The databases of PubMed, Web of Science, Cochrane, Scopus and World Health Organization were searched from January 2009 to December 2022. Based on inclusion criteria, nine studies included in this systematic review. Cochrane data extraction form was used for the systematic review and the article’s information was summarized using the modified Jadad scale. Results: The interventions provided by the nurses included: virtual and face-to-face educational programs, written information resources, palliative care, motivational interview, case management, home visit, and care services, along with face-to-face and telephone follow-up. These interventions have a significant positive effect on the quality of life and management of high-risk behaviors, disease management, symptoms and complications, adherence to treatment, immune function, and mental health in patients with HIV/AIDS. Conclusion: The results of the present study show that despite the fact that the interventions have a purely psychological content and can be done with various methods, they are able to have positive consequences in physical, psychological, behavioral, and laboratory health in HIV/AIDS patients. HIV/AIDS Nurses Psychology Systematic review ==== Body pmcIntroduction Despite the ongoing attempts to cure and prevent AIDS/HIV, this disease remains one of the major health challenges worldwide.1 AIDS/HIV affects all aspects of a person’s quality of life, including physical, psychological, social, and spiritual aspects.2Patients with HIV/AIDS experience significant medically and psychologically suffering.3 There is a two-way and complex relationship between psychological health and the problems of these patients. HIV and associated infections can cause nerve damage.4 On the other hand, psychological health problems can be caused by antiviral treatment or the social stigma, stress, and economic and social consequences associated with treatment process.5 Psychological problems can negatively affect the adherence to antiviral treatment and development of AIDS and consequently poorer health outcomes.6 HIV patients compared to the general population are more likely to develop psychological health disorders such as depression, anxiety, suicide, and drug abuse.7 So, performing psychological interventions is needed.8 Since nurses are more firmly and continuously connected to the patients, they have the main role in the care of patients with AIDS/HIV.9,10 Therefore, their performance may affect the patients’ treatment and satisfaction.11One of the traditional roles and responsibilities of nurses is to provide psychological support. In holistic nursing care, psychological support is essential for healing of patients.12 Psychological support includes any support to help them improve their psychological, cognitive, emotional, and behavioral well-being. Psychological support is provided by a wide range of professional groups, peers, and informal providers, both in the clinical setting and in the community.13 Psychological support helps patients make informed decisions, better cope with disease, and better deal with discrimination. Moreover, psychological support improves the quality of life of patients with HIV and prevents further transmission of HIV infection. Besides, this type of support is important for patients with AIDS/HIV to follow antiretroviral therapy and ongoing counseling to strengthen adherence to treatment regimens.2 Searching the literature showed that various psychological interventions have been performed by nurses to support patients with AIDS/HIV, but no summary of these interventions is available. The implications of these interventions are also unclear. Therefore, this study was performed aimed to review and summarize the psychological interventions provided by nurses to patients with AIDS/HIV. This study increases our knowledge of the psychological interventions which nurses can directly provide to patients with AIDS/HIV. The present study also increases nurses’ awareness of the primary and secondary consequences of these interventions. Moreover, by knowing the possible consequences of each intervention, nurses can choose the appropriate intervention according to the available options and the patients’ preferences. The novelty and innovative aspects of this study is that, to the best of our knowledge, this is the first systematic review which summarizes and highlights the outcomes of psychological interventions performed by nurses on various aspects of the lives of patients with AIDS/HIV. This study also distinguished the types of psychological intervention approaches which are commonly used by nurses. Moreover, the results of psychological interventions were summarized. Our conclusions from the collected data provide positive strategies and new practical evidence for psychological support of patients with AIDS/HIV that can be applied in clinical practice. This review was a comprehensive, detailed, and systematic search of the literature. This study aimed to systematically review the interventional studies which investigated the effectiveness of psychosocial interventions delivered by nurses to patients with HIV/AIDS. Material and Methods This systematic review was performed based on Cochrane’s handbook of systematic reviews of interventional studies.14 The PRISMA statement (Preferred Reporting Items for Systematic Reviews and Meta-analysis) was used in this study.15 Search strategy was based on PICO (Participants, intervention, comparison, and outcomes). The participants were patients with HIV/AIDS, the intervention was psychological interventions, comparisons were routine management (i.e., routine care and standard medical care) or no intervention, and outcomes were the effects of the intervention on the physical, psychological, social, and other dimensions of the life of patients with AIDS/HIV. Literature searches were performed in the databases of PubMed, Web of Science, Scopus, Cochrane, and the World Health Organization clinical trials registration system) from 2009 to December 2022. The seminars, conferences,congresses, and journals were more searched. If needed, the researcher contacted the corresponding author for access to the full text of the article. Keywords were based on the mesh that included: Psychological intervention, Mental intervention, Nursing, HIV, AIDS, acquired immune deficiency syndrome, and Human immunodeficiency virus and combination of them using Boolean operators (OR, NOT, AND) (Table 1). Table 1 An example of a database search strategy Database Search strategy Web of science Search field; title; search term ("psychological intervention" or "mental intervention") AND ("HIV" OR "AIDS" OR "Human immunodeficiency virus" OR "acquired immune deficiency syndrome") Limits; restricted to articles, nursing, English, open access PubMed Search: ("psychological intervention"[Title/Abstract] OR "mental intervention"[Title/Abstract]) AND nurse*[Title/Abstract] AND (HIV[Title/Abstract] OR AIDS[Title/Abstract] OR "Human immunodeficiency virus"[Title/Abstract] OR "acquired immune deficiency syndrome"[Title/Abstract] ") Limits: restricted to Randomized Controlled Trial, Controlled Clinical Trial, Clinical Trial, pragmatic Controlled Trial Congress, full text, English Scopus Article title, abstract, keyword; ("psychological intervention" OR "mental intervention") AND (HIV OR "Human immunodeficiency virus") OR ("AIDS OR "acquired immune deficiency syndrome") Limits; nursing, psychology, social science, article, English, open access Cochran Title/abstract, keyword: ("psychological intervention" OR "mental intervention") AND nurse AND ("("HIV" OR "AIDS" OR "Human immunodeficiency virus" OR "acquired immune deficiency syndrome") Limited; trials Other citations from the original articles and systematic reviews were also searched and evaluated. After the initial evaluation, duplicate studies were excluded. Then, the abstracts of the articles were reviewed and those which did not coincide with the inclusion criteria were excluded. The full text of the remaining studies was re-evaluated. Finally, nine studies were analyzed (Table 2). Table 2 Characteristics of included studies Author (year), Country Total sample size Type of intervention in the experimental group Type of intervention in the control group Finding Details Duration of intervention Blank et al16 (2011), USA 238 Providing counseling and coordination of home and mental health services 12 months Routine care Significant reduction in viral load Blank et al17 (2014), USA 238 Home care services and coordination between services 12 months Routine care Moderate to excellent changes in CD4+T cell and viral load Improving the quality of life in the physical and mental dimension Côté et al18 (2015), Canada 179 Virtual training and follow-up (motivational skills, emotion recognition, and management skills, problem solving, and communication with others) 8 weeks (4 sessions of 20-30 minutes) In-person counseling (providing a list of drug-related websites and side effects) Improve adherence to treatment Increase self-efficacy Reducing stress Increasing positive attitude Average increase in the level of social support Reducing the discomfort associated with the symptoms of the disease Eller et al19 (2013), South Africa 222 Symptom management training with practical training A 30-minute session Nutrition and supportive care training Reducing the number of depression symptoms, severity and its effects Positive change in the use of self-care and self-management strategies Hanrahan et al20 (2011) United States American 238 An advanced nursing care model for face-to-face and telephone case management at home 12 months Routine care Improvement of symptoms of depression, mental problems Improving the quality of life related to health in the physical dimension Holstad et al21 (2011), USA 203 Motivational group interview 8 sessions Training sessions (Nutrition and stress control and health complications) Reducing risky behaviors Increasing the use of protective equipment during sex Better adherence to treatment Lowther et al22 (2015), Africa 120 Palliative care intervention including physical, emotional and spiritual 6 sessions over 4 months Routine care Reduction of pain (in the dimension of ability to share feelings and feeling valuable in life, feeling relaxed) Reducing psychological complications Reduce worries Enhancing the family's ability to plan for the future Increasing the quality of life in the mental dimension Increasing the quality of life in the physical dimension Madhombiro23 (2020), Africa 235 Motivational Interviewing blended with brief Cognitive Behavioral Therapy 8 to 10 sessions four Enhanced Usual Care sessions based on the alcohol treatment module from the World Health Organization MH GAP intervention guide) Reduction of alcohol use disorder recognition test score Decreased viral load Improving quality of life and performance Nkhoma et al24 (2015), Southeastern Africa 182 Educational intervention includes face-to-face training, pamphlet presentation, instruction booklet, and telephone follow-up A 30-minute session Pain medication control training Decreased pain intensity Better pain management Reducing pain interference with daily life. Increase patient and family knowledge about pain management Increasing the motivation of the family to manage the patient's pain Improving the quality of life Wang et al25 (2010), China 116 Home visit and telephone contact by a nurse 8 months Routine care Increased adherence to treatment Improved quality of life Reduced symptoms of depression The inclusion criteria were: Original articles with psychological interventions delivered directly by the nurse, English articles published in between 2009 and 2022, randomized clinical trial, experimental or Semi-experimental study, and access to the full-text file. Also, the exclusion criteria were: pilot study, articles published in several sources, articles in the low-quality source, letters to the editor, articles without abstract, studies that were not conducted on patients with HIV/AIDS, and a study in which psychological interventions were not implemented directly by nurses. In this review, all studies have been conducted based on 4 stages of PRISMA (Figure 1). Figure 1 Preferred reporting items systemic review (PRISMA) flowchart Eight RCTarticles, one semi-experimental and one cluster randomized clinical trial (cRCT) study were found. Four interventions were implemented in the patients’ home and five interventions in health care settings. Cochrane data extraction form was used. The two authors independently extracted the article information and agreed upon it after discussion. This form included the author’s first name, year of publication, research location, interventional measures and study findings. Initially, several sessions were held to match the rating of the two evaluators. Before the study, the kappa coefficient of agreement between the two evaluators was calculated which was 0.87. In the case of a difference of opinion between the two evaluators, the opinion of third evaluator was used. Two reviewers independently evaluated the quality of the articles and analyzed abstracts and limitations of the studies. Before reviewing the articles, the names of the authors and journals were eliminated. The article information was summarized using the modified JADAD tool. This checklist includes 8 questions assessing different sections of the article. A score of 0 to 4 was considered as weak study, 4 to 6 as moderate, and scores of ≥ 6 as strong study26 (Table 3). Then, a summary of the articles was presented in multidimensional tables. Also, the mean score and P value were evaluated and compared. Table 3 Evaluating and scoring the quality of reviewed articles Author Pointing to being random Proper randomization A blind investigation Blinding (single blind=0.5 double blind=1) Refers to trial interruptions and crashes Refer to the inclusion and exclusion criteria Reference to the method of investigating & unwanted side effects Reference to the statistical analysis method Total points Quality level Yes 1 No 0 Yes 1 No -1 Not mentioned 0 Yes 1 No 0 Yes 1 No -1 Not mentioned 0 Yes 1 No 0 Yes 1 No 0 Yes 1 No 0 Yes 1 No 0 Blank et al16 * * * 0.5 * * * * 7.5 H Blank et al17 * * * 0.5 * * * * 7.5 H Côté et al18 * * * * * * * * 4 M Eller et al19 * * * 0.5 * * * * 5.5 M Hanrahan et al20 * * * 0.5 * * * * 6.5 H Holstad et al21 * * * * * * * * 5 H Lowther et al22 * * * * * * * * 4 M Madhombiro et al 23 * * * 0.5 * * * * 5.5 M Nkhoma et al24 * * * * * * * * 8 H Wang et al25 * * * * * * * * 6 H Quality level: H=High; M=Moderate. Results Since psychological support for patients with AIDS/HIV includes any support provided to help HIV/AIDS patients improve psychological, cognitive, emotional, and behavioral well-being, the results of the studies which performed the interventions according to this definition are explained in this section. Accordingly, a total of 1975 participants started interventions and 1687 included in the analyses at the final follow-up, however, the numbers were sometimes unclearly reported. Participants included adults of any age diagnosed with HIV/AIDS. One study involved only women.21The sample size varied from 11625 to 23816,17,20 participants. Women involved 41.77% of the total population. The mean age of the participants was 42/49 years. Also, 40% of these studies were conducted in America, 40% in Africa, 10% in Europe and 10% in Asia. The articles were evaluated by the modified Jadad scale. The assessment included the effect of randomization, appropriate randomization, blind study, appropriate blinding, inclusion and exclusion criteria, adverse complications, and statistical analysis.26 Of the 10 studies extracted, 1 was a quasi-experimental study, 8 were randomized clinical trials, and 1 was cluster randomized clinical trial. Also, 40% (n=4) had moderate quality and 60% (n=6) had high quality (Table 3). Nurses’ interventions in the above mentioned studies included virtual intervention and training of emotional management skills, communication skills, motivational and problem-solving skills, home care and counseling services, symptom and side effects management services, palliative care services, case management, and motivational interviewing with cognitive behavioral therapy. The patients were followed-up by phone or in-person. The duration of Interventions varied from one training session to 12 months. The duration of follow-up varied from 8 weeks to 24 months. In six studies, the effects of the designed psychological intervention were compared with routine care group.16,17,20,22,23,25 in other studies, the effects of two designed interventions were compared compared.18,19,21,24 The study units in 9 studies were male and female patients with AIDS/HIV and in one study only women with AIDS/HIV were included. Interventions were performed on patients with different ages and stages of the disease and different underlying diseases.23 Accordingly, 3 studies reported a decrease in the severity of depression and its symptoms after the intervention.19,20,25 Also, 3 studies emphasized the improvement of patients’ adherence to the treatment regimen and the use of antiviral medications after the intervention,18,21,25 but in one of these three studies (semi-experimental study with moderate tool score),18despite the improvement in adherence to treatment, there was no significant difference between the intervention and the control groups. One article suggested that conducting a group motivational interview and a problem-solving program led to a reduction in risky behaviors in women and increased use of contraceptive methods during sexual intercourse.21 Blank et al also found that home-based care programs, counseling on medical and psychological issues and effective coordination between health care providers reduce the viral load and increase the CD4 + T cell count.16,17Concerning the effectiveness of nursing interventions on patients’ pain and managing it, some studies showed that pain intensity can be reduced by receiving palliative care and face-to-face and phone-based patient education.22,24 Moreover, the quality of life, especially in the mental and physical dimensions, significantly increased after the quantitative and qualitative improvement of the nurse’s interaction with the patient at home.17,20,25A study also showed that nurses’ use of motivational interviews along with cognitive behavioral therapy can reduce the alcohol use disorder recognition test score in patients with AIDS/HIV who consume alcohol even after six months. Also, this supportive intervention can improve the viral load, quality of life and performance of these patients.23 In addition to the main results, some studies reported secondary outcomes. Accordingly, providing patient education (motivational skills, recognizing emotional and management skills, problem-solving, and communicating with others) and virtual follow-up can improve patients’ self-efficacy and reduce perceived stress18,19,22,24; this intervention increases perceived social support and reduce the incidence of complications associated with the disease. Also, it improves a positive attitude and reduces anxiety.18 Face-to-face training and providing pamphlets and educational booklets to the patient and family, as well as telephone follow-up make the patient’s pain less interfering with daily life and increase patient and family knowledge and motivation for pain management.24 Also, providing patients with palliative care compared to routine care makes them less likely to experience psychological distress, reduces their anxiety, increases family capacity for future planning, and improves their quality of life.22 Theoretical and practical training in the management of disease symptom makes patients use positive self-care and self-management strategies to improve their mental health.19 Discussion Given the significant impact of psychological health conditions on physical health, medical care, quality of life, and risk of HIV transmission, this study was conducted to review and summarize the psychological interventions performed by nurses on patients with AIDS/HIV. To achieve this goal, eight RCTs, one semi-experimental, and one cRCT study were reviewed and evaluated. The results of this systematic review showed that interventions such as virtual or face-to-face educational programs, written information resources, palliative care, motivational interviewing, case management, motivational interviewing with cognitive-behavioral therapy, and home care with face-to-face or telephone follow-up can positively affect the quality of life, manage disease and symptoms, increase patients’ adherence to treatment, improve their mental and functional health, and reduce patients’ risky behaviors. These psychological interventions can be done by nurses independently because they have the skills to deliver them. This is an important point because it is associated with improving patient-related outcomes and achieving health policy goals, and development of nurses’ professional roles. Since this study is the first study to answer the research question “what are nursing psychological interventions in patients with AIDS/HIV and what are the consequences?”, therefore, the findings of other similar studies were used to compare the results: The results of the Borgmann and Schmidt showed that psychological interventions improved the quality of life of men with prostate cancer. They found that the patients who participated in psychological interventions had a higher quality of life at the end of the interventions. This improvement was significant in the physical dimension, but there was no evidence to support the significant effect of the interventions in the psychological dimension.27 Anderson and Ozakinci in a review study examined the impact of psychological interventions on the quality of life of patients with chronic conditions; they reported the significant effect of such interventions on the quality of life.28 However, the results of a systematic review study by Timmer et al contradict the present study. They evaluated the impact of psychological interventions on patients with inflammatory bowel disease, and reported that of the 21 clinical trials reviewed, there was no evidence to support the effectiveness of psychiatric interventions on quality of life in adult patients.29 In the study of MacKenzie et al there was no evidence to support the effectiveness of psychological interventions on the quality of life of patients undergoing strabismus surgery.30 In the present study, the impact of psychological interventions on high-risk behaviors of patients with AIDS/HIV was reported in a moderate-quality study. Risky sexual behaviors appear to be related to the psychological health of patients with AIDS/HIV. One study found that women with negative emotions were not able to cope with stress due to the desire to use condoms in sexual intercourse. There was also a significant relationship between hopelessness and loneliness and sexual risk in homosexual men.31 Another study which examined the relationship between depression and condom use among 278 Chinese women found that 62% of participants had severe depression and were less likely to use condoms consistently.32 These results emphasize that psychological interventions may reduce the risk of sexual behaviors and thus decrease HIV transmission. Meader et al also reported that women with drug abuse change their high-risk behaviors, such as sexual and injective behaviors, if they receive psychological interventions. However, there was no significant difference between women receiving psychological intervention and those receiving routine care.33 Carvalho et al also reported the limited effect of behavioral interventions on condom use in women with HIV.34 However, given that only one study examined high-risk behaviors in women, no definite conclusions can be made on the impact of psychological interventions to increase protected sex behaviors among patients. Kisely et al in a study with moderate quality found that psychological intervention, especially those with the cognitive-behavioral approach, will be effective on heart pain in people without coronary artery disease.35 Another study with good quality reported that psychological interventions along with medical treatments play an important role for patients to manage pain.36 The benefits of using psychological approaches include increasing pain self-control, improving pain management, reducing pain-related disability, and reducing emotional stress through self-monitoring, and behavioral and cognitive therapy. Implementing these interventions can help patients more control pain. Also, the skills trained through psychological interventions enable patients to actively manage their disease and provide valuable skills that patients can apply in their life.37 However, Ziehm et al in their systematic review stated that psychological interventions didn’t decrease the pain of patients undergoing open-heart surgery.38 Psychological interventions can also affect different aspects of psychological health in patients with AIDS/HIV. In addition to physical problems such as pain and lack of energy, psychological health problems such as depression and anxiety are also common among patients with AIDS/HIV.39 The prevalence of mood disorders or depressive symptoms is approximately 33% and the prevalence of anxiety is about 20%.40 Thus, psychological interventions appear to be associated with improved depression, anxiety, and distress. Consistent with this finding, a review study showed that psychological interventions leads to a short, mid, and long-term reduction in the severity of depression in diabetic patients.41 Ziehm et al also reported that psychological interventions can reduce the psychological stress experienced in patients undergoing open-heart surgery.38 In contrast to the findings of the present study, the results of a systematic review did not confirm the effectiveness of psychological interventions on the improvement of psychological health of patients with sickle cell anemia.42 Also, Natale et al.didn’t confirm the impact of psychological interventions on major depression in hemodialysis patients.43 Fisher et al also found no significant difference between the psychological intervention group and the usual care group in anxiety and depression among children with chronic and recurrent pain.44 The efficacy of psychological interventions on laboratory markers of patients with AIDS/HIV was also seen in two studies. They reported that providing medical and home care services to patients results in reduced viral load and improved CD4 + T cell. Since psychological factors (such as depressive symptoms and stress) affect immune function such as decreasing CD4 + T cell and increasing viral load,45therefore, psychological interventions with a positive impact on depression and anxiety in these patients may be influenced by laboratory results and immune function. Also, since psychological interventions lead to increased adherence to treatment, improvements in laboratory markers can be considered one of the therapeutic consequences of adherence. Locher et al also agree with the present findings regarding the impact of psychological interventions on viral load and adherence to treatment.46 Moreover, Chew et al also found that psychological interventions compared to routine care may lead to improved HbA1C outcomes in diabetic patients. Since the reviewed articles have low-quality levels, they emphasized that stronger evidence is needed to confirm the impact of psychological interventions on HbA1C levels.47 Madhombiro et al also reported similar results between the groups receiving the psychological intervention and routine care in terms of viral load and CD4 + T cell. Their finding contradicts the results of the present study.48 According to the results of the present study, psychological interventions also increase adherence to treatment. Adherence to treatment in patients with AIDS/HIV is important because it improves health and well-being and reduces the risk of infection transmission. Failure adherence to treatment is associated with psychological health problems and stressors, such as depression49 and anxiety.50 The results of a meta-analysis study showed that psychological factors are strongly associated with adherence to treatment.51 People with psychological health problems due to cognitive and behavioral problems such as fatigue, hopelessness, decreased motivation and concentration may have more difficulty in seeking treatment.52 Psychological interventions primarily focus on psychological or social factors rather than just on medical factors. Therefore, psychological therapies may be a priority for adherence to treatment. Goldbeck et al also emphasize the positive effect of psychological interventions on diet in children with cystic fibrosis.53 These differences in different studies can be attributed to differences in the power and quality of studies, and intervention protocols (provider, skill level, content, style, duration of intervention, etc.), and target groups. One of the strengths of this study is that it was conducted comprehensively because several databases and reviewed articles published from 2009 to 2022 were searched. Also, to the best of our knowledge, this systematic review is one of the first records to document the psychological interventions directly provided by nurses to patients with AIDS/HIV. So it can be a guideline for future research. Limitations of this review study include lack of access to full-text articles despite contact with authors and lack of knowledge of authors in non-English languages. The results of this study may also be influenced by the bias in the studies and a considerable degree of heterogeneity between them. In general, given the limited literature in this area, it is needed for further clinical trials on the research question. Meta-analysis studies are also recommended to gain a clear understanding of the results. Conclusions There is evidence that nurses are able to independently design and implement effective psychological interventions for patients with AIDS/HIV. This is a fundamental step for developing the professional roles of nurses and improving the care of AIDS/HIV patients. These interventions lead to the achievement of health goals in the physical, psychological, behavioral and laboratory fields. However, the clinical impact of these results is unclear because this systematic review only summarized and reported results qualitatively. Therefore, it is not possible to measure the intensity of the interventions from this systematic review. So, it is suggested to conduct meta-analysis studies to determine the intensity of the effect of each of these interventions on specific aspects of the problems of patients with AIDS/HIV. Acknowledgments The authors would like to thank all the authors of the articles used in this review study. Also, the central library authorities of Mashhad University of Medical Sciences is appreciated for providing access to related studies from various databases. Authors’ Contribution Conceptualization: Malihe Davoudi, Abbas Heydari, Zahra Sadat Manzari. Data curation: Malihe Davoudi. Formal analysis: Malihe Davoudi, Zahra Sadat Manzari. Funding acquisition: Abbas Heydari. Methodology: Malihe Davoudi, Abbas Heydari, Zahra Sadat Manzari. Project administration: Abbas Heydari. Supervision: Abbas Heydari, Zahra Sadat Manzari. Writing–original draft: Malihe Davoudi. Writing–review & editing: Abbas Heydari, Zahra Sadat Manzari. COI-statement There is no conflict of interest. Data Availability All data generated or analyzed during this study are included in this published article. Ethical Approval Ethical considerations have been observed in all stages of research, including the study and collection of data, their documentation, analysis, and publication. Funding The financial support of this study was provided by the Research and Technology Vice-Chancellor of Mashhad University of Medical Sciences. Research Highlights What is the current knowledge? HIV/AIDS patients experience significant medically and psychologically suffering One of the traditional responsibilities and roles of nurses is psychological support for these patients. We do not know what interventions nurses can take directly and independently to provide psychological support to these patients and what are the consequences. What is new here? Virtual and face-to-face training programs, written information resources, palliative care, motivational interviewing, case management, visits and home care services along with face-to-face and telephone follow-up can be done directly by nurses. Psychological interventions have a significant positive effect on quality of life and management of high-risk behaviors, disease management, and adherence to treatment, immune system function and psychological health in patients with HIV/AIDS. ==== Refs References 1 World Health Organization (WHO). Global HIV & AIDS Statistics—Fact Sheet. 2019. Available from: https://www.unaids.org/en/resources/fact-sheet. 2 Paul AA, Premraj FC. Psychosocial problems and its impact faced by the HIV/AIDS infected patients. IOSR J Humanit Soc Sci 2008; Available from: https://www.iosrjournals.org/iosr-jhss/papers/Conf.17004/Volume-3/7.%2040-45.pdf. 3 Pahuja M, Merlin J, Selwyn PA, Sullivan M. Issues in HIV/AIDS in Adults in Palliative Care. 2021. Available from: https://www.uptodate.com/contents/issues-in-hiv-aids-in-adults-in-palliative-care#:~:text=Associated%20conditions%20%E2%80%93%20Co%2Doccurring%20pain,and%20substance%20use%20%5B71%2C111%2C112%5D. 4 Owe-Larsson B Säll L Salamon E Allgulander C HIV infection and psychiatric illness Afr J Psychiatry (Johannesbg) 2009 12 2 115 28 10.4314/ajpsy.v12i2.43729 19582313 5 Yi S Chhoun P Suong S Thin K Brody C Tuot S AIDS-related stigma and mental disorders among people living with HIV: a cross-sectional study in Cambodia PLoS One 2015 10 3 e0121461 10.1371/journal.pone.0121461 25806534 6 Buckingham E Schrage E Cournos F Why the Treatment of mental disorders is an important component of HIV prevention among people who inject drugs Adv Prev Med 2013 2013 690386 10.1155/2013/690386 23401785 7 Hughes E Bassi S Gilbody S Bland M Martin F Prevalence of HIV, hepatitis B, and hepatitis C in people with severe mental illness: a systematic review and meta-analysis Lancet Psychiatry 2016 3 1 40 8 10.1016/s2215-0366(15)00357-0 26620388 8 Chuah FLH Haldane VE Cervero-Liceras F Ong SE Sigfrid LA Murphy G Interventions and approaches to integrating HIV and mental health services: a systematic review Health Policy Plan 2017 32 Suppl 4 iv27 iv47 10.1093/heapol/czw169 29106512 9 Dumitru G Irwin K Tailor A Updated federal recommendations for HIV prevention with adults and adolescents with HIV in The United States: the pivotal role of nurses J Assoc Nurses AIDS Care 2017 28 1 8 18 10.1016/j.jana.2016.09.011 27839911 10 Tunnicliff SA Piercy H Bowman CA Hughes C Goyder EC The contribution of the HIV specialist nurse to HIV care: a scoping review J Clin Nurs 2013 22 23-24 3349 60 10.1111/jocn.12369 24131477 11 Wood JH Chambers M White SJ Nurses’ knowledge of and attitude to electroconvulsive therapy J ect 2007 23 4 251 4 10.1097/yct.0b013e31813e0692 18090698 12 Papathanassoglou ED Psychological support and outcomes for ICU patients Nurs Crit Care 2010 15 3 118 28 10.1111/j.1478-5153.2009.00383.x 20500650 13 The Australasian Society for HIV, Viral Hepatitis and Sexual Health Medicine (ASHM). Standards for Psychological Support for Adults with HIV. 2010. Available from: https://www.bhiva.org/file/BbShtfyMFNKCz/Standards_for_psychological_support_for_adults_living_with_HIV.pdf. 14 Cumpston M Li T Page MJ Chandler J Welch VA Higgins JP Updated guidance for trusted systematic reviews: a new edition of the Cochrane Handbook for Systematic Reviews of Interventions Cochrane Database Syst Rev 2019 10 ED000142 10.1002/14651858.ed000142 31643080 15 Moher D Liberati A Tetzlaff J Altman DG Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement PLoS Med 2009 6 7 e1000097 10.1371/journal.pmed.1000097 19621072 16 Blank MB Hanrahan NP Fishbein M Wu ES Tennille JA Ten Have TR A randomized trial of a nursing intervention for HIV disease management among persons with serious mental illness Psychiatr Serv 2011 62 11 1318 24 10.1176/ps.62.11.pss6211_1318 22211211 17 Blank MB Hennessy M Eisenberg MM Increasing quality of life and reducing HIV burden: the PATH + intervention AIDS Behav 2014 18 4 716 25 10.1007/s10461-013-0606-x 24000053 18 Côté J Godin G Ramirez-Garcia P Rouleau G Bourbonnais A Guéhéneuc YG Virtual intervention to support self-management of antiretroviral therapy among people living with HIV J Med Internet Res 2015 17 1 e6 10.2196/jmir.3264 25563775 19 Eller LS Kirksey KM Nicholas PK Corless IB Holzemer WL Wantland DJ A randomized controlled trial of an HIV/AIDS Symptom Management Manual for depressive symptoms AIDS Care 2013 25 4 391 9 10.1080/09540121.2012.712662 22880943 20 Hanrahan NP Wu E Kelly D Aiken LH Blank MB Randomized Clinical trial of the effectiveness of a home-based advanced practice psychiatric nurse intervention: outcomes for individuals with serious mental illness and HIV Nurs Res Pract 2011 2011 840248 10.1155/2011/840248 21935499 21 Holstad MM DiIorio C Kelley ME Resnicow K Sharma S Group motivational interviewing to promote adherence to antiretroviral medications and risk reduction behaviors in HIV infected women AIDS Behav 2011 15 5 885 96 10.1007/s10461-010-9865-y 21165692 22 Lowther K Selman L Simms V Gikaara N Ahmed A Ali Z Nurse-led palliative care for HIV-positive patients taking antiretroviral therapy in Kenya: a randomised controlled trial Lancet HIV 2015 2 8 e328 34 10.1016/s2352-3018(15)00111-3 26423375 23 Madhombiro M Kidd M Dube B Effectiveness of a psychological intervention delivered by general nurses for alcohol use disorders in people living with HIV in Zimbabwe: a cluster randomized controlled trial J Int AIDS Soc 2020 23 12 e25641 10.1002/jia2.25641 33314786 24 Nkhoma K, Seymour J, Arthur A. An educational intervention to reduce pain and improve pain management for malawian people living with HIV/AIDS and their family carers: a randomized controlled trial. J Pain Symptom Manage 2015;50(1):80-90.e4. 10.1016/j.jpainsymman.2015.01.011 25 Wang H Zhou J Huang L Li X Fennie KP Williams AB Effects of nurse-delivered home visits combined with telephone calls on medication adherence and quality of life in HIV-infected heroin users in Hunan of China J Clin Nurs 2010 19 3-4 380 8 10.1111/j.1365-2702.2009.03048.x 20500277 26 Lempesi E Koletsi D Fleming PS Pandis N The reporting quality of randomized controlled trials in orthodontics J Evid Based Dent Pract 2014 14 2 46 52 10.1016/j.jebdp.2013.12.001 24913524 27 Borgmann H Schmidt S [Psychosocial interventions for men with prostate cancer] Urologe A 2015 54 6 863 6 10.1007/s00120-015-3858-4 26081817 28 Anderson N Ozakinci G Effectiveness of psychological interventions to improve quality of life in people with long-term conditions: rapid systematic review of randomised controlled trials BMC Psychol 2018 6 1 11 10.1186/s40359-018-0225-4 29587884 29 Timmer A Preiss JC Motschall E Rücker G Jantschek G Moser G Psychological interventions for treatment of inflammatory bowel disease Cochrane Database Syst Rev 2011 2 CD006913 10.1002/14651858.CD006913.pub2 21328288 30 MacKenzie K Hancox J McBain H Ezra DG Adams G Newman S Psychosocial interventions for improving quality of life outcomes in adults undergoing strabismus surgery Cochrane Database Syst Rev 2016 2016 5 CD010092 10.1002/14651858.CD010092.pub4 27171652 31 Su X Zhou AN Li J Shi LE Huan X Yan H Depression, loneliness, and sexual risk-taking among HIV-negative/unknown men who have sex with men in China Arch Sex Behav 2018 47 7 1959 68 10.1007/s10508-017-1061-y 29147806 32 Hong Y Li X Fang X Zhao R Depressive symptoms and condom use with clients among female sex workers in China Sex Health 2007 4 2 99 104 10.1071/sh06063 17524287 33 Meader N Li R Des Jarlais DC Pilling S Psychosocial interventions for reducing injection and sexual risk behaviour for preventing HIV in drug users Cochrane Database Syst Rev 2010 2010 1 CD007192 10.1002/14651858.CD007192.pub2 20091623 34 Carvalho FT Gonçalves TR Faria ER Shoveller JA Piccinini CA Ramos MC Behavioral interventions to promote condom use among women living with HIV Cochrane Database Syst Rev 2011 9 CD007844 10.1002/14651858.CD007844.pub2 21901711 35 Kisely SR Campbell LA Skerritt P Yelland MJ Psychological interventions for symptomatic management of non-specific chest pain in patients with normal coronary anatomy Cochrane Database Syst Rev 2010 1 CD004101 10.1002/14651858.CD004101.pub3 20091559 36 Jensen MP Psychosocial approaches to pain management: an organizational framework Pain 2011 152 4 717 25 10.1016/j.pain.2010.09.002 21168972 37 Roditi D Robinson ME The role of psychological interventions in the management of patients with chronic pain Psychol Res Behav Manag 2011 4 41 9 10.2147/prbm.s15375 22114534 38 Ziehm S Rosendahl J Barth J Strauss BM Mehnert A Koranyi S Psychological interventions for acute pain after open heart surgery Cochrane Database Syst Rev 2017 7 7 CD009984 10.1002/14651858.CD009984.pub3 28701028 39 Morrison MF Petitto JM Ten Have T Gettes DR Chiappini MS Weber AL Depressive and anxiety disorders in women with HIV infection Am J Psychiatry 2002 159 5 789 96 10.1176/appi.ajp.159.5.789 11986133 40 Evans DL Charney DS Lewis L Golden RN Gorman JM Krishnan KR Mood disorders in the medically ill: scientific review and recommendations Biol Psychiatry 2005 58 3 175 89 10.1016/j.biopsych.2005.05.001 16084838 41 Baumeister H Hutter N Bengel J Psychological and pharmacological interventions for depression in patients with diabetes mellitus: an abridged Cochrane review Diabet Med 2014 31 7 773 86 10.1111/dme.12452 24673571 42 Anie KA Psychological complications in sickle cell disease Br J Haematol 2005 129 6 723 9 10.1111/j.1365-2141.2005.05500.x 15952997 43 Natale P Palmer SC Ruospo M Saglimbene VM Rabindranath KS Strippoli GF Psychosocial interventions for preventing and treating depression in dialysis patients Cochrane Database Syst Rev 2019 12 12 CD004542 10.1002/14651858.CD004542.pub3 31789430 44 Fisher E Law E Dudeney J Palermo TM Stewart G Eccleston C Psychological therapies for the management of chronic and recurrent pain in children and adolescents Cochrane Database Syst Rev 2018 9 9 CD003968 10.1002/14651858.CD003968.pub5 30270423 45 Ironson G O’Cleirigh C Fletcher MA Laurenceau JP Balbin E Klimas N Psychosocial factors predict CD4 and viral load change in men and women with human immunodeficiency virus in the era of highly active antiretroviral treatment Psychosom Med 2005 67 6 1013 21 10.1097/01.psy.0000188569.58998.c8 16314608 46 Locher C Messerli M Gaab J Gerger H Long-term effects of psychological interventions to improve adherence to antiretroviral treatment in HIV-infected persons: a systematic review and meta-analysis AIDS Patient Care STDS 2019 33 3 131 44 10.1089/apc.2018.0164 30844307 47 Chew BH Vos RC Metzendorf MI Scholten RJ Rutten GE Psychological interventions for diabetes-related distress in adults with type 2 diabetes mellitus Cochrane Database Syst Rev 2017 9 9 CD011469 10.1002/14651858.CD011469.pub2 28954185 48 Madhombiro M Musekiwa A January J Chingono A Abas M Seedat S Psychological interventions for alcohol use disorders in people living with HIV/AIDS: a systematic review Syst Rev 2019 8 1 244 10.1186/s13643-019-1176-4 31661030 49 Gonzalez JS Batchelder AW Psaros C Safren SA Depression and HIV/AIDS treatment nonadherence: a review and meta-analysis J Acquir Immune Defic Syndr 2011 58 2 181 7 10.1097/QAI.0b013e31822d490a 21857529 50 Safren SA O’Cleirigh C Tan JY Raminani SR Reilly LC Otto MW A randomized controlled trial of cognitive behavioral therapy for adherence and depression (CBT-AD) in HIV-infected individuals Health Psychol 2009 28 1 1 10 10.1037/a0012715 19210012 51 Langebeek N Gisolf EH Reiss P Vervoort SC Hafsteinsdóttir TB Richter C Predictors and correlates of adherence to combination antiretroviral therapy (ART) for chronic HIV infection: a meta-analysis BMC Med 2014 12 142 10.1186/preaccept-1453408941291432 25145556 52 Wagner GJ Goggin K Remien RH Rosen MI Simoni J Bangsberg DR A closer look at depression and its relationship to HIV antiretroviral adherence Ann Behav Med 2011 42 3 352 60 10.1007/s12160-011-9295-8 21818528 53 Goldbeck L Fidika A Herle M Quittner AL Psychological interventions for individuals with cystic fibrosis and their families Cochrane Database Syst Rev 2014 2014 6 CD003148 10.1002/14651858.CD003148.pub3 24941199
PMC010xxxxxx/PMC10352635.txt
==== Front J Caring Sci J Caring Sci J Caring Sci TBZMED Journal of Caring Sciences 2251-9920 2251-9920 Tabriz University of Medical Sciences 10.34172/jcs.2023.30626 Original Article Occupational Challenges of Intensive Care Nurses During the COVID-19 Pandemic: A Qualitative Study https://orcid.org/0000-0002-4791-7866 Yousefi Mahdi 1 2 https://orcid.org/0000-0003-4968-6448 Ebrahimi Zahra 3 https://orcid.org/0000-0002-5486-7203 Bakhshi Mahmoud 4 5 * https://orcid.org/0000-0002-8292-7381 Fazaeli Somayeh 6 1Social Determinant of Health Research Center, Mashhad University of Medical Sciences, Mashhad, Iran 2Department of Health Economics and Management, School of Health, Mashhad University of Medical Sciences, Mashhad, Iran 3Department of Management, North Tehran Branch, Islamic Azad University, Tehran, Iran 4Department of Medical-Surgical Nursing, School of Nursing and Midwifery, Mashhad University of Medical Sciences, Mashhad, Iran 5Nursing and Midwifery Care Research Center, Mashhad University of Medical Sciences, Mashhad, Iran 6Department of Medical Records and Health Information Technology, School of Paramedical Sciences, Mashhad University of Medical Sciences, Mashhad, Iran * Mahmoud Bakhshi [email protected] 6 2023 12 3 2023 12 2 110115 20 1 2022 21 4 2022 © 2023 The Author(s). 2023 https://creativecommons.org/licenses/by-nc/4.0/ This work is published by Journal of Caring Sciences as an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by-nc/4.0/). Non-commercial uses of the work are permitted, provided the original work is properly cited. http://jcs.tbzmed.ac.ir Introduction: The coronavirus disease 2019 (COVID-19) has been spreading rapidly as a pandemic and posed numerous challenges to healthcare workers (HCWs), especially nurses. This study aimed to investigate the occupational challenges experienced by intensive care unit (ICU) nurses in caring for patients with COVID-19. Methods: This qualitative study was conducted using a conventional content analysis method in September and October 2020. The study environment was the ICU wards dedicated to the patients of COVID-19 in a large hospital in east of Iran. The participants were selected by purposeful sampling method, and data were collected using semi-structured interviews with 17 nurses working in the COVID-19 ICUs. Data analysis was done with MAXQDA 2020. Results: The data analysis led to the extraction of 6 main categories and 17 sub-categories. The main challenges included "payment system", "human resource management", "consumable resource supply", "psychological and ethical distress", "personal or family problems", and "staff motivation and welfare issues". Conclusion: Considering the key and important role of nurses in the healthcare system, particularly during the COVID-19 pandemic, it is necessary to increase their motivation by applying a fair and non-discriminatory payment system and paying special attention to psychological issues. Also, managerial support and provision of required facilities and manpower have a significant impact on reducing their occupational challenges in caring for patients with COVID-19. Qualitative study Nurses Intensive care COVID-19 ==== Body pmcIntroduction The COVID-19 disease was first reported in China, but it had spread widely throughout the world.1 The high percentage of deaths caused by COVID-19 had faced many countries with a huge health challenge.2 Despite preventive measures, as well as personal and public precautions, a large number of people are still infected with COVID-19 all around the world.3 Hospital admissions and hospitalizations of patients have increased following the COVID-19 pandemic.4 The prognosis of patients with coronavirus infection are very variable. The rate of intensive care unit (ICU) admission among patients with coronavirus infection varies from 3% to 100%.5 Therefore, ICU admission play an important role in the care and treatment of patients with COVID-19.6 Nurses are the frontline healthcare workers (HCWs) in ICU and have a significant professional responsibility in caring of critical patients.7 The emergence of COVID-19 has put unprecedented pressure on the healthcare system and HCWs. It has potentially affected nurses’ performance and mental health and even influenced their lives.8Nurses spend about 86% of their time in direct contact with ICU patients and are faced with a variety of challenges that impose severe physical and mental strains on them during the provision of care for patients with COVID-19.9,10 Reports indicate that nurses experience the highest levels of anxiety among the HCWs.11 The main source of nurses’ anxiety during the COVID-19 epidemic is found the fear of their own infection and also transmitting the disease to others.9 Other identified causes include the lack of protection facilities, lack of access to diagnostic tests for screening, fear of transmitting the virus to others at work, feeling of insufficient support, and being deployed in an unfamiliar ward or unit.12 Maben and Bridges pointed out that the use of personal protective equipment during long shifts leads to severe fatigue of nurses and is an important communication barrier for proper communication between nurses and patients.13 These factors lead to nurses’ depression, low mood, absenteeism, apathy, and poor performance, which will eventually result in patient dissatisfaction.14 Therefore, due to the rapid spread of COVID-19 disease, the largest hospital in eastern Iran, was designated as the main referral hospital for these patients. There were 1200 nurses working in this hospital, 82% of which have been involved in caring for patients with COVID-19. Considering the increasing need for ICU beds to admit and care for patients with COVID-19 or other possible emerging disease, it seems necessary to identify the challenges and concerns of nurses working in these units. Therefore, this qualitative study aimed to explore the occupational challenges experienced by ICU nurses in caring for patients with COVID-19. Materials and Methods This qualitative study was conducted based on a conventional content analysis approach. The research environments were two inpatient departments with a capacity of 300 beds, which were under the supervision of a large hospital in east of Iran. In addition to internal and infectious ICUs, the surgical and open-heart surgery ICUs were also dedicated to care of COVID-19 patients. These centers were equipped and used for hospitalization of COVID-19 patients after complete evacuation. The participants included the nurses who worked in ICUs dedicated to COVID-19 patients. Seventeen nurses were selected using purposive sampling and were interviewed. The inclusion criteria were willingness to participate in the study, having at least 3 months of experience working in the ICUs of COVID-19 patients, and having the ability to convey rich experiences regarding to care of COVID-19 patients. Data were collected using in-depth semi-structured interviews during September and October 2020. Due to the dangerous nature of the coronavirus and to maintain the researchers’ and nurses’ health status, telephone interviews were done. The researcher after explaining the study objectives, determined the appropriate time to conduct the interview with the participants’ agreement. All interviews were recorded electronically using a digital voice recorder. In first, a general open-ended question was asked from the interviewees: “Please tell me about your experiences of theoccupational challenges related to caring for COVID-19 patients in ICU?”. The interviews continued according to the responses provided by the participants, and then deeper questions were asked about such issues as workplace conditions, resource supply conditions, payment system, and communication issues. The average duration of the interviews was 20 min and ranged from 15 to 35 minutes. The interviews and data collection continued until saturation was reached, when no more new data was obtained from the interviews. The data analysis was done concurrently with data collection using MAXQDA 2020, based on inductive qualitative content analysis as mentioned by Graneheim & Lundman.15 Initially, the interviews were transcribed and then reviewed several times meticulously until a general understanding was obtained from the interviews. In the first stage, the descriptive codes were identified and agreed upon them by the researchers through discussion. In the second step, the classification and labeling of the data was done by identifying the appropriate sentences and comments. In the third stage, the deductive analysis was carried out by reading and thoughtful re-reading of the text to integrate the statements. Finally, the precise perspectives of interviewees and the importance of evidence were achieved by organizing the integrated statements for addressing the purpose of the study. Four criteria of credibility, confirmability, dependability, and transferability by Lincoln and Guba were used to ensure data trustworthiness.16 To this end, the following strategies were used: (a) purposeful sampling of participants; (b) Interview with nurses with different levels of management and work shifts; (c) Considering enough time to conduct interviews; (d) Continuous review and comparison of data and concepts in terms of similarities and differences; (e) member checking; and (f) Providing detailed data analysis and deep and rich description of research concepts. The final analysis was reviewed by team members as well as by other researcher who was familiar with the research methodology. Results In this study, 17 participants were interviewed, of which 76.5% were female. The age range of nurses was 30-42 years with an average age of 37 years. Also, their mean work experience was 12 years. The analysis of data led to the identification of six categories as occupationalchallenges of ICU nurses in caring for COVID-19 patients. The main categories included payment system challenges; personal and familial challenges; human resource management challenges; consumable resource supply deficiency; staff motivation and welfare issues; and psychological and ethical distress (Table 1). Table 1 Main and subcategories extracted from the data Main category Sub-categories Payment system challenges Payment irrespective of performance Dissatisfaction with the payments Dissatisfaction with the payment mechanism Personal and familial challenges Fear of getting infected with COVID-19 Problems caused by family members being infected with COVID-19 Tension in family relationships Human resource management challenges high volume workload Compulsory employment in Corona sectors Inadequate efficiency of novice nurses Inadequate preparedness for crisis management Consumable resource supply deficiency Problem in drug supply Medical supplies shortage Inadequate quality of consumables Staff motivation and welfare issues Lack of support from managers and officials Inadequate amenities psychological and ethical distress job stress and emotional distress related to COVID-19 Ethical conflicts related to work environment problem Payment System Challenges This topic refers to how to pay the salaries and benefits of the nurses who take care of the COVID-19 patients in the ICU. One of the main complaints of nurses was financial issues that were mentioned in the form of dissatisfaction about evaluation and performance-based payment, dissatisfaction with discrimination in the payment, delays in the payments, non-priority of payments to nurses working in COVID-19 wards, and the need for more financial support. The statements of the interviewees in this regard were as follows: Nurse:” Our biggest need as nurses is related to financial issues. They don’t pay the special corona payment on time. It paid with a delay, and our per-cases aren’t paid since last year”. (Participant No. 2). Nurse: “I see at least difference between the salaries of HCWs working in coronavirus wards in comparison with those working in other parts of the hospital”. (Participant No. 1). Personal and Familial Challenges This category included the fear of contracting the infection (for self and family members), disruption in friendly communication with family and colleagues. The statements of some participants regarding this concept are given below: Nurse: “The early days of the crisis, it had a great effect. We were so scared that we wouldn’t have contact with our children. It was very difficult for us. I personally hadn’t hugged my children for almost 2 months. I could hardly see my mother and we were really troubled during this time” (Participant No. 1). Nurse: “This disease really caused a crisis for me and my family; my child got sick on one side and I was depressed on the other side. Family members also made me upset because they used to say that you made the child ill” (Participant No. 15). Human Resource Management Challenges Some sub-categories were high workload of nursing staff, imposing unexpected extra shifts, forcing some nurses to work in the COVID-19 ICU, and lacking sufficient preparedness to deal with the crisis. The reasons for the high workload included the high ratio of patients to nurses, employment of inexperienced and novice staff, tightness of shift work schedules, absence and sick leave of a colleague, and training the new staff. Some of the nurses’ statements in this category were as follows: Nurse: “The training of personnel who came from other wards was very hard. Many of them even didn’t know how to work with a common pumps in the ward …. They didn’t know the ventilator alarms because they hadn’t worked with a ventilator in the past”. (Participant No. 17). Nurse: “The work schedule that was arranged for us was dense. It has been 4 or 5 months that none of ICU nurses had a request plan. If I asked for 4 or 5 days off, it wouldn’t be agreed and it was said that we are currently in a critical situation. If we didn’t come to work for even one day, we would be considered absent.” (Participant No. 9). Consumable Resource Supply Deficiency The basic concepts in this category included shortage of drugs especially expensive drugs, improper management of medication supply in the wards and hospital, lack of personal protective equipment (e.g., N95 mask, hand rub solution, gloves), lack of equipment (e.g., non-invasive ventilation mask), poor quality of personal protective equipment, and use of reusable clothing. Nursestated about the lack of consumption: “…There were scarcities in supplies and equipment. If we wanted masks, there weren’t enough. We wanted to change our gloves from patient to patient, sometimes it really was impossible because they didn’t give us more than 4 to 5 pairs of gloves” (Participant No. 14). Sometimes the shortage of medicine and consumables were related to the lack of coordination between the relevant authorities. In this regard, one other nursesaid: “To take a new medicine like Remdesivir, it was needed that one physician coordinate with another physician or pharmacy manager... and then we called the pharmacy but they said it hadn’t been coordinated, and it must be informed in written form”. (Participant No. 5). Staff Motivation and Welfare Issues Among the main concepts mentioned in this category were inadequate support and not attention to HCWs requests by ward and hospital officials, inappropriate reactions to professional protests, allocation of incentive items with low practicality, lack of welfare facilities. The use of personal protective equipment had a negative effect on the performance of nursing staff. Participants commented on the mentioned challenge as follows: Nurse: “With the exception of the head nurse, we did not see anyone as a patron. We felt as if we had fallen into a pit. I wish we could say, for example, the director of the hospital or the director of a part of the hospital is next to me; we really didn’t have that support” (Participant No. 8). Nurse: “To start work, for example, the nurses were protesting to the conditions of the changing room and bedroom. They were protesting for improper condition of sleeping places. Because the number of nurses added to us was much, our changing room wasn’t large enough” (Participant No. 6). Psychological and Ethical Distress The basic concepts extracted in this category included the concern of additional costs to the patient’s companion and family, pressure and stress on the nurse regarding the provision of medicine and equipment out of the hospital by the patient’s companion, mental and emotional disorders due to the patient’s death or colleague’s infection, increased stress due to high mortality, concern with the transmission of the disease to companion and others, insufficient insurance support to pay for expensive medications, and mental and physical fatigue caused by high-volume workload.. The high mortality percentage of patients and the infection and death of colleagues had a negative mental effect. In this regard, Nurse commented, “…The hardest job was managing our emotions against the fact that we saw the deaths of 7 or 8 of our colleagues. We cared for many of our colleagues. This emotional management was really hard for us” (Participant No. 7). Nurse: “Some drugs weren’t available for a while, such as Ribavirin, and we had to tell the patient’s companion that it was very difficult to get. Many of the companions had to buy it from outside the hospital, but they didn’t have the money and thought that it would be solved by transferring it to us…” (Participant No. 6). Discussion The findings of this study showed the occupational challenges of nurses working in COVID-19 ICUs. The major challenges included payment system, human resource management, consumable resource supply, psychological and ethical distress, personal or family problems, and staff motivation and welfare issues. Nurses have a great responsibility in ICU. Due to the expansion of their health care role, workload has increased in pandemic conditions and more attention to them is necessary.7 One of the important occupational challenges that was identified in the present study was the problems related to the payment system for ICU nurses working with COVID-19 patients. Consistent with the results of this study, it was also revealed that non-timely payments, discrimination in incentive payments, dissatisfaction with the amount and mechanism of payments were among the issues raised in other studies.17,18 The COVID-19 pandemic has caused a significant decrease on hospital incomes due to the cancellation of elective surgeries, provision of personal protective equipment, provision of welfare facilities for HCWs, and the necessity of paying attention to various aspects of public health and public education. This disruption in the process of production and consumption of financial resources in medical centers poses problems for the health system. Financial issues may negatively affect the performance of HCWs. Due to the fact that COVID-19 reference hospitals face financial constraints, this lack of funding can have a significant impact on staff’s payments and lead to their dissatisfaction.19 Although the moral aspect of caring causes HCWs to continue serving patients and clients despite financial pressures, the effect of staff’s financial satisfaction on the quality of care should not be overlooked. The peculiarity of care marked by a great number of patients in isolation led to an exponential increase in the nurses’ workload. They had to deal with providing care in a unique way, however, without previous specific experience. For instance, since the number of professionals in the room had to be reduced, the nurses had to stay inside longer and carry out a greater number of interventions that, in other situations, were carried out by other professionals.7 In such a situation, motivational rewards are materially one of the solutions used in the hospital to encourage and motivate nurses. Nevertheless, from the nurses’ point of view, these rewards have not been distributed fairly in some cases. The findings of this study showed that nurses experienced high levels of stress and anxiety when caring for COVID-19 patients due to the unknown nature of this disease and its high mortality rate. The other concerns of nurses included the fear of transmitting the disease to the colleagues and family members due to contracting COVID-19 disease. Huang et al conducted a study, in which they interviewed nurses working in the COVID-19 emergency room. Accordingly, one of the biggest fears of the nurses was found to be the risk of being infected by themselves and their families.20 Since nurses’ physical and mental health is directly related to the quality of their performance regarding patient care, it is necessary to try to reduce the stress of these members of society, especially those working in the ICU. Consequently, it seems essential to provide adequate training in psychological skills to deal with anxiety and other emotional problems, required financial and human resources, regular counseling services, and attention to nurses’ problems related to the COVID-19 crisis to increase nurses’ level of professional commitment and help them overcome interpersonal conflicts.21 Certainly, a crisis such as the current one (caused by COVID-19) will face the healthcare systems with a challenge in supplying consumable resources. In order to comply with the principles of personal protection while caring for the patient, it is necessary for the nurse to ensure the quality and provision of personal protection facilities and equipment.22 According to the results of the present study, one of the challenges was related to the supply of consumable resources. The shortage of supplies and its effect on the performance of nurses have also been mentioned in previous studies.23,24 Participants reported lack of personal protective equipment (e.g., N95 mask, gloves, and coverall). Appropriate use and supply chain management play a key role in optimizing and availability of personal protective equipment. Preparation of clear protocols and evidence-based guidelines on the use of personal protective equipment, ensuring adequate supply and checking their quality are important requirements in crises and pandemics.25 Nurses pointed that the personal protective equipment has inadequate quality and they felt tired after using it for a long time while taking care of the patients. This finding was consistent with the results of studies conducted during the outbreak of the Middle East respiratory syndrome (MERS) and Ebola.26,27 Finally, the last category of data analysis pointed to staff motivation and welfare issues. In this study, nursing staff considered the visit and the presence of the officials in the ICU as a factor of their encouragement. Also, they expressed dissatisfaction with the lack of a suitable resting place, and other amenities. Lack of proper resting place and adequate amenities and support have previously been reported by nurses during the prevalence of H1N1 influenza and MERS.28 Based on the results of a study conducted in China, the lack of support for HCWs was identified as a source of stress for nurses during the outbreak of COVID-19 disease.29 According to the findings of some studies, listening to employees’ problems is found to be a sign of supporting emotionally, valuing them, and increasing motivation and productive activity. It has been reported that in addition to financial and incentive payments, managers’ empathetic and sincere communication with nurses, showing trust, developing open climate, providing positive feedback, and rational supporting of nurses in their conflicts with others are effective in motivating and encouraging nurses to act voluntarily, especially in critical situations.30,31 Due to the heavy burden of COVID-19 disease on the nursing community, the nurses welcomed the interviews and provided the necessary cooperation. This led to a deep understanding of their work experiences, and therefore, achieving reliable and comprehensive data. Due to the nature of qualitative studies, the sample size of this study was limited. Due to the prevalence of coronavirus at the time of the study, it was not possible to conduct face-to-face interviews. In order to respect the principles of infection control and preserve the researcher and participant’s health, the interviews were conducted over the phone. Although in qualitative studies, the researcher bias may influence the findings. However, in the present study, analysis was confirmed by another researcher who was not involved in the initial data analysis. Conclusion Based on the results of the present study, exposure to the COVID-19 pandemic and similar crises posed several occupational challenges for ICU nurses, which are mainly related to the payment system, resource management, motivation, and their psycho-emotional issues. Therefore, some solutions were suggested to address these challenges and problems are apply the fair and just payment systems; increase motivation through incentive payments; periodic screening; proper management of human resources and consumption; provide required facilities and resources; pay attention to the psychological needs, and managers’ and officials’ support. Acknowledgments The researchers would like to appreciate the Vice Chancellor for Research of Mashhad University of Medical Sciences for the financial and moral support of this research with the registered number of 990014. Moreover, they express their gratitude to all participants in this study. Authors’ Contribution Conceptualization: Mahmoud Bakhshi, Zahra Ebrahimi. Data curation: Mahdi Yousefi, Zahra Ebrahimi. Formal analysis: Mahmoud Bakhshi, Mahdi Yousefi, Zahra Ebrahimi. Methodology: Mahmoud Bakhshi. Project administration: Mahdi Yousefi. Supervision: Mahdi Yousefi, Mahmoud Bakhshi. Writing–original draft: Zahra Ebrahimi,Mahmoud Bakhshi, Mahdi Yousefi. Writing–review & editing: Mahmoud Bakhshi, Mahdi Yousefi, Zahra Ebrahimi, Somayeh Fazaeli. COI-statement The authors have no conflicts of interest to declare. Data Availability The datasets are available from the corresponding author on reasonable request. Ethical Approval This study was approved by the Ethics Committee of Mashhad University of Medical Sciences, Mashhad, Iran. (Ethical code: IRMUMSREC.1399.021). Informed consent was obtained from all subjects, and they were informed of the right to leave the study at any time. Moreover, all participants were assured of anonymity and confidentiality in this study. Funding Financial resources were provided by Mashhad University of Medical Sciences. Research Highlights What is the current knowledge? COVID-19 is one of the diseases that causes a lot of stress to intensive care nurses. Efforts are being made to understand the different aspects of the impact of this disease on caregivers. What is new here? The payment system for ICU nurses should be fair and should not be affected by the decrease in hospital income during the COVID-19 pandemic and similar crises. Improper management and lack of resources have a negative effect on the mental and physical conditions and performance of ICU nurses. It is necessary to pay attention to the motivational and welfare issues of ICU nurses during the COVID-19 pandemic and similar crises. ==== Refs References 1 Zhu H Wei L Niu P The novel coronavirus outbreak in Wuhan, China Glob Health Res Policy 2020 5 6 10.1186/s41256-020-00135-6 32226823 2 Torales J O’Higgins M Castaldelli-Maia JM Ventriglio A The outbreak of COVID-19 coronavirus and its impact on global mental health Int J Soc Psychiatry 2020 66 4 317 20 10.1177/0020764020915212 32233719 3 Aghajani M Taghadosi M Mirbagher Ajorpaz N Intuitive decision-making by Iranian nurses of patients with COVID-19: a qualitative study J Caring Sci 2022 11 3 154 62 10.34172/jcs.2022.04 36247041 4 Birkmeyer JD Barnato A Birkmeyer N Bessler R Skinner J The impact of the COVID-19 pandemic on hospital admissions in the United States Health Aff (Millwood) 2020 39 11 2010 7 10.1377/hlthaff.2020.00980 32970495 5 Abate SM Ahmed Ali S Mantfardo B Basu B Rate of intensive care unit admission and outcomes among patients with coronavirus: a systematic review and meta-analysis PLoS One 2020 15 7 e0235653 10.1371/journal.pone.0235653 32649661 6 Sadeghi A Eslami P Dooghaie Moghadam A Pirsalehi A Shojaee S Vahidi M COVID-19 and ICU admission associated predictive factors in Iranian patients Caspian J Intern Med 2020 11 Suppl 1 512 9 10.22088/cjim.11.0.512 33425268 7 Fernández-Castillo RJ González-Caro MD Fernández-García E Porcel-Gálvez AM Garnacho-Montero J Intensive care nurses’ experiences during the COVID-19 pandemic: a qualitative study Nurs Crit Care 2021 26 5 397 406 10.1111/nicc.12589 33401340 8 Labrague LJ De Los Santos JAA COVID-19 anxiety among front-line nurses: predictive role of organisational support, personal resilience and social support J Nurs Manag 2020 28 7 1653 61 10.1111/jonm.13121 32770780 9 Butler R, Monsalve M, Thomas GW, Herman T, Segre AM, Polgreen PM, et al. Estimating time physicians and other health care workers spend with patients in an intensive care unit using a sensor network. Am J Med 2018; 131(8): 972.e9-972.e15. 10.1016/j.amjmed.2018.03.015 10 Mo Y Deng L Zhang L Lang Q Liao C Wang N Work stress among Chinese nurses to support Wuhan in fighting against COVID-19 epidemic J Nurs Manag 2020 28 5 1002 9 10.1111/jonm.13014 32255222 11 Luo M Guo L Yu M Jiang W Wang H The psychological and mental impact of coronavirus disease 2019 (COVID-19) on medical staff and general public - a systematic review and meta-analysis Psychiatry Res 2020 291 113190 10.1016/j.psychres.2020.113190 32563745 12 Lai J Ma S Wang Y Cai Z Hu J Wei N Factors associated with mental health outcomes among health care workers exposed to coronavirus disease 2019 JAMA Netw Open 2020 3 3 e203976 10.1001/jamanetworkopen.2020.3976 32202646 13 Maben J Bridges J COVID-19: supporting nurses’ psychological and mental health J Clin Nurs 2020 29 15-16 2742 50 10.1111/jocn.15307 32320509 14 Lv Y, Yao H, Xi Y, Zhang Z, Zhang Y, Chen J, et al. Social Support Protects Chinese Medical Staff from Suffering Psychological Symptoms in COVID-19 Defense. 2020. Available from: https://ssrn.com/abstract=3559617. 15 Graneheim UH Lundman B Qualitative content analysis in nursing research: concepts, procedures and measures to achieve trustworthiness Nurse Educ Today 2004 24 2 105 12 10.1016/j.nedt.2003.10.001 14769454 16 Connelly LM Trustworthiness in qualitative research Medsurg Nurs 2016 25 6 435 6 30304614 17 Spiotta AM Crosa R Letter to the editor: two perspectives on the COVID-19 pandemic nobody is talking about-and it’s costing lives World Neurosurg 2020 139 723 10.1016/j.wneu.2020.05.004 32426072 18 Yousefi M Ebrahimi Z Fazaeli S The experiences of nurses of infectious and non- infectious wards of caring COVID-19 patients in a big hospital in Iran: a qualitative study Iran J Nurs Midwifery Res 2022 27 1 35 40 10.4103/ijnmr.IJNMR_459_20 35280194 19 Roshanzadeh M, Jamalinik M, Hasheminik M, Tajabadi A. Stigma of COVID-19: the basic challenge in health economics. Iran Occupational Health 2020; 17(1): 137-41. [Persian]. 20 Huang L Lin G Tang L Yu L Zhou Z Special attention to nurses’ protection during the COVID-19 epidemic Crit Care 2020 24 1 120 10.1186/s13054-020-2841-7 32220243 21 Duran S Celik I Ertugrul B Ok S Albayrak S Factors affecting nurses’ professional commitment during the COVID-19 pandemic: a cross-sectional study J Nurs Manag 2021 29 7 1906 15 10.1111/jonm.13327 33794061 22 Saffari M Vahedian-Azimi A Mahmoudi H Nurses’ experiences on self-protection when caring for COVID-19 patients J Mil Med 2020 22 6 570 9 10.30491/jmm.22.6.570 23 Heydari A Vafaee Najar A Bakhshi M Resource management among intensive care nurses: an ethnographic study Mater Sociomed 2015 27 6 390 4 10.5455/msm.2015.27.390-394 26889097 24 Malelelo-Ndou H Ramathuba DU Netshisaulu KG Challenges experienced by health care professionals working in resource-poor intensive care settings in the Limpopo province of South Africa Curationis 2019 42 1 e1 e8 10.4102/curationis.v42i1.1921 25 Nayna Schwerdtle P Connell CJ Lee S Plummer V Russo PL Endacott R Nurse expertise: a critical resource in the COVID-19 pandemic response Ann Glob Health 2020 86 1 49 10.5334/aogh.2898 32435602 26 Kang HS Son YD Chae SM Corte C Working experiences of nurses during the Middle East respiratory syndrome outbreak Int J Nurs Pract 2018 24 5 e12664 10.1111/ijn.12664 29851209 27 Smith MW Smith PW Kratochvil CJ Schwedhelm S The psychosocial challenges of caring for patients with Ebola virus disease Health Secur 2017 15 1 104 9 10.1089/hs.2016.0068 28192056 28 Kim Y Nurses’ experiences of care for patients with Middle East respiratory syndrome-coronavirus in South Korea Am J Infect Control 2018 46 7 781 7 10.1016/j.ajic.2018.01.012 29502886 29 Hong S Ai M Xu X Wang W Chen J Zhang Q Immediate psychological impact on nurses working at 42 government-designated hospitals during COVID-19 outbreak in China: a cross-sectional study Nurs Outlook 2021 69 1 6 12 10.1016/j.outlook.2020.07.007 32919788 30 Patrick A Laschinger HK The effect of structural empowerment and perceived organizational support on middle level nurse managers’ role satisfaction J Nurs Manag 2006 14 1 13 22 10.1111/j.1365-2934.2005.00600.x 16359442 31 Mohammadi F Farjam M Gholampour Y Sohrabpour M Oshvandi K Bijani M Caregivers’ perception of the caring challenges in coronavirus crisis (COVID-19): a qualitative study BMC Nurs 2021 20 1 102 10.1186/s12912-021-00607-1 34147101
PMC010xxxxxx/PMC10352636.txt
==== Front J Caring Sci J Caring Sci J Caring Sci TBZMED Journal of Caring Sciences 2251-9920 2251-9920 Tabriz University of Medical Sciences 10.34172/jcs.2023.30508 Short Communication Acceptance of COVID-19 Vaccine and Related Factors in Iran: A Cross-sectional Study https://orcid.org/0000-0001-9623-129X Abdollai Mostafa 1 Conceptualization Data curation Funding acquisition Investigation Methodology Project administration Resources Validation https://orcid.org/0000-0003-0612-7427 Ayar Ayub 1 Data curation Investigation Methodology Resources Validation https://orcid.org/0000-0001-7732-1599 Khorashadizadeh Mohammad 2 Formal analysis https://orcid.org/0000-0001-7361-0123 Kouhpeikar Hamideh 3 *Supervision Validation Visualization Writing – original draft Writing – review & editing 1Department of Nursing, Tabas School of Nursing, Birjand University of Medical Sciences, Birjand, Iran 2Department of Statistics, University of Birjand, Birjand, Iran 3Department of Hematology and Blood Bank, Tabas School of Nursing, Birjand University of Medical Sciences, Birjand, Iran * Hamideh Kouhpeikar [email protected] 6 2023 26 2 2023 12 2 7983 12 1 2022 23 6 2022 © 2023 The Author(s). 2023 https://creativecommons.org/licenses/by-nc/4.0/ This work is published by Journal of Caring Sciences as an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by-nc/4.0/). Non-commercial uses of the work are permitted, provided the original work is properly cited. http://jcs.tbzmed.ac.ir Introduction: One of the most important strategies to control COVID-19 pandemic is vaccination. Effective vaccination coverage is necessary to control this pandemic. Therefore, in this study we investigated acceptance of COVID-19 vaccine and associated factors among Iranian population. Methods: A cross-sectional study conducted through Pors Line in South Khorasan Province of Iran. 1043 people participated in this study. Results were analyzed with SPSS software version 13. Results: 85.2% of the participants wanted to receive the vaccine. Vaccine acceptance was higher in participants that were over 41 years old. Moreover, rate of vaccine acceptance was higher in men than women. Major concern about vaccination was fear of its side effects. Vaccine acceptance increased with increasing education level. Conclusion: Results of this study showed that one of the most important reasons for vaccine rejection is the fear of vaccine side effects. COVID-19 Vaccine Pandemic Corona ==== Body pmcIntroduction COVID-19 vaccine is the most effective approach to control pandemic.1 Studies showed that if a vaccine is available, high and effective coverage of vaccination needs to be accepted by the individuals.2 In addition to the problems in vaccine distribution in different countries, vaccine hesitancy is one of the most challenges ahead, as World Health Organization (WHO) identifying it as one of the top 10 global health threats in 2019. Although clinical trials have shown that vaccines are effective and safe, there are various reasons for vaccine hesitancy, including distrust of governments, concerns about vaccine side effects and misinformation about vaccines in the media.3-6 A global study reports that in 90% of countries people are hesitant about getting the vaccine and generally hesitancy about vaccine varied between 8% to 15%.7-11 Therefore, in each country the public health officials must pay special attention to this issue. low vaccine acceptance can slow down the vaccination and delay control of COVID-19 pandemic. It also imposes an enormous burden on governments and medical staff. In Iran, as in other countries, there is misinformation about the vaccine which affects the vaccine acceptance. So it is necessary to conduct studies to identify the factors affecting vaccine acceptance and provide solutions to improve public acceptance. Accordingly, this study designed to investigate COVID-19 vaccine acceptance and associated factors among Iranian population. Materials and Methods We did a cross-sectional study that conducted between 28 February 2021 and 18 March 2021. PASS software version 11 was used to determine the sample size. So considering a margin of error 5% and confidence level 95%, the sample size was determined 1043. Using proportional stratified sampling method, the participants were selected from 5 city (Birjand, Tabas, Nehbandan, Ferdows and Boshrouyieh) of South Khorasan province of Iran. Using Cochran’s formula (with precision of 0.01 and 95% confidence level) validity and reliability was determined (Cronbach’s alpha= 0.78). This survey was produced by Pors Line, an Iranian online survey platform (https://survey.porsline.ir/s/76HZY9K). We shared the link of survey through social media (WhatsApp and Telegram channel). The questions of questionnaire consisted of several sections: demographic characteristics of participants, vaccine acceptance, barriers of vaccine acceptance, people’s attitudes about the coronavirus and coronavirus vaccine. Exclusion criteria were people over 85 years and under 18 years. Analysis of data were conducted with SPSS software version 13. Normal distributions were evaluated using Kolmogorov–Smirnov test. The analytical procedure consisted of two tests. First, chi-square test was used to examine relationship between vaccine acceptance and demographic characteristics. Also simple linear regression was used to investigate role of barriers on vaccine acceptance. Significance level was considered to be 0.05. Descriptive statistics were presented as frequencies (n) and percentage (%). Results Total number of participants in this study was 1043. The highest number of respondents was belonged to Tabas city 663 (63.6). The majority of respondents were in the age range 19-30 years. 442 (42.4) of participants, were male and 601 (57.6) were female. Most of the participants were married 728 (69). Moreover most respondents had a bachelor’s degree 447 (42.9). Most of the participants in this study were employed in government jobs 39.5% and 38.6% of them were unemployed. 831(79.7) of the participants had no comorbidities and most of them had no history of infection with COVID-19. The majority of people stated that COVID-19 is a serious and important disease 910 (87.3) and they said that they were more likely to get COVID-19 633 (60.7). Table 1 indicates acceptance of COVID-19 vaccine among the participants. Most of respondents strongly agreed with this subject: {If you know that getting the COVID-19 vaccine will protect you against this disease, be sure to get vaccinated} 555(53.3) and 332(31.9) of them agreed. This indicates that 888 (85.2) of participants were sure about self-vaccination and only 154 (14.8) of them disagreed or they had not decided. In the second and third questions was asked that: 886 (85) of people agreed with these questions: If you know that the COVID-19 vaccine will protect your family and community from disease, be sure to get vaccinated and If you know that the COVID-19 vaccine will return the community to normal, you will be vaccinated. Furthermore in response to {I follow all health protocols and inject the vaccine} 675 (64.8) of respondents strongly agreed or agreed. 372 (35.7) of participants preferred to use internal vaccine and only 209 (20.1) of them preferred foreign vaccine, 310 (29.8) did not decide about it and 150 (14.4) of participants said type of vaccine is not important. Most of participants said that if a vaccine is confirmed by the WHO, they will inject it 764 (73.3) which shows people’s trust in World Health Organization (WHO). Results indicated that vaccination acceptance was higher in participants that were over 41 years old. Moreover, the rate of vaccine admission was higher in men than women and in proportion to the increase in education level, rate of vaccine acceptance also increases. Table 1 Acceptance of COVID-19 vaccine Items Strongly disagree Disagree Neither agree nor disagree Agree Strongly agree If you know that getting the COVID-19 vaccine will protect you against this disease, be sure to get vaccinated. 32 3.1 40 3.8 82 7.9 333 31.9 556 53.3 If you know that the COVID-19 vaccine will protect your family and friends from the disease, be sure to get vaccinated. 29 2.8 33 3.2 75 7.2 319 30.6 587 56.3 If you know that COVID-19 vaccine protects other people in the community against this disease, be sure to get vaccinated. 26 2.5 32 3.1 77 7.4 332 31.8 576 55.2 If you know that the COVID-19 vaccine will return the community to normal, you will be vaccinated. 21 2 27 2.6 67 6.4 290 27.8 638 61.2 I follow all health protocols and receive the vaccine. 47 4.5 94 9 226 21.7 352 33.7 324 31.1 Table 2 shows the barriers associated with acceptance of COVID-19 vaccination. One of the concerns in most people about vaccine injections was the side effects of vaccine. In our study, in response to the question {I’m worried about the side effects of the COVID-19 vaccine}, most of participants agreed or strongly agreed 806 (77.3). I think COVID-19 vaccine is dangerous for me: only 396 (38) of respondents agreed with this sentence, and about 361 (34.7) of them were neither agree nor disagree, and 219 (21) were opposed to this. Another thing that is considered to be one of the barriers associated with COVID-19 vaccination is the cost of the vaccine, also fear of getting COVID-19 through vaccine injection. In our study, more than 312 (30) of people agreed with these sentences: I’m worried to get COVID-19 from the vaccine and I have to charge much money for this vaccine. Also a small percentage of respondents were afraid of injections 166 (16). 378 (36.3) of respondents were agree or strongly agree with this sentence: I believe in natural remedies and traditional medicine to treat COVID-19 disease. It seems that the role of religious beliefs as an obstacle in vaccination is very low and only 70 (6.8) of people agreed or strongly agreed with: I do not get the COVID-19 vaccine because of my religious beliefs. Most participants were opposed or strongly opposed with this sentence: I do not get the vaccine because I am not at high risk 661 (63.4). Table 2 Barriers of COVID-19 vaccine Items Strongly disagree Disagree Neither agree nor disagree Agree Strongly agree I'm worried about the side effects of the COVID-19 vaccine 23 2.2 69 6.6 145 13.9 360 34.5 446 42.8 I think the COVID-19 vaccine is dangerous for me 65 6.2 219 21 362 34.7 240 23 157 15.1 I'm worried to get COVID-19 from the vaccine 99 5.9 271 26 315 30.2 228 21.8 130 12.5 I have to pay a lot of money for this vaccine 131 12.6 264 25.3 317 30.4 217 20.8 114 10.9 I'm afraid of injections 348 33.4 400 38.4 128 12.3 113 10.8 54 5.2 I believe in natural remedies and traditional medicine to treat coronary heart disease 166 15.9 202 19.4 297 28.4 229 22 149 14.3 I do not get the COVID-19 vaccine because of my religious beliefs 438 42 385 36.9 149 14.3 42 4 29 2.8 I do not get the vaccine because I am not at high risk 271 26 390 37.4 223 21.4 111 10.6 48 4.6 Simple linear regression has been used to investigate role of barriers on vaccine acceptance. Table 3 shows result of Simple linear regression: According to the obtained results, the following linear relationship can be expressed based on regression analysis: Acceptance of vaccine = 4.17-0.16 × Barriers of vaccine acceptance the above relationship shows if the barriers increase one unit, vaccine acceptance will decrease 0.16 (R2 = 0.03, F= 41.08). Table 3 Results of simple regression model (dependent variable: vaccine acceptance) Unstandardized coefficients Beta standard coefficients T value P value B Standard error Constant 3.07 0.11 29.49 0.0001 Barriers to vaccination -0.06 0.02 -0.06 -2.18 .030 Encouragements for vaccination 0.24 0.02 0.42 14.33 0.0001 F test 127.27 P value 0.0001 Kolmogorov-Smirnov test (P value) 0.32 Durbin-Watson 1.98 R square 0.197 Adjusted R square 0.195 Discussion Results of this study showed that the rate of vaccine acceptance among respondents was 85.2% which was higher than other studies. In a global study that conducted in 19 countries, 13142 participants entered in this study to determine acceptance of the COVID-19 vaccine. 71.5% of respondents said that they would be likely to take vaccine if an effective and safe vaccine was available. 48.1% of participants reported that they would inject the vaccine if their employer recommended it. In Asian countries such as China, Korea and Singapore, more than 80% of people wanted to take COVID-19 vaccine.12 In a study conducted in France, 25% of the French adult population did not want to take COVID-19 vaccine and the main reason they stated was that the vaccine was not safe.13These differences in vaccine acceptance between different countries can delay control of the COVID-19 pandemic. Vaccine acceptance is influenced by beliefs of the people and social factors. For example, in a recent study in several countries, the authors found that Africans experienced medical distrust due to racial discrimination and were less likely to receive the vaccine. Also it has been found that in middle-income countries people are more likely to receive the vaccine.14 Results of our study showed that the majority of participants who wanted to receive the vaccine were over 41 years old and also the rate of vaccination was higher in men than women. In line with this evidence in a global study by Lazarus et al older people were more likely to get the vaccine whereas respondents that were in the age range of 25-54 and 55-64 years reported that they would inject the vaccine if their employer recommended it.12 In compliance with our results, Troiano and Nardi indicated that vaccine acceptance was higher in men than women.15 Whereas Lazarus et al reported that this rate was higher in women than men.12 The results of our study showed that as the level of education increases, the same rate of vaccine acceptance among participants increases which may be related to increasing the level of knowledge and awareness of individuals. In line with this evidence, Salali and Uysal in Turkey reported that low levels of education were associated with low vaccine acceptance.16 Vaccine acceptance in people who had not been infected with COVID-19 and participants with a history of COVID-19 disease was same. In line with this, Lazarus et al showed that there was no difference between two groups in receiving vaccine.12 Various factors affect the acceptance of vaccine and people have concerns about vaccine. Recognizing these concerns and designing appropriate strategies to solve them increases overall coverage of the COVID-19 vaccine. One of the most important reasons for rejecting the vaccine is the concerns about side effects of vaccine.1 Results of our study showed that the most important reason for refraining from vaccination was concern about its side effects. Contrary to our results, a recent study conducted in Iran and several countries showed that the most important reason for rejecting the vaccine was lack of confidence in the vaccine.14 In our study, about 77.3% of participants were concerned about the side effects of the vaccine, and this seems to be one of the most important barriers to vaccination. One study by Akarsu et al indicated that the most reasons of COVID-19 vaccine rejection were fear of vaccine, side effects and distrust in vaccine efficacy.17In compliance with previous study, Pugliese-Garcia et al reported that fear of infection with the COVID-19 through vaccination and the belief that the vaccine was ineffective were among the most important reasons for not receiving the vaccine.18Lazarus et al reported that religious motivation was a negative factor in vaccine acceptance,12 while the results of our study showed that religious beliefs are not barriers of vaccine acceptance. One of our hypotheses was that people with comorbidities would be more likely to receive the vaccine, and the results showed that vaccine acceptance in this group was higher than healthy individuals. In a study by Harapan et al in Indonesia, people at higher risk for COVID-19 were more likely to be vaccinated.2 Also, one of the unique features of our study was evaluating of people’s attitude about type of vaccine. The results showed that most people were willing to receive the Iranian vaccine. Our study had some limitations, such as the use of only online questionnaires instead of face-to-face interview Also another limitation was the small number of the studied population, so it is suggested that future studies be conducted in a larger statistical population. Conclusion Results of this study showed that there are several factors affects COVID-19 vaccine acceptance. One of the most important factor is the fear of the COVID-19 vaccine side effects. Age, sex and level of education also affected vaccine acceptance. Results of this study can promote vaccine acceptance. Acknowledgements This work was supported by a grant from Birjand University of Medical Sciences. Author’s Contribution Conceptualization: Mostafa Abdollai. Data curation: Mostafa Abdollai, Ayub Ayar. Formal analysis: Mohammad Khorashadizadeh. Funding acquisition: Mostafa Abdollai. Investigation: Mostafa Abdollai, Ayub Ayar. Methodology: Mostafa Abdollai, Ayub Ayar. Project administration: Mostafa Abdollai. Resources: Mostafa Abdollai, Ayub Ayar. Supervision: Hamideh Kouhpeikar. Validation: Mostafa Abdollai, Ayub Ayar. Hamideh Kouhpeikar. Visualization: Hamideh Kouhpeikar. Writing–original draft: Hamideh Kouhpeikar. Writing–review & editing: Hamideh Kouhpeikar. COI-statement The authors declare that they have no conflict of interest. Data Availability The findings of this study will be available upon request from the corresponding author. Ethical Approval This study was approved by the ethics committee of Birjand University of Medical Sciences (IR.BUMS.REC.1399.485). Funding This study was funded by Birjand University of Medical Sciences. Research Highlights What is the current knowledge? COVID-19 vaccine is the best approach to control pandemic. If a vaccine is available, high and effective coverage of vaccination needs to be accepted by the individuals. What is new here? The fear of COVID-19 vaccine side effects was the main barrier of vaccine acceptance. ==== Refs References 1 Magadmi RM Kamel FO Beliefs and barriers associated with COVID-19 vaccination among the general population in Saudi Arabia BMC Public Health 2021 21 1 1438 10.1186/s12889-021-11501-5 34289817 2 Harapan H Wagner AL Yufika A Winardi W Anwar S Gan AK Acceptance of a COVID-19 vaccine in Southeast Asia: a cross-sectional study in Indonesia Front Public Health 2020 8 381 10.3389/fpubh.2020.00381 32760691 3 Lurie N Saville M Hatchett R Halton J Developing Covid-19 Vaccines at Pandemic Speed N Engl J Med 2020 382 21 1969 1973 10.1056/NEJMp2005630 32227757 4 Askarian M, Fu L, Taghrir MH, Borazjani R, Shayan Z, Taherifard E, et al. Factors Affecting COVID-19 Vaccination Intent Among Iranians: COVID-19 Vaccination Acceptance. 2020. Available at SSRN: https://ssrn.com/abstract=3741968. 5 Larson HJ Clarke RM Jarrett C Eckersberger E Levine Z Schulz WS Measuring trust in vaccination: a systematic review Hum Vaccin Immunother 2018 14 7 1599 609 10.1080/21645515.2018.1459252 29617183 6 Xiao X Wong RM Vaccine hesitancy and perceived behavioral control: a meta-analysis Vaccine 2020 38 33 5131 8 10.1016/j.vaccine.2020.04.076 32409135 7 Chan EY Cheng CK Tam GC Huang Z Lee PY Willingness of future A/H7N9 influenza vaccine uptake: a cross-sectional study of Hong Kong community Vaccine 2015 33 38 4737 40 10.1016/j.vaccine.2015.07.046 26226564 8 Lane S MacDonald NE Marti M Dumolard L Vaccine hesitancy around the globe: analysis of three years of WHO/UNICEF Joint Reporting Form data-2015-2017 Vaccine 2018 36 26 3861 7 10.1016/j.vaccine.2018.03.063 29605516 9 Alabbad AA Alsaad AK Al Shaalan MA Alola S Albanyan EA Prevalence of influenza vaccine hesitancy at a tertiary care hospital in Riyadh, Saudi Arabia J Infect Public Health 2018 11 4 491 9 10.1016/j.jiph.2017.09.002 28988776 10 Cunningham RM Minard CG Guffey D Swaim LS Opel DJ Boom JA Prevalence of vaccine hesitancy among expectant mothers in Houston, Texas Acad Pediatr 2018 18 2 154 60 10.1016/j.acap.2017.08.003 28826731 11 Giambi C Fabiani M D’Ancona F Ferrara L Fiacchini D Gallo T Parental vaccine hesitancy in Italy–results from a national survey Vaccine 2018 36 6 779 87 10.1016/j.vaccine.2017.12.074 29325822 12 Lazarus JV Ratzan SC Palayew A Gostin LO Larson HJ Rabin K A global survey of potential acceptance of a COVID-19 vaccine Nat Med 2021 27 2 225 8 10.1038/s41591-020-1124-9 33082575 13 Ward JK Alleaume C Peretti-Watel P The French public’s attitudes to a future COVID-19 vaccine: the politicization of a public health issue Soc Sci Med 2020 265 113414 10.1016/j.socscimed.2020.113414 33038683 14 Asadi Faezi N Gholizadeh P Sanogo M Oumarou A Mohamed MN Cissoko Y Peoples’ attitude toward COVID-19 vaccine, acceptance, and social trust among African and Middle East countries Health Promot Perspect 2021 11 2 171 8 10.34172/hpp.2021.21 34195040 15 Troiano G Nardi A Vaccine hesitancy in the era of COVID-19 Public Health 2021 194 245 51 10.1016/j.puhe.2021.02.025 33965796 16 Salali GD, Uysal MS. COVID-19 vaccine hesitancy is associated with beliefs on the origin of the novel coronavirus in the UK and Turkey. Psychol Med. 2020: 1-3. 10.1017/s0033291720004067 17 Akarsu B Canbay Özdemir D Ayhan Baser D Aksoy H Fidancı İ Cankurtaran M While studies on COVID-19 vaccine is ongoing, the public’s thoughts and attitudes to the future COVID-19 vaccine Int J Clin Pract 2021 75 4 e13891 10.1111/ijcp.13891 33278857 18 Pugliese-Garcia M Heyerdahl LW Mwamba C Nkwemu S Chilengi R Demolis R Factors influencing vaccine acceptance and hesitancy in three informal settlements in Lusaka, Zambia Vaccine 2018 36 37 5617 24 10.1016/j.vaccine.2018.07.042 30087047
PMC010xxxxxx/PMC10352637.txt
==== Front J Caring Sci J Caring Sci J Caring Sci TBZMED Journal of Caring Sciences 2251-9920 2251-9920 Tabriz University of Medical Sciences 10.34172/jcs.2023.31742 Original Article The Relationship between COVID-19 Exposure Risk and Burnout in Prehospital Emergency Medical Technicians https://orcid.org/0000-0002-5961-968X Javanmardi Karim 1 Conceptualization Data curation Formal analysis Investigation Methodology Validation Visualization Writing – original draft Writing – review & editing https://orcid.org/0000-0002-5399-0277 Gilani Neda 2 Formal analysis Methodology Project administration Supervision Validation Visualization Writing – original draft Writing – review & editing https://orcid.org/0000-0002-3771-5152 Ghafourifard Mansour 1 Formal analysis Methodology Supervision Validation Visualization Writing – original draft Writing – review & editing https://orcid.org/0000-0002-2432-1904 Dadashzadeh Abbas 1 *Conceptualization Formal analysis Funding acquisition Investigation Methodology Project administration Supervision Validation Visualization Writing – original draft Writing – review & editing https://orcid.org/0000-0003-1304-4726 Dehghannejad Javad 1 Conceptualization Data curation Validation Visualization Writing – original draft Writing – review & editing https://orcid.org/0000-0001-5142-109X Feyzollahzade Hossein 1 Project administration Validation Visualization Writing – original draft Writing – review & editing 1Department of Medical Surgical Nursing, School of Nursing & Midwifery, Tabriz University of Medical Sciences Tabriz, Iran 2Department of Statistics and Epidemiology, Faculty of Health, Tabriz University of Medical Sciences, Tabriz, Iran * Abbas Dadashzadeh [email protected] 6 2023 13 3 2023 12 2 123128 10 6 2022 19 8 2022 © 2023 The Author(s). 2023 https://creativecommons.org/licenses/by-nc/4.0/ This work is published by Journal of Caring Sciences as an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by-nc/4.0/). Non-commercial uses of the work are permitted, provided the original work is properly cited. http://jcs.tbzmed.ac.ir Introduction: Exposure to coronavirus disease 2019 (COVID-19) has caused many physical and psychological effects on front-line healthcare workers (HCWs). This study aimed to assess the relationship between the exposure risk to COVID-19 disease and burnout in prehospital emergency medical technicians (EMTs). Methods: In this correlational study, 335 prehospital EMTs were selected by random sampling method from the 49 stations of emergency medical services in the northwest of Iran. Data were collected using a questionnaire developed by the world health organization for the risk assessment and management of exposure of health care workers to COVID-19. Moreover, Pines burnout measure was used for the assessment of participants’ burnout. Data were analyzed using SPSS version 13. Results: Results showed that 30.7 % of prehospital EMTs had a high burnout score against COVID-19 disease. The prehospital EMTs who had a high occupational exposure risk experienced a high risk of burnout (P=0.03). The results of the linear regression analysis showed that prehospital EMTs who had a low exposure risk of COVID-19 had a low burnout score (β=-9.30; P<0.001), and those who had less than 10 years of work experience showed less burnout (β=-10.54; P<0.001). Conclusion: According to the results, the exposure risk to COVID-19 increases the prehospital EMT’s burnout. As a result, reducing the exposure risk to COVID-19 by providing adequate access to personal protective equipment (PPE), development of training and following standards and protocols can be effective in controlling burnout in HCWs. Burnout COVID-19 Emergency medical service Exposure Prehospital emergency ==== Body pmcIntroduction Healthcare workers (HCWs) are at high risk for coronavirus disease 2019 (COVID-19) due to prolonged exposure to large numbers of people infected with this virus.1Prehospital emergency medical technicians (EMTs) play a main role in initiating disease isolation precautions, providing emergency care, and transporting of patients to a hospital or medical center for receiving further treatment.2 However, these personnel are potentially exposed to the COVID-19 virus3 because of first contact with infected or suspected individuals,4 performing aerosol-generating procedures such as cardiopulmonary resuscitation (CPR), suctioning, and tracheal intubation for infected patients.5Therefore, fear of infection with COVID-19 is considered a main concern for prehospital EMTs who have a contact with patients and provide the ambulance services.1 In addition, working in the long shifts alongside with a shortage of personal protective equipment (PPE) could leads to the further stress and discomfort among them.6 Therefore, working in this challenging context could result in physical, emotional and psychological problems and eventually leads to burnout.7 Burnout is a syndrome caused by chronic work-related stress, which is associated with symptoms such as physical and emotional fatigue, feeling of professional failure, and a negative attitude towards work.8It can lead to many negative consequences such as job dissatisfaction, poor quality of care, and decreased efficiency of healthcare services.9 In the caring field, burnout means loss of feeling and interest in the patients and providing improper and non-standard care for them.10The COVID-19 pandemic has raised many challenges for HCWs1and affected the performance and job satisfaction of the front-line HCWs.11 In a study on prehospital emergency medical personnel, Heidari et al found that personnel who directly cared for the patients with COVID-19 had higher symptoms of burnout.12 Another study conducted by Hadian et al showed that prehospital emergency medical personnel caring for the COVID-19 patients are suffering from a range of psychiatric problems.13 Also, a study in Taiwan showed that prehospital personnel who cared for COVID-19 patients had higher burnout scores.14 Healthcare providers who directly care for COVID-19 patients often see patients’ suffering and dying which could lead to their emotional distress and fatigue.15 Moreover, it could have harmful effects on the physical health and quality of life of healthcare providers and can lead to their absenteeism. Finally, these challenges reduce the quality of care delivered to the patients.16 There are two large cities (Tabriz and Urmia) in the northwest of Iran. In these two cities, the emergency medical service (EMS) systems annually receive more than 700,000 calls which more than 150,000 of them lead to ambulance services. With the COVID-19 outbreak, the number of EMSs were increased very sharply. In this regard, Natalzia et al reported that 63.7% of the patients during this pandemic suffered from an unstable situation.17Another study showed that cardiopulmonary arrests had increased by 77.4% during the COVID-19 pandemic. These patients needed the EMS personnel help to start CPR as soon as possible.18 During the COVID-19 pandemic, controlling the exposure risk of the disease and improving the job satisfaction of personnel is of great importance. Due to the novelty of the disease and since there is no evidence about the relationship between exposure risk to the COVID-19 disease and burnout in prehospital EMTs, this study aimed to assess the relationship between the exposure risk to the COVID-19 disease and burnout among prehospital EMTs. Materials and Methods This correlational study was conducted in two large cities (Tabriz and Urmia) located in the northwest of Iran. This study was conducted between May 2020 and April 2021. A total of 335 prehospital EMTs were selected by random sampling method from the 49 stations of EMS. The main inclusion criteria included working for at least six months as an emergency care provider to the COVID-19 patients. Data collection tools consisted of three tools. The first tool is related to the demographic and job characteristics. The second questionnaire was adapted from a questionnaire developed by the World Health Organization (WHO) for the assessment of exposure risk and management of COVID-19 by HCWs. The questionnaire consists of three parts: community exposure to the COVID-19 virus, occupational exposure to the COVID-19 virus, and infection prevention and control (IPC) measure in contact with suspected or infected COVID-19 patients.19,20 The questionnaire assesses the type of activity in which the HCW is engaged. Moreover, it measures the level of risk based on the low-risk or high-risk events. In a subscale of community exposure and occupational exposure, if an HCW responds “yes” to any of the activities reported in the scale, the person is considered to high exposure risk to the COVID-19 virus. If an HCW select the response of “always, as recommended” to any of the IPC measures when caring for a confirmed COVID-19 patient, the person was considered at the low risk of the COVID-19 virus infection. If an HCW responds to other options, the person was considered high risk for the COVID-19 virus infection.21 The burnout measure which has 21 items was used to assess the physical, emotional, and mental exhaustion of the participants.22 Each item is scored based on a 7-point Likert scale ranging from 1 “Never” to 7 “Always”. The higher scores indicate more severe symptoms of burnout. The burnout score is obtained from the mean of the responses to all items. Burnout scores are categorized into four groups: no burnout (≤ 2.9), risk of burnout (3-3.9), burnout present (4- 4.9), and clinically depressed (≥ 5). Moreover, burnout score was classified into two levels: low risk (score less than 50%) and high risk (score above 50%). The content validity of the questionnaire was confirmed by 30 different samples. The reliability of the questionnaire was calculated in the range of 0.91-0.93 through Cronbach’s alpha coefficient.23 The present questionnaire was used in the study of Johns to measure the burnout of caregivers of patients with HIV and AID.24 The collected data were analyzed using descriptive and inferential statistics such as ANOVA, chi-square, Pearson correlation coefficient and regression using SPSS version 13 (SPSS Inc., Chicago, Ill., USA) software. A P value < 0.05 was considered as significant. Results In terms of demographic and job characteristics, most of the prehospital EMTs (44.8 %) were between 30-39 years old with a mean (SD) age of 32.81(6.81). The mean work experience of prehospital EMTs was 8.41 (6.15) years. During 13 months after the onset of the pandemic, 60.3% of the prehospital EMTs were infected with COVID-19. The average number of prehospital services related to suspected or infected COVID-19 patients was less than 3 service in 24 hours, with an average of 30 minutes of contact with each COVID-19 patient (Table 1). Table 1 Demographic characteristics of the participants (N=335) Variable No. (%) Age 29≥ 128 (38.2) 30-39 150 (44.8) 40-49 49 (14.6) 50≤ 8 (2.4) Mean (SD) 32.81(6.81) Work experience (years) ≤10 225 (67.2) >10 110 (32.8) Mean (SD) 8.41(6.15) Marital status Single 95 (28.4) Married 230 (68.7) Divorced 10 (3) History of COVID-19 infection Yes 202 (60.3) No 133 (39.7) Average number of suspected or infected COVID-19 patients admitted during a 24-hour shift 3> 151 (45.5) 3-5 96 (28.7) 6-10 78 (23.3) 10< 10 (3) Mean duration of contact with each COVID-19 patient 15 min 56 (16.7) 30 min 104 (31) 45 min 85 (25.4) 1 h 63 (18.8) Over 1 h 27 (8.1) In 55.2% of prehospital EMTs, the exposure risk of the COVID-19 disease was high and 30.7 % of the personnel had high burnout score. The standardized mean of the total exposure risk and the total burnout score was 53.71% and 39.83 %, respectively (Table 2). The mean and standard deviation of the burnout was 3.62 ± 0.93. According to the results, 41.9 % of personnel were at risk of burnout, and 32.7 % had burnout (Table 3). Table 2 Distribution of the exposure risk of COVID-19 and burnout (N=335) Variable N (%) Mean (SD) Community exposure Low risk 105 (31.3) 82.39 (27.68) High risk 230 (68.7) Occupational exposure Low risk 47 (14) 73.97 (23.73) High risk 288 (86) Adherence to infection prevention measures Low risk 203 (606) 46.50 (24.13) High risk 132 (39.4) Total score of exposure risk Low risk 150 (44.8) 53.71 (19.24) High risk 185 (55.2) Total burnout score Low risk 232 (69.3) 39.83 (19.80) High risk 103 (30.7) Table 3 tribution of burnout levels (N=335) Variable No. (%) Total burnout score No burnout (≤2.9) 84 (25.3) Risk of burnout (3–3.9) 139 (41.9) Burnout present (4–4.9) 83 (25.0) Clinically depressed (≥5) 26 (7.8) Mean (SD) 3.62(0.93) In this study, the prehospital EMTs who had a high occupational exposure risk had a high risk of burnout (P=0.03). Participants who followed the infection prevention measures more (46.57%), were at a lower risk of burnout (P=0.005). 19.70 % of participants who had a high total exposure risk score to COVID-19 showed a high risk of burnout (P=0.001). However, there was no significant difference in the community exposure risk (Table 4). Table 4 The relationship between exposure risk levels to COVID-19 and burnout (N=335) Variable Low risk High risk r P value No. (%) No. (%) Community exposure 0.06 0.23 Low risk 79 (23.58) 159 (47.46) High risk 26 (7.76) 68 (20.30) Occupational exposure 0.161a 0.03 Low risk 39 (11.64) 199 (59.40) High risk 7 (2.09) 87 (25.97) Following infection prevention measures 0.244a 0.001 Low risk 156 (46.57) 82 (24.48) High risk 46 (13.73) 48 (14.33) Total exposure risk score 0.272a 0.001 Low risk 121 (36.12) 117 (34.92) High risk 28 (8.35) 66 (19.70) r=Pearson's correlation coefficient, a Significant correlation at the level of 0.05. The analysis by Pearson’s correlation showed a positive and significant association between occupational exposure score, following infection prevention measures score, and total exposure risk score of the COVID-19 disease and burnout. The correlation between the total exposure risk score of the COVID-19 disease and burnout was r=0.27 (P<0.001) (Table 4). The results of the linear regression showed that prehospital EMTs who had a low exposure risk of COVID-19 showed a low burnout score (β=-9.30; P<0.001), and those who had less than 10 years of work experience had less burnout (β=-10.54; P<0.001). Moreover, burnout was higher in those who had a COVID-19 history (β=4.85; P=0.02) (Table 5). Table 5 Univariate and multivariate linear regression between the exposure risk of COVID-19 and socio-demographic features with burnout Variable Univariate Multivariate β (CI: 95%) P value β (CI: 95%) P value Exposure risk of COVID-19 High risk Reference Low risk -9.30 (-13.48- -5.11) <0.001 -8.54 (-12.60- -4.49) <0.001 Work experience >10 Reference ≤10 -10.54(43.30-50.53) <0.001 -10.23(-14.52 - -5.95) lt;0.001 COVID-19 history No Reference Yes 4.85 (0.50-9.21) 0.02 4.64 (0.52- 8.77) 0.02 Discussion According to the results, the majority of the prehospital EMTs infected with COVID-19, which is similar to studies conducted in this regard.12,14 In line with previous studies, the prehospital EMTs had a high exposure risk to COVID-19.19,25 Since these personnel are at the frontline of the medical system against the pandemic3 and they provide care in uncertain environments for patients with unclear medical histories,26they have a high exposure risk in the workplace. However, COVID-19 disease can be controlled with adequate access to PPE and providing a high-quality care. In this study, 30.7% of prehospital EMTs were at high risk of burnout during the COVID-19 pandemic. In this regard, a study by Kakemam et al showed that 31.5% of nurses had a high risk of burnout against COVID-19.6 In another study conducted in Japan, the rate of burnout in HCWs who were in close contact with COVID-19 patients was reported at 31.4%.27 Another study in Wuhan, China, showed that physicians and nurses experienced high levels of burnout symptoms, anxiety, and insomnia during the COVID-19 pandemic.28These findings are similar to our study and it can be explained by the negative physical and psychological effects of the disease on the personnel health. According to a study done in Iran, 64.6% of nurses working in the hospitals of Shiraz during COVID-19 pandemic had severe burnout.29Another study in India found a higher prevalence of burnout (44.6%) among HCWs during the pandemic30 as compared with our study. In addition to the physical and psychological effects of the COVID-19 pandemic. It seems that the lack of health information and inadequate access to PPE could lead to increasing burnout.11 Therefore, due to the high rate of burnout in prehospital EMTs in the COVID-19 crisis, emotional support and psychological counseling could play an important role in minimizing it. Also, providing self-care techniques, financial support, training and employment of staff, can be effective in maintaining nurses’ mental health.27,31 Xiong et al showed that improving nurses’ self-efficacy in dealing with infectious diseases such as COVID-19 is an important factor in reducing their psychological stress.32 Our study also showed that personnel who were at high exposure risk of COVID-19 disease had higher burnout. This result is similar to a study conducted in Saudi Arabia by Al Sulais et al.33 Moreover, another study in Iran showed that nurses’ burnout have been increased by 39% during the outbreak of the COVID-19 pandemic, and their efficiency and job performance have been decreased by 20%.34This may be attributed to the fear of infection with COVID-19. Thus, paying attention to the psychological needs of healthcare providers is recommended. Contrary to our findings, the results of a study in China showed that front-line HCWs caring for COVID-19 patients experienced lower burnout than those working in the usual wards.7 According to the literature review, providing basic safety requirements and an adequate access to PPE could reduce the exposure risk of COVID-19. Moreover, early identification of HCWs with lower job satisfaction and higher burnout and providing supportive measures for them could help to decrease the personnel’s burnout.35,36 In this study, prehospital EMTs with professional experience and who had a COVID-19 history showed a higher burnout score. In this regard, Zhang et al showed that when the staffs’ work experience is increased, their job satisfaction is decreased.36 Similar to our findings, a study by Hoseinabadi et al showed that burnout was high among those who cared for infected patients for a long time and among those who had a history of COVID-19.37 These results could be attributed to the psychological stress caused by this. This research was conducted in the Northwest of Iran. Therefore, the findings could not be generalizable to the whole country. Conducting other studies in other cities is recommended. Moreover, we only studied the personnel working in prehospital emergency services. It is suggested to compare the attitudes of personnel working in the prehospital and hospital settings. Conclusion Prehospital EMTs are at high risk of burnout against COVID-19 disease. According to the results, the exposure risk to this disease increases the prehospital EMT’s burnout. As a result, reducing the exposure risk to COVID-19 with providing adequate access to PPE, development of training and following standards and protocols can be effective in controlling burnout in HCWs. Moreover, emotional support of the prehospital EMTs by managers alongside with engaging in mindfulness techniques, such as breathing exercises and meditation could be helpful. Acknowledgments This project was derived from the master’s thesis conducted at Tabriz University of Medical Sciences, Faculty of Nursing and Midwifery. Hereby, the authors would like to thank all participants in this study. Authors’ Contribution Conceptualization: Karim Javanmardi, Abbas Dadashzadeh, Javad Dehghannejad. Data curation: Karim Javanmardi, Javad Dehghannejad. Formal analysis: Karim Javanmardi, Neda Gilani, Mansour Ghafourifard, Abbas Dadashzadeh. Funding acquisition: Abbas Dadashzadeh. Investigation: Karim Javanmardi, Abbas Dadashzadeh. Methodology: Karim Javanmardi, Neda Gilani, Mansour Ghafourifard, Abbas Dadashzadeh. Project administration: Abbas Dadashzadeh, Hossein Feyzollahzade, Neda Gilani. Supervision: Neda Gilani, Mansour Ghafourifard, Abbas Dadashzadeh. Validation: Karim Javanmardi, Neda Gilani, Mansour Ghafourifard, Abbas Dadashzadeh, Javad Dehghannejad, Hossein Feyzollahzade. Visualization: Karim Javanmardi, Neda Gilani, Mansour Ghafourifard, Abbas Dadashzadeh, Javad Dehghannejad, Hossein Feyzollahzade. Writing–original draft: Karim Javanmardi, Neda Gilani, Mansour Ghafourifard, Abbas Dadashzadeh, Javad Dehghannejad, Hossein Feyzollahzade. Writing–review & editing: Karim Javanmardi, Neda Gilani, Mansour Ghafourifard, Abbas Dadashzadeh, Javad Dehghannejad, Hossein Feyzollahzade. COI-statement The authors have no conflicts of interest to declare. Data Availability The datasets are available from the corresponding author on reasonable request. Ethical Approval The approval was obtained from the ethical review board of Tabriz University of Medical Sciences (IR. TBZMED.REC.1399.1079). The objective of the study was explained to all participants and they were assured of the confidentiality of collected data. The written consent form was obtained from all participants. Funding This research was funded by the Master Research Project from Tabriz University of Medical Sciences. Research Highlights What is the current knowledge? Occupational burnout in prehospital EMTs reduces the efficiency and causes physical and psychological complications. Because of first contact with infected or suspected individuals, performing the high risk activities such as CPR, tracheal intubation or suctioning for infected, prehospital EMTs are potentially exposed to the COVID-19 virus. Fear of infection is considered a main concern for these personnel and has caused more physical and psychological side effects in them. What is new here? Half of the prehospital EMTs showed a high exposure risk of the COVID-19 disease. Moreover, one-third of them had high burnout scores. There was a significant correlation between the exposure risk to COVID-19 and burnout among prehospital EMTs. ==== Refs References 1 Labrague LJ de Los Santos JAA Fear of COVID-19, psychological distress, work satisfaction and turnover intention among frontline nurses J Nurs Manag 2021 29 3 395 403 10.1111/jonm.13168 32985046 2 Baru A Sultan M Beza L The status of prehospital care delivery for COVID-19 patients in Addis Ababa, Ethiopia: The study emphasizing adverse events occurring in prehospital transport and associated factors PLoS One 2022 17 2 e0263278 10.1371/journal.pone.0263278 35104287 3 Gibson C Ventura C Collier GD Emergency medical services resource capacity and competency amid COVID-19 in the United States: preliminary findings from a national survey Heliyon 2020 6 5 e03900 10.1016/j.heliyon.2020.e03900 32368629 4 Sprague RM, Ladd M, Ashurst JV. EMS resuscitation during contamination while wearing PPE. In: StatPearls. Treasure Island, FL: StatPearls Publishing; 2022. 5 Al-Shaqsi S Models of international emergency medical service (EMS) systems Oman Med J 2010 25 4 320 3 10.5001/omj.2010.92 22043368 6 Kakemam E Chegini Z Rouhi A Ahmadi F Majidi S Burnout and its relationship to self-reported quality of patient care and adverse events during COVID-19: a cross-sectional online survey among nurses J Nurs Manag 2021 29 7 1974 82 10.1111/jonm.13359 33966312 7 Wu Y Wang J Luo C Hu S Lin X Anderson AE A comparison of burnout frequency among oncology physicians and nurses working on the frontline and usual wards during the COVID-19 epidemic in Wuhan, China J Pain Symptom Manage 2020 60 1 e60 e5 10.1016/j.jpainsymman.2020.04.008 32283221 8 Cocchiara RA Peruzzo M Mannocci A Ottolenghi L Villari P Polimeni A The use of yoga to manage stress and burnout in healthcare workers: a systematic review J Clin Med 2019 8 3 284 10.3390/jcm8030284 30813641 9 Ogresta J Rusac S Zorec L Relation between burnout syndrome and job satisfaction among mental health workers Croat Med J 2008 49 3 364 74 10.3325/cmj.2008.3.364 18581615 10 Sheini-Jaberi P, Baraz-Pordanjani S, Beiranvand S. Relationship between self-esteem and burnout in nurses. J Clin Nurs Midwifery 2014; 3(3): 52-62. [Persian]. 11 De los Santos JAA, Labrague LJ. Impact of COVID-19 on the psychological well-being and turnover intentions of frontline nurses in the community: a cross-sectional study in the Philippines. medRxiv [Preprint]. August 6, 2020. Available from: https://www.medrxiv.org/content/10.1101/2020.08.05.20167411v1. 12 Heidari M Heydarpoor S Yadollahi S Sheikhi RA Aliakbari F Evaluation of anxiety and professional competence of prehospital emergency medical personnel in COVID-19 pandemics Disaster Emerg Med J 2022 7 3 150 6 10.5603/DEMJ.a2022.0012 13 Hadian M Jabbari A Abdollahi M Hosseini E Sheikhbardsiri H Explore pre-hospital emergency challenges in the face of the COVID-19 pandemic: a quality content analysis in the Iranian context Front Public Health 2022 10 864019 10.3389/fpubh.2022.864019 36062086 14 Chang YT Hu YJ Burnout and health issues among prehospital personnel in Taiwan fire departments during a sudden spike in community COVID-19 cases: a cross-sectional study Int J Environ Res Public Health 2022 19 4 2257 10.3390/ijerph19042257 35206444 15 Alharbi J Jackson D Usher K The potential for COVID-19 to contribute to compassion fatigue in critical care nurses J Clin Nurs 2020 29 15-16 2762 4 10.1111/jocn.15314 32344460 16 Shariati A Rahmani Anaraki H Parvareshmasoud M Hesam M Asayesh H Mousavi SM Relationship between burnout and nurses’ job characteristics J Res Dev Nurs Midwifery 2022 12 1 47 55 17 Natalzia P Murk W Thompson JJ Dorsett M Cushman JT Reed P Evidence-based crisis standards of care for out-of-hospital cardiac arrests in a pandemic Resuscitation 2020 156 149 56 10.1016/j.resuscitation.2020.07.021 32758516 18 Baldi E Sechi GM Mare C Canevari F Brancaglione A Primi R Out-of-hospital cardiac arrest during the COVID-19 outbreak in Italy N Engl J Med 2020 383 5 496 8 10.1056/NEJMc2010418 32348640 19 Ashinyo ME Dubik SD Duti V Amegah KE Ashinyo A Larsen-Reindorf R Healthcare workers exposure risk assessment: a survey among frontline workers in designated COVID-19 treatment centers in Ghana J Prim Care Community Health 2020 11 2150132720969483 10.1177/2150132720969483 33213266 20 Bani-Issa WA Al Nusair H Altamimi A Hatahet S Deyab F Fakhry R Self-report assessment of nurses’ risk for infection after exposure to patients with coronavirus disease (COVID-19) in the United Arab Emirates J Nurs Scholarsh 2021 53 2 171 9 10.1111/jnu.12625 33476482 21 Do BN Tran TV Phan DT Nguyen HC Nguyen TTP Nguyen HC Health literacy, eHealth literacy, adherence to infection prevention and control procedures, lifestyle changes, and suspected COVID-19 symptoms among health care workers during lockdown: online survey J Med Internet Res 2020 22 11 e22894 10.2196/22894 33122164 22 Takai M Takahashi M Iwamitsu Y Ando N Okazaki S Nakajima K The experience of burnout among home caregivers of patients with dementia: relations to depression and quality of life Arch Gerontol Geriatr 2009 49 1 e1 5 10.1016/j.archger.2008.07.002 18703239 23 Malakh-Pines A, Aronson E, Kafry D. Burnout: From Tedium to Personal Growth. New York: Free Press; 1981. 24 Johns BR. The relationship between personality hardiness and burnout among informal caregivers of people with HIV and AIDS. University of Cincinnati; 1998. 25 Lee J Kim M Estimation of the number of working population at high-risk of COVID-19 infection in Korea Epidemiol Health 2020 42 e2020051 10.4178/epih.e2020051 32660216 26 Alshammaria A Baila JB Parilloa SJ PPE misuse and its effect on infectious disease among EMS in Saudi Arabia J Stud Res 2019 8 1 51 8 10.47611/jsr.v8i1.592 27 Matsuo T Kobayashi D Taki F Sakamoto F Uehara Y Mori N Prevalence of health care worker burnout during the coronavirus disease 2019 (COVID-19) pandemic in Japan JAMA Netw Open 2020 3 8 e2017271 10.1001/jamanetworkopen.2020.17271 32749466 28 Lai J Ma S Wang Y Cai Z Hu J Wei N Factors associated with mental health outcomes among health care workers exposed to coronavirus disease 2019 JAMA Netw Open 2020 3 3 e203976 10.1001/jamanetworkopen.2020.3976 32202646 29 Kamali M Kalateh Sadati A Khademi MR Ghahramani S Zarei L Ghaemi SZ Burnout among nurses during coronavirus disease 2019 outbreak in Shiraz Galen Med J 2020 9 e1956 10.31661/gmj.v9i0.1956 34466619 30 Khasne RW Dhakulkar BS Mahajan HC Kulkarni AP Burnout among healthcare workers during COVID-19 pandemic in India: results of a questionnaire-based survey Indian J Crit Care Med 2020 24 8 664 71 10.5005/jp-journals-10071-23518 33024372 31 Chegini Z Arab-Zozani M Rajabi MR Kakemam E Experiences of critical care nurses fighting against COVID-19: a qualitative phenomenological study Nurs Forum 2021 56 3 571 8 10.1111/nuf.12583 33895986 32 Xiong H Yi S Lin Y The psychological status and self-efficacy of nurses during COVID-19 outbreak: a cross-sectional survey Inquiry 2020 57 46958020957114 10.1177/0046958020957114 32900271 33 Al Sulais E Mosli M AlAmeel T The psychological impact of COVID-19 pandemic on physicians in Saudi Arabia: a cross-sectional study Saudi J Gastroenterol 2020 26 5 249 55 10.4103/sjg.SJG_174_20 32496223 34 Karimi Johani R, Taghilou H, Karimi Johani F, Jafarzadeh Gharajag Z, Babapour Azam L. Investigating the relationship between burnout and job performance in the corona epidemic from the perspective of nurses. Quarterly Journal of Nursing Management 2020; 9(4): 27-33. [Persian]. 35 Zhang SX Liu J Afshar Jahanshahi A Nawaser K Yousefi A Li J At the height of the storm: healthcare staff’s health conditions and job satisfaction and their associated predictors during the epidemic peak of COVID-19 Brain Behav Immun 2020 87 144 6 10.1016/j.bbi.2020.05.010 32387345 36 Zhang SX Chen J Afshar Jahanshahi A Alvarez-Risco A Dai H Li J Succumbing to the COVID-19 pandemic-healthcare workers not satisfied and intend to leave their jobs Int J Ment Health Addict 2022 20 2 956 65 10.1007/s11469-020-00418-6 33437225 37 Sarboozi Hoseinabadi T, Kakhki S, Teimori G, Nayyeri S. Burnout and its influencing factors between frontline nurses and nurses from other wards during the outbreak of coronavirus disease COVID-19 in Iran. Invest Educ Enferm 2020; 38(2). 10.17533/udea.iee.v38n2e03
PMC010xxxxxx/PMC10352638.txt
==== Front J Caring Sci J Caring Sci J Caring Sci TBZMED Journal of Caring Sciences 2251-9920 2251-9920 Tabriz University of Medical Sciences 10.34172/jcs.2023.30679 Original Article A Mobile Application to Assist Alzheimer’s Caregivers During COVID-19 Pandemic: Development and Evaluation https://orcid.org/0000-0002-5654-1987 Amiri Parastoo 1 Resources Project administration Validation Visualization Writing – original draft Writing – review & editing https://orcid.org/0000-0002-8081-6290 Gholipour Maryam 1 Visualization Writing – original draft Writing – review & editing https://orcid.org/0000-0001-5212-6148 Hajesmaeel-Gohari Sadrieh 2 *Supervision Validation Writing – original draft Writing – review & editing https://orcid.org/0000-0002-5430-3758 Bahaadinbeigy Kambiz 3 Supervision Writing – review & editing 1Student Research Committee, Kerman University of Medical Sciences, Kerman, Iran 2Medical Informatics Research Center, Institute for Futures Studies in Health, Kerman University of Medical Sciences, Kerman, Iran 3Digital Health Team, The Australian College of Rural and Remote Medicine, Brisbane, QLD, Australia * Sadrieh Hajesmaeel-Gohari [email protected] 6 2023 20 5 2023 12 2 129135 13 3 2022 05 4 2022 © 2023 The Author(s). 2023 https://creativecommons.org/licenses/by-nc/4.0/ This work is published by Journal of Caring Sciences as an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by-nc/4.0/). Non-commercial uses of the work are permitted, provided the original work is properly cited. http://jcs.tbzmed.ac.ir Introduction: Access to healthcare services for patients with Alzheimer’s disease (AD) was limited during the COVID-19 pandemic. A mobile application (app) can help overcome this limitation for patients and caregivers. Our study aims to develop and evaluate an app to help caregivers of patients with AD during COVID-19. Methods: The study was performed in three steps. First, a questionnaire of features required for the app design was prepared based on the interviews with caregivers of AD patients and neurologists. Then, questionnaire was provided to neurologists, medical informatics, and health information management specialists to identify the final features. Second, the app was designed using the information obtained from the previous phase. Third, the quality of the app and the level of user satisfaction were evaluated using the mobile app rating scale (MARS) and the questionnaire for user interface satisfaction (QUIS), respectively. Results: The number of 41 data elements in four groups (patient’s profile, COVID-19 management and control, AD management and control, and program functions) were identified for designing the app. The quality evaluation of the app based on MARS and user satisfaction evaluation based on QUIS showed the app was good. Conclusion: This is the first study that focused on developing and evaluating a mobile app for assisting Alzheimer’s caregivers during the COVID-19 pandemic. As the app was designed based on users’ needs and covered both information about AD and COVID-19, it can help caregivers perform their tasks more efficiently. Alzheimer’s disease COVID-19 Caregiver Mobile application ==== Body pmcIntroduction Dementia is the loss of memory and other mental capabilities, which is severe enough to affect daily activity. It is caused by physical changes in the brain.1 Alzheimer’s disease (AD) is the most common type of dementia which is characterized by a gradual decrease in mental ability and neurological, psychological, and behavioral disorders.2 This chronic disease decreases the quality of life and imposes an economic burden on the family.3 Currently, about 50 million people worldwide live with AD, and about ten million new cases are reported annually. Although age is the strongest risk factor for dementia, and this disease often occurs in people over 60 years, it is not limited to this age and may affect younger people.4 AD is one of the main causes of death globally and is reported to be the sixth cause of death in the world. AD has significantly impacted the affected person, caregivers, and society.5 Taking care of these people impose huge financial burdens and has many psychological consequences.6-8 Patients with AD depend on caregivers (often family members) to carry out their daily activities. Currently, about 8.9 million family members help these patients as caregivers. The exacerbation of the disease increases the caregivers’ workload and affects their emotional and psychological health.9 Caregivers will spend about 15.3 billion hours caring for people with AD in 2020.6 This number of hours indicates the high workload for caregivers. The management of chronic diseases such as dementia largely depends on the capability of family members to take on and accomplish the role of care. Nowadays, many efforts have been made to reduce the workload and stress of caregivers.10 Mobile phones play an important role in supporting caregivers to perform their daily care activities by providing up-to-date and accessible information.11 Mobile-based applications (apps) are effective for various healthcare tasks such as monitoring symptoms, tracking treatment progress, and training in chronic diseases.12 Studies showed the impact of mobile apps to assist patients with AD. Providing a reminder system in these apps could enable people with low memory to remember all their tasks, which may help prevent the rapid progression of the disease.13,14 These apps also may help manage destructive behavior and improve cognitive function by continuing social interactions.15,16 Some of these apps can assist in decreasing the stress and anxiety of family caregivers by estimating the probability of wandering using geolocation, monitoring the patient in and around the domestic in real-time, as well as helping care administration and services by healthcare providers.17 Today, the death rate of patients with AD has increased in the world with the event of the COVID-19widespread, perhaps due to limitations in access to healthcare services.18 Mobile phone use could overcome limitations in physical communication and in-person access to healthcare services during this widespread.19 To the best knowledge of the authors, there is no study on the development and evaluation of a mobile-based app to help caregivers of patients with AD, especially during COVID-19. Developing a mobile-based app that provides information about all aspects of AD and COVID-19 can reduce the need for face-to-face visits to access healthcare services and assist caregivers in caring for patients appropriately. Therefore, the current study purposes to develop and evaluate a mobile app to assist caregivers of patients with AD during the COVID-19 widespread. Materials and Methods This research was a cross-sectional study with a mixed-method research design performed in three steps: identifying the features needed to design the app, developing the app, and evaluating the app. Identifying the Features In this step, a semi-structured interview was conducted. The questions of this interview were designed based on a review of the literature and the opinion of neurologists and medical informatics experts. Four questions were asked from the participants: What characteristics of the patients should be included in the app? What information about COVID-19 management is required? What information about AD management is required? What other features should this app have? Participants were 30 caregivers and eight neurologists. The interviews, which were performed by one of the authors (P.A.), continued to reach data saturation. Each interview that lasted for 40 min was recorded for the analysis. Informed consent was obtained from the participants to record the interviews. The content of the interviews was transcribed and entered into MAXQDA software. MAXQDA software (version 2018) was used for the qualitative content analysis. Based on the information items obtained from the results of the interviews, a questionnaire consisting of 41 items in four categories (patient profile, seven items; management and control of COVID-19, 11 items; management and control of AD, 14 items; and application functions, nine items) was designed. Each item had three choices: ‘essential’, ‘not essential’, and ‘useful but not essential’. A blank row was used to add critical comments by experts at the end of the structured questionnaire. The content validity of the questionnaire was evaluated by four experts (two health information management and two medical informatics). Cronbach’s alpha was calculated 0.87 for the reliability of the questionnaire. The questionnaire was distributed among four neurologists, three medical informatics specialists, and three health information management specialists to determine the content of the app. The content validity ratio (CVR) was used to analyze the questionnaire. CVR value was calculated for each item and compared with the Lawshe table value.20 According to this table, each item that obtained the CVR of more than 0.62 was accepted and used in the design of the app. App Development In this step, the app was developed based on the gathered data from the previous step. The online app maker named Puzzley (https://puzzley.ir) was used for developing the app. This app maker enables users to create their own Android app without needing programming knowledge, with the least time and cost. Apps that are made with this mobile app maker are not different from other types of apps and can be easily published in app stores. App Evaluation In this step, the quality of the app was evaluated using the mobile app rating scale (MARS) with the participation of two medical informatics specialists in a day. MARS contains 23 items in four sections for objective quality, including engagement (five items), functionality (four items), aesthetics (three items), information quality (seven items), and a section for subjective quality (four items).21 Each item has been designed based on a five-point scale (1=inadequate, 2=poor, 3=acceptable, 4=good, and 5=excellent). For some items, not applicable choice is provided. The mean score of each section was calculated separately. The level of user satisfaction was evaluated using the questionnaire for user interface satisfaction (QUIS). After two weeks of use, fifteen caregivers evaluated the app using QUIS in a week. This questionnaire has 27 items in five sections, including the overall reaction (six items), screen (four items), terminology and information (six items), learning (six items), and general app capabilities (five items). Each item in the QUIS has a 10-point score in the range of 0 to 9.22 The mean score of each section and an overall mean of the app were calculated separately. The total mean scores of 0 to 3, 3.1 to 6, and 6.1 to 9 were considered as weak, medium, and good, separately. The validity and reliability of Persian version of QUIS were confirmed in the previous study.23 Results Identifying the Features The demographic information of the study participants is presented in Table 1. Eight neurologists and 30 caregivers participated in this study with an age range of 35 to 60 and 30 to 70 years old, respectively. The work experience of the neurologists and caregivers was between 5 and 15 and between 6 and 23 years old. Half of the caregivers did not have an academic degree. Table 1 Demographic characteristic of the participants in interview Variable No. (%) Neurologist Age (year) 35-45 5 (62.5) 46-60 3 (37.5) Work experience (year) 5-10 6 (75) 11-15 2 (25) Caregiver Age (year) 30-40 8 (26.7) 41-50 12 (40) 51-60 6 (20) 61-70 4 (13.3) Level of education High school diploma 15 (50) Bachelor’s 10 (33.4) Master’s 5 (16.6) Work experience as a caregiver (year) 6-15 22 (73.3) 16-23 8 (26.7) The interview results were 41 items in four categories, including “patient profile”, “management and control of COVID-19”, “management and control of Alzheimer’s disease” and “application functions”. All the items were considered necessary because the value of CVR was more than 0.62 (Table 2). Table 2 Perspective of specialists on the necessity of care-educational information and features Data elements Specialists’ perspective Patient profile Gender 0.99 Age 0.95 BMI 0.65 Level of education 0.69 Disease stage 0.99 Address 0.99 Caregiver’s relationship with the patient 0.87 Management and control of COVID-19 Definition of COVID-19 0.99 Ways to prevent and deal with COVID-19 0.86 Symptoms of COVID-19 0.88 Side effects of COVID-19 0.80 Early measures in case of COVID-19 0.99 Information on COVID-19 vaccines 0.98 Necessary care after vaccination 0.98 Strategies to control and reduce stress caused by COVID-19 0.97 COVID-19 in patients with AD 0.93 Transfer of COVID-19 from caregivers to an AD 0.86 Reliable news websites related to COVID-19 0.79 Management and control of AD Introduction to AD 0.88 Alzheimer's from onset to the end 0.78 Progression of the disease and the role transformation of the caregiver 0.89 Recommendations in the early stages of AD 0.90 Recommendations in the middle stage of AD 0.89 Recommendations in the final stage of AD 0.90 Eating and drinking, the early stages of AD 0.80 Eating and drinking, the end of the disease 0.96 Behavioral changes 0.97 Pharmacological safety 0.85 Home security 0.86 Other diseases 0.77 Other care matters 0.78 Care of caregivers 0.89 Applications functions Initial diagnosis of the possibility of COVID-19 0.90 Identification of COVID-19 specific medical centers 0.87 Identification of COVID-19 vaccine centers 0.84 Virtual visit 0.98 Reminder for when to visit the doctor 0.99 Reminders for when to take medication 0.99 Identifying high-risk areas 0.97 Searching 0.86 App settings 0.78 App Development Based on the results obtained from the previous phase, the app was developed (Figure 1). The content of the application was in Persian.The language of the application could be changed to Persian or English in the settings. The app size was 29 MB and can be installed on the Android platform. The demographic information about the patient with Alzheimer’s was entered into the patient profile. Medical information and instructions related to COVID-19 were placed for the management and control of COVID-19 based on WHO resources and expert opinions. The necessary information about Alzheimer’s disease for caregivers was placed in the management and control of AD section based on authoritative academic sources and the “Dard Ashna” website.24 Figure 1 App pages: a) Home page, b)COVID-19 diagnosis, c) All COVID-19 specialized medical centers in Iran As shown in Figure 1, in the “Diagnosis of COVID-19” section, by entering the symptoms of COVID-19 such as dry cough, shortness of breath, fever, chills, body temperature, sore throat, and underlying non-AD, the possibility of COVID-19 disease is determined. In the “Virtual Visit” section, active virtual physicians are listed by entering the type of specialization and location. By clicking on each of their names, the telecommunication hours with them can be informed. In the section “COVID-19 Special Medical Centers”, users are informed of all active centers to serve COVID-19 patients. In the section “COVID-19 Vaccine Special Health Centers”, all the centers with their exact addresses are displayed by selecting the city. Due to the rapid spread of this virus, it is better to avoid going to high-risk areas announced by the news. For this reason, in the “Identification of high-risk areas” section, the user is notified of all virus-infected areas by specifying her/his location. Controlling and managing the behavioral changes of patients with AD are among the significant problems for their caregivers. In “the behavioral change management” section, the user can obtain the necessary information to control any behavioral changes. In the “Searching” section, caregivers can search for more educational information about AD and how to care for this type of patient, as well as COVID-19. In “Settings”, caregivers can control how pages are moved, font type, color, and size of the information. App Evaluation The quality and user satisfaction of the app were evaluated. The results of the app quality evaluation are shown in Figure 2. In the objective quality section of MARS, the “Information” dimension attained the highest mean score, and the “aesthetic” dimension attained the lowest mean score. The total mean score of the app was 4.09. Figure 2 Results of the app evaluation with MARS Demographic information of 15 participant caregivers in evaluating user satisfaction is provided in Table 3. Table 3 Participants' demographic information Variable No. (%) Age (year) 30-50 9 (60) 51-70 6 (40) Level of education High school diploma 7 (46.7) Bachelor’s 5 (16.7) Master’s 3 (20) Work experience as caregiver (year) 9-15 8 (53.3) 16-21 7 (46.7) Table 4 reveals the average opinions of caregiver’s in assessing the care-educational app. In all the assessed sections, a mean score of more than six was achieved; thus, the participants generally believed that the app was well. The “Screen” section achieved the highest mean score and the “learning” section had the minimum. Table 4 Results of the app evaluation with QUIS Questions about each section Mean (SD) Overall reaction to the app General use of the app 8.30 (0.11) Ease of use of the app 8.68 (1.51) How the user feels about using the app 8.35 (1.23) General design of the app 8.50 (1.64) Consistent use of the app 7.78 (0.26) Settings feature of the app 7.52 (1.30) Total 8.18 (1.09) Screen Reading characters on the screen 8.91 (0.80) Using clear statements to simplify tasks 8.54 (0.16) Organization of information 7.90 (1.45) Sequence of screens 8.55 (1.20) Total 8.47 (0.90) Terminology and information used in the app Use of terms throughout the system 7.58 (0.25) Task-related terminology 8.20 (1.30) Position of messages on the screen 8.68 (1.82) Prompts for input 7.95 (0.60) App messages to complete user’s tasks 7.84 (1.70) Error messages 8.91 (0.82) Total 8.19 (1.08) Learning Learning to operate the system 7.96 (1.22) Exploring new features by trial and error 8.99 (0.55) Remembering names and use of commands 8.65 (0.21) Straightforward task performance 7.43 (0.11) Help messages on the screen 8.25 (1.90) Supplemental reference materials 7.60 (1.10) Total 8.15 (0.85) App capabilities App speed 8.12 (0.71) System reliability 8.53 (0.25) Number of app specifications 8.90 (1.40) Correcting user’s mistakes when inputting data 7.91 (1.13) Designed for all levels of users 8.26 (0.42) Total 8.34 (0.78) Total 8.26 (0.92) Discussion This study focused on developing and evaluating a mobile app for assisting Alzheimer’s caregivers during the COVID-19 pandemic. This app can be used by the Alzheimer’s patients in the first stage of the disease and all the caregivers of Alzheimer’s patients at every stage. The app was designed in four sections, including “patient profile”, “management and control of COVID-19”, “management and control of AD”, and “application functions.” Among different sections of the objective quality in MARS, the “information” section obtained the highest, and the “aesthetic” section obtained the lowest mean scores. Among different sections of QUIS, the “screen” section achieved the highest mean score, and the “learning” section gained the lowest mean score. Identifying the Features and App Developing The app was designed in four sections, including “patient profile,” “management and control of COVID-19”, “management and control of AD”, and “application functions.” Similar to our study, other studies have shown that the information needs of patients with dementia and their caregivers are about characteristics of the disease, disease management, and self-care of caregivers.25However,during the COVID-19pandemic, the needs of caregivers of dementia patients have become broader and have been included protecting the patients from COVID-19 infection, managing the patients if they were infected, managing the changes in the daily activities of patients, managing comorbidities, diet management, and managing behavioral changes.26 On the other hand, the content review of the mobile apps for dementia showed that most of the apps were educational to increase awareness about dementia. Additionally, these apps presented practical caregiving information to increase the quality of life and support caregivers.27 Another study demonstrated that the apps designed for AD mainly have information about caregiving and disease management. Only a few apps offer prevention, early detection, disease monitoring, financial and legal issues, and organization promotion.28 Apps designed only for AD caregivers have features that mainly include tracking patients, daily task management, and monitoring patients and their surrounding environment. Besides, some apps provide mental support for caregivers, educational information, and a platform for communication of caregivers with each other.29 We used the online app maker named Puzzley for developing the app. Other studies used different programming languages to design the mobile health apps.30,31 However, Puzzley enables users to create their Android app quickly without programming knowledge. App Evaluation The results of app evaluation using MARS showed that the app had higher quality in terms of information and lower quality in terms of aesthetics. Compared to our study, other studies that have reviewed and evaluated the quality of apps in dementia27 and AD28 using MARS have revealed that apps have higher quality in functionality and lower quality in engagement. These differences may be because the two mentioned studies have evaluated several apps, while our study evaluated only a designed app. The information used in the app that we designed was gathered from standard and reliable resources and proved by Alzheimer’s specialists.24 Since we tried to consider the users’ needs in designing the app, the engagement section could gain an acceptable mean score compared to other studies. Aesthetic criteria include graphics, layout, and visual appeal. Our app could not get a high mean score in terms of aesthetics. Since aesthetics is one of the factors in the acceptability and continuity use of a mobile app,32 it should be considered and improved for this app. The results of app evaluation using QUIS showed that users had the highest satisfaction with the screen and the lowest satisfaction with the learning capabilities of the app. Similar to this study, a work that evaluated a self-care app for multiple sclerosis with QUIS showed that the “screen” and “learning” sections achieved the highest and lowest mean scores, respectively.30 As a usability principle, information required to use the app should be noticeable and accessible when needed, and users should not have to remember information for using different parts of the app. Moreover, the help section should be provided in an app design to support users in completing their tasks efficiently.33 Since these principles affect users’ satisfaction in using the app, they should be considered in the app design. Moreover, the use of some principles, such as Myer’s multimedia principles that focus on several points in designing educational multimedia, can affect more involvement and improve learning.34 This study has a limitation. The app development was performed only on the Android operating system. The Android operating system has covered most of the mobile operating system’s market.35 Nevertheless, to be more popular, it would be better to develop the app on iOS as well. Conclusion To the best knowledge of the authors, this study was the first one that focused on the development and evaluation of a mobile app for assisting Alzheimer’s caregivers during the COVID-19 pandemic. Since this app was designed based on user’s needs and covered both information about Alzheimer’s and COVID-19, it can be helpful for caregivers in terms of doing their tasks more efficiently. The evaluation results of the designed app also showed that this app was at a good level in terms of quality and user satisfaction. Acknowledgements We thank all the experts and caregivers that helped us and contributed to this study. Authors’ Contribution Conceptualization: Parastoo Amiri, Sadrieh Hajesmaeel-Gohari, Kambiz Bahaadinbeigy. Data curation: Parastoo Amiri, Sadrieh Hajesmaeel-Gohari, Maryam Gholipour. Formal analysis: Parastoo Amiri, Sadrieh Hajesmaeel-Gohari, Maryam Gholipour. Funding acquisition: Sadrieh Hajesmaeel-Gohari. Investigation: Parastoo Amiri, Maryam Gholipour. Methodology: Parastoo Amiri, Sadrieh Hajesmaeel-Gohari, Maryam Gholipour. Project administration: Sadrieh Hajesmaeel-Gohari, Kambiz Bahaadinbeigy. Resources: Parastoo Amiri. Software: Parastoo Amiri. Supervision: Sadrieh Hajesmaeel-Gohari, Kambiz Bahaadinbeigy. Validation: Parastoo Amiri, Sadrieh Hajesmaeel-Gohari. Visualization: Parastoo Amiri, Maryam Gholipour. Writing–original draft: Parastoo Amiri, Sadrieh Hajesmaeel-Gohari, Maryam Gholipour. Writing–review & editing: Parastoo Amiri, Sadrieh Hajesmaeel-Gohari, Maryam Gholipour, Kambiz Bahaadinbeigy. COI-statement The authors declare that there is no conflict of interest. Data Availability Not applicable. Ethical Approval The Ethics Committee of Kerman University of Medical Sciences approved this research with ethical code IR.KMU.REC.1400.381. Funding This study was funded by the Kerman University of Medical Sciences with research ID 400000546. Research Highlights What is the current knowledge? AD is the common dementia disease. Access to healthcare services for patients with AD was limited during the COVID-19 pandemic. Developing a mobile application (app) can help overcome this limitation for the patients and caregivers. What is new here? The mobile app was designed based on user’s needs and covered both information about AD and COVID-19. ==== Refs References 1 Bliss RM Wood M Food discoveries for brain fitness Agric Res 2010 58 6 18 21 2 Cummings JL Koumaras B Chen M Mirski D Effects of rivastigmine treatment on the neuropsychiatric and behavioral disturbances of nursing home residents with moderate to severe probable Alzheimer’s disease: a 26-week, multicenter, open-label study Am J Geriatr Pharmacother 2005 3 3 137 48 10.1016/s1543-5946(05)80020-0 16257816 3 Rice DP Fox PJ Max W Webber PA Lindeman DA Hauck WW The economic burden of Alzheimer’s disease care Health Aff (Millwood) 1993 12 2 164 76 10.1377/hlthaff.12.2.164 8375811 4 World Health Organization (WHO). Dementia. United States: WHO; 2022. Available from: https://www.who.int/news-room/fact-sheets/detail/dementia. Accessed December 20, 2022. 5 Qiu C Kivipelto M von Strauss E Epidemiology of Alzheimer’s disease: occurrence, determinants, and strategies toward intervention Dialogues Clin Neurosci 2009 11 2 111 28 10.31887/DCNS.2009.11.2/cqiu 19585947 6 Mecca AP van Dyck CH Alzheimer’s & dementia: the journal of the Alzheimer’s association Alzheimers Dement 2021 17 2 316 7 10.1002/alz.12190 33047474 7 Dillehay RC Sandys MR Caregivers for Alzheimer’s patients: what we are learning from research Int J Aging Hum Dev 1990 30 4 263 85 10.2190/2p3j-a9ah-hhf4-00rg 2204604 8 Zhu CW Ornstein KA Cosentino S Gu Y Andrews H Stern Y Medicaid contributes substantial costs to dementia care in an ethnically diverse community J Gerontol B Psychol Sci Soc Sci 2020 75 7 1527 37 10.1093/geronb/gbz108 31425587 9 Sansoni J Anderson KH Varona LM Varela G Caregivers of Alzheimer’s patients and factors influencing institutionalization of loved ones: some considerations on existing literature Ann Ig 2013 25 3 235 46 10.7416/ai.2013.1926 23598807 10 Hepburn KW Tornatore J Center B Ostwald SW Dementia family caregiver training: affecting beliefs about caregiving and caregiver outcomes J Am Geriatr Soc 2001 49 4 450 7 10.1046/j.1532-5415.2001.49090.x 11347790 11 Gupta G, Gupta A, Jaiswal V, Ansari MD. A review and analysis of mobile health applications for Alzheimer patients and caregivers. In: 2018 Fifth International Conference on Parallel, Distributed and Grid Computing (PDGC). Solan, India: IEEE; 2018. p. 171-5. 10.1109/pdgc.2018.8745995 12 Guo Y Yang F Hu F Li W Ruggiano N Lee HY Existing mobile phone apps for self-care management of people with Alzheimer disease and related dementias: systematic analysis JMIR Aging 2020 3 1 e15290 10.2196/15290 32012045 13 Chen H Soh Y A cooking assistance system for patients with Alzheimers disease using reinforcement learning Int J Inf Technol 2017 23 2 1 12 14 Habash ZA, Hussain W, Ishak W, Omar MH. Android-based application to assist doctor with Alzheimer’s patient. In: Proceedings of the 4th International Conference on Computing and Informatics. ICOCI; 2013. 15 Désormeaux-Moreau M Michel CM Vallières M Racine M Poulin-Paquet M Lacasse D Mobile apps to support family caregivers of people with Alzheimer disease and related dementias in managing disruptive behaviors: qualitative study with users embedded in a scoping review JMIR Aging 2021 4 2 e21808 10.2196/21808 33861207 16 Brown EL Ruggiano N Li J Clarke PJ Kay ES Hristidis V Smartphone-based health technologies for dementia care: opportunities, challenges, and current practices J Appl Gerontol 2019 38 1 73 91 10.1177/0733464817723088 28774215 17 Moreira H, Oliveira R, Flores N. STAlz: remotely supporting the diagnosis, tracking and rehabilitation of patients with Alzheimer’s. In: 2013 IEEE 15th International Conference on e-Health Networking, Applications and Services (Healthcom 2013). Lisbon, Portugal: IEEE; 2013. 10.1109/HealthCom.2013.6720743 18 Murray ML Access to health services to reduce morbidity and mortality Int J Childbirth 2019 8 4 212 5 10.1891/2156-5287.8.4.212 19 Torous J Keshavan M COVID-19, mobile health and serious mental illness Schizophr Res 2020 218 36 7 10.1016/j.schres.2020.04.013 32327314 20 Lawshe CH A quantitative approach to content validity Pers Psychol 1975 28 4 563 75 21 Stoyanov SR Hides L Kavanagh DJ Zelenko O Tjondronegoro D Mani M Mobile app rating scale: a new tool for assessing the quality of health mobile apps JMIR Mhealth Uhealth 2015 3 1 e27 10.2196/mhealth.3422 25760773 22 Chin JP, Diehl VA, Norman KL. Development of an instrument measuring user satisfaction of the human-computer interface. In: Proceedings of the SIGCHI Conference on Human Factors in Computing Systems. Association for Computing Machinery (ACM); 1988. p. 213-8. 10.1145/57167.57203 23 Ayatollahi H Hasannezhad M Saneei Fard H Kamkar Haghighi M Type 1 diabetes self-management: developing a web-based telemedicine application Health Inf Manag 2016 45 1 16 26 10.1177/1833358316639456 28691565 24 Dardashna. 2021. Available from: https://dardashna.ir/. Accessed September 18, 2021. 25 Soong A Au ST Kyaw BM Theng YL Tudor Car L Information needs and information seeking behaviour of people with dementia and their non-professional caregivers: a scoping review BMC Geriatr 2020 20 1 61 10.1186/s12877-020-1454-y 32059648 26 Vaitheswaran S Lakshminarayanan M Ramanujam V Sargunan S Venkatesan S Experiences and needs of caregivers of persons with dementia in India during the COVID-19 pandemic-a qualitative study Am J Geriatr Psychiatry 2020 28 11 1185 94 10.1016/j.jagp.2020.06.026 32736918 27 Chelberg GR Neuhaus M Mothershaw A Mahoney R Caffery LJ Mobile apps for dementia awareness, support, and prevention - review and evaluation Disabil Rehabil 2022 44 17 4909 20 10.1080/09638288.2021.1914755 34034601 28 Choi SK Yelton B Ezeanya VK Kannaley K Friedman DB Review of the content and quality of mobile applications about Alzheimer’s disease and related dementias J Appl Gerontol 2020 39 6 601 8 10.1177/0733464818790187 30049239 29 Kim E Baskys A Law AV Roosan MR Li Y Roosan D Scoping review: the empowerment of Alzheimer’s disease caregivers with mHealth applications NPJ Digit Med 2021 4 1 131 10.1038/s41746-021-00506-4 34493819 30 Mokhberdezfuli M Ayatollahi H Naser Moghadasi A A smartphone-based application for self-management in multiple sclerosis J Healthc Eng 2021 2021 6749951 10.1155/2021/6749951 34221301 31 Ghazisaeedi M, Sheikhtaheri A, ‏Dalvand‎ H, Safari A. Design and evaluation of an applied educational smartphone-based ‎program for caregivers of children with cerebral palsy. J Clin Res Paramed Sci 2015; 4(2): 128-39. [Persian]. 32 Malik A Suresh S Sharma S Factors influencing consumers’ attitude towards adoption and continuous use of mobile applications: a conceptual model Procedia Comput Sci 2017 122 106 13 10.1016/j.procs.2017.11.348 33 Nielsen J. 10 Usability Heuristics for User Interface Design. USA: Nielsen Norman Group; 2020. Available from: https://www.nngroup.com/articles/ten-usability-heuristics/. Accessed December 20, 2022. 34 Mayer RE Moreno R A cognitive theory of multimedia learning: implications for design principles J Educ Psychol 1998 91 2 358 68 35 Statista. Mobile Operating Systems’ Market Share Worldwide from 1st Quarter 2009 to 4th Quarter 2022. United States: Statista; 2022. Available from: https://www.statista.com/statistics/272698/global-market-share-held-by-mobile-operating-systems-since-2009. Accessed December 20, 2022.
PMC010xxxxxx/PMC10352639.txt
==== Front J Caring Sci J Caring Sci J Caring Sci TBZMED Journal of Caring Sciences 2251-9920 2251-9920 Tabriz University of Medical Sciences 10.34172/jcs.2023.30120 Original Article The Effect of Aromatherapy with Citrus aurantium Aroma on Pain after Orthopedic Surgery: A Randomized Clinical Trial https://orcid.org/0000-0001-9056-8414 Bargi Sepideh 1 Methodology Project administration Validation Visualization Writing – original draft Writing – review & editing https://orcid.org/0000-0002-6447-186X Bahraminejad Nasrin 2 *Funding acquisition Investigation Methodology Project administration Supervision Validation Visualization Writing – original draft Writing – review & editing https://orcid.org/0000-0002-9087-2341 Jafari Samineh 3 Methodology Validation Visualization Writing – original draft Writing – review & editing https://orcid.org/0000-0002-6404-9044 Fallah Ramezan 4 Methodology Project administration Validation Visualization Writing – original draft Writing – review & editing 1Department of Nursing, School of Nursing and Midwifery, Zanjan University of Medical Sciences, Zanjan, Iran 2Social Determinant of Health Research Center, School of Nursing and Midwifery, Zanjan University of Medical Science, Zanjan, Iran 3Department of Pharmacognosy, School of Pharmacy, Zanjan University of Medical Sciences, Zanjan, Iran 4Department of Biostatistics and Epidemiology, School of Medicine, Zanjan University of Medical Sciences, Zanjan, Iran * Nasrin Bahraminejad [email protected] 6 2023 06 5 2023 12 2 116122 14 12 2021 14 7 2022 © 2023 The Author(s). 2023 https://creativecommons.org/licenses/by-nc/4.0/ This work is published by Journal of Caring Sciences as an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by-nc/4.0/). Non-commercial uses of the work are permitted, provided the original work is properly cited. http://jcs.tbzmed.ac.ir Introduction: Postoperative pain is one of the most common physiological and psychological stress in patients that disrupts body function and can endanger patients’ health. This study aims to determine the effect of aromatherapy with Citrus aurantium essential on pain after orthopedic surgery. Methods: This randomized clinical trial was performed on 60 candidates for orthopedic surgery. Patients were selected through convenience sampling and divided into intervention and control groups through randomized block allocation. If the visual analogue scale (VAS) score was above 3, patients in the intervention group received aromatherapy with C. aurantium essential and the patients in the control group received a placebo (almond oil). VAS was used to measure pain. Data analysis was performed using independent t test, paired t test, and analysis of variance with repeated measures using SPSS software version 13. Results: Mean (SD) of pain intensity after intervention in experimental and control groups within 4, 8, and 12 hours after surgery was 7.30 (1.23) vs. 7.90 (0.99), 5.30 (0.98) versus 5.53 (0.68) and 2.53 (0.9) vs. 3.60 (0.77) respectively. The findings indicated that there was a significant difference in mean pain intensity between the experimental and control groups at 4 and 12 hours after surgery. Use of analysis of variance with repeated measures test with taking into account the interaction of time and group also showed a significant difference in mean pain intensity between the two experimental and control groups. Conclusion: Aromatherapy with Citrus aurantium essential can be effective in reducing mild to moderate pain after orthopedic surgery. Further studies are recommended to confirm this finding. Pain Aromatherapy Orthopedic surgery Citrus aurantium ==== Body pmcIntroduction The management of acute postoperative pain remains a significant challenge for physicians and nurses, especially in low- and middle-income countries.1 Postoperative pain can cause physiological and psychological complications such as fear, anxiety and feelings of helplessness.2 Failure to control postoperative pain can activate the sympathetic nervous system and increase myocardial work and oxygen demand.3 Thereby contributing to ischemia, myocardial infarction, increasing morbidity and mortality of the patients. Ineffective postoperative pain control can also lead to economic and medical problems, including the increase in hospitalization time, the need for hospital re-admission, and patients’ dissatisfaction with medical care.4 Therefore, assessment of pain intensity after surgery, its effective control and treating pain complications are of great importance. Despite new advance in measures have been taken to control postoperative pain but the pain is an unresolved health-care challenge after surgery.5Based mostly on in hospital evidence, 86% of patients experience pain after surgery.6In non-western countries data are scared but prevalence goes up 73% to 96.66 %, which was shown in Tanzania and India.7,8In a study by Hadavi et al 40% of patients were dissatisfied with postoperative pain control in Iran.9 Orthopedic surgery results in moderate to severe pain in a majority of patients.10 Dissatisfaction with pain management in patients undergoing orthopedic surgery is common.11 Whiles, patient satisfaction is a valuable criterion in health care outcomes process and is used to improve the provided quality of care.12 According to the standards of the Commission for the Approval of Health Services Providers, pain is considered to be the fifth vital sign and should be assessed regularly from the time of admission to the discharge of the patients.13 Pharmacological interventions used to relieve postoperative pain are mainly focused on the prescription of opiate and non-steroidal anti-inflammatory drug (NSAID).14 Opioids are used as the first line of treatment for postoperative pain relief.15 Morphine is a common opiate pain reliever used to treat pain.16 Overall, systemic opioid use has been associated with complications such as nausea, vomiting, constipation, itching, and respiratory depression.17 NSAIDs can also cause skin reactions, renal complications including analgesic nephropathy and gastrointestinal complications such as peptic ulcers.18 In the last two decades, concerns about narcotics complications,19 the inability to optimally relieve pain5, increase in medical costs, and the length of hospital stay20 have led to the use of non-pharmacological approaches including music,21massage,22and aromatherapy.23 The effect of different fragrances on pain has been studied by different researchers.24,25 One of the aromatic volatile oils that is widely used in aromatherapy, is oil extracted from Citrus aurantium plant. C. aurantium is native to tropical Asia but it also grows in all tropical and subtropical regions.26 The constituents of C. aurantium essential oil are more than ten, most of which are linalool, neryl acetate, limonen, beta-pinene, myrcene and alpha-terpinylacetate.27 The myrcene is a monoterpene composition and has analgesic, sodium channel blocking and muscle relaxant effects.28 Limonene in C. aurantium essential oil inhibits the activity of prostaglandins by controlling cyclooxygenase I and II and thus is effective in reducing pain.29 No specific side effects related to the use of C. aurantium essential oil has been reported through studies.30 Namazi et al through an experimental study used the effect of C. aurantium essential oil on pain intensity of the active phase of labor in nulliparous women. The results of the study showed that mean pain was significantly decreased in the group treated with C. aurantium essential oil compared to the control group.31 In another study, Sharifipour et al used the effect of C. aurantium essential oil on anxiety after the cesarean section on 80 women undergoing surgery. The results of this study also showed a significant decrease in anxiety in the intervention group compared with the control group.23 A study conducted by Chen and Xie in China also showed that postoperative aromatherapy with C. aurantium was effective in relieving pain in patients undergoing gastrectomy.32So considering the importance of postoperative pain relief and easy and cost-effective availability of C. aurantium essential oil and few studies about the effect of C. aurantium essential oil on postoperative acute pain, this study was developed to investigate the effect of aromatherapy with C. aurantium aroma on pain after orthopedic surgery. Materials and Methods The current study was a double-blind clinical trial (registration code: IRCT20140304016843N13). The study was carried out on 60 patients undergoing elective orthopedic lower limb surgery (femur or tibia) referring to Ayatollah Mousavi Hospital, Zanjan, Iran between October 29, 2018, and February 9, 2019). The sample size was determined according to a confidence interval of 95% and test power of 80% and according to a similar study33 and considering the probability of sample loss in the experimental group (n = 30) and control group (n=30). The patients chosen by convenience sampling were divided by random allocation method into experimental and control groups. The sequence of sample allocation in the two groups was determined as follows: First, the letters A and B were assigned to each of the two groups under study, and then the size of the blocks (4, twice the number of groups) was determined. To avoid bias - the choice of the block size was not mentioned with the aid of research assistants. After itemizing approximately all modes of 4 blocks (AABB, ABAB, BABA, BBAA, ABBA, BAAB) and assigning a number to each one, based on a random number table, 15 blocks were selected to increase the number of samples to 60 patients (30 patients in the experimental group, 30 patients in the control group). It should be noted that the rooms of patients treated with the aroma of orange spring were separated from the rooms of patients treated with a placebo (sweet almond oil) in order to avoid mixing the aroma of spring orange and sweet almond oil, which inadvertently leads to contamination of samples in the control group. Blinding was performed for patients who participate in the study and the researcher assistant who performed the intervention in the study, namely patients in the experimental and control groups were in separate rooms without any contact, and C. aurantium was introduced to them as drug A and almond oil as drug B. Though, the fragrance of C. aurantium may have been familiar to them. But none of the patients knew they were in the experimental or control group. Also, for the researcher assistant who performed the intervention, C. aurantium was again introduced as drug A and almond oil as drug B. To avoid researcher bias, pain measurement after the intervention was performed with the help of another researcher assistant. Those patients who met all the inclusion criteria were included in this study. The inclusion criteria were as follow: having orthopedic surgery, no surgical history, full alertness, and postoperative co-operation, self-reported a normal sense of smell and visual state to see the visual analogue scale (VAS) for pain intensity; being 15 years old and over; having no history of herbal allergies, psychological disorder, coagulation disorders, diabetes, and respiratory problems, and addiction according to the patient record. Exclusion criteria included having postoperative complications (such as hemorrhage, and hematoma at the site of operation), the need for postoperative oxygen therapy, and the patient’s unwillingness to participate after performing the first intervention. Before surgery, patients selected for the experimental group was evaluated for allergy to C. aurantium. To do this, a drop of C. aurantium essential oil was poured into the patient’s wrist, and it was immediately dressed to reduce inhalation. After two minutes the dressing was removed. None of the participants showed allergic reactions. During the intervention phase, in the experimental group, if the patients who underwent lower orthopedic surgery, their pain score by using the VAS instrument was more than 3 after 4,8,12 hours after surgery, 4 drops of C. aurantium essential oil was poured on a cotton ball and the patient was asked to inhale it for 5 minutes at a distance of 20 cm. Then, after 20 minutes, the pain intensity was measured again on the VAS. In the control group, similar to the intervention group at 4, 8 12 hours after surgery, the severity of pain was assessed using the VAS instrument. If their postoperative pain score was more than 3, they were treated with sweet almond oil and then 20 minutes later, similar to the experimental group; their pain intensity was assessed and recorded. Patients in both groups also received routine medications to relieve pain. C. aurantium essential oil 10% and sweet almond oil (as a placebo) were supplied from the Ayat essence company, Iran. The amount of essential oil was based on a literature review34 and consultation with an herbalist. The instrument used for data collection comprises two parts: The first part consists of demographic data and contextual variables such as age, sex, marital status, educational level, occupation, place of residence, and underlying diseases such as sinusitis and allergy was completed by interview. The second part consisted of the VAS tool. Studies in Iran and outside Iran confirmed the validity and reliability of the VAS tool.35,36 Analyzing data was carried out by employing SPSS version 13. An independent t-test was used to compare the pain intensity between the experimental and control groups before the intervention. An analysis of variance with repeated measures was used to compare the pain intensity before and after the intervention in both experimental and control groups (during three stages of intervention). Generalized estimating equations models were used to evaluate the effect of intervention time and the interaction effect of intervention time on pain intensity changes in the experimental and control groups. The level of statistical significance was set at P<0.05) (Figure 1). Figure 1 Flow chart of the study Results The mean (SD) age of the participants in the experimental and control groups was 38.73 and 42.83 years, respectively. Most of the patients were married in the experimental group 60 (18) and in the control group 70 (21). The majority of patients in the experimental group 70 (21) and the control group 63.3 (19) had no history of hospitalization. In addition, concerning the type of surgery, the majority 83.3 (25) underwent surgery in the femur area. As Table 1 shows, participants in the experimental and control groups did not have significant differences in terms of contextual and demographic variables. According to the results the mean of pain in both experimental and control groups before intervention in each the three stages of 4, 8, and 12 hours after the operation was not significant. However, the mean of pain after intervention in all three stages decreased in the experimental group compared to the control group. So that, the mean (SD) of pain intensity in the first 4 hours after surgery in the experimental group compared to the control group was 7.30 (1.23) vs. 7.90 (0.99), at 8 hours after surgery was 5.30 (0.98) versus 5.53 (0.68) and at 12 hours after surgery was 2.53 (0.9) vs. 3.60 (0.77). Moreover, there was a significant difference in mean pain intensity between the two groups at 4 and 12 hours after surgery (Table 2). Use of repeated measurement test with taking into account the interaction of time and group also showed a significant difference in mean pain intensity between two experimental and control groups. Table 1 Demographic and contextual variable of participant in the experimental and control group Variables Mean (SD) P value a Experimental group (n=30) Control group (n=30) Habitat Urban 22 (73.7) 19 (63.3) 0.58 Rural 8 (26.7) 11 (36.7) Surgical site Femur 25 (83.3) 5 (16.7) 0.99 Tibia 25 (83.3) 5 (16.7) Health insurance Yes 24 (80) 26 (86.7) 0.73 No 6 (20) 4 (13.3) Income Relatively adequate 8 (26.7) 7 (23.5) 0.99 Inadequate 22 (73.3) 23 (76.7) Surgeon A 23 (76.7) 22 (73.3) 0.76 B 7 (23.3) 8 (26.7) History of hospitalization Yes 9 (30) 11 (36.7) 0.58 No 21 (70) 19 (63.3) Type of analgesic Morphine 10 (33.3) 12 (40) 0.59 Pethidine 20 (66.7) 18 (60) aChi- squared test. Table 2 The mean and standard deviation in experimental and control groups in terms of time measurement Time Group Group Time Hour Experimental group (n=30) Control group (n=30) Before intervention After intervention Before intervention After intervention Mean (SD) CV% Mean (SD) CV% Mean (SD) CV% Mean (SD) CV% F Df P F Df P F Df P 4 hours after surgery 8.23(0.89) 10.9 7.30(1.23) 16.9 8.10(1.09) 13.5 7.90(0.99) 12.6 8.15 1 0.005 56.84 1 0.001 61.70 2 0.001 8hours after surgery 5.77(0.81) 14.2 5.30(0.98) 18.6 5.7(0.70) 12.3 5.53(0.68) 12.3 12 hours after surgery 4.40(0.72) 16.5 2.53(0.90) 35.5 4.37(0.49) 11.2 3.60(0.77) 12.4 Discussion This study aimed to evaluate the effect of aromatherapy with C. aurantium on postoperative orthopedic (lower limb) pain. The findings of the study indicated that aromatherapy with C. aurantium had a significant effect on postoperative pain reduction of the patients. Of course according to the findings, the highest reduction in pain score was related to 12 hours after surgery (1.87 in the experimental group compared to 0.77 in control group). Based on these findings, the use of aromatherapy with C. aurantium seems to be more effective when the pain intensity is mild and moderate. The study by Sharifipour et al on the effects of aromatherapy with C. aurantium and Salvia officinalis oil on the pain after cesarean section at 4, 8 and 12 hours after surgery showed that aromatherapy with both C. aurantium and S. officinalis had significant and equal effects on pain relief after cesarean section which is in line with the results of the present study.34 Namazi et al study on the effect of C. aurantium on pain intensity of active phase of labor also showed that the use of C. aurantium essential oils in women reduced the intensity of labor pain at different stages of labor.31 Yip and Tam in their study showed that massage with combined aromatherapy of C. aurantium and ginger essential oils was effective in relieving moderate to intensive knee pain.37 The antinociceptive activities of C. aurantium can be explained in two ways. First, the aroma-induced odor appears to stimulate the olfactory nerve cells and subsequently the limbic system. Depending on the type of aroma, the neurons release different neurotransmitters. These neurotransmitters include enkephalin, endorphin, noradrenaline, and serotonin which are capable of altering the feelings in humans through odors. On the other hand, myrcene present in C. aurantium is a monoterpene compound which is antinociceptive, sodium channel blocking and muscle relaxant. Also, limonene in C. aurantium essential oil restrains the activity of prostaglandins by inhibiting cyclooxygenase I and II and by this way, it is effective in reducing pain.28,29 As the findings showed, the pain reduction was more significant in the experimental group but the changes in pain intensity in the control group were also significant at 12 hours after surgery. One of the causes of pain reduction in the control group can be due to the passage of time. On the other hand, pain reduction in the control group, including the effects of placebo on pain relief, is due to the role of psychological factors in pain relief. However, the present study confirms the results of previous studies on the effect of C. aurantium on acute pain.32,34 Based on the definition of acceptable effect size to consider the anti-nociceptive effect of any treatment that is equivalent to a 30% reduction in pain severity,38 it can be said, though the use of C. aurantium caused a significant difference in pain intensity in the experimental group compared to the control group, but the anti-nociceptive effects were not clinically significant in the first 4 and 8 hours postoperatively, and the most anti-nociceptive effect occurred at 12 hours postoperatively (when the patient’s pain was mild to moderate). Therefore, due to the acceptable size effect on pain relief, the therapeutic effects of C. aurantium especially during the first 4 and 8 hours after surgery were not significant. The therapeutic methods of anti-nociceptive effects of C. aurantium especially in the first 4 and 8 hours after the operation were not significant and it can be said that aromatherapy with C. aurantium has been used to reduce pain along with other medication interventions, and most when the intensity of pain is at a moderate or moderate level. One of the advantages of this study is that C. aurantium has public acceptance, and another advantage compare with similar studies is that the current study determined the pain intensity level affected by C. aurantium. While other studies nearly deems general nociceptive effect of C. aurantium. Considering the impact of gender (participants were only men) and age on pain experience, the results of this study cannot be generalized to all types of acute pain, female gender, all age groups, and cultures. Further studies on the anti-nociceptive effects of C. aurantium in other patients and chronic pain and cancer pain are recommended. Conclusion Aromatherapy with C. aurantium aroma had positive effect on reducing orthopedic postoperative pain mainly in mild to moderate pain. Therefore, this intervention can be applied by nurses to decrease postoperative pain along with other medication interventions. Acknowledgments The present study is based on a master thesis in nursing. We would like to express our deep sense of gratitude to the Vice Chancellor for Research and Technology of Zanjan University of Medical Sciences, the staff of Ayatollah Mousavi Hospital and the participants in this study. Author’s Contribution Conceptualization: Sepideh Bargi, Nasrin Bahraminejad, Samineh Jafari, Ramezan Fallah. Data curation: Sepideh Bargi. Formal analysis: Sepideh Bargi, Nasrin Bahraminejad, Ramezan Fallah. Funding acquisition: Nasrin Bahraminejad. Investigation: Nasrin Bahraminejad. Methodology: Sepideh Bargi, Nasrin Bahraminejad, Samineh Jafari, Ramezan Fallah. Project administration: Nasrin Bahraminejad Sepideh Bargi. Software: Ramezan Fallah. Supervision: Nasrin Bahraminejad, Samineh Jafari. Validation: Sepideh Bargi, Nasrin Bahraminejad, Samineh Jafari, Ramezan Fallah. Visualization: Sepideh Bargi, Nasrin Bahraminejad, Samineh Jafari, Ramezan Fallah. Writing–original draft: Sepideh Bargi, Nasrin Bahraminejad, Samineh Jafari, Ramezan Fallah. Writing–review & editing: Sepideh Bargi, Nasrin Bahraminejad, Samineh Jafari, Ramezan Fallah. COI-statement The authors declared no conflict of interest in this study. Data Availability The datasets are available from the corresponding author on reasonable request. Ethical Approval The present study has been extracted from master degree dissertation. The ethical permission for the study was obtained by the Ethics Committee of the Zanjan University of Medical Sciences (ethics code: IR.ZUMS.REC.1397.14). In this way, the researcher first provided the participants with the necessary information about the study, its aims, the right to leave the study at any stage of the intervention, and the confidentiality of their personal data, and after they met the inclusion criteria, written informed consent was obtained from them before initiation of the study. Funding This study was funded by Zanjan University of Medical Sciences, Zanjan, Iran. Research Highlights What is the current knowledge? Aromatherapy is one of the complementary therapy. Aromatherapy has been applied to alleviate pain via olfactory stimulation as well as changing physiological parameters. What is new here? Aromatherapy with the aroma of Citrus aurantium was effective in reducing postoperative pain. To improve postoperative pain management, aromatherapy with Citrus aurantium can be used along with drug treatments after orthopedic surgery. ==== Refs References 1 Eshete MT Baeumler PI Siebeck M Tesfaye M Wonde D Haileamlak A The views of patients, healthcare professionals and hospital officials on barriers to and facilitators of quality pain management in Ethiopian hospitals: a qualitative study PLoS One 2019 14 3 e0213644 10.1371/journal.pone.0213644 30870467 2 Baratta JL Schwenk ES Viscusi ER Clinical consequences of inadequate pain relief: barriers to optimal pain management Plast Reconstr Surg 2014 134 4 Suppl 2 15S 21S 10.1097/prs.0000000000000681 25254999 3 Ledowski T Reimer M Chavez V Kapoor V Wenk M Effects of acute postoperative pain on catecholamine plasma levels, hemodynamic parameters, and cardiac autonomic control Pain 2012 153 4 759 64 10.1016/j.pain.2011.11.002 22305628 4 Garimella V Cellini C Postoperative pain control Clin Colon Rectal Surg 2013 26 3 191 6 10.1055/s-0033-1351138 24436674 5 Gan TJ Poorly controlled postoperative pain: prevalence, consequences, and prevention J Pain Res 2017 10 2287 98 10.2147/jpr.s144066 29026331 6 Gan TJ Habib AS Miller TE White W Apfelbaum JL Incidence, patient satisfaction, and perceptions of post-surgical pain: results from a US national survey Curr Med Res Opin 2014 30 1 149 60 10.1185/03007995.2013.860019 24237004 7 Ndebea AS van den Heuvel SAS Temu R Kaino MM van Boekel RLM Steegers MAH Prevalence and risk factors for acute postoperative pain after elective orthopedic and general surgery at a tertiary referral hospital in Tanzania J Pain Res 2020 13 3005 11 10.2147/jpr.s258954 33244259 8 Singh PK Saikia P Lahakar M Prevalence of acute post-operative pain in patients in adult age-group undergoing inpatient abdominal surgery and correlation of intensity of pain and satisfaction with analgesic management: a cross-sectional single institute-based study Indian J Anaesth 2016 60 10 737 43 10.4103/0019-5049.191686 27761037 9 Hadavi M, Hosseinkhani Z, Saghafi Z. Patient satisfaction from postoperative pain relief in Ali-Ebne-Abitaleb hospital in Rafsanjan - 2012. Community Health Journal 2017; 7(1): 1-8. [Persian]. 10 Maheshwari AV Blum YC Shekhar L Ranawat AS Ranawat CS Multimodal pain management after total hip and knee arthroplasty at the Ranawat Orthopaedic Center Clin Orthop Relat Res 2009 467 6 1418 23 10.1007/s11999-009-0728-7 19214642 11 Jacobs CA Christensen CP Factors influencing patient satisfaction two to five years after primary total knee arthroplasty J Arthroplasty 2014 29 6 1189 91 10.1016/j.arth.2014.01.008 24534535 12 Kolivand P Ali Hakami M Eghbali F Monjamed Z Sharifi F Patient satisfaction of the quality of patient teaching in Khatam Alanbia hospital Neurosci J Shefaye Khatam 2015 3 3 25 31 10.18869/acadpub.shefa.3.3.25 13 Berry PH Dahl JL The new JCAHO pain standards: implications for pain management nurses Pain Manag Nurs 2000 1 1 3 12 10.1053/jpmn.2000.5833 11706454 14 Gupta A Kaur K Sharma S Goyal S Arora S Murthy RS Clinical aspects of acute post-operative pain management & its assessment J Adv Pharm Technol Res 2010 1 2 97 108 22247838 15 Woldehaimanot TE Eshetie TC Kerie MW Postoperative pain management among surgically treated patients in an Ethiopian hospital PLoS One 2014 9 7 e102835 10.1371/journal.pone.0102835 25033399 16 Wiffen PJ Wee B Moore RA Oral morphine for cancer pain Cochrane Database Syst Rev 2016 4 4 CD003868 10.1002/14651858.CD003868.pub4 27105021 17 Baldini A, Von Korff M, Lin EH. A review of potential adverse effects of long-term opioid therapy: a practitioner’s guide. Prim Care Companion CNS Disord 2012; 14(3): PCC.11m01326. 10.4088/PCC.11m01326 18 Negro S Martín A Azuara L Sánchez Y Barcia E Compatibility and stability of ternary admixtures of tramadol, haloperidol, and hyoscine N-butyl bromide: retrospective clinical evaluation J Palliat Med 2010 13 3 273 7 10.1089/jpm.2009.0187 20078209 19 Overdyk FJ Postoperative opioids remain a serious patient safety threat Anesthesiology 2010 113 1 259 60 10.1097/ALN.0b013e3181e2c1d9 20 Dusek JA Griffin KH Finch MD Rivard RL Watson D Cost savings from reducing pain through the delivery of integrative medicine program to hospitalized patients J Altern Complement Med 2018 24 6 557 63 10.1089/acm.2017.0203 29474095 21 Ganesan P Manjini KJ Bathala Vedagiri SC Effect of music on pain, anxiety and physiological parameters among postoperative sternotomy patients: a randomized controlled trial J Caring Sci 2022 11 3 139 47 10.34172/jcs.2022.18 36247036 22 Khalili E Molavynejad S Adineh M Haghighizadeh MH The effect of Thai massage on the severity of pain in patients with unstable angina: a randomized controlled clinical trial J Caring Sci 2023 12 1 73 8 10.34172/jcs.2023.30150 37124410 23 Sharifipour F Bakhteh A Mirmohammad Ali M Effects of Citrus aurantium aroma on post-cesarean anxiety Iran J Obstet Gynecol Infertil 2015 18 169 12 20 10.22038/ijogi.2015.6131 24 Arjmand GR, Bahraminezhad N, Jafari S, Faghihzadeh S, Lotfinia I, Shimia M. The effect of inhalation of aromatherapy lavender (essential oil) on post-operative discectomy pain J Adv Med Biomed Res 2017; 25(110): 83-93. [Persian]. 25 Amini A Bahraminejad N Jafari S Kamali K The effect of aromatherapy with Rosa damascena essence on postoperative pain in inguinal hernia repair: a randomized clinical trial Nurs Midwifery Stud 2020 9 3 117 23 10.4103/nms.nms_103_19 26 Saonere Suryawanshi JA An overview of Citrus aurantium used in treatment of various diseases Afr J Plant Sci 2011 5 7 390 5 27 Suntar I Khan H Patel S Celano R Rastrelli L Suntar I, Khan H, Patel S, Celano R, Rastrelli LAn overview on Citrus aurantium L: its functions as food ingredient and therapeutic agent Oxid Med Cell Longev 2018 2018 7864269 10.1155/2018/7864269 29854097 28 Abbasnejad, M, Keramat B, Esmaili Mahani, S, Rezaeezade-Roukerd M. Effect of hydro-methanolic extract of sour orange flowers, Citrus aurantium, on pentylentetrazole induced seizure in male rats. J Babol Univ Med Sci 2012; 14(5): 20-8. [Persian]. 29 Cheraghi J Valadi A Effects of anti-nociceptive and anti-inflammatory component of limonene in herbal drugs Iran J Med Aromat Plants Res 2010 26 3 415 22 10.22092/ijmapr.2010.6803 30 Bent S Padula A Neuhaus J Safety and efficacy of Citrus aurantium for weight loss Am J Cardiol 2004 94 10 1359 61 10.1016/j.amjcard.2004.07.137 15541270 31 Namazi M, Amiraliakbari S, Mojab F, Talebi A, Alavi Majd H, Jannesari S. Investigating the effect of Citrus aurantium aroma on severity of labor pain during active phase of labor. Complementary Medicine Journal 2014; 4(1): 733-44. [Persian]. 32 Chen L Xie T Effect of Citrus aurantium on pain intensity and blood pressure in gastrectomy Indian J Pharm Sci 2022 84 2 174 79 10.36468/pharmaceutical-sciences.spl.471 33 Olapour A Behaeen K Akhondzadeh R Soltani F Razavi FA Bekhradi R The effect of inhalation of aromatherapy blend containing lavender essential oil on cesarean postoperative pain Anesth Pain Med 2013 3 1 203 7 10.5812/aapm.9570 24223363 34 Sharifipour F Mirmohammad Ali M Hashemzadeh M Comparison of the effect of Citrus aurantium and Salvia officinalis aroma on post-cesarean section pain Iran J Obstet Gynecol Infertil 2017 20 2 41 9 10.22038/ijogi.2017.8713 35 Mudgalkar N Bele SD Valsangkar S Bodhare TN Gorre M Utility of numerical and visual analog scales for evaluating the post-operative pain in rural patients Indian J Anaesth 2012 56 6 553 7 10.4103/0019-5049.104573 23325940 36 Fadaizadeh L Emami H Samii K Comparison of visual analogue scale and faces rating scale in measuring acute postoperative pain Arch Iran Med 2009 12 1 73 5 19111034 37 Yip YB Tam AC An experimental study on the effectiveness of massage with aromatic ginger and orange essential oil for moderate-to-severe knee pain among the elderly in Hong Kong Complement Ther Med 2008 16 3 131 8 10.1016/j.ctim.2007.12.003 18534325 38 Younger J McCue R Mackey S Pain outcomes: a brief review of instruments and techniques Curr Pain Headache Rep 2009 13 1 39 43 10.1007/s11916-009-0009-x 19126370
PMC010xxxxxx/PMC10352640.txt
==== Front J Caring Sci J Caring Sci J Caring Sci TBZMED Journal of Caring Sciences 2251-9920 2251-9920 Tabriz University of Medical Sciences 10.34172/jcs.2023.31964 Review Article Scales and Interventions for Resilience among Treatment-Seeking Patients with Depression: A Systematic Review https://orcid.org/0000-0003-1005-7556 Jayakrishnan K 1 Conceptualization Data curation Formal analysis Investigation Methodology Project administrat Project administration Writing – original draft Writing – review & editing https://orcid.org/0009-0000-6856-8207 Baruah Arunjyoti 2 Data curation Methodology Supervision Validation Visualization Writing – review & editing https://orcid.org/0000-0002-0351-087X Kumar Pankaj 3 Methodology Validation Visualization Writing – review & editing https://orcid.org/0000-0003-4632-0344 Javeth Athar 4 *Conceptualization Data curation Formal analysis Investigation Methodology Project administration Writing – original draft 1College of Nursing, AIIMS Kalyani, West Bengal, India 2Department of Psychiatric Nursing, LGBRIMH, Tezpur, Assam, India 3Department of Psychiatry, AIIMS Patna, Patna, Bihar, India 4College of Nursing, AIIMS Kalyani, West Bengal, India * Athar Javeth [email protected] 6 2023 21 5 2023 12 2 8493 03 3 2023 07 5 2023 © 2023 The Author(s). 2023 https://creativecommons.org/licenses/by-nc/4.0/ This work is published by Journal of Caring Sciences as an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by-nc/4.0/). Non-commercial uses of the work are permitted, provided the original work is properly cited. http://jcs.tbzmed.ac.ir Introduction: Resilience is an ability of an individual to effectively adjust and thrive in adverse stressful conditions. Resilience has protective and compensatory effects against depression. Evaluating resilience clinically and modifying it among patients with depression hugely impacts their prognosis. We aimed to explore different clinical scales for measuring resilience as well as interventions used with an intent to improve resilience among patients with depression. Methods: A systematic literature review was conducted by searching PubMed central, Biomed central, and google scholar, using relevant MeSH keywords. The population of interest were the patients who were clinically diagnosed with Bipolar or Unipolar Depression and the population were not restricted to any country. Clinical scales for evaluation and interventions for resilience among patients with depression were set as an outcome of the study. Randomized controlled trials (RCTs), Quasi-experimental studies, observational studies, and narrative reviews were considered relevant research designs for extraction. Results: A total of 8689 articles were identified and 13 articles were included in the final review, which yielded five scales that have been identified and have been used to evaluate resilience among the patients who are clinically diagnosed with depression and six different interventions for building resilience among patients with depression. Conclusion: Resilience-building interventions will not only act as a preventive measure against depression but also help in promoting recovery and sustaining remission after a depressive episode. Clinical evaluation of resilience and management will significantly support boosting emotional experience. Psychological resilience Interventions Scales Depression ==== Body pmcIntroduction Sigmund Freud once explained the concept of vulnerability through a crystal principle, which takes the example of a crystal that shatters once it falls on the ground only through the weakest structural planes (intrinsic cleavages), even when that is invisible. Patients with mental illness are also hypothesized to have similar weak structures in their mind, for example, individuals with depression may have undergone multitude of adversities in their life, but the symptoms may arise with an insignificant cause or without any significant change.1 So various mental illnesses are believed to arise from an individual’s particular psychological weaknesses. While resilience is a concept understood to be psychologically adapting to stress and adversities to maintain emotional homeostasis.2,3 Resilience is a multidimensional construct; it cannot be understood linearly. For simplicity, it could be termed as a phenomenon that gives the ability an individual to psychologically bounce back from adversity. This particular ability seems to vary between patients, as seen through genetic studies,4 Deoxyribonucleic acid (DNA) studies showing differences in stress responses being mediated through genetic factors in the reactivity from the sympathetic nervous system,5and the lowered reuptake of serotonin being mediated by a single base substitution in the long form of 5-HTTLPR gene promoting the risk of depression.6 Psychological factors such as having loving caretakers, a positive worldview, positive emotional regulation, a better coping system, sound social support, seeing problems in a positive view, spiritual support, attention to physical health, and disciplined focus help to deal with adversities better than otherwise.7 A resilient individual will talk to health professionals regarding depressive symptoms and will seek help. Building resilience in depressive individuals will encourage them to talk about depressive symptoms with significant others.8,9 In a cross-sectional study done on 100 depressive and recurrent depressive disorder patients from ten hospitals in Tokyo, took the subjects who were eligible if they were outpatients 18 years of age or older, who fulfilled the criteria for a depressive episode (F32) or recurrent depressive disorder (F33) and who were capable of providing informed consent. It was reported that lower levels of resilience showed higher depressive symptomatology. The study also found as the severity of depression increased, irrespective of assessing objectively or subjectively, both were negatively associated with resilience.10 More interventions concerning building resilience are necessary for depressive disorders and reduction of depressive symptomatology. In fact, in a pilot study that was done for the improvement of resilience levels and reduction of depressive symptoms in college students, during a stressful period, the intervention involved a month-long psychoeducational intervention of weekly two-hour sessions, along with a cognitive behavioral therapy component. The intervention on assessment at the end of the four-week intervention revealed that there is a significant improvement in resilience scores and a significant reduction in depressive symptomatology. The study indicates that non-pharmacological interventions could be promoted for improvement of resilience and in turn, reduction of symptoms.11 A systematic review would yield extensive literature search for screening out the specific group of scales and intervention under the preview of the study. Moreover, there are no systematic reviews which enlist the scales and interventions of resilience which has been administered specifically to treatment seeking depressive patients. By understanding this specific group of interventions for building resilience among clinically diagnosed depressive individuals, the nurses receive greater understanding for managing individuals with depression effectively. It is understood that depression is a grave disorder, and the challenges that the health professional has to deal with could be resolved majorly by an improved resilience level. Resilience could be assessed easily and enhanced non-invasively. Moreover, the studies done for the assessment and development of resilience, especially with individuals who are clinically diagnosed with depressive disorders are limited. So, this review tries to explore systematically the clinical scales as well as interventions that have been used in studies for the assessment and development of resilience, specifically on patients who were clinically diagnosed with depression. This systematic review will include inclusive evidence about the clinical scales and interventions necessary for resilience in depression. Materials and Methods The systematic review has been conducted as per the criteria of Preferred Reporting Items of Systematic Review (PRISMA)12 and was registered in the PROSPERO registry for systematic review (reg no. CRD42022308942). Data Sources and Article Selection The search strategy adopted for recognizing all the available and pertinent articles was done by utilizing three separate medical databases (PubMed, Google Scholar, and Science Direct) with the keywords (Resilience or Resiliencies or Resiliency or Resiliency traits or psychological resilience or psychological resiliencies or psychological resiliency) AND (depression or bipolar depression or unipolar depression or depressive disorder or primary depression or clinical depression) AND (interventions or interventional or methods or management). The literature search was limited to articles published from 1990 to date, only in English Language. The current review had identified 8689 articles from different databases, out of which, 4111 articles were retrieved for screening, which found that only 86 articles were qualifying the eligibility criteria. 19 articles were excluded as their full texts could not be retrieved. Further screening showed that only 13 studies were eligible for inclusion to the review (Figure 1). Figure 1 Prisma flowchart of literature search showing identification, screening and final inclusion of articles Inclusion and Exclusion Criteria Inclusion criteria for the articles are: Population: Patients who are clinically diagnosed with Bipolar and Unipolar Depression. The population is not restricted to any country. Outcome: Clinical scales for evaluation, and interventions for enhancing resilience among patients with depression. Setting: Hospital admissions, clinical setting, primary health care. Study designs: Randomized controlled trials (RCTs), Quasi-experimental, observational studies, narrative reviews Exclusion criteria for the articles are: Studies that show the presence of depression or anxiety as a co-morbidity to other medical illnesses. Studies published in any language other than English Studies that were done on paediatric patients The stages of article selection were: Stage 1: Identification: All the articles that were retrievable from the selected databases were exported to Mendeley Reference Software. The articles which were found to be duplicates were removed initially through the software. Stage 2: Screening: Two independent reviewers screened the title and abstracts of the retrieved articles and excluded the studies which were not conducted on individuals who are clinically diagnosed with depressive disorder or on measures of resilience. Any disputes by the two reviewers were sorted by the third reviewer. Stage 3: Eligibility: The remaining articles whose full texts could not be retrieved were also excluded. The full text of the articles for the rest was read by the reviewers and checked for its eligibility criteria. The articles found to be ineligible were excluded. Stage 4: Quality Assessment: The quality of the selected articles was assessed with Joanna Briggs Institute Critical appraisal tools. Three RCTs were assessed with the Joanna Briggs Institute (JBI) Checklist of Randomised Control Trials which had 13 items assessing the randomization, blinding, allocation, similarity in outcome assessment between groups, use of proper statistical analyses, etc. Three Quasi-Experimental studies were assessed with the JBI Checklist of Quasi-Experimental designs which had 9 items assessing the clarity of cause and effect, homogeneity of participants, presence of control, usage of proper statistical analyses, etc. The rest of the eight observational studies were assessed with the JBI Checklist for Analytical Cross-Sectional Studies which had 8 items assessing the clarity of sample description, setting, objectivity in measuring outcomes, identification of confounding variables, checking validity, and reliability of measurement tools and usage of proper statistical analyses, etc. Responses like ‘Yes’, ‘No’, ‘Unclear’, or ‘Not Applicable’ was provided for the rating of each item of the quality appraisal tools. Results The different stages of the retrieval, checking eligibility, and selection of articles have been depicted in a PRISMA flowchart (Figure 1). A total number of 8689 articles were identified from three different databases, i.e., PubMed (2122), Google Scholar (1250), and Science Direct (5317). Of the identified articles 4578 articles were excluded before the screening, the majority of the articles (4467) could not be retrieved from the database itself and 111 articles were found to be duplicate articles, giving 4111 articles for screening. Subsequently, 4025 articles were excluded from the study either due to the absence of any resilience measures or conducted in individuals who had not been clinically diagnosed with depression, or both. Out of the remaining 86 articles that were screened, 19 articles were further excluded as the full text of the articles could not be retrieved. Then, the 67 articles were reviewed thoroughly for assessing their eligibility by the reviewers. Further, on assessment it was noticed that 21 other studies were also not done on individuals who are clinically diagnosed with depression, 19 studies were done on individuals who were all below the age of 18 years, and eight studies were narrative reviews that did not mention any clinical scale or intervention, 2 studies were done on animals, 1 study was in Spanish and 3 other articles were found to be duplicates.13 The final 13 studies that were included for analysis were done in 8 different countries like the USA (three),11,14,15 Italy (three)16–18 India (two),19,20 Australia (one),21 Sweden (one),22 Egypt (one),23 Netherland (one)24 and Thailand (one).25 The data synthesis is done by following the extracted studies for review and put under the headings: Details of publication (Author and year of publication) Type of study (design) Mode of Completion (Self-report or interview schedule) Subject Characteristics (type of participants) Name of the clinical scale used (measuring scale) The data that has been extracted after the review of the included articles regarding the clinical scales and interventions for resilience among patients with depression, has been presented in a form of narrative synthesis (Table 1). Table 1 Characteristics of articles that yielded the clinical scales for resilience among patients with depression Author & Year Type of study Mode of completion Subject characteristics Scales identified McCann et al25 (2017) RCT Self-reporting Participants who are diagnosed with moderate depression who are currently hospitalised within the age group of 18-60years The Resilience Scale (G. M. Wagnild & Young, 1993) Vlasova et al14 (2018) Observational study Self-reporting Clinical diagnosed individuals with major depressive disorder who are aged 60years and above CD-RISC 25 Wu et al15 (2019) Observational study Self-reporting Participants who are adults≥60 years with major depressive disorder CD-RISC 25 Siddarth et al11 (2019) Observational study Self-reporting Participants who are adults≥60 years with major depressive disorder CD-RISC 25 Priyadarsini and Rohini19 (2017) Pre-test, post-test and follow-up without control group design Self-reporting Participants were adults who were clinically diagnosed with Unipolar depression between 18-45 years of age BURS Sawle et al21 (2015) Observational study Self-reporting First group are young patients (17-25 years) who are diagnosed with psychosis and their current principal familial caregiver (23-57 years) Second group are young patients (15-25 years) who are diagnosed with depressive disorder and their current principal familial caregiver (23-54 years) ER89 Favale et al (2020)16 Observational study Self-reporting Participants with the clinical diagnosis of unipolar depressive disorder, bipolar disorder and schizoaffective disorder reporting major current depressive episode ages between 18-75 years CD-RISC 10 Priyadarsini and Rohini20 (2015) Two group quasi-experimental study Self-reporting Participants were with mild level of depression in the age range of 20-45 years BURS Ferrari et al (2016)24 Two-arm, double-blinded RCT Self-reporting Participants who were between the age of 18 to 65 years and had been diagnosed with first time or recurrent major depressive disorder The Resilience Scale (G. M. Wagnild & Young, 1993) Collazzoni et al17 (2020) Observational Study Self-reporting Patients with the primary diagnosis of major depressive disorder within the age range from 19 to 64 years RSA Hassnin Eita and Mohamed Aboshereda23 (2021) Quasi-experimental design two groups study Self-reporting Participants are medically diagnosed with depression in the age range 18 to 65 years CD-RISC 25 Rossetti et al18 (2017) Observational Study Self-reporting Participants were clinically diagnosed depression with the mean age of 42.14 RSA Abbreviations: RCT, randomized controlled trial; RSA, Resilience Scale for Adult; CD-RISC, Connor-Davidson Resilience Scale; BURS, Bharathiar University Resilience Scale; ER89, Ego-Resiliency Scale. Clinical Scales for Measuring Resilience Among Individuals with Depression This search yielded six different resilience scales that have been utilized for the assessment of individuals with clinical depression. The research articles comprised two Randomised control studies, three quasi-experimental studies, and eight observational studies that yielded the six resilience scales. The Resilience Scale by G.M. Wagnild and Young This scale measures components of resilience in different domains of young patients’ lives, ranging from planning and thinking ahead to the level of independence. The scale is a 25-item Likert scale, with each item rated with a seven-point rating, from 1 (disagree) to 7 (agree). The scores range from 25 to 175, suggesting higher scores with higher resilience. The scale also consists of three sub-components, i.e., Personal Competence, Health and wellness, and Acceptance of self and life.24,25 Connor Davidson Resilience Scale (CD-RISC) CD-RISC is a self-reporting 25-item rating scale, developed by Kathryn M. Conner and Jonathan R.T. Davidson, for testing resilience in all age groups. It is a five-point scale ranging from not at all true (0) to true nearly all the time (4). The scores range from 0 to 100, indicating higher resilience for higher scores. CD-RISC is officially authorized with three versions of the same, such as CD-RISC-25, CD-RISC-10, and CD-RISC-2. The CD-RISC-25 reflects on five factors such as high standards, tenacity, and competence (8 items), handling negative emotions, trusting one’s instincts, and perceived benefits of stress (7 items), positive attitude to change, and secure relationships (5 items), perceived control (3 items) and finally spirituality (2 items).14-16,23,26,27 Bharathiar University Resilience Scale (BURS) BURS is a 30-item self-reporting Likert scale with 5-point response options, ranging from “not at all appropriate” (1) to “most appropriate” (5). The 30 items are set as personal statements which the participants are expected to mark as the most appropriate response in their regard. BURS is considered to measure seven domains of resilience, such, (i) duration of getting back to normalcy, (ii) perception of the effect of past negative events, (iii) response to risk factors, (iv) response to negative events, (v) openness to experience, (vi) flexibility and (vii) confidence in coping with future. The total scores of all the items are summed up to establish the level of psychological resilience of the respondent, with the scores ranging from 30 to 150.19,20 Ego-resiliency Scale (ER89) This scale has been created by Block and Kremen in 1989, to measure psychological resilience. Psychological resilience is the adaptability of the mind to bounce back from any negative emotional experience and be able to function in flexibility under stressful environmental states. ER89 is a 14-item Likert scale with four-point response options, such as, does not apply at all (1) to applies very strongly (4). The range of scores goes from 14-56, with higher scores indicating higher psychological resilience.21 Resilience Scale for Adults (RSA) RSA is one of the resilience scales which approaches the measurement of resilience directly. This scale constitutes six factors for the measurement of resilience, with four factors for personal characteristics, that is, perception of self, planned future, social competence, and structured style; one factor for family characteristics, i.e., family cohesion and one factor for social characteristics, i.e., social resources. The scale is a 33-item scale with a seven-point semantic differential rating scale, with two polar opposite attributes at both ends of the scale for each item. For example, appreciating my qualities and despising my qualities would be the two ends of the response option and varies with each item. The range of scores is from 33 to 231, with higher resilience indicated by a higher score.17,28,29 There are many more scales which has been used to assess resilience,30but in the review, these were the only scales that have been mentioned to be used on adults who are clinically diagnosed depressive patients (Table 2). Table 2 Clinical scales for resilience for patients with depression Scales Population No. of items No. of dimensions Dimensions The Resilience Scale (G. M. Wagnild & Young, 1993) Moderate and recurrent major depressive disorders 25 3 i) Personal Competence, ii) Health and wellness and iii) Acceptance of self and life CD-RISC 25 Major depressive disorders 25 5 i) High standards, tenacity and competence, ii) handling negative emotions, trusting one’s instincts, and perceived benefits of stress iii) positive attitude to change and secure relationships, iv) perceived control and v) spirituality BURS Unipolar depression and mild level of depression 30 7 i) Duration of getting back to normalcy, ii) perception of effect of past negative events, iii) response to risk factors, iv) response to negative events, v) openness to experience, vi) flexibility and vii) confidence in coping with future. ER89 All levels of depression 14 1 - CD-RISC 10 Unipolar, bipolar and schizoaffective disorders 10 1 - RSA Major depressive disorder 33 6 i) Personal characteristics ( (a) perception of self, (b) planned future, (c) social competence and (d) structured style), ii) Family characteristics (family cohesion) and iii) social characteristics (social resources.) Abbreviations: RSA, Resilience Scale for Adult; CD-RISC, Connor-Davidson Resilience Scale; BURS, Bharathiar University Resilience Scale; ER89, Ego-Resiliency Scale. Interventions for Improving Resilience Among Patients with Depression In the case of interventions, the review identified six different interventions from three randomized control trials and three Quasi-experimental studies. These interventions have proven to be effective in raising resilience among patients with clinically diagnosed depression. (Table 3) Table 3 Characteristics of articles that has identified the interventions for improving resilience among depressive patients Author & Year Type of study Subject characteristics Interventions Main findings of the study McCann et al25 (2017) RCT Participants who are diagnosed with moderate depression who are currently hospitalised within the age group of 18-60 years CBT based guided self help bibliotherapy Reading, writing and activities based to challenge negative thoughts and behaviours to enhance resilience Ekbäck et al22(2021) Multi-Center RCT Adolescents and young adults between age of 15 to 22 years, who have attended clinic for the diagnosis of major depressive disorder or persistent depressive disorder. TARA Mindfulness based interventions focusing on emotional self-regulation than acceptance of emotional experience Priyadarsini and Rohini19 (2017) Pre-test, post-test and follow-up without control group design Participants were adults who were clinically diagnosed with unipolar depression between 18-45 years of age Pranayama Anuloma Viloma is a breathing practice believed to restore autonomic nervous system imbalances, balance of pineal gland and activate the frontal lobe to provide tranquillity, clarity and concentration Priyadarsini and Rohini20 (2015) Two group quasi-experimental study Patients with mild depression aged between 20-45 years Carnatic music and pranayama Atana, Mohanam and Neelambari were 3 ragas included as Carnatic music for intervention along with pranayama Ferrari et al24(2016) Two-arm, double-blinded RCT Participants who were between the age of 18 to 65 years and had been diagnosed with first time or recurrent major depressive disorder Attentional bias modification Image based modification of negative attention towards positive or neutral attention for improving resilience Hassnin Eita and Mohamed Aboshereda23 (2021) Quasi-experimental design two groups study Participants are medically diagnosed with depression in the age range 18 to 65 years Resilience Training based Nursing Intervention Pre-written manual involving psychoeducation and group discussion on relevant areas for improving emotion regulation and resilience. Abbreviations: RCT, randomized controlled trial; RSA, Resilience Scale for Adult; CD-RISC, Connor-Davidson Resilience Scale; BURS, Bharathiar University Resilience Scale; ER89, Ego-Resiliency Scale; CBT, cognitive behavioural therapy; TARA, Training for Awareness, Resilience and Action. CBT-Based Guided Self Help Bibliotherapy Guided self-help would provide a modest level of support by a coach or a clinician in completing a therapy. For instance, Bibliotherapy is a creative method to inculcate reading specific texts to treat. Cognitive behavioural therapy (CBT) has been a proven effective therapy in the conventional treatment regimen for reducing depressive symptoms. So, CBT-based Bibliotherapy which utilizes a guided self-help mode seems to be effective in enhancing resilience among patients with depression. It contains eight modules, with reading, writing, and activities to be completed in one week, for eight weeks. The completion of the assignments would require participants to challenge negative thoughts and behaviors to improve their resilience.25 Training for Awareness, Resilience and Action (TARA) TARA is a neuropsychiatric intervention based upon the mindfulness approach, yet primarily focusing on emotional self-regulation than acceptance of emotional experience. The intervention consists of 12 sessions, that is completely online delivered, and has been divided into four modules, such as, (i) calming down and creating a sense of safety (1st-3rd session)- yoga-based movement and breathing exercise are trained primarily to reduce amygdala hyperactivity, (ii) Attending to and caring about our inner experience (4th-6th session)- emotional labeling and interoceptive attention is trained for shifting attention from negative self-referencing to present moment sensory awareness, (iii) recognizing, regulating and communicating emotions (7th to 9th session)-practicing of empathetic listening, effective communication and compassionate responses to reduce interpersonal stress and regulation of emotions and (iv) core values, goal setting and committed action (10th to 12th session) – training to identify and recognize the experiential avoidance to identify the core values and be guided in participant’s actions in their lives. Overall, this intervention aims to provide an ability for patients with depression, to have cognitive control over emotional experiences, in dealing with day-to-day life activities.22,31 Pranayama Pranayama is considered to be a basic component of Hatha Yoga, which primarily consists of control of Prana (vital energy needed for the survival of our physical self), by breathing techniques. Anuloma Viloma is a type of pranayama, which is believed to restore autonomic nervous system imbalances, maintain the balance of the pineal gland and even activate the frontal lobe of the brain to provide tranquillity, clarity, and concentration. So, Anuloma Viloma provides calmness and better awareness without any side effects. The practice is provided for 30 minutes, followed by 15 minutes of discussion with the participants. Long-term practice of pranayama would initiate a process known as telencephalisation, which is shifting towards conscious breathing from regular unconscious breathing leading to the involvement of the cerebral cortex and surrounding areas of the brain, even concerned with emotions.19 Carnatic Music and Pranayama In this intervention, both Carnatic music and Pranayama was utilized together. For Carnatic music, three ragas were chosen, namely, Atana, Mohanam, and Neelambari. Each raga was played to the participants for 10 minutes each and was combined with 15 minutes of Pranayama after. The Intervention was given for 12 sessions, with two sessions per week for 6 weeks. Each session was of 45 minutes duration overall. The ragas were played through recorded flute music and Pranayama was similar to the previous intervention mentioned. The study has assumed that the vibrations in the ragas were able to resonate with participants’ moods and health. The concept of Raga Chikitsa that is, healing through the use of raga is believed to balance nature that is in imbalance and help in the healing process.20 Attentional Bias Modification This intervention is based on the idea that patients with depression have a more attentional bias toward negative information and difficulty disengaging, as compared to positive or neutral information. The negative attentional bias has been linked with ineffective emotion regulation in states of stress and decreased resilience. So, through attentional bias modification, it is expected to alter the symptoms of depressive disorders. Participants will engage in eight training sessions for a period of two weeks, in which 50 pairs of pictures are created that have negative and positive images or scenes, equally. A white fixation arrow is introduced during the display of the contrasting images to the participants, and 90% of the time, the arrow is placed over the positive images, modifying the attention towards the same. Regular breaks and feedback are included for the sustenance of motivation in the participants.24 Resilience Training-Based Nursing Intervention This intervention was of eight sessions with one session of one hour each per week for eight weeks. Each session would include a revision (10 minutes), practice of relaxation technique (10 minutes), structured presentation (15 minutes), group discussion (15 minutes) and finally ending with a summarisation and feedback by the participants (15 minutes). Each session involved different topics for presentation which was structured for uniformity and was consistent with a pre-written manual. The first session involved training on topics regarding different aspects of depression (causes, signs, symptoms, fears, and stigma), the second session was on resilience (factors, process, and relationship with depression), and the third and fourth sessions aimed to build resilience, whereas the fifth session focussed on building and utilizing healthy coping strategies, the sixth session dealt with increasing social support, whereas the seventh session aimed to enhance problem-solving techniques and flexibility and finally the eight sessions focussed on practicing stress management techniques, increasing positivity and dealing difficult emotions.23 Discussion From our systematic review, we identified 13 articles that met our inclusion criteria for the review. In this, there were three RCTs, three quasi-experimental studies, and seven observational studies. From the identified articles, we have yielded five different clinical scales as well as six interventions that were especially utilized on patients who were clinically diagnosed with depressive disorders for resilience. The most extensively used scale seems to be CD-RISC 25, mostly because it has strong psychometric properties and its applicability in clinical settings. The CD-RISC 25 is sensitive to clinical interventions as it identifies resilient characteristics that are enhanced by an individual’s adaptive pursuits.32Apart from the five scales that were identified, the Brief resilience scale was also accounted for by the reviewers, but its non-applicability in clinical scenarios led us to remove it from the analysis. Even then, Brief Resilience Scale was just a six-item rating scale where the participants are expected to respond to a five-point response option. The authors believed that six items are sufficient to assess resilience, as resilience could never be assessed directly. They have utilized coping styles, social relationships, and health-related outcomes as their factors to evaluate the level of resilience.33,34 Different measures differ in the means of assessing resilience, as resilience by default is a complex construct to appraise. For example, as we consider that early childhood trauma and neglect being a central concept of maladaptation, and further poor resilience in individuals,10 certain studies have shown that some neglected and abused children did show better resilience in different areas of functioning and positive adjustment towards particular developmental tasks.35-38 So, the idea of measuring resilience on account of their protective factors and developmental tasks is questionable, as the evaluation of the resilience of individuals with similar adversities would show varied resilience levels.39 On account of featuring prevalence of resilience among populations with common stressors could show diversity in values, i.e., different scales would infer the results differently.40 Whereas, in a network analysis, risk and protective factors for the remitted depressive patients were explored, such as residual symptomatology, emotional regulation, cognitive control, and resilience. This previous study showed that resilience was taking a central role in connecting all the factors and proving to be a key factor in connecting all the other risk and protective factors. Resilience stands out to be a successful coping factor when it comes to the stress from remission of depression.41 Therefore, the health team members must be vigilant to assess the patients with depression regarding the levels of resilience and its impact on symptomatology. By focussing on resilience, the patient would also participate in the treatment process actively and generate a sense of insight to his advantage.42 In the case of interventions, six different interventions were identified after a comprehensive search from three RCTs and quasi-experimental studies each. The review did not show any pharmacological agents as interventions to improve resilience, even though there are pharmacological enhancements of neurochemical systems for resilience also being reviewed. The effect of pharmacology on depressive patients is yet to be proven.43 The non-pharmacological interventions that have been identified in the review, are also not measured for their level of effectiveness in our review, even though individual studies have shown their effect on significantly improving resilience in depressive individuals. The Pranayama and Carnatic music are the interventions that would require prior training and certification, or expert assistance to conduct in its apposite manner. The rest of the interventions include self-help assignments for reading, writing, and comprehension, otherwise, previously structured manuals are being followed, which are easier to replicate by any member of the health team. TARA program also includes yogic practices and breathing techniques, regardless, the authors have not mentioned any prior requirement of specific training for the same. Nonetheless, the authors have mentioned that the whole program of TARA needs to replicate with TARA-trained facilitators.22 Attention bias modification treatment (ABMT) seems to be a promising intervention concerning depressive symptomatology. As ABMT seems to be a more focused and limited sort of intervention, the extraneous factors affecting the outcome also seem to be less, as compared to a CBT or Pharmacological intervention. Moreover, the control group of ABMT is also very tight, as they are also subjected to similar cues, number of sessions, or effects which might show a lesser group effect size of ABMT, as compared to its counterparts.44 The effect of resilience is also evident in the quality of life among the depressive population.45-48 So deliberate assessment of resilience, as a part of the treatment protocol, needs to be mandated, along with the inclusion of non-pharmacological interventions, led by nursing personnel in a psychiatric setting for patients with depression, shall act as an effective change in depressive symptomatology, and more importantly the quality of life. Health team personnel must also take up research projects involving interventions for resilience or systematic reviews involving meta-analyses of the same to add to our evidence base for the enhancement of resilience in a clinically depressed population. Limitations of the study were that only three databases were considered and google scholar is more of a search engine than a database, which narrowed the scope of review, lack of access to some full texts that could have been relevant, and not involving grey literature in the review. Conclusion The systematic review was able to synthesize clinical scales and interventions that have been used to improve resilience among patients with clinical depression, in the last two decades. Five clinical scales and six non-pharmacological interventions were identified through the review. Among patients, enhancing resilience could improve their prognosis and boost their quality of life. Considering the 13 articles that were included in the review, we can conclude that the latest interventions for resilience have the potential to bring well-being to the soaring population of depressive disorders. Acknowledgments We are extremely thankful to Dr. Vishnu Renjith and Ms. Neethu Maria Joseph for retrieving full texts of inaccessible articles and proof reading our article. Authors’ Contribution Conceptualization: K Jayakrishnan, Athar Javeth. Data curation: K Jayakrishnan, Arunjyoti Baruah, Athar Javeth. Formal analysis: K Jayakrishnan, Athar Javeth. Investigation: K Jayakrishnan, Athar Javeth. Methodology: K Jayakrishnan, Arunjyoti Baruah, Pankaj Kumar, Athar Javeth. Project administration: K Jayakrishnan. Software: K Jayakrishnan, Athar Javeth. Supervision: Arunjyoti Baruah. Validation: Arunjyoti Baruah, Pankaj Kumar. Visualization: K Jayakrishnan, Arunjyoti Baruah, Pankaj Kumar. Writing–original draft: K Jayakrishnan, Athar Javeth. Writing–review & editing: K Jayakrishnan, Arunjyoti Baruah, Pankaj Kumar. COI-statement All the authors declare no conflicts of interest with research or writing of the paper. Data Availability The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request. Ethical Approval We did not require ethical approval as neither data has been collected nor any intervention has been conducted on humans/animals. Funding No funding has been received for this systematic review. Research Highlights What is the current knowledge? Resilience is found to be lowering, with increasing depressive severity Resilient individual has more probability to seek treatment than otherwise Building resilience could improve the prognosis of depression as well as their quality of life. There are no systematic reviews that has enlisted the scales and interventions of resilience, specifically administered to treatment seeking depressive patients. What is new here? Five different scales which has been administered on treatment seeking depressive patients in hospital settings has been identified. Six separate non-pharmacological interventions for building resilience on clinically diagnosed depressive individuals were highlighted. ==== Refs References 1 Freud S, Strachey J, Gay P. New Introductory Lectures on Psycho-Analysis. WW Norton & Company; 1989. p. 253. 2 Palmiter D, Alvord M, Dorlen R, Comas-Diaz L, Luther SS, Maddi SS, et al. Building Your Resilience. American Psychological Association; 2012. Available from: https://www.apa.org/topics/resilience/building-your-resilience. Accessed August 9, 2022. 3 Elisei S Sciarma T Verdolini N Anastasi S Resilience and depressive disorders Psychiatr Danub 2013 25 Suppl 2 S263 7 23995190 4 Feder A Nestler EJ Charney DS Psychobiology and molecular genetics of resilience Nat Rev Neurosci 2009 10 6 446 57 10.1038/nrn2649 19455174 5 Southwick SM Vythilingam M Charney DS The psychobiology of depression and resilience to stress: implications for prevention and treatment Annu Rev Clin Psychol 2005 1 255 91 10.1146/annurev.clinpsy.1.102803.143948 17716089 6 Caspi A Hariri AR Holmes A Uher R Moffitt TE Genetic sensitivity to the environment: the case of the serotonin transporter gene and its implications for studying complex diseases and traits Am J Psychiatry 2010 167 5 509 27 10.1176/appi.ajp.2010.09101452 20231323 7 Southwick SM Charney DS The science of resilience: implications for the prevention and treatment of depression Science 2012 338 6103 79 82 10.1126/science.1222942 23042887 8 Aigner M Förster-Streffleur S Prause W Freidl M Weiss M Bach M What does the WHOQOL-Bref measure? Measurement overlap between quality of life and depressive symptomatology in chronic somatoform pain disorder Soc Psychiatry Psychiatr Epidemiol 2006 41 1 81 6 10.1007/s00127-005-0997-8 16341612 9 Shumye S Belayneh Z Mengistu N Health related quality of life and its correlates among people with depression attending outpatient department in Ethiopia: a cross sectional study Health Qual Life Outcomes 2019 17 1 169 10.1186/s12955-019-1233-7 31703701 10 Seok JH Lee KU Kim W Lee SH Kang EH Ham BJ Impact of early-life stress and resilience on patients with major depressive disorder Yonsei Med J 2012 53 6 1093 8 10.3349/ymj.2012.53.6.1093 23074107 11 Siddarth P Laird K Vlasova R St Cyr N Lavretsky H Clinical correlates of resilience factors in geriatric depression Am J Geriatr Psychiatry 2018 26 3 S139 10.1016/j.jagp.2018.01.169 12 Moher D Liberati A Tetzlaff J Altman DG Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement PLoS Med 2009 6 7 e1000097 10.1371/journal.pmed.1000097 19621072 13 JBI. Critical Appraisal Tools. South Australia. Available from: https://jbi.global/critical-appraisal-tools. Accessed August 19, 2022. 14 Vlasova RM Siddarth P Krause B Leaver AM Laird KT St Cyr N Resilience and white matter integrity in geriatric depression Am J Geriatr Psychiatry 2018 26 8 874 83 10.1016/j.jagp.2018.04.004 29803529 15 Wu P Lavretsky H Siddarth P Neurocognitive correlates of resilience in geriatric depression Am J Geriatr Psychiatry 2018 26 3 S139 40 10.1016/j.jagp.2018.01.170 16 Favale D Bellomo A Lotti P Bradascio C Ventriglio A Role of hope and resilience in the outcome of depression and related suicidality Psychology 2020 11 3 428 35 10.4236/psych.2020.113028 17 Collazzoni A Stratta P Pacitti F Rossi A Santarelli V Bustini M Resilience as a mediator between interpersonal risk factors and hopelessness in depression Front Psychiatry 2020 11 10 10.3389/fpsyt.2020.00010 32184740 18 Rossetti AC Papp M Gruca P Paladini MS Racagni G Riva MA Stress-induced anhedonia is associated with the activation of the inflammatory system in the rat brain: restorative effect of pharmacological intervention Pharmacol Res 2016 103 1 12 10.1016/j.phrs.2015.10.022 26535964 19 Priyadarsini AC Rohini NS Pranayama for the management of depression and enhancement of resilience Indian J Posit Psychol 2017 8 3 351 5 20 Priyadarsini AC Rohini NS Effect of camatic music and pranayama on depression and resilience in mild depressives Indian Journal of Health & Wellbeing 2015 6 9 849 53 21 Sawle G Lennings C Heard R Carer attachment moderates resilience for young patients experiencing psychosis or depression Universal Journal of Psychology 2015 3 6 176 84 10.13189/ujp.2015.030604 22 Ekbäck E Granåsen G Svärling R Blomqvist I Henje E Clinical effectiveness of training for awareness resilience and action online compared to standard treatment for adolescents and young adults with depression: study protocol and analysis plan for a pragmatic, multi-center randomized controlled superiority trial Front Psychiatry 2021 12 674583 10.3389/fpsyt.2021.674583 34707516 23 Hassnin Eita L Mohamed Aboshereda H Resilience training based nursing intervention: its effect on depressive symptoms, coping styles and resilience among patients with depression Egypt J Health Care 2021 12 2 11 27 24 Ferrari GR Becker ES Smit F Rinck M Spijker J Investigating the (cost-) effectiveness of attention bias modification (ABM) for outpatients with major depressive disorder (MDD): a randomized controlled trial protocol BMC Psychiatry 2016 16 1 370 10.1186/s12888-016-1085-1 27809880 25 McCann TV Songprakun W Stephenson J Effectiveness of a guided self-help manual in strengthening resilience in people diagnosed with moderate depression and their family caregivers in Thailand: a randomised controlled trial Issues Ment Health Nurs 2017 38 8 655 62 10.1080/01612840.2017.1313913 28471254 26 Velickovic K Rahm Hallberg I Axelsson U Borrebaeck CAK Rydén L Johnsson P Psychometric properties of the Connor-Davidson Resilience Scale (CD-RISC) in a non-clinical population in Sweden Health Qual Life Outcomes 2020 18 1 132 10.1186/s12955-020-01383-3 32398074 27 Laird KT Lavretsky H Wu P Krause B Siddarth P Neurocognitive correlates of resilience in late-life depression Am J Geriatr Psychiatry 2019 27 1 12 7 10.1016/j.jagp.2018.08.009 30262406 28 Stratta P. The Italian Validation Study of the Resilience Scale for Adults (RSA). Paolo Stratta - Academia. USA; [Internet]. Available from: https://www.academia.edu/18398592/The_Italian_validation_study_of_the_Resilience_Scale_for_Adults_RSA_. Accessed March 11, 2022. 29 Rossetti MC Tosone A Stratta P Collazzoni A Santarelli V Guadagni E Different roles of resilience in depressive patients with history of suicide attempt and no history of suicide attempt Braz J Psychiatry 2017 39 3 216 9 10.1590/1516-4446-2016-2045 28538755 30 Windle G Bennett KM Noyes J A methodological review of resilience measurement scales Health Qual Life Outcomes 2011 9 8 10.1186/1477-7525-9-8 21294858 31 Henje Blom E Tymofiyeva O Chesney MA Ho TC Moran P Connolly CG Feasibility and preliminary efficacy of a novel RDoC-based treatment program for adolescent depression: “Training for Awareness Resilience and Action” (TARA)-a pilot study Front Psychiatry 2016 7 208 10.3389/fpsyt.2016.00208 28138319 32 Connor KM Davidson JR Development of a new resilience scale: the Connor-Davidson Resilience Scale (CD-RISC) Depress Anxiety 2003 18 2 76 82 10.1002/da.10113 12964174 33 de Holanda Coelho GL Hanel PH Cavalcanti TM Rezende AT Gouveia VV Brief Resilience Scale: testing its factorial structure and invariance in Brazil Univ Psychol 2016 15 2 397 408 10.11144/Javeriana.upsy15-2.brst 34 Liu N Liu S Yu N Peng Y Wen Y Tang J Correlations among psychological resilience, self-efficacy, and negative emotion in acute myocardial infarction patients after percutaneous coronary intervention Front Psychiatry 2018 9 1 10.3389/fpsyt.2018.00001 29410632 35 Toth SL Cicchetti D Toth SL, Cicchetti DA developmental psychopathology perspective on child maltreatmentIntroduction Child Maltreat 2013 18 3 135 9 10.1177/1077559513500380 23886641 36 Flores E Cicchetti D Rogosch FA Predictors of resilience in maltreated and nonmaltreated Latino children Dev Psychol 2005 41 2 338 51 10.1037/0012-1649.41.2.338 15769190 37 Cicchetti D Rogosch FA Lynch M Holt KD Resilience in maltreated children: processes leading to adaptive outcome Dev Psychol 1993 5 4 629 47 10.1017/s0954579400006209 38 Kaufman J Cook A Arny L Jones B Pittinsky T Problems defining resiliency: illustrations from the study of maltreated children Dev Psychol 1994 6 1 215 29 10.1017/s0954579400005964 39 Haskett ME Nears K Ward CS McPherson AV Diversity in adjustment of maltreated children: factors associated with resilient functioning Clin Psychol Rev 2006 26 6 796 812 10.1016/j.cpr.2006.03.005 16797807 40 Luthar SS Sexton CC Maternal drug abuse versus maternal depression: vulnerability and resilience among school-age and adolescent offspring Dev Psychol 2007 19 1 205 25 10.1017/s0954579407070113 41 Hoorelbeke K Marchetti I De Schryver M Koster EH The interplay between cognitive risk and resilience factors in remitted depression: a network analysis J Affect Disord 2016 195 96 104 10.1016/j.jad.2016.02.001 26878206 42 Ozawa C Suzuki T Mizuno Y Tarumi R Yoshida K Fujii K Resilience and spirituality in patients with depression and their family members: a cross-sectional study Compr Psychiatry 2017 77 53 9 10.1016/j.comppsych.2017.06.002 28628806 43 Wu G Feder A Cohen H Kim JJ Calderon S Charney DS Understanding resilience Front Behav Neurosci 2013 7 10 10.3389/fnbeh.2013.00010 23422934 44 Hakamata Y Lissek S Bar-Haim Y Britton JC Fox NA Leibenluft E Attention bias modification treatment: a meta-analysis toward the establishment of novel treatment for anxiety Biol Psychiatry 2010 68 11 982 90 10.1016/j.biopsych.2010.07.021 20887977 45 Laird KT Krause B Funes C Lavretsky H Psychobiological factors of resilience and depression in late life Transl Psychiatry 2019 9 1 88 10.1038/s41398-019-0424-7 30765686 46 Pardeller S Kemmler G Hoertnagl CM Hofer A Associations between resilience and quality of life in patients experiencing a depressive episode Psychiatry Res 2020 292 113353 10.1016/j.psychres.2020.113353 32771836 47 Post F Pardeller S Frajo-Apor B Kemmler G Sondermann C Hausmann A Quality of life in stabilized outpatients with bipolar I disorder: associations with resilience, internalized stigma, and residual symptoms J Affect Disord 2018 238 399 404 10.1016/j.jad.2018.05.055 29909303 48 Lee D Cha B Park CS Kim BJ Lee CS Lee SJ Effects of resilience on quality of life in patients with bipolar disorder J Affect Disord 2017 207 434 41 10.1016/j.jad.2016.08.075 27969568
PMC010xxxxxx/PMC10352644.txt
==== Front Respirol Case Rep Respirol Case Rep 10.1002/(ISSN)2051-3380 RCR2 Respirology Case Reports 2051-3380 John Wiley & Sons, Ltd Chichester, UK 10.1002/rcr2.1191 RCR21191 Case Report Case Reports An effective case of bronchoscopic balloon dilatation for tuberculous bronchial stenosis BALLOON DILATION FOR BRONCHIAL STENOSIS Ichikawa et al. Ichikawa Yukari https://orcid.org/0009-0008-3406-5554 1 [email protected] Kurokawa Koji 1 Furusho Shiho 1 Nakatsumi Yasuto 1 Yasui Masahide 2 Katayama Nobuyuki 1 1 Department of Respiratory Medicine Kanazawa Municipal Hospital Kanazawa Japan 2 Department of Respiratory Medicine National Hospital Organization Nanao Hospital Nanao Japan * Correspondence Yukari Ichikawa, Department of Respiratory Medicine, Kanazawa Municipal Hospital, 3‐7‐3 Heiwa‐machi, Kanazawa, Ishikawa 921‐8105, Japan. Email: [email protected] 18 7 2023 8 2023 11 8 10.1002/rcr2.v11.8 e0119112 4 2023 09 7 2023 © 2023 The Authors. Respirology Case Reports published by John Wiley & Sons Australia, Ltd on behalf of The Asian Pacific Society of Respirology https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an open access article under the terms of the http://creativecommons.org/licenses/by-nc-nd/4.0/ License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non‐commercial and no modifications or adaptations are made. Abstract Endobronchial tuberculosis often causes bronchial stenosis. Balloon dilation is a minimally invasive and effective bronchoscopic intervention for bronchial stenosis; however, reports on balloon dilation in older individuals are limited. We present a case of a 77‐year‐old woman with endobronchial tuberculosis and clarify the efficacy and safety of balloon dilation. She presented with dyspnea, right lung atelectasis, and respiratory failure 55 days after initiation of antituberculosis therapy. We performed bronchoscopic balloon dilatation for the right main bronchial stenosis. Consequently, respiratory failure rapidly improved. Chest computed tomography (CT) showed improved lung atelectasis; however, severe bronchial stenosis and rhonchi persisted. Therefore, we performed a second balloon dilatation. CT 3 months after the first balloon dilation showed right upper bronchial stenosis and right lung middle lobe atelectasis. Restenosis was absent 21 months after third balloon dilatation. Bronchoscopic balloon dilation is effective for restenosis with repeated treatment and can be safely performed in older individuals. Endobronchial TB (EBTB) causes bronchial stenosis; therefore, various endoscopic interventions may be required. Balloon dilatation is less invasive and has a high restenosis rate compared with the other interventions, and repeated performance is considered safe in older patients. This study presents the case of an older patient with tuberculous bronchial stenosis treated with repeated balloon dilatation to aid clinicians in persistent atelectasis and respiratory failure. balloon dilatation bronchial stenosis bronchial tuberculosis source-schema-version-number2.0 cover-dateAugust 2023 details-of-publishers-convertorConverter:WILEY_ML3GV2_TO_JATSPMC version:6.3.2 mode:remove_FC converted:18.07.2023 Ichikawa Y , Kurokawa K , Furusho S , Nakatsumi Y , Yasui M , Katayama N . An effective case of bronchoscopic balloon dilatation for tuberculous bronchial stenosis. Respirology Case Reports. 2023;11 :e01191. 10.1002/rcr2.1191 Associate Editor: Phan Nguyen ==== Body pmcINTRODUCTION Tuberculosis (TB) is a global health concern and a common disease in older individuals in Japan. The proportion of those aged 80 years and above among the total notified TB cases was 29.9% (n = 3440/11,519), and of those aged 70 years and above was 63.5% (n = 7314), as reported by the Ministry of Health, Labour and Welfare in Japan in 2021. Endobronchial TB (EBTB) causes bronchial stenosis; therefore, various endoscopic interventions may be required. 1 Balloon dilatation is less invasive and has a high restenosis rate compared with the other interventions, and repeated performance is considered safe in older patients. This study presents the case of an older patient with tuberculous bronchial stenosis treated with repeated balloon dilatation to aid clinicians in persistent atelectasis and respiratory failure. CASE REPORT A 77‐year‐old woman had chronic productive cough and appetite loss and lost 10 kg of weight in 3 months. Chest computed tomography (CT) revealed multiple granular shadows in all lung fields, nodular shadows in the right upper lobe, and bronchial stenosis with thickening of the bronchial lumen from the right main bronchus to the bronchus intermedius. Acid‐fast bacterial staining of sputum was smear‐positive, and polymerase chain reaction for TB was positive. The patient was diagnosed with pulmonary and endobronchial tuberculosis and was admitted to the isolation ward of our hospital. Physical examination revealed temperature, 37°C; blood pressure, 134/81 mmHg; pulse rate, 90 beats/min; and oxygen saturation, 97%. Her breath sounds were rhonchi. Right lower lobe atelectasis and respiratory failure appeared on the 55th day of anti‐TB medications. One week later, CT showed obstruction of the right main bronchus (Figure 1A, D) and exacerbation of right atelectasis on chest X‐ray (Figure 2A); therefore, she underwent flexible bronchoscopic balloon dilation under general anaesthesia. Bronchoscopy (1 T‐260; Olympus Corporation, Tokyo, Japan) revealed that the right main bronchus was completely obstructed (Figure 3A). The bronchial stenosis was treated using CRE™ Pulmonary Balloon Dilator (No. 5033) (Boston Scientific Corporation, Marlborough, MA, USA) with the pressure increased to 3 atm to dilate the right main bronchus for 30 s. After the balloon dilatation was repeated thrice, the bronchial lumen expanded (Figure 3B), and the right atelectasis improved within 1 week after the balloon dilatation (Figure 2B, C). Severe bronchial stenosis persisted, and bronchoscopy was performed again 2 weeks after the first balloon dilation. The stenosis of the bronchus intermedius was treated by increasing the pressure to 3 atm for 30 s and 9 atm for 30 s. Three months after the first balloon dilation, CT showed the right main bronchus restenosis (Figure 1B, E). The patient underwent the third balloon dilatation. Bronchoscopic high‐pressure (9 atm) balloon dilatation was performed six times for the right main bronchial stenosis (Figure 3C) and thrice for the bronchus intermedius stenosis, with 30 s per cycle. The bronchoscope was passed through the bronchial stenosis following the procedure (Figure 3D). Restenosis was not observed from 4 to 21 months after the third balloon dilatation (Figure 1C, F). FIGURE 1 The course of chest computed tomography & virtual 3D bronchial model. (A) Before the first balloon dilatation, computed tomography (CT) showed that the right main bronchus was almost completely obstructed. (B) Before the third balloon dilatation, CT showed that the right main bronchus stenosis persisted. (C) Four months after the third balloon dilatation, right main bronchus stenosis was improved. Virtual 3D bronchial model constructed from CT slices. (D) Before the first balloon dilatation. (E) Before the third balloon dilatation. (F) Four months after the third balloon dilatation. FIGURE 2 The course of chest x‐ray. (A) Before the first balloon dilatation, right atelectasis is visible. (B) One day after the first balloon dilatation, right atelectasis was almost improved. (C) One week after the first balloon dilatation, right atelectasis disappeared. FIGURE 3 Bronchoscopic image. The black arrow shows the right main bronchus; the white arrow shows the left main bronchus. (A) Before the first balloon dilatation, the right main bronchus was almost completely obstructed. (B) After the first balloon dilatation, the right main bronchial lumen expanded. (C) Before the third balloon dilatation, restenosis of the right main bronchus was observed. (D) After the third balloon dilatation, stenosis of the right main bronchus improved. DISCUSSION Bronchoscopic intervention can be less invasive than surgical treatment and is helpful for tuberculous endobronchial stenosis. Various interventions, such as balloon dilatation, laser resection, electrocautery, cryotherapy, and stent implantation have been proposed. 1 , 2 Balloon dilation is less effective in cases of bronchial stenosis complicated by bronchomalacia, and stent implantation may be useful in some cases. 1 However, balloon dilation alone seems to be sufficient in many cases. 3 , 4 In this case, after the first balloon dilatation, atelectasis improved, but bronchial stenosis persisted. Despite the second balloon dilatation, bronchial stenosis was severe; therefore, a paediatric Dumon stent (diameter 9 mm) could not be implanted. Single or multiple balloon dilatations were successful in 82 patients (73%) with tuberculous tracheobronchial stenosis, with a mean follow‐up of 30.3 months. 3 Lee followed up 131 patients with tracheobronchial stenosis due to tuberculosis for 5 years and found that 19 patients (29.7%) underwent balloon dilatation alone. 2 Repeated use of balloon dilatation resulted in avoiding the need for stent placement, resulting in a less invasive approach for the patient and reducing the risk of complications such as granulation tissue growth and stent migration. We performed three balloon dilatations over 3 months and observed no recurrence for 21 months subsequently. Cho et al. reported that symptoms recurred from 1 day to 113 months (mean, 13 months) after repeated balloon dilatation. 3 Restenosis may occur after a long period, and further long‐term observations are therefore necessary in this case. In previous reports, patients had a mean age of 37 years 3 and 50 ± 18 years. 2 Limited reports are available on older patients. This case report suggests that repeated balloon dilation may be minimally invasive, even in elderly patients. There is no established standard technique for balloon dilation; therefore, balloon pressure and dilation duration vary. In many cases, the balloon pressure was 3–5 atm with the highest pressure being 16 atm. 5 The reported dilation duration varied widely, and was 10–30 s in many cases, but the longest time was 40 min. 4 , 5 Fu et al. reported that by increasing the duration of balloon dilatation alone at a high pressure of 14 atm and a duration of 40 min, restenosis was not observed, and complications were absent. 4 Further consideration should be given to the optimal duration of dilatation required to avoid restenosis, as well as the safety of balloon dilatation with higher pressure and a longer time. The procedure depends on the state of the bronchial walls and each patient's respiratory condition; therefore, it is difficult to standardize the balloon pressure, dilatation duration, and frequency. We report a case in which bronchoscopic balloon dilation was effective in treating tuberculous bronchial stenosis. Balloon dilation quickly improves the patient's symptoms and is effective for restenosis with repeated treatment, suggesting that it can be safely performed in older patients. Evaluation of the long‐term prognosis is warranted. AUTHOR CONTRIBUTIONS Yukari Ichikawa cured for the patient, prepared the data, and edited the manuscript. Koji Kurokawa and Shiho Furusho cured for the patient. Yasuto Nakatsumi, Masahide Yasui reviewed the manuscript. Nobuyuki Katayama drafted and reviewed the manuscript. All authors approved the final manuscript. CONFLICT OF INTEREST STATEMENT None declared. ETHICS STATEMENT The authors declare that appropriate written informed consent was obtained for the publication of this manuscript and accompanying images. ACKNOWLEDGMENTS We would like to thank Editage (www.editage.com) for English language editing. DATA AVAILABILITY STATEMENT The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions. ==== Refs REFERENCES 1 Mondoni M , Repossi A , Carlucci P , Centanni S , Sotgiu G . Bronchoscopic techniques in the management of patients with tuberculosis. Int J Infect Dis. 2017;64 :27–37. 10.1016/j.ijid.2017.08.008 28864395 2 Lee KCH , Tan S , Goh JK , Hsu AAL , Low SY . Long‐term outcomes of tracheobronchial stenosis due to tuberculosis (TSTB) in symptomatic patients: airway intervention vs. conservative management. J Thorac Dis. 2020;12 :3640–3650. 10.21037/JTD-20-670 32802443 3 Cho YC , Kim JH , Park JH , Shin JH , Ko HK , Song HY , et al. Tuberculous tracheobronchial strictures treated with balloon dilation: a single‐center experience in 113 patients during a 17‐year period. Radiology. 2015;277 :286–293. 10.1148/radiol.2015141534 25955577 4 Fu EQ , Jin FG . Novel bronchoscopic balloon dilation for patients with bronchostenosis caused by bronchial tuberculosis: a case report. J Med Case Reports. 2014;24 (8 ):225. 10.1186/1752-1947-8-225 5 Fang Y , You X , Sha W , Xiao H . Bronchoscopic balloon dilatation for tuberculosis‐associated tracheal stenosis: a two case report and a literature review. J Cardiothorac Surg. 2016;29 (11 ):21. 10.1186/s13019-016-0417-z
PMC010xxxxxx/PMC10352645.txt
==== Front Open Forum Infect Dis Open Forum Infect Dis ofid Open Forum Infectious Diseases 2328-8957 Oxford University Press US 10.1093/ofid/ofad331 ofad331 Major Article AcademicSubjects/MED00290 Geographic Targeting of COVID-19 Testing to Maximize Detection in Los Angeles County https://orcid.org/0000-0001-8875-2415 Jia Katherine M Department of Epidemiology, Center for Communicable Disease Dynamics, Harvard T.H. Chan School of Public Health, Boston, Massachusetts, USA https://orcid.org/0000-0001-9511-6142 Kahn Rebecca Department of Epidemiology, Center for Communicable Disease Dynamics, Harvard T.H. Chan School of Public Health, Boston, Massachusetts, USA Fisher Rebecca Los Angeles County Department of Public Health, Acute Communicable Disease Program, Los Angeles, California, USA Balter Sharon Los Angeles County Department of Public Health, Acute Communicable Disease Program, Los Angeles, California, USA https://orcid.org/0000-0003-1504-9213 Lipsitch Marc Department of Epidemiology, Center for Communicable Disease Dynamics, Harvard T.H. Chan School of Public Health, Boston, Massachusetts, USA Department of Immunology and Infectious Diseases, Harvard T.H. Chan School of Public Health, Boston, Massachusetts, USA Correspondence: Katherine M. Jia, MSc, Department of Epidemiology, Center for Communicable Disease Dynamics, Harvard T.H. Chan School of Public Health, 677 Huntington Avenue, Boston, MA 02115 ([email protected]). Potential conflicts of interest. R.K. discloses consulting fees from the Pan American Health Organization. M.L. has received consulting fees from Janssen and Merck, honoraria from Bristol Myers Squibb and Sanofi Pasteur, and institutional grant support from Pfizer. All other authors report no potential conflicts. 7 2023 27 6 2023 27 6 2023 10 7 ofad33104 5 2023 21 6 2023 23 6 2023 18 7 2023 © The Author(s) 2023. Published by Oxford University Press on behalf of Infectious Diseases Society of America. 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs licence (https://creativecommons.org/licenses/by-nc-nd/4.0/), which permits non-commercial reproduction and distribution of the work, in any medium, provided the original work is not altered or transformed in any way, and that the work is properly cited. For commercial re-use, please contact [email protected] Abstract Background Many severe acute respiratory syndrome coronavirus 2 infections have not been detected, reported, or isolated. For community testing programs to locate the most cases under limited testing resources, we developed and evaluated quantitative approaches for geographic targeting of increased coronavirus disease 2019 testing efforts. Methods For every week from December 5, 2021, to July 23, 2022, testing and vaccination data were obtained in ∼340 cities/communities in Los Angeles County, and models were developed to predict which cities/communities would have the highest test positivity 2 weeks ahead. A series of counterfactual scenarios were constructed to explore the additional number of cases that could be detected under targeted testing. Results The simplest model based on most recent test positivity performed nearly as well as the best model based on most recent test positivity and weekly tests per 100 persons in identifying communities that would maximize the average yield of cases per test in the following 2 weeks and almost as well as the perfect knowledge of the actual positivity 2 weeks ahead. In the counterfactual scenario, increasing testing by 1% 2 weeks ahead and allocating all tests to communities with the top 10% of predicted positivity would yield a 2% increase in detected cases. Conclusions Simple models based on current test positivity can predict which communities may have the highest positivity 2 weeks ahead and hence could be allocated with more testing resources. COVID-19 community testing Morris-Singer Fund U01CA261277 US National Cancer Institute of the National Institutes of Health Department of Health and Human Services 10.13039/100000016 ==== Body pmcDuring the coronavirus disease 2019 (COVID-19) pandemic, infected individuals without access to testing and unaware of infection status may have further transmitted the disease. Therefore, testing is a key disease control strategy. However, from February 2020 to September 2021, it was estimated that only 1 out of 4 severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infections in the United States was detected and reported [1]. When cases surge, community testing (eg, free tests available for residents provided by the government partnering with local test providers) may be scaled up to detect as many cases as possible and minimize further transmissions. With limited resources, testing resources are often prioritized to those who are most likely positive (eg, showing symptoms or exposed) [2]. Du et al. [3] formalized test allocation as an optimization problem and showed that more cases could be detected if allocation was optimized based on symptom severity and age group. However, our interest is in prioritizing the communities most likely to have the highest test positivity, as this is the quantity that predicts how many cases will be identified per unit testing effort. Within a jurisdiction, a test allocation strategy pertaining to communities should be developed. Motivated by the need for an efficient test allocation strategy to identify as many cases as possible under resource constraints, we explored how basic testing data and simple statistical models could be used to inform targeted testing. We used data from cities/communities in Los Angeles (LA) County to develop models to predict test positivity in each city/community 2 weeks ahead. We also hypothesized that targeted testing could detect more cases overall by prioritizing the cities/communities with the highest model-predicted test positivity compared with nontargeted testing. METHODS Study Population LA County is the largest county in the United States by population size and has a total of 346 cities/communities as statistical areas for data collection (hereafter referred to as communities) [4]. We obtained daily testing and vaccination data for each community corresponding to individual residential address from the LA County Department of Public Health. Long Beach and Pasadena were not included in the data because they have their own health departments. The 2018 population estimates were used as denominators for calculating the cumulative incidence per 100 residents and the number of tests per 100 residents [5], while the 2019 population estimates were used to calculate vaccination coverage [6]. Testing Data All cases were determined using the case definition from The Council of State and Territorial Epidemiologists (CSTE) [7, 8]. A confirmed case could be defined by a positive nucleic acid amplification test result, and a probable case could be defined by a positive diagnostic antigen test performed by a provider certified under the Clinical Laboratory Improvement Amendments [9]. Confirmed (polymerase chain reaction) and probable cases (antigen) were summed together as the total number of positive tests in the data set, which was used as the numerator for computing test positivity. LA County cases included reinfections, defined as a repeat positive ≥90 days after a previous confirmed case. Testing data were compiled from multiple data sources including electronic lab reporting and medical provider reporting. However, at-home and over-the-counter tests were not reported to the Department of Public Health. Inclusion and Exclusion Criteria Data from June 13, 2021, to July 23, 2022, were included in the analysis, covering both the Delta and Omicron waves (Supplementary Figure 1). Communities were highly heterogeneous in population sizes (interquartile range [IQR], 3592–41 342; range, 0–220 424). We also excluded communities in the lowest 5% of population sizes (<300 individuals) because of the small number of tests conducted and therefore the highly unstable positivity. In addition, weeks with <50 tests were excluded to avoid extreme values in test positivity, but other weeks with ≥50 tests were included in the same community. Targeted Testing Strategies Three hypothetical strategies of geographic targeting for intensified COVID-19 testing were evaluated. Communities were selected for intensified testing based on the following strategies: (1) model-predicted test positivity 2 weeks forward, (2) random selection, or (3) perfect knowledge into the future—that is, the observed test positivity 2 weeks forward (not feasible in practice). For each week, the top 10% of communities with the highest predicted or observed positivity were selected (except for random selection) for intensified testing, with the same percent increase in testing for all the selected communities. We constructed counterfactual scenarios to evaluate the strategies by calculating the additional number of cases detected and number needed to test (NNT) per case detected in the targeted communities (see descriptions below). Model-based geographic targeting was expected to yield more additional cases compared with random selection but fewer cases compared with perfect knowledge into the future. Statistical Models for Model-Based Geographic Targeting For the model-based geographic targeting for strategy 1, we developed regression models using the most recent week (t) to predict weekly test positivity for each community 2 weeks ahead (t + 2) and select communities with top 10% predicted positivity for intensified testing. Predictors in the models included (1) test positivity of the most recent week (t), (2) the most recent 3-week average test positivity (t−2,t−1 and t), (3) testing rate of the most recent week (weekly tests per 100 persons at t), (4) cumulative proportion of residents (aged 12 years and older) who were fully vaccinated as vaccines were widely available for this group from mid-May 2021 [10], and (5) cumulative proportion of residents who tested positive relative to the population (per 100 persons). Variable definitions are in Supplementary Table 1. The 3 statistical models were as follows: (1) the “simplest” model with 1 predictor—the most recent positivity (%posi,t), (2) the “reduced” model with 2 predictors (%posi,t and testsper100i,t), and (3) the “full” model with all predictors. Variables in the reduced model were selected because they were more accessible compared with other variables, while the variable (%posi,t) in the simplest model was selected because this variable was the strongest single predictor among all variables. All models were also evaluated by replacing the most recent positivity (%posi,t) with the average positivity over the past 3 weeks (Avg%posi,t−2tot). To account for the temporal autocorrelation of the residuals, different specifications were tested: (1) ordinary least squares (OLS), (2) second-order autoregressive (AR-2) process, and (3) mixed-effects models with community-specific random intercepts. Model Development and One-Step Forward Forecasting A total of 58 weeks of data were used in this analysis. From June 13, 2021, to December 25, 2021 (Delta wave: 28 weeks), data were used for developing the initial models, which were tested on the first week of the Omicron period. One-step forward forecasting was performed by refitting the models every week for the Omicron wave (December 26, 2021, to July 23, 2022: 30 weeks) using all previous data (both Delta and Omicron) to predict 2 weeks forward from each week of testing (Supplementary Figure 1). The Delta- and Omicron-predominant periods were defined based on sequencing results from the California Department of Public Health [11]. The steps in model development and evaluation are outlined in Supplementary Figure 2A. Model Evaluation To explore how many additional cases could have been detected under targeted testing, a series of counterfactual scenarios was constructed (Supplementary Figure 2B): (1) for each week t, 10% of communities with the highest predicted positivity 2 weeks ahead (%posi,t+2^ for week t+2) were selected; (2) the total number of tests in LA County increased by 1%, and all the additional tests were allocated to the selected communities in t+2, such that all the selected communities had tests increased by a common multiplier m% compared with tests in t; (3) assuming that positivity is unaffected by the relatively small increase in testing, the hypothetical number of cases could be modeled by multiplying the observed test positivity in t+2 with the hypothetical number of tests allocated in step 2. The total number of cases in LA County 2 weeks ahead is given by, for targeted communities j and untargeted communities i, Σj(100+m)%⋅%posj,t+2⋅testsj,t+Σi%posi,t+2⋅testsi,t. The models were evaluated based on 2 metrics: (1) the total number of additional cases 2 weeks ahead, Σjm%⋅%posj,t+2⋅testsj,t, and (2) the NNT for each case, Σjtestsj,t+2Σjcasesj,t+2, in the selected communities j. Evaluation of Targeted Testing Strategies For the other 2 strategies (ie, random selection and perfect knowledge of future positivity), the same evaluation procedures were applied as with the model-based strategy. Specifically, we used the 2 metrics defined above to evaluate the 3 testing strategies for each of the 30 weeks from December 26, 2021, to July 23, 2022. We expected that random allocation of additional tests (ie, 10% of the communities were randomly selected for intensified testing) would be the worst-case scenario that gave a lower bound on the number of additional cases that could be detected, while selecting the communities based on the perfect knowledge of the future test positivity would give an upper bound. The best model was the model that can maximize the number of additional cases (defined in metric 1) given the number of additional tests. Analyses were conducted using R 3.6.2 (R Foundation for Statistical Computing, Vienna, Austria) [12]. Data and R codes are available at https://github.com/c2-d2/LAC_testing. RESULTS Descriptive Statistics A total of 295 communities were included in the training set (the Delta period) and 303 in the validation set (the Omicron period). On average, each community administered 8 tests weekly per 100 persons for the Delta period and 9 tests weekly per 100 persons for the Omicron period. Weekly positivity was 2.8% on average for the Delta period and 7.8% for the Omicron period. Model Predictions Current positivity (%posi,t) and 3-week average positivity (Avg%posi,t−2tot) were the strongest predictors for positivity 2 weeks ahead (Supplementary Table 2). Supplementary Figure 3 shows estimates from all models that include %posi,t (and Avg%posi,t−2tot in Supplementary Figure 4). From December 26, 2021, to July 23, 2022 (the Omicron wave), the simplest model with test positivity alone (%posi,t) correctly predicted ∼48% of communities with top 10% positivity 2 weeks ahead. During the model evaluation period, the best model that maximized the average additional cases was the mixed-effects model with %posi,t and testsper100i,t as predictors, but this model performed only slightly better than the simplest model by successfully predicting ∼49% of the communities. In general, predictive performances were similar across models (Supplementary Figure 5). Additional Cases Under Targeted Testing In the counterfactual scenario, testing was increased for the next 2 weeks in communities with the top 10% predicted positivity. Throughout the testing period, 2% more cases could be detected in a week on average if testing were increased by 1%, and all the additional tests were allocated to the selected communities. In general, the number of additional cases was greater in weeks with higher positivity (Figure 1). For example, 4000 more cases could be detected in the week of January 16, 2022, if 1% more tests (n = 17 200) were performed. Of note, in Figure 1, the curves for the simplest and best models are indistinguishable in most weeks, slightly below the best possible model performance (using perfect knowledge of future positivity). Figure 1. Number of additional cases in a week under the counterfactual scenario where tests were increased by 1% in LA County compared with 2 weeks back, with all additional tests allocated to communities in the top 10% of predicted positivity. The blue line (the uppermost line) represents the number of additional cases if communities were ranked based on perfect knowledge of future test positivity, %posi,t+2. While requiring knowledge of the future and thus not practicable, this line serves as the upper bound of the best possible prediction. The purple line represents the yields if communities were randomly selected and serves as the lower bound. The red and green lines represent the number of additional cases that could be detected if communities were selected based on the best and the simplest models, respectively. Test positivity was based on testing data from LA County Department of Public Health. Compared with selecting communities based on perfect knowledge of future positivity, %posi,t+2, the simplest model using most recent positivity, %posi,t, could detect as many as 89% of achievable additional cases (Table 1). Table 1 shows the performance of the simplest and best models. Supplementary Table 3 shows results for all models. Table 1. Mean Number of Additional Cases and Number Needed to Test in the Selected Communities Under Different Test Allocation Strategies, December 26, 2021, to July 23, 2022 No. of Additional Cases NNT Mean No. of Additional Cases Relative to Perfect Knowledge of Future Positivity Mean NNT Relative to Perfect Knowledge of Future Positivity Simplest model most recent positivity (%posi,t) only (OLS) 985 0.887 19.3 1.263 Best model most recent positivity (%posi,t) + most recent tests per 100 (testsper100i,t) (mixed-effects model with community-specific intercepts) 994 0.895 19.0 1.241 Perfect knowledge of future positivity (%posi,t+2) 1110 1.000 15.3 1.000 Random selection 565 0.509 43.8 2.864 Abbreviations: NNT, number needed to test; OLS, ordinary least squares. Reduction in NNT in the Communities Under Intensified Testing Targeting communities with top 10% predicted positivity could reduce the NNT in those communities. If tests were increased by 1% in LA County and all tests were allocated to selected communities under the simplest or the best model, NNT in the selected communities could be reduced by ∼50% on average compared with randomly selecting the communities for increased testing (Figure 2). Note that this NNT considers the 10% selected communities with increased testing only. Figure 2 and Supplementary Table 3 show the NNT under different test allocation strategies. Figure 2. NNT in the selected communities under intensified testing, December 26, 2021–July 23, 2022. The blue line represents NNT in the targeted communities selected based on perfect knowledge of future test positivity, %posi,t+2, which serves as the lower bound for NNT (ie, the most efficient scenario). The purple line (the uppermost line) represents the expected NNT under random selection of communities for targeted testing (ie, the least efficient) and serves as the upper bound. Abbreviation: NNT, number needed to test. DISCUSSION We found that test positivity was temporally correlated—high positivity in the most recent week indicated a high positivity 2 weeks ahead. Through our counterfactual scenarios, we showed that by targeting communities with high positivity in the most recent week, detected cases could increase by 1% to 3% 2 weeks later, on average about 2% given a 1% increase in testing, meaning that these simple approaches can roughly double the yield of additional tests. The simplest model using test positivity alone performed as well as the best model in finding more positive cases and nearly as well as perfect knowledge of future positivity. This simple method could inform a timely prioritization of testing resources to identify cases and mitigate future transmissions. By defining the goal of identifying communities with high future test positivity, we do not make the claim that test positivity is a good indicator of actual disease burden, as this relationship has been shown to be quite dependent on multiple factors; rather, we optimize for test positivity because it is the quantity that predicts the number of cases that can be identified per unit testing effort. This approach is robust to changes of positivity during different phases of an epidemic. That the communities with the highest positivity in the most recent week were likely to have comparatively high positivity in the next 2 weeks was true in both cases when test positivity for overall LA County was on an uptrend or a downtrend. To illustrate this phenomenon, Supplementary Figure 6 shows the epidemic curves for communities that had the highest cumulative percent testing positive. At the start of the Omicron wave, positivity increased exponentially, but it declined slowly after the peak. Therefore, communities with the highest positivity were likely to continue to top the list in terms of positivity in the next few weeks, even when overall LA County positivity was on a downtrend. If positivity continued to decrease in the next few weeks in a community being targeted, due to either increased testing or reduced incidence, then that area would eventually be dropped from the list as the model would pick up some other community with a higher predicted positivity. The model corrected itself by making new predictions every week and updating the list based on the most recent data. The simple method is ready to use by any local public health department. This method does not require modeling expertise or complex inaccessible data and provides quick and useful predictions on where the positivity could be highest in the next 2 weeks. However, test allocation strategies informed by complex dynamic transmission models that predict the full epidemic trajectory would require detailed data for model parameterization. Empirical data are often unavailable to account for factors including vaccination coverage, frequency of breakthrough infections, waning immunity, and compliance with mitigation measures. Using this simple method, nearly 90% of additional cases could be detected when compared with perfect knowledge of future positivity. In this case, it is difficult to imagine that the resources devoted to more elaborate approaches would be justified for a further 10% improvement in outcome, which is the upper bound based on perfect knowledge of the future. In addition, this simple approach could substantially improve test efficiency, as the targeted areas based on the simplest model had half the NNT of randomly selected areas. More generally, NNT is a useful measure of test efficiency and could be widely used to inform test allocation strategy. There is a trade-off between accuracy and the prediction horizon—the longer the prediction horizon, the less accurate the prediction. Out of logistical considerations, the models were used to predict positivity 2 weeks ahead, as it might take ≥2 weeks for the implementation team to increase testing in a community. Model performance decreased slightly when the prediction horizon increased to 3 weeks, but the model would still be useful for informing test allocation (Supplementary Figure 5, Supplementary Table 3). This study has policy implications beyond the COVID-19 pandemic, and our model could be applied in other infectious disease outbreaks. In general, targeted testing is needed to save resources by prioritizing tests to those with the greatest needs and to maximize the efficiency of testing [13]. For other diseases like HIV where testing services are limited in some countries, targeted testing is also recommended by prioritizing communities with high HIV positivity [14]. We proposed and verified a straightforward approach to target geographical communities using a relatively minimal set of data (eg, the number of tests per 100 persons and positivity by statistical areas) to prioritize testing for transmissible diseases. If identifying subpopulations with inadequate testing, finer data disaggregated by age, race and ethnicity, or other demographic characteristics would be needed to detect any disparity in access to tests or test positivity [15]. If such data were available, the test allocation strategy could take any disparity into account and address undertesting by looking at test positivity in those subgroups. An ecological analysis on testing trends by ZIP code in New York City showed the total number of tests per capita was higher in ZIP codes with higher percentages of the White population during the early stage of the pandemic (March 2 to April 6, 2020) [16]. Since August 2020, all certified laboratories are required to report demographics for every test [17], but those fields may have a high proportion of data missing unless there is a proper enforcement of the mandate [15]. This study has some limitations. The first limitation is that we did not include at-home or over-the-counter tests as they were not reported to the Department of Public Health. Impacts of the unreported tests depend on positivity and how well the positivity correlates with community testing results. The model evaluation period from December 26, 2021, to July 23, 2022, covered a time period in which at-home tests were more widely available, and the models performed well during that period. If temporal correlation between positivity in the communities is weakened by the wide availability of at-home tests over the course of the pandemic, we would expect the efficiency of the targeting strategy to decrease. Therefore, this approach is expected to be more useful when at-home testing is rare or in settings where the testing rate is more stable over time. Second, we assumed that positivity was unaffected by the number of tests. This assumption was plausible when tests and cases increased by a small amount compared with the large number of untested infections. A 1% increase for testing in LA County was equivalent to a 12%–45% increase of testing in the top 10% communities varying by week. We were not able to test this assumption empirically (ie, to test whether the 1% increase in tests was large enough to bring down the positivity). In Supplementary Figure 7, percent change in positivity is plotted against percent change in tests (compared with the previous weeks) for communities with top 10% cumulative incidence. However, in the majority of weeks from December 26, 2021, to July 23, 2022, test positivity did not decline when the number of tests increased from 0% to 50% compared with the previous week (upper bound corresponding to 1% increase in overall LA County). In practice, if positivity decreased as a result of intensified testing, the chance of the community being targeted in the coming week would also decrease. Ultimately, an increase in tests should be determined by resource availability, and here the 1% increase served as an example only. Third, we did not explore any disparity in testing and positivity across race, ethnicity, or socioeconomic status. Our simple goal was to predict in which communities the additional tests would yield the most positive cases. Further analysis would be needed to identify any systematic disparity in testing and to guide test allocation. Importantly, our model is just one piece of information that can be used to inform decisions and may be considered in conjunction with other factors such as community vulnerability and high-priority population subgroups. Future research could consider how to incorporate these additional factors into quantitative approaches. Fourth, we did not consider the spatial correlation of the positivity between communities. However, given the good performance of the model with test positivity as the sole predictor (the simplest model), spatial correlation may not add much to the predictive performance. Another limitation is that the location indicator of each test corresponded to the residential address of each individual, rather than the testing site. This indicator added complexity to implementing targeted testing, as the communities where people lived might not be the same communities as were tested. Finally, we excluded small communities with a population size below the 5th percentile (ie, <300) and weeks with <50 tests. However, in practice, health authorities should look into reasons for the high positivity in those geographic areas, and more tests may be needed. CONCLUSIONS We showed that simple models based on current positivity and proportion of tests relative to population size are useful for identifying communities with high positivity in the next 2 weeks. In particular, more tests can be allocated to communities with the highest current positivity. Using this approach to allocate tests could detect more cases and save resources. Although targeting communities with the highest predicted positivity is shown to be an efficient strategy, health authorities should be cognizant of the community vulnerability and any systematic disparity within the communities when allocating tests. Supplementary Material ofad331_Supplementary_Data Click here for additional data file. Acknowledgments We thank Phoebe Danza, Remy Landon, Chelsea Foo, the ACDC Morning Data Team (Dr. Katie Chun, Harry Persaud, Tuff Witarama, Mark Johnson, Laureen Masai, Zoe Thompson, Dr. Ndifreke Etim), and Dr. Paul Simon from the LA County Department of Public Health for providing data and valuable support on this work. We also thank Kathy L. Brenner from Harvard T.H. Chan School of Public Health and Dr. Samantha Jones from the Fellowships & Writing Center of Harvard University for providing support in writing the manuscript. Financial support. This work was supported in part by the Morris-Singer Fund, as well as Award Number U01CA261277 from the US National Cancer Institute of the National Institutes of Health (M.L.). Surveillance efforts were supported by a cooperative agreement awarded by the Department of Health and Human Services (HHS) with funds made available under the Coronavirus Preparedness and Response Supplemental Appropriations Act, 2020 (P.L. 116-123); the Coronavirus Aid, Relief, and Economic Security Act, 2020 (the “CARES Act”; P.L. 116-136); the Paycheck Protection Program and Health Care Enhancement Act (P.L. 116-139); the Consolidated Appropriations Act and the Coronavirus Response and Relief Supplement Appropriations Act, 2021 (P.L. 116-260); and/or the American Rescue Plan of 2021 (P.L. 117-2). Author contributions. R.K., M.L., R.F., and S.B. developed the concept for the study. K.J., R.K., and M.L. developed and conducted the analysis. K.J. drafted the article. K.J., R.K., M.L., and R.F. edited the article. S.B. provided public health expertise that contributed to the development of the research and interpretation of the data in the article. All authors reviewed and finalized the article. Statement of data availability. Data are available at https://github.com/c2-d2/LAC_testing. Patient consent. Not applicable. All data were aggregated to statistical area units for public health surveillance. This study does not include factors necessitating patient consent. Supplementary Data Supplementary materials are available at Open Forum Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author. ==== Refs References 1 Centers for Disease Control and Prevention . Estimated COVID-19 burden. 2021. Available at: https://www.cdc.gov/coronavirus/2019-ncov/cases-updates/burden.html. Accessed August 13, 2022. 2 Health Canada . Priority strategies to optimize testing and screening for COVID-19 in Canada: report. 2021. Available at: https://www.canada.ca/content/dam/hc-sc/documents/services/drugs-health-products/covid19-industry/medical-devices/testing-screening-advisory-panel/reports-summaries/priority-strategies/priority-strategies-eng.pdf. Accessed August 13, 2022. 3 Du J LJB , LeeS, ZhouX, DempseyW, MukherjeeB. Optimal diagnostic test allocation strategy during the COVID-19 pandemic and beyond. Stat Med 2022; 41 :310–27.34697824 4 County of Los Angeles . Countywide statistical areas (CSA). 2022. Available at: https://data.lacounty.gov/datasets/countywide-statistical-areas-csa/. Accessed August 13, 2022. 5 County of Los Angeles . July 1, 2018 population estimates, prepared by hedderson demographic services for Los Angeles County Internal Services Department (ISD). 2019. 6 County of Los Angeles . COVID-19 vaccine. 2022. Available at: publichealth.lacounty.gov/media/coronavirus/vaccine/vaccine-dashboard.htm. Accessed October 4, 2022. 7 Centers for Disease Control and Prevention . Coronavirus disease 2019 (COVID-19) 2021 case definition. 2021. Available at: https://ndc.services.cdc.gov/case-definitions/coronavirus-disease-2019-2021/. Accessed August 13, 2022. 8 Council of State and Territorial Epidemiologists . Update to the standardized surveillance case definition and national notification for 2019 novel coronavirus disease (COVID-19). 2020. Available at: https://cdn.ymaws.com/www.cste.org/resource/resmgr/ps/positionstatement2020/Interim-20-ID-02_COVID-19.pdf. Accessed October 3, 2022. 9 LA County Department of Public Health . Data source documentation. 2022. Available at: http://dashboard.publichealth.lacounty.gov/covid19_surveillance_dashboard/. Accessed August 13, 2022. 10 Wallace M , WoodworthKR, GarganoJW, et al The Advisory Committee on Immunization Practices’ interim recommendation for use of Pfizer-BioNTech COVID-19 vaccine in adolescents aged 12–15 years—United States, May 2021. MMWR Morb Mortal Wkly Rep 2021; 70 :749–52.34014913 11 California Department of Public Health . Variants in California. 2022. Available at: https://covid19.ca.gov/variants/. Accessed August 13, 2022. 12 R Core Team . R: A Language And Environment for Statistical Computing. R Foundation for Statistical Computing; 2021. Available at: https://www.R-project.org/. Accessed August 13, 2022. 13 Amirian ES . Prioritizing COVID-19 test utilization during supply shortages in the late phase pandemic. J Public Health Policy 2022; 43 :320–4.35414693 14 World Health Organization . Assessment of HIV testing services and antiretroviral therapy service disruptions in the context of COVID-19: lessons learned and way forward in sub-Saharan Africa. 2021. Available at: https://apps.who.int/iris/rest/bitstreams/1393960/retrieve. Accessed August 13, 2022. 15 Servick K . ‘Huge hole’ in COVID-19 testing data makes it harder to study racial disparities. 2020. Available at: https://www.science.org/content/article/huge-hole-covid-19-testing-data-makes-it-harder-study-racial-disparities. Accessed August 13, 2022. 16 Lieberman-Cribbin W , AlpertN, FloresR, TaioliE. Analyzing disparities in COVID-19 testing trends according to risk for COVID-19 severity across New York City. BMC Public Health 2021; 21 :1717.34548041 17 Centers for Disease Control and Prevention . 07/31/2020: lab advisory: update on COVID-19 laboratory reporting requirements. 2022. Available at: https://www.cdc.gov/csels/dls/locs/2020/update-on-covid-19-reporting-requirements.html. Accessed August 13, 2022.
PMC010xxxxxx/PMC10352646.txt
==== Front Open Forum Infect Dis Open Forum Infect Dis ofid Open Forum Infectious Diseases 2328-8957 Oxford University Press US 10.1093/ofid/ofad325 ofad325 Major Article AcademicSubjects/MED00290 Case-Control Study to Estimate the Association Between Tdap Vaccination During Pregnancy and Reduced Risk of Pertussis in Newborn Infants in Peru, 2019–2021 https://orcid.org/0000-0001-6843-3206 Juscamayta-López Eduardo Centro Nacional de Salud Pública, Instituto Nacional de Salud, Lima, Peru Facultad de Salud Pública y Administración (GA, AGL), Universidad Peruana Cayetano Heredia, Lima, Peru Valdivia Faviola Centro Nacional de Salud Pública, Instituto Nacional de Salud, Lima, Peru Soto María Pía Centro Nacional de Salud Pública, Instituto Nacional de Salud, Lima, Peru Horna Helen Centro Nacional de Salud Pública, Instituto Nacional de Salud, Lima, Peru Pajuelo Mónica Facultad de Ciencias y Filosofía, Universidad Peruana Cayetano Heredia, Lima, Peru Correspondence: E. Juscamayta-López, MSc ([email protected]). Potential conflicts of interest. All authors: No reported conflicts of interest. 7 2023 27 6 2023 27 6 2023 10 7 ofad32512 4 2023 19 6 2023 23 6 2023 18 7 2023 © The Author(s) 2023. Published by Oxford University Press on behalf of Infectious Diseases Society of America. 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs licence (https://creativecommons.org/licenses/by-nc-nd/4.0/), which permits non-commercial reproduction and distribution of the work, in any medium, provided the original work is not altered or transformed in any way, and that the work is properly cited. For commercial re-use, please contact [email protected] Abstract Background Despite widespread vaccination, pertussis has re-emerged as a serious public health concern worldwide. Since 2017, Peru has experienced an increase in pertussis cases exhibiting a higher risk of severity and death in young infants. Thus, a dose of the tetanus, diphtheria, and acellular pertussis (Tdap) vaccine is recommended for pregnant women in the third trimester. Although evidence suggests the maternal Tdap vaccine is safe and effective, its association with a reduced risk of pertussis in developing countries remains poorly investigated. Methods We conducted a case-control study to evaluate the association between Tdap vaccination during pregnancy and reduction in the risk of pertussis among infants aged <2 months in Peru. Pertussis cases and controls treated in healthcare facilities nationwide between 2019 and 2021 and confirmed by real-time polymerase chain reaction were included. The controls were randomly selected from test-negative patients. Odds ratios (ORs) and vaccine effectiveness (VE) were calculated using a multiple logistic regression model and 1 − (OR) × 100%, respectively. Results Fifty cases and 150 controls were included in the analysis. The mothers of 4% of cases and 16.7% of controls received Tdap vaccination during pregnancy, resulting in an OR of 0.19 (95% confidence interval [CI], .04–.86) and VE of 81% (95% CI, 14%–96%) for preventing pertussis in infants. Conclusions Peruvian infants <2 months old whose mothers received the Tdap vaccine in the third trimester of pregnancy had a significantly lower risk of pertussis. The Tdap vaccination is thus an effective intervention to reduce the burden of pertussis in at-risk populations. Administration of the maternal Tdap vaccine during the third trimester of pregnancy reduced the risk of pertussis in infants younger than 2 months of age in Peru. pertussis maternal immunization Peru pregnancy Tdap National Institute of Health of Peru FONDECYT 10.13039/501100002850 CIENCIACTIVA 10.13039/501100010751 EF033-235-2015 D43 TW007393 Fogarty International Center 10.13039/100000061 US 10.13039/100018390 National Institutes of Health 10.13039/100000002 ==== Body pmcPertussis (whooping cough) is a highly contagious respiratory infectious disease of major public health concern [1]. Despite universal introduction of pertussis vaccines, countries around the world, including Peru, have reported an increased incidence of the disease in the last 2 decades, primarily in infants [2, 3]. The resulting estimated global burden of pertussis is 5.1 million cases and >85 000 deaths in children under 1 year of age [4]. The re-emergence of pertussis has been explained by various factors, including improved disease detection via more-sensitive polymerase chain reaction (PCR)-based diagnostic tests, decreased vaccine-induced immunity, the transition from whole-cell vaccines to less-effective acellular vaccines, and pathogen adaptation [5]. The re-emergence of pertussis has been observed in all age groups [6]. However, infants younger than 3 months of age who are unvaccinated or only partially immunized are at increased risk of severe complications and death due to pertussis [7]. Adults and adolescents are considered the most important sources of infection due to waning immunity and asymptomatic transmission in this group [8]. The household is the most commonly identified source of infection in hospitalized infants under 6 months of age, with the mother representing 39% (95% confidence interval [CI], 33%–45%), the father representing 16% (95% CI, 12%–21%), and grandparents representing 5% (95% CI, 2%–10%) [9]. Various strategies have been proposed to protect the vulnerable population of newborns, including the cocooning strategy, in which postpartum mothers and close contacts are vaccinated against pertussis; however, implementation of this strategy is challenging, and evidence of its effectiveness is limited [10]. In 2011, the United States recommended that all women receive a single dose of the tetanus, diphtheria, and acellular pertussis (Tdap) vaccine in the third trimester of pregnancy [11], which the World Health Organization (WHO) considers to be the most cost-effective strategy for protecting infants during this susceptible period [12]. In addition to the United States [13], this strategy has been instituted in Argentina [14], the United Kingdom [15], Australia, Belgium, and Spain [16]. In Peru, the Ministry of Health recommended that beginning in January 2019, pregnant women should receive a single dose of the Tdap vaccine between weeks 27 and 36 of pregnancy as a temporary protective measure for infants until they receive the first dose of pertussis vaccine according to the national vaccination program [17]. This recommendation was based on evidence of maternal antibody transfer across the placenta, which is maximal in the 34th week of gestation [18]. In a randomized, double-blind, placebo-controlled clinical trial, vaccination of pregnant women with Tdap in the third trimester produced higher levels of antibodies against pertussis in neonates and 2-month-old infants compared with infants whose mothers received a placebo, suggesting potential passive protection of infants during the period of high risk of pertussis morbidity and mortality [19]. In addition, some studies reported that the sum of more than 1 antibody against pertussis at levels greater than detectable contribute to an increased protective effect [20, 21]. Preliminary data from the United Kingdom and Australia indicate that infants born to mothers who received Tdap vaccination during pregnancy have a reduced risk of pertussis at an early age [22]. It is interesting to note that a recent study found that although Tdap vaccination of pregnant women is highly effective in protecting infants under 3 months of age against severe pertussis (odds ratio [OR] = 0.06, P = .004, 95% CI = .01–.41; vaccine effectiveness [VE] = 94%, 95% CI = 59%–99%), but it has lower effectiveness at preventing mild disease (OR = 0.31, P = .026, 95% CI = .11–.87; VE = 69%, 95% CI = 13%–89%) [23]. This suggests that acellular pertussis vaccination might not be enough to prevent transmission of the disease and to disrupt pertussis circulation, as suggested by another study using nonhuman models [24]. Overall, although studies have demonstrated the safety of the Tdap vaccine [25] and its protective effect against pertussis in infants, the durability of passively acquired antibody is unclear, and precise serological correlation of protection against the disease are not yet known [26]. Differences in maternal vaccine type and suggested optimal timing of Tdap vaccine administration (between 16 and 39 weeks of gestation) could affect the estimated measure of association [25]. Furthermore, the available studies were mainly conducted in developed countries with differing epidemiological contexts and sociodemographic characteristics and in which acellular pertussis vaccines (aP) were implemented as part of the vaccination schemes, whereas whole-cell vaccines (wP) are used in Peru. The response of Tdap-induced antibodies in women immunized with wP in infancy may change from those immunized with aP and affect the effectiveness of the Tdap maternal vaccination in preventing infant pertussis [26]. These differences in vaccine type also represent a selective pressure for important allelic variations in the circulating Bordetella pertussis strains [27], so it is possible that the maternal acellular Tdap vaccine may not provide adequate protection against pertussis in the high-risk Peruvian population. Finally, no studies have examined the association between Tdap vaccination in pregnant women and the reduction in the risk of pertussis in infants in Peru. Such studies would be particularly relevant from a public health standpoint to monitor and strengthen evidence of prevention of infection and complications associated with pertussis in susceptible populations, as well as for evidence-based decision making by public health professionals (to update prevention strategies) and women themselves regarding vaccination during pregnancy. In this context, the present study examined whether administration of the maternal Tdap vaccine between 27 and 36 weeks of gestation is associated with a reduction in the risk of pertussis in infants <2 months of age treated in healthcare facilities nationwide. METHODS Study Design and Population A case-control study was conducted in Peru to evaluate whether maternal Tdap vaccination is associated with a lower risk of pertussis in infants <2 months old. The study population consisted of infants under 2 months of age with suspected pertussis treated in healthcare facilities nationwide between 2019 and 2021, with clinical samples sent to the National Institute of Health of Peru (INS-Peru) for confirmatory diagnosis and epidemiological surveillance. According to the national vaccination program, the first dose of the diphtheria, tetanus, pertussis, hepatitis B, and Haemophilus influenzae type B pentavalent combination vaccine (DTP-HvB-Hib) should be administered at 2 months of age, which is why this age was selected as the limit, because any protective effect of the DTP-HvB-Hib vaccine must be excluded. Inclusion and Exclusion Criteria Cases were defined as infants aged <2 months, treated in healthcare facilities nationwide between 2019 and 2021, who were positive for B pertussis deoxyribonucleic acid (DNA) by multitarget quantitative PCR (qPCR) at INS-Peru. The multitarget qPCR method was developed by Tatti et al [28] and combines a Singleplex assay targeting pertussis toxin subunit S1 (ptxS1) and a multiplex assay based on the insertion sequences IS481, pIS1001, and hIS1001 for the detection and differentiation of B pertussis, Bordetella parapertussis, and Bordetella holmesii. This multitarget assay exhibits superior performance and specificity for the diagnosis of B pertussis infection compared with other PCR-based assays [28, 29]. The inclusion criteria for the controls were as follows: infants <2 months old, treated in healthcare facilities nationwide between 2019 and 2021, with negative multitarget qPCR results for DNA of B pertussis or other Bordetella species. Cases or controls were excluded from the analysis if the maternal Tdap vaccination status was unknown, they had been vaccinated with the first dose of DTP-HvB-Hib pentavalent combination vaccine, or if the B pertussis multitarget qPCR test was indeterminate. Cases and controls were included in the analysis according to the above-mentioned eligibility criteria and selected through the Clinical, Epidemiological, and Laboratory Investigation Form for Pertussis. This form accompanies the clinical samples of infants aged <2 months with suspected pertussis treated in healthcare facilities nationwide that are sent to the INS-Peru for diagnostic confirmation and epidemiological surveillance. According to Peru's Ministry of Health, definition of pertussis clinical case in infants is defined as any acute respiratory infection with cough and at least 1 of the following symptoms: paroxysmal cough, inspiratory whoop, posttussive vomiting, apnea, or cyanosis [2]. The confirmed pertussis case definition was based on the criteria reported by the WHO (in 2000) and the US Centers for Disease Control and Prevention (in 2010) [30]. The controls were randomly selected from the total population of eligible noncases who were multitarget qPCR-negative for B pertussis DNA. Exposure and Collection of Data The exposure variable was maternal Tdap vaccination during pregnancy and coded as a binary variable indicating whether the infant's mother received a single dose of Tdap vaccine during the period of 27 to 36 weeks of gestation. Information on cases and controls was obtained from the Clinical, Epidemiological, and Laboratory Research Form for Pertussis. The form is filled out with clinical and epidemiological information including Tdap vaccination status of patients' mothers that is collected from clinical history and immunization registries. Demographic, clinical, and epidemiological variables were analyzed in both groups, including sex, infant age, days since symptom onset, geographic region, clinical symptoms of pertussis (paroxysmal cough, stridor, and vomiting after cough), hospitalization, length of hospital stay, complications due to pneumonia, maternal age at delivery, and maternal Tdap vaccination during pregnancy. Although the first dose of the DTP-Hbv-Hib vaccine is given starting at 2 months of age, it was verified whether infants had received the vaccine. The number of days since symptom onset was calculated as the difference between the date the symptoms began and the date of sampling. The geographic region was coded as a binary variable, within Lima or outside Lima, because there are differences in the maternal Tdap vaccination rate in Lima compared with other regions of the country [31]. The age of the infant was calculated as the difference between the infant's date of birth and the date of sampling. The mother's age at delivery was defined as the difference between the mother's and the infant's dates of birth. Information on status of vaccination with the first dose of the pentavalent DTP-HvB-Hib and maternal Tdap vaccine during pregnancy was validated using the vaccination records of the Ministry of Health of Peru, within the framework of pertussis epidemiological surveillance. Infants' mothers were classified as unvaccinated if Tdap vaccination occurred in a previous or posterior pregnancy. Sample Size Calculation This study analyzed 50 cases and 150 controls (N = 200), assuming a proportion of exposed cases and controls of 17% and 71%, respectively [32]. This should give 100% power to detect an OR = 0.08 with a 95% CI of 79.34%–96.49%. The minimum sample size was calculated using Epidat v4.3. Statistical Analyses Statistical analyses, including regression models, were performed using the Stata/MP v.15.10 program, considering a 95% CI and two-tailed P < .05 as statistically significant. Bivariate analyses were performed between each covariable and outcome using the χ² test, Fisher's exact test, and Wilcoxon signed-rank test, as appropriate. The association between maternal Tdap vaccination during pregnancy and pertussis in infants younger than 2 months of age was evaluated using a multiple logistic regression model, estimating the OR, adjusted for confounding variables such as sex, infant age, geographic region, and mother's age at delivery. The Tdap VE was calculated as 1− OR × 100%. Sex and age of the infant and mother at delivery were included as a priori confounders [33]. The region variable was selected to adjust for temporal and spatial variations in vaccine coverage [32]. Ethics Statement This study was reviewed and approved by the Ethics in Research Committee of the INS-Peru (reference number OT-024-19) and Universidad Peruana Cayetano Heredia (reference number 103935). Patient Consent Statement Written informed consent for participation was not required due to the retrospective nature of this study, in accordance with national legislation and institutional requirements. RESULTS A total of 626 newborn infants with suspected pertussis were identified whose nasopharyngeal swab samples and clinical/epidemiological forms were referred to the INS-Peru for pertussis surveillance and diagnostic confirmation by multitarget qPCR between 2019 and 2021. From this group, 20 participants were excluded because they were older than 2 months of age, DTP-Hbv-Hib vaccination history of 2 cases and 13 controls was not available, and mothers of 9 cases and 101 controls did not register Tdap vaccine status. Thus, these participants were not considered in the study, according to the inclusion and exclusion criteria. A total of 50 cases and 150 controls who were randomly selected from B pertussis-negative noncases were included in the analysis (Figure 1). This study analyzed the information contained in the form and the results of sample processing. The characteristics of the selected participants and their mothers are described in Table 1. Among all study participants (N = 200), males were predominant (52.0%). The median age of participants was 42 days (range, 11–61 days), and they presented classic pertussis symptoms such as paroxysmal cough (89.5%), stridor (47.5%), and vomiting after coughing (45.5%). The median number of days with symptoms at which the sample was taken was 6 days (range, 0–33 days). The mothers of 27 infants had been vaccinated with Tdap during gestation (Table 1). Figure 1. Flowchart of study participant selection. *A total of 150 controls were randomly selected from eligible noncases who were multitarget quantitative polymerase chain reaction (qPCR) negative for Bordetella pertussis DNA (n = 431). DTP-HvB-Hib, diphtheria, tetanus, pertussis, hepatitis B, and Haemophilus influenzae type B pentavalent combination vaccine; Tdap, tetanus toxoid, reduced diphtheria toxoid, and acellular pertussis vaccine. Table 1. Characteristics of Newborn Infants in the Study Population, 2019–2021 (N = 200) Characteristic n (%) Infant Characteristicsa Sex  Male 104 (52.0)  Female 95 (47.5) Age, Weeks  1 to <2 3 (1.5)  2 to <3 12 (6.0)  3 to <4 15 (7.5)  4 to <5 26 (13.0)  5 to <6 43 (21.5)  6 to <7 33 (16.5)  7 to <8 39 (19.5)  8 to <9 29 (14.5) Age, median (IQR), days 42 (11–61) Days since symptom onset, median (IQR), days 6 (0–33) Geographic Region  Lima 109 (54.5)  Outside Lima 91 (45.5) Clinical Symptoms  Paroxysmal Cough …  No 18 (9.0)  Yes 179 (89.5)  Stridor …  No 69 (34.5)  Yes 95 (47.5)  Vomiting After Coughing …  No 85 (42.5)  Yes 91 (45.5) Hospitalization  No 12 (6.0)  Yes 181 (90.5) Duration of hospital stay, median (IQR), days 5 (0–32) Pneumonia  No 97 (48.5)  Yes 55 (27.5) Maternal Characteristics  Age at delivery, median (IQR), years 27 (14–48) Maternal Tdap Vaccination During Pregnancy  Unvaccinated 173 (86.5)  Vaccinated 27 (13.5) Abbreviations: IQR, interquartile range; Tdap, tetanus toxoid, reduced diphtheria toxoid, and acellular pertussis vaccine. a Some values may not add to 100% due to missing data. Table 2 shows the results of bivariate analysis of the characteristics of infants and mothers between cases and controls. No statistically significant differences were found in the median age of infants between cases and controls (41.5 days vs 42 days; P = .780). The proportion of cases and controls exhibited statistically significant differences with regard to maternal Tdap vaccination during pregnancy (P = .029). Four percent of cases (2 of 50) and 16.7% of controls (25 of 150) had mothers who were vaccinated with Tdap during pregnancy (P = .029). The median age of mothers in the case and control groups was 27 years (range, 15–46 years) and 26 years (range, 14–48 years), respectively (P = .670). Ninety-two percent of cases (46 of 50) were hospitalized, compared with 90% of controls (135 of 150), with no statistically significant difference (P = .975). A total of 58% of controls resided in the Lima region, compared with 44% of cases (P = .085) (Table 2). The distributions according to sex (P = .390), clinical features (paroxysmal cough [P = .570], stridor [P = .235], vomiting after coughing [P = .087]), median length of hospital stay (P = .152), and complications due to pneumonia (P = .598) showed no statistically significant differences between cases and controls (Table 2). Table 2. Bivariate Analysis Comparing Newborn Infant Characteristics Between Cases and Controls, 2019–2021 (N = 200) Characteristic Cases (n = 50) n (%) Controls (n = 150) n (%) P Value Infant Characteristicsa Sex … … .390b  Male 23 (46.0) 81 (54.0) …  Female 26 (52.0) 69 (46.0) … Age, Weeks … … .700c  1 to <2 1 (2.0) 2 (1.3) …  2 to <3 1 (2.0) 11 (7.3) …  3 to <4 5 (10.0) 10 (6.7) …  4 to <5 7 (14.0) 19 (12.7) …  5 to <6 11 (22.0) 32 (21.3) …  6 to <7 11 (22.0) 22 (14.7) …  7 to <8 8 (16.0) 31 (20.7) …  8 to <9 6 (12.0) 23 (15.3) … Age, median (IQR), days 41.5 (13–61) 42 (11–61) .780d Days since symptom onset, median (IQR), days 7 (1–33) 6 (0–33) .409d Geographic Region … … .085b  Lima 22 (44.0) 87 (58.0) …  Outside Lima 28 (56.0) 63 (42.0) … Clinical Symptoms … … … Paroxysmal Cough … … .570c  No 3 (6.0) 15 (10.0) …  Yes 46 (92.0) 133 (88.7) …  Stridor … … .235b  No 14 (28.0) 55 (36.7) …  Yes 27 (54.0) 68 (45.3) …  Vomiting After Coughing … … .087b  No 15 (30.0) 70 (46.7) …  Yes 26 (52.0) 65 (43.3) … Hospitalization … … .975b  No 3 (6.0) 9 (6.0) …  Yes 46 (92.0) 135 (90.0) … Duration of hospital stay, median (IQR), days 6 (1–32) 5 (0–31) .152d Pneumonia … … .598b  No 23 (46.0) 74 (49.3) …  Yes 11 (22.0) 44 (29.3) … Maternal Characteristics … … …  Age at delivery, median (IQR), years 27 (15–46) 26 (14–48) .670d Maternal Tdap Vaccination During Pregnancy … … .029c  Unvaccinated 48 (96.0) 125 (83.3) …  Vaccinated 2 (4.0) 25 (16.7) … Abbreviations: IQR, interquartile range; Tdap, tetanus toxoid, reduced diphtheria toxoid, and acellular pertussis vaccine. a Some values may not add to 100% due to missing data. b P value calculated by χ2 test. c P value calculated by Fisher exact test. d P value calculated by Wilcoxon rank-sum test. Table 3 shows the crude and adjusted models used to predict the ORs of associations between each independent factor and pertussis. The unadjusted OR for the association between the maternal Tdap vaccine administered between 27 and 36 weeks of pregnancy and the reduced risk of pertussis in infants younger than 2 months of age was 0.21 (95% CI, .05–.91) (Table 3). After adjusting for the confounding variables sex, infant age, geographic region, and mother's age at delivery, infants younger than 2 months of age born to mothers who received the Tdap vaccine between 27 and 36 weeks of pregnancy (OR = 0.19; 95% CI, .04–.86; P = .031) had an 81% lower risk of pertussis than infants younger than 2 months of age born to mothers who did not receive the Tdap vaccine, indicating 81% effectiveness of the Tdap vaccine in preventing pertussis in infants (95% CI, 14%–96%). Table 3. Regression Models Predicting Reduced Risk of Pertussis Infection Among Newborn Infants in Peru, 2019–2021 (N = 200) Characteristic Crude Modela Adjusted Modelb OR (95% CI) P Value P Value Sex  Male Ref. Ref.  Female 1.33 (.70–2.53) .391 1.51 (0.76–2.98) .236 Infant age 1.00 (.97–1.02) .922 1.01 (0.98–1.03) .637 Geographic Region  Lima Ref. Ref.  Outside Lima 1.76 (0.92–3.35) 0.087 1.85 (0.94–3.66) .076 Maternal age at delivery 1.01 (.97–1.06) .632 1.02 (0.97–1.08) .343 Maternal Tdap Vaccination During Pregnancy  Unvaccinated Ref. Ref.  Vaccinated .21 (.05–.91) .038 0.19 (0.04–0.86) .031 Abbreviations: CI, confidence interval; OR, odds ratio; Ref., reference; Tdap, tetanus toxoid, reduced diphtheria toxoid, and acellular pertussis vaccine. a Analyzed by logistic regression. b Analyzed by multiple logistic regression, adjusted per all listed variables. After adjusting for confounders, no associations were observed between pertussis and sex (OR = 1.51; 95% CI, .76–2.98; P = .236), infant age (OR = 1.01; 95% CI, .98–1.03; P = .637), geographic region (OR = 1.85; 95% CI, .94–3.66; P = .076), or maternal age at delivery (OR = 1.02; 95% CI, .97–1.08; P = .343) (Table 3). DISCUSSION The sustained increase in morbidity and mortality associated with whooping cough that has occurred since 2017, mainly in infants under 3 months of age, led the Ministry of Health of Peru to implement Tdap vaccination in pregnant mothers in 2019 [2, 7]. We present the first case-control study in Peru to evaluate whether administration of maternal Tdap vaccine between 27 and 36 weeks of gestation is associated with a reduced risk of pertussis in infants younger than 2 months of age. Our results demonstrate that infants <2 months old whose mothers were vaccinated with Tdap in the third trimester of pregnancy had an 81% lower risk of pertussis than infants born to mothers who did not receive the vaccine, suggesting that this strategy is effective (VE = 81%; 95% CI, 14%–96%) for protecting the infant during the first months of life and before primary immunization [3]. It is interesting to note that these results are consistent with those obtained in a case-control study from Argentina, another country using the whole-cell vaccine for primary pertussis immunization, where the Tdap vaccine was 80.7% (95% CI, 52.1%–92.2%) effective in preventing pertussis in infants <2 months of age [34]. In a study that analyzed 58 cases of infants younger than 8 weeks of age with pertussis in comparison with 55 controls, mothers of 10 infants with whooping cough (17%) had received the vaccine during pregnancy, compared with 39 mothers of 55 controls (71%), indicating a significant association between maternal prenatal vaccination and pertussis in infants (OR = 0.07; 95% CI, .03–.19), with a VE = 93% (95% CI, 81%–97%) after adjustment for sex, geographic area, and birth period [32]. Our results are also consistent with those reported by a study based on a laboratory-confirmed case coverage method as part of the pertussis surveillance system in England, where the authors found that the introduction of maternal Tdap vaccination provided protection against pertussis in infants younger than 2 months of age (VE = 90%; 95% CI, 82%–95%) and reduced disease-associated cases, hospitalizations, and deaths [35]. Despite differences between these studies in terms of design, population, locality, and type of vaccine used in primary immunization, the evaluation of the maternal Tdap vaccination strategy in Peru shows a comparable and high VE, probably due to the protection conferred to the infant through passive transfer of antibodies and the indirect effect of protecting the mother from whooping cough and potentially reducing the risk of transmission of the disease to the infant [32]. The main limitation of this study is the lack of control over the type of data recorded in the Clinical, Epidemiological, and Laboratory Investigation Form, which can lead to bias and misclassification, potentially resulting in over- or underestimation of the OR and VE [36]. However, validation of information, including maternal Tdap vaccination history, reduced the risk of recall bias and exposure misclassification. The analysis based on information from the form also made it impossible to control for other potential confounders, including mothers' breastfeeding status, gestational age, number of people in the household, daycare attendance, smoking, and maternal education [37]. For example, mothers who choose to be vaccinated may exhibit different characteristics—such as educational attainment—than those who do not [38], which could introduce a protective bias into the effect of maternal vaccination. According to a study by Quinn et al [39], the mother's educational attainment and exposure to school-age children at home are associated with pertussis in infants. Because the controls presented pertussis symptomatology (Table 2) and were selected based on negative PCR diagnostic results for DNA of B pertussis or other Bordetella species to avoid potential cross-protective effects, there was a possibility of misclassification of this group due to imperfections of the molecular test. However, no statistically significant differences were found between cases and controls in terms of the number of days with symptoms when sampled for PCR (P = .409) (Table 2), suggesting a low probability that the controls were false negatives. On the other hand, there may have been bias in the selection of cases, because patients who died of severe pertussis before being diagnosed would be excluded, thus biasing the protective effect of the Tdap vaccine toward cases with mild disease [36]. Nevertheless, in this study, more than 90% (46 of 50) of the cases were hospitalized, including nonsevere cases and those with complications of pneumonia (Table 2), suggesting that this effect was limited. Because this study uses test-negative controls, some mild cases of pertussis are likely being missed. This is explained because individuals who come to healthcare facilities usually are more likely to have severe symptoms that suggest Tdap vaccination is effective for protecting the Peruvian infants against severe pertussis. However, this medical help-seeking behavior is affected not only by severity of symptoms, but also by others factors including age, gender, socioeconomic status, access to healthcare, proximity to testing facilities, personality, and insurance coverage. Several studies have supported that theses bias are controlled by a test-negative design, and it can generate valid estimates based on factors that lead individuals to come to healthcare facilities, which are the same on both those who test positive and those who test negatives [40]. Although severe pertussis has been associated with hospitalization, definition of pertussis severity is complex, and information such as duration of stay, level of hospitalization, use of oxygen, and intravenous therapy as well as clinical features including presence of complications is required. Furthermore, pertussis severity must be evaluated using these variables and based on a scoring system that has to be validated to be applicable to hospitalized children in specified country settings [41]. CONCLUSIONS Despite these limitations, this study analyzed valuable information at the national level and provides robust evidence demonstrating that maternal Tdap administration during pregnancy is a highly effective intervention for protecting infants against pertussis, particularly for the Peruvian population at higher risk. It is thus crucial to communicate this information to both pregnant women and health professionals to promote greater acceptance of the vaccine at the national level and in low- and middle-income countries that recommend the whole-cell vaccine in their primary immunization schedule. Acknowledgments We thank the Oficina General de Tecnologías de la Información of the Ministry of Health of Peru for validating the vaccination history of infants and mothers. We are very grateful to all health personnel of Peru, especially the members of the National Institute of Health, for their dedication to providing continued pertussis surveillance. Author contributions. EJ-L planned and designed the study. FV, MPS, and HH contributed to data collection. EJ-L performed the analyses, interpretated the data, and drafted the manuscript. MPS and MP contributed to writing the manuscript and provided ideas for the submission. All authors reviewed and approved the final manuscript. Financial support. This work was supported by the National Institute of Health of Peru . EJ-L is a completing an Epidemiological Research Doctorate at Universidad Peruana Cayetano Heredia under FONDECYT/CIENCIACTIVA Scholarship EF033-235-2015 and supported by Training Grant D43 TW007393, awarded by the Fogarty International Center of the US National Institutes of Health. ==== Refs References 1 Fu P , WangC, TianH, KangZ, ZengM. Bordetella pertussis infection in infants and young children in Shanghai, China, 2016–2017: clinical features, genotype variations of antigenic genes and macrolides resistance. Pediatr Infect Dis J 2019; 38 :370–376.30882726 2 Centro Nacional de Epidemiología . [Prevención y Control de Enfermedades/Ministerio de Salud de Perú. Alerta Epidemiológica No AE-010-2019]. Available at: https://www.dge.gob.pe/portal/docs/alertas/2019/AE010.pdf. Accessed 7 March 2023. 3 Kandeil W , van den EndeC, BungeEM, JenkinsVA, CeregidoMA, GuignardA. A systematic review of the burden of pertussis disease in infants and the effectiveness of maternal immunization against pertussis. Exp Rev Vaccines 2020; 19 :621–38. 4 Yeung KHT , DuclosP, NelsonEAS, HutubessyRCW. An update of the global burden of pertussis in children younger than 5 years: a modelling study. Lancet Infect Dis 2017; 17 :974–80.28623146 5 Chiappini E , StivalA, GalliL, de MartinoM. Pertussis re-emergence in the post-vaccination era. BMC Infect Dis 2013; 13 :151.23530907 6 Maertens K , CaboréRN, HuygenK, HensN, Van DammeP, LeuridanE. Pertussis vaccination during pregnancy in Belgium: results of a prospective controlled cohort study. Vaccine 2016; 34 :142–50.26592142 7 Castillo ME , BadaC, del AguilaO, et al Detection of Bordetella pertussis using a PCR test in infants younger than one year old hospitalized with whooping cough in five Peruvian hospitals. Int J Infect Dis 2015; 41 :36–41.26523641 8 Althouse BM , ScarpinoSV. Asymptomatic transmission and the resurgence of Bordetella pertussis. BMC Med 2015; 13 :146.26103968 9 Wiley KE , ZuoY, MacartneyKK, McIntyrePB. Sources of pertussis infection in young infants: a review of key evidence informing targeting of the cocoon strategy. Vaccine 2013; 31 :618–25.23200883 10 Forsyth K , PlotkinS, TanT, von KönigCHW. Strategies to decrease pertussis transmission to infants. Pediatrics 2015; 135 :e1475–82.25963002 11 Centers for Disease Control and Prevention (CDC) . Updated recommendations for use of tetanus toxoid, reduced diphtheria toxoid and acellular pertussis (Tdap) vaccine from the advisory committee on immunization practices, 2010. MMWR Morb Mortal Wkly Rep 2011; 60 :13–5.21228763 12 World Health Organization . Pertussis vaccines: WHO position paper—September 2015. Available at: https://www.who.int/publications-detail-redirect/WHO-WER9035. Accessed 12 January 2023. 13 Centers for Disease Control and Prevention (CDC) . Updated recommendations for use of tetanus toxoid, reduced diphtheria toxoid, and acellular pertussis vaccine (Tdap) in pregnant women—advisory committee on immunization practices (ACIP), 2012. MMWR Morb Mortal Wkly Rep 2013; 62 :131–5.23425962 14 Vizzotti C , NeyroS, KatzN, et al Maternal immunization in Argentina: a storyline from the prospective of a middle income country. Vaccine 2015; 33 :6413–9.26277071 15 Department of Health, United Kingdom . Vaccination against pertussis (whooping cough) for pregnant women. September 2021. Available at: https://www.gov.uk/government/publications/vaccination-against-pertussis-whooping-cough-for-pregnant-women. Accessed 11 July 2023. 16 Campbell H , GuptaS, DolanGP, et al Review of vaccination in pregnancy to prevent pertussis in early infancy. J Med Microbiol 2018; 67 :1426–56.30222536 17 Ministerio de Salud de Perú . [Norma Técnica Sanitaria N° 141-MINSA/2018/DGIESP: Norma Técnica de Salud que establece el Esquema Nacional de Vacunación]. Available at: https://www.gob.pe/institucion/minsa/normas-legales/178240-719-2018-minsa. Accessed 21 December 2022. 18 Healy CM . Vaccines in pregnant women and research initiatives. Clin Obstet Gynecol 2012; 55 :474–86.22510631 19 Munoz FM , BondNH, MaccatoM, et al Safety and immunogenicity of tetanus diphtheria and acellular pertussis (Tdap) immunization during pregnancy in mothers and infants: a randomized clinical trial. JAMA 2014; 311 :1760–9.24794369 20 Kohberger RC , JemioloD, NoriegaF. Prediction of pertussis vaccine efficacy using a correlates of protection model. Vaccine 2008; 26 :3516–21.18495303 21 Gall SA , MyersJ, PichicheroM. Maternal immunization with tetanus—diphtheria–pertussis vaccine: effect on maternal and neonatal serum antibody levels. Am J Obstet Gynecol 2011; 204 :334.e1–5. 22 Harriman K , WinterK. Pertussis vaccine uptake during pregnancy: we need to do better in the U.S. Prev Med 2014; 67 :320–1.25034730 23 Saul N , WangK, BagS, et al Effectiveness of maternal pertussis vaccination in preventing infection and disease in infants: the NSW public health network case-control study. Vaccine 2018; 36 :1887–92.29501321 24 Warfel JM , ZimmermanLI, MerkelTJ. Acellular pertussis vaccines protect against disease but fail to prevent infection and transmission in a nonhuman primate model. Proc Natl Acad Sci U S A 2014; 111 :787–92.24277828 25 Furuta M , SinJ, NgESW, WangK. Efficacy and safety of pertussis vaccination for pregnant women—a systematic review of randomised controlled trials and observational studies. BMC Preg Child 2017; 17 :390. 26 Healy CM , RenchMA, SwaimLS, et al Kinetics of maternal pertussis-specific antibodies in infants of mothers vaccinated with tetanus, diphtheria and acellular pertussis (Tdap) during pregnancy. Vaccine 2020; 38 :5955–61.32709433 27 Bart MJ , HarrisSR, AdvaniA, et al Global population structure and evolution of Bordetella pertussis and their relationship with vaccination. mBio 2014; 5 :e01074. 28 Tatti KM , SparksKN, BoneyKO, TondellaML. Novel multitarget real-time PCR assay for rapid detection of Bordetella species in clinical specimens. J Clin Microbiol 2011; 49 :4059–66.21940464 29 Roorda L , BuitenwerfJ, OssewaardeJM, van der ZeeA. A real-time PCR assay with improved specificity for detection and discrimination of all clinically relevant Bordetella species by the presence and distribution of three insertion sequence elements. BMC Res Notes 2011; 4 :11.21255383 30 von König C-HW . Pertussis diagnostics: overview and impact of immunization. Exp Rev Vaccines 2014; 13 :1167–74. 31 Dirección General de Intervenciones Estratégicas en Salud Pública (DGIESP)/Ministerio de Salud de Perú . [Situación actual de las coberturas en vacunación]. Available at: https://www.mesadeconcertacion.org.pe/storage/documentos/2022-03-11/coberturas-en-vacunacion-segunda-reunion-de-la-secretaria-ejecutiva-de-la-comision-intergubernamental-en-salud-cigs-2022.pdf. Accessed 9 March 2023. 32 Dabrera G , AmirthalingamG, AndrewsN, et al A case-control study to estimate the effectiveness of maternal pertussis vaccination in protecting newborn infants in England and Wales, 2012–2013. Clin Infect Dis 2015; 60 :333–7.25332078 33 Petousis-Harris H , JiangY, YuL, et al A retrospective cohort study of safety outcomes in New Zealand infants exposed to tdap vaccine in utero. Vaccines (Basel) 2019; 7 :147.31614582 34 Romanin V , AcostaAM, Del Valle JuarezM, et al Maternal vaccination in Argentina: tetanus, diphtheria, and acellular pertussis vaccine effectiveness during pregnancy in preventing pertussis in infants <2 months of age. Clin Infect Dis 2020; 70 :380–7.30877308 35 Amirthalingam G , AndrewsN, CampbellH, et al Effectiveness of maternal pertussis vaccination in England: an observational study. Lancet 2014; 384 :1521–8.25037990 36 Kopec JA , EsdaileJM. Bias in case-control studies. A review. J Epidemiol Commun Health 1990; 44 :179–86. 37 Liu BC , HeW-Q, NewallAT, et al Effectiveness of acellular pertussis vaccine in older adults: nested matched case-control study. Clin Infect Dis 2020; 71 :340–50.31504309 38 Donaldson B , JainP, HolderBS, LindseyB, ReganL, KampmannB. What determines uptake of pertussis vaccine in pregnancy? A cross sectional survey in an ethnically diverse population of pregnant women in London. Vaccine 2015; 33 :5822–8.26409139 39 Quinn HE , SnellingTL, HabigA, ChiuC, SpokesPJ, McIntyrePB. Parental Tdap boosters and infant pertussis: a case-control study. Pediatrics 2014; 134 :713–20.25225136 40 Vandenbroucke JP , BrickleyEB, Vandenbroucke-GraulsCMJE, PearceN. A test-negative design with additional population controls can be used to rapidly study causes of the SARS-CoV-2 epidemic. Epidemiology 2020; 31 :836.32841988 41 Marshall H , ClarkeM, RasiahK, et al Predictors of disease severity in children hospitalized for pertussis during an epidemic. Pediatr Infect Dis J 2015; 34 :339–45.25260040
PMC010xxxxxx/PMC10352648.txt
==== Front Open Forum Infect Dis Open Forum Infect Dis ofid Open Forum Infectious Diseases 2328-8957 Oxford University Press US 10.1093/ofid/ofad360 ofad360 Major Article AcademicSubjects/MED00290 Food Insecurity Is Associated With Low Tenofovir Diphosphate in Dried Blood Spots in South African Persons With HIV Hirsh Molly L Augusta University/University of Georgia Medical Partnership, Medical College of Georgia, Athens, Georgia, USA Edwards Jonathan A Division of Infectious Diseases, Department of Medicine, Emory University School of Medicine, Atlanta, Georgia, USA School of Health and Social Care, University of Lincoln, Lincoln, United Kingdom Robichaux Chad Division of Infectious Diseases, Department of Medicine, Emory University School of Medicine, Atlanta, Georgia, USA Brijkumar Jaysingh Department of Medicine, University of KwaZulu-Natal, Durban, South Africa Moosa Mahomed-Yunus S Department of Medicine, University of KwaZulu-Natal, Durban, South Africa Ofotokun Igho Division of Infectious Diseases, Department of Medicine, Emory University School of Medicine, Atlanta, Georgia, USA Johnson Brent A Department of Biostatistics and Computational Biology, University of Rochester, Rochester, New York, USA Pillay Selvan Department of Medicine, University of KwaZulu-Natal, Durban, South Africa Pillay Melendhran Department of Medicine, National Health Laboratory Service, Durban, South Africa Moodley Pravi Department of Medicine, National Health Laboratory Service, Durban, South Africa Sun Yan V Department of Epidemiology, Rollins School of Public Health, Emory University, Atlanta, Georgia, USA Liu Chang Department of Epidemiology, Rollins School of Public Health, Emory University, Atlanta, Georgia, USA Dudgeon Mathew R Department of Epidemiology, Rollins School of Public Health, Emory University, Atlanta, Georgia, USA Ordoñez Claudia Department of Epidemiology, Rollins School of Public Health, Emory University, Atlanta, Georgia, USA Kuritzkes Daniel R Division of Infectious Diseases, Brigham and Women's Hospital, Harvard Medical School, Boston, Massachusetts, USA Sunpath Henry Department of Medicine, University of KwaZulu-Natal, Durban, South Africa Morrow Mary Department of Biostatistics and Bioinformatics, Colorado School of Public Health, Aurora, Colorado, USA Anderson Peter L Department of Pharmaceutical Sciences, Skaggs School of Pharmacy and Pharmaceutical Sciences, University of Colorado Anschutz Medical Campus, Aurora, Colorado, USA Ellison Lucas Department of Pharmaceutical Sciences, Skaggs School of Pharmacy and Pharmaceutical Sciences, University of Colorado Anschutz Medical Campus, Aurora, Colorado, USA Bushman Lane R Department of Pharmaceutical Sciences, Skaggs School of Pharmacy and Pharmaceutical Sciences, University of Colorado Anschutz Medical Campus, Aurora, Colorado, USA Marconi Vincent C Division of Infectious Diseases, Department of Medicine, Emory University School of Medicine, Atlanta, Georgia, USA Department of Medicine, National Health Laboratory Service, Durban, South Africa Emory Vaccine Center, Emory University, Atlanta, Georgia, USA https://orcid.org/0000-0003-1242-1745 Castillo-Mancilla Jose R Division of Infectious Diseases, Department of Medicine, University of Colorado Anschutz Medical Campus, Aurora, Colorado, USA Correspondence: Jose R. Castillo-Mancilla, MD, Division of Infectious Diseases, Department of Medicine, University of Colorado Anschutz Medical Campus, 12700 E 19th Ave, Aurora, CO 80045 ([email protected]). Potential conflicts of interest. V. C. M. has received investigator-initiated research grants paid to institution and consultation fees from Eli Lilly, Bayer, Gilead Sciences, and ViiV, outside the current work. All other authors report no potential conflicts. 7 2023 13 7 2023 13 7 2023 10 7 ofad36024 4 2023 05 7 2023 18 7 2023 © The Author(s) 2023. Published by Oxford University Press on behalf of Infectious Diseases Society of America. 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs licence (https://creativecommons.org/licenses/by-nc-nd/4.0/), which permits non-commercial reproduction and distribution of the work, in any medium, provided the original work is not altered or transformed in any way, and that the work is properly cited. For commercial re-use, please contact [email protected] Abstract Background Food insecurity has been linked to suboptimal antiretroviral therapy (ART) adherence in persons with HIV (PWH). This association has not been evaluated using tenofovir diphosphate (TFV-DP) in dried blood spots (DBSs), a biomarker of cumulative ART adherence and exposure. Methods Within a prospective South African cohort of treatment-naive PWH initiating ART, a subset of participants with measured TFV-DP in DBS values was assessed for food insecurity status. Bivariate and multivariate median-based regression analysis compared the association between food insecurity and TFV-DP concentrations in DBSs adjusting for age, sex, ethnicity, medication possession ratio (MPR), and estimated glomerular filtration rate. Results Drug concentrations were available for 285 study participants. Overall, 62 (22%) PWH reported worrying about food insecurity and 44 (15%) reported not having enough food to eat in the last month. The crude median concentrations of TFV-DP in DBSs differed significantly between those who expressed food insecurity worry versus those who did not (599 [interquartile range {IQR}, 417–783] fmol/punch vs 716 [IQR, 453–957] fmol/punch; P = .032). In adjusted median-based regression, those with food insecurity worry had concentrations of TFV-DP that were 155 (95% confidence interval, −275 to −35; P = .012) fmol/punch lower than those who did not report food insecurity worry. Age and MPR remained significantly associated with TFV-DP. Conclusions In this study, food insecurity worry is associated with lower TFV-DP concentrations in South African PWH. This highlights the role of food insecurity as a social determinant of HIV outcomes including ART failure and resistance. We established that worrying about food insecurity was significantly associated with lower tenofovir diphosphate concentrations in dried blood spots for persons with HIV in South Africa. adherence antiretroviral therapy dried blood spots food insecurity tenofovir diphosphate Emory University Center for AIDS Research P30AI050409 National Institute of Allergy and Infectious Diseases 10.13039/100000060 National Institutes of Health 10.13039/100000002 [ADReSS] R01 AI098558 ==== Body pmcSouth Africa has the highest number of people with human immunodeficiency virus (PWH) and fourth-highest prevalence of human immunodeficiency virus (HIV) globally, with an estimated 7.8 million cases in 2020 [1]. Several drivers of this high prevalence have been proposed, including social determinants of health (SDoH) such as low income and food insecurity [2]. Of note, South Africa has experienced an increase in food insecurity between 2020 and 2022; food insecurity presently affects >20% of its population due to natural disasters, political unrest, supply chain challenges, and the coronavirus disease 2019 pandemic [3, 4]. Defined as the lack of regular access to safe and nutritious foods for normal growth and development as well as an active and healthy life, food insecurity is a recognized determinant of HIV outcomes via several proposed pathways [5]. These include poor medication absorption from malnutrition, as well as poor medication adherence due to fears or actual experience of increased ART side effects in the absence of adequate nutritional intake; cost-reducing behaviors such as selling medication or skipping clinic visits for other competing priorities; and associated mental health comorbidities such as depression and anxiety [6–8]. Access to antiretroviral therapy (ART) has increased the proportion of PWH who are engaged in care and have achieved viral suppression [9]. However, durable ART adherence is still required for sustained viral suppression. Poor adherence to ART medications can lead to adverse HIV outcomes such as viremia, virologic failure, antiretroviral resistance, and HIV transmission [10]. While ART alone is known to reduce disease burden, SDoH (including food insecurity) are also influential in determining a patient's access to, and empowerment in, utilizing these medications effectively over time, during a lifelong course of disease management [2, 8, 11–17]. Research on ART adherence has usually relied on measures that are subject to reporting bias and cannot confirm medication ingestion [18]. Antiretroviral concentrations in urine, hair, and dried blood spots (DBSs) can address some of these limitations, as they can serve as both measures of adherence and pharmacokinetics (ie, absorption, distribution, metabolism, and excretion) in a single test. Tenofovir diphosphate (TFV-DP) in DBSs—a quantitative measure of cumulative ART adherence to tenofovir-based regimens—is easily collected and is a highly informative ART adherence measure that is predictive of HIV outcomes including viral suppression, future viremia, drug resistance, virologic failure, and endothelial/immune activation [10, 18–20]. Prior research in the United States (US) has demonstrated an inverse association between food insecurity and pharmacologically measured ART adherence quantified using antiretroviral drug concentrations in hair [21]. Another study has linked income inequality with ART concentrations measured in DBSs [13]. To date, no studies have evaluated the relationship between food insecurity and a DBS-based adherence biomarker in a resource-limited setting. To address this gap, we evaluated whether food insecurity was associated with drug concentrations of TFV-DP in DBSs in South African PWH. METHODS Patient Consent Statement All patients provided informed written consent prior to their participation in the study. The study was approved by the University of KwaZulu-Natal (KZN) Biomedical Research Ethics Committee and the Emory University and MassGeneral Brigham institutional review boards. Overview A secondary data analysis of the KZN HIV/AIDS Drug Resistance Surveillance Study (ADReSS) in South Africa was conducted [22, 23]. The primary exposures of interest were self-reported measures of food insecurity in the last month. The primary outcome of interest in this subanalysis was cumulative ART adherence, measured using TFV-DP in DBSs. Covariates of interest included sociodemographic and clinical factors such as age, sex, race, ethnicity, education, income, wealth index, employment, distance to HIV clinic, psychological distress, medication possession ratio (MPR), and estimated glomerular filtration rate (eGFR). Univariate, bivariate, and multivariate relationships identified significant associations between exposure variables and TFV-DP concentration using an α = .05 level with no adjustments for multiple comparisons. Significantly associated covariates with both the primary exposure and outcome variables were used in a stratified analysis to explore potential effect modification or confounders. Study Design The parent study ADReSS was an observational nested case-control study conducted in KZN, South Africa, to assess ART response and resistance among those initiating first-line ART (efavirenz/tenofovir disoproxil fumarate/emtricitabine). Study enrollment of 1000 ART-naive patients took place at 2 clinics, R. K. Khan Hospital and Bethesda District Hospital Clinic, between July 2014 and September 2016, as previously described [10, 23]. Inclusion criteria required participants to be ≥18 years of age when first-line ART was initiated consisting of tenofovir disoproxil fumarate (TDF) and emtricitabine, plus efavirenz (a single-tablet regimen) or nevirapine, which were provided by South African Health authorities. PWH were excluded if they had any prior ART except for single-dose nevirapine for HIV vertical transmission prevention. Participants were then followed prospectively with HIV RNA measurement (viral load [VL]) obtained at 6 months, 12 months, and annually thereafter to identify cases and controls. Cases were participants who had met criteria for virologic failure, defined as HIV-1 VL >1000 copies/mL after receiving ≥5 months of ART, and controls were participants without virologic failure (participants who had VL ≤1000 copies/mL) matched 2:1 by study site, sex, closest age, and duration of ART (Figure 1). At the follow-up study visit where case-control status was determined, DBS samples were obtained to quantify TFV-DP concentrations. At the same visit, participants completed surveys with questions regarding the primary exposure of interest—food insecurity—as well as secondary covariates of interest such as socioeconomic status, stigma, ART challenges, and mental health assessments. Participants remained blinded to their virologic failure status until completion of the survey; subsequently, participants followed clinic protocols for adherence counseling and a confirmation of VL. This temporary participant blinding was intended to avoid any influence on survey responses, without delay in appropriate care. Additional data such as VL, CD4+ T-cell count, and serum creatinine were collected throughout the study duration at 6-month intervals. Figure 1. At the initial study visit, baseline characteristics were noted such as sex, age, race, and study site. Testing was obtained for participants' baseline CD4+ T-cell count and serum creatinine. Subsequent blood samples for plasma human immunodeficiency virus type 1 RNA viral load (VL) and CD4+ T-cell count were obtained at 6 months, 12 months, and annually thereafter. Virologic failure status was defined in participants whose VL values exceeded 1000 copies/mL in at least 2 readings, 2 months apart, after a minimum of 5 months of antiretroviral therapy (ART). Participants with virologic failure were matched 1:2 to participants without virologic failure (<1000 copies/mL) by age, sex, site, and ART duration. Participants completed a follow-up study visit survey, where questions regarding socioeconomic status, stigma, ART challenges, and mental health assessments were included. At this visit, dried blood spots were also obtained to quantify tenofovir diphosphate concentrations. Subsequently, participants followed clinic protocols for adherence counseling and a confirmation of VL. *To assess for virologic failure. Abbreviations: ART, antiretroviral therapy; eGFR, estimated glomerular filtration rate; TFV-DP, tenofovir diphosphate; VL, viral load. Data Collection Whole blood for DBSs was collected in ethylenediaminetetraacetic acid tubes and 25 µL was spotted 5 times onto 903 protein saver cards. The concentrations of TFV-DP in DBSs were quantified from a 3-mm punch using a validated liquid chromatography–tandem mass spectrometry assay with a limit of quantification of 25 fmol/sample [24]. Blood for HIV VL, CD4+ T-cell count, and blood chemistries were obtained per local standard of care [22]. Participants responded to a questionnaire containing adapted validated psychosocial scales (including the World Health Organization [WHO] wealth index [25, 26], Household Food Insecurity Access Scale measures [modified] [27], and psychological distress screening using the Kessler-10 scale [28, 29]). Food security was assessed with the following 3 questions: “In the past 4 weeks, did you worry that you and your family would not have enough food?”; “In the past 4 weeks, was the amount of food you and your family has to eat enough, sometimes not enough, or often not enough to eat?”; and “In the past 4 weeks, how many times did you and anyone in your family go an entire day and night without food because there was not enough food?” For each food security measure, respectively, any participant answer indicating having ever worried, ever not had enough to eat, or ever gone 24 hours without food in the last 4 weeks was considered as food insecure. Participants with a Kessler-10 score >20 were considered psychologically distressed. The MPR, which was estimated using the number of days' supply for all pharmacy fills of ART medication in a particular time period divided by the number of days in that period multiplied by 100, was corrected with an adjustment for values >100% caused by early refills and was used to estimate participants' medication adherence [30]. Collected response variables were composited into variables of interest for analysis in R software (version 2022.02.2) (R Foundation for Statistical Computing, Vienna, Austria, 2017). Data Analysis The primary study outcome of this secondary analysis was TFV-DP in DBS samples. Concentrations of TFV-DP in DBSs that were below the limit of quantification were imputed to 12.5 fmol/punch, as previously described [31]. Serum creatinine measured at baseline (closest value to enrollment within a period 1 year prior to enrollment and up to 90 days after) were used to calculate eGFR using the Chronic Kidney Disease–Epidemiology Collaboration (CKD-EPI) 2021 formula [32]. Preliminary analyses examined the relationship between concentrations of TFV-DP in DBSs and food insecurity, as well as other variables of interest using Wilcoxon rank-sum tests for categorical variables and Spearman correlation for continuous variables (P < .05). The relationships between predictors, food insecurity measures, and participant characteristics were analyzed using either χ2 or Fisher exact test for categorical characteristics, as appropriate, and Wilcoxon rank-sum tests for continuous variables. Crude differences in TFV-DP concentrations compared by covariates of interest were then calculated using median-based regression. Standard error was calculated using the bootstrapping method. An adjusted multivariate model using median-based regression was created to include covariates that were significantly associated with TFV-DP concentration in bivariate analyses. Additional variables decided based on prior literature (sex and ethnicity) were also added into the model [10]. The adjusted median-based regression model measuring the association between food insecurity and TFV-DP concentration thus included age, sex, ethnicity, and MPR. Potential confounders and effect modifiers were identified with stratified analysis. While eGFR had fewer participant data available, a sensitivity analyses was conducted to account for it in the adjusted regression model as it remained of interest based on the literature [32]. An additional version of the adjusted model was created without MPR to determine its impact on the relationship between food insecurity and TFV-DP concentration. An additional analysis was conducted with an aggregated food insecurity measure, which combined 2 of the food security measures that asked about food insecurity worry in the last month and lack of food in the last month. Responses indicating whether participants had ever worried and/or had a lack of food were scaled categorically with the following options: food insecure, partially food insecure, or food secure. This aggregated food insecurity measure was then compared to TFV-DP in crude and adjusted models using the same model covariates: age, sex, ethnicity, and MPR. RESULTS The parent cohort enrolled a total of 1000 treatment-naive PWH who initiated first-line ART [22]. This subanalysis included data from 285 PWH in whom TFV-DP concentrations in DBS were available, 92 (32%) of whom had virologic failure. Demographic and baseline clinical characteristics of the study participants are shown in Table 1. Overall, 166 (58%) of the participants were women, 275 (96%) were Black including 2 participants who identified as mixed race, 220 (77%) identified as ethnically Zulu, 209 (73%) reported having an income, and 168 (59%) were employed. The median age at first visit was 35 (interquartile range [IQR], 26–38) years, the number of years of education attained was 11 (IQR, 9–12), and the wealth index resolved on 2 principal component analyses was 0.2 and −0.7. Clinically, 22 (8%) participants screened positive for psychological distress, 30 (10%) had an eGFR <90 mL/minute/1.73 m², and 106 (37%) had an MPR <85%. Examining secondary outcome measures, the median baseline CD4+ T-cell count was 286 (IQR, 171–386) cells/µL, which increased to 371 (IQR, 223–520) cells/µL at the time concurrent with DBS testing. Table 1. Participant Characteristics in the HIV AIDS Drug Resistance Surveillance Study (ADReSS) Cohort Nested Case-Control Subset (N = 285) Characteristic No. (%) or Median (IQR) Food insecurity worry in last month  Never experienced 207 (72)  Experienced 62 (22)  Missinga 16 (6) Enough to eat in last month  Always 235 (83)  Not always 44 (15)  Missing 6 (2) 24 h without food in last month  Never experienced 266 (93)  Experienced 14 (5)  Missing 5 (2) Age at first visit, y 31 (26–38) Sex  Male 119 (42)  Female 166 (58) Race  Black or mixed race 275 (96)  Indian 10 (4) Ethnicity  Zulu 220 (77)  Xhosa 33 (12)  Other 22 (8)  Missing 10 (4) Education, y 11 (9–12) Income  No 71 (25)  Yes 209 (73)  Missing 5 (2) Wealth indicesa,b  Wealth index 1 0.2 (−1.1 to 1.5)  Wealth index 2 −0.7 (−0.9 to 0.5) Employment  Unemployed 99 (35)  Employed 168 (59)  Other 13 (4)  Missing 5 (2) Time to clinic arrival, min  <30 79 (28)  30–60 178 (62)  >60 22 (8)  Missing 6 (2) Psychological distressc  No 254 (89)  Yes 22 (8)  Missing 9 (3) eGFRd, mL/min/1.73 m²  >90 210 (74)  60–90 26 (9)  <60 4 (1)  Missinga 45 (16) MPRe, %  ≥100 96 (34)  85–99 82 (29)  <85 106 (37)  Missing 1 (0.3) TFV-DP concentrationf, fmol/punch 696 (443–938) CD4+ T-cell count at follow-up, cells/μL 371 (223–520) HIV RNA load at follow-up, copies/mL 40 (24–150) Virologic failure  No 193 (68)  Yes 92 (32) Abbreviations: eGFR, estimated glomerular filtration rate; HIV, human immunodeficiency virus; IQR, interquartile range; MPR, medication possession ratio; TFV-DP, tenofovir diphosphate. a More than 5% of responses missing. b Wealth indices missing 17 responses each (6%). c Psychological distress was considered in participants scoring ≥20 on the Kessler-10 scale. d eGFR missing in 166 participants (58%). e MPR percentage missing 1 response (<1%). f TFV-DP concentrations below the lower limit of quantification were imputed to 12.5 fmol/punch. The median TFV-DP concentration in DBS for the whole cohort was 696 (IQR, 443–938) fmol/punch (Table 1). In the month prior to being surveyed, 62 (22%) participants reported experiencing worry about food insecurity and 44 (15%) participants reported that they had insufficient food to eat. An additional 14 (5%) participants reported going 24 hours without eating due to insufficient food; given this limited sample size of affirmative responses, this variable was not included in further analyses. Bivariate analyses indicated a significant association between food insecurity worry and TFV-DP concentrations in DBS (Table 2). Study participants who reported food insecurity worry had a lower median TFV-DP concentration compared to those without food insecurity worry (599 [IQR, 417–783] fmol/punch vs 716 [IQR, 453–957] fmol/punch; P = .031). PWH who reported insufficient food in the last month also had lower TFV-DP concentrations compared to those who always had enough to eat, although the association was not statistically significant (613 [IQR, 480–838] fmol/punch vs 699 [IQR, 419–943] fmol/punch; P = .361). Significant associations were additionally found in bivariate analyses comparing median TFV-DP concentrations to both age and MPR. The median TFV-DP concentration was positively correlated with age (ρ = 0.26; P < .001). Median TFV-DP concentrations were lower among participants with <85% MPR compared to those with at least 100% MPR (602 vs 789 fmol/punch; P < .001). No significant associations were found between TFV-DP and sex, race, ethnicity, education, wealth indices, employment, time to clinical arrival, psychological distress, eGFR, or having enough to eat in the last month (P > .09 for all variables; Table 2). Table 2. Participant Characteristics in Relation to Tenofovir Diphosphate Concentrations in Dried Blood Spots Characteristic TFV-DP Concentration in DBS (fmol/punch) Median (IQR) or Correlation Coefficient P Value Food insecurity worry in last month .031*  Never experienced 716 (453–957)  Experienced 599 (417–783) Enough to eat in last month .361  Always 699 (419–943)  Not always 613 (480–838) Age at first visit, y 0.26 <.001* Sex .298  Male 717 (488–1006)  Female 668 (435–894) Race .988  Black 692 (448–937)  Indian 774 (137–1203) Ethnicity .371  Zulu 667 (418–935)  Xhosa 741 (552–913)  Other 746 (462–1001) Education, y −0.04 .486 Income .947  No 670 (447–938)  Yes 696 (443–938) Wealth index 1a 0.1 .093 Wealth index 2a 0.03 .599 Employment .713  Unemployed 630 (405–928)  Employed 707 (501–940)  Other 658 (496–895) Time to clinic arrival, min .223  <30 664 (415–826)  30–60 713 (462–972)  >60 595 (396–912) Psychological distress .308  No 700 (445–944)  Yes 642 (407–782) eGFRa, mL/min/1.73 m2 .204  >90 702 (478–941)  60–90 739 (455–1022)  <60 265 (13–610) MPR, % <.001*  ≥100 789 (586–1036)  85–99 677 (476–930)  <85 602 (302–856) Data are presented as median (IQR) or Spearman correlation coefficient. P values compare mean values of TFV-DP between groups, calculated using Wilcoxon rank-sum tests for categorical variables and Spearman rank correlation for continuous variables. Abbreviations: DBS, dried blood spots; eGFR, estimated glomerular filtration rate; IQR, interquartile range; MPR, medication possession ratio; TFV-DP, tenofovir diphosphate. a More than 5% missing data. * P < .05. To identify potential confounders or effect modifiers on the association between food insecurity and concentrations of TFV-DP in DBS, further analyses were conducted. In bivariate analyses examining differences in covariates among measures of food insecurity, significant associations were found between food insecurity worry and age, ethnicity, wealth, and employment status (Supplementary Table 1). There were no significant associations found between food insecurity worry and sex, race, education, income, time to clinic arrival, psychological distress, eGFR, or MPR (Supplementary Table 1). Based on this, age was considered a potential effect modifier or confounder. Using median-based regression modeling, Table 3 shows the crude and adjusted relationships between food insecurity, covariates of interest, and median TFV-DP concentrations in DBS. Similar to the Wilcoxon rank-sum results in Table 2, food insecurity worry in the last month was crudely associated with lower concentrations of TFV-DP, although without statistical significance (−112 [95% confidence interval {CI}, −232 to 8] fmol/punch; P = .069). In an adjusted model that included age, sex, ethnicity, and MPR, food insecurity worry became significantly associated with median TFV-DP concentration, showing that PWH who reported food insecurity worry had a median concentration of TFV-DP in DBS that was 155 fmol/punch lower compared to those who reported no worry (95% CI, −275 to −35 fmol/punch; P = .012). Age remained significantly associated with TFV-DP in both crude and adjusted models. Both models showed that median TFV-DP in DBS was 16 fmol/punch higher for every increasing year of age (95% CI, 9–22; P < .001, adjusted model). MPR <85% also was significantly associated with a decrease in TFV-DP concentration compared to an MPR of 100% in both crude and adjusted models (crude model: −183 [95% CI, −294 to −72] fmol/punch, P = .002; adjusted model: −151 [95% CI, −271 to −31] fmol/punch, P = .015). Table 3. Median-Based Regression Models of Food Insecurity and Tenofovir Diphosphate Concentrations in Dried Blood Spots Characteristic Median Difference in TFV-DP in DBS (fmol/punch) Single Variable Median Regressiona (95% CI) (n = 285) Adjusted Median Regression Modelb (95% CI) (n = 258) Sensitivity Analyses Adjusted Median Regression Model With eGFRc (95% CI) (n = 215) Adjusted Median Regression Model With MPR Removedd (95% CI) (n = 259) Food insecurity worry in last monthe  Never Ref Ref Ref Ref  Ever −112 (−232 to 8) −155 (−275 to −35)* −195 (−320 to −71)* −139 (−232 to −46)* Enough to eat in last month  Always Ref … … …  Not always −77 (−192 to 38) … … … Age at first visit (per y) 16 (10–23)* 16 (9–22)* 16 (8–23)* 16 (10–22)* Sex  Male Ref Ref Ref Ref  Female −47 (−135 to 41) −40 (−157 to 78) −35 (−155 to 86) −23 (−137 to 91) Race  Black Ref … … …  Indian 62 (−497 to 621) … … … Ethnicity  Zulu Ref Ref Ref Ref  Xhosa 71 (−55 to 197) 42 (−123 to 207) 55 (−113 to 222) 96 (−68 to 261)  Other 39 (−158 to 234) 71 (−162 to 303) 87 (−130 to 304) 93 (−116 to 302) Education (y) 3 (−24 to 31) … … … Income  No Ref … … …  Yes 26 (−107 to 160) … … … Wealth index 1e 11 (−11 to 33) … … … Wealth index 2e 15 (−14 to 45) … … … Employment  Employed Ref … … …  Unemployed −76 (−194 to 41) … … …  Other −48 (−280 to 184) … … … Time to clinic arrival (min)  <30 Ref … … …  30–60 45 (−64 to 154) … … …  >60 −60 (−266 to 146) … … … Psychological distress  No Ref … … …  Yes −29 (−196 to 138) … … … eGFR (mL/min/1.73 m2)e  >90 Ref … Ref …  60–90 8 (−195 to 211) … −40 (−281 to 201) …  <60 −182 (−834 to 470) … −350 (−1261 to 561) … MPR (%)  ≥100 Ref Ref Ref …  85–99 −91 (−257 to 75) −47 (−173 to 81) −74 (−211 to 64) …  <85 −183 (−294 to −71)* −151 (−271 to −31)* −137 (−263 to −11)* … Abbreviations: CI, confidence interval; DBSs, dried blood spots; eGFR, estimated glomerular filtration rate; MPR, medication possession ratio; Ref, reference group; TFV-DP, tenofovir diphosphate. a Single variable median regression: crude relationship between individual variables and TFV-DP concentration. b Adjusted median-based regression: relationship between food insecurity and TFV-DP concentration adjusted for age, sex, ethnicity, and MPR. c Adjusted median-based regression with eGFR: relationship between food insecurity and TFV-DP concentration adjusted for age, sex, ethnicity, MPR, and eGFR. d Adjusted median-based regression with MPR removed: relationship between food insecurity and TFV-DP concentration adjusted for age, sex, and ethnicity. e More than 5% missing data. * P < .05. In a sensitivity analysis that added eGFR to the adjusted model, though with fewer total participant observations (n = 215), food insecurity, age, and MPR <85% remained significant predictors of decreased TFV-DP concentration (Table 3). No significant crude associations were found between median TFV-DP concentration and having enough to eat in the last month, sex, race, ethnicity, education, income, wealth indices, employment, time to clinic arrival, psychological distress, or eGFR. In another adjusted model excluding MPR, food insecurity worry still accounted for 139 fmol/punch lower concentrations of TFV-DP (95% CI, −232 to −46 fmol/punch; P = .004) when compared to no food insecurity worry (Table 3). In a bivariate analysis comparing an aggregated food insecurity measure to median TFV-DP concentration, a stepwise reduction in median TFV-DP concentration was noted among increasing severity of food insecurity (food secure, 716 [IQR, 450–961] fmol/punch vs partially food insecure, 611 [IQR, 423–754] fmol/punch vs food insecure, 595 [IQR, 441–787] fmol/punch; P = .079). In an adjusted analysis, food-insecure participants had a median 129 fmol/punch lower concentration compared to those who were food secure (95% CI, −238 to −18 fmol/punch; P = .023). DISCUSSION In this study, we identified an inverse association between TFV-DP in DBS, a measure for cumulative ART adherence and exposure, and the experience of food insecurity in PWH in South Africa who were taking first-line TDF-based therapy. This is consistent with a previous US-based study where food insecurity was associated with lower cumulative ART drug concentrations quantified in hair [21]. Furthermore, the study found that the relationship between food insecurity and ART concentrations persisted both with and without the inclusion of MPR in the model, suggesting that low medication adherence may only partially explain our observations and that an objective measure of adherence and exposure may more readily identify these differences [21]. For example, previous studies have proposed that food insecurity may impact pharmacokinetic drug absorption if participants are unable to take medications with a meal, which can be particularly impactful for tenofovir absorption [21, 33–36]. Collectively, our findings support the premise that SDoH (in this case, food insecurity) could adversely impact drug adherence and bioavailability and on that addressing these factors using novel adherence measures could improve ART adherence and clinical outcomes. Compared to the worry of not having enough food, the relationship between TFV-DP and actually having enough food in the last month was less robust, although demonstrating similar directionality. In a sensitivity analysis where food insecurity measures were aggregated, this directionality was maintained: An increased number of reported food insecurity measures corresponded with a decreased TFV-DP concentration. However, with food insecurity worry being the strongest singular predictor of the food insecurity measures, this suggests that the component of worrying about food security may play a role in ART adherence. This worry may be a contextual indication of longer-term wealth instability versus the more focused measure of whether food insecurity was directly experienced in the last 4-week period. It is also plausible that worry may act as a surrogate for mental health comorbidities such as stress, anxiety, or depression, which have been implicated in bidirectional relationships with both food insecurity and ART adherence [2, 6, 7, 29, 37]. In this study, though only a small proportion of participants reported psychological distress, those participants tended to have lower concentrations of TFV-DP. Furthermore, it is possible that psychological distress among participants may have been undercounted due to varied sensitivity of the Kessler-10 scale among the Black South African demographic [29]. Regarding clinical translation, several inferences can be derived from our findings. In our cohort, food-insecure participants had a median TFV-DP concentration that was 155 fmol/punch lower than those who had no food insecurity worry (median concentration of TFV-DP, ∼700 fmol/punch). This reduction is consistent with approximately missing 1 TDF dose per week [38]. Moreover, extrapolating to previously published literature, PWH with similar TFV-DP concentrations to those who experienced food insecurity worry in our cohort were at greater risk for various adverse HIV outcomes such as loss of viral suppression, future viremia, or virologic failure with HIV drug resistance [10, 19, 39]. In a US study, PWH with TFV-DP between 700 and 1250 fmol/punch had 33 times greater odds of viral suppression than those with TFV-DP <350 fmol/punch. In comparison, PWH with TFV-DP between 351 and 699 fmol/punch, analogous to the concentration observed in our food insecurity worry group, had only 9 times greater odds of suppression than the study reference group—almost a 4-fold reduction [31]. Similarly, in a South African cohort of postpartum women with HIV, study participants with TFV-DP concentrations in DBS in the 350–699 fmol/punch range had higher odds of future viremia (∼3.4) compared to a reference group of women with TFV-DP >1850 fmol/punch, whereas those with TFV-DP concentrations >700 fmol/punch did not show significant odds of future viremia [39]. Beyond the individual impact of food insecurity on ART adherence, our findings could have concrete public health implications, as focusing on access to food in PWH would not only improve their overall health by allowing proper nutrition, but would also impact their adherence and could lead to a reduction in HIV transmission. Our study did not identify the previously observed associations between eGFR, sex, race, or ethnicity and TFV-DP concentrations in DBS [39, 40]. This may be due to a limited sample size and decreased heterogeneity among groups of study participants, reducing the power of this analysis. In this study, >95% of participants identified as Black. Additionally, ethnicities were local to the South African population, and did not include Latinx or Hispanic participants as traditionally studied in a US context. Previous studies describing race, ethnicity, and TFV-DP in DBS have, to our knowledge, only been described among participants in the US, where additional sociostructural predictors may have influenced the relationship [21]. However, this does not impact our conclusions. Future studies where the association of this (and other) SDoH in more diverse populations are needed. Among the strengths of our study are the large clinical cohort from where this substudy is derived, the wide range of SDoH collected, and the use of a novel and validated adherence biomarker that has been utilized effectively in South African cohorts [10, 33]. Limitations include the lack of availability of other known predictors of TFV-DP drug concentrations such as hematocrit, body mass index, and the temporal relationship between meals and ART dosing [40, 41]. Additional predictors of ART adherence such as substance abuse, depression and anxiety, traditional or herbal medicine use, HIV stigma and discrimination, and healthcare worker dissatisfaction were also either unavailable or underreported, potentially due to social desirability bias among these stigmatized participants [2, 7, 11]. The use of MPR as a proxy for ART adherence also has its limitations, due to imperfect sensitivity and assumption that all pills distributed by the pharmacy were consumed appropriately and consistently [42]. Larger structural issues not addressed in our study, such as medication access, pharmacy stockouts, drug prices, and food costs [2, 11, 43], should also be examined in the future. In conclusion, food insecurity worry is significantly associated with low cumulative ART adherence and bioavailability. While an overall high proportion of adherence to ART regimens is reported in sub-Saharan Africa, SDoH remain influential in an individual patient's likelihood to remain adherent and effectively absorb their treatment [44]. The WHO has recognized that holistic treatment of PWH must address SDoH, including a focus on nutritionally based interventions where applicable [45]. The observed rise in the prevalence of food insecurity should be considered within HIV intervention planning to better meet the needs of PWH at risk for suboptimal ART concentrations. Future research delineating the mechanisms through which food insecurity impacts ART adherence and exposure could help prevent adverse HIV treatment outcomes [10]. Supplementary Material ofad360_Supplementary_Data Click here for additional data file. Supplementary Data Supplementary materials are available at Open Forum Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author. Notes Financial support. This work was supported by the Emory University Center for AIDS Research for salary support (grant number P30AI050409 to V. C. M.) and the National Institute of Allergy and Infectious Diseases, National Institutes of Health for salary support and funding the parent study (the HIV AIDS Drug Resistance Surveillance Study [ADReSS]) (grant number R01 AI098558 to V. C. M. and D. R. K). ==== Refs References 1 Joint United Nations Programme on HIV/AIDS . South Africa country fact sheet. Available at: https://www.unaids.org/en/regionscountries/countries/southafrica. Accessed 5 February 2022. 2 Heestermans T , BrowneJL, AitkenSC, VervoortSC, Klipstein-GrobuschK. Determinants of adherence to antiretroviral therapy among HIV-positive adults in sub-Saharan Africa: a systematic review. BMJ Glob Health 2016; 1 :e000125. 3 Statistics South Africa . Towards measuring the extent of food security in South Africa: An examination of hunger and food adequacy. 2019:34 . Available at: https://www.statssa.gov.za/publications/03-00-14/03-00-142017.pdf. Accessed 5 February 2022. 4 van der Berg S , PatelL, BridgmanG. Food insecurity in South Africa: evidence from NIDS-CRAM wave 5. Dev South Afr 2022; 39 :722–37. 5 United Nations Food and Agriculture Organization . Hunger and food insecurity. Available at: https://www.fao.org/hunger/en/. Accessed 8 August 2022. 6 Weiser SD , YoungSL, CohenCR, et al Conceptual framework for understanding the bidirectional links between food insecurity and HIV/AIDS. Am J Clin Nutr 2011; 94 :1729s–39s.22089434 7 Weiser SD , SheiraLA, PalarK, et al Mechanisms from food insecurity to worse HIV treatment outcomes in US women living with HIV. AIDS Patient Care STDS 2020; 34 :425–35.32941054 8 Young S , WheelerAC, McCoySI, WeiserSD. A review of the role of food insecurity in adherence to care and treatment among adult and pediatric populations living with HIV and AIDS. AIDS Behav 2014; 18 (Suppl 5 ):S505–15.23842717 9 Center for Disease Control and Prevention. HIV/AIDS care and treatment programs in South Africa. 2011. Available at: https://www.cdc.gov/globalhealth/countries/southafrica/what/antiretroviral-treatment.htm. Accessed 26 January 2023. 10 Castillo-Mancilla JR , EdwardsJA, BrijkumarJ, et al Tenofovir diphosphate levels in dried blood spots are associated with virologic failure and resistance to first-line therapy in South Africa: a case-control cohort study. J Int AIDS Soc 2021; 24 :e25849. 11 Naidoo P , PremduttR. Do challenges still exist amongst HIV/AIDS patients in managing their condition? A cross-sectional study of 297 participants in the Ethekwini Metro of KwaZulu-Natal, South Africa. Afr J AIDS Res 2019; 18 :198–204.31575338 12 Peltzer K , PengpidS. Socioeconomic factors in adherence to HIV therapy in low- and middle-income countries. J Health Popul Nutr 2013; 31 :150–70.23930333 13 Vernon F , MorrowM, MaWhinneyS, et al Income inequality is associated with low cumulative antiretroviral adherence in persons with human immunodeficiency virus. Open Forum Infect Dis 2020; 7 :ofaa391. 14 Weiser SD , TullerDM, FrongilloEA, SenkunguJ, MukiibiN, BangsbergDR. Food insecurity as a barrier to sustained antiretroviral therapy adherence in Uganda. PLoS One 2010; 5 :e10340. 15 McCoy SI , NjauPF, FaheyC, et al Cash vs. food assistance to improve adherence to antiretroviral therapy among HIV-infected adults in Tanzania. AIDS 2017; 31 :815–25.28107221 16 Tirivayi N , KoetheJR, GrootW. Clinic-based food assistance is associated with increased medication adherence among HIV-infected adults on long-term antiretroviral therapy in Zambia. J AIDS Clin Res 2012; 3 :171.23227443 17 Franke MF , MurrayMB, MuñozM, et al Food insufficiency is a risk factor for suboptimal antiretroviral therapy adherence among HIV-infected adults in urban Peru. AIDS Behav 2011; 15 :1483–9.20714923 18 Nachega JB , MarconiVC, van ZylGU, et al HIV treatment adherence, drug resistance, virologic failure: evolving concepts. Infect Disord Drug Targets 2011; 11 :167–74.21406048 19 Jennings L , RobbinsRN, NguyenN, et al Tenofovir diphosphate in dried blood spots predicts future viremia in persons with HIV taking antiretroviral therapy in South Africa. AIDS 2022; 36 :933–940.35131960 20 Castillo-Mancilla J , MaWhinneyS, CoyleR, et al Cumulative ART exposure is associated with endothelial and immune activation in HIV [poster 657]. In: Conference on Retroviruses and Opportunistic Infections, Seattle, WA, 4–7 March 2019. 21 Leddy AM , SheiraLA, TamrazB, et al Food insecurity is associated with lower levels of antiretroviral drug concentrations in hair among a cohort of women living with human immunodeficiency virus in the United States. Clin Infect Dis 2020; 71 :1517–23.31608363 22 Brijkumar J , JohnsonBA, ZhaoY, et al A packaged intervention to improve viral load monitoring within a deeply rural health district of South Africa. BMC Infect Dis 2020; 20 :836.33176715 23 Brijkumar J , EdwardsJA, JohnsonBA, et al Comparing effectiveness of first-line antiretroviral therapy between peri-urban and rural clinics in KwaZulu-Natal, South Africa. HIV Med 2022; 23 :727–37.35023287 24 Zheng JH , RowerC, McAllisterK, et al Application of an intracellular assay for determination of tenofovir-diphosphate and emtricitabine-triphosphate from erythrocytes using dried blood spots. J Pharm Biomed Anal 2016; 122 :16–20.26829517 25 Nachega JB , ChaissonRE, GoliathR, et al Randomized controlled trial of trained patient-nominated treatment supporters providing partial directly observed antiretroviral therapy. AIDS 2010; 24 :1273–80.20453627 26 Shin S , MuñozM, EspirituB, et al Psychosocial impact of poverty on antiretroviral nonadherence among HIV-TB coinfected patients in Lima, Peru. J Int Assoc Physicians AIDS Care (Chic) 2008; 7 :74–81.18319510 27 Coates J , SwindaleA, BilinskyP. Household Food Insecurity Access Scale (HFIAS) for measurement of household food access: indicator guide (v.3). 2007. Available at: https://www.fantaproject.org/sites/default/files/resources/HFIAS_ENG_v3_Aug07.pdf. Accessed 5 February 2022. 28 Kessler RC , AndrewsG, ColpeLJ, et al Short screening scales to monitor population prevalences and trends in non-specific psychological distress. Psychol Med 2002; 32 :959–76.12214795 29 John S , KearnsR, JohnsonBA, et al Religious beliefs and depression: psychosocial factors affecting HIV treatment outcomes in South Africa. New Voices Psychol 2016; 12 :2–20.29937923 30 Canfield SL , ZuckermanA, AnguianoRH, et al Navigating the Wild West of medication adherence reporting in specialty pharmacy. J Manag Care Spec Pharm 2019; 25 :1073–7.31556829 31 Castillo-Mancilla JR , MorrowM, CoyleRP, et al Tenofovir diphosphate in dried blood spots is strongly associated with viral suppression in individuals with human immunodeficiency virus infections. Clin Infect Dis 2019; 68 :1335–42.30137238 32 Inker LA , EneanyaND, CoreshJ, et al New creatinine- and cystatin C–based equations to estimate GFR without race. N Engl J Med 2021; 385 :1737–49.34554658 33 Anderson PL , LiuAY, Castillo-MancillaJR, et al Intracellular tenofovir-diphosphate and emtricitabine-triphosphate in dried blood spots following directly observed therapy. Antimicrob Agents Chemother 2018; 62: e01710-17. 34 Republic of South Africa, National Department of Health . 2019 ART clinical guidelines. 2019. https://health.gov.za/wp-content/uploads/2020/11/2019-art-guideline.pdf 35 FHI 360 . Tenofovir, lamivudine, and dolutegravir (TLD) transition. 2019. Available at: https://www.fhi360.org/sites/default/files/media/documents/linkages-tld-transition-information.pdf. Accessed 5 February 2022. 36 Kearney BP , FlahertyJF, ShahJ. Tenofovir disoproxil fumarate: clinical pharmacology and pharmacokinetics. Clin Pharmacokinet 2004; 43 :595–612.15217303 37 Whittle HJ , PalarK, SeligmanHK, NapolesT, FrongilloEA, WeiserSD. How food insecurity contributes to poor HIV health outcomes: qualitative evidence from the San Francisco Bay Area. Soc Sci Med 2016; 170 :228–36.27771206 38 Castillo-Mancilla JR , ZhengJH, RowerJE, et al Tenofovir, emtricitabine, and tenofovir diphosphate in dried blood spots for determining recent and cumulative drug exposure. AIDS Res Hum Retroviruses 2013; 29 :384–90.22935078 39 Odayar J , OrrellC, PhillipsTK, et al Use of tenofovir diphosphate levels to predict viremia during the postpartum period in women living with human immunodeficiency virus (HIV): a nested case-control study. Clin Infect Dis 2022; 75 :761–7.34979553 40 Coyle RP , MorrowM, ColemanSS, et al Factors associated with tenofovir diphosphate concentrations in dried blood spots in persons living with HIV. J Antimicrob Chemother 2020; 75 :1591–8.32211787 41 Baxi SM , GreenblattRM, BacchettiP, et al Common clinical conditions—age, low BMI, ritonavir use, mild renal impairment—affect tenofovir pharmacokinetics in a large cohort of HIV-infected women. AIDS 2014; 28 :59–66.24275255 42 Sangeda RZ , MoshaF, ProsperiM, et al Pharmacy refill adherence outperforms self-reported methods in predicting HIV therapy outcome in resource-limited settings. BMC Public Health 2014; 14 :1035.25280535 43 Drammeh W , HamidN, RohanaA. Determinants of household food insecurity and its association with child malnutrition in sub-Saharan Africa: a review of the literature. Curr Res Nutr Food Sci 2019; 7 . 44 Bijker R , JiamsakulA, KityoC, et al Adherence to antiretroviral therapy for HIV in sub-Saharan Africa and Asia: a comparative analysis of two regional cohorts. J Int AIDS Soc 2017; 20 :21218.28362063 45 World Health Organization . Nutrition and HIV/AIDS. Vol. EB116/12. 2005. Available at: https://apps.who.int/gb/ebwha/pdf_files/EB116/B116_12-en.pdf. Accessed 5 February 2022.
PMC010xxxxxx/PMC10352649.txt
==== Front Open Forum Infect Dis Open Forum Infect Dis ofid Open Forum Infectious Diseases 2328-8957 Oxford University Press US 10.1093/ofid/ofad296 ofad296 Major Article AcademicSubjects/MED00290 Staphylococcus aureus Bacteremia in Pediatric Patients: Uncovering a Rural Health Challenge Whittington Kyle J Department of Pediatrics, Washington University School of Medicine, St. Louis, Missouri, USA Malone Sara M Department of Pediatrics, Washington University School of Medicine, St. Louis, Missouri, USA Department of Surgery, Washington University School of Medicine, St. Louis, Missouri, USA Hogan Patrick G Department of Pediatrics, Washington University School of Medicine, St. Louis, Missouri, USA Ahmed Faria Department of Pediatrics, Washington University School of Medicine, St. Louis, Missouri, USA Flowers JessieAnn Department of Pediatrics, Washington University School of Medicine, St. Louis, Missouri, USA Milburn Grace Department of Pediatrics, Washington University School of Medicine, St. Louis, Missouri, USA Morelli John J Department of Pediatrics, Washington University School of Medicine, St. Louis, Missouri, USA Newland Jason G Department of Pediatrics, Washington University School of Medicine, St. Louis, Missouri, USA https://orcid.org/0000-0002-8602-5879 Fritz Stephanie A Department of Pediatrics, Washington University School of Medicine, St. Louis, Missouri, USA Correspondence: Stephanie A. Fritz, MD, MSCI, 660 S. Euclid Avenue, MSC 8116-43-10, St Louis, MO 63110-9872 ([email protected]); or Kyle Whittington, MD, 660 S. Euclid Avenue, MSC 8116-43-10, St Louis, MO 63110-9872 ([email protected]). Potential conflicts of interest. All authors: no reported conflicts. 7 2023 31 5 2023 31 5 2023 10 7 ofad29610 3 2023 25 5 2023 30 5 2023 18 7 2023 © The Author(s) 2023. Published by Oxford University Press on behalf of Infectious Diseases Society of America. 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs licence (https://creativecommons.org/licenses/by-nc-nd/4.0/), which permits non-commercial reproduction and distribution of the work, in any medium, provided the original work is not altered or transformed in any way, and that the work is properly cited. For commercial re-use, please contact [email protected] Abstract Background Staphylococcus aureus bacteremia poses significant risk for morbidity and mortality. This may be exacerbated in rural populations facing unique health challenges. Methods To investigate factors influencing S. aureus bacteremia outcomes, we conducted a retrospective cohort study of children admitted to St. Louis Children's Hospital (SLCH) from 2011 to 2019. Exposures included rurality (defined by the Rural-Urban Continuum Code), Area Deprivation Index, and outside hospital (OSH) admission before SLCH admission. The primary outcome was treatment failure, a composite of 90-day all-cause mortality and hospital readmission. Results Of 251 patients, 69 (27%) were from rural areas; 28 (11%) were initially admitted to an OSH. Treatment failure occurred in 39 (16%) patients. Patients from rural areas were more likely to be infected with methicillin-resistant S. aureus (45%) vs urban children (29%; P = .02). Children initially admitted to an OSH, vs those presenting directly to SLCH, were more likely to require intensive care unit–level (ICU) care (57% vs 29%; P = .002), have an endovascular source of infection (32% vs 12%; P = .004), have a longer duration of illness before hospital presentation (4.1 vs 3.0 days; P = .04), and have delayed initiation of targeted antibiotic therapy (3.9 vs 2.6 days; P = .01). Multivariable analysis revealed rural residence (adjusted odds ratio [aOR], 2.3; 95% CI, 1.1–5.0), comorbidities (aOR, 2.9; 95% CI, 1.3–6.2), and ICU admission (aOR, 3.9; 95% CI, 1.9–8.3) as predictors of treatment failure. Conclusions Children from rural areas face barriers to specialized health care. These challenges may contribute to severe illness and worse outcomes among children with S. aureus bacteremia. This study investigated the impact of rurality in children with Staphylococcus aureus bacteremia. Children from rural areas were more likely to be infected with methicillin-resistant S. aureus. Rural residence, comorbidities, and need for ICU admission were associated with treatment failure. Area Deprivation Index bacteremia rural health Rural-Urban Continuum Code Staphylococcus aureus Washington University St. Louis Children’s Hospital 10.13039/100009068 National Institutes of Health 10.13039/100000002 National Center for Advancing Translational Sciences 10.13039/100006108 UL1-TR002345 TL1-TR002344 Agency for Healthcare Research and Quality 10.13039/100000133 R01-HS024269 ==== Body pmcIndividuals living in underserved rural areas have been designated as a “health disparity population” by the National Institute on Minority Health and Health Disparities [1]. Adult residents of rural areas have worse health behaviors and health outcomes and access health services less frequently [2–4]. Mortality rates are higher among rural-dwelling individuals compared with their urban counterparts, even when controlling for poverty and age [2, 3]. A similar trend has been demonstrated in children; rural children have an annual mortality rate of 63 per 100 000 compared with 50 per 100 000 in urban children [2]. Alarmingly, the rural-urban disparity is growing, with the mortality gap increasing 5-fold from 1969 to 2009 [2]. The drivers of these poor outcomes are numerous and broad, including cultural, socioeconomic, and structural dynamics [5, 6]. Since 2010, more than 100 rural hospitals have closed [7]. Rural hospitals have fewer board-certified medical specialists and may lack on-site access to specialized diagnostic and therapeutic modalities, particularly for pediatric patients [8–10]. Despite these established health challenges, there is a paucity of research regarding the health outcomes of children in rural populations [11, 12]. Staphylococcus aureus bacteremia leads to significant morbidity and mortality in children. The rate of infection ranges from 1.5 to 3.5 per 1000 hospitalizations [13, 14]. Infection results in prolonged hospitalization, posing risk for complications; 10% of patients with S. aureus bacteremia develop septic emboli and metastatic infection [13, 14]. Additionally, these children have an increased risk of death, with mortality ranging from 2% to 15% [15–18]. Studies from Australia, the United Kingdom, and the United States demonstrate that infectious diseases (ID) consultation for S. aureus bacteremia improves management and outcomes; however, children residing in rural areas have limited access to pediatric ID subspecialists [4, 18–21]. As the impact of rural residence on pediatric S. aureus bacteremia outcomes is unknown, we investigated the influence of rural residence and socioeconomic deprivation on outcomes in children with S. aureus bacteremia. METHODS Setting and Patients This retrospective cohort study comprised 385 unique pediatric patients ranging from 0 to 24 years of age hospitalized with S. aureus bacteremia from January 2011 to December 2019 at St. Louis Children's Hospital (SLCH). SLCH is a 402-bed tertiary care hospital with ∼13 000 admissions annually, serving patients from all 50 states and >80 countries, with a primary service region covering 6 states. Patients were either admitted directly to SLCH or admitted initially to an outside hospital (OSH; for a minimum of 24 hours) before being transferred to SLCH. Children with community-associated or community-onset health care-associated infections whose blood cultures were obtained within 48 hours of hospital admission (to an OSH or SLCH) and who were positive for S. aureus were eligible. Patients with hospital-onset infections were excluded (definitions provided in the Supplementary Data). Patients with positive blood cultures for which antibiotics were not prescribed (per provider notes and laboratory comments) were also excluded. Patient Consent This study was approved by the Washington University institutional review board with waiver of informed consent. Data Collection Electronic medical record review was performed to collect demographic and clinical factors that may be associated with S. aureus bacteremia and outcomes (Supplementary Data). Administration of targeted antibiotic therapy was based on S. aureus susceptibility: cefazolin, nafcillin, or oxacillin for methicillin-susceptible S. aureus (MSSA) and vancomycin, ceftaroline, or daptomycin for methicillin-resistant S. aureus (MRSA). Sufficient antibiotic therapy was defined as treatment with an antibiotic with potential antistaphylococcal activity, though not targeted (eg, clindamycin for MRSA or ceftriaxone for MSSA). Study data were managed with REDCap [22, 23]. Exposures and Outcomes The primary objective of this study was to evaluate the impact of rural residence on clinical outcomes in pediatric patients with S. aureus bacteremia. The primary outcome was treatment failure, defined as a composite of all-cause 90-day mortality and 90-day hospital readmission in patients diagnosed with S. aureus bacteremia, congruent with other research in this area [24]. Secondary outcomes included length of hospitalization, length of bacteremia, and endovascular focus of infection. Exposures included rurality, primary admission to an OSH before transfer to SLCH, and Area Deprivation Index (ADI), as described below. Rural-Urban Continuum Code The first exposure evaluated in this study was the Rural-Urban Continuum Code (RUCC). The RUCC is a validated definition of rurality used by the US Department of Agriculture (USDA), which considers the population size as well as nearness to an urban center (Table 1). The RUCC ranges from 1 to 9, where 1 is the most urban and 9 is the most rural. The traditional rural/urban cutoff is between 3 and 4, with codes 1–3 considered urban and codes 4–9 considered rural [25]. Table 1. Rural-Urban Continuum Code (RUCC) Code Description 1 Counties in metropolitan areas of 1 million population or more 2 Counties in metropolitan areas of 250 000 to 1 million population 3 Counties in metropolitan areas of fewer than 250 000 populationa 4 Urban population of 20 000 or more, adjacent to a metropolitan area 5 Urban population of 20 000 or more, not adjacent to a metropolitan area 6 Urban population of 2500 to 19 999, adjacent to a metropolitan area 7 Urban population of 2500 to 19 999, not adjacent to a metropolitan area 8 Completely rural or less than 2500 urban population, adjacent to a metropolitan area 9 Completely rural or less than 2500 urban population, not adjacent to a metropolitan area United States Department of Agriculture Economic Research Service, Rural-Urban Continuum Codes 2019 [25]. a The common cutoff for rural and urban is between RUCC 3 and RUCC 4. Outside Hospital Admission The second exposure evaluated was primary admission to an OSH before transfer to SLCH. A child was determined to have their primary admission at an OSH if they were admitted to any outlying hospital for a minimum of 24 hours before transfer to SLCH. Area Deprivation Index The third exposure evaluated was the ADI [26]. The ADI uses 9-digit zip codes corresponding to census block groups, allowing for the characterization of deprivation in small tracts with similar demographic and geographic features. This methodology uses a composite of variables (eg, education, employment, income, housing quality) to give a rank-based score quantifying disadvantage. As SLCH serves a multistate region in the Midwestern United States, our study used the national percentile, which assigns 1 as the least disadvantaged and 100 as the most disadvantaged [26]. We categorized percentiles into quartiles: Quartile 1 reflects the least disadvantaged 25% of the nation (ie, ADI 1–25), while quartile 4 represents the most disadvantaged 25% of the nation (ie, ADI 76–100). Statistical Analysis Descriptive statistics characterized the study population. Means and standard deviations were computed for data that were normally distributed; medians and interquartile ranges (IQRs) were computed for non–normally distributed data. Categorical variables were compared using chi-square analysis; continuous variables were analyzed using either independent-sample t tests or Mann-Whitney U tests. P values of ≤.05 were considered significant. Backward stepwise logistic regression was performed to analyze factors associated with treatment failure. Variables were initially selected based on statistical significance in univariate analysis, demographics, and expert input (Supplementary Data). At each step, variables were retained based on P values <.05. All statistical analyses were performed in SPSS, version 27, for Windows (IBM SPSS, Chicago, IL, USA). RESULTS A total of 385 unique pediatric patients with S. aureus bacteremia were identified. We excluded 29 due to having a blood culture deemed to be a contaminant and 105 with hospital-onset infections. Therefore, 251 patients were included in this study. Within this cohort, 148 (59%) children had a community-associated infection, while 103 (41%) had a health care–associated, community-onset infection. Patients were predominantly White (67%) and male (62%) (Table 2). Patients were more frequently diagnosed with MSSA bacteremia (66%) than MRSA bacteremia (34%). The median distance from the patients’ homes to SLCH (IQR) was 31 (0–131) miles, and the median distance to an infectious diseases specialist (IQR) was 23 (0–104) miles. Twenty-eight children were diagnosed with bacteremia without an identified source (11%), while 223 (89%) had an additional source of infection including skin and soft tissue, pulmonary, musculoskeletal, endovascular, and hardware- and central line–associated infections. Of note, for children with central line–associated infections, the median time to removal of their infected central line (IQR) was 6 (2–36) days. Overall, 39 (16%) children experienced treatment failure within 90 days of initial hospital admission: 29 patients were readmitted within 90 days, and 12 died. Table 2. Patient Characteristics by Rural vs Urban Residence (per Rural-Urban Continuum Code Classification) Variable Total (n = 251), No. (%) Rural (n = 69), No. (%) Urban (n = 182), No. (%) P Age, mean (SD), y 7.9 (±5.5) 8.0 (±5.3) 7.8 (±5.6) .76 Sex  Female 94 (38) 22 (32) 72 (40) .26  Male 157 (62) 47 (68) 110 (60) Race  White 167 (67) 61 (88) 106 (58) <.001  African American and “other” racesa 84 (33) 8 (12) 74 (42) Staphylococcal susceptibility  MRSA 84 (34) 31 (45) 53 (29) .02  MSSA 167 (66) 38 (55) 129 (71) Initial admission to SLCH vs OSH  SLCH 223 (89) 54 (78) 169 (93) .001  OSH 28 (11) 15 (22) 13 (7) Distances  Distance from patient's home to SLCH, median (IQR), mi 31 (11–112) 119 (80–161) 19 (10–40) <.001  Distance from patient's home to nearest pediatric ID specialist, median (IQR), mi 23 (10–92) 105 (73–155) 15 (7–28) <.001 Infection entityb  Bacteremia without focus 28 (11) 6 (9) 22 (12) .45  Central line–associated infection 42 (17) 10 (15) 32 (18) .56  Musculoskeletal infection 122 (49) 38 (55) 84 (46) .21  Endovascular focus 36 (14) 16 (23) 20 (11) .01  Pulmonary infection 31 (12) 8 (12) 23 (13) .82  Skin and soft tissue infectionc 36 (14) 13 (19) 23 (13) .21  Other diagnosis (urinary tract infection/pyelonephritis, gastrointestinal tract, central nervous system) 29 (12) 6 (9) 23 (13) .38 Severity of illness  Duration of symptoms before initial hospitalization, mean (SD), d 3.14 (±2.8) 3.4 (±2.6) 3.1 (±2.8) .38  Complicated bacteremia (≥3 d)d 176 (70) 52 (75) 124 (68) .26  Required a surgical procedure 109 (43) 32 (46) 77 (43) .56  Time to surgical debridement (osteomyelitis only, n = 77), mean (SD), d 5.2 (±3.9) 4.8 (±3.9) 5.5 (±4) .52  ICU admission 80 (32) 22 (32) 58 (32) .99   Required ventilator support 46 (18) 15 (22) 31 (17) .39   Required inotropic support 33 (13) 10 (15) 23 (13) .70  Presence of instrumentation at infection site 33 (13) 5 (7) 28 (15) .09  Comorbiditiese 120 (48) 34 (49) 86 (47) .78   Structural heart condition 15 (6) 7 (10) 8 (4) .09   Cystic fibrosis 6 (2) 3 (4) 3 (2) .20   Malignancy 11 (4) 5 (7) 6 (3) .20 Outcomes  Duration of bacteremia, mean (SD), d 2.6 (±2.3) 2.7 (±2) 2.6 (±2.4) .67  Duration of hospitalization (including stay at OSH), mean (SD), d 13 (±26) 11 (±9) 14 (±30) .47  Musculoskeletal infection complications (n = 122)f 25 (20) 18 (21) 7 (18) .70  Treatment failureg 39 (16) 16 (23) 23 (13) .04   90-d mortality 12 (5) 5 (7) 7 (4) .26   90-d readmission 29 (12) 12 (17) 17 (9) .08 Means and SDs were computed for data that were normally distributed; medians and interquartile ranges were computed for non–normally distributed data. Abbreviations: ICU, intensive care unit; ID, infectious diseases; IQR, interquartile range; MRSA, methicillin-resistant S. aureus; MSSA, methicillin-susceptible S. aureus; OSH, outside hospital; SLCH, St. Louis Children's Hospital. a African American 74, Asian 5, biracial 2, Pacific Islander 1, Native American 1, other not specified 1. b Categories are not mutually exclusive (eg, a patient could have skin infection, pneumonia, and osteomyelitis); P value of chi-square analysis is 1 entity vs all other entities. c Including infections resulting from skin breakdown (eg, burns). d Complicated bacteremia was defined as the patient having 1 or more of these factors: duration of bacteremia >3 days, fever >72 hours, metastatic disease, or endocarditis. e Comorbidities include severe prematurity, congenital anomalies, malignancy, cystic fibrosis, structural heart conditions, etc. f Musculoskeletal infection complications included chronic osteomyelitis, pathologic fracture, chronic pain or limp, and leg length discrepancy. g Treatment failure: a composite of 90-day all-cause mortality and 90-day all-cause hospital readmission. Rural vs Urban Of 251 patients, 69 (27%) lived in an area designated as rural by the RUCC (Table 2). Patients from rural areas were predominantly White (88%) compared with urban children (58% White; P ≤ .001). Age and sex did not differ significantly between the groups. Significant comorbidities (eg, malignancy, congenital heart disease, cystic fibrosis, and bowel abnormalities) were present similarly between urban-dwelling (47%) and rural-dwelling (49%) children. Fifteen rural children (22%) initially presented to an OSH, while 54 (78%) presented directly to SLCH. In comparison, 13 (7%) urban children presented initially to an OSH, and 169 (93%) presented directly to SLCH (P = .001). Of the rural children who presented directly to SLCH, 27 of 54 (50%) had a significant comorbidity. Children from rural areas were more likely to present with MRSA infection (45%) compared with urban children (29%, P = .02). The median distance rural patients traveled to SLCH (IQR) was 119 (80–161) miles, compared with 19 (10–40) miles traveled by urban patients. The median distance to a pediatric infectious diseases physician (IQR) was 105 (73–155) miles for rural children and 15 (7–28) miles for urban children. Endovascular infection was diagnosed in 23% of rural children compared with 11% of urban children (P = .01). Treatment failure was significantly higher among rural children (23%) compared with urban children (13%; P = .04). Primary OSH Admission vs Entire Admission at SLCH Twenty-eight (11%) of 251 children were admitted to an OSH (for at least 24 hours) before being transferred to SLCH. These OSHs ranged from small community hospitals with limited pediatric resources (23 patients) to medium-sized academic institutions with access to pediatric infectious diseases specialists (5 patients). These children spent an average (SD) of 2.9 (1.7) days at the OSH before transfer to SLCH (Table 3). Of the 28 children initially admitted to an OSH, 15 (54%) were from rural areas. Children initially presenting to an OSH did not differ significantly in age, sex, or race compared with children admitted to SLCH for the entirety of their hospitalization. Patients who were initially admitted to an OSH had a significantly higher incidence of MRSA infection (57%) vs those initially admitted to SLCH (30%; P = .005). Patients transferred from an OSH lived significantly farther from SLCH than children presenting directly to SLCH (median [IQR], 134 [33–235] miles vs 27 [0–113] miles, respectively; P < .001) or to pediatric infectious diseases specialists (116 [2–230] miles vs 21 [0–82] miles, respectively; P < .001). Children who were transferred had a significantly higher incidence of endovascular infection (32%) compared with those presenting directly to SLCH (12%; P = .004) and were significantly more likely to require ICU-level care (57% vs 29%, respectively; P = .002), ventilator support (43% vs 15%; P ≤ .001), and inotropic support (29% vs 11%; P = .01). These patients also had a significantly longer mean duration of symptoms [SD] before initial hospitalization (4.1 [3.1] days) compared with children initially presenting to SLCH (3 [2.7] days; P = .04). The median time to infectious diseases consultation (IQR) was 4 (3–6) days for children initially admitted to an OSH and 2 (1–4) for children presenting directly to SLCH (P = .002). Optimal antibiotic management was also delayed for children first admitted to an OSH. The mean number of days to sufficient antibiotic therapy (SD) was 2.6 (2.9) days for children transferred from an OSH and 1.5 (2) days for children initially admitted to SLCH (P = .01). The mean number of days to targeted antibiotic therapy (SD) was 3.9 (2.8) days for children transferred from an OSH and 2.6 (2.4) days for children initially admitted to SLCH (P = .01). Children initially admitted to an OSH had a longer total duration of bacteremia (mean [SD], 3.6 [2.7] days) than children first admitted to SLCH (mean [SD], 2.5 [2.2] days; P = .02) and a longer total duration of hospitalization (median length of stay [IQR], 13 [7–21] days vs 7 [5–13] days, respectively; P = .03). Overall, treatment failure was similar between children transferred from an OSH and those initially admitted to SLCH (18% and 15%, respectively). Table 3. Characteristics of Patients Admitted Initially to an Outside Hospital vs St. Louis Children's Hospital Variable Total (n = 251), No. (%) OSH (n = 28), No. (%) SLCH (n = 223), No. (%) P Age, mean (SD), y 7.9 (±5.5) 9.2 (±5.9) 7.7 (±5.5) .17 Sex  Female 94 (38) 14 (50) 80 (26) .15  Male 157 (62) 14 (50) 143 (64) Race  White 167 (67) 22 (79) 143 (64) .13  African American and “other” racesa 84 (33) 6 (21) 80 (36) Staphylococcal susceptibility  MRSA 84 (34) 16 (57) 68 (30) .005  MSSA 167 (66) 12 (43) 155 (70) Rural vs urban residence  Rural 69 (28) 15 (54) 54 (24) .001  Urban 182 (72) 13 (46) 169 (76) Distances  Distance from patient's home to SLCH, median (IQR), mi 31 (11–112) 134 (33–235) 27 (0–113) <.001  Distance from patient's home to nearest pediatric ID specialist, median (IQR), mi 23 (10–92) 116 (2–230) 21 (0–82) <.001 Infection entityb  Bacteremia without focus 28 (11) 4 (14) 24 (11) .58  Central line–associated infection 42 (17) 2 (7) 40 (18) .15  Musculoskeletal infection 122 (49) 14 (50) 108 (48) .88  Endovascular focus 36 (14) 9 (32) 27 (12) .004  Pulmonary infection 31 (12) 6 (21) 25 (11) .12  Skin and soft tissue infectionc 36 (14) 4 (14) 32 (14) .99  Other diagnosis (urinary tract infection/pyelonephritis, gastrointestinal tract, central nervous system) 29 (12) 4 (14) 25 (11) .63 Severity of illness  Duration of symptoms before initial hospitalization, mean (SD), d 3.1 (±2.8) 4.1 (±3.1) 3.0 (±2.7) .04  Complicated bacteremia (≥3 d)d 176 (70) 22 (79) 154 (69) .30  Required surgical procedure 109 (43) 13 (46) 96 (43) .73  Time to surgical debridement (osteomyelitis only, n = 77), mean (SD), d 5.2 (±3.9) 7.6 (±5.5) 4.8 (±3.6) .09  ICU admission 80 (32) 16 (57) 64 (29) .002   Required ventilator support 46 (18) 12 (43) 34 (15) <.001   Required inotropic support 33 (13) 8 (29) 25 (11) .01  Presence of instrumentation at infection site 33 (13) 6 (21) 27 (12) .17  Comorbiditiese 120 (48) 12 (43) 108 (48) .58   Structural heart condition 15 (6) 2 (7) 13 (6) .78   Cystic fibrosis 6 (2) 1 (4) 5 (2) .70   Malignancy 11 (4) 0 (0) 11 (5) .20 Diagnostics  Echocardiogram (any) 46 (18) 7 (25) 39 (18) .33  Echocardiogram (following 3 positive cultures) 35 (34) 12 (39) 23 (32) .54  Time to echocardiogram, median (IQR), d 3 (2–6) 5.5 (2–14) 3 (2–4) .04  All appropriate labsf 112 (45) 15 (54) 97 (44) .31  Blood culture proof of cureg 234 (93) 27 (96) 207 (93) .47  Time to obtain radiology study (osteomyelitis only, n = 77), mean (SD), d 3 (±3.2) 3.77 (±3.4) 2.9 (±3.1) .34 Outcomes  ID consult obtained 182 (73) 24 (86) 158 (71) .10  Time to ID consultation, median (IQR), d 2 (1–4) 4 (3–6) 2 (1–4) .002  Empiric antibiotic therapy sufficient for any S. aureus type (includes OSH) 188 (75) 19 (68) 169 (76) .36  Days to empiric antibiotic therapy sufficient for any S. aureus type, mean (SD) (n = 188) 1.6 (±2) 2.3 (±2.2) 1.5 (±1.9) .06  Empiric antibiotic therapy sufficient for MSSA (includes OSH) 239 (95) 27 (96) 212 (95) .75  Days to empiric antibiotic therapy sufficient for MSSA, mean (SD) (n = 239) 1.5 (±1.9) 2.1 (±2.1) 1.5 (±1.9) .10  Days to initiating sufficient antibiotic therapy (including OSH), mean (SD)h 1.7 (±2.2) 2.6 (±2.9) 1.5 (±2) .01  Days treated with sufficient antibiotics for S. aureus bacteremia,h mean (SD) 38 (±52) 35 (±27) 38 (±54) .72  Received targeted antibiotic therapyi 221 (88) 26 (93) 195 (87) .41 Days to targeted antibiotic therapy, mean (SD) (n = 221) 2.8 (±2.4) 3.9 (±2.8) 2.6 (±2.4) .01  Duration of bacteremia, median (SD), d 2.6 (±2.3) 3.6 (±2.7) 2.5 (±2.2) .02  Duration of hospitalization, median (IQR), d 8 (5–14) 13 (7–21) 7 (5–13) .03  Treatment failurej 39 (16) 5 (18) 34 (15) .72   90-d mortality 12 (5) 1 (4) 11 (5) .75   90-d readmission 29 (12) 4 (14) 25 (11) .63 Means and SDs were computed for data that were normally distributed; medians and interquartile ranges were computed for non–normally distributed data. Abbreviations: CBC, complete blood count; CRP, c-reactive protein; ESR, erythrocyte sedimentation rate; ICU, intensive care unit; ID, infectious diseases; IQR, interquartile range; MRSA, methicillin-resistant S. aureus; MSSA, methicillin-susceptible S. aureus; OSH, outside hospital; SLCH, St. Louis Children’s Hospital. a African American 74, Asian 5, biracial 2, Pacific Islander 1, Native American 1, other not specified 1. b Categories are not mutually exclusive (eg, a patient could have skin infection, pneumonia, and osteomyelitis); P value of chi-square analysis is 1 entity vs all other entities. c Including infections resulting from skin breakdown (eg, burns). d Complicated bacteremia was defined as the patient having 1 or more of these factors: duration of bacteremia >3 days, fever >72 hours, metastatic disease, or endocarditis. e Comorbidities include severe prematurity, congenital anomalies, malignancy, cystic fibrosis, structural heart problems, etc. f Appropriate labs includes CBC, ESR, and CRP for all patients and vancomycin trough and creatinine for children who received 3 doses or 2 days of vancomycin. g Proof of cure is 2 consecutive negative cultures following a positive culture. h Sufficient therapy: antibiotic therapy with antisstaphylococcal activity, but not targeted therapy. i Targeted therapy: antibiotic therapy based on S. aureus susceptibility. For MSSA, targeted therapy includes cefazolin, nafcillin, and oxacillin. For MRSA, targeted therapy includes vancomycin, ceftaroline, and daptomycin. j Treatment failure: a composite of 90-day all-cause mortality and 90-day all-cause hospital readmission. Area Deprivation Index Of the 251 patients included in the study, 15 (6%) resided in ADI quartile 1 (ie, the least disadvantaged 25%), 51 (20%) in quartile 2, 70 (28%) in quartile 3, and 115 (46%) in quartile 4 (ie, the most disadvantaged) (Table 4). Of the 69 rural children per RUCC designation, 46 (66%) resided in an area classified as ADI quartile 4, while 38% of urban children resided in quartile 4 (P ≤ .001). None of the children residing in rural areas were categorized into quartile 1. Children living in more disadvantaged areas were more likely to be diagnosed with MRSA, while children living in more advantaged areas were more likely to be diagnosed with MSSA. The MRSA incidence increased across quartiles: 8% in quartile 1, 25% in quartile 2, 27% in quartile 3, and 40% in quartile 4 (P = .007). ADI was not associated with treatment failure. Table 4. Factors Associated With Area Deprivation Index Variable Total (n = 251), No. (%) ADI 1 (n = 15), No. (%) ADI 2 (n = 51), No. (%) ADI 3 (n = 70), No. (%) ADI 4 (n = 115), No. (%) P Race  White 167 (100) 13 (8) 34 (21) 52 (32) 65 (39) .02  African American and other racesa 84 (100) 2 (2) 16 (19) 18 (21) 50 (58) Staphylococcal susceptibility  MRSA 84 (100) 2 (2) 9 (11) 25 (30) 48 (57) .007  MSSA 167 (100) 13 (8) 42 (25) 45 (27) 67 (40) Patient resides >70 mi from a pediatric infectious disease physician  Yes 77 (100) 1 (1) 8 (10) 26 (34) 42 (55) .005  No 174 (100) 14 (8) 43 (25) 44 (25) 73 (42) Rural vs urban residence  Rural 69 (100) 0 (0) 4 (6) 19 (28) 46 (66) <.001  Urban 182 (100) 15 (8) 47 (26) 51 (28) 69 (38) Outcomes  Treatment failure 39 (100) 2 (5) 6 (15) 15 (39) 16 (41) .44   90-d mortality 12 (5) 0 (0) 3 (6) 5 (7) 4 (4) .54   90-d readmission 29 (12) 2 (13) 4 (8) 10 (14) 13 (11) .74 ADI percentiles were categorized into quartiles: Quartile 1 reflects the least disadvantaged 25% of the nation (ie, ADI 1–25), while quartile 4 represents the most disadvantaged 25% of the nation (ie, ADI 76–100). Abbreviations: ADI, Area Deprivation Index; MRSA, methicillin-resistant S. aureus; MSSA, methicillin-sensitive S. aureus. a Other races: African American 74, Asian 5, biracial 2, Pacific Islander 1, Native American 1, other not specified 1. Multivariable Logistic Regression Analysis In the multivariable model (Table 5), treatment failure was associated with rural residence (adjusted odds ratio [aOR], 2.3; 95% CI, 1.1–5.0), comorbidities (aOR, 2.9; 95% CI, 1.3–6.2), and need for ICU admission (aOR, 3.9; 95% CI, 1.9–8.3). Table 5. Factors Associated With Treatment Failure, Multivariable Logistic Regression Model Covariate aOR (95% CI) Residence  Rural 2.3 (1.1–5.0)  Urban Ref Age, ya 0.9 (0.9–1.0) Comorbiditiesb  Yes 2.9 (1.3–6.2)  No Ref Intensive care unit admission  Yes 3.9 (1.9–8.3)  No Ref Hosmer Lemeshow test = 0.92; Nagelkerke R2 = 0.194 (the model explains nearly 20% of the variation of the outcome). Other variables that were included but did not remain in the final model included race, sex, endovascular focus of infection, duration of symptoms before initial hospitalization, antibiotic susceptibility (MSSA vs MRSA), and initial admission to an OSH. Abbreviations: aOR, adjusted odds ratio; MRSA, methicillin-resistant S. aureus; MSSA, methicillin-sensitive S. aureus. a aOR represents each year of age increase. b Comorbidities include severe prematurity, congenital anomalies, malignancy, cystic fibrosis, or structural heart problem. DISCUSSION Pediatric S. aureus bacteremia is a serious infection that can lead to significant morbidity and mortality. Rural health systems face many challenges, particularly for patients needing a higher level of medical care. These challenges include a paucity of resources to provide specialized care, including lack of access to subspecialists and inability to perform specialized diagnostic and imaging studies or surgical procedures. This is especially true for pediatric patients and may lead to a delay in diagnosis and ultimately delayed treatment [4, 10, 27–31]. Thus, patients residing in rural areas who present to a local hospital may require transfer to larger tertiary care centers for the management of invasive infections. This study aimed to determine the impact of rural residence and admission to an OSH before transfer to SLCH on the clinical outcomes of children with S. aureus bacteremia. Importantly, we found that children residing in rural areas were more likely to experience treatment failure. Additionally, even when controlling for comorbidities, we found that primary admission to an OSH was correlated with a higher level of acuity upon admission to SLCH. Lastly, we determined that children with S. aureus bacteremia living in rural areas and areas with higher levels of deprivation had a higher incidence of MRSA infection. These findings underscore the urgent need to address the significant health disparities faced by children residing in rural areas to ultimately improve child health. Our models demonstrated that treatment failure was more than twice as likely among children residing in rural areas compared with urban-dwelling children. This finding was independent of initial presentation to an OSH, suggesting that the factors driving the association between rurality and treatment failure are multifactorial. These factors likely include underlying systemic issues of disadvantage, more so than where one presents for care. This aligns with prior studies demonstrating that children with cancer living in rural areas had worse survival outcomes [32]. Moreover, half of the children residing in rural areas presented directly to SLCH for care. Of these children presenting directly to SLCH, half had underlying comorbidities, including malignancy or cystic fibrosis, which placed them at increased risk for severe infection. Importantly, these children have an established relationship with subspecialists at the academic medical center. In prior studies, given concerns regarding lack of local resources, parents of children with complex health needs have described a desire to present directly to the specialty care center, even for emergent care, rather than seeking care from a local hospital [31]. Our analysis demonstrated that patients who were transferred to SLCH after initial admission to an OSH had a higher acuity of illness, frequently requiring ICU admission, as well as a higher incidence of endovascular infection. These children also had a more prolonged illness before presenting for medical care, presenting a full day after the onset of illness. This delayed presentation has been previously described among rural populations in Australia and the United States and is likely attributable to access to care, lack of specialists, and social and economic factors (eg, missed time from work, disruption to family routine, poverty, and transportation) [4, 31, 33, 34]. Furthermore, these children had a delay in the management of their S. aureus bacteremia, including a longer time from hospital admission to diagnostic studies and initiation of targeted antibiotics. As many of the OSH did not have an infectious diseases specialist, infectious diseases consultation, which has been demonstrated to improve the quality of care and patient outcomes, was also delayed [18–21]. Overall, delays in diagnostic evaluation and treatment likely contributed to prolonged bacteremia, a predisposing factor for the development of endovascular or metastatic infection, and hence a longer duration of hospitalization [13, 14, 20]. These delays highlight the need for partnerships between community hospitals and tertiary care centers to avoid these undesirable outcomes. Among our population of pediatric patients with S. aureus bacteremia, the incidence of MRSA, compared with MSSA, infection was higher among children from rural areas. Antibiotic overuse has been demonstrated to drive antimicrobial resistance. Two separate studies conducted using pediatric Medicaid claims data from Kentucky and West Virginia found that antibiotic prescription rates were highest among rural-dwelling children [35, 36]. Thus, antibiotic overuse could be a driver of the higher incidence of MRSA detected among children from rural areas in our study population. Moreover, previous research has attributed higher rates of MRSA infection to lower socioeconomic status and associated living conditions [16, 37–41]. Thus, the higher incidence of MRSA infection among rural children may also be attributable to living conditions associated with overall lower socioeconomic status. Indeed, a large proportion of our rural patients resided in areas within ADI quartile 4, the most “disadvantaged” group. Similar to the present study, in a study of children with cystic fibrosis in Alabama, rural residence was correlated with a higher level of deprivation, as determined by the ADI. Moreover, this study demonstrated that children living in deprived areas had a 2-fold increased risk for MRSA infection compared with those not living in deprived areas [39]. In sum, the finding of higher incidence of MRSA infection among children residing in rural areas can impact the treatment of children presenting with an illness for which S. aureus is a likely pathogen. Rural physicians, or physicians at tertiary medical centers caring for patients from rural areas, need to be aware that this patient population is at an increased risk for MRSA infection, and thus empiric antimicrobial therapy should include coverage for MRSA. The strengths of this study include applying multiple approaches (including rurality, outlying hospital care, and socioeconomic deprivation) to understand factors driving treatment failure among children with S. aureus bacteremia. While previous studies of pediatric rural health disparities have been conducted among children with chronic conditions (eg, malignancy and cystic fibrosis), this study evaluated outcomes among children with acute infections [4, 31, 39]. Several limitations are also of note. The first is the use of the RUCC as our indicator of rurality. The RUCC is a rural classification used by the USDA Economic Research Service to characterize “trends in nonmetro areas that are related to population density and metro influence” [25]. While this indicator is a useful baseline method, it was not created with health care in mind. The ideal classification system would focus on access to health care (eg, hospitals, primary care physicians, subspecialists), considerations for pediatric patients (as an adult subspecialist may not be equipped to care for children), recreational facilities and parks, and healthy food options. The second limitation is the potential bias that children of a higher acuity were transferred to SLCH for care, and those experiencing less severe illness may have been successfully treated at their local community hospital. However, the infrastructure to conduct clinical outcomes research at community hospitals is limited, and we were not able to obtain data regarding the overall incidence of S. aureus bacteremia in children at these outlying hospitals. To fully understand the clinical characteristics and management of children with S. aureus bacteremia and associated outcomes, a prospective multicenter study comprising community hospitals and tertiary care centers is needed. Third, this study was conducted at a single center and thus may not be generalizable to other regions of the United States or the world, particularly countries with differing health systems. Fourth, as there is no consensus regarding outcomes across studies of pediatric S. aureus bacteremia, we selected the composite of all-cause 90-day mortality and 90-day hospital readmission as our primary outcome measure, an outcome used in adult studies [20, 24, 42]. As mortality is rare among children with S. aureus bacteremia, determining an alternative, more optimal, measure would be of great benefit to the field. Finally, the retrospective nature of this project limited our data analysis to existing documentation, which could be mitigated through a prospective, multicenter study. CONCLUSIONS This study revealed a collision of social determinants of health impacting rural children, including a willingness to access care, the threshold that families use to determine when to seek care, and the ability to access pediatric subspecialists, diagnostics, and treatment. These factors intermingle with a potentially life-threatening illness to create a complex medical scenario. Children with S. aureus bacteremia from rural or resource-deprived areas, as well as those admitted to outlying hospitals, are at risk for adverse outcomes. A contributing factor to the state of health in rural areas is lack of research funding; only 1% of the National Institutes of Health budget is allocated to rural health, although nearly 20% of the US population lives in these areas. Strategies to address health disparities among rural populations are desperately needed. Highlighted by the recent COVID-19 pandemic, telemedicine allows patients to seek care from hours away in their own homes, in their primary care provider's office, or at a local hospital [43]. This solution is by no means perfect, with limited exam capacity and rural areas that often lack access to quality broadband internet [44]. However, access to specialists through telemedicine is undeniably valuable and has been shown to be acceptable to patients [45–49]. To overcome rural health disparities, specialized physicians in large academic centers can take multiple actions. First, foster professional relationships with their rural colleagues, allowing for phone or email consultation, opportunities for telemedicine consultation, and established referral partners. Second, develop clinical practice guidelines and educational opportunities with rural primary care physicians. Third, advocate for a nursing coordinator who can act as a liaison between local health care providers and specialists at the academic medical center [4]. These care coordinators can help to prevent delays in care, assist families in the challenges of navigating a large metropolitan health center, and ensure appropriate follow-up after hospital discharge, ultimately yielding improved health outcomes. Supplementary Material ofad296_Supplementary_Data Click here for additional data file. Acknowledgments Financial support. This work was supported by the Children's Discovery Institute of Washington University and St. Louis Children’s Hospital; National Institutes of Health (NIH)/National Center for Advancing Translational Sciences (grant numbers UL1-TR002345 and TL1-TR002344 to K.J.W.); and the Agency for Healthcare Research and Quality (AHRQ; grant number R01-HS024269 to S.A.F.). Disclaimer. These funding sources had no role in the design or conduct of the study; collection, management, analysis, or interpretation of the data; preparation, review, or approval of the manuscript; or the decision to submit the manuscript for publication. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH or AHRQ. Supplementary Data Supplementary materials are available at Open Forum Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author. ==== Refs References 1 Alvidrez J , CastilleD, Laude-SharpM, RosarioA, TaborD. The National Institute on Minority Health and Health Disparities research framework. Am J Public Health 2019; 109 (Suppl 1 ):S16–20.30699025 2 Singh GK , SiahpushM. Widening rural-urban disparities in all-cause mortality and mortality from major causes of death in the USA, 1969–2009. J Urban Health 2014; 91 :272–92.24366854 3 Matthews KA , CroftJB, LiuY, et al Health-related behaviors by urban-rural county classification—United States, 2013. MMWR Surveill Summ 2017; 66 :1–8. 4 Bristow S , PowerT, JacksonD, UsherK. Conquering the great divide: rural mothers of children with chronic health conditions accessing specialist medical care for their children. Collegian 2020; 27 :319–26. 5 Douthit N , KivS, DwolatzkyT, BiswasS. Exposing some important barriers to health care access in the rural USA. Public Health 2015; 129 :611–20.26025176 6 Elson LE , LukeAA, BarkerAR, McBrideTD, Joynt MaddoxKE. Trends in hospital mortality for uninsured rural and urban populations, 2012–2016. J Rural Health 2021; 37 :318–27.32472709 7 The Cecil G. Sheps Center for Health Services Research at The University of North Carolina at Chapel Hill . Rural hospital closures. 2021. Available at: https://www.shepscenter.unc.edu/programs-projects/rural-health/rural-hospital-closures/. Accessed 1 May, 2021. 8 Paez K , SchurC, ZhaoL, LucadoJ. A national study of nurse leadership and supports for quality improvement in rural hospitals. Am J Med Qual 2013; 28 :127–34.22822169 9 O’Hanlon CE , KranzAM, DeYoreoM, MahmudA, DambergCL, TimbieJ. Access, quality, and financial performance of rural hospitals following health system affiliation. Health Aff (Millwood) 2019; 38 :2095–104.31794306 10 Bennett CL , EspinolaJA, SullivanAF, et al Evaluation of the 2020 pediatric emergency physician workforce in the US. JAMA Netw Open 2021; 4 :e2110084. 11 Pasquali SK , LamWK, ChiswellK, KemperAR, LiJS. Status of the pediatric clinical trials enterprise: an analysis of the US ClinicalTrials.gov registry. Pediatrics 2012; 130 :e1269–77.23027172 12 Davis TC , ArnoldCL, MillsG, MieleL. A qualitative study exploring barriers and facilitators of enrolling underrepresented populations in clinical trials and biobanking. Front Cell Dev Biol 2019; 7 :74.31114788 13 Hamdy RF , DonaD, JacobsMB, GerberJS. Risk factors for complications in children with Staphylococcus aureus bacteremia. J Pediatr 2019; 208 :214–20.e2.30879729 14 Kumarachandran G , JohnsonJK, ShirleyDA, GraffunderE, HeilEL. Predictors of adverse outcomes in children with Staphylococcus aureus bacteremia. J Pediatr Pharmacol Ther 2017; 22 :218–26.28638305 15 Cardenas-Comfort C , KaplanSL, VallejoJG, McNeilJC. Follow-up blood cultures in children with Staphylococcus aureus bacteremia. Pediatrics 2020; 146 :e20201821. 16 Iwamoto M , MuY, LynfieldR, et al Trends in invasive methicillin-resistant Staphylococcus aureus infections. Pediatrics 2013; 132 :e817–24.24062373 17 Friedland IR , du PlessisJ, CilliersA. Cardiac complications in children with Staphylococcus aureus bacteremia. J Pediatr 1995; 127 :746–8.7472829 18 Duguid RC , Al ReesiM, BartlettAW, PalasanthiranP, McMullanBJ. Impact of infectious diseases consultation on management and outcome of Staphylococcus aureus bacteremia in children. J Pediatric Infect Dis Soc 2021; 10 :569–75.33355663 19 Saunderson RB , GouliourisT, CartwrightEJ, et al Impact of infectious diseases consultation on the management of Staphylococcus aureus bacteraemia in children. BMJ Open 2014; 4 :e004659. 20 Campbell AJ , Al YazidiLS, PhuongLK, et al Pediatric Staphylococcus aureus bacteremia: clinical spectrum and predictors of poor outcome. Clin Infect Dis 2022; 74 :604–13.34089594 21 Whittington KJ , MaY, ButlerAM, et al The impact of infectious diseases consultation for children with Staphylococcus aureus bacteremia. Pediatr Res 2022; 92 :1598–605.35982140 22 Harris PA , TaylorR, ThielkeR, PayneJ, GonzalezN, CondeJG. Research Electronic Data Capture (REDCap)—a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform 2009; 42 :377–81.18929686 23 Harris PA , TaylorR, MinorBL, et al The REDCap Consortium: building an international community of software platform partners. J Biomed Inform 2019; 95 :103208. 24 Tong SYC , LyeDC, YahavD, et al Effect of vancomycin or daptomycin with vs without an antistaphylococcal beta-lactam on mortality, bacteremia, relapse, or treatment failure in patients with MRSA bacteremia: a randomized clinical trial. JAMA 2020; 323 :527–37.32044943 25 United States Department of Agriculture ERS . Rural-urban continuum codes. 2019. Available at: https://www.ers.usda.gov/data-products/rural-urban-continuum-codes.aspx. Accessed January 20, 2020. 26 Kind AJH , BuckinghamWR. Making neighborhood-disadvantage metrics accessible—the neighborhood atlas. N Engl J Med 2018; 378 :2456–8.29949490 27 Cyr ME , EtchinAG, GuthrieBJ, BenneyanJC. Access to specialty healthcare in urban versus rural US populations: a systematic literature review. BMC Health Serv Res 2019; 19 :974.31852493 28 Mayer ML . Disparities in geographic access to pediatric subspecialty care. Matern Child Health J 2008; 12 :624–32.17879148 29 Pletcher BA , RimszaME, CullWL, ShipmanSA, ShugermanRP, O'ConnorKG. Primary care pediatricians’ satisfaction with subspecialty care, perceived supply, and barriers to care. J Pediatr 2010; 156 :1011–5.e1. 30 Bolin JN , BellamyGR, FerdinandAO, et al Rural healthy people 2020: new decade, same challenges. J Rural Health 2015; 31 :326–33.25953431 31 Walling EB , FialaM, ConnollyA, DrevenakA, GehlertS. Challenges associated with living remotely from a pediatric cancer center: a qualitative study. J Oncol Pract 2019; 15 :e219–29.30702962 32 Tarnasky AM , OlivereLA, LedbetterL, TracyET. Examining the effect of travel distance to pediatric cancer centers and rurality on survival and treatment experiences: a systematic review. J Pediatr Hematol Oncol 2021; 43 :159–71.33625091 33 Nielsen M , D’AgostinoD, GregoryP. Addressing rural health challenges head on. Mo Med 2017; 114 :363–6.30228634 34 Probst JC , BarkerJC, EndersA, GardinerP. Current state of child health in rural America: how context shapes children's health. J Rural Health 2018; 34 (Suppl 1) :s3–12.27677973 35 Wattles BA , VidwanNK, FeyginY, JawadKS, CreelLM, SmithMJ. Antibiotic prescribing to Kentucky Medicaid children, 2012–2017: prescribing is higher in rural areas. J Rural Health 2022; 38 :427–32.33978987 36 Kilgore JT , LanataMM, WillisJM, et al Utilization of West Virginia pediatric Medicaid claims data to guide outpatient antimicrobial stewardship interventions. J Pediatric Infect Dis Soc 2022; 11 :172–6.34939655 37 Hota B , EllenbogenC, HaydenMK, AroutchevaA, RiceTW, WeinsteinRA. Community-associated methicillin-resistant Staphylococcus aureus skin and soft tissue infections at a public hospital: do public housing and incarceration amplify transmission? Arch Intern Med 2007; 167 :1026–33.17533205 38 Milstone AM , CarrollKC, RossT, ShangrawKA, PerlTM. Community-associated methicillin-resistant Staphylococcus aureus strains in pediatric intensive care unit. Emerg Infect Dis 2010; 16 :647–55.20350379 39 Oates GR , HarrisWT, RoweSM, et al Area deprivation as a risk factor for methicillin-resistant Staphylococcus aureus infection in pediatric cystic fibrosis. Pediatr Infect Dis J 2019; 38 :e285–9.31568067 40 See I , WessonP, GualandiN, et al Socioeconomic factors explain racial disparities in invasive community-associated methicillin-resistant Staphylococcus aureus disease rates. Clin Infect Dis 2017; 64 :597–604.28362911 41 Casey JA , CosgroveSE, StewartWF, PollakJ, SchwartzBS. A population-based study of the epidemiology and clinical features of methicillin-resistant Staphylococcus aureus infection in Pennsylvania, 2001–2010. Epidemiol Infect 2013; 141 :1166–79.22929058 42 Hamdy RF , HsuAJ, StockmannC, et al Epidemiology of methicillin-resistant Staphylococcus aureus bacteremia in children. Pediatrics 2017; 139 :e20170183. 43 Demeke HB , MeraliS, MarksS. Trends in use of telehealth among health centers during the COVID-19 pandemic—United States, June 26–November 6, 2020. MMWR 2021; 70 :240–4.33600385 44 Perrin A . Digital Gap Between Rural and Nonrural America Persists. Pew Research Center; 2019. 45 Burnham JP , FritzSA, YaegerLH, ColditzGA. Telemedicine infectious diseases consultations and clinical outcomes: a systematic review. Open Forum Infect Dis 2019; 6 (12 ):ofz517. 46 Evers EC , FritzSA, ColditzGA, BurnhamJP. Perceptions of telemedicine and costs incurred by a visit to a general infectious diseases clinic: a survey. Open Forum Infect Dis 2022; 9 (3 ):ofab661. 47 Tully L , CaseL, ArthursN, SorensenJ, MarcinJP, O'MalleyG. Barriers and facilitators for implementing paediatric telemedicine: rapid review of user perspectives. Front Pediatr 2021; 9 :630365. 48 Nyoni T , EversEC, PérezM, et al Perceived barriers and facilitators to the adoption of telemedicine infectious diseases consultations in Southeastern Missouri hospitals. J Telemed Telecare. 2023:1357633x221149461. 49 Burnham JP , FritzSA, ColditzGA. Implementation of telemedicine infectious diseases consultation in a rural hospital using the active implementation framework. Open Forum Infect Dis 2022; 9 (11 ):ofac523.
PMC010xxxxxx/PMC10352650.txt
==== Front Pulm Circ Pulm Circ 10.1002/(ISSN)2045-8940 PUL2 Pulmonary Circulation 2045-8932 2045-8940 John Wiley and Sons Inc. Hoboken 10.1002/pul2.12268 PUL212268 Research Article Research Articles Inhaled iloprost is an effective alternative therapy for persistent pulmonary hypertension in newborns YILDIRIM Yıldırım Şükran http://orcid.org/0000-0002-9678-1571 1 [email protected] 1 Istanbul Prof. Dr. Cemil Tascioglu City Hospital, Neonatal Intensive Care Unit University of Health Sciences Istanbul Sisli Turkey * Correspondence Şükran Yıldırım, University of Health Sciences, Istanbul Prof. Dr. Cemil Tascioglu City Hospital, Neonatal Intensive Care Unit, Sisli, Istanbul, Turkey. Email: [email protected] 18 7 2023 7 2023 13 3 10.1002/pul2.v13.3 e1226829 6 2023 18 2 2023 05 7 2023 © 2023 The Authors. Pulmonary Circulation published by John Wiley & Sons Ltd on behalf of Pulmonary Vascular Research Institute. https://creativecommons.org/licenses/by-nc/4.0/ This is an open access article under the terms of the http://creativecommons.org/licenses/by-nc/4.0/ License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes. Abstract Persistent pulmonary hypertension of the newborn (PPHN) is one of the diseases of the neonate with severe potential morbidity and mortality. Inhaled iloprost, a stable prostacyclin analog, has been suggested as an alternative treatment for inhaled nitric oxide (iNO). However, more data on neonates' dosing, setting, and effectiveness still needs to be solved. This study suggests using inhaled iloprost as rescue therapy for PPHN based on our experience. This was a retrospective study. The data from medical records of six newborns diagnosed with PPHN and had received inhaled iloprost from December 2019 to April 2022 were collected. Demographic and clinical features, dosing regimen, changes in oxygenation index, echocardiographic findings, and mortality were evaluated. The inhalation dose was 2−4 mcg/dose, and 3−48 inhalations per day were applied over 2−7 days. Inhaled iloprost was effective in all patients. No side effects were attributable to inhaled iloprost, and no mortality was recorded. Our experience suggests that inhaled iloprost can be used as a first‐line therapy in newborn infants with PPHN when iNO is unavailable. However, there are large fluctuations in the oxygenation index due to the setting. inhaled iloprost persistent pulmonary hypertension of the newborn prostacyclin None source-schema-version-number2.0 cover-dateJuly 2023 details-of-publishers-convertorConverter:WILEY_ML3GV2_TO_JATSPMC version:6.3.2 mode:remove_FC converted:18.07.2023 Yıldırım Ş . Inhaled iloprost is an effective alternative therapy for persistent pulmonary hypertension in newborns. Pulm Circ. 2023;13 :e12268. 10.1002/pul2.12268 ==== Body pmcINTRODUCTION Persistent pulmonary hypertension of the newborn (PPHN) results from a failure of the newborn's normal circulatory transition, characterized by marked hypoxemia secondary to right‐to‐left extrapulmonary shunting of deoxygenated blood. Generally, it is a disease of the term or near‐term neonates. The incidence of PPHN ranges from 0.4 to 6.8 per 1000 live births, 1 with an associated mortality of 4− 60%. 2 , 3 Also, significant long‐term morbidities of up to 25% are reported. 2 The pathophysiological process of PPHN may involve acute pulmonary vasoconstriction, pulmonary vascular remodeling, pulmonary vascular hypoplasia, or pulmonary intravascular obstruction. 4 , 5 , 6 Echocardiography is necessary to rule out cyanotic congenital heart disease. 7 A specific pulmonary artery pressure is defined for adult primary pulmonary hypertension, not PPHN. Right‐to‐left shunt without congenital heart disease is enough for diagnosing PPHN, regardless of the pulmonary arterial pressure. 5 Inhaled nitric oxide (iNO) is the only approved pulmonary vasodilator for treating PPHN. Nevertheless, it does not improve survival, and ~40% of neonates fail to respond. 8 It is still unclear whether iNO is safer and more effective than other vasodilators delivered by inhalation, such as inhaled prostacyclin. 8 Therefore, investigation of the efficacy and safety of other potential therapeutic agents, such as pulmonary vasodilators like prostanoids, phosphodiesterase inhibitors like sildenafil and milrinone, and endothelin antagonists like bosentan is ongoing. 9 , 10 , 11 , 12 Acquired neonatal diseases like meconium aspiration syndrome, asphyxia, sepsis, transient tachypnea of the newborn, and respiratory distress syndrome may cause PPHN as well as congenital diseases like diaphragmatic hernia, alveolar capillary dysplasia, surfactant protein defects and cardiac defects. 13 Increased pulmonary vascular resistance is the final result of the above. 14 Prostacyclin, a naturally occurring prostaglandin, is a potent vasodilator with antithrombotic, antiproliferative, and anti‐inflammatory effects. 15 The convenience for the prostacyclin analogs to treat PPHN is evident. 16 , 17 Iloprost is an analog of prostacyclin with more excellent chemical stability, making it practical for treatment purposes. 18 Prostacylin has been used and studied over time for PPHN. 19 Several reports have addressed inhaled iloprost treatment for pulmonary hypertension in the newborn literature, 19 , 20 , 21 , 22 , 23 , 24 , 25 , 26 , 27 , 28 , 29 , 30 , 31 , 32 , 33 , 34 , 35 all of which emphasize the need for further research. So, to contribute to the literature, six patients with PPHN treated with inhaled iloprost because of lack of accessibility to iNO are presented in detail in this article, and potential obstacles for iloprost inhalation use are discussed. MATERIALS AND METHODS We retrospectively analyzed the records of 6 patients with PPHN treated with inhaled iloprost at neonatal intensive care at the University of Health Sciences, Prof. Dr. Cemil Tascioglu City Hospital, between December 2019 and April 2022. Demographic and clinical features of neonates with persistent pulmonary hypertension are presented (Table 1). The eligibility criteria for inhaled iloprost treatment in persistent pulmonary hypertension on clinical basis of our unit are shown (Source file 1). Inhaled iloprost (20 μg/2 mL, Ventavis, Bayer, Leverkusen, Germany) was administered at a dose of 2−4 mcg (microgram) every 1/2−8 h according to the protocol used for inhaled iloprost treatment in the absence of nitric oxide for newborns with persistent pulmonary hypertension (Supporting Information: Source file 2) by integrating a nebulizer into the ventilator circuit close to humidifier (Supporting Information: Video 1). Note that the eligibility criteria and the protocol are performed on a clinical basis in our unit, not part of a prospective research study. Table 1 Demographic and clinical features of neonates with persistent pulmonary hypertension. Patient no 1 2 3 4 5 6 Birthweight (g) 3380 3905 2250 2320 3215 3400 Delivery method C/S NVD C/S C/S C/S NVD Gender Male Female Female Male Male Female Gestational age 35 w 6 d 38 w 34 w 36 w 2 d 38 w 39 w Diagnose Infant of a diabetic mother, cerebral ventriculomegaly, resuscitation, and intubation in the delivery room Convulsion, Chronic intrauterine hypoxia? No pregnancy follow‐up, intubated at postnatal 9 h Infant of a mother with SARS‐CoV‐2 pneumonia,a resuscitation, and intubation in the delivery room, acute renal failure, grade 4 IVH Infant of a mother with gestational hypertension, resuscitation and intubation in the delivery room Infant of a diabetic mother, birth trauma, resuscitation, and intubation in the delivery room, convulsions Perinatal asphyxia, meconium aspiration syndrome, resuscitation and intubation in the delivery room Oxygenation index before iloprost treatment 9 18 21 23 9 15 Age at referral (h) 38 31 51 9 44 2.5 Pulmonary artery pressure (basal and at the end of therapy) (mm/Hg) 45/25 60/NA 45/20 60/45 55/18 35/25 Iloprost treatment duration (h) 156 61 205 77 152 52 Maximal dose (mcg/day) 14 12 70 48 192 36 Cumulative dose (mcg) 62 66 230 92 510 64 Inotrope − + + + + + Surfactant (100 mg/kg) 2 doses (−) (considered CCHD at first) 1 dose 1 dose 2 doses 2 doses Response to surfactant FiO2 remained >40% after two doses − No use, FiO2 remained 100% No use, FiO2 remained 100% FiO2 remained >40% after two doses FiO2 remained >40% after two doses Chest X‐ray Hyperaeration Hyperaeration Patchy infiltrates in the left lung, hyperaeration in the right lung Hyperaeration Normal Patchy infiltrates Sedative + + + + + + Hospitalization (d) 19 25 27 10 14 14 Abbreviations: CCHD, congenital cyanotic heart disease; C/S, cesarean section; d, day; g, gram; h, hour; IVH, intraventricular hemorrhage; MAS, meconium aspiration syndrome; mcg, microgram; NVD, normal vaginal delivery; OI, oxygenation index; PPHN, persistent pulmonary hypertension of the newborn; RDS, respiratory distress syndrome; w, week. a The baby was SARS‐CoV‐2 negative. John Wiley & Sons, Ltd. Echocardiography demonstrated a right‐to‐left shunt or a bidirectional shunt and no structural anomaly in all patients. Immediate echocardiographic confirmation of a drop in pulmonary arterial pressure after inhaled iloprost therapy was not available but was performed during or at the end of the treatment in 5 patients (Table 1). Pulmonary arterial pressure was measured from the systolic measures from tricuspid regurgitation. Milrinone infusion was administered to the patient 3 to alleviate the afterload because of very high ventilatory pressures, and sildenafil was to patient 5 after a rising oxygenation index due to 4 mcg/dose of iloprost. Inhaled iloprost was escalated to 4 mcg/dose for patients 3 and 5 after the initial dose because of failure to respond to 2 mcg/dose. The sequence of therapies that might affect the oxygenation index during the treatment is also revealed (Figures 1, 2, 3, 4, 5, 6). Figure 1 Oxygenation index of patient 1. Figure 2 Oxygenation index of patient 2. Figure 3 Oxygenation index of patient 3. Figure 4 Oxygenation index of patient 4. Figure 5 Oxygenation index of patient 5. Figure 6 Oxygenation index of patient 6. RESULTS Inhaled iloprost decreased fiO2 and improved oxygen saturation in all patients (Table 1, Figures 1, 2, 3, 4, 5, 6). The only issue to consider about the therapy was the fluctuations in the oxygenation index. (Figures 1, 2, 3, 4, 5, 6). No side effects were attributable to inhaled iloprost, and no mortality was recorded. DISCUSSION iNO is the only approved therapy for PPHN. 8 In Turkey, iNO is delivered by a private company. Sometimes the devices are occupied, so alternative treatments have to be considered now and then. Inhaled iloprost is suggested as a sole or adjunctive therapy in the literature. It is readily available and has comparable clinical effects to iNO. 20 , 21 , 22 , 23 , 24 , 25 , 26 , 27 , 28 , 29 , 30 , 31 , 32 , 33 , 34 , 35 Due to ethical issues, conducting randomized trials of inhaled iloprost is challenging, so the clinical experience has gained prominence. In terms of dosage and timing, there are no specific guidelines for the inhaled treatment of iloprost. It has a biological half‐life of 20−30 min in humans. 36 For the treatment of adult PAH, a dose of 2.5−7.5 mcg, 6−9 times a day, with a maximum amount of 45 mcg per day, was approved in 2004. 37 The same doses are used in children with success. 38 For neonates, different dosages are suggested in the literature; 0.2−2.5 mcg/kg/dose every 1−6 h in general. 19 , 20 , 21 , 22 , 23 , 24 , 25 , 26 , 27 , 28 , 29 , 30 , 31 , 32 , 33 , 34 , 35 A report of continuous inhalation has also been made. 33 Delivering one dose of inhalation lasts approximately 5 min in our setting, and the clinical effect lasts 1−2 h at most at the beginning of treatment. Unfortunately, this causes unstable oxygen saturation. To avoid this instability, the caregiver has to react quickly to fiO2 and oxygen saturation and decide the dosing interval. Although preferable, continuous inhalation was impossible in this setting. Techniques needed to be developed to give constant inhalation and increase the dosage, like iNO, which could provide a more stable oxygen saturation and hemodynamics. Another issue to consider is the imprecision of the actual amount of the drug delivered to the baby using an integrated nebulizer to the ventilator circuit. A certain amount may be lost in the setting. Also, all the patients mentioned above have somewhat gone through acute or chronic hypoxia and developed PPHN due to acute pulmonary vasoconstriction or pulmonary vascular remodeling mechanisms. It needs to be clarified if it works for PPHN caused by other reasons. 1 , 4 , 5 , 6 As for the side effects, headache, cough, and dizziness were reported in adults and children receiving inhaled iloprost. Also, intravenous iloprost was reported to cause hypotension. 39 , 40 For the neonates, no side effects attributable to inhaled iloprost were reported. 19 , 20 , 21 , 22 , 23 , 24 , 25 , 26 , 27 , 28 , 29 , 30 , 31 , 32 , 33 , 34 , 35 A maximal dose of 192 mcg/day and a cumulative dose of 510 mcg/day caused no side effects in our patients, which means very high maximal doses compared to adults are well tolerated in neonates. Two patients received inotrope for hypotension but were already on inotropic support before iloprost. Another two had inotropes to raise the mean arterial pressure to pulmonary arterial pressure, not because of iloprost‐induced hypotension. So, we concluded that the adverse effects of inhaled iloprost are negligible in the neonatal population. 19 , 20 , 21 , 22 , 23 , 24 , 25 , 26 , 27 , 28 , 29 , 30 , 31 , 32 , 33 , 34 , 35 Inhaled iloprost is a reliable alternative for infants with persistent pulmonary hypertension when iNO is unavailable and may be used as adjunctive therapy along with other pulmonary vasodilators. 20 , 21 , 22 , 23 , 24 , 25 , 26 , 27 , 28 , 29 , 30 , 31 , 32 , 33 , 34 , 35 Compared to iNO, the pros of inhaled iloprost are the lower cost and availability, and the cons are swings in oxygenation (Figures 1, 2, 3, 4, 5, 6). In conclusion, our clinical experience supports that inhaled iloprost might be an alternative drug treatment for PPHN and provides safety and efficacy data that is insufficient for now in the literature. Well‐designed trials are warranted to remedy the lack of evidence. AUTHOR CONTRIBUTIONS Only one author and no guarantor. I declare that I participated in the design, execution, and analysis of the paper by Sukran Yildirim entitled “Inhaled iloprost is an effective alternative therapy for persistent pulmonary hypertension in newborns” that I have seen and approved the final version and that it has neither been published nor submitted elsewhere. I also declare that I have no conflict of interest other than any noted in the cover letter to the editor. CONFLICT OF INTEREST STATEMENT The author declares no conflict of interest. ETHICS STATEMENT N/A. Supporting information Supporting information. Click here for additional data file. Supporting information. Click here for additional data file. Supplementary video 1 caption: Nebulizer integration to a ventilator for inhaled iloprost delivery. Click here for additional data file. DATA AVAILABILITY STATEMENT The data supporting this study's findings are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions. ==== Refs REFERENCES 1 Fuloria M , Aschner JL . Persistent pulmonary hypertension of the newborn. Semin Fetal Neonatal Med. 2017;22 (4 ):220–226. 10.1016/j.siny.2017.03.004 28342684 2 Konduri GG , Vohr B , Robertson C , Sokol GM , Solimano A , Singer J , Ehrenkranz RA , Singhal N , Wright LL , Van Meurs K , Stork E , Kirpalani H , Peliowski A , Johnson Y . Early inhaled nitric oxide therapy for term and near‐term newborn infants with hypoxic respiratory failure: neurodevelopmental follow‐up. J Pediatr. 2007;150 (3 ):235–240.e1. 10.1016/j.jpeds.2006.11.065 17307536 3 Walsh‐Sukys MC , Tyson JE , Wright LL , Bauer CR , Korones SB , Stevenson DK , Verter J , Stoll BJ , Lemons JA , Papile LA , Shankaran S , Donovan EF , Oh W , Ehrenkranz RA , Fanaroff AA . Persistent pulmonary hypertension of the newborn in the era before nitric oxide: practice variation and outcomes. Pediatrics. 2000;105 (1 Pt 1 ):14–20. 10.1542/peds.105.1.14 10617698 4 Lakshminrusimha S . The pulmonary circulation in neonatal respiratory failure. Clin Perinatol. 2012;39 (3 ):655–683. 10.1016/j.clp.2012.06.006 22954275 5 Cabral JEB , Belik J . Persistent pulmonary hypertension of the newborn: recent advances in pathophysiology and treatment. J Pediatr. 2013;89 (3 ):226–242. 10.1016/j.jped.2012.11.009 6 Storme L , Aubry E , Rakza T , Houeijeh A , Debarge V , Tourneux P , Deruelle P , Pennaforte T . Pathophysiology of persistent pulmonary hypertension of the newborn: impact of the perinatal environment. Arch Cardiovasc Dis. 2013;106 (3 ):169–177. 10.1016/j.acvd.2012.12.005 23582679 7 Mathew B , Lakshminrusimha S . Persistent pulmonary hypertension in the newborn. Children. 2017;4 (8 ):63. 10.3390/children4080063 28788074 8 Barrington KJ , Finer N , Pennaforte T , Altit G . Nitric oxide for respiratory failure in infants born at or near term. Cochrane Database Syst Rev. 2017;2017 (1 ):CD000399. 10.1002/14651858.CD000399.pub3 9 Shivanna B , Gowda S , Welty SE , Barrington KJ , Pammi M . Prostanoids and their analogues for the treatment of pulmonary hypertension in neonates. Cochrane Database Syst Rev. 2019;2019 (10 ):CD012963. 10.1002/14651858.CD012963.pub2 10 Perez KM , Laughon M . Sildenafil in term and premature infants: a systematic review. Clin Ther. 2015;37 (11 ):2598–2607.e1. 10.1016/j.clinthera.2015.07.019 26490498 11 Bassler D , Kreutzer K , McNamara P , Kirpalani H . Milrinone for persistent pulmonary hypertension of the newborn. Cochrane Database Syst Rev. 2010;2010 (11 ):CD007802. 10.1002/14651858.CD007802.pub2 21069698 12 Maneenil G , Thatrimontrichai A , Janjindamai W , Dissaneevate S . Effect of bosentan therapy in persistent pulmonary hypertension of the newborn. Pediat Neonatol. 2018;59 (1 ):58–64. 10.1016/j.pedneo.2017.02.003 13 Singh Y , Lakshminrusimha S . Pathophysiology and management of persistent pulmonary hypertension of the newborn. Clin Perinatol. 2021;48 (3 ):595–618. 10.1016/j.clp.2021.05.009 34353582 14 Singh Y , Tissot C . Echocardiographic evaluation of transitional circulation for the neonatologists. Front Pediatr. 2018;6 :140. 10.3389/fped.2018.00140 29868528 15 Moncada S , Gryglewski R , Bunting S , Vane JR . An enzyme isolated from arteries transforms prostaglandin endoperoxides to an unstable substance that inhibits platelet aggregation. Nature. 1976;263 (5579 ):663–665. 10.1038/263663a0 802670 16 Tuder RM , Cool CD , Geraci MW , Wang J , Abman SH , Wright L , Badesch D , Voelkel NF . Prostacyclin synthase expression is decreased in lungs from patients with severe pulmonary hypertension. Am J Respir Crit Care Med. 1999;159 (6 ):1925–1932. 10.1164/ajrccm.159.6.9804054 10351941 17 Christman BW , McPherson CD , Newman JH , King GA , Bernard GR , Groves BM , Loyd JE . An imbalance between the excretion of thromboxane and prostacyclin metabolites in pulmonary hypertension. N Engl J Med. 1992;327 (2 ):70–75. 10.1056/NEJM199207093270202 1603138 18 Krause W , Krais T . Pharmacokinetics and pharmacodynamics of the prostacyclin analogue iloprost in man. Eur J Clin Pharmacol. 1986;30 (1 ):61–68. 10.1007/BF00614197 2423342 19 Eronen M , Pohjavuori M , Andersson S , Pesonen E , Raivio KO . Prostacyclin treatment for persistent pulmonary hypertension of the newborn. Pediatr Cardiol. 1997;18 (1 ):3–7. 10.1007/s002469900099 8960484 20 Kelly LK , Porta NFM , Goodman DM , Carroll CL , Steinhorn RH . Inhaled prostacyclin for term infants with persistent pulmonary hypertension refractory to inhaled nitric oxide. J Pediatr. 2002;141 (6 ):830–832. 10.1067/mpd.2002.129849 12461501 21 Ehlen M , Wiebe B . Iloprost in persistent pulmonary hypertension of the newborn. Cardiol Young. 2003;13 (4 ):361–363. 10.1017/s1047951103000726 14694957 22 Concheiro Guisán A , Sousa Rouco C , Suárez Traba B . Iloprost inhalado: una alternativa terapéutica para la hipertensión pulmonar persistente del recién nacido [Inhaled iloprost: a therapeutic alternative for persistent pulmonary hypertension of the newborn]. An Pediatr (Barc). 2005;63 (2 ):175–176. 10.1157/13077463 16045880 23 De Luca D , Zecca E , Piastra M , Romagnoli C . Iloprost as ‘rescue’ therapy for pulmonary hypertension of the neonate. Pediatric Anesthesia. 2007;17 (4 ):394–395. 10.1111/j.1460-9592.2006.02104.x 17359413 24 Chotigeat U , Jaratwashirakul S . Inhaled iloprost for severe persistent pulmonary hypertension of the newborn. J Med Assoc Thai. 2007;90 (1 ):167–170.17621749 25 Limsuwan A , Wanitkul S , Khosithset A , Attanavanich S , Samankatiwat P . Aerosolized iloprost for postoperative pulmonary hypertensive crisis in children with congenital heart disease. Int J Cardiol. 2008;129 (3 ):333–338. 10.1016/j.ijcard.2007.08.084 18096256 26 Eifinger F , Sreeram N , Mehler K , Huenseler C , Kribs A , Roth B . Aerosolized iloprost in the treatment of pulmonary hypertension in extremely preterm infants: a pilot study. Klin Padiatr. 2008;220 (2 ):66–69. 10.1055/s-2007-984370 17710738 27 Verma S , Lumba R , Kazmi SH , Vaz MJ , Prakash SS , Bailey SM , Mally PV , Randis TM . Effects of inhaled iloprost for the management of persistent pulmonary hypertension of the newborn. Am J Perinatol. 2022;39 (13 ):1441–1448. 10.1055/s-0040-1722653 33477175 28 Gokce I , Kahveci H , Turkyilmaz Z , Adakli B , Zeybek C . Inhaled iloprost in the treatment of pulmonary hypertension in very low birth weight infants: a report of two cases. J Pak Med Assoc. 2012;62 (4 ):388–391.22755287 29 Avila‐Alvarez A , Bravo‐Laguna MC , Bronte LD , Cerro MJD . Inhaled iloprost as a rescue therapy for transposition of the great arteries with persistent pulmonary hypertension of the newborn. Pediatr Cardiol. 2013;34 (8 ):2027–2029. 10.1007/s00246-012-0575-2 23143352 30 Kahveci H , Yilmaz O , Avsar UZ , Ciftel M , Kilic O , Laloglu F , Ozturk K . Oral sildenafil and inhaled iloprost in the treatment of pulmonary hypertension of the newborn. Pediatr Pulmonol. 2014;49 (12 ):1205–1213. 10.1002/ppul.22985 24420987 31 Yilmaz O , Kahveci H , Zeybek C , Ciftel M , Kilic O . Inhaled iloprost in preterm infants with severe respiratory distress syndrome and pulmonary hypertension. Am J Perinatol. 2014;31 (4 ):321–326. 10.1055/s-0033-1348949 23797963 32 Chotigeat U , Champrasert M , Khorana M . Iloprost inhalation for the treatment of severe persistent pulmonary hypertension of the newborn, experience at QSNICH. J Med Assoc Thai. 2014;97 (Suppl 6 ):S89–S94. 33 Dykes JC , Torres M , Alexander PJ . Continuous inhaled iloprost in a neonate with d‐transposition of the great arteries and severe pulmonary arterial hypertension. Cardiol Young. 2016;26 (3 ):571–573. 10.1017/S1047951115001250 26220108 34 Unal S , Aktas S , Aksu M , Hirfanoglu IM , Atalay Y , Turkyilmaz C . Iloprost instillation in two neonates with pulmonary hypertension. JCPSP. 2017;27 (4 ):257–259.28492159 35 Kim SH , Lee HJ , Kim NS , Park HK . Inhaled iloprost as a First‐Line therapy for persistent pulmonary hypertension of the newborn. Neonatal Med. 2019;26 (4 ):191–197. 10.5385/nm.2019.26.4.191 36 Ewert R , Schäper C , Halank M , Gläser S , Opitz CF . Inhalative iloprost—pharmacology and clinical application. Expert Opin Pharmacother. 2009;10 (13 ):2195–2207. 10.1517/14656560903164228 19663611 37 Humbert M , Sitbon O , Simonneau G . Treatment of pulmonary arterial hypertension. N Engl J Med. 2004;351 (14 ):1425–1436. 10.1056/NEJMra040291 15459304 38 Ivy DD , Doran AK , Smith KJ , Mallory GB , Beghetti M , Barst RJ , Brady D , Law Y , Parker D , Claussen L , Abman SH . Short‐ and long‐term effects of inhaled iloprost therapy in children with pulmonary arterial hypertension. JACC. 2008;51 (2 ):161–169. 10.1016/j.jacc.2007.09.031 18191742 39 Kingman M , Archer‐Chicko C , Bartlett M , Beckmann J , Hohsfield R , Lombardi S . Management of prostacyclin side effects in adult patients with pulmonary arterial hypertension. Pulm Circ. 2017;7 (3 ):598–608. 10.1177/2045893217719250 28632002 40 Janjindamai W , Thatrimontrichai A , Maneenil G , Chanvitan P , Dissaneevate S . Effectiveness and safety of intravenous iloprost for severe persistent pulmonary hypertension of the newborn. Indian Pediatr. 2013;50 (10 ):934–938. 10.1007/s13312-013-0263-1 23798625
PMC010xxxxxx/PMC10352651.txt
==== Front Open Forum Infect Dis Open Forum Infect Dis ofid Open Forum Infectious Diseases 2328-8957 Oxford University Press US 10.1093/ofid/ofad334 ofad334 Major Article AcademicSubjects/MED00290 Editor's Choice Distribution, Trends, and Antimicrobial Susceptibility of Bacteroides, Clostridium, Fusobacterium, and Prevotella Species Causing Bacteremia in Japan During 2011–2020: A Retrospective Observational Study Based on National Surveillance Data https://orcid.org/0000-0002-7576-2654 Kajihara Toshiki Antimicrobial Resistance Research Center, National Institute of Infectious Diseases, Tokyo, Japan Yahara Koji Antimicrobial Resistance Research Center, National Institute of Infectious Diseases, Tokyo, Japan Kitamura Norikazu Antimicrobial Resistance Research Center, National Institute of Infectious Diseases, Tokyo, Japan Hirabayashi Aki Antimicrobial Resistance Research Center, National Institute of Infectious Diseases, Tokyo, Japan Hosaka Yumiko Antimicrobial Resistance Research Center, National Institute of Infectious Diseases, Tokyo, Japan Sugai Motoyuki Antimicrobial Resistance Research Center, National Institute of Infectious Diseases, Tokyo, Japan Correspondence: Toshiki Kajihara, MD, Phd, Antimicrobial Resistance Research Center, National Institute of Infectious Diseases, 4-2-1 Aoba-cho Higashimurayama, Tokyo 189-0002, Japan ([email protected]). Potential conflicts of interest. All authors: no reported conflicts. 7 2023 03 7 2023 03 7 2023 10 7 ofad33405 4 2023 22 6 2023 29 6 2023 18 7 2023 © The Author(s) 2023. Published by Oxford University Press on behalf of Infectious Diseases Society of America. 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs licence (https://creativecommons.org/licenses/by-nc-nd/4.0/), which permits non-commercial reproduction and distribution of the work, in any medium, provided the original work is not altered or transformed in any way, and that the work is properly cited. For commercial re-use, please contact [email protected] Abstract Background The increasing prevalence of anaerobic bacteremia is a major concern worldwide and requires longitudinal monitoring. Methods We present one of the largest and longest longitudinal studies on the prevalence and antimicrobial resistance of Bacteroides, Clostridium, Fusobacterium, and Prevotella spp. isolated from blood culture samples using national comprehensive surveillance data in Japan during 2011–2020 as part of the Japan Nosocomial Infections Surveillance. Results Data for 41 949 Bacteroides spp., 40 603 Clostridium spp., 7013 Fusobacterium spp., and 5428 Prevotella spp. isolates were obtained. The incidences of bacteremia caused by Bacteroides fragilis, Clostridium perfringens, and Fusobacterium nucleatum significantly increased during the period (P < .0001). Among the 20 species analyzed, 18 showed no significant changes in susceptibility over time, including B. fragilis, C perfringens, and F. nucleatum. However, resistance to clindamycin increased in B. thetaiotaomicron (P = .0312), and resistance to ampicillin increased in B. ovatus (P = .0008). Conclusions Our comprehensive national surveillance data analysis demonstrated a continuous increase in the incidence of anaerobic bacteremia, particularly in B. fragilis, C. perfringens, and F. nucleatum. This may be linked to the increasing number of colorectal cancer cases or advancing methods for species identification and susceptibility testing, requiring cautious interpretation. The discovery of an upsurge in anaerobic bacteremia and potential alterations in susceptibility highlights the necessity for more extensive studies in this field. This study using comprehensive national surveillance data demonstrated a continuous increase in the incidence of anaerobic bacteremia and potential alterations of antimicrobial susceptibility in Japan during 2011-2020, highlighting the necessity for more extensive studies in this field. anaerobe antimicrobial resistance bacteremia bloodstream infections Research Program on Emerging and Re-emerging Infectious Diseases Japan Agency for Medical Research and Development 10.13039/100009619 21fk0108604j0001 ==== Body pmcAnaerobic bacteria continue to be important causative pathogens of bacteremia, which frequently leads to severe life-threatening conditions [1]. The difficulty in isolating these bacteria leads to delayed diagnosis and treatment. Routine susceptibility testing for anaerobic bacteria should be considered for specific clinical situations and monomicrobial infections [2]. Bacteremia caused by anaerobes is a re-emerging infectious disease. The incidence of anaerobic bacteria among all bacteria from positive blood cultures varies by country: 1.6% in Italy, 4.1% in Singapore, 3% in Iowa and Burlington in the United States in 2000, and 10.4% in Minnesota in the United States in 2004 [3–6]. In the United States, the incidence of bacteremia caused by anaerobes decreased by 45% between 1974 and 1988 [7], but Lassmann et al. reported a 30% increase in 2 hospitals during 1993–2004 [5]. In Italy, a slight upward trend was noted in anaerobic blood infections between 2016 and 2020 [3]. Antibiotic resistance among anaerobic microorganisms has significantly increased in recent decades [8, 9], and the resistance rates vary widely by region. Veloo et al. reported that 9.6% of Bacteroides isolates from Kuwait and 4% from Belgium were resistant to meropenem [10]. In the B. fragilis group, resistance to penicillin occurred in 80%–90% of isolates, and resistance to amoxicillin-clavulanate rose from 0.8% in 1992% to 6.2% in 2010–2011 [9]. The most significant change in Bacteroides spp. in recent years has been an increase in resistance to clindamycin (CLDM) by up to 30%–50% [11, 12]. In vitro susceptibility testing for Bacteroides isolates reliably predicts patient response to therapy [13]. Further studies are needed to evaluate trends in incidence and antimicrobial resistance and to inform prescribing and antimicrobial stewardship strategies. In Japan, comprehensive surveillance data have been collected in a national antimicrobial resistance surveillance program—the Japan Nosocomial Infections Surveillance (JANIS)—in which all routine microbiological test results are being collected for all sample types from both symptomatic and asymptomatic patients from hundreds or thousands of participating hospitals since 2000 [14]. However, comprehensive national surveillance data have not yet been utilized for anaerobic bacteria studies, and only local surveillance in the Kansai region has been conducted for 4 months [15]. In this study, we focused on Bacteroides, Clostridium, Fusobacterium, and Prevotella spp. and evaluated their incidence rate, distribution trend, and antimicrobial susceptibility, using comprehensive national data from JANIS for the period 2011–2020. METHODS Data Preparation and Tabulation All inpatient and outpatient data from January 2011 to December 2020 were extracted from the JANIS database, comprising all routine microbiological diagnostic tests (including culture-positive and culture-negative results) and antimicrobial susceptibility testing results [14]. A total of 2167 hospitals across Japan submitted their data to the JANIS database in 2020. These included 46 of 52 (88.5%) hospitals with >900 beds, 287 of 349 (82.2%) hospitals with 500–899 beds, 1031 of 2130 (48.4%) hospitals with 200–499 beds, and 803 of 5769 (13.9%) hospitals with <200 beds. We specifically targeted Bacteroides spp., Clostridium spp., Fusobacterium spp., and Prevotella spp. due to their high crude mortality [16]. We used a Java toolkit to extract the data of isolates of Bacteroides spp. (B. fragilis, B. thetaiotaomicron, B. vulgatus, B. uniformis, B. ovatus, B. caccae, Parabacteroides distasonis, and other Bacteroides), Clostridium spp. (C. perfringens, C. septicum, and other Clostridium), Fusobacterium spp. (F. nucleatum, F. necrophorum, F. mortiferum, F. varium, and other Fusobacterium), and Prevotella spp. (P. oralis, P. meraninogenica, P. buccae, P. bivia, P. intermedia, P. denticola, P. loescheii, P. corporis, P. ruminicola, and other Prevotella), which were isolated from blood samples and subjected to antimicrobial susceptibility testing for ampicillin (ABPC), ampicillin-sulbactam (SBT/ABPC), piperacillin-tazobactam (TAZ/PIPC), CLDM, cefmetazole (CMZ), cefotaxime (CTX), imipenem (IPM), and meropenem (MEPM). The antimicrobial susceptibility testing data in JANIS comprises minimum inhibitory concentration (MIC) data. Participating hospitals in JANIS may employ various CLSI methods, but for this study, we used the CLSI 2020 criteria to interpret the MIC data for susceptibility, employing breakpoints. The data were organized using the “one isolate per patient for each species” deduplication algorithm of the World Health Organization's Global Antimicrobial Surveillance System [17, 18]. The collection of data on susceptibility testing methods, such as Walkaway, Vitek II, or Dryplate Eiken, was also implemented for each participating hospital in JANIS and subsequently incorporated into the tabulated results. Statistical Analysis The statistical significance of the differences in proportions was tested using the Pearson chi-square or Fisher exact test (when the minimum count in a contingency table was <5). To account for multiple comparisons, we separately applied the Benjamini-Hochberg false discovery rate correction [19] for each species. To maintain a false discovery rate <5% for each species, the significance threshold was established. The Cochran-Armitage trend test was used to test for any trend in the incidence (ie, number of anaerobic bacteremia cases divided by the total number of patients who underwent blood culture testing) across the years. The level of significance was set at P < .05. All statistical analyses were performed using R software (version 4.0.5) and JMP Pro (version 13; SAS Institute, Cary, NC, USA). Patient Consent Patient identifiers were de-identified by each hospital before data submission to JANIS. The anonymous data stored in the JANIS database were exported and analyzed. The protocol of this study was approved by the Ministry of Health, Labor and Welfare (approval number: 0425–3) according to Article 32 of the Statistics Act and in accordance with the Helsinki Declaration. The requirement for informed consent was waived by the Ministry of Health, Labor and Welfare (approval number: 0425–3). RESULTS Trends and Incidence of Bacteremia due to Bacteroides, Clostridium, Fusobacterium, and Prevotella Species The annual number of patients with bacteremia due to Bacteroides, Clostridium, Fusobacterium, and Prevotella spp. and the number of patients from whom blood samples were collected from January 2011 to December 2020 are shown in Table 1. A total of 13 118 386 blood samples were collected from patients during this period, with 40 841, 40 214, 6978, and 5367 patients diagnosed with Bacteroides, Clostridium, Fusobacterium, and Prevotella bacteremia, respectively. Table 1. Distribution and Trend of Bacteremia Caused by Bacteroides, Clostridium, Fusobacterium, and Prevotella spp. in Japan Between 2011 and 2020 Total 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 No. of participating hospitals 593 660 745 883 1435 1653 1795 1947 2075 2167 No. of patients who underwent blood culture testing 13 118 386 565 036 662 045 783 151 915 448 1 403 631 1 577 451 1 672 373 1 811 038 1 930 940 1 797 273 No. of Bacteroides isolates 41 949 1794 2041 2533 2844 4328 4852 5205 5874 6109 6369 No. of bacteremia episodes 40 841 1760 2007 2477 2788 4236 4720 5076 5690 5903 6184 Incidence of bacteremia (per 100 000 tested patients) 319.8 317.5 308.3 323.4 310.7 308.3 307.6 311.2 324.3 316.4 354.4 Male sex, % 56.8 59.3 59.9 57.0 58.3 58.3 57.3 56.2 57.0 54.8 54.9 Unknown sex 1176 64 59 74 106 117 151 130 156 149 170 Age, mean ± SD, y 73.4 ± 14.9 70.9 ± 15.3 70.8 ± 15.7 71.8 ± 15.5 71.7 ± 15.7 72.6 ± 14.9 73.2 ± 14.4 74.1 ± 14.9 74.3 ± 14.4 74.4 ± 14.9 75.0 ± 14.6 Unknown age 1465 19 22 24 53 197 164 225 249 263 249 No. of Clostridium isolates 40 603 1351 1670 2220 2752 4087 4929 5385 5956 5961 6292 No. of bacteremia episodes 40 214 1336 1655 2192 2725 4048 4887 5344 5899 5898 6230 Incidence of bacteremia (per 100 000 tested patients) 309.5 239.1 252.2 283.5 300.6 291.2 312.5 322.0 328.9 308.7 350.1 Male sex, % 53.9 54.6 57.8 56.0 55.8 55.7 53.0 52.9 52.7 52.8 53.6 Unknown sex 1360 46 43 76 125 167 163 178 197 181 184 Age, mean ± SD, y 79.0 ± 12.5 77.0 ± 14.0 77.3 ± 13.2 77.7 ± 12.8 78.1 ± 11.9 78.6 ± 12.7 79.1 ± 12.2 79.4 ± 12.4 79.5 ± 12.4 79.5 ± 12.3 79.9 ± 12.2 Unknown age 1615 9 20 18 52 179 190 261 291 306 289 No. of Fusobacterium isolates 7013 247 275 426 500 704 798 902 948 1055 1158 No. of bacteremia episodes 6978 247 275 422 498 704 791 897 944 1049 1151 Incidence of bacteremia (per 100 000 tested patients) 53.5 43.7 41.5 54.4 54.6 50.2 50.6 53.9 52.3 54.6 64.4 Male sex, % 65.0 69.9 65.0 67.7 63.8 64.8 64.8 66.0 65.0 64.3 62.0 Unknown sex 197 8 9 16 12 19 27 30 23 25 28 Age, mean ± SD, y 68.0 ± 18.4 65.4 ± 18.0 68.6 ± 16.6 66.8 ± 18.4 66.1 ± 18.6 68.0 ± 17.8 67.4 ± 18.0 67.6 ± 19.1 68.1 ± 18.4 69.5 ± 18.0 69.2 ± 19.0 Unknown age 183 1 0 4 3 15 19 30 35 39 37 No. of Prevotella isolates 5428 231 257 373 446 596 629 686 689 753 768 No. of bacteremia episodes 5367 229 257 367 439 588 614 677 693 745 758 Incidence of bacteremia (per 100 000 tested patients) 41.4 40.9 38.8 47.6 48.7 42.5 39.9 41.0 38.0 39.0 42.7 Male sex, % 57.7 56.2 54.7 55.0 58.3 60.1 57.7 59.3 55.4 58.3 58.6 Unknown sex 147 10 3 7 15 21 13 23 14 19 22 Age, mean ± SD, y 71.4 ± 15.8 69.8 ± 14.8 70.6 ± 15.5 70.0 ± 17.4 70.5 ± 15.8 71.6 ± 15.5 71.9 ± 15.2 71.8 ± 15.4 71.0 ± 16.7 72.7 ± 15.5 71.4 ± 15.7 Unknown age 104 2 0 0 5 6 8 12 28 28 15 The incidence of bacteremia due to Bacteroides spp., Clostridium spp., Fusobacterium spp., and Prevotella spp. was 354, 350, 64, and 43/100 000 patients, respectively, who underwent blood culture testing in 2020 (Figure 1). The incidence of bacteremia caused by Bacteroides spp., Clostridium spp., and Fusobacterium spp. significantly increased from 2011 to 2020 by 11.7% (P < .0001), 52.2% (P < .0001), and 47.4% (P < .0001), respectively. In 2016, the incidence rate of bacteremia caused by Clostridium spp. was similar to that caused by Bacteroides spp. The annual incidence from 2011 to 2020 stratified by species is shown in Figure 2, where only B. fragilis showed a 1.12-fold increase in incidence among Bacteroides spp. (P < .0001, blue in Figure 2A); C. perfringens and F. nucleatum showed a significant increase (P < .0001, blue in Figure 2B and C). Figure 1. Combined annual incidence of bacteremia caused by Bacteroides, Clostridium, Fusobacterium, and Prevotella spp. during 2011–2020. Figure 2. Distribution and individual annual incidences of bacteremia caused by Bacteroides, Clostridium, Fusobacterium, and Prevotella spp. during 2011–2020. A, Bacteroides spp., (B) Clostridium spp., (C) Fusobacterium spp., and (D) Prevotella spp. Notable Susceptibility Patterns to Antimicrobial Agents Only some of the isolates submitted to the JANIS database (32.9% [13 788/41 949] of all Bacteroides spp., 33.6% [13 658/40 603] of all Clostridium spp., 31.7% [2220/7013 isolates] of all Fusobacterium spp., and 36.6% [1985/5428 isolates] of all Prevotella spp.) were subjected to antimicrobial susceptibility testing. The susceptibility trends of Bacteroides spp., Clostridium spp., Fusobacterium spp., and Prevotella spp. are summarized in Figure 3. All Bacteroides isolates remained highly susceptible to CMZ (gray line), IPM (light blue line), MEPM (green line), SBT/ABPC (dark blue line), and TAZ/PIPC (brown line). All Clostridium isolates remained highly susceptible to ABPC (blue line), CMZ, CTX (yellow line), IPM, MEPM, SBT/ABPC, and TAZ/PIPC. The resistance rates of Prevotella spp. to ABPC, CTX, and CLDM (orange line) were higher (62.7%, 23.3%, and 29.8% in 2020) than those to the others. The resistance rates of Bacteroides, Clostridium, Fusobacterium, and Prevotella spp., except for P. ruminicola during 2011 and 2020, and the comparison between the early (2011–2015) and late (2016–2020) phases are shown in Table 2. P. ruminicola was excluded because the number of isolates subjected to antimicrobial susceptibility testing was <30. Significant increases in resistance rates were observed for B. thetaiotaomicron to CLDM and B. ovatus to ABPC (P = .0312 and P = .0008). Conversely, the resistance rates of P. distasonis to IPM and P. loescheii to ABPC (P = .0008 and P < .0001) significantly decreased. Figure 3. Individual antimicrobial resistance rates of Bacteroides, Clostridium, Fusobacterium, and Prevotella spp. during 2011–2020. A, Bacteroides spp., (B) Clostridium spp., (C) Fusobacterium spp., and (D) Prevotella spp. The following antibiotics are shown: ABPC is indicated by the blue line, CLDM is indicated by the orange line, CMZ is indicated by the gray line, CTX is indicated by the yellow line, IPM is indicated by the light blue line, MEPM is indicated by the green line, SBT/ABPC is indicated by the dark blue line, and TAZ/PIPC is indicated by the brown line. Abbreviations: ABPC, ampicillin; CLDM, clindamycin; CMZ, cefmetazole; CTX, cefotaxime; IPM, imipenem; MEPM, meropenem; SBT/ABPC, sulbactam/ampicillin; TAZ/PIPC, tazobactam/piperacillin. Table 2. Resistance Rates of Anaerobic Bacteria to Eight Antibiotics During 2011–2020 and Comparison With the Findings Obtained During 2011–2015 and 2016–2020 Total, % (No.) Early Period, % (No.) Late Period, % (No.) Species 2011–2020 2011–2015 2016–2020 Difference, % Adjusted P Value* Bacteroides fragilis ABPC 96.8 (7167/7401) 96.6 (2190/2268) 97.0 (4977/5133) +0.4 n.s. CLDM 32.7 (2557/7824) 32.1 (733/2284) 32.9 (1824/5540) +0.8 n.s. CMZ 5.6 (401/7127) 6.6 (132/2009) 5.3 (269/5118) −1.3 n.s. CTX 38.1 (758/1991) 35.9 (208/579) 39.0 (550/1412) +3.1 n.s. IPM 2.0 (152/7625) 2.6 (58/2259) 1.8 (94/5366) −0.8 n.s. MEPM 3.6 (253/7126) 3.5 (71/2056) 3.6 (182/5070) +0.1 n.s. SBT/ABPC 3.3 (242/7224) 3.6 (73/2052) 3.3 (169/5172) −0.3 n.s. TAZ/PIPC 1.5 (82/5344) 1.5 (21/1368) 1.5 (61/3976) 0 n.s. Bacteroides thetaiotaomicron ABPC 97.6 (2008/2058) 97.3 (619/636) 97.7 (1389/1422) +0.4 n.s. CLDM 51.5 (1128/2190) 46.7 (307/657) 53.6 (821/1533) +7.1 0.0312 CMZ 32.3 (652/2016) 33.9 (205/605) 31.7 (447/1411) −2.2 n.s. CTX 64.0 (311/486) 57.0 (86/151) 67.2 (225/335) +10.2 n.s. IPM 1.5 (32/2142) 1.2 (8/663) 1.6 (24/1479) +0.4 n.s. MEPM 1.8 (35/1944) 1.5 (8/551) 1.9 (27/1393) +0.4 n.s. SBT/ABPC 4.0 (80/2015) 3.3 (19/572) 4.2 (61/1443) +0.9 n.s. TAZ/PIPC 2.8 (40/1432) 2.6 (9/350) 2.9 (31/1082) +0.3 n.s. Bacteroides caccae ABPC 95.5 (299/313) 95.2 (100/105) 95.7 (199/208) +0.5 n.s. CLDM 35.5 (119/335) 32.7 (37/113) 36.9 (82/222) +3.2 n.s. CMZ 7.7 (21/272) 8.5 (8/94) 7.3 (13/178) −1.2 n.s. CTX 35.5 (27/76) N/A 39.3 (22/56) N/A N/A IPM 2.4 (8/328) 2.8 (3/109) 2.3 (5/219) −0.5 n.s. MEPM 3.4 (10/296) 4.7 (4/85) 2.8 (6/211) −1.9 n.s. SBT/ABPC 3.8 (11/289) 4.4 (4/90) 3.5 (7/199) −0.9 n.s. TAZ/PIPC 3.1 (7/223) 3.3 (2/60) 3.1 (5/163) −0.2 n.s. Bacteroides vulgatus ABPC 95.5 (779/816) 93.1 (176/189) 96.2 (603/627) +3.2 n.s. CLDM 42.4 (342/806) 44.6 (82/184) 41.8 (260/622) −2.8 n.s. CMZ 5.3 (42/797) 7.1 (13/184) 4.7 (29/613) −2.4 n.s. CTX 35.2 (62/176) 36.2 (17/47) 34.9 (45/129) −1.3 n.s. IPM 1.3 (11/816) 1.6 (3/185) 1.3 (8/628) −0.3 n.s. MEPM 0.8 (6/757) 0.6 (1/169) 0.9 (5/588) +0.3 n.s. SBT/ABPC 3.1 (24/778) 1.8 (3/166) 3.4 (21/612) +1.6 n.s. TAZ/PIPC 2.0 (12/593) 2.7 (3/112) 1.9 (9/481) −0.8 n.s. Parabacteroides distasonis ABPC 94.2 (537/570) 93.3 (194/208) 94.8 (343/362) +1.5 n.s. CLDM 39.1 (218/557) 41.3 (76/184) 38.1 (142/373) −3.2 n.s. CMZ 25.2 (130/515) 26.5 (43/162) 24.6 (87/353) −1.9 n.s. CTX 47.5 (58/122) 45.5 (20/44) 48.7 (38/78) +3.2 n.s. IPM 6.0 (36/605) 11.3 (24/213) 3.1 (12/392) −8.2 0.0008 MEPM 1.5 (8/509) 1.2 (2/166) 1.7 (4/341) +0.5 n.s. SBT/ABPC 16.1 (87/542) 21.1 (38/180) 13.5 (49/362) −7.6 n.s. TAZ/PIPC 5.8 (23/398) 6.9 (7/102) 5.4 (16/298) −1.5 n.s. Bacteroides uniformis ABPC 94.3 (549/582) 92.1 (152/165) 95.2 (397/417) +3.1 n.s. CLDM 49.7 (286/576) 43.8 (64/146) 51.6 (222/430) +7.8 n.s. CMZ 9.7 (53/549) 9.7 (14/145) 9.7 (39/404) 0 n.s. CTX 51.1 (69/135) 48.9 (22/45) 52.2 (47/90) +3.3 n.s. IPM 1.0 (6/586) 1.3 (2/156) 0.9 (4/430) −0.4 n.s. MEPM 0.9 (5/567) 1.3 (2/154) 0.7 (3/413) −0.6 n.s. SBT/ABPC 2.9 (16/556) 4.8 (7/145) 2.2 (9/411) −2.6 n.s. TAZ/PIPC 0.9 (4/428) 1.1 (1/91) 0.9 (3/337) −0.2 n.s. Bacteroides ovatus ABPC 93.3 (567/608) 87.3 (172/197) 96.1 (395/411) +8.8 .0008 CLDM 39.8 (229/576) 35.2 (64/182) 41.9 (165/394) +6.7 n.s. CMZ 23.0 (139/605) 20.2 (39/193) 24.3 (100/412) +4.1 n.s. CTX 46.7 (64/137) 46.5 (20/43) 46.8 (44/94) +0.3 n.s. IPM 1.8 (11/607) 1.0 (2/203) 2.2 (9/403) +1.2 n.s. MEPM 1.2 (7/568) 1.2 (2/166) 1.2 (5/402) 0 n.s. SBT/ABPC 3.3 (19/575) 3.3 (6/183) 3.3 (13/392) 0 n.s. TAZ/PIPC 2.6 (11/417) 0.9 (1/115) 3.3 (10/302) +2.4 n.s. Clostridium perfringens ABPC 1.8 (138/7696) 1.8 (40/2180) 1.8 (98/5516) 0 n.s. CLDM 9.8 (794/8064) 9.5 (203/2137) 10.0 (591/5927) +0.5 n.s. CMZ 0.3 (19/7344) 0.3 (6/1922) 0.2 (13/5422) −0.1 n.s. CTX 0.4 (7/1843) 0.6 (3/507) 0.3 (4/1336) −0.3 n.s. IPM 0.3 (21/7802) 0.3 (6/2096) 0.3 (15/5706) 0 n.s. MEPM 0.1 (7/7204) 0.2 (3/1901) 0.1 (4/5303) −0.1 n.s. SBT/ABPC 0.3 (24/7410) 0.2 (3/1900) 0.4 (21/5470) +0.2 n.s. TAZ/PIPC 0.1 (6/5367) 0 (0/1202) 0.1 (6/4165) +0.1 n.s. Clostridium septicum ABPC 2.5 (4/157) 1.7 (1/59) 3.1 (3/98) +1.4 n.s. CLDM 17.6 (26/148) 18.9 (10/53) 16.8 (16/95) −2.1 n.s. CMZ 0 (0/138) 0 (0/49) 0 (0/89) 0 n.s. CTX 0 (0/40) N/A N/A N/A N/A IPM 0 (0/142) 0 (0/50) 0 (0/92) 0 n.s. MEPM 0 (0/141) 0 (0/50) 0 (0/91) 0 n.s. SBT/ABPC 0 (0/144) 0 (0/49) 0 (0/95) 0 n.s. TAZ/PIPC 0 (0/97) N/A 0 (0/71) N/A N/A Fusobacterium nucleatum ABPC 3.6 (27/759) 6.2 (14/225) 2.4 (13/534) −3.8 n.s. CLDM 1.6 (12/757) 1.4 (3/221) 1.7 (9/536) +0.3 n.s. CMZ 0.4 (3/730) 1.4 (3/212) 0 (0/518) −1.4 n.s. CTX 1.0 (2/198) 1.8 (1/55) 0.7 (1/143) −1.1 n.s. IPM 0 (0/732) 0 (0/222) 0 (0/510) 0 n.s. MEPM 0.1 (1/724) 0.5 (1/219) 0 (0/505) −0.5 n.s. SBT/ABPC 0.5 (4/740) 1.4 (3/209) 0.2 (1/531) −1.2 n.s. TAZ/PIPC 0.7 (4/566) 2.2 (3/137) 0.2 (1/429) −2.0 n.s. Fuobacterium necrophorum ABPC 4.7 (11/235) 7.1 (6/84) 3.3 (5/151) −3.8 n.s. CLDM 2.8 (7/246) 3.3 (3/91) 2.6 (4/155) −0.7 n.s. CMZ 1.8 (4/221) 3.9 (3/76) 0.7 (1/145) −3.2 n.s. CTX 5.5 (3/55) N/A 0 (0/34) N/A N/A IPM 0.8 (2/242) 1.2 (1/81) 0.6 (1/161) −0.6 n.s. MEPM 0.4 (1/234) 1.1 (1/89) 0 (0/145) −1.1 n.s. SBT/ABPC 0.4 (1/251) 1.1 (1/91) 0 (0/160) −1.1 n.s. TAZ/PIPC 0.6 (1/157) 0 (0/49) 0.9 (1/109) +0.9 n.s. Fusobacterium mortiferum ABPC 20.4 (44/216) 23.3 (14/60) 19.2 (30/156) −4.1 n.s. CLDM 1.8 (4/221) 3.2 (2/63) 1.3 (2/158) −1.9 n.s. CMZ 1.5 (3/205) 1.7 (1/58) 1.4 (2/147) −0.3 n.s. CTX 14.9 (7/47) N/A 11.4 (4/35) N/A N/A IPM 1.4 (3/218) 1.6 (1/62) 1.3 (2/156) −0.3 n.s. MEPM 1.4 (3/216) 0 (0/65) 2.0 (3/151) +2.0 n.s. SBT/ABPC 2.9 (6/210) 3.4 (2/58) 2.6 (4/152) −0.8 n.s. TAZ/PIPC 4.0 (6/151) 5.6 (2/36) 3.5 (4/115) −2.1 n.s. Fusobacterium varium ABPC 42.2 (70/166) 40 (18/45) 43.0 (52/121) +3.0 n.s. CLDM 37.8 (62/164) 28.9 (13/45) 41.2 (49/119) +12.3 n.s. CMZ 3.3 (5/153) 5.1 (2/39) 2.6 (3/114) −2.5 n.s. CTX 2.4 (1/42) N/A 3.2 (1/31) N/A N/A IPM 4.2 (7/165) 2.4 (1/42) 4.9 (6/123) +2.5 n.s. MEPM 1.3 (2/152) 2.3 (1/44) 0.9 (1/108) −1.4 n.s. SBT/ABPC 2.5 (4/161) 5.0 (2/40) 1.7 (2/121) −3.3 n.s. TAZ/PIPC 3.2 (4/126) N/A 3.1 (3/97) N/A N/A Prevotella oralis ABPC 73.9 (224/276) 78.4 (87/111) 70.9 (117/165) −7.5 n.s. CLDM 32.6 (95/291) 33.0 (37/112) 32.4 (58/179) −0.6 n.s. CMZ 10.4 (25/240) 11.4 (10/88) 9.7 (15/152) −1.7 n.s. CTX 38.9 (35/90) 41.9 (13/31) 37.3 (22/59) −4.6 n.s. IPM 1.1 (3/276) 1.0 (1/102) 1.1 (2/174) +0.1 n.s. MEPM 0.4 (1/260) 1.0 (1/105) 0 (0/155) −1.0 n.s. SBT/ABPC 3.4 (9/268) 3.0 (3/99) 3.6 (6/169) +0.6 n.s. TAZ/PIPC 2.2 (4/181) 3.4 (2/58) 1.6 (2/123) −1.8 n.s. Prevotella melaninogenica ABPC 70.3 (142/202) 74.7 (68/91) 66.7 (74/111) −8.0 n.s. CLDM 29.0 (61/210) 31.9 (30/94) 26.7 (31/116) −5.2 n.s. CMZ 11.5 (21/183) 16.7 (14/84) 7.1 (7/99) −9.6 n.s. CTX 36.5 (19/52) N/A N/A N/A N/A IPM 1.4 (3/215) 2.0 (2/98) 0.9 (1/117) −1.1 n.s. MEPM 2.1 (4/191) 2.5 (2/80) 1.8 (2/111) −0.7 n.s. SBT/ABPC 3.7 (7/189) 6.5 (5/77) 1.8 (2/112) −4.7 n.s. TAZ/PIPC 1.4 (2/139) 2.0 (1/51) 1.1 (1/88) −0.9 n.s. Prevotella buccae ABPC 54.3 (120/221) 59.6 (34/57) 52.4 (86/164) −7.2 n.s. CLDM 25.6 (57/223) 29.1 (16/55) 24.4 (41/168) −4.7 n.s. CMZ 2.3 (5/215) 5.4 (3/56) 1.3 (2/159) −4.1 n.s. CTX 19.6 (9/46) N/A 19.4 (6/31) N/A N/A IPM 0.4 (1/231) 0 (0/58) 0.6 (1/173) +0.6 n.s. MEPM 1.9 (4/210) 4.3 (2/47) 1.2 (2/163) −3.1 n.s. SBT/ABPC 0.5 (1/203) 0 (0/46) 0.6 (1/157) +0.6 n.s. TAZ/PIPC 0 (0/153) N/A 0 (0/129) N/A N/A Prevotella bivia ABPC 73.4 (127/173) 63.8 (30/47) 77.0 (97/126) +13.2 n.s. CLDM 35.1 (65/185) 21.6 (11/51) 40.3 (54/134) +18.7 n.s. CMZ 2.9 (5/175) 7.0 (3/43) 1.5 (2/132) −5.5 n.s. CTX 11.9 (7/59) N/A 10.5 (4/38) N/A N/A IPM 1.2 (2/173) 4.5 (2/44) 0 (0/124) −4.5 n.s. MEPM 0 (0/156) 0 (0/33) 0 (0/124) 0 n.s. SBT/ABPC 0.6 (1/171) 2.8 (1/36) 0 (0/135) −2.8 n.s. TAZ/PIPC 0 (0/126) N/A 0 (0/103) N/A N/A Prevotella intermedia ABPC 41.0 (64/156) 40.0 (20/50) 41.5 (44/106) +1.5 n.s. CLDM 13.9 (23/166) 13.5 (7/52) 14.0 (16/114) +0.5 n.s. CMZ 0 (0/143) 0 (0/43) 0 (0/100) 0 n.s. CTX 2.7 (1/37) N/A N/A N/A N/A IPM 0 (0/164) 0 (0/52) 0 (0/112) 0 n.s. MEPM 0 (0/146) 0 (0/40) 0 (0/106) 0 n.s. SBT/ABPC 0 (0/151) 0 (0/42) 0 (0/109) 0 n.s. TAZ/PIPC 0 (0/110) N/A 0 (0/84) N/A N/A Prevotella denticola ABPC 71.9 (46/64) N/A 65.1 (28/43) N/A N/A CLDM 28.3 (17/60) N/A 27.3 (12/44) N/A N/A CMZ 1.8 (1/55) N/A 2.6 (1/39) N/A N/A CTX N/A N/A N/A N/A N/A IPM 0 (0/58) N/A 0 (0/39) N/A N/A MEPM 1.6 (1/61) N/A 2.3 (1/43) N/A N/A SBT/ABPC 1.6 (1/62) N/A 2.3 (1/43) N/A N/A TAZ/PIPC 0 (0/45) N/A 0 (0/34) N/A N/A Prevotella loescheii ABPC 78.0 (96/123) 94.4 (67/71) 55.8 (29/52) −38.6 <0.0001 CLDM 40.7 (57/140) 49.3 (36/73) 31.3 (21/67) −18.0 n.s. CMZ 11.0 (10/91) 9.8 (5/51) 12.5 (5/40) +2.7 n.s. CTX 46.2 (24/52) N/A N/A N/A N/A IPM 1.5 (2/135) 2.9 (2/68) 0 (0/67) −2.9 n.s. MEPM 1.6 (2/122) 3.0 (2/66) 0 (0/66) −3.0 n.s. SBT/ABPC 5.6 (7/126) 9.7 (6/62) 1.5 (1/64) −8.2 n.s. TAZ/PIPC 0 (0/51) N/A N/A N/A N/A Abbreviations: ABPC, ampicillin; CLDM, clindamycin; CMZ, cefmetazole; CTX, cefotaxime; IPM, imipenem; MEPM, meropenem; N/A, not analyzed because the total number of tests was <30 for the period; n.s., not significant; SBT/ABPC, sulbactam/ampicillin; TAZ/PIPC, tazobactam/piperacillin. DISCUSSION This study showed a continuous increase in the incidence of anaerobic bacteremia, particularly in B. fragilis, C. perfringens, and F. nucleatum. However, most species (18 of 20) showed no significant changes in susceptibility over the study duration, including B. fragilis, C perfringens, and F. nucleatum. Notably, resistance to CLDM increased in B. thetaiotaomicron, and resistance to ABPC increased in B. ovatus. Our results provide a comprehensive overview of the epidemiology of the 4 anaerobic species that caused bacteremia in Japan. Dorsher et al. reported that the incidence rate of anaerobic bacteremia decreased over a 15-year period in Minnesota the United States in 1991 [7]. In contrast, some studies have reported increasing bacteremia caused by anaerobic bacteria [3, 5, 20]. Consistent with these reports, in the present study, we observed an increase in the incidence of anaerobic bacteremia caused by Bacteroides, Clostridium, and Fusobacterium spp. Note that we could not show the incidence of anaerobic bacteremia in terms of the number of hospitalizations because the number of admissions is not mandatory in the voluntary-based JANIS database. Therefore, we used the number of patients who underwent blood culture tests to evaluate the incidence of anaerobic bacteremia. Table 1 demonstrates that the incidence of anaerobic bacteremia is generally higher in males than in females. Male sex was identified as a risk factor for the development of anaerobic bacteremia [21]. Notably, the occurrence of Fusobacterium spp. is particularly elevated in males, with rates in the range of 64.3%–69.9% (Table 1). Mason et al. have reported that smoking increases the abundance of Fusobacterium spp., especially F. nucleatum, in both periodontally healthy and diseased individuals [22]. In Japan, the percentage of male smokers has been considerably higher than that of female smokers (35.9% vs 13.6% in 2022) [23]. Hence, smoking habits may be associated with the incidence of bacteremia caused by Fusobacterium spp. Figure 2 illustrates an increase in bacteremia caused by B. fragilis, C. perfringens, and F. nucleatum. This trend may be attributed to the rising incidence of colorectal cancer. Kwong et al. reported significant associations between colorectal cancer and bloodstream infections caused by Streptococcus gallolyticus, B. fragilis, F. nucleatum, Peptostreptococcus spp., C. perfringens, and other anaerobic bacteria [24]. In Japan, the annual incidence of colorectal cancer continues to increase [25]. The surge in surgical procedures involving the small intestine, colon, rectum, anus, gall bladder, and pancreas may also contribute to this trend [26]. According to a report from the JANIS SSI section, in 2020 B. fragilis ranked as the third most frequent pathogen in colon surgeries and the fourth most frequent pathogen in rectal surgeries (source: https://janis.mhlw.go.jp/report/open_report/2021/3/5/SSI_Open_Report_202100.xls). Another potential factor could be advancements in species identification and susceptibility testing methods, such as the Rapid ID 32A API, the system for microorganism identification, susceptibility testing (eg, WalkAway and BD Phoenix), and matrix-assisted laser desorption/ionization time-of-flight mass spectrometry, which are commonly utilized in Japanese hospital laboratories. However, the impact of these methods could not be verified because information regarding the species identification practices at each hospital was not recorded in the JANIS system. The breakdown of 4 anaerobic bacteria causing bacteremia showed little change from 2011 to 2020, except for an increase in Fusobacterium nucleatum from 36.0% to 44.5% (P < .001) (Supplementary Figure 1). This increase may be associated with colorectal cancer and colorectal surgeries [24]. On the other hand, the breakdown (Supplementary Figure 1C) showed a relative decrease in F. necrophorum, primarily associated with nongut infections [27, 28], from 17.8% in 2011 to 13.1% in 2020 (P < .001), which corresponds to the increase in F. nucleatum primarily associated with the gut. We also analyzed the annual trends in resistance rates for various antimicrobials and compared the data from 2011–2015 with 2016–2020 (Figure 3, Table 2). Metronidazole (MNZ) was not included due to the absence of susceptibility data in the JANIS database, although MNZ injections have been used since 2014. Notably, high resistance rates of Bacteroides spp. and Prevotella spp. to ABPC were observed (>90% and >60%, respectively, in Figure 3). Eitel et al. have identified the cepA, cfxA, and cfiA genes as β-lactamase genes detected in the B. fragilis group. Tran et al. and Hashimoto et al. reported that the cfxA gene, specifically associated with resistance to ABPC, has been predominantly found in Bacteroides spp. isolates from Japanese patients [29, 30]. In contrast, carbapenems retained high activity against these 4 bacteria, with Bacteroides spp., Clostridium spp., Fusobacterium spp., and Prevotella spp. showing resistance rates of <4%, 0.5%, 1%, and 3%, respectively, to meropenem. Among the Bacteroides species, the resistance rate of B. fragilis to meropenem has been relatively high (up to 3.6%). It was reported that 9.6% of Bacteroides isolates in Kuwait and 4% in Belgium were resistant to meropenem, whereas none of the clinical isolates from Germany, Turkey, Hungary, Croatia, and the Netherlands showed resistance to carbapenems [31]. Snydman et al. reported that the resistance rate of carbapenems against B. fragilis in United States was low (1.1%–2.5%) [32]. In Canada, the resistance rate of gram-negative anaerobic bacteria to imipenem during 2012–2019 was reported to be <4% [33]. However, in China, 18.2% and 29.5% of B. fragilis isolates were found to be resistant to imipenem and meropenem, respectively, during 2009–2015 [34]. SBT/ABPC and TAZ/PIPC also showed low resistance rates against all 4 bacteria. These results are similar to those obtained in Canada, Argentina, and European countries [3, 35–38]. The resistance rate of B. thetaiotaomicron to CLDM, CTX, and CMZ was higher than that of B. fragilis (Table 2), consistent with previous reports of higher resistance to cephalosporins and CLDM in Korea, Canada, Argentina, and European countries [9, 35, 39, 40]. Kierzkowska et al. reported a higher resistance rate to CLDM in non-fragilis Bacteroides during 2013–2017 than during 2007–2012 [41]. The observed resistance to CLDM in Bacteroides, Fusobacterium, and Prevotella spp. was found to be related to the presence of ermF genes [42, 43]. This study revealed a significant increase in the resistance rate to CLDM by B. thetaiotaomicron. The resistance rate of C. perfringens to CLDM was higher (7.3%–11.6%) than to other antimicrobials. The resistance rates of Prevotella spp. to ABPC (62.8%), CLDM (29.8%), IPM, and SBT/ABPC were similar to those observed in Italy [3]. This study had several limitations. First, the JANIS surveillance system relies on voluntary participation. Thus, the number of participating hospitals can vary each year. Second, JANIS did not collect information on how species identification is conducted in each hospital. Third, JANIS has not collected strains but has rather collected data on species, specimens, and antimicrobial susceptibility reported by the participating hospitals. Fourth, drug susceptibility tests were conducted only for 30%–40% of the strains isolated in these hospitals, and there were no data on MNZ susceptibility tests. Additionally, each participating hospital used their own microbiological diagnostic and antimicrobial susceptibility testing instruments, which could introduce variability in the results. Despite the limitations of this study, it provides the largest longitudinal overviews using national surveillance data of culturing and antimicrobial susceptibilities of 41 949 Bacteroides, 40 603 Clostridium, 7013 Fusobacterium, and 5428 Prevotella isolates from >2000 hospitals collected for 10 years. The incidence of anaerobic bacteremia, particularly B. fragilis, C. perfringens, and F. nucleatum, has been continuously increasing, which may be attributed to the rising number of patients with colorectal cancer or the advancing methods for species identification and susceptibility testing, requiring cautious interpretation. The resistance rate of B. thetaiotaomicron to CLDM and CTX was significantly increased. These results could help guide empirical therapies for anaerobic bacteremia, and the findings of increased anaerobic bacteremia and possible changes in susceptibility highlight the need for further extensive and diverse studies in this field. Supplementary Material ofad334_Supplementary_Data Click here for additional data file. Acknowledgments We are grateful to all the hospitals that participated and contributed data to JANIS as well as Editage (www.editage.jp) for English language editing. Financial support. This study was supported by the Research Program on Emerging and Re-emerging Infectious Diseases from the Japan Agency for Medical Research and Development under grant number 21fk0108604j0001 to M.S. Supplementary Data Supplementary materials are available at Open Forum Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author. ==== Refs References 1 Goldstein EJ . Anaerobic bacteremia. Clin Infect Dis 1996; 23 (Suppl 1 ):S97–101.8953114 2 Brook I . Spectrum and treatment of anaerobic infections. J Infect Chemother 2016; 22 :1–13.26620376 3 Di Bella S , AntonelloRM, SansonG, et al Anaerobic bloodstream infections in Italy (ITANAEROBY): a 5-year retrospective nationwide survey. Anaerobe 2022; 75 :102583. 4 Diekema DJ , BeekmannSE, ChapinKC, MorelKA, MunsonE, DoernGV. Epidemiology and outcome of nosocomial and community-onset bloodstream infection. J Clin Microbiol 2003; 41 :3655–60.12904371 5 Lassmann B , GustafsonDR, WoodCM, RosenblattJE. Reemergence of anaerobic bacteremia. Clin Infect Dis 2007; 44 :895–900.17342637 6 Ng LSY , KwangLL, RaoS, TanTY. Anaerobic bacteraemia revisited: species and susceptibilities. Ann Acad Med Singap 2015; 44 :13–8.25703492 7 Dorsher CW , RosenblattJE, WilsonWR, IlstrupDM. Anaerobic bacteremia: decreasing rate over a 15-year period. Rev Infect Dis 1991; 13 :633–6.1925283 8 Hecht DW . Prevalence of antibiotic resistance in anaerobic bacteria: worrisome developments. Clin Infect Dis 2004; 39 :92–7.15206059 9 Karlowsky JA , WalktyAJ, AdamHJ, BaxterMR, HobanDJ, ZhanelGG. Prevalence of antimicrobial resistance among clinical isolates of Bacteroides fragilis group in Canada in 2010-2011: CANWARD surveillance study. Antimicrob Agents Chemother 2012; 56 :1247–52.22203594 10 Veloo ACM , BaasWH, HaanFJ, CocoJ, RossenJW. Prevalence of antimicrobial resistance genes in Bacteroides spp. and Prevotella spp. Dutch clinical isolates. Clin Microbiol Infect 2019; 25 :1156.e9–13. 11 Boyanova L , KolarovR, MitovI. Recent evolution of antibiotic resistance in the anaerobes as compared to previous decades. Anaerobe 2015; 31 :4–10.24875330 12 Cobo F , Rodriguez-GrangerJ, Perez-ZapataI, SampedroA, AliagaL, Navarro-MariJM. Antimicrobial susceptibility and clinical findings of significant anaerobic bacteria in southern Spain. Anaerobe 2019; 59 :49–53.31103531 13 Nguyen MH , YuVL, MorrisAJ, et al Antimicrobial resistance and clinical outcome of Bacteroides bacteremia: findings of a multicenter prospective observational trial. Clin Infect Dis 2000; 30 :870–6.10852736 14 Kajihara T , YaharaK, HirabayashiA, et al Japan nosocomial infections surveillance (JANIS): current status, international collaboration, and future directions for a comprehensive antimicrobial resistance surveillance system. Jpn J Infect Dis 2021; 74 :87–96.32863357 15 Shimura S , WatariH, KomatsuM, et al Antimicrobial susceptibility surveillance of obligate anaerobic bacteria in the Kinki area. J Infect Chemother 2019; 25 :837–44.31427200 16 Wilson JR , LimayeAP. Risk factors for mortality in patients with anaerobic bacteremia. Eur J Clin Microbiol Infect Dis 2004; 23 :310–6.15024626 17 Kajihara T , YaharaK, StellingJ, et al Comparison of de-duplication methods used by WHO Global Antimicrobial resistance Surveillance System (GLASS) and Japan Nosocomial Infections Surveillance (JANIS) in the surveillance of antimicrobial resistance. PLoS One 2020; 15 :e0228234. 18 World Health Organization . Global Antimicrobial Resistance Surveillance System (GLASS) Guide to Preparing Aggregated Antimicrobial Resistance Data Files. World Health Organization; 2016. 19 Benjamini Y , HochbergY. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc B 1995; 57 :289–300. 20 Sarvari KP , RaczNB, BurianK. Epidemiology and antibiotic susceptibility in anaerobic bacteraemia: a 15-year retrospective study in south-eastern Hungary. Infect Dis 2022; 54 :16–25. 21 Ngo JT , ParkinsMD, GregsonDB, et al Population-based assessment of the incidence, risk factors, and outcomes of anaerobic bloodstream infections. Infection 2013; 41 :41–8.23292663 22 Mason MR , PreshawPM, NagarajaHN, et al The subgingival microbiome of clinically healthy current and never smokers. ISME J 2015; 9 :268–72.25012901 23 Odani S , TabuchiT. Prevalence and denial of current tobacco product use: combustible and heated tobacco products, Japan, 2022. Prev Med Rep 2022; 30 :102031. 24 Kwong TNY , WangX, NakatsuG, et al Association between bacteremia from specific microbes and subsequent diagnosis of colorectal cancer. Gastroenterology 2018; 155 :383–8.29729257 25 Foundation for Promotion of Cancer Research . Cancer statistics in Japan. 2022. Available at: https://ganjoho.jp/public/qa_links/report/statistics/pdf/cancer_statistics_2022.pdf . Accessed March 8, 2023. 26 Marubashi S , TakahashiA, KakejiY, et al Surgical outcomes in gastroenterological surgery in Japan: report of the National Clinical Database 2011-2019. Ann Gastroenterol Surg 2021; 5 :639–58.34585049 27 Kristensen LH , PragJ. Lemierre's syndrome and other disseminated Fusobacterium necrophorum infections in Denmark: a prospective epidemiological and clinical survey. Eur J Clin Microbiol Infect Dis 2008; 27 :779–89.18330604 28 Riordan T . Human infection with Fusobacterium necrophorum (necrobacillosis), with a focus on Lemierre's syndrome. Clin Microbiol Rev 2007; 20 :622–59.17934077 29 Hashimoto T , HashinagaK, KomiyaK, HiramatsuK. Prevalence of antimicrobial resistant genes in Bacteroides spp. isolated in Oita Prefecture, Japan. J Infect Chemother 2023; 29 :284–8.36473684 30 Tran CM , TanakaK, WatanabeK. PCR-based detection of resistance genes in anaerobic bacteria isolated from intra-abdominal infections. J Infect Chemother 2013; 19 :279–90.23338012 31 Veloo ACM , TokmanHB, Jean-PierreH, et al Antimicrobial susceptibility profiles of anaerobic bacteria, isolated from human clinical specimens, within different European and surrounding countries. A joint ESGAI study. Anaerobe 2020; 61 :102111. 32 Snydman DR , JacobusNV, McDermottLA, et al Trends in antimicrobial resistance among Bacteroides species and Parabacteroides species in the United States from 2010-2012 with comparison to 2008-2009. Anaerobe 2017; 43 :21–6.27867083 33 Forbes JD , KusJV, PatelSN. Antimicrobial susceptibility profiles of invasive isolates of anaerobic bacteria from a large Canadian reference laboratory: 2012-2019. Anaerobe 2021; 70 :102386. 34 Gao Q , WuS, XuT, ZhaoX, HuangH, HuF. Emergence of carbapenem resistance in Bacteroides fragilis in China. Int J Antimicrob Agents 2019; 53 :859–63.30831235 35 Fernandez-Canigia L , LitterioM, LegariaMC, et al First national survey of antibiotic susceptibility of the Bacteroides fragilis group: emerging resistance to carbapenems in Argentina. Antimicrob Agents Chemother 2012; 56 :1309–14.22232282 36 Seifert H , DalhoffA; PRISMA Study Group. German multicentre survey of the antibiotic susceptibility of Bacteroides fragilis group and Prevotella species isolated from intra-abdominal infections: results from the PRISMA study. J Antimicrob Chemother 2010; 65 :2405–10.20851813 37 Soki J , WyboI, HajduE, et al A Europe-wide assessment of antibiotic resistance rates in Bacteroides and Parabacteroides isolates from intestinal microbiota of healthy subjects. Anaerobe 2020; 62 :102182. 38 Wybo I , Van den BosscheD, SoetensO, et al Fourth Belgian multicentre survey of antibiotic susceptibility of anaerobic bacteria. J Antimicrob Chemother 2014; 69 :155–61.24008826 39 Nagy E , UrbanE, NordCE, et al Antimicrobial susceptibility of Bacteroides fragilis group isolates in Europe: 20 years of experience. Clin Microbiol Infect 2011; 17 :371–9.20456453 40 Yim J , LeeY, KimM, et al Antimicrobial susceptibility of clinical isolates of Bacteroides fragilis group organisms recovered from 2009 to 2012 in a Korean hospital. Ann Lab Med 2015; 35 :94–8.25553287 41 Kierzkowska M , MajewskaA, MlynarczykG. Trends and impact in antimicrobial resistance among Bacteroides and Parabacteroides species in 2007-2012 compared to 2013-2017. Microb Drug Resist 2020; 26 :1452–7.32407191 42 Kierzkowska M , MajewskaA, Szymanek-MajchrzakK, et al In vitro effect of clindamycin against Bacteroides and Parabacteroides isolates in Poland. J Glob Antimicrob Resist 2018; 13 :49–52.29129778 43 Sood A , RayP, AngrupA. Anaerobic gram-negative bacteria: role as a reservoir of antibiotic resistance. Antibiotics (Basel) 2023; 12 :942.37237845
PMC010xxxxxx/PMC10352687.txt
==== Front J Audiol Otol J Audiol Otol JAO Journal of Audiology & Otology 2384-1621 2384-1710 The Korean Audiological Society and Korean Otological Society 37340774 10.7874/jao.2022.00542 jao-2022-00542 Original Article Recovery From Otitis Media and Associated Factors Among 1- to 6-Year-Old Children in South India: A Longitudinal Study http://orcid.org/0000-0003-4978-796X Harinath Sathya 1 http://orcid.org/0000-0002-6884-3632 Lakshmanan Somu 2 http://orcid.org/0000-0002-1008-9403 James Saji 3 http://orcid.org/0000-0003-0162-6702 Maruthy Sandeep 4 1 Department of Audiology, Sri Ramachandra Faculty of Audiology and Speech Language Pathology, Sri Ramachandra Institute of Higher Education and Research (DU), Chennai, India 2 Department of ENT, Head and Neck Surgery, Sri Ramachandra Institute of Higher Education and Research (DU), Chennai, India 3 Department of Pediatrics, Sri Ramachandra Institute of Higher Education and Research (DU), Chennai, India 4 Department of Audiology, All India Institute of Speech and Hearing, Mysuru, India Address for correspondence Sathya Harinath, MSc Department of Audiology, Sri Ramachandra Faculty of Audiology and Speech Language Pathology, Sri Ramachandra Institute of Higher Education and Research (DU), Chennai - 600116, Tamil Nadu, India Tel +91-044-24768027 Fax +91-044-24767008 E-mail [email protected] 7 2023 22 6 2023 27 3 139144 7 12 2022 22 1 2023 24 2 2023 Copyright © 2023 The Korean Audiological Society and Korean Otological Society 2023 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. Background and Objectives This study was aimed at assessing recovery from otitis media (OM) and variables associated with it among 1- to 6-year-old children. Subjects and Methods We assessed 87 children with OM otologically and audiologically. Medicines were prescribed, and medication compliance was ensured. The children were followed up after 3 months to judge the status of OM as resolved or recurrent. Data were statistically analyzed to derive the risk of recurrence of OM with effusion (OME) and acute OM by degree of hearing loss, type of tympanogram, age group, and sex. Results The overall recurrence rate was 26%. The risk of recurrence was higher for OME (odds ratio [OR]=4.33; 95% confidence interval [CI]: 1.90 to 9.83); at AC auditory brainstem peak V responses up to 40 dBnHL (OR=5.20; 95% CI: 2.05 to 13), 50 dBnHL (OR=3.47; 95% CI: 0.5 to 23), and 60 dBnHL (OR=16.09; 95% CI: 4.36 to 1.2); in B (OR= 3.16; 95% CI: 1.36 to 7.33) and C tympanograms (OR=2.83; 95% CI: 0.70 to 11.41); and in the age group of 5-6 years (OR=8, 95% CI: 2.23 to 28). The risk of recurrence of OM did not differ between male and female patients. Conclusions The rate of recurrence was comparable to or lower than that reported in the pediatric population of other countries. The findings suggest that children with OME, severe pathology, or age of 5-6 years require more attention and frequent monitoring to minimize the risk of recurrence. Recurrence Recovery Otitis media Tympanogram Conductive hearing loss ==== Body pmcIntroduction Otitis media (OM) is the most common pathology seen in children, next only to the common cold. The global annual incidence of acute OM is 10.85% and half of them are reported to occur under the age of 5 years [1]. OM is known to show bimodal peak prevalence; one peak before 2 years of age and the other peak at 5 to 7 years of age [2,3]. If left untreated, acute OM can result in chronic suppurative OM (CSOM), mastoiditis, labyrinthitis, petrositis, facial nerve paralysis, meningitis, subdural abscess, extradural abscess, and thrombophlebitis [4]. OM normally does not occur just once. In fact, multiple episodes of the disease are quite common. Klein, et al. [5] reported that 33% of all their subjects had three or more episodes of OM by 3 years, 24% by 2 years, and 17% by 18 months. The ratio of unilateral to bilateral OM is 2:3 [6]. To understand the nature of acute OM, Mandel, et al. [7] followed 148 children (aged 1 to 8 years) by weekly otoscopy for a period of 6 months. They found that OM peaks in December and March, and the duration of a new episode was short. Gibney, et al. [8] followed 31 Aboriginal children (aged less than 8 years) with acute OM for 3 weeks or longer till the infection resolved. Results showed that 70% of the children had persistent sign of OM [8]. In a study in the Oxford area, 95 full-term infants were tracked for their tympanometry findings every month for 3 years. They found that children susceptible to OM with effusion tend to have more episodes of effusion rather than increased duration of effusion [9]. In general, most children who develop OM experience the disease for more than 2 months before spontaneous resolution [10]. Even if acute OM persists for a short duration, the associated effusion in the middle ear may persist for weeks or even months, often without clinical signs. Roland, et al. [11] reported the median number of days before the resolution of OM to be 72 days. In up to 30% of the children with acute OM, fluid remains in the ear for 3 to 12 months. The important predictors of outcomes in OM are age, severity of the disease, and nasopharyngeal colonization patterns [12]. Children whose symptoms failed to improve early in the course of the disease were the ones who were younger and had more severe disease. Colonization with Streptococcus pneumoniae was associated with more severe OM than that with other pathogens like Moraxella catarrhalis and Haemophilus influenzae. A high incidence of OM is reported among the Indian population [13]. Jacob, et al. [14] found OM in 17.6% of the 284 children aged 6 to 10 years. Specifically, the incidence of OM with effusion was found to be 3.06%, while the incidence of acute OM was 0.65%. The exact reason for high incidence of OM in the Indian population is not very clear. However, genetic factors, difference in the Eustachian tube, poor socioeconomic standards, poor nutrition, and lack of health education have been speculated as the contributing factors [15,16]. According to Beery, et al. [16], though Eustachian tubes in the Indian population allow better ventilation of the middle ear cavity, they have poorer protective function making middle ear an easy target for bacterial invasion from the throat. Irrespective of the reasons, the high incidence of OM deserves serious consideration in Indian children. Dhingra [15] even reported a significant difference between rural (46 per thousand) and urban populations (16 per thousand) in the prevalence of OM in India. Although OM is a highly prevalent pathology in children of India, the nature of OM and the course of recovery is not yet explored in the country. Considering that the genetic, sociocultural, environmental, and economic factors in India are different compared to the developed countries, one can expect that the course of recovery is different in the cohort here compared to that in the developed countries. The recovery with reference to infection as well as hearing status needs exploration and such exploration warrants longitudinal follow-up of the subjects. American Academy of Paediatrics (AAP) [17] recommends that children with OM should be re-examined at 3 or 6 months until the effusion is no longer present and should identify children with risk of hearing impairment or other complications. However, no such time schedule or protocol for follow-up is in practice in India. Hence, the present study longitudinally followed young children diagnosed with OM up to 3 months of diagnosis. The follow-up examinations were done to track the status of the infection, findings in tympanometry, and the hearing sensitivity. The study attempted to derive the relationship between risk of OM and the characteristics of the subject, otological findings, tympanometric findings, and hearing sensitivity. The aim of this study was to eventually make recommendations for follow-up protocol in cases of OM. Subjects and Methods The study examined children with OM during their first episode and after 3 months to understand the nature of recurrence of infection and its effects on hearing sensitivity. The Institutional Ethics Committee of Ramachandra Institute of Higher Education and Research where the study was conducted had approved the study (Ref: IEC-NI/16/JUL/54/48). Written informed consent was obtained form all the parents of children for evaluation. Screening for OM The study population was recruited from outpatient units of ENT, Pediatrics, and Audiology of Sri Ramachandra Hospital. Children in the age range of 1 to 6 years were screened for their motor development, speech-language development, and ear infections. Children with normal development but showing clinical signs and symptoms suggestive of OM were the potential participants of the study. Distorted or missing cone of light, air bubbles in the middle ear, fluid in middle ear, and dull and bulging tympanic membrane were considered the signs of OM [18]. Children with chronic suppurative OM, congenital or late onset sensorineural hearing impairment, anomalies of the external ear, congenital conductive hearing impairment, developmental delay, cerebral palsy, autism spectrum disorders, mental subnormality, genetic syndrome, cleft lip and palate, history of surgical intervention for OM with effusion were excluded from the study. The otorhinolaryngologist examined each ear otoscopically to check for any obstruction in the ear canal. Children with wax in the ear canal were sent for wax removal. Based on the specific clinical signs observed, otorhinolaryngologist diagnosed the presence and type of OM in that ear. Test procedure Baseline evaluation An experienced otorhinolaryngologist visually inspected each ear of the child using a microscope. The observed otoscopic signs were noted down and a corresponding diagnosis was made. The child was then evaluated by an experienced (more than 15 years) audiologist to determine the middle ear status and hearing sensitivity. Middle ear status was assessed using tympanometry and acoustic reflexes. A calibrated GSI 39 immittance meter (Grason-Stadler, Eden Prairie, MN, USA) was used for the purpose. A probe tone of 226 Hz was used to derive the admittance while pressure in the ear canal was swept from +200 daPa to -400 daPa. The resultant tympanogram was noted down for its peak static admittance, peak pressure, gradient, equivalent ear canal volume, and the type of tympanogram. Tympanograms were classified as per the criteria given by Feldman [19]. The hearing thresholds were derived by tracking thresholds of auditory brainstem responses (ABRs) in air conduction (AC-ABR) and bone conduction (BC-ABR) modalities. ABRs were recorded using Neuro-Audio AEP equipment (version 10, Neurosoft, Ivanovo, Russia). The children were tested within 2 days of the identification of OM and the thresholds were tracked in ears with OM. If found necessary, sedative drug was given to make the child sleep. The electrode sites were FPz (positive), ipsilateral mastoid (negative), and contralateral mastoid (ground). Click-evoked ABR was recorded as per the stimulus and acquisition parameters given by Katz, et al. [20]. The audiologist visually inspected the recorded waveform to mark wave I, III, and V. ABR threshold was defined as the lowest intensity at which wave V was recordable. If peak V was present at 20 dBnHL, it was considered as hearing sensitivity within normal limits [21]. BC-ABR was meant to ensure that hearing loss, if any, is of conductive type, and there is no sensorineural hearing loss. To record BC-ABR, clicks were presented through a bone vibrator B 71 placed on the forehead. In order to elicit ear-specific BC-ABR, the non-test ear was masked by delivering broadband noise at 50 dBSPL through TDH-39 headphone. All the other stimulus and acquisition parameters remained as that of AC-ABR. The method of deriving BC-ABR threshold and BC hearing sensitivity was also same as that of AC-ABR. The treatment of OM All the children who completed audiological evaluation were medically treated by the Otorhinolaryngologist. Medication (antibiotics, antihistamines, and decongestants) was prescribed for 5 to 7 days and its dosage depended on the severity of infection. The parents were counseled regarding the risk factors of OM. The children were monitored telephonically for consumption of the medicines as per the prescription. All the children strictly complied with the prescribed treatment. Follow-up testing The participants were re-evaluated at the end of 3 months after their first episode. Otoscopy, immittance evaluation, and ABR (AC & BC) were repeated during the follow-up evaluation. Based on the results of otoscopy and impedance audiometry at 3 months follow-up, the middle ear status was diagnosed as either “recurrent” or “resolved.” It was considered “resolved,” if the tests revealed normal findings. Otherwise, it was considered “recurrent.” In the present study, recurrent OM was operationally defined as an episode of OM after 3 months of their initial episode [22]. A total of 1,040 children were screened, of whom, 130 children were found to have unilateral/bilateral acute OM or OM with effusion (OME). However, the parents of only 114 children gave informed consent for further evaluation. In all these 114 children, it was the first episode of OM. On testing with AC-ABR, 1 child was found to have auditory neuropathy spectrum disorder and therefore was excluded from the study. In 3 children, BC-ABR thresholds were elevated, indicating the presence of sensorineural hearing loss. They were also excluded from the study. Of the 110 children, 23 did not turnup for the follow-up evaluation, resulting in 87 children who completed the entire study protocol. Analysis The data distribution was tested using Shapiro-Wilk test of normality. Owing to non-normal distribution, non-parametric tests were used for statistical analysis. Odds ratio with 95% confidence interval was derived to find the risk of recurrence of OM and chi-square test was used to assess the significance of difference in recurrence of OM across age groups, sex, types of middle ear pathologies, types of tympanogram, and degree of conductive hearing loss. Results Of the 110 children enrolled in the study, 23 children did not come for follow-up evaluation mainly due to the distance of travel to the hospital. Totally 87 children with data of type of OM, tympanogram, and AC-ABR threshold at two points constitute the participants of the present study. Of the 87 children, 75 children had bilateral OM, 11 children had unilateral OM, and 1 child had acute otitis media (AOM) on one ear and chronic otitis media (COM) on the other ear. Table 1 shows the impact of the type of OM, degree of conductive hearing loss, type of tympanogram, age, and sex on the recurrence of OM. In ears with OME, the recurrence rate was 38% (rounded off to the nearest whole number), whereas it was 12% in acute OM. Results showed a significantly higher risk of recurrence in OME compared to acute OM (p<0.0001). When examining the relationship between the degree of conductive hearing loss and OM recurrence, the risk of recurrence was significantly lower (p=0.005) for AC-ABR threshold was 20 and 30 dBnHL compared to 40, 50, and 60 dBnHL. Similarly, in the relationship between the type of tympanogram and OM recurrence, the results showed a significantly higher risk of recurrence in ears with B and Cs type tympanograms compared to C and As type tympanograms (p=0.006). Regarding the child’s age and OM recurrence, the results showed a significantly higher risk of recurrence of OM in the 5 to 6-year age groups compared to other groups (p=0.005). However, in terms of the child’s sex and the recurrence of OM, there was no significant difference in the risk of recurrence between the two sex (p=0.155). Discussion In the current study, the participants were assessed twice with an interval of 3 months. It was found that 26% of the cases had OM on both occasions. This was true in spite of all of them pursuing the prescribed medical treatment without fail. Whether it is termed as persistence of OM or recurrence of OM is debatable. However, it is operationally referred to as “recurrence” in the current study, in line with the earlier studies [22]. Recurrence rate of 26% found in the current study is lower than that reported in pediatric population of Taiwan (33% during a 1 year period) [23], Finland (28% during 1 year period) [24], and higher than indigenous children of Australia (18%). Across various earlier studies conducted in the other countries, the recurrence varies from 9% to 73% [8,22-26], attributable to differences in compliance with treatment, socioeconomic status, climatic conditions, and genetic factors. Compared to other countries higher incidence of OM has been reported among the Indian population [14,27]. Yet, the recurrence of OM appears comparable to or even lower than some of the other countries. In the current study, we found that the risk of recurrence was significantly higher in ears with OME compared to that with AOM. The risk of recurrence of OME found in the current study (38%) was comparable to that reported by some of the earlier studies (35%) [28,29] and lower than that (50%) reported by Zielhuis, et al. [30]. The auditory deprivation caused by hearing loss secondary to OM is shown to result in deficits in cochlear and neural structures [31,32], deviations in auditory brainstem responses [33], and poor speech in noise perception [34,35]. The higher recurrence rate in ears with OME suggests that this group is at a greater risk of auditory deprivation during the developmental age. Therefore, ears with OME need greater attention and closer supervision in terms of more frequent follow-ups to minimize the recurrence compared to those with AOM. The risk of recurrence was found to be higher in ears with higher degree of hearing loss and in ears with B or Cs type tympanogram. Higher degree of hearing loss and, B or Cs type tympanogram are indications of greater damage to middle ear, in turn suggestive of more severe pathology. This hints at the direct association between severity of OM and its probability of recurrence. It also reflects the importance of audiological test findings in predicting the prognosis and planning the course of management in cases with OM. Ears with higher degree of hearing loss and ears with B or Cs type tympanogram need closer supervision and more frequent follow-ups than the other cases with OM. Earlier studies had revealed the characteristics of hearing loss that result from OM [36,37]. However, the current study is the first one to show the relationship between audiological findings and recurrence of OM. The current study also assessed the association between demographic variables (age and sex) and the risk of recurrence of OM. Results showed higher risk of recurrence in 5 to 6-year-old group compared to the younger age group. Although the exact reason for the finding is not clear, we suspect that it is because of the higher incidence of other inflammations such as tonsillitis or adenoiditis in this group [38]. Some of the earlier studies have revealed the recurrence of OM in children up to 12 years of age [3,39] with maximum prevalence at 5 years of age. The comparison between males and females revealed no significant difference in the recurrence of OM. The incidence and prevalence of OM are shown to vary between the two sex but the risk of recurrence appears to be comparable. Neurosoft, Russia for equipment support Table 1. Effect of type of OM, degree of conductive hearing loss, type of tympanogram, age, and sex on recurrence of otitis media Parameter No. of ears OR (95% CI) p Total Recurrent OM OM resolved Type of OM <0.0001  AOM 74 9 65 Reference  OME 88 33 55 4.33 (1.90 to 9.83) AC-ABR threshold (dBnHL) 0.005  20 46 2 44 0.24 (0.04 to 1.13)  30 62 10 52 Reference  40 38 19 19 5.20 (2.05 to 13.16)  50 5 2 3 3.47 (0.51 to 23.47)  60 11 9 2 16.09 (4.38 to 124.00) Type of tympanogram 0.006  As 9 0 9 -  B 81 29 52 3.16 (1.36 to 7.33)  C 60 9 51 Reference  Cs 12 4 8 2.83 (0.70 to 11.41) Age (yrs) 0.005  1 to 2 35 3 32 Reference  >2 to 3 13 3 10 3.20 (0.55 to 18.42)  >3 to 4 23 1 22 0.48 (0.04 to 4.97)  >4 to 5 21 5 16 3.33 (0.70 to 15.73)  >5 to 6 70 30 40 8.00 (2.23 to 28.61) Sex 0.155  Male 92 19 73 Reference  Female 70 23 47 1.88 (0.92 to 3.82) OM, otitis media; OR, odds ratio; CI, confidence interval; AOM, acute otitis media; OME, otitis media with effusion; AC-ABR, auditory brainstem response in air conduction Conflicts of Interest The authors have no financial conflicts of interest. Author Contributions Conceptualization: all authors. Data curation: Sathya Harinath. Formal analysis: Sathya Harinath. Investigation: Sathya Harinath. Methodology: all authors. Supervision: Somu Lakshmanan, Saji James, Sandeep Maruthy. Visualization: Somu Lakshmanan, Saji James, Sandeep Maruthy. Writing—original draft: Sathya Harinath. Writing—review & editing: Somu Lakshmanan, Saji James, Sandeep Maruthy. Approval of final manuscript: all authors. ==== Refs REFERENCES 1 Monasta L Ronfani L Marchetti F Montico M Vecchi Brumatti L Bavcar A Burden of disease caused by otitis media: systematic review and global estimates PLoS One 2012 7 e36226 22558393 2 Zielhuis GA Rach GH van den Bosch A van den Broek P The prevalence of otitis media with effusion: a critical review of the literature Clin Otolaryngol Allied Sci 1990 15 283 8 2203566 3 Zhang Q Wei J Xu M Zhang Q Zhang X Zhang Z Prevalence of otitis media with effusion among children in Xi’an, China: a randomized survey in China’s mainland Ann Otol Rhinol Laryngol 2011 120 617 21 22032077 4 Vergison A Dagan R Arguedas A Bonhoeffer J Cohen R Dhooge I Otitis media and its consequences: beyond the earache Lancet Infect Dis 2010 10 195 203 20185098 5 Klein JO Teele DW Mannos R Menyuk P Rosner BA Otitis media with effusion during the first three years of life and development of speech and language Recent Advances in Otitis Media with Effusion Lim DJ Toronto: B.C Decker Inc 1984 332 4 6 Engel J Anteunis L Volovics A Hendriks J Marres E Risk factors of otitis media with effusion during infancy Int J Pediatr Otorhinolaryngol 1999 48 239 49 10402121 7 Mandel EM Doyle WJ Winther B Alper CM The incidence, prevalence and burden of OM in unselected children aged 1-8 years followed by weekly otoscopy through the “common cold” season Int J Pediatr Otorhinolaryngol 2008 72 491 9 18272237 8 Gibney KB Morris PS Carapetis JR Skull SA Smith-Vaughan HC Stubbs E The clinical course of acute otitis media in high-risk Australian Aboriginal children: a longitudinal study BMC Pediatr 2005 5 16 15955251 9 Hogan SC Stratford KJ Moore DR Duration and recurrence of otitis media with effusion in children from birth to 3 years: prospective study using monthly otoscopy and tympanometry BMJ 1997 314 350 3 9040332 10 Zielhuis GA Rach GH van den Broek P The natural course of otitis media with effusion in preschool children Eur Arch Otorhinolaryngol 1990 247 215 21 2375863 11 Roland PS Finitzo T Friel-Patti S Brown KC Stephens KT Brown O Otitis media. Incidence, duration, and hearing status Arch Otolaryngol Head Neck Surg 1989 115 1049 53 2765220 12 Hotomi M Yamanaka N Samukawa T Suzumot M Sakai A Shimada J Treatment and outcome of severe and non-severe acute otitis media Eur J Pediatr 2005 164 3 8 15549378 13 McShane D Mitchell J Middle ear disease, hearing loss and educational problems of American Indian children J Am Indian Educ 1979 19 7 11 14 Jacob A Rupa V Job A Joseph A Hearing impairment and otitis media in a rural primary school in South India Int J Pediatr Otorhinolaryngol 1997 39 133 8 9104621 15 Dhingra PL Diseases of Ear, Nose and Throat 2nd ed New Delhi Elsevier 1998 16 Beery QC Doyle WJ Cantekin EI Bluestone CD Wiet RJ Eustachian tube function in an American Indian population Ann Otol Rhinol Laryngol Suppl 1980 89 3 Pt 2 28 33 17 Harlor AD Jr Bower C Committee on Practice and Ambulatory Medicine Section on Otolaryngology-Head and Neck Surgery Hearing assessment in infants and children: recommendations beyond neonatal screening Pediatrics 2009 124 1252 63 19786460 18 Isaacson G Otoscopic diagnosis of otitis media Minerva Pediatr 2016 68 470 7 27196119 19 Feldman AS Tympanometry: application and interpretation Ann Otol Rhinol Laryngol 1976 85 2 Suppl 25 Pt 2 202 8 1267349 20 Katz J Chasin M English K Hood LJ Tillery KL Handbook of Clinical Audiology 7th ed Philadelphia, PA Wolters Kluwer 2015 21 Roeser RJ Valente M Hosford-Dunn H Audiology Diagnosis 2nd ed New York Thieme 2007 22 Damoiseaux RA Rovers MM Van Balen FA Hoes AW de Melker RA Long-term prognosis of acute otitis media in infancy: determinants of recurrent acute otitis media and persistent middle ear effusion Fam Pract 2006 23 40 5 16107490 23 Wang PC Chang YH Chuang LJ Su HF Li CY Incidence and recurrence of acute otitis media in Taiwan’s pediatric population Clinics (Sao Paulo) 2011 66 395 9 21552661 24 Pukander J Karma P Sipilä M Occurrence and recurrence of acute otitis media among children Acta Otolaryngol 1982 94 479 86 7180421 25 Marchisio P Cantarutti L Sturkenboom M Girotto S Picelli G Dona D Burden of acute otitis media in primary care pediatrics in Italy: a secondary data analysis from the Pedianet database BMC Pediatr 2012 12 185 23190626 26 Liese JG Silfverdal SA Giaquinto C Carmona A Larcombe JH Garcia-Sicilia J Incidence and clinical presentation of acute otitis media in children aged <6 years in European medical practices Epidemiol Infect 2014 142 1778 88 24330917 27 Chadha SK Sayal A Malhotra V Agarwal AK Prevalence of preventable ear disorders in over 15,000 schoolchildren in northern India J Laryngol Otol 2013 127 28 32 23218257 28 Triglia JM Roman S Nicollasi R [Serous otitis media] J Pediatrie Pueric 2004 17 83 100 French 29 Fiellau-Nikolajsen M Tympanometry in three-year-old children: prevalence and spontaneous course of MEE Ann Otol Rhinol Laryngol Suppl 1980 89 3 Pt 2 223 7 6778312 30 Zielhuis GA Rach GH van den Broek P Screening for otitis media with effusion in preschool children Lancet 1989 1 311 4 2563465 31 Papp Z Rezes S Jókay I Sziklai I Sensorineural hearing loss in chronic otitis media Otol Neurotol 2003 24 141 4 12621323 32 Webster DB Webster M Effects of neonatal conductive hearing loss on brain stem auditory nuclei Ann Otol Rhinol Laryngol 1979 88 5 Pt 1 684 8 496200 33 Maruthy S Mannarukrishnaiah J Effect of early onset otitis media on brainstem and cortical auditory processing Behav Brain Funct 2008 4 17 18384677 34 Sandeep M Jayaram M Effect of early otitis media on speech identification Australian and New Zealand Journal of Audiology 2008 30 38 49 35 Tomlin D Rance G Long-term hearing deficits after childhood middle ear disease Ear Hear 2014 35 e233 42 25127328 36 Jayaram M WHO project report 1. Assessment of etiological factors of conductive hearing loss Mysore All India Institute of Speech and Hearing 2007 37 Parsram K Jalvi R NIHH-WHO project. Assessment of etiological factors for conductive hearing impairment in general population in Mumbai (all age groups) Mumbai Ali Yavar Jung National Institute of Speech & Hearing Disabilities 2007 38 Marseglia GL Poddighe D Caimmi D Marseglia A Caimmi S Ciprandi G Role of adenoids and adenoiditis in children with allergy and otitis media Curr Allergy Asthma Rep 2009 9 460 4 19814919 39 Mark A Matharu V Dowswell G Smith M The point prevalence of otitis media with effusion in secondary school children in Pokhara, Nepal: a cross-sectional study Int J Pediatr Otorhinolaryngol 2013 77 1523 9 23899700
PMC010xxxxxx/PMC10352688.txt
==== Front J Audiol Otol J Audiol Otol JAO Journal of Audiology & Otology 2384-1621 2384-1710 The Korean Audiological Society and Korean Otological Society 37340775 10.7874/jao.2022.00521 jao-2022-00521 Original Article Validation of the Sinhala Version of Tinnitus Handicap Inventory http://orcid.org/0000-0003-3962-0592 Rodrigo Asiri 1 http://orcid.org/0000-0002-1398-7831 Abayabandara-Herath Thilini 2 1 Victorian Institute of Forensic Mental Health, Thomas Embling Hospital, Melbourne, Australia 2 Faculty of Medicine, University of Kelaniya, Ragama, Sri Lanka Address for correspondence Asiri Rodrigo, MD, PhD Victorian Institute of Forensic Mental Health, Thomas Embling Hospital, Melbourne, Australia E-mail [email protected] 7 2023 22 6 2023 27 3 128132 29 11 2022 19 1 2023 3 2 2023 Copyright © 2023 The Korean Audiological Society and Korean Otological Society 2023 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. Background and Objectives Tinnitus is a common and disabling condition that largely remains undertreated in Sri Lanka. Currently, standardized tools that assess and monitor the treatment of tinnitus or the distress it causes are unavailable in either of the two main vernacular languages prevalent in Sri Lanka. The Tinnitus Handicap Inventory (THI) is used internationally to measure tinnitus-induced distress and to monitor treatment efficacy. In this study, we validated the Sinhala version of the THI (THI-Sin). Subjects and Methods The THI was translated into Sinhala and back translated into English and finalized by independent translators. The THI-Sin questionnaire and the 12-item General Health Questionnaire (GHQ-12) and Visual Analog Scale of tinnitus annoyance (VAS) were administered to 122 adults who visited the otolaryngology clinic of Colombo North Teaching Hospital, Ragama, Sri Lanka Results THI-Sin scores showed satisfactory internal consistency (Cronbach’s α=0.902) and were significantly correlated with the GHQ-12 and VAS scores. Factor analysis of the THI-Sin confirmed a three-factorial structure, which did not correspond to the original THI subscales. Conclusions We observed significant reliability and validity of the THI-Sin tool for evaluation of tinnitus-induced handicaps among the Sinhalese-speaking population of Sri Lanka. Tinnitus Tinnitus Handicap Inventory Validation Sinhala Sri Lanka ==== Body pmcIntroduction Tinnitus is a perception of noise or ringing in the ear(s), which affects 10%–20% of the adult population during their lifetime [1]. Tinnitus can seriously affect role functioning and quality of life [1]. It is often associated with psychological conditions such as anxiety and depression [2]. Comorbidities will add to the burden of tinnitus and further exacerbate tinnitus; hence, the association is bidirectional [3-5]. Severe annoyance associated with tinnitus is linked to increased suicide [2]. While prevalence of tinnitus in Sri Lanka is not known, it is estimated that least 2 million Sri Lankans experience tinnitus during their lifetime [6]. A Sri Lankan study found that 82.5% reported functional impairment due to tinnitus and 61.5% were found to have depression [6]. Tinnitus remains an undertreated condition in Sri Lanka in spite of its high prevalence, and associated impairment at least in part due to failure to identify the impact of tinnitus [6]. Papitsi, et al. [7] reported that questionnaires are the easiest and most effective tool for quantifying the impact of tinnitus in day-to-day life. Tinnitus Handicap Inventory (THI), is a widely used questionnaire to measure distress caused by tinnitus and efficacy of its treatment [8-13]. The THI has high internal consistency, test-retest reliability, and construct, concurrent and discriminant validity [14]. THI has been translated into various languages and it has demonstrated adequate reliability and validity in all these translations. There are currently no standardized questionnaires that have been validated in Sri Lanka for the assessment of tinnitus or the distress it causes. There are two main vernaculars in Sri Lanka—Sinhala and Tamil. This study aimed to validate the THI among a Sinhala-speaking study population in Sri Lanka. A validated Sinhala version of the tool would enable Sri Lankan clinicians to measure severity of and impairment of tinnitus patients and assess treatment success. This would also assist Sri Lankan researchers to study this area which has very little local literature. Subjects and Methods Ethical consideration Permission to validate THI in Sri Lanka was obtained from the original author of the THI, Dr. Newman. Ethics approval to conduct this study was granted by the Ethics Review Committee of the Faculty of Medicine, University of Kelaniya, Sri Lanka (P/239/09/2017). Informed consent was obtained from all the participants. Translation of the THI, English into Sinhala The process of translation followed the steps recommended by Hall, et al. [15]. These steps included preparation, translating the source language into the target language, translating the target language back into the source language, committee review, field testing, reviewing, and finalizing the translation. The original English version of THI was translated independently into the Sinhala language by three bilingual health professionals. The three translators made a pooled version from the translations. A professional translator blinded to the original questionnaire back translated the Sinhala THI to English. The original questionnaire and the back translation were compared for coherence and necessary adjustments were made. The draft Sinhala translation was provided to 20 audiology students of the Faculty of Medicine, University of Kelaniya, Sri Lanka, who speak Sinhala as their native language. They were asked to comment on the coherence of the questioning and further improvements. Their comments were used in finalizing the Sinhalese version of THI (THI-Sin). All changes made to the Sinhala version even at the later stages were also back translated to English and assessed for coherence. Participants Finalized Sinhala THI was administered to 122 adult patients conveniently selected from the outpatient otolaryngology clinic at the Colombo North Teaching Hospital, one of the biggest tertiary care hospitals in Sri Lanka with 1,442 beds. Clinic attendees with complaints of chronic unilateral or bilateral tinnitus lasting for at least 6 months were invited for the study and written consent obtained from all participants. Ten conveniently selected participants were asked to comment on the coherence of the questionnaire and further ways to improve the questions. Study measures Study participants were invited to complete the THI-Sin and 12 items of the General Health Questionnaire (GHQ-12) as well as Visual Analog Scale (VAS) of self-perceived tinnitus annoyance. Questionnaires were distributed among the participating patients after obtaining informed consent. Questionnaires were collected on the same day. THI The THI is a self-report measure with 25 items to assess tinnitus-related functional impairment (11 items), catastrophic thinking (five items), and emotional responses (nine items). Each question can be answered “yes” (4 points), “sometimes” (2 points), or “no” (0 points), with a worst possible total of 100 points. GHQ-12 The GHQ-12 is a commonly used measure to identify potential non-psychotic mental health problems. The GHQ-12 is used to assess short-term psychological disorders/distress in the community or general hospital settings [16]. The GHQ-12 has high internal consistency; test-retest reliability; and construct, concurrent, and discriminant validity. VAS The VAS is a valid and effective measure of tinnitus severity with self-perceived tinnitus annoyance [17]. This scale consisted of a horizontal 5 cm line with marked endpoints designated as not annoying and extremely annoying which corresponded with scores of 0 and 5, respectively. Statistical analysis Statistical Package for the Social Sciences (SPSS) version 22 (IBM Corp., Armonk, NY, USA) was used to analyze data. Item characteristics were explored with frequency, mean and standard deviation (SD) of the item as well as item-total correlation while means scale of the item-item correlations were computed to determine scale characteristics. Internal consistency and reliability of the THI was assessed using Cronbach’s alpha. Criterion validity was calculated by using a person’s correlation coefficients. Principal component analysis with varimax rotation was used to validate the THI-Sin factor structure. A factor analysis was performed adopting the principal component factor analysis with varimax rotation to examine whether the data could verify three subscales in the original version of THI proposed by Newman, et al. [18]. Results Of the 122 patients studied, the majority were males (66, 54.1%). The age of participants ranged from 18 to 83 years with average of 53.8 years (SD, ±16.15). Mean total THI score was 50.9 (SD, ±23.35) and almost half of the participants (44.3%) had severe handicap due to tinnitus. Most of participants (66.4%) scored above cut-off of GHQ-12 suggesting psychiatric caseness. THI-Sin has good internal consistency with Cronbach’s alpha coefficient of 0.902. Coefficients for functional, emotional, and catastrophic were 0.793, 0.776, and 0.742, respectively. Table 1 presents item-total statistics which indicates removal of any item would result in a lower coefficient. THI-Sin has a satisfactory item homogeneity with mean inter-item correlation of 0.353. The lowest correlations was in item 7 (0.292) while the highest in item 4 (0.629). Table 1 also presents means, standard deviations, and corrected item-total correlations of the 25 items of THI-Sin. The highest mean scores were recorded in item 17 (“Do you feel that your tinnitus has placed stress on your relationship?”) and item 9 (“Does your tinnitus interfere with your ability to enjoy social activities?”). Lowest mean scores were recorded in item 6 (“Do you complain a great deal about your tinnitus?”) and item 4 (“Does your tinnitus make you feel confused?”). Strong positive Spearman’s correlations for the total THI-Sin score and VAS (ρ=0.90, p<0.01) and GHQ-12 (ρ=0.34, p<0.01) confirmed convergent validity. Suitability of data and sample adequacy for factor analysis were confirmed by Bartlett’s test of sphericity (df=300, p<0.001) and Kaiser-Meyer-Olkin test (MSAs 0.816). Table 2 presents a three-factor solution with first extracted factor with an eigenvalue of 7.796 explaining 31.185% of the variance. Second and third extracted factors recorded eigenvalues of 2.852 and 1.613 which explained 11.409% and 6.452% of variance, respectively. Selected patients were interviewed regarding their experiences with THI-Sin. All patients reported that all the items in THI-Sin were clear to them and further modifications are not necessary. Discussion The results of this study demonstrate that the THI-Sin is a reliable and valid tool to evaluate tinnitus-related distress and handicap. It has excellent internal consistency, satisfactory construct and criterion validity which are similar to the original version [18]. Psyhological morbidity reported in this sample is similar to the prevalence reported in Sri Lankan and international studies conducted in individuals with tinntius [6,19-28]. Severity of tinnitus in our sample is largely similar to other validation studies [20,28]. Higher item scores for stress on relationship and difficulty in social activities could be due to collectivistic nature of the Sri Lankan society. Minimal endorsement of items on complaining about tinnitus and feeling confused due to tinnitus may be due to stigma related to the illness. Latter phenomena will have an implication on treatment seeking behaviour in Sri Lanka. Component analysis and factor loadings demonstrated three factors loadings of the 25 items, although these factors were different from subscales proposed by Newman, et al. [18]. The first factor loads on 11 items, all five items of catastrophic subscale, three factors each from emotional and functional subscales relating to mental health aspect such as depression, anxiety, anhedonia, helplessness, hopelessness, desperation, inability to cope and feeling trapped. Second factor loads on eight items, six items from functional scale and two items from emotional scale relating to impact of tinnitus on social and daily activities including socialization, relationships. The third factor loads on six items, four items from functional sub-scale and two items from emotional subscale relating to emotional and cognitive consequences of tinnitus such as poor concentration, irritability, and tiredness. Our factor analysis results are largely similar to analysis performed on the Polish version of the THI [24]. Factor analyses of several versions of THI including Cantonese, Danish, Italian, and Persian revealed unifactorial structures [20,22,25,28,29]. However, it should be noted that the external validity of THI-Sin may be limited due to probable sample bias with our study population consisting of an urban and semi urban population with relatively high health literacy which is not representative of whole of Sri Lanka. Further studies will be needed to assess the test-retest reliability, which when confirmed, will add important value to THI-Sin. In spite of this, THI-Sin is a measure with good psychometric properties which can be easily administered to quantify the impact of tinnitus in Sri Lankan clinical and research settings. We acknowledge the support extended by otolaryngology surgeon Dr. M. Sheriff and the staff of the otolaryngology clinic in Colombo North Teaching Hospital. Further, we would like to offer our special thanks to Ms. Ramesha Madushani for her valuable support in the data collection of this study. Table 1. Corrected item-total correlations, internal consistency, reliability and means of the THI items Mean Std. deviation Corrected item-total correlation Cronbach’s alpha if Item deleted THI Item 1 1.60 0.822 0.406 0.900 THI Item 2 1.83 0.907 0.295 0.902 THI Item 3 1.71 0.861 0.447 0.899 THI Item 4 1.58 0.814 0.629 0.895 THI Item 5 2.04 0.916 0.486 0.898 THI Item 6 1.56 0.763 0.528 0.897 THI Item 7 1.70 0.900 0.292 0.902 THI Item 8 1.73 0.847 0.625 0.895 THI Item 9 2.49 0.818 0.437 0.899 THI Item 10 1.72 0.839 0.627 0.895 THI Item 11 2.12 0.887 0.443 0.899 THI Item 12 2.32 0.839 0.491 0.898 THI Item 13 1.93 0.901 0.515 0.897 THI Item 14 1.86 0.859 0.611 0.895 THI Item 15 2.31 0.855 0.469 0.898 THI Item 16 1.74 0.824 0.623 0.895 THI Item 17 2.75 0.609 0.437 0.899 THI Item 18 2.30 0.872 0.603 0.895 THI Item 19 1.75 0.819 0.582 0.896 THI Item 20 2.22 0.851 0.373 0.900 THI Item 21 1.78 0.861 0.555 0.896 THI Item 22 2.10 0.831 0.557 0.896 THI Item 23 2.00 0.837 0.637 0.895 THI Item 24 2.11 0.874 0.335 0.901 THI Item 25 2.09 0.894 0.347 0.901 THI, Tinnitus Handicap Inventory Table 2. Rotated factor loadings of the three-factor solution Scale Item no and item Factor 1 Factor 2 Factor 3 F 24. Does your tinnitus get worse when you are under stress? 0.772 E 25. Does your tinnitus make you feel insecure? 0.770 E 22. Does your tinnitus make you feel anxious? 0.670 C 23. Do you feel that you can no longer cope with your tinnitus? 0.605 C 19. Do you feel that you have no control over your tinnitus? 0.524 E 21. Because of your tinnitus, do you feel depressed? 0.495 C 11. Because of your tinnitus, do you feel that you have a terrible disease? 0.411 C 5. Because of your tinnitus, do you feel desperate? 0.392 F 12. Does your tinnitus make it difficult for you to enjoy life 0.375 C 8. Do you feel as though you cannot escape your tinnitus? 0.334 F 4. Does your tinnitus make you feel confused? 0.318 E 3. Does your tinnitus make you angry? 0.792 F 1. Because of your tinnitus, is it difficult for you to concentrate? 0.725 F 7. Because of your tinnitus, do you have trouble falling asleep at night? 0.586 F 13. Does your tinnitus interfere with your job or household responsibilities? 0.584 F 2. Does the loudness of your tinnitus make it difficult to hear people? 0.516 E 16. Do you complain a great deal about your tinnitus? 0.387 F 9. Does your tinnitus interfere with your ability to enjoy social activities? 0.353 F 17. Do you feel that your tinnitus has placed stress on your relationship? 0.294 F 15. Because of your tinnitus, is it difficult for you to read? 0.747 F 18. Do you feel it difficult to focus your attention away from your tinnitus? 0.634 F 14. Because of your tinnitus, do you find that you are often irritable? 0.602 F 20. Because of your tinnitus, do you feel tired? 0.590 E 10. Because of your tinnitus, do you feel frustrated? 0.543 E 6. Does your tinnitus make you upset? 0.538 Scale categories F, C, and E represent tinnitus-related functional impairment (11 items), catastrophic thinking (5 items), and emotional responses (9 items), respectively. Conflicts of Interest The authors have no financial conflicts of interest. Author Contributions Conceptualization: Asiri Rodrigo, Thilini Abayabandara-Herath. Data curation: Thilini Abayabandara-Herath. Formal analysis: Asiri Rodrigo, Thilini Abayabandara-Herath. Methodology: Asiri Rodrigo, Thilini Abayabandara-Herath. Project administration: Asiri Rodrigo, Thilini Abayabandara-Herath. Supervision: Asiri Rodrigo. Validation: Asiri Rodrigo. Writing—original draft: Asiri Rodrigo, Thilini Abayabandara- Herath. Writing—review & editing: Asiri Rodrigo, Thilini Abayabandara- Herath. Approval of final manuscript: Asiri Rodrigo, Thilini Abayabandara-Herath. ==== Refs REFERENCES 1 Wallhäusser-Franke E Delb W Balkenhol T Hiller W Hörmann K Tinnitus-related distress and the personality characteristic resilience Neural Plast 2014 2014 370307 25120934 2 Salviati M Bersani FS Terlizzi S Melcore C Panico R Romano GF Tinnitus: clinical experience of the psychosomatic connection Neuropsychiatr Dis Treat 2014 10 267 75 24550676 3 Stobik C Weber RK Münte TF Walter M Frommer J Evidence of psychosomatic influences in compensated and decompensated tinnitus Int J Audiol 2005 44 370 8 16078732 4 Malouff JM Schutte NS Zucker LA Tinnitus-related distress: a review of recent findings Curr Psychiatry Rep 2011 13 31 6 21080115 5 Bartels H Pedersen SS van der Laan BF Staal MJ Albers FW Middel B The impact of type D personality on health-related quality of life in tinnitus patients is mainly mediated by anxiety and depression Otol Neurotol 2010 31 11 8 19816233 6 Abayabandara-Herath T Rodrigo A Psychological morbidity and resilience associated with tinnitus in a Sri Lankan population SL J Psychiatry 2021 12 10 4 7 Papitsi I Balatsouras DG Makris ID Koukoutsis G Kaberos A Tzavara C Validation of the Greek version of tinnitus handicap inventory Audiol Res 2020 10 39 48 33704121 8 Langguth B Landgrebe M Kleinjung T Sand GP Hajak G Tinnitus and depression World J Biol Psychiatry 2011 12 489 500 21568629 9 Shekhawat GS Searchfield GD Kobayashi K Stinear CM Prescription of hearing-aid output for tinnitus relief Int J Audiol 2013 52 617 25 23859059 10 Roland LT Lenze EJ Hardin FM Kallogjeri D Nicklaus J Wineland AM Effects of mindfulness based stress reduction therapy on subjective bother and neural connectivity in chronic tinnitus Otolaryngol Head Neck Surg 2015 152 919 26 25715350 11 Wilson MB Kallogjeri D Joplin CN Gorman MD Krings JG Lenze EJ Ecological momentary assessment of tinnitus using smartphone technology: a pilot study Otolaryngol Head Neck Surg 2015 152 897 903 25676150 12 Wise K Kobayashi K Searchfield GD Feasibility study of a game integrating assessment and therapy of tinnitus J Neurosci Methods 2015 249 1 7 25863140 13 Zobay O Palmer AR Hall DA Sereda M Adjamian P Source space estimation of oscillatory power and brain connectivity in tinnitus PLoS One 2015 10 e0120123 25799178 14 Kuk FK Tyler RS Russell D Jordan H The psychometric properties of a tinnitus handicap questionnaire Ear Hear 1990 11 434 45 2073977 15 Hall DA Zaragoza Domingo S Hamdache LZ Manchaiah V Thammaiah S Evans C A good practice guide for translating and adapting hearing-related questionnaires for different languages and cultures Int J Audiol 2018 57 161 75 29161914 16 Abeysena H Jayawardana P Peiris U Rodrigo A Validation of the Sinhala version of the 12-item general health questionnaire J Postgrad Inst Med 2014 1 E8 17 Adamchic I Langguth B Hauptmann C Tass PA Psychometric evaluation of visual analog scale for the assessment of chronic tinnitus Am J Audiol 2012 21 215 25 22846637 18 Newman CW Jacobson GP Spitzer JB Development of the tinnitus handicap inventory Arch Otolaryngol Head Neck Surg 1996 122 143 8 8630207 19 Monzani D Genovese E Marrara A Gherpelli C Pingani L Forghieri M Validity of the Italian adaptation of the tinnitus handicap inventory; focus on quality of life and psychological distress in tinnitus-sufferers Acta Otorhinolaryngol Ital 2008 28 126 34 18646574 20 Mahmoudian S Shahmiri E Rouzbahani M Jafari Z Keyhani M Rahimi F Persian language version of the “tinnitus handicap inventory”: translation, standardization, validity and reliability Int Tinnitus J 2011 16 93 103 22249867 21 Tobias CA Llanes EG Chiong C Validity of a Filipino translation of the tinnitus handicap inventory Int Tinnitus J 2012 17 64 9 23906830 22 Kam AC Cheung AP Chan PY Leung EK Wong TK van Hasselt CA Psychometric properties of the Chinese (Cantonese) tinnitus handicap inventory Clin Otolaryngol 2009 34 309 15 19673977 23 Der C Alzérreca E San Martín JT Román L Zamorano I Malhue J National linguistic validation of the tinnitus handicap inventory (THI). Assessment of disability caused by tinnitus in Chilean Spanish-speaking population Int Tinnitus J 2012 17 146 51 24333886 24 Wrzosek M Szymiec E Klemens W Kotyło P Schlee W Modrzyńska M Polish translation and validation of the tinnitus handicap inventory and the tinnitus functional index Front Psychol 2016 7 1871 27965609 25 Meng Z Zheng Y Liu S Wang K Kong X Tao Y Reliability and validity of the Chinese (Mandarin) tinnitus handicap inventory Clin Exp Otorhinolaryngol 2012 5 10 6 22468196 26 Schmidt LP Teixeira VN Dall’Igna C Dallagnol D Smith MM Brazilian Portuguese language version of the “tinnitus handicap inventory”: validity and reproducibility Braz J Otorhinolaryngol 2006 72 808 10 17308834 27 Aksoy S Firat Y Alpar R The tinnitus handicap inventory: a study of validity and reliability Int Tinnitus J 2007 13 94 8 18229787 28 Jalali MM Soleimani R Fallahi M Aghajanpour M Elahi M Psychometric properties of the Persian version of the tinnitus handicap inventory (THI-P) Iran J Otorhinolaryngol 2015 27 83 94 25938079 29 Zachariae R Mirz F Johansen LV Andersen SE Bjerring P Pedersen CB Reliability and validity of a Danish adaptation of the tinnitus handicap inventory Scand Audiol 2000 29 37 43 10718675
PMC010xxxxxx/PMC10352689.txt
==== Front J Audiol Otol J Audiol Otol JAO Journal of Audiology & Otology 2384-1621 2384-1710 The Korean Audiological Society and Korean Otological Society 36423622 10.7874/jao.2022.00192 jao-2022-00192 Case Report Tone-Burst Auditory Brainstem Response and Cortical Potentials in Diagnosis of Syndromic Auditory Neuropathy Spectrum Disorder http://orcid.org/0000-0002-4391-4959 Kaf Wafaa A. 1 http://orcid.org/0000-0001-8567-0203 Reiter Samantha 1 2 http://orcid.org/000-0003-4795-4349 Brodeur Amanda 3 http://orcid.org/0000-0003-4332-7911 White-Minnis Letitia 1 http://orcid.org/0000-0003-2521-2870 Deal William 4 1 Department of Communication Sciences and Disorders, Missouri State University, Springfield, MO, USA 2 St. Cloud Veterans Affairs Health Care System, St. Cloud, MN, USA 3 Department of Biomedical Sciences, Missouri State University, Springfield, MO, USA 4 Department of Psychology, Missouri State University, Springfield, MO, USA Address for correspondence Wafaa A. Kaf, MD, PhD Department of Communication Sciences and Disorders, Missouri State University, 901 S. National Ave., Springfield, MO 65897, USA Tel +1-417-836-4456 Fax +1-417-836-4242 E-mail [email protected] 7 2023 24 11 2022 27 3 153160 21 4 2022 2 9 2022 19 9 2022 Copyright © 2023 The Korean Audiological Society and Korean Otological Society 2023 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. In this study, we report our findings of comprehensive evaluation in a man with syndromic craniofacial features, cognitive impairment, and hearing loss. The patient underwent psychological and genetic testing and screening for 133 genetic mutations associated with hearing loss, as well as extensive audiological evaluation to assess the auditory pathway between the middle ear and the auditory cortex. Psychological testing showed moderate cognitive impairment. Genetic testing did not reveal a genetic mutation for hearing loss. Audiological evaluation revealed mixed hearing loss and signs of auditory neuropathy spectrum disorder (ANSD) despite absence of otoacoustic emissions and an absent click-evoked auditory brainstem response (ABR) without recording of cochlear microphonics (CM). ANSD was characterized by abnormal speech discrimination, bilateral robust CM to 2,000 Hz tone-burst (TB) ABR, and abnormal left thalamocortical and cortical pathways diagnosed based on auditory middle latency and cortical N1-P2 responses. These behavioral and electrophysiological findings suggest post-synaptic ANSD at the brainstem level. An abnormal left thalamocortical auditory pathway may be attributable to the combined effect of lack of neural synchrony secondary to ANSD mainly on the left and/or brain injury. The findings in this study support the use of TB ABR and auditory cortical potentials in the ANSD test protocol and in patients with craniofacial anomalies. Syndromic hearing loss Tone-burst ABR Cortical potentials Genetic screening ==== Body pmcIntroduction Individuals with craniofacial, syndromic features are at high risk of having auditory neuropathy spectrum disorders (ANSD) [1] as well as auditory processing disorders [2] that affect perceptual abilities. Therefore, these individuals should undergo comprehensive audiological evaluation, including both behavioral and electrophysiologic measures, to assess the entire auditory pathway from the middle ear to the auditory cortex. This study examined a 53-year-old male, hereafter referred to as PR, with known mixed hearing loss, cognitive impairment, facial dysmorphology, and involvement across various systems. A comprehensive audiological behavioral and electrophysiological assessment of the middle and inner ear and the central auditory pathway was administered to identify the possible site(s) of lesion and to determine if ANSD was present. Psychological testing was conducted to evaluate PR’s current cognitive status, including the intelligence quotient (IQ). In addition, genetic screening was completed to determine a possible mutation for common syndromic hearing loss. This case study of PR’s unique presentation adds to the knowledge base of craniofacial dysmorphologies associated with hearing loss. The study findings demonstrate detection of ANSD in the presence of mixed hearing loss, allowing for accurate diagnosis and appropriate intervention. Case Report History Case history was drawn from previous medical records and a structured interview with PR and his caregiver. We placed special emphasis on audiologic and otologic history, related medical conditions, and family history. Based on reports from PR’s mother, prenatal history was unremarkable; there were no complications during pregnancy and fetal development appeared to progress as expected. PR’s mother denies use of drugs or alcohol during the pregnancy. PR was born full-term via vaginal delivery. He was breech, with no other complications. His birth weight was 2.3 kg, which is considered low birth weight or small for a full-term gestational age [3]. His mother reported that PR was late to meet various developmental milestones, such as sitting up, walking, and talking. PR’s mother reported that he sat up at 10 months, walked at 20 months, and began talking (1-2 words) at about 24 months. PR’s mother recalled that his pediatrician determined that he had delayed bone growth and he was diagnosed with LeggCalve-Perthes disease (avascular necrosis of the proximal femoral head) in his left hip. Karyotype at age four revealed normal male chromosomes. During childhood, PR appeared to have normal hearing sensitivity. Hearing loss was first diagnosed at approximately age 43. PR was diagnosed with moderate cognitive impairment, but the IQ was unknown. Given the craniofacial anomalies and cognitive impairment, PR had a computed tomography scan at age 18 for detailed anatomic descriptions and detection of associated intracranial lesions or anomalies of the central nervous system, and the results were normal. He had bilateral keratoconus and had a corneal transplant in the right eye in 1994, at the age of 31. At the age of 43, PR has experienced bilateral chronic Eustachian tube dysfunction, resulting in recurrent serous otitis media in the left ear and bilateral high-frequency hearing loss. He has had several tympanostomy and pressure equalizing tubes in the left ear. He was first identified as having bilateral mixed hearing loss at 43 years of age and was fit with behind-the-ear hearing aids. Prior to the age of 43, history was negative for otitis media. When we evaluated PR in 2015, he was 51 years old, and his family reported a significant decline in his speech processing and understanding abilities. PR has a parental history of age-related and noise-induced hearing loss. PRs history is negative for noise exposure. Procedures A comprehensive evaluation was performed, including physical, audiological, psychological, and genetic evaluation. Physical examination PR is 160 cm tall, weighs 63.5 kg, and has a head circumference of 54.6 cm. His height is slightly below the 5th percentile while his weight is at the 5th percentile [4]. His head circumference falls at the 25th percentile of head circumference for a height and gender matched group [5]. PR displays hemispatial neglect toward his left side at the head area. He also displays musculoskeletal manifestations, including centralized obesity, weak muscle tone, abdominal hernias of unknown type, short stature (legs, arms, hands, head), uneven leg length most likely due to his Legg-Calve-Perthes disease, and flat feet. Clinical examination revealed facial dysmorphology, including: micrognathia (small jaw), microstomia (small mouth), smooth philtrum, large forehead, and low-set ears. Audiological evaluation Basic and behavioral audiological evaluation Assessment was conducted twice in April 2015 and in May 2016. Otoscopic examination and 226 Hz tympanometry, using a Grason-Stadler Middle Ear Analyzer (GSI Inc., Eden Prairie, MN, USA), were conducted to assess outer and middle ear status. Behavioral audiometric test battery, including pure tone and speech audiometry, and hearing speech-in-noise and processing fast speech using QuickSIN (Etymotic Research, Elk Grove Village, IL, USA) test and a 40% Time Compressed Speech Test (TCST) at the most comfortable level, respectively. Otoacoustic emissions (OAE), both distortion-product (DPOAE) and transient evoked (TEOAE), were collected, using a Natus Bio-Logic Navigator Pro Scout SPORT PC-based diagnostic OAE system (Natus Medical Inc, Middleton, WI, USA), to assess outer hair cell (OHC) function. OAE testing assessed frequencies from 500-4,000 Hz for click TEOAE and from 750-8,000 Hz for DPOAE. DPOAE responses were elicited with two “primary” frequencies (f1 and f2) at an f2/f1 ratio of 1.21, with f2 frequencies varied from 750 to 8,000 Hz with the resolution of 3 points per octave. DPOAE level (at 2f1-f2) was measured with L1=65 dB SPL and L2=55 dB SPL. Responses were considered present if the signal-to-noise ratio (SNR) was ≥6 dB [6]. Electrophysiologic audiological evaluation Auditory brainstem response (ABR) to click and tone-burst (TB) stimuli, auditory middle latency response (AMLR), and cortical N1-P2 complex were recorded from both ears. All recordings were conducted using the Intelligent Hearing System-Smart-Evoked Potential (IHS-SmartEP; Intelligent Hearing Systems, Miami, FL, USA) that was calibrated following manufacturer specifications. All recordings were conducted using 2-channel recording, with disposable electrodes were placed on the PR’s head with the non-inverting electrode at the high forehead (Fz) for the ABR and the AMLR recordings and on the vertex (Cz) for the cortical N1-P2 complex recording. The reference, inverting electrodes were placed on each mastoid (M1 and M2), and the common ground electrode on the mid forehead (Fpz). Electrode impedances were equivalent between electrodes and were <5 kΩ. PR was quiet during ABR recording and, he was kept awake and alert during AMLR and N1-P2 recordings, respectively. The ABR was recorded to clicks and TB stimuli (1,000 and 2,000 Hz), condensation and rarefaction polarities. Stimuli were delivered at a rate of 21.1/s and an intensity of 90 dB nHL to an insert earphone (ER-3A). For TB ABR, 2,000 Hz frequency was chosen because it has the best hearing threshold with the least amount of ABG. Blackman window with a 2-1-2 envelope was used to enhance neural synchronization, specifically in individuals with neural desynchronization as in our case. Responses were amplified 100,000 times and filtered (10–3,000 Hz). Each ABR trace was derived from the average of 2,068 sweeps. The AMLR was recorded to an 80 dB nHL alternating click presented at a rate of 17.1/s. The window was set to 60 ms. The AMLR were amplified 75,000 times, filtered (3–1,500 Hz), and responses were collected to 1,530 sweeps. For the N1-P2 complex, 85 dB nHL, alternating clicks were presented at 1.1/s, responses were amplified (50,000 times) and filtered (1-30 Hz), and the window was set a 500 ms window. Each N1-P2 trace was collected to 100 sweeps. All responses, ABR, AMLR and N1-P2, were replicated, and the averaged responses were labeled. Psychological evaluation A comprehensive psychological evaluation was completed to assess intellectual ability, attention, language function, and IQ using the Stanford-Binet scale, 5th edition [7]. This testing also assessed the general ability to reason, solve problems, visualize, and recall information in various forms. Testing was conducted by a licensed psychologist at the Learning Diagnostic Clinic on the university campus. Genetic evaluation Genetic evaluation was conducted by obtaining a family history and genetic testing with OtoSCOPE, a Next Generation Sequencing panel. The Molecular Otolaryngology and Renal Research Laboratories (MORL) at the University of Iowa conducted the OtoSCOPE genetic testing on a research basis. This genetic screening assesses 133 genes known to cause hearing loss, both non-syndromic and syndromic (for review, visit https://morl.lab.uiowa.edu/genes-included-otoscope-v9). The MORL team completed the data analysis. The analytical sensitivity was >99% for regions sequenced with >10× depth of coverage; 99.63% of the 1.208 Mbp targeted were covered at >10× depth of coverage; 9,850,609 sequencing reads (100 bp, paired end) were aligned. Basic and behavioral audiological findings Otoscopy in the current study (2015 and 2016) revealed a clear view of the right canal and tympanic membrane (TM) and all landmarks were visible. The left ear showed a partially occluded view of the left TM, with a pressure equalizing tube placed. Fig. 1 depicts annual pure tone thresholds and word recognition scores (WRS) from PR’s medical records from 2010-2014 and from the current study in 2015 and 2016. Pure tone thresholds were relatively stable from 2010 to 2012, with bilateral high frequency sensorineural hearing loss (SNHL) possibly due to early presbycusis. A significant decline in hearing occurred in the left ear, and a conductive component was noted in 2014, with progression of hearing loss mainly in the left from 2015 to 2016. Speech reception threshold (SRT) for the right ear declined slightly from 2010 to 2014 (15 dB to 25 dB), while left ear SRT declined significantly (30 dB to 55 dB). Results of WRS showed a similar trend from 2010 to 2016, with the right ear remaining stable (88-84%) but the left ear declining significantly from 76% in 2010 to 48% in 2016. A large SNR loss was evident on the QuickSIN test, bilaterally, with more SNR loss on the left (22.5 dB) than the right (12 dB). Also, PR has demonstrated a borderline TCST score on the right ear (78%) and a poor score in the left ear (64%) below the normal cutoff score of 82%. When tested with standard 226 Hz tympanometry, results showed Jerger Type A tympanograms for the right ear and Jerger Type B tympanograms with normal ear canal volume for the left ear, consistent with otitis media with effusion (OME) build up behind an occluded pressure equalizing tube; similar findings were also obtained in 2012 and 2014. Acoustic reflex thresholds were absent ipsilaterally and contralaterally, bilaterally. The PI-PB function test in 2015 and 2016 showed a significant rollover of 0.4 for the right ear and 0.47 for the left ear, suggesting a retrocochlear site of lesion [8]. Both TEOAEs and DPOAEs were absent at all frequencies, bilaterally. Absent OAEs typically indicate a cochlear loss due to OHC damage; however, PR’s extensive history of bilateral, recurrent middle ear issues and his current left mixed hearing loss should have impacted OAE results. Electrophysiologic audiological findings Fig. 2 displays the recorded click ABR. Click ABR produced no distinguishable waveforms with any polarity, and there was no evidence of cochlear microphonics (CM), bilaterally. In contrast, 2,000 Hz TB ABR to condensation and rarefaction polarities, as shown in Fig. 3A, revealed presence of robust CM waves and poor ABR morphology for both ears. The presence of the CM waves was confirmed using two measurements. First, when we clenched the tube of the insert earphone (control condition) the CM disappeared (bottom traces of Fig. 3A), confirming that the recorded responses were not artifact. Second, when the right 2,000 Hz TB ABR was recorded at different intensities (90, 80, and 70 dB nHL) the recorded CM waves at these intensities showed no latency shift, as shown in Fig. 3B, indicating their intensity-independent, and this confirms that the recorded responses are truly CM waves and not a neural response [9]. To evaluate the thalamocortical level of PR’s auditory pathway, click AMLR and N1-P2 responses were recorded in 2015 and 2016. Fig. 4A shows the AMLR Na-Pa response in 2015 displaying grossly normal morphology and amplitude (2.9 μV) on the right. The left ear Na-Pa response was noisier, poorer, and slightly smaller (2.1 μV), with slightly delayed Pa latency (27.6 ms) compared to the right Pa latency (22.4 ms). Absolute latency of Na wave was within normal bilaterally (15.6 ms right; 16.8 ms left). The N1-P2 response was recorded, while an assistant was asking PR some questions to keep him awake and alert during the recording. Fig. 4B shows that the response morphology, latency (N1=114 ms; P2=150 ms) and N1- P2 response amplitude were within normal on the right. In contrast, the left N1-P2 response morphology is abnormal with significantly smaller N1-P2 amplitude and delayed P2 latency (183 ms) than the right. These AMLR and N1-P2 findings are consistent with left auditory thalamocortical deficit. Psychological evaluation The psychological evaluation was conducted by a licensed psychologist. Results revealed that PR’s nonverbal IQ was 43, his verbal IQ was 46, and his full-scale IQ was 42. All scores fall into the moderately delayed range. The full-scale IQ of 42 is <0.1 percentile in comparison to age matched adults. These results indicate that nonverbal and verbal abilities are equally developed. Genetic evaluation Identification of specific gene mutations may further support the diagnosis of ANSD, and to help determine if the lesion is pre- or post-synaptic [10]. A five-generation family pedigree did not reveal a mode of inheritance for PR’s syndromic features. The multidisciplinary Hearing Group at the University of Iowa, performed the OtoSCOPE next-generation sequencing panel and determined that no plausible variants, including either single nucleotide or copy number variants, were identifiable to explain the deafness phenotype in this subject. Discussion This study examined a subject suspected of having syndromic hearing loss to determine whether the hearing loss is due to ANSD and to identify the specific syndrome, if possible. Given the heterogeneous syndromic hearing loss and ANSD population [11], a comprehensive audiological behavioral and electrophysiological test battery was used to assess the integrity of the central auditory pathway up to the level of the auditory cortex. In addition, genetic testing using OtoSCOPE screening for genetic mutations of hearing loss was performed. We also conducted psychological evaluation because our subject has cognitive impairment. Audiological evaluation shows signs of ANSD Results of speech audiometry revealed poor WRS that are disproportionate with the degree of hearing loss, and not related to cognitive impairment. Also, the significant rollover of speech recognition scores when speech was presented at loud sound is consistent with a possible neural hearing loss. The use of OAE testing with the absent responses was not helpful in our test battery to detect or rule out ANSD, because our subject had recurrent OME and repeated pressure equalizing tube placement. Therefore, recording ABR to both condensation and alternating polarity to assess CM response is recommended [12]. Our finding of absent or indiscernible click ABR waves possibly due to presence of CHL and/or is indicative of a neural pathology. However, the lack of recording CM response in the absence of ABR responses to click stimulus cannot confirm the diagnosis of ANSD [10]. On the contrary, recording CM with 2,000 Hz TB ABR supports the presence of ANSD bilaterally in our subject. To confirm the presence of the recorded CM response, absolute latency of the recorded responses at 2,000 Hz TB ABR showed no latency shift of the recorded waves with decreasing stimulus level, confirming present CM that is intensity-independent [9]. In addition, when the insert earphone tube was clenched, as a control condition, the CM response disappeared. These findings indicate that PR’s SNHL component and other auditory manifestations are due to a syndromic ANSD and not a cochlear lesion. These findings highlight the importance of recording TB ABR when click ABR does not show CM responses in suspected cases with ANSD [13]. The presence of abnormal left AMLR and N1-P2 response amplitudes and the delayed latency of the AMLR wave Pa and the P1 wave of the N1-P2 response supports the presence of abnormal left auditory thalamocortical and cortical pathway. These abnormal findings may be attributed to a combination of a post-synaptic ANSD due to lack of neural synchrony at the auditory nerve level and thus atypical onset precision neural input into the cortical level [14], cognitive problem, and/or brain damage as a possible cause of PR’s hemifacial neglect. In addition, abnormal auditory cortical responses are biomarkers to predict poor behavioral speech perception in difficult-to-test individuals who cannot reliably participate in behavioral speech testing, and to determine the most appropriate amplification option in individuals with ANSD. Furthermore, the grossly abnormal left cortical recorded responses suggest that PR may have a problem discriminating rapid spectro-temporal transitions and auditory closure processing. These were demonstrated in our subject as manifested by significant speech rollover and severe deficits in hearing speech in less favorable listening environments such as fast speech as in TCST and background babbling noise as with QuickSIN test. On the other hand, PR’s poor scores on TCST supports processing issues involving both low-level peripheral processing, as shown with abnormal click and TB ABR findings, and high-level processing speed ability as shown with abnormal thalamocortical and cortical findings [15]. Genetic testing did not detect mutations Through genetic testing, identification of specific gene mutations may further support the diagnosis of ANSD. The OtoSCOPE DNA sequencing findings in this study showed no mutation in any of the 133 screened genes; however, SNHL is genetically extremely heterogeneous. According to the genetic team, although the subject’s SNHL is not due to a mutation in any of these 133 genes, the subject’s SNHL may be due to mutations in novel gene(s) or a mutation in a different gene that was not evaluated and/or age-related hearing loss. While the OtoSCOPE platform includes the most common genes known to cause non-syndromic and syndromic hearing loss, more than 110 genetic loci have been mapped and many hearing losscausing genes remain to be discovered. In conclusion, the use of TB ABR to condensation and rarefaction polarities and auditory cortical N1-Pe potentials should be clinically adopted as part of the test protocol when assessing individuals suspected of having ANSD, a lesion that has proven difficult to identify with genetic testing or with both OAE and click ABR, especially in individuals with recurrent middle ear issues. Because ANSD affects auditory processing and perceptual abilities, auditory cortical potentials should be recorded as a biomarker to also determine the best intervention course, and site-of-lesion. Though genetic assessments did not identify a particular syndrome or a mutation for ANSD, the comprehensive audiologic test results increased our confidence in labeling this a case of syndromic ANSD, associated with left thalamocortical and cortical deficits due to lack of neural synchrony and faulty neural input to the auditory cortex. The authors would like to thank the research subject and his family for their participation in this study. The authors would also like to acknowledge Dr. Richard Smith and the Molecular Otolaryngology and Renal Research Laboratory team at the University of Iowa for providing genetic testing and interpretation of results. The authors would like to thank Ms. Rose Milcic, for helping with formatting and proofreading of the paper. Fig. 1. Puretone thresholds and word recognition scores (WRS) for PR show bilateral mild-to-moderate mixed hearing loss, sloping to moderate-to-severe high frequency loss. From 2010 to 2016, left ear thresholds progressed from mild-to-moderate mixed loss with fair WRS (76%) to moderate-to-severe mixed loss with poor WRS in 2014 and 2015 (52%–48%). WRS was not measured in 2010 and bone conduction thresholds were also not measured in 2010 and 2012. Fig. 2. Click auditory brainstem response (ABR) recorded waveforms at 90 dB nHL and 21.1/s clicks for condensation, rarefaction, and alternating polarities from both ears. The click ABR morphology was poor, no distinguishable ABR waves with any polarity, and no evidence of cochlear microphonics, bilaterally. The abnormal click ABR findings on the left are due to the mixed hearing loss and the presence of otitis media with effusion (OME). For the ABR findings on the right, although the degree of hearing loss mainly at 4,000 Hz does not justify the absent responses to clicks, it is possible that previous history of recurrent OME may be the cause of the absent click ABRs. Fig. 3. Tone-burst (TB) auditory brainstem response (ABR) recorded waveforms at 2,000 Hz for condensation and rarefaction from both ears. A: 2,000 Hz TB ABR at 90 dB nHL and 21.1/s for condensation and rarefaction polarities from the left ear and the right ear. There were no distinguishable TB ABR waveforms, but there were present cochlear microphonics (CM) waves, bilaterally (two repeated traces and the averaged trace at each polarity). The control traces revealed disappearance of CM waves when we clenched the tube of the insert earphone, confirming that the recorded CM responses were not artifact. B: 2,000 Hz TB ABR recorded at different intensities (90, 80, and 70 dB nHL). Results showed no latency shift with decreasing intensity levels, confirming that the recorded waves are CM responses, not neural responses. These findings are consistent with auditory neuropathy spectrum disorders. Fig. 4. Suprathreshold click auditory middle latency response (AMLR) and auditory late-latency cortical (N1-P2) response recorded from both ears. A: The AMLR. The Na-Pa response displays grossly normal morphology and amplitude on the right, but noisier, poorer, with slightly delayed Pa latency on the left. B: The slow cortical N1-P2 response. The response morphology, N1-P2 amplitude, and N1 and P2 latencies are within normal on the right, but abnormal on the left (smaller amplitude and delayed P2 latency). Overall, these AMLR and N1- P2 findings are suggestive of an abnormality mainly involving the left thalamocortical pathways. Ethics Statement The Institutional Review Board of Missouri State University approved this study (IRB-FY2016-234) and PR’s guardian signed a consent form for participation. Conflicts of interest The authors have no financial conflicts of interest. Author Contributions Conceptualization: Amanda Brodeur, Wafaa A. Kaf, Samantha Reiter, Letitia White-Minnis. Data curation: all authors. Formal analysis: Amanda Brodeur, Wafaa A. Kaf, Samantha Reiter. Funding acquisition: Amanda Brodeur, Wafaa A. Kaf (donated genetic testing and funding from the Graduate College of Missouri State University). Investigation: Amanda Brodeur, William Deal, Wafaa A. Kaf, Samantha Reiter. Methodology: all authors. Project administration: Wafaa A. Kaf, Samantha Reiter. Resources: all authors. Supervision: Amanda Brodeur, Wafaa A. Kaf, Samantha Reiter, Letitia White-Minnis. Validation: William Deal, Wafaa A. Kaf. Visualization: Wafaa A. Kaf, Samantha Reiter. Writing—original draft: Wafaa A. Kaf, Samantha Reiter. Writing—review & editing: all authors. Approval of final manuscript: all authors. ==== Refs REFERENCES 1 Abdul Wahid SN Md Daud MK Sidek D Abd Rahman N Mansor S Zakaria MN The performance of distortion product otoacoustic emissions and automated auditory brainstem response in the same ear of the babies in neonatal unit Int J Pediatr Otorhinolaryngol 2012 76 1366 9 22770594 2 Ma X McPherson B Ma L Behavioral signs of (central) auditory processing disorder in children with nonsyndromic cleft lip and/or palate: a parental questionnaire approach Cleft Palate Craniofac J 2016 53 147 56 25647518 3 Martin JA Hamilton BE Osterman MJ Driscoll AK Mathews TJ Births: final data for 2015 Natl Vital Stat Rep 2017 66 1 4 Fryar CD Gu Q Ogden CL Anthropometric reference data for children and adults: United States, 2007-2010 Vital Health Stat 2012 11 10 5 5 Bushby KM Cole T Matthews JN Goodship JA Centiles for adult head circumference Arch Dis Child 1992 67 1286 7 1444530 6 Hall JW A clinician’s guide to OAE measurement and analysis AudiologyOnline 2015 Aug 24 [Epub]. Available from https://www.audiologyonline.com/articles/clinician-s-guide-to-oae-14981 7 Roid GH Stanford-Binet intelligence scales 5th ed Itasca, IL Riverside Publishing 2003 8 Jerger J Jerger S Diagnostic significance of PB word functions Arch Otolaryngol 1971 93 573 80 5314647 9 Zhang M Effects of stimulus intensity on low-frequency toneburst cochlear microphonic waveforms Audiol Res 2013 3 e3 26557341 10 Santarelli R Information from cochlear potentials and genetic mutations helps localize the lesion site in auditory neuropathy Genome Med 2010 2 91 21176122 11 De Siati RD Rosenzweig F Gersdorff G Gregoire A Rombaux P Deggouj N Auditory neuropathy spectrum disorders: from diagnosis to treatment: literature review and case reports J Clin Med 2020 9 1074 32290039 12 Dallos P Cheatham MA Production of cochlear potentials by inner and outer hair cells J Acoust Soc Am 1976 60 510 2 993471 13 Mobley KJ Gibson E Tone burst evoked potentials: clinical applications AudiologyOnline 2000 Sep 6 [Epub]. Available from: http://www.audiologyonline.com/articles/tone-burst-evoked-potentialsclinical-1279 14 Dimitrijevic A Starr A Bhatt S Michalewski HJ Zeng FG Pratt H Auditory cortical N100 in pre- and post-synaptic auditory neuropathy to frequency or intensity changes of continuous tones Clin Neurophysiol 2011 122 594 604 20822952 15 Pickora-Fuller MK Processing speed and timing in aging adults: psychoacoustics, speech perception, and comprehension Int J Audiol 2003 42 Suppl 1 S59 67 12918611
PMC010xxxxxx/PMC10352690.txt
==== Front J Audiol Otol J Audiol Otol JAO Journal of Audiology & Otology 2384-1621 2384-1710 The Korean Audiological Society and Korean Otological Society 37340773 10.7874/jao.2023.00017 jao-2023-00017 Original Article Feasibility of Speech Testing Using Wireless Connection in Single-Sided Cochlear Implant Users http://orcid.org/0000-0001-9243-9392 Bae Seong Hoon 1 http://orcid.org/0000-0001-7893-1866 Jung Youngrak 2 http://orcid.org/0000-0002-1056-733X Hur Ji Hye 2 http://orcid.org/0000-0001-8714-2834 Kim Jeong Ha 2 http://orcid.org/0000-0001-9493-3458 Choi Jae Young 2 1 Department of Otorhinolaryngology, Gangnam Severance Hospital, Yonsei University College of Medicine, Seoul, Korea 2 Department of Otorhinolaryngology, Yonsei University College of Medicine, Seoul, Korea Address for correspondence Jae Young Choi, MD, PhD Department of Otorhinolaryngology, Yonsei University College of Medicine, Severance Hospital, Yonsei University Health System, 50 Yonsei-ro, Seodaemun-gu, Seoul 03722, Korea Tel +82-2-2228-3484 Fax +82-2-393-0580 E-mail [email protected] 7 2023 22 6 2023 27 3 133138 9 1 2023 7 2 2023 16 2 2023 Copyright © 2023 The Korean Audiological Society and Korean Otological Society 2023 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. Background and Objectives The speech tests used to evaluate language performance in patients with bilateral deafness (BiD) and cochlear implant (CI) are problematic if applied to patients with single-sided deafness (SSD) because normal ear hearing should be excluded. Thus, we investigated the feasibility of using wireless connection to evaluate speech intelligibility of the CI ear in patients with SSD. Subjects and Methods Patients with BiD and SSD were administered the word recognition scores (WRS) and speech intelligibility tests using an iPad-based wireless connection and conventional methods. To exclude normal side hearing in patients with SSD, masking noise and “plugged and muffed” method were used in the WRS and speech intelligibility tests, respectively. Results In patients with BiD, the WRS and speech intelligibility tests results using wireless connection and conventional methods were similar. In patients with SSD, the WRS using masking noise in the normal hearing ear was similar to that of using wireless connection. However, 3 of 11 patients with SSD showed under-masked results if using the “plugged and muffed” method. Conclusions Speech intelligibility testing using wireless connection is a convenient and reliable method for evaluating CI performance in patients with SSD. The “plugged and muffed” method is not recommended for evaluating CI performance in patients with SSD. Cochlear implants Bilateral hearing loss Unilateral hearing loss Hearing tests Speech intelligibility ==== Body pmcIntroduction Single-sided deaf (SSD) patients experience difficulties with speech recognition in noisy environment and sound localization. SSD occurs in 12–27 per 100,000 individuals, and is usually due to idiopathic sudden sensorineural hearing loss [1]. Cochlear implant (CI) can be a good treatment option for these patients and is supported by considerable evidence [2-4]. The hearing in noise capability of SSD patients significantly improves by using CI compared to that obtained with the contralateral routing of signal or bone-anchored hearing aid [2]. The effectiveness of CI in SSD patients is usually evaluated using the hearing in noise test (HINT); in this test, decreased signal to noise ratio implies that the patients achieved improvements in binaural summation and squelch effect [5]. Indeed, HINT is a very important test for assessing the symptoms of hearing difficulties in SSD patients. Conversely, the speech tests used in bilateral deaf (BiD) patients with CI to evaluate language performance are problematic when applied to SSD patients. To exclude the normal hearing ear, researchers use a masking noise or the “plugged and muffed” method [6-9]. These methods are effective, but they are time- and labor-consuming and require specific environments and devices. More importantly, the normal hearing ear is always a possible confounding factor, as over- and under-masking issues are present when using masking noise and the “plugged and muffed” method, respectively. Consequently, we used a wireless connection to the CI in SSD patients to perform speech tests. The recorded words or sentences were directly transmitted to the CI device via a wireless connection, thus completely excluding the normal hearing ear during the speech test. We hypothesized that this method would allow reliable evaluation of speech intelligibility in SSD patients after CI surgery. The BiD group result with the conventional method was compared to that with the test with the wireless connection to investigate the reliability of the novel testing method. Then, the SSD group results were analyzed to validate the conventional method using masking noise or the “plugged and muffed” method. Subjects and Methods Patients Retrospectively enrolled in the study were patients who visited our institution after CI for follow-up speech audiogram and speech intelligibility tests between January 1, 2021, and June 1, 2022. The patients concurrently underwent wireless connection testing as well as conventional auditory tests as a clinical routine. A total of 23 patients were enrolled; of these, 15 SSD patients (mean pure-tone threshold at 0.5, 1, 2, 3 kHz for better ear <50 dB) who underwent unilateral CI constituted the SSD group, and 8 BiD patients (mean pure-tone threshold at 0.5, 1, 2, 3 kHz for better ear >70 dB) with unilateral CI served as the control group (BiD group). The Severance Hospital (Seoul, Korea) Institutional Review Board approved this study (project number 1-2021-0044). Informed consent was waived because of the retrospective nature of the study. Conventional auditory tests Auditory performances were evaluated before surgery, and annually as possible after switching on of the device. During the follow-up period after operation, aided pure-tone audiometry and speech audiometry were performed in the sound field in the soundproof booth. The sound field consisted of two loudspeakers located at a distance of 1 m and at ±45° from the subject’s head. In SSD patients, a masking noise was applied to the normal hearing ear during pure-tone audiometry and speech audiometry using a headphone. The masking noise was 40 dB louder than the average pure-tone threshold of 500, 1000, 2000, and 4000 Hz in the normal hearing ear. The word recognition score (WRS) was assessed during speech audiometry, using 50 phonetically balanced monosyllabic words at the most comfortable loudness level [10]. Eight of 15 SSD patients and all the BiD patients conducted the speech audiometry test. To assess speech intelligibility, Categories of Auditory Performance (CAP) score, consonant discrimination, vowel discrimination, mono/disyllabic words (MSW/DSW), and sentence perception tests under auditory-only listening conditions were performed. The tests were performed with samples of words or sentences (modified SNUH Speech Perception Test) in a noiseless room environment at a 65 dB stimulation level, with the sample words or phrases pronounced by a single audiologist positioned 1 m away. All of BiD patients conducted the speech intelligibility tests twice with the conventional method and with a wireless connection. For 11 of 15 SSD patients, a speech intelligibility test was conducted twice with the “plugged and muffed” and wireless connection methods. Speech tests using wireless connection The enrolled patients had two different implanted devices: MedEL (Innsbruck, Austria) and Cochlear (Sydney, Australia). In the case of the Med-EL device, a neck loop receiver was needed to connect to the CI device. We used the Bluetooth function on an iPad to transmit the recorded sound signal from the iPad to the speech processor. After connection, the audiologist determined the most comfortable volume level of the device. The WRS was assessed in the most comfortable volume level, using the recorded 50 phonetically balanced monosyllabic words used in conventional speech audiometry in the soundproof booth. As there are currently no validated iPad-based Korean-language words for the speech perception tests, the same set of words/sentences used in the conventional tests were used for this assessment. Statistical analysis The Wilcoxon signed rank test was used to compare the conventional method and the wireless connection method in the same group. Statistical analyses were conducted using SPSS 25.0 (IBM Corp., Armonk, NY, USA) and visualized using PRISM 8.0 (GraphPad Software, San Diego, CA, USA). A p-value <0.05 was considered statistically significant. Result Results of wireless connection tests in the BiD group To evaluate the reliability of the speech test using wireless connection, we first compared the test results of the BiD group using the conventional test method with those using the wireless connection (Table 1 and Fig. 1). Because BiD groups have no bias due to contralateral ear in tests, the test results of the two methods should be similar if the wireless method is comparable to the conventional method. The WRS in the sound field in the soundproof booth was similar to that in the test using a wireless connection. The maximal difference was 12% between the two methods. In speech intelligibility test, the maximal differences between the two methods were 10%, 10%, and 4% in MSW, DSW, and sentence, respectively. There were no significant differences in the result of the two methods when using pairwise statistical analysis. These results support that the wireless method is comparable to the conventional method. Results of wireless connection tests in the SSD group Next, we evaluated the WRS of the SSD group by applying the masking noise to the normal hearing ear. The result was similar to that using a wireless connection. The maximal difference between the two methods was 14%. There was no statistically significant difference between the methods when using pairwise analysis. These results suggested that the masking noise in the soundproof booth was effective in inhibiting the normal hearing ear in the SSD group. However, there are noticeable different results between the two methods when they were applied to speech intelligibility tests. The speech intelligibility test was conducted using two different methods (“plugged and muffed” and wireless connection) in SSD patients. The results showed statistically significant differences in several categories (Fig. 2). Specifically, CAP score (p=0.031), vowel discrimination (p=0.012), and MSW (p=0.031) were significantly poorer when using a wireless connection. In addition, there were three patients who showed a large gap of more than 10% in sentence scores. Their sentence scores were 38%, 100%, and 88%, respectively in conventional test with plugged and muffed method. However, in wireless connection test, their sentence scores were under 10%. Given that their speech intelligibility results using a wireless connection were commonly poor in all subtests compared with those in the “plugged and muffed” method, an under-masked result was suspected. Discussion Our study proved that speech tests using a wireless connection to CI devices have an advantage in the prevention of under-masking compared to the “plugged and muffed” method. In the BiD group, there was no significant difference in test results between the conventional methods and the wireless connection. Similarly, there were no differences between the two methods in the SSD group upon applying the masking noise in the normal hearing ear. However, the speech intelligibility test for SSD patients using the “plugged and muffed” method showed significantly better results which are suspected of under-masking in several categories compared with the wireless connection method (Fig. 3). The wireless connection to the CI device is intended to help patients hear the directly transmitted signal from electronic devices such as the telephone or television. With the development of smartphones and their applications, wireless connection has become more widely applied to CI patients. We tried this method to evaluate CI patients’ speech intelligibility, and to determine the exact function of the implanted ear while excluding the normal ear. As the exclusion of a contralateral side hearing was de facto in the BiD group, the results should be similar between the two methods if the speech test using a wireless connection was accurate. Indeed, given the similar results for the BiD group among the conventional method and the wireless connection, the WRS and speech intelligibility test using the wireless connection seem reliable. Applying masking noise to the normal hearing ear seems effective, given that the WRS of the SSD group showed similar results regardless of the method used. However, the “plugged and muffed” method showed under-masked results in 3 of 11 SSD patients. Although the number of subjects was too small to identify statistically significant risk factors for the under-masking, the under-masked subjects showed poor CI speech intelligibility in common. Residual hearing did not cause the under-masking because the three under-masked subjects were deaf in the operated ear. The “plugged and muffed” method can attenuate 44–66 dB of sound [11]. Given our results, this method does not provide a consistent degree of sound attenuation. Therefore, although it is easier than applying masking noise, the “plugged and muffed” method is not recommended in SSD patients. Compared with the conventional method, wireless connection seems to be a simple and reliable method to inhibit the normal hearing ear. Meanwhile, several groups have tried directly connected audiometric testing via electrical cable connection [4,12,13]. The direct connection system is expected to be equivalent to the wireless connection in inhibiting the normal hearing ear. Yet, to the best of our knowledge, the feasibility of speech testing using wireless connection had not been tested prior to this study. With the advancement of technology, wireless connections are projected to replace electrical cables in the near future. Thus, we expect the result of this study can support the rationale for speech tests using a wireless connection. This study has some limitations. Mainly, there are no validated iPad-based Korean-language words for the speech perception tests. In the future, such tests should be developed and validated not only for hearing-impaired patients but also for the general population, similar to the existing English and Japanese versions [14,15]. In addition, this study is not including sound localization and hearing function in noise which are major functional advantages of CI in SSD. Future studies with these tests can suggest concrete evidence of the advantage of the wireless connection method compared to the “plugged and muffed” method. In conclusion, speech intelligibility testing using a wireless connection is a convenient and reliable method for evaluating CI performance in SSD patients. In addition, the “plugged and muffed” method is not recommended for evaluating CI performance in SSD patients because of this technique’s high rate of under-masking. This study was supported by a faculty research grant of Yonsei University College of Medicine for 6-2018-0086 (granted to J. Y. Choi). This work was supported by the National Research Foundation of Korea(NRF) grant funded by the Korea government(MSIP) (2020R1A2C3005787). Fig. 1. Word recognition score results of all groups (upper) and speech intelligibility test results of BiD patients (lower). In the graphs on the bottom, the median (thick line) and interquartile range (thin error bar) are marked. n=8 in both group. WRS: word recognition score, SF: sound field, BiD, bilateral deafness; SSD, single-side deafness; WC, wireless connection; ns, not significant; MSW, monosyllabic words; DSW, disyllabic words. Fig. 2. Speech intelligibility of the single-sided deafness group (n=11) with the “plugged and muffed” method and a wireless connection. The red hollow circles indicate the subjects who showed more than a 10% difference in sentence score between the two methods. *p<0.05. CAP, categories of auditory performance; Vowel, vowel discrimination; Consonant, consonant discrimination; MSW, monosyllabic words; DSW, disyllabic words; Plug, plugged and muffed method; WC, wireless connection; ns, not significant. Fig. 3. The graphical summary of this study. Blue arrows indicate similar results between the two methods. Red arrows indicate significant different result between the two methods. WRS, word recognition score using conventional method; Wireless WRS, word recognition score using wireless connection method; SIT, speech intelligibility tests; Wireless SIT, speech intelligibility tests using wireless connection method. Table 1. Information of the patients Patient Op age (yr) Sex PTA better ear (dB) Deaf duration (yr) CI device Follow-up (yr) CI using time per day (h) WRS (%) (masking) WRS (%) (wireless connection) BiD1 36 F 110 Progressive Cochlear 0.5 13 72 74 BiD2 66 M 71 Progressive Cochlear 0.5 13 62 74 BiD3 55 M 88 Progressive Med-EL 5 12 64 58 BiD4 28 M 101 Progressive Med-EL 1 12 62 64 BiD5 23 F 78 Progressive Cochlear 1.5 13 56 54 BiD6 60 F 74 Progressive Med-EL 3 12 72 76 BiD7 52 F 89 Progressive Cochlear 2 12 82 88 BiD8 60 M 106 Progressive Med-EL 5 12 66 56 SSD1 28 M 13 2 Cochlear 4 13 70 76 SSD2 38 M 9 0.5 Med-EL 0.5 4 4 0 SSD3 62 M 35 5 Med-EL 3.5 16 52 38 SSD4 56 M 21 16 Cochlear 1 3 26 20 SSD5 65 M 22 16 Med-EL 0.5 12 44 38 SSD6 54 M 48 1 Med-EL 2 15 38 44 SSD7 66 F 29 30 Med-EL 0.5 0 0 0 SSD8 55 M 40 2 Cochlear 0.7 13 44 32 Op age, operative age; PTA, mean pure-tone threshold for 0.5, 1, 2, 3 kHz; CI, cochlear implant; WRS, word recognition score; BiD, bilateral deafness; SSD, single-sided deafness Conflicts of Interest The authors have no financial conflicts of interest. Author Contributions Conceptualization: Jae Young Choi. Data curation: Ji Hye Hur, Youngrak Jung, Jeong Ha Kim. Formal analysis: Seong Hoon Bae. Funding acquisition: Seong Hoon Bae. Methodology: Seong Hoon Bae. Project administration: Youngrak Jung. Visualization: Seong Hoon Bae. Writing— original draft: Jae Young Choi, Seong Hoon Bae. Writing—review & editing: Jae Young Choi, Seong Hoon Bae. Approval of final manuscript: all authors. ==== Refs REFERENCES 1 Baguley DM Bird J Humphriss RL Prevost AT The evidence base for the application of contralateral bone anchored hearing aids in acquired unilateral sensorineural hearing loss in adults Clin Otolaryngol 2006 31 6 14 16441794 2 Arndt S Aschendorff A Laszig R Beck R Schild C Kroeger S Comparison of pseudobinaural hearing to real binaural hearing rehabilitation after cochlear implantation in patients with unilateral deafness and tinnitus Otol Neurotol 2011 32 39 47 21068690 3 Arndt S Laszig R Aschendorff A Hassepass F Beck R Wesarg T Cochlear implant treatment of patients with single-sided deafness or asymmetric hearing loss HNO 2017 65 Suppl 2 98 108 28188428 4 Friedmann DR Ahmed OH McMenomey SO Shapiro WH Waltzman SB Roland JT Jr Single-sided deafness cochlear implantation: candidacy, evaluation, and outcomes in children and adults Otol Neurotol 2016 37 e154 60 26756150 5 Schleich P Nopp P D’Haese P Head shadow, squelch, and summation effects in bilateral users of the MED-EL COMBI 40/40+ cochlear implant Ear Hear 2004 25 197 204 15179111 6 Zeitler DM Sladen DP DeJong MD Torres JH Dorman MF Carlson ML Cochlear implantation for single-sided deafness in children and adolescents Int J Pediatr Otorhinolaryngol 2019 118 128 33 30623849 7 Deep NL Gordon SA Shapiro WH Waltzman SB Roland JT Jr Friedmann DR Cochlear implantation in children with single-sided deafness Laryngoscope 2021 131 E271 7 32065422 8 Buss E Dillon MT Rooth MA King ER Deres EJ Buchman CA Effects of cochlear implantation on binaural hearing in adults with unilateral hearing loss Trends Hear 2018 22 2331216518771173 29732951 9 Galvin JJ 3rd Fu QJ Wilkinson EP Mills D Hagan SC Lupo JE Benefits of cochlear implantation for single-sided deafness: data from the House Clinic-University of Southern California-University of California, Los Angeles Clinical Trial Ear Hear 2019 40 766 81 30358655 10 Byun SW Frequencies of Korean syllables and the distribution of syllables of PB word list Korean J Otorhinolaryngol-Head Neck Surg 2003 46 737 41 11 Abel SM Odell P Sound attenuation from earmuffs and earplugs in combination: maximum benefits vs. missed information Aviat Space Environ Med 2006 77 899 904 16964737 12 Finke M Strauß-Schier A Kludt E Büchner A Illg A Speech intelligibility and subjective benefit in single-sided deaf adults after cochlear implantation Hear Res 2017 348 112 9 28286233 13 Chan JC Freed DJ Vermiglio AJ Soli SD Evaluation of binaural functions in bilateral cochlear implant users Int J Audiol 2008 47 296 310 18569102 14 Gallun FJ Seitz A Eddins DA Molis MR Stavropoulos T Jakien KM Development and validation of portable automated rapid testing (PART) measures for auditory research Proc Meet Acoust 2018 33 050002 30627315 15 Nishio SY Tono T Iwaki T Moteki H Suzuki K Tsushima Y Development and validation of an iPad-based Japanese language monosyllable speech perception test (iCI2004 monosyllable) Acta Otolaryngol 2021 141 267 72 33320029
PMC010xxxxxx/PMC10352691.txt
==== Front J Audiol Otol J Audiol Otol JAO Journal of Audiology & Otology 2384-1621 2384-1710 The Korean Audiological Society and Korean Otological Society 37461800 10.7874/jao.2023.00045 jao-2023-00045 Original Article Hearing Screening Alternative Using a Website-Based Application http://orcid.org/0000-0002-7041-2890 Rahim Tety Hadiaty 1 http://orcid.org/0000-0001-5781-0563 Sunjaya Deni Kurniadi 2 http://orcid.org/0000-0003-3684-4144 Hilmanto Dany 3 http://orcid.org/0000-0002-3196-3067 Hasansulama Wijana 4 http://orcid.org/0000-0001-6333-7792 Putra Frans Zefanya 5 1 Doctoral Study Program, Faculty of Medicine, Universitas Padjadjaran, Bandung, Indonesia 2 Faculty of Medicine, Universitas Padjadjaran, Dr. Hasan Sadikin General Hospital, Bandung, Indonesia 3 Department of Pediatrics, Faculty of Medicine, Universitas Padjadjaran, Dr. Hasan Sadikin General Hospital, Bandung, Indonesia 4 Department of Otolaryngology, Faculty of Medicine, Universitas Padjadjaran, Dr. Hasan Sadikin General Hospital, Bandung, Indonesia 5 School of Electrical Engineering and Informatics, Bandung Institute of Technology, Bandung, Indonesia Address for correspondence Tety Hadiaty Rahim, MD Doctoral Study Program, Faculty of Medicine, Universitas Padjadjaran, Jl. Prof. Eyckman No. 38, Pasteur, Sukajadi, Bandung, West Java 40161, Indonesia Tel +62-8122101045 Fax +62-22 2040984 E-mail [email protected] 7 2023 10 7 2023 27 3 123127 9 2 2023 26 3 2023 11 4 2023 Copyright © 2023 The Korean Audiological Society and Korean Otological Society 2023 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. Background and Objectives Indonesians encounter several barriers to regular functional hearing assessment. Hearing loss screening is only provided by tertiary-care hospitals that are not reachable by people in remote regions. This study aimed to develop a website-based hearing screening application that is accessible and inexpensive. Subjects and Methods This comparative study was conducted between July and August 2022 in the Otolaryngology Clinic of Muhammadiyah Bandung Hospital with noise levels below 50 dB. The hearing screening was conducted using a website-based application (www.Screenout.id) and audiogram as the gold standard method. On ScreenOut, patients heard sounds with frequencies at 500, 1,000, 2,000, 4,000, and 8,000 Hz and sound intensity of each frequency at 35, 55, and 75 dB using earphones. Results A total of 133 participants were enrolled in our study. ScreenOut showed high sensitivity, specificity, accuracy, positive predictive value, and negative predictive value (90.9%, 98.9%, 93.6%, 99.4%, and 84.8%, respectively). Regarding hearing threshold, a very strong correlation was found between ScreenOut and audiogram, ranging between r=0.843 and r=0.899. Aside from that, there was no significant difference in hearing threshold values between ScreenOut and audiogram. Conclusions Many advantages of the ScreenOut were found, including low-cost, accessibility, and easy-to-use interface, making it favorably used in low–middle-income countries such as Indonesia. Website-based hearing test Hearing screening Hearing disorder Audiogram ==== Body pmcIntroduction Indonesians encounter several barriers to having their hearing function assessed regularly, even though hearing is one of the five important senses. Audiogram is the gold standard for examining hearing function, but not all hospitals in Indonesia provide audiogram examinations due to its expensive cost and audiogram procurement required subsidies from the government [1,2]. Aside from that, there is also an assumption that hearing screening is not a priority since hearing loss is not an emergent condition needing prompt medical care. Due to the reasons above, the screening can only be conducted in tertiary-care hospitals located in Indonesia’s urban area. To access those services, patients spend more for transportation and accommodation causing economic burden for most patients considering that the number of low-income population in Indonesia is as many as 26.16 million people (9.54%) [3,4]. An alternative hearing screening method, that is accessible and affordable by Indonesians, is needed so that hearing screening can be accessed equally. A website-based application hearing screening is proposed as a new alternative in Indonesia because of its accessibility. The proposed method is relatively inexpensive, simple, and feasible since it does not require complex equipment and the skill of trained health workers that audiogram requires. Many studies have been carried out to develop a smartphone-based hearing loss screening method to be able to conduct early hearing loss screening examinations with easy access and low cost [1,5-11]. Studies developing smartphone-based or website-based hearing screening found that both methods had sensitivity and specificity that are comparable to audiograms [6,10,12-14]. In Indonesia, there is no previous study that develops a smartphone-based or a website-based hearing screening. The aim of our study is to develop a website-based application hearing screening model that is comparable to an audiogram and is able to overcome the barriers that occur in Indonesia. In the future, we hope that this website-based application can be proposed as an alternative hearing screening in Indonesia to detect hearing loss. Subjects and Methods Study design and subject enrollment This comparative study, approved by the Ethics Committee of Hasan Sadikin General Hospital (1043/UN6.KEP/EC/2021), was conducted from July 2022 to August 2022. All participating subjects provided written informed consent. Subjects were enrolled consecutively and randomly. The inclusion criteria were subjects diagnosed with hearing loss on anamnesis and physical examination and subjects with normal hearing who agreed to participate in the study by signing informed consent. The exclusion criteria were subjects who did not agree to participate in the study. Procedure Examination of hearing loss using audiometry is usually conducted in a sound-attenuating room with a relatively low sound intensity, at 25 dB. However, examination of hearing loss using ScreenOut in our study was conducted in an examination room in the Otolaryngology Clinic of Muhammadiyah Bandung Hospital, with a sound intensity of under 50 dB. When calibrated, the sound intensity was proportional to the sound intensity in a sound-attenuating room [15]. One study examined hearing loss in an empty examination room using a calibrated sound intensity that was proportional to the sound intensity in a sound-attenuating room. The study employed 35 dB as the lowest sound intensity tested in an empty examination room [16]. Thus, we used a sound intensity of 35 dB as the lowest sound intensity referencing the previous published results by other researchers [15,16]. Sound production for website-based hearing test A sound recording that is proportional to the sound frequency of the audiogram for the hearing screening test was performed in the recording studio. A sound generator was used to produce sound with frequencies at 500, 1,000, 2,000, 4,000, and 8,000 Hz and the sound intensity of each frequency at 35, 55, and 75 dB. The audio results from the sound generator that had been successfully recorded were then converted into files in WAV format, with the sample rate of 44,100 kHz, 24 bit. The hearing screening module was built according to the Business Process Model and Notation (BPMN). Hearing test The hearing screening was conducted using a website-based application (www.Screenout.id) and audiogram (OSCILLA SM.-950; Oscilla, Aarhus, Denmark) as the gold standard. All subjects were examined first using ScreenOut and followed by audiogram. The screening was carried out on the right ear first and followed by the left ear. On the website-based hearing test, patients heard sounds from the website by using earphones/headphones with frequencies of 500, 1,000, 2,000, 4,000, and 8,000 Hz and sound intensity at each frequency of 35, 55, and 75 dB. The order of frequency and intensity of the sound was: 500, 1,000, 2,000, 4,000, 8,000, 1,000, and 500 Hz (35, 55, 75 dB). The increase in sound intensity intervals in this study; namely from 35 dB, 55 dB, and 75 dB; is based on the classification of hearing loss by the American National Standards Institute, in which we took the average of each classification. The classification of hearing loss based on the American Standards Institute [17] is as follows: 1) mild hearing loss: 26 to 40 dB; 2) moderate hearing loss: 41 to 55 dB; and 3) severe hearing loss: 71 to 90 dB. Statistical analysis Hearing test data from audiogram and ScreenOut were analyzed using SPSS, Version 18 (SPSS Inc., Chicago, IL, USA). Analysis of sensitivity, specificity, accuracy, negative predictive value (NPV), positive predictive value (PPV), and Spearman rank correlation coefficient was performed. Results A total of 133 subjects were included in this study. Fiftynine subjects (43.36%) were male and 74 subjects (55.64%) were female. Subjects’ ages ranged from 8 to 84 years, with a median of 50 years (Table 1). Screening for hearing loss was conducted using ScreenOut website-based application and then compared with an audiogram. Hearing loss examination was measured in the frequency range of 500 Hz to 8,000 Hz. The results of the study are presented in the following table (Tables 1-3). Table 2 illustrates that the results of measuring the hearing threshold through examination through ScreenOut applications of various frequencies had lower median values when compared to audiogram examination. There was a very strong correlation of the hearing threshold between ScreenOut and audiogram, ranging from r=0.843 to r=0.899. Statistical analysis showed that hearing loss screening with ScreenOut had high sensitivity, specificity, accuracy, PPV, and NPV of 90.9%, 98.9%, 93.6%, 99.4%, and 84.8%, respectively (Table 3). Furthermore, the correlation analysis revealed a very strong correlation between ScreenOut and audiogram. Aside from that, there was no significant difference in hearing threshold values between ScreenOut and audiogram (mean difference [95% CI]=-2.80 [-4.61; -0.996] on the right ear and mean difference [95% CI]=-5.88 [-7.65; -4.12] on the left ear). Discussion In Indonesia, research developing a website-based application for early detection of hearing loss has never been carried out. Our research is essential in developing a website-based application with high sensitivity, specificity, and accuracy as compared to audiogram, which is the gold standard for hearing loss screening. Aside from that, in Indonesia, there is no alternative method of hearing loss detection other than audiogram, that can be reliably used as a reference for further examination or hearing loss therapy. A website-based application as an alternative to audiograms for hearing screening offers the opportunity to detect hearing loss in a simpler and more inexpensive manner. These features are suitable to use in developing countries such as Indonesia with a maldistribution of health facilities, a large number of low-income population, and geographical constraints to access health services. This website-based application can pose as the first screening tool, as the results of hearing screening can act as the basis for referrals so that patients receive further hearing loss examinations and therapy. Smartphone-based applications and websites for hearing loss screening have been developed by several developed countries, but this has never been developed in Indonesia [8-10,12]. Uhear, Audcal, Audicus, and Easy Hear are several smartphone-based applications that have been validated for hearing loss screening [12]. The website-based hearing screening application that we have developed is called ScreenOut. Applications for hearing screening as an alternative to audiograms should be comparable to audiograms in terms of sensitivity, specificity, and accuracy. Our study found that the cutoff value of 35 dB on ScreenOut had a sensitivity and specificity of 90.9% and 98.9%, respectively to detect hearing loss. This value is excellent when compared to other mobile-based applications. Uhear, a phone-based application for hearing screening, demonstrated that when screening for moderate or severe hearing loss in adults (pure tone average of >40 dB), high sensitivity (98.2%–100%) was achieved. However, the specificity varied (60.0%–82.1%) if screening was carried out in an environment with a noise level of around 40–50 dBA (quiet room) [9,18]. Ambient noise levels had a significant impact on the accuracy of Uhear [18]. The sensitivity of Uhear remained high across all test settings, whereas the specificity decreased in the waiting room setting (noise >50 dBA) and increased in a soundproof room (noise <40 dBA) [9]. Early detection of hearing loss has proven to be beneficial in early treatment and identification of etiologies. ScreenOut’s high specificity and sensitivity make it a proficient tool for early detection of hearing loss. This research is important because, in Indonesia, there is no alternative method for early hearing loss detection other than audiogram. Meanwhile, audiogram is relatively expensive and usually only available in large hospitals in Indonesia. Consequently, it is inaccessible to the majority of Indonesia population since there is a high number of low-income population especially in rural areas. Aside from that, Indonesia’s geographical condition, that of an archipelago with islands, also poses an obstacle to accessing hearing screening services. Due to the aforementioned reasons, there is a need for hearing loss detection method that can be easily accessed by all Indonesian citizens at a relatively low price and without the need for audiologist. We hope that an alternative screening method for hearing loss can be created through our study so that hearing loss can be detected early with better prognosis. In conclusion, ScreenOut is an excellent screening tool for hearing disorders due to its low-cost, accessibility, and easy-to-use interface. The present study had some limitations. The test is not accessible without internet access that it cannot be used in more remote areas. There is no standardization of the use of earphones/headphones which can affect the results. In conclusion, a website-based application for hearing assessment may aid people in low-resource settings to be able to detect hearing loss earlier. Our study showed that ScreenOut website was able to screen hearing loss with high sensitivity and specificity. ScreenOut can serve as a screening test before the patient is tested further with the more advanced examinations and can provide early warning of hearing loss. Further studies are needed for regular website calibration and for providing standardized operational procedures to implement ScreenOut in remote areas. This study received funding from Academic Leadership Grant (ALG), Universitas Padjadjaran. The funding bodies had no role in the design of the study and collection, analysis, and interpretation of data, or in writing the manuscript. We would like to thank Abida Hasna Laila, M.D. for her linguistic contribution to the manuscript. Table 1. Subjects’ age and sex distribution Age (yr) Sex Total Female Male 7-16 1 4 5 17-26 25 12 37 27-36 4 4 8 37-46 9 5 14 47-56 13 6 19 57-66 7 18 25 67-76 11 4 15 77-86 4 6 10 Total 74 59 133 Table 2. Hearing threshold of each examined frequencies on audiogram and website-based application ScreenOut (n=266) Frequency (Hz) Hearing threshold (dB) p-value Correlation coefficient (r) Audiogram ScreenOut 500 45 (15-110) 35 (35-80) <0.001 0.843 1,000 40 (10-110) 35 (35-80) <0.001 0.847 2,000 40 (5-110) 35 (35-80) <0.001 0.860 4,000 40 (10-110) 35 (35-80) <0.001 0.857 8,000 55 (0-100) 55 (35-80) <0.001 0.899 Values are presented as median (range). Table 3. Comparison of hearing screening results between audiogram and website-based application ScreenOut Hearing disorder on ScreenOut Hearing disorder on audiogram Total Validity Reliability (Kappa index) Positive Negative Positive 160 1 161 Sensitivity=90.9% 86.3% Negative 16 89 105 Specificity=98.9% Accuracy=93.6% PPV=99.4% NPV=84.8% LR+=81.8 LR-=0.09 Total 176 90 266 PPV, positive predictive value; NPV, negative predictive value; LR, likelihood ratio Conflicts of Interest The authors have no financial conflicts of interest. Author Contributions Conceptualization: Tety Hadiaty Rahim. Data curation: Tety Hadiaty Rahim, Deni Kurniadi Sunjaya. Formal analysis: Deni Kurniadi Sunjaya. Funding acquisition: Dany Hilmanto. Investigation: Tety Hadiaty Rahim, Deni Kurniadi Sunjaya, Wijana Hasansulama. Methodology: Tety Hadiaty Rahim, Deni Kurniadi Sunjaya, Dany Hilmanto. Project administration: Dany Hilmanto. Resources: Tety Hadiaty Rahim. Software: Frans Zefanya Putra. Supervision: Deni Kurniadi Sunjaya, Wijana Hasansulama. Validation: Tety Hadiaty Rahim, Deni Kurniadi Sunjaya. Writing—original draft: Tety Hadiaty Rahim. Writing—review & editing: all authors. Approval of final manuscript: all authors. ==== Refs REFERENCES 1 Saliba J Al-Reefi M Carriere JS Verma N Provencal C Rappaport JM Accuracy of mobile-based audiometry in the evaluation of hearing loss in quiet and noisy environments Otolaryngol - Head Neck Surg 2017 156 706 11 28025906 2 Best V Mason CR Swaminathan J Roverud E Kidd G Jr Use of a glimpsing model to understand the performance of listeners with and without hearing loss in spatialized speech mixtures J Acoust Soc Am 2017 141 81 91 28147587 3 Maulana A Benita T Typologi of island city in Indonesia IOP Conf Ser Earth Environ Sci 2017 79 012026 4 Ali PB Solikha DA Arifi MD Siahaan RGM Firdaus MZ Ariteja S White Book: National Health System Reformation [Internet] Jakarta The Ministry of National Development Planning Indonesia 2023 [cited 2023 Jul 03]. Available from: https://perpustakaan.bappenas.go.id/e-library/file_upload/koleksi/migrasi-data-publikasi/file/Policy_Paper/Buku%20Putih%20Reformasi%20SKN.pdf 5 Almufarrij I Dillon H Dawes P Moore DR Yeung W Charalambous AP Web- and app-based tools for remote hearing assessment: a scoping review Int J Audiol 2022 Jun 9 [Epub]. 10.1080/14992027.2022.2075798 6 Bauer MA Sales A Teixeira AR Morsch P Lessa AH Bós ÂJG Development and accuracy of a hearing screening application Braz J Otorhinolaryngol 2021 87 711 7 32620319 7 Bright T Pallawela D Validated smartphone-based apps for ear and hearing assessments: a review JMIR Rehabil Assist Technol 2016 3 e13 28582261 8 Corona AP Ferrite S Bright T Polack S Validity of hearing screening using hearTest smartphone-based audiometry: performance evaluation of different response modes Int J Audiol 2020 59 666 73 32134341 9 Peer S Fagan JJ Hearing loss in the developing world: evaluating the iPhone mobile device as a screening tool S Afr Med J 2015 105 35 9 26046161 10 Pickens AW Robertson LD Smith ML Zheng Q Song S Headphone evaluation for app-based automated mobile hearing screening Int Arch Otorhinolaryngol 2018 22 358 63 30357066 11 Shrimal Y Nandurkar A Self-reported hearing status and audiometric thresholds among college students using headphones J Otolaryngol ENT Res 2021 13 60 8 12 Irace AL Sharma RK Reed NS Golub JS Smartphone-based applications to detect hearing loss: a review of current technology J Am Geriatr Soc 2021 69 307 16 33341098 13 Wasmann JW Pragt L Eikelboom R Swanepoel W Digital approaches to automated and machine learning assessments of hearing: scoping review J Med Internet Res 2022 24 e32581 34919056 14 Sandström J Swanepoel DW Laurent C Umefjord G Lundberg T Accuracy and reliability of smartphone self-test audiometry in community clinics in low income settings: a comparative study Ann Otol Rhinol Laryngol 2020 129 578 84 31965808 15 Storey KK Muñoz K Nelson L Larsen J White K Ambient noise impact on accuracy of automated hearing assessment Int J Audiol 2014 53 730 6 24909592 16 Bean BN Roberts RA Picou EM Angley GP Edwards AJ Automated audiometry in quiet and simulated exam room noise for listeners with normal hearing and impaired hearing J Am Acad Audiol 2022 33 6 13 34034339 17 Anastasiadou S Al Khalili Y Hearing loss [Internet] Treasure Island, FL StatPearls Publishing 2022 [cited 2023 Mar 25]. Available from: https://www.ncbi.nlm.nih.gov/books/NBK542323/ 18 Szudek J Ostevik A Dziegielewski P Robinson-Anagor J Gomaa N Hodgetts B Can uHear me now? Validation of an iPod-based hearing loss screening test J Otolaryngol Head Neck Surg 2012 41 Suppl 1 S78 84 22569055
PMC010xxxxxx/PMC10352692.txt
==== Front J Audiol Otol J Audiol Otol JAO Journal of Audiology & Otology 2384-1621 2384-1710 The Korean Audiological Society and Korean Otological Society 36793227 10.7874/jao.2022.00353 jao-2022-00353 Original Article Health-Related Quality of Life in Children With Cochlear Implants From Parents’ Perspective http://orcid.org/0000-0003-0227-4198 Rochd Sara 1 http://orcid.org/0000-0001-8700-1046 Benhoummad Othmane 2 http://orcid.org/0000-0001-8028-0417 Lakhdar Youssef 1 http://orcid.org/0000-0002-7965-0502 Salhi Salma 1 http://orcid.org/0000-0002-0003-4185 Lhadj Mohamed Amine Ait 1 http://orcid.org/0000-0001-7410-9672 Rochdi Youssef 1 http://orcid.org/0000-0002-9783-724X Raji Abdelaziz 1 1 Faculty of Medicine and Pharmacy of Marrakech, Caddi Ayyad University, Marrakech, Morocco 2 Faculty of Medicine and Pharmacy, Ibn Zohr University, Agadir, Morocco Address for correspondence Sara Rochd, MD Faculty of Medicine and Pharmacy of Marrakech, Caddi Ayyad University, 274 Semlalia, Marrakech 40000, Morocco Tel +212666034973 Fax +212524432887 E-mail [email protected] 7 2023 17 2 2023 27 3 115122 2 8 2022 11 9 2022 20 10 2022 Copyright © 2023 The Korean Audiological Society and Korean Otological Society 2023 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. Background and Objectives To evaluate the health-related quality of life (HRQoL) in parents of children with cochlear implants and assess influencing factors. These data can enable practitioners to support patients and their families in utilizing the cochlear implant and its benefits to the fullest extent. Subjects and Methods A retrospective descriptive and analytic study was conducted at the Implantation center Mohammed VI. Parents of cochlear implant patients were asked to fill out forms and answer a questionnaire. Participants included parents of children <15 years old who underwent unilateral cochlear implantation between January 2009 and December 2019, presenting with bilateral severe to profound neurosensory deafness. Participants completed the children with cochlear implantation: parent’s perspective (CCIPP) HRQoL questionnaire. Results The mean age of the children was 6.49±2.55 years. The mean time between implantation for each patient and this study was calculated as 4.33±2.05 years. There was a positive correlation between this variable and the following subscales: communication, well-being and happiness, and the process of implantation. For these subscales, the score was higher as the delay was greater. Parents of children who received speech therapy before implantation were more satisfied on the following subscales: communication, general functioning, well-being and happiness, implantation process, implantation effectiveness, and support for the child. Conclusions The HRQoL is better in families of children who received their implant at an early age. This finding raises awareness of the importance of systemic screening in newborns. Cochlear Implantation HRQoL Congenital deafness CCIPP ==== Body pmcIntroduction According to a Center for Disease Control and Prevention report published in 2019, Congenital hearing loss represents 1.7 per 1,000 newborn [1] in the US, and the prevalence of hearing loss among children between 6-19 years of age was 14.9% of children based on the work of Niskar, et al. [2] As it is well known, hearing loss in children leads to other disabilities in 40%-50% of cases [3]. The cochlear implantation (CI) for congenital deafness in children has revolutionized the otological field in so many ways, permitting the acquisition of the hearing sense, the development of communication skills, and integration of social life, with a lesser risk of developing additional conditions [3,4]. Multiple health-related quality of life (HRQoL) questionnaires have been developed over the years and some of them were specific to the CI. These questionnaires were both addressed to the pediatric population after the surgery and to their parents [4]. O’Neill, et al. [7] developed the “The children with cochlear implantation: parent’s perspective (CCIPP),” a closed set specific CI survey based on the parent’s point of view. It is divided into 10 subscales. The CCIPP is the most used questionnaire by implantation teams [8]. The aim of our study is to evaluate the HRQoL in parents of children with CI and to assess the factors that might influence it, so we, as practitioners, would act on them to allow the patients and their families to take full advantage of the CI and its benefits in their daily lives. Subjects and Methods A total of 70 chilren were included. The parents gave verbal consent to participate in this study. This study was conducted in accordance with the Declaration of Helsinki. We conducted a retrospective descriptive and analytic study. This study included all children <15 years old when implanted, presenting a bilateral severe to profound neurosensory deafness who underwent unilateral CI between January 2009 and December 2019 in our implantation center, at least 2 years after their implantation surgery. The exclusion criteria as follows: • >15 years old, • With neurosensory deafness associated with abnormalities, • Cerebral palsy, • Patients who underwent bilateral implantations, and reimplantations. • Plus the Absolute contraindications:  - Major inner ear malformations,  - Complete cochlear ossification,  - Cochlear nerve agenesis,  - And in case of major anesthetic risk (implantation under local anesthesia is impossible in children). Patients data was collected from medical files in the same center; data like patient’s history, epidemiology and patient characteristics. The HRQoL questionnaire used was the Nottingham Pediatric Cochlear Implant Program “CCIPP” (Nottingham University Hospital, Nottingham, United Kingdom). The questionnaire was composed of 10 subscales: communication, general functioning, self-reliance, well-being and happiness, social relationship, decision to the implant, education, the process of implantation, supporting the child, and effect of implantation. The parents’ answers to the questionnaire were rated as strongly agree, agree, neither agree nor disagree, disagree, and strongly disagree. The parents completed the form in a special consult after agreeing to participate in the study, using the help of the doctors in charge of the study. We used IBM SPSS 25.0 (IBM Corp., Armonk, NY, USA), to perform the statistical analysis. Frequencies, mean, median, standard deviation, and percentages were used to assess the data distribution. The Spearman nonparametrical test was used to search for the correlations between descriptive data of the study and the relationship with the change of the subscales scores. A value of p<0.05 was considered statistically significant. Results We collected the data from 70 patients; the mean age was 6.49±2.55 years ranging from 1 to 12 years old. The most found age interval was between 5 and 7 years old children. There were more female than male patients representing 57.1% of cases. Our patients’ families were mostly of low social and economical levels living in urban areas, according to our country’s classifications. Parents had a secondary academic level in most cases and 10% of them were unschooled and 21.4% made it to primary school. Few mothers were working and representing 7.1% of all cases. Two sub-groups represented the medical coverage: patients with health insurance (31.4%) and patients with social security (a national program for destitute families) representing 65.7%, finally there were 2.9% of patients with no medical coverage (Table 1). The patient’s history found was mostly a history of meningitis in 8 cases, neonatal jaundice in 2 cases, and history of cranial trauma in 2 cases. The child’s deafness was discovered in most cases when patients presented language acquisition disorder and the age of the first consult was before 2 years old. All our patients benefited from a tonal and/or behavioral audiometry, an otoacoustic emission and acoustic evoked potential under general anesthesia, a CT scan and an MRI before the decision to implant. They also benefited from a psychological speech therapy consult that helped us know their psychological condition and their communication mode. The indication to implant was decided upon the age and the audiometric findings. The distribution of patients according to their demographic characteristics and the status of the child is shown in Table 1, with the number and percentages of patients in each category. The time between the implantation and this study was calculated in years in each patient with a mean of 4.33±2.05 years ranging of from 2 to 9 years. There was a positive correlation between this variable and some scores of subscales: communication (p=0.002), well-being and happiness (p<0.001), and the process of implantation (p=0.001), the score was higher as the delay was greater (Fig. 1). Fig. 2 is a box-plot chart with mean scores in all quality of life domains. The statistical study objectified a correlation between age and well-being and happiness (p=0.005), the process of implantation (p=0.029) subscales. The younger the patients, the more satisfied were the parents. Gender was not correlated to any subscale score. The mother’s profession was correlated to the process of implantation (p=0.003) and the supporting the child (p=0.028) subscales, stay at home mothers were more satisfied in terms of this study. Parents of children who received before implantation a speech therapy were more satisfied in terms of communication (p=0.043), general functioning (p=0.012), well-being and happiness (p=0.048), the process of implantation (p=0.002), effect of implantation (p=0.017) and supporting the child (p=0.008) subscales. They were highly more satisfied than the parents of children who did not benefit from a speech therapy before the implantation (Fig. 3). This was also the case for parents of children who were implanted early compared to those who had late implantation in the well-being and happiness (p=0.037) and education (p=0.004) subscales (Table 2 and Fig. 4). Table 2 regroups the nonparametric correlation results between the different variables and the subscales of the questionnaire. The rest of the plot’s representation of correlated variables to different subscale scores is presented in Fig. 5. The CCIPP subscales were intercorrelated for most of them, except for the effect of implantation subscale which was not correlated with any other questionnaire subscale. Supporting the child was not correlated to the well-being and happiness subscale (p=0.095) (Table 3). Table 3 regroups the correlations between the different subscales of the questionnaire. Discussion In a study about the history of CI, Ramsden stated that after concentrating on conductive hearing loss in past decades, otology is entering a new era of sensory hearing loss [3]. CI has known a big evolution in the last century, it started in the 17th century with Duchenne of Boulogne when he first stimulated an ear with electricity. Then many years later Dr. House managed to invent the very first single electrode cochlear implant, afterwards with the development of technology and science the multi-electrode implant we have nowadays have seen the light of life [4]. The main purpose of CI is to provide sound awareness in deaf patients by stimulating different areas of the cochlea [5]. It provides speech comprehension and high intelligibility, even conversations over the phone [4]. All these researches allowed to revolutionize the otology world, making members of families with neurosensory deafness acquire the hearing sense and thus the ability to speak and have a normal social integration. The burden and stress of having a disabled child and the mental pressure resulting from it, is a real problem for parents, considering the social, intellectual, and behavioral abnormalities arising from lacking an essential sense for a so-called normal life [6]. Multiple studies have treated this subject and the conclusion is that parents with impaired children were more luckily to develop mental health issues [6-8]. Vieira, et al. [9] conducted a study about the family’s perspective on the cochlear implant and concluded that for the parents the cochlear implant alone meant a better future for their children, not considering that the main actor in the processes of the child’s rehabilitation and the success of this whole process is the family itself. Which makes these kinds of studies important; understanding the family’s difficulties and the factors that might influence the quality of life in this families [10,11]. Studies were conducted over the years to assess the HRQoL in children with a cochlear implant using different questionnaires; most of them used the parent’s perspective instead of considering the children’s point of view [12]. We used the CCIPP, the same questionnaire was used by multiple studies published [8,9,12-17]. The correlation between the questionnaire subscales was evaluated, and it was found that the communication subscale was positively correlated with most subscales in our study, the same results were obtained by other researchers [8,12-14,17], and according to Kumar, et al. [17], this is explained by the fact that from the parent’s point of view a better communication skills implicates better self-reliance, a better education, better social interaction, and mostly achieving greater happiness. On the other hand, Alkhatani, et al. [15] findings were different. This could be due to the small sample they studied, as only the social relationship subscale was correlated to the well-being and happiness subscale. The other important parameter that was studied, was the time of use of the cochlear implant before the parent’s submission of the CCIPP questionnaire, in all different papers published the longer the time frame the higher the parent’s satisfaction [9,13-17,18,19], and our results match those found in the literature review. We also found out that the longer the children had their implant the highly satisfied were the parents with the process of implantation and the decision to implant subscales. In a systematic review conducted in 2021, only studied in few papers studied their correlation with the questionnaire results [12], two out of these studies, Alkhatani, et al. [15] and Vieira, et al. [9], used the CCIPP questionnaire and it was found that the age of implantation was associated with a greater quality of life score, our findings corroborate the ones in the literature. Consequently discovering the congenital deafness by systematic screening and by raising awareness about congenital deafness, would allow us to implant children at a younger age and hence get better results. Contrastingly as was stated by Vermi Sli Peker, et al. [20], demographic characteristics and patients’ history should be taken into account as they might influence the parent’s perspective and would help us point out the variables to act on. In this study, we also found that parents of children who received hearing aid and speech therapy before the CI were more satisfied than others. We did not find any previous studies discussing this parameter. The limitation of this study is mostly the questionnaire used to assess the HRQoL from the parent’s perspective, as the questionnaire has no consensual calculation process nor an interpretation scale [14,16]. Overall, parents were satisfied with the CI results in their children [17,19,21-28], but more studies should be done on this subject to have a better knowledge of the factor that might influence the HRQoL, and act on the findings to enhance the effect of the CI. In conclusion, parents’ expectations from the CI are not only for their children to acquire hearing sense, but also to have better communication skills, education level, and integrate the social life. In our study, the HRQoL is found to be better in families in which children were implanted at an early age, the younger the children (<5) the greater were the score satisfaction (p<0.001); this finding raises awareness of the importance of systemic screening in newborns. Psychological support should also be encouraged for both children and parents to allow better social integration. None Fig. 1. Box-plots for the time (from implantation to the study) and subscale scores. A: Communication scores (p=0.002). B: Well-being and happiness scores (p<0.001). C: The process of implantation scores (p=0.001) Fig. 2. Box-plot chart with mean scores in all quality of life domains. *Subscales: 1, communication; 2, general functioning; 3, self-reliance; 4, well-being and happiness; 5, social relationships; 6, education; 7, process of implantation; 8, effects of implantation; 9, decision to implant; 10, supporting the child. Fig. 3. Plot for the history of receiving speech therapy and different subscale scores. Fig. 4. Plots for implantation age and the education score and well-being of happiness score. Fig. 5. Box-plots for the age of appearance and the self-reliance score (A), the history of using hearing aids and the self-reliance score (B), communication mode and the self-reliance score (C), psychological assessment and well-being and happiness score (D), schooling status and the effect of implantation score (E). Table 1. Patients’ demographic and clinical status characteristics (n=70) Variables n (%) Age  <5 16 (22.9)  5-7 23 (32.9)  7-9 15 (21.4)  9-11 10 (14.3)  ≥11 6 (8.6) Gender  Male 30 (42.9)  Female 40 (57.1) Origine  Rural 33 (47.1)  Urban 37 (52.9) Socio-economical level  Low 35 (50.0)  Average 34 (48.6)  High 1 (1.4) Parents’ academic level  Unschooled 7 (10.0)  Primary 15 (21.4)  Secondary 32 (45.7)  University level 16 (22.9) Mother’s profession  Housewife 65 (92.9)  Working 5 (7.1) Health insurance  Social security 46 (65.7)  Health insurance 22 (31.4)  None 2 (2.9) Related marriage  No 42 (60.0)  Yes 28 (40.0) Neonatal jaundice  No 68 (97.1)  Yes 2 (2.9) Meningitis  No 62 (88.6)  Yes 8 (11.4) Cranial trauma  No 68 (97.1)  Yes 2 (2.9) Family similar cases  No 58 (82.9)  Yes 12 (17.1) Age of appearance  Congenital deafness 23 (32.8)  Perilingual deafness 36 (51.4)  Postlingual deafness 11 (15.7) Circumstances of discovery  Systematic screening 4 (5.7)  Doctor/pediatrician 8 (11.4)  Language acquisition disorder 58 (82.9) Age of first consultation  At birth/systematic screening (yr) 3 (4.3)  <2 44 (62.9)  2-4 16 (22.9)  ≥4 7 (10.0) Tonal or behavioral audiometry results  Moderate deafness 3 (4.3)  Severe deafness 38 (54.3)  Profound deafness 28 (40.0)  Total deficiency 1 (1.4) Auditory evoked potential  Presence of wave V 3 (4.3)  No wave V 67 (95.7) History of using a hearing aid  No 66 (94.3)  Yes 4 (5.7) History of receiving speech therapy  No 22 (31.4)  Yes 48 (68.6) Schooling status  Unschooled 12 (17.1)  Schooled 58 (82.9) Communication mode  Oral 49 (70.0)  Gestures 19 (27.1)  None 2 (2.9) Psychological assessment  Personality disorder 3 (4.3)  Behavioral problems 23 (32.9)  Relationship disorders 25 (35.7)  Mood disorder 13 (18.6)  Autism spectrum disorders 6 (8.6) Age indication  Prelingually deaf children 59 (84.3)  After the age of 5 if oral communication skills 11 (15.7) Audiometric indication  Profound deafness with prosthetic gain not allowing language development 34 (48.6)  Severe deafness, if the discrimination is less than or equal to 50%. 33 (47.1) Frequent hearing fluctuations and/or impact on the child’s language 3 (4.3) Implantation age (yr)  ≤5 58 (82.9)  >5 12 (17.1) Percentages may not total 100% due to rounding Table 2. Nonparametric correlation results between different variables and subscale scores Communication General functioning Self-reliance Well-being and happiness Social relationship Decision to implant Education Process of implantation Effect of implantation Supporting the child Age 0.005* 0.029* Sex Parents profession 0.003* 0.028* Health insurance 0.019* 0.003* 0.002* 0.014* 0.013* Age of appearance 0.030* History of the child receiving a hearing aid 0.024* History of the child receiving therapy speech 0.043* 0.012* 0.048* 0.002* 0.017* 0.008* Schooling status 0.049* Communication mode 0.005* Psychological assessment 0.029* Implantation age 0.037* 0.004* The time between implantation and this study 0.002* <0.001* 0.001* * significant correlation if p<0.05. Empty cells imply correlations are not significant Table 3. Nonparametric correlation results between different questionnaire subscale scores Communication General functioning Self-reliance Well-being and happiness Social relationships Education Effects of implantation Supporting the child Communication - <0.001 <0.001 <0.001 <0.001 0.001 0.889 0.001 General functioning - <0.001 <0.001 <0.001 <0.001 0.938 0.001 Self-reliance - <0.001 <0.001 <0.001 0.538 0.004 Well-being and happiness - <0.001 <0.001 0.793 0.095 Social relationships - <0.001 0.339 <0.001 Education - 0.436 <0.001 Effects of implantation - 0.003 Supporting the child - Significant correlation if p<0.05 Conflicts of Interest The authors have no financial conflicts of interest. Author Contributions Conceptualization: Sara Rochd, Othmane Benhoummad. Data curation: Salma Salhi, Mohammed Amine Ait Lhadj. Formal analysis: Sara Rochd. Methodology: Sara Rochd, Youssef Lakhdar. Supervision: Othmane Benhoummad, Abdelaziz Raji. Visualization: Youssef Rochdi. Writing—original draft: Sara Rochd. Writing—review & editing: Othmane Benhoummad, Abdelaziz Raji. Approval of final manuscript: all authors. ==== Refs REFERENCES 1 Centers for Disease Control and Prevention 2019 summary of diagnostics among infants not passing hearing screening [Internet] Atlanta, GA CDC 2021 [cited 2022 Jan 20]. Available from: URL: https://www.cdc.gov/ncbddd/hearingloss/2019-data/06-diagnostics.html 2 Niskar AS Kieszak SM Holmes A Esteban E Rubin C Brody DJ Prevalence of hearing loss among children 6 to 19 years of age: the Third National Health and Nutrition Examination Survey JAMA 1998 279 1071 5 9546565 3 Omidvar S Jeddi Z Doosti A Hashemi SB Cochlear implant outcomes in children with attention-deficit/hyperactivity disorder: comparison with controls Int J Pediatr Otorhinolaryngol 2020 130 109782 31785496 4 Hoffman MF Cejas I Quittner AL Health-related quality of life instruments for children with cochlear implants: development of child and parent-proxy measures Ear Hear 2019 40 592 604 30059365 5 Meserole RL Carson CM Riley AW Wang NY Quittner AL Eisenberg LS Assessment of health-related quality of life 6 years after childhood cochlear implantation Qual Life Res 2014 23 721 33 6 Kobosko J Geremek-Samsonowicz A Skarżyński H [Mental health problems of mothers and fathers of the deaf children with cochlear implants] Otolaryngol Pol 2014 68 135 42 Polish 24837909 7 O’Neill C Lutman ME Archbold SM Gregory S Nikolopoulos TP Parents and their cochlear implanted child: questionnaire development to assess parental views and experiences Int J Pediatr Otorhinolaryngol 2004 68 149 60 14725981 8 Fortunato-Tavares T Befi-Lopes D Bento RF Andrade CR Children with cochlear implants: communication skills and quality of life Braz J Otorhinolaryngol 2012 78 15 25 9 Vieira SS Dupas G Chiari BM Cochlear implant: the family’s perspective Cochlear Implants Int 2018 19 216 24 29363411 10 Ramsden RT History of cochlear implantation Cochlear Implants Int 2013 14 Suppl 4 S3 5 24533753 11 Moller AR Cochlear and brainstem implants Basel Karger 2006 12 Michelson RP The results of electrical stimulation of the cochlea in human sensory deafness Ann Otol Rhinol Laryngol 1971 80 914 9 5127759 13 Syed IH Awan WA Syeda UB Caregiver burden among parents of hearing impaired and intellectually disabled children in Pakistan Iran J Public Health 2020 49 249 56 32461932 14 Sessa B Sutherland H Addressing mental health needs of deaf children and their families: the national deaf child and adolescent mental health service Psychiatrist 2013 37 175 8 15 Alkhatani B Parents’ perspectives on cochlear implantation results for deaf children or children with hearing loss in Saudi Arabia Am Ann Deaf 2021 165 510 26 33678717 16 Silva JM Campos PD Moret ALM Influencing variables in the quality of life of children with cochlear implants: a systematic review Codas 2021 33 e20190153 33950145 17 Kumar R Warner-Czyz A Silver CH Loy B Tobey E American parent perspectives on quality of life in pediatric cochlear implant recipients Ear Hear 2014 36 269 78 18 Stefanini MR Morettin M Zabeu JS Bevilacqua MC Moret AL Parental perspectives of children using cochlear implant Codas 2014 26 487 93 25590912 19 Almeida RP Matas CG Couto MI Carvalho AC Quality of life evaluation in children with cochlear implants Codas 2015 27 29 36 25885194 20 Vermi Sli Peker S Demi R Korkmaz F Cukurova I Quality of life and parental care burden in cochlear implanted children: a case-control study Int J Pediatr Otorhinolaryngol 2020 136 110164 32570061 21 Brewis B le Roux T Schlemmer K Nauta L Vinck B Health-related quality of life in South African children who use cochlear implants Int J Audiol 2020 59 132 9 31516047 22 Byčkova J Simonavičienė J Mickevičienė V Lesinskas E Evaluation of quality of life after paediatric cochlear implantation Acta Med Litu 2018 25 173 84 30842707 23 Schorr EA Roth FP Fox NA Quality of life for children with cochlear implants: perceived benefits and problems and the perception of single words and emotional sounds J Speech Lang Hear Res 2009 52 141 52 18664684 24 Tokat T Çatlı T Başaran Bozkurt E Atsal G Muderris T Olgun L Parents’ view on quality of life after cochlear implantation in children with auditory neuropathy J Int Adv Otol 2019 15 338 44 31846909 25 Silva JM Yamada MO Guedes EG Moret ALM Factors influencing the quality of life of children with cochlear implants Braz J Otorhinolaryngol 2020 86 411 8 30898483 26 Dempsey M Simões-Franklin C Walshe P Glynn F Viani L A retrospective review of parents’ perceptions of the impact of bilateral cochlear implants on their child’s quality of life Cochlear Implants Int 2021 22 303 10 34126866 27 Artières-Sterkers F Mondain M Aubry K Bordure P Bozorg-Grayeli A Deguine O The French National Cochlear Implant Registry (EPIIC): results, quality of life, questionnaires, academic and professional life Eur Ann Otorhinolaryngol Head Neck Dis 2020 137 Suppl 1 S57 63 32792302 28 Archbold S Sach T O’neill C Lutman M Gregory S Outcomes from cochlear implantation for child and family: parental perspectives Deaf Educ Int 2008 10 120 42
PMC010xxxxxx/PMC10352693.txt
==== Front J Audiol Otol J Audiol Otol JAO Journal of Audiology & Otology 2384-1621 2384-1710 The Korean Audiological Society and Korean Otological Society 36423621 10.7874/jao.2022.00276 jao-2022-00276 Case Report Endoscopic Management of Glomus Tympanicum Tumor: Report of Three Cases and Review of the Literature http://orcid.org/0000-0002-5967-7054 Fountarlis Athanasios Luca http://orcid.org/0000-0003-2312-7218 Hajiioannou Jiannis http://orcid.org/0000-0002-4501-3257 Lachanas Vasileios http://orcid.org/0000-0001-6316-4991 Tsitiridis Ioannis http://orcid.org/0000-0002-2393-4012 Saratziotis Athanasios http://orcid.org/0000-0002-4834-2953 Alagianni Aggeliki http://orcid.org/0000-0003-1711-6811 Skoulakis Charalampos Department of Otorhinolaryngology, University General Hospital of Larisa, Mezourlo, Larisa, Greece Address for correspondence Athanasios Luca Fountarlis, MD, MSc Department of Otorhinolaryngology, University General Hospital of Larissa, Mezourlo, 41110, Larisa, Greece Tel +30-6980411541 Fax +30-241-0282050 E-mail [email protected] 7 2023 24 11 2022 27 3 145152 27 6 2022 3 9 2022 17 9 2022 Copyright © 2023 The Korean Audiological Society and Korean Otological Society 2023 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. Glomus tympanicum are benign tumors of vascular origin, arising from the neural crest cells and located on promontory. The treatment of choice is surgical excision of the lesion. Traditionally, it is performed under microscopic observation. With the introduction of endoscopes in the field of ear surgeries, an endoscopic approach has also evolved. Herein, we present case reports of three patients diagnosed with glomus tympanicum tumors who were operated on using an endoscopic approach. A review of the literature is also performed. The mass was completely excised in all patients, and there were no signs of recurrence at the follow-up at least a year later. Endoscopic ear surgery is a safe and effective method of managing glomus tympanicum tumors. Its main limitation is the tumor size; however, in most cases, tumors of stages I to II as per the Glasscock-Jackson classification and types A1 to B1 according to the modified Fisch-Mattox classification can be completely removed endoscopically. Careful preoperative selection of patients warrants the best outcomes. Glomus tympanicum Endoscopic surgical procedure Paraganglioma Conductive hearing loss Tinnitus ==== Body pmcIntroduction Glomus tympanicum tumors (GTT), also called middle ear paragangliomas, are benign tumors of vascular origin that arise from neural crest cells located on the promontory [1]. They most commonly present with pulsatile tinnitus, though hearing loss, pain and cranial neuropathies could also present in more advanced stages [2,3]. The treatment of choice consists of complete surgical excision via a transcanal or a postauricular approach [3,4]. This was traditionally done under microscopic vision, nevertheless, with the introduction of endoscopes in ear surgery, a total endoscopic tumor excision has been attempted. The first total endoscopic middle ear neoplasm removal was reported by Marchioni, et al. [5] in 2013 and since then a few other reports have been published. We present here three cases of GTT treated via a total endoscopic transcanal approach. A review of the relevant literature is also performed. Case Report Case 1 A 74-year-old female patient was referred to our clinic with a clinical diagnosis of glomus tympanicum. She had a 3-year history of pulsatile tinnitus and fullness sensation in the left ear, as well as earache for the last 3 months. Otoscopy showed a reddish mass posterior to the tympanic membrane (Fig. 1A). Pure-tone audiometry revealed a left-sided mild sensorineural hearing loss in the 3-4 kHz frequency (Fig. 1B) while her tympanogram was A type. CT showed a mass measuring 6.1×4.8 mm adherent to the promontory (Fig. 1C). The mass was classified A1, according to the modified Fisch-Mattox (mFM) classification [6], and I, according to Glasscock-Johnson (GJ) classification [7] (Table 1). Surgical excision of the tumor was performed endoscopically and it was sent for pathological examination, which confirmed the diagnosis of GTT. No complications occurred and the patient showed no signs of recurrence after 2 years (Fig. 1D, Table 2). Case 2 A 56-year-old female patient presented to the outpatient setting, complaining about pulsatile tinnitus, ear fullness, and a sensation of hearing loss. A red mass located in the middle ear appeared in otoscopy (Fig. 2A). Pure-tone audiometry revealed a left-sided moderately severe conductive hearing loss (CHL) in all frequencies (Fig. 2B) whereas her tympanogram was B type. Nonspecific findings of soft tissue density in the left middle ear and mastoid were shown in CT. MRI revealed a mass in the left middle ear, in touch with the promontory, enhancing post gadolinium (Fig. 2C). A clinical diagnosis of GTT (mFM class Β1, GJ class II) was made based on clinical examination and MRI, and the patient was planned for surgery. The mass was removed endoscopically and uneventfully, and the histopathological examination confirmed the diagnosis. The patient showed no signs of recurrence on a 14-month follow-up, although CHL persisted (Fig. 2D, Table 2). Otoscopy revealed a retraction pocket adherent to the promontory and the patient was offered a tympanoplasty, which she refused. Case 3 A 45-year-old female patient was referred to our clinic with a 2-year history of pulsatile tinnitus and hearing loss. A red mass located in the middle ear, attached anteriorly to a thinned tympanic membrane, appeared in otoscopy (Fig. 3A). Pure-tone audiometry revealed a left-sided moderate CHL in all frequencies (Fig. 3B), whereas her tympanogram was B type. The diagnosis of GTT (mFM class A2, GJ class I) was made based on clinical examination and imaging (Fig. 3C), and the patient was offered surgical excision. The mass was removed endoscopically along with the part of the thinned tympanic membrane which was firmly attached to the tumor therefore a cartilage tympanoplasty was performed to fix the perforation. Histopathological examination confirmed the diagnosis. The patient showed no signs of recurrence on a 13-month follow-up, although air conduction thresholds remained stable (Fig. 3D, Table 2). Surgical technique All procedures were performed under general anesthesia. None of the patients underwent preoperative angiography and embolization. A 3 mm 0° endoscope was used throughout the whole procedure and a 3 mm 30° endoscope was used at the end to check for tumor residues. Traditional ear surgery instruments, as well as instruments from the Panetti endoscopic ear surgery (EES) set, were used. The patient was placed in the supine position with the head turned 45° to the right. A solution of 1:100.000 epinephrine was injected into the external ear canal skin and the tympanomeatal flap was raised and elevated from the annulus. The tympanic membrane was then detached from the malleus and the tumor was removed from the promontory with a suction Rosen knife. In the first and third patients, the tumors were removed en bloc, while in the second one via piecemeal resection, due to its larger size and proximity to the ossicles. Hemostasis was achieved by leaning the monopolar diathermy tip against a suction catheter (Frazier suction tube, 5 French, coated with a Nelaton catheter), which was pushed on the tumor and its feeding vessels in the first two cases. The power was set at 18 W. In the third case, an ophthalmic bipolar diathermy (25G Diathermy probe DSP, Alcon/Grieshaber, Schaffhausen, Switzerland) was used. The endoscope was cleared with warm natural saline, when the image became obscure due to blood clots, even though the use of adrenaline-soaked cotton balls prevented this from frequently happening. After removal of the tumor, myringoplasty was performed with the use of a tragal cartilage graft, due to small perforations in the tympanic membrane. No bone drilling was applied throughout the whole procedure. Discussion EES has gained popularity over the last years, due to its advantages, which include a wide field of view, magnified vision, and the ability to visualize around corners [5,8,9]. As a result, its indications have expanded from cholesteatoma and stapes surgery to more complex procedures such as middle ear tumor excision and lateral skull base surgery, either via a total endoscopic approach or in combination with microscopic surgery [5,10,11]. GTT have as well been managed through an endoscopic approach and, since 2013, 13 studies have been published, to our knowledge, including 95 patients (Tables 3 and 4). An exclusive endoscopic excision was achieved in all patients, apart from one in the series of Noel and Sajjadi [12] and three in the series of Kileen, et al. [1]. The reasons, for which a postauricular microscopic conversion was needed, were tumor size in three cases and excessive hemorrhage in one case. One of the main challenges of GTT surgery is indeed its vascularity, which can result in significant bleeding. The most common methods, used for coagulation were bipolar cautery [4,5,13,14] and lasers. Argon plasma coagulation (APC) [13], diode [1], CO2 [1,3-5,15], and potassium titanyl phosphate (KTP) [1,2] lasers have all been used to achieve hemostasis. In two of our cases, a type of monopolar coagulation was used, by leaning the monopolar diathermy tip against a suction catheter. Generally, the use of monopolar cautery against the promontory is not advocated, as there is a risk of thermal injury in the fluids of the cochlea. In our two cases and in one case by Teh, et al. [16] that used the same method, hearing did not deteriorate postoperatively, nevertheless, monopolar cautery should be used with caution. For this reason, we set the power at a low value (18 W) and used it intermittently. To minimize this risk, in our third case, we used an ophthalmic bipolar diathermy. Of note, in one patient, in the series of Kileen, et al. [1], conversion to microscopic surgery via a transcanal approach was required to control bleeding. This problem was solved with the use of a “three-hand technique” in eight patients, in the series of Fermi, et al. [14]. It involves the use of a second surgeon that holds an extra instrument, to assist the primary surgeon. Total resection of the tumor was obtained in all studies presented, apart from three patients in the series of Fermi, et al. [14], because the tumors were closely related to the carotid artery. As for the complications, the most common one was tympanic membrane perforation in five patients [2,4,10,14,15], dysgeusia in four just from the series of Fermi, et al. [14], and hearing loss in one patient [5]. In our three cases, no major complications were recorded. Hearing thresholds remained stable or improved in the majority of patients, in the studies retrieved. A hearing deterioration of more than 10 dB was recorded in six patients and it was associated with tympanic membrane perforation in half of them. In our cases, two patients had CHL in all frequencies, preoperatively, which did not improve despite surgery. The reason in the second case was the medialization of the tympanic membrane, resulting in a retraction pocket adherent to the promontory. In the third case, no apparent reason was found, yet the patient declined surgical exploration. Operative time ranged from 45 to 248 minutes, but the higher values were affected by cases, in which conversion to microscopic surgery was undergone. EES has, on one hand, a positive effect on operative time, since postauricular approach and canalplasty can be spared. On the other hand, the use of just one instrument for dissection and hemostasis and the need of cleaning the endoscope tip from blood clots, prolong the procedure. Finally, tumor size is the main limitation of EES. The size of GTT in the studies retrieved from the literature review ranges from I to II in the GJ classification and from A1 to B1 in the mFM classification. Tumors extending into the mastoid or in the external auditory canal are more challenging to be managed in a total endoscopic approach. Careful preoperative evaluation and the ability to convert in microscopic surgery, when needed, are of utmost importance in such cases. EES is a safe and effective method of managing GTT. Its main limitation is tumor size, although tumors ranging from I to II in GJ classification and from A1 to B1 in mFM classification, can be completely removed endoscopically, in most cases. Hemostasis can be achieved with the use of lasers or electrocautery. Careful preoperative selection of patients, that could be managed endoscopically, ensures the best outcomes for the patients. None Fig. 1. Case 1. A: Endoscopic image (Glasscock-Jackson classification I). B: Preoperative pure-tone audiogram. C: CT image showing the tumor (arrow) adherent to the promontory. D: Postoperative pure-tone audiogram. Fig. 2. Case 2. A: Endoscopic image (Glasscock-Jackson classification II). B: Preoperative pure-tone audiogram. C: T1-weighted MRI image with contrast enhancement showing the tumor (dotted arrow). D: Postoperative pure-tone audiogram. Fig. 3. Case 3. A: Endoscopic image (Glasscock-Jackson classification I). The tympanic membrane is retracted and adherent to the tumor (white arrow). B: Preoperative pure-tone audiogram. C: CT image showing the tumor (dotted arrow). D: Postoperative pure-tone audiogram. Table 1. Glomus tympanicum classification Class Definition Glasscock-Jackson classification [7]  1 Tumor completely visible on otoscopy  2 Tumor completely filling the middle ear cavity  3 Tumors filling the middle ear cavity and extending into the mastoid  4 Tumor extending through the tympanic membrane into the external auditory canal Modified Fisch-Mattox classification [6]  A1 Tumor limited to the middle ear cavity and completely visible on otoscopic examination  A2 Tumor limited to the middle ear cavity but the margins are not visible on otoscopy – may extend to the Eustachian tube and/or to the posterior mesotympanum  B1 Tumor filling the middle ear cavity with extension into the hypotympanum and tympanic sinus  B2 Tumor filling the middle ear cavity, extending into the mastoid and medially to the mastoid segment of the facial nerve  B3 Tumor filling the middle ear cavity, extending into the mastoid with erosion of carotid canal Table 2. Characteristics of the three cases of glomus tympanicum tumors treated via a total endoscopic transcanal approach Case 1 Case 2 Case 3 Audiometric data (PTA, dB) (preoperative - postoperative) 23.75-27.5 61.25-63.75 43.75-46.25 Tumor size (CT) (mm) 6.1×4.8 15×13 3.6×3.7 Perioperative findings  Chorda tympani Preserved Preserved Preserved  Ossicular status Mobile – intact Mobile – intact Mobile – intact  Tumor location & extension Limited to the promontory Extension to sinus tympani, ET and hypotympanum Limited to the promontory and posterior mesotympanum  Feeding vessels identified 1 posteroinferiorly NI 1 posteroinferiorly  Facial nerve involvement No No No  Operative time (min) 66 139 184 Resolution of symptoms Yes HL persisted HL persisted PTA, pure tone average; ET, Eustachian tube; NI, non-identified; HL, hearing loss Table 3. Literature review: study demographics and tumor classification Study Type of study Country No. of patients Age mean (range)* F/M GJ classification mFM classification Marchioni, et al. [5] Case series Italy 3 33, 57, 67 2/1 NA 1 Α1, 2 Β1 Daneshi, et al. [13] Case series Iran 13 54 (44-68) 9/4 6 (I), 7 (II) NA Pollak and Soni [3] Case report USA 1 74 1/0 (I) A1 O’Connell, et al. [9] Two case reports USA 2 53, 64 0/2 1 (I), 1 (II) NA Killeen, et al. [1] Case series USA, Brazil 14 61.6 (34-82) 13/1 13 (I), 1 (II) 1 A1, 13 A2 Noel and Sajjadi [12] Case series USA 5 57.2 (38-77) 2/3 NA NA Okhi and Kikuchi [8] Case report Japan 1 51 1/0 (I) A1 Teh, et al. [16] Case report Malaysia 1 53 1/0 (II) NA Vicario-Quiñones, et al. [15] Two case reports Spain 2 70, 74 2/0 2 (I) NA Kaul, et al. [2] Case series USA 8 NA (43-83) NA NA NA Fermi, et al. [14] Case series Italy, Egypt 30 56.6 (22-82) 25/5 14 (I), 16 (II) 11 A1, 10 A2, 9 B1 Pradhan, et al. [10] Case report India 1 50 1/0 NA NA Quick, et al. [4] Case series Australia 10 45.5 (25-69) 9/1 5 (I), 5 (II) 5 A1, 2 A2, 3 B1 Total 91 66/17 43 (I), 31 (II) 20 A1, 25 A2, 14 B1 * Age in case reports and in small case series (≤3 patients) is presented separately for each case. GJ classification, Glasscock-Jackson classification; mFM classification, modified Fisch-Mattox classification; NA, not available Table 4. Literature review: results of endoscopic ear surgery Study Excusive endoscopic approach Complete removal Audiologic results Complications Operative time (min) Hemostasis technique Time of follow-up (mo) mean (range) Recurrence Marchioni, et al. [5] 3/3 3/3 1↑ 1↓ 1= HL NA 3 BC, 1 CO2L 10.7 (5-18) No Daneshi, et al. [13] 13/13 13/13 10↑ 0↓ 3= None 60 (45-120) 9 BC, 4 APC-L 20 (NA) No Pollak and Soni [3] 1/1 1/1 1= None NA CO2L 8 No O’Connell, et al. [9] 2/2 2/2 1↑ 1= None 52, 96 NA None No Killeen, et al. [1] 11/14 14/14 8↑ 3↓ 1= None 108.1 (58-248) KTP-L, diode-L, CO2L, temporary packing 11.1 (NA) No Noel and Sajjadi [12] 4/5 5/5 5= None NA KTP-L 23.2 (12-38) No Okhi and Kikuchi [8] 1/1 1/1 1= NA 66 Temporary packing 36 No Teh, et al. [16] 1/1 1/1 1= NA NA Monopolar suction diathermy 1 No Vicario-Quiñones, et al. [15] 2/2 1/1* 1↓ 1NA 1 TM perforation NA CO2L 6 No Kaul, et al. [2] 8/8 8/8 NA 1 temporary tinnitus NA KTP-L NA NA Fermi, et al. [14] 30/30 27/30 4↑ 3↓ 13= 1 TM perforation, 4 dysgeusia 115.2 (NA) BC, diamond burr 38.1 (18-103) No Pradhan, et al. [10] 1/1 1/1 NA TM perforation 110 NA 12 No Quick et, al. [4] 10/10 10/10 1↑ 2↓ 6= 1 TM perforation 98 (60-160) 4 BC, 7 CO2L 10 (4-25) No Total 87/91 87/90 * In one case it was not reported whether the tumor was completely removed. ↑, hearing improvement; ↓, hearing deterioration; =, hearing stable (changes up to 10 dB were considered as hearing stable); HL, hearing loss; TM, tympanic membrane; BC, bipolar cautery; CO2L, CO2 laser; APC-L, argon plasma coagulation laser; KTP-L, potassium titanyl phosphate laser; diode-L, diode laser; NA, not available Ethics Statement Written informed consent was obtained from the patients, for publication of these case reports and of the accompanying images. Conflicts of interest The authors have no financial conflicts of interest. Author Contributions Conceptualization: Athanasios Luca Fountarlis, Charalampos Skoulakis. Data curation: Vasileios Lachanas, Ioannis Tsitiridis. Formal analysis: Athanasios Luca Fountarlis, Jiannis Hajiioannou. Methodology: Athanasios Saratziotis, Aggeliki Alagianni. Supervision: Aggeliki Alagianni, Jiannis Hajiioannou, Charalampos Skoulakis. Visualization: Athanasios Luca Fountarlis, Athanasios Saratziotis. Writing— original draft: Athanasios Luca Fountarlis. Writing—review & editing: Jiannis Hajiioannou. Approval of final manuscript: all authors. ==== Refs REFERENCES 1 Killeen DE Wick CC Hunter JB Rivas A Wanna GB Nogueira JF Endoscopic management of middle ear paragangliomas: a case series Otol Neurotol 2017 38 408 15 28192382 2 Kaul VF Filip P Schwam ZG Wanna GB Nuances in transcanal endoscopic surgical technique for glomus tympanicum tumors Am J Otolaryngol 2020 41 102562 32563784 3 Pollak N Soni RS Endoscopic excision of a tympanic paraganglioma: training the next generation of ear surgeons World J Otorhinolaryngol Head Neck Surg 2017 3 160 2 29516061 4 Quick ME Acharya A Friedland P Kong JHK Saxby AJ Patel NP Endoscopic management of early stage middle ear paragangliomas-an Australian case series Otol Neurotol 2021 42 e1677 82 34121083 5 Marchioni D Alicandri-Ciufelli M Gioacchini FM Bonali M Presutti L Transcanal endoscopic treatment of benign middle ear neoplasms Eur Arch Otorhinolaryngol 2013 270 2997 3004 23377229 6 Jackson CG Glasscock ME 3rd Harris PF Glomus Tumors. Diagnosis, classification, and management of large lesions Arch Otolaryngol 1982 108 401 10 6284098 7 Fisch U Mattox D Paragangliomas of the temporal bone In: Microsurgery of the skull base 1st ed Stuttgart/New York Georg Thieme Verlag 1988 148 281 8 Ohki M Kikuchi S A small glomus tympanicum tumor resected by minimally invasive transcanal endoscopic approach Case Rep Otolaryngol 2019 2019 5780161 31360569 9 O’Connell BP Hunter JB Carlson ML Isaacson B Rivas A Wanna GB Utility of an ultrasonic aspirator in transcanal endoscopic resection of temporal bone paraganglioma J Neurol Surg B Skull Base 2017 78 96 8 28180050 10 Pradhan S Chappity P Nayak A Pradhan P Parida PK Exclusive endoscopic transcanal approach to lateral skull base lesions: institutional experience of 3 cases J Otol 2021 16 55 60 33505451 11 Ridge SE Shetty KR Lee DJ Current trends and applications in endoscopy for otology and neurotology World J Otorhinolaryngol Head Neck Surg 2021 7 101 8 33997719 12 Noel JE Sajjadi H KTP-laser-assisted endoscopic management of glomus tympanicum tumors: a case series Ear Nose Throat J 2018 97 399 402 30540890 13 Daneshi A Asghari A Mohebbi S Farhadi M Farahani F Mohseni M Total endoscopic approach in glomus tympanicum surgery Iran J Otorhinolaryngol 2017 29 305 11 29383310 14 Fermi M Ferri G Bayoumi Ebaied T Alicandri-Ciufelli M Bonali M Badr El-Dine M Transcanal endoscopic management of glomus tympanicum: multicentric case series Otol Neurotol 2021 42 312 8 33351561 15 Vicario-Quiñones F Rojas-Lechuga MJ Berenguer J Larrosa Díaz F Exclusive transcanal endoscopic approach to glomus tympanicum: experience in two cases Acta Otorrinolaringol Esp (Engl Ed) 2020 71 321 3 31514961 16 Teh CS Azrina A Fadzilah I Prepageran N Transcanal endoscopic excision of glomus tympanicum: a case report Med J Malaysia 2020 75 189 90 32281609
PMC010xxxxxx/PMC10352694.txt
==== Front J Audiol Otol J Audiol Otol JAO Journal of Audiology & Otology 2384-1621 2384-1710 The Korean Audiological Society and Korean Otological Society 36791797 10.7874/jao.2022.00311 jao-2022-00311 Case Report A Case of Unilateral Otologic Symptoms as Initial Manifestations of Granulomatosis With Polyangiitis http://orcid.org/0000-0001-5917-5794 Batinović Franko 1 http://orcid.org/0000-0002-2076-0512 Martinić Marina Krnić 1 http://orcid.org/0000-0002-0942-8092 Durdov Merica Glavina 2 http://orcid.org/0000-0002-6480-4812 Sunara Davor 1 1 Department of Otorhinolaryngology with Head and Neck Surgery, University Hospital Center Split, Split, Croatia 2 Department of Pathology, Forensic Medicine and Cytology, University Hospital Center Split, Split, Croatia Address for correspondence Franko Batinović, MD Department of Otorhinolaryngology with Head and Neck Surgery, University Hospital Center Split, Spinčićeva 1, 21000, Split, Croatia Tel +385-7-719-5505 Fax +385-21-389-563 E-mail [email protected] 7 2023 16 2 2023 27 3 161167 19 7 2022 23 8 2022 20 10 2022 Copyright © 2023 The Korean Audiological Society and Korean Otological Society 2023 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. Granulomatosis with polyangiitis (GPA) is a chronic and systematic autoimmune condition characterized by granuloma formation and necrotizing vasculitis of small to medium-sized vessels. GPA initially presents itself as respiratory and renal symptoms. Although temporal bone involvement is not uncommon, an otologic disorder is rarely the initial symptom. We present a case of a 36-year-old man who presented with unilateral ear pain, hearing loss, and facial palsy. After a series of diagnostics and temporal bone and chest imaging, he was diagnosed with GPA with multiorgan involvement. Cyclophosphamide and methylprednisolone relieved the patient’s ear pain and partially improved his hearing, facial palsy, and overall clinical condition. Although uncommon, systemic GPA may cause initial otologic symptoms and should not be dismissed as a possible cause of an otologic disease resistant to standard therapy. Granulomatosis with polyangiitis Facial palsy Hearing loss Otalgia ==== Body pmcIntroduction Granulomatosis with polyangiitis (GPA, previously called Wegener’s granulomatosis) is a rare, systemic, antineutrophil cytoplasmic antibody (ANCA)-associated form of vasculitis whose lesions usually affect the respiratory tract and kidneys [1]. This is an uncommon disease of an unknown etiology with the prevalence of five cases per 100,000 in the European population and equal distribution among sexes [2,3]. In general, two different forms of GPA have been described: the localized form, usually limited to the upper airway, and the systemic form, with predominant renal and pulmonary involvement [1]. Although upper airway symptoms are present in 70%-100% of GPA patients at the time of diagnosis and may be the only symptoms in the localized form [1,2], ear disorders are much less frequently the first and only manifestations of GPA, and usually occur as a consequence of sino-nasal involvement [2,3]. On the other hand, facial palsy, either alone or in combination with hearing loss, is rarely the presenting feature of the disease [4]. We present a clinical case of systemic GPA initially presenting as nocturnal otalgia, unilateral peripheral facial palsy (PFP), and severe mixed hearing loss (MHL). Case Report Presenting concerns A 36-year-old male patient arrived at the ENT Emergency Department (EED) with extreme pain in his left ear and a long history of left acute otitis media and noise exposure. The patient reported several weeks of persistent otalgia, left-sided hearing loss, and a sudden-onset left-sided facial weakness. Clinical findings and diagnostic assessments On examination, only erythema of the external auditory canal and a purulent effusion from the perforated left eardrum was detected. A mild peripheral left-sided facial palsy was detected and categorized as House-Brackmann (HB) grade 2/6. The function of other cranial nerves was normal. The initial neurological screening done before presentation in the EED was unremarkable. With the preliminary diagnosis of acute otitis media complicated with facial palsy, the patient was hospitalized for intravenous antibiotic and corticosteroid therapy (ceftriaxone and methylprednisolone) and further investigation. Initial laboratory tests were ordered and showed mild leukocytosis and mild elevation of C-reactive protein (CRP) levels up to 23.6 mg/L (Table 1). An initial audiological examination was conducted. Pure tone audiometry (PTA) documented normal hearing on the right ear up to 2 kHz and then severe sensorineural hearing loss up to 70 dB. On the left ear, severe mixed MHL up to 2 kHz and then profound hearing loss with no measurable response within the limits of the audiometer was recorded (Fig. 1A). A fiberoptic nasopharyngoscopy revealed only a small mucosal erosion on the right lateral nasal wall and was otherwise inconspicuous. For further investigation, a multi-slice computed tomography (MSCT) of the temporal bones was ordered. Scans of the left temporal bone showed a total opacification of the mastoid air cells and the tympanic cavity with ossicular lysis but preserved bone septations. The ossicular chain erosion on the left was also detected (Fig. 2). We decided to watchful waiting because the differential diagnosis was not clarified. We found the conservative therapy more appropriate than to engage in tympanoplasty with a vague working diagnosis. Treatment and follow-up Antibiotic therapy was initiated with ceftriaxone 2 g/day intravenously (IV), methylprednisolone 40 mg/day IV, and topical ciprofloxacin without improvement. On the fourth day of hospitalization, the patient began coughing severely and developed a continuous fever of 39.5°C. The nocturnal ear pain became unbearable (9 out of 10 on the visual analog pain scale). The facial nerve palsy deteriorated to HB grade 4/6 (Fig. 3) and his CRP level significantly rose from 23.6 mg/L to 155 mg/L (Table 1). A plain chest X-ray was ordered and showed multiple nodules on both lung fields. A chest MSCT revealed numerous cavitary lung masses measuring up to 3.8 cm and mediastinal lymphadenopathy (Fig. 4). Urine cultures were negative. Laboratory screening for possible autoimmune diseases was ordered and showed significant elevation of cytoplasmic-pattern of ANCA (c-ANCA) measuring 59 IU/mL, while myeloperoxidase-ANCA (p-ANCA) and antinuclear antibodies (ANA) were negative which pointed to a possible systematic autoimmune disorder (Table 1). The patient, therefore, underwent a kidney ultrasound examination which was unremarkable. The otologic surgery was the patient’s last resort, so the pulmonologist did the bronchoscopy with lavage and blind biopsy of the tracheal carina. Finally, the histological finding of the transbronchial biopsy confirmed GPA. In the underlying stroma, there was a dense infiltration of plasma cells, lymphocytes, and eosinophils, as well as a necrotic granuloma (Fig. 5A). A necrotic small vessel was surrounded by palisading histiocytes and a few giant multinuclear cells in the center of a granuloma (Fig. 5B). The final diagnosis of GPA was made three weeks after the onset of symptoms. Treatment with 1,000 mg IV cyclophosphamide (CYC), along with 800 mg IV methylprednisolone daily, at 7-day intervals, was initiated in consultation with a rheumatologist. Antibiotic treatment was discontinued. One week after the initiation of immunosuppressive therapy, the symptoms of cough, fever, and ear pain diminished. Control ear examination revealed a thickened, pale, and nontransparent left eardrum with a retraction pocket in the Prussak’s space, without effusion. Facial palsy improved from HB grade 4 to grade 2. A control PTA recorded a mild improvement from severe MHL to moderate MHL up to 2 kHz and profound hearing loss with no measurable response within the limits of the audiometer on the left ear (Fig. 1B). The patient was transferred to the Pulmonology Department where the methylprednisolone dosage was tapered slowly for the next 15 months. The rheumatologist limits the risk of relapse with CYC (1,000 mg/day IV), which was continued every 2 weeks for 1 month (day 1, day 15, day 30), then 500 mg/day IV every 3 weeks. No significant side effects were reported on immunosuppressive agents (Fig. 6). In the 2-month follow-up, there was no significant hearing improvement on the left ear on PTA. Tympanometry documented a shallow A-type tympanogram on the left ear. The facial nerve function remained HB grade 2. On a 6-month follow-up, the hearing thresholds were unchanged, and his facial palsy could still be categorized as HB grade 2. Discussion Classically, the ELK (ear, nose, and throat; lung; and kidney) acronym is used to describe the usual clinical involvement of the three systems in the systemic form of GPA [2]. According to the American College of Rheumatology, if at least two of the four criteria; 1) sinus involvement; 2) alteration in pulmonary radiology; 3) urinary sediment with hematuria or red cell casts, and; 4) histology with the presence of perivascular granulomas, are met the diagnosis of GPA can be determined with 88.2% and 92.0% sensitivity and specificity, respectively [1,5]. Our patient’s pulmonary involvement, bronchial findings, and elevated c-ANCA are sufficient criteria for GPA diagnosis. The presence of c-ANCA is observed in more than 90% of patients with GPA, but a negative result does not exclude the diagnosis [1]. At a large tertiary academic referral center, Kiessling, et al. [6] retrospectively analyzed 29 patients with skull base GPA. Twelve patients tested positive for c-ANCA and PR3 and eleven had facial weakness, while only four had MHL. In contrast to their study, our case showed audiometric verification of MHL. Furthermore, we documented gradation and improvement of PFP and MHL. According to the literature, otologic manifestations appear in 6% to 56% of patients suffering from GPA [3]. Serous otitis media with conductive hearing loss is the most common middle ear disorder found in GPA. GPA patients, on the other hand, are less likely to have sensorineural hearing loss, vertigo, or PFP with hypoacusis, which has been identified as a possible indicator of GPA activity [3,4]. GPA rarely presents as an isolated ear disorder, especially without the preceding sino-nasal involvement [2,3] and even more infrequently as facial palsy [4]. So, the presenting symptoms of facial palsy, ear suppuration, and MHL initially did not point to GPA. Rather an acute otitis media with or without cholesteatoma was primarily suspected. Sahyouni, et al. [7] reported a series of 11 patients who presented at a neurotology clinic with otologic symptoms with no previous diagnosis of GPA. In that series, 10 patients presented with hypoacusis, more than half of which were bilateral. Upon audiometric examination, only 1 patient had unilateral MHL with otalgia and facial palsy [7]. Wierzbicka, et al. [8] noted only 1 case of unilateral MHL as the first symptom in 7 patients with GPA, the other 6 had bilateral severe MHL with hearing improvement after ipsilateral paracentesis and steroid therapy. Mur, et al. [9] described a patient with GPA-induced serous otitis media and unilateral PFP which was resolved with spontaneous remission of paresis after tympanostomy tube placement. These studies depict several treatment options found effective in patients with GPAs with MHL and PFP. On the other hand, our therapy of choice was immunosuppressive therapy which also showed a beneficial effect on MHL and PFP. Facial palsy, in association with GPA, is noticed in about 5% of patients [4] and it is commonly caused by compression in the middle ear facial nerve course or vasculitis of the vasa nervorum [3,4]. Unilateral PFP has been reported in advanced local disease [6,8] but it is extremely rarely the presenting feature. Moreover, two case reports described ipsilateral facial nerve paralysis with bilateral sensorineural hearing loss as initial GPA presentation [10,11]. On the other hand, we showed the combination of ipsilateral PFP and unilateral severe MHL as presenting signs of the GPA. The standard treatment for the systemic form of GPA is an immunosuppressive such as CYC or rituximab, along with corticosteroids in high doses [1]. Kim, et al. [10] described significant effect of rituximab on hearing loss and PFP. In our case, pulse doses of 1,000 mg/day IV CYC and 10 mg/kg/day IV methylprednisolone were given for the first 7 days. Classically, the doses were adjusted for condition, comorbidities, and kidney function to obtain better tolerability without a decrease in efficacy. After 7-day therapy, PFP and hearing thresholds partially improved. Although commonly, CYC and corticosteroids are a mainstay of GPA therapy, relapses are common (up to 93%) [2]. The outcomes of GPA initially presenting with otologic dysfunction and afterward progressing to the generalized form is poor and mortality, without treatment, reaches 90% within 2 years [2,8]. Therefore, clinical suspicion, early diagnosis, and treatment are of utmost importance [2]. The GPA presenting with otologic manifestations can be very insidious and present a challenge for the ENT specialist. Because GPA is a multisystem disease with variable symptomatology, patients must be treated holistically. In our case, the ENT specialist included a multidisciplinary team of physicians (neurology, radiology, pathology, pulmonology, and rheumatology specialists) who effectively addressed the patient’s diagnostic and treatment needs. This case is limited by its incomplete data collection due to missing documentation from pulmonary control and detailed rheumatology treatment and specific tests such as audiometric data after 6 month follow-up. However, the apparent strength of our case is an important consideration that GPA should be a differential diagnosis of prolonged and painful acute otitis media resistive to standard treatments in adults. Finally, a multidisciplinary approach is mandatory in the treatment of patients with the otologic manifestation of GPA. In summary, although ENT symptoms are expected during a course of localized or systemic GPA, GPA presenting with unilateral MHL, PFP, and ear pain is uncommon and can present a clinical challenge. Ear pain and effusion with accompanying PFP not responding to antibiotic treatment should raise the concern of the otologic GPA, and a holistic approach to the patient must be included. The authors would like to thank Vana Košta, MD, PhD, neurologist; Josipa Kokeza, MD, pulmonologist; Ivona Božić, MD, rheumatologist; Sanja Lovrić Kujundžić, MD, PhD, radiologist, for their treatment advice and technical assistance. We would also like to thank professor Dalibora Rako for the professional proofreading of the case in English. Fig. 1. Initial audiogram (A) delineating normal hearing on the right ear up to 2 kHz and then severe sensorineural hearing loss up to 70 dB. On the left ear, severe mixed hearing loss at 250 Hz to 2 kHz and then profound hearing loss with no measurable response within the limits of the audiometer is recorded. Post-treatment audiogram (B) revealed mild improvement of severe mixed hearing loss in the left ear with residual profound sensorineural hearing loss with no responses within the limits of the audiometer at 2 kHz to 8 kHz. Fig. 2. Coronal (A) and axial (B) CT scans of the temporal bone demonstrate total opacification of the mastoid air cells (B, asterisk) and the tympanic cavity with ossicular lysis but preserved bone septations. The arrow shows ossicular chain erosion (A and B, arrow). The right temporal bone with normal appearance of the middle ear ossicles and mastoid cells on coronal and axial CT scans. Fig. 3. Peripheral facial palsy on the fourth day of hospitalization. At rest (A), the patient has slight facial asymmetry, especially around the corners of the lips. When asked to look surprised (B), the left frontalis does not contract. When closing his eyes (C), the left lids cannot be completely approximated. When attempting to smile (D), the patient experiences clear weakness of the left side of his face with his mouth pulling to the right. Fig. 4. Multi-slice CT of the lungs on the fourth day of hospitalization. Coronal (A) views of the lungs showing mediastinal lymphadenopathy (filled arrow). Sagital (B) and axial (C) views reveal basal areas of poorly defined consolidation and multiple peripheral nodules (empty arrows). Fig. 5. Granulomatosis with polyangiitis in the tracheal mucosa. A: Fragments of the tracheal mucosa are covered with respiratory and metaplastic squamous epithelium. Dense infiltration of plasma cells, lymphocytes, and eosinophils with a necrotic granuloma (arrow) can be seen in the underlying stroma (hematoxylin and eosin stain, magnification ×200). Scale bar: 200 µm. B: In the center of a granuloma is a necrotic small vessel surrounded by palisading histiocytes (empty arrow) and a few giant multinuclear cells (filled arrow) (hematoxylin and eosin stain, magnification ×400). Scale bar: 100 µm. Fig. 6. Timeline of the case. HB, House-Brackmann grade; PTA, pure tone audiometry; MSCT, multi-slice CT; ANA, antinuclear antibodies; c-ANCA, cytoplasmic-pattern of antineutrophil cytoplasmic antibody; CRP, C-reactive protein; GPA, granulomatosis with polyangiitis. Table 1. Significant laboratory test results Test Result Normal value Initial (June 11, 2017)  Hemoglobin (g/dL) 137 138-185  Hematocrit (%) 0.386 0.415-0.530  MCV (fL) 84.2 83.0-97.2  Leukocytes (×109/L) 10.4 3.4-9.7  Lymphocytes (%) 23.8 20-46  Platelets (×109/L) 282 158-424  Urea (mmol/L) 7.0 2.8-8.3  Creatinine (µmol/L) 83 64-104  CRP (mg/L) 23.6 <5  Urine analysis Normal On hospitalization (June 18, 2017)  Hemoglobin (g/dL) 144 138-185  Hematocrit (%) 0.405 0.415-0.530  Leukocytes (×109/L) 14.0 3.4-9.7  CRP (mg/L) 155 <5  ANA Negative Negative  c-ANCA (IU/mL) 59 Positive >25  p-ANCA (AU/mL) 2 Negative <20 MCV, mean corpuscular volume; CRP, C-reactive protein; ANA, antinuclear antibodies; c-ANCA, cytoplasmic-pattern of antineutrophil cytoplasmic antibody; p-ANCA, myeloperoxidase-antineutrophil cytoplasmic antibody Ethics Statement This case report was approved by the Ethical Committee of the University Hospital of Split (No. 2181-147/01/06/M.S.-22-02). Written informed consent for data and image publication was obtained. Conflicts of Interest The authors have no financial conflicts of interest. Author Contributions Conceptualization: Franko Batinović, Davor Sunara. Data curation: Davor Sunara, Merica Glavina Durdov. Investigation: Franko Batinović, Davor Sunara. Methodology: Merica Glavina Durdov. Software: Marina Krnić Martinić. Supervision: Davor Sunara, Marina Krnić Martinić. Validation: Merica Glavina Durdov. Visualization: Franko Batinović, Merica Glavina Durdov. Writing—original draft: Franko Batinović, Marina Krnić Martinić. Writing—review & editing: Davor Sunara, Merica Glavina Durdov. Approval of final manuscript: all authors. ==== Refs REFERENCES 1 Comarmond C Cacoub P Granulomatosis with polyangiitis (Wegener): clinical aspects and treatment Autoimmun Rev 2014 13 1121 5 25149391 2 Greco A Marinelli C Fusconi M Macri GF Gallo A De Virgilio A Clinic manifestations in granulomatosis with polyangiitis Int J Immunopathol Pharmacol 2016 29 151 9 26684637 3 Takagi D Nakamaru Y Maguchi S Furuta Y Fukuda S Otologic manifestations of Wegener’s granulomatosis Laryngoscope 2002 112 1684 90 12352687 4 Ferri E Armato E Capuzzo P Cavaleri S Ianniello F Early diagnosis of Wegener’s granulomatosis presenting with bilateral facial paralysis and bilateral serous otitis media Auris Nasus Larynx 2007 34 379 82 17350198 5 Pakalniskis MG Berg AD Policeni BA Gentry LR Sato Y Moritani T The many faces of granulomatosis with polyangiitis: a review of the head and neck imaging manifestations AJR Am J Roentgenol 2015 205 W619 29 26587951 6 Kiessling PT Marinelli JP Peters PA DeLone DR Lane JI Koster MJ Cranial base manifestations of granulomatosis with polyangiitis Otolaryngol Head Neck Surg 2020 162 666 73 32178578 7 Sahyouni R Moshtaghi O Abouzari M Le P Birkenbeuel J Cheung D A case series of granulomatosis with polyangiitis primarily diagnosed by otological manifestations Ann Otol Rhinol Laryngol 2019 128 263 6 30486667 8 Wierzbicka M Szyfter W Puszczewicz M Borucki L Bartochowska A Otologic symptoms as initial manifestation of Wegener granulomatosis: diagnostic dilemma Otol Neurotol 2011 32 996 1000 21725265 9 Mur T Ghraib M Khurana JS Roehm PC Granulomatosis with polyangiitis presenting with bilateral hearing loss and facial paresis OTO Open 2019 3 2473974X18818791 10 Kim SH Jung AR Kim SI Yeo SG Refractory granulomatosis with polyangiitis presenting as facial paralysis and bilateral sudden deafness J Audiol Otol 2016 20 55 8 27144236 11 Koenen L Elbelt U Olze H Zappe S Dommerich S Granulomatosis with polyangiitis in a patient with polydipsia, facial nerve paralysis, and severe otologic complaints: a case report and review of the literature J Med Case Rep 2022 16 291 35897050
PMC010xxxxxx/PMC10352704.txt
==== Front RSC Adv RSC Adv RA RSCACL RSC Advances 2046-2069 The Royal Society of Chemistry d3ra01981k 10.1039/d3ra01981k Chemistry Formation of typical disinfection by-products (DBPs) during chlorination and chloramination of polymyxin B sulfate† † Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d3ra01981k https://orcid.org/0000-0002-2929-6111 Wei Xingya a Han Bangjun a Gu Renzheng a Geng Weimin a Gao Naiyun b a School of Municipal and Ecological Engineering, Shanghai Urban Construction Vocational College Shanghai 200438 China [email protected] b College of Environmental Science and Engineering, Tongji University Shanghai 200092 China 18 7 2023 12 7 2023 18 7 2023 13 31 2153721544 26 3 2023 8 7 2023 This journal is © The Royal Society of Chemistry 2023 The Royal Society of Chemistry https://creativecommons.org/licenses/by-nc/3.0/ Disinfection by-products (DBPs) formed in chlorination and chloramination are proved to be cytotoxic and genotoxic and arouse increasing attention. However, previous studies of DBP precursors mainly focused on free amino acids (AAs) and few papers evaluated DBPs' formation potential of combined AAs. This study demonstrated that typical carbonaceous (C-) DBPs, trihalomethanes (THMs) and typical nitrogenous (N-) DBPs, dichloroacetonitrile (DCAN), trichloroacetonitrile (TCAN) and trichloronitromethane (TCNM) could be formed during chlorination and chloramination of polymyxin B sulfate (PBS), a common polypeptide antibiotic working against Gram-negative bacterial infections. The effects of major parameters, including disinfectant dose, contact time, solution pH, temperature, bromide concentration and chloramination mode were evaluated in batch experiments. Different kinds of DBPs exhibited different characteristics as disinfectant dose or contact time increased. Solution pH and temperature affected the formation of DBPs greatly. The formation pathways of different DBPs from chlor(am)ination of PBS were also proposed. Combined AAs, such as PBS, were proved to be important precursors of DBPs during disinfections. Polymyxin B Sulfate (PBS), like free amino acids, could be the precursor of typical C-DBPs (CF) and N-DBPs (DCAN, TCAN and TCNM) during chlorination and chloramination. Shanghai Education Development Foundation 10.13039/501100003024 No. 18CGB11 Shanghai Municipal Education Commission 10.13039/501100003395 No. 18CGB11 pubstatusPaginated Article ==== Body pmc1 Introduction Chlorination and chloramination are the two most widely used disinfection methods in water treatment, and can effectively remove many pathogens like bacteria and viruses and guarantee the biological safety of drinking water.1 However, various kinds of disinfection by-products (DBPs) have been discovered during disinfection processes since early 1970s.2 DBPs are mainly divided into two classes, the carbonaceous disinfection by-products (C-DBPs) and the nitrogenous disinfection by-products (N-DBPs), in subsequent studies.3 Trihalomethanes (THMs, including chloroform (CF), bromodichloromethane (BDCM), dibromochloromethane (DBCM), and bromoform (BF)), haloacetic acids, halogenated aldehydes and halofuranone are common C-DBPs, and N-DBPs mainly include haloacetonitriles (such as dichloroacetonitrile, DCAN and trichloroacetonitrile, TCAN), halonitromethanes (such as trichloronitromethane, TCNM), haloacetamides, cyanogen halides and nitrosamines.4 DBPs, especially N-DBPs were proved to exhibit great cytotoxicity and genotoxicity to animals and human beings, which aroused increasing attention and research.5 A proven effective method to control the formation of DBPs was to remove the precursors of DBPs (DBPPs).6,7 Dissolved organic matters (DOM) were the major DBPPs. Specifically, N-DBPs were mainly formed from dissolved organic nitrogen (DON), which accounted for about 10% of the total DOM.8,9 Among DON, amino acids (AAs) are the earliest and most studied precursors of C-DBPs and N-DBPs, which reported to account for 20–75% of total DON in runoffs.10,11 AAs occurs in two forms in natural water, free AAs and combine AAs (dipeptide, polypeptide, protein). Most of previous studies focused on free AAs, which however constitute only averaging 5.9% on a molar basis of total AAs in natural water.12 Thus, it is necessary to investigate the formation of DBPs from combined AAs during disinfections.13 Another relevant factor is that source waters were increasingly polluted by chemical residues and discharges, and the largest amounts of chemical contaminants were pharmaceuticals and personal care products (PPCPs).14 Among PPCPs, antibiotics are a class of drugs widely used and present worldwide. Particularly, polypeptide antibiotics are thought to be high likely to form C-DBPs and N-DBPs during disinfections, due to being a type of combined AAs. As a typical polypeptide antibiotic, polymyxin B sulfate (PBS) has been detected in wastewater15 and was selected as the target contaminant in this experiment. The main characteristics of PBS are listed in Table 1. Important chemical/physical properties of PBS Name Formula Chemical structure Molecule weight (g mol−1) CAS number Solubility in water (mg mL−1) Log P Polymyxin B sulfate (PBS) C55H96N16O13·2H2SO4 1385.61 1405-20-5 50 2.55 The conventional water treatment process cannot significantly remove DBPs or DBPPs.7 Hence, deep water treatment processes, including advanced oxidation, adsorption, membrane technology, biological treatment, and combination processes such as ozonation and biologically activated carbon process, have been widely studied.16–18 To the authors' knowledge, there have been no examinations of DBPs formation potential for PBS. Based on the analysis of molecular structure of PBS, it is likely that typical C-DBPs (THMs) and N-DBPs (DCAN, TCAN and TCNM) could be formed during chlorination and chloramination of PBS,6,13 which was demonstrated by a pre-experiment. The objective of this study was to evaluate the yields of C-DBPs (THMs) and N-DBPs (DCAN, TCAN and TCNM) during chlorination and chloramination of PBS. Batch experiments were carried out to investigate the impacts of major parameters, including disinfectant dose, contact time, solution pH, temperature, bromide concentration and chloramination mode. And formation pathways of different DBPs from chlor(am)ination of PBS were also proposed. 2 Experimental 2.1 Chemicals All experimental chemicals were at least analytical reagent (AR). PBS was supplied by Aladdin Industrial Inc. (Shanghai, China). Sodium hypochlorite solution (NaClO, free chlorine >5%), ammonium chloride (NH4Cl), sodium thiosulfate (Na2S2O3) and other chemicals were purchased from Sinopharm Chemical Reagent Co., Ltd (Shanghai, China). Standard solutions of different DBPs were obtained from Sigma-Aldrich (St Louis, Missouri, USA). Ultrapure deionized water, obtained from a Millipore Milli-Q Academic Ultra Pure Water Purification System (resistivity >18.2 MΩ, Billerica, MA, USA), was used to prepare all solutions in batch experiments. 2.2 Experimental procedure The chlor(am)ination experiments were carried out in full and sealed 40 mL amber glass bottles under dark condition (experimental procedure see Fig. S1 in ESI†). The concentration of PBS ([PBS]) was kept at 10 μM. NaOCl was diluted to prepare free chlorine stock solutions. Monochloramine (NH2Cl) solutions were made by mixing equal volumes of NaOCl and NH4Cl at a weight ratio of 4 mg L−1 Cl2 to 1 mg L−1 N–NH4+.19 Free chlorine and total chlorine (as Cl2, mg L−1) were measured by DPD-FAS titration before used.13 The residual disinfectant was rapidly quenched set times by Na2S2O3 at twice molarity to initial chlorine or monochloramine concentration.20 All DBPs tests were performed in at least duplicate. The common condition for disinfection experiments was summarized as: chlorine or monochloramine dose as [Cl2]/[PBS] = 50, contact time as t = 24 h, pH = 7, temperature as T = 25 °C, and without bromide. Each batch experiments varied one parameter from the common condition. Based on the practical conditions of disinfection processes in water treatment, the parameters were set as follows:13 chlorine or monochloramine dose ([Cl2]/[PBS] was 10, 20, 30, 40, 50, 70 and 100), contact time (t was 0.5 h, 1 h, 2 h, 3 h, 6 h, 12 h, 24 h, 48 h and 72 h), pH (5, 6, 7, 8 and 9), temperature (T was 15 °C, 25 °C, and 35 °C), bromide concentration ([Br−]/[Cl2] was 0.01, 0.02, 0.05, 0.1, 0.2, and 0.5). Different chloramine application modes were also evaluated: (1) “chloramination” (added with preformed monochloramine); (2) “preammonification” (added free chlorine to premixed solutions containing PBS and NH4Cl); (3) “prechlorination” (added NH4Cl after 5 min, 10 min and 15 min chlorination of PBS).6,20 2.3 Analytical methods The concentrations of free chlorine and total chlorine (Cl2, mg L−1) were quantified by a Pocket Colorimeter™ II Chlorine (HACH). The yields of THMs, DCAN, TCAN, and TCNM were determined based on USEPA method 524.2, by using a Purge – Trap sample concentrator (Eclipse 4660, OI, USA) and a Gas Chromatograph – Mass Spectrometry (GC-MS, QP2010, Shimadzu, Japan). The injector temperature was set at 190 °C. The column oven temperature was held at 35 °C for 10 min, increased to 72 °C at 7 °C min−1 and held for 1 min, and then increased to 200 °C at 40 °C min−1 and held for 1 min. The ion source temperature was set at 200 °C. Detection limits for CF, BDCM, DBCM, BF, DCAN, TCAN, and TCNM were 0.061, 0.071, 0.087, 0.095, 0.10, 0.15, and 0.25 μg L−1, respectively. The bromine substitution factor (BSF) was introduced to evaluate the proportion of the DBPs which partially or totally brominated, and BSF(THMs) can be calculated as eqn (1).1 3 Results and discussion 3.1 Effect of disinfectant dose Fig. 1(a) shows DBPs formation of PBS at different chlorine dosages. As chlorine dose increased, the yield of CF increased gradually, while the concentration of DCAN decreased continuously, though DCAN kept as the main product. Specifically, DCAN concentrations were 28.84 μg L−1 and 20.20 μg L−1 when [Cl2]/[PBS] was 10 and 100, respectively. The difference between formation of CF and DCAN is mainly because of their different stability, and the presence of residual chlorine was found to accelerate the hydrolytic degradation of DCAN.21 The yield of TCAN increased firstly and then decreased, which reached a maximum of 10.31 μg L−1 when [Cl2]/[PBS] = 70. This was mainly due to the instability of TCAN, and it exhibited the result under combined action of TCAN formation and hydrolysis. The regularity of TCNM formation was the same as that of CF. Fig. 1 Effect of chlorine (a) or monochloramine (b) dose on DBPs formation from PBS and consumption of disinfectants (c), as chlorine concentration. It can be seen from Fig. 1(b) that the effect of monochloramine dose on DBPs formation showed different characteristics. The yields of CF, DCAN and TCNM all increased with disinfectant dosage (the same trend as previous studies22,23) and CF could catch up with DCAN during chloramination. Particularly, TCNM only can be detected when [Cl2]/[PBS] ≥50. The yield of DCAN during chloramination was much lower than that of chlorination, and TCAN cannot be detected in chloramination no matter what the chloramine ratio was. This was mainly because of the weaker oxidation ability of chloramine, which showed slow rate to form free chlorine. However, yield of CF was even more than that during chlorination. In addition, the concentration of an intermediate product, dichloromethane (DCM) almost kept the same at all monochloramine doses, meaning that the conversion efficiency of CF was not one of major limiting factors. Fig. 1(c) shows the consumption of disinfectants after 24 h reaction. With the increase of chlorine dose, there was a large amount of chlorine residual in the solution, accelerating the further hydrolysis of DCAN and leading to the decrease trend of DCAN generation at last. In addition, as disinfectant dose increased, the consumption of monochloramine gradually exceeded that of chlorine. 3.2 Effect of contact time As shown in Fig. 2, the formations of different DBPs showed different characteristics as contact time increased, both for chlorination and chloramination. The yields of CF increased continuously with reaction time during chlor(am)ination. The eventually concentration of CF was close after 72 h reaction, 9.35 μg L−1 and 10.71 μg L−1 for chlorination and chloramination, respectively. However, the concentrations of DCAN increased firstly and then decreased, which was mainly due to the hydrolytic degradation of DCAN by residual disinfectant.24 Moreover, the inflection point occurred at 24 h for chlorination and 48 h for chloramination, which was because of the different rate to form free chlorine. TCAN only can be detected during chlorination, and it increased in 24 h reaction and changed little then. The yields of TCNM during chlor(am)ination showed similar trend with TCAN in chlorination, which may be because of the weak stability of TCNM as reported before.25 Fig. 2 Effect of chlorination (a) or chloramination (b) time on DBPs formation from PBS and consumption of disinfectants (c), as chlorine concentration. It can be seen from Fig. 2(c), the consumption of disinfectant showed similar trend with increase of DBPs in the initial stage. Particularly, it led to the fast formation of DCAN in 2 h chlorination and rapid increase of CF in 6 h chloramination. The consumption of free chlorine was larger than that of monochloramine within 3 h reaction, and then the consumption of monochloramine increased continuously to exceed that of free chlorine, which mainly due to the low stability of monochloramine. 3.3 Effect of pH During chlorination (Fig. 3(a)), the yield of CF gradually increased with increasing pH, the same trend as previous study.26 Particularly, it was 0.19 μg L−1 and 25.30 μg L−1 for pH = 5 and pH = 9, respectively. This is probably because that alkaline environment was conducive to the transformation of the intermediate products and eventually generation of CF.13 Conversely, the formation of DCAN and TCAN decreased obviously as pH increased, the same result with previous report.27 For DCAN, it was 32.61 μg L−1 and 2.46 μg L−1 at pH 5 and 8, respectively. What is more, TCAN cannot be detected at pH 8 and 9. This was mainly because the relatively unstable haloacetonitriles (including DCAN and TCAN) could easily hydrolyze under alkaline conditions.28 Fig. 3 Effect of pH on DBPs formation from PBS during chlorination (a) or chloramination (b). Fig. 3(b) shows that the yields of CF, DCAN and TCNM increased when pH changed from 5 to 6, and then decreased rapidly as pH increased. The yields of different DBPs were maximum when pH = 6, which was different from previous study in which the maximum yields came out at pH 7.6 Moreover, DCAN and TCNM were undetectable when pH reached 9 and 8, respectively. The formation trend of CF, DCAN and TCNM during chloramination was mainly resulted from two aspects: (1) the facilitation of N-DBPs hydrolysis under alkaline conditions; (2) the great impact of pH on the stability and effectivity of monochloramine.29 Under alkaline conditions, monochloramine keeps stable and is difficult to hydrolyze to release free chlorine. 3.4 Effect of temperature The yields of CF increased gradually with increasing temperature both in chlorination and chloramination (Fig. 4), the same with previous study.6 This was mainly due to the acceleration of formation rate of CF by increasing temperature. The formation of DCAN showed the same trend with CF during chlor(am)ination, which was, however, different from previous study.6 This was probably because that the improvement in generation of DCAN was greater than that in hydrolysis of DCAN, as temperature increased. Moreover, the yields of TCAN during chlorination also increased with temperature, owing to the same reason as DCAN. Fig. 4 Effect of temperature on DBPs formation from PBS during chlorination (a) or chloramination (b). However, the formation of TCNM presented different patterns. There was no TCNM detected at 15 °C in both disinfection processes. During chlorination, the yield of TCNM decreased as temperature increased from 25 °C to 35 °C. On the contrast, the concentration of TCNM at 35 °C was larger than that at 25 °C in chloramination. This was because TCNM had low stability and easy to hydrolyze. The higher temperature can both promote the generation and hydrolysis of TCNM, which had opposite influences on TCNM concentration and eventually led to different result. 3.5 Effect of bromide The effect of bromide on formation potential of THMs was investigated by adding different concentrations of bromide ([Br−]/[Cl2] = 0.01, 0.02, 0.05, 0.1, 0.2 and 0.5, respectively) during chlorination and chloramination. The results were shown in Fig. 5. Fig. 5 Effect of bromide on THMs formation from chlorination (a) or chloramination (b) of PBS and corresponding bromine substitution factor (BSF) (c). It can be seen that the total yields of THMs significantly increased with the increasing concentration of bromide, both in chlorination and chloramination, the same result with previous studies.20,23 The rate of increase in yield of THMs in chloramination was obviously higher than that in chlorination, which mainly due to the slower efficiency to release free chlorine. Particularly, the concentration of CF decreased continuously while BF had the opposite pattern, and the yields of BDCM and DBCM increased firstly and then decreased. For chlorination, the maximum values of BDCM and DBCM occurred at [Br−]/[Cl2] = 0.05 and 0.2, respectively. These all indicated the process of bromine substitution for chlorine. The BSF(THMs) was calculated in order to evaluate the effect of bromide ions. As shown in Fig. 5(c), BSF(THMs) increased with the increasing bromide concentration both for chlorination and chloramination, the same as previous report.30 BSF(THMs) for chlorination was higher than that for chloramination and the value gap gradually increased with increasing bromide concentration, which also because of the slower reaction rate of chloramine. Particularly, when [Br−]/[Cl2] = 0.5, BSF(THMs)s for chlorination and chloramination were 0.96 and 0.77, respectively. Moreover, the addition of bromide increased the overall toxicity of THMs, since brominated DBPs are more toxic than chlorinated DBPs. In addition, as concentration of bromide increased during chlorination or chloramination, the yield of DCAN decreased obviously and neither TCAN nor TCNM can be detected. 3.6 Effect of chloramination mode The effects of different chloramination modes (chloramination, prechlorination and preammonification) on the formation potential of DBPs were shown in Fig. 6. Fig. 6 Effect of different chloramination modes (A) chloramination; (B) preammonification; (C) prechlorination on DBPs formation from PBS. The yields of CF under different chloramination modes were almost the same. This was probably because that CF formation during chlorination and chloramination did not differ significantly when [Cl2]/[PBS] was 50 (2.97 μg L−1 and 3.69 μg L−1, respectively). The concentrations of DCAN were ranged as: 15 min prechlorination < chloramination < preammonification ∼10 min prechlorination < 5 min prechlorination. This was the result of a combination of two factors: the stronger generation capability of DCAN for chlorination; the rapid consumption of chlorine before the post-addition of amine reduced subsequent chloramine production. TCAN cannot be detected under all chloramination modes. This was mainly because that the prechlorination time was not long enough to form a stable amount of TCAN. TCNM only can be detected in chloramination. This was mainly due to the slow generation and low stability of TCNM. The large amount of free chlorine during prechlorination and during the initial reaction after post-addition of chlorine in preammonification accelerated the hydrolysis of TCNM. 3.7 Proposed formation pathways of DBPs from chlori(am)nation of PBS The molecule of PBS formed by the dehydration condensation of 11 amino acids with a carboxylic acid, containing a seven-membered ring and a long-chain branch. The molecular structure can be divided into three major categories: 5 short branched chains of hydrocarbon groups (Part A), 5 short branched chains containing free amino acids (Part B) and 11 peptide bonds plus linked hydrocarbon groups (Part C). Referring to previous studies,4,6,13,25,31 the proposed pathways for PBS to generate different DBPs during chlorination and chloramination are shown in Fig. 7 (except that TCAN was only detected in chlorination). Fig. 7 Proposed formation pathways of DBPs from PBS during chlori(am)nation. The hydrocarbon groups in Part A were easily oxidized by HOCl to form halogenated hydrocarbons, some of which eventually formed CF.6 The hydrocarbon groups in Part B and Part C can also be oxidized to form CF (Fig. 7, Pathway 2 (ref. 31) and Pathway 4 (ref. 13)) though relatively difficult, especially for Part C. For the –CH2–CH2–NH2 group in short chain branches (Part B), substitution and elimination reactions can be carried out under oxidation of HOCl to generate –CH2–C <svg xmlns="http://www.w3.org/2000/svg" version="1.0" width="23.636364pt" height="16.000000pt" viewBox="0 0 23.636364 16.000000" preserveAspectRatio="xMidYMid meet"><metadata> Created by potrace 1.16, written by Peter Selinger 2001-2019 </metadata><g transform="translate(1.000000,15.000000) scale(0.015909,-0.015909)" fill="currentColor" stroke="none"><path d="M80 600 l0 -40 600 0 600 0 0 40 0 40 -600 0 -600 0 0 -40z M80 440 l0 -40 600 0 600 0 0 40 0 40 -600 0 -600 0 0 -40z M80 280 l0 -40 600 0 600 0 0 40 0 40 -600 0 -600 0 0 -40z"/></g></svg> N and finally form DCAN and TCAN (Fig. 7, Pathway 1 (ref. 4)). In addition, the amino group in –CH2–CH2–NH2 group can also be oxidized by HOCl to form a nitro group (–NO2) and finally form TCNM (Fig. 7, Pathway 3 (ref. 25)). For the peptide bond moiety in the molecular structure of PBS (Part C), DBPs such as CF, DCAN, TCAN and TCNM can also be generated through different pathways under the function of HOCl (Fig. 7, Pathway 4 (ref. 13)). Considering the difficulty of each reaction, the oxidation of hydrocarbon groups in Part A was the major pathway to form CF, and dominant formation pathways of DCAN/TCAN and TCNM were Pathway 1 and Pathway 3, respectively. 4 Conclusions The study demonstrated that PBS, like free amino acids, could be the precursor of typical C-DBPs and N-DBPs during chlorination and chloramination. DCAN was the major DBPs from chlorination of PBS, while CF could catch up with DCAN during chloramination. TCAN could only be detected during chlorination. Different kinds of DBPs exhibited different characteristics as disinfectants dose or contact time increased, which was mainly due to the different oxidation ability of disinfectants and different stability of the products. Particularly, as chlorine dose increased for chlorination, the yields of CF and TCNM increased continuously, the concentration of DCAN decreased persistently, and the formation of TCAN increased firstly and then decreased. It was found that pH and temperature both had great effect on DBPs formation from chlorination and chloramination of PBS, which was mainly because of the effectiveness of disinfectants and the stability of DBPs. As for chloramination, the yields of different DBPs reached maximum values when pH = 6. The concentrations of CF and DCAN increased gradually with temperature. The yields and BSF of THMs gradually increased with the concentration of bromide, both in chlorination and chloramination. In order to reduce the formation of DBPs, the disinfection process can be adjusted to the following conditions: selecting chloramination instead of chlorination, maintaining neutral condition, appropriately lowing reaction temperature and removing bromine from solution. However, the impact of other substances in natural water bodies was not considered in batch experiments, and further research on the control method of DBPs formation is still needed to be carried out in the future. Author contributions Xingya Wei was in charge of the study and performed the preparation, experiments, analysis and writing. Bangjun Han and Renzheng Gu provided helpful suggestions and discussions. Weimin Geng participated in the collection and analysis of the data. Naiyun Gao helped revise the manuscript. All authors have given approval to the final version of the manuscript. Conflicts of interest There are no conflicts to declare. Supplementary Material RA-013-D3RA01981K-s001 This work was financially supported by “Chenguang Program” supported by Shanghai Education Development Foundation and Shanghai Municipal Education Commission (No. 18CGB11) and 2022 Research Programs supported by Shanghai Urban Construction Vocational College (No. cjky202233). ==== Refs Notes and references Tawk A. Deborde M. Labanowski J. Gallard H. Water Res. 2015 76 132 142 25813488 Rook J. J. Water Treat. Exam. 1974 23 234 243 Krasner S. W. Weinberg H. S. Richardson H. S. Pastor S. J. Chinn R. Sclimenti M. J. Onstad G. D. Thruston A. D. Environ. Sci. Technol. 2006 40 7175 7185 17180964 Bond T. Huang J. Templeton M. R. Graham N. Water Res. 2011 45 4341 4354 21705040 Muellner M. G. Wagner E. D. McCalla K. Richardson S. D. Woo Y. T. Plewa M. J. Environ. Sci. Technol. 2007 41 645 651 17310735 Wei X. Gao N. Li C. J. Environ. Chem. Eng. 2017 5 4397 4405 Krasner S. W. Philos. Trans. R. Soc., A 2009 367 4077 4095 Lee W. Westerhoff P. Croue J. P. Environ. Sci. Technol. 2007 41 5485 5490 17822121 Shah A. D. Mitch W. A. Environ. Sci. Technol. 2012 46 119 131 22112205 Trehy M. L. Yost R. A. Miles C. J. Environ. Sci. Technol. 1986 20 1117 1122 Frank H. Patrick S. Peter W. Hannes F. Biogeochemistry 2000 50 137 161 Dotson A. Westerhoff P. J. - Am. Water Works Assoc. 2009 101 101 115 Li C. Gao N. Chu W. Bond T. Wei X. Chem. Eng. J. 2017 307 487 495 Mompelat S. Bot B. L. Thomas O. Environ. Int. 2009 35 803 814 19101037 Wang N. Gao J. Wang Q. Xiao S. Zhuang G. J. Hazard. Mater. 2022 431 128616 35359112 Guzzella L. Feretti D. Monarca S. Water Res. 2002 36 4307 4318 12420936 Zhang L. Ma P. Dai L. Li S. Yu W. Guan J. Catal. Sci. Technol. 2021 11 3834 3844 Zhang L. Li X. Chen S. Guan J. Guo Y. Yu We. Catal. Commun. 2023 176 106627 Mitch W. A. Sedlak D. L. Environ. Sci. Technol. 2002 36 588 595 11878371 Yang X. Shang C. Westerhoff P. Water Res. 2007 41 1193 1200 17270234 Reckhow D. A. Platt T. L. MacNeill A. L. McClellan J. N. J. Water Supply: Res. Technol.-AQUA 2001 50 1 13 Sun X. Che L. Lu Y. Sun L. Zhao S. Water Sci. Technol.: Water Supply 2015 15 142 149 Wang F. Gao B. Ma D. Li R. Sun S. Yue Q. Wang Y. Li Q. Environ. Sci. Pollut. Res. 2016 23 1576 1583 Zhang H. Qu J. LIU H. Wei D. J. Environ. Sci. 2009 21 54 61 Yang X. Shen Q. Guo W. Peng J. Liang Y. Chemosphere 2012 88 25 32 22425029 Singh K. P. Rai P. Pandey P. Sinha S. Environ. Sci. Pollut. Res. 2012 19 113 127 Chu W. Yao D. Deng Y. Sui M. Gao N. J. Hazard. Mater. 2017 327 153 160 28064143 Glezer V. Harris B. Tal N. Iosefzon B. Lev O. Water Res. 1999 33 1938 1948 Jafvert C. T. Valentine R. L. Environ. Sci. Technol. 1992 26 577 586 Bond T. Huang J. Graham N. J. D. Templeton M. R. Sci. Total Environ. 2014 470 469 479 24176694 Yang X. Shang C. Lee W. Westerhoff P. Fan C. Water Res. 2008 42 2329 2339 18222525
PMC010xxxxxx/PMC10352705.txt
==== Front Korean J Clin Oncol Korean J Clin Oncol Korean Journal of Clinical Oncology 1738-8082 2288-4084 Korean Society of Surgical Oncology 37449395 10.14216/kjco.23004 kjco-19-1-18 Original Article Impact of Nrf2 overexpression on cholangiocarcinoma treatment and clinical prognosis https://orcid.org/0000-0002-3565-6064 Lee Huisong 1 https://orcid.org/0000-0001-9894-7603 Min Seog Ki 1 https://orcid.org/0000-0001-8772-9686 Cho Min-Sun 2 https://orcid.org/0000-0002-7975-2672 Lee Hyeon Kook 1 1 Department of Surgery, Ewha Womans University College of Medicine, Seoul, Korea 2 Department of Pathology, Ewha Womans University College of Medicine, Seoul, Korea Correspondence to: Hyeon Kook Lee, Department of Surgery, Ewha Womans University College of Medicine, 1071 Anyangcheon-ro, Yangcheon-gu, Seoul 07985, Korea, Tel: +82-2-2650-5694, Fax: +82-2-2644-7984, E-mail: [email protected] 6 2023 30 6 2023 19 1 1826 17 4 2023 1 6 2023 19 6 2023 Copyright © 2023 Korean Society of Surgical Oncology 2023 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. Purpose Nrf2 regulates antioxidant protein expression and protects against drug toxicity and oxidative stress, whereas Keap1 controls Nrf2 activity. The Keap1-Nrf2 pathway affects the prognosis of various cancers, however, its effect on cholangiocarcinoma chemoresistance and prognosis remains unclear. This study aimed to determine whether the Keap1-Nrf2 pathway affects chemoresistance and prognosis of distal cholangiocarcinoma. Methods We investigated the correlation between Nrf2 and Keap1 expression and clinical characteristics and prognosis in 91 patients with distal cholangiocarcinoma who underwent curative surgery. Immunohistochemical staining was performed on paraffin blocks using primary antibodies against Nrf2 and Keap1. The relationship between Keap1 and Nrf2 protein expression levels, and clinical characteristics and prognosis was examined. Results Nrf2 expression was not associated with overall survival in patients who did not receive adjuvant chemotherapy (P=0.994). Among patients receiving adjuvant chemotherapy, the Nrf2 low expression group had a significantly longer median overall survival than the Nrf2 high expression group in Kaplan-Meier survival analysis (P=0.019). In multivariate analysis, high expression of Nrf2 was confirmed as an independent poor prognostic factor in the group receiving adjuvant chemotherapy (P=0.041). Conclusion This study suggests that Nrf2 overexpression reduces the efficacy of adjuvant chemotherapy in distal cholangiocarcinoma. Keap1 protein Nrf2 protein Cholangiocarcinoma Adjuvant chemotherapy Pancreatoduodenectomy ==== Body pmcINTRODUCTION Adjuvant chemotherapy is a crucial treatment option for decreasing the risk of cancer recurrence after surgical resection for malignancies such as cholangiocarcinoma and pancreatic cancer. Despite the availability of several chemotherapeutic agents that target different steps of cancer cell growth and proliferation, the efficacy of adjuvant chemotherapy varies depending on the type of cancer. Biliary tract cancer, including cholangiocarcinoma, are less responsive to adjuvant chemotherapy than other malignancies. Randomized controlled trials and systematic reviews of previous studies have suggested that adjuvant chemotherapy has a limited effect on extrahepatic bile duct cancer [1–3]. However, despite its limited efficacy, adjuvant chemotherapy has been demonstrated to enhance survival in some patients with biliary tract cancer [4–6]. One potential factor that may affect chemotherapy resistance and prognosis in cholangiocarcinoma is the Keap1 (Kelch-like ECH-associated protein)-Nrf2 (NF-E2 p45-related factor 2) pathway. The Keap1-Nrf2 pathway acts as a double-edged sword in cancer cells. Nrf2 activity protects cells and makes them resistant to oxidative and electrophilic stresses, whereas elevated Nrf2 activity helps in cancer cell survival and proliferation [7]. Unregulated NRF2 confers high-level resistance to anticancer drugs and reactive oxygen species and directs cancer cells toward metabolic reprogramming [8]. Therefore, Nrf2 has been studied as a potential therapeutic target molecule in cancer. This pathway plays a role in the production of antioxidant proteins that protect cells from oxidative stress and drug toxicity. However, cancer cells can hijack Nrf2 activity to support their malignant growth, making Nrf2 a potential therapeutic target [9–13]. Previous studies have demonstrated that the Keap1-Nrf2 pathway affects the prognosis of various cancers, including gallbladder cancer and pancreatic cancer [14–16]. However, the effect of the Keap1-Nrf2 pathway on chemoresistance and prognosis of cholangiocarcinoma remains unclear. Therefore, this study aimed to investigate whether the Keap1-Nrf2 pathway affects chemoresistance and prognosis in distal cholangiocarcinoma. METHODS This study adhered to the principles outlined in the Declaration of Helsinki. The study was approved by the Institutional Ethics Review Board of Ewha Womans University Mokdong Hospital (IRB No. EUMC 2015-07-023) and the informed consent was waived. Patients Of the 111 patients who underwent curative surgery for distal cholangiocarcinoma at a single center, 91 were enrolled in the study, while 20 were excluded because of distant lymph node metastasis, double primary cancer, or palliative surgery. All the included patients were histologically confirmed to have cholangiocarcinoma. Given that the objective of this study was to ascertain the correlation between Nrf2 and Keap1 expression and clinical characteristics and prognosis, the clinical data of each patient were further scrutinized. The inquiry encompassed an array of variables, such as sex, age, comorbidities (diabetes and hypertension), American Society of Anesthesiologists score, height, weight, body mass index, surgery date, discharge date, type of surgery, status of minimally invasive surgery, status of resection margin, degree of cell differentiation, lesion size, TNM stage (American Joint Committee on Cancer 7th edition). Additionally, the number of affected lymph nodes, tumor-positive lymph nodes, lymphovascular invasion, perineural invasion, preoperative chemotherapy, postoperative chemotherapy, postoperative radiotherapy, recurrence, recurrence site, death, and survival time were examined. Surgery In this study, two surgical methods, bile duct segmental resection and pancreatoduodenectomy, were used to treat distal cholangiocarcinoma [17]. Bile duct segmental resection involves the removal of the affected segment of the bile duct, along with regional lymph node dissection. This procedure is typically used when the tumor is localized to a small area of the bile duct and has not spread to nearby organs. On the other hand, pancreatoduodenectomy, also known as the Whipple procedure, is a more extensive surgery that involves the removal of the head of the pancreas, the duodenum, the gallbladder, and the bile duct, along with regional lymph node removal. This procedure is typically used when the tumor has spread to intrapancreatic bile duct or nearby organs. The choice of surgery depends on various factors, including the location and extent of the tumor and the overall health of the patient. We administered adjuvant chemotherapy and radiation therapy to the selected patients after surgery. The primary indication for these treatments was lymph node metastasis, followed by consideration of the patient’s age and overall health status, including underlying comorbidities. If a patient refused cancer treatment or radiation therapy, it was not administered. Immunohistochemistry stain Surgically resected specimens were used to create paraffin blocks containing multiple cores from both tumor and normal tissue. Immunohistochemical staining was performed on these blocks using DAKO Autostainer Plus (Agilent Technologies) and primary antibodies against Nrf2 and Keap1. Paraffin sections were deparaffinized using xylene and rehydrated with graded alcohol. Antigen retrieval was achieved by boiling the sections in 0.1 M citric acid buffer (pH 6.0) for 10 minutes using a decloaking chamber (Biocare Medical). Peroxidase was blocked for 10 minutes using peroxidase-blocking solution. The REAL EnVision Detection System, Peroxidase/DAB+ (K5007; DAKO) was used to visualize the staining. The primary antibodies used were anti-Nrf2 (sc-722; Santa Cruz Biotech, 1:100 dilution) and anti-Keap1 (10503-2-AP; ProteinTech, 1:200). All the antibodies were incubated for 30 minutes at room temperature [18]. Statistics The degree of expression for each protein was scored by two experts who read the pathological slides for immunohistochemistry staining, and any discrepancies were resolved through consensus discussion. A 4-value intensity score (ranging from 0–3 positive) and the percentage of reactivity extent were used to determine the expression level of Keap1 and Nrf2. The respective values were multiplied together, which resulted in a possible score range from 0 to 300. This range was then categorized into two distinct levels of expression using a threshold score of 100. Scores below this threshold (<100) were classified as low or absent expression, while scores equal to or above this threshold (≥100) were classified as high expression. An analysis was then carried out to explore the relationship between the expression levels of Keap1 and Nrf2 and various clinical data. Categorical variables are expressed as numbers (percentages), while continuous variables are expressed as medians (ranges). The chi-square test and Mann-Whitney U test were used to compare categorical and continuous variables, respectively. Survival was calculated using the Kaplan-Meier method and Cox proportional hazard model from the date of surgical treatment, and differences in survival were assessed using the log-rank test. Statistical significance was set at P<0.05. RESULTS Demographics A total of 91 patients with distal cholangiocarcinoma who underwent curative surgery between January 2000 and December 2012 were included in this study. The median age of the patients was 69 years (range, 47–88 years), and the median follow-up period was 23.9 months (range, 3.6–176.2 months). We have categorized the overall patient demographics based on the administration of adjuvant chemotherapy. As a result, there was a statistically significant difference in the presence of lymph node metastasis between the group that received adjuvant chemotherapy and the group that did not. Additionally, the group that did not receive adjuvant chemotherapy was found to have a higher average age. This aligns with our institution’s approach, as mentioned in the methodology, of administering adjuvant chemotherapy to relatively younger patients who have lymph node metastasis or are in a good overall condition. There were no significant differences observed between the two groups in other variables (Table 1). And we summarized the differences among patient groups based on the expression levels of Nrf2 and Keap1 in Tables 2 and 3, respectively. The patients’ demographic characteristics showed that 31 patients belonged to the Keap1 low expression group, 60 patients belonged to the Keap1 high expression group, 51 patients belonged to the Nrf2 low expression group, and 40 patients belonged to the Nrf2 high expression group. The association between Keap1 and Nrf2 expression was not statistically significant (P=0.868). None of the patients had received neoadjuvant chemotherapy. Postoperative radiotherapy was not administered to any patients in the Keap1 low expression group, while nine (13%) patients in the Keap1 high expression group received radiotherapy (P=0.025). However, no significant differences were observed in sex, age, tumor stage, cell differentiation, adjuvant chemotherapy, resection margin status, operation time, or hospital stay (Table 2). Based on the hypothesis of our study, we assumed that the degree of Nrf2 expression could potentially impact the effectiveness of chemotherapy. Therefore, we categorized patients according to whether they had undergone adjuvant chemotherapy, and then compared the demographics while differentiating the differences in Nrf2 expression. As a result, no statistically significant differences were found in any of the categories (Table 3). There was no predetermined regimen for the chemotherapy, and it was administered by an oncology specialist. The types of chemotherapy varied greatly, with four patients receiving oral chemotherapy such as Xeloda, five patients receiving 5-fluorouracil (5-FU) and cisplatin, three patients treated with 5-FU and leucovorin, six patients administered with gemcitabine, one patient with cisplatin, one patient with epirubicin, and three patients treated at other hospitals for which the regimen was unknown. Given the variety of chemotherapy regimens, it was not possible to conduct an analysis of the differences in Kea1 and Nrf2 expression for each regimen. Immunohistochemistry Immunohistochemistry staining with Keap1 and Nrf2 specific antibodies was performed successfully for all cases by an experienced pathologist. Staining intensity and reactivity extension values were evaluated by two experts, and a consensus was reached. Cases with multiplication result of intensity score and reactivity extension value less than 100 were classified as having low or absent Keap1 and Nrf2 expression, whereas cases with multiplication result score of 100 or more were categorized as having high Keap1 and Nrf2 expression (Fig. 1). Survival For patients who did not receive adjuvant chemotherapy, there was no significant difference in overall survival between the Keap1 low expression group and Keap1 high expression group or between the Nrf2 low expression group and Nrf2 high expression group. For patients who received adjuvant chemotherapy, the median overall survival was significantly longer for the Nrf2 low expression group (31.3 months) than for the Nrf2 high expression group (21.1 months) (P=0.019) (Fig. 2). However, there was no significant difference between the Keap1 low and Keap1 high expression groups. On multivariate analysis, Nrf2 high expression was found to be an independent prognostic factor in patients with distal cholangiocarcinoma who received adjuvant chemotherapy (Table 4). DISCUSSION The Keap1-Nrf2 pathway is a regulatory system that modulates the activity of Nrf2, a protein responsible for controlling antioxidant expression and offers protection against drug toxicity and oxidative stress. The Keap1 protein governs Nrf2 activity, and the Keap1-Nrf2 pathway has been demonstrated to influence the prognosis of various cancers [19]. And there are bright and dark sides of Keap1-Nrf2 pathway. While it plays a protective role in normal cells by producing antioxidative enzymes and preventing malignant transformation, cancer cells often exploit this pathway to protect themselves against the oxidative stress of chemotherapy and support their malignant growth (Fig. 3) [20]. In cholangiocarcinoma, the effect of this pathway on chemoresistance and prognosis remains unclear. Nonetheless, studies have indicated that unregulated Nrf2 provides high-level resistance to anticancer drugs and reactive oxygen species, steers cancer cells toward metabolic reprogramming [10]. Keap1 deletion hastens mutant K-ras/p53-driven cholangiocarcinoma, emphasizing the complex interplay between this pathway and carcinogenesis [21]. Cholangiocarcinoma is linked to a worse prognosis compared to other cancers [22]. Prognostic factors for cholangiocarcinoma include location, margin status, vascular invasion, lymph node metastases, extension to the gallbladder, histology, gender, and serum albumin and bilirubin levels [4]. Among these factors, poor response to adjuvant chemotherapy is a primary contributor to the dismal prognosis observed in cholangiocarcinoma [2,23]. However, recent randomized controlled trials have shown that adjuvant chemotherapy can improve relapse-free survival in biliary tract cancer patients, including those with cholangiocarcinoma [24]. The findings of the present study indicated a correlation between Nrf2 overexpression and poor prognosis in patients with distal cholangiocarcinoma, with Nrf2 overexpression emerging as a significant predictor of chemoresistance. Furthermore, Nrf2 has been identified as a potential factor influencing the therapeutic efficacy of adjuvant chemotherapy following surgery. Notably, Nrf2 has been acknowledged as a potential cancer treatment target in other carcinomas, such as glioblastoma. However, the Keap1-Nrf2 pathway is crucial in normal cells, and overcoming the challenges associated with therapeutically targeting this pathway presents a considerable obstacle [25]. With advances in genetic analysis technology, precision medicine has emerged as a promising approach to the treatment of various carcinomas. However, cholangiocarcinoma presents unique challenges, owing to its heterogeneity and, diverse gross and histological features. While applying precision medicine to cholangiocarcinoma is considered difficult, recent efforts have been made to identify and target specific mutations in this cancer [26,27]. Nrf2 is a transcription factor that responds to environmental stimuli, primarily oxidative stress. Under oxidative stress, Keap1 dissociates from Nrf2, allowing Nrf2 to migrate to the nucleus and promote the production of cytoprotective enzymes. When a chemotherapeutic agent enters the cell, Keap1 dissociates from Nrf2, which fosters the translation of antioxidant and drug-transporting enzymes. This mechanism can inhibit various chemotherapeutic agents, including 5-fluorouracil and doxorubicin [12,25]. Previous studies have reported Nrf2 activity rates of 25%, 53%, and 61% in pancreatic and gallbladder cancers, respectively [14,15,28]. These rates are similar to those found in the present study. Translational research, aimed at bridging the gap between fundamental scientific discoveries and clinical applications, serves as a crucial component in understanding and addressing chemoresistance in cholangiocarcinoma. This study suggests that the Keap1-Nrf2 pathway may play a clinically significant role in cholangiocarcinoma development. The results of this study demonstrated that Nrf2 overexpression is associated with poor prognosis in patients with distal cholangiocarcinoma. Consequently, targeting Nrf2 may enhance the chemotherapy efficacy in cholangiocarcinoma patients. These findings could contribute to the development of novel treatments targeting the Keap1-Nrf2 pathway in cholangiocarcinoma, potentially improving patient outcomes and survival rates. Recent studies have proposed that inhibiting Nrf2 expression in vitro can augment chemotherapy effectiveness, underscoring the importance of this study as it suggests a novel approach to cholangiocarcinoma treatment [29,30]. By identifying patients who may benefit from effective chemotherapy or boost chemotherapeutic effects through Nrf2 suppression, patient survival rates can be improved. However, it is essential to note that the study’s sample size was limited, and various chemotherapy regimens were not investigated. In this study, only approximately one-fourth of the patients received anticancer treatment. However, it is worth noting that recent studies, such as the BILCAP trial, have reported the effectiveness of anticancer treatments, leading to an increasing emphasis on their use [6]. As the field of chemotherapy for cholangiocarcinoma continues to evolve and treatment options become more established, it is anticipated that future research can be conducted with more sophisticated study designs. Furthermore, there is a possibility of subjective evaluation of immunohistochemical staining results by researchers. Additionally, due to their diverse nature, antioxidative enzymes, which are products of the Keap1-Nrf2 pathway, were not analyzed in this study, complicating the assessment. Further studies considering these limitations are necessary to corroborate the findings of this study. In conclusion, recent research on the Keap1-Nrf2 pathway has demonstrated the feasibility of selecting effective chemotherapy based on Nrf2 expression levels in patients with various cancers, indicating a new direction for carcinoma treatment by selecting patients for effective chemotherapy or by enhancing the chemotherapeutic effect through Nrf2 inhibition. This study revealed that Nrf2 expression in distal cholangiocarcinoma was similar to that in other cancers, confirming the possibility that Keap1 and Nrf2 expression is associated with chemotherapy effects. Fig. 1 Immunohistochemistry results of Keap1 and Nrf2 proteins (×200), with high magnification images in small squares (×400). (A) Negative Keap1 expression. (B) Low Keap1 expression. (C) High Keap1 expression. (D) Negative Nrf2 expression. (E) Low Nrf2 expression. (F) High Nrf2 expression. Fig. 2 The Kaplan-Meyer overall survival (OS) analysis based on Keap1 and Nrf2 expression levels in patients with distal cholangiocarcinoma. (A, B) OS according to the expression level of Keap1 without and with adjuvant chemotherapy, respectively. (C, D) OS according to the expression level of Nrf2 without and with adjuvant chemotherapy, respectively. Fig. 3 Keap1-Nrf2 pathway modulates the production of antioxidant enzymes. There are bright and dark sides of the Keap1-Nrf2 pathway. While it plays a protective role in normal cells by producing antioxidative enzymes and preventing malignant transformation, cancer cells often exploit the pathway to protect themselves against the oxidative stress of chemotherapy and support their malignant growth. Table 1 Characteristics of patients according to adjuvant chemotherapy Characteristic Adjuvant chemotherapy (n=23) No adjuvant chemotherapy (n=68) P-valuea) Nrf2 expression, No. (%)  Low or absent 12 (52) 39 (57) 0.665  High 11 (48) 29 (43) Keap1 expression, No. (%)  Low or absent 8 (35) 23 (34) 0.933  High 15 (65) 45 (66) Sex, No. (%)  Male 15 (65) 47 (69) 0.729  Female 8 (35) 21 (31) Age (yr), median (range) 64 (47–82) 70 (48–88) 0.078 Operation type, No. (%)  Pancreatoduodenectomy 19 (83) 54 (79) 0.739  CBD segmental resection 4 (17) 14 (21) T stage, No. (%)  T1 or T2 12 (52) 36 (53) 0.949  T3 or T4 11 (11) 32 (47) N stage, No. (%)  N0 9 (39) 46 (68) 0.016  N1 14 (61) 22 (32) Cell differentiation, No. (%)  WD 4 (17) 13 (19) 0.809  MD or PD 19 (83) 53 (81) Radiation therapy, No. (%)  Yes 3 (13) 6 (9) 0.558  No 20 (87) 62 (91) Resection margin, No. (%)  R0 22 (96) 58 (85) 0.188  R1 1 (4) 10 (15) Operation time (min), median (range) 465 (240–695) 452.5 (245–740) 0.841 Hospital duration (day), median (range) 20 (8–57) 20 (6–78) 0.421 CBD, common bile duct; WD, well differentiated; MD, moderately differentiated; PD, poorly differentiated. a) Chi-square test, Fisher exact test or Mann-Whitney U test. Table 2 Characteristics of patients according to the expression of Keap1 and Nrf2 Characteristic Keap1 low expression (n=31) Keap1 high expression (n=60) P-valuea) Nrf2 low expression (n=51) Nrf2 high expression (n=40) P-valuea) Nrf2 expression, No. (%) 0.868  Low or absent 17 (55) 34 (57)  High 14 (45) 26 (43) Keap1 expression, No. (%) 0.868  Low or absent 17 (33) 14 (35)  High 34 (67) 26 (65) Sex, No. (%) 0.676 0.735  Male 22 (71) 40 (67) 34 (67) 28 (70)  Female 9 (29) 20 (33) 17 (33) 12 (30) Age (yr), median (range) 70 (53–88) 69 (47–82) 0.633 70 (47–88) 67 (48–82) 0.597 Operation type, No. (%) 0.101 0.299  Pancreatoduodenectomy 28 (90) 45 (75) 43 (84) 30 (75)  CBD segmental resection 3 (10) 15 (25) 8 (16) 10 (25) T stage, No. (%) 0.549 0.642  T1 or T2 15 (48) 33 (55) 28 (55) 20 (50)  T3 or T4 16 (52) 27 (45) 23 (45) 20 (50) N stage, No. (%) 0.306 0.612  N0 21 (68) 34 (57) 32 (63) 23 (58)  N1 10 (32) 26 (43) 19 (37) 17 (43) Cell differentiation, No. (%) 0.239 0.493  WD 8 (26) 9 (15) 11 (22) 6 (15)  MD or PD 23 (74) 49 (82) 40 (78) 32 (80) Adjuvant chemotherapy, No. (%) 0.933 0.665  Yes 8 (26) 15 (25) 12 (24) 11 (28)  No 23 (74) 45 (75) 39 (76) 29 (73) Radiation therapy, No. (%) 0.025 0.999  Yes 0 9 (15) 5 (10) 4 (10)  No 31 (100) 51 (85) 46 (90) 36 (90) Resection margin, No. (%) 0.743 0.203  R0 28 (90) 52 (87) 47 (92) 33 (83)  R1 3 (10) 8 (13) 4 (8) 7 (18) Operation time (min), median (range) 460 (285–740) 455 (240–720) 0.738 460 (240–740) 430 (245–720) 0.323 Hospital duration (day), median (range) 23 (7–56) 20 (6–78) 0.511 19 (7–78) 22.5 (6–57) 0.181 CBD, common bile duct; WD, well differentiated; MD, moderately differentiated; PD, poorly differentiated. a) Chi-square test, Fisher exact test or Mann-Whitney U test. Table 3 Characteristics of patients according to expression of Nrf2 with or without chemotherapy Characteristic With chemotherapy Without chemotherapy Nrf2 low expression (n=15) Nrf2 high expression (n=8) P-valuea) Nrf2 low expression (n=47) Nrf2 high expression (n=21) P-valuea) Keap1 expression, No. (%) 0.999 0.541  Low or absent 5 (33) 3 (38) 17 (36) 6 (29)  High 10 (67) 5 (63) 30 (64) 15 (71) Sex, No. (%) 0.999 0.989  Male 10 (67) 5 (63) 32 (68) 15 (71)  Female 5 (33) 3 (38) 15 (32) 6 (29) Age (yr), median (range) 66 (47–82) 61 (5–77) 0.722 70 (52–88) 70 (48–82) Operation type, No. (%) 0.589 0.336  Pancreatoduodenectomy 13 (87) 6 (75) 39 (83) 15 (71)  CBD segmental resection 2 (13) 2 (75) 15 (17) 6 (29) T stage, No. (%) 0.400 0.951  T1 or T2 9 (60) 3 (38) 25 (53) 11 (52)  T3 or T4 6 (40) 5 (63) 22 (47) 10 (48) N stage, No. (%) 0.999 0.656  N0 6 (40) 3 (38) 31 (66) 15 (71)  N1 9 (60) 5 (63) 16 (34) 6 (29) Cell differentiation, No. (%) 0.999 0.739  WD 3 (20) 1 (13) 10 (21) 3 (14)  MD or PD 12 (80) 7 (88) 36 (77) 17 (81) Radiation therapy, No. (%) 0.999 0.363  Yes 2 (13) 1 (13) 3 (6) 3 (14)  No 13 (87) 7 (88) 44 (94) 18 (86) Resection margin, No. (%) 0.348 0.157  R0 15 (100) 7 (88) 42 (89) 16 (76)  R1 0 1 (13) 5 (11) 5 (24) Operation time (min), median (range) 455 (240–610) 500 (285–695) 0.723 450 (245–740) 455 (260–720) 0.947 Hospital duration (day), median (range) 19 (10–57) 21.5 (8–48) 0.628 20 (7–78) 21 (6–56) 0.590 CBD, common bile duct; WD, well differentiated; MD, moderately differentiated; PD, poorly differentiated. a) Chi-square test, Fisher exact test, or Mann-Whitney U test. Table 4 Univariate and multivariate analyses using Cox regression proportional hazard model of OS after adjuvant chemotherapy in distal bile duct cancer patients Univariate analysis of OS Multivariate analysis of OSa) HR (95% CI) P-value HR (95% CI) P-value Keap1 high expression 1.332 (0.453–3.921) 0.603 Nrf2 high expression 3.506 (1.174–10.476) 0.025 3.579 (1.056–12.134) 0.041 Pancreatoduodenectomy 1.371 (0.721–2.605) 0.336 Female sex 0.818 (0.278–2.402) 0.715 0.758 (0.207–2.775) 0.675 Age 0.994 (0.945–1.045) 0.816 1.014 (0.956–1.076) 0.636 T stage (T3 or T4) 1.339 (0.495–3.621) 0.565 Lymph node metastasis 1.127 (0.408–3.111) 0.818 Moderately or poorly differentiation 1.974 (0.445–8.760) 0.371 Radiation therapy 1.044 (0.294–3.705) 0.947 1.450 (0.368–5.720) 0.596 R1 resection 3.136 (0.377–26.090) 0.290 2.484 (0.215–28.717) 0.466 OS, overall survival; HR, hazard ratio; CI, confidence interval. a) Sex, age, radiation therapy and variables with P<0.3 by univariate analysis were included. No potential conflict of interest relevant to this article was reported. FUNDING None. ==== Refs REFERENCES 1 Neoptolemos JP Moore MJ Cox TF Valle JW Palmer DH McDonald AC Effect of adjuvant chemotherapy with fluorouracil plus folinic acid or gemcitabine vs observation on survival in patients with resected periampullary adenocarcinoma: the ESPAC-3 periampullary cancer randomized trial JAMA 2012 308 147 56 22782416 2 Wang ML Ke ZY Yin S Liu CH Huang Q The effect of adjuvant chemotherapy in resectable cholangiocarcinoma: a meta-analysis and systematic review Hepatobiliary Pancreat Dis Int 2019 18 110 6 30470543 3 Ebata T Hirano S Konishi M Uesaka K Tsuchiya Y Ohtsuka M Randomized clinical trial of adjuvant gemcitabine chemotherapy versus observation in resected bile duct cancer Br J Surg 2018 105 192 202 29405274 4 Wirasorn K Ngamprasertchai T Khuntikeo N Pakkhem A Ungarereevittaya P Chindaprasirt J Adjuvant chemotherapy in resectable cholangiocarcinoma patients J Gastroenterol Hepatol 2013 28 1885 91 23829232 5 Siebenhuner AR Seifert H Bachmann H Seifert B Winder T Feilchenfeldt J Adjuvant treatment of resectable biliary tract cancer with cisplatin plus gemcitabine: a prospective single center phase II study BMC Cancer 2018 18 72 29325521 6 Primrose JN Fox RP Palmer DH Malik HZ Prasad R Mirza D Capecitabine compared with observation in resected biliary tract cancer (BILCAP): a randomised, controlled, multicentre, phase 3 study Lancet Oncol 2019 20 663 73 30922733 7 Deshmukh P Unni S Krishnappa G Padmanabhan B The Keap1-Nrf2 pathway: promising therapeutic target to counteract ROS- mediated damage in cancers and neurodegenerative diseases Biophys Rev 2017 9 41 56 28510041 8 Taguchi K Yamamoto M The KEAP1-NRF2 system in cancer Front Oncol 2017 7 85 28523248 9 Huang CF Zhang L Ma SR Zhao ZL Wang WM He KF Clinical significance of Keap1 and Nrf2 in oral squamous cell carcinoma PLoS One 2013 8 e83479 24386210 10 No JH Kim YB Song YS Targeting nrf2 signaling to combat chemoresistance J Cancer Prev 2014 19 111 7 25337579 11 Sporn MB Liby KT NRF2 and cancer: the good, the bad and the importance of context Nat Rev Cancer 2012 12 564 71 22810811 12 Wang XJ Sun Z Villeneuve NF Zhang S Zhao F Li Y Nrf2 enhances resistance of cancer cells to chemotherapeutic drugs, the dark side of Nrf2 Carcinogenesis 2008 29 1235 43 18413364 13 Ren P Ye H Dai L Liu M Liu X Chai Y Peroxiredoxin 1 is a tumor-associated antigen in esophageal squamous cell carcinoma Oncol Rep 2013 30 2297 303 24009050 14 Lister A Nedjadi T Kitteringham NR Campbell F Costello E Lloyd B Nrf2 is overexpressed in pancreatic cancer: implications for cell proliferation and therapy Mol Cancer 2011 10 37 21489257 15 Solis LM Behrens C Dong W Suraokar M Ozburn NC Moran CA Nrf2 and Keap1 abnormalities in non-small cell lung carcinoma and association with clinicopathologic features Clin Cancer Res 2010 16 3743 53 20534738 16 Kim JH Bogner PN Ramnath N Park Y Yu J Park YM Elevated peroxiredoxin 1, but not NF-E2-related factor 2, is an independent prognostic factor for disease recurrence and reduced survival in stage I non-small cell lung cancer Clin Cancer Res 2007 13 3875 82 17606720 17 Kim N Lee H Min SK Lee HK Bile duct segmental resection versus pancreatoduodenectomy for middle and distal common bile duct cancer Ann Surg Treat Res 2018 94 240 6 29732355 18 Chang LC Fan CW Tseng WK Chen JR Chein HP Hwang CC Immunohistochemical study of the Nrf2 pathway in colorectal cancer: Nrf2 expression is closely correlated to Keap1 in the tumor and Bach1 in the normal tissue Appl Immunohistochem Mol Morphol 2013 21 511 7 23455180 19 Islam SS Qassem K Islam S Parag RR Rahman MZ Farhat WA Genetic alterations of Keap1 confers chemotherapeutic resistance through functional activation of Nrf2 and Notch pathway in head and neck squamous cell carcinoma Cell Death Dis 2022 13 696 35945195 20 Baird L Yamamoto M The molecular mechanisms regulating the KEAP1-NRF2 pathway Mol Cell Biol 2020 40 e00099 20 32284348 21 Nabeshima T Hamada S Taguchi K Tanaka Y Matsumoto R Yamamoto M Keap1 deletion accelerates mutant K-ras/p53-driven cholangiocarcinoma Am J Physiol Gastrointest Liver Physiol 2020 318 G419 27 31961719 22 Park CY Choi SH Kwon CI Cho JH Jang SI Lee TH What is the better surgical treatment option for recurrent common bile duct stones? Ann Surg Treat Res 2020 99 329 36 33304860 23 Horgan AM Amir E Walter T Knox JJ Adjuvant therapy in the treatment of biliary tract cancer: a systematic review and meta-analysis J Clin Oncol 2012 30 1934 40 22529261 24 Liotta L Quante M Adjuvant chemotherapy with Capecitabine as new standard for resected cholangiocarcinomas: a look at the BILCAP trial Z Gastroenterol 2018 56 839 40 29986358 25 Zhu J Wang H Fan Y Lin Y Zhang L Ji X Targeting the NF-E2-related factor 2 pathway: a novel strategy for glioblastoma (review) Oncol Rep 2014 32 443 50 24926991 26 Jain A Javle M Molecular profiling of biliary tract cancer: a target rich disease J Gastrointest Oncol 2016 7 797 803 27747093 27 Shiao MS Chiablaem K Charonian V Ngamphaiboon N Jinawath N Emergence of intrahepatic cholangiocarcinoma: how high-throughput technologies expedite the solutions for a rare cancer type Front Genet 2018 9 309 30158952 28 Shibata T Kokubu A Gotoh M Ojima H Ohta T Yamamoto M Genetic alteration of Keap1 confers constitutive Nrf2 activation and resistance to chemotherapy in gallbladder cancer Gastroenterology 2008 135 1358 68 18692501 29 Jiang ZY Lu MC You QD Nuclear factor erythroid 2-related factor 2 (Nrf2) inhibition: an emerging strategy in cancer therapy J Med Chem 2019 62 3840 56 30444366 30 Zhang B Ma Z Tan B Lin N Targeting the cell signaling pathway Keap1-Nrf2 as a therapeutic strategy for adenocarcinomas of the lung Expert Opin Ther Targets 2019 23 241 50 30556750
PMC010xxxxxx/PMC10352706.txt
==== Front Korean J Clin Oncol Korean J Clin Oncol Korean Journal of Clinical Oncology 1738-8082 2288-4084 Korean Society of Surgical Oncology 37449398 10.14216/kjco.23007 kjco-19-1-38 Case Report Supra-ampullary duodenectomy in a patient with positive distal resection margin after subtotal gastrectomy for gastric cancer: a case report https://orcid.org/0000-0003-3980-3639 Lee Kyung-Goo 1 https://orcid.org/0009-0006-0036-3794 Jeong Jin Ho 1 https://orcid.org/0009-0000-7708-1257 Joo Jong Eun 2 https://orcid.org/0000-0001-6565-0854 Kim Hyun Beom 3 1 Department of Surgery, Myongji Hospital, Hanyang University College of Medicine, Goyang, Korea 2 Department of Pathology, Myongji Hospital, Hanyang University College of Medicine, Goyang, Korea 3 Department of Radiology, Myongji Hospital, Hanyang University College of Medicine, Goyang, Korea Correspondence to: Kyung-Goo Lee, Department of Surgery, Myongji Hospital, Hanyang University College of Medicine, 55 Hwasu-ro 14beon-gil, Deokyang-gu, Goyang 10475, Korea, Tel: +82-31-810-5445, Fax: +82-31-969-0500, E-mail: [email protected] 6 2023 30 6 2023 19 1 3842 21 1 2023 7 6 2023 18 6 2023 Copyright © 2023 Korean Society of Surgical Oncology 2023 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. Resection margin involvement after curative intent resection for gastric cancer results in a poor prognosis and deprives the patient of the chance for a cure. Reoperation to achieve an R0 status should guarantee tolerable morbidity and achievement of negative margins. We performed laparoscopic distal gastrectomy with extracorporeal Billroth II reconstruction in a 56-year-old woman with gastric cancer following neoadjuvant chemotherapy. Scattered cancer cells were observed in the proximal and distal resection margins on immunohistochemical staining for cytokeratin. Two weeks postoperatively, remnant total gastrectomy and supra-ampullary duodenectomy were performed. Before reoperation, percutaneous transhepatic gallbladder drainage and angiocatheter placement outside the ampulla of Vater (AoV) via the cystic duct were performed to avoid pancreaticoduodenectomy and to obtain the maximal distal margin. Duodenal transection was performed 1 cm above the AoV. The resected duodenum was 4 cm in length. The patient had no postoperative complications and received adjuvant chemotherapy 1 month after the reoperation. Reoperation Margin Ampulla of Vater Stomach neoplasm Gastrectomy ==== Body pmcINTRODUCTION Resection margin (RM) involvement after curative intent resection for gastric cancer results in a poor prognosis and deprives the patient of the chance for cure. Although intraoperative frozen biopsy is a reliable and widely used method, the false-negative rate of that is 0.5% to 4.7% [1,2]. However, there are few cases of reoperation to achieve a negative RM because it should guarantee tolerable morbidity and achievement of negative margins [3]. Recently, we performed remnant total gastrectomy and supra-ampullary duodenectomy in a patient with positive proximal and distal RMs after distal gastrectomy for gastric cancer. Herein, we describe the perioperative procedure and management. Informed consent was obtained from the patient for being included in the study. CASE REPORT A 56-year-old woman with gastric cancer and no other relevant medical history visited the outpatient clinic. She had undergone three cycles of neoadjuvant chemotherapy with FLOT (docetaxel, oxaliplatin, fluorouracil, and leucovorin) regimen in another hospital 1 month ago. Esophagogastroduodenoscopy before chemotherapy had revealed an ulcerative lesion with signet ring cell carcinoma in the lower body of the stomach. Abdominal computed tomography (CT) and positron emission tomography-CT before chemotherapy had shown gastric wall thickening involving the lower body and antrum, with fluorodeoxyglucose uptake. Preoperative positron emission tomography-CT revealed that the previously noted hypermetabolic lesion in the gastric wall was nearly normalized. There was no evidence of lymph node or distant metastases (Fig. 1). Laparoscopic distal gastrectomy with D2 lymph node dissection and extracorporeal Billroth II reconstruction was performed. No enlarged lymph nodes or peritoneal seeding was observed. Proximal transection was performed 1 cm below the esophagogastric junction on the lesser curvature side, and distal transection was performed 2 cm distal to the pylorus. The gross proximal and distal margins were of 3.5 cm and 2 cm, respectively. Frozen biopsy revealed a negative RM at both ends (Fig. 2). The final pathological diagnosis was ypT3N2M0 with the negative result of washing cytology. The specimen was of Borrmann type IV cancer involving the whole stomach. Some single cancer cells were observed on immunohistochemical staining for cytokeratin in the proximal and distal RMs (Fig. 3). Two weeks postoperatively, open remnant total gastrectomy and supra-ampullary duodenectomy were performed. Before reoperation, percutaneous transhepatic gallbladder drainage (PTGBD) and angiocatheter placement outside the ampulla of Vater (AoV) via the cystic duct were performed to avoid pancreaticoduodenectomy (PD) and obtain the maximal distal margin (Fig. 4). The operative findings revealed moderate adhesion at the superior pancreatic border and gastrojejunostomy site. Duodenal transection was performed 1 cm above the AoV through palpation of the angiocatheter, following kocherization and detachment of the duodenum from the pancreas. Invagination of the duodenal stump followed the duodenal transection. The resected duodenum was 4 cm long (Fig. 5). Two Jackson-Pratt (JP) drains were inserted around the duodenal stump and behind the esophagojejunostomy. The final pathology revealed a negative RM at both ends, and there was no carcinoma in the resected duodenum. On a postoperative day 4, upper gastrointestinal series and fistulography by injecting contrast material through the angiocatheter revealed no anastomotic leakage. On postoperative day 8, abdominal CT was performed and no significant fluid collection was found. Accordingly, the JP drain placed behind the esophagojejunostomy and the angiocatheter were removed. On postoperative day 9, the JP drain around the duodenal stump was removed, and the patient was discharged without complications. The PTGBD was removed after clamping for 4 days. The patient received three cycles of adjuvant chemotherapy with the FLOT regimen 6 weeks after the initial operation. She was followed up every 3 months with carcinoembryonic antigen, carbohydrate antigen 19-9, and chest and abdominal CT. Recurrence with peritoneal seeding was detected 12 months later, and the patient has been receiving palliative chemotherapy. DISCUSSION On analyzing RMs following intraoperative frozen biopsy for gastric adenocarcinoma, the risk factors for false-negative results were found to be neoadjuvant chemotherapy, signet ring cell carcinoma, diffuse type, and Borrmann type IV carcinoma [1,2]. In our case, all of these risk factors were present. In initial esophagogastroduodenoscopy, there were no specific findings other than ulcerative lesion and fold thickening around the lesion in lower body greater curvature because the gross type was Borrmann type IV, in which cancer cells spread along the submucosa and muscle layer. This was supported by pathologic findings. We focused on the proximal RM intraoperatively because Borrmann type IV cancer often involves the entire stomach. Thus, we transected the stomach extracorporeally for palpation of wall thickening. Although it seemed that RM was less than 5 cm recommended in the guideline, we deferred the decision of RM to frozen section because there is no evidence about RM except utilization of frozen section [4]. However, false-negative results were observed at both ends. In most studies, the number of R1 cases is 2% to 8%, and reoperation was performed in few of them [3]. For patients with microscopically positive RMs, there is some debate about the survival benefit of reoperation to achieve R0 status in advanced stage although most authors recommended reoperation in the early stage [5]. Unfortunately, our patient had a ypT3N2M0 tumor, and recurrence was detected 12 months postoperatively. However, it is clear that reoperation is the only treatment option to achieve R0 status with a chance of cure. Therefore, clinicians should explore perioperative management, including optimal operation methods, to prevent severe complications and achieve R0 resection, especially in medically fit patients. With respect to gastric cancer with esophageal invasion, the abdominal-transhiatal approach is feasible if the length of the esophageal resection is within 6 cm. No survival benefit of extended aggressive surgery was observed for esophageal involvement of ≥3 cm in subcardia cancer [6]. In terms of gastric cancer with duodenal invasion, Kakeji et al. [7] suggested that the first portion of the duodenum is 4 to 5 cm in length, which is sufficient to resect 3 to 4 cm of the duodenum beyond the pylorus, and PD can be considered when the lesion extends >3 cm beyond the pyloric ring. However, the postoperative morbidity rate of PD was 23% to 74%, the pancreatic leak rate was 24.5%, and the mortality rate was 1.4% in the pooled analysis [8]. Yonemura et al. [9] reported no survival benefit of PD for tumors with duodenal invasion. On the other hand, Ajisaka et al. [10] concluded that curative resection including PD improved the prognosis of patients with duodenal invasion, except in nodal type duodenal invasion, which involves nodal metastatic lesions around the pancreatic head and is the longest duodenal invasion type with a mean length of 3.6 cm [10]. In our case, CT before reoperation revealed that the length from the duodenal stump to the AoV was approximately 3 cm, and accordingly, we planned further duodenal resection, not PD. However, it is difficult to guarantee no injury of AoV during detachment of the duodenum from the pancreas for maximal achievement of duodenal margin beyond 4 cm. This confidence is even more necessary in reoperation. Recently, Di Saverio et al. [11] introduced pancreas-sparing, ampulla-preserving D1–D2 duodenectomy for emergency treatment of major duodenal perforations. They performed cholecystectomy and tube placement from the cystic duct to the duodenal lumen outside the AoV for proximal duodenectomy with preservation of the AoV. We transformed this surgical technique to a preoperative intervention to avoid pancreatico-duodenectomy and obtain the maximal distal margin. A drainage catheter was placed in the duodenum outside the AoV via the cystic duct after PTGBD was attempted. However, the cystic duct was too thin for the drainage catheter to pass through in this patient. Finally, PTGBD and angiocatheter placement were performed in the duodenum outside the AoV via the cystic duct. These were useful not only for duodenectomy but also for postoperative management. Although gallstone formation is one of the common postoperative complications of gastric cancer surgery, cholecystectomy was not performed in this patient because the role of prophylactic cholecystectomy remains controversial [12,13]. The total length of the resected duodenum was 6 cm, which was sufficient for the free margin of tumors with duodenal invasion. In summary, supra-ampullary duodenectomy can be considered the most suitable treatment in terms of a chance for cure in patients with positive distal margin after subtotal gastrectomy for gastric cancer. Preoperative percutaneous catheter placement in the duodenum, outside the AoV, via the cystic duct, can aid re-excision to obtain sufficient margins and relieve the burden on the operator in terms of postoperative management. Fig. 1 Preoperative evaluation. (A) Esophagogastroduodenoscopy before neoadjuvant chemotherapy. An ulcerative lesion (arrow) in the lower body of the stomach was observed. (B) Coronal view of abdominal computed tomography (CT) following upper gastrointestinal series before neoadjuvant chemotherapy. The CT scan shows gastric wall thickening involving the lower body and antrum. Positron emission tomography-CT (C) before neoadjuvant chemotherapy and (D) before the operation. The previously observed hypermetabolic lesions in the gastric wall appeared nearly normalized. Fig. 2 Specimen of the stomach after subtotal gastrectomy following neoadjuvant chemotherapy. Fig. 3 Postoperative pathology of the resection margin. (A) Hematoxylin and eosin (H&E) staining of the proximal resection margin specimen (×100). Some atypical cells (arrow) are identified in fibrotic muscle layer on permanent section retrospectively, but it is not easy to realize poorly cohesive carcinoma cell infiltration on frozen section. (B) Cytokeratin staining of the proximal resection margin specimen (×40). Some neoplastic cells (arrow) showing positive reaction for cytokeratin on immunohistochemical stain are observed in submucosa and muscle layer. (C) H&E staining of the distal resection margin specimen (×100). Focal infiltration of neoplastic cells (arrow) is seen in duodenal muscular layer but distinctive tumor cells are not well identified in distal duodenal margin. (D) Cytokeratin staining of the distal resection margin specimen (×100). A few neoplastic cells (arrow) showing positive reaction for cytokeratin on immunohistochemical stain are observed in duodenal resection margin. Fig. 4 Percutaneous angiocatheter placement (arrow) in the duodenum, outside the ampulla of Vater, and via the cystic duct. Fig. 5 Specimen after supra-ampullary duodenectomy. (A) Duodenal transection was performed 1 cm above the angiocatheter at the ampulla of Vater level. (B) The resected duodenum is 4 cm long. No potential conflict of interest relevant to this article was reported. FUNDING None. ==== Refs REFERENCES 1 McAuliffe JC Tang LH Kamrani K Olino K Klimstra DS Brennan MF Prevalence of false-negative results of intraoperative consultation on surgical margins during resection of gastric and gastroesophageal adenocarcinoma JAMA Surg 2019 154 126 32 30422226 2 Spicer J Benay C Lee L Rousseau M Andalib A Kushner Y Diagnostic accuracy and utility of intraoperative microscopic margin analysis of gastric and esophageal adenocarcinoma Ann Surg Oncol 2014 21 2580 6 24806114 3 Aurello P Magistri P Nigri G Petrucciani N Novi L Antolino L Surgical management of microscopic positive resection margin after gastrectomy for gastric cancer: a systematic review of gastric R1 management Anticancer Res 2014 34 6283 8 25368226 4 Berlth F Kim WH Choi JH Park SH Kong SH Lee HJ Prognostic impact of frozen section investigation and extent of proximal safety margin in gastric cancer resection Ann Surg 2020 272 871 8 32833759 5 Morgagni P La Barba G Colciago E Vittimberga G Ercolani G Resection line involvement after gastric cancer treatment: handle with care Updates Surg 2018 70 213 23 29934732 6 Bozzetti F Bignami P Bertario L Fissi S Eboli M Surgical treatment of gastric cancer invading the oesophagus Eur J Surg Oncol 2000 26 810 4 11087650 7 Kakeji Y Korenaga D Baba H Watanabe A Tsujitani S Maehara Y Surgical treatment of patients with gastric carcinoma and duodenal invasion J Surg Oncol 1995 59 215 9 7630166 8 Roberts P Seevaratnam R Cardoso R Law C Helyer L Coburn N Systematic review of pancreaticoduodenectomy for locally advanced gastric cancer Gastric Cancer 2012 15 Suppl 1 S108 15 21870150 9 Yonemura Y Ooyama S Matumoto H Kamata T Kimura H Takegawa S Pancreaticoduodenectomy in combination with right hemicolectomy for surgical treatment of advanced gastric carcinoma located in the lower half of the stomach Int Surg 1991 76 226 9 1685729 10 Ajisaka H Fujita H Kaji M Maeda K Yabushita K Konishi K Treatment of patients with gastric cancer and duodenal invasion Int Surg 2001 86 9 13 11890346 11 Di Saverio S Segalini E Birindelli A Todero S Podda M Rizzuto A Pancreas-sparing, ampulla-preserving duodenectomy for major duodenal (D1–D2) perforations Br J Surg 2018 105 1487 92 30024637 12 Murata A Okamoto K Muramatsu K Kubo T Fujino Y Matsuda S Effects of additional laparoscopic cholecystectomy on outcomes of laparoscopic gastrectomy in patients with gastric cancer based on a national administrative database J Surg Res 2014 186 157 63 24135376 13 Bencini L Marchet A Alfieri S Rosa F Verlato G Marrelli D The Cholegas trial: long-term results of prophylactic cholecystectomy during gastrectomy for cancer-a randomized-controlled trial Gastric Cancer 2019 22 632 9 30244294
PMC010xxxxxx/PMC10352707.txt
==== Front Korean J Clin Oncol Korean J Clin Oncol Korean Journal of Clinical Oncology 1738-8082 2288-4084 Korean Society of Surgical Oncology 37449392 10.14216/kjco.23001 kjco-19-1-1 Editorial Inflammatory and nutritional markers in patients with resectable pancreatic cancer https://orcid.org/0000-0003-2502-0086 Jung Hae Il Department of Surgery, Soonchunhyang University Cheonan Hospital, Cheonan, Korea Correspondence to: Hae Il Jung, Department of Surgery, Soonchunhyang University Cheonan Hospital, 31 Suncheonhyang 6-gil, Dongnam-gu, Cheonan 31151, Korea, Tel: +82-41-570-3635, Fax: +82-41-570-0129, E-mail: [email protected] 6 2023 30 6 2023 19 1 12 26 6 2023 28 6 2023 Copyright © 2023 Korean Society of Surgical Oncology 2023 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. ==== Body pmcPancreatic ductal adenocarcinoma (PDAC) is known to be one of the most aggressive malignancies with a very poor prognosis. Various prognostic factors of PDAC have been widely studied including age, sex, tumor size, regional lymph node and distant metastasis, carbohydrate antigen 19-9, inflammatory markers like C-reactive protein (CRP) and nutritional markers, such as albumin, ferritin, weight loss [1]. Inflammation plays a crucial role in the progression of malignancy. The repetitive tissue injury and resulting inflammation promote cell proliferation and increase the risk of DNA damage, thereby contributing to neoplastic development. Since Virchow first described the interaction between inflammation and malignancy, CRP which is one of the inflammatory indicators has been reported as prognostic factors in various cancers [2]. Additionally, nutritional status such as serum albumin, and prealbumin has been found to be correlated with the prognosis of various malignancies, including pancreatic cancer [1,3,4]. Albumin, the most abundant protein in the human serum, serves as a valuable indicator of nutritional status. It constitutes about 60% of the serum proteins by weight. Its hepatic synthesis is affected by osmotic colloid pressure and inflammatory states, but also, by nutritional status and hormones [5]. In patients with cancer, serum albumin has been found to be an independent prognostic factor for survival in various cancers such as melanoma, colorectal, pancreatic, lung, gastric, and breast cancer [3,5–7]. Furthermore, prealbumin has a shorter half-life which is about 2 to 3 days, and is considered more sensitive and accurate biomarker than albumin. Recent studies have reported that prealbumin is an independent predictor in cancer patients [3]. Recently, the CRP to prealbumin ratio (CPAR), which is an inflammatory and nutritional marker, has been investigated as a potential prognostic indicator in various cancers [6,8]. In this issue of the Korean Journal of Clinical Oncology, Kwon et al. [9] demonstrated the utility of CPAR in predicting the prognosis of PDAC. The authors aimed to identify patients who were at high risk of early recurrence in cases of resectable PDAC. Their findings showed that the patients with high CPAR were significantly associated with an increased risk of early recurrence. However, it is important to interpret these results with caution due to the limitations of the study, such as its retrospective design, small sample size, and short follow-up period. Additionally, the study did not provide information on the preoperative treatment of nutritional status and subgroup analysis based on the stage. Despite these limitations, the use of CPAR may have clinical utility in predicting the early recurrence of PDAC. Further large-scale studies are warranted to validate these findings and establish the broader applicability of CPAR as a prognostic marker. Hae Il Jung is an editorial board member of the journal but was not involved in the peer reviewer selection, evaluation, or decision process of this article. No other potential conflicts of interest relevant to this article were reported. FUNDING None. ==== Refs REFERENCES 1 Xie Q Wang L Zheng S Prognostic and clinicopathological significance of C-reactive protein to albumin ratio in patients with pancreatic cancer: a meta-analysis Dose Response 2020 18 15593258 20931290 2 Nam YH Park MS Lee SM Preoperative C-reactive protein as a prognostic factor for recurrence after surgical resection of biliary tract cancer Korean J Clin Oncol 2015 11 101 5 3 Park YM Seo HI Noh BG Kim S Hong SB Lee NK Clinical impact of serum prealbumin in pancreaticobiliary disease Korean J Clin Oncol 2022 18 61 5 36945244 4 Qiao W Leng F Liu T Wang X Wang Y Chen D Prognostic value of prealbumin in liver cancer: a systematic review and meta-analysis Nutr Cancer 2020 72 909 16 31507226 5 Fan Y Sun Y Man C Lang Y Preoperative serum prealbumin level and adverse prognosis in patients with hepatocellular carcinoma after hepatectomy: a meta-analysis Front Oncol 2021 11 775425 34746015 6 Huang Z Cai P Zhao Y Niu D Xu F Lai Y Preoperative C-reactive protein to prealbumin ratio is independently associated with prognosis in patients with resectable colorectal cancer J Surg Oncol 2022 125 1238 50 35174885 7 Kang YH Park JW Ryoo SB Jeong SY Park KJ Impact of nutritional screening index on perioperative morbidity after colorectal cancer surgery as a independent predictive factor Korean J Clin Oncol 2017 13 118 25 8 Matsunaga T Miyata H Sugimura K Motoori M Asukai K Yanagimoto Y Prognostic significance of C-reactive protein-to-prealbumin ratio in patients with esophageal cancer Yonago Acta Med 2019 63 8 19 32158328 9 Kwon CH Seo HI Kim DU Han SY Kim S Lee NK Clinical significance of C-reactive protein-to-prealbumin ratio in predicting early recurrence in resectable pancreatic cancer Korean J Clin Oncol 2023 19 11 7 37449394
PMC010xxxxxx/PMC10352708.txt
==== Front Korean J Clin Oncol Korean J Clin Oncol Korean Journal of Clinical Oncology 1738-8082 2288-4084 Korean Society of Surgical Oncology 37449397 10.14216/kjco.23006 kjco-19-1-32 Case Report Pancreatic metastasis from papillary thyroid cancer: a case report and literature review https://orcid.org/0000-0003-4726-8461 Song Sang Hwa 1 https://orcid.org/0000-0003-1089-2033 Hur Young Hoe 1 https://orcid.org/0000-0001-8457-7314 Cho Chol Kyoon 12 https://orcid.org/0000-0002-0368-5389 Koh Yang Seok 12 https://orcid.org/0000-0001-7242-4855 Park Eun Kyu 3 https://orcid.org/0000-0002-8636-5726 Kim Hee Joon 3 https://orcid.org/0009-0005-8397-6520 Shin Sang Hoon 3 https://orcid.org/0009-0003-2713-8026 Yu Sung Yeol 3 https://orcid.org/0000-0003-1290-1837 Oh Chae Yung 3 1 Department of Surgery, Chonnam National University Hwasun Hospital, Hwasun, Korea 2 Department of Surgery, Chonnam National University Medical School, Gwangju, Korea 3 Department of Surgery, Chonnam National University Hospital, Gwangju, Korea Correspondence to: Chol Kyoon Cho, Department of Surgery, Chonnam National University Hwasun Hospital, Chonnam National University Medical School, 160 Baekseo-ro, Dong-gu, Gwangju 61469, Korea, Tel: +82-62-220-6451, Fax: +82-62-227-1635, E-mail: [email protected] 6 2023 30 6 2023 19 1 3237 15 5 2023 20 6 2023 21 6 2023 Copyright © 2023 Korean Society of Surgical Oncology 2023 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. Pancreatic metastasis from papillary thyroid cancer (PTC) is extremely rare; only 18 cases have been reported in the literature. However, several reviews have highlighted similar characteristics between metastatic and primary pancreatic tumors. The patient was a 51-year-old male with a history of total thyroidectomy, modified radical neck dissection, and radioactive iodine ablation for PTC in 2014. Nodules suspected of metastasis were found in both lungs on chest computed tomography (CT). However, after 6 months, a follow-up chest CT showed no increase in size; thus, a follow-up observation was planned. Six years after his initial diagnosis, abdominal CT and pancreas magnetic resonance imaging revealed a 4.7 cm cystic mass with a 2.5 cm enhancing mural nodule in the pancreas tail. We diagnosed the pancreatic lesion as either metastatic cancer or primary pancreas cancer. The patient underwent distal pancreato-splenectomy. After surgery, the pathological report revealed that the mass was metastatic PTC. Pancreatic metastasis from PTC indicates an advanced tumor stage and poor prognosis. However, pancreatectomy can increase the survival rate when the lesion is completely resectable. Therefore, surgical resection should be considered as a treatment for pancreatic metastasis from PTC. Papillary thyroid cancer Neoplasm metastasis Pancreas Case reports ==== Body pmcINTRODUCTION Papillary thyroid cancer (PTC) is the most common type of well-differentiated endocrine malignancy [1]. PTC has an excellent prognosis with a 5-year disease-specific survival rate of 98%. The main route of metastasis for PTC is local spread to the lymph nodes of the neck [1]. Around 5% of patients have systemic metastases, most commonly to the lungs and bones [1]. PTC metastasis to the pancreas is extremely rare. To date, only 18 cases have been reported in the literature [2–11]. According to the literature, diagnosis is difficult due to the occurrence of pancreatic metastasis after a long period and lack of specific imaging findings and clinical standards. Some reports have revealed similar characteristics between pancreatic metastatic tumors and primary pancreatic tumors. Here, we report a patient who underwent surgical resection for pancreatic metastasis detected 6 years after the first PTC diagnosis. Informed consent was waived because the study had a retrospective nature and the analysis used anonymous clinical data. CASE REPORT The patient was a 51-year-old male with a history of total thyroidectomy and modified radical neck dissection for PTC in 2014. According to the surgical findings, a tumor lesion was found to invade the trachea, and tracheal shaving was additionally performed. Postoperative pathological diagnosis indicated PTC (TNM classification: pT3N1b) requiring radioactive iodine (RAI) ablation. Three months after RAI ablation, a recurrent mass was found in the trachea on imaging study, and tracheal mass resection was performed. Nodules suspected of metastasis were found in both lungs on chest computed tomography (CT). However, the follow-up chest CT examination after 6 months did not detect an increase in size, and follow-up of the patient was decided. Six years after the initial diagnosis, the patient was referred for surgery for a pancreatic tail mass discovered incidentally on an abdominal CT performed to diagnose the cause of uncontrolled diabetes. All laboratory findings were within normal limits except for a high glucose level (485 mg/dL). The levels of tumor markers were high, with a carbohydrate antigen 19-9 level of 110.8 U/mL (normal range, 0–39 U/mL) and a carcinoembryonic antigen level of 10.54 ng/mL (normal range, 0–5 ng/mL). A high thyroglobulin level of 350 ng/mL (normal range, 1.4–78 ng/mL) was also detected. Abdominal CT showed a 4.7 cm cystic mass with an enhancing mural nodule in the pancreatic tail, rule out malignant cystic neoplasm (Fig. 1). Moreover, pancreas magnetic resonance imaging (MRI) showed pancreas tail with a well-circumscribed dominant cystic mass of 5.6×5.0 cm with a 2.6 cm, mural-enhancing solid portion on the posterior wall with diffusion restriction (Fig. 2). The positron emission tomography-CT (PET-CT) scan showed multiple tiny to small nodules without significant hypermetabolism in both lungs, with a slight size increase or with newly developed lesions, and a 5.8 cm cystic mass in the pancreatic tail, with somewhat focal hypermetabolism in the solid portion (SUV 2.5) (Fig. 3). Based on the imaging findings, we diagnosed the pancreatic lesion as either metastatic cancer or primary pancreatic cancer, and we planned for distal pancreato-splenectomy (DPS). During surgery, the pancreatic tail mass and mesocolon were abutting, therefore, DPS including local excision of mesocolon was performed. Following resection, gross histology of the pancreatic tail mass showed small hematomas and necrotic lesions in the cystic mass. A 2.5 cm-sized solid portion was detected in the cystic mass (Fig. 4). Immunohistochemical analysis showed that the solid portion was positive for CK7, TTF-1, thyroglobulin, and BRAF V600E (3+). The solid portion of the pancreatic tail cystic mass was confirmed as metastatic PTC based on histopathological findings. The patient had an uneventful recovery after the operation and abdomen and chest CT was performed at the scheduled follow-up (at 6-month intervals) after surgery. The nodules suspected of metastasis in both lungs showed non-specific interval changes on serial follow-up chest CT, and there was no recurrence or metastasis of the tumor 3 years after the pancreatic surgery. Abdominal CT was performed to confirm tumor recurrence and metastasis. In addition, chest CT was used to check whether the size of the mass had increased, and an increase in thyroglobulin level was observed. If these findings occur, chemotherapy with a tyrosine kinase inhibitor (sorafenib) can be planned. DISCUSSION PTC is considered as a low-grade malignancy. Metastases are most often found in the cervical lymph nodes. Distant metastases are uncommon and usually occur in the bones, lungs, and thoracic lymph nodes [1]. According to the literature, metastasis of PTC to the pancreas is extremely rare. Only 18 cases have been reported thus far (Table 1). Patients can be diagnosed with pancreatic metastasis from PTC at various ages (mean age, 59 years), and it occurs mostly in males (male 15 cases and female 4 cases). Most patients are diagnosed with PTC at the initial diagnosis [2–5, 8–10] (tall cell PTC 4 cases [7,11] and follicular variant PTC 2 cases [5]) and undergo RAI ablation [2,3,5,6,9,10]. It has been observed that it takes an average of 7 years for pancreatic metastasis to occur. In our case, involving a 51-year-old male patient, similar to the previously reported literature, the patient was initially diagnosed with PTC (TNM classification: pT3N1b), and after 6 years, pancreatic metastasis was incidentally detected, further MRI and PET-CT were performed. According to the literature, diagnosis is difficult due to the occurrence of pancreatic metastasis after a long period and lack of specific imaging findings and clinical standards. In some studies, the characteristics of pancreatic metastatic tumors are similar to those of primary pancreatic tumors. Metastatic tumors clinically mimic primary pancreatic tumors and have been reported to appear as a solitary pancreatic mass in imaging studies. Based on CT findings, metastatic tumors typically show a well-circumscribed lesion pattern with contrast enhancement. In addition, endoscopic ultrasound (EUS) usually reveals a hypoechoic hypodense pattern, and these characteristics are similar to those of primary pancreatic tumors [4,11,12]. Therefore, it is necessary to determine whether the pancreatic lesion is a primary or metastatic tumor. EUS-guided biopsy may be an appropriate diagnostic method, which has a sensitivity of 80% to 90%, a specificity of nearly 100%, and an accuracy in diagnosing metastatic lesions of 89% [8,12]. However, in our case, we observed a cystic mass with an enhancing mural nodule on imaging, mimicking a cystic pancreatic tumor. Therefore, EUS-guided biopsy was not performed, and surgical resection was performed under a diagnosis of a primary or metastatic pancreatic tumor based on imaging findings. A review of the literature found that surgery was performed in 10 out of 18 cases of pancreatic metastasis from PTC, indicating that the surgical method differed depending on the tumor’s location. For example, pancreaticoduodenectomy (PD) [8,9] or enucleation [5,11] was performed when the tumor was located in the pancreatic head, and distal pancreatectomy (DP) [5,6] or DPS [3,4] was performed when the tumor was located in the body or tail. In our case, DPS was performed because the tumor’s location was in the tail. In the cases reported by Stein et al. [11], Rossi et al. [8], and Tramontin et al. [9], metastasis to the pancreatic head was found at an average of 7.3 years after the first diagnosis. PD was performed as a treatment, and no recurrence or metastasis to other organs after surgery was reported in all of them. In 2010, Borschitz et al. [5] reported the presence of a tumor in the pancreatic head. Excision was performed for pancreatic metastasis from PTC, and lymph node metastasis occurred 42 months after surgery. When DP (Borschitz et al. [5], Meyer and Behrend [6]) or DPS (Ren et al. [3], Angeles-Angeles et al. [4], and Stein et al. [11]) was performed because the tumor was located in the body or tail, metastasis was found at an average of 9.2 years after the first diagnosis. Among these cases, multiple metastases were found after surgery in two cases. Meyer and Behrend [6] reported a tumor that recurred with death 53 months after surgery, and Angeles-Angeles et al. [4] reported a tumor that recurred 9 months after surgery with death 3 years later. In two cases, there was no recurrence after surgery for pancreatic metastasis from PTC (Ren et al. [3] and Stein et al. [11]). In our case, metastasis to the pancreatic tail was found 6 years after the first diagnosis, DPS was performed on this lesion, and no recurrence or metastasis was found until 3 years after surgery. Two cases were reported in which no surgical treatment was performed for pancreatic metastasis from PTC. Alzahrani et al. [13] reported a case of sorafenib administration instead of surgical treatment for a PTC patient with pancreatic metastasis. In the case reported by Cho et al. [2], the patient was 81 years old, and systemic chemotherapy was performed instead of surgical treatment because of multiple organ metastases. Whether surgery for metastatic pancreatic tumors benefits patients’ prognosis is controversial because of the high morbidity and mortality after pancreatectomy. However, when the metastatic pancreatic lesion is a single metastatic tumor, pancreatectomy can increase the 5-year survival rate up to 31% [14]. Therefore, pancreatectomy may be beneficial for treating a patient without multiple organ metastases [6]. In 2006, Reddy and Wolfgang [15] reported the following patient selection criteria: primary cancer type associated with successful outcomes, control of the primary cancer site, isolated metastases, resectability of the metastasis, and patient fitness to tolerate pancreatectomy [15]. In summary, pancreatic metastasis from PTC is extremely rare, which indicates the advanced stage of the tumor, and the prognosis is very poor. However, pancreatectomy can increase the survival rate when the metastatic lesion is completely resectable. Therefore, surgical resection should be considered as a treatment for pancreatic metastasis from PTC. Fig. 1 Abdominal computed tomography showed a 4.7-cm-sized pancreatic tail cystic mass and a 2.3-cm-sized enhancing mural nodule in the cystic mass lumen. Fig. 2 Abdominal magnetic resonance imaging (enhance) showed a 2.3-cm-sized enhancing mural nodule (arrow in A, T1WI) with a 5.6-cm-sized well-circumscribed cystic mass in the pancreatic tail (arrow in B, T2WI). Fig. 3 Positron emission tomography-computed tomography showed a 5.8-cm-sized cystic mass in pancreatic tail, with somewhat focal hypermetabolism in solid portion (SUV 2.5). Fig. 4 Surgical specimen of the pancreas revealed small hematomas and necrotic lesions in the cystic mass (A; arrow) with a 2.5 cm-sized solid portion (B; arrow). Table 1 Summary of cases of pancreatic metastasis from PTC First author Age (yr) Sex Histology/stage RAI Imaging study Imaging characteristics Location in pancreas Years from metastasis Treatment Recurrence or metastasis after surgery Stein [11] 39 F PTC/NA No ERCP, CT Hypervascular on arteriography 5 cm isodense lesion Head 7 NA NA 53 M TCPTC/NA Yes MRI 3×4 cm, well circumscribed Head 1 PD NA 67 M TCPTC/NA Yes PETCT, EUS Hypermetabolic, hypodense mass 1.5×1.1 cm well defined hypoechoic, homogeneous mass Head 7 NA NA 82 M PTC/NA Yes EUS NA Neck 5 NA NA 67 F FVPTC/pT2N1c Yes CT, MRCP 1.8×1.5 cm hypovascular lesions, T1 and T2 hypointense, faint enhancement, well-circumscribed Neck 7 NA NA 66 M PTC/NA NA PET-CT, CT 6.2×5.8 cm heterogeneously enhancing mass with clear border hypometabolic Body and tail 11 DPS NA 46 M TCPTC/NA Yes PET-CT 3 cm well-circumscribed Head 3 Pancreatic head mass resection No recurrence Meyer [6] 67 M PTC/pT4N0 Yes CT Capsulated cystic mass Head 5 DP 53 mo Angeles-Angeles [4] 72 M PTC/NA NA CT 8.5 cm non encapsulated, well circumscribed Body and tail NA DPS NA Borschitz [5] 61 M PTC/pT3N1 Yes PETCT, MRI Hypermetabolic isodense well circumscribed lesion Body 15 DP 9 mo 44 F FVPTC/pT3N1a Yes PETCT, MRI Hypermetabolic, T1 hypointense, T2 intermediate well-circumscribed Head 10 Enucleation 42 mo Alzahrani [13] 55 M PTC/pT4aN1b Yes PETCT, MRI 1.7 cm well circumscribed hypermetabolic Head 8 Sorafenib No surgery Davidson [7] 84 F TCPTC/pT3N1b Yes PETCT, CT Hypermetabolic, well circumscribed 1.1 cm enhancing mass Body 2 NA NA Cho [2] 81 M PTC/NA Yes PETCT, MRCP Hypermetabolic lesion 1×0.8 cm T1 hypointense, slightly T2 hyperintense, diffusion restriction, peripheral enhancement Head/body 10 Systemic treatment NA Ren [3] 47 M PTC/NA Yes US, CT 4×3 cm pancreatic space occupying lesions with main ductal dilation Body and tail 0 DPS No recurrence Rossi [8] 60 M PTC/NA No CT, MRI, EUS 2 cm hypoechoic lesion and intraductal growth Head 15 PD NA Tramontin [9] 73 M PTC/pT4aN1b Yes PETCT 2.8 cm mass Head 6 PD NA Wong [10] 75 M PTC/NA Yes PETCT Hypodense, well circumscribed lesion with atrophy of distal pancreas hypermetabolic Body and tail 7 NA NA Present case 51 M PTC/pT3N1b Yes CT, MRCP, PETCT Well circumscribed dominant cystic mass of 5.6 cm with a 2.6 cm mural enhancing solid portion on the posterior wall with diffusion restriction, hypermetabolic Tail 6 DPS No recurrence PTC, papillary thyroid cancer; RAI, radioactive iodine; F, female; M, male; NA, not available; ERCP, endoscopic retrograde cholangiopancreatography; CT, computed tomography; TCPTC, tall cell PTC; MRI, magnetic resonance imaging; PD, pancreaticoduodenectomy; PET-CT, positron emission tomography-CT; EUS, endoscopic ultrasound; MRCP, magnetic resonance cholangiopancreatography; DPS, distal pancreato-splenectomy; FVPTC, follicular variant PTC; DP, distal pancreatectomy; US, ultrasound. No potential conflict of interest relevant to this article was reported. FUNDING None. ==== Refs REFERENCES 1 Toniato A Boschin I Casara D Mazzarotto R Rubello D Pelizzo M Papillary thyroid carcinoma: factors influencing recurrence and survival Ann Surg Oncol 2008 15 1518 22 18324441 2 Cho M Acosta-Gonzalez G Brandler TC Basu A Wei XJ Simms A Papillary thyroid carcinoma metastatic to the pancreas: case report Diagn Cytopathol 2019 47 214 7 30479026 3 Ren H Ke N Tan C Wang X Cao W Liu X Unusual metastasis of papillary thyroid cancer to the pancreas, liver, and diaphragm: a case report with review of literature BMC Surg 2020 20 82 32321510 4 Angeles-Angeles A Chable-Montero F Martinez-Benitez B Albores-Saavedra J Unusual metastases of papillary thyroid carcinoma: report of 2 cases Ann Diagn Pathol 2009 13 189 96 19433299 5 Borschitz T Eichhorn W Fottner C Hansen T Schad A Schadmand-Fischer S Diagnosis and treatment of pancreatic metastases of a papillary thyroid carcinoma Thyroid 2010 20 93 8 20025539 6 Meyer A Behrend M Is pancreatic resection justified for metastasis of papillary thyroid cancer? Anticancer Res 2006 26 3B 2269 73 16821600 7 Davidson M Olsen RJ Ewton AA Robbins RJ Pancreas metastases from papillary thyroid carcinoma: a review of the literature Endocr Pract 2017 23 1425 9 29144798 8 Rossi G Petrone MC Schiavo Lena M Doglioni C Pecorelli N Falconi M Pancreatic metastasis of papillary thyroid carcinoma with an intraductal growth pattern Endoscopy 2020 52 E452 3 32396960 9 Tramontin MY Faria PA Nascimento CM Barbosa CA Barros MF Barros AR Cholestatic syndrome as initial manifestation of pancreatic metastasis of papillary thyroid carcinoma: case report and review Arch Endocrinol Metab 2020 64 179 84 32236313 10 Wong BZ Dickie G Garcia P Scott D Pattison DA 124I-PET/CT-guided diagnosis and personalized treatment of metastatic papillary thyroid cancer to the pancreas Clin Nucl Med 2021 46 337 9 33492854 11 Stein R Harmon TS Harmon CE Kuo E Ozdemir S Pancreatic metastasis from papillary thyroid carcinoma: case report and literature review Hell J Nucl Med 2021 24 140 8 34352049 12 Waters L Si Q Caraway N Mody D Staerkel G Sneige N Secondary tumors of the pancreas diagnosed by endoscopic ultrasound-guided fine-needle aspiration: a 10-year experience Diagn Cytopathol 2014 42 738 43 24554612 13 Alzahrani AS AlQaraawi A Al Sohaibani F Almanea H Abalkhail H Pancreatic metastasis arising from a BRAF(V600E)-positive papillary thyroid cancer: the role of endoscopic ultrasound-guided biopsy and response to sorafenib therapy Thyroid 2012 22 536 41 22435913 14 Soyluk O Selcukbiricik F Erbil Y Bozbora A Kapran Y Ozbey N Prognostic factors in patients with papillary thyroid carcinoma J Endocrinol Invest 2008 31 1032 7 19169062 15 Reddy S Wolfgang CL The role of surgery in the management of isolated metastases to the pancreas Lancet Oncol 2009 10 287 93 19261257
PMC010xxxxxx/PMC10352709.txt
==== Front Korean J Clin Oncol Korean J Clin Oncol Korean Journal of Clinical Oncology 1738-8082 2288-4084 Korean Society of Surgical Oncology 37449396 10.14216/kjco.23005 kjco-19-1-27 Original Article Is bone mineral density a prognostic factor in postmenopausal women with luminal A breast cancer? https://orcid.org/0000-0001-5100-0017 Lee Seungju 1 https://orcid.org/0000-0001-7717-7734 Kim Hyun Yul 1 https://orcid.org/0000-0002-9647-8556 Jung Youn Joo 1 https://orcid.org/0000-0002-4119-9445 Kang Seok-Kyung 1 https://orcid.org/0000-0002-0503-984x Kim Jee Yeon 2 https://orcid.org/0000-0003-3982-8350 Yun Mi Sook 3 1 Department of Surgery, Pusan National University Yangsan Hospital, Pusan National University School of Medicine, Yangsan, Korea 2 Department of Pathology, Pusan National University Yangsan Hospital, Pusan National University School of Medicine, Yangsan, Korea 3 Division of Biostatistics, Research Institute for Convergence of Biomedical Science and Technology, Pusan National University Yangsan Hospital, Yangsan, Korea Correspondence to: Youn Joo Jung, Department of Surgery, Pusan National University Yangsan Hospital, Pusan National University School of Medicine, 20 Geumo-ro, Mulgeum-eup, Yangsan 50612, Korea, Tel: +82-55-360-2124, Fax: +82-55-360-2154, E-mail: [email protected] 6 2023 30 6 2023 19 1 2731 18 4 2023 23 6 2023 28 6 2023 Copyright © 2023 Korean Society of Surgical Oncology 2023 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. Purpose Several studies are concerned about the association between bone mineral density (BMD) and the risk of breast cancer in postmenopausal women, but it is controversial. Therefore, we evaluated whether BMD might be a risk factor for recurrences, or metastases in menopausal luminal A breast cancer patients. Methods In this retrospective study, data of 348 patients with luminal A breast cancer who received treatment at Pusan National University Yangsan Hospital between 2012 and 2016 were analyzed. Patients were divided into two groups: normal BMD and low BMD including osteopenia or osteoporosis in preoperative examination. Patients were also divided into three groups according to BMD changes: no change in BMD; improvement in BMD, and deterioration in BMD. Events were defined as recurrence, occurrence of contralateral breast cancer, and metastasis to any other organ. Results Preoperative examination revealed normal BMD in 129 of 348 patients and low BMD in 219 patients. During a median follow-up period of 78 months, only 14 patients (4.0%) experienced recurrences, distant metastases, or occurrences of contralateral breast cancer. Five-year disease-free survival rate was 98.2% for 219 patients with low BMD and 95.0% for 129 patients with normal BMD (P=0.33). Disease-free survival at 5 years was 97.0% for the no change in the BMD group, 94.6% for the BMD improvement group, and 98.4% for the BMD deterioration group (P=0.79). Conclusion In this study, BMD had no statistically significant associations on recurrences, metastases, or incidences of contralateral breast cancer in postmenopausal patients with luminal A breast cancer. Breast neoplasms Bone density Aromatase inhibitors ==== Body pmcINTRODUCTION Hormone receptor-positive (HR+) breast cancer is the most prevalent subtype of breast cancer [1,2]. Estrogen exposure has an impact on risk factors for HR+ breast cancer, including hormone replacement therapy, early menarche, and late menopause [3–5]. Furthermore, as estrogen controls bone turnover, increased bone mineral density (BMD) has been thought to be a sign of persistent estrogen exposure [3,4,6–8]. Therefore, it has been hypothesized that high BMD is associated with a worse prognosis for breast cancer. A previous study has shown that postmenopausal breast cancer patients with low BMD have lower rates of local and distant recurrences than patients with normal BMD [9]. The subtype of breast cancer was not classified in that study, making it difficult to apply its results to all postmenopausal patients with breast cancer. Due to widespread use of aromatase inhibitors, postmenopausal breast cancer patients are especially susceptible to bone density loss. Several studies have demonstrated that aromatase inhibitor-treated breast cancer patients experience an elevated incidence of osteoporotic fractures [10–12]. Therefore, bone density reduction should not be neglected, as previous research has shown that increased bone density is a poor prognostic factor for breast cancer. The objective of the present study was to examine the associations between BMD and breast cancer recurrences among postmenopausal patients with luminal A breast cancer. Associations of breast cancer recurrences with BMD changes during the follow-up period were also evaluated. METHODS Patients This retrospective analysis evaluated 348 postmenopausal patients with luminal A breast cancer who were treated at Pusan National University Yangsan Hospital from 2012 to 2016. Luminal A breast cancer was defined if the following criteria were met: (1) HR positivity; (2) human epidermal growth factor receptor 2 negativity; and (3) Ki-67 percentage less than 15%. Menopause was defined as the absence of menstruation for more than a year combined with elevated follicle-stimulating hormone (>25 IU/mL) levels in the blood. Dual energy X-ray absorptiometry was used to measure BMDs of the lumbar spine, total femur, and femoral neck. According to measured BMD and World Health Organization (WHO) criteria, those with T scores ≥ −1.0, −2.5 to −1.0, and ≤ −2.5 were defined as normal, osteopenia, and osteoporosis patients, respectively. Patients were divided into two groups in the preoperative examination: those with normal BMD and those with low BMD (including osteopenia and osteoporosis). Changes in BMD over the course of the follow-up period were also analyzed according to WHO criteria. The bone density of patients was measured every year after surgery, and it decreased when the change in bone density was lower than before surgery, increased when it improved, and remained unchanged if there was no change. Patients with osteopenia were given calcium medications, while those with osteoporosis were given denosumab. Data such as age at diagnosis, body mass index (kg/m2), histology, tumor size, tumor grade, lymph node status, stage, the type of surgery, chemotherapy, radiation therapy, and endocrine therapy were collected. Cancers were staged according to the breast cancer anatomic stage guidelines of the 8th American Joint Committee on Cancer. Patient status was monitored every 3 to 6 months for the first 5 years following the initial treatment and then annually thereafter. This study was approved by the Institutional Review Board of Pusan National University, Korea (IRB No. 05-2022-154). Informed consent is not needed in this study. Statistical analysis Means of continuous data were compared between those with normal BMD and those with low BMD using the t-test and Mann-Whitney test. Categorical data between two groups were analyzed using the Pearson chi-square test. Events were defined as recurrence in the ipsilateral breast, chest wall or axillary, supraclavicular, infraclavicular, or internal mammary nodes, occurrences of contralateral breast cancer, and metastasis to any other organ. Disease-free survival was calculated from the date of surgery to the date of event. Kaplan-Meier survival function and log-rank test were used to confirm difference in survival rates between the two groups (normal BMD and low BMD). All data were analyzed using SPSS version 23 (SPSS Inc.). RESULTS Among 348 patients diagnosed with postmenopausal luminal A type breast cancer, preoperative examination confirmed normal BMD in 129 patients and low BMD in 219 patients. Table 1 summarizes patient and tumor characteristics of the two groups. The normal BMD group had younger patients (55.82±6.86 years) than the low BMD group (61.76±7.95 years) (P<0.001) (Table 1). Body mass index did not differ significantly between the two groups (Table 1). During the treatment and follow-up period, 46 patients in the normal BMD group (35.7%) and 22 patients in the low BMD group (10.0%) showed decrease in BMD (Table 1). All patients received endocrine therapy. In the low BMD group, 47 patients (21.5%) switched from aromatase inhibitor to tamoxifen. In the normal BMD group, only 15 patients (11.6%) made the switch (Table 2). During a median follow-up period of 78 months, only 14 patients (4.0%) had recurrences, distant metastases, or occurrences of contralateral breast cancer. Four patients experienced recurrences: one in the normal BMD group and three in the low BMD group. Six patients were found to have distant metastases: four in the normal BMD group and two in the low BMD group. Contralateral breast cancer was detected in five patients: four in the normal BMD group and one in the low BMD group. Disease-free survival rate at 5 years was 98.2% for 219 patients with low BMD and 95.0% for 129 patients with normal BMD (P=0.33) (Fig. 1). In addition, disease-free survival at 5 years was 97.0% for those whose BMD did not change during treatment, 94.6% for those with BMD improvement, and 98.4% for those with BMD deterioration (P=0.79) (Fig. 2). DISCUSSION Estrogen is known to affect bone turnover, and continuous estrogen exposure has been linked to breast cancer [3,4,6]. Thus, it has been hypothesized that high BMD is associated with breast cancer. High BMD has been regarded as a negative prognostic factor for breast cancer. Several studies have found that postmenopausal women with a high BMD have an increased risk of developing breast cancer [3,7,8,13,14]. However, such results could not be applied to all postmenopausal women because previous studies included patients who voluntarily underwent routine health examinations. In addition, because they did not classify subtypes of breast cancer, it was difficult to apply their findings to all patients with breast cancer. In a prospective cohort study, Fraenkel et al. [8] have hypothesized that BMD could serve as a biomarker for breast cancer risk. However, other studies have indicated that there is no correlation between breast cancer and BMD [15–19]. Consequently, although numerous studies have been conducted to clarify this issue, there are no conclusive results. A systemic review has also found that the association between BMD and breast cancer risk is still debatable [20]. This study found no association between BMD and breast cancer recurrence or metastasis (Fig. 1). It also found that bone density change during the follow-up period was unrelated to the prognosis of patients (Fig. 2). Additionally, there were more young women in the normal BMD group than in the low BMD group (Table 1). Therefore, patients with normal BMD are more likely to undergo breast-conserving surgery with radiation therapy than mastectomy. In addition, chemotherapy was administered more frequently in the group with normal BMD. In the low BMD group, there was a higher rate of switching to tamoxifen than in the normal BMD group (Table 2). This might be due to progression of bone density loss and aromatase inhibitor-related side effects such as musculoskeletal pain. After treatment with an aromatase inhibitor, 35.7% of patients in the normal BMD group experienced a decrease in BMD. Moreover, only a few patients showed an increase in BMD during the follow-up period. This decrease in BMD can cause osteoporotic fractures as suggested in previous studies [10–12]. In this study, seven patients with osteoporosis had fractures despite receiving treatment for bone density reduction. Therefore, patients confirmed to have low bone density should be actively treated to prevent fractures. Due to the small number of patients with events, it was difficult to achieve statistical significance in this study, which exclusively focused on luminal A breast cancer. In addition, when determining how BMD changes affected prognosis, bias was introduced because patients with low BMD were treated with calcium medications to increase BMD. It would be preferred to examine the association between bone density and breast cancer prognosis using multicenter data with a more extended follow-up period. In conclusion, BMD had no statistically significant associations on recurrences, metastases, or incidences of contralateral breast cancer in postmenopausal patients with luminal A breast cancer. In addition, BMD change during treatment showed no statistically significant associations with breast cancer recurrences, metastases, or contralateral occurrences. Fig. 1 Disease-free survival rates in normal bone mineral density (BMD) and low bone mineral density groups. Fig. 2 Disease-free survival rate according to bone mineral density (BMD) change during treatment. Table 1 Patient’s characteristics and pathology according to bone mineral density Variable Bone mineral density P-value Normal (n=129) Low (n=219) Age (yr) 55.82±6.86 61.76±7.95 <0.001 Body mass index (kg/m2) 24.72±3.38 24.47±3.37 0.502 Bone mineral density change <0.001  Decrease 46 (35.7) 22 (10.0)  No change 83 (64.3) 160 (73.1)  Increase 0 37 (16.9) Tumor histology 0.160  Invasive ductal carcinoma 105 (81.4) 194 (88.6)  Invasive lobular carcinoma 13 (10.1) 12 (5.5)  Others 11 (8.5) 13 (5.9) pT 0.005  1 86 (66.7) 128 (58.5)  2 35 (27.1) 84 (38.4)  3 7 (5.4) 4 (1.8)  4 1 (0.8) 3 (1.4) pN 0.385  0 93 (72.0) 153 (69.8)  1 25 (19.4) 49 (22.4)  2 6 (4.7) 14 (6.4)  3 5 (3.9) 3 (1.4) Stage 0.322  I 74 (57.4) 106 (48.4)  IIA 28 (21.7) 66 (30.1)  IIB 14 (10.9) 26 (11.9)  III 13 (10.1) 21 (9.6) Grade 0.748  1 52 (40.3) 80 (36.5)  2 70 (54.3) 128 (58.5)  3 7 (5.4) 11 (5.0) Lymphovascular invasion 0.280  Yes 21 (16.3) 46 (21.0)  No 108 (83.7) 173 (79.0) Values are presented as mean±standard deviation or number (%). Table 2 Treatments of patients according to bone mineral density Variable Bone mineral density P-value Normal (n=129) Low (n=219) Operation 0.063  Breast-conserving surgery 104 (80.6) 157 (71.7)  Mastectomy 25 (19.4) 62 (28.3) Radiation therapy 0.007  Yes 114 (88.4) 168 (76.7)  No 15 (11.6) 51 (23.3) Chemotherapy 0.001  Yes 102 (79.1) 137 (62.6)  No 27 (20.9) 82 (37.4) Endocrine therapy 0.007  Aromatase inhibitor 74 (57.4) 135 (61.6)  Tamoxifen 35 (27.1) 33 (15.1)  Aromatase inhibitor → tamoxifen 15 (11.6) 47 (21.5)  Tamoxifen → aromatase inhibitor 5 (3.9) 4 (1.8) Values are presented as number (%). No potential conflict of interest relevant to this article was reported. FUNDING This study was supported by a 2022 research grant from Pusan National University Yangsan Hospital. ==== Refs REFERENCES 1 Kang SY Kim YS Kim Z Kim HY Kim HJ Park S Breast cancer statistics in Korea in 2017: data from a Breast Cancer Registry J Breast Cancer 2020 23 115 28 32395372 2 Parise CA Bauer KR Brown MM Caggiano V Breast cancer subtypes as defined by the estrogen receptor (ER), progesterone receptor (PR), and the human epidermal growth factor receptor 2 (HER2) among women with invasive breast cancer in California, 1999–2004 Breast J 2009 15 593 602 19764994 3 Howell A Anderson AS Clarke RB Duffy SW Evans DG Garcia-Closas M Risk determination and prevention of breast cancer Breast Cancer Res 2014 16 446 25467785 4 Santen RJ Yue W Wang JP Estrogen metabolites and breast cancer Steroids 2015 99 Pt A 61 6 25168343 5 Samavat H Kurzer MS Estrogen metabolism and breast cancer Cancer Lett 2015 356 2 Pt A 231 43 24784887 6 Zhang Y Kiel DP Kreger BE Cupples LA Ellison RC Dorgan JF Bone mass and the risk of breast cancer among postmenopausal women N Engl J Med 1997 336 611 7 9032046 7 Hadji P Gottschalk M Ziller V Kalder M Jackisch C Wagner U Bone mass and the risk of breast cancer: the influence of cumulative exposure to oestrogen and reproductive correlates. Results of the Marburg breast cancer and osteoporosis trial (MABOT) Maturitas 2007 56 312 21 17049767 8 Fraenkel M Novack V Mizrakli Y Koretz M Siris E Norton L Bone mineral density in women newly diagnosed with breast cancer: a prospective cohort study NPJ Breast Cancer 2022 8 21 35177701 9 Zambetti A Tartter PI Bone mineral density is a prognostic factor for postmenopausal Caucasian women with breast cancer Breast J 2013 19 168 72 23406171 10 Lee S Yoo JI Lee YK Park JW Won S Yeom J Risk of osteoporotic fracture in patients with breast cancer: meta-analysis J Bone Metab 2020 27 27 34 32190606 11 Lee YK Lee EG Kim HY Lee Y Lee SM Suh DC Osteoporotic fractures of the spine, hip, and other locations after adjuvant endocrine therapy with aromatase inhibitors in breast cancer patients: a meta-analysis J Korean Med Sci 2020 35 e403 33258332 12 Waqas K Lima Ferreira J Tsourdi E Body JJ Hadji P Zillikens MC Updated guidance on the management of cancer treatment-induced bone loss (CTIBL) in pre-and postmenopausal women with early-stage breast cancer J Bone Oncol 2021 28 100355 33948427 13 Kim BK Choi YH Song YM Park JH Noh HM Nguyen TL Bone mineral density and the risk of breast cancer: a case-control study of Korean women Ann Epidemiol 2014 24 222 7 24360852 14 Kalder M Jager C Seker-Pektas B Dinas K Kyvernitakis I Hadji P Breast cancer and bone mineral density: the Marburg Breast Cancer and Osteoporosis Trial (MABOT II) Climacteric 2011 14 352 61 21413864 15 Tremollieres F Ribot C Bone mineral density and prediction of non-osteoporotic disease Maturitas 2010 65 348 51 20079983 16 Cauley JA Song J Dowsett SA Mershon JL Cummings SR Risk factors for breast cancer in older women: the relative contribution of bone mineral density and other established risk factors Breast Cancer Res Treat 2007 102 181 8 17028986 17 Kerlikowske K Shepherd J Creasman J Tice JA Ziv E Cummings SR Are breast density and bone mineral density independent risk factors for breast cancer? J Natl Cancer Inst 2005 97 368 74 15741573 18 Stewart A Kumar V Torgerson DJ Fraser WD Gilbert FJ Reid DM Axial BMD, change in BMD and bone turnover do not predict breast cancer incidence in early postmenopausal women Osteoporos Int 2005 16 1627 32 15782281 19 Nagel G Peter RS Klotz E Brozek W Concin H Bone mineral density and breast cancer risk: results from the Vorarlberg Health Monitoring & Prevention Program and meta-analysis Bone Rep 2017 7 83 9 29018837 20 Zain NM Seriramulu VP Chelliah KK Bone mineral density and breast cancer risk factors among premenopausal and postmenopausal women a systematic review Asian Pac J Cancer Prev 2016 17 3229 34 27509955
PMC010xxxxxx/PMC10352710.txt
==== Front Korean J Clin Oncol Korean J Clin Oncol Korean Journal of Clinical Oncology 1738-8082 2288-4084 Korean Society of Surgical Oncology 37449393 10.14216/kjco.23002 kjco-19-1-3 Original Article Clinical influence of neoadjuvant chemoradiotherapy on immunonutritional status in locally advanced rectal cancer https://orcid.org/0000-0002-6423-4434 Lee Soohyeon https://orcid.org/0000-0002-6559-0369 Kang Dong Hyun https://orcid.org/0000-0001-5617-0365 Ahn Tae Sung https://orcid.org/0009-0007-8954-9671 Jo Dong Hee https://orcid.org/0009-0009-2842-2030 Kim Eunhyeon https://orcid.org/0000-0003-3567-6687 Baek Moo Jun Department of Surgery, Soonchunhyang University Cheonan Hospital, Soonchunhyang University College of Medicine, Cheonan, Korea Correspondence to: Moo Jun Baek, Department of Surgery, Soonchunhyang University Cheonan Hospital, Soonchunhyang University College of Medicine, 31 Suncheonhyang 6-gil, Dongnam-gu, Cheonan 31151, Korea, Tel: +82-41-570-3633, Fax: +82-41-571-0129, E-mail: [email protected] 6 2023 30 6 2023 19 1 310 22 6 2023 28 6 2023 29 6 2023 Copyright © 2023 Korean Society of Surgical Oncology 2023 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. Purpose Cancer patients receiving various anti-cancer treatments commonly experience malnutrition, and many studies have reported that nutritional status is associated with survival and prognosis. Although standard neoadjuvant chemoradiotherapy (CRT) is commonly used in patients with locally advanced rectal cancer owing to its tumor-downsizing and downstaging effects, there is a lack of research on the impact of patients’ nutritional status on the efficacy of neoadjuvant CRT. Methods We investigated the immunonutritional markers before and after long-course neoadjuvant CRT in 131 patients diagnosed with locally advanced rectal cancer from March 2013 to March 2022. Results We divided the patients into two groups: a low prognostic nutritional index (PNI) with a cutoff value of 50.92, and a high PNI. In both groups, significant decreases in lymphocyte count and PNI and an increase in neutrophil-to-lymphocyte ratio (NLR) were observed before and after CRT (P<0.001). Furthermore, a higher proportion of patients experienced adverse effects in the low PNI group than in the high PNI group (76.6% in low PNI vs. 54.8% in high PNI, P=0.013). The most commonly reported CRT-induced adverse effect was lower gastrointestinal tract toxicity. Conclusion By measuring the PNI and NLR without additional tests prior to starting neoadjuvant CRT in patients with locally advanced rectal cancer, it is possible to predict the risk of acute adverse effects caused by CRT. Additionally, providing external nutritional support to reduce the immunonutritional changes that occur during CRT can decrease side effects and potentially increase treatment compliance. Rectal neoplasms Chemoradiotherapy Nutritional status ==== Body pmcINTRODUCTION Rectal cancer is the third most common cause of cancer-related deaths worldwide [1]. Surgical methods for rectal cancer vary depending on the size and location of the tumor and the degree of infiltration into the surrounding tissues, including transanal local excision and transabdominal resection. In the case of locally advanced rectal cancer (LARC), preoperative chemoradiotherapy (CRT) is also an important treatment modality. Standard neoadjuvant CRT has shown the expected effects in patients, including tumor downsizing, downstaging, and sphincter preservation. Pathological complete response rates range from 15% to 38%; however, the associated adverse effects cannot be ignored [2]. The common adverse effects of maintenance CRT include loss of appetite, nausea, fecal incontinence, and anal pain. Severe adverse effects include fistula formation and an increased risk of postoperative anastomotic leakage. In particular, fecal incontinence caused by radiotherapy (RT) has been shown to have a significant negative impact on the patients’ quality of life [3]. Nutritional status is a significant prognostic factor in cancer patients, and even patients who were initially well-nourished can easily experience malnutrition due to cancer-induced metabolic dyshomeostasis. Malnutrition can affect immune function, physical performance, and overall quality of life, as well as negatively impact the efficacy of anti-cancer treatments including chemotherapy and radiation therapy. It has been reported that up to 10%–20% of deaths in cancer patients are attributed to malnutrition rather than for the tumor itself [4,5]. In colorectal cancer, nutritional status is particularly crucial as malnutrition may increase the risk of anastomotic leakage and delay the recovery of intestinal functions. Consequently, prolonged hospitalization, increased postoperative complications, and reduced treatment response and survival rates are observed in these patients [6]. As mentioned above, while CRT is a useful treatment, its associated adverse effects cannot be ignored. In cancer patients commonly affected by malnutrition, the impact of nutritional status on various anti-cancer treatments, as well as its relationship with survival rates, is being studied in lung, cervical, breast, and rectal cancers. However, research on this topic is lacking. Therefore, this study aimed to investigate the differences in adverse effects of neoadjuvant CRT based on the nutritional status of patients with LARC who received neoadjuvant CRT and to examine the changes in nutritional status before and after CRT. METHODS Study population This retrospective study was conducted at Soonchunhyang University Cheonan Hospital between March 2013 and March 2022. We included patients who were histologically diagnosed with LARC and received long-course CRT. The following exclusion criteria were applied: (1) patients who did not undergo surgery after CRT; (2) patients with incomplete medical records; and (3) patients with psychiatric conditions requiring medication that could affect treatment compliance. In total, 131 patients were enrolled in this study. Preoperative RT was delivered at a dose of 5,000 to 5,040 cGy in 25 to 28 fractions. Concurrent chemotherapy was administered with either oral capecitabine or 5-fluorouracil/leucovorin according to the National Comprehensive Cancer Network guidelines. This study complied with the principles of the Declaration of Helsinki. The study protocol was reviewed and approved by the Institutional Review Board of the Soonchunhyang University Cheonan Hospital (IRB No. 2022-06-022). The informed consent was waived because this design is a retrospective study. Data collection We extracted information from the medical records, including age at diagnosis, sex, body weight, body mass index (BMI), medical history, CRT regimen, CRT duration, American Society of Anesthesiologists (ASA) physical status classification, and Charlson Comorbidity Index (CCI). The clinical cancer stage was determined based on imaging tests performed before CRT initiation, and radiation therapy records, surgical procedures, and postoperative histopathological results were reviewed. Blood tests conducted within 1 month before CRT initiation and within 2 months after completion were used to assess serum albumin, hemoglobin, neutrophil count, lymphocyte count, carcinoembryonic antigen levels, and carbohydrate antigen 19-9 levels. The neutrophil-to-lymphocyte ratio (NLR) was calculated as the neutrophil count divided by the lymphocyte count. The pathological tumor regression grade (TRG) was determined based on the histopathological results after surgery using the Dworak grading system, which categorizes TRG into five grades: complete regression (TRG4), near-complete regression (TRG3), moderate regression (TRG2), minimal regression (TRG1), and no regression (TRG0). The adverse effects of CRT reported by the patients during the CRT period until 2 months after completion were classified according to the Acute Radiation Scoring criteria of the toxicity criteria of the Radiation Therapy Oncology Group (RTOG). Statistical analyses The chi-square test, Fisher exact test, and Mann-Whitney U test were used to compare groups using the Statistical Package for the Social Sciences 26.0 (IBM Corp.). The cutoff value for the pre-CRT PNI was calculated using adverse effect-dependent receiver operating characteristic curves. All patients were divided into two groups according to the PNI cutoff value. Univariate analysis was used to analyze the relationship between each variable and acute adverse effects of neoadjuvant CRT. Multivariate analysis adjusted for age, sex, ASA grade, CCI, pre-CRT BMI (kg/m2), clinical cancer stage, pre-CRT PNI, pre-CRT NLR, PNI change value, and NLR change value was performed using multivariate logistic regression. Statistical significance was set at P<0.05. RESULTS In total, 131 patients were included. Based on the presence or absence of adverse effects, the pre-CRT PNI cutoff value was 50.92 (sensitivity 86.0%, specificity 49.4%, area under the curve=0.676). Table 1 summarizes the characteristics of the low PNI and high PNI groups based on the cutoff value of the pre-CRT PNI. High PNI patients had a significantly lower mean age than those with low PNI (P<0.001). Furthermore, when comparing the ASA grades before surgery, the percentage of patients with ASA grade 3 differed between the low (19.1%) and high (6.0%) PNI groups. Differences were considered statistically significant. The severity of comorbidities, as indicated by a CCI score of 5 or higher, was considered severe. In the low PNI group, 89.4% of the patients had a CCI score of 5 or higher, which was significantly higher than that in the high PNI group (P<0.001). There were no differences between the two groups in terms of the clinical cancer stage before CRT or tumor markers. The total amount of radiation varied slightly depending on the timing of treatment; however, the dose ranged from 5,000 to 5,040 cGy. The total duration of radiation therapy was 36 days, and there was no difference between the two groups. The chemotherapy regimens used were leucovorin/5-fluorouracil and capecitabine, and there were no statistically significant differences. Three patients (2.3% of the total patients) discontinued CRT, and all three had a high PNI. The reasons for discontinuation were deterioration of the general condition, loss of appetite, uncontrolled anal pain, and fecal incontinence. There was no statistically significant difference in body weight and BMI before and after neoadjuvant CRT (Table 2). However, both the low and high PNI groups showed statistically significant differences in lymphocyte count, PNI, and NLR (P<0.001) (Fig. 1). The neutrophil count also showed a significant difference between the total patient group and the high PNI patients (P<0.001). Hemoglobin tended to decrease post-CRT compared with pre-CRT, but the difference was not statistically significant. Serum albumin showed no statistical significance, but it decreased from 4.41±0.29 to 4.35±0.32 in the high PNI group (P=0.103), while it increased from 3.74±0.39 to 3.84±0.45 in the low PNI group (P=0.117). Despite these changes in serum albumin levels, the significant decrease in the PNI in all groups was attributed to a significant decrease in the lymphocyte count. Similarly, the NLR increased in both groups for the same reason (P<0.001). According to the medical records, adverse effects reported by patients during CRT were classified according to the RTOG toxicity criteria (Table 3). Symptoms such as anorexia, nausea, and vomiting were classified under the upper gastrointestinal category, whereas fecal incontinence, rectal discomfort, and proctitis were classified under the lower gastrointestinal category. Other categories included the genitourinary, hematologic, and central nervous system categories, resulting in a total of five categories. The proportion of patients who experienced adverse effects was higher in the low PNI group than in the high PNI group, and the difference between the two groups was statistically significant (76.6% vs. 54.8%, P=0.013). In other words, the risk of CRT-induced adverse effects in the high PNI group was 0.370 times lower than that in the low PNI group (odds ratio, 0.370; 95% confidence interval, 0.166–0.824). The most common adverse effects were related to the lower gastrointestinal tract, which was consistent in both groups. However, in the low PNI group, more cases were classified as RTOG grade 1 than RTOG grade 2, which required medication or intervention, whereas the high PNI group had more cases classified as RTOG grade 2. The most commonly performed surgical method in both groups was low anterior resection, and all six cases of the Miles operation were in the low PNI group. The number of patients who underwent transanal excision was second highest in each group, with 17.0% in the low PNI group and 10.7% in the high PNI group (Table 4). The difference in surgical methods between the two groups was statistically significant (P=0.014). The pathological stage and TRG were compared based on the pathology report. Evaluation was conducted using Dworak TRG, and both groups had the highest proportion of moderate regression (TRG2), with slightly more patients with complete regression (TRG4) in the high PNI group (11.9%) than in the low PNI group (8.5%), but it was not statistically significant. Table 5 presents the results of univariate and multivariate analyses conducted on variables that may influence acute adverse effects. In the univariate analysis, the PNI change value (P=0.050) and NLR change value (P=0.040), representing the difference in values before and after CRT, were found to be significant. However, ASA grade and CCI, despite showing a statistically significant difference when comparing low PNI and high PNI, were not found to have an impact on acute adverse effects. In the multivariate analysis, the pre-CRT PNI (P=0.014) and PNI change (P=0.003) were found to be statistically significant. However, NLR showed no meaningful P-value in both the pre-CRT measurement and the change value. DISCUSSION We observed two main findings in this study. First, in patients with LARC, there was a noticeable decrease in lymphocyte count, leading to a decrease in the PNI and an increase in NLR before and after the initiation of long-course CRT. Second, we found that a lower PNI before the initiation of CRT was associated with an increased incidence of acute adverse effects. The prevalence of malnutrition among patients with colorectal cancer is 39.3%, and in patients receiving RT and CRT, the proportion of malnourished patients increases from 44% to 88% [7]. During CRT, fat-free mass is lost, and CRT-induced toxicity occurs, resulting in a decrease in quality of life and decreased survival rates [8,9]. CRT-induced toxicities can be classified into acute and late toxicities; severe late toxicity is associated with weight loss during CRT, whereas the presence of acute toxicity does not increase the risk of late toxicity [10]. We distinguished CRT-induced toxicities in patients with rectal cancer based on symptoms reported by the patients and the use of medications, as recorded in the medical records following the RTOG criteria. Diarrhea was the most common toxicity, followed by upper gastrointestinal toxicities such as anorexia and nausea. These symptoms directly lead to a decrease in dietary intake and worsen malnutrition. The PNI was initially designed to assess the immunonutritional status of patients with gastrointestinal cancer and has since been used for various cancer types. The PNI has been validated as a predictor of postoperative complications and overall survival in patients with colorectal cancer who undergo surgery [11,12]. Serum albumin, which is used to calculate PNI, is a marker of nutritional status. Due to the systemic inflammatory response to tumors, albumin synthesis is suppressed, leading to a rapid decrease in serum albumin levels in response to malnutrition [13]. Lymphocytes play an important role in the host cytotoxic immune response and reflect the systemic inflammatory response to tumors [14]. NLR has also been identified as an independent prognostic factor for progression-free survival and overall survival in colorectal cancer patients receiving neoadjuvant CRT [12]. Similar to our study, other studies have shown a decrease in PNI levels post-CRT compared with pre-CRT in patients with rectal cancer. They also mentioned that the pre-CRT PNI had an impact on overall survival and disease-free survival [15]. Okugawa et al. [16] did not find a significant correlation between the pre-CRT PNI and adverse CRT effects, but a low pre-CRT PNI was an independent risk factor for the ineffectiveness of CRT. However, as the data are not shown, it is difficult to provide a detailed interpretation. In our study, we did not examine the survival rate; however, the acute adverse effects of CRT were more common in the low-PNI group than in the high-PNI group. This finding is consistent with a study conducted in patients with cervical cancer, which showed that as nutritional status worsens, the adverse effects of CRT worsen, leading to a decrease in treatment completion [17]. This is thought to be related to skeletal muscle loss caused by anti-cancer treatments [18]. Studies targeting cervical, head, and neck cancers have shown that clinical nutritional support reduces or prevents the adverse effects of CRT, positively affecting quality of life and prognosis [19,20]. In our study, a significant decrease in lymphocytes was observed before and after CRT; however, the difference in serum albumin levels was not significant. Serum albumin level can predict neutropenia during CRT, but it reflects systemic influences rather than nutritional status [21]. Therefore, serum albumin level alone may be limited as an indicator of malnutrition and the occurrence of adverse effects during CRT. According to the classification based on the pre-CRT PNI, patients with a high PNI were younger, indicating a difference in immunonutritional status according to age. The age difference between the groups also led to statistically significant differences in the CCI and ASA grades. Previous studies have shown that age affects acute RT toxicity [22]. Contrary to our expectations, there were no statistically significant changes in body weight or BMI before and after CRT. Without examining specific changes in body composition, such as changes in weight and BMI, it is not possible to definitively conclude that there were no changes in the patients’ nutritional status. After surgery, the local excision rate and pathological TRG changes were examined to determine whether there were differences in the response to radiation therapy based on the pre-CRT PNI. A P-value of 0.228 indicated no significant difference in the pathological TRG between the two groups. However, in terms of the surgical method, only the low PNI group had patients who underwent Miles’ operation (n=6, 12.8%), whereas the high PNI group had the highest proportion of low anterior resections. Patients with high PNI showed a more significant decrease in PNI and an increase in NLR than those with low PNI, with a greater decrease in neutrophil and lymphocyte counts. This suggests a more pronounced systemic immune response, which may be associated with a higher rate of lower gastrointestinal toxicities (RTOG grade 2) including diarrhea. However, the exact mechanism is unknown, and a detailed explanation is needed regarding the significant differences in PNI and NLR before and after CRT, with higher PNI values before treatment associated with larger differences. This study aimed to minimize bias from external factors through a prospective design and to provide an explanation for these findings. The limitations of our study are as follows: First, it was a retrospective study based on a small number of patients and medical records, and there was a lack of information on patients’ nutritional intake during the CRT period, making it difficult to exclude these variables. Second, the two chemotherapy drug regimens used in CRT varied among patients, leading to uncontrolled bias. In this study, it is observed that the cutoff value of PNI has high sensitivity, resulting in few false negative results. However, there is a limitation of low specificity. Therefore, in subsequent studies, it is necessary to select a cutoff value that not only has high sensitivity but also high specificity. Nevertheless, through this study, we confirmed statistically significant changes in PNI and NLR in patients with LARC receiving neoadjuvant CRT, and we found that measuring pre-CRT PNI and the PNI change that occur during CRT can predict the occurrence of acute adverse effects. A prospective study is needed to investigate the relationship between changes in PNI due to external nutritional support and acute adverse effects. However, it can be expected that reducing PNI decline through external nutritional supplementation would decrease acute adverse effects and potentially improve treatment compliance. Fig. 1 Spaghetti plot and box plot of PNI (A) and NLR (B) before and after neoadjuvant CRT. When analyzed with total 131 patients, left image demonstrates the difference in PNI before and after neoadjuvant CRT, while the right image shows the difference in NLR using Spaghetti plot and box plot. There were statistically significant difference (P<0.001). PNI, prognostic nutritional index; NLR, neutrophil-to-lymphocyte ratio; CRT, chemoradiotherapy. Table 1 Demographic and clinical characteristics of patients Characteristic Type Total (n=131) Low PNI (n=47) High PNI (n=84) P-value Age (yr), mean±SD (range) 70.1±10.7 (44–89) 75.4±8.3 (58–89) 67.1±10.8 (44–88) <0.001 Age group (yr) <65 40 (30.5) 7 (14.9) 33 (39.3) <0.004 ≥65 91 (69.5) 40 (85.1) 51 (60.7) Sex Male 93 (71.0) 34 (72.3) 59 (70.2) 0.799 Female 38 (29.0) 13 (27.7) 25 (29.8) ASA grade 1 25 (19.1) 6 (12.8) 19 (22.6) 0.039 2 92 (70.2) 32 (68.1) 60 (71.4) 3 14 (10.7) 9 (19.1) 5 (6.0) Hypertension Yes 63 (48.1) 25 (53.2) 38 (45.2) 0.382 No 68 (51.9) 22 (46.8) 46 (54.8) Diabetes mellitus Yes 35 (26.7) 16 (34.0) 19 (22.6) 0.156 No 96 (73.3) 31 (66.0) 65 (77.4) Other medical history Yes 40 (30.5) 19 (40.4) 21 (25.0) 0.066 No 91 (69.5) 28 (59.6) 63 (75.0) CCI Mild (1–2) 4 (3.1) 0 4 (4.8) <0.001 Moderate (3–4) 39 (29.8) 5 (10.6) 34 (40.5) Severe (≥5) 88 (67.2) 42 (89.4) 46 (54.7) Clinical T stage T1 1 (0.8) 1 (2.1) 0 0.221 T2 5 (3.8) 0 5 (6.0) T3 50 (38.2) 18 (38.3) 32 (38.1) T4 75 (57.3) 28 (59.6) 47 (56.0) Clinical N stage N0 35 (26.7) 13 (27.6) 22 (26.2) 0.912 N1 77 (58.8) 28 (59.6) 49 (58.3) N2 19 (14.5) 6 (12.8) 13 (15.5) CEA (ng/mL) 7.79±9.45 8.77±12.64 7.25±7.11 0.630 CA19-9 (U/mL) 14.38±19.58 15.27±14.41 13.64±21.62 0.128 Chemoradiotherapy regimen LV+5FU 89 (67.9) 35 (74.5) 54 (62.3) 0.231 Capecitabine 42 (32.1) 12 (25.5) 30 (35.7) Radiotherapy Duration (day) 36.0±6.0 36.0±4.7 36.0±6.6 0.220 Radiotherapy dose 180 cGy×28 fractions 51 (38.9) 23 (48.9) 28 (33.3) 0.079 200 cGy×25 fractions 80 (61.1) 24 (51.1) 56 (66.7) Radiotherapy discontinued Yes 3 (2.3) 0 3 (3.6) 0.553 No 128 (97.7) 47 (100.0) 81 (96.4) Change of BMI ≤–5 17 (13.0) 6 (12.7) 11 (13.1) 0.103 >–5 to <5 97 (74.0) 31 (66.0) 66 (78.6) ≥5 17 (13.0) 10 (21.3) 7 (8.3) PNI, prognostic nutritional index; SD, standard deviation; ASA, American Society of Anesthesiologists; CCI, Charlson Comorbidity Index; CEA, carcinoembryonic antigen; CA19-9, carbohydrate antigen 19-9; LV+5-FU, leucovorin/5-fluorouracil; BMI, body mass index. Table 2 Changes in laboratory tests and immunonutritional markers according to neoadjuvant chemoradiotherapy Index Total patient Low PNI High PNI Pre-CRT Post-CRT P-value Pre-CRT Post-CRT P-value Pre-CRT Post-CRT P-value Body weight (kg) 61.44±10.97 61.47±10.87 0.619 57.63±10.34 58.24±10.56 0.174 63.57±10.79 63.28±10.68 0.377 Body mass index (kg/m2) 0.810 0.155 0.402  Underweight (<18.5) 11 (8.4) 8 (6.1) 10 (21.3) 7 (14.9) 1 (1.2) 1 (1.2)  Normal (18.5 to <23) 59 (45.0) 62 (47.3) 20 (42.6) 21 (44.7) 39 (46.4) 41 (48.8)  Overweight (23 to <25) 23 (17.6) 23 (17.6) 5 (10.6) 6 (12.8) 18 (21.4) 17 (20.2)  Obesity (≥25) 38 (29.0) 38 (29.0) 12 (25.5) 13 (27.7) 26 (31.0) 25 (29.8) Serum albumin (g/dL) 4.17±0.46 4.16±0.44 0.851 3.74±0.39 3.84±0.45 0.117 4.41±0.29 4.35±0.32 0.103 Hemoglobin (g/dL) 13.67±11.85 12.36±1.78 0.203 11.61±1.92 11.74±1.54 0.476 14.83±14.63 12.70±1.82 0.186 Neutrophil count (cells/mm3) 4,638±1,739 3,897±1605 <0.001 4,547±1892 4,179±1,859 0.248 4,688±1,656 3,740±1,432 <0.001 Lymphocyte count (cells/mm3) 2,156±700 1,151±441 <0.001 1,707±538 1,070±388 <0.001 2,407±655 1,197±464 <0.001 PNI 52.47±6.23 47.38±5.04 <0.001 45.96±3.95 43.71±5.15 0.001 56.11±3.84 49.44±3.62 <0.001 NLR 2.35±1.14 3.91±2.65 <0.001 2.86±1.38 4.58±3.35 0.001 2.07±0.85 3.54±2.10 <0.001 Values are presented as mean±standard deviation or number (%). CRT, chemoradiotherapy; PNI, prognostic nutritional index; NLR, neutrophil-lymphocyte ratio. Table 3 Neoadjuvant chemoradiotherapy-induced adverse effects Adverse effects Score Total Low PNI High PNI P-value Radiotherapy induced adverse effects 0.013  No 49 (37.4) 11 (23.4) 38 (45.2)  Yes 82 (62.6) 36 (76.6) 46 (54.8)   Upper gastrointestinal (n=23) 1 18 (22.0) 11 (30.6) 7 (15.3) 2 5 (6.1) 3 (8.3) 2 (4.3)   Lower gastrointestinal (n=50) 1 19 (23.2) 9 (25.0) 10 (21.7) 2 31 (37.8) 8 (22.2) 23 (50.0)   Genitourinary (n=7) 1 2 (2.4) 1 (2.8) 1 (2.2) 2 5 (6.1) 2 (5.5) 3 (6.5)   Hematologic (n=1) 1 - - - 2 - - - 3 1 (1.2) 1 (2.8) -   Central nervous system (n=1) 1 - - - 2 1 (1.2) 1 (2.8) - Values are presented as number (%). The RTOG (Radiation Therapy Oncology Group) grading system assigns a numerical score to each adverse effect based on its severity. The scores range from 0 to 5, with higher scores indicating more severe side effects. Table 4 Differences in postoperative pathologic stage and tumor regression grade between two groups classified by PNI Index Type Total Low PNI High PNI P-value Surgery name Low anterior resection 102 (77.9) 31 (66.0) 71 (84.5) 0.014 Hartmann’s operation 3 (2.3) 1 (2.1) 2 (2.4) Miles’ operation 6 (4.6) 6 (12.8) 0 Transanal excision 17 (12.9) 8 (17.0) 9 (10.7) Palliative loop ileostomy 3 (2.3) 1 (2.1) 2 (2.4) Pathologic stage No residual tumor 13 (9.9) 5(10.6) 8 (9.5) 0.941 1 31 (23.7) 12 (25.5) 19 (22.6) 2 43 (32.8) 16 (34.1) 27 (32.2) 3 25 (19.1) 6 (12.8) 19 (22.6) 4 2 (1.5) 1 (2.1) 1 (1.2) No data 17 (13.0) 7 (14.9) 10 (11.9) Tumor regression grade 0 6 (4.6) 1 (2.1) 5 (5.9) 0.228 1 31 (23.7) 15 (31.9) 16 (19.0) 2 64 (48.8) 19 (40.4) 45 (53.6) 3 16 (12.2) 8 (17.0) 8 (9.5) 4 14 (10.7) 4 (8.5) 10 (11.9) PNI, prognostic nutritional index. Table 5 Univariate and multivariate analysis according to acute adverse effects of CRT Variable Univariate analysis Multivariate analysis OR (95% CI) P-value OR (95% CI) P-value Age 1.172 (0.546–2.514) 0.684 0.887 (0.361–2.178) 0.794 Sex 2.022 (0.881–4.643) 0.097 1.658 (0.680–4.045) 0.267 ASA grade 0.222  ASA grade (1) 1.908 (0.777–4.681) 0.158  ASA grade (2) 0.923 (0.249–3.417) 0.905 CCI 0.508 Pre-CRT BMI 0.289  Pre-CRT BMI (1) 0.402 (0.079–2.036) 0.271  Pre-CRT BMI (2) 0.204 (0.036–1.157) 0.073  Pre-CRT BMI (3) 0.381 (0.072–2.020) 0.257 Clinical cancer stage 0.694  Clinical cancer stage (1) 0.643 (0.053–7.832) 0.729  Clinical cancer stage (2) 0.912 (0.080–10.425) 0.941 Pre-CRT PNI 0.978 (0.924–1.037) 0.461 0.900 (0.827–0.978) 0.014 Pre-CRT NLR 0.930 (0.682–1.269) 0.648 0.889 (0.598–1.321) 0.561 PNI change valuea) 0.926 (0.858–1.000) 0.050 0.848 (0.761–0.945) 0.003 NLR change valueb) 1.232 (1.010–1.504) 0.040 1.155 (0.926–1.440) 0.202 We compare stratification variable based on a reference variable (ASA grade 1, underweight and clinical cancer stage I). e.g., (1) is compared to grade 2, and (2) is compared to grade 3 based on ASA grade 1. CRT, chemoradiotherapy; OR, odds ratio; CI, confidence interval; ASA, American Society of Anesthesiologists physical status classification; CCI, Charlson Comorbidity Index; BMI, body mass index; PNI, prognostic nutritional index; NLR, neutrophil-to-lymphocyte ratio. a) PNI change value=(post-CRT PNI value–pre-CRT PNI value). b) NLR change value=(post-CRT NLR value–pre-CRT NLR value). No potential conflict of interest relevant to this article was reported. FUNDING This study was supported by the Soonchunhyang University Research Fund and a National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIT) (grant nos. 2021 R1G1A1094891). ==== Refs REFERENCES 1 Sung H Ferlay J Siegel RL Laversanne M Soerjomataram I Jemal A Global Cancer Statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries CA Cancer J Clin 2021 71 209 49 33538338 2 Maas M Nelemans PJ Valentini V Das P Rodel C Kuo LJ Long-term outcome in patients with a pathological complete response after chemoradiation for rectal cancer: a pooled analysis of individual patient data Lancet Oncol 2010 11 835 44 20692872 3 Neibart SS Manne SL Jabbour SK Quality of life after radiotherapy for rectal and anal cancer Curr Colorectal Cancer Rep 2020 16 1 10 32632345 4 Lee DU Fan GH Hastie DJ Addonizio EA Suh J Prakasam VN The clinical impact of malnutrition on the postoperative outcomes of patients undergoing colorectal resection surgery for colon or rectal cancer: propensity score matched analysis of 2011–2017 US hospitals Surg Oncol 2021 38 101587 33915485 5 Muscaritoli M Arends J Bachmann P Baracos V Barthelemy N Bertz H ESPEN practical guideline: clinical nutrition in cancer Clin Nutr 2021 40 2898 913 33946039 6 Unsal D Mentes B Akmansu M Uner A Oguz M Pak Y Evaluation of nutritional status in cancer patients receiving radiotherapy: a prospective study Am J Clin Oncol 2006 29 183 8 16601440 7 Hebuterne X Lemarie E Michallet M de Montreuil CB Schneider SM Goldwasser F Prevalence of malnutrition and current use of nutrition support in patients with cancer JPEN J Parenter Enteral Nutr 2014 38 196 204 24748626 8 Chung E Lee HS Cho ES Park EJ Baik SH Lee KY Prognostic significance of sarcopenia and skeletal muscle mass change during preoperative chemoradiotherapy in locally advanced rectal cancer Clin Nutr 2020 39 820 8 30928250 9 Meyer F Fortin A Wang CS Liu G Bairati I Predictors of severe acute and late toxicities in patients with localized head-and-neck cancer treated with radiation therapy Int J Radiat Oncol Biol Phys 2012 82 1454 62 21640495 10 Nourissat A Bairati I Samson E Fortin A Gelinas M Nabid A Predictors of weight loss during radiotherapy in patients with stage I or II head and neck cancer Cancer 2010 116 2275 83 20187097 11 Paku M Uemura M Kitakaze M Fujino S Ogino T Miyoshi N Impact of the preoperative prognostic nutritional index as a predictor for postoperative complications after resection of locally recurrent rectal cancer BMC Cancer 2021 21 435 33879101 12 Yang J Xu H Guo X Zhang J Ye X Yang Y Pretreatment inflammatory indexes as prognostic predictors for survival in colorectal cancer patients receiving neoadjuvant chemoradiotherapy Sci Rep 2018 8 3044 29445100 13 Yeun JY Kaysen GA Factors influencing serum albumin in dialysis patients Am J Kidney Dis 1998 32 6 Suppl 4 S118 25 14 Xu F Meng C Yang Z Li H Gao J Sun L Prognostic nutrition index predicts short-term surgical complications in patients with rectal cancer after laparoscopic surgery Front Surg 2022 9 1000108 36386497 15 Mizutani C Matsuhashi N Tomita H Takahashi T Suetsugu T Tajima JY Predictive value of the prognostic nutritional index in neoadjuvant chemoradiotherapy for rectal cancer Cancer Diagn Progn 2022 2 38 48 35400011 16 Okugawa Y Toiyama Y Oki S Ide S Yamamoto A Ichikawa T Feasibility of assessing prognostic nutrition index in patients with rectal cancer who receive preoperative chemoradiotherapy JPEN J Parenter Enteral Nutr 2018 42 998 1007 29786882 17 Laura FC Lucely CP Tatiana GC Roberto JL Dulce GI Arturo PS Handgrip strength, overhydration and nutritional status as a predictors of gastrointestinal toxicity in cervical cancer patients: a prospective study Nutr Cancer 2022 74 2444 50 35023398 18 Wang HB Xu XT Tian MX Ding CC Tang J Qian Y Prognostic values of the prognostic nutritional index, geriatric nutritional risk index, and systemic inflammatory indexes in patients with stage IIB-III cervical cancer receiving radiotherapy Front Nutr 2023 10 1000326 36937347 19 Cox S Powell C Carter B Hurt C Mukherjee S Crosby TD Role of nutritional status and intervention in oesophageal cancer treated with definitive chemoradiotherapy: outcomes from SCOPE1 Br J Cancer 2016 115 172 7 27328311 20 Qiu Y You J Wang K Cao Y Hu Y Zhang H Effect of whole-course nutrition management on patients with esophageal cancer undergoing concurrent chemoradiotherapy: a randomized control trial Nutrition 2020 69 110558 31526964 21 Yamano T Tomita N Sato T Hayakawa K Kamikonya N Matoba S Influence of chemoradiotherapy on nutritional status in locally advanced rectal cancer: prospective multicenter study Nutrition 2020 77 110807 32402940 22 Monroe AT Reddy SC Gibbs GL White GA Peddada AV Factors associated with radiation-induced nausea and vomiting in head and neck cancer patients treated with intensity modulated radiation therapy Radiother Oncol 2008 87 188 94 18237801
PMC010xxxxxx/PMC10352711.txt
==== Front Korean J Clin Oncol Korean J Clin Oncol Korean Journal of Clinical Oncology 1738-8082 2288-4084 Korean Society of Surgical Oncology 37449394 10.14216/kjco.23003 kjco-19-1-11 Original Article Clinical significance of C-reactive protein-to-prealbumin ratio in predicting early recurrence in resectable pancreatic cancer https://orcid.org/0000-0003-2277-1570 Kwon Chae Hwa 1 https://orcid.org/0000-0002-4132-7662 Seo Hyung Il 12 https://orcid.org/0000-0002-7208-7753 Kim Dong Uk 13 https://orcid.org/0000-0002-0256-9781 Han Sung Yong 13 https://orcid.org/0000-0003-3268-1763 Kim Suk 14 https://orcid.org/0000-0003-1972-2719 Lee Nam Kyung 14 https://orcid.org/0000-0002-1731-0430 Hong Seung Baek 14 https://orcid.org/0000-0002-3312-788X Ahn Ji Hyun 15 https://orcid.org/0000-0002-4165-3054 Park Young Mok 12 https://orcid.org/0000-0002-7764-9516 Noh Byung Gwan 12 1 Biomedical Research Institute, Pusan National University Hospital, Busan, Korea 2 Department of Surgery, Pusan National University Hospital, Pusan National University School of Medicine, Busan, Korea 3 Department of Internal Medicine, Pusan National University Hospital, Pusan National University School of Medicine, Busan, Korea 4 Department of Radiology, Pusan National University Hospital, Pusan National University School of Medicine, Busan, Korea 5 Department of Pathology, Pusan National University Hospital, Pusan National University School of Medicine, Busan, Korea Correspondence to: Hyung Il Seo, Department of Surgery, Pusan National University School of Medicine and Biomedical Research Institute, Pusan National University Hospital, 179 Gudeok-ro, Seo-gu, Busan 49241, Korea, Tel: +82-51-240-7238, Fax: +82-51-247-1365, E-mail: [email protected] 6 2023 30 6 2023 19 1 1117 18 5 2023 18 6 2023 18 6 2023 Copyright © 2023 Korean Society of Surgical Oncology 2023 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. Purpose Resectable pancreatic ductal adenocarcinoma (PDAC) has a high risk of recurrence after curative resection; despite this, the preoperative risk factors for predicting early recurrence remain unclear. This study therefore aimed to identify preoperative inflammation and nutrition factors associated with early recurrence of resectable PDAC. Methods From March 2021 to November 2021, a total of 20 patients who underwent curative resection for PDAC were enrolled in this study. We evaluated the risk factors for early recurrence within 1 year by univariate and multivariate analyses using Cox hazard proportional regression. The cutoff values for predicting recurrence were examined using receiver operating characteristic (ROC) curves. Results In our univariate and multivariate analyses, C-reactive protein (CRP), CRP-albumin ratio, and CRP-prealbumin ratio, as well as sex and age, were significant independent prognostic factors for early recurrence in PDAC. However, known inflammatory factors (neutrophil-to-lymphocyte and platelet-to-lymphocyte ratios), nutritional factors (albumin, prealbumin, ferritin, vitamin D), and inflammatory-nutritional factors (Glasgow Prognostic Score, modified Glasgow Prognostic Score, albumin-bilirubin) showed no association with early recurrence. In addition, using cutoff values by ROC curve analysis, a high preoperative CRP level of >5 mg/L, as well as high CRP-to-albumin (>5.3) and CRP-to-prealbumin (>1.3) ratios showed no prognostic value. Conclusion Our results showed that inflammatory and perioperative nutritional factors, especially CRP-to-prealbumin ratio, have significant associations with early recurrence after curative resection in resectable PDAC. Therefore, for such patients, a cautious approach is needed when inflammation and poor nutritional status are present. Pancreatic neoplasms C-reactive protein Prealbumin Prognosis ==== Body pmcINTRODUCTION Pancreatic ductal adenocarcinoma (PDAC) is a highly aggressive malignancy with a poor prognosis; additionally, recurrence after curative resection remains a major challenge in the management of this disease. Despite advances in diagnostic and therapeutic modalities, the 5-year survival rate for PDAC patients after curative resection remains low, at approximately 20% to 25% [1,2]. Although neoadjuvant chemotherapy (NAC) has shown promising results in borderline resectable PDAC, its benefits are controversial [3]. Early recurrence after curative resection is particularly concerning, as upfront surgery is unlikely to be successful. Therefore, identifying reliable predictors of early recurrence in resectable PDAC patients is crucial for optimal treatment planning and improving patient outcomes. Several prognostic factors such as tumor size, lymph node metastasis, histologic grade, vascular invasion, and perineural invasion have been known to affect outcomes in PDAC [1]. However, these factors have limitations as preoperative predictive factors, as they can only be obtained through postoperative pathologic analyses. Recently, carbohydrate antigen 19-9 (CA19-9) has attracted interest as a prognostic indicator for PDAC, although its usefulness is still debatable [1,4,5]. Although various molecular biomarkers such as gene mutations, exosomes, and miRNA have been identified as potential predictors, there are only a few targets that have scientific bases and practical clinical utility [1,6]. It is also difficult to use them in clinical settings, because discovery of biomarkers is time-consuming, expensive, and resource-intensive. Thus, non-invasive, time-efficient, and cost-effective biomarkers are needed as preoperative predictors of early PDAC recurrence. Recently, some researchers have focused on the potential role of inflammatory and nutritional factors as preoperative predictors of early recurrence in PDAC [7–10]. Inflammatory and preoperative nutritional factors can be used as relatively simple and accessible preoperative predictors. In this study, we aim to explore the correlation between inflammatory and preoperative nutritional factors and early recurrence in resectable PDAC, and to identify patients who may not be suitable candidates for upfront surgery. METHODS Patients Twenty patients who underwent upfront surgery (pancreaticoduodenectomy or radical antegrade modular pancreaticosplenectomy) with resectable PDAC at Pusan National University Hospital between March and December of 2021 were included. Cases with lymph node metastases other than to regional lymph nodes, for which NAC was performed, those who had R2 resections, those with disease recurrence within 3 months post-surgery, and those with other types of malignant diseases (PDAC origin from intraductal papillary mucinous neoplasm, neuroendocrine tumor, mucinous adenocarcinoma, adenosquamous carcinoma and signet ring cell, and other similar malignancies) were excluded. Recurrence within 1 year was defined as early recurrence, and this was found in nine cases. The median duration of follow-up was 15 months. This study was approved by the Pusan National University Hospital Institutional Review Board at the Clinical Trial Center (IRB No. 2303-007-124), and written informed consent was obtained from all participants. Clinical data collection The hemato-chemistry results were based on those performed at the first visit to our center for PDAC evaluation. Pathology results were described by a pathologist specializing in hepatobiliary pancreatic disease. The resection margin status defined a margin of <1 mm as an R1 resection. Preoperative resectability based on chest computed tomography (CT), abdomen CT, magnetic resonance imaging, endoscopic ultrasound, and positron emission tomography-CT was independently determined by three radiologists, three gastroenterologists, and two surgeons who majored in hepatobiliary pancreatic disease. During follow-ups, imaging studies (abdomen and chest CT) and tumor marker tests (CA19-9 and carcinoembryonic antigen [CEA]) were performed every 3–4 months to confirm recurrence. Adjuvant chemotherapy was performed in 17 patients (FOLFIRINOX [15 patients] and gemcitabine and capecitabine [two patients]). Efficacy of inflammation and nutrition status: CRP-albumin/CRP-prealbumin The reference ranges of each value were as follows: C-reactive protein (CRP; <0.5 mg/dL), albumin (3.3–5.2 g/dL), prealbumin (20– 40 mg/dL), total vitamin D (25-OH vitamin D2+D3) (>20 ng/mL), ferritin (15–332 ng/mL), CEA (<5 ng/mL), and CA19-9 (<39 U/mL). CRP-to-albumin and CRP-to-prealbumin ratios were used to evaluate the correlation between inflammation-preoperative nutrition and early recurrence. In addition, the area under the receiver operating characteristic curve (AUC) value was obtained using the receiver operating characteristic curve (ROC; cutoff value of CRP-to-albumin ratio: 5.3, AUC: 72.2%; CRP-to-prealbumin ratio: 1.3, AUC: 61.6%). Statistical analysis Clinical and pathological characteristics of the early recurrence and non-recurrence groups were compared and analyzed using the Mann-Whitney U and chi-square tests for continuous and categorical variables, respectively. Univariate and multivariate analyses were conducted using a Cox proportional hazards model. The optimal cutoff values of CRP-to-albumin and CRP-to-prealbumin ratios were determined by analyzing the ROC curves. The AUC was calculated to determine the discriminatory abilities of the indices. Statistical analyses were performed using R software (version 4.2.1). The R packages “moonbook”, “survminer”, “survival”, “plotROC”, “pROC”, and “rocNIT” were used. The statistical significance was set at P<0.05. RESULTS The characteristics of the 20 patients with PDAC with early recurrence (n=9) and non-recurrence (n=11) are summarized in Table 1. Age and platelet-to-lymphocyte ratio (PLR) showed statistically significant differences between the two groups (P=0.037 and P=0.056, respectively). Although CRP level (2.7 vs. 0.6, P=0.102), CRP-to-albumin ratio (5.7 vs. 1.6, P=0.119), and CRP-to-prealbumin ratio (1.08 vs. 0.21, P=0.177) tended to be higher in patients with early recurrence than in patients with non-recurrence, the differences were not statistically significant. Other preoperative inflammatory and nutritional factors, such as neutrophil-to-lymphocyte ratio (NLR), PLR; levels of albumin and prealbumin, vitamin D levels, ferritin levels, ALBI grade, Glasgow Prognostic Score, and modified Glasgow Prognostic Score did not exhibit significant differences between the groups. No significant associations were found in terms of other clinicopathological factors such as tumor size, T stage, N stage, metastatic lymph node ratio, lymphovascular invasion, perineural invasion, tumor differentiation, margin status, and levels of CEA and CA19-9 as well. Table 2 presents the results of a univariate analysis of preoperative factors associated with early recurrence of PDAC using Cox proportional hazards regression. Sex, age, CRP, CRP-to-albumin ratio, and CRP-to-prealbumin ratio were found to be statistically significant prognostic factors for early recurrence. Female patients had a higher risk of recurrence (hazard ratio [HR]=4.094, P=0.038), and older age was also associated with a higher risk of early recurrence (HR=1.086, P=0.049). Higher levels of CRP, CRP-to-albumin ratio, and CRP-to-prealbumin ratio were associated with a higher risk of early recurrence (HR=1.058, P=0.026; HR=1.027, P=0.019; HR=1.156, P=0.010, respectively). Of these, CRP-to-prealbumin ratio was the most significant prognostic factor. In our multivariate analysis, sex, age, CRP level, CRP-to-albumin ratio, and CRP-to-prealbumin ratio were revealed to be related to early recurrence, with HRs of 59.866 (95% confidence interval [CI]= 2.019–1,774.79, P=0.018), 1.353 (95% CI=1.011–1.812, P=0.042), 20.216 (95% CI=1.567–260.754, P=0.021), 0.140 (95% CI=0.026–0.745, P=0.021), and 25.798 (95% CI=1.598–416.38, P=0.022), respectively–suggesting that these factors are independent preoperative predictors for early recurrence in PDAC patients. Furthermore, the optimal cutoff values of preoperative CRP-to-albumin and CRP-to-prealbumin ratios for predicting early recurrence were set to 5.3 (AUC, 71.2%) and 1.3 (AUC, 61.6%), respectively, by ROC curve analysis (Fig. 1), and patients were divided into two groups according to each cutoff value. Univariate analysis using Cox hazard proportional regression showed that a CRP level of >5 mg/L (HR=7.770, 95% CI=1.857–32.520, P=0.005), a CRP-to-albumin ratio of >5.3% (HR=5.201, 95% CI=1.346–20.100, P=0.017) and a CRP-to-prealbumin ratio of >1.3% (HR=10.380, 95% CI=2.178–49.440, P=0.003) were associated with early recurrence (Table 3). However, in multivariate analysis, these factors based on cutoff values did not show prognostic significance for early recurrence. DISCUSSION Even if PDAC is resectable, the overall prognosis remains poor. Although the benefits of borderline resectable PDAC have been established, the debate over whether to perform upfront surgery or NAC in cases of resectable PDAC is ongoing. Surgeons are continuously struggling with the possibilities of missing the timing of surgery due to progression during NAC, or performing unnecessary surgery after early recurrence following upfront surgery. Moreover, PDAC is a very heterogeneous disease, and the uniformity of treatment is still questionable. Thus, by identifying preoperative biomarkers that can detect a high risk of early recurrence after upfront surgery in resectable PDAC, clinicians can better stratify patients, improving their outcomes. In this study, we found that sex, age, CRP, CRP-to-albumin ratio and CRP-to-prealbumin ratio were significant preoperative predictors of early recurrence in PDAC patients. Although several factors have been identified as potential predictors of early recurrence of PDAC, the relationship between age and sex and PDAC prognosis remains unclear. The reason for this may reflect underlying differences in tumor biology or patient characteristics. One possible explanation for the sex- and age-based differences in recurrence risk is the influence of hormonal and immune factors. Sex hormones such as estrogen and testosterone have been shown to play a role in modulating the immune response [11], and may affect the progression of PDAC. In addition, aging is associated with changes in the immune system, which may contribute to the higher risk of recurrence in older patients [12]. Recently, Xu et al. [13] reported that in the case of older age, poor preoperative nutritional status and slow recovery after upfront surgery prevented or delayed adjuvant treatment, which could affect the prognosis. Due to the small number of enrolled patients in this study, subgroup analysis by age and sex was not possible; thus, further studies are needed to propose a clear reason for this. Inflammation and nutritional status are closely interrelated in cancer patients. Inflammation plays a critical role in the development and progression of cancer, as well as in the deterioration of nutritional status [14]. Increasing evidence has proven that inflammation markers such as CRP, NLR, and PLR, as well as preoperative nutritional markers such as albumin, prealbumin, ferritin, and vitamin D are known to be related to carcinogenesis and the progression of cancer [15–17]. However, in this study, of these markers, CRP was the only independent predictor of early recurrence in PDAC. CRP, an acute phase marker of inflammation, is mainly synthesized in the liver and induced by proinflammatory cytokines including interleukin-6 and tumor necrosis factor-α, and is commonly elevated in many cancer patients [18]. Albumin and prealbumin are used to assess nutritional status, and are decreased during inflammation due to the transfer of plasma proteins to the reactants of the acute phase, such as CRP [19]. Recent studies have reported that CRP-to-albumin and CRP-to-prealbumin ratios, using the combination of inflammation and preoperative nutritional factors, were significant predictors of various cancer prognoses [20–22]. In this study, not only CRP, but also CRP-to-albumin and CRP-to-prealbumin ratios were found to be independent prognostic factors of early recurrence of PDAC, while other preoperative inflammatory and nutritional factors were not found to have significant associations. CRP-to-albumin ratio has been reported as a prognostic factor for pancreatic cancer [9,20,23]. Compared to albumin, prealbumin has a shorter half-life of about 2 to 3 days, and is considered to be a more sensitive biomarker for assessing changes in nutritional status over a shorter period of time [24]. Prealbumin levels can decrease rapidly in response to nutritional deficiency or inflammation, which makes it accurate biomarker for monitoring nutritional status. Recent studies have reported that prealbumin is superior to albumin and an independent predictor in cancer patients [24–26]. In our previous study, we found that preoperative low prealbumin levels predicted advanced disease while confirming poor nutritional status in patients with low body mass index and hemoglobin levels who had hepatobiliary pancreatic malignant disease [27]. Based on these findings, the combination of CRP and prealbumin may prove to be a useful biomarker for predicting recurrence and prognosis, but its prognostic value has not yet been elucidated in PDAC patients. This study, to the best of our knowledge, is the first to demonstrate the CRP-to-prealbumin ratio is an independent predictor of early recurrence in PDAC. In this study, CRP-to-albumin and CRP-to-prealbumin ratios did not show significant prognostic effects when using cutoff values of 5.3 and 1.3, respectively. Other groups have suggested different cutoff values of the CRP-to-albumin ratio for predicting survival and monitoring chemotherapy responses in PDAC patients [20–23]. Estimation of accurate cutoff values needs to be done using large number of patients. It is also important to carefully consider the interpretation of the cutoff value used in each study. Nevertheless, in this study, CRP-to-albumin and CRP-to-prealbumin were still identified as independent prognostic factors for the early recurrence of PDAC. In addition, the relationship between inflammation and preoperative nutritional status may reflect certain aspects of tumor biology. However, several limitations do remain inherent to our study, such as its small sample size and the fact that it was conducted at a single center with a single operator. In addition, the inclusion of PDACs in various locations may have introduced some heterogeneity to our findings. In conclusion, our study suggests that markers of inflammation and preoperative nutritional status may serve as useful predictors of early recurrence in cases of resectable PDAC, in particular the CRP-to-prealbumin ratio. These findings highlight the importance of evaluating patients’ inflammatory and nutritional statuses prior to deciding on an appropriate treatment approach. In cases with high risks of early recurrence, NAC could be considered instead of upfront surgery. Further studies are needed to confirm these findings, and to identify optimal treatment strategies for patients with resectable PDAC. Fig. 1 The receiver operating characteristic curves and area under curve (AUC) of C-reactive protein (CRP)-to-albumin ratio (A) and CRP-to-prealbumin ratio (B) for early recurrence. The optimal cutoff values were 5.3 (specificity 90.9%, sensitivity 55.6%) and 1.3 (specificity 100%, sensitivity 44.4%), respectively. Table 1 Characteristics of 20 PDAC patients with early recurrence and non-recurrence Characteristic Type Non-recurrence (n=11) Recurrence (n=9) P-value Sex Female 2 5 0.203 Male 9 4 Age (yr), median (range) 63 (56–78) 77 (59–83) 0.037a) Size (cm), median (IQR) 3.8 (3.00–5.15) 4.1 (4.00–4.50) 0.675 T stage T3 5 6 0.619 T2 6 3 N stage N1, N2 8 8 0.736 N0 3 1 LN ratio, median (IQR) 4.17 (1.11–12.48) 15.00 (5.88–22.22) 0.139 LVI Positive 8 8 0.736 Negative 3 1 PNI Positive 11 9 NA Negative 0 0 Differentiation Poor 2 2 1.000 Well, moderate 9 7 Margin status ≥1 mm 6 6 0.927 <1 mm 5 3 CEA ≥5 ng/mL 3 4 0.742 <5 ng/mL 8 4 CA19-9 ≥39 U/L 4 6 0.369 <39 U/L 7 3 CRP (mg/L), median (IQR) 0.60 (0.40–1.80) 2.70 (0.60–16.20) 0.102 NLR, median (IQR) 2.26 (1.66–3.91) 2.21 (1.68–2.56) 0.603 PLR, median (IQR) 150.8 (120.4–170.3) 100.6 (89.0–130.8) 0.056 Albumin <3.3 or >5.2 mg/dL 0 0 NA 3.3–5.2 mg/dL 11 9 Prealbumin 20 mg/dL 2 4 0.330 20–40 mg/dL 9 4 Vitamin D ≤20 ng/mL 5 5 0.756 >20 ng/mL 6 4 Ferritin ≤15 or ≥332 ng/mL 1 3 0.432 15–332 ng/mL 10 6 CRP-to-albumin ratio, median (IQR) 1.60 (1.05–3.95) 5.70 (1.50–35.40) 0.119 CRP-to-prealbumin ratio, median (IQR) 0.21 (0.16–0.68) 1.08 (0.21–11.70) 0.177 ALBI grade 2 1 1 1.000 1 10 8 GPS 1 5 6 0.619 0 6 3 mGPS 1 4 6 0.369 0 7 3 PDAC, pancreatic ductal adenocarcinoma; IQR, interqurtile range; LN, lymph node metastasis; LVI, lymphovascular invasion; PNI, perineural invasion; CEA, carcinoembryonic antigen; CA19-9, carbohydrate antigen 19-9; CRP, C-reactive protein; NLR, neutrophil-to-lymphocyte ratio; PLR, platelet-to-lymphocyte ratio; ALBI, albumin-bilirubin; GPS, Glasgow Prognostic Score; mGPS, modified GPS; NA, not applicable. a) P<0.05. Table 2 Univariate and multivariate analyses of preoperative risk factors for early recurrence Univariate analysis Multivariate analysis HR (95% CI) P-value HR (95% CI) P-value Sex (female) 4.094 (1.083–15.47) 0.038a) 59.866 (2.019–1,774.79) 0.018a) Age 1.086 (1.000–1.179) 0.049a) 1.353 (1.011–1.812) 0.042a) Size 0.926 (0.697–1.230) 0.596 T stage 1.816 (0.452–7.294) 0.400 N stage 2.163 (0.799–5.856) 0.129 LN ratio (%) 1.058 (0.994–1.126) 0.077 LVI (positive) 2.842 (0.353–22.890) 0.326 PNI (positive) NA NA Differentiation (poor) 1.259 (0.260–6.101) 0.775 Margin status (≥1 mm) 1.597 (0.398–6.417) 0.509 CEA (≥5 ng/mL) 1.759 (0.471–6.577) 0.401 CA19-9 (≥39 U/L) 2.586 (0.637–10.490) 0.184 CRP 1.058 (1.007–1.112) 0.026a) 20.216 (1.567–260.754) 0.021a) NLR 0.856 (0.476–1.540) 0.604 PLR 0.986 (0.968–1.005) 0.153 Albumin 1.021 (0.164–6.346) 0.982 Prealbumin 0.904 (0.804–1.018) 0.095 Vitamin D (≤20 ng/mL) 1.233 (0.330–4.599) 0.756 Ferritin (≤15 or ≥332 ng/mL) 2.957 (0.717–12.200) 0.134 CRP-to-albumin ratio (%) 1.027 (1.004–1.049) 0.019a) 0.140 (0.026–0.745) 0.021a) CRP-to-prealbumin ratio (%) 1.156 (1.035–1.292) 0.010b) 25.798 (1.598–416.381) 0.022a) ALBI grade 1.449 (0.180–11.640) 0.727 GPS 2.177 (0.542–8.741) 0.272 mGPS 2.851 (0.708–11.480) 0.140 HR, hazard ratio; CI, confidence interval; LN, lymph node metastasis; LVI, lymphovascular invasion; PNI, perineural invasion; CEA, carcinoembryonic antigen; CA19-9, carbohydrate antigen 19-9; CRP, C-reactive protein; NLR, neutrophil-to-lymphocyte ratio; PLR, platelet-to-lymphocyte ratio; ALBI, albumin-bilirubin; GPS, Glasgow Prognostic Score; mGPS, modified GPS; NA, not applicable. a) P<0.05. b) P<0.01. Table 3 Univariate and multivariate analyses of risk factors based on cutoff value for early recurrence Univariate analysis Multivariate analysis HR (95% CI) P-value HR (95% CI) P-value CRP (≥5 mg/L) 7.770 (1.857–32.520) 0.005b) 0.503 (0.034–7.526) 0.618 CRP-to-albumin ratio (≥5.3) 5.201 (1.346–20.100) 0.017a) 5.66E–09 (0–inf) 0.999 CRP-to-prealbumin ratio (≥1.3) 10.380 (2.178–49.440) 0.003b) 3.41E+09 (0–inf) 0.999 HR, hazard ratio; CI, confidence interval; CRP, C-reactive protein. a) P<0.05. b) P<0.01. No potential conflict of interest relevant to this article was reported. FUNDING This work was supported by clinical research grant from Pusan National University Hospital in 2023. ==== Refs REFERENCES 1 Vincent A Herman J Schulick R Hruban RH Goggins M Pancreatic cancer Lancet 2011 378 607 20 21620466 2 Seppanen H Juuti A Mustonen H Haapamaki C Nordling S Carpelan-Holmstrom M The results of pancreatic resections and long-term survival for pancreatic ductal adenocarcinoma: a single-institution experience Scand J Surg 2017 106 54 61 27130378 3 Seufferlein T Ettrich TJ Treatment of pancreatic cancer-neoadjuvant treatment in resectable pancreatic cancer (PDAC) Transl Gastroenterol Hepatol 2019 4 21 30976724 4 Azizian A Ruhlmann F Krause T Bernhardt M Jo P Konig A CA19-9 for detecting recurrence of pancreatic cancer Sci Rep 2020 10 1332 31992753 5 Binicier OB Pakoz ZB CA 19-9 levels in patients with acute pancreatitis due to gallstone and metabolic/toxic reasons Rev Assoc Med Bras (1992) 2019 65 965 70 31389506 6 Giannis D Moris D Barbas AS Diagnostic, predictive and prognostic molecular biomarkers in pancreatic cancer: an overview for clinicians Cancers (Basel) 2021 13 1071 33802340 7 Fu YJ Li KZ Bai JH Liang ZQ C-reactive protein/albumin ratio is a prognostic indicator in Asians with pancreatic cancers: a meta-analysis Medicine (Baltimore) 2019 98 e18219 31770284 8 Zang Y Fan Y Gao Z Pretreatment C-reactive protein/albumin ratio for predicting overall survival in pancreatic cancer: a meta-analysis Medicine (Baltimore) 2020 99 e20595 32502031 9 Liu Z Jin K Guo M Long J Liu L Liu C Prognostic value of the CRP/Alb ratio, a novel inflammation-based score in pancreatic cancer Ann Surg Oncol 2017 24 561 8 10 Lu J Xu BB Zheng ZF Xie JW Wang JB Lin JX CRP/prealbumin, a novel inflammatory index for predicting recurrence after radical resection in gastric cancer patients: post hoc analysis of a randomized phase III trial Gastric Cancer 2019 22 536 45 30377862 11 Orzołek I Sobieraj J Domagała-Kulawik J Estrogens, cancer and immunity Cancers (Basel) 2022 14 2265 35565393 12 Foster AD Sivarapatna A Gress RE The aging immune system and its relationship with cancer Aging health 2011 7 707 18 22121388 13 Xu Y Zhang Y Han S Jin D Xu X Kuang T Prognostic effect of age in resected pancreatic cancer patients: a propensity score matching analysis Front Oncol 2022 12 789351 35433408 14 Diakos CI Charles KA McMillan DC Clarke SJ Cancer-related inflammation and treatment effectiveness Lancet Oncol 2014 15 e493 503 25281468 15 An X Ding PR Li YH Wang FH Shi YX Wang ZQ Elevated neutrophil to lymphocyte ratio predicts survival in advanced pancreatic cancer Biomarkers 2010 15 516 22 20602543 16 Aliustaoglu M Bilici A Seker M Dane F Gocun M Konya V The association of pre-treatment peripheral blood markers with survival in patients with pancreatic cancer Hepatogastroenterology 2010 57 640 5 20698242 17 Geng Y Qi Q Sun M Chen H Wang P Chen Z Prognostic nutritional index predicts survival and correlates with systemic inflammatory response in advanced pancreatic cancer Eur J Surg Oncol 2015 41 1508 14 26343824 18 Gabay C Kushner I Acute-phase proteins and other systemic responses to inflammation N Engl J Med 1999 340 448 54 9971870 19 Ingenbleek Y Young VR Significance of transthyretin in protein metabolism Clin Chem Lab Med 2002 40 1281 91 12553432 20 Fan Z Fan K Gong Y Huang Q Yang C Cheng H The CRP/albumin ratio predicts survival and monitors chemotherapeutic effectiveness in patients with advanced pancreatic cancer Cancer Manag Res 2019 11 8781 8 31632137 21 Kinoshita A Onoda H Imai N Iwaku A Oishi M Tanaka K The C-reactive protein/albumin ratio, a novel inflammation-based prognostic score, predicts outcomes in patients with hepatocellular carcinoma Ann Surg Oncol 2015 22 803 10 25190127 22 Liu X Sun X Liu J Kong P Chen S Zhan Y Preoperative C-reactive protein/albumin ratio predicts prognosis of patients after curative resection for gastric cancer Transl Oncol 2015 8 339 45 26310380 23 Murakawa M Yamamoto N Kamioka Y Kamiya M Kobayashi S Ueno M Clinical implication of pre-operative C-reactive protein-albumin ratio as a prognostic factor of patients with pancreatic ductal adenocarcinoma: a single-institutional retrospective study In Vivo 2020 34 347 53 31882498 24 Unal D Orhan O Eroglu C Kaplan B Prealbumin is a more sensitive marker than albumin to assess the nutritional status in patients undergoing radiotherapy for head and neck cancer Contemp Oncol (Pozn) 2013 17 276 80 24596514 25 Kawai H Ota H Low perioperative serum prealbumin predicts early recurrence after curative pulmonary resection for non-small-cell lung cancer World J Surg 2012 36 2853 7 22948197 26 Han WX Chen ZM Wei ZJ Xu AM Preoperative pre-albumin predicts prognosis of patients after gastrectomy for adenocarcinoma of esophagogastric junction World J Surg Oncol 2016 14 279 27809860 27 Park YM Seo HI Noh BG Kim S Hong SB Lee NK Clinical impact of serum prealbumin in pancreaticobiliary disease Korean J Clin Oncol 2022 18 61 5 36945244
PMC010xxxxxx/PMC10352712.txt
==== Front RSC Adv RSC Adv RA RSCACL RSC Advances 2046-2069 The Royal Society of Chemistry d3ra01581e 10.1039/d3ra01581e Chemistry Chitosan-coated halloysite nanotube magnetic microspheres for carcinogenic colorectal hemorrhage and liver laceration in albino rats Majeed Sajid ab Qaiser Muhammad af Shahwar Dure a https://orcid.org/0000-0003-1025-5800 Mahmood Khalid b Ahmed Nadeem c https://orcid.org/0000-0002-2303-3804 Hanif Muhammad a Abbas Ghulam d https://orcid.org/0000-0002-3639-0026 Shoaib Muhammad Harris e Ameer Nabeela a Khalid Muhammad g a Department of Pharmaceutics, Faculty of Pharmacy, Bahauddin Zakariya University Multan Pakistan [email protected] b Institute of Chemical Sciences, Bahauddin Zakariya University Multan Pakistan [email protected] c Center for Excellence in Molecular Biology, University of Punjab Pakistan d Faculty of Pharmacy, GOVT College University Faisalabad Pakistan e Faculty of Pharmacy, University of Karachi Pakistan f Drug Testing Laboratory Punjab Multan Pakistan g Department of Chemistry, Khwaja Fareed University of Engineering and Information Technology Rahim Yar Khan Pakistan 18 7 2023 12 7 2023 18 7 2023 13 31 2152121536 10 3 2023 16 6 2023 This journal is © The Royal Society of Chemistry 2023 The Royal Society of Chemistry https://creativecommons.org/licenses/by-nc/3.0/ Carcinogenic colorectal hemorrhage can cause severe blood loss and longitudinal ulcer, which ultimately become fatal if left untreated. The present study was aimed to formulate targeted release gemcitabine (GC)-containing magnetic microspheres (MM) of halloysite nanotubes (MHMG), chitosan (MCMG), and their combination (MHCMG). The preparation of MM by magnetism was confirmed by vibrating sample magnetometry (VSM), the molecular arrangement of NH2, alumina, and silica groups was studied by X-ray diffraction (XRD) and energy-dispersive spectroscopy (EDS), the hollow spherical nature of the proposed MM was observed by scanning electron microscopy (SEM), functional groups were characterized by Fourier transform infrared (FTIR) spectroscopy and thermochemical modification was studied by thermogravimetric analysis (TGA). In vitro thrombus formation showed a decreasing trend of hemostatic time for MMs in the order of MHMG3 < MCMG3 < MHCMG7, which was confirmed by whole blood clotting kinetics. Interestingly, rat tail amputation and liver laceration showed 3 folds increased clotting efficiency of optimized MHCMG7 compared to that of control. In vivo histopathological studies and cell viability assays confirmed the regeneration of epithelial cells. The negligible systemic toxicity of MHCMG7, more than 90% entrapment of GC and high % release in alkaline medium made the proposed MM an excellent candidate for the control of hemorrhage in colorectal cancer. Conclusively, the healing of muscularis and improved recovery of the colon from granulomas ultimately improved the therapeutic effects of GC-containing MMs. The combination of both HNT and CTS microspheres made them more targeted. Sever blood lose in carcinogenic colorectal hemorrhage due to longitudinal ulcer, ultimately become fatal if left untreated. Bahauddin Zakariya University 10.13039/100007713 Unassigned pubstatusPaginated Article ==== Body pmc1 Introduction Colorectal cancer (CRC) is the 3rd most common cancer in men and 2nd most common cancer in women, with an increasing average of 10% per year across the world. The onset of CRC is strictly increasing due to aging, smoking, alcohol abuse, poor diet, diabetes mellitus, and increased body mass index (BMI). Treatments such as cancer chemotherapy immunotherapy, surgery, radiation, and target drug delivery are the only available options, but they still have many drawbacks with reoccurrence. Therefore, there is a necessity to develop a treatment that not only has a therapeutic effect against colorectal cancer but also equally effective in reducing colorectal hemorrhage.1 The sustained effects of microspheres containing biodegradable polymers and their micro composites with clay-like minerals having ability to release a drug specifically in the colon make the MMs most suitable due to their increased surface area and pH-sensitive release.2 Grafting magnetic microspheres with chitosan not only reduces their toxicity but also decreases the self-flocculation of magnetic microspheres.3 The stability of dosage form, production process conditions, oxidation, hydrolysis, and the effect of gastrointestinal (GIT) pH are some limitations of simple microspheres, which can be overcome by preparing magnetic microspheres (MM). High mechanical resistance, ease of manufacturing, thermal stability, and excellent shelf life made MMs more targeted than simple microspheres.4,5 Halloysite nanotubes are effective as catalytic supports, pollutant adsorbents, and nanocarriers for functional compounds with biological and chemical activities because of their morphological properties.6 The average length of halloysite nanotubes (HNTs) is 2.01 m, and their inner and outer diameters, respectively, range from 6 to 60 nm and 30 to 160 nm. In this context, it is important to emphasize that halloysite is appropriate for biomedical and pharmaceutical applications due to its biocompatibility and low toxicity, which were seen in both unicellular and multicellular species.7 The catalytic application,8 drug release carrier,7 and restoration properties9 of HNTs were already reported in previous studies. The problem of limited functional group binding of MMs was resolute by the addition of biodegradable polymers such as chitosan and halloysite nanotubes.10 Chitosan (CTS) can be a possible option due to its strong chelating nature and ability to interact with various cross-linkers for microspheres preparation.11 Biomedical polymers have the advantages of being degraded in the GIT, as already demonstrated, and additionally, they can protect colorectal anticancer drugs from the acidic pH. CTS and its complex with halloysite nanotubes (HNTs) can be used for the preparation of Ms. The HNTs have gained interest in nanotechnology applications due to their freely available and tubular structure. HNTs are negatively charged natural polymers having a tube or spherical sheet-like structure obtained from residues, and composed of inner alumina octahedrons and outer silicate tetrahedrons layers. MMs are constituted by the interaction of magnetic materials with natural and synthetic polymers. MMs are supramolecular particles with size less than 40 to 170 μm, and the ferromagnetic properties of the magnetic field, 0.5–0.8 tesla, can be a better option for the targeted release of colorectal cancer drugs.12,13 In our previous study, the CTS–HNT complex showed a positive response to control the clotting time.14 As far as we are aware, there has not been any research done on the targeted release of gemcitabine (GC) for the treatment of colorectal cancer under the direction of the CTS–HNT magnetic complex that would have enhanced hemostatic qualities and more desirable cell regeneration potential. Colorectal cancer is the most advanced type of cancer worldwide. The mechanism of action of GC is to inhibit DNA synthesis as it is a nucleoside analog during the S-phase of the cell cycle. The GC is incorporated into DNA, and thus inhibits DNA replication in a cancerous cell. Moreover, it also promotes the apoptosis of cancerous cells. Considering the foregoing work, the goal of the current work was to construct reproducible core/shell microspheres consisting of Fe3O4 with CTS and HNTs, loaded with GC. The application of the co-precipitation technique, analysis of magnetism, confirmation of thermodynamic surface characteristics, and hemostatic properties of the prepared MM made the study more interesting.15,16 The size and zeta potential of the prepared microspheres were also measured, and their toxicity was determined using a cell viability assay. In vitro drug release profiles were characterized according to the drug loading and release procedure. In addition, the rat tail amputation and liver laceration activity of prepared MM and histopathological studies were also performed to confirm its uses for the targeted release of anticancer drugs in colorectal patients. 2. Materials and methods 2.1. Materials Gemcitabine was provided by Novartis Pharma (Pvt.) Ltd Karachi. Halloysite nanotubes (MW ≈ 610.52 Da) with an average length of 2.01 μm, chitosan (MW ≈ 180–310 kDa), iron(iii) hydrochloride hexahydrate (FeCl3·6H20, MW ≈ 270.195 Da), iron(ii) hydrochloride tetrahydrate (FeCl2·4H2O, MW ≈ 198.751 Da), 32% ammonia (MW ≈ 35.04 Da), Triton X-100, and 25 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES)-buffered saline of HPLC grade used for the cell viability studies were purchased from Sigma Aldrich, Germany. Glutaraldehyde (MW ≈ 100.11 Da) was purchased from BDH, UK; toluene, paraffin, span 80 (MW ≈ 1310 Da) as an emulsifier, petroleum ether (MW ≈ 82.2 Da), acetic acid (MW ≈ 60.052 Da), ethanol (MW ≈ 46.07 Da), methanol (MW ≈ 32.042 Da), para-phenylenediamine (PPD, 99.8%), sodium tetraborate, potassium dihydrogen phosphate (KH2PO4), sodium hydroxide, orthophosphoric acid, and a membrane filter (0.22–0.45 μm) were procured from Merck Darmstadt, Germany. Reverse osmosis, deionized, and double-distilled water were gifted by the Drugs Testing laboratory of Multan, Pakistan. 2.2. Methods 2.2.1. Magnetic microspheres (MM) with and without HNTs A slight modification of the previously reported co-precipitation process was used for synthesis by Ma Dai et al.17 Briefly, an aqueous suspension of FeCl3·6H2O and FeCl2·4H2O in a ratio of 2.4 : 1.0 was prepared at 60 °C in a controlled nitrogen environment. Ammonia solution (32%) was added dropwise after achieving the alkaline pH 9 with 1 M NaOH. The final mixture was heated at 70 °C for 4 h. This resulted in Fe3O4 microspheres, which were washed thrice with double-distilled water, dried at 60 °C in a vacuum oven, and stored in a well-sealed container for further use. A similar process was repeated with the addition of different concentrations of the preactivated HNT suspension (shown in Table 1) in the aqueous suspension of Fe3O4 and resulted in HNTs/Fe3O4 (MHM1 to MHM5) microspheres that were washed with water and stored in sealed containers.18 Different formulations of gemcitabine (GC) loaded magnetic microspheres (MM) with their physicochemical properties and TGA profile of 50 mg of each sample with RM (residual mass wt%), DM (degraded mass wt%) and LM (mass loss wt%) at 180 °C and 800 °C Gemcitabine loaded MM Formulation code HNT (%) CTS (%) Fe3O4 microspheres (% yield) Hausner's ratio Gemcitabine loading (%) FeCl2 & FeCl3 (2.4 : 1) HNTs/Fe3O4/GC MHMG1 15 0 85 1.12 ± 0.01 49.90 ± 1.20 MHMG2 17 0 83 1.10 ± 0.02 60.95 ± 1.06 MHMG3 19 0 81 1.01 ± 0.03 72.24 ± 0.98 MHMG4 21 0 79 1.02 ± 0.02 51.75 ± 0.94 MHMG5 23 0 77 1.02 ± 0.01 65.46 ± 0.89 CTS/Fe3O4/GC MCMG1 0 15 85 1.03 ± 0.01 40.25 ± 0.78 MCMG2 0 17 83 1.02 ± 0.02 56.37 ± 0.96 MCMG3 0 19 81 1.01 ± 0.02 65.21 ± 1.03 MCMG4 0 21 79 1.02 ± 0.02 65.08 ± 0.97 MCMG5 0 23 77 1.00 ± 0.01 70.71 ± 0.86 CTS/Fe3O4/HNTs/GC MHCMG1 12 12 76 1.02 ± 0.02 44.20 ± 3.23 MHCMG2 11 12 77 1.03 ± 0.01 55.57 ± 2.83 MHCMG3 10 12 78 1.03 ± 0.01 56.46 ± 2.56 MHCMG4 10 14 76 1.04 ± 0.02 57.08 ± 2.87 MHCMG5 10 15 75 1.20 ± 0.02 65.97 ± 2.88 MHCMG6 10 16 74 1.10 ± 0.01 75.54 ± 3.04 MHCMG7 15 15 70 1.01 ± 0.02 90.34 ± 2.76 MHCMG8 14 14 72 1.02 ± 0.01 80.14 ± 3.22 MHCMG9 13 13 74 1.04 ± 0.02 79.93 ± 3.45 Thermogravimetric parameters Materials LM180/wt% RM800/wt% DM800/wt% MHMG3 05 ± 0.5 50 ± 0.5 45 ± 0.5 MCMG3 10 ± 0.5 35 ± 0.5 55 ± 0.5 MHCMG7 06 ± 0.5 10 ± 0.5 15 ± 0.5 2.2.2. Chitosan MMs with and without HNTs Chitosan (CTS) containing MM (CTS/Fe3O4) were prepared by an already reported method of E. Türkeş et al. with slight modifications.19 Briefly, 2% m/v CTS solution in 2% acetic acid was added into 1 g of already prepared Fe3O4 microspheres (CTS/Fe3O4). A paraffin and span 80 (100 : 8, emulsifier) mixture was added into CTS/Fe3O4 with continuous stirring at 1500 rpm at 40 °C for 2 h. Finally, 25% glutaraldehyde saturated with toluene (GST) was added for crosslinking purposes with continuous stirring at 60 °C for 1 h. This resulted in CTS/Fe3O4microspheres being centrifuged and washed with petroleum ether, ethanol, and distilled water respectively. Dried microspheres of CTS/Fe3O4 (MCM1 to MCM5) were stored in a well-sealed container for further use.20 A similar procedure was used for the preparation of CTS/Fe3O4/HNTs with the addition of HNTs (MHM1 to MHM5) and a combination of both CTS and HNTs (MHCM1 to MHCM9), as shown in Table 1. The % yield was calculated using eqn (1):1 “Wmg” is the weight of the CTS/HNTs microspheres and “wtmg” is the weight of all contents used in preparation. 2.2.3. Physicochemical properties Bulk density and tapped density of gemcitabine (GC)-unloaded HNTs/Fe3O4, CTS/Fe3O4, and CTS/Fe3O4/HNTs, and GC-loaded CTS/Fe3O4/GC, HNTs/Fe3O4/GC, and CTS/Fe3O4/HNTs/GC microspheres were calculated using a simple glass apparatus. Briefly, 1000 mg of all three types of MM were used for bulk and tapped densities, Carr's index, and angle of repose:2 3 4 5 Hausner's ratio = ρtap/ρb 6 tan ∅ = 2 h/d where “m” is the powder mass in g, “Vb” and “Vt”are powder bulk volume and tapped volume in cm−3, ρtap and ρtap are the tapped and bulk density, and h and d are the cone height and diameter of the powder heap respectively. For powder having good flow, Hausner's ratio should be less than 1.25. 2.2.4. Electromagnetic properties The electromagnetic behaviors of MHM (1–5), MCM (1–5), MHCM (1–9), MHMG (1–5), MCMG (1–5), and MHCMG (1–9) microspheres were calculated using a vibrating sample magnetometer (7400 series). Briefly, 100 mg sample powder was placed in a Teflon sample holder and magnetic field 1 μ at 25 °C was applied and difference of magnetism among GC-loaded and GC-unloaded microspheres was observed using a cryogenic limited device (PPMS) of a vibrating sample magnetometer under 10 kOe magnetic fields at room temperature, providing a powerful investigation device having a capacity of low temperature 1 K to 7 tesla magnetic fields. Magnetic contents in the solution were assayed at 248 nm by AAS and % of contents can be calculated using eqn (2):7 “PPm” is the result shown by the instrument and “mL” is the volume for digestion, and 100 mg weight of each sample was used. 2.2.5. Structural and techno-functional properties The structure of all proposed microspheres and their components, i.e., Fe3O4, CTS, and HNTs, with and without GC was confirmed by X-ray diffraction (XRD) studies. First, 100 mg samples were placed in a copper holder for their exposure to 40 mA × 40 kV radiation by using a wide-angle instrument (JDS 3525, Jeol Japan),21 and differences in the peaks were observed. The thermal stability of all samples was investigated using a ZTY-ZP type thermal analyzer (model TGA Q500, Hullhorst, Germany). Briefly, 50 mg of each sample was used and heated from 50 to 800 °C with a temperature variation rate of 10 °C min−1 in an N2 environment. The degraded mass was calculated using the following formula:8 DM800 wt% = 100 − (LM180 wt% + RM800 wt%) where DM800 wt% is the degraded mass at 800 °C, LM180 wt% is the loss mass at 180 °C and RM800 wt% is the residual mass of samples at 800 °C respectively.7 Functionalization was confirmed using a diamond-based ATR-FTIR spectrometer (Bruker alpha, Germany). An average of 12 scans with 4 cm−1 resolution were calculated and the spectra ranging from 4000 to 400 cm−1 were reported at room temperature. Elemental and morphological analyses were performed using an energy-dispersive spectrometer (EDS; Jeol Japan) equipped with a detector model elemental analyzer (EX-541/5JMU), while the morphology was observed using a scanning electron microscope (SEM) (Jeol Japan detector, JSM-6380A). The samples were placed in an aluminum mount, with adhesive on both sides, and carbon tape was used for electrical conductivity. Palladium and gold (40 : 60%) were used as coating materials for 30 s and 48 mA current with 38 kV voltage and an average of three experiments was reported.20 2.2.6. Size and surface charge characterization The size of all microspheres was analyzed by the already reported method of Liu M. et al.,27 using Zetasizer Nano-z (Malvern Instrument Ltd., Worcestershire, UK) with disposable folded capillary cells. Briefly, a dispersion of 10 mg sample in 1 mL distilled water was prepared, vertexed at room temperature and mean diameter and size distribution of the prepared microspheres were measured by dynamic light scattering at 633 nm at 36 ± 0.5 °C. The effect of different pH conditions on zeta potential changes of prepared microspheres was measured by the method reported by Hosseinzadeh H. et al.51 Then, 10 mg mL−1 suspension of all formulations were prepared in 0.1 M (HCL) having pH 1.2 and 50 mM phosphate buffer having pH 4.5, 6.0, 6.8 and 7.4. Zeta potential changes were measured and each experiment was repeated thrice for average results.22 A light microscope was used to keep track of the microsphere suspension's condition (LM; model Axioscope A1, Carl Zeiss Microscopy GmbH, Jena, Germany). Important characteristics of interest included microsphere morphology (such as size) and a propensity to exist in a liquid media. Additionally, the light microscope was utilized to see how microspheres' size altered throughout wet milling.23 The morphological study using optical microscopy was done to determine how the microspheres' size and shape related to one another. The microscopy photos display spherical particles with what appears to be internal granulation, and they also demonstrated homogeneity in terms of morphology among many microsphere components. The optical microscopic image of a single agglomerated particle of powder material with an optical spectrum showed the distribution of elements over the particle area.24 2.2.7. Cell viability studies Cell viability studies of the prepared CTS/Fe3O4/HNTs/GC (MHCMG7) microspheres were performed using the resazurin assay technique by already reported of S. Anoopkumar-Dukie et al.25 Briefly, Caco-2 cells were seeded in a 24-well plate at a density of 25 000 cells/well cultured under the controlled conditions of 95% relative humidity and 5% CO2 at 37 °C for 14 days. Old MEM was replaced with fresh MEM every 48 h. MEM interference was minimized with Fetal Bovine Serum (FBS). Incubated cells were assayed by HEPES buffered saline (HBS) at pH 7.4 (25 mM). Freshly prepared 0.5 and 1% suspensions of MHCMG7 microspheres were replaced with already present white MEM. Microspheres containing wells were gestated, for 3 and 24 h under the controlled condition of relative humidity and CO2. Fresh MEM and 0.2% Triton X-100 were used as positive and negative controls respectively. Then, 250 μL 2.2 mM resazurin solution was added in pre-washed cells containing fresh HBS buffer at 37 °C for 3 h. Fluorescence and metabolism of resazurin in Caco-2 cells were observed at 540 and 590 nm and fluorescence was observed by M200 Tecan Infinite Grading, Austria. A similar procedure was repeated for 24 h incubation samples. The following equation was used for the calculation of toxicity:9 2.2.8. Processing variables and in vitro study Optimized formulations from MHM, MCM, and MHCM microspheres were selected for loading the drug (GC). Approximately 50 mg microspheres were immersed with 1 mg mL−1 aqueous solution of GC with continuous stirring at 90 rpm for 24 h.26 The concentration of GC from a microsphere-containing suspension was measured at different time intervals at 269 nm. The percentage released of GC was calculated in 100 mM phosphate buffers of pH 5.2, 6.8, and 7.4 at 37 °C using a USP II paddle apparatus (Erweka, Darmstadt Germany). The filtered supernatant of 1 mL was collected at predetermined time points, the same volume was replaced with a fresh medium and the concentration of GC was calculated at 269 nm. Drug release results were elaborated and understood by different kinetics models such as Korsmeyer–Peppas, Higuchi, first-order, and zero-order:10 F = K0 × t 11 ln(1 − F) = −K1t 12 F = Kh × t1/2 13 Mt/M∞ = K3tn where K0, K1, Kh, and K3 are zero-order, first-order, Higuchi dissolution and Korsmeyer–Peppas rate constants respectively. t is the time and n is the release exponent. The dissolution efficiency was measured by ratio of % age of the area under dissolution curve concerning time from 0 to 100% of area of the rectangle. The % GC loading, entrapment efficiency (EE), cumulative percentage release, and dissolution efficiency were calculated using eqn (9)–(12), respectively.14 15 16 17 (AUC)To and Q100T are area under the curve and rectangle with 100% dissolution values measured. 2.2.9. In vitro thrombin, prothrombin, and thrombus formation The hemostatic activity of optimized formulations, i.e., MHMG3, MCMG3, and MHCMG7MM, was studied by the method already reported by Chenglong et al.1 Briefly, five different groups of 10 mL blood samples were used in such a way that group 1 was considered false positive without treatment, and the remaining four were treated with MHMG3, MCMG3, and MHCMG7MM. Then, 3.2% m/v sodium citrate was added to the sample blood and centrifuged at 2000 rpm at 4 °C for 30 min and aPTT (activated partial thromboplastin time) reagent was mixed in a ratio of 1 : 1 and incubated at 37 °C for 15 min. The mixture was added in a test tube containing 25 mM CaCl2 (used for recalcification) and the calculated amount of prepared microspheres. The coagulation time in seconds was observed in aPTT using a stopwatch. The same procedure was repeated for the calculation of the prothrombin test (PTT). The thrombin effect of MHMG3, MCMG3, and MHCMG7 microspheres was measured with 3.2% citrated fresh rat blood by the method reported by Sun et al.,22 with slight modifications. Five groups for test samples were made. Each group has four glass tubes, and 0.5 mL, 3.2% sodium citrated fresh blood diluted with 10 mL deionized water was added in each glass tube to start thrombus formation. This resulted in the thrombus being soaked in a 37% formaldehyde solution for 15 min at 25 °C and dried at 50 °C for constant weight. 2.2.10. Stability study of MHCMG7 Stability studies of MHCMG7 were carried out in acidic and alkaline media under controlled conditions of room temperature and relative humidity conditions according to the ICH guidelines, and the estimated shelf life was determined using Rgui software. 2.2.11. Rat liver laceration and tail amputation test The hemostatic effect of MCMG3, MHMG3, and MHCMG7 was evaluated by in vivo rat liver laceration and rat tail amputation experiments according to animal guidelines of the Ethical Committee of Baha Uddin Zakariya University, Multan, and ARRIVE animal guidelines.28 For this purpose, adult rats (n = 6) with an average weight of 174 ± 5 g were anesthetized by keeping them in a chloroform jar for 2 to 3 min. The rat's chest was cut with a 1 cm × 0.5 cm deep incision in the right lobe of the liver, the sample of 0.2 g MHMG3 was instantly placed over the incised area, and then the incision was pressed with medical gauze until the bleeding stops. The blood over the sterile medical gauze was weighed using a precision balance (Mettler Toledo). Meanwhile, the same procedure was repeated for MHMG3 and MHCMG7. Rat without any sample was treated as blank.29 The rat tail was cut 1 cm from the tip. The first drop of blood should be discarded using sterile medical gauze. A pre-weighed sample of MCMG3 was administered at the cut site, and blood was poured over the gauze. The spilled blood was absorbed with medical gauze every 20 s. To assess the actual amount of blood loss, the amount of blood in grams was measured until the bleeding stopped and compared to the control. Meanwhile, the same procedure was repeated for MHMG3 and MHCMG7. Rat without treatment was treated as blank.30 2.2.12. Histopathological examination of rabbit's colon Albino rabbits (1 to 1.5 kg) were divided into 5 groups as control (group 1), MHMG3 (group 2), MCMG3 (group 3), MHCMG7 (group 4), and GC market available brand® (group 5) with an equal number (12 in each group). Then, 24 h fasted rabbits were used for cancer induction. Cancer was induced by ingesting 30 mg kg−1 dimethylhydrazine once a week for 4–6 weeks. Dimethylhydrazine cause hyperplasia of intestinal mucosa resulting in severe granulomas in the colon.31 Two rabbits of all groups were sacrificed, dissected, and removed from colon and formation of cancer cells was examined. The number of aberrant crypt foci (ACF) per cm2, granulomas (G), and crypt abscesses (CA) were measured after a different period and found a rapid decrease in ACF per cm2 before and after the administration of MHCMG7. Histopathological studies showed cancer development in the colon mucosa of control group as compared to the normal control group.32 2.2.13. Statistical analysis All the results are interpreted using Microsoft Excel and GraphPad. To select the optimized formulations of CTS, HNTs, and CTS/HNTs MM, and to study the effects of variables, ANOVA was used (p < 0.05). 3. Results and discussion In this work, the phenomenon of magnetism was investigated for colon-targeted delivery of GC. MM are supramolecular particles that can flow through capillaries without obstructing them when a magnetic field of between 0.5 and 0.8 tesla was applied. However, they are also sufficiently prone to being caught in microscopic arteries and drawn into adjacent tissues. MM of CTS and HNTs were prepared by a solvent emulsion evaporation technique using TSG as a cross-linker. In Fig. 1, a chemical scheme for materials is proposed. According to step 1, Fe3O4 was dissolved in 32% ammonia solution whose pH was adjusted to 9 with a 1 M NaOH solution. Meanwhile, MHM was prepared after adding HNTs to the prepared solution at 70 °C under stirring for 4 h. According to step 2, Fe3O4, paraffin, and span 80 were dispersed in a 2% acetic acid solution at 1500 rpm for 2 h. Meanwhile, MCM was prepared after adding CTS into prepared solution in the presence of toluene-saturated glutaraldehyde (TSG) at 60 °C, while MHCM was prepared by a similar method that is reported in step 2. The resulting MHCM was washed with ethanol and distilled water. Fig. 1 Chemical scheme for the synthesis of microspheres (MHM, MCM, and MHCM), (step 1) synthesis of HNT magnetic microspheres (MHM) in the presence of an ammonia solution, (step 2) synthesis of CTS magnetic microspheres (MCM) in the presence of toluene-saturated glutaraldehyde (TSG), and (step 3) synthesis of CTS and HNT magnetic microspheres (MHCM) with TSG. All types of MM showed excellent flow, which was confirmed by Hausner's ratio, Carr's index, and the angle of repose acceptable values. Hausner's ratio was 1.01 ± 0.03, 1.01 ± 0.02 and 1.01 ± 0.02, Carr's index was 10 ± 2, 12 ± 2 and 11 ± 2, and the angle of repose was 18 ± 0.5, 19 ± 0.5 and 17 ± 0.5 of the optimized formulation i.e., MHMG3, MCMG3, and MHCMG7, respectively. The percentage yield of unloaded MM was in between the range of 39.19 ± 2.18% to 43.84 ± 2.01% while the loading of GC was 89.14 ± 2.17 to 93.62 ± 2.17%, which was confirmed by an already constructed standard curve of increasing concentration of pure GC at 269 nm. MHM3 and MHMG3 showed the highest bulk density, lowest tapped density, decreased Carr's index, and increased angle of repose due to the cylindrical shapes of HNTs, which created wide spaces among the microspheres. MHMG3 was an optimized formulation among the HNT-containing MM due to its maximum percentage loading and dominant hollow structure. In the case of MCMG3, the presence of CTS made the significant loading of GC due to its spherical and pH-dependent nature. Finally, the combined HNT- and CTS-containing MHCMG7 have both properties of the hallowed nature and positive charge at a lower pH providing a percentage loading of 90.34 ± 2.76%.33 The comparative results of the tapped density of optimized MHCM7-unloaded and GC-loaded microspheres were 0.64 ± 0.02 g cm−3 and 0.92 ± 0.06 g cm−3, respectively. The increase in tapped density may be due to the higher molecular weight of GC, which increased the mass of resulting microspheres and decreased wide space among the particles. The increased value of Carr's index and the decreased value of the angle of repose of the optimized formulation having both HNTs and CTS may be due to the overlapping of CTS on the surface of HNTs, which made them smooth. Excellent flow properties of the optimized microspheres were further used for the oral drug delivery system, which can release GC in the colon and is more targeted for the treatment of colorectal cancer.33 3.1. Electromagnetism of microspheres Fig. 2(A)–(C) shows the VSM analyses of GC-loaded HNT-containing MHMG (1 to 5), CTS-containing MCMG (1 to 5), and both HNT- and CTS-containing MHCMG (1 to 9) microspheres. The hysteresis curves obtained for MHMG showed the function of the magnetic field applied at room temperature with superparamagnetic properties. Magnetization and values of HNT-containing microspheres ranged from 14.07 to 25.35 emu g−1, which may be due to concentration difference between HNTs and GC.34 A decrease in magnetization was observed by increasing the concentration of HNTs due to overlapping of HNTs on Fe3O4. Fig. 2(B) shows a slight decrease in magnetization of CTS-containing MCMG from 20.32 to 16.79 emu g−1, which was attributed to the formation of complete spherical shapes of microspheres. In the case of MHCMG, the increased values of magnetization may be due to the complete interaction of negatively charged HNTs and positively charged CTS, which ultimately frees the availability of Fe3O4 for magnetization. The loading of GC into MM has very little effect on magnetization. The magnetization curve showed a small hysteresis loop and a slight coercivity as a typical characteristic of magnetization and the remanence values approach zero. It was scrutinized that the increase in HNT concentrations in MHM microspheres had superparamagnetic properties. MHCMG (1 to 9) microsphere magnetization ranged from 22.54 to 25.35 emu g−1, respectively as shown in Fig. 2(A). Optimized formulation MHCMG7 microspheres showed maximum magnetization values of 25.35 emu g−1 and were excellent candidates for controlled release of GC-like drugs on targeted sites, as shown in Fig. 2(D).35 Fig. 2 VSM analyses of different MM (A) HNTs/Fe3O4 (MHMG1 to MHMG5), (B) CTS/Fe3O4 (MCMG1 to MCMG5) (C) CTS/Fe3O4/HNTs (MHCMG1 to MHCMG9), and (D) comparative VSM analysis of optimized formulation MCMG3, MHMG3, MHCMG7 and MHCMG7 GC (gemcitabine loaded) microspheres. 3.2. Size and surface charge studies Fig. 3(A) shows the effect of different concentrations of polymers and temperature conditions of HNT, CTS, and both CTS GC-containing MHMG3, MCMG3, and MHCMG7 respectively.36 The size of the HNT-containing microspheres was 135 ± 5.91 to 157 ± 5.86 μm, while in the case of CTS-containing microspheres, it also increased from 123 ± 6.56 μm to 140 ± 6.97 with a PDI value less than 0.5. The decrease in the size of MCMG3 as compared to MHMG3 may be due to the formation of complete complexes in the form of covalent bond of chitosan on the outer surface of Fe3O4. In case of combined HNTs and CTS, MHCMG7 showed further increase in the particle size from 50 to 70 μm with PDI values of 0.453, which may be due to cross-linking of HNTs with chitosan in the presence of TSG making them excellent for the targeted release of anticancer drugs.37,38 Zeta potential changes of all formulations are reported in Fig. 3(B), the positive charge in all formulations was observed in acidic environments, i.e., from pH 1.2 to 4.5, while the negative behavior in alkaline media is attributed to the enhanced permeation of GC from negative charge-carrying mucous membrane.39,40 The presence of CTS in MCMG3 formulations showed the highest changes in the acidic medium as compared to HNT-containing MHMG3 and MHCMG7, which may be due to the presence of an amino group with a positive charge. In the case of alkaline pH, MHCMG7 formulations had the highest values of zeta potential attributed to the formation of complete sphere complexes, which made them an excellent candidate for the release of anticancer drugs in an alkaline medium like colon.40Table 2 shows the optical microscopic image of the microspheres dispersed in distilled water.41 These results indicated the morphological and size changes before milling of microspheres, i.e., 190.5 μm of MHCMG7, while after milling, the size of microspheres reduced to about 84 μm.42 Fig. 3 (A) Zetasizer particle size of MHMG3, MCMG3, and MHCMG7 microspheres was calculated and reduction in particle size confirmed the formation of the complex. (B) Zeta potential and ion pairing changes were observed and a positive charge was found in an acidic medium, which converted into a negative charge in an alkaline buffer at pH 6, 6.8, and 7.4. (C) Hemostatic effect of MHMG3, MCMG3, and MHCMG7, which shows 2.5 folds increase in thrombin formation of MHCMG7. (D) Cell viability and metabolic activity were observed from 100% to 0% at pH 7.4 for 3 h and 24 h, PC; positive control and NC; negative control, respectively. Size determination of microspheres i.e., MHMG3, MCMG3, and MHCMG7 using optical microscopy, part a and b showed results before milling while c and d showed results with size reduction after milling Before milling (Y−, 190.5 μm) Sample Diameter X (μm) Y = X CF (X − Y−)2 MHMG3 29 130.5 2682.22 161.49 MCMG3 56 252.4 18090.25 134.55 MHCMG7 42 189.2 22052.25 148.56   After milling (Y−, 84 μm) MHMG3 22 99.3 3844 62.32 MCMG3 16 72.2 4624 68.35 MCMG7 18 81.4 4356 66.66 3.3. Structural and techno-functional properties Fig. 4(A) presents the X-ray diffraction patterns of GC-loaded MHMG, MCMG, and MHCMG microspheres, and distinct peaks were observed. The diffraction line of MHMG3 at 15.89° (2θ) and MCMG3 at 19.59° (2θ) showed the crystalline nature of the reported microspheres. The presence of HNTs decreased the 2θ values, while inverse was observed in case of CTS, i.e., 15.89 to 19.5. Furthermore, characteristic distinct peaks at 2θ values were 17.59°, 24.41°, 25.03°, 26.89°, 27.15° and 28.74° for MCMG (1 to 5), while they were slightly shifted by increasing Fe3O4. The diffractogram peak of MHCMG7 at 7.85°, and 18.69° (2θ) and GC-loaded MHCMG7 were observed at 14.47°, 19.58°, 20.95°, 24.31°, 26.03°, 36.15° and 38.56° (2θ), respectively. The diffraction peak of CTS disappeared in the XRD pattern of MHCM3 and MHCMG7 microspheres due to entrapment of Fe3O4 in CTS microspheres. The relative peak intensities and peak position indicated the successful introduction of CTS, HNTs, and GC into Fe3O4 microspheres.43 Fig. 4 Analysis of gemcitabine-loaded and -unloaded different MMs: (A) XRD analysis define the diffraction line ranging from 14.47° (2θ) to 26.31° (2θ). (B) TGA analysis showing the rapid weight loss during heating from 30 to 170 °C and a further decrease at 220 °C. Complete denaturation magnetic microspheres observed before 600 °C. (C) DTG at 225 °C and 480 °C, MCMG3 at 260 °C and 415 °C, MHMG3 61 °C, 280 °C and 530 °C and drug loading MHCMG7 at 66 °C and 510 °C. (D) FTIR analysis results. Fig. 4(B) shows the TGA and weight loss of MHMG3, MCMG3, and MHCMG7. Rapid weight loss during heating from 30 to 180 °C of all the formulations was observed due to presence of water molecules in the compositions. Stability in the weight loss was observed from 180 to 300 °C due to formation of stable structures among HNTs and Fe3O4 in the case of MHMG3, while CTS and Fe3O4 in MCMG3. The sudden decrease in weight loss was started after 220 °C in case of MHCMG7, and two peaks at 225 °C and 480 °C were observed showing the decomposition of covalent bonds between HNTs and CTS. Finally, at 550 °C, complete decomposition of MHCMG7 was observed attributed to the stability of GC-loaded microspheres, which made them more acceptable for drug-loaded purposes. Fig. 4(C) shows the DTG analysis results of all prepared microspheres, supported by TGA results.44 The initial water loss of all samples was observed in DTG peaks at a temperature of 180 °C. The initial large peak at 66 °C and 510 °C in MHCMG7 was due to denaturation. The complete decomposition of organic chitosan and structure loss was observed at 180–400 °C. However, significant reduction in weight loss of organic compounds may be due to the stable structure of HNTs. Other factors such as de-hydroxylation of Al(OH)3 in HNTs made the complex more stable at a temperature between 400 and 500 °C.43 Fig. 4(D) briefly describes the FTIR spectra in the 4000–400 cm−1 wavenumber range of MHMG3, MCMG3, and MHCMG7 microspheres. HNTs show peaks at 911 cm−1 and 3695 cm−1 due to the deformation and stretching of the O–H group respectively.45 Two peaks of HNTs were observed at 1121 cm−1 and 756 cm−1 due to Si–O group stretching. Chitosan spectra at peaks of 1039 cm−1, 3400 cm−1, 1680 cm−1, 1180 cm−1, and 820 cm−1 were due to the C <svg xmlns="http://www.w3.org/2000/svg" version="1.0" width="13.200000pt" height="16.000000pt" viewBox="0 0 13.200000 16.000000" preserveAspectRatio="xMidYMid meet"><metadata> Created by potrace 1.16, written by Peter Selinger 2001-2019 </metadata><g transform="translate(1.000000,15.000000) scale(0.017500,-0.017500)" fill="currentColor" stroke="none"><path d="M0 440 l0 -40 320 0 320 0 0 40 0 40 -320 0 -320 0 0 -40z M0 280 l0 -40 320 0 320 0 0 40 0 40 -320 0 -320 0 0 -40z"/></g></svg> C stretching vibration (CH3OH), NH stretching, CO, and C–O group stretching respectively. Slight shifts in NH bending from 1556 cm−1 to 1552 cm−1 were observed due to cross-linking with glutaraldehyde. A new absorption peak was observed at 1656 cm−1, indicating the formation of (CN) by CTS and glutaraldehyde. The absorption peak at 3716 cm−1 and 3786 cm−1 is related to Al(OH)2 stretching. Two peaks at 967 cm−1 and 1115 cm−1 were assigned to Si–O and O–H group deformation. The incorporation of HNTs was confirmed by an absorption band at 543 cm−1 due to the deformation of Al–O–Si, and bands at 3673 cm−1 and 3714 cm−1 were attributed to O–H stretching of HNTs. A slight shift of 549 cm−1, 3691 cm−1 and 3720 cm−1 was observed after HNT incorporation.46 All the above-mentioned characteristic spectra of Fe3O4, HNTs, CTS, and GC confirmed successful incorporation into MHMG3, MCMG3, and MHCMG7 microspheres. The absorption spectra of GC were observed at 2936 cm−1 due to CH2 stretching, and at 1681 cm−1, and 1061 cm−1 due to stretching of the C–O group. The absorption peak was observed at 1757 cm−1 due to the uredo group in GC. The formation of a broadband at 3050–3600 cm−1 was due to strong hydrogen bond between GC and MCMG3.46,47 Fig. 5(A)–(D) confirm that the spherical shape of MMs becomes irregular or wrinkled during observation due to the shrinkage of particles upon drying.43 The presence of the N-acetyl glucosamine group in CTS molecules made their shape irregular as the concentrations of CTS increased in CTS/Fe3O4, CTS (25%)/Fe3O4, and CTS (50%)/Fe3O4. The halloysite nanotubes were porous interior hybrid microspheres that overlapped loosely together. Many stripes were noted to form on the surface of the MHCMG7 microspheres.48 The uneven surface of MCMG3 microspheres was reported due to the absorption of MHMG3 on the surface. A smooth surface of simple and CTS-containing Fe3O4 microspheres turned slightly rough after the addition of an increasing concentration of HNTs from HNTs/Fe3O4, HNTs (25%)/Fe3O4, and HNTs (50%)/Fe3O4. Change in the surface and formation of many strips on the surface may be due to the tube-like structure of HNTs. The sphericity of the MM becomes worse with higher HNTs. Loose overlapped structure and large surface area were favorable for drug loading into MHCMG7.17 Fig. 5(E)–(H) shows the elemental analysis of MHMG3, MCMG3, MHCMG7 (without drug loading), and MHCMG7 (drug-loaded). The presence of iron, oxygen, carbon, potassium, chlorine, sodium, aluminum, and silicon was determined and compared.48,49 The increasing ratio of carbon and oxygen was in the order of MCMG3 > MHMG3 > MHCMG7 microspheres, while chlorine and potassium were only observed in MHMG3 microspheres, which may be due to HNTs. The Al and Fe ratio decreased with the addition of HNTs and CTS due to their complex formation with Fe3O4.43,45 Fig. 5 SEM analysis of MHMG3 (A), MCMG7 (B) MHCMG7 (C), and drug loading MHCMG7. (D) EDS analysis of MHM3 (E), MCMG7 (F) MHCMG7 (G), and drug loading MHCMG7 (H). 3.4. Processing variables and in vitro studies Fig. 6 shows the percentage release of GC from MHMG3, MCMG3, and MHCMG7 at acidic as well alkaline pH 1.2, 5.2, and 7.4, respectively. All formulations showed a maximum percentage release at 7.4 due to the presence of chitosan. There is no chemical interaction between GC and CTS. GC was physically loaded in MHMG3, MCMG3, and MHCMG7. The pH-dependent nature of CTS with GC was already reported in the literature. At a low pH, CHT-free NH2 groups were charged, causing the repulsion of polymer chains. This will result in the controlled release of GC.38 NH2 groups of CTS-containing formulations were protonated at a slightly acidic pH, while presence of HNTs provided the favorable controlled percentage release of GC at an alkaline pH due to their electrostatic repulsion and drug loading in the HNT lumen,50 as shown in Table 1. The first burst release, i.e., 45% of GC from MHCMG7 microspheres within 6 h at pH 7.4 confirmed the presence of GC at the surface of optimized microspheres. After 96 h, more than 80% of GC release at pH 7.4 confirmed the sustained release behavior of GC, which made the formulation ideal for the treatment of colorectal cancer.51 In Fig. 7, the stability of optimized MHCMG7 was calculated according to Rgui and the estimated shelf life was 30 months. Fig. 6 (A) Gemcitabine-loaded MMs, i.e., MCMG3, MHMG3, and MHCMG7 showing cumulative percentage release of 72.28%, 86.67%, and 95.78% after 96 h at pH 7.4 respectively. (B) Gemcitabine-loaded MMs of MCMG3, MHMG3, and MHCMG7 showing cumulative percentage release of 65.53%, 80.84%, and 90.98% after 96 h at pH 5.2 respectively. (C) Gemcitabine-loaded microspheres of MCMG3, MHMG3, and MHCMG7 showing cumulative percentage release of 53.26%, 59.77%, and 66.59% after 96 h at pH 1.2 respectively. (D) Cumulative percentage release of gemcitabine-loaded MHCMG7 microspheres at different pH values. Fig. 7 Stability study and estimated shelf life of MHCMG7. 3.5. Plasma coagulation Table 3 shows the comparison of all formulations of GC-loaded MM. The maximum hemostatic effect of MHCMG7 as compared to MHMG3 and MCMG3 was observed due to the formation of CTS and HNT complexes, as shown in Fig. 3(C). The results indicated that the MM reduces the time for aggregation of not only erythrocytes but also platelets. The clotting time of control (without treatment) was 232 ± 6.08 s, which was reduced to 220 ± 5.93 s in the case of MHMG3. MCMG3 clotted the blood in 140 ± 6.08 s, which was further reduced to 89 ± 7.13 s by using MHCMG7. Similar behavior of whole blood clotting was observed by Sun et al.22 The thrombogenic activity of all microspheres was significantly shorter than the control sample, which may be due to the formation of intrinsic blood coagulation (fibrin formation), and their interaction with negatively charged HNTs. The weight of the whole blood clot was 82 ± 5.08, 100 ± 4.99, 194 ± 5.99, and 216 ± 5.04 mg for control, MHMG3, MCMG3, and MHCMG7, respectively. An increase in the clotted blood weight and roughness of the beads' surface (clotted blood) was observed by using the optimized MHCMG7 due to rapid interaction of the positively charged amino group with negatively charged RBCs. Eventually, rapid clotting process of optimized MHCMG7 made the formulation excellent for the control of hemorrhage process in the case of colorectal cancer patients. GC loaded MHMG3, MCMG3, and MHCMG7 release profile and entrapment efficiency with application of various kinetic release models Formulations Zero-order kinetics First order kinetics Higuchi's model Korsmeyer–Peppas model K 0 (h−1) R 2 K 1 (h−1) R 2 K H (h1) R 2 n R 2 MHMG3 7.259 0.8371 0.145 0.9978 22.539 0.9740 0.492 0.9698 MCMG3 9.85 0.4298 0.264 0.9508 28.056 0.9665 0.548 0.9442 MHCMG7 6.062 0.2498 0.236 0.8884 23.799 0.9729 0.310 0.9846 3.6. Rat liver laceration and tail amputation test Under ARRIVE animal guidelines,52 which were approved by the Animal Ethics Committee of Baha Uddin Zakariya University, hemostatic assays were carried out on the rat tail and rat liver.28Fig. 8 and Table 4 showed that the bleeding volume and bleeding time were significant parameters. Due to more effective blood coagulation caused by the positive surface charge of CTS, bleeding duration in case of MCMG3 was dramatically reduced by 1.5 folds as compared to the control. Due to epithelial regeneration and higher collagen deposition at the damaged region, MHMG3 demonstrated only a 0.8-fold increase in efficiency compared to the control.27 Interestingly, MHCMG7 demonstrated three times greater hemostatic effectiveness because of the interaction between HNTs and CHT MM. The bleeding volume and bleeding duration were found to be 0.03 g and 15 s, respectively, as opposed to that of the control, 3.33 ± 0.03 g and 290 s. Fig. 8 (A) Rat liver laceration test showing improved clotting and hemostatic ability of MHCMG7 than MHMG3 and MCMG3, while in part (B) rat tail amputation test with bleeding time is shown. Effect of gemcitabine loaded different magnetic microspheres on hemostatic time and weight of whole blood clot in rat liver laceration and rat tail amputation test Gemcitabine loaded formulation Haemostasis time (s) Whole blood clot weight (mg) Rat liver laceration (weight of blood g) Rat tail amputation (weight of blood g) Control 232 ± 6.08 82.4 ± 5.08 3.33 ± 0.03 1.33 ± 0.03 MHMG3 147 ± 5.93 144.7 ± 4.99 1.35 ± 0.03 1.02 ± 0.03 MCMG3 216 ± 6.08 216.2 ± 5.04 1.26 ± 0.03 1.00 ± 0.03 MHCMG7 89 ± 7.13 194.6 ± 5.99 0.03 ± 0.001 0.50 ± 0.003 Rats (n = 6) were used to investigate the hemostatic effectiveness of MCMG3, MHMG3, and MHCMG7 using a tail amputation procedure. In Fig. 8(B), 250 s, 200 s, and 30 s hemostatic time were noted for MCMG3, MHMG3, and MHCMG7 respectively, which clearly showed 6.6 times increased hemostatic ability of MHCMG7. From Table 4, it is clear that the blood weight in a gram of the control, MCMG3, MHMG3, and MHCMG7 are 1.330, 1.020, 1.000, and 0.500, respectively. Similar findings were also mentioned in our previous work by Hanif et al.14 3.7. Histopathological studies Different stages of colorectal cancer development and therapeutic effects are shown in Fig. 9(A)–(H). The formulation of MHCMG7MM showed the best results and rapid development of normal colon cells as compared to the gemcitabine reference formulation. A moderate-to-mild intracellular positive response was observed in glass slides from the MM. The positive reaction was responded to the dose of gemcitabine loaded in MHCMG7. Different stages of development and treatment of colorectal cancer are shown in Fig. 9(A)–(H). The number of aberrant crypt foci (ACF) was measured after a different period and an increased ACF number was found in Fig. 9(A)–(D) due to dimethylhydrazine. Well-differentiated glandular adenocarcinoma tissue is shown in the image in Fig. 9(A) and (B), while poorly differentiated glandular adenocarcinoma homogeneous tissue is observed in the image in Fig. 9(C) and (D). Fig. 9(E)–(H) shows the schematic regeneration of normal body cells after therapy with GC-loaded MHCMG7MM for up to 12 weeks. Dense stroma image patches cluster to loose stroma clusters were observed during the development of normal body cells.53 Healing of mucosa, regeneration of epithelial cells, and recovery from colorectal cancer were successfully investigated by histopathological examination. Fig. 9 Different stages of development and treatment of colorectal cancer from image (A) to (H). The number of aberrant crypt foci (ACF) per cm2, granulomas (G), and crypt abscesses (CA) were measured after a different period and an increase in ACF number is found in (A) to (D) due to dimethylhydrazine. Well-differentiated granuloma (G) tissue is shown in (A) and (B) while poorly differentiated glandular adenocarcinoma homogeneous tissue is observed in (G) and (H). (E) to (H) show the schematic of the regeneration of normal body cells after therapy with GC-loaded MHCMG7 magnetic microspheres for up to 12 weeks. (LP: lamina propria, MM: muscularis mucosa, MS: muscularis, CC: columnar cells, COL: crypts of Lieberkühn, GC: goblet cells). 3.8. Cell viability study and safety profile The resazurin assay was used to determine the cytotoxic activity of biological materials in formulation MHCMG7, as shown in Fig. 3(D).25 For positive control MEM, MHCMG7 (0.5%), MHCMG7 (1.0%), and negative control (Triton X-100), respectively, the percentage viability was 99, 96, 94, and 0%. The findings demonstrated that MCMG3 produced superior outcomes and served as an essential medium for the growth of Caco-2 cells, while MHMG3 also improved epithelial cell regeneration and produced results that were within acceptable limits. However, the two together improved cell development because CTS provided more nutrients. In contrast to MCMG3 and MHMG7, MHCMG7 demonstrated greater biocompatibility, biodegradability, and enhanced cell regeneration. As a negative control, Triton X-100 eliminated all the Caco-2 cells. Fig. 3(D) used absorbance measurements collected at 3 and 24 h intervals to calculate the percentage viability of MHCMG7 (0.5%) and MHCMG7 (1.0%).54 The biocompatibility and biodegradability of natural polymer chitosan and clay halloysite nanotubes were considered safe for use. Target drug delivery of MM also minimized the toxic effect of carcinogenic drugs. The chitosan-coated halloysite MM is widely used in biomedical applications due to its therapeutic effect on carcinogenic colorectal hemorrhage.54 4. Conclusion Treatment of carcinogenic hemorrhage can be possible using the developed controlled-release (CR) strategy of GC. Synthesized CR and targeted release of GC-containing magnetic microspheres (MM) of halloysite nanotubes (MHMG), chitosan (MCMG), and their combination (MHCMG) by co-emulsion techniques proved a good alternative to intravascular use of GC. The spherical shape of CTS microspheres and the irregular shape of HNT microspheres proved to be better solution for oral drug delivery of GC. The combination of both HNT and CTS microspheres made them more targeted in case of optimized MHCMG7 formulation. Decreased hemostatic time and whole blood clotting time with the regeneration of epithelial cells are more applicable in general and in cancerous patients. Magnetic microspheres can be considered a carrier for the release of anticancer drugs. In the future perspective, MHCMG7 magnetic microspheres loaded with GC may be applied for various cancers such as breast cancer and bladder cancer along with radiotherapy respectively. Abbreviations GC Gemcitabine CTS Chitosan FeCl3·6H20 Iron(iii) hydrochloride hexahydrate FeCl2·4H2O Iron(ii) hydrochloride tetrahydrate MM Magnetic microsphere HNTs Halloysite nanotubes MHMG Magnetic microspheres of HNTs MCMG Magnetic microspheres of CTS MHCMG Magnetic microspheres of HNTs and CTS VSM Vibrating sample magnetometer XRD X-Ray diffraction EDS Energy dispersive spectroscopy SEM Scanning electron microscopy FTIR Fourier transform infrared spectroscopy TGA Thermogravimetric analysis CRC Colorectal cancer BMI Body mass index GIT Gastrointestinal tract CTS Chitosan HEPES 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid PPD para-phenylenediamine KH2PO4 Sodium tetraborate, potassium dihydrogen phosphate GST Glutaraldehyde saturated toluene aPTT Activated partial thromboplastin time PTT Prothrombin test ACF Aberrant crypt foci ACF Aberrant crypt foci Ethical statement All animal procedures performed in accordance with guidelines for care and use of laboratory animals of Bahauddin Zakariya University and experiments were approved by the animal Ethical Committee of Faculty of Pharmacy, Bahauddin Zakariya University, Multan. Author contributions Muhammad Hanif was the supervisor of all experiments. Khalid Mahmood was the supplier of chemicals and reagents. The writer, reviewer, and critical analyzer were Muhammad Qaiser and Dure Shahwar. Sajid Majeed and Muhammad Qaiser was the analyst that perform the analysis of formulation and in vivo activities. Muhammad Harris Shoaib was the scientist who develop the chemical reaction. Ghulam Abbas and Nabeela Ameer help with manuscript writing. Conflicts of interest The authors report no conflicts of interest. The authors alone are responsible for the content and writing of this article. Supplementary Material The authors acknowledge the financial support received from Bahauddin Zakariya University, Multan Pakistan, and Drugs Testing Laboratory Punjab Pakistan for their support and encouragement in carrying out this work. ==== Refs References Dai C. Liu C. Wei J. Hong H. Zhao Q. Biomaterials 2010 31 7620 7630 20656346 De Silva R. Pasbakhsh P. Goh K. Chai S.-P. Ismail H. Polym. Test. 2013 32 265 271 Kakar S. Batra D. Singh R. Nautiyal U. J. Acute Dis. 2013 2 1 12 Kiralp S. Topcu A. Bayramoğlu G. Arıca M. Y. Toppare L. Sens. Actuators, B 2008 128 521 528 Pich A. Bhattacharya S. Adler H. J. P. Wage T. Taubenberger A. Li Z. van Pee K. H. Böhmer U. Bley T. Macromol. Biosci. 2006 6 301 310 16572475 Cavallaro G. Chiappisi L. Pasbakhsh P. Gradzielski M. Lazzara G. Appl. Clay Sci. 2018 160 71 80 Lisuzzo L. Cavallaro G. Milioto S. Lazzara G. Polym 2020 12 1766 Lisuzzo L. Cavallaro G. Milioto S. Lazzara G. J. Colloid Interface Sci. 2022 608 424 434 34626986 Lisuzzo L. Cavallaro G. Milioto S. Lazzara G. Appl. Clay Sci. 2021 213 106231 Yao J. Wang Q. Wang Y. Zhang Y. Zhang B. Zhang H. Desalin. Water Treat. 2015 55 1293 1301 Li Q. He T. Zhang Y.-Q. Wu H. Liu J. Qi Y. Lei Y. Chen H. Sun Z. Peng C. ACS Sustainable Chem. Eng. 2019 7 17039 17046 Pandi K. Viswanathan N. Meenakshi S. Int. J. Biol. Macromol. 2019 132 600 605 30902717 Eskandarloo H. Arshadi M. Abbaspourrad A. ACS Sustainable Chem. Eng. 2018 6 14561 14573 Arshad M. S. Qaiser M. Mahmood K. Shoaib M. H. Ameer N. Ramzan N. Hanif M. Abbas G. Int. J. Biol. Macromol. 2022 314 323 Ward C. J. DeWitt M. Davis E. W. ACS Appl. Nano Mater. 2012 209 238 Englehart M. S. Cho S. D. Tieu B. H. Morris M. S. Underwood S. J. Karahan A. Muller P. J. Differding J. A. Farrell D. H. Schreiber M. A. J. Trauma Acute Care Surg. 2008 65 884 892 Ma W. Dai J. Dai X. Da Z. Yan Y. Desalin. Water Treat. 2016 57 4162 4173 Wang Q. Zhang J. Mu B. Fan L. Wang A. Carbohydr. Polym. 2014 102 877 883 24507359 Türkeş E. Sağ Açıkel Y. Int. J. Environ. Sci. Technol. 2019 17 1281 1294 Zhang W. Hou C. Li Y. Zhang Q. Wang H. J. Colloid Interface Sci. 2020 558 115 122 31585220 Sohrabijam Z. Zamanian A. Saidifar M. Nouri A. J. P. M. S. Colloids Surf., B 2015 11 282 286 Sun L. Chen Y. Zhou Y. Guo D. Fan Y. Guo F. Zheng Y. Chen W. Asian J. Pharm. 2017 12 418 423 Lee J. Sah H. Pharmaceutics 2022 14 1540 35893796 Shekhovtsov V. Glass Ceram. 2022 79 282 286 Anoopkumar-Dukie S. Carey J. Conere T. O'sullivan E. Van Pelt F. Allshire A. Br. J. Radiol. 2005 78 945 947 16177019 Kaur T. Kaur S. Kaur P. Int. J. Appl. Pharm. 2017 9 60 65 Liu M. Shen Y. Ao P. Dai L. Liu Z. Zhou C. RSC Adv. 2014 4 23540 23553 Du Sert N. P. Ahluwalia A. Alam S. Avey M. T. Baker M. Browne W. J. Clark A. Cuthill I. C. Dirnagl U. Emerson M. PLoS Biol. 2020 18 e3000411 32663221 Yan R.-R. Xue D. Su C. Xu Y. Gong J.-S. Liu Y.-L. Jiang M. Geng Y. Lv G.-Z. Xu Z.-H. Colloids Surf., B 2022 218 112770 Zheng C. Liu J. Bai Q. Quan Y. Li Z. Chen W. Gao Q. Zhang Y. Lu T. Mater. Des. 2022 111116 Rogers A. E. Herndon B. J. Newberne P. M. Cancer Res. 1973 33 1003 1009 4703115 Yoshida H. Shimazu T. Kiyuna T. Marugame A. Yamashita Y. Cosatto E. Taniguchi H. Sekine S. Ochiai A. J. Gastrointest. Cancer 2018 21 249 257 Sharif S. Abbas G. Hanif M. Bernkop-Schnürch A. Jalil A. Yaqoob M. Colloids Surf., B 2019 184 110527 Türkeş E. Açıkel Y. S. Int. J. Environ. Sci. Technol. 2020 17 1281 1294 Luo S. J. Zhang P. Mei Y. A. Chang J. B. Yan H. J. Appl. Polym. Sci. 2016 133 1 5 Podzus P. Daraio M. Jacobo S. Phys. B 2009 404 2710 2712 Li G.-y. Jiang Y.-r. Huang K.-l. Ding P. Chen J. J. Alloys Compd. 2008 466 451 456 Parsian M. Unsoy G. Mutlu P. Yalcin S. Tezcaner A. Gunduz U. Eur. J. Pharmacol. 2016 784 121 128 27181067 Molaei A. Amadeh A. Yari M. Afshar M. R. Mater. Sci. Eng., C 2016 59 740 747 Dhawan S. Singla A. K. Sinha V. R. AAPS PharmSciTech 2004 5 122 128 Rodrigues E. M. Calvert N. D. Crawford J. C. Liu N. Shuhendler A. J. Hemmer E. Small 2022 2107130 Mandal D. Sathiyamoorthy D. Rao V. G. Fusion Eng. Des. 2012 87 7 12 Xiong X. Wang Y. Zou W. Duan J. Chen Y. J. Chem. 2012 2013 1 8 Chyiy L. Saravanan M. Int. J. Pharm. Pharm. Sci. 2015 7 203 206 Xie Y. Qian D. Wu D. Ma X. J. Chem. Eng. 2011 168 959 963 Luo P. Zhao Y. Zhang B. Liu J. Yang Y. Liu J. Water Res. 2010 44 1489 1497 20003999 Chang P. R. Yu J. Ma X. Anderson D. P. Carbohydr. Polym. 2011 83 640 644 Devi N. Dutta J. Int. J. Biol. Macromol. 2019 127 222 231 30641190 Kim M. Jee S. C. Sung J.-S. Kadam A. A. Int. J. Biol. Macromol. 2018 118 228 237 29913193 Kamble R. Ghag M. Gaikawad S. Panda B. K. J. Adv. Sci. Res. 2012 3 25 29 Hosseinzadeh H. Atyabi F. Dinarvand R. Ostad S. N. Int. J. Nanomed. 2012 7 1851 Kilkenny C. Browne W. J. Cuthill I. C. Emerson M. Altman D. G. J. Pharmacol. Pharmacother. 2010 1 94 99 21350617 Reddy B. S. Hirose Y. Lubet R. Steele V. Kelloff G. Paulson S. Seibert K. Rao C. V. Cancer Res. 2000 60 293 297 10667579 Venkatesan P. Puvvada N. Dash R. Kumar B. P. Sarkar D. Azab B. Pathak A. Kundu S. C. Fisher P. B. Mandal M. Biomaterials 2011 32 3794 3806 21392822
PMC010xxxxxx/PMC10352713.txt
==== Front Respirol Case Rep Respirol Case Rep 10.1002/(ISSN)2051-3380 RCR2 Respirology Case Reports 2051-3380 John Wiley & Sons, Ltd Chichester, UK 10.1002/rcr2.1192 RCR21192 Case Report Case Reports Eosinophilic pneumonia developed after dupilumab administration in a patient with atopic dermatitis EOSINOPHILIC PNEUMONIA DUE TO DUPILUMAB Kanata et al. Kanata Kei https://orcid.org/0009-0007-8019-7243 1 [email protected] Shirai Toshihiro https://orcid.org/0000-0002-2352-7580 2 Ichijo Koshiro 1 Uehara Masahiro 1 1 Department of Respiratory Medicine Shimada General Medical Center Shizuoka Japan 2 Department of Respiratory Medicine Shizuoka General Hospital Shizuoka Japan * Correspondence Kei Kanata, Department of Respiratory Medicine, Shimada General Medical Center, 1200‐5 Noda, Shimada, Shizuoka Prefecture 427‐8502, Japan. Email: [email protected] 18 7 2023 8 2023 11 8 10.1002/rcr2.v11.8 e0119211 4 2023 29 6 2023 © 2023 The Authors. Respirology Case Reports published by John Wiley & Sons Australia, Ltd on behalf of The Asian Pacific Society of Respirology. https://creativecommons.org/licenses/by-nc/4.0/ This is an open access article under the terms of the http://creativecommons.org/licenses/by-nc/4.0/ License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes. Abstract A 63‐year‐old woman with refractory atopic dermatitis started treatment with dupilumab. She developed a cough 4 days later, sputum, and a slight fever 2 weeks later. Laboratory test results showed a blood eosinophil count of 7360/μL. Chest x‐ray and computed tomography scan showed infiltrative shadows with surrounding consolidation of both upper lobes. Bronchoalveolar lavage fluid eosinophil count was increased (50.0%), and histopathological findings were consistent with numerous eosinophilic infiltrations. Treatment with prednisolone 30 mg/day (0.5 mg/kg/day) was initiated. Her symptom resolved, and the shadow of the lung fields improved. There have been no reported cases of eosinophilic pneumonia diagnosed 7 weeks after the administration of dupilumab for atopic dermatitis. A 63‐year‐old woman with refractory atopic dermatitis developed a cough 4 days later, sputum, and a slight fever 2 weeks later. Chest x‐ray and computed tomography scan showed infiltrative shadows with surrounding consolidation of both upper lobes. Bronchoalveolar lavage fluid eosinophil count was increased (50.0%), and histopathological findings were consistent with numerous eosinophilic infiltrations. atopic dermatitis dupilumab eosinophilic pneumonia IL‐4/13 source-schema-version-number2.0 cover-dateAugust 2023 details-of-publishers-convertorConverter:WILEY_ML3GV2_TO_JATSPMC version:6.3.2 mode:remove_FC converted:18.07.2023 Kanata K , Shirai T , Ichijo K , Uehara M . Eosinophilic pneumonia developed after dupilumab administration in a patient with atopic dermatitis. Respirology Case Reports. 2023;11 :e01192. 10.1002/rcr2.1192 Associate Editor: Kazuyuki Nakagome ==== Body pmcINTRODUCTION Dupilumab is a monoclonal antibody that specifically binds to the interleukin (IL)‐4/13 receptor. 1 In the past literature, there have been a few reports of eosinophilic pneumonia as a side effect of dupilumab. We experienced a case of eosinophilic pneumonia in a patient with atopic dermatitis, which developed after dupilumab administration. CASE REPORT A 63‐year‐old woman diagnosed with atopic dermatitis had regularly attended the dermatology department of our hospital. She had been receiving budesonide for comorbid cough variant asthma for 10 years with controlled cough symptoms. However, she had no history of paediatric or typical asthma, chronic sinusitis, other allergic diseases, or prolonged cough after upper respiratory infections. Despite receiving antihistamines and steroid ointment administration, her skin condition did not improve, thus dupilumab was added for refractory atopic dermatitis in November of a certain year. She developed a cough 4 days later, sputum, and a slight fever 2 weeks later. A baseline blood eosinophil count of 610/μL increased to 1770/μL after 2 weeks and then continued to increase (Figure 1). Based on the diagnosis of worsened cough variant asthma, a nearby doctor changed budesonide to budesonide/formoterol. However, the symptoms did not improve, and the patient refused to receive dupilumab after the second dosing. In January of the following year, she visited the respiratory medicine department for an investigation. A physical examination revealed a body temperature of 37.7°C, and coarse crackles were heard in both upper lungs. There was no high‐grade fever, numbness, or weakness suggestive of vasculitis. Laboratory test results on admission showed a blood eosinophil count of 7360/μL, C‐reactive protein (CRP) of 3.56 mg/dL, total serum IgE of 2.4 IU/mL, and myeloperoxidase anti‐neutrophil cytoplasmic antibody (MPO‐ANCA) of negative (Figure 1). Specific IgE antibodies to 16 common inhalant allergens were negative. FIGURE 1 Changes in type‐2/eosinophilic biomarkers before and after dupilumab treatment. CRP, C‐reactive protein; TARC, thymus and activation‐regulated chemokine. Despite normal radiological findings prior to the dupilumab dosing, chest x‐ray and CT scan showed infiltrative shadows with surrounding consolidation of both upper lobes (Figure 2), and antibiotics were ineffective. On the 4th day after hospitalization, a bronchoscopy was performed, followed by bronchoalveolar lavage (BAL) and transbronchial lung biopsy. The BAL eosinophil count was remarkably increased (50.0%). Histopathological findings were consistent with numerous eosinophilic infiltration (Figure 1). Treatment with prednisolone 30 mg/day (0.5 mg/kg/day) was initiated. Changes in type‐2/eosinophilic biomarkers are shown in Figure 1. After the start of prednisolone treatment, her symptoms resolved, the shadow of the lung fields improved, and her blood eosinophil count decreased to 490/μL 3 days after the prednisolone administration. Prednisolone dose was tapered over approximately 1–2 weeks and finished 124 days after starting treatment. No recurrence was observed after that. FIGURE 2 Chest x‐ray and computed tomography (CT) scan showing infiltrative shadows with surrounding consolidation of both upper lobes (upper panel) and histopathological examination findings of numerous eosinophilic infiltration (lower panel). (HE staining × 200). DISCUSSION Dupilumab is a monoclonal antibody against the human IL‐4/13 receptor. It specifically binds to the IL‐4 receptor alpha subunit, which is common to both the IL‐4 and IL‐13 receptor complex, and inhibits IL‐4 and IL‐13 signalling, thereby broadly suppressing type 2 inflammatory responses. It is effective against treatment‐resistant bronchial asthma, atopic dermatitis, and chronic rhinosinusitis in patients with nasal polyps. 1 , 2 In a previous phase 3 clinical trial, approximately 0.16% (2/1263) of adult patients with moderate‐to‐severe uncontrolled asthma developed severe eosinophilic pneumonia and discontinuation of dupilumab. 1 Nishiyama et al. measured the serum cytokine levels in two cases of dupilumab‐associated eosinophilic pneumonia and found elevated IL‐5, but not in patients who did not develop dupilumab‐associated eosinophilic pneumonia. 3 Kurihara et al. reported two similar cases with high‐grade eosinophilia and emphasized fever and dyspnea as initial symptoms of eosinophilic pneumonia. 4 Eger et al. described four oral corticosteroid (OCS)‐dependent asthma patients in whom IL‐5 pathway biologics were switched to dupilumab leading to the development of eosinophilic pneumonia in one patient. 5 These findings suggest that not only mild eosinophilia caused by IL‐4/13 blocking but also severe eosinophilia caused by IL‐5 might be associated with the development of eosinophilic pneumonia. Inappropriate distribution of eosinophils due to blocked IL‐4 and IL‐13 seems to be a likely mechanism in this case. However, the origin of IL‐5 or the relationship between blood eosinophilia and pulmonary eosinophilic infiltration is still unknown. Further studies are needed to clarify these. Eger et al. hypothesized that the OCS‐dependent severe asthma patient who developed eosinophilic pneumonia with hypereosinophilia after switching to dupilumab might originally have had a latent ANCA‐negative eosinophilic granulomatosis with polyangiitis (EGPA). 5 This patient did not have severe asthma nor require OCS before dupilumab treatment. In addition, this patient had neither peripheral neuropathy, sinusitis, renal lesion, gastrointestinal lesions, nor other skin lesions than atopic dermatitis. Kurihara's two cases also had a similar degree of fever to this patient. However, we cannot exclude ANCA‐negative EGPA or its pre‐stage, entirely. Drug‐induced eosinophilic pneumonia, other than dupilumab's pharmacological effect, was also possible. However, we did not examine the immunological test such as DLST. Taken together, we think it is difficult to give a definitive interpretation of the mechanisms of eosinophilic pneumonia of this case. In the past literature, there have been a few reports of eosinophilic pneumonia after the administration of dupilumab. However, there have been no reported cases of eosinophilic pneumonia diagnosed 7 weeks after the administration of dupilumab for atopic dermatitis. The absence of respiratory symptoms before dupilumab treatment may have contributed to the early discontinuation of the treatment. However, it still took more than a month to be diagnosed since symptoms developed, suggesting a delay in diagnosis. Therefore, eosinophilic pneumonia should be considered as a differential diagnosis in the presentation of fever or dyspnea after dupilumab administration for atopic dermatitis. CONFLICT OF INTEREST STATEMENT None declared. ETHICS STATEMENT The authors declare that appropriate written informed consent was obtained for the publication of this manuscript and accompanying images. DATA AVAILABILITY STATEMENT The data that support the findings of this study are available from the corresponding author upon reasonable request. ==== Refs REFERENCES 1 Castro M , Corren J , Pavord ID , Maspero J , Wenzel S , Rabe KF , et al. Dupilumab efficacy and safety in moderate‐to‐severe uncontrolled asthma. N Engl J Med. 2018;378 :2486–2496. 10.1056/NEJMoa1804092 29782217 2 Bachert C , Han JK , Desrosiers M , Hellings PW , Amin N , Lee SE , et al. Efficacy and safety of dupilumab in patients with severe chronic rhinosinusitis with nasal polyps (LIBERTY NP SINUS‐24 and LIBERTY NP SINUS‐52): results from two multicentre, randomized, double‐blind, placebo‐controlled, parallel‐group phase 3 trials. Lancet. 2019;394 :1638–1650. 10.1016/S0140-6736(19)31881-1 31543428 3 Nishiyama Y , Koya T , Nagano K , Abe S , Kimura Y , Shima K , et al. Two cases of dupilumab‐associated eosinophilic pneumonia in asthma with eosinophilic chronic rhinosinusitis: IL‐5‐driven pathology? Allergol Int. 2022;71 :548–551. 10.1016/j.alit.2022.03.005 35443910 4 Kurihara M , Masaki K , Matsuyama E , Fujioka M , Hayashi R , Tomiyasu S , et al. How can dupilumab cause eosinophilic pneumonia? Biomolecules. 2022;12 :1743. 10.3390/biom12121743 36551171 5 Eger K , Pet L , Weersink EJM , Bel EH . Complications of switching from anti‐IL‐5 or anti‐IL‐5R to sarilumab in corticosteroid‐dependent severe asthma. J Allergy Clin Immunol Pract. 2021;9 :2913–2915. 10.1016/j.jaip.2021.02.042 33676050
PMC010xxxxxx/PMC10352714.txt
==== Front Virus Res Virus Res Virus Research 0168-1702 1872-7492 Elsevier S0168-1702(23)00104-1 10.1016/j.virusres.2023.199142 199142 Article Molecular evolutionary analyses of the fusion protein gene in human respirovirus 1 Takahashi Tomoko a Akagawa Mao b Kimura Ryusuke cd Sada Mitsuru bc Shirai Tatsuya c Okayama Kaori b Hayashi Yuriko b Kondo Mayumi e Takeda Makoto f Ryo Akihide g Kimura Hirokazu [email protected] bc⁎ a Iwate Prefectural Research Institute for Environmental Science and Public Health, Morioka-shi, Iwate 020-0857, Japan b Department of Health Science, Gunma Paz University Graduate School of Health Sciences, Takasaki-shi, Gunma 370-0006, Japan c Advanced Medical Science Research Center, Gunma Paz University Research Institute, Shibukawa-shi, Gunma 377-0008, Japan d Department of Bacteriology, Gunma University Graduate School of Medicine, Maebashi-shi, Gunma 371-8514, Japan e Department of Clinical Engineering, Faculty of Medical Technology, Gunma Paz University, Takasaki-shi, Gunma 370-0006, Japan f Department of Microbiology, Graduate School of Medicine and Faculty of Medicine, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan g Department of Microbiology, Yokohama City University School of Medicine, Yokohama-shi, Kanagawa 236-0004, Japan ⁎ Corresponding author at: Department of Health Science, Gunma Paz University Graduate School of Health Sciences, Takasaki-shi. [email protected] 09 6 2023 8 2023 09 6 2023 333 19914222 1 2023 26 4 2023 31 5 2023 © 2023 The Author(s) 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). HighLights • The phylogenetic analyses of the full-length F genes in HRV1 strains collected from various countries were performed. • HRV1 strains showed three major lineages in time-scaled phylogenetic tree of F genes. • No positive selection sites were found in F protein, whereas numerous negative selection sites were identified. • Incompatibilities between predicted epitopes in this study and NT-Ab binding might contribute to HRV1 reinfection. Few evolutionary studies of the human respiratory virus (HRV) have been conducted, but most of them have focused on HRV3. In this study, the full-length fusion (F) genes in HRV1 strains collected from various countries were subjected to time-scaled phylogenetic, genome population size, and selective pressure analyses. Antigenicity analysis was performed on the F protein. The time-scaled phylogenetic tree using the Bayesian Markov Chain Monte Carlo method estimated that the common ancestor of the HRV1 F gene diverged in 1957 and eventually formed three lineages. Phylodynamic analyses showed that the genome population size of the F gene has doubled over approximately 80 years. Phylogenetic distances between the strains were short (< 0.02). No positive selection sites were detected for the F protein, whereas many negative selection sites were identified. Almost all conformational epitopes of the F protein, except one in each monomer, did not correspond to the neutralising antibody (NT-Ab) binding sites. These results suggest that the HRV1 F gene has constantly evolved over many years, infecting humans, while the gene may be relatively conserved. Mismatches between computationally predicted epitopes and NT-Ab binding sites may be partially responsible for HRV1 reinfection and other viruses such as HRV3 and respiratory syncytial virus. Keywords Human respirovirus 1 Molecular evolutionary analyses Fusion protein gene B cell conformational epitope Abbreviations HRV Human respirovirus F protein fusion protein HN haemagglutinin-neuraminidase NT-Ab neutralising antibodies RSV respiratory syncytial virus F gene fusion gene BMCMC Bayesian Markov chain Monte Carlo ESS effective sample sizes HPDs highest posterior densities ML marginal likelihood BSPs Bayesian skyline plots dN non-synonymous substitution rates dS synonymous substitution rates SLAC single-likelihood ancestor counting FEL fixed effects likelihood IFEL internal fixed-effects likelihood FUBAR fast unconstrained Bayesian approximation MEME mixed-effects model of evolution 3D three-dimensional BRV bovine respiratory virus ==== Body pmc1 Introduction Human respirovirus 1 (formerly called human parainfluenza virus 1, HRV1) is an RNA virus that belongs to the genus Respirovirus of the family Paramyxoviridae. HRV1 is a causative agent of acute respiratory diseases, such as common colds, acute laryngotracheobronchitis (croup), bronchiolitis, and pneumonia, and is distributed world-wide as the most prevalent type of the former human parainfluenza virus as well as HRV3 (Henrickson, 2003; Karron, 2007). Epidemiological studies showed that HRV is a causative agent for croup in children under five years of age, among which approximately 26–74% experience HRV1 infection (Denny et al., 1983). HRV1 reinfection and HRV3 may occur throughout life; however, the reinfection mechanisms are not exactly known (Henrickson, 2003). The HRV1 genome encodes six genes that are translated into seven proteins (Karron, 2007). Among these, fusion protein (F protein) and haemagglutinin-neuraminidase (HN) proteins are the major viral antigens (Karron, 2007). In particular, the F protein consists of a homotrimer and may be associated with infection of airway epithelial cells in the host (Karron, 2007). Moreover, the existence of two conformations of the F protein, prefusion and postfusion, have been confirmed (Yin et al., 2006). However, detailed F protein structure is not well understood (Shao et al., 2021). Antibody responses are central to acquired immunity against viral infections. Epitopes are classified into two categories: conformational and linear epitopes (Van Regenmortel, 2001). Linear epitopes are continuous amino acid sequences of the primary antigen structure. Conformational epitopes are composed of discontinuous residues that are in proximity on the protein three-dimensional (3D) structure. Both epitopes are recognised by the immune system, triggering the production of antibodies (Sharon et al., 2014; Collins and Karron, 2013; Van Regenmortel, 2001). A previous report showed that over 90% of B cell epitopes are conformational, and only a few are linear (Van Regenmortel, 2001). In contrast, due to an explicit distinction between antigenicity and immunogenicity, these epitopes in antigenic proteins may not be adequately recognised by a neutralising antibody (NT-Ab) (Sharon et al., 2014; Collins and Karron, 2013). Our previous data suggested that the computationally predicted conformational epitopes in the respiratory syncytial virus (RSV) and HRV F proteins do not correspond to the NT-Ab binding sites of these proteins (Aso et al., 2020; Saito et al., 2021). However, to the best of our knowledge, the relationship between conformational and linear epitopes and the NT-Ab binding sites of the HRV1 F protein is not known. Moreover, the detailed phylogeny of the viral fusion (F) gene is unknown. Therefore, detailed molecular evolutionary analyses of the HRV1 F gene were performed in strains collected globally, using bioinformatic technologies. 2 Materials and methods 2.1 Strains used in this study To better understand the molecular evolution of the HRV1 F gene, nucleotide sequences, including the full-length coding region of the gene (positions 5060–6727; 1668 nt for the hPIV1/USA/ATCC VR-94/1957 strain; GenBank accession No. JQ901971) was retrieved from GenBank on 11 June 2019. Among these, sequences from strains with confirmed information on the detected/isolated years and regions were selected. In addition, data of strains that displayed ambiguity with undermined sequences (e.g., N, Y, R, and V) were omitted from the dataset, providing data from 71 strains for the analysis. Homologous sequences were identified using Clustal Omega (Sievers and Higgins, 2021). When three or more similar sequences were present, only two among them were randomly retained, which reduced the final sequence set to those from 66 strains. Temporal signal analysis of the sequences from 66 HRV1 strains was performed to determine whether the dataset was suitable for molecular clock analysis. Maximum likelihood method was used to generate a phylogenetic tree using molecular evolutionary genetics analysis version 7.0 (MEGA 7; for bigger datasets) software. The data were analysed using TempEst software (version 1.5.3) (Rambaut et al., 2016). All data are presented in Supplementary Table S1. These sequences were aligned using MAFFT version 7 (Katoh and Standley, 2013) and subsequently trimmed to 1668 nt based on the prototype F gene sequences. 2.2 Time-Scaled phylogenetic analysis and phylodynamic analyses using the bayesian markov chain monte carlo method (BMCMC) To investigate the evolution of HRV1 strains, a time-scaled phylogenetic analysis of full-length sequences of the HRV1 F gene was conducted using the Bayesian Markov chain Monte Carlo (BMCMC) method in BEAST (version 2.4.8) (Bouckaert et al., 2014). To select a suitable substitution model, jModelTest program (version 2.1.10) was used (Darriba et al., 2012). The path-sampling implemented in the BEAST package was used to determine the best of four clock models (strict clock, exponential relaxed clock, log-normal relaxed clock, and random local clock) and three prior tree models (coalescent constant population, coalescent exponential population, and coalescent Bayesian skyline). The TrN + I substitution model, log-normal of the relaxed clock model, and coalescent exponential from the tree prior model were used for BMCMC analysis of all strains. An BMCMC tree was constructed using BEAST software with the obtained strains and selected models. We analysed the MCMC chains for the 100,000,000 steps with sampling performed after every 2000 steps. To confirm convergence, the effective sample sizes (ESS) were evaluated using Tracer (version 1.6), and values above 200 were considered acceptable. After burn-in of the first 10% of the trees, a maximum clade credibility tree was generated using TreeAnnotator (version 2.4.8) in the BEAST package. The BMCMC phylogenetic tree was visualised using FigTree (version 1.4.03), and the 95% highest posterior densities (HPDs) of all internal nodes were computed. Moreover, strain clustering in the constructed phylogenetic tree of the HRV1 F gene followed the illustrated tree topology. Simultaneously, the evolutionary rates of the 66 HRV1 strains and strains of each lineage determined by the BMCMC phylogenetic tree were estimated using the BMCMC method, and the values were confirmed using the Tracer software. The marginal likelihood (ML) values for model selection and the detailed parameters of the BMCMC analyses are shown in Supplemental Tables S2 and S3. The statistics calculated by Tracer for each dataset are listed in Supplementary Tables S2–S6. Statistical analysis for comparing evolutionary rates between the lineages was performed using the Kruskal–Wallis test with the EZR software (Kanda, 2013). The evolutionary rates sampled every 2000 steps from the MCMC chains after discarding the 10% burn-in (45,001 samples) were used for statistical analysis. Statistical significance was defined as p < 0.05. Past genome population dynamics of the HRV1 F gene were examined using Bayesian skyline plots (BSPs) in BEAST. A coalescent Bayesian skyline was selected as the prior tree model. 2.3 Phylogenetic distance calculation The phylogenetic distances among all HRV1 strains were analysed to estimate F gene diversity. A phylogenetic tree of all HRV1 strains was constructed using the ML method with MEGA7 software (Kumar et al., 2016), and branch reliability was supported by 1000 bootstrap replications. The jModelTest program was used to select the best substitution model for the ML method. Subsequently, the phylogenetic distance of the ML tree was calculated using Patristic (Fourment and Gibbs, 2006). 2.4 Selective pressure analyses The selective pressure sites for the F protein of HRV1 were analysed by calculating the non-synonymous (dN) and synonymous (dS) substitution rates at each amino acid site using the Datamonkey web server (https://www.datamonkey.org/) (Weaver et al., 2018). Single-likelihood ancestor counting (SLAC), fixed effects likelihood (FEL), internal fixed-effects likelihood (IFEL), fast unconstrained Bayesian approximation (FUBAR) (Murrell et al., 2013), and the mixed-effects model of evolution (MEME) (Murrell et al., 2012) were used to estimate positive selection sites, whereas, SLAC, FEL, IFEL, and FUBAR were used to predict negative selection sites. The positive (dN/dS > 1) and negative (dN/dS < 1) selection was determined based on the p values (p < 0.05) for SLAC, FEL, IFEL, and MEME and on the posterior probability values (> 0.9) for FUBAR. 2.5 Modelling of three-dimensional structure of the HRV1 F protein Experimentally validated 3D structure of the HRV1 F protein is not available. Hence, we employed a homology modelling method to construct trimeric structural models of the prefusion F protein of HRV1 for representative strains from each group, determined using the BMCMC phylogenetic tree (prototype, ATCC VR-94/USA/1957 strain, JQ901971; lineage 1, HPIV1/WI/629–008/1997 strain, JQ901978; lineage 2, HPIV1/WI/629–007/1997 strain, JQ901979; and lineage 3, HPIV1/USA/629-D02161/2009 strain, KF687308). The cryo-electron microscopy structure of HRV3 F protein (Protein Data Bank accession ID: 6MJZ) was selected as the template based on the results from BLAST web server (Shao et al., 2021). The amino acid sequences of each strain and template were aligned using MAFFT. The percentage sequence identity of each strain to the template was calculated using Clustal Omega. Based on the template sequence, 3D structures were constructed using Modeller software (version 10.2). The generated models were assessed by Ramachandran plot analyses using WinCoot implemented in the CCP4 package, and the models with the best scores were selected. Energy minimisation of the generated structures was performed using GROMOS96, which was implemented in Swiss PDB Viewer (version 4.1.0) (Guex and Peitsch, 1997). 2.6 Analyses of conformational and linear epitopes and amino acid substitution sites To accurately analyse the pressure of human immune defence against the natural state of the HRV1 F protein, epitopes in the trimeric prefusion state were predicted. The conformational epitopes of the constructed models were analysed using Disco-Tope (version 2.0) (Kringelum et al., 2012), ElliPro (Ponomarenko et al., 2008), SEMA (Shashkova et al., 2022) and SEPPA (version 3.0) (Zhou et al., 2019) with cut-off values of −3.7, 0.5, 0.76, and 0.064, respectively. The accuracy of the analyses was also supported by the consensus sites predicted by more than three of the four methods, and regions with residues close to two of the sites on the trimeric structure models were determined as conformational epitopes. Subsequently, the linear epitopes were analysed using LBtope (Singh et al., 2013) and BECEPS (Ras Carmona et al., 2021), BepiPred (version 2.0) (Galgonek et al., 2017) and ABCpred (Saha and Raghava, 2006). Cut-off values were set as 80% (LBtope), 0.5 (BECEPS), 0.5 (BepiPred 2.0), and 0.51 (ABCpred), respectively. Regions that had more than 10 continuous amino acids and were estimated in common by at least three of the four methods were regarded as linear epitopes. Finally, the predicted and previously identified epitopes were mapped onto the constructed, pre-fusion F protein models using PyMOL (version 2.3) (WL, 2002.). 3 Results 3.1 Time-Scaled phylogenetic analysis and phylodynamic analysis of the HRV 1 F gene using the BMCMC method To estimate the time-scaled evolution of the HRV1 F gene, a phylogenetic tree was constructed using the BMCMC method. In this study, we used only the sequences from HRV1 strains (66 strains) that were detected in humans, because sequence data from bovine respiratory virus (BRV) type 1, which may be a common ancestor of both BRV and HRV, were not available. Before constructing the BMCMC tree, the temporal signal of the dataset was estimated using TempEst (version 1.5.3). The plots of root-to-tip genetic distance against sampling time exhibited a positive correlation between genetic divergence and sampling time, and the R square value was calculated as 0.87 (Figure S1). These results suggest that the dataset of the 66 HRV1 strains was adequate for molecular clock analysis. Hence, we used this dataset to carry out the BMCMC method. As shown in Fig. 1, a common ancestor of the HRV1 prototype strain (hPIV1/USA/ATCC_VR-94_1957; GenBank accession No. JQ901971) and other existing HRV1 strains diverged in 1957 (95% HPD, 1956–1957), resulting ultimately in the formation of three major lineages 1–3. After the first divergence in 1957, strains belonging to lineage 1 further diverged from a common ancestor of strains belonging to the three lineages in 1992 (95% HPD, 1989–1994), and the opposite side of lineage 1 diverged into lineages 2 and 3 in 1994 (95% HPD, 1991–1996). Currently, strains belonging to lineage 3 are widespread and form several clusters.Fig. 1 Time-scaled evolutionary tree of the full length Human respirovirus 1 (HRV1) fusion gene constructed by the Bayesian Markov chain Monte Carlo (BMCMC) method. The scale bar represents time (years). Green bars indicate the 95% highest posterior density (HPD) for each branch year. Fig 1 Next, the evolutionary rate of HRV1 F gene was estimated (Table 1). The evolutionary rate of all strains was estimated to be 8.504 × 10−4 substitutions/site/year (s/s/y) (95% HPD, 7.003 × 10−4 to 1.0008 × 10−3 s/s/y). The calculations for each of the above lineages showed that the evolutionary rate of strains belonging to lineage 2 was 6.580 × 10−4 s/s/y (95% HPD, 4.784 × 10−4 to 8.4595 × 10−4 s/s/y), and that of strains belonging to lineage 3 was 1.205 × 10−3 s/s/y (95% HPD, 7.159 × 10−4 to 1.6866 × 10−3 s/s/y). The evolutionary rate of the strains in lineage 1 with the same detection year (1997) was not calculated. The evolutionary rate of strains belonging to lineage 3 was significantly higher than that of strains belonging to lineage 2 (p < 2 − 16), which may indicate that strains belonging to lineage 3 are more adapted to humans, although the mechanisms underlying the significance of these values are not known.Table 1 Evolutionary rates of all HRV1 strains and each lineage. Table 1 Evolutionary rates (95% HPD) (substitutions/site/year) Effective sample size All strains (66 strains) 8.504 × 10−4 (7.003 × 10−4 to 1.0008 × 10−3) 220 Lineage 1 (6 strains) ― ― Lineage 2 (23 strains) 6.580 × 10−4 (4.784 × 10−4 to 8.4595 × 10−4) 4053 Lineage 3 (36 strains) 1.205 × 10−3 (7.159 × 10−4 to 1.6866 × 10−3) 954 3.2 Phylodynamics of the HRV1 F gene using the bayesian skyline plot (BSP) analysis method As shown in Fig. 2, the phylodynamics of the F gene in HRV1 strains were analysed using the BSP analysis method to detect fluctuations in effective population size. The genome population size of all the strains doubled between 1995 and 2008 (Fig. 2A). Similarly, strains belonging to lineage 2 showed a two-fold increase in genome population around 2003 and 2008 (Fig. 2B). In contrast, in lineage 3, a steep increase in genome population size was observed, even though it occurred only once around 2008 (Fig. 2C). Because the detection year of the strains belonging to lineage 1 was the same (1997), we could not calculate the genome population size of this lineage. To summarise the results of these BSP analyses, the rapid increase in genome population size around 2008 for all strains was speculated to be mainly due to an increase in the genome population size of lineage 3.Fig. 2 Bayesian skyline plot for the Human respirovirus 1 (HRV1) fusion gene. Each panel illustrates the phylodynamics of all 66 strains (a), lineage 2 (b), and lineage 3 (c). Y and x-axes indicate the effective population size and time in years, respectively. Thick black lines show the median values over time; thin blue lines represent the 95% highest posterior density (HPD) intervals. Fig 2 3.3 Phylogenetic distances calculation of the HRV1 F gene The phylogenetic distance and distribution between strains were evaluated based on their nucleotide sequences. A histogram of the distances between the sequence pairs of all the strains revealed a bimodal distribution (Fig. 3). Furthermore, histograms of lineages 1 and 2 showed bimodal distributions, although the apparent phylogenetic distances of lineage 1 may not represent the phylogenetic distances of the actual lineage, owing to the small sample size (Fig. 3B and 3C). In contrast, the histogram of lineage 3 showed a unimodal pattern (Fig. 3D). The mean distance (± SD) between each pair of F gene sequence in the 66 HRV1 strains examined in this study was 0.018575 ± 0.01227. The results of the study showed that lineages 1, 2 and 3 had phylogenetic distances of 0.0022 ± 0.0012, 0.0073 ± 0.0030, and 0.0092 ± 0.0058, respectively. Thus, the phylogenetic distance for each of the lineage was less than 0.02, suggesting conservation of the F gene sequence.Fig. 3 Distribution of phylogenetic distances between the full-length sequences of the fusion gene of all Human respirovirus 1 (HRV1) strains. Each panel illustrates the histogram of all 66 strains (a), lineage 1 (b), lineage 2 (c), and lineage 3 (d). The y-axis and x-axis indicate the number of sequence pairs and phylogenetic distances, respectively. Fig 3 3.4 Homology modelling To visualise the relationship between NT-Ab binding sites and predicted B-cell conformational epitopes, we constructed the homotrimer (chains A, B, and C) of the HRV1 pre-fusion protein structure (Fig. 4). The amino acid sequence of the template covered amino acids 24–98 and 126–550 in each strain (Fig. 5). In this range, the amino acid residues on the protein surface were the same between representative strains of lineages 1–3. Moreover, from homologous analysis using Clustal Omega, the percentage sequence identity value of the prototype strain against the representative strains was high (96.8%). Hence, we presented the prototype structural model alone and showed the sites where amino acid substitutions occurred in representative strains from lineages 1 to 3 (Figs. 4 and 5). Both the prototype and lineage 1 representative strains had the same sequence identity (44.3%) as the template.Fig. 4 Structural models of the prefusion protein of USA/1957 strain. Chains of the trimeric structures are given in white (chain A), light grey (chain B), and dark grey (chain C). The conformational and linear epitopes have been indicated in green and blue, respectively. Areas where the conformational and linear epitopes overlap are shown in purple. Experimentally identified epitopes (neutralising antibody binding sites) are indicated in red, and those overlapping with the conformational epitope are coloured yellow. Fig 4 Fig. 5 The Human respirovirus 1 (HRV1) fusion (F) protein chain A amino acid sequences investigated in this study. The amino acid residue numbers used for constructing HRV1 F protein three-dimensional (3D) structure are highlighted in yellow. The amino acid residues of neutralising antibodies (NT-Ab) binding sites and mutation are highlighted in orange and light green, respectively. The black and green line boxes in the sequences are the conformational epitope and linear epitope sites, respectively. Blue line boxes in the sequence indicate negative selection sites detected by all four methods (single-likelihood ancestor counting, fixed effects likelihood, internal fixed-effects likelihood, and fast unconstrained Bayesian approximation). Fig 5 3.5 Selective pressure analyses of HRV1 F protein The rates of dS and dN substitutions were estimated using the DataMonkey web server to identify the positive and negative selection sites of F proteins in all 66 strains (Table S7). Only one method (FUBAR) predicted a positive selection site (amino acid residue 5), and the other four methods identified no positive selection sites. Thus, a positive selection site of the F protein is absent. In contrast, many negative selection sites were identified (Table S7). Among them, four negative selection sites (amino acids 150, 382, 460, and 473) were detected using all methods employed (Fig. 5). 3.6 Analyses of B-Cell epitopes and amino acid substitution sites The amino acid substitution sites, NT-Ab binding sites, and predicted epitopes are shown in the HRV1 F chain A amino acid sequences and the trimeric structural model of the prototype (Figs. 4 and 5). First, the amino acid substitution sites in the F protein chain A among the prototype strain and the representative strains of lineages 1, 2, and 3 were compared. Seventeen amino acid substitutions were common in lineages 1, 2, and 3 (Fig. 5 and Table S8). An amino acid substitution unique to lineage 1 is present in Glu5Lys. Moreover, four amino acid residues (Thr493Lys, Val526Thr, Met545Ile, and Arg546Lys) showed substitutions that were unique to lineage 3. However, some of the common substitution regions of these amino acid substitution sites are not located on the surface of the 3D structure. Similarly, uncommon substitution residues between lineages 1 and 3 were not present in the 3D structural model. Thus, only seven residues (Glu63Gln, Ile155Val, Leu163Phe, Asn184Asp, Arg338Lys, Arg410Lys, and Arg442Gly) of the common substitution regions in each lineage were located on the surface of the 3D structure model. Next, conformational and linear B-cell epitopes on HRV1 F protein were analysed. Six conformational epitope sites and seven linear epitope sites were identified for the prototype HRV1 F protein chain A (18 conformational epitopes and 21 linear epitopes in the trimeric structure) (Figs. 4 and 5 and Table S9). No amino acid substitutions were found at these sites of strains in lineages 1, 2, or 3, whereas only one residue substitution (Glu63Gln) was found near one of the predicted conformational epitope sites (Fig. 5). Notably, in the HRV1 F protein chain A, five of the six conformational epitope sites and all the linear epitope sites failed to coincide with the experimentally determined NT-Ab binding sites, whereas only one residue of the conformational epitope (aa 473) coincided (Fig. 5). This mismatch may be a possible mechanism by which HRV1 can reinfect humans. 4 Discussion Evolutionary studies of HRV have been reported (Bose et al., 2019; Mao et al., 2012; Mizuta et al., 2014), but most of these focus on HRV3 (Mao et al., 2012; Mizuta et al., 2014). A few reports regarding the HRV1 F gene have been published involving domestic or partial F gene analyses (Ambrose et al., 1995; Aso et al., 2020). To study the detailed molecular evolution of the full-length F gene in HRV1 strains from various countries, we performed time-scaled phylogenetic, genome population size, and selective pressure analyses on the gene, as well as antigenicity analysis of the F protein. From the time-scaled phylogenetic tree, constructed using the BMCMC method, it was estimated that the common ancestor of the HRV1 F gene diverged in 1957 and that their progenies continuously evolved and formed three lineages (lineages 1–3, Fig. 1). Strains belonging to lineage 3 predominate in various countries. Second, phylodynamic analyses using the BSP method showed that the genome population size of the F gene doubled over approximately 80 years. Third, phylogenetic distances among the strains were short (< 0.02; Fig. 3). Finally, no positive selection sites were detected in the F protein, whereas many sites were identified as negative selection sites. Moreover, five sites of the six conformational epitopes and all linear epitopes in each chain of the F protein lacked correspondence to the NT-Ab binding sites (Figs. 4 and 5). These results suggest that despite the apparent conservation, the HRV1 F gene has evolved over many years. Yet, the conformational and linear epitopes did not correspond to the NT-Ab binding sites in either the pre or postfusion forms of the protein. This mismatch may be partially responsible for HRV1 reinfection and may extend to related viruses, such as HRV3 and RSV. A phylogenetic analysis of HRV1 F gene was performed using the BMCMC method. It revealed that this gene continuously evolved and formed three lineages with many clusters (Fig. 1). Our previous report showed full-length F protein genes in HRV1 among patients with acute respiratory infections in Eastern Japan during 2011–2015 (Tsutsui et al., 2017). Phylogenetic analyses using the ML method showed that HRV1 strains formed three lineages and that the lineage 3 strains were dominant during the investigation period (2011–2015). This finding is consistent with the results of the present study. However, the time-scaled phylogeny of the HRV1 F gene was not assessed in the previous study (Tsutsui et al., 2017). Here, time-scaled phylogenetic analyses were done using the BMCMC method. As a result, the divergence year of a common ancestor and each lineage was estimated (Fig. 1). Although the analysed gene was distinct, our previous report suggested that the HRV1 haemagglutinin-neuraminidase (HN) glycoprotein gene (full-length) isolated from Yamagata prefecture, in northern Japan, was classified into two lineages and formed many clusters using different phylogenetic analysis methods, including neighbour-joining and ML methods (Mizuta et al., 2011, 2014). Thus, to the best of our knowledge, the present study may be the first time-scaled phylogenetic analysis of the HRV1 F gene based on full-length sequences from globally collected strains. However, the present study has some limitations, including, the relatively small number of strains used. This is due to the paucity of studies on HRV1 molecular epidemiology. Another limitation is selection bias owing to the limited number of countries studying HRV1 and HRV3 (Aso et al., 2020). The evolutionary rate of the HRV1 F gene has also been estimated. The mean evolutionary rate was 8.504 × 10−4 s/s/y. This is similar to the rates reported for the F genes of HRV3 and RSV (Aso et al., 2020). Furthermore, the evolutionary rate of HRV1 strains belonging to lineage 3 was found to be faster than that of lineage 2, whereas a previous evolutionary study on the HRV3 F gene did not find this difference (Aso et al., 2020). Moreover, these findings were not found for the RSV F gene, a virus belonging to a different genus and species (Saito et al., 2021). Thus, these findings were only observed for HRV1 F gene, which to the best of our knowledge, is the first report of lineage differences in evolutionary rate. In addition, the rapid evolutionary rate reflects short generation times and/or strong positive selection, which may generate a phenotype that is more adapted to the host (Collins and Karron, 2013). Together, the strains belonging to lineage 3 were more adaptive to humans and could become dominant strains, although our study did not address the mechanisms underlying the difference in the evolutionary rate. The mean phylogenetic distance of the F gene HRV1 strains was approximately 0.02 (Fig. 3). This agrees with our previous study of Japanese strains, which reported a phylogenetic distance for the HRV1 F gene of 0.026 (Tsutsui et al., 2017). Moreover, the distribution of distances in our study was similar to that in previous reports on the HRV3 F gene (Aso et al., 2020) as well as the HRV3 HN gene (Mao et al., 2012; Takahashi et al., 2018). These data suggest that the diversity of various viruses carrying the F gene may be similar and restricted within each species. With its high conservation and pivotal role in entry, the HRV1 F protein can be an attractive target for prophylaxis therapy, as it does for RSV F protein (Battles and McLellan, 2019). Phylodynamic analyses of HRV1 F gene (Fig. 2) showed that the genome population of the gene has doubled over approximately 80 years. These fluctuations corresponded to the emergence of strains belonging to lineages 2 and 3 (Figs. 1 and 2). Our previous report showed that the genome population size of HRV3 F gene increased only once between 2000 and 2010 (Aso et al., 2020). Thus, genome fluctuations between HRV1 and 3 were different. The selective pressure analyses of the HRV1 F gene did not reveal any positive selection sites for the present strains, whereas many negative selection sites were identified. In general, positive selection sites reflect escape from host defence mechanisms, such as cellular or humoral immunity (Barreiro and Quintana-Murci, 2010). Thus, HRV1 F protein may not be affected by such defence mechanisms. Conversely, negative selection sites may act to prevent deterioration of antigenicity (Holmes, 2013; Loewe, 2008). These findings may reflect the essential role the HRV1 F protein plays in host cell infection. Similar findings have been reported for the HRV3 F protein (Henrickson, 2003; Tsutsui et al., 2017). The negative selection sites of the HRV1 F protein were clustered near predicted epitopes, based on the 3D structural modelling. This indicates that these sites play important roles, for example, in the cellular receptor binding domain. Finally, we analysed the amino acid substitutions and conformational and linear epitopes in the HRV1 F protein and evaluated their relationship with the NT-Ab binding sites (Figs. 4 and 5). Notably, amino acid substitutions were not estimated for computationally predicted epitopes or NT-Ab binding sites. Furthermore, almost all experimentally determined NT-Ab binding sites were incompatible with the computationally estimated conformational epitopes and linear epitopes. Antibody responses play a pivotal role in virus neutralisation or elimination, and antigenicity and immunogenicity differ explicitly (Collins and Karron, 2013). Conformational and linear epitopes may stimulate the production of antiviral antibodies (Lo et al., 2021; Sharon et al., 2014). In contrast, predicted epitope sites on antigenic proteins may not be adequately recognised by NT-Abs, which might mean that these predicted epitopes have weak potential for producing NT-Abs (Collins and Karron, 2013; Lo et al., 2021; Sharon et al., 2014). Thus, incompatibilities between NT-Ab binding sites and predicted epitopes may indicate low immunogenicity of the F protein and may be partially responsible for HRV1 reinfection, as has been reported for reinfections with HRV3 and RSV (Aso et al., 2020; Saito et al., 2021). However, the paucity of antigenic/antibody complex structures may lead to a mismatch between the NT-Ab binding sites and the predicted epitopes. Hence, the interpretation of the computationally conducted epitope analyses in this study may be limited, although conformational and linear epitopes were investigated using multiple computational methods for increased accuracy. 5 Conclusions In this study, molecular evolutionary analyses of HRV1 F gene were performed based on full-length sequences collected globally. The time-scaled phylogenetic tree generated by the BMCMC method estimated that the common ancestor of the HRV1 F gene diverged in 1957, and that their progenies have continuously evolved and formed three lineages. The phylodynamic analyses using the BSP method showed that the genome population size of the F gene doubled over approximately 80 years. The phylogenetic distances among the strains were short with no positive selection sites. Almost all conformational and linear epitopes in the F protein did not correspond to NT-Ab binding sites. These results showed that the HRV1 F gene has evolved over many years, although the gene may be relatively conserved. Moreover, incompatibility between predicted epitopes and the NT-Ab binding sites in both the pre and postfusion forms of the protein may be responsible for HRV1 virus reinfection, as well as reinfections with HRV3 and RSV. Funding This work was supported by a commissioned project for 10.13039/100009619 Research on Emerging and Reemerging Infectious Diseases from the Japan Agency for Medical Research and Development (AMED; grant number JP23fk0108661). Data availability statement The data sets generated and analysed during the current study are available from the corresponding author upon reasonable request. CRediT authorship contribution statement Tomoko Takahashi: Methodology, Data curation, Writing – original draft. Mao Akagawa: Methodology, Visualization. Ryusuke Kimura: Methodology, Visualization. Mitsuru Sada: Methodology, Writing – original draft, Visualization. Tatsuya Shirai: Writing – original draft. Kaori Okayama: Writing – original draft. Yuriko Hayashi: Writing – original draft. Mayumi Kondo: Methodology. Makoto Takeda: Writing – review & editing. Akihide Ryo: Writing – review & editing. Hirokazu Kimura: Conceptualization, Writing – original draft, Writing – review & editing. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Appendix Supplementary materials Image, application 1 Acknowledgments We thank Ms. Miki Kawaji for the skilful support in figure preparation. Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.virusres.2023.199142. ==== Refs References Ambrose M. Hetherington S. Watson A. Scroggs R. Portner A. Molecular evolution of the F glycoprotein of human parainfluenza virus type 1 J. Infect. Diseases 171 4 1995 851 856 7706810 Aso J. Kimura H. Ishii H. Saraya T. Kurai D. Matsushima Y. Nagasawa K. Ryo A. Takizawa H. Molecular evolution of the fusion protein (F) gene in human respirovirus 3 Front. Microbiol. 10 2020 3054 32010105 Barreiro L.B. Quintana-Murci L. From evolutionary genetics to human immunology: how selection shapes host defence genes Nat. Rev. Genet 11 1 2010 17 30 19953080 Battles M.B. McLellan J.S. Respiratory syncytial virus entry and how to block it Nat. Rev. Microbiol. 17 2019 233 245 30723301 Bose M.E. Shrivastava S. He J. Nelson M.I. Bera J. Fedorova N. Halpin R. Town C.D. Lorenzi H.A. Amedeo P. Gupta N. Noyola D.E. Videla C. Kok T. Buys A. Venter M. Vabret A. Cordey S. Henrickson K.J. Sequencing and analysis of globally obtained human parainfluenza viruses 1 and 3 genomes PLoS. ONE 14 7 2019 e0220057 Bouckaert R. Heled J. Kühnert D. Vaughan T. Wu C.-H. Xie D. Suchard M.A. Rambaut A. Drummond A.J. BEAST 2: a software platform for Bayesian evolutionary analysis PLoS. Comput. Biol. 10 4 2014 e1003537 Collins P.L. Karron R.A. Respiratory Syncytial Virus and Metapneumovirus Knipe D.M. Howley P. Fields Virology 6th ed. 2013 Lippincott Williams & Wilkins Philadelphia 1086 1123 Darriba D. Taboada G.L. Doallo R. Posada D. jModelTest 2: more models, new heuristics and parallel computing Nat. Methods 9 8 2012 772-772 Denny F.W. Murphy T.F. Clyde W.A. Jr Collier A.M. Henderson F.W. Senior R. Sheaffer C. Conley W. III Christian R. Croup: an 11-year study in a pediatric practice Pediatrics 71 6 1983 871 876 6304611 Fourment M. Gibbs M.J. PATRISTIC: a program for calculating patristic distances and graphically comparing the components of genetic change BMC. Evol. Biol. 6 1 2006 1 5 16388682 Galgonek J. Vymetal J. Jakubec D. Vondrášek J. Amino Acid Interaction (INTAA) web server Nucl. Acids. Res. 45 W1 2017 W388 W392 28472475 Guex N. Peitsch M.C. SWISS-MODEL and the Swiss-Pdb Viewer: an environment for comparative protein modeling Electrophoresis 18 15 1997 2714 2723 9504803 Henrickson K.J. Parainfluenza viruses Clin. Microbiol. Rev. 16 2 2003 242 264 12692097 Holmes E.C. Virus evolution Knipe D.M. Howley P. Fields Virology 6th ed. 2013 Lippincott Williams & Wilkins Philadelphia 286 313 Kanda Y. Investigation of the freely available easy-to-use software 'EZR' for medical statistics Bone Marrow Transpl. 48 3 2013 452 458 Karron R.A. Parainfluenza viruses Fields Virol. 2007 1497 1526 Katoh K. Standley D.M. MAFFT multiple sequence alignment software version 7: improvements in performance and usability Mol. Biol. Evol. 30 4 2013 772 780 23329690 Kringelum J.V. Lundegaard C. Lund O. Nielsen M. Reliable B cell epitope predictions: impacts of method development and improved benchmarking PLoS. Comput. Biol. 8 12 2012 e1002829 Kumar S. Stecher G. Tamura K. MEGA7: molecular evolutionary genetics analysis version 7.0 for bigger datasets Mol. Biol. Evol. 33 7 2016 1870 1874 27004904 Lo Y.-T. Shih T.-C. Pai T.-W. Ho L.-P. Wu J.-L. Chou H.-Y. Conformational epitope matching and prediction based on protein surface spiral features BMC. Genomics 22 Suppl_2 2021 116-116 Loewe L. Negative Selection Nat. Educ. 1 1 2008 59 Mao N. Ji Y. Xie Z. Wang H. Wang H. An J. Zhang X. Zhang Y. Zhu Z. Cui A. Xu S. Shen K. Liu C. Yang W. Xu W. Human parainfluenza virus-associated respiratory tract infection among children and genetic analysis of HPIV-3 strains in Beijing, China PLoS. ONE 7 8 2012 e43893 22937119 Mizuta K. Saitoh M. Kobayashi M. Tsukagoshi H. Aoki Y. Ikeda T. Abiko C. Katsushima N. Itagaki T. Noda M. Kozawa K. Ahiko T. Kimura H. Detailed genetic analysis of hemagglutinin-neuraminidase glycoprotein gene in human parainfluenza virus type 1 isolates from patients with acute respiratory infection between 2002 and 2009 in Yamagata prefecture, Japan Virol. J. 8 2011 533-533 Mizuta K. Tsukagoshi H. Ikeda T. Aoki Y. Abiko C. Itagaki T. Nagano M. Noda M. Kimura H. Molecular evolution of the haemagglutinin-neuraminidase gene in human parainfluenza virus type 3 isolates from children with acute respiratory illness in Yamagata prefecture Japan. J. Med. Microbiol 63 Pt 4 2014 570 577 Murrell B. Moola S. Mabona A. Weighill T. Sheward D. Kosakovsky Pond S.L. Scheffler K. FUBAR: a fast, unconstrained bayesian approximation for inferring selection Mol. Biol. Evol. 30 5 2013 1196 1205 23420840 Murrell B. Wertheim J.O. Moola S. Weighill T. Scheffler K. Kosakovsky Pond S.L. Detecting individual sites subject to episodic diversifying selection PLoS. Genet. 8 7 2012 e1002764 Ponomarenko J. Bui H.-H. Li W. Fusseder N. Bourne P.E. Sette A. Peters B. ElliPro: a new structure-based tool for the prediction of antibody epitopes BMC. Bioinformatics 9 1 2008 1 8 18173834 Rambaut A. Lam T.T. Max Carvalho L. Pybus O.G. Exploring the temporal structure of heterochronous sequences using TempEst (formerly Path-O-Gen) Virus. Evol 2 1 2016 vew007-vew007 Ras Carmona A. Pelaez Prestel H.F. Lafuente E.M. Reche P.A. BCEPS: a Web Server to Predict Linear B Cell Epitopes with Enhanced Immunogenicity and Cross-Reactivity Cells 10 10 2021 Saha S. Raghava G.P.S. Prediction of continuous B-cell epitopes in an antigen using recurrent neural network Proteins 65 1 2006 40 48 16894596 Saito M. Tsukagoshi H. Sada M. Sunagawa S. Shirai T. Okayama K. Sugai T. Tsugawa T. Hayashi Y. Ryo A. Detailed evolutionary analyses of the F gene in the respiratory syncytial virus subgroup A Viruses 13 12 2021 2525 34960794 Shao N. Liu B. Xiao Y. Wang X. Ren L. Dong J. Sun L. Zhu Y. Zhang T. Yang F. Genetic characteristics of human parainfluenza virus types 1–4 from patients with clinical respiratory tract infection in China Front. Microbiol 2021 12 Sharon J. Rynkiewicz M.J. Lu Z. Yang C.Y. Discovery of protective B-cell epitopes for development of antimicrobial vaccines and antibody therapeutics Immunology 142 1 2014 1 23 24219801 Shashkova T.I. Umerenkov D. Salnikov M. Strashnov P.V. Konstantinova A.V. Lebed I. Shcherbinin D.N. Asatryan M.N. Kardymon O.L. Ivanisenko N.V. SEMA: antigen B-cell conformational epitope prediction using deep transfer learning bioRxiv 2022 Sievers, F., Higgins, D.G., 2021. The Clustal Omega Multiple Alignment Package. Vol. 2231. Humana Press, pp. 3–16. Singh H. Ansari H.R. Raghava G.P.S. Improved method for linear B-cell epitope prediction using antigen's primary sequence PLoS. ONE 8 5 2013 e62216-e62216 Takahashi M. Nagasawa K. Saito K. Maisawa S.-i. Fujita K. Murakami K. Kuroda M. Ryo A. Kimura H. Detailed genetic analyses of the HN gene in human respirovirus 3 detected in children with acute respiratory illness in the Iwate Prefecture Japan. Infection,. Genetics. and. Evolution 59 2018 155 162 Tsutsui R. Tsukagoshi H. Nagasawa K. Takahashi M. Matsushima Y. Ryo A. Kuroda M. Takami H. Kimura H. Genetic analyses of the fusion protein genes in human parainfluenza virus types 1 and 3 among patients with acute respiratory infections in Eastern Japan from 2011 to 2015 J. Med. Microbiol 66 2 2017 160 168 28266286 Van Regenmortel M.H. Antigenicity and immunogenicity of synthetic peptides Biologicals 29 2001 209 213 11851317 Weaver S. Shank S.D. Spielman S.J. Li M. Muse S.V. Kosakovsky Pond S.L. Datamonkey 2.0: a modern web application for characterizing selective and other evolutionary processes Mol. Biol. Evol. 35 3 2018 773 777 29301006 WL D. The PyMOL Molecular Graphics System 2002 DeLano Scientific San Carlos, CA Yin H.-S. Wen X. Paterson R.G. Lamb R.A. Jardetzky T.S. Structure of the parainfluenza virus 5 F protein in its metastable, prefusion conformation Nature 439 7072 2006 38 44 16397490 Zhou C. Chen Z. Zhang L. Yan D. Mao T. Tang K. Qiu T. Cao Z. SEPPA 3.0—enhanced spatial epitope prediction enabling glycoprotein antigens Nucleic. Acids. Res. 47 W1 2019 W388 W394 31114919
PMC010xxxxxx/PMC10352715.txt
==== Front RSC Adv RSC Adv RA RSCACL RSC Advances 2046-2069 The Royal Society of Chemistry d3ra03259k 10.1039/d3ra03259k Chemistry The role of electrostatic potential in the translocation of triangulene across membranes† † Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d3ra03259k https://orcid.org/0009-0006-6392-1650 Tang Xiaofeng a‡ Li Youyun b‡ Li Qianyan a Yu Jinhui a Bai Han ac a Yunnan Cancer Hospital, The Third Affiliated Hospital of Kunming Medical University Kunming People's Republic of China [email protected] [email protected] b The Second Affiliated Hospital of Kunming Medical University Kunming People's Republic of China c School of Physics and Astronomy, Yunnan University Kunming People's Republic of China ‡ Xiaofeng Tang and Youyun Li contributed equally to this work. 18 7 2023 12 7 2023 18 7 2023 13 31 2154521549 16 5 2023 30 6 2023 This journal is © The Royal Society of Chemistry 2023 The Royal Society of Chemistry https://creativecommons.org/licenses/by-nc/3.0/ Triangulene and its derivatives show broad application prospects in the fields of biological imaging and biosensing. However, its interaction with cell membranes is still poorly studied. In this study, classical molecular dynamics simulations were used to adjust the electrostatic potential of triangulene to observe its interactions with cell membranes. We found that electrostatic potential not only affects the behavior as it enters the cell membrane, but also spatial distribution within the cell membrane. The angle distribution of inside-0 and all-0 triangulene when penetrating the membrane is more extensive than that of ESP triangulene. However, inside-0 triangulene could cross the midline of the cell membrane and prefers to stay in the upper leaflet, while all-0 triangulene and ESP triangulene can reach the lower leaflet. These findings can help us regulate the distribution of nanoparticles in cells, so as to design functional nanoparticles that conform to the requirements. The electrostatic potential can affect the angle and spatial distribution in the translocation of triangulene across membranes. Yunnan Provincial Science and Technology Department 10.13039/501100008871 202101AY070001-164 pubstatusPaginated Article ==== Body pmc1. Introduction Recent years have witnessed a spurt of progress in nanotechnology, and nanomaterials have shown increasing potential in biomedical fields such as targeted drug delivery, radiosensitization, and bioimaging.1–4 The cell membrane is the first barrier for the interaction between biological systems against nanoparticles, and its interaction with nanoparticles will be the beginning of a series of subsequent reactions.5–7 Hence, figuring out how nanoparticles interact with the cell membrane is crucial. It has been demonstrated that surface charge of nanoparticles has the potential to regulate the way they interact with cell membranes. For example, the periodically distributed electrostatic potential of C2N nanosheets can prevent it from damaging the cell membrane.8 Moreover, our recent research has found that the more polarized electrostatic potential of graphene quantum dots makes it harder to enter the cell membrane.9 Besides, because of the highly polar nature of the head group of phospholipid molecules of cell membranes, it has been recognized that modulating the electrostatic potential (EP) of nanomaterials is one of the most effective methods to regulate the nanoparticle–cell membrane interactions.10,11 Therefore, it is essential to understand how electrostatic potential (EP) of nanomaterials plays a role in the translocation across membranes. This can help us regulate the location of nanoparticles in the cell and design the required functional nanoparticles. Triangulene is one of the most famous triplet-ground-state benzenoid hydrocarbons with interesting biological properties.12–14 Yu created a benchmark dataset of 25 magnetic systems with nonlocal spin densities that can more accurately predict electronic ground state for triangulene analogues.15 Shen introduced a new intramolecular radical–radical coupling approach and successfully synthesize two fused triangulene dimers efficiently.16 Arikawa synthesized a kinetically-stabilized nitrogen-doped triangulene cation derivative.17 Besides, atomic force microscopy (AFM) appears as an invaluable experimental technique, allowing the measurement of the mechanical strength of biomolecular complexes to provide a quantitative characterization of their interaction properties from a single molecule perspective.18 The above research indicates that stable triangulene may be synthesized in the future, and tools such as AFM may help us analyze the interaction process between triangulene and cell membrane in experiments.19 Recent research has found that the uniformly distributed electrostatic potential allows the system to have a small permanent dipole moment that blocks the electronic transition in the light excitation such that the electronic transition can only be carried out between adjacent carbon atoms.14 According to the calculation of density functional theory (DFT), the derivatives of triangulene have high affinity for DNA helix.20–23 Nevertheless, there are different mechanisms in healthy tissues and tumours. Unlike the non-toxic buckled form in healthy tissues, in tumours with relatively low pH values, the derivatives can intercalate DNA double stranded base pairs in a planar form to disrupt biological processes, showing their potential anti-cancer properties.13 In addition, the derivatives of triangulene can be used for specifically labeling cell mitochondria, lysosomes and other organelles with cell structures well preserved, thereby allow for various applications in the fields of biosensing and imaging due their excellent photo-physical properties.24,25 However, the interaction mechanism between the electrostatic potential of triangulene and cell membrane is unrevealed, which limits its application in biomedicine. In this study, we simulated the translocation of triangulene with different electrostatic potential in the cell membrane. Since 1-palmitoyl-2-oleoylphosphatidylcholine (POPC) is among the primary constituents of cellular membranes, it is often used to mimic eukaryote cell membrane.26 Classical molecular dynamics (MD) simulations were employed to illustrate the binding dynamics of triangulene to the 1-palmitoyl-2-oleoylphosphatidylcholine (POPC) membrane. We constructed three groups of triangulene with different potentials, one of which had edges with potential polarity (inside-0), another group of evenly distributed potentials (ESP), and the third group with their potentials ignored (all-0). 2. Models and methods 2.1. System setup As shown in Fig. 1, we constructed three groups of triangulene with different potentials. The net charge of each triangulene is 0e (shown in the S1†), and the force field parameters are derived from previous studies.27,28 The first group of triangulene atomic charges are calculated by DFT which is called ESP.29 The second group of triangulene atomic charges are acquired from ref. 16, the partial charges of hydrogen atoms and the linked carbon atoms were set to +0.115e and −0.115e, while carbon atoms away from the edge remained neutral.28 The third group of triangulene atomic charges were assigned neutral charge. The electrostatic potential of triangulene are calculated by Adaptive Poisson–Boltzmann Solver (APBS).30–32 The triangulene are placed on a plane parallel to the x–y plane, about 3 nm from the center of the POPC membrane. The side length of triangulene is 0.925 nm. In the MD simulation, 188 POPC lipids were used with 94 lipids in each layer. The POPC membrane was equilibrated at 100 ps in NPT ensemble. The solvent added to the system is the TIP3P-water molecules, and the size of the box is 7.64 nm × 7.64 nm × 10 nm.33 The whole system was filled with 0.15 mol L−1 NaCl (44 Na+ and Cl−) to mimic the physiological environment (shown in the S2†). Fig. 1 The initial placement position of the system. (a) Side view (b) top view. (c), (d), and (e) Corresponding to surface distribution of electrostatic potential of ESP triangulene, inside-0 triangulene and all-0 triangulene, respectively. 2.2. Simulation parameters All MD simulations were performed using the GROMACS package (version 4.6.7) with a time step of 2 fs. The parameters of the POPC lipid membrane were derived from the Charmm36 force field.34 The minimized system by the steepest descent method accepted a 2 ns NPT ensemble pre-equilibration. The v-rescale thermostat and semiisotropic Parrinello–Rahman barostat matained the pressure of the system at 1 atm and the temperature at 310 K.35,36 The short-range electrostatic and van der Waals interaction cut off distance was set to 1.2 nm, and the long-range electrostatic interaction was treated by the particle-mesh Ewald (PME) algorithm.37 The SETTLE algorithm was used for the water model, and all heavy atoms connected to H atoms were constrained using the LINCS algorithm.38,39 More details are available in our previous search.9 2.3. Potential of mean force (PMF) calculation By analyzing the three different electrostatic potential of the triangulene through the membrane process, we can calculate the free energy, thus explaining why the triangulene will finally stay. To compute the free energy profile along the Z axis we employ an umbrella sampling method.40 Pulling triangulene from 0 nm (center of mass of POPC membrane) to 3.4 nm along the Z axis, with an umbrella sampling window of 0.1 nm. The constraining force constant was 2000 kJ mol−1 nm−2. Notedly, the constraining force was only used for umbrella sampling to calculate PMF, while not imposed on the triangulation during MD simulation. The PMF curve is obtained by g_wham the tool.41 3. Results and discussion 3.1. Angular distribution of triangulene translocation across the POPC membrane The whole system was divided into three regions along the Z axis, 5 nm to 2 nm was defined as region I (bulk water region), 2 nm to 0 nm was region II (upper leaflet), 0 to −2 nm was region III (lower leaflet). Relationship between angular distribution of triangulene and energy barrier was investigated. We found that ESP triangulene is easier to enter the membrane at an angle of 90° when permeating the membrane (Fig. 2b). However, inside-0 triangulene was more extensive when translocating into the membrane. It had the lowest energy barrier from 45 to 90° for inside-0 triangulene to mitigate from bulk water to upper leaflet (Fig. 2c). The all-0 triangulene was more effortless to enter the membrane at angles of 70, 90, 110 and 120° (Fig. 2d). As shown in Fig. 2, ESP triangulene and all-0 triangulene can get into the membrane and can ulteriorly cross the middle of the membrane to the lower leaflet, with the lowest energy barrier at 0.8 nm and −0.8 nm. While, the inside-0 triangulene can only get the upper leaflet of the membrane and preferred to remain in the upper layer region. Fig. 2 (a) Density distribution of POPC membrane, water and phosphate group, defining the boundary of membrane at 2 nm. (b), (c), and (d) respectively show the distribution of angle (formed by ESP, inside-0 and all-0 triangulene surfaces and membrane surfaces) and the energy barrier along the Z axis during translocation. 3.2. The spatial distribution of triangulene in the system These three types of triangulene moved irregularly outside the membrane, whereas their behaviors changed obviously after entering the membrane (Fig. 3). The inside-0 triangulene had been slightly fluctuated and kept in region II (upper leaflet). ESP triangulene had been stayed mostly in region II, occasionally passing through the midline to the region III (lower leaflet). All-0 triangulene moved back and forth along the POPC membrane's center, and the time between the region II and region III was roughly equal. The results showed that the electrostatic potential of triangulene would have a significant effect on its spatial distribution in the POPC membrane. Fig. 3 The spatial distribution of triangulene in the POPC membrane. (a) ESP triangulene. (b) Inside-0 triangulene. (c) All-0 triangulene. 3.3. The potential of mean force (PMF) of triangulene translocating into POPC To further explain the reasons above spatial distribution occurs, the mean force potential (PMF) from 5 nm to −2 nm of triangulene was calculated (Fig. 4). The ΔW1, ΔW2 and ΔW3 were defined as the PMF of ESP triangulene, inside-0 triangulene and all-0 triangulene in the midline of the membrane minus the lowest in the region II. The ESP triangulene and all-0 triangulene needed to overcome 4.84 kJ mol−1 (ΔW1), 4.51 kJ mol−1 (ΔW3) to reach the lower leaflet. Inside-0 triangulene needed to overcome 7.15 kJ mol−1 (ΔW2), which explains why the inside-0 triangulene is only stayed in the upper leaflet of the membrane. However, ESP triangulene and all-0 triangulene can reach the lower leaflet of the membrane. From 5 nm to −2 nm, the energy at 0.8 nm is the lowest, so triangulene finally stay at 0.8 nm. Fig. 4 The potential of mean force (PMF) of triangulene translocating into POPC along the z-axis direction. 4. Conclusion In this study, the effect of electrostatic potential on the interaction between triangulene and cell membrane were described by setting three kinds of triangulene models with different electrostatic potential distributions. We can conclude that: (1) the angle of ESP triangulene penetrated POPC membrane is mainly 90°, and can cross the midline of the membrane to the opposite side. The angle distribution of inside-0 triangulene penetrated is more extensive than that of ESP. The energy barrier is lower when penetrating the membrane from 45° to 90°, which makes it easier to penetrate the membrane, but it is difficult to cross the central line of the cell membrane. The angle distribution of all-0 triangulene penetrate membrane is more discrete, the energy barrier is lower at 70°, 90°, 110° and 120°, easy to penetrate into the membrane, and can also cross the cell membrane midline to reach the lower leaflet of the POPC membrane; (2) through PMF calculation, it is found that ESP, inside-0 and all-0 triangulene respectively need to overcome 4.84 kJ mol−1 (ΔW1), 7.15 kJ mol−1 (ΔW2) and 4.51 kJ mol−1 (ΔW3). Therefore, inside-0 triangulene can hardly reach the lower leaflet due to the high energy barrier; (3) among these three systems, the energy barrier of the triangulene system is the lowest at 0.8 nm or −0.8 nm, that is the reason they finally stays at there. Our research suggests the feasibility of regulating the interaction between triangulene and cell membrane through electrostatic potential, which would be helpful for further application of triangulene in biomedicine. Abbreviations MD Classical molecular dynamics POPC 1-Palmitoyl-2-oleoylphosphatidylcholine DFT Density functional theory EP Electrostatic potential APBS Adaptive Poisson–Boltzmann solver Author contributions The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Conflicts of interest The authors declare no competing financial interest. Supplementary Material RA-013-D3RA03259K-s001 We acknowledge financial support by Science, Technology Department of Yunnan Province (202101AY070001-164). ==== Refs References Chen X. Zhang W. Chem. Soc. Rev. 2017 46 734 760 27942638 Wang H. Chen Q. Zhou S. Chem. Soc. Rev. 2018 47 4198 4232 29667656 Kang H. Gravier J. Bao K. Wada H. Lee J. H. Baek Y. El Fakhri G. Gioux S. Rubin B. P. Coll J. L. Choi H. S. Adv. Mater. 2016 28 8162 8168 27414255 Xiong H. M. Adv. Mater. 2013 25 5329 5335 24089351 Wang X. Wang X. Bai X. Yan L. Liu T. Wang M. Song Y. Hu G. Gu Z. Miao Q. Chen C. Nano Lett. 2019 19 8 18 30335394 Fontana F. Lindstedt H. Correia A. Chiaro J. Kari O. K. Ndika J. Alenius H. Buck J. Sieber S. Makila E. Salonen J. Urtti A. Cerullo V. Hirvonen J. T. Santos H. A. Adv. Healthcare Mater. 2020 9 2000529 Zou D. Wu Z. Yi X. Hui Y. Yang G. Liu Y. Tengjisi Wang H. Brooks A. Wang H. Liu X. Xu Z. P. Roberts M. S. Gao H. Zhao C. X. Proc. Natl. Acad. Sci. U. S. A. 2023 120 e2214757120 36574680 Zhang S. Liu L. Duan G. Zhao L. Liu S. Zhou B. Yang Z. ACS Appl. Mater. Interfaces 2019 11 34575 34585 31469275 Tang X. Zhang S. Zhou H. Zhou B. Liu S. Yang Z. Nanoscale 2020 12 2732 2739 31951244 Chen P. Huang Z. Liang J. Cui T. Zhang X. Miao B. Yan L. T. ACS Nano 2016 10 11541 11547 27936576 Pagnout C. Jomini S. Dadhwal M. Caillet C. Thomas F. Bauda P. Colloids Surf., B 2012 92 315 321 Arikawa S. Shimizu A. Shiomi D. Sato K. Shintani R. J. Am. Chem. Soc. 2021 143 19599 19605 34767718 Leung E. Pilkington L. I. Naiya M. M. Barker D. Zafar A. Eurtivong C. Reynisson J. MedChemComm 2019 10 1881 1891 32952994 Zhang N. Feng W. Wen H. Feng N. Sheng H. Huang Z. Wang J. Molecules 2023 28 3744 37175153 Yu H. Sun J. Heine T. J. Chem. Theory Comput. 2023 19 3486 3497 37263582 Shen T. Dijkstra D. Farrando-Perez A. Boj P. G. Villalvilla J. M. Quintana J. A. Zou Y. Hou X. Wei H. Li Z. Sun Z. Diaz-Garcia M. A. Wu J. Angew. Chem., Int. Ed. 2023 62 e202304197 Arikawa S. Shimizu A. Shiomi D. Sato K. Takui T. Sotome H. Miyasaka H. Murai M. Yamaguchi S. Shintani R. Angew. Chem., Int. Ed. 2023 62 e202302714 Lostao A. Lim K. Pallares M. C. Ptak A. Marcuello C. Int. J. Biol. Macromol. 2023 238 124089 36948336 Escorihuela E. Concellón A. Marín I. Kumar V. J. Herrer L. Moggach S. A. Vezzoli A. Nichols R. J. Low P. J. Cea P. Serrano J. L. Martín S. Mater. Today Chem. 2022 26 101067 Reynisson J. Schuster G. B. Howerton S. B. Williams L. D. Barnett R. N. Cleveland C. L. Landman U. Harrit N. Chaires J. B. J. Am. Chem. Soc. 2003 125 2072 2083 12590535 Pothukuchy A. Ellapan S. Gopidas K. R. Salazar M. Bioorg. Med. Chem. Lett. 2003 13 1491 1494 12668019 Kotar A. Wang B. Shivalingam A. Gonzalez-Garcia J. Vilar R. Plavec J. Angew. Chem., Int. Ed. 2016 55 12508 12511 Yu H. Sun J. Heine T. J. Chem. Theory Comput. 2023 19 3486 3497 37263582 Bosson J. Gouin J. Lacour J. Chem. Soc. Rev. 2014 43 2824 2840 24500211 Bauer C. Duwald R. Labrador G. M. Pascal S. Lorente P. M. Bosson J. Lacour J. Rochaix J. D. Org. Biomol. Chem. 2018 16 919 923 29334105 Mijajlovic M. Wright D. Zivkovic V. Bi J. X. Biggs M. J. Colloids Surf., B 2013 104 276 281 Liang L. Kong Z. Kang Z. Wang H. Zhang L. Shen J. W. ACS Biomater. Sci. Eng. 2016 2 1983 1991 33440534 Cohen-Tanugi D. Grossman J. C. Nano Lett. 2012 12 3602 3608 22668008 Gapurenko O. A. Starikov A. G. Minyaev R. M. Minkin V. I. J. Comput. Chem. 2015 36 2193 2199 26393417 Baker N. A. Sept D. Joseph S. Holst M. J. McCammon J. A. Proc. Natl. Acad. Sci. U. S. A. 2001 98 10037 10041 11517324 Dolinsky T. J. Czodrowski P. Li H. Nielsen J. E. Jensen J. H. Klebe G. Baker N. A. Nucleic Acids Res. 2007 35 W522 W525 17488841 Mamidi A. S. Ray A. Surolia N. Front. Bioeng. Biotechnol. 2019 7 240 31608276 Jorgensen W. L. Chandrasekhar J. Madura J. D. Impey R. W. Klein M. L. J. Chem. Phys. 1983 79 926 935 Huang J. Mackerell A. D. J. Comput. Chem. 2013 34 2135 2145 23832629 Bussi G. Donadio D. Parrinello M. J. Chem. Phys. 2007 126 014101 17212484 Parrinello M. Rahman A. J. Appl. Phys. 1998 52 7182 7190 Darden T. York D. Pedersen L. J. Chem. Phys. 1993 98 10089 10092 Miyamoto S. Kollman P. A. J. Comput. Chem. 2010 13 952 962 Hess B. Bekker H. Fraaije J. G. E. M. J. Comput. Chem. 1997 18 1463 1472 Torrie G. M. Valleau J. P. J. Chem. Phys. 1977 66 1402 1408 Hub J. S. J. Chem. Phys. 2015 6 3713 3720
PMC010xxxxxx/PMC10352717.txt
==== Front Virus Res Virus Res Virus Research 0168-1702 1872-7492 Elsevier S0168-1702(23)00101-6 10.1016/j.virusres.2023.199139 199139 Article Porcine reproductive and respiratory syndrome virus regulates lipid droplet accumulation in lipid metabolic pathways to promote viral replication Yang Yunlong ab Luo Yizhuo ab Yi Songqiang c Gao Qi ab Gong Ting ad Feng Yongzhi a Wu Dongdong ad Zheng Xiaoyu ad Wang Heng ab Zhang Guihong abd Sun Yankuo [email protected] abd⁎ a Guangdong Provincial Key Laboratory of Zoonosis Prevention and Control, College of Veterinary Medicine, South China Agricultural University, Guangzhou 510642, PR China b Maoming Branch, Guangdong Laboratory for Lingnan Modern Agriculture, Maoming 525000, China c Agricultural Technology Extension Center of Jiangxi Province, Nanchang, China d Key Laboratory of Animal Vaccine Development, Ministry of Agriculture and Rural Affairs, PR China ⁎ Corresponding author at: Guangdong Provincial Key Laboratory of Zoonosis Prevention and Control, College of Veterinary Medicine, South China Agricultural University, Guangzhou 510642, PR China. [email protected] 07 6 2023 8 2023 07 6 2023 333 19913911 2 2023 16 5 2023 19 5 2023 © 2023 The Authors. Published by Elsevier B.V. 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Highlights • PRRSV infection promotes LD accumulation. • DGAT1 inhibitor, BAY11–7082, and MH reduce PRRSV-induced LD accumulation. • Reducing PRRSV-induced LD accumulation reduces NF-κB signaling. • Reducing LDs and NF-κB signaling pathways reduces IL-1β and IL-8 up-regulation by PRRSV. • Reducing LDs and NF-κB signaling pathway reduces PRRSV replication. Porcine reproductive and respiratory syndrome (PRRS) is a severe respiratory disease caused by porcine reproductive and respiratory syndrome virus (PRRSV) that can lead to the abortion of pregnant sows and decreased boar semen quality. However, the mechanisms of PRRSV replication in the host have not yet been fully elucidated. As lipid metabolism and lipid droplets (LDs) have been reported to play important roles in the replication of various viruses, we aimed to explore the mechanisms through which LDs affect PRRSV replication. Laser confocal and transmission electron microscopy revealed that PRRSV infection promoted intracellular LD accumulation, which was significantly reduced by treatment with the NF-κB signaling pathway inhibitors BAY11–7082 and metformin hydrochloride (MH). In addition, treatment with a DGAT1 inhibitor significantly reduced the protein expression of Phosphorylated NF-ΚB P65and PIκB and the transcription of IL-1β and IL-8 in the NF-κB signaling pathway. Furthermore, we showed that the reduction of the NF-κB signaling pathway and LDs significantly reduced PRRSV replication. Together, the findings of this study suggest a novel mechanism through which PRRSV regulates the NF-κB signaling pathway to increase LD accumulation and promote viral replication. Moreover, we demonstrated that both BAY11–7082 and MH can reduce PRRSV replication by reducing the NF-κB signaling pathway and LD accumulation. This study lays a theoretical foundation for research on the mechanism of PRRS prevention and control, as well as the research and development of antiviral drugs. Keywords PRRSV Nf-κb signaling pathway Lipid droplet Lipid metabolism ==== Body pmc1 Introduction Porcine reproductive and respiratory syndrome (PRRS) is a highly prevalent disease that has affected the global pig industry for decades (Zhang et al., 2020). This infectious disease is caused by porcine reproductive and respiratory syndrome virus (PRRSV) and can cause severe respiratory disease in infected newborn piglets and young pigs, the abortion of pregnant sows, and the deterioration of boar semen quality (Han et al., 2017). PRRS was first identified in the United States in the 1880s before becoming endemic in North America and Europe, and gradually spreading to Asia (Zhang et al., 2018). PRRSV, of the genus Arterivirus in the family Arteriviridae (Kuhn et al., 2016), is a single-stranded enveloped RNA virus (Lunney et al., 2016) with a genome of approximately 15.4 kb in size including 12 open reading frames (Chand et al., 2012). Based on its genetic characteristics, PRRSV can be divided into two genotypes: type 1 (PRRSV-1) and type 2 (PRRSV-2) (Murtaugh et al., 2010). Unfortunately, commercial vaccines against PRRSV currently provide limited protection and there are no effective therapeutic drugs; therefore, it is important to study the pathogenic mechanism of PRRSV in order to develop novel strategies to prevent and control the disease. Lipid droplets (LDs) are dynamic organelles that are mainly composed of triglycerides and cholesteryl esters (Preuss et al., 2019), and are involved in a variety of biological processes, including lipid storage and transport, metabolism, and protein storage and degradation (Bartz et al., 2007; Li et al., 2012; Xu et al., 2018). After LDs have been produced by the endoplasmic reticulum, their size and cellular position change constantly within a small range (E.A. Monson et al., 2021). Many recent studies have demonstrated that pathogens can induce LD accumulation after infecting host cells and it has been shown that host LDs can also regulate the life cycle of viruses (Roingeard and Melo, 2017). For example, the core protein of hepatitis C virus (HCV) interacts with diacylglycerol acyltransferase (DGAT1) to promote LD accumulation in the perinuclear region, while the inhibition of DGAT1 activity can reduce the production of infectious virions (Herker et al., 2010). LD formation in host cells infected with rotavirus is very important for viral replication, and blocking LD accumulation can significantly reduce the number of progenies produced by rotavirus proliferation (Crawford and Desselberger, 2016). In addition, recent studies have shown that LDs can be used as a platform to recruit viral proteins, accelerate viral assembly, and increase viral replication (Zhao et al., 2022). Although PRRSV infection has been shown to increase the number of LDs in African green monkey kidney epithelial cells (Marc-145) and porcine alveolar macrophages (PAMs) (Yu et al., 2022), the molecular mechanisms through which LDs regulate PRRSV replication and PRRSV regulates LDs accumulation remain unclear. Metformin hydrochloride (MH) is a broad-spectrum biguanide drug that is used to treat type 2 diabetes mellitus and is a potential anticancer drug that has become a research hotspot (Flory and Lipska, 2019). MH mainly reduces hepatic gluconeogenesis and improves cellular glucose uptake (Kim et al., 2020), and can reduce the expression of adipogenic genes and LD accumulation and activate Adenosine 5′-monophosphate (AMP)-activated protein kinase, thereby blocking the differentiation of human preadipose tissue (Moreno-Navarrete et al., 2011). In addition, studies have shown that MH can inhibit the NF-κB signaling pathway (Zhang et al., 2021). However, the effect of MH regulation of the NF-κB signaling pathway on intracellular LD accumulation following PRRSV infection remains unclear. In this study, we investigated the regulatory relationships between PRRSV infection, LD accumulation, and the NF-κB signaling pathway, and demonstrate a mechanism of mutual regulation that provides new avenues for the theoretical study of anti-PRRSV therapies. 2 Materials and methods 2.1 Cell culture and viral strain African green monkey kidney epithelial cells (Marc-145) were treated with 10% fetal bovine serum (FBS; Gibco, Grand Island, NY, USA), 100 U/mL penicillin, and 100 μg/mL streptomycin sulfate in Dulbecco's modified Eagle medium (DMEM; Gibco, Grand Island, NY, USA) and grown at 37 °C with 5% CO2. Porcine alveolar macrophages (PAMs) were treated with 10% FBS, 100 U/mL penicillin, 50 μg/mL streptomycin, and 0.25 μg/mL amphotericin in RPMI 1640 medium (Gibco, Waltham, MA, USA) and grown at 37 °C with 5% CO2. The highly pathogenic PRRSV-JXA1 strain (GenBank accession No. EF112445.1; lineage 8.7) generously provided by Professor Tian Kegong (Tian et al., 2007) was propagated in Marc-145 cells. Viral titers were determined in Marc-145 cells and calculated using the Reed-Muench method (Yang et al., 2021). 2.2 Reagents and antibodies Reagents used in this study are as follows: Oil Red O (Sigma, SL, MO, USA), 4′, 6-diamino-2-phenylindole (DAPI; Beyotime, China, C1006), anti-fluorescence quencher (Beyotime, Shanghai, China), Radio Immunoprecipitation Assay Lusis Buffer (RIPA Lusis buffer; Beyotime, Shanghai, China), protease phosphatase inhibitor mixture (Beyotime, Shanghai, China), BCA protein concentration assay kit (Beyotime, Shanghai, China), SDS-PAGE protein loading buffer (5 ×) (Beyotime, Shanghai, China), NF-κB agonist Betulinic aid (BetA; MCE, Monmouth Junction, NJ, USA), NF-κB inhibitor BAY11–7082 (MCE, Monmouth Junction, NJ, USA), MH (MCE, Monmouth Junction, NJ, USA), LD inhibitor DGAT1 inhibitor (Sigma, SL, MO, USA), Cell counting kit-8 (CCK-8; NCM, Suzhou, China), RNA Fastagen Kit (Fastagen, Shanghai, China), HiScript II 1st strand cDNA synthesis kit (+ gDNAwiper; Vazyme, Nanjing, China), ChamQ universal SYBR qPCR master mix (Vazyme, Nanjing, China), and AceQ Universal U+Probe Master Mix V2 (Vazyme, Nanjing, China). Antibodies used in this study are as follows: NF-κB p65 (Cell Signaling Technology, Danvers, MA, USA, 1:1000), IκB (Cell Signaling Technology, Danvers, MA, USA, 1:1000), MyD88 (Proteintech, Chicago, IL, USA, 1:1000), Phosphorylated NF-κB p65 (Cell Signaling Technology, Danvers, MA, USA, 1:1000), pIκB (Cell Signaling Technology, Danvers, MA, USA, 1:1000), IL-1β(Cell Signaling Technology, Danvers, MA, USA, 1:1000), IL-8 (Cell Signaling Technology, Danvers, MA, USA, 1:1000), and β-tubulin (Abmart, Shanghai, China, 1:5000), Mouse monoclonal antibodies against the PRRSV N protein were purchased from MEDIAN Diagnostics (Korea, 1:1000), IRDye® 800CW goat anti-rabbit IgG antibodies (LI-COR Biosciences, Lincoln, NE, USA, 1:10,000), IRDye® 800CW goat anti-mouse IgG antibodies (highly cross-sorbent) (LI-COR Biosciences, Lincoln, NE, USA, 1:10,000), and CoraLite488 goat anti-mouse IgG antibodies (Proteintech, Chicago, IL, USA, 1:300). 2.3 Immunofluorescence assay Marc-145 cells were grown in a 24-well plate under a cover glass. When the cells reached 70–80% confluence, they were inoculated with 1 MOI PRRSV and then incubated with the DGAT1 inhibitor, BAY11–7082, and MH at 37 °C with 5% CO2 for 72 h. After the medium had been discarded, the cells were fixed with 4% paraformaldehyde for 30 min at 25 °C and cell membranes were permeabilized with 0.1% Triton X-100 for 20 min. Next, the cells were blocked with 5% bovine serum albumin (BSA) at 37 °C for 1 h, washed with PBS three times, and incubated with monoclonal antibodies against PRRSV N protein for 1 h at 37 °C. After three washes with PBS, the cells were incubated with CoraLite488 goat anti-mouse IgG secondary antibodies diluted with 2% BSA for 1 h at 37 °C in the dark. Nuclei were stained with DAPI for 5 min at room temperature. The cells were washed three times with PBS, treated with an anti-fluorescence quenching agent, and observed using a laser scanning confocal microscope (Olympus, Tokyo, Japan) or their fluorescence was observed using an inverted fluorescence microscope (Nikon, Japan). 2.4 Lipid droplet staining Marc-145 cells were grown in a 24-well plate under a cover glass. When the cells reached 70–80% confluence, they were inoculated with 1 MOI PRRSV and incubated with the DGAT1 inhibitor, BAY11–7082, and MH at 37 °C with 5% CO2 for 72 h. After the medium had been discarded, the cells were fixed with 4% paraformaldehyde for 30 min at room temperature and cell membranes were permeabilized with 0.1% Triton X-100 for 20 min. Next, the cells were blocked with 5% BSA at 37 °C for 1 h, washed with PBS three times, and incubated with monoclonal antibodies against PRRSV N protein for 1 h at 37 °C. After three washes with PBS, the cells were incubated with CoraLite488 goat anti-mouse IgG secondary antibodies diluted with 2% BSA for 1 h at 37 °C in the dark and stained with 0.3% oil red O (diluted in 60% isopropanol) for 15 min at room temperature. Nuclei were stained with DAPI for 5 min at room temperature. The cells were washed three times with PBS, treated with an anti-fluorescence quenching agent, and observed using a laser scanning confocal microscope (Olympus, Tokyo, Japan). 2.5 Transmission electron microscopy (TEM) TEM was performed on PRRSV-infected and control group Marc-145 cells 48 h after PRRSV infection, as described previously (Dias et al., 2020). Cell monolayers were digested with trypsin, fixed, and embedded in paraffin. Cell morphology was observed using a Talos L120C TEM (FEI, Czech Republic). 2.6 Cell viability analysis Cell viability was determined using a CCK-8 assay kit. Cells were seeded into 96-well plates at 1 × 104 cells per well and treated with the DGAT1 inhibitor (10, 25, 50, 100, and 200 μM), BAY11–7082 (2, 5, 10, 20, and 40 μM), and MH (20, 50, 80, 100, and 200 μM) when the cells reached 70–80% confluence. After 48 h, 10 μL of CCK-8 reagent was added to each well and incubated for 1 h at 37 °C. Absorbance was measured at OD450 nm using a microplate reader (BioTek, Vermont, USA). Relative cell viability was calculated as follows: cell survival (%) = [OD (drug) − OD (blank)/OD (control) − OD (blank)] × 100%. 2.7 Western blot analysis Cells from the PRRSV infection, drug treatment, and control groups were lysed using RIPA buffer supplemented with 1% protease inhibitors, phosphatase inhibitors, and EDTA for 30 min on ice. After centrifugation at 4 °C for 30 min, the supernatant was collected and total protein concentration was determined using a BCA protein concentration assay kit. The protein samples were denatured by boiling at 100 °C for 15 min after the addition of SDS-PAGE buffer containing DL-dithiothreitol. Proteins were separated using SDS-PAGE and transferred onto nitrocellulose membranes using a Trans-Blot Turbo fast transfer system (Bio-Rad, Hercules, CA, USA). The membranes were blocked with 5% non-fat milk for 1 h at 37 °C, incubated with primary antibodies for 1 h at 37 °C, and then washed three times (5 min each) with washing buffer (TBS containing 0.1% Tween20). After incubation with IRDye® 800CW goat anti-mouse IgG secondary antibodies for 1 h at 37 °C, the membranes were washed three times with TBST wash buffer and results were analyzed using an infrared imaging system (Azure Biosystems, Dublin, CA, USA). 2.8 Real-time quantitative PCR (RT-qPCR) Total RNA was extracted from cells in each treatment group using an RNA Fastagen Kit and reverse-transcribed into cDNA using a HiScript II 1st Strand cDNA Synthesis Kit (+ gDNAwiper). RT-qPCR was performed using ChamQ Universal SYBR qPCR master mix kit and specific primers for glycerol 3-phosphate dehydrogenase (GAPDH), IL-1β, and IL-8 in a CFX96™ Real-time System PCR instrument (Bio-Rad, CA, USA), with GAPDH as the reference gene. The relative mRNA expression levels of IL-1β and IL-8 were calculated using the 2-△△CT method. The CT value of the PRRSV-JXA1 Nsp9 gene was detected using qRT-PCR with AceQ Universal U+Probe master mix V2 and JXA1 Nsp9 specific primers and probes. All experiments were performed in triplicate. The sequences of the gene-specific primers and probes are listed in Table 1.Table 1 RT-qPCR primer and probe sequences. Table 1Gene Primer sequence (5 '−3′) IL-1β-F GGAAGACAAATTGCATGG IL-1β-R CCCAACTGGTACATCAGCAC IL-8-F AGGACAAGAGCCAGGAAG IL-8-R CTGCACCTTCACACAGAGC Nsp9-F CCTGCAATTGTCCGCTGGTTTG Nsp9-R GACGACAGGCCACCTCTCTTAG JXA1-Nsp9-Probe ACTGCTGCCACGACTTACTGGTCACGCAGT GAPDH-F CCTTCCGTGTCCCTACTGCCAAC GAPDH-R GACGCCTGCTTCACCACCTTCT 2.9 Statistical analysis All statistical analyses were performed using GraphPad Prism 8.0.1 (GraphPad Software, San Diego, CA, USA). Differences between groups were determined using one-way ANOVA. All data were expressed as the mean ± standard deviation (SD) of at least three independent replicates. Two-tailed t-tests were used to analyze significant differences between groups. Statistical significance was defined as: *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. 3 Results 3.1 PRRSV infection promotes LD accumulation To investigate the effect of PRRSV infection on LD accumulation, Marc-145 cells were infected with 1 MOI of PRRSV and LD accumulation was observed using laser confocal microscopy after 48 h. LD accumulated in the PRRSV-infected group as compared to the control group (Fig. 1A). It has been reported that PAMs are the main target cells of PRRSV infection in vivo (Lu et al., 2020). Therefore, we performed TEM on PRRSV-infected and uninfected PAMs to investigate the effect of PRRSV on LD accumulation. As expected, PRRSV infection promoted LD accumulation in PAMs (Fig. 1B, C). Together, these results indicate that PRRSV infection can promote LD accumulation.Fig. 1 PRRSV infection promotes LD accumulation. (A) LD observation using a laser confocal microscope. After infection with 1 MOI PRRSV for 48 h, Marc-145 cells were incubated with oil red dye, treated with monoclonal antibodies against PRRSV N protein and DAPI, and subjected to IFA assays. Quantification of the fluorescence intensity of lipid droplets from 10 different confocal fields is shown on the right. ****, P < 0.0001. (B) TEM observation of PAMs without PRRSV infection. (C) TEM observation of LDs in PAMs infected with 1 MOI PRRSV for 48 h. Lipid droplets are indicated using white arrows. Fig 1 3.2 DGAT1 inhibitor, BAY11–7082, and MH reduce PRRSV-induced LD accumulation Next, we investigated the potential cytotoxicity of various concentrations of the NF-κB agonist BetA, LD inhibitor targeting DGAT1, NF-κB inhibitor BAY11–7082, and MH on Marc-145 cell activity using CCK-8 assays (Fig. 2A–D).Compared to the control group, Marc-145 cell viability decreased in a dose-dependent manner after 48 h of incubation with different concentrations of the DGAT1 inhibitor, BAY11–7082, and MH. Consequently, 40 μM BetA, 10 μM DGAT1 inhibitor, 10 μM BAY11–7082, and 80 μM MH were used in subsequent experiments based on their maximum safe concentrations in Marc-145 cells. Compared to the control group, 40 μM BetA can activate the NF-κB pathway, and 10 μM DGAT1 inhibitor, 10 μM BAY11–7082, and 80 μM MH can inhibit the NF-κB pathway (Fig. 2E). When we further explored the effects of the BetA, DGAT1 inhibitor, BAY11–7082, and MH on LD accumulation in PRRSV-infected cells, we found that treatment with 40 μM BetA increased LD accumulation to the level of and PRRSV infection (Fig. 2F). In addition, 10 μM DGAT1 inhibitor reduced,10 μM BAY11–7082 and 80 μM MH reduced LD accumulation induced by PRRSV 72 h after infection (Fig. 2F). This suggests that PRRSV may promote the accumulation of lipid droplets by activating the NF-κB pathway.Fig. 2 DGAT1 inhibitor, BAY11–7082 and MH inhibit LD accumulation induced by PRRSV infection. (A) Cytotoxicity of BetA (10, 20, 30, 40 and 50 μm) in Marc-145 cells detected using CCK-8 assays after 48 h. (B) Cytotoxicity of DGAT1 inhibitor (10, 25, 50, 100 and 200 μm) in Marc-145 cells detected using CCK-8 assays after 48 h. (C) Cytotoxicity of BAY11–7082 (2, 5, 10, 20 and 40 μm) in Marc-145 cells detected using CCK-8 assays after 48 h. (D) Cytotoxicity of MH (20, 50, 80, 100 and 200 μm) in Marc-145 cells detected using CCK-8 assays after 48 h. (E) NF-κB p65, IκB, Phosphorylated NF-κB p65, and pIκB protein expression in Marc-145 cells treated with the BetA, and Marc-145 cells infected with 1 MOI PRRSV were treated with 10 μM DGAT1 inhibitor, 10 μM BAY11–7082 and 80 μM MH detected using western blotting after 0, 12, 24, 48, 60, and 72 h. (F) Observation of LDs using a laser confocal microscope. Marc-145 cells were treated with 40 μM BetA for 72 h, and Marc-145 cells infected with 1 MOI PRRSV were treated with 10 μM DGAT1 inhibitor, 10 μM BAY11–7082 and 80 μM MH for 72 h, respectively. Quantification of the fluorescence intensity of lipid droplets from 10 different confocal fields is shown on the right. ****, P < 0.0001. ns, no significance. Fig 2 3.3 Reducing PRRSV-induced LD accumulation reduces NF-κB signaling To further explore the relationship between LDs and the NF-κB signaling pathway, PRRSV-infected cells were treated with 10 μM DGAT1 inhibitor and the protein expression levels of NF-Κb, NF-κB p65, IκB, MyD88, Phosphorylated NF-κB p65, and pIκB were measured. Phosphorylated NF-κB p65 and pIκB protein expression were significantly increased after PRRSV infection compared to the control group but were significantly reduced by treatment with the DGAT1 inhibitor (Fig. 3A–C). NF-Κb p65, IκB, MyD88,Phosphorylated NF-κB p65, and pIκB protein expression was also detected in PRRSV-infected cells after treatment with 10 μM BAY11–7082 and 80 μM MH. Interestingly, Phosphorylated NF-κB p65 and pIκB protein levels were significantly reduced in cells treated with BAY11–7082 and MH compared to the PRRSV infection group (Fig. 3D, E). Together, these results indicate that reducing LD accumulation can reduce the activation of the NF-κB signaling pathway in PRRSV-infected cells.Fig. 3 Reducing PRRSV-induced LD accumulation reducing NF-κB signaling pathway. (A) NF-κB p65, IκB, MyD88, Phosphorylated NF-κB p65, and pIκB protein expression in Control (no PRRSV-infected) Marc-145 cells detected using western blotting 0, 12, 24, 48, 60, and 72 h. (B) NF-κB p65, IκB, MyD88, Phosphorylated NF-κB p65, and pIκB protein expression in PRRSV-infected Marc-145 cells detected using western blotting 0, 12, 24, 48, 60, and 72 h after PRRSV infection. (C) NF-κB p65, IκB, MyD88, Phosphorylated NF-κB p65, and pIκB protein expression in PRRSV-infected Marc-145 cells treated with the DGAT1 inhibitor detected using western blotting 0, 12, 24, 48, 60, and 72 h after PRRSV infection. (D) NF-κB p65, IκB, MyD88, Phosphorylated NF-κB p65, and pIκB protein expression in PRRSV-infected Marc-145 cells treated with the BAY11–7082 detected using western blotting 0, 12, 24, 48, 60, and 72 h after PRRSV infection. (E) NF-κB p65, IκB, MyD88, Phosphorylated NF-κB p65, and pIκB protein expression in PRRSV-infected Marc-145 cells treated with the MH detected using western blotting 0, 12, 24, 48, 60, and 72 h after PRRSV infection. Tubulin expression was used as a positive control. Fig 3 3.4 Reducing LDs and NF-κB signaling pathways reduces IL-1β and IL-8 up-regulation by PRRSV To determine the effect of reducing LD accumulation or downregulating the NF-κB signaling pathway on the transcription of inflammatory cytokines induced by PRRSV infection, Marc-145 cells and PAMs infected with PRRSV were treated with the DGAT1 inhibitor, BAY11–7082, and MH, and the transcriptional levels of IL-1β and IL-8 were detected. PRRSV infection significantly upregulated the transcription of IL-1β and IL-8 in Marc-145 cells and PAMs, while the DGAT1 inhibitor, BAY11–7082, and MH significantly decreased their transcription compared to PRRSV-infected cells (Fig. 4A–D). The protein expression levels of IL-1β and IL-8 in Marc-145 cells (Fig. 4E) and PAMs (Fig. 4F) demonstrated the same results. These findings indicate that reducing LD accumulation and NF-κB signaling significantly reduces the transcription of IL-1β and IL-8 upregulated by PRRSV infection.Fig. 4 Reducing LDs and NF-κB signaling pathways, reduces IL-1β and IL-8 up-regulation by PRRSV infection. (A) RT-qPCR analysis of IL-1β mRNA levels 0, 12, 24, 48, 60, and 72 h after PRRSV infection in Marc-145 cells treated with and without the DGAT1 inhibitor, BAY11–7082, and MH. Data represent the mean ± SD of three independent experiments. **p < 0.01, ***p < 0.001, and ****p < 0.0001. (B) RT-qPCR analysis of IL-8 mRNA levels 0, 12, 24, 48, 60, and 72 h after PRRSV infection in Marc-145 cells treated with and without the DGAT1 inhibitor, BAY11–7082, and MH. Data represent the mean ± SD of three independent experiments. **p < 0.01, ***p < 0.001, and ****p < 0.0001. (C) RT-qPCR analysis of IL-1β mRNA levels 0, 12, 24, 48, 60, and 72 h after PRRSV infection in PAMs treated with and without the DGAT1 inhibitor, BAY11–7082, and MH. Data represent the mean ± SD of three independent experiments. **p < 0.01, ***p < 0.001, and ****p < 0.0001. (D) RT-qPCR analysis of IL-8 mRNA levels 0, 12, 24, 48, 60, and 72 h after PRRSV infection in PAMs treated with and without the DGAT1 inhibitor, BAY11–7082, and MH. Data represent the mean ± SD of three independent experiments. **p < 0.01, ***p < 0.001, and ****p < 0.0001. (E) Western blot analysis of IL-1β and IL-8 protein expression 0, 12, 24, 48, 60, and 72 h after PRRSV infection in Marc-145 cells treated with and without the DGAT1 inhibitor, BAY11–7082, and MH. (F) Western blot analysis of IL-1β and IL-8 protein expression 0, 12, 24, 48, 60, and 72 h after PRRSV infection in Marc-145 cells treated with and without the DGAT1 inhibitor, BAY11–7082, and MH. Fig 4 3.5 Reducing LDs and NF-κB signaling pathway reduces PRRSV replication Finally, we explored the roles of LDs and the NF-κB signaling pathway in PRRSV replication, as well as the effects of the DGAT1 inhibitor, BAY11–7082, and MH using RT-qPCR, western blotting, IFA and TCID50. RT-qPCR showed that the DGAT1 inhibitor-, BAY11–7082-, and MH-treated groups displayed significantly lower PRRSV N mRNA levels 48, 60, and 72 h after viral infection (Fig. 5A). Similarly, western blot assays revealed that the DGAT1 inhibitor, BAY11–7082, and MH significantly reduced PRRSV N protein expression 48, 60, and 72 h after PRRSV infection (Fig. 5B). IFA showed that the DGAT1 inhibitor, BAY11–7082, and MH significantly reduced PRRSV N protein fluorescence 72 h after viral infection (Fig. 5C). Furthermore, the DGAT1 inhibitor, BAY11–7082, and MH significantly reduced the virus titer of PRRSV. Taken together, these results indicate that reducing the induction of LD accumulation and the NF-κB signaling pathway by PRRSV infection could significantly reduce PRRSV replication.Fig. 5 Reducing LD accumulation and NF-κB signaling reduces PRRSV replication. (A) N gene transcription levels in PRRSV-infected Marc-145 cells treated with and without the DGAT1 inhibitor, BAY11–7082, and MH detected using RT-qPCR after 0, 12, 24, 48, 60, and 72 h. Data represent the mean ± SD of three independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. (B) PRRSV N protein expression in negative control and PRRSV-infected Marc-145 cells treated with and without the DGAT1 inhibitor, BAY11–7082, and MH detected using western blotting after 0, 12, 24, 48, 60, and 72 h. Tubulin expression was used as a positive control. (C) IFA of PRRSV N protein fluorescence in negative control and PRRSV-infected Marc-145 cells treated with and without the DGAT1 inhibitor, BAY11–7082, and MH after 72 h. Quantification of the fluorescence intensity of N protein from 10 different confocal fields is shown on the right. ****, P < 0.0001. (D) Marc-145 cells were incubated with PRRSV strains (MOI of 1 TCID50 per cell) treated with and without the DGAT1 inhibitor, BAY11–7082, and MH. Supernatants were collected 12, 24, 48, 60, 72 h after inoculation for virus titer determination. Fig 5 4 Discussion LDs are ubiquitous in animals, plants, bacteria, and fungi (Walther and Farese, 2012), and are the main site of neutral lipid storage in cells (Zhang and Liu, 2017). LDs participate in lipid metabolism and homeostasis, and their cores can promote energy metabolism and membrane synthesis during nutrient deficiency or growth through lipolysis or autophagy (Olzmann and Carvalho, 2019). It has been reported that viruses can hijack cellular lipid metabolism to produce LDs with phospholipid layers and surface proteins that can promote virus assembly and budding (Villareal et al., 2015). For example, dengue virus (DENV) infection can promote the accumulation of LDs and thus the encapsulation of the viral genome, and can increase the number of LDs in cells, with LD inhibition significantly reducing DENV replication (Samsa et al., 2009). Similarly, the capsid protein of Japanese encephalitis virus (JEV) can colocalize with LDs to promote the release of mature viral particles (Ishida et al., 2019). DGAT1 and DGAT2 enzymes are essential for catalyzing triglyceride biosynthesis during LD biogenesis (Yen et al., 2008); therefore, inhibiting DGAT1 activity or knocking down DGAT using RNAi have been shown to significantly inhibit the production of infectious HCV virions (Herker et al., 2010). It has also been demonstrated that LD induction corresponds to enhanced type I and -III IFN production in infected cells, and that enhanced LD accumulation reduces herpes simplex virus 1 (HSV-1) and Zika virus (ZIKV) viral replication (EA Monson et al., 2021). In this study, we found that PRRSV infection promoted LD accumulation and that reducing LD accumulation using a DGAT1 inhibitor significantly reduced PRRSV replication. Together, our results suggest that LDs may act as energy-demanding elements or synthesis sites during PRRSV replication, and that the synthesis and metabolism of LDs may be an effective way to regulate PRRSV replication. NF-κB is a key transcription factor that regulates inflammatory cytokines involved in viral replication, apoptosis, inflammation, and various immune diseases (Barnabei et al., 2021). The activation of the NF-κB signaling pathway is part of the stress response, which is stimulated by factors including bacterial infection, viral infection, and the release of proinflammatory cytokines (Liu et al., 2017). PRRSV has been reported to activate NF-κB signaling and increase the release of IL-1β, IL-6, IL-8, and TNF-α cytokines (Ke et al., 2019). BAY11–7082 is a synthetic IκB kinase-β antagonist, BAY11–7082 allows IκB to protect NF-κB and prevent its translocation into the nucleus by inhibiting degradation of I-κB kinase-β. (Lang et al., 2021). In addition, it has been shown that NF-κB can activate the expression of SREBP1a protein containing the NF-κB response element in macrophages, which subsequently induces lipogenesis and IL-1β production (Kawai and Akira, 2010). Similarly, exogenous fatty acids (FAs) can activate the NF-κB signaling pathway and SREBP-1c protein expression to form LDs in hepatocytes (Jung et al., 2013). These findings further confirm the link between innate immune response and lipid metabolism and LDs, which requires NF-κB-dependent mechanisms. Therefore, to further explore whether PRRSV replication is related to the mutual regulation of LDs and NF-κB signaling pathway, we used BAY11–7082 and MH to reduce the NF-κB signaling pathway to detect PRRSV replication. Both BAY11–7082 and MH significantly reduced NF-κB p65 and IκB protein phosphorylation and the transcription of IL-1β and IL-8, and significantly reduced LD accumulation and PRRSV replication. In addition, we found that the DGAT1 inhibitor prevented LD accumulation in PRRSV-infected cells, reduced the transcription of NF-κB signaling pathway components and related inflammatory factors (IL-1β and IL-8), and significantly reduced PRRSV replication. Taken together, these findings indicate that reducing LD accumulation and the NF-κB signaling pathway could significantly reduce PRRSV replication and the transcription of the inflammatory cytokines IL-1β and IL-8. In conclusion, this study demonstrated that PRRSV infection induces LD accumulation and that reducing LD accumulation can significantly reduce PRRSV replication and the NF-κB signaling pathway following PRRSV infection while downregulating IL-1β and IL-8 transcription. Furthermore, reducing the NF-κB signaling pathway significantly reduced PRRSV replication and LD accumulation, suggesting that LD accumulation and NF-κB signaling interact to regulate PRRSV replication. Finally, we demonstrated that MH can pharmacologically block LD accumulation and NF-κB signaling to reduce PRRSV replication. These findings suggest a novel pathway through which PRRSV regulates LD accumulation to promote viral replication and provide a new strategy for the prevention and control of PRRSV infection. Funding This work was supported by the 10.13039/501100001809 National Natural Science Foundation of China (grant number 32102704) , Start-up Research Project of Maoming Laboratory (2021TDQD002) , and China Agriculture Research System of MOF and MARA (cars-35). CRediT authorship contribution statement Yunlong Yang: Writing – original draft, Writing – review & editing. Yizhuo Luo: Investigation, Writing – review & editing. Songqiang Yi: Investigation, Writing – review & editing. Qi Gao: Investigation, Writing – review & editing. Ting Gong: Investigation, Writing – review & editing. Yongzhi Feng: Visualization, Writing – review & editing. Dongdong Wu: Visualization, Writing – review & editing. Xiaoyu Zheng: Visualization, Writing – review & editing. Heng Wang: Visualization, Writing – review & editing. Guihong Zhang: Conceptualization, Writing – review & editing. Yankuo Sun: Conceptualization, Writing – review & editing. Declaration of Competing Interest The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. Data availability Data will be made available on request. ==== Refs References Barnabei L. Laplantine E. Mbongo W. Rieux-Laucat F. Weil R. NF-κB: at the borders of autoimmunity and inflammation Front. Immunol. 12 2021 716469 10.3389/fimmu.2021.716469 Bartz R. Li W.H. Venables B. Zehmer J.K. Roth M.R. Welti R. Anderson R.G. Liu P. Chapman K.D. Lipidomics reveals that adiposomes store ether lipids and mediate phospholipid traffic J. Lipid Res. 48 2007 837 847 10.1194/jlr.M600413-JLR200 17210984 Chand R.J. Trible B.R. Rowland R.R. Pathogenesis of porcine reproductive and respiratory syndrome virus Curr. Opin. Virol. 2 2012 256 263 10.1016/j.coviro.2012.02.002 22709514 Crawford S.E. Desselberger U. Lipid droplets form complexes with viroplasms and are crucial for rotavirus replication Curr. Opin. Virol. 19 2016 11 15 10.1016/j.coviro.2016.05.008 27341619 Dias S. Soares V.C. Ferreira A.C. Sacramento C.Q. Fintelman-Rodrigues N. Temerozo J.R. Teixeira L. Nunes da Silva M.A. Barreto E. Mattos M. de Freitas C.S. Azevedo-Quintanilha I.G. Manso P. Miranda M.D. Siqueira M.M. Hottz E.D. Pão C. Bou-Habib D.C. Barreto-Vieira D.F. Bozza F.A. Souza T. Bozza P.T. Lipid droplets fuel SARS-CoV-2 replication and production of inflammatory mediators PLoS Pathog. 16 2020 e1009127 10.1371/journal.ppat.1009127 Flory J. Lipska K. Metformin in 2019 JAMA 321 2019 1926 1927 10.1001/jama.2019.3805 31009043 Han J. Zhou L. Ge X. Guo X. Yang H. Pathogenesis and control of the Chinese highly pathogenic porcine reproductive and respiratory syndrome virus Vet. Microbiol. 209 2017 30 47 10.1016/j.vetmic.2017.02.020 28292547 Herker E. Harris C. Hernandez C. Carpentier A. Kaehlcke K. Rosenberg A.R. Farese R.V. Ott M. Jr Efficient hepatitis C virus particle formation requires diacylglycerol acyltransferase-1 Nat. Med. 16 2010 1295 1298 10.1038/nm.2238 20935628 Ishida K. Goto S. Ishimura M. Amanuma M. Hara Y. Suzuki R. Katoh K. Morita E. Functional correlation between subcellular localizations of japanese encephalitis virus capsid protein and virus production J. Virol. 93 2019 e00612 e00619 10.1128/JVI.00612-19 31315991 Jung T.W. Youn B.S. Choi H.Y. Lee S.Y. Hong H.C. Yang S.J. Yoo H.J. Kim B.H. Baik S.H. Choi K.M. Salsalate and adiponectin ameliorate hepatic steatosis by inhibition of the hepatokine fetuin-A Biochem. Pharmacol. 86 2013 960 969 10.1016/j.bcp.2013.07.034 23948064 Kawai T. Akira S. The role of pattern-recognition receptors in innate immunity: update on Toll-like receptors Nat. Immunol. 11 2010 373 384 10.1038/ni.1863 20404851 Ke H. Lee S. Kim J. Liu H.C. Yoo D. Interaction of PIAS1 with PRRS virus nucleocapsid protein mediates NF-κB activation and triggers proinflammatory mediators during viral infection Sci. Rep. 9 2019 11042 10.1038/s41598-019-47495-9 31363150 Kim H.S. Ren G. Kim T. Bhatnagar S. Yang Q. Bahk Y.Y. Kim J.A. Metformin reduces saturated fatty acid-induced lipid accumulation and inflammatory response by restoration of autophagic flux in endothelial cells Sci. Rep. 10 2020 13523 10.1038/s41598-020-70347-w 32782332 Kuhn J.H. Lauck M. Bailey A.L. Shchetinin A.M. Vishnevskaya T.V. Bào Y. Ng T.F. LeBreton M. Schneider B.S. Gillis A. Tamoufe U. Diffo J. Takuo J.M. Kondov N.O. Coffey L.L. Wolfe N.D. Delwart E. Clawson A.N. Postnikova E. Bollinger L. Lackemeyer M.G. Radoshitzky S.R. Palacios G. Wada J. Shevtsova Z.V. Jahrling P.B. Lapin B.A. Deriabin P.G. Dunowska M. Alkhovsky S.V. Rogers J. Friedrich T.C. O'Connor D.H. Goldberg T.L. Reorganization and expansion of the nidoviral family arteriviridae Arch. Virol. 161 2016 755 768 10.1007/s00705-015-2672-z 26608064 Lang Y. Chu F. Liu L. Zheng C. Li C. Shen D. Liu S. Zhang W. Cui L. Zhu J. Potential role of BAY11-7082, a NF-κB blocker inhibiting experimental autoimmune encephalomyelitis in C57BL/6J mice via declining NLRP3 inflammasomes Clin. Exp. Immunol. 207 2021 378 386 10.1093/cei/uxab022 Li Z. Thiel K. Thul P.J. Beller M. Kühnlein R.P. Welte M.A. Lipid droplets control the maternal histone supply of Drosophila embryos Curr. Biol. 22 2012 2104 2113 10.1016/j.cub.2012.09.018 23084995 Liu T. Zhang L. Joo D. Sun S.C. NF-κB signaling in inflammation Signal Transduct. Target. Ther. 2 2017 17023 10.1038/sigtrans.2017.23 – 29158945 Lu Y. Zhang Y. Xiang X. Sharma M. Liu K. Wei J. Shao D. Li B. Tong G. Olszewski M.A. Ma Z. Qiu Y. Notch signaling contributes to the expression of inflammatory cytokines induced by highly pathogenic porcine reproductive and respiratory syndrome virus (HP-PRRSV) infection in porcine alveolar macrophages Dev. Comp. Immunol. 108 2020 103690 10.1016/j.dci.2020.103690 Lunney J.K. Fang Y. Ladinig A. Chen N. Li Y. Rowland B. Renukaradhya G.J. Porcine reproductive and respiratory syndrome virus (PRRSV): pathogenesis and interaction with the immune system Annu. Rev. Anim. Biosci. 4 2016 129 154 10.1146/annurev-animal-022114-111025 26646630 Monson E.A. Trenerry A.M. Laws J.L. Mackenzie J.M. Helbig K.J. Lipid droplets and lipid mediators in viral infection and immunity FEMS Microbiol. Rev. 45 2021 fuaa066 10.1093/femsre/fuaa066 33512504 Monson E.A. Crosse K.M. Duan M. Chen W. O'Shea R.D. Wakim L.M. Carr J.M. Whelan D.R. Helbig K.J Intracellular lipid droplet accumulation occurs early following viral infection and is required for an efficient interferon response Nat. Commun. 12 1 2021 4303 10.1038/s41467-021-24632-5 34262037 Moreno-Navarrete J.M. Ortega F.J. Rodríguez-Hermosa J.I. Sabater M. Pardo G. Ricart W. Fernández-Real J.M. OCT1 Expression in adipocytes could contribute to increased metformin action in obese subjects Diabetes 60 2011 168 176 10.2337/db10-0805 20956498 Murtaugh M.P. Stadejek T. Abrahante J.E. Lam T.T. Leung F.C. The ever-expanding diversity of porcine reproductive and respiratory syndrome virus Virus Res. 154 1–2 2010 18 30 10.1016/j.virusres.2010.08.015 20801173 Olzmann J.A. Carvalho P. Dynamics and functions of lipid droplets Nat. Rev. Mol. Cell Biol. 20 2019 137 155 10.1038/s41580-018-0085-z 30523332 Preuss C. Jelenik T. Bódis K. Müssig K. Burkart V. Szendroedi J. Roden M. Markgraf D.F. A new targeted lipidomics approach reveals lipid droplets in liver, muscle and heart as a repository for diacylglycerol and ceramide species in non-alcoholic fatty liver Cells 8 2019 277 10.3390/cells8030277 30909521 Roingeard P. Melo R.C. Lipid droplet hijacking by intracellular pathogens Cell. 19 Microbiol. 2017 10.1111/cmi.12688 Samsa M.M. Mondotte J.A. Iglesias N.G. Assunção-Miranda I. Barbosa-Lima G. Da Poian A.T. Bozza P.T. Gamarnik A.V. Dengue virus capsid protein usurps lipid droplets for viral particle formation PLoS Pathog. 5 2009 e1000632 10.1371/journal.ppat.1000632 Tian K. Yu X. Zhao T. Feng Y. Cao Z. Wang C. Hu Y. Chen X. Hu D. Tian X. Liu D. Zhang S. Deng X. Ding Y. Yang L. Zhang Y. Xiao H. Qiao M. Wang B. Hou L. Wang X. Yang X. Kang L. Sun M. Jin P. Wang S. Kitamura Y. Yan J. Gao G.F. Emergence of fatal PRRSV variants: unparalleled outbreaks of atypical PRRS in China and molecular dissection of the unique hallmark PLoS One 2 2007 e526 10.1371/journal.pone.0000526 17565379 Villareal V.A. Rodgers M.A. Costello D.A. Yang P.L. Targeting host lipid synthesis and metabolism to inhibit dengue and hepatitis C viruses Antiviral Res. 124 2015 110 121 10.1016/j.antiviral.2015.10.013 26526588 Walther T.C. Farese R.V. Jr Lipid droplets and cellular lipid metabolism Annu. Rev. Biochem. 81 2012 687 714 10.1146/annurev-biochem-061009-102430 22524315 Xu S. Zhang X. Liu P. Lipid droplet proteins and metabolic diseases Biochim. Biophys. Acta Mol. Basis Dis. 1864 2018 1968 1983 10.1016/j.bbadis.2017.07.019 28739173 Yang C. Zuo Q. Liu X. Zhao Q. Pu H. Gao L. Zhao L. Guo Z. Lin Y. Liu J. Bi J. Yin G. Small molecule screening identified cepharanthine as an inhibitor of porcine reproductive and respiratory syndrome virus infection in vitro by suppressing integrins/ILK/RACK1/PKCα/NF-κB signalling axis Vet. Microbiol. 255 2021 109016 10.1016/j.vetmic.2021.109016 Yen C.L. Stone S.J. Koliwad S. Harris C. Farese R.V. Jr Thematic review series: glycerolipids. DGAT enzymes and triacylglycerol biosynthesis J. Lipid Res. 49 2008 2283 2301 10.1194/jlr.R800018-JLR200 18757836 Yu P.W. Fu P.F. Zeng L. Qi Y.L. Li X.Q. Wang Q. Yang G.Y. Li H.W. Wang J. Chu B.B. Wang M.D. EGCG restricts PRRSV proliferation by disturbing lipid metabolism Microbiol. Spectr. 10 2 2022 e0227621 10.1128/spectrum.02276-21 Zhang C. Liu P. The lipid droplet: a conserved cellular organelle Protein Cell 8 2017 796 800 10.1007/s13238-017-0467-6 28913786 Zhang M. Du T. Long F. Yang X. Sun Y. Duan M. Zhang G. Liu Y. Zhou E.M. Chen W. Chen J. Platycodin D suppresses type 2 porcine reproductive and respiratory syndrome virus in primary and established cell lines Viruses 10 2018 657 10.3390/v10110657 30469357 Zhang Y. Liu W. Zhong Y. Li Q. Wu M. Yang L. Liu X. Zou L. Metformin corrects glucose metabolism reprogramming and NLRP3 inflammasome-induced pyroptosis via inhibiting the TLR4/NF-κB/PFKFB3 signaling in trophoblasts: implication for a potential therapy of preeclampsia Oxid. Med. Cell. Longev. 2021 2021 1806344 10.1155/2021/1806344 Zhao J. Zeng Z. Chen Y. Liu W. Chen H. Fu Z.F. Zhao L. Zhou M. Lipid droplets are beneficial for rabies virus replication by facilitating viral budding J. Virol. 96 2022 e0147321 10.1128/JVI.01473-21
PMC010xxxxxx/PMC10352718.txt
==== Front Virus Res Virus Res Virus Research 0168-1702 1872-7492 Elsevier S0168-1702(23)00102-8 10.1016/j.virusres.2023.199140 199140 Article Emergence of a novel genetic lineage ‘A/ASIA/G-18/2019′ of foot and mouth disease virus serotype A in India: A challenge to reckon with Mohapatra Jajati Keshari [email protected] ⁎ Dahiya Shyam Singh Subramaniam Saravanan Rout Manoranjan Biswal Jitendra Kumar Giri Priyabrata Nayak Vinayak Singh Rabindra Prasad ICAR-National Institute on Foot and Mouth Disease, International Centre for FMD, Arugul, Bhubaneswar 752050, Odisha, India ⁎ Corresponding author at: ICAR-National Institute on Foot and Mouth Disease, International Centre for FMD, Arugul, Bhubaneswar 752050, Odisha, India. [email protected] 10 6 2023 8 2023 10 6 2023 333 19914020 4 2023 30 5 2023 30 5 2023 © 2023 The Authors. Published by Elsevier B.V. 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Highlights • VP1 region nucleotide sequence for 26 Indian FMD virus serotype A strains determined. • Emergence of a novel lineage ‘A/ASIA/G-18/2019′ in India and Bangladesh. • The node age for the lineage was predicted to be 2014 through tMRCA analysis. • Rate of evolution of the VP1 region estimated to be 6.747 × 10−3 substitutions/site/year. • AIND27/2011 could be more appropriate vaccine strain in the scenario than AIND40/2000. Foot and mouth disease (FMD) has engendered large scale socioeconomic crises on numerous occasions owing to its extreme contagiousness, transboundary nature, complicated epidemiology, negative impact on productivity, trade embargo, and need for intensive surveillance and expensive control measures. Emerging FMD virus variants have been predicted to have originated and spread from endemic Pool 2, native to South Asia, to other parts of the globe. In this study, 26 Indian serotype A isolates sampled between the year 2015 and 2022 were sequenced for the VP1 region. BLAST and maximum likelihood phylogeny suggest emergence of a novel genetic group within genotype 18, named here as ‘A/ASIA/G-18/2019′ lineage, that is restricted so far only to India and its eastern neighbour, Bangladesh. The lineage subsequent to its first appearance in 2019 seems to have displaced all other prevalent strains, in support of the phenomenon of ‘genotype/lineage turnover’. It has diversified into two distinct sub-clusters, reflecting a phase of active evolution. The rate of evolution of the VP1 region for the Indian serotype A dataset was estimated to be 6.747 × 10−3 substitutions/site/year. India is implementing a vaccination centric FMD control programme. The novel lineage showed good antigenic match with the proposed vaccine candidate A IND 27/2011 when tested in virus neutralization test, while the existing vaccine strain A IND 40/2000 showed homology with only 31% of the isolates. Therefore, in order to combat this challenge of antigenic divergence, A IND 27/2011 could be the preferred strain in the Indian vaccine formulations. Keywords FMD virus Serotype A Phylogeny Vaccine matching Novel lineage A/ASIA/G-18/2019 ==== Body pmc1 Introduction Foot and mouth disease (FMD) is considered globally to be one of the major transboundary animal diseases having far reaching socio-economic consequences. The causative agent, FMD virus (FMDV), is a small, non-enveloped, positive-strand RNA virus classified within the genus Aphthovirus in the family Picornaviridae. A total of seven immunologically and genetically distinguishable FMDV serotypes, such as A, O, Asia 1, C, Southern African Territories (SAT)−1, SAT-2, and SAT-3 exist and innumerable genetic and antigenic variants within each serotype are being reported continuously (Grubman and Baxt, 2004). Based on the pattern of prevalence of specific serotypes and topotypes/lineages, the global burden of FMDV can be subdivided into 7 endemic virus pools (Paton et al., 2009). Typically for an RNA virus, random mutations resulting from error-prone RNA replication accumulate constantly giving rise to new lineages from time to time, incidental with a change in phenotype of the virus (Duffy, 2018). At times it can pose serious threat to the ongoing vaccination centric disease control programme due to emergence and expansion of antigenic variants capable of causing breakthrough infection. There have been instances where the lineages have transgressed their geographic niche leading to inter-pool, occasionally intercontinental, virus dispersal (Bachanek-Bankowska et al., 2018; Knowles et al., 2005; Singanallur et al., 2022; Valarcher et al., 2009). Also, in a particular endemic region, waxing and waning of lineages have been observed warranting frequent change in the vaccine strains (Bari et al., 2014; Mahapatra et al., 2016; Mattion et al., 2004; Mohapatra et al., 2018). Such epidemiological trends underline the importance of continued surveillance. In India, FMD with prevalence of serotypes O, A, and Asia1, is thought to be one of the most important infectious diseases in the livestock health sector and therefore its control has been given priority in the policy decisions (Subramaniam et al., 2022). India shares international land border with several South Asian countries, home to the virus Pool 2. Countrywide biannual mass vaccination is at the fulcrum of the WOAH endorsed official FMD control programme operating in India. ICAR-National Institute on FMD (ICAR-NIFMD), formerly named as ICAR-Directorate of FMD (ICAR-DFMD), being the national referral laboratory for FMD, is entrusted with the responsibility of identifying the appropriate vaccine strains (Mohapatra et al., 2018). FMD epidemiological studies have benefited immensely from capsid coding (P1) region sequence analysis, especially that of the VP1 region (Knowles and Samuel, 2003). Due to the considerable antigenic heterogeneity displayed by serotype A FMDV, it becomes increasingly important to characterise the antigenic behaviour of every single field outbreak strain. This study describes the genetic and antigenic characterization of FMDV serotype A viruses isolated from field outbreaks in India between 2015 and 2022, and reports emergence of a novel lineage, named here as ‘A/ASIA/G-18/2019′. An appropriate vaccine strain having adequate antigenic match with the contemporary field strains is always desirable for effective vaccination campaign and progressive regional disease control. Phylogenetic analysis and vaccine matching of field viruses therefore bear great significance particularly for countries heading towards or presumed to be at stage 3 of the progressive control pathway that implement countrywide aggressive vaccination and surveillance strategy to eliminate the circulation of FMDV. 2 Materials and methods 2.1 Serotype identification and virus isolation in cell culture From April 2014 to December 2022, a total of 8041 clinical samples were collected from FMD suspected cases and processed for the identification of FMD virus serotype using serotype differentiating sandwich ELISA (Bhattacharya et al., 1996) and reverse transcription-multiplex PCR (RT-mPCR) (Giridharan et al., 2005). The clarified tissue homogenates were given three blind passes in BHK-21 cells for isolation of the virus. 2.2 RNA extraction, reverse transcription, PCR and nucleotide sequencing From the infected cell culture supernatant, confirmed for FMDV serotype A (Table 1), total RNA was extracted using the QIAamp Viral RNA Mini Kit (Qiagen, Germany). Reverse transcription was carried out using MMuLV reverse transcriptase (Promega, USA) with antisense primer NK61 (5′-GACATGTCCTCCTGCATCTG, Knowles and Samuel, 1995) at 42°C for 1 h. PCR amplification of the VP1/2A coding region was performed using pfu DNA polymerase (Thermo Fisher Scientific, USA) with primer pair AVP3–562F (5′TACCAGATCACCCACGGGAAGGC, Pandey et al., 2014) and NK61. Cycle sequencing reaction of gel-purified PCR products (about 860 bp) was assembled using the Bigdye V3.1 terminator kit. Both sense primer AVP3–562F and antisense primer NK61 were run separately in the ABI 3500 genetic analyser (Applied Biosystems, USA) to generate the VP1 coding (1D) region sequence.Table 1 History of FMD virus serotype A field isolates sequenced in this study. Table 1S No Isolate ID Place of outbreak Date of collection Host Accession no. 1 A/IND29(82)/2015 Ramanagara, Karnataka 2015 Cattle OQ378364 2 A/IND15(78)/2015 Meerut Cantt, Uttar Pradesh 14–04–2015 Cattle OQ378365 3 A/IND15(79)/2015 Meerut Cantt, Uttar Pradesh 14–04–2015 Cattle OQ378366 4 A/IND28/2015 Meerut Cantt, Uttar Pradesh 2015 NA OQ378367 5 A/IND111(256)/2015 Bangalore Urban, Karnataka 16–10–2015 Pig OQ378368 6 A/IND111(259)/2015 Bangalore Urban, Karnataka 16–10–2015 Pig OQ378369 7 A/IND111(262)/2015 Bangalore Urban, Karnataka 16–10–2015 Pig OQ378370 8 A/IND10(24)/2016 Izatnagar,Uttar Pradesh 2016 Cattle OQ378371 9 A/IC67/2019 Satara, Maharashtra Jan-2019 Cattle OQ378372 10 A/IC68/2019 Satara, Maharashtra Jan-2019 Cattle OQ378373 11 A/IC69/2019 Satara, Maharashtra Jan-2019 Cattle OQ378374 12 A/IC16/2021 Muzaffarnagar, Uttar Pradesh Jan-2021 Cattle OQ378375 13 A/IC18/2021 Muzaffarnagar, Uttar Pradesh Jan-2021 Cattle OQ378376 14 A/IC22/2021 Medziphema,Nagaland 29–01–2021 Mithun OQ378377 15 A/IC154/2021 Kupwara, Srinagar, J & K 09–06–2021 Cattle OQ378378 16 A/IC161/2021 Kupwara, Srinagar, J & K 09–06–2021 Cattle OQ378379 17 A/IC164/2021 Kupwara, Srinagar, J & K 09–06–2021 Cattle OQ378380 18 A/IC165/2021 Kupwara, Srinagar, J & K 09–06–2021 Cattle OQ378381 19 A/IC167/2021 Kupwara, Srinagar, J & K 09–06–2021 Cattle OQ378382 20 A/IC341/2021 Ludhiana, Punjab July-2021 Cattle OQ378383 21 A/IC370/2021 Brajrajnagar, Odisha 07–08–2021 Cattle OQ378384 22 A/IC655/2021 Bangalore Urban, Karnataka 12–06–2021 Cattle OQ378385 23 A/IC793/2021 Chikkaballapur, Karnataka 15–07–2021 Cattle OQ378386 24 A/IC794/2021 Chikkaballapur, Karnataka 15–07–2021 Cattle OQ378387 25 A/IC195/2022 Jajpur, Odisha 24–02–2022 Cattle OQ378388 26 A/IC196/2022 Jajpur, Odisha 24–02–2022 Cattle OQ378389 2.3 Sequence alignment and phylogenetic analysis In order to see the genetic relationship of the novel lineage with isolates from other countries, a nucleotide BLAST analysis was performed using the Basic Local Alignment Search Tool (https://blast.ncbi.nlm.nih.gov/Blast.cgi, accessed on 01–02–2023). Phylogenetic analysis was performed using the MEGA 11 software (Molecular Evolutionary Genetics Analysis; www.megasoftware.net) (Tamura et al., 2021). 1D region sequences of Indian FMDV serotype A strains retrieved from either GenBank or the nucleotide sequence database of the Institute along with that generated in this study for 26 FMDV serotype A isolates (GenBank ac. no. OQ378364 to OQ378389) from field outbreaks between 2015 and 2022 were included in the analysis (Table 1). Also, comparative phylogenetic analysis was carried out including sequence of contemporary serotype A viruses from other countries retrieved from the GenBank. Multiple sequence alignment was performed using the MUSCLE algorithm available in MEGA. Based on Bayesian Information Criterion scores, the best fit Hasegawa–Kishono–Yano (HKY) model with gamma-distributed rates of variables amongst sites (+G) was used to infer the evolutionary relationship. Phylogenetic trees were constructed using the maximum likelihood statistical method inferred from 1000 replicates using bootstrap resampling. 2.4 Spatio-temporal analysis The BEAST 1.10.4 programme (Suchard et al., 2018) was used to perform Bayesian Markov chain Monte Carlo (MCMC) analyses in order to determine the approximate age of the dataset and evolutionary rate. A HKY+G nucleotide substitution model was assumed under a relaxed uncorrelated lognormal (UCLN) clock and the exponential coalescent model. Using marginal likelihoods calculated by path sampling (PS) and stepping-stone sampling (SS), the clock and coalescent models were chosen (Baele et al., 2012). To ensure adequate parameter convergence, the MCMC chain was run 200 million times. Using Tree Annotator, the posterior tree distributions were summarised. The 95% highest probability density (HPD) values represent the statistical uncertainty in the data. The MCC consensus tree was plotted using Fig Tree version 1.4.4. Based on the MCC phylogeny, divergence times (time to the most recent common ancestor, tMRCA) and nucleotide substitution rates were calculated for significant nodes, with statistical error expressed as the 95% highest probability density (95% HPD). To monitor geographic location states along the evolutionary tree, Bayesian stochastic search variable selection (BSSVS) was used (Lemey et al., 2009). The Bayes factor (BF) analysis available in SPREAD 1.0.7 (Bielejec et al., 2011) was used to test the most significant pathway of distribution. 2.5 Selection pressure analysis The nature and strength of selection pressure acting upon the VP1 region alignment was evaluated as the ratio of non-synonymous (dN) to synonymous (dS) mutations (dN/dS). Individual sites were assessed for evidence of purifying or diversifying selection pressure using two separate methods available through DataMonkey (Weaver et al., 2018). A likelihood approach, the single-likelihood ancestor (SLAC), was used to identify sites under pervasive selection pressure. For determining sites under episodic diversifying selection, the Mixed Effects Model of Evolution (MEME) was employed. Strong evidence of selection was accepted at significance levels (p <0.1). 2.6 Antigenic analysis The antigenic relationship of the field outbreak isolates with the in use vaccine strain A IND 40/2000 and the proposed alternate vaccine strain A IND 27/2011 was evaluated in two-dimensional virus neutralization test (2D-VNT) using bovine vaccinal serum (BVS) raised against the vaccine strains as described earlier (Mohapatra et al., 2018; Rweyemamu et al., 1978). The mean one-way antigenic relationship coefficient, ‘r1’ value (titre of BVS against heterologous field virus/titre against homologous vaccine virus) was determined from two runs of 2D-VNT. A BVS was titrated against both homologous and heterologous virus to find out the log10 SN50 titres of the serum that neutralised exactly 100 TCID50 virus in 50% of the wells from regression lines. An ‘r1’ value of greater than 0.3 indicates the field isolate is sufficiently similar antigenically to the vaccine strain, whereas value less than 0.3 suggests that the vaccine strain is unlikely to confer protection against challenge with the field isolate (Mohapatra et al., 2018; Rweyemamu, 1984). 3 Results 3.1 Prevalence and distribution of FMDV Serotype A A total of 8041 FMD suspected clinical samples referred between 2014 and 2022 were tested using sandwich ELISA and multiplex PCR to confirm 1505 number of FMD outbreaks in India (Table 2). Serotype O has invariably been the most prevalent one amongst the three circulating serotypes, which was detected in about 90% of the incidences. Serotype A accounted for only 2.72% of the total number of outbreaks recorded over the last nine years (2014–2022). Outbreaks with involvement of serotype A have been reported across the regions within the country except the central part of India. Interestingly, no incidence due to serotype A virus could be detected during 2014–2015 and 2016–2018. The highest number of outbreaks due to serotype A was recorded during 2021 when the country experienced a surge in overall number of outbreaks. Serotype A virus could be isolated from 21 clinical samples using BHK-21 cells.Table 2 Proportion and regional distribution of outbreaks due to FMD virus serotype A during the period 2014–2022. Table 2Period Total no. of outbreaks (total no. of clinical samples tested) Total no. of outbreaks due to serotype A (no. of clinical samples positive for serotype A)a States that reported outbreaks due to FMDV serotype A 2014–15 (April-March) 76 (182) 0 – 2015–16 (April-March) 252 (671) 6 (11) Karnataka, Tamil Nadu, Uttar Pradesh, Uttarakhand 2016–17 (April-March) 150 (523) 0 – 2017–18 (April-March) 149 (520) 0 – 2018 (April-December) 351 (2396) 0 – 2019 (January-December) 52 (306) 1 (2) Maharashtra 2020 (January-December) 45 (215) 6 (28) Assam, Meghalaya 2021 (January-December) 378 (2824) 25 (87) Karnataka, Tamil Nadu, Uttar Pradesh, Maharashtra, Punjab, Jammu and Kashmir, Odisha, Nagaland 2022 (January-December) 52(404) 3 (6) Odisha, Assam, Mizoram Total 1505 (8041) 41 (134) a Multiple samples collected from the same outbreak were subjected to testing. 3.2 Phylogenetic analysis In the 1D nucleotide sequence based maximum likelihood phylogenetic tree (Fig. 1), four distinct genotypes in India (2, 10, 16, and 18) were clearly evident with a high boot strap support (> 80%). The genotypes differed by at least 15% from one another in 1D nucleotide sequence. After 2001, all the isolates grouped in genotype 18, being distributed amongst two distinct lineages such as ‘VP359-deletion lineage’ and ‘non-deletion group’. However, 18 sequences for viruses isolated during 2019–2022 out of the 26 resolved in this study did not cluster in any of these lineages. This genetically distinct novel group, first isolated in 2019, was designated here as ‘A/ASIA/G-18/2019′ lineage. Isolates of this lineage exhibited a mean nucleotide difference of more than 15% with genotypes 2 (18.9%), 10 (15.1%), and 16 (17.3%). Within genotype 18, the viral sequences of the novel group showed a 9% and 10.6% mean nucleotide difference from the ‘VP359 deletion lineage’ and ‘non-deletion group’, respectively. The nucleotide difference amongst the A/ASIA/G-18/2019 lineage strains varied from 0.0 to 6.6%. Within the A/ASIA/G-18/2019 lineage, two distinct sub-clusters were evident. The sub-cluster 1 comprised 11 isolates collected from the states of Maharashtra in 2019, Punjab, Nagaland, and Karnataka in 2021, and Odisha in 2021 and 2022. Seven isolates collected from the states of Uttar Pradesh and Jammu and Kashmir during 2021 grouped in the sub-cluster 2.Fig. 1 Maximum likelihood phylogenetic tree reconstructed based on VP1 coding sequence of FMD virus serotype A isolates from India. Sequences generated in this study are identified with red-filled triangles. Bootstrap values >70% are shown next to the nodes. Fig 1 Sequences of two isolates representing two subclusters of the A/ASIA/G-18/2019 lineage were used for BLAST search. A maximum nucleotide similarity of 98% with an isolate from Bangladesh collected in 2020 was observed. In addition, in global phylogeny it was evident that A/ASIA/G-18/2019 lineage has distinct genetic identity and it forms a monophyletic cluster of its own in which none of the global sequences except the 2020 strains from Bangladesh grouped (Fig. 2).Fig. 2 Maximum likelihood evolutionary relationship inferred based on VP1 coding sequence of contemporary FMD virus serotype A strains from other countries and from India. Bootstrap values >70% are shown next to the nodes. Fig 2 3.3 Variation in VP1 amino acid (aa) sequence All the isolates of A/ASIA/G-18/2019 lineage had 639 nucleotides long VP1 region coding for 213 aa (Fig. 3). Receptor binding motif ‘RGD’ at VP1 aa position 144–146 and Leucine (L) residue immediately downstream of the motif at position 147 in the βG-βH loop were fully conserved in all except in one sequence where L147 was replaced by S147. At four positions, the novel lineage showed variation from both the vaccine strains, A IND 40/2000 and the proposed candidate strain A IND 27/2011. Additionally, at nine positions, in some of the sequences of the novel lineage, variation was noticed in comparison to both the vaccine strains. At seven positions, the novel lineage showed variation exclusively from A IND 40/2000 as compared to three positions where they varied from A IND 27/2011. Studies using monoclonal and polyclonal antibody neutralization resistant mutants have identified 22 aa sites in the VP1 region such as 83, 137, 139, 141, 143, 145, 147–155, 158, 170, 173, 199, 201, 205 and 209 that are critical to FMDV serotype A antigenicity (Das et al., 2016). Out of these 22 sites, all the strains of the novel lineage varied from A IND 40/2000 at three positions (83 S-D; 154 I-V and 170 D-T), while at two positions from A IND 27/2011 (139 P-A and 155 T-A). In addition, at three more positions (147 L-L/S; 152 A-A/V and 201 H—H/Y), a few of the strains of the novel lineage showed variation from both the vaccine strains. Between the sub-cluster 1 and 2 within the novel lineage, mutually exclusive aa signatures at certain positions such as 32 (V-I), 43 (G-Q), 44 (N-S), 108 (S-N) and at 134 (S-N) were observed, respectively.Fig. 3 Alignment of deduced VP1 amino acid sequence of strains belonging to subcluster-1 and 2 of the novel A/ASIA/G-18/2019 lineage with that of the vaccine strain A IND 40/2000 and alternate vaccine strain A IND 27/2011. A dot indicates similarity with that of A IND 40/2000 and variations are shown as single letter amino acid codes. Fig 3 3.4 Spatio-temporal analysis Marginal likelihoods estimation was used to test several clock and coalescent model combinations, with the combination of an uncorrelated lognormal clock and an exponential prior providing the best fit for the serotype A data set. We reconstructed MCC (Fig. 4) to estimate tMRCA as well as the evolutionary rates of each of the defined genetic groups. The overall rate of molecular evolution of the VP1 region for Indian isolates was estimated to be 6.747 × 10−3 nucleotide substitutions/site/year (s/s/y) with a 95% credibility interval of 5.599–7.95 × 10−3 s/s/y. The mutation rate at the third codon position was estimated to be 1.666 (95% HPD 1.538–1.801), which is the fastest evolving compared to codon positions 1 and 2, which had a combined rate of 0.667 (95% HPD 0.599–0.731).Fig. 4 Maximum clade credibility (MCC) evolutionary relationships of FMD virus serotype A VP1 coding sequences from different states of India. Branches are coloured according to the most probable location state of their descendant nodes. Horizontal branch lengths are drawn to scale, with the bar at the bottom indicating years. The A/ASIA/G-18/2019 lineage is shown within the red box. Fig 4 The mean tMRCA of Indian serotype A sequences was estimated to be 1964 with a 95% HPD of 1955–1972. The common branch leading to genotypes 2, 10, and 18 diverged approximately in 1969, while the one that leads to genotype 16 diverged in 1972. Genotype 10 emerged from the common ancestor around 1971, and genotypes 2 and 18 emerged from the common ancestor around 1975. Furthermore, genotype 18 diverged from the common ancestor in 1986 and the VP359-deletion group within genotype 18 diverged in 1996. The node age of A/ASIA/G-18/2019 lineage was determined to be 2014. With respect to A/ASIA/G-18/2019 lineage, the state of Maharashtra received the highest marginal support (43.3%) for root state posterior probabilities (RSSP), where the new lineage was first documented in 2019. The states of Karnataka and West Bengal received RSSPs of 17% each. BSSVS analysis was conducted using states as the geographical units on A/ASIA/G-18/2019 lineage isolates (Fig. 5). Isolates of this new lineage spread to Uttar Pradesh and Nagaland after first appearing in Maharashtra in 2019. From Uttar Pradesh and Nagaland, the virus was transmitted to Jammu and Kashmir and Karnataka, respectively. Further migration to Punjab and Odisha occurred from Karnataka. All the transmission events had a BF support of more than 3 and high posterior probability values.Fig. 5 Spatiotemporal dynamics predicted for the A/ASIA/G-18/2019 lineage found in India. The arrows between the locations indicate the migration route of A/ASIA/G-18/2019 lineage. Only the propagation paths with BF> 3 are shown. States are numbered as per predicted chronology of virus dissemination event. Fig 5 3.5 Selection pressure To examine the selection pressure on the serotype A virus, measures of non-synonymous to synonymous changes per site were calculated for the whole data set and different genetic groups. In general, the mean ratio of nonsynonymous to synonymous substitutions (dN/dS) was found to be low and varied from 0.151 to 0.250, and most of the nucleotide substitutions were synonymous. VP359-deletion group of genotype 18 had a high dN/dS ratio, while A/ASIA/G-18/2019 lineage had the lowest. Overall, the dN/dS ratio of the serotype A data set was determined to be 0.175. Furthermore, the majority of the sites (66% of the codon positions) were under purifying selection, and only 3 sites, viz., 24, 134, and 171, were detected to be under pervasive positive selection with statistical significance (p < 0.1). Codon position 65 in genotype 16, position 171 in the non-deletion group and 24 and 45 in the deletion group of genotype 18 were found to be under positive selection pressure. Within genotype 18, purifying selection was found to be more profound in the deletion group (31% of the codon positions) than the non-deletion group (15% of the codon positions). Episodic selection seems to play a significant role in shaping the evolution of VP1, as fifteen sites, viz., 22, 33, 43, 45, 95, 99, 134, 140, 153, 152, 169, 170, 171, 173, and 201, were found to be under episodic selection. 3.6 Vaccine matching of viruses from the novel genetic lineage Taking into account the emerging antigenic diversity of the field viruses, a suitable alternate FMDV serotype A candidate vaccine strain, A IND 27/2011 was proposed based on its antigenic relatedness with the circulating field strains (Mohapatra et al., 2018). In the present study, thirteen isolates of the A/ASIA/G-18/2019 lineage were subjected to antigenic matching with both in use vaccine strain A IND 40/2000 and the alternate strain A IND 27/2011. Only 31% of the isolates had an r value >0.3 with A IND 40/2000 indicating its poor antigenic coverage for the novel lineage as well. On the other hand, A IND 27/2011 demonstrated comparatively better antigenic homology (100%) with the recent isolates (Table 3).Table 3 Antigenic relationship ‘r1’ value of A/ASIA/G-18/2019 lineage isolates with the in use and proposed vaccine strain. Table 3FMD Virus Isolate Current vaccine strain (A IND 40/2000) Proposed vaccine strain (A IND 27/2011) A/IC154/2021 0.1 0.925 A/IC161/2021 0.41 0.933 A/IC164/2021 0.103 1 A/IC165/2021 0.21 1 A/IC167/2021 0.567 1 A/IC341/2021 0.264 0.941 A/IC370/2021 0.517 0.745 A/IC655/2021 0.584 0.611 A/IC793/2021 0.064 0.58 A/IC794/2021 0.224 1 A/IC195/2022 0.19 0.70 A/IC196/2022 0.16 0.50 A/IC197/2022 0.18 0.42 4 Discussion amongst the three Eurasian serotypes such as O, A and Asia 1 prevalent in India, historically, serotype O is involved in the majority of the field outbreaks (Subramaniam et al., 2022). Proportion of outbreaks attributed to serotype A is significantly lower than that due to serotype O, despite considerable genetic and antigenic diversity associated with serotype A virus population (Kitching, 2005). Serotype A virus incidence was reported to be intermittently absent in some of the years during the study period. Notably in 2018, when the FMD outbreaks in the country reached an epidemic scale of 351 outbreaks during a single year and 2396 clinical samples were tested, no incidence due to serotype A could be confirmed in the laboratory (Table 2). Such an observation possibly suggests genuine absence of serotype A incidence rather than a flawed surveillance system. However such finding is hard to generalise since chance of missing out an incidence of serotype A due to sampling bias can't be ruled out unless every single FMD suspected case is investigated in time. A detailed phylogenetic analysis from a global perspective is imperative to gain insights into geographic distribution, transmission and evolution of FMDV (Samuel and Knowles, 2001). After 2015, analysis of nucleotide sequence generated for the Indian FMDV serotype A field strains in relation to those from other parts of the globe have not been reported. In the 1D nucleotide sequence based maximum likelihood phylogenetic tree inferred for the serotype A Indian viruses (Fig. 1), four distinct genotypes (2, 10, 16, and 18) within the Asia topotype were clearly evident as described in our earlier study (Das et al., 2016; Mohapatra et al., 2011). All 26 FMDV serotype A isolates collected from field outbreaks in India between 2015 and 2022 clustered in genotype 18, which has been the only genotype present in the country since 2001. An interesting phenomenon of ‘genotype/lineage turnover’ with a period of co-circulation of multiple genotypes followed by exclusive dominance of one when all other resident genotypes have been supplanted by the emerging variant has been noticed for serotype A in India (Das et al., 2016; Mohapatra et al., 2011; Tosh et al., 2003). Eight sequences determined in this study for viruses isolated during 2015–2016 grouped in the VP359-deletion group. Between 2016 and 2018, none of the reported field outbreaks revealed involvement of serotype A virus, possibly an interlude signalling progressive disappearance of the resident lineages and entry into a phase of ‘lineage turnover’. Eighteen isolates collected during 2019–2022 clustered within genotype 18, but were found to be genetically distinct from both ‘VP359-deletion group’ and ‘non-deletion group’ that used to co-circulate in India. This novel genetic group, designated as A/ASIA/G-18/2019 lineage, was detected first in 2019 in an outbreak in Maharashtra state. It was subsequently reported from as many as eight states of India and has diversified into two discrete sub-clusters with time suggesting an active phase of evolution. Exclusive presence of this novel lineage since 2019 and no documentation of the two lineages of genotype 18 that previously occupied the same geo-ecological niche is once again reflective of the remarkable phenomenon of ‘lineage turnover’, rather than an oscillatory trajectory of ‘emergence-subdued circulation-reemergence’ of viral lineages. Both BLAST and global phylogenetic clustering suggests the novel A/ASIA/G-18/2019 lineage is so far restricted to only the two neighbouring countries such as India and Bangladesh (Fig. 2). Nucleotide homology between isolates from Bangladesh reported during 2020 and sub-cluster1 of the A/ASIA/G-18/2019 lineage was found to be about 98% indicating an epidemiological link and common ancestry. Although the exact country of origin couldn't be ascertained, this lineage has firmly secured its foothold in South Asian region, home to endemic Pool 2. Formal/informal cross border trade in livestock/livestock products can't be precluded as one of the primary contributing factors to exchange of viruses between the neighbouring countries which share vast international land border. Previously, geographical contiguity and livestock trade have been identified as key drivers for easing virus dispersal between and within countries of the Western and Southern Asia regions (Di Nardo et al., 2021). In the past, new FMDV lineages had shown geographical expansion beyond their ecosystems of origin over a short period of time and had quickly supplanted earlier dominating lineages (Brito et al., 2017). For instance, genotype 18 (also referred to as A/ASIA/G-VII) established itself as one of the dominant strains after spreading from Pool 2 to Saudi Arabia and Iran in 2015 and further into Armenia, Turkey, and northern Israel during 2017 (Singanallur et al., 2022). Taken together, there is no denying that the emerging A/ASIA/G-18/2019 lineage is presently a concern at least to the South Asian region and could be perceived as a threat to other parts of Asia since historically serotype A genetic variants have generally surfaced with a changed antigenic profile, thereby necessitating change in the vaccine strain (Jangra et al., 2005; Mohapatra et al., 2008 & 2018). The overall rate of molecular evolution of the VP1 region for Indian serotype A virus estimated here as 6.747 × 10−3substitutions/site/year (s/s/y) with a 95% credibility interval of 5.599–7.95 × 10−3 s/s/y is slightly higher than the estimates for the global serotype A dataset, 4.26 × 10−3 s/s/y (Tully and Fares, 2008), and for the Kenyan serotype A isolates, 4.22 × 10−3 s/s/y (Wekesa et al., 2014). In an earlier study, the overall rate of molecular evolution of the P1 region for Indian isolates (1977–2013) was estimated to be 4.96 × 10−3 substitutions/site/year (s/s/y) with a 95% credibility interval of 4.17–5.85 × 10−3 s/s/y (Das et al., 2016). The rate estimated here is found to be lower than the mean evolutionary rate of lineages Sea-97 (1.2 × 10−2 s/s/y, Bae et al., 2021), Iran-05 (1.25 × 10−2; Jamal et al., 2011) (1.3 × 10−2; Di Nardo et al., 2021), and G-VII clade C (1.1 × 10−2 s/s/y, Bachanek-Bankowska et al., 2018). Although such variation in rate is far too complex to reason, there could be a number of ecological and evolutionary variables acting together to modulate the transmissibility, magnitude of selection pressure, and pace and course of induced evolution. Factors such as density and variety of susceptible animal population, diversity of indigenous breeds, mixed farming set up, variable level of herd immunity, untraceable animal movement and porous international borders permeable to trade in livestock in the country, to name a few, could possibly be impacting the rate of evolution of the virus. The mean tMRCA of Indian serotype A sequences was estimated to be 1964 with a 95% HPD of 1955–1972, which is in good agreement with the estimate reported earlier based on P1 region sequences (Das et al., 2016). The node age of A/ASIA/G-18/2019 lineage was predicted to be 2014 in this study, although the first field virus belonging to this lineage was isolated only during 2019 (Fig. 4). The state of Maharashtra received the highest marginal support (43.3%) for RSSP, where the lineage was first documented. The states of Karnataka and West Bengal received RSSPs of 17% each. Interestingly however, the new lineage is yet to be reported from the state of West Bengal. BSSVS analysis was conducted using states as the geographical units and isolates of the new lineage spread to as many as seven states after Maharashtra (Fig. 5). The timeline of reporting of outbreaks from various states could be correlated with the predicted virus transmission pathway between the states (Table 1 & Fig. 5). However, sampling biases due to inconsistent effort in sampling represent an important caveat in reconstruction of such spatial network of virus transmission and therefore should be interpreted cautiously (Di Nardo et al., 2021). Out of the 22 residues in VP1 known to be critical to the antigenic sites of serotype A virus (refer to Das et al., 2016), the novel lineage varied from the in use vaccine strain A IND 40/2000 at three positions (83 S-D; 154 I-V and 170 D-T), while at two positions from the alternate strain A IND 27/2011 (139 P-A and 155 T-A) (Fig. 3). Although it is difficult to pin point variations in the capsid region which can be linked to antigenic diversity in general (Das et al., 2016), the observed variations might have played a role in the antigenic divergence of serotype A field isolates. Nonetheless, it is logical to determine the whole P1 region sequence in order to link the antigenic phenotype with difference in the aa sequences since many vital antigenic sites are located outside of the VP1. Majority of the sites in VP1 were observed to be under purifying selection. Codon position 65 in genotype 16, position 171 in the non-deletion group and 24 and 45 in the VP359-deletion group were found to be under positive selection pressure. These sites are not yet identified as antigenically critical. The importance of selection pressure is not known but might be playing a role in virus adaptation or in fixation of compensatory mutations to preserve the architectural integrity of the capsid in response to substitutions accrued elsewhere. amongst the fifteen sites viz., 22, 33, 43, 45, 95, 99, 134, 140, 153, 152, 169, 170, 171, 173, and 201, found to be under episodic selection, five sites (153, 154, 170, 173, and 201) located on the surface exposed loops or carboxy termini are known to be antigenically critical (Das et al., 2016). Also, position 170 and 173 located on βH–βI loop have been detected to be under positive selection in earlier analyses including entire P1 region data (Das et al., 2016; Tosh et al., 2003). Many codons experience purifying selection for the major part of their evolutionary space, with bursts of strong positive selection within particular lineages. Such sites may experience transient positive selection, followed by purifying selection to maintain the erstwhile residue, and are likely to contribute to adaptive evolution (Grueber et al., 2014). Preventive biannual vaccination is the mainstay of India's official FMD control programme (Pattnaik et al., 2012). The effectiveness of the vaccination campaign is severely constrained by a poor antigenic match between field isolates and vaccine strain (Jangra et al., 2005; Maradei et al., 2013). Surprisingly, A IND 40/2000 vaccine strain failed to offer optimum antigenic coverage to isolates of its own genotype since 2012, indicating considerable intra-genotypic antigenic variation to an extent that prompted selection of an alternate candidate vaccine strain (Mohapatra et al., 2018). Reportedly, serotype A FMDV that are prevalent in one particular area share a similar genetic makeup but differ from one another in their antigenic characteristics thereby affecting the effectiveness of the vaccination campaign (Ludi et al., 2014; Seeyo et al., 2020). Virus exposure or vaccine induced pre-existing partial immunity in the host population is one of the key determinants that drive FMDV evolution in endemic settings and can catalyse rapid and recurring changes in the antigenic profile of circulating strains (Domingo et al., 1980). The geographic expansion of genotype 18 from Pool 2 to the Middle East countries and the poor vaccine matching results with the available FMD vaccines have been the cause of serious concern in the past (Waters et al., 2018; Singanallur et al., 2022). Therefore, monitoring the antigenic behaviour of the field viruses becomes more important in the context of emergence of novel lineages. In total, thirteen isolates of A/ASIA/G-18/2019 lineage were subjected to antigenic matching with both existing (A IND 40/2000) and proposed alternate (A IND 27/2011) vaccine strains. Only 31% of the isolates exhibited an ‘r1’ value of >0.3 with the current vaccine strain, while all of them showed optimum antigenic relatedness with the alternate vaccine strain (Table 3). A IND 27/2011, thus, should be the strain of choice for inclusion in the Indian vaccine formulation in the prevailing field situation. 5 Conclusions The sudden upsurge in the FMD outbreaks due to serotype A virus in 2021, although inexplicable, could possibly be linked with emergence and expansion of a genetically distinct lineage A/ASIA/G-18/2019 within genotype 18 in India. Both BLAST and phylogenetic tree topology confirmed that the lineage is circulating only in India and its eastern neighbour, Bangladesh. The novel lineage seems to have established itself firmly in India apparently superseding all other strains that had been around for about two decades. Historically, genetic lineages of serotype A have been shown to emerge with a concomitant drift in the antigenic spectrum. The isolates of the new lineage showed good antigenic match with the proposed vaccine candidate, A IND 27/2011. The currently used vaccine strain A IND 40/2000, on the other hand, failed to provide optimum antigenic coverage to the field isolates since 2012, including those tested from the new genetic lineage. Therefore, it would be prudent to phase out the vaccine strain A IND 40/2000 being used in the Indian vaccination programme in favour of the alternate strain, A IND 27/2011. Since serotype A viruses originating from the endemic Pool 2 have shown a trend of westward spread up to the Middle East in the past, it is advisable to maintain intensive surveillance to understand the spatio-temporal dynamics of the A/ASIA/G-18/2019 lineage for effecting risk assessment on a continuous basis. CRediT authorship contribution statement JKM and RPS conceived and designed the project. JKM, SSD, MR, JKB, PG and VN performed the laboratory experiments. JKM and SS analysed the data. JKM and SS wrote the original draft of the manuscript. RPS, SSD and JKB reviewed and edited the final manuscript. All authors read and approved the manuscript. Declaration of Competing Interest The authors declare that they have no known competing financialinterestsor personal relationships that could have appeared to influence the work reported in this paper. Data availability Data will be made available on request. Acknowledgements We are thankful to the 10.13039/501100001503 Indian Council of Agricultural Research and Department of Animal Husbandry & Dairying, Govt. of India for providing necessary facilities and funding support. ==== Refs References Bachanek-Bankowska K. Di Nardo A. Wadsworth J. Henry E.K.M. Parlak Ü. Timina A. Mischenko A. Qasim I.A. Abdollahi D. Sultana M. Hossain M.A. King D.P. Knowles N.J. Foot-and-mouth disease in the Middle East caused by an A/ASIA/G-VII virus lineage, 2015-2016 Emerg. Infect. Dis. 24 2018 1073 1078 10.3201/eid2406.170715 29774839 Bae S. Li V. Hong J. Kim J.N. Kim H. Phylogenetic and evolutionary analysis of foot-and-mouth disease virus A/ASIA/Sea-97 lineage Virus Genes 57 2021 443 447 10.1007/s11262-021-01848-7 34260046 Baele G. Lemey P. Bedford T. Rambaut A. Suchard M.A. Alekseyenko A.V. Improving the accuracy of demographic and molecular clock model comparison while accommodating phylogenetic uncertainty Mol. Biol. Evol. 29 2012 2157 2167 10.1093/molbev/mss084 22403239 Bari F.D. Parida S. Tekleghiorghis T. Dekker A. Sangula A. Reeve R. Haydon D.T. Paton D.J. Mahapatra M. Genetic and antigenic characterisation of serotype A FMD viruses from East Africa to select new vaccine strains Vaccine 32 2014 5794 5800 10.1016/j.vaccine.2014.08.033 25171846 Bhattacharya S. Pattnaik B. Venkataramanan R. Development and application of sandwich enzyme-linked immunosorbent assay (ELISA) for type identification of foot-and-mouth disease (FMD) virus in direct field materials Indian J. Anim. Sci. 66 1996 1201 1209 Bielejec F. Rambaut A. Suchard M.A. Lemey P. SPREAD: spatial phylogenetic reconstruction of evolutionary dynamics Bioinforma. Oxf. Engl. 27 2011 2910 2912 10.1093/bioinformatics/btr481 Brito B.P. Rodriguez L.L. Hammond J.M. Pinto J. Perez A.M. Review of the global distribution of foot-and-mouth disease virus from 2007 to 2014 Transbound. Emerg. Dis. 64 2017 316 332 10.1111/tbed.12373 25996568 Das B. Mohapatra J.K. Pande V. Subramaniam S. Sanyal A. Evolution of foot-and-mouth disease virus serotype A capsid coding (P1) region on a timescale of three decades in an endemic context Infect. Genet. Evol. J. Mol. Epidemiol. Evol. Genet. Infect. Dis. 41 2016 36 46 10.1016/j.meegid.2016.03.024 Di Nardo A. Ferretti L. Wadsworth J. Mioulet V. Gelman B. Karniely S. Scherbakov A. Ziay G. Özyörük F. Parlak Ü. Tuncer-Göktuna P. Hassanzadeh R. Khalaj M. Dastoor S.M. Abdollahi D. Khan E.-U.-H. Afzal M. Hussain M. Knowles N.J. King D.P. Evolutionary and ecological drivers shape the emergence and extinction of foot-and-mouth disease virus lineages Mol. Biol. Evol. 38 2021 4346 4361 10.1093/molbev/msab172 34115138 Domingo E. Davila M. Ortín J. Nucleotide sequence heterogeneity of the RNA from a natural population of foot-and-mouth disease virus Gene 11 1980 333 346 6260578 Duffy S. Why are RNA virus mutation rates so damn high? PLoS Biol 16 2018 e3000003 10.1371/journal.pbio.3000003 Giridharan P. Hemadri D. Tosh C. Sanyal A. Bandyopadhyay S.K. Development and evaluation of a multiplex PCR for differentiation of foot-and-mouth disease virus strains native to India J. Virol. Method. 126 2005 1 11 10.1016/j.jviromet.2005.01.015 Grubman M.J. Baxt B. Foot-and-mouth disease Clin. Microbiol. Rev. 17 2004 465 493 10.1128/CMR.17.2.465-493.2004 15084510 Grueber C.E. Wallis G.P. Jamieson I.G. Episodic positive selection in the evolution of avian toll-like receptor innate immunity genes PLoS ONE 9 2014 e89632 10.1371/journal.pone.0089632 24595315 Jamal S.M. Ferrari G. Ahmed S. Normann P. Curry S. Belsham G.J. Evolutionary analysis of serotype A foot-and-mouth disease viruses circulating in Pakistan and Afghanistan during 2002-2009 J. Gen. Virol. 92 2011 2849 2864 10.1099/vir.0.035626-0 21813704 Jangra R.K. Tosh C. Sanyal A. Hemadri D. Bandyopadhyay S.K. Antigenic and genetic analyses of foot-and-mouth disease virus type A isolates for selection of candidate vaccine strain reveals emergence of a variant virus that is responsible for most recent outbreaks in India Virus Res. 112 2005 52 59 10.1016/j.virusres.2005.03.021 16022900 Kitching R.P. Global epidemiology and prospects for control of foot-and-mouth disease Curr. Top. Microbiol. Immunol. 288 2005 133 148 10.1007/3-540-27109-0_6 15648177 Knowles N.J. Samuel A.R. Molecular epidemiology of foot-and-mouth disease virus Virus Res. 91 2003 65 80 10.1016/s0168-1702(02)00260-5 12527438 Knowles, N.J., Samuel, A.R., 1995. Polymerase chain reaction amplification and cycle sequencing of the 1D (VP1) gene of foot and mouth disease viruses. Report of the Session of the Research Group of the Standing Technical Committee of the European Commission for the Control of Foot-and-Mouth Disease held jointly with the FMD Sub group of the Scientific Veterinary Committee of the Commission of the European Community, Mödling, Vienna, Austria, 19-22 September 1994. Rome: FAO, Rome: FAO, 45–53. Knowles N.J. Samuel A.R. Davies P.R. Midgley R.J. Valarcher J.F. Pandemic strain of foot-and-mouth disease virus serotype O Emerg. Infect. Dis. 11 2005 1887 1893 10.3201/eid1112.050908 16485475 Lemey P. Rambaut A. Drummond A.J. Suchard M.A. Bayesian phylogeography finds its roots PLoSComput. Biol. 5 2009 e1000520 10.1371/journal.pcbi.1000520 Ludi A.B. Horton D.L. Li Y. Mahapatra M. King D.P. Knowles N.J. Russell C.A. Paton D.J. Wood J.L.N. Smith D.J. Hammond J.M. Antigenic variation of foot-and-mouth disease virus serotype A J. Gen. Virol. 95 2014 384 392 10.1099/vir.0.057521-0 24187014 Mahapatra M. Statham B. Li Y. Hammond J. Paton D. Parida S. Emergence of antigenic variants within serotype A FMDV in the Middle East with antigenically critical amino acid substitutions Vaccine 34 2016 3199 3206 10.1016/j.vaccine.2016.02.057 27016651 Maradei E. Malirat V. Beascoechea C.P. Benitez E.O. Pedemonte A. Seki C. Novo S.G. Balette C.I. D'Aloia R. La Torre J.L. Mattion N. Toledo J.R. Bergmann I.E. Characterization of a type O foot-and-mouth disease virus re-emerging in the year 2011 in free areas of the Southern Cone of South America and cross-protection studies with the vaccine strain in use in the region Vet. Microbiol. 162 2013 479 490 10.1016/j.vetmic.2012.10.035 23182909 Mattion N. König G. Seki C. Smitsaart E. Maradei E. Robiolo B. Duffy S. León E. Piccone M. Sadir A. Bottini R. Cosentino B. Falczuk A. Maresca R. Periolo O. Bellinzoni R. Espinoza A. Torre J.L. Palma E.L. Reintroduction of foot-and-mouth disease in Argentina: characterisation of the isolates and development of tools for the control and eradication of the disease Vaccine 22 2004 4149 4162 10.1016/j.vaccine.2004.06.040 15474705 Mohapatra J.K. Das B. Rout M. Sreenivasa B.P. Subramaniam S. Sanyal A. Pattnaik B. Alternate vaccine strain selection in the wake of emerging foot-and-mouth disease virus serotype A antigenic variants in India Vaccine 36 2018 3191 3194 10.1016/j.vaccine.2018.04.090 29735325 Mohapatra J.K. Hemadri D. Rao T.V.S. Sreenivasa B.P. Subramaniam S. Sanyal A. Periyasamy T.R. Singh N.K. Pattnaik B. Venkataramanan R. Assessment of suitability of two serotype A candidate vaccine strains for inclusion in FMD vaccine in India Vet. Microbiol. 131 2008 65 72 10.1016/j.vetmic.2008.02.020 18394827 Mohapatra J.K. Subramaniam S. Pandey L.K. Pawar S.S. De A. Das B. Sanyal A. Pattnaik B. Phylogenetic structure of serotype A foot-and-mouth disease virus: global diversity and the Indian perspective J. Gen. Virol. 92 2011 873 879 10.1099/vir.0.028555-0 21228130 Pandey L.K. Mohapatra J.K. Subramaniam S. Sanyal A. Pande V. Pattnaik B. Evolution of serotype A foot-and-mouth disease virus capsid under neutralizing antibody pressure in vitro Virus Res. 181 2014 72 76 10.1016/j.virusres.2014.01.009 24452141 Paton D.J. Sumption K.J. Charleston B. Options for control of foot-and-mouth disease: knowledge, capability and policy Philos. Trans. R. Soc. Lond. B. Biol. Sci. 364 2009 2657 2667 10.1098/rstb.2009.0100 19687036 Pattnaik B. Subramaniam S. Sanyal A. Mohapatra J.K. Dash B.B. Ranjan R. Rout M. Foot-and-mouth disease: global status and future road map for control and prevention in India Agric. Res. 1 2012 132 147 10.1007/s40003-012-0012-z Rweyemamu M.M. Antigenic variation in foot-and-mouth disease: studies based on the virus neutralization reaction J. Biol. Stand. 12 1984 323 337 10.1016/s0092-1157(84)80013-x 6207176 Rweyemamu M.M. Booth J.C. Head M. Pay T.W. Microneutralization tests for serological typing and subtyping of foot-and-mouth disease virus strains J. Hyg. (Lond.) 81 1978 107 123 10.1017/s002217240005381x 211160 Samuel A.R. Knowles N.J. Foot-and-mouth disease type O viruses exhibit genetically and geographically distinct evolutionary lineages (topotypes) J. Gen. Virol. 82 2001 609 621 10.1099/0022-1317-82-3-609 11172103 Seeyo K.B. Nishi T. Kawaguchi R. Ungvanijban S. Udon R. Fukai K. Yamakawa M. Rukkwamsuk T. Evolution of antigenic and genetic characteristics of foot-and-mouth disease virus serotype A circulating in Thailand, 2007-2019 Virus Res. 290 2020 198166 10.1016/j.virusres.2020.198166 Singanallur N.B. Eblé P.L. Ludi A.B. Statham B. Bin-Tarif A. King D.P. Dekker A. Vosloo W. A vaccine based on the A/ASIA/G-VII lineage of foot-and-mouth disease virus offers low levels of protection against circulating viruses from the A/ASIA/Iran-05 lineage Viruses 14 2022 97 10.3390/v14010097 35062300 Subramaniam S. Mohapatra J.K. Sahoo N.R. Sahoo A.P. Dahiya S.S. Rout M. Biswal J.K. Ashok K.S. Mallick S. Ranjan R. Jana C. Singh R.P. Foot-and-mouth disease status in India during the second decade of the twenty-first century (2011-2020) Vet. Res. Commun. 46 2022 1011 1022 10.1007/s11259-022-10010-z 36190601 Suchard M.A. Lemey P. Baele G. Ayres D.L. Drummond A.J. Rambaut A. Bayesian phylogenetic and phylodynamic data integration using BEAST 1.10 Virus Evol. 4 2018 10.1093/ve/vey016 vey016 Tamura K. Stecher G. Kumar S. MEGA11: molecular evolutionary genetics analysis version 11 Mol. Biol. Evol. 38 2021 3022 3027 10.1093/molbev/msab120 33892491 Tosh C. Hemadri D. Sanyal A. Bandyopadhyay S.K. Genetic and antigenic analysis of two recently circulating genotypes of type A foot-and-mouth disease virus in India: evidence for positive selection in the capsid-coding genes Arch. Virol. 148 2003 853 869 10.1007/s00705-002-0968-2 12721795 Tully D.C. Fares M.A. The tale of a modern animal plague: tracing the evolutionary history and determining the time-scale for foot and mouth disease virus Virology 382 2008 250 256 10.1016/j.virol.2008.09.011 18945462 Valarcher J.F. Knowles N.J. Zakharov V. Scherbakov A. Zhang Z. Shang Y.J. Liu Z.X. Liu X.T. Sanyal A. Hemadri D. Tosh C. Rasool T.J. Pattnaik B. Schumann K.R. Beckham T.R. Linchongsubongkoch W. Ferris N.P. Roeder P.L. Paton D.J. Multiple origins of foot-and-mouth disease virus serotype Asia 1 outbreaks, 2003-2007 Emerg. Infect. Dis. 15 2009 1046 1051 10.3201/eid1507.081621 19624919 Waters R. Ludi A.B. Fowler V.L. Wilsden G. Browning C. Gubbins S. Statham B. Bin-Tarif A. Mioulet V. King D.J. Colenutt C. Brown E. Hudelet P. King D.P. Efficacy of a high-potency multivalent foot-and-mouth disease virus vaccine in cattle against heterologous challenge with a field virus from the emerging A/ASIA/G-VII lineage Vaccine 36 2018 1901 1907 10.1016/j.vaccine.2018.02.016 29506922 Weaver S. Shank S.D. Spielman S.J. Li M. Muse S.V. Kosakovsky Pond S.L. Datamonkey 2.0: a modern web application for characterizing selective and other evolutionary processes Mol. Biol. Evol. 35 2018 773 777 10.1093/molbev/msx335 29301006 Wekesa S.N. Sangula A.K. Belsham G.J. Muwanika V.B. Heller R. Balinda S.N. Masembe C. Siegismund H.R. Genetic diversity of serotype A foot-and-mouth disease viruses in Kenya from 1964 to 2013; implications for control strategies in eastern Africa Infect. Genet. Evol. J. Mol. Epidemiol. Evol. Genet. Infect. Dis. 21 2014 408 417 10.1016/j.meegid.2013.12.006
PMC010xxxxxx/PMC10352719.txt
==== Front Virus Res Virus Res Virus Research 0168-1702 1872-7492 Elsevier S0168-1702(23)00106-5 10.1016/j.virusres.2023.199144 199144 Article Sentinel plot surveillance of cotton leaf curl disease in Pakistan- a case study at the cultivated cotton-wild host plant interface Iqbal Muhammad Javed ab$ Zia-Ur-Rehman Muhammad b$ Ilyas Muhammad a$ Hameed Usman b Herrmann Hans Werner a Chingandu Nomatter a Manzoor Muhammad Tariq b Haider Muhammad Saleem b Brown Judith K. [email protected] a⁎ a School of Plant Sciences, The University of Arizona, 1140 E South Campus Drive, Tucson, AZ 85721 USA b Faculty of Agricultural Sciences, University of the Punjab, New Campus Canal Road Lahore, Pakistan ⁎ Corresponding author: Forbes Building, RM 431, 1140 E South Campus drive, Tucson, AZ 85721 USA. [email protected] $ Each primary author contributed equally but differently. 07 6 2023 8 2023 07 6 2023 333 1991447 4 2023 31 5 2023 1 6 2023 © 2023 The Author(s) 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Highlights • Sentinel plot and commercial field monitoring reveals recent shifts in CLCuD complexes in Pakistan. • Evidence of altered prevalence of CLCuD-helper virus genomic variability. • Previously unreported CLCuD-associated plant host species discovered in cotton-vegetable and uncultivated species. • Early-detection predicted re-emergence of CLCuD-associated, cotton leaf curl Multan virus-Rajasthan. A sentinel plot case study was carried out to identify and map the distribution of begomovirus-betasatellite complexes in sentinel plots and commercial cotton fields over a four-year period using molecular and high-throughput DNA ‘discovery’ sequencing approaches. Samples were collected from 15 study sites in the two major cotton-producing areas of Pakistan. Whitefly- and leafhopper-transmitted geminiviruses were detected in previously unreported host plant species and locations. The most prevalent begomovirus was cotton leaf curl Kokhran virus-Burewala (CLCuKoV-Bu). Unexpectedly, a recently recognized recombinant, cotton leaf curl Multan virus-Rajasthan (CLCuMuV-Ra) was prevalent in five of 15 sites. cotton leaf curl Alabad virus (CLCuAlV) and cotton leaf curl Kokhran virus-Kokhran, ‘core’ members of CLCuD-begomoviruses that co-occurred with CLCuMuV in the ‘Multan’ epidemic were detected in one of 15 sentinel plots. Also identified were chickpea chlorotic dwarf virus and ‘non-core’ CLCuD-begomoviruses, okra enation leaf curl virus, squash leaf curl virus, and tomato leaf curl New Delhi virus. Cotton leaf curl Multan betasatellite (CLCuMuB) was the most prevalent CLCuD-betasatellite, and less commonly, two ‘non-core’ betasatellites. Recombination analysis revealed previously uncharacterized recombinants among helper virus-betasatellite complexes consisting of CLCuKoV, CLCuMuV, CLCuAlV and CLCuMuB. Population analyses provided early evidence for CLCuMuV-Ra expansion and displacement of CLCuKoV-Bu in India and Pakistan from 2012-2017. Identification of ‘core’ and non-core CLCuD-species/strains in cotton and other potential reservoirs, and presence of the now predominant CLCuMuV-Ra strain are indicative of ongoing diversification. Investigating the phylodynamics of geminivirus emergence in cotton-vegetable cropping systems offers an opportunity to understand the driving forces underlying disease outbreaks and reconcile viral evolution with epidemiological relationships that also capture pathogen population shifts. Keywords begomoviruses cotton leaf curl disease Geminiviridae sentinel plot case study whitefly vector ==== Body pmc1 Introduction Cotton is the most important fiber and oil crop worldwide. In South Asia, commercial cotton production is at risk for infection by cotton leaf curl disease (CLCuD), annually, since the initial outbreak in the early 1990s (Briddon & Markham, 2000). Cotton leaf curl disease is caused by one or more begomovirus species or strains, referred to as the ‘core’ CLCuD complex that consists of five begomoviruses and their strains, belonging to the genus, Begomovirus (family, Geminiviridae) (Brown, 2020). The genus Begomovirus contains the greatest number of species (>400) within the Geminiviridae (https://talk.ictvonline.org/taxonomy/). Worldwide, begomoviruses are transmitted by members of the whitefly Bemisia tabaci (Genn.) cryptic species group (de Moya et al., 2019) in a circulative, non-propagative manner (Brown & Czosnek, 2002). Begomoviruses have a circular DNA genome encapsidated in a twinned, icosahedral particle (Walker et al., 2021), and based on genome components, they are divided into two types, referred to as bipartite and monopartite. Most bipartite begomoviruses, or those that have two genomic components referred to as the DNA-A and DNA-B component, each of 2.5–2.7 kb in size, respectively, occur in the western hemisphere. The bipartite DNA-A component encodes the coat protein (CP) on the virion sense strand, with some species also encoding a pre-coat (AV2), while the replication-associated protein (Rep), transcriptional activator protein (TrAP), replication enhancer protein (REn), and AC4 protein (when present) are encoded on the complementary sense strand. The DNA-B component encodes the nuclear shuttle protein (NSP) and movement protein (MP) on the virion and complementary sense strands, respectively (Mansoor et al., 1999). In Asia, most begomoviruses have a monopartite genome, or DNA-A of ∼2.7-2.8 kb in size, and for the most part, the coding and non-coding regions of the monopartite genome are comparable to bipartite begomoviruses. In addition, most monopartite viruses are recognized as ‘helper viruses’ on which a type of non-viral molecule relies for replication, movement, encapsidation and whitefly transmission (Briddon & Stanley, 2006), referred to as betasatellites. Betasatellites have an adenine (A) rich region, a βC1 gene, and a highly conserved sequence of about 235 nucleotides known as the satellite conserved region (SCR). One or more helper begomoviruses can serve as the casual virus of cotton leaf curl disease, aided by at least one betasatellite (Mansoor et al., 2003a). Symptoms of CLCuD are manifest as foliar- and vein-thickening, curling, shortening of internodes, development of enations (outgrowths) on the underside of the leaf, and overall stunting of the plant. The first major outbreak occurred in 1989-1990, when high incidences of leaf curl disease symptoms were observed throughout the Punjab Province, Pakistan. The initial CLCuD outbreak in Kokhran, a township near Multan from where it spread following widespread adoption of S-12, a cotton variety found to be highly susceptible to colonization by the whitefly vector, Bemisia tabaci (Genn.) cryptic species complex and CLCuD-begomoviruses (Zubair et al., 2017). The causal agent of the 1990’s ‘Multan’ epidemic was identified as cotton leaf curl Multan virus (CLCuMuV), now referred to as the CLCuMuV–Faisalabad strain (CLCuMuV-Fai), and its associated cotton leaf curl Multan betasatellite (CLCuMuB) (Mansoor et al., 2003a). The CLCuMuV-Fai reached epidemic proportions and was spread by the whitefly vector to the major cotton-growing areas of Pakistan during the next several years. During the late 1990s, CLCuMuV-Fai-resistant commercial cotton varieties were released, and fiber production was restored to near pre-epidemic levels (Rahman et al., 2002). However, during 2001 the CLCuMuV-Fai-resistant varieties developed symptoms of leaf curl disease, initially in cotton fields in the tehsil of Burewala (Mansoor, Amin, et al., 2003). The causal agent was identified as a previously uncharacterized recombinant, named CLCuKoV-Burewala (CLCuKoV-Bu) and a previously unidentified-associated recombinant CLCuMuB (Amrao et al., 2010). The CLCuKoV-Bu and CLCuMuB complex spread throughout Punjab Province, displacing CLCuMuV, and becoming the predominant leaf curl strain associated with leaf curl disease in cotton throughout Pakistan, and in the nearby Rajasthan and Haryana states and in the Punjab region of India by 2012 (Rajagopalan et al., 2012). Previous studies have identified begomoviruses from cotton plants exhibiting leaf curl symptoms in Pakistan and India, including cotton leaf curl Alabad virus (CLCuAlV), cotton leaf curl Kokhran virus- Kokhran (CLCuKoV-Ko), cotton leaf curl Multan virus- Rajasthan (CLCuMuV-Ra), papaya leaf curl virus (PaLCuV), and tomato leaf curl Bangalore virus (ToLCBaV) (Briddon & Markham, 2000; Mansoor et al., 2003a; Zhou et al., 1998) and a satellite-like molecule cotton leaf curl Multan betasatellite (CLCuMuB) (Briddon et al., 2003, 2014). Unexpectedly, geminiviruses previously considered to restricted to vegetable crop hosts, have been identified from cotton exhibiting leaf curl disease. Among them, tomato leaf curl New Delhi virus (ToLCNDV) and chickpea chlorotic dwarf virus were detected in mixed infection with CLCuKoV-Bu in cotton plants (Manzoor et al., 2014; Zaidi et al., 2016). The genomic variability, geographical- and host-distribution, and the phylodynamics of geminivirus-betasatellite complexes was investigated in cotton-vegetable agroecosystems in the major cotton producing areas in Pakistan. Leaf samples of suspect and previously recognized plant hosts were collected from cotton, vegetable, ornamental, and uncultivated or ‘wild’ plants/weeds during sentinel plot and field surveys. Fifteen study sites in the Punjab and Sindh Provinces with a prior history of leaf curl epidemics were sampled annually, June to September. Also, symptomatic leaf samples were collected from commercial cotton fields in 2017 during a routine survey in Vehari (Punjab Province). Full-length geminivirus genome (n=236) and betasatellite (n=411) sequences were determined from symptomatic cotton and non-cotton samples. Their relative frequency and distribution were compared with results of previous CLCuD-surveys conducted from ∼1990-onward, the timeframe during which two historical outbreaks reached epidemic proportion in Pakistan, caused by ‘Multan’ (CLCuMuV-Fai) and ‘Burewala’ (CLCuKoV-Bu), respectively. Based on preliminary reports of a suspected, impending third outbreak species detected at low frequency in cotton fields in 2015 (Zubair et al., 2017), samples of symptomatic cotton leaves were collected in the vicinity of Vehari (Punjab Province) in 2017. Sequencing results revealed the prevalence of CLCuMuV-Ra, a strain now recognized as the causal begomovirus associated with the third cotton leaf curl disease in Pakistan and in some locations in India (Ahmed et al., 2021). 2 Materials and Methods 2.1 Cultivation and maintenance of sentinel plots Sentinel plots (276 × 236 ft) were established and sampled from 2010 to 2014 at study sites located in Lahore, Multan, and Vehari. The main cotton-producing regions in Pakistan are near Multan and Vehari, whereas, in Lahore, cotton is cultivated on a smaller scale. Known or suspect begomovirus plant hosts were planted in sentinel plots to bell pepper, chili, cotton, cucumber, okra, sponge gourd (Luffa cylindrica), squash, and tomato. Selected cotton varieties, or ‘differentials’ that are known to be differentially susceptible to cotton leaf curl disease complex species and strains, respectively. These observations are on official reporting from official government records, the plant breeding community, and are acknowledged by farmers. The differentials selected for planting in sentinel plots were: MNH886, MNH814, MNH786 and CIM496. In addition, uncultivated, or ‘wild’ plant species (Table S1-S3) growing in the vicinity of sentinel plots were encouraged to flourish within or nearby the sentinel plots to attract natural whitefly infestation and recruit cotton leaf curl disease-associated as well as other prospective plant viruses in the environment. The sentinel plot locations were selected as the study sites to represent different geographical locations within the districts of the two most important provinces Punjab and Sindh, where cotton is produced in Pakistan. 2.2 Sample collection and nucleic acid isolation Leaf samples exhibiting symptoms reminiscent of leaf curl disease (Figure S1) were collected from the sentinel plots established for the project and from commercial cotton fields. Samples were collected early and late during the regular cotton-growing season, beginning when the cotton ‘differentials’ planted in sentinel plots reached the 12-14 leaf stage. Sentinel plots and commercial cotton fields were surveyed for symptoms characteristic of begomoviral infection. Samples were collected from sentinel plots established in Vehari, Multan and Lahore planted to cotton, vegetable crops, and wild plant species (uncultivated/weeds), cotton-breeding plots located in Vehari, Multan and Sakrand, and commercial cotton fields in the Punjab and Sindh Provinces. Commercial cotton fields in the Punjab province were located in Bahawalnagar, Bahawalpur, Burewala, Dera Ghazi Khan, Faisalabad, Layyah, Lodhran, Multan, Muzaffargarh, Rahimyar Khan, Sahiwal, and Vehari, whereas the fields in the Sindh Province were located in Ghotki and Sakrand. During 2017, symptomatic cotton leaf samples were collected from several commercial cotton fields near Vehari, Punjab province. Newly developing leaves of cotton plants exhibiting characteristic leaf curl disease symptoms were collected from cotton, vegetables, and alternate hosts growing near cotton fields. Leaf samples were transported in an ice-chest to the Faculty of Agricultural Sciences (University of the Punjab, Lahore) and stored at -20°C or -80°C. Total nucleic acids were isolated from 100 mg of leaf tissue per plant sample using the cetyltrimethylammonium bromide (CTAB) method (Doyle and Doyle, 1990). Purified DNA was shipped by international courier (under APHIS PPQ permit to J.K.B.) to the School of Plant Sciences (The University of Arizona, Tucson, USA). To identify the most prevalent geminiviruses and betasatellites, DNA were analyzed by rolling circle amplification (RCA), polymerase chain reaction (PCR) amplification, cloning, and Sanger sequencing, and/or by high throughput ‘discovery’ DNA sequencing (HTS). 2.3 DNA sequencing, assembly and geminivirus-betasatellite identification Total nucleic acids were subjected to rolling-circle amplification (RCA) of circular DNA using the Illustra TempliPhi Amplification Kit (GE Healthcare Life Sciences). The enriched DNA was digested with selected restriction enzymes, EcoRI, HindIII, KpnI, NcoI, PstI, SacI, SalI, XbaI, and XhoI (ThermoFisher Scientific) to obtain a ∼2.8 kilobase pair (kbp) for geminiviral genomes and ∼1.4 kbp for satellite molecules. A fragment of the expected size, corresponding to a full-length geminiviral genome (2.8 kbp) or betasatellite molecule (half unit length, 1.4 kbp), were cloned into the plasmid vector pGEM TM-3ZF(+) (ThermoFisher Scientific), pre-digested with the respective enzyme compatible for the cloned insert. Plasmids containing an insert of the expected size were subjected to capillary DNA (Sanger) sequencing at The University of Arizona Genetics Core (UAGC) facility (Tucson, USA). The RCA-enriched DNA samples that did not yield a digested betasatellite product were used as template for PCR amplification (Brown et al., 2017) with degenerate primers (Zia-Ur-Rehman et al., 2013). The amplicons were cloned into the pGEM-T Easy plasmid vector (Promega) and subjected to capillary DNA (Sanger) sequencing. Provisional identification was determined by BLASTn (https://blast.ncbi.nlm.nih.gov/Blast.cgi) analysis (similarity) of sequences available in the GenBank database. To establish begomovirus and betasatellite species identity, pairwise distance analysis was carried out using the SDT v1.2 algorithm (Muhire et al., 2014). Further, samples were subjected to ‘discovery’ DNA sequencing (n=∼285 samples) with the Illumina Hi-Seq platform. Library construction, quantification, and sequencing were carried out at the University of Arizona Genetic Core, Tucson, USA. The nucleotide reads were assembled de novo with SPAdes (Bankevich et al., 2012) and NCBI BLASTn, respectively. Assembly of geminivirus- and betasatellite reads was carried out using the de novo assembly software SOAPdenovo2 (Luo et al., 2012). Provisional identification was determined using BLASTn against the sequences available in the GenBank database, as described above. Representative full-length geminiviral-betasatellite sequences were deposited in the NCBI GenBank database. The Accession numbers are provided in Table S1. The nomenclature adheres to most recent guidelines published by the Geminiviridae Study Group, International Committee on Taxonomy of Viruses (ICTV) (Zerbini et al., 2022) 2.4 Population expansion analysis of cotton leaf curl viruses Population expansion analysis of full-length begomoviral genome sequences was carried out using the Tajima's D (Tajima, 1989), Fu, and Li's tests (Fu & Li, 1993). Full-length sequences were aligned with the MUSCLE algorithm (Edgar, 2004) and analyzed individually with DNASP Tajima's D, Fu and Li's F, and Fu and Li's D algorithms, respectively (Simonsen et al., 1995). 2.5 Phylogenetic analysis Phylogenetic analyses were carried out by comparing sequences determined in this study with representative core and non-core geminivirus and betasatellite references available in the NCBI GenBank database. Representative begomovirus and betasatellite reference sequences were selected from well-known isolates representing prevalent and rare species and strains (Brown lab curated database). Reference sequences for previously identified recombinants of CLCuMuB were selected based on historical descriptions (Akhtar et al., 2014; Amin et al., 2006; Zubair et al., 2017). The species and strain identifications of helper virus and betasatellite sequences reported here were assigned based on the results of pairwise nucleotide identity analysis using SDT software (Muhire et al., 2014), and in consultation with the 2021 ICTV master species list, version 3 (https://ictv.global/msl). Sequences were aligned using MUSCLE (Edgar, 2004) in Geneious Prime (v2021.2.2) software (https://www.geneious.com). Phylogenetic analysis was carried out using the Maximum likelihood (ML) method in RAxML-HPC2 using the XSEDE command in the CIPRES portal https://www.phylo.org/) (Miller et al., 2010) and the trees was reconstructed using the GTRGAMMA phase model, with 1000 bootstrap iterations. When many sequences shared high nucleotide identity based on SDT analysis, the groups that clustered in the same clade or sub-clade were collapsed to enable visualization of the rarer relationships on the tree. The ML trees were drawn with FigTree (http://tree.bio.ed.ac.uk/software/figtree/) and manual editing was carried out in Inkscape (http://www.download3.co/ic2/inkscape/index.php?kw=inkscape%20software). 2.6 Recombination analysis The full-length genome sequences of CLCuAlV (n=29), CLCuKoV (n=382), and CLCuMuV (n=163) were aligned (MUSCLE) with representative begomoviral taxa as reference sequences (GenBank) and subjected to recombination analysis using RDP software, version 5.05 that consists of the software suite: Geneconv, Bootscan, Maximum Chi Square, Chimaera, Sister Scan and 3Seq algorithms (Martin et al., 2021). Results were considered statistically significant, for P-values lower than the Bonferroni-corrected α=0.05 cut-off. Similarly, recombination analysis of betasatellite sequences was carried out using representative betasatellite as references (GenBank). Only predicted recombination events that were statistically supported by all seven algorithms are reported here. 3 Results 3.1 Cotton leaf curl Kokhran virus-Burewala and cotton leaf curl Multan betasatellite prevalence Commercial cotton fields in 15 study sites in the Punjab and Sindh Provinces were surveyed for cotton leaf curl disease symptoms to guide sample collection. The results of a routine follow-up 2017 sampling of commercial cotton fields in Vehari are also reported here. Both full-length and partial sequences of begomovirus and betasatellite genomes resulted from PCR amplification and sanger sequencing of the cloned amplicons, and/or those determined from Illumina ‘discovery’ sequencing were considered in the BLASTn analysis conducted to establish provisional identification and are reported as prevalence data archived in the project database (available on request). Only full-length geminivirus and betasatellite sequences were considered in the pairwise distance, phylogenetic, and recombination analyses reported here (Fig. 1). The sequences for isolates occurring in symptomatic cotton plants were assembled into 236 full-length genomes. Virus isolates for which full-length genome sequences were determined were identified to species and strain by pairwise percentage nucleotide identity analysis using the SDT algorithm and representative reference sequences available in GenBank (Brown et al., 2015; Muhire et al., 2014).Fig. 1 Geographical distribution of cotton-infecting geminiviruses identified in cotton plants sampled from sentinel plots and commercial cotton fields at the study sites in Pakistan during 2010-2014 and from Sahiwal during 2017. The virus sequences were determined by cloning and sequencing, followed by Sanger sequencing and/or by Illumina DNA sequencing. Isolates are represented by full-length sequences (n=236) of different species and strains. Viral species were identified as chickpea chlorotic dwarf virus (CpCDV), chili leaf curl virus (ChLCV), cotton leaf curl Alabad virus (CLCuAlV), cotton leaf curl Multan virus (CLCuMuV), cotton leaf curl Kokhran virus (CLCuKoV), squash leaf curl virus (SLCV), and tomato leaf curl new Delhi virus (ToLCNDV). The different viral species are color-coded by species and the number of sequences per species in each geographical location is indicated parenthetically. Fig. 1: The CLCuKoV-Bu was identified as the most predominant cotton-infecting species identified, occurring at all 15 study sites for the four-year study (Fig. 1). Somewhat unexpectedly, CLCuMuV, the primary causal agent of the Multan epidemic in Pakistan (1990’s) was identified in cotton plants and in several other host species at five of the fifteen study sites, in Multan and Vehari districts of the Punjab Province, and the Ghotki and Sakrand districts in the Sindh Province. Several other begomoviruses, CLCuKoV-Ko in Sahiwal (Punjab) and CLCuAlV in Multan (Punjab) that occurred at low prevalence during the Multan epidemic era were also detected in this study (Fig. 1). Finally, follow-on sampling in 2017 from commercial cotton fields in Sahiwal (Punjab) near Vehari district revealed the occurrence of CLCuMuV-Rajasthan strain(CLCuMuV-Ra), first reported infecting cotton plants at a low frequency in Pakistan and India in 2000s (Kirthi et al., 2004; Shahid et al., 2007). In this study, several geminiviruses previously unassociated with leaf curl disease in cotton were identified for the first time, and included the leaf hopper transmitted mastrevirus, chickpea chlorotic dwarf virus (CpCDV) and the bipartite begomovirus, ToLCNDV, both which have been previously, widely reported from different vegetable crop species (Table S3) (Manzoor et al., 2014; Zubair et al., 2017). Further, several geminiviruses were identified infecting cotton and in previously unreported locations for the first time. The ToLCNDV was detected in cotton plants in Burewala, Layyah, Multan, and Vehari, while CpCDV was identified in Multan and Rahim Yar Khan (Fig. 1), and the New World squash leaf curl virus (SLCV), a bipartite begomovirus endemic to the southwestern U.S and Mexico, was detected in cotton plants in Multan and Layyah. A number of vegetable and weed species have been reported as CLCuKoV and CLCuMuV hosts in Pakistan. In this study, CLCuAlV was detected from okra plants for the first time from a commercial field (Bahawalpur) and the sentinel plot in Burewala (Table S3). And in Sahiwal, the CLCuKoV-Ko, was detected in symptomatic cotton plants, which has not been reported since the ‘Multan’ epidemic, underscoring the persistence of this historical isolate in a major cotton-growing district in the Punjab Province. Finally, diverse betasatellite species and variants were detected in symptomatic cotton plants in this study (Fig. 2) and included cotton leaf curl Multan betasatellite (CLCuMuB) (15 of 15 locations), bhendi yellow vein mosaic betasatellite (BYVMB) (Multan, Lahore, Vehari, Sakrand) and chili leaf curl betasatellite (ChLCuB) (Vehari) co-occurred in cotton with previously recognized ‘core’ CLCuD begomoviruses (Table S1). The full-length geminivirus helper genome and representative betasatellite sequences were deposited in the NCBI GenBank database. Sample information, isolate identification, and the GenBank Accession number for each sequence are provided (Table S1-S2).Fig. 2 Geographical distribution of betasatellites (n=411; full-length sequences) identified in cotton samples from sentinel plots and commercial cotton fields at study sites in Pakistan during 2010-2014, and from Sahiwal during 2017. Betasatellite identification was based on alignment of PCR-amplified, cloned virus/betasatellite sequences (Sanger or Illumina sequencing) with selected GenBank reference sequences. The betasatellites were identified as bhendi yellow vein mosaic betasatellite (BYVB), chilli leaf curl betasatellite (ChLCuB), and cotton leaf curl Multan betasatellite (CLCuMuB), and are color-coded by species, with the number of sequences per species by geographical location indicated parenthetically. Fig. 2: 3.2 Viruses associated with previously or recently unreported symptomatic host plant species Identification of geminivirus species associated with symptomatic sentinel plot plant species revealed the frequent detection of CLCuKoV-Bu in cotton differentials established in sentinel plots to recruit begomovirus species from the ‘environment’ (Tables 1, 3). During the four-year sampling period, among the 105 isolates detected, 94 (89%) were identified as CLCuKoV-Bu, whereas the minor species identified were CLCuAlV, OELCuV and ToLCNDV (Table 1).Table 1 Begomoviruses detected in leaf curl cotton differentials in sentinel plots located in Lahore, Multan, and Vehari, were identified as cotton leaf curl Alabad virus (CLCuAlV), cotton leaf curl Kokhran virus-Burewala (CLCuKoV-Bu), okra enation leaf curl virus (OELCuV), and tomato leaf curl New Delhi virus (ToLCNDV). The number of isolates identified, per host species, is indicated parenthetically. Table 1:Plant host Location Virus Cotton Lahore CLCuKoV-Bu (4) Cotton Lahore CLCuKoV-Bu (9), OELCuV (1) Cotton Multan CLCuKoV-Bu (10), ToLCNDV (1) Cotton Multan CLCuKoV-Bu (4), CLCuAlV (1) Cotton Multan CLCuKoV-Bu (5), ToLCNDV (1) Cotton Multan CLCuKoV-Bu (8) Cotton Vehari CLCuKoV-Bu (11), ToLCNDV (1), OELCuV (1) Cotton Vehari CLCuKoV-Bu (2) Cotton Vehari CLCuKoV-Bu (37) Cotton Vehari CLCuKoV-Bu (4) Several begomoviruses and/or a mastrevirus were detected in sentinel plot plants at a relatively low frequency, respectively (Tables 2, 4), and were identified as: CLCuKoV-Bu in L. cylindrica (Zia-Ur-Rehman et al., 2013), squash, and cucumber, CpCDV from tomato (Zia-Ur-Rehman et al., 2015), okra (Zia-Ur-Rehman et al., 2017) and cucumber (Hameed et al., 2017), okra enation leaf curl virus (OELCuV) from cotton (Hameed et al., 2014), CLCuAlV from cotton and okra, and ToLCNDV from cotton and vegetable crops.Table 2 The begomoviruses detected in vegetable and weed species, in or nearby sentinel plots in Lahore, Multan, and Vehari, were identified as cotton leaf curl Alabad virus (CLCuAlV), cotton leaf curl Kokhran virus-Burewala (CLCuKoV-Bu). The number of isolates representing the different viruses is indicated parenthetically. Table 2Virus Location Host CLCuKoV-Bu Lahore Okra (1) CLCuKoV-Bu Vehari Okra (1), Luffa (1) CLCuKoV-Bu Vehari Squash (1), Cucumber (1) CLCuAlV Multan Okra (1) CLCuAlV Multan Okra (1) CLCuAlV Multan Okra (1) CLCuAlV Vehari (Burewala) Okra (1) Twenty-four complete genome sequences were determined for isolates from symptomatic commercial cotton fields sampled in Multan, Vehari and Sakrand, and identified as CLCuAlV, CLCuKoV-Bu, SLCV, ToLCNDV, whereas one isolate was identified as the mastrevirus, CpCDV (Table 3).Table 3 Viruses identified in commercial symptomatic cotton fields in Multan, Sakrand, and Vehari, were identified as chickpea chlorotic dwarf virus (CpCDV), cotton leaf curl Alabad virus (CLCuAlV), cotton leaf curl Kokhran virus-Burewala (CLCuKoV-Bu), cotton leaf curl Kokhran virus-Shahdadpur (CLCuKoV-Sha), cotton leaf curl Multan virus-Rajasthan (CLCuMuV-Ra), squash leaf curl virus (SLCV), and tomato leaf curl New Delhi virus (ToLCNDV). Table 3Host Location Virus Cotton Vehari CLCuKoV-Bu (1) Cotton Vehari CLCuKoV-Bu (6), CLCuMuV-Ra (3), CpCDV (1) Cotton Sakrand CLCuKoV-Bu (1), CLCuMuV-Ra (2) Cotton Sakrand CLCuKoV-Bu(1),CLCuKoV-Sh(1),CLCuMuV-Ra(2), ToLCNDV (1) Cotton Multan CLCuAlV (3), SLCV (1) Cotton Multan CLCuKoV-Bu (1) 3.3 Mixed geminivirus infections Geminiviruses commonly occur in mixed infections in the same host plant, an observation that suggests they do not cross-protect against one another. Exploiting this unusual feature among plant viruses has facilitated inter-specific, and intra-specific recombination as well as recombination between different genera, which in turn, promotes geminivirus diversification and adaptation to new environmental challenges (Rybicki, 1994). In this study, mixed geminivirus infections were detected routinely in sentinel plot and commercial cotton fields (Table 4) in bitter gourd, chili, cotton, cucumber, and okra plants. In the sentinel plots CLCuAlV and ToLCNDV were co-occurring in bitter gourd plants, while chili leaf curl virus (ChLCuV) and ToLCNDV were detected in chili plants. The CLCuKoV-Bu, ToLCNDV, and cherry tomato leaf curl virus (CToLCV) were identified in cotton and a new strain of PaLCV was discovered for the first time (Table S3). Also, in cucumber plants CLCuKoV-Bu, squash leaf curl China virus (SLCCNV), and ToLCNDV were co-existing, while bhendi yellow vein mosaic virus (BYVMV), CLCuAlV, and OELCuV occurred in mixed infection in okra plants. Similarly, in commercial cotton fields, mixed infections were common, and involved CLCuKoV-Bu and ToLCNDV, or CLCuKoV-Bu, SLCV, or CLCuAlV, or CLCuMuV with either BYVMV, CpCDV, OELCuV, pedilanthus leaf curl virus (PeLCV) (proposed new strain of PaLCV) Table S3), and ToLCNDV, (partial sequences, not shown). Finally, mixed infection was detected in commercial okra fields consisting of BYVMV and/or OELCuV with ToLCNDV.Table 4 The geminiviruses identified in mixed infections in cotton, vegetable, and weed species in or near the established sentinel plots and commercial cotton field sampled were bhendi yellow vein mosaic virus (BYVMV), cherry tomato leaf curl virus (CToLCV), chickpea chlorotic dwarf virus (CpCDV), chili leaf curl virus (ChLCuV), cotton leaf curl Alabad virus (CLCuAlV), cotton leaf curl Kokhran virus-Burewala (CLCuKoV-Bu), cotton leaf curl Kokhran virus-Shahdadpur (CLCuKoV-Sha), cotton leaf curl Multan virus-Rajasthan (CLCuMuV-Ra), okra enation leaf curl virus (OELCuV), pedilanthus leaf curl virus (PeLCV), squash leaf curl virus (SLCV), squash leaf curl China virus (SLCCNV), tomato leaf curl New Delhi virus (ToLCNDV), and okra enation leaf curl virus (OELCuV). Table 4Host Location Sample Type Virus Okra Multan Sentinel CLCuAlV, OELCuV Okra Lahore Sentinel BYVMV, OELCuV Cucumber Burewala Sentinel SLCCNV, ToLCNDV Cucumber Vehari Sentinel CLCuKoV-Bu, ToLCNDV Okra Burewala Sentinel BYVMV, CLCuKoV-Bu, CLCuAlV Okra Lahore Sentinel CLCuKoV-Bu, BYVMV Chili Multan Sentinel ChLCuV, ToLCNDV Bitter gourd Multan Sentinel CLCuAlV, ToLCNDV Cotton Multan Sentinel CLCuKoV-Bu, ToLCNDV, CToLCV Cotton Layyah Field CLCuKoV-Bu, SLCV Cotton Multan Field CLCuKoV-Bu, ToLCNDV Cotton Multan Field CpCDV, ToLCNDV Cotton Vehari Field CLCuKoV-Bu, ToLCNDV Cotton Multan Field CLCuKoV-Bu, ToLCNDV Okra Arifwala Field CLCuAlV, BYVMV Okra Bahawalpur Field CLCuAlV, OELCuV, ToLCNDV Duranta repens Lahore Field PeLCV, ToLCNDV, CLCuAlV Cotton Vehari Germplasm CLCuAlV, CpCDV Cotton Sakrand Germplasm CLCuKoV-Bu, CLCuMuV-Ra 3.4 Comparison of the host range of cotton leaf curl complex identified in this study with previously published records A literature search of all previously reported hosts of CLCuD revealed that in addition to cotton, geminivirus-betasatellite complexes have been associated with diverse vegetable crop and wild plant species, with the potential to serve as persistent or transient disease reservoirs (Table S3). Among the plant species sampled in this study, cotton harbored the greatest number of geminiviruses, which consisted of BGYMV, CLCuAlV, CLCuKoV-Bu, CLCuKoV-Ko, CLCuMuV-Dar, CLCuMuV-Ra, CLCuMuV-Sha, CpCDV, hollyhock leaf curl virus (HoLCV), OELCuV, pedilanthus yellow leaf curl virus (PeYLCV), and SLCV. In contrast, okra plants harbored the second largest number of geminiviruses, which consisted of BYVMV, CLCuAlV, CLCuKoV-Bu, and OELCuV. All of the latter viruses have been previously reported from cotton and okra (Table S3). The most prevalent begomovirus detected from cotton and vegetable crops was CLCuKoV, which was detected from symptomatic chili, cotton, cucumber, holly hock, luffa, okra, papaya, soybean, Ricinus sp., squash, Xanthium sp. bean and several ornamental plants (Table S3). Interestingly, the previously recognized CLCuMuV-Pakistan (CLCuMuV-Pk) and CLCuMuV-Fai strains were prevalent in cotton and Sida sp. plants, whereas, the CLCuAlV species was detected exclusively in cotton and okra plants, and the CLCuKoV-Ko species was found in one cotton sample. Among the geminiviruses reported in this survey, ToLCNDV exhibited the most extensive host range. The ToLCNDV host plants consisted primarily of bitter gourd, chili, cotton, squash, ornamentals, and uncultivated (wild) species (partial sequences were not included in the phylogenetic or SDT analyses) (Table S3). Also, several plant species identified as begomovirus hosts in this study, have previously been reported (Table S3). However, several previously unreported species were identified as begomovirus hosts, including HoLCV and PeLCV detected in cotton, PaLCV in Croton sp., CLCuKoV-Bu, SLCCNV, and ToLCNDV in cucumber, CLCuAlV and CLCuKoV-Bu in okra, CLCuMuV in Sida sp., and CLCuKoV-Bu in squash (Table S3). 3.5 Recombination analysis of geminiviruses and betasatellites Recombination analysis for CLCuKoV (n=382), CLCuMuV (n=163), and CLCuAlV (n=29) using RDP5 (Martin et al., 2021), predicted events supported by all seven algorithms (Fig. 3) identifying one or more events among ‘core’ leaf curl species. Two major predicted recombination events were identified in the coat protein (cp) coding region, involving recombination between different isolates of CLCuKoV-Bu. The breakpoints for event 1 were located at nucleotide coordinates 180-1132, and involved 215 of 382 CLCuKoV-Bu isolates, whereas the event 2 breakpoints were located at nucleotide coordinates 219-731 involving 39 CLCuKoV-Bu isolates. Further, six additional recombination events were predicted among CLCuMuV isolates (Figure 3; events 3 to 8). These consisted of events 5 and 6, predicted in the cp with breakpoints located at nucleotide coordinates 327 – 887, and at 449 – 871, for 19 and 39 of 163 isolates, respectively. Four events were predicted in rep, of which 3 (event no. 3,4 and 7) involved CLCuMuV and CLCuKoV-Bu with breakpoints at nucleotide coordinates 1509 –2720, 1660–2221 and 1857–2397 respectively. Events 3 and 4 involved 52 isolates each, and Event 7 involved 12 of 163 CLCuMuV isolates. A fourth event, enumerated as 8, predicted recombination within the CLCuMuV rep coding region for 46 isolates, with breakpoints at nucleotide coordinates 1834-2218. Importantly, the minor parent involved in this predicted recombination event was CLCuKoV-Ko.Fig. 3 Predicted recombination events involving cotton leaf curl Kokhran virus-Burewala (CLCuKoV-Bu) (Events 1 and 2), cotton leaf curl Multan Virus (CLCuMuV) (Events 3 to 8), cotton leaf curl Kokhran virus- Kokhran (CLCuKoV-Ko) (Event 9) and cotton leaf curl Alabad virus (CLCuAlV) (Event 10). For each event, the number of predicted recombinants is indicated parenthetically. For reference, the characteristic genome organization of a monopartite begomovirus is provided for reference, at the top of the figure. Fig. 3: Two additional predicted recombination events, 9 and 10, involved CLCuKoV-Ko and CLCuAlV. Here, the CLCuKoV-Ko isolates detected in cotton plants harbored predicted recombinants between CLCuKoV-Ko (major parent) and CLCuKoV-Bu (minor parent), with breakpoint positions at nucleotide coordinates 1861-2729. Also, among the seven CLCuAlV isolates identified in this study, two were predicted recombinants that harbored a fragment of CLCuMuV cp, with breakpoints at nucleotide coordinates 540-949. (Fig. 3). Recombination analysis of the betasatellite sequences revealed several predicted recombination events. Among CLCuMuB isolates, a number of predicted recombination sites involved breakpoints located within the satellite-conserved region (SCR). Thirty-three of these were predicted recombinants of Tomato leaf curl Rajasthan betasatellite (ToLCuRaB, KP892648), and all SCR region (∼235nt) recombinants were contributed by a ToLCuRaB (minor parent). Isolates of the newly emergent ToLCuRaB, reported herein, were previously classified as two taxa, referred to as tomato leaf curl betasatellite (ToLCuB) and/or chili leaf curl betasatellite (ChLCuB) (2016) (Briddon et al., 2017). However, based on pairwise distance analysis (SDT v1.2 algorithm) (Muhire et al., 2014) (datasheet S3) they have been merged into one taxon, named ToLCuRaB, with the ICTV code 2020.009P (Fiallo-Olivé & Navas-Castillo, 2020). Two BYMB isolates from cotton were predicted recombinants of CLCuMuB (minor parent), with the breakpoints located at nucleotide coordinates 843-1158, a region containing part of the A-rich region and of the SCR. 3.6 Expansion analysis of cotton leaf curl viruses The Tajima's D test, which provides an estimate of genetic diversity and population expansion or contraction, was used to analyze the three most prevalent begomoviral species genomes, CLCuKoV-Bu, CLCuMuV-Fai, and CLCuMuV-Ra that includes CLCuKoV-Bu and CLCuMuV-Ra, which were responsible for the second epidemic and the third or current outbreak, respectively (Table 5). Two of the three viruses, CLCuKoV-Bu, CLCuMuV-Ra showed signals of population expansion, while significantly negative values were assigned to CLCuMuV-Fai, the causal begomovirus of the Multan epidemic.Table 5 Genetic diversity among cotton leaf curl viral genomes estimated by Tajima's D, Fu, and Li's D, and Fu & Li's F tests. The number of plant samples associated with each begomoviral genome is shown parenthetically, for cotton leaf curl Kokhran virus-Burewala (CLCuKoV-Bu), cotton leaf curl Multan virus-Faisalabad (CLCuKoV-Fai), cotton leaf curl Multan virus-Rajasthan (CLCuKoV-Ra). Table 5:Virus (No. of sequences) Tajima's D Fu & Li's D Fu & Li's F CLCuKoV-Bu (292) -2.65709 (p < 0.001) -9.56213 (p < 0.02) -6.91397 (p < 0.02) CLCuMuV-Fai (56) -1.99661 (P < 0.05) -3.14670 (P < 0.05) -3.23268 (P < 0.02) CLCuMuV-Ra (22) -2.48720 (P < 0.001) -3.63838 (P < 0.02) -3.84498 (P < 0.02) 3.7 Phylogenetic analysis of full-length ‘core’ cotton leaf curl virus genomes Phylogenetic analysis of the full-length genome sequences and representative GenBank sequences that included the CLCuD-associated ‘core’ species, CLCuKoV and CLCuMuV and strains thereof, CLCuKoV-Bu, CLCuKoV-Ko and CLCuMuV-Darwini (CLCuMuV-Dar), CLCuMuV-Fai, CLCuMuV-Hibiscus (CLCuMuV-Hib), CLCuMuV-Hisar (CLCuMuV-His), CLCuMuV-Pk, CLCuMuV-Ra, CLCuMuV-Sha respectively. In addition to representing the historically recognized CLCuD-associated virus strains, several were identified in this study. Phylogenetic analysis (ML) of the full-length CLCuKoV-Bu genome sequences indicated that all were closely related to previously-documented CLCuKoV-Bu isolates (Fig. 5). Only one isolate of CLCuKoV-Ko (1261_CLCuKoV-Ko_Sahiwal_2013) grouped in the sister clade that also contained an isolate of CLCuKoV-Ko (HQ257374) from India. Further, this relationship was supported by the pairwise distance analysis that showed the two isolates shared 97.6% nucleotide identity. The latter two CLCuKoV-Ko isolates grouped most closely with members of the CLCuKoV-Bu clade that harbored isolates identified as the same strain, despite low bootstrap support (16%), suggesting they could be recombinant sequences. Previously, Cotton leaf curl Kokhran virus-Shahdadpur (CLCuKoV-Sha) was considered a recognized strain of CLCuKoV (Brown et al., 2015). Based on a sequence comparison with other CLCuKoV-Sha-like GenBank references, the isolate (herein) was determined to be a strain of CLCuMuV, based on 95.6% and 91-93.4%shared nucleotide identity with CLCuMuV (MG373556) and CLCuKoV (KY797661, HQ257374), respectively. This CLCuMuV isolate was therefore included in the CLCuMuV species-specific phylogenetic analysis of isolates and strains of CLCuMuV identified here, with representative sequences available in the GenBank database (Fig. 6). Phylogenetic analysis (Maximum Likelihood) of the full-length genome CLCuMuV sequences (determined here) grouped with representative GenBank reference sequences for the three recognized CLCuMuV strains, CLCuMuV-Sha, CLCuMuV-Ra and CLCuMuV-Dar. Thus, all CLCuMuV isolates identified in this study were similar to those reported previously from cotton in Pakistan and/or India (Fig. 6). However, the CLCuMuV-Ra isolates exhibited some extents of divergence, with the isolates from Vehari and Sahiwal (Punjab Province) grouping separate from the Sakrand isolates (Sindh Province), while the Vehari isolates grouped with reference GenBank Accessions reported in 2021 (Ahmed et al., 2021). The CLCuMuV-Ra, Punjab isolates shared 99.5-99.7% pairwise nucleotide identity with GenBank Accessions (MK357244, MK357255, MT037031, MT037033), compared to Sakrand isolates of CLCuMuV-Ra field isolates that exhibited the greatest divergence, with 93.9- 94.4% pairwise nucleotide identity with available CLCuMuV-Ra Pakistan isolates (n=48). 3.8 Phylogenetic analysis of full-length CLCuD-betasatellite sequences Phylogenetic analysis (ML) of betasatellite sequences determined in this study, with representative GenBank reference sequences, indicated that 140 isolates grouped with BYVMB, ChLCuB, and CLCuMuB, with robust bootstrap support. In addition, six of the isolates identified here as a new species, clustered within the CLCuMuB sister clade with which they shared <91% pairwise nucleotide identity (Briddon et al., 2017) with previously reported betasatellites (Fig. 7a, spreadsheet S1). Finally, BYVMB sequences showed extensive intra-specific diversity, and grouped within one of 2 sister clades (Fig. 7a). Several CLCuMuB betasatellites (this study) were identified as the following predicted recombinants, CLCuMuB-Multan (CLCuMuB-Mul), CLCuMuB-Burewala (CLCuMuB-Bu), CLCuMuB-Shahdadpur (CLCuMuB-Sha) and CLCuMuB-Vehari (CLCuMuB-Veh), associated with the CLCuD landscape (Akhtar et al., 2014; Amin et al., 2006; Zubair et al., 2017). Because the working cut-off for betasatellite ‘strain’ demarcation has not been established by the previous or presently revised classification (Briddon et al., 2008, 2017; Fiallo-Olivé & Navas-Castillo, 2020), herein, the term “recombinant” has been applied, instead of ‘strain’. Based on results of phylogenetic and recombination analyses of CLCuMuB sequences, recombinants grouped in one of several distinct clades. In this study, the CLCuMuB-Bur predicted recombinant found to be highly prevalent, was also associated with CLCuD-begomoviruses during the Burewala epidemic (Amin et al., 2006). The additional isolates contributed by this study has revealed more extensive divergence among the previously reported references CLCuMuB-Bur sequences, splitting the latter into several subclades (Akhtar et al., 2014; Amin et al., 2006; Zubair et al., 2017). A number of divergent betasatellite recombinant sequences determined in this study grouped in previously un-resolved sister clades, indicating they are newly discovered and potentially emergent CLCuMuB-Bu recombinants (Fig. 7b). 3.9 Identification of helper begomovirus and betasatellite species/strains Based on the shared pairwise nucleotide identity estimates, the working cut-off for betasatellite ‘species’ delineation has been revised from 78% (Briddon et al., 2008) to 91% (Briddon et al., 2017), resulting in 61 proposed species classified within the genus, Betasatellite, and additional modifications have been proposed (Fiallo-Olivé & Navas-Castillo, 2020). Consequently, the nomenclature used for most betasatellite GenBank Accessions is still based on the initially-established species cut-off. In this study, the resultant taxa and nomenclature for BYVMB, ChLCuB, and CLCuMuB sequences determined here and GenBank references, has been based on pairwise distance analysis (SDT algorithm) to guide species demarcation. These revised ‘type species’ accessions (Briddon et al., 2017) were then used as reference sequences (datasheets S1-S3). Based on this analysis 6 isolates determined in this study were found to share 86.6%-90% pairwise nucleotide identity with CLCuMuB previously described isolates from Pakistan (LT827054, KR013746). Collectively, the sequences represent three previously undescribed betasatellite species (Table 6). All 6 isolates encode a characteristic βC1 ORF, A-rich region, and SCR. These previously unrecognized betasatellite species were associated with symptomatic commercial cotton fields in Lodhran, and Luffa cylindrica plants from the sentinel plots in Lahore. The latter associated helper begomovirus sequences have been submitted to GenBank (Table S1, S2).Table 6 Proposed species and name for previously unreported betasatellite identified in this study. Table 6Sample No. Sample name, host, location, and year Closest match and percent nucleotide identity (SDT) Proposed species and name 1 CV25b12401p_LuYMB_Luffa _Lahore_2011 88.7% with CLCuMuB (LT827054) Luffa yellow mosaic betasatellite 2a 104_CLCuLoB _Cotton_Lodhran_2012 89.9% with CLCuMuB (LT827054) Cotton leaf curl Lodhran betasatellite 2b CVb3104_CLCuLoB _Cotton_Lodhran_2011 89.9% with CLCuMuB (LT827054) Cotton leaf curl Lodhran betasatellite 2c 108_CLCuLoB _Hollyhock_Islamabad_2012 89.9% with CLCuMuB (LT827054) Cotton leaf curl Lodhran betasatellite 2d CLCuV_108_CLCuLoB _Hollyhock_Islamabad_2012 90% with CLCuMuB (LT827054) Cotton leaf curl Lodhran betasatellite 3 CV28b16205p_CuMB _Cucumber_Lahore_2011 86.6% with CLCuMuB KR013746 Cucumber mosaic betasatellite The full-length begomovirus (n=1500) and betasatellite genomes (n=1200), with the respective plant hosts and year have been collated and archived into the J.K. Brown laboratory databases. The SDT analysis delineated eight additional betasatellite species, 2 additional begomovirus species and 10 additional begomovirus strains (datasheets S1-S4). The GenBank Accession number assigned to each begomoviral and betasatellite sequence determined here and/or of available GenBank sequences have been archived in the Brown laboratory database. Results of the analyses of the curated database sequences supports the revision of species or strain boundaries and revised nomenclature proposed in this report (Table S3 and Datasheets S1-S4). 4 Discussion 4.1 Geminivirus distribution in sentinel plots and commercial fields Sentinel plants have proven useful for monitoring new and emergent plant pathogens, can inform the status of new or impending outbreak pathogens, and for calibrating forecasting systems (Hobbs et al., 2010; Sikora et al., 2014). The goal of this study was to conduct large-scale, long-term surveillance of cotton-infecting begomoviruses in two major cotton-growing provinces of Pakistan, to assess the distribution of known CLCuD complexes, and achieve early detection of potentially high-risk emergent or re-emergent species. This surveillance approach involved the combined sampling of commercial cotton fields and sentinel plots in 15 cotton-growing districts, including Multan and Vehari (Burewala), which were the epicenters of the cotton leaf curl disease (CLCuD) epidemics in Pakistan and India during 1989-1999 and 2002-2015 (Sattar et al., 2017). Previously known and newly discovered begomoviruses and betasatellite strains and/or species were identified in commercial cotton fields and sentinel plots planted to four differentially susceptible cotton varieties and vegetable crop species. Wild, uncultivated weed species established and/or naturally-occurring in the sentinel plots also harbored previously known and newly discovered isolates. These results corroborate a multitude of studies carried out in Pakistan that have shown the importance of wild species as reservoirs of ‘non-core CLCuD’ begomoviral species, some that also infect cotton. The CLCuKoV-Bu strain was the predominant begomovirus identified in cotton agroecosystems in both the Punjab and Sindh Provinces based on surveillance from 2010 to 2014 (Sattar et al., 2017). This may not be surprising, given its prior association with the second CLCuD epidemic that began near the township of Burewala (Amrao, Amin, et al., 2010), now recognized as the epicenter of the ‘Burewala epidemic’ that prevailed from 2001-2014 (Sattar et al., 2013). The CLCuKoV-Bu has been recognized as a resistance-breaking phenotype, most likely due to selection following release of resistant varieties developed to combat CLCuMuV, and which spread rapidly and prevailed throughout cotton-growing regions of Pakistan (Sattar et al., 2017) and India (Rajagopalan et al., 2012) until 2014-2015. Population genetics analysis for the predominant begomovirus genome sequences determined here, has revealed that CLCuKoV-Bu has continued to expand. During this four-year same timeframe and apparently onward, the recombinant CLCuMuV-Ra (Table 5) recognized in 2008, as a minor component of the CLCuD complex (Kumar et al., 2010), has emerged to become a predominant strain. Notably, since the release of leaf curl-resistant cotton in the mid-1990s, CLCuMuV and the core CLCuD complex consisting of CLCuAlV, CLCKoV-Ko and CLCuMuV-Fai, which were discovered during the Multan epidemic, have rarely been detected in cotton fields in Pakistan since then (Zubair et al., 2017). In this study, CLCuMuV and several ‘non-core’ CLCuD-associated begomoviruses were identified in the ‘differential’ sentinel plots established in Lahore, Multan, Vehari (Punjab) and Sakrand (Sindh). The MNH886 MNH814, MNH786 and CIM496 cotton differentials were selected with prior knowledge of their differential susceptibilities to CLCuMuV and CLCuKoV-Bu ‘core’ CLCuD complex. The use of differentially-susceptible cotton genotypes in this study has demonstrated that they were useful indicator hosts for the epidemic-associated ‘core’ begomoviruses, making them useful for annual surveillance to identify the predominant members of the CLCuD complex in the Punjab and Sindh Provinces. 4.2 Emergence of a new cotton leaf curl Multan virus-Ra strain Tajima's D analysis revealed that the now-emergent CLCuMuV-Ra, identified as a minor component of the leaf curl complex in Pakistan and India as early as 2014, is undergoing genetic expansion, while geographic expansion is also observed in this study and previous studies (Zubair et al., 2017; Islam et al., 2018; Biswas et al., 2020; Ahmed et al., 2021). In a study (Zubair et al., 2017) CLCuMuV variants, including a previously identified Rajasthan strain of the ‘Multan virus’ and core begomoviruses known since the CLCuD ‘Multan epidemic, were identified in cultivated cotton fields. Thus, between the first and second epidemics, the causal CLCuD species and strains were thought to be displaced, unknowingly prevailed in the environment until now. Interestingly, in a more recent study (Ahmed et al., 2021) a new strain of Multan species, CLCuMuV-Ra, has been identified from cotton in the Punjab Province. Similarly, a survey of begomoviruses associated with whiteflies in cotton fields in the Punjab and Sindh Provinces (Islam et al., 2018) has revealed that the whitefly vector already harbored the ‘new’ CLCuMuV-Ra isolate. Equally striking, has been the emergence and spread of CLCuMuV-Ra in northwest India from 2012-2020 (Biswas et al., 2020; Datta et al., 2017), at about the same time this once rare strain was more frequently detected in Pakistan cotton during 2010-2017 (this report). Accordingly, the number of GenBank accessions corresponding to the new CLCuMuV-Ra strain, increased from 105 to 156 between January 2015 to September 2021. Overall, results indicate that CLCuMuV-Ra has become the predominant core CLCuD species/strain and has displaced CLCuKo-Bur as the predominant CLCuD-begomovirus in both India and Pakistan. The periodic shift in CLCuD dynamics appears to be primarily attributable to unrealized differences in susceptibility among the cotton varieties released to combat resistance to extant viruses and strains circulating in the environment, resulting in great vulnerability to the selection of CLCuD variants through exposure to the newly-released varieties. The cultivated cotton varieties that predominated during the surveillance period (this report), were selected based on CLCuKoV-Bu tolerance/resistance. Despite the release continuous release of tolerant/resistant varieties by breeding programs, the genetic diversity is overall extremely narrow, and at the same time, the varieties planted in nearly all cotton production areas consist of the same few varieties. Thus, cultivation of narrow genetic diversity throughout the cotton belt in the Punjab Province has led to widespread crop failure in years when whitefly populations are high, due to selection of new virus resistance-breaking variants and their rapid spreading across vast acreages in months to several years. As cotton varieties begin to fail, the tendency has been to increase the frequency of insecticide application. Further, when insecticide use is not rotated effectively, resistance to compounds can develop rapidly among the different whitefly vector haplotypes. Shifts in the predominant whitefly haplotype vector(s) of begomoviruses in cotton-vegetable systems in Pakistan can lead to the accelerated spread of certain begomovirus variants over others (Paredes-Montero et al., 2019; Shah et al., 2020). In this study, CLCuMuV-Ra isolates extant in Sakrand (Sindh) shared 96.8% to 97.9% pairwise nucleotide identity with certain begomoviral GenBank accessions that had been associated with whiteflies collected in 2013 from an unknown location in Pakistan (GenBank Accessions MH555070, MH555071, MH555072, MH560503). These isolates shared 94.0%-94.7% pairwise nucleotide identity with other CLCuMuV-Ra genome sequences identified in the Punjab Province, later reported by Zubair et al. in 2017 (Zubair et al., 2017), Islam et al. in 2018 (Islam et al., 2018) and Ahmed et al. in 2021 (Ahmed et al., 2021), all sharing 98%-100% nucleotide identity (datasheet S4). The CLCuMuV-Ra isolates from Punjab (this study), shared the highest nucleotide identify with CLCuMuV-Ra isolates reported in 2021 by Ahmed et al. from Punjab Province. These observations support the hypothesis that the ‘original Multan epidemic’ (first) ‘core’ begomoviral species persist in the environment, even at low frequencies, where they continue to diversify and recombine in response to cotton genotype/germplasm, characteristically, one step behind the next begomovirus resistance-breaking variant. Finally, a somewhat distinct variant of the CLCuMuV-Ra strain that has become the predominant variant in Sindh Province, appear to be evolving independently from the now widespread CLCuMuV-Ra. This observation is supported by the substantial number of CLCuMuV sequences determined here, and was not foreshadowed by CLCuMuV-Ra sequences, of which only a few have been deposited in the GenBank database. In this study, the CLCuD-viruses that predominated during the Multan epidemic, namely, CLCuMuV-Fai, CLCuKoV-Ko, and CLCuAlV, were not identified in commercial cotton field samples largely, however, CLCuMuV and CLCuAlV were detected in symptomatic cotton differentials in sentinel plots. This indicates that viruses previously responsible for widespread outbreaks are persistent in susceptible plant species, most likely also including vegetables, tropical fruit trees, and uncultivated hosts. The genetic background of cotton genotypes sampled in commercial fields are largely unknown. Without this information, it is not possible to identify or predict which varieties or germplasm source(s) may be susceptible to newly diversifying CLCuD species and strains. Incorporating such knowledge into breeding programs could guide germplasm choices for commercial planting including new or existing varieties with resistance to known species or strains. Knowledge of the different genetic backgrounds of cotton varieties planted in the different cotton-growing locations, would complement surveillance activities that identify the predominant geminiviruses circulating in sentinel plot cotton differentials and commercial cotton fields. A coordinated effort to translate surveillance data to guide the selection of virus-specific tolerant/resistant cotton varieties in each subsequent year would be a step in the right direction toward curbing outbreaks caused by already known species and strains of CLCuD (Ahmed et al., 2021; Zubair et al., 2017). 4.3 The prevalence of recombinants and other leaf curl variants Previous examples of recombination among CLCuD-associated begomovirus-betasatellite complexes have pointed to the propensity among geminiviruses to undergo recombination, potentially involving all coding and intergenic regions (Qadir et al., 2019). Frequent recombination among CLCuMuV and CLCuMuB variants at different breakpoints has been observed often (Farooq et al., 2021; Zubair et al., 2017). In this study, recombination was predicted to contribute importantly to the evolution of diverse viruses and strains by recombination. Most recombination events occurred in replication-associated protein (Rep) and coat protein (CP) encoding regions of CLCuAlV, CLCuMuV and CLCuKoV-Bu (Fig. 3). Recombination events were predicted in new CLCuMuV variants that possessed segments of cp and rep of CLCuKoV-Bu (different isolates). Results indicated that CLCuKoV-Bu was the predominant ‘core’ CLCuD-associated virus, previously recognized as the resistance-breaking, recombinant involving CLCuMuV and CLCuKo-Ko (Amrao et al., 2010; Sattar et al., 2017), which is consistent with a recent report (Ahmed et al., 2021). However, the recombination events predicted here are distinct from those associated with the recombinants that involved the cp or rep, respectively (Zubair et al., 2017). A single CLCuKoV-Ko-like (OL436149) recombinant was identified in this study, whereas two CLCuAlV recombinants were identified with the break points located in the viral cp, making them unique from CLCuAlV variants associated with the initial ‘core’ CLCuD complex species in the first or “Multan’ epidemic. Finally, an insertion (mutation) was identified in the C2 ORF of a CLCuKoV-Ko isolate sequenced here that results in ‘early’ stop codon in the ORF. This mutation is reminiscent of that identified in the truncated C2 ORF of the resistance-breaking CLCuKoV-Bu strain (Amrao, Amin, et al., 2010). Evolution of the latter truncated CLCuKoV-Bu C2 ORF is posited to have facilitated resistance-breaking in cotton varieties released to combat the CLCuMuV pandemic, which led to the emergence of CLCuKoV-Bu, causal virus of the second CLCuD outbreak in Pakistan (Amrao, Amin, et al., 2010). Among plant viruses, geminiviruses, in particular, are known have relatively high rates of mutation (Duffy & Holmes, 2009) and to diversify by recombination with co-infecting viruses and their strains (Martin et al., 2021; Syller, 2012). The extensive genomic variability evident among the begomovirus-betasatellite isolates studied here, is consistent with the anticipated ongoing diversification of this group of plants viruses within their South Asian center of diversification. Based on the observations presented here, the establishment of a country-wide surveillance program to regularly track the dynamic diversification of the CLCuD-geminivirus-betasatellite complexes in commercial cotton and wild reservoirs, and in strategically located sentinel plots, can provide valuable insights into the dynamics of CLCuD in Pakistan cotton-vegetable cropping systems, and inform the strategic selection of specific cotton varieties that are made available to farmers each year. The annual establishment of sentinel plots for CLCuD surveillance would facilitate early-recognition of new or re-emergent virus variants with a potential for resistance- breaking, increased virulence, and/or host range shifts. Similarly informative, parallel tracking of the predominant whitefly vector haplotypes, which occupy different environmental niches (Paredes-Montero et al., 2019; Paredes-Montero et al., 2020, Paredes-Montero et al., 2021; Paredes-Montero et al., 2021; Shah et al., 2020), can vary in begomovirus-transmission competencies (Pan et al., 2018, Pan et al., 2020, Shahid et al., 2023) and differ in the propensity for developing insecticide resistance, would go far to inform knowledge-driven CLCuD management practices that result in reduced damage to the cotton crop in Pakistan. 4.4 Geminivirus diversity and host range in cotton-vegetable cropping systems Until the ‘Multan’ epidemic occurred, the etiological agent of CLCuD, now known to consist of a complex of geminiviruses, primarily begomovirus-betasatellite-alphasatellite complexes, was unidentified (Briddon et al., 2001). During the CLCuKoV-Bu epidemic, several non-core CLCuD-associated viruses previously reported solely from vegetable crops, were detected in cotton, and included CpCDV (Manzoor et al., 2014), ToLCNDV tomato leaf curl Gujarat virus (Zaidi et al., 2015), and OELCV (Hameed et al., 2014) (Table S3), providing evidence of host-shifting and of their persistence in the environment. In addition, cotton was identified as the host of exotic, New World bipartite SLCV in commercial cotton fields (Table 3,4). Although SLCV is endemic to the Americas (Brown & Nelson, 1986, 1989), it has been reported from vegetable crops in the Middle East during 2003-2012, owing to its presumed introduction on ornamental plants (Lapidot et al., 2014). Also, the bipartite abutilon mosaic virus (AbMV) that originated in the West Indies and was detected in cotton in Pakistan for the first time, which is distinct from the AbMV isolate identified in India from an Abutilon sp. (Jyothsna et al., 2013). The discovery of SLCV and AbMV in cotton was unexpected because neither has been reported to infect cotton in North America and the Caribbean Islands where they are endemic, respectively, even though cotton is widely cultivated in both locales (Jyothsna et al., 2013; Lapidot et al., 2014). Finally, also unexpectedly, four previously known begomoviruses, HoLCV, OELCuV, PaLCV, and PeLCV, were identified in cotton for the first time. The CLCuMuV and CLCuKoV host range was discovered to be much broader than previously known, herein leading to the first detection report of CLCuKoV-Bur in cucumber, Luffa cylindrica, and okra. In addition, previous studies have reported CLCuKoV-Bu capable of infecting Hibiscus rosa-sinensis, L. cylindrica, papaya, Solanum melongena, Ricinus communis, and Xanthium strumarium (Fareed et al., 2012; Kumar et al., 2010; Mubin et al., 2012; Zhou et al., 1998; Zia-Ur-Rehman et al., 2013). These collective observations reveal that the CLCuKoV-Bur host range spans at least several plant families. Also, here, CLCuMuV was detected in Sida micrantha, hibiscus, okra, papaya, and tomato plants, all viruses previously reported as hosts (Shahid et al., 2007; Sinha et al., 2013; Zhou et al., 1998). These results further support the potential for host range expansion among CLCuD viruses, especially those associated with host reservoirs that bridge the gap between cropping cycles from which new outbreaks could be initiated. 4.5 Shift in the predominant CLCuD begomovirus-betasatellite complex in Pakistan Many begomoviruses and strains thereof, are known to cause CLCuD, however, until this report, all recognized helper begomoviruses have been accompanied by the same betasatellite, CLCuMuB. Since the Multan epidemic, a unique recombinant of CLCuMuB, named CLCuMuB-Bu, has prevailed. It is a predicted recombinant that harbors the satellite-conserved region (SCR), apparently derived from the tomato leaf curl betasatellite (ToLCB) (Amin et al., 2006). However, in this study, cotton leaf curl Multan betasatellite-Shahdadpur (CLCuMuB-Sha), a distinct CLCuMuB variant that until now, was previously reported exclusively in the Sindh Province (Akhtar et al., 2014). A sequence comparison of the latter betasatellites, revealed that the newly-emergent CLCuMuB-Sha harbors a fragment of the ToLCB SCR that is 24 bases shorter in length than found in CLCuMuB-Bu (Amrao, Akhter, et al., 2010). Among the betasatellites analyzed in this study, CLCuMuB isolates exhibited the highest rate of evolution (Fig. 4, 7b). This is most likely attributable to its widespread prevalence in cotton and non-cotton host plants throughout Pakistan, and an inherent high frequency of recombination. Phylogenetic and recombination analysis of betasatellites also revealed several prevalent divergent betasatellites (Fig. 4, 7a). Among the 79 CLCuMuB isolates sequenced in this study, 37 were nearly identical to the prototype CLCuMuB betasatellite from the Multan epidemic, indicating its persistence in cotton. However, the phylogeny of the remaining predicted CLCuMuB recombinants, revealed that they formed several sister groups (Fig. 7b). Further, the unexpected prevalence of the previously undiscovered betasatellite, BYVMB, in cotton, points to a major evolutionary step in diversification of CLCuD-associated betasatellites, previously dominated by CLCuMuB. The extant prevalence of genetically-divergent and recombinant isolates of BYVMB and of previously undiscovered CLCuMuB and ChLCuB variants, portend yet another potentially major shift associated with recent selection events among the members of this disease complex. In addition, 33 recombinant isolates of CLCuMuB were identified in this study and found to contain a ToLCRaB-like SCR region (∼235 bases) (7 RDP methods; significant P-value of >0.05 for each) with a hairpin sequence and structure reminiscent of wild type ToLCRaB. Alignment of the latter sequences with another predicted recombinant, also containing a fragment of the SCR region (Amin et al., 2006; Amrao, Akhter, et al., 2010; Zubair et al., 2017) suggests that the ToLCRaB and recombinant CLCuMuB isolates (herein) share a common evolution history. The results reported here therefore provide robust evidence for a surprising shift in the predominant CLCuD-associated betasatellite species in Pakistan and India during 2010-2017, the newly emergent CLCuMuB recombinant that has apparently displaced CLCuMB, which has prevailed since 1990, until its recent apparent displacement.Fig. 4 Predicted recombination involving cotton leaf curl Multan betasatellite (CLCuMuB) (Event 1) and bhendi yellow vein mosaic betasatellite (BYVMB) (Event 2). For each event, the number of isolates harboring predicted recombinants is indicated parenthetically. For reference, the characteristic genome organization of a betasatellite is shown at the top of the figure. Fig. 4: Fig. 5 The cotton leaf curl Kokhran virus (CLCuKoV) tree was reconstructed by Maximum likelihood analysis using 1000 bootstrap iterations for 146 sequences of the isolates determined in this study (blue-colored font) and selected GenBank reference sequences (n=21). The CLCuKoV isolates were identified to the species and strain level, respectively, based on pairwise distance analysis with the SDT v1.2 algorithm. The tree was rooted with the cotton leaf curl Gezira virus (CLCuGeV) sequence, Accession no. NC038444, endemic to sub-Saharan Africa. Fig. 5: Fig. 6 The cotton leaf curl Multan virus (CLCuMuV) tree was reconstructed by Maximum likelihood analysis using 1000 bootstrap iterations for 11 sequences of isolates determined in this study (blue-colored font) and selected GenBank reference sequences (n=63). The CLCuMuV isolates were identified to the species and strain level, respectively, based on pairwise distance analysis with the SDT v1.2 algorithm. The tree was rooted with the cotton leaf curl Gezira virus (CLCuGeV) sequence, Accession no. NC038444, endemic to sub-Saharan Africa. Fig. 6: Fig. 7a The phylogenetic tree was reconstructed for betasatellite sequences by Maximum likelihood analysis using 1000 bootstrap iterations, for 140 sequences determined in this study (blue color font) and selected reference sequences available in the GenBank database (n=73). CLCuMuB isolates were identified to the species level based on pairwise distance analysis with the SDT v1.2 algorithm. The tree was rooted with the cotton leaf curl Gezira betasatellite (CLCuGeB) sequence, Accession no. NC006935, endemic to sub-Saharan Africa. Fig. 7: Fig. 7b The cotton leaf Multan betasatellite (CLCuMuB) tree was reconstructed by Maximum likelihood analysis using 1000 bootstrap iterations for 80 sequences of isolates determined in this study (blue-colored font) and selected GenBank reference sequences (n=46). The CLCuMuB isolates were identified to the species level based on pairwise distance analysis with the SDT v1.2 algorithm. The tree was rooted with the cotton leaf curl Gezira betasatellite (CLCuGeB) sequence, Accession no. NC006935, endemic to sub-Saharan Africa. Fig. 7a: 4.6 The benefits of ongoing surveillance for to inform leaf curl disease management A surveillance strategy that exploits the benefits of sentinel plot-commercial field monitoring can clearly provide valuable information for facilitating the selection of cotton varieties with tolerance or resistance to prevailing CLCuD viruses. Breeding programs are consistently late releasing/providing resilient cotton planting material, a dilemma that could be remedied by regular surveillance to determine the predominant CLCuD helper and betasatellite species and strains. A recent breeding strategy has incorporated MAC7 from the USDA cotton germplasm collection into breeding programs in response to the CLCuKoV-Bu epidemic (Aslam et al., 2022; Rahman et al., 2017; Zaidi et al., 2020). Knowledge of the susceptibility or breadth of resistance in the breeding lines that incorporate this valuable resistance, is imperative, to manage strategic deployment of the resulting varieties, and avoid sole reliance on them for long-term resistance. Sentinel plot surveillance of susceptible and resistant cotton germplasm, of malvaceous and other vegetable crop species, and potential wild host reservoirs of CLCuD-core complex, has demonstrated the persistence of previously known geminivirus-betasatellite variants, while also underscoring the occurrence of minor helper virus-satellite combinations that contribute to genomic and biological variability and therefore, feasibly, to diversification. A comprehensive surveillance approach that incorporates routine sampling of sentinel plots and commercial cotton to track CLCuD-virus and whitefly vector prevalence and distribution would contribute valuably to effective CLCuD management. Integrating surveillance observations into regional and national cotton breeding program logistics, and into annual recommendations of commercial cotton varieties for farmers would go far to abate CLCuD-episodic outbreaks, while also combatting the emergence of new CLCuD-begomovirus-betasatellite variants that pose an ongoing challenge to cotton production in Pakistan and India. Funding The authors acknowledge the Pakistan-U.S. Cotton Productivity Enhancement Program-ICARDA, funded by United States Department of Agriculture-Agricultural Research Service (USDA-ARS) Agreement No. 58-6402-0-178F to MSH, and USDA-ARS Special Cooperative Agreement Nos. 58-6402-0-544 and 58-6402-2-763 to JKB for funding to support this research. The authors acknowledge the support of Cotton Incorporated Project #06-829 to JKB for providing continuous support for this research. Author Contributions This research was conceptualized by JKB, MZ and MSH. MZ, UH, MJI, MTM and MSH organized experimental fields in Pakistan and collect samples. NC carried out helper component annotation and molecular and bioinformatics analyses, and primer design. HWH oversaw and carried out high throughput sequencing data analysis of helper components and betasatellites. MJI and MI analyzed the data and MJI and JKB wrote original manuscript draft. All authors have reviewed and edited the manuscript. JKB and SH co-supervised and co-coordinated the project. All authors have read and agreed to the manuscript contents. Data Statement Sequences determined in the study are available in the GenBank NCBI database (https://www.ncbi.nlm.nih.gov/). Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Appendix Supplementary materials Datasheet S1: SDT and species demarcation of CLCuMuB Table S1: Hepler and betasatellite accessions Figure S1: Symptomatic plants Table S2: Betasatellite accessions Table S3: Host range of begomoviruses Datasheet S2: SDT and species demarcation of BYVMB Datasheet S3: SDT and species demarcation of ChLCuB Datasheet S4: SDT and species demarcation of begomovirus isolates of Pakistan Data availability Sequences determined in the study are available in the GenBank NCBI database. Acknowledgement The authors acknowledge the technical contributions of Ms. Sofia Avelar and Dr. Noma Chingandu for annotation and analyses of the begomoviral-betasatellite sequences used for aspects of the study, and for the design of some of the primers used in this study (also, see Reference Brown et al. 2017). The authors wish to acknowledge the provincial and federal administrated cotton research stations in Vehari, Multan, Sakrand, and members of the Faculty of Agricultural Sciences, University of Punjab Lahore for technical assistance to establish and maintain the sentinel plots and facilitated sample collections in commercial cotton fields located at study sites. Table S1: A list of plant samples, geographical locations, collection year, geminiviruses and betasatellites isolates identification and accession numbers; Table S2: List of betasatellites (in addition to Table S2) isolated from the samples during study with Partial or un-amplified helpers.; Table S3: Host range of begomoviruses/strains in Pakistan, collected from GenBank, literature and this study. Nomenclature for given begomovirus species is verified/corrected by pairwise nucleotide homology analysis and suggestion regarding update in existing demarcations and nomenclature Datasheet S1: Percentage pairwise nucleotide identity data of cotton leaf curl Multan betasatellite determined by SDT v1.2 using available accessions in GenBank and isolates of this study, also suggestions regarding update in nomenclature and species demarcation. Datasheet S2: Percentage pairwise nucleotide identity data of bhendi yellow vein mosaic betasatellite determined by SDT v1.2 using available accessions in GenBank and isolates of this study, also suggestions regarding update in nomenclature and species demarcation. Datasheet S3: Percentage pairwise nucleotide identity data of chili leaf curl betasatellite determined by SDT v1.2 using available accessions in GenBank and isolates herein and proposed updated nomenclature and species demarcation. Datasheet S4: Percentage pairwise nucleotide identity data of Begomovirus isolates of Pakistan determined by SDT v1.2 using available accessions in GenBank and representative isolates of this study, also suggestions regarding update in nomenclature and species demarcation. Figure S1: pictures of various symptomatic plants collected during this study. Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.virusres.2023.199144. ==== Refs References Ahmed N. Amin I. Zaidi S.S.-A. Rahman S.U. Farooq M. Fauquet C.M. Mansoor S. Circular DNA enrichment sequencing reveals the viral/satellites genetic diversity associated with the third epidemic of cotton leaf curl disease Biology Methods and Protocols 6 1 2021 bpab005 10.1093/biomethods/bpab005 33884305 Akhtar S. Tahir M.N. Baloch G.R. Javaid S. Khan A.Q. Amin I. Briddon R.W. Mansoor S. Regional Changes in the Sequence of Cotton Leaf Curl Multan Betasatellite Viruses 6 5 2014 5 10.3390/v6052186 Amin I. Mansoor S. Amrao L. Hussain M. Irum S. Zafar Y. Bull S.E. Briddon R.W. Mobilisation into cotton and spread of a recombinant cotton leaf curl disease satellite Archives of Virology 151 10 2006 2055 2065 10.1007/s00705-006-0773-4 16732497 Amrao L. Akhter S. Tahir M.N. Amin I. Briddon R.W. Mansoor S. Cotton leaf curl disease in Sindh province of Pakistan is associated with recombinant begomovirus components Virus Research 153 1 2010 161 165 10.1016/j.virusres.2010.07.003 20621137 Amrao L. Amin I. Shahid M.S. Briddon R.W. Mansoor S. Cotton leaf curl disease in resistant cotton is associated with a single begomovirus that lacks an intact transcriptional activator protein Virus Research 152 1 2010 153 163 10.1016/j.virusres.2010.06.019 20600387 Aslam M.Q. Naqvi R.Z. Zaidi S.S.-A. Asif M. Akhter K.P. Scheffler B.E. Scheffler J.A. Liu S.-S. Amin I. Mansoor S Analysis of a tetraploid cotton line Mac7 transcriptome reveals mechanisms underlying resistance against the whitefly Bemisia tabaci Gene 820 2022 146200 10.1016/j.gene.2022.146200 Bankevich A. Nurk S. Antipov D. Gurevich A.A. Dvorkin M. Kulikov A.S. Lesin V.M. Nikolenko S.I. Pham S. Prjibelski A.D. Pyshkin A.V. Sirotkin A.V. Vyahhi N. Tesler G. Alekseyev M.A. Pevzner P.A. SPAdes: A New Genome Assembly Algorithm and Its Applications to Single-Cell Sequencing Journal of Computational Biology 19 5 2012 455 477 10.1089/cmb.2012.0021 22506599 Biswas K.K. Bhattacharyya U.K. Palchoudhury S. Balram N. Kumar A. Arora R. Sain S.K. Kumar P. Khetarpal R.K. Sanyal A. Mandal P.K. Dominance of recombinant cotton leaf curl Multan-Rajasthan virus associated with cotton leaf curl disease outbreak in northwest India PLOS ONE 15 4 2020 e0231886 10.1371/journal.pone.0231886 Briddon R.W. Akbar F. Iqbal Z. Amrao L. Amin I. Saeed M. Mansoor S. Effects of genetic changes to the begomovirus/betasatellite complex causing cotton leaf curl disease in South Asia post-resistance breaking Virus Research 186 2014 114 119 10.1016/j.virusres.2013.12.008 24361351 Briddon R.W. Brown J.K. Moriones E. Stanley J. Zerbini M. Zhou X. Fauquet C.M. Recommendations for the classification and nomenclature of the DNA-β satellites of begomoviruses Archives of Virology 153 4 2008 763 781 10.1007/s00705-007-0013-6 18247103 Briddon R.W. Bull S.E. Amin I. Idris A.M. Mansoor S. Bedford I.D. Dhawan P. Rishi N. Siwatch S.S. Abdel-Salam A.M. Brown J.K. Zafar Y. Markham P.G. Diversity of DNA β, a satellite molecule associated with some monopartite begomoviruses Virology 312 1 2003 106 121 10.1016/S0042-6822(03)00200-9 12890625 Briddon R.W. Mansoor S. Bedford I.D. Pinner M.S. Saunders K. Stanley J. Zafar Y. Malik K.A. Markham P.G. Identification of DNA Components Required for Induction of Cotton Leaf Curl Disease Virology 285 2 2001 234 243 10.1006/viro.2001.0949 11437658 Briddon R.W. Markham P.G. Cotton leaf curl virus disease Virus Research 71 1 2000 151 159 10.1016/S0168-1702(00)00195-7 11137169 Briddon, R. W., Navas-Castillo, J., & Fiallo-Olivé, E. (2017). ICTV taxonomic proposal 2016.021 a-kP. A. v2. Tolecusatellitidae. Create the Tolecusatellitidae, a new family of single-stranded DNA satellites with two genera. https://ictv.global/proposals-16/2016.021a-kP.A.v2.Tolecusatellitidae.pdf. Briddon R.W. Stanley J. Subviral agents associated with plant single-stranded DNA viruses Virology 344 1 2006 198 210 10.1016/j.virol.2005.09.042 16364750 Brown J.K. Cotton leaf curl disease complex (leaf curl disease of cotton) Invasive Species Compendium 2020 CABI Wallingford, UK 10.1079/ISC.16813.20210200739 Brown J.K. Czosnek H. Whitefly transmission of plant viruses Advances in Botanical Research 36 2002 65 100 10.1016/S0065-2296(02)36059-2 Brown J.K. Nelson M.R. Whitefly-borne viruses of melons and lettuce in Arizona Phytopathology® 76 1986 236 239 10.1094/Phyto-76-236 Brown J.K. Nelson M.R. Characterisation of watermelon curly mottle virus, a geminivirus distinct from squash leaf curl virus Annals of Applied Biology 115 2 1989 243 252 10.1111/j.1744-7348.1989.tb03383.x Brown J.K. Ur-Rehman M.Z. Avelar S. Chingandu N. Hameed U. Haider S. Ilyas M. Molecular diagnostic development for begomovirus-betasatellite complexes undergoing diversification: A case study Virus Research 241 2017 29 41 10.1016/j.virusres.2017.04.014 28438632 Brown J.K. Zerbini F.M. Navas-Castillo J. Moriones E. Ramos-Sobrinho R. Silva J.C.F. Fiallo-Olivé E. Briddon R.W. Hernández-Zepeda C. Idris A. Malathi V.G. Martin D.P. Rivera-Bustamante R. Ueda S. Varsani A. Revision of Begomovirus taxonomy based on pairwise sequence comparisons Archives of Virology 160 6 2015 1593 1619 10.1007/s00705-015-2398-y 25894478 Datta S. Budhauliya R. Das B. Gopalakrishnan R. Sharma S. Chatterjee S. Vanlalhmuaka Raju P.S. Veer V Rebound of Cotton leaf curl Multan virus and its exclusive detection in cotton leaf curl disease outbreak, Punjab (India), 2015 Scientific Reports 7 1 2017 1 10.1038/s41598-017-17680-9 28127051 de Moya R.S. Brown J.K. Sweet A.D. Walden K.K.O. Paredes-Montero J.R. Waterhouse R.M. Johnson K.P. Nuclear Orthologs Derived from Whole Genome Sequencing Indicate Cryptic Diversity in the Bemisia tabaci (Insecta: Aleyrodidae) Complex of Whiteflies Diversity 11 9 2019 9 10.3390/d11090151 Duffy S. Holmes E.C. Validation of high rates of nucleotide substitution in geminiviruses: Phylogenetic evidence from East African cassava mosaic viruses The Journal of General Virology 90 Pt 6 2009 1539 1547 10.1099/vir.0.009266-0 19264617 Edgar R.C. MUSCLE: Multiple sequence alignment with high accuracy and high throughput Nucleic Acids Research 32 5 2004 1792 1797 10.1093/nar/gkh340 15034147 Fareed M.I. Tahir M. Briddon R.W. First report of Cotton leaf curl Burewala virus infecting Ricinus communis Journal of Plant Pathology 94 4 2012 Supplement https://www.cabdirect.org/cabdirect/abstract/20143208884 Farooq T. Umar M. She X. Tang Y. He Z. Molecular phylogenetics and evolutionary analysis of a highly recombinant begomovirus, Cotton leaf curl Multan virus, and associated satellites Virus Evolution 7 2 2021 veab054 10.1093/ve/veab054 34532058 Fiallo-Olivé, E., & Navas-Castillo, J. (2020). ICTV taxonomic proposal 2020.009P. Create 61 species and abolish three species in the genus Betasatellite (Tolecusatellitidae). Fu Y.X. Li W.H. Statistical tests of neutrality of mutations Genetics 133 3 1993 693 709 10.1093/genetics/133.3.693 8454210 Hameed A. Tahir M.N. Amin I. Mansoor S. First Report of Tomato leaf curl New Delhi virus and a Tomato yellow leaf curl Thailand betasatellite Causing Severe Leaf Curl Disease of Potato in Pakistan Plant Disease 101 6 2017 1065 10.1094/PDIS-09-16-1335-PDN Hameed U. Zia-Ur-Rehman M. Herrmann H.-W. Haider M.S. Brown J.K. First Report of Okra enation leaf curl virus and Associated Cotton leaf curl Multan betasatellite and Cotton leaf curl Multan alphasatellite Infecting Cotton in Pakistan: A New Member of the Cotton Leaf Curl Disease Complex Plant Disease 98 10 2014 1447 10.1094/PDIS-04-14-0345-PDN Hobbs H.A. Herman T.K. Slaminko T.L. Wang Y. Nguyen B.T. McCoppin N.K. Domier L.L. Hartman G.L. Occurrences of Soybean Viruses, Fungal Diseases, and Pests in Illinois Soybean Rust Sentinel Plots Plant Health Progress 11 1 2010 44 10.1094/PHP-2010-0827-01-BR Islam W. Lin W. Islam S.U. Arif M. Li X. Yang Y. Ding X. Du Z. Wu Z. Genetic diversity of begomoviruses in Pakistan captured through a vector based survey Microbial Pathogenesis 118 2018 91 97 10.1016/j.micpath.2018.03.019 29548695 J D.J Isolation of plant DNA from faesh tissue Focus 12 1990 13 15 Jyothsna P. Haq Q.M.I. Jayaprakash P. Malathi V.G. Molecular Evidence for the Occurrence of Abutilon mosaic virus, A New World Begomovirus in India Indian Journal of Virology 24 2 2013 284 288 10.1007/s13337-013-0139-y 24426288 Kirthi N. Priyadarshini C.G.P. Sharma P. Maiya S.P. Hemalatha V. Sivaraman P. Dhawan P. Rishi N. Savithri H.S. Genetic variability of begomoviruses associated with cotton leaf curl disease originating from India Archives of Virology 149 10 2004 2047 2057 10.1007/s00705-004-0352-5 15669112 Kumar J. Kumar A. Roy J.K. Tuli R. Khan J.A. Identification and molecular characterization of begomovirus and associated satellite DNA molecules infecting Cyamopsis tetragonoloba Virus Genes 41 1 2010 118 125 10.1007/s11262-010-0482-7 20405195 Lapidot M. Gelbart D. Gal-On A. Sela N. Anfoka G. Haj Ahmed F. Abou-Jawada Y. Sobh H. Mazyad H. Aboul-Ata A.-A.E. Kamal El-Attar A. Ali-Shtayeh M.S. Jamous R.M. Polston J.E. Duffy S Frequent migration of introduced cucurbit-infecting begomoviruses among Middle Eastern countries Virology Journal 11 1 2014 181 10.1186/1743-422X-11-181 25300752 Luo R. Liu B. Xie Y. Li Z. Huang W. Yuan J. He G. Chen Y. Pan Q. Liu Y. Tang J. Wu G. Zhang H. Shi Y. Liu Y. Yu C. Wang B. Lu Y. Han C. …Wang J. SOAPdenovo2: An empirically improved memory-efficient short-read de novo assembler GigaScience 1 1 2012 18 10.1186/2047-217X-1-18 23587118 Mansoor S. Amin I. Iram S. Hussain M. Zafar Y. Malik K.A. Briddon R.W. Breakdown of resistance in cotton to cotton leaf curl disease in Pakistan Plant Pathology 52 6 2003 784 10.1111/j.1365-3059.2003.00893.x Mansoor S. Briddon R.W. Bull S.E. Bedford I.D. Bashir A. Hussain M. Saeed M. Zafar Y. Malik K.A. Fauquet C. Markham P.G. Cotton leaf curl disease is associated with multiple monopartite begomoviruses supported by single DNA β Archives of Virology 148 10 2003 1969 1986 10.1007/s00705-003-0149-y 14551819 Mansoor S. Khan S.H. Bashir A. Saeed M. Zafar Y. Malik K.A. Briddon R. Stanley J. Markham P.G. Identification of a Novel Circular Single-Stranded DNA Associated with Cotton Leaf Curl Disease in Pakistan Virology 259 1 1999 190 199 10.1006/viro.1999.9766 10364503 Manzoor M.T. Ilyas M. Shafiq M. Haider M.S. Shahid A.A. Briddon R.W. A distinct strain of chickpea chlorotic dwarf virus (genus Mastrevirus, family Geminiviridae) identified in cotton plants affected by leaf curl disease Archives of Virology 159 5 2014 1217 1221 10.1007/s00705-013-1911-4 24212888 Martin D.P. Varsani A. Roumagnac P. Botha G. Maslamoney S. Schwab T. Kelz Z. Kumar V. Murrell B. RDP5: A computer program for analyzing recombination in, and removing signals of recombination from, nucleotide sequence datasets Virus Evolution 7 1 2021 veaa087 10.1093/ve/veaa087 33936774 Miller M.A. Pfeiffer W. Schwartz T. Creating the CIPRES Science Gateway for inference of large phylogenetic trees In Proceedings of the Gateway Computing Environments Workshop (GCE) 2010 New Orleans, LA 1 8 Mubin M. Akhtar S. Amin I. Briddon R.W. Mansoor S. Xanthium strumarium: A weed host of components of begomovirus–betasatellite complexes affecting crops Virus Genes 44 1 2012 112 119 10.1007/s11262-011-0662-0 21969121 Muhire B.M. Varsani A. Martin D.P. SDT: A Virus Classification Tool Based on Pairwise Sequence Alignment and Identity Calculation PLOS ONE 9 9 2014 e108277 10.1371/journal.pone.0108277 Pan L.-L. Chi Y. Liu C. Fan Y.-Y. Liu S.-S. Mutations in the coat protein of a begomovirus result in altered transmission by different species of whitefly vectors Virus Evol. 6 1 2020 veaa014 10.1093/ve/veaa014 Pan L.-L. Cui X.-Y. Chen Q.-F. Wang X.-W. Liu S.-S. Cotton leaf curl disease: which whitefly Is the vector? Phytopathology® 108 10 2018 1172 1183 10.1094/PHYTO-01-18-0015-R 29714092 Paredes-Montero J.R. Hameed U. Zia-Ur-Rehman M. Rasool G. Haider M.S. Herrmann H.-W. Brown J.K. Demographic Expansion of the Predominant Bemisia tabaci (Gennadius) (Hemiptera: Aleyrodidae) Mitotypes Associated With the Cotton Leaf Curl Virus Epidemic in Pakistan Annals of the Entomological Society of America 112 3 2019 265 280 10.1093/aesa/saz002 Paredes-Montero J.R. Ibarra M.A. Arias-Zambrano M. Peralta E.L. Brown J.K. Phylo-biogeographical distribution of whitefly Bemisia tabaci (Insecta: Aleyrodidae) mitotypes in Ecuador Ecosphere 11 6 2020 e03154 10.1002/ecs2.3154 Paredes-Montero J.R. Haq Q.M.I. Mohamed A.A. Brown J.K. Phylogeographic and SNPs analyses of Bemisia tabaci B mitotype populations reveal only two of eight haplotypes are invasive Biology 10 10 2021 10 10.3390/biology10101048 Qadir R. Khan Z.A. Monga D. Khan J.A. Diversity and recombination analysis of Cotton leaf curl Multan virus: A highly emerging begomovirus in northern India BMC Genomics 20 1 2019 274 10.1186/s12864-019-5640-2 30954067 Rahman M. Hussain D. Zafar Y. Estimation of Genetic Divergence among Elite Cotton Cultivars–Genotypes by DNA Fingerprinting Technology Crop Science 42 6 2002 2137 2144 10.2135/cropsci2002.2137 Rahman M.- Khan A.Q. Rahmat Z. Iqbal M.A. Zafar Y. Genetics and Genomics of Cotton Leaf Curl Disease, Its Viral Causal Agents and Whitefly Vector: A Way Forward to Sustain Cotton Fiber Security. Frontiers in Plant Science 8 2017 https://www.frontiersin.org/articles/10.3389/fpls.2017.01157 Rajagopalan P.A. Naik A. Katturi P. Kurulekar M. Kankanallu R.S. Anandalakshmi R. Dominance of resistance-breaking cotton leaf curl Burewala virus (CLCuBuV) in northwestern India Archives of Virology 157 5 2012 855 868 10.1007/s00705-012-1225-y 22307170 Rybicki E.P. A phylogenetic and evolutionary justification for three genera of Geminiviridae Archives of Virology 139 1–2 1994 49 77 10.1007/BF01309454 7826215 Sattar M.N. Kvarnheden A. Saeed M. Briddon R.W. Cotton leaf curl disease – an emerging threat to cotton production worldwide Journal of General Virology 94 4 2013 695 710 10.1099/vir.0.049627-0 23324471 Sattar M.N. Qurashi F. Iqbal Z. Saleem Haider M. Molecular characterization of Hollyhock leaf curl virus and associated DNA-satellites infecting Malva parviflora in Pakistan Canadian Journal of Plant Pathology 39 2 2017 229 234 10.1080/07060661.2017.1336490 Shah R. Al-Sadi A.M. Scott I.M. AlRaeesi A. AlJahdhami A.A. Insecticide resistance monitoring in whitefly (Bemisia tabaci) (Hemiptera: Aleyrodidae) in Oman Journal of Asia-Pacific Entomology 23 4 2020 1248 1254 10.1016/j.aspen.2020.09.018 Shahid M.S. Mansoor S. Briddon R.W. Complete nucleotide sequences of cotton leaf curl Rajasthan virus and its associated DNA β molecule infecting tomato Archives of Virology 152 11 2007 2131 2134 10.1007/s00705-007-1043-9 17703290 Shahid M.S. Paredes-Montero J.R. Ashfaq M. Al-Sadi A.M. Brown J.K. Native and non-native Bemisia tabaci NAFME haplotypes can be implicated in dispersal of endemic and introduced begomoviruses in Oman Insects 14 3 2023 3 10.3390/insects14030268 Sikora E.J. Allen T.W. Wise K.A. Bergstrom G. Bradley C.A. Bond J. Brown-Rytlewski D. Chilvers M. Damicone J. DeWolf E. Dorrance A. Dufault N. Esker P. Faske T.R. Giesler L. Goldberg N. Golod J. Gómez I.R.G. Grau C. …Zidek J. A Coordinated Effort to Manage Soybean Rust in North America: A Success Story in Soybean Disease Monitoring Plant Disease 98 7 2014 864 875 10.1094/PDIS-02-14-0121-FE 30708845 Simonsen K.L. Churchill G.A. Aquadro C.F. Properties of statistical tests of neutrality for DNA polymorphism data Genetics 141 1 1995 413 429 10.1093/genetics/141.1.413 8536987 Sinha V. Kumar A. Bhatnagar D. Khan J.A. Association of Cotton leaf curl Multan virus and its satellite molecules with leaf curl disease of papaya in India New Disease Reports 27 1 2013 9 10.5197/j.2044-0588.2013.027.009 Syller J. Facilitative and antagonistic interactions between plant viruses in mixed infections Molecular Plant Pathology 13 2 2012 204 216 10.1111/j.1364-3703.2011.00734.x 21726401 Tajima F. Statistical method for testing the neutral mutation hypothesis by DNA polymorphism Genetics 123 3 1989 585 595 10.1093/genetics/123.3.585 2513255 Walker P.J. Siddell S.G. Lefkowitz E.J. Mushegian A.R. Adriaenssens E.M. Alfenas-Zerbini P. Davison A.J. Dempsey D.M. Dutilh B.E. García M.L. Harrach B. Harrison R.L. Hendrickson R.C. Junglen S. Knowles N.J. Krupovic M. Kuhn J.H. Lambert A.J. Łobocka M. …Zerbini F.M. Changes to virus taxonomy and to the International Code of Virus Classification and Nomenclature ratified by the International Committee on Taxonomy of Viruses (2021) Archives of Virology 166 9 2021 2633 2648 10.1007/s00705-021-05156-1 34231026 Zaidi S.S.A. Iqbal Z. Amin I. Mansoor S. First Report of Tomato leaf curl Gujarat virus, a Bipartite Begomovirus on Cotton Showing Leaf Curl Symptoms in Pakistan Plant Disease 99 11 2015 1655 10.1094/PDIS-02-15-0195-PDN Zaidi S.S.-A. Naqvi R.Z. Asif M. Strickler S. Shakir S. Shafiq M. Khan A.M. Amin I. Mishra B. Mukhtar M.S. Scheffler B.E. Scheffler J.A. Mueller L.A. Mansoor S Molecular insight into cotton leaf curl geminivirus disease resistance in cultivated cotton (Gossypium hirsutum) Plant Biotechnology Journal 18 3 2020 691 706 10.1111/pbi.13236 31448544 Zaidi S.S.-A. Shafiq M. Amin I. Scheffler B.E. Scheffler J.A. Briddon R.W. Mansoor S. Frequent occurrence of tomato leaf curl New Delhi virus in cotton leaf curl disease affected cotton in Pakistan PLOS ONE 11 5 2016 e0155520 10.1371/journal.pone.0155520 Zerbini F.M. Siddell S.G. Mushegian A.R. Walker P.J. Lefkowitz E.J. Adriaenssens E.M. Alfenas-Zerbini P. Dutilh B.E. García M.L. Junglen S. Krupovic M. Kuhn J.H. Lambert A.J. Łobocka M. Oksanen H.M. Robertson D.L. Rubino L. Sabanadzovic S. Simmonds P. …Varsani A. Differentiating between viruses and virus species by writing their names correctly Archives of Virology 167 4 2022 1231 1234 10.1007/s00705-021-05323-4 35043230 Zhou X. Liu Y. Robinson D.J. Harrison B.D. Four DNA-A variants among Pakistani isolates of cotton leaf curl virus and their affinities to DNA-A of geminivirus isolates from okra The Journal of General Virology 79 Pt 4 1998 915 923 10.1099/0022-1317-79-4-915 9568988 Zia-Ur-Rehman M. Hameed U. Ali C.A. Haider M.S. Brown J.K. First Report of Chickpea chlorotic dwarf virus Infecting Okra in Pakistan Plant Disease 101 7 2017 1336 10.1094/PDIS-11-16-1626-PDN Zia-Ur-Rehman M. Hameed U. Herrmann H.-W. Iqbal M.J. Haider M.S. Brown J.K. First Report of Chickpea chlorotic dwarf virus Infecting Tomato Crops in Pakistan Plant Disease 99 9 2015 1287 10.1094/PDIS-02-15-0202-PDN Zia-Ur-Rehman M. Herrmann H.-W. Hameed U. Haider M.S. Brown J.K. First Detection of Cotton leaf curl Burewala virus and Cognate Cotton leaf curl Multan betasatellite and Gossypium darwinii symptomless alphasatellite in Symptomatic Luffa cylindrica in Pakistan Plant Disease 97 8 2013 1122 10.1094/PDIS-12-12-1159-PDN Zubair M. Zaidi S.S.-A. Shakir S. Farooq M. Amin I. Scheffler J.A. Scheffler B.E. Mansoor S Multiple begomoviruses found associated with cotton leaf curl disease in Pakistan in early 1990 are back in cultivated cotton Scientific Reports 7 1 2017 1 10.1038/s41598-017-00727-2 28127051
PMC010xxxxxx/PMC10352720.txt
==== Front Asian Pac J Cancer Prev Asian Pac J Cancer Prev APJCP Asian Pacific Journal of Cancer Prevention : APJCP 1513-7368 2476-762X West Asia Organization for Cancer Prevention Iran 37116139 10.31557/APJCP.2023.24.4.1181 Research Article Risk Assessment of Alcohol Consumption for Oral Cancer: A Case-Control Study in Patients Attending the National Cancer Institute (Apeksha Hospital, Maharagama) of Sri Lanka Edirisinghe Sajith Tilal 1* Devmini Thisara 1 Pathmaperuma Shanaka 1 Weerasekera Manjula 2 De Silva Kanishka 1 Liyanage Indunil 1 Niluka Malith 1 Madushika Kasuni 1 Deegodagamage Sandeepani 1 Wijesundara Chanuka 1 Rich Alison 3 De Silva Harsha 3 Hussaini Haizal 3 De Silva Dulmini Kavindya 4 Yasawardene Surangi 1 1 Department of Anatomy, Faculty of Medical Sciences, University of Sri Jayewardenepura, Gangodawila, Nugegoda, Sri Lanka. 2 Department of Microbiology, Faculty of Medical Sciences, University of Sri Jayewardenepura, Gangodawila, Nugegoda, Sri Lanka. 3 Department of Oral Diagnostic and Surgical Sciences, Faculty of Dentistry, University of Otago, Dunedin, New Zealand. 4 National Cancer Institute of Sri Lanka, Sri Lanka. * For Correspondence: [email protected] 2023 24 4 11811185 6 10 2022 21 4 2023 https://creativecommons.org/licenses/by-nc/4.0/ This work is licensed under a Creative Commons Attribution-Non Commercial 4.0 International License. (https://creativecommons.org/licenses/by-nc/4.0/) Background: Oral squamous cell cancer (OSCC) is one of the commonest cancers in Sri Lanka. Objectives: This study aimed to determine the use of alcohol, its duration and consuming pattern in relation to the risk of developing OSCC in patients attending the National Cancer Institute of Sri Lanka. Methods: A case-control study was carried out on 105 patients with a histologically confirmed primary OSCC and 210 age-sex matched controls. Information on alcohol consumption was obtained via an interviewer-administered questionnaire. Results: Participants who had consumed alcohol at some point in their life had a 3.8-fold risk of developing OSCC (p=0.000). Current consumers had a higher risk compared to who have consumed previously. Former consumers had a lower risk of developing OSCC compared to current consumers. Individuals who had consumed alcohol for more than 20 years had a greater risk [Odds ratio (OR)=4.69] of developing OSCC compared to those who had consumed alcohol for less than ten years (OR=3.25). Those who consumed the locally-made illicit liquor (Kasippu) had the greatest risk (OR=8.45; p<0.05) of developing OSCC when considering the type of alcohol consumed. Conclusions: Alcohol consumption is a risk factor for OSCC. The OSCC risk increased with longer duration of alcohol use, the consumption of locally-made illicit liquor and current consumers of alcohol. Key Words Squamous cell carcinoma alcohol drinking risk factors Sri Lanka ==== Body pmcIntroduction Oral cancers comprise of tumours involving the oral cavity, pharynx, and salivary glands. Tumours in the oral cavity are an emerging type of cancer in various parts of the world. They accounted for an estimated 354,864 new cancer patients (2.1% of all new cancer patients) and 177,384 deaths (1.9% of all deaths) in 2018 while another study reports that the standard incidence of OSCC in the world was 4 per 100,000 people (Salehiniya and Raei, 2020; Bray et al., 2018). Squamous cell carcinoma is the commonest malignant epithelial neoplasm in the oral cavity and accounts for more than 90% of oral malignant lesions (Warnakulasuriya, 2009). In the Indian subcontinent and other parts of Asia, oral squamous cell carcinoma (OSCC) is one of the commonest forms of cancer. In South-Central Asia to which Sri Lanka belongs, it is the third most common type of cancer (Petersen, 2003). According to data from GLOBOCAN 2018, when considering both the incidence and the mortality data of lip and oral cancer, South Central Asia accounts for the highest age-standardized incidence rates and mortality rates compared to the rest of the Asian, African and European data (Bray et al., 2018). The National Cancer Control Program of Sri Lanka reported that 1941 new cases were diagnosed in the year 2011(NCCP, 2011), while 2199 new patients were detected in 2014(NCCP, 2014). These data imply that there is a rise in newly diagnosed OSCC cases. A cross sectional study carried out in 7 of the 9 provinces in Sri Lanka between 2005 and 2006 revealed the overall prevalence of alcohol consumption to be 23.7%. It had a higher prevalence among the urban population, reaching up to 29.5% (Katulanda et al., 2014). Given the high prevalence of alcohol consumption, and irritation effect of alcohol on oral mucosa it may be worthwhile to study the possible relationship between alcohol consumption and the presence of OSCC. Therefore, the objective of this study was to determine the risk of alcohol use for OSCC in patients attending the National Cancer Institute of Sri Lanka (Apeksha Hospital), Maharagama. Materials and Methods The sample size was calculated based on the literature relating to case-control studies in two independent groups (Daly and Bourke, 2000) and a case-control study was carried out on 105 patients over the age of 18 years with a histologically confirmed primary OSCC recruited from Apeksha Hospital, Maharagama. The patients were either being treated at the institute at the time of recruitment or were awaiting clinic visits at the same institute. Two hundred and ten age-matched (± 5 years from the age of each case) and gender-matched individuals attending the general clinics at Colombo South Teaching Hospital and community were selected as controls using an open advertisement over twelve months. Controls were excluded of the disease, after inspecting for oral lesions for any oral malignant or pre malignant lesions. The patient to control ratio was 1: 2. Ethical approval for the study was granted by the Ethics Review Committee, University of Sri Jayewardenepura, Colombo South Teaching Hospital (29/16) and Apeksha Hospital, Maharagama. Informed and written consent was obtained from all participants. Patients who do not gave the informed consent and did not belong to ICD-O-3 site codes: 00 to 06 categories were excluded from the study population. Both cases and controls were selected randomly in order to reduce the confounding errors. An interviewer-administered questionnaire was given to the participants. The questionnaire comprised two main sections i.e., demographic details and risk factor assessment. The main demographic details gathered were age, sex, gender, education status, occupation and monthly income of the study population. The second section of the questionnaire was used to collect data on alcohol consumption including duration, quantity, pattern, frequency and type of alcohol consumed. Consumption of alcohol along with other two risk factors i.e., smoking and betel chewing was also assessed as these two habits are also considered to be main risk factors for OSCC. All interviews were conducted by the primary investigator and trained five pre-intern medical officers who completed pre-survey calibration to minimize inter-observer variability. Statistical software SPSS version 21.0 was used for general statistical analysis. The odds ratios (OR) with 95% confidence intervals (CI) were calculated to assess the risk factor. Statistical significance for associations was calculated using the chi-square test, with statistical significance set at p<0.05. Results Out of the 105 oral cancer patients, 80 were males and 25 were females. The patients’ age ranged from 35-85 years and they had a mean age of 59.67±15.50 years. The mean age of the male patients was 61.22±15.75 years, while that of the female patients was 54.72±13.80 years. The control group (n=210) had 160 males and 50 females within an age range of 40-82 years (mean 61±12) with the males having a mean age of 61±13 years and females 60±12 years. Among the patients, 49.5% (52/105) were educated up to General Certificate of Education - Ordinary Level while 40.4% (82/210) of the control group were educated up to the same level. In the sample, 82 (82/105, 78.1%) patients and 102 (102/210, 48.5%) controls had consumed alcohol at some point during their life. The overall OR for alcohol consumption was 3.77, with a 95% CI of 2.20-6.45. The risk was statistically significant (p=0.000). Twenty-four patients claimed that they do not drink alcohol anymore but were former alcohol consumers. Fifty-eight patients were current alcohol consumers. The former alcohol consumers had a lower risk of developing OSCC compared to the current consumers. Both former and current alcohol consumption showed a significant association with oral cancer (Table 1). There was a linear association between the duration of alcohol intake and OSCC. Of the total population, 23 patients claimed to have consumed alcohol for more than 20 years. Patients who had consumed alcohol for more than 20 years had a greater risk (OR=4.69) of developing OSCC compared to the people who had consumed alcohol for less than ten years (OR=3.25) (Table 1). Out of the total patients who consumed alcohol, 33 claimed that they consumed hard liquor while 25 claimed to consume beer. When considering the type of ethanol, wine consumption showed the lowest risk (OR=1.76), although the data were not statistically significant (p=0.282). Among patients those who consumed the locally-made illicit liquor (Kasippu)(18/105) had the greatest risk (OR=8.45; p<0.05) of developing OSCC (Table 2). People who consumed beer (OR=3.45) and liquor (OR=3.68) shared almost an equal risk of developing OSCC (p<0.05). When considering the 3 main risk factors together, 47.6% (50/105) of patients were engaged in smoking, consuming alcohol and using smokeless tobacco, while among the controls, the consumers of all three risk factors were 28% (59/210). The patients who smoked and consumed alcohol were 62.8% (66/105), while was 60.9% (64/105) combined the use of alcohol and betel chewing. The OR was statistically significant and was 27.54 for the combination of all three risk factors together. The combination of alcohol and smoking showed a lower risk (OR = 5.67) compared to the combined use of alcohol and betel chewing (OR = 13.12) (Table 3). Table 1 Association between Alcohol Consumption and duration with OSCC Patients Controls Odds Ratio (95% CI) N % N % p value Alcohol usage Former 24 22.8 48 22.8 2.34 (1.20-4.56) p=0.01 Current 58 55.2 54 25.7 5.04 (2.81-9.03) p<0.05 Never 23 21.9 108 51.4 1 Duration of alcohol consumption < 10 years of alcohol 27 25.7 39 18.5 3.25 (1.67-6.32) p<0.05 10-20 years of alcohol 32 30.4 40 19.0 3.75 (1.96-7.17) p<0.05 >20 years of alcohol 23 21.9 23 10.9 4.69 (2.25-9.76) p<0.05 No consumption 23 21.9 108 51.4 1 Table 2 Association between the Types of Alcohol and OSCC Alcohol type Patients Controls Odds Ratio (95% CI) N % N % p value Beer 25 23.8 34 16.1 3.45 (1.74-6.84) p<0.05 Wine 6 5.7 16 7.6 1.76 (0.62-4.98) p=0.282 Liquor 33 31.4 42 20 3.68 (1.94-7.00) p<0.05 Kasippu 18 17.1 10 4.7 8.45 (3.45-20.67) p<0.05 None 23 21.9 108 51.4 1 Table 3 Association between Combined Use of Alcohol, Smoking, and the Betel Chewing with OSCC Type Patients Controls Odds Ratio (95% CI) N % N % p value Combined use of smoking, 50 47.6 59 28 27.54 (6.41-118.19) alcohol and chewing betel p=0.000 Not smoking or using alcohol or chewing betel 2 0.01 65 30.9 1 Combined use of smoking and alcohol 66 62.8 77 36.6 5.67 (2.90-11.07) p=0.000 Not smoking or using alcohol 13 12.3 86 40.9 1 Combined use of alcohol and chewing betel 64 60.9 79 37.6 13.12 (5.01-34.32) p=0.000 Not using alcohol and chewing betel 5 0.02 81 38.5 1 Discussion Sri Lanka accounts for the world’s 5th place and the 4th place among Asian countries in the incidence of lip and oral cavity cancers (Bray et al., 2018). Oral cancer is more frequent in men than in women (Vithana et al., 2021). According to the National Cancer Control Programme of Sri Lanka, oral cancer is the commonest type of cancer found among Sri Lankan males, and the eighth most common type among females (NCCP, 2014). According to data published by the National Institutes of Health in 2018, male patients have a two to six times greater risk than female patients which could be due to their higher intake of tobacco and alcohol (NIH, 2018). Alcohol has gained attention as a significant risk factor for oral cancer (Ribeiro et al., 2015). Furthermore, it has shown a synergistic cancer-promoting effect with tobacco and betel quid. Ethanol and its metabolites such as acetaldehyde are known carcinogenic agents present in alcoholic beverages (Cancer, 2010; Schwartz et al., 2001). It has been suggested that alcohol alters cellular metabolism and enhances the entry of carcinogenic substances into exposed cells (McCoy, 1978). However, the impact of use of alcohol alone on carcinogenesis is debatable. This may be due to the combined intake of tobacco and alcohol by most of the study subjects (Ram et al., 2011). The present study showed that current alcohol consumers have a higher risk of having OSCC (OR=5.04; 95% CI=2.81-9.03; p<0.05) compared to former drinkers (OR=2.34; 95% CI=1.20-4.56; p=0.01). However, a study conducted in Italy and Switzerland found that the risk of having an OSCC is higher among former drinkers compared to current drinkers (OR=1.9; 95% CI=1.3-2.7)(Franceschi et al., 2000). These contrasting findings may be due to the fact that the people in the former drinking category had consumed excessive amounts of alcohol for longer periods prior to cessation. This demonstrates the importance of considering the duration and quantity of alcohol consumed by the individual in addition to their drinking status at the time of the study. Further type of alcohol may also have contributed to these findings. Studies have shown that the risk increases in individuals when the amount of alcohol drinking increases(Bagnardi et al., 2001; Wynder and Stellman, 1977; Silverman Jr and Griffith, 1972). A study carried out in Brazil has highlighted that the risk of oral cancer increases with the increasing frequency of alcohol consumption (OR=3.25; 95% CI=1.03-10.22) (Andrade et al., 2015). A similar finding was published in a study conducted in Spain. The study observed an increased risk for OSCC in heavy drinkers (OR=5.04; 95% CI=1.84-13.85) (Moreno-Lopez et al., 2000). This is in keeping with the findings of the present study where the relative risk of having an OSCC was higher in those who had consumed alcohol for a longer duration. In the present study, we also noted that those who consume locally-made illicit liquor (Kasippu) have a higher risk of having OSCC (OR=8.45; 95% CI=3.45-20.67; p<0.05)). Kasippu is a locally made hard liquor that is made by adding different locally found ingredients without adhering to a standard protocol; thus, the alcohol content may vary. A Brazilian study showed that the consumption of distilled beverages was associated with oral cancer (OR=5.87; 95% CI=3.65-9.44) (Andrade et al., 2015). A pooled analysis of three case-control studies done in the United States of America, Italy, and China emphasized that the risk of OSCC is high in locally made distilled alcohol products because of the higher alcohol content in these drinks (Macfarlane et al., 1995). However, some studies have that the quantity and the duration of consumption of alcohol are more important than the type of alcohol consumed (Reidy et al., 2011; McDowell, 2006). The present study also revealed that smoking, alcohol and betel chewing in combination has a very high risk of developing oral cancer (OR=27.54; 95% CI=6.41-118.19, p<0.01) compared to the consumption of alcohol alone (OR=3.77; 95% CI=2.20-6.450, p=0.000). A Brazilian study has also shown a risk increase by almost ten times in the development of OSCC (OR=9.65; 95% CI=1.57-59.08) when smoking, alcohol and betel are used in combination (Andrade et al., 2015; Edirisinghe et al., 2022). The association between the risk habits of betel quid chewing, tobacco and alcohol use was demonstrated in a recent study conducted in Sri Lanka which also showed a combined synergistic effect of the three habits (Sumithrarachchi et al., 2021). In the present study, it was found that the simultaneous consumption of tobacco and alcohol significantly increases the risk of oral cancer. Similarly, a study conducted in New York City has highlighted that the simultaneous consumption of tobacco and alcohol increases the risk of oral cancer by six to fifteen times (Cruz et al., 2007). It is important to mention that it is difficult to measure the amount of consumption of alcohol by individuals because of the differences in the frequency of intake, the variation of the alcohol level of the drinks consumed by individuals and recall bias. Although the present study has several limitations inherent to case-control studies such as recall bias, the advantages include the sample size, heterogeneity of distribution of exposures and a detailed assessment of lifestyle habits. A future study could also focus on the histological staging of OSCC and the site of the cancer associated with alcohol consumption. The association between gender and carcinogenic risk of alcohol consumption was not assessed in the present study and is a potential question to be answered in future studies. In conclusion, there is a dose-dependent effect of alcohol consumption on OSCC with the locally-made illicit liquor (Kasippu) having the greatest risk for the development of OSCC. A study conducted in Sri Lanka to assess awareness about oral cancers among people revealed that only 43% were aware of oral cancers. In addition, the participants had less awareness about the association between betel chewing and OSCC as compared to that between oral cancer and tobacco smoking as well as alcohol consumption(Sumithrarachchi et al., 2021). Therefore, oral cancer prevention programs in Sri Lanka need to be strengthened to address risk factors including alcohol consumption as the public education and control of alcohol consumption appears to reduce the risk of developing OSCC. Author Contribution Statement Edirisinghe ST1(MBBS/PhD), Conceived and designed the analysis; Collected the data; Contributed data or analysis tools; Performed the analysis; Wrote the paper. Devmini KA(MBBS), Collected the data and data entry, Wrote the paper; Pathmaperuma SD(MBBS), Collected the data and data entry Wrote the paper; Weerasekera M (MBBS/PhD), Conceived and designed the analysis; Collected the data; Contributed data or analysis tools; Performed the analysis; Proofread the paper; De Silva DK(MBBS), Conceived and designed the analysis; Performed the analysis; Wrote the paper; Liyanage I (MBBS), Collected the data and data entry; Niluka M (MBBS), Collected the data and data entry; Madushika K (MBBS), Collected the data and data entry; Deegodagamage S1(MBBS), Collected the data and data entry; Wijesundara C(MBBS), Collected the data and data entry; Rich AM (BDS/PhD), Proofread the paper; De Silva H (BDS/PhD), Proofread the paper; Hussaini HM (BDS/PhD), Proofread the paper; De Silva K (MBBS/MD), Proofread the paper; Yasawardene SG (MBBS/PhD) Conceived and designed the analysis; Collected the data; Contributed data or analysis tools; Performed the analysis; Proofread the paper. Acknowledgements Funding This paper includes the outcome of a subsection of a large study funded by the university research grant, University of Sri Jayewardenepura ASP/01/RE/MED/2016/46, ASP/01/RE/MED/2017/62 and a grant from the Centre for Cancer Research (Number 003/2017), University of Sri Jayewardenepura. Ethical approval Ethical approval for the study was granted by the Ethics Review Committee, University of Sri Jayewardenepura (29/16), Colombo South Teaching Hospital and Apeksha Hospital, Maharagama. Conflicts of interest None ==== Refs References Andrade JOM Santos CAdST Oliveira MC Associated factors with oral cancer: a study of case control in a population of the Brazil’s Northeast Rev Bras Epidemiol 2015 18 894 905 26982303 Bagnardi V Blangiardo M La Vecchia C A meta-analysis of alcohol drinking and cancer risk Br J Cancer 2001 85 1700 5 11742491 Bray F Ferlay J Soerjomataram I Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries CA Cancer J Clin 2018 68 394 424 30207593 Cruz GD Shulman LC Kumar JV The cultural and social context of oral and pharyngeal cancer risk and control among Hispanics in New York J Health Care Poor Underserved 2007 18 833 17982210 Edirisinghe ST Weerasekera M De Silva DK The Risk of Oral Cancer among Different Categories of Exposure to Tobacco Smoking in Sri Lanka Asian Pac J Cancer Prev 2022 23 2929 35 36172654 Franceschi S Levi F Dal Maso L Cessation of alcohol drinking and risk of cancer of the oral cavity and pharynx Int J Cancer 2000 85 787 90 10709096 Katulanda P Ranasinghe C Rathnapala A Prevalence, patterns and correlates of alcohol consumption and its’ association with tobacco smoking among Sri Lankan adults: a cross-sectional study BMC Public Health 2014 14 1 10 24383435 Macfarlane G Zheng T Marshall J Alcohol, tobacco, diet and the risk of oral cancer: a pooled analysis of three case-control studies Eur J Cancer B Oral Oncol 1995 31 181 7 McCoy G A biochemical approach to the etiology of alcohol related cancers of the head and neck Laryngoscope 1978 88 59 62 619212 McDowell JD An overview of epidemiology and common risk factors for oral squamous cell carcinoma Otolaryngol Clin North Am 2006 39 277 94 16580911 Moreno-Lopez L Esparza-Gomez G Gonzalez-Navarro A Risk of oral cancer associated with tobacco smoking, alcohol consumption and oral hygiene: a case-control study in Madrid, Spain Oral Oncol 2000 36 170 4 10745168 NCCP Cancer Incidence Data 2011 https://www.nccp.health.gov.lk/storage/post/pdfs/Cancer_Incidence_Data_2011.pdf NCCP Cancer Incidence Data 2014 https://www.nccp.health.gov.lk/storage/post/pdfs/Cancer_Incidence_in_Sri_Lanka_2014.pdf NIH Oral Cancer Incidence (New Cases) by Age, Race, and Gender 2018 https://www.nidcr.nih.gov/research/data-statistics/oral-cancer/incidence Petersen PE The World Oral Health Report 2003: continuous improvement of oral health in the 21st century–the approach of the WHO Global Oral Health Programme Community Dent Oral Epidemiol 2003 31 3 24 15015736 Ram H Sarkar J Kumar H Oral cancer: risk factors and molecular pathogenesis J Maxillofac Oral Surg 2011 10 132 37 22654364 Reidy J McHugh E Stassen L A review of the relationship between alcohol and oral cancer Surgeon 2011 9 278 83 21843823 Ribeiro ILA Medeiros JJd Rodrigues LV Factors associated with lip and oral cavity cancer Rev Bras Epidemiol 2015 18 618 29 26247186 Salehiniya H Raei M Oral cavity and lip cancer in the world: An epidemiological review Biomed Res Ther 2020 7 3898 905 Schwartz SM Doody DR Fitzgibbons ED Oral squamous cell cancer risk in relation to alcohol consumption and alcohol dehydrogenase-3 genotypes Cancer Epidemiol Biomarkers Prev 2001 10 1137 44 11700261 Silverman Jr S Griffith M Smoking characteristics of patients with oral carcinoma and the risk for second oral primary carcinoma J Am Dent Assoc 1972 85 637 40 4505593 Sumithrarachchi SR Pemasiri WUC Pathiranage AD Betel quid, smoking and alcohol dependency among patients with oral potentially malignant disorders and oral cancer in Sri Lanka; a preliminary case-control study Asian Pac J Cancer Biol 2021 6 207 12 Vithana PVC Dheerasinghe DS Handagiripathira HM Sri Lankan Patterns of Female Cancers: Incidence and Mortality Over 1995-2010 Asian Pac J Cancer Care 2021 6 27 33 Warnakulasuriya S Global epidemiology of oral and oropharyngeal cancer Oral Oncol 2009 45 309 16 18804401 Wynder EL Stellman SD Comparative epidemiology of tobacco-related cancers Cancer Res 1977 37 4608 22 562711
PMC010xxxxxx/PMC10352721.txt
==== Front Asian Pac J Cancer Prev Asian Pac J Cancer Prev APJCP Asian Pacific Journal of Cancer Prevention : APJCP 1513-7368 2476-762X West Asia Organization for Cancer Prevention Iran 37116135 10.31557/APJCP.2023.24.4.1143 Research Article Sarcopenia as a Predictive Factor for Recurrence of Hepatocellular Carcinoma Following Radiofrequency Ablation Jaruvongvanich Varin Thamtorawat Somrach Saiviroonporn Pairash Pisanuwongse Arin Siriwanarangsun Palanan * Department of Radiology, Faculty of Medicine Siriraj Hospital, Mahidol University, Bangkok, Thailand. * For Correspondence: [email protected] 2023 24 4 11431150 5 8 2022 6 4 2023 https://creativecommons.org/licenses/by-nc/4.0/ This work is licensed under a Creative Commons Attribution-Non Commercial 4.0 International License. (https://creativecommons.org/licenses/by-nc/4.0/) Background: Sarcopenia is a skeletal muscle mass deficiency and a potential prognostic factor for the recurrence of hepatocellular carcinoma (HCC). Objective: To determine whether sarcopenia correlates with the recurrence rate of HCC after curative radiofrequency ablation (RFA) in early and very early HCC. Methods: We retrospectively reviewed 669 HCC patients who underwent their first curative RFA at Siriraj hospital from 2011 to 2020. Fifty-six patients who were diagnosed with HCC by triple-phase CT scan and had complete response on follow-up CT were included. All patients underwent skeletal muscle index (SMI) assessment at level L3 vertebra and sarcopenia was defined by the cut-off values of 52.4 cm2/m2 for men and 38.5 cm2/m2 for women. We compared patients with and without sarcopenia. Time to recurrence was evaluated by the Kaplan-Meier method. Univariate and multivariate Cox regression analysis was performed. Results: Sarcopenia was present in 37 of 56 patients (66.1%). There was no significant difference between groups except body mass index (BMI) (P<0.001) and serum alanine aminotransferase (ALT) (P=0.035). There was a promising result indicating the difference of time to recurrence between each group (P=0.046) and potential association of sarcopenia with HCC recurrence (HR=2.06; P=0.052). The Child-Pugh score and tumor number were independent risk factors for HCC recurrence (HR=2.04; P=0.005 and HR=2.68; P=0.017, respectively). Conclusion: Sarcopenia is a potential prognostic factor for recurrence of HCC in Thai patients who underwent RFA. A larger study is required to properly confirm this association. Key Words Sarcopenia hepatoma Low muscle mas muscle atrophy skeletal muscle index HCC RFA ==== Body pmcIntroduction Hepatocellular carcinoma (HCC) was one of the most common cause of cancer-related death in the Asia-Pacific region and one with the highest mortality rate worldwide (Kew, 2010; Zhu et al., 2016). There are many new treatment options in the Barcelona Clinic Liver Cancer (BCLC) guideline that have helped improve survival rates.(Bruix et al., 2005; Bruix et al., 2011; Bruix et al., 2016) Selection of the most appropriate treatment regimen is determined by disease factors (number and size of the HCC as well as portal vein thrombosis and metastasis), liver capacity factors (serum bilirubin, portal hypertension), and patient’s performance factor. The computed tomography (CT) and magnetic resonance imaging (MRI) are powerful tools to diagnose, stage, and guide treatment decisions in HCC. Radiofrequency ablation (RFA) is a treatment of choice in very early and early stage HCC patients that have preserved liver function. Sarcopenia is a condition of decreased skeletal muscle mass two standard deviations below healthy adults (Baumgartner et al., 1998; Prado et al., 2008; Fearon et al., 2011). The condition is associated with many adverse outcomes such as physical disability (Janssen et al., 2002; Kim et al., 2016), falls in the elderly (Landi et al., 2012), osteoporosis (Ahedi et al., 2014), prolonged hospital stay and re-admission (Gariballa and Alessa, 2013), and poor quality of life (Boutin et al., 2015). Sarcopenia is often observed in cancer patients and the elderly (Cruz-Jentoft et al., 2010; Cruz-Jentoft et al., 2014). Studies have suggested that sarcopenia is an independent predictor of survival in esophageal, gastric, pancreatic, lung, breast, and urinary bladder cancer. Furthermore, sarcopenia is associated with poor post-treatment outcome such as post-operative complications, post-treatment infection and delayed recovery (Prado et al., 2008; Gibson et al., 2015; Jones et al., 2015; Shachar et al., 2016; Deng et al., 2018; Simonsen et al., 2018). Sarcopenia can be measured using the middle upper arm muscle by anthropometry (Fearon et al., 2011) or appendicular skeletal muscle area by dual energy X-ray absorptiometry (DXA) (Prado et al., 2008; Fearon et al., 2011; Shepherd, 2016; Buckinx et al., 2018; Scafoglieri and Clarys, 2018; Walowski et al., 2020). In addition, the European Working Group on Sarcopenia in Older People (EWGSOP) suggests that sarcopenia is not only low muscle mass but also low muscular strength because of the non-linear the relationship between muscle mass and strength (Cruz-Jentoft et al., 2010). The EWGSOP group have classified sarcopenia into pre-sarcopenia, sarcopenia, and severe sarcopenia, reflecting the severity of pathology in muscle mass, strength, and physical performance (Cruz-Jentoft et al., 2010; Sergi et al., 2016). Kamachi et al., (2016) reported a correlation between sarcopenia measured by CT scan and the recurrence of HCC in patients with hepatitis C-related cirrhosis after curative resection or RFA. Sarcopenia and high pre-operative α-fetoprotein (AFP) (>40 ng/mL) were independent risk factors for the recurrence of HCC. Similarly, Kobayashi et al., (2019) found that pre-operative sarcopenic obesity and advanced stage of the HCC were independent risk factors for HCC recurrence and death after hepatectomy. Moreover, a meta-analysis by Chang et al., (2018) also reported a significant association between sarcopenia and tumor recurrence and all-cause mortality. Therefore, we aimed to explore the association between sarcopenia and the recurrence of HCC after RFA. Materials and Methods This study was a retrospective observational study approved by the Siriraj Institutional Review Board (IRB). Patient Samples and data collection We searched the electronic database to identify 669 HCC patients who were treated with radiofrequency ablation for the first time at the Siriraj Center of Interventional Radiology (SiCIR), Siriraj hospital between January, 2011 and August, 2020. Inclusion criteria All HCC patients who underwent RFA during January, 2011 to August, 2020 and has all of the following conditions were included: 1. Complete data of CT scan of the upper abdomen/liver prior to RFA procedure in the patient archiving and communication system (PACS) system of Siriraj hospital. The time between the CT study and RFA is less than 120 days. 2. All patients were classified as very early or early stages according to the BCLC prior to RFA (Bruix et al., 2005; Bruix et al., 2011; Bruix et al., 2016; Omata et al., 2017; European Association for the Study of the Liver. Electronic address and European Association for the Study of the, 2018). 3. Complete response in the follow up imaging (CT scan or MR scan of the upper abdomen/liver) performed 4-8 weeks after the procedure. Complete response is defined as no detectable intratumoral arterial enhancement (no viable tumor) and no new evidence of HCC in another liver segment by CT or MRI (Vincenzi et al., 2015; Liver, 2019). 4. Follow-up CT or MRI every 3-6 months, as appropriate. Tumor progression or recurrence is defined according to the mRECIST criteria (Eisenhauer et al., 2009; Lencioni and Llovet, 2010; Liver, 2015; Vincenzi et al., 2015; Liver, 2019). Exclusion criteria 1) History of other cancer types 2) History of prior treatment other than RFA such as liver transplant or transarterial chemoembolization 3) Incomplete clinical data in the electronic medical record 4) No obtainable CT images before or after RFA 5) Significant CT artifact which could obscure interpretation or interfere with the analysis of skeletal muscle area After the inclusion and exclusion criteria were applied, 56 patients were included. A total of 610 cases were excluded due to following conditions; presence of other cancer (60 cases), no pre-treatment imaging (180 cases), underwent other treatment prior to radiofrequency ablation (356 cases), no available follow-up imaging (4 cases), incomplete/inadequate radiofrequency ablation or had residual tumor after treatment (12 cases), and poor pretreatment imaging quality due to metallic artifact (1 case). The subjects were divided into sarcopenic and non-sarcopenic groups according to their skeletal muscle areas (SMI of less than 52.4 cm2/m2 in male and 38.5 cm2/m2 in female measured by CT at the L3 vertebral body level). The electronic medical record of each subject was reviewed for patient demographic data, performance status, laboratory findings, disease parameters, radiologic examination details, treatment, and post-treatment information. Time to recurrence/progression was defined as the interval from the date of curative RFA (complete response) to date of detectable radiological tumor recurrence. Treatment procedure: RFA The patients underwent RFA according to the BCLC classification and the Thailand guidelines for management of hepatocellular carcinoma 2019 (Bruix et al., 2005; Bruix et al., 2011; Liver, 2015; Bruix et al., 2016; Liver, 2019). RFA was performed using the 15-cm LeVeen (Boston Scientific) or StarBurst Talon (Balmer Medical) radiotherapeutic needle electrode with a 2.0- to 5.0-cm ablation zone diameter fit for each lesion under local anesthesia and an intravenous sedation. An electrode was inserted percutaneously into the lesion assisted by ultrasonography or CT. Then, thermal power was delivered and adjusted according to the standard protocol until the target impedance was reached or echogenic cloud was observed. Before the procedural termination, an adequate ablative safety margin of at least 5 mm away from the tumor border was confirmed to assure complete tumor necrosis (Goldberg et al., 2003; Kim et al., 2006; Nakazawa et al., 2007; Teng et al., 2015). Image analysis Skeletal muscle area evaluation The CT images data were acquired via the eFilm Workstation 3.1 and then analyzed in semi-automatic manner using Siriraj hospital’s SiSarcopenia 3.0. Axial CT slice with 1.25-mm to 1.5-mm thickness was chosen at 3rd lumbar vertebral level. The target muscle area consists of psoas muscles, paraspinal muscles (erector spinae, multifidus, and quadratus lumborum), and abdominal wall muscles (transversus abdominis, external and internal obliques, and rectus abdominis). Semi-automated outlining of the aforementioned muscle areas was done by including the pixel with attenuation range of -29 to +150 HU for muscular selection (Harimoto et al., 2013; Kamachi et al., 2016; Yabusaki et al., 2016; Chang et al., 2018). Manual exclusion was done to make the perfect outline of the abdominal and back muscles as shown in Figure 1. Automatic calculation into cross-sectional area in cm2 was made. The skeletal muscle index was then obtained by the following equation. Skeletal muscle index SMI)cm2/m2=Lean tissue at L3 vertical level (cm2)Height cm2 Sarcopenia was defined as an SMI of less than 52.4 cm2/m2 in male and 38.5 cm2/m2 in female. HCC recurrence evaluation Recurrence of HCC was made according to the tumor progression definition from the mRECIST criteria (Eisenhauer et al., 2009; Lencioni and Llovet, 2010; Liver, 2015; Vincenzi et al., 2015; Liver, 2019). This was a condition when new intrahepatic or extrahepatic HCC is found after RFA in this study. The diagnosis of HCC recurrence was based on imaging or pathology result (if any). By radiologic diagnosis, the lesion must have had a longest diameter of at least 1 cm and an arterial enhancement pattern typical of HCC with washout in portal venous or late venous phase. In addition, a highly suspicious lesion that led to further treatment including surgery or RFA was counted as recurrent disease. If the lesion did not show characteristic enhancement of HCC, interval increase in size of at least 1 cm was also considered as disease recurrence. A suspicious finding that did not meet any of the aforementioned recurrence criteria was counted as an equivocal lesion. Statistical analysis Descriptive statistics were used to summarize demographic data such as age, sex, comorbidities, BMI, laboratory findings as well as tumor stage, size, and number. For categorical variables including sex, comorbidities, tumor stage, tumor size, and tumor number, Pearson’s Chi-squared test or Fisher’s exact test was used. The obtained data were presented as frequency and percentage. Continuous data with normal distribution such as age, BMI, albumin, and prothrombin time were evaluated with independent-samples T-test and shown as mean ± S.D. The Mann-Whitney U test was used to assess continuous data with non-normal distribution such as AST, ALT, AFP, and total bilirubin which were demonstrated in median and range. The Cox proportional hazards model was used to analyze associations between variables. Time to recurrence was evaluated using the Log–rank or Breslow test (Kaplan-Meier method). Any variables identified as significant (p value <0.05) or showing a value of P <0.10 in univariate analysis with the abovementioned tests were deemed as candidates for multivariate Cox regression analysis and the results were displayed as hazard ratios (HRs) with 95% confidence interval (CI). All statistical analyses were performed using SPSS version 18.0. A P value that is less than 0.05 was considered statistically significant. Results Patient characteristics The patients’ characteristics, laboratory data, and other associated information are shown in Table 1. Among the 56 patients, 37 (66.1%) had sarcopenia and 19 patients (33.9%) did not. The distribution of skeletal muscle index is shown in Figure 2. No significant difference is observed between these two groups in terms of age, sex, comorbidities, and laboratory findings. Only the BMI and serum ALT level were significantly different between the sarcopenic and non-sarcopenic groups (P<.001 and P=0.035, respectively). The means of the BMI were 23.2 kg/m2 and 28.6 kg/m2 and the median for the serum ALT were 27 U/L in sarcopenic and 41 U/L in the non-sarcopenic groups. The pre-treatment disease parameters involving BCLC stage, tumor number, and tumor size, revealed no significant difference between these two groups (p value>0.56). Recurrence of HCC Throughout the 10-year observational interval, there were 39 patients who had recurrent cancer and 17 who did not (recurrent rate of 69.6 percent). The means and standard deviations of the skeletal muscle index in male patients with and without HCC recurrence are 48.26 ± 9.38 cm2/m2 and 47.36 ± 7.68 cm2/m2, respectively. The means and standard deviations of the skeletal muscle index in female patients with and without HCC recurrence are 38.10 ± 7.59 cm2/m2 and 38.65 ± 5.54 cm2/m2, respectively. There is no significant difference between recurrent and non-recurrent groups (P=0.777 for men; P=0.879 for women). Time to recurrence The median time to recurrence of HCC was 17.6 months (95% CI, 7.2-28.0 months) in sarcopenic patients and 36.7 months (95% CI, 33.7-39.6 months) in non-sarcopenic patients (P=0.046). (Figure 3) Recurrence rates after 1, 3, and 5 years were 43.0%, 61.4%, and 100%, respectively, for patients with sarcopenia, and 21.1%, 49.5%, and 88.0%, respectively, for those without sarcopenia. The Child-Pugh score (P=0.018) and tumor number (P=0.022) were significantly associated with recurrence of HCC. After multivariate analysis, only the Child-Pugh score (HR, 2.04; 95% CI, 1.23-3.36; P=0.005) and tumor number (HR, 2.68; 95% CI, 1.20-5.99; P=0.017) were significantly and independently associated with recurrence of HCC (Table 2). Figure 1 Skeletal Muscle Mass Evaluation before (A) and after (B) Cross-Sectional Muscular Delineation. The skeletal muscle mass including psoas, paraspinal, and abdominal wall muscles were manually outlined and calculated at the L3 vertebral level using a range of -29 to +150 HU Table 1 Clinicopathological Characteristics of Patients with and without Sarcopenia Characteristics Non-sarcopenia, n=19 Sarcopenia, n=37 p value Patient characteristics and profiles Age, years, mean ± S.D. 59.6 ± 8.6 62.1 ± 11.7 0.419 Male: Female, n (%) 12 (63.2%): 7 (36.8%) 28 (75.7%): 9 (24.3%) 0.326 Comorbidities Cirrhosis, n (%) 15 (78.9%) 33 (89.2%) 0.423 Hepatitis B infection, n (%) 11 (57.9%) 14 (37.8%) 0.153 Hepatitis C infection, n (%) 4 (21.1%) 14 (37.8%) 0.203 Diabetes mellitus, n (%) 7 (36.8%) 9 (24.3%) 0.326 BMI, kg/m2, mean ± S.D. 28.6 ± 3.3 23.2 ± 3.1 <0.001 Albumin, g/dL, mean ± S.D. 3.9 ± 0.6 4.0 ± 0.5 0.743 Total bilirubin, mg/dL, median (range) 0.8 (0.2-3.9) 0.7 (0.3-2.7) 0.597 AST, U/L, median (range) 50 (30-155) 41.5 (16-315) 0.344 ALT, U/L, median (range) 41 (15-119) 27 (10-204) 0.035 PT, second, mean ± S.D. 13.5 ± 1.4 13.5 ± 1.4 0.948 AFP, IU/mL, median (range) 9.2 (1.1-989.8) 6.8 (0.8-2519.0) 0.734 Disease parameters before treatment BCLC stage, n (%) 0.61 0 8 (42.1%) 13 (35.1%) A 11 (57.9%) 24 (64.9%) Tumor number, n (%) 1 Single 16 (84.2%) 31 (83.8%) Two 3 (15.8%) 6 (16.2%) Tumor size, n (%) 0.703 ≤ 30 mm 17 (89.5%) 31 (83.8%) > 30 mm 2 (10.5%) 6 (16.2%) AFP, α-fetoprotein; ALT, alanine aminotransferase; AST, aspartate aminotransferase; BCLC, the Barcelona Clinic Liver Cancer; BMI, body mass index (kg/m2); PT, prothrombin time (second) Figure 2 Distribution of Skeletal Muscle Index (cm2/m2) at the L3 Vertebral Level Table 2 Univariate and Multivariate Cox Proportional Hazard Models for Recurrence of HCC Variable Univariate Multivariate Hazard ratio (95% CI) p value Hazard ratio (95% CI) p value Patient characteristics and profiles Age > 75 years 1.47 (0.52-4.16) 0.474 Men 1.35 (0.65-2.80) 0.425 BMI > 25 kg/m2 0.68 (0.36-1.29) 0.234 Diabetes mellitus 0.46 (0.20-1.05) 0.065 Albumin < 3.5 g/dL 1.63 (0.76-3.47) 0.207 Total bilirubin > 1.0 mg/dL 1.63 (0.84-3.17) 0.149 AST > 35 U/L 1.03 (0.50-2.14) 0.935 ALT > 40 U/L 1.30 (0.69-2.46) 0.42 PT 1.09 (0.86-1.39) 0.473 AFP 1.0 (0.99-1.00) 0.408 Child-Pugh score 1.75 (1.10-2.77) 0.018 2.04 (1.23-3.36) 0.005 Disease parameters (before treatment) Tumor size > 30 mm 1.30 (0.54-3.15) 0.558 Tumor number 2.52 (1.14-5.55) 0.022 2.68 (1.20-5.99) 0.017 BCLC stage A 1.17 (0.60-2.28) 0.639 Sarcopenia parameter Sarcopenia 1.68 (0.85-3.32) 0.137 2.06 (0.99-4.27) 0.052 AFP, α-fetoprotein; ALT, alanine aminotransferase; AST, aspartate aminotransferase; BCLC, the Barcelona Clinic Liver Cancer; BMI, body mass index (kg/m2); PT, prothrombin time (second) Figure 3 Time to Recurrence after Curative Radiofrequency Ablation of Patients with and without Sarcopenia Discussion We observed a high incidence of sarcopenia in patients with early stage HCC (37 in 56 patient or 66.1 percent) compared with a report by Prado et al., (2008) (15%). Our result is consistent with a study in HCC patients by Kamachi et al. (2016) (61/92; 66.3 percent). Using the sarcopenia cut-off values of 52.4 cm2/m2 in male and 38.5 cm2/m2 in female, we observed no significant difference in recurrence rate between sarcopenic and non-sarcopenic groups (HR 2.06; P=0.052). Multivariate Cox analysis suggested that sarcopenia may be associated with HCC recurrence in Thai patients (P=0.052). This could be due to insufficient sample size to demonstrate the substantial difference between these two groups. However, the calculated p values for sarcopenia are only near the cut-off point of statistical significance. Nevertheless, sarcopenia is a promising factor of becoming independent risk factor for HCC recurrence in Thai population. The Child-Pugh score and tumor number of HCC were confirmed to be independently associated with HCC recurrence (P=0.005 and P=0.017, respectively), consistent with a 1973 study by Pugh et al. (1973) stating the Child-Pugh score as a prognostic predictor in cirrhotic patients. The score reflects both clinical and laboratory conditions that are important factors in determining treatment outcome. Increasing in number of HCC to more than one is one of the critical features in distinguishing between the very early and early stages according to BCLC classification. Our study confirms the higher recurrence rate (HR, 2.68; P=0.017). The result also correlates well with several studies which show connection of higher stage and poorer outcome under the concept of the larger/more number of the tumor, the poorer outcome (Llovet et al., 1999; Sala et al., 2005; Colecchia, 2014; Kao et al., 2015; Forner et al., 2018). Despite the association of tumor number with HCC recurrence, other features such as tumor size of more than 3 cm or BCLC stage A were not associated with tumor recurrence in this study (P=0.558 and P=0.639, respectively). This could be due to advancement in RFA technique which could be done safely in case of tumor size of more than 3 cm without a change in prognosis. Consistent with many studies, we observed that higher BMI is significantly correlated with the absence of sarcopenia (P<0.001) (Harimoto et al., 2013; Levolger et al., 2015; Voron et al., 2015; Kamachi et al., 2016; Yabusaki et al., 2016; Ha et al., 2018). In general, BMI varies directly with body weight (kg), but inversely with height squared (m2). Therefore, such calculation tells us roughly how the patient might look like in general. However, caution is still needed in this regard since the mentioned body weight can represent both fat and muscle or even ascites. Because muscular mass weighs more than fat, a person with a high BMI can be muscular with short stature or fat with tall stature. In our study, the high BMI in HCC patients might reflect the higher muscular mass patient (BMI, 28.6 ± 3.3 kg/m2 in non-sarcopenia and 23.2 ± 3.1 kg/m2 in sarcopenia; P<0.001). The association between serum ALT and low muscle mass has been a topic of research interest (Le Couteur et al., 2010; Vespasiani-Gentilucci et al., 2018; Bekkelund and Jorde, 2019). A study by Vespasiani-Gentilucci et al. (2018) that utilized a different criteria of sarcopenia at the leg reported that decreased ALT level is associated with frailty, disability, and sarcopenia in elderly. Though ALT is mainly found in hepatocytes, it is also aggregated in muscle, heart, adipose tissue, intestines, prostate, and brain (Panteghini, 1990; Ozer et al., 2008; Liu et al., 2014). Thus, we postulate that low ALT is probably due to reduced quantities released from muscles into the bloodstream in the presence of sarcopenia. Another possible explanation is that lower BMI in sarcopenic patients portrays reduced risk of developing fatty liver disease (Fan et al., 2018), and so does the probability of increased ALT level (Chen et al., 2008; Miyake et al., 2013; Wang et al., 2013; Loomis et al., 2016; Wang et al., 2016). Le Couteur et al. examined elderly men and found a dissimilar relationship between serum ALT and lean body mass measured by DXA (Le Couteur et al., 2010). We suspect that this discrepancy could in part be from different population groups in each study, including genders and underlying diseases or the method of sarcopenia measurement. This retrospective study has several limitations. It was conducted in a single institution using a small sample size. This reduced our ability to detect meaningful associations among the variable of interest. There are several different definitions of sarcopenia, including the one used in this study. To the best of our knowledge, there is no consensus definition of sarcopenia in the Thai population. So, we decided to use the cut-off values of 52.4 cm2/m2 in men and 38.5 cm2/m2 in women, which have been used in the Japanese population. Finally, we did not evaluate muscle function or strength. In conclusion, sarcopenia measured by skeletal muscle index (SMI) (cm2/m2) in CT scan is a potential risk factor for recurrence of HCC in patients who underwent radiofrequency ablation. To confirm this association, larger studies are required. Author Contribution Statement Jaruvongvanich V: Conceptualization, methodology, data curation, investigation, resources, validation, visualization, writing - original draft, review & editing; Thamtorawat S: Methodology, validation, visualization, writing – review & editing, supervision; Saiviroonporn P: Conceptualization, validation; Pisanuwongse A: Validation, proofreading; Siriwanarangsun P: Conceptualization, methodology, data curation, investigation, resources, validation, visualization, writing - original draft, review & editing, supervision. Acknowledgements The authors wish to thank Julaporn Pooliam from the Research Network Division, Research Department at Siriraj hospital for her assistance with the statistical analysis. Approval The research was not approved by any scientific body or a part of student thesis. Ethical declaration The study was approved by the Siriraj Institutional Review Board (IRB) with certificate of approval (COA) Si 092/2020. Availability of data The datasets analyzed during this study are not publicly available due to privacy and ethical concerns Any conflict of interest The authors declare that they have no potential conflicts of interest. ==== Refs References Ahedi H Aitken D Scott D The association between hip muscle cross-sectional area, muscle strength, and bone mineral density Calcif Tissue Int 2014 95 64 72 24829114 Baumgartner RN Koehler KM Gallagher D Epidemiology of Sarcopenia among the Elderly in New Mexico Am J Epidemiol 1998 147 755 63 9554417 Bekkelund SI Jorde R Alanine Aminotransferase and Body Composition in Obese Men and Women Dis Markers 2019 2019 1695874 31534560 Boutin RD Yao L Canter RJ Sarcopenia: Current Concepts and Imaging Implications AJR Am J Roentgenol 2015 205 W255 66 26102307 Bruix J Reig M Sherman M Evidence-Based Diagnosis, Staging, and Treatment of Patients With Hepatocellular Carcinoma Gastroenterology 2016 150 835 53 26795574 Bruix J Sherman M Management of hepatocellular carcinoma: an update Hepatology 2011 53 1020 2 21374666 Bruix J Sherman M Management of hepatocellular carcinoma Hepatology 2005 42 1208 36 16250051 Buckinx F Landi F Cesari M Pitfalls in the measurement of muscle mass: a need for a reference standard J Cachexia Sarcopenia Muscle 2018 9 269 78 29349935 Chang KV Chen JD Wu WT Association between Loss of Skeletal Muscle Mass and Mortality and Tumor Recurrence in Hepatocellular Carcinoma: A Systematic Review and Meta-Analysis Liver Cancer 2018 7 90 103 29662836 Chen ZW Chen LY Dai HL Relationship between alanine aminotransferase levels and metabolic syndrome in nonalcoholic fatty liver disease J Zhejiang Univ Sci B 2008 9 616 22 18763311 Colecchia A Prognostic factors for hepatocellular carcinoma recurrence World J Gastroenterol 2014 20 5935 24876717 Cruz-Jentoft AJ Baeyens JP Bauer JM Sarcopenia: European consensus on definition and diagnosis: Report of the European Working Group on Sarcopenia in Older People Age Ageing 2010 39 412 23 20392703 Cruz-Jentoft AJ Landi F Schneider SM Prevalence of and interventions for sarcopenia in ageing adults: a systematic review Report of the International Sarcopenia Initiative (EWGSOP and IWGS) Age Ageing 2014 43 748 59 25241753 Deng CY Lin YC Wu JS Progressive Sarcopenia in Patients With Colorectal Cancer Predicts Survival AJR Am J Roentgenol 2018 210 526 32 29364725 Eisenhauer EA Therasse P Bogaerts J New response evaluation criteria in solid tumours: revised RECIST guideline (version 1 1) Eur J Cancer 2009 45 228 47 19097774 European Association for the Study of the Liver. Electronic address eee EASL Clinical Practice Guidelines: Management of hepatocellular carcinoma J Hepatol 2018 69 182 236 29628281 Fan R Wang J Du J Association between body mass index and fatty liver risk: A dose-response analysis Sci Rep 2018 8 15273 30323178 Fearon K Strasser F Anker SD Definition and classification of cancer cachexia: an international consensus Lancet Oncol 2011 12 489 95 21296615 Forner A Reig M Bruix J Hepatocellular carcinoma Lancet 2018 391 1301 14 29307467 Gariballa S Alessa A Sarcopenia: prevalence and prognostic significance in hospitalized patients Clin Nutr 2013 32 772 6 23395102 Gibson DJ Burden ST Strauss BJ The role of computed tomography in evaluating body composition and the influence of reduced muscle mass on clinical outcome in abdominal malignancy: a systematic review Eur J Clin Nutr 2015 69 1079 86 25782424 Goldberg SN Charboneau JW Dodd GD Image-guided tumor ablation: proposal for standardization of terms and reporting criteria Radiology 2003 228 335 45 12893895 Ha Y Kim D Han S Sarcopenia Predicts Prognosis in Patients with Newly Diagnosed Hepatocellular Carcinoma, Independent of Tumor Stage and Liver Function Cancer Res Treat 2018 50 843 51 28882021 Harimoto N Shirabe K Yamashita YI Sarcopenia as a predictor of prognosis in patients following hepatectomy for hepatocellular carcinoma Br J Surg 2013 100 1523 30 24037576 Janssen I Heymsfield SB Ross R Low relative skeletal muscle mass (sarcopenia) in older persons is associated with functional impairment and physical disability J Am Geriatr Soc 2002 50 889 96 12028177 Jones KI Doleman B Scott S Simple psoas cross-sectional area measurement is a quick and easy method to assess sarcopenia and predicts major surgical complications Colorectal Dis 2015 17 O20 6 25328119 Kamachi S Mizuta T Otsuka T Sarcopenia is a risk factor for the recurrence of hepatocellular carcinoma after curative treatment Hepatol Res 2016 46 201 8 26223826 Kao WY Chao Y Chang CC Prognosis of Early-Stage Hepatocellular Carcinoma: The Clinical Implications of Substages of Barcelona Clinic Liver Cancer System Based on a Cohort of 1265 Patients Medicine (Baltimore) 2015 94 e1929 26512620 Kew MC Epidemiology of chronic hepatitis B virus infection, hepatocellular carcinoma, and hepatitis B virus-induced hepatocellular carcinoma Pathol Biol (Paris) 2010 58 273 7 20378277 Kim H Suzuki T Saito K Long-term effects of exercise and amino acid supplementation on muscle mass, physical function and falls in community-dwelling elderly Japanese sarcopenic women: A 4-year follow-up study Geriat Gerontol Int 2016 16 175 81 Kim YS Rhim H Cho OK Intrahepatic recurrence after percutaneous radiofrequency ablation of hepatocellular carcinoma: analysis of the pattern and risk factors Eur J Radiol 2006 59 432 41 16690240 Kobayashi A Kaido T Hamaguchi Y Impact of Sarcopenic Obesity on Outcomes in Patients Undergoing Hepatectomy for Hepatocellular Carcinoma Ann Surg 2019 269 924 31 29064889 Landi F Liperoti R Russo A Sarcopenia as a risk factor for falls in elderly individuals: results from the ilSIRENTE study Clin Nutr 2012 31 652 8 22414775 Le Couteur DG Blyth FM Creasey HM The association of alanine transaminase with aging, frailty, and mortality J Gerontol Series A Biol Sci Med Sci 2010 65 712 7 Lencioni R Llovet JM Modified RECIST (mRECIST) assessment for hepatocellular carcinoma Semin Liver Dis 2010 30 52 60 20175033 Levolger S van Vledder MG Muslem R Sarcopenia impairs survival in patients with potentially curable hepatocellular carcinoma J Surg Oncol 2015 112 208 13 26266324 Liu Z Que S Xu J Alanine aminotransferase-old biomarker and new concept: a review Int J Med Sci 2014 11 925 35 25013373 Llovet JM Brú C Bruix J Prognosis of hepatocellular carcinoma: the BCLC staging classification Semin Liver Dis 1999 19 329 38 10518312 Loomis AK Kabadi S Preiss D Body Mass Index and Risk of Nonalcoholic Fatty Liver Disease: Two Electronic Health Record Prospective Studies J Clin Endocrinol Metabolism 2016 101 945 52 Miyake T Kumagi T Hirooka M Body mass index is the most useful predictive factor for the onset of nonalcoholic fatty liver disease: a community-based retrospective longitudinal cohort study J Gastroenterol 2013 48 413 22 22933183 Nakazawa T Kokubu S Shibuya A Radiofrequency ablation of hepatocellular carcinoma: correlation between local tumor progression after ablation and ablative margin AJR Am J Roentgenol 2007 188 480 8 17242258 Omata M Cheng AL Kokudo N Asia-Pacific clinical practice guidelines on the management of hepatocellular carcinoma: a 2017 update Hepatol Int 2017 11 317 70 28620797 Ozer J Ratner M Shaw M The current state of serum biomarkers of hepatotoxicity Toxicology 2008 245 194 205 18291570 Panteghini M Aspartate aminotransferase isoenzymes Clin Biochem 1990 23 311 9 2225456 Prado CMM Lieffers JR McCargar LJ Prevalence and clinical implications of sarcopenic obesity in patients with solid tumours of the respiratory and gastrointestinal tracts: a population-based study Lancet Oncol 2008 9 629 35 18539529 Pugh RN Murray-Lyon IM Dawson JL Transection of the oesophagus for bleeding oesophageal varices Br J Surg 1973 60 646 9 4541913 Sala M Forner A Varela M Prognostic prediction in patients with hepatocellular carcinoma Semin Liver Dis 2005 25 171 80 15918146 Sergi G Trevisan C Veronese N Imaging of sarcopenia Eur J Radiol 2016 85 1519 24 27117135 Shachar SS Williams GR Muss HB Prognostic value of sarcopenia in adults with solid tumours: A meta-analysis and systematic review Eur J Cancer 2016 57 58 67 26882087 Shepherd J Evaluation of Sarcopenia by DXA Clin Rev Bone Mineral Metabolism 2016 14 45 9 Simonsen C de Heer P Bjerre ED Sarcopenia and Postoperative Complication Risk in Gastrointestinal Surgical Oncology: A Meta-analysis Ann Surg 2018 268 58 69 29373365 Teng W Liu KW Lin CC Insufficient ablative margin determined by early computed tomography may predict the recurrence of hepatocellular carcinoma after radiofrequency ablation Liver Cancer 2015 4 26 38 26020027 Vespasiani-Gentilucci U De Vincentis A Ferrucci L Low Alanine Aminotransferase Levels in the Elderly Population: Frailty, Disability, Sarcopenia, and Reduced Survival J Gerontol A Biol Sci Med Sci 2018 73 925 30 28633440 Vincenzi B Di Maio M Silletta M Prognostic Relevance of Objective Response According to EASL Criteria and mRECIST Criteria in Hepatocellular Carcinoma Patients Treated with Loco-Regional Therapies: A Literature-Based Meta-Analysis PLoS One 2015 10 e0133488 26230853 Voron T Tselikas L Pietrasz D Sarcopenia Impacts on Short- and Long-term Results of Hepatectomy for Hepatocellular Carcinoma Ann Surg 2015 261 1173 83 24950264 Walowski CO Braun W Maisch MJ Reference Values for Skeletal Muscle Mass - Current Concepts and Methodological Considerations Nutrients 2020 12 33374515 Wang L Guo J Lu J Risk factor compositions of nonalcoholic fatty liver disease change with body mass index in males and females Oncotarget 2016 7 35632 42 27248665 Wang Z Xu M Peng J Prevalence and associated metabolic factors of fatty liver disease in the elderly Exp Gerontol 2013 48 705 9 23721951 Yabusaki N Fujii T Yamada S Adverse impact of low skeletal muscle index on the prognosis of hepatocellular carcinoma after hepatic resection Int J Surg 2016 30 136 42 27154615 Zhu RX Seto WK Lai CL Epidemiology of Hepatocellular Carcinoma in the Asia-Pacific Region Gut Liver 2016 10 332 9 27114433
PMC010xxxxxx/PMC10352722.txt
==== Front Asian Pac J Cancer Prev Asian Pac J Cancer Prev APJCP Asian Pacific Journal of Cancer Prevention : APJCP 1513-7368 2476-762X West Asia Organization for Cancer Prevention Iran 37116166 10.31557/APJCP.2023.24.4.1413 Research Article Prognostic Value of β-Catenin and L1CAM Expressions in Type I Endometrial Carcinoma Manule Yolanda 1 Miskad Upik Anderiani 12* Masadah Rina 1 Nelwan Berti 1 Cangara Muhammad Husni 1 Mardiati Mardiati 2 1 Department of Pathology, Faculty of Medicine, Hasanuddin University, Makassar, Indonesia. 2 Anatomical Pathology Laboratory, Hasanuddin University Hospital, Makassar, Indonesia. * For Correspondence: [email protected] 2023 24 4 14131417 8 1 2023 23 4 2023 https://creativecommons.org/licenses/by-nc/4.0/ This work is licensed under a Creative Commons Attribution-Non Commercial 4.0 International License. (https://creativecommons.org/licenses/by-nc/4.0/) Objective: The aim of this study is to evaluate the expression of β-catenin and L1CAM in the type I of Endometrial Carcinoma. Material and Methods: This study was an analytical study with a cross-sectional design using 49 samples of type I Endometrial Carcinoma. Immunohistochemical method was used to evaluate the expression of β-catenin and L1CAM related to two significant prognostic parameters i.e., lymphovascular space invasion (LVSI) and metastases event of type I Endometrial Carcinoma samples. Results: From all samples collected, based on the presence of LVSI, there were 17 cases (34.7%) with LVSI and 32 (65.3%) no LVSI. Among them, there were 13 cases that included lymph node or omental samples in type I Endometrial Carcinoma, 5 (38.5%) cases of metastasis, and 8 (61.5%) cases that did not metastasize. The statistical results showed that there was a significant correlation between β-catenin and L1CAM expressions examined from tumor cells with lymphovascular space invasion and the presence of metastases in the type I Endometrial Carcinoma (p <0.05). Conclusion: This study suggest that the positive expression of β-catenin together with L1CAM can participate in the development of tumor cells in type I Endometrial Carcinoma, in its ability to involve lymphovascular space invasion, and metastases to other sites. Our results indicate that both of β-catenin and L1CAM are prominent biomarkers for the prognosis of type I Endometrial Carcinoma. Key Words type I Endometrial Carcinoma β-catenin L1CAM immunohistochemistry ==== Body pmcIntroduction Endometrial carcinoma is a malignancy of the female genital tract, which is the endometrial lining, with a high attack rate. It is the sixth most commonly occurring cancer in women worldwide and the fifteenth most commonly occurring cancer overall, of which there were over 380,000 new cases in 2018 (Bray et al., 2018). The American Cancer Society estimated that there were 63,230 new cases and 11,350 deaths in 2018 (Siegel et al., 2018). In Western countries, it is also a common malignancy and had the highest rate of endometrial cancer in 2018. Even in developing countries, the incidence of Endometrial Carcinoma is increasing (Bray et al., 2018). There are two main types of clinicopathology of Endometrial Carcinoma, namely : type I is low grade and estrogen-related, known as Endometrioid Endometrial Carcinoma (EEC) and type II is Nonendometrioid Endometrial Carcinoma (NEEC) which commonly occurs in aged women (Morice et al., 2016). Type I Endometrial Carcinoma has an 80% incidence of endometrial cancer. This malignancy is an endometrial gland neoplasm that gives an image of acinar, papillary, or partially solid formation. It is the same as precursor lesions, that type I Endometrial Carcinoma can develop from atypical hyperplasia / endometrioid intraepithelial neoplasia, due to excessive “unopposed estrogen” stimulation (Morice et al., 2016; Sanderson et al., 2017). The role of using biological markers to identify tumor progression to an advanced stage is needed in the initial assessment of the patient to monitor the possibility of tumor aggressiveness and determine the patient’s prognosis, which in the end can identify the appropriate and effective management strategies for sufferers. Besides its known role in several malignancies, β-catenin through Canonical Wnt / β-catenin pathway also plays an important role in Endometrial Carcinoma tumorigenesis, by activating the target gene through stabilization of β-catenin in the nucleus (Coopes et al., 2018; Sanderson et al., 2017). Nowadays, it has been mentioned by several studies on the existence of L1 cell adhesion molecule (L1CAM) as a biomarker against predictive potential and helps to identify Endometrial Carcinoma with poor outcomes (Bosse et al., 2014). Where L1CAM expression shown in Endometrial Carcinoma, is associated with the aggressiveness of histology subtypes, advanced stages, the occurrence of distant metastases, and poor prognosis (Bosse et al., 2014; Geels et al., 2016). There are several prognostic parameters in endometrial cancer, including: age, parity, histological type, histological grade, myometrial invasion, lymphovascular invasion and lymph node metastasis (Sorbe, 2012; la Rubia et al., 2020). In this study, the correlation of β-catenin and L1CAM expressions was evaluated with two prognostic parameters of Endometrial Carcinoma included in histopathology report, that are lymphovascular space invasion (LVSI) and the presence of metastases. LVSI is an important prognostic factor and it is independent of histological grade or depth of myometrial invasion, whereas lymph node metastases are also significant independent prognostic factors for poor survival (Sorbe, 2012). Materials and Methods This study was conducted at the Anatomical Pathology Laboratory, Hasanuddin University Hospital Makassar. The population of this study as the inclusion criteria was resection tissue from endometrium and diagnosed as type I Endometrial Carcinoma grade 1, grade 2, and grade 3, from hematoxylin and eosin staining slides, including LVSI and metastases status contained in histopathological reports in the hospital. There were 49 samples that met the inclusion criteria consisting of 17 cases of type I Endometrial Carcinoma with LVSI, 32 cases with no LVSI and of these there were 13 samples that included either lymph node or omental tissue. Immunohistochemical procedures were performed using β-catenin monoclonal antibodies (catalog No. GTX34442, dilution 1:50; Gene Tex Laboratory) and L1CAM polyclonal antibody (catalog No.A00729-1, dilution 1:50; Boster Biological Technology). Immunohistochemical staining results were evaluated using a light microscope by two pathologists and researchers. L1CAM and β-catenin immunoexpression are expressed in semi-quantitative estimates with a scoring system, namely: Evaluation of L1CAM expression, was evaluated by presentation of stained areas on tumor cell membranes: There was no colored area (score 0); Colored area <10% (score 1); Colored area 11-50% (score 2); Colored area >50% (score 3) and score interpretations are divided into two categories, namely: 0-1 score = negative expression; and 2-3 scores = positive expression (Zeimet et al., 2013). Scoring of β-catenin expression in the cytoplasm and cell nucleus, by evaluating the intensity of β-catenin color: There are no normal/stained epithelial cells (score 0); Weak (score 1); Moderate (score 2); Strong (score 3); and evaluation the percentage of the colored area: There is no colored area (score 0); Colored area <10% (score 1); Colored areas are 10-25% (score 2); Colored areas 25-50% (score 3); and Colored area > 50% (score 4); then summed up from the intensity scores and colored area presentations, and will get a total score range of 0-7, which in turn the interpretations are divided into two categories as follows: 0-4 = negative expression; and 5-7 = positive expression (Florescu et al., 2016). This study was an analytical study with a cross-sectional design. Bivariate analysis in the form of Chi-square test was used. Results Table 1 shows the characteristics of the samples in the study. The average age of type I Endometrial Carcinoma patients is 51 years old. The age of patients younger than 50 years old were 21 (42.9%), and older than 50 years old were 28 (57.1%). The sample distribution based on the diagnosis of histopathological grading showed that there were 15 cases of type I Endometrial Carcinoma grade 1 (30.6%), 20 cases grade 2 (40.8%), and 14 cases grade 3 (28.6%). Meanwhile, of all the samples collected, there were only 13 cases that sent samples in the form of lymph node and omentum for evaluation of metastases presence in type I Endometrial Carcinoma : 5 metastatic cases, and 8 other cases without metastases. Microscopic evaluation of β-catenin and L1CAM expressions with immunohistochemical staining showed positive expression of β-catenin in the nucleus and cytoplasm, whilst L1CAM positive expression appears on the cell membrane (Figure 1-A,1-B). Microscopic evaluation of the expression of β-catenin and L1CAM on tumor cells in the lymphovascular space invasion area, both showed positive expression in almost all type I Endometrial Carcinoma samples with lymphovascular space invasion (Figure 2-A,2-B). In this study, a microscopic evaluation of β-catenin and L1CAM expressions by immunohistochemical staining of type I Endometrial Carcinoma that metastases to lymph node or omentum was also carried out. Microscopic examination of all samples revealed positive expression of β-catenin and L1CAM (Figure 2-C,2-D). Table 2 shows the results of statistical tests of β-catenin and L1CAM expression scores with lymphovascular space invasion correlation. It results in p = 0.0001 (p <0.05), which means there are significant differences between β-catenin and L1CAM expression scores with the presence of lymphovascular space invasion of type I Endometrial Carcinoma. The table indicates that positive expression of both β-catenin and L1CAM influences the level of lymphovascular invasion. Compared to one or two proteins with a negative expression, which show less influence on level of lymphovascular space invasion. The results of the statistical tests shown in table 3 is the correlation between the β-catenin and L1CAM expression scores with the occurrence of metastasis, which shows the results of p = 0,0001 (p <0.05). This indicates that there are significant differences between β-catenin and L1CAM expression scores with the presence of metastatic events in type I Endometrial Carcinoma. The table also shows that positive expression of both β-catenin and L1CAM strongly influence the metastatic events. Compared to one protein with a negative expression, that more less influences the presence of metastases. Figure 1 A). Positive expression of β-catenin on the nucleus and cell cytoplasm of type I endometrial carcinoma Grade 2 (200x). B). Positive expression of L1CAM on cell membranes of type I endometrial carcinoma Grade 2 (200x) Table 1 Characteristics of the Sample Characteristics Number % Age (n = 49) < 50 21 42.9 ≥ 50 28 57.1 Mean 51.1 51.1 Histopathological Grade (FIGO) Grade 1 15 30.6 Grade 2 20 40.8 Grade 3 14 28.6 LVSI Yes 17 34.7 No 32 65.3 (n = 13) Lymph node or omental metastases Yes 5 38.5 No 8 61.5 Table 2 β-catenin and L1CAM Expressions in Type I Endometrial Carcinoma based on the Presence of Lymphovascular Space Invasion (LVSI) β-catenin and L1CAM expression interpretations LVSI LVSI p-value Yes No β-catenin positif and L1CAM positive 14 (82.4%) 0 (0%) <0.05* β-catenin positif and L1CAM negative 2 (11.8%) 0 (0%) β-catenin negative and L1CAM positive 0 (0%) 0 (0%) β-catenin negative and L1CAM negative 1 (5.9%) 32 (100%) Total no. (%) 17 (100) 32 (100) LVSI, Lymphovascular Space Invasion; * Chi-Square Test Table 3 β-catenin and L1CAM Expressions in Type I Endometrial Carcinoma based on the Presence of Lymph Node or Omental Metastases β-catenin and L1CAM expression interpretations Metastases No metastases p-value β-catenin positif and L1CAM positive 4 (80.0%) 0 (0%) <0.05* β-catenin positif and L1CAM negative 0 (0%) 0 (0%) β-catenin negative and L1CAM positive 1 (20.0%) 0 (0%) β-catenin negative and L1CAM negative 0 (0%) 8 (100%) Total no. (%) 5 (100) 8 (100) *, Chi-Square Test Figure 2 A). β-catenin positive expression of tumor cells in the area of ​​Lymphovascular invasion (200x). B). L1CAM positive expression of tumor cells in the area of ​​lymphovascular invasion (100x). C). Positive expression of β-catenin in type I endometrial carcinoma that metastases to the omentum (100x). D). L1CAM positive expression in type I endometrial carcinoma metastases to the omentum (100x) Discussion Endometrial Carcinoma patients if detected at an early stage will have a better prognosis, with a life expectancy of around 85% in 5 years (Notaro et al., 2016; Van der Putten et al., 2017). Therefore, the ability to detect patients with a high risk of Endometrial Carcinoma can have a major impact on sufferers. Recent studies revealed currently promising prognostic immunohistochemistry markers whether these markers can help to assist the gynecological oncology surgeon selecting the adequate surgical extent, where has been evaluated the correlation with LVSI as the important markers for adjuvant treatment strategy decisions (Weinberger et al., 2019). The finding in this study demonstrate there was a correlation between positive expressions of β-catenin and L1CAM with LVSI in type I Endometrial Carcinoma. Further in this study, an association between the positive expression of β-catenin and L1CAM with the presence of metastases in the lymph nodes and omentum was found. Previous studies have reported that positive expression of L1CAM immunohistochemical examination results has a strong relationship with the depth of myometrial invasion and the presence of lymphovascular invasion in Endometrial Carcinoma (Dellinger et al., 2016; Van Gool et al., 2016). Another study also stated that there was a relationship between L1CAM expression and the presence of lymphovascular invasion in Endometrial Carcinoma. In addition, there is a close relationship between L1CAM expression with the involvement or incidence of lymph nodes metastases (de Freitas et al., 2018; Geels et al., 2016). Whereas a previous research study conducted by Florescu et al., (2016), suggested that there was a relationship between β-catenin expression and tumor staging, degree of differentiation, and myometrial invasion. Instead, it was found that there were no association between β-catenin expression and metastatic involvement to lymph nodes. On the other hand, there was another study that had suggested that Endometrial Carcinoma patients with β-catenin mutations were significantly more likely to have tumors with pathological characteristics that generally had a lower clinical risk of recurrence (lower histologic grade, less incidence of deep myometrial invasion, and less incidence of lymphovascular space invasion (Kurnit et al., 2017)). L1CAM induces EMT in several cancers, including Endometrial Carcinoma. Previous studies have reported that L1CAM and β-catenin, both have prognostic values and can be used as prognostic markers in patients with Endometrial Carcinoma, mentioned that L1CAM upregulation events in Endometrial Carcinoma were induced by β-catenin and TGFβ / SLUG (Pfeifer et al., 2010; Giordano and Cavallaro, 2020). Further explained, L1CAM is the target gene of the Wnt / β-catenin signaling pathway, where β-catenin accumulation in the nucleus shows the same location as L1CAM. Therefore, it explains that the transcriptional β-catenin / TCF-LEF complex is assumed to be a direct regulator of L1CAM expression (Pfeifer et al., 2010; Giordano and Cavallaro, 2020). It was also stated that tumor cells from Endometrial Carcinoma present the process of Epithelial-mesenchymal transition (EMT) induced by TGF-β, which then up-regulates L1CAM with vimentin, both of which down-regulation of E-Chaderin, which this mechanism depends on the transcription factor Slug (Zeimet et al., 2013). Pfeifer et al., (2010) conducted a research study in which the L1CAM gene had 2 functionally active promoters namely Slug and β-catenin which are involved in the transcriptional regulation of L1CAM. Pfeifer carried out the L1CAM gene analysis in detail with the PCR technique, and concluded that the regulation of L1CAM expression in tumor cells in Endometrial Carcinoma was carried out by two different promoter regions, which Slug transcription factors were the relevant transcription factors in this L1CAM regulation. Several previous studies mentioned above support the findings that are found in this study, where there is a significant correlation between β-catenin and L1CAM expression in type I Endometrial Carcinoma that associates with lymphovascular space invasion and metastasis events. In conclusion, the results of this study suggest that there is a tendency for overexpression of β-catenin along with L1CAM which can increase the progressivity of tumor cells in type I Endometrial Carcinoma in its ability to involve lymphovascular space invasion and metastases to other tissues. Thus, both β-catenin and L1CAM have prognostic values and can be used as prognostic markers in patients with Endometrial Carcinoma. Author Contribution Statement The methodology was planned and designed by YM, RM and UM; YM, BN, MHC and M were involved in data gathering, processing, and reporting. UM and YM, conducted a comprehensive conceptual and editorial evaluation; the finalization of the article was amended and approved by all of the contributors. Acknowledgements The Anatomical Pathology Laboratory at Hasanuddin University Hospital contributed to the feasibility of this study in part. Study Approval The research committee of the Faculty of Medicine at Hasanuddin University approved this project. Ethical approval The Ethics Committee of the Faculty of Medicine granted informed consent for this study (Protocol #UH18060363, Archive No. 442/H4.8.4.5.31/PP36-KOMETIK). Availability of Data On reasonable request, the associated author will release the datasets used in this work. Conflict of Interest All contributors report having no competing interests. ==== Refs References Bosse T Nout RA Stelloo E L1 Cell Adhesion Molecule Is A Strong Predictor for Distant Recurrence and Overall Survival In Early Stage Endometrial Cancer: Pooled PORTEC Trial Results Eur J Cancer 2014 50 2602 10 25126672 Bray F Ferlay J Soerjomataram I Global Cancer Statistics 2018: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers In 185 Countries CA Cancer J Clin 2018 68 394 424 30207593 Coopes A Henry CE Llamosas E An Update of Wnt Signalling in Endometrial Cancer and Its Potential as A Therapeutic Target Endocr Relat Cancer 2018 25 R647 62 De Freitas D Aguiar FN Anton C L1 Cell Adhesion Molecule (L1CAM) Expression In Endometrioid Endometrial Carcinomas: A Possible Pre-operative Surrogate of Lymph Vascular Space Invasion PLoS One 2018 13 1 10 Dellinger TH Smith DD Ouyang C L1CAM Is An Independent Predictor of Poor Survival In Endometrial Cancer - An Analysis of The Cancer Genome Atlas (TCGA) Gynecol Oncol 2016 141 336 40 26861585 Florescu MM Pirici D Simionescu CE E-cadherin and β-catenin Immunoexpression In Endometrioid Endometrial Carcinoma Rom J Morphol Embryol 2016 57 1235 40 28174789 Geels YP Pijnenborg JMA Gordon BBM L1CAM Expression is Related to Non-Endometrioid Histology, and Prognostic for Poor Outcome in Endometrioid Endometrial Carcinoma Pathol Oncol Res 2016 22 863 8 26891628 Giordano M Cavallaro U Different Shades of L1CAM In The Pathophysiology of Cancer Stem Cells J Clin Med 2020 9 1502 32429448 Kurnit KC Kim GN Fellman BM CTNNB1 (Beta-Catenin) Mutation Identifies Low Grade, Early Stage Endometrial Cancer Patients At Increased Risk of Recurrence Mod Pathol 2017 30 1032 41 28281553 la Rubia EC de Martinez-Garcia E Dittmar G Prognostic Biomarkers In Endometrial Cancer: A Systematic Review and Meta-Analysis J Clin Med 2020 9 1 20 Morice P Leary A Creutzberg C Endometrial Cancer Lancet 2016 387 1094 108 26354523 Notaro S Reimer D Duggan-Peer M Evaluating L1CAM Expression In Human Endometrial Cancer Using qRT-PCR Oncotarget 2016 7 40221 33 27233077 Pfeifer M Schirmer U Geismann C L1CAM Expression In Endometrial Carcinomas Is Regulated by Usage of Two Different Promoter Regions BMC Mol Biol 2010 11 1 14 20070894 Sanderson PA Critchley HOD Williams ARW New Concepts for An Old Problem: The Diagnosis of Endometrial Hyperplasia Hum Reprod Update 2017 23 232 54 27920066 Siegel RL Miller KD Jemal A Cancer Statistics, 2018 CA Cancer J Clin 2018 68 7 30 29313949 Sorbe B Predictive and Prognostic Factors in Definition of Risk Groups in Endometrial Carcinoma ISRN Obstet Gynecol 2012 2012 1 8 Van der Putten LJM van Hoof R Tops BBJ Molecular Profiles of Benign and (Pre)Malignant Endometrial Lesions Carcinogenesis 2017 38 329 35 28203752 Van Gool IC Stelloo E Nout RA Prognostic Significance of L1CAM Expression and Its Association with Mutant p53 Expression in High-Risk Endometrial Cancer Mod Pathol 2016 29 174 81 26743472 Weinberger V Bednarikova M Hausnerova J A Novel Approach to Preoperative Risk Stratification in Endometrial Cancer: The Added Value of Immunohistochemical Markers Front Oncol 2019 9 1 13 30761267 Zeimet AG Reimer D Huszar M L1CAM In Early-Stage Type I Endometrial Cancer: Results of A Large Multicenter Evaluation J Natl Cancer Inst 2013 105 1142 50 23781004