Carla et al. Respiratory Research (2020) 21:81 https://doi.org/10.1186/s12931-020-01337-9 STUDY PROTOCOL Open Access Acute respiratory distress syndrome subphenotypes and therapy responsive traits among preclinical models: protocol for a systematic review and meta-analysis Adrien Carla1, Bruno Pereira2, Hanifa Boukail1, Jules Audard1,3, Nathalie Pinol-Domenech4, Manuela De Carvalho4, Raiko Blondonnet1,3, Ruoyang Zhai3, Dominique Morand1, Céline Lambert2, Vincent Sapin3,5, Lorraine B. Ware6,7, Carolyn S. Calfee8, Julie A. Bastarache6,7,9, John G. Laffey10,11, Nicole P. Juffermans12, Lieuwe D. Bos12, Antonio Artigas13, Patricia R. M. Rocco14, Michael A. Matthay8, Daniel F. McAuley15, Jean-Michel Constantin16, Matthieu Jabaudon1,3,6* and for the ESICM Translational Biology Group of the Acute Respiratory Failure section Abstract Background: Subphenotypes were recently reported within clinical acute respiratory distress syndrome (ARDS), with distinct outcomes and therapeutic responses. Experimental models have long been used to mimic features of ARDS pathophysiology, but the presence of distinct subphenotypes among preclinical ARDS remains unknown. This review will investigate whether: 1) subphenotypes can be identified among preclinical ARDS models; 2) such subphenotypes can identify some responsive traits. Methods: We will include comparative preclinical (in vivo and ex vivo) ARDS studies published between 2009 and 2019 in which pre-specified therapies were assessed (interleukin (IL)-10, IL-2, stem cells, beta-agonists, corticosteroids, fibroblast growth factors, modulators of the receptor for advanced glycation end-products pathway, anticoagulants, and halogenated agents) and outcomes compared to a control condition. The primary outcome will be a composite of the four key features of preclinical ARDS as per the American Thoracic Society consensus conference (histologic evidence of lung injury, altered alveolar-capillary barrier, lung inflammatory response, and physiological dysfunction). Secondary outcomes will include the single components of the primary composite outcome, net alveolar fluid clearance, and death. MEDLINE, Embase, and Cochrane databases will be searched electronically and data from eligible studies will be extracted, pooled, and analyzed using random-effects models. Individual study reporting will be assessed according to the Animal Research: Reporting of In Vivo Experiments guidelines. Meta-regressions will be performed to identify subphenotypes prior to comparing outcomes across subphenotypes and treatment effects. (Continued on next page) * Correspondence: mjabaudon@chu-clermontferrand.fr 1 Department of Perioperative Medicine, CHU Clermont-Ferrand, Clermont-Ferrand, France 3 GReD, CNRS UMR 6293, INSERM U1103, Université Clermont Auvergne, Clermont-Ferrand, France Full list of author information is available at the end of the article © The Author(s). 2020 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits 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/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Carla et al. Respiratory Research (2020) 21:81 Page 2 of 10 (Continued from previous page) Discussion: This study will inform on the presence and underlying pathophysiological features of subphenotypes among preclinical models of ARDS and should help to determine whether sufficient evidence exists to perform preclinical trials of subphenotype-targeted therapies, prior to potential clinical translation. Systematic review registration: PROSPERO (ID: CRD42019157236). Keywords: Systematic review protocol, Acute respiratory distress syndrome, Preclinical, Subphenotypes Background soluble tumor necrosis factor receptor (sTNFr)-1). These The acute respiratory distress syndrome (ARDS) is a life- two phenotypes are sometimes referred to as “endo- threatening clinical syndrome of rapid-onset pulmonary types” because they seem defined by distinct degrees of failure characterized by a dysregulated inflammatory re- inflammation, but their underlying mechanisms remain sponse that results in severe respiratory failure and the largely unknown. However, these subphenotypes were need for mechanical ventilation [1]. Although the preva- stable over the first 3 days of enrolment in both RCTs, lence of ARDS is low (1–5/10,000 inhabitants), its inci- suggesting that they are linked to true biological pro- dence is high in critically ill patients, as the syndrome is cesses. The hyper-inflammatory subphenotype demon- present in approximately 10% of patients upon admission strated significantly worse outcomes when compared to to the intensive care unit (ICU) and in 24% of ICU pa- the hypo-inflammatory subphenotype, with higher mor- tients under mechanical ventilation support [2]. Despite tality and lesser ventilator-free and organ failure-free decades of research, mortality due to ARDS remains high, days. The analysis of plasma biomarkers in patients with with significant long-term comorbidity and reduced qual- ARDS previously enrolled in the MARS prospective ity of life in survivors [2]. Various pharmacotherapeutic cohort also identified two biological subphenotypes, which agents have failed to improve ARDS outcomes and were very similar to those found earlier and were again current treatment strategy is largely supportive, based on associated with distinct mortality rates [10]. In the AL- optimized ventilatory settings [1]. A key reason for the VEOLI trial, where a lower positive end-expiratory pres- lack of specific pharmacological therapy is likely due to sure (PEEP) strategy was compared to a higher PEEP the heterogeneity within ARDS, which is diagnosed by a strategy, the two subphenotypes demonstrated a distinct constellation of signs as defined by the Berlin criteria [3]. response in terms of major outcomes, suggesting the po- The clinical and biological heterogeneity within syn- tential for using this subphenotypic classification to iden- dromes such as sepsis ARDS makes it essential to identify tify therapy responsive traits. This hypothesis was further more homogeneous subphenotypes when investigating confirmed through secondary analysis of the fluids and potential therapies [4, 5]. There has been recent recogni- catheters treatment trial (FACTT) in which the two in- tion of distinct subphenotypes (on the basis of clinical/bio- flammatory subphenotypes had distinct effects of the ran- chemical variables, natural history, disease manifestation, domly assigned fluid management strategy on 90-day and/or treatment response without any implication about mortality [11]. The potential for the inflammation-based mechanism) and endotypes (defined by a distinct func- subphenotypic classification was again demonstrated in a tional/pathobiological mechanism) within patients with secondary analysis of the hydroxymethylglutaryl-CoA ARDS [4, 6, 7]. Therefore, while benefit has been demon- reductase inhibition with simvastatin in acute lung injury strated in numerous preclinical studies for many candidate to reduce pulmonary dysfunction–2 (HARP-2) trial, a therapies, most have failed to translate to improved patient multicenter trial investigating the potential of simvastatin outcomes in randomized clinical trials (RCTs), suggesting as an anti-inflammatory therapy for ARDS. Subphenotyp- that the appropriate subset of patients to target with the ing based on the hypo- versus hyper-inflammatory profile novel therapies may not have been correctly identified [8]. identified a therapy response trait as evidenced by the The retrospective analysis of data from two large mul- improved survival of patients classified as hyper- ticenter RCTs (the “ventilation with lower tidal volumes inflammatory and randomized to simvastatin [12]. as compared with traditional tidal volumes for acute Translation of basic research findings to clinical prac- lung injury and the acute respiratory distress syndrome tice remains daunting because of the heterogeneity and [ARMA] and “assessment of low tidal volume and ele- complexity of ARDS [13]. A substantial number of pre- vated end-expiratory volume to obviate lung injury” [AL- clinical models of experimental lung injury have been VEOLI] trials) [9] has first identified at least two ARDS used to investigate mechanisms of lung injury and test subphenotypes within clinical ARDS which were differ- novel therapies [14–16]. Although none of these experi- entiated by the presence of shock, metabolic acidosis mental models fully reproduces all features of human and a higher inflammatory status (interleukin (IL)-6 and lung injury, most of them are based on the reproduction, Carla et al. Respiratory Research (2020) 21:81 Page 3 of 10 in animals, of known risk factors for ARDS, such as pul- Table 1 Preclinical models of acute respiratory distress monary or extrapulmonary sepsis, lipid embolism such syndrome included in the review as when secondary to bone fracture, and acid aspiration, Oleic acid among other clinical risks [14, 15, 17]. Recently, more Lipopolysaccharides translational models of acute lung injury have been used, Acid aspiration such as the ex vivo human lung preparation [18–21]. Al- Hyperoxia though the nature of the insult leading to acute lung in- Saline lavage jury in preclinical models grossly influences one or more features of human ARDS, the presence of subphenotypes Pulmonary ischemia/reperfusion among experimental models of ARDS has never been Nonpulmonary ischemia/reperfusion systematically assessed to date. Intravenous bacteria Intrapulmonary bacteria Study question Peritonitis We hypothesize that ARDS subphenotypes may exist Cecal ligation and puncture among preclinical ARDS studies and that they may have distinct treatment response to select therapies. Trauma/Shock To test this hypothesis, we designed a systematic re- Mechanical ventilation (ventilator-induced lung injury) view and meta-analysis of controlled preclinical trials in Chemical or chemotherapeutic injury (such as detergents) experimental, − in vivo (animal) and ex vivo (human Pancreatitis lung preparation) -, models of ARDS. The primary goal Burns of the review is to compare the key features of preclin- ical ARDS, as defined by the American Thoracic Society (ATS) consensus conference (histologic evidence of lung Preclinical model eligibility criteria injury, altered alveolar-capillary barrier, lung inflamma- We will include preclinical in vivo animal and ex vivo hu- tory response, and physiological dysfunction) [15], in man models of experimentally induced acute lung injury lung-injured experimental groups treated or not with that mimic at least some aspects of the pathophysiology of any of some prespecified candidate therapies. humans with ARDS according to the ATS consensus cri- teria [15]. The inclusion of a wide range of acute lung in- Methods jury models should enhance the generalizability of our Protocol and registration study findings. Acute lung injury in experimental models The systematic review and meta-analysis protocol origi- may be induced by several methods (Table 1). These in- nated from discussions between our team of translational clude intravenous, intratracheal or intraperitoneal admin- research scientists (AC, JA, RB, DM, VS, JMC, MJ), infor- istration of bacteria or endotoxin, induction of injury by mation system specialists (NPD, MDC), a statistical and the ventilator (ventilator-induced lung injury), chemical methodologist (BP), and the European Society of Intensive agents (oleic acid, hydrochloric acid), trauma, shock (for Care Medicine (ESICM) Translational Biology Group of example hemorrhagic or induced by sepsis), or remote the Acute Respiratory Failure section. The protocol was organ injury (for example pancreatitis, ischemia- registered on PROSPERO (ID: CRD42019157236) in reperfusion). We will exclude neonatal animal models of December 2019. acute lung injury because the mechanisms of disease, and efficacy of treatments, are likely to be different in this spe- Study eligibility criteria cific setting. We will include controlled comparative studies (ran- domized and non-randomized) of preclinical acute lung Interventions injury/ARDS published from 2009 to 2019 that evaluate Preclinical intervention groups will include animals and the efficacy and safety outcomes of a number of prede- ex vivo human lungs from studies that examine the fined therapies (stem cells, fibroblast growth factors, IL- therapeutic use of stem cells (and their derived microve- 2, IL-10, beta-agonists, corticosteroids, modulators of sicles), fibroblast growth factors, IL-2, IL-10, beta- the receptor for advanced glycation end-products agonists, corticosteroids, RAGE modulators, anticoagu- (RAGE) pathway, anticoagulants, and halogenated lants, and halogenated agents such as isoflurane, sevo- agents). In order to collect sufficient data to test our hy- flurane, and desflurane (Table 2). These interventions potheses, it was decided a priori to include only studies have been selected a priori because they are the main in which at least two features, out of the four key fea- areas of interest of our research groups [22–30]. In tures of preclinical ARDS proposed by the 2011 Official addition to studies of therapies that are administered ATS Workshop Report [15], are assessed. after lung injury is modeled, experiments using Carla et al. Respiratory Research (2020) 21:81 Page 4 of 10 Table 2 Predefined list of select therapies included in the meta-analysis Class Example of therapies Interleukins IL-10, IL-2 Cell therapy MSCs, MSC-derived microvesicles Adrenergic beta-agonists Salbutamol, albuterol, sultanol, proventil Growing factors Fibroblast growing factors, keratinocyte growth factor, endothelial growth factor RAGE modulators Anti-RAGE antibody, recombinant soluble RAGE, gene knockout, endogenous secretory RAGE Corticosteroids Glucocorticoids, mineralocorticoids, 3-oxo steroid, 17-ketosteroids, domolene, cortifan, epicortisol, komed-hc, heb-cort, prednisone, encortone, panasol, deltacortisone, decortisyl, decortin, predeltin, orasone, prednidib, prednisone, dacortin, cortancyl, solupred, deltasone, pronisone, sterapred, encorton, panafcort, delta-dome, dehydrocortisone, prednisonum, lisacort, ultracorten, winpred, metacortandracin, meticorten, prednisolone, predate, prednisolone, di-adreson f, delta-1- hydrocortisone, prednisolonum, sterane, meti derm, metacortandralone, predonine Anticoagulant agents Heparin, antithrombin, plasminogen activator, fibrinolytic agents, tissue factor pathway inhibitor, lipoprotein-associated coagulation inhibitor, thrombin-thrombomodulin complex, thrombin receptor, activated protein C, drotrecogin alfa activated Halogenated agents Sevoflurane, desflurane, isoflurane IL interleukin; MSCs mesenchymal stem (stromal) cells; RAGE receptor for advanced glycation end-products pretreatment will also be included since they are clinic- Because the assessment of our composite endpoint ally relevant for the prevention of ARDS. Moreover, must reflect that the timing of experimental induction of studies using a co-treatment or multiple injurious hits ARDS may vary across studies, the primary endpoint will will be included. be analyzed in categories of time from the induction of ARDS of less than 6 h, between 6 and 24 h, and greater Comparison than 24 h. These time points for measurement were se- The preclinical comparison group will include data from lected since the development of inflammation and acute studies that have had experimental acute lung injury in- lung injury occurs over time and contributes to death duced but have not been administered any of the prespe- and morbidity in this population [2]. cified therapies listed in the previous paragraph. This will allow to perform effect size calculations in our Preclinical secondary endpoints meta-analysis to examine how efficacious and safe these Secondary endpoints will include death, net alveolar therapies are in preclinical ARDS in general, and in pre- fluid clearance, and each individual component of the clinical ARDS subphenotypes in particular. In addition, primary composite endpoint. The same categories of we will use other control groups, such as healthy or time will be used for primary and secondary endpoints. sham-injured controls, to examine the severity of pre- Where reported, the occurrence of adverse events will clinical ARDS. also be documented for each included study. For animal models that use infectious induction of acute lung in- Preclinical primary endpoint jury, we will describe pathogen clearance (for example The primary endpoint is a composite of the four key fea- number of bacterial colony-forming units). tures that define preclinical ARDS, with specific mea- surements defined by the consensus criteria proposed by Information sources the ATS [15]. Search strategies and equation of searching have been These include histological evidence of tissue injury developed and tested with the help of two librarians (usually based on the lung injury score), alteration of the using the Peer Review of Electronic Search Strategies alveolar-capillary barrier (for example increased concen- (PRESS) template (see Additional file 1 for the represen- tration of high molecular weight proteins in bronchoal- tative search strategy) [31]. We will search on MED- veolar fluid), measures of the inflammatory response in LINE, Embase, and Cochrane databases. Search the lung (for example pulmonary leukocytes or neutro- strategies will use a combination of controlled vocabu- phils, cytokines), and measures of physiological dysfunc- lary (for example corticoids, mesenchymal stem cells, tion such as the alveolar-arterial gradient of oxygen acute lung injury, respiratory distress syndrome) and concentration and the partial pressure of arterial oxygen keywords (for example mesenchymal stromal cells, acute (PaO2) to fraction of inspired oxygen (FiO2) ratio. These lung injury, acute/adult respiratory distress syndrome). quantitative features and measurements are summarized Vocabulary and syntax will be adjusted across the data- in Table 3. We will focus on the “very relevant” mea- bases. Two recently published animal filters [32, 33], val- surements defined by the consensus criteria [15]. idated for PubMed/MEDLINE and Embase and Carla et al. Respiratory Research (2020) 21:81 Page 5 of 10 Table 3 Acute lung injury features and measurements Features Relevant measurements Histological evidence of lung injury Accumulation of neutrophils in the alveolar/interstitial space Formation of hyaline membranes Proteinaceous debris in alveolar space Proteinaceous debris in alveolar space Thickening of the alveolar wall Injury by a standardized histology score Evidence of hemorrhage Areas of atelectasis Gross macroscopic changes such as discoloration of the lungs Alteration of the alveolar capillary barrier Increased extravascular lung water content Accumulation of protein/tracer in airspaces/extravascular space Total BAL protein concentration BAL concentration of high molecular weight proteins Vascular filtration coefficient Translocation of a protein from the airspaces into plasma Increased lung lymph flow, lymph protein concentration Inflammation BAL total neutrophil counts Lung MPO activity Concentrations of cytokines and chemokines in lung tissue, in BAL fluid, and/or in plasma Physiological dysfunction Hypoxemia Increased A-a oxygen difference BAL bronchoalveolar lavage; MPO myeloperoxidase amended slightly, will be used to increase relevancy. for validated data entry; (2) audit trails for tracking data These filters will be adjusted for use in the other data- manipulation and export procedures; (3) automated bases where a validated filter is unavailable. Language export procedures for seamless data downloads to com- will be restricted to English studies and no studies pub- mon statistical packages; and (4) procedures for import- lished before 2009 will be included. The bibliographies ing data from external sources. Specific data elements of included studies and relevant reviews will also be are listed in Table 4. hand searched for further preclinical studies. Assessment of risk of bias Study selection The risk of bias will be evaluated by two reviewers (AC, The titles and abstracts of search results will be screened MJ) for each included preclinical study. Since no vali- independently by two investigators (AC, MJ). The full text dated and standardized risk of bias checklist exists for of all potentially eligible studies will be retrieved and preclinical studies, we will describe the biases of the in- reviewed for eligibility. Disagreements between reviewers cluded studies using the Cochrane risk of bias assess- will be resolved by consensus or by a third member of the ment [36] and the Risk Of Bias In Nonrandomized systematic review team (JA). Reasons for exclusion of po- Studies - of Interventions (ROBINS-I) [37] tools. Items tentially eligible studies will be recorded to enable a trans- in this tool include assessments for concealment of allo- parent selection process and to be in accordance with the cation, random sequence generation, blinding of Preferred Reporting Items for Systematic Reviews and personnel and the endpoint measurements, and com- Meta-Analyses (PRISMA) guidelines developed for proper pleteness of endpoint reporting. Each bias criterion will reporting of clinical systematic reviews [34]. The Preferred be assigned a value of low, high, or unclear risk of bias Reporting Items for Systematic Reviews and Meta- for each included study [36]. Analyses Protocols (PRISMA-P) checklist for this review protocol is available as Additional file 2. Assessment of construct validity and external validity We will also record features that will facilitate judge- Data collection and process and data items ments of construct validity and external validity [38]. Data will be extracted by four members (AC, JA, RZ, Construct validity in preclinical research concerns the MJ) of the research team and collected into standard- extent to which an experimental system accurately ized, pilot-tested forms using REDCap electronic data models a clinical entity. These will include: age, gender, capture tools hosted at CHU Clermont-Ferrand [35]. weight, species, and strain of animal; presence of comor- REDCap (Research Electronic Data Capture) is a secure, bidities; type of acute lung injury model; timing, dose web-based application designed to support data capture and mode of administration of treatment; and use of co- for research studies, providing: (1) an intuitive interface interventions (for example fluid resuscitation, type of Carla et al. Respiratory Research (2020) 21:81 Page 6 of 10 Table 4 Data collection elements Categories Main items Study characteristics DOI, Num ID (specific number ID for the purpose of the meta-analysis), title, author, year, country and journal of publication, country of study, mono/multicentric Study population (experimental model) Animal species or ex vivo model, age, gender, strain, weight, and comorbidity Type of acute lung injury model As defined in Table 1, with timing Intervention Type (see Table 2), number, timing, mode of administration, dose Co-intervention Resuscitation fluids, antibiotics, mechanical ventilation parameters, mode of anesthesia, drugs, etc. Preclinical endpoints As defined in Table 3 Risk of bias assessments (Cochrane risk of bias Randomization, allocation, blinding and completeness of follow-up tool) Quality of reporting of individual preclinical ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines elements studies DOI digital object identifier anesthesia, use of antibiotics for infectious acute lung in- systematic review will be presented at relevant confer- jury models) (Table 5). External validity in preclinical re- ences and published in a peer-reviewed journal. search concerns the extent to which cause and effect relationships holdup under varied conditions [39], such Data analysis as the use of a multicenter preclinical design. Statistical analyses will be performed using Stata (version 15; StataCorp, College Station, US), Comprehensive Meta-Analysis (version 2; Biostat, Englewood, NJ), and Description of reporting Mplus [41] softwares. For descriptive analyses, data will We will describe the quality of reporting of the included be presented as mean and standard deviation or median preclinical studies using the elements of the Animal Re- and interquartile range, according to statistical distribu- search: Reporting of In Vivo Experiments (ARRIVE) tion. To address the non-independence of data due to guidelines. The ARRIVE guidelines were developed to study effect, random-effects models will be performed enhance the transparent and comprehensive reporting of rather than fixed-effect as certain experimental parame- research methods and results for in vivo animal experi- ters can have wide variation. The dichotomous end- ments [40]. points from each included study will be pooled and described as relative risks (RR) and 95% confidence in- Ethics and dissemination tervals incorporating a random effects modeling ap- This study does not require ethical approval. Data col- proach with the use of forest plots for presentation of lected through this systematic review will be managed the data [42]. Continuous endpoints will be pooled using by the ESICM Translational Biology Group and made the ratio of weighted means method with inverse available upon reasonable request. The results of this variance random effects modeling and Hedges’s Table 5 Elements of construct validity and external validity Category Items Animal species and strain (for mice) Human lung preparation, mouse, rat, sheep, pig, rabbit, dog or others Animal age Young, middle-aged, and mature, depending on species Animal gender Male, female, both Comorbidity Yes or no, type Model of acute lung injury As defined in Table 1 Severity of lung injury Lung injury score Intervention type The drug, mode of administration and dose Timing of intervention regarding to acute lung injury induction Before lung injury, 0 to 1, 1 to 6, and > 6 h after lung injury Type of control Injured, non-injured Use of co-interventions Resuscitation fluids, use of antibiotics, type of anesthesia, mechanical ventilation strategy Number of study centers Single versus multi-center Carla et al. Respiratory Research (2020) 21:81 Page 7 of 10 standardized mean differences (SMD) [43]. When neces- possible, to assess the effects of the nature, sequence, and sary, the RRs reported in individual studies will be con- timing of all the injurious hits that are used in studies verted into SMD [44]. The statistical heterogeneity of based on multiple-hit models of experimental ARDS. included preclinical studies will be measured using the I2 [45], which will be used to measure heterogeneity with Discussion 25, 50 and 75% indicating low, moderate and high het- The results of this systematic review and meta-analysis erogeneity, respectively. Egger’s test and funnel plots will will inform translational and clinical scientists, re- be used to assess publication bias. In the absence of bias, searchers, and clinicians internationally regarding the studies will be distributed evenly around the mean effect existing preclinical evidence for subphenotypes among size because of random sampling error. preclinical, in vivo or ex vivo models of ARDS. Such data Then, meta-regression will be conducted, with covari- will be crucial to confirm the potential of personalized ates determined according to the aforementioned analyses ARDS medicine [6, 8, 46], i.e. the promise to deliver the and to clinical and/or biological relevance, and methodo- right -targeted- treatment to the right subject. These logical issues such as the categories of time from ARDS data will also inform on the potential value of testing induction. Particular attention will be paid to the study of candidate therapies (including therapies that have multicollinearity and interactions between covariates 1) already been tested in previous studies) that target dis- studying the relationships between the covariates and 2) tinct functional and/or pathobiological processes (such evaluating the impact of adding or removing variables to as lung epithelial injury and alveolar inflammation [9, or from the multivariable model. Whenever possible in 47]) in select experimental models that might best appropriate data, latent class analyses will be carried out mimic ARDS subphenotypes described in patients to to identify subphenotypes within experimental models of date [6]. For example, although it seems plausible that ARDS, associated with distinct outcomes and differential tracheal instillation with hydrochloric acid causes prom- responses to therapy. Statistical analyses will be conducted inent injury to the alveolar epithelium and live bacteria including all controlled comparative studies, with the type (such as Streptococcus pneumoniae) causes major alveo- of study considered as stratification variable (randomized lar inflammation profiles, it remains unknown whether and non-randomized). these models might have reliable value to represent clin- When the number of studies and sample size will per- ical subphenotypes of higher inflammation and of higher mit, analyses will be performed by study type. Several degree of lung epithelial injury, respectively [6]. subgroup analyses to examine preclinical heterogeneity Our review is timely since there is an increasing will be conducted on the primary endpoint. These ana- amount of research dedicated to the identification of lyses will include “clinical”/biological features (such as ARDS subphenotypes, their underlying mechanisms, and the type of experimental model; animal age, gender, their potential responsive treats [4, 7, 9–12, 48–52]. The weight, species, and strain; presence of comorbidities; results from this study should therefore be useful in cause of experimental ARDS; severity of the ARDS order to generate hypotheses on whether some therapies model; type of tested therapies (Table 2); type of con- (see Table 2) might differentially benefit to some pre- trols; type of anesthesia; use of co-interventions, antibi- clinical subphenotypes, as compared to others. In case of otics, and mechanical ventilation settings; multiple-hit positive results, this might open new perspectives on versus single-hit model of ARDS) and methodological preclinical precision therapies, with the ultimate goal of features (such as single versus multi-center study, pres- future clinical translation. However, the predefined, lim- ence of an a priori sample size calculation). These sub- ited number of therapies investigated through our review group analyses will be exploratory and the results will be (see Table 2) is a limitation that should be acknowl- interpreted with caution. In addition, subphenotypes edged; this limited number reflects both our specific re- that may be identified among preclinical studies will be search topics of interest and a balance between scientific compared to those already described in Humans (and validity and feasibility. Our results will inform on those that could be identified in the future) using the whether the nature of the lung injurious hit used in pub- same set of variables than in patients such as plasma lished experimental models of ARDS could identify sub- levels of some proinflammatory cytokines, when applic- phenotypes with distinct degrees of severity and able [6, 7]. treatment effects. Yet, we might not be able to distin- Sensitivity analyses will be conducted to assess how in- guish such subphenotypes using individual subject data cluding and excluding studies influences the results. More analysis in a given experimental model, such as recently precisely, sensitivity analyses will be performed to measure reported in a secondary analysis of a murine cecal the impact of high heterogeneity or a methodological ligation and puncture sepsis model, with one subpheno- quality estimated too low, and risk of bias assessments. In type characterized by faster deterioration and more se- addition, sensitivity analyses will be performed, when vere inflammation than the other and distinct responses Carla et al. Respiratory Research (2020) 21:81 Page 8 of 10 to some interventions (immediate versus delayed antibi- Availability of data and materials otics and fluids) [53]. Data collected through this systematic review will be managed by the ESICM Translational Biology Group and made available upon reasonable request. In a broader perspective, we also hope this review will increase our knowledge on, and identify challenges and Ethics approval and consent to participate barriers related to, the conduct of preclinical ARDS Not applicable. studies. Ultimately, it will also inform and enrich future preclinical and clinical precision, subphenotype-based, Consent for publication Not applicable. research that should aid the translation of novel therap- ies for ARDS. Competing interests The authors declare that they have no competing interests. Supplementary information Author details Supplementary information accompanies this paper at https://doi.org/10. 1 Department of Perioperative Medicine, CHU Clermont-Ferrand, 1186/s12931-020-01337-9. Clermont-Ferrand, France. 2Biostatistics Unit, Department of Clinical Research and Innovation (DRCI), CHU Clermont-Ferrand, Clermont-Ferrand, France. 3 Additional file 1. Representative Search Strategy. GReD, CNRS UMR 6293, INSERM U1103, Université Clermont Auvergne, Additional file 2. Preferred reporting items for systematic review and Clermont-Ferrand, France. 4Université Clermont Auvergne, Health Library, meta-analysis protocols (PRISMA-P) checklist. Clermont-Ferrand, France. 5Department of Medical Biochemistry and Molecular Biology, CHU Clermont-Ferrand, Clermont-Ferrand, France. 6 Division of Allergy, Pulmonary, and Critical Care Medicine, Department of Medicine, Vanderbilt University Medical Center, Nashville, TN, USA. Abbreviations 7 Department of Pathology, Microbiology, and Immunology, Vanderbilt ALVEOLI: Assessment of low tidal volume and elevated end-expiratory vol- University Medical Center, Nashville, TN, USA. 8Division of Pulmonary and ume to obviate Lung injury; ARDS: Acute respiratory distress syndrome; Critical Care Medicine, Departments of Medicine and Anesthesia, ARMA: Ventilation with lower tidal volumes as compared with traditional Cardiovascular Research Institute, University of California San Francisco, San tidal volumes for acute lung injury and the acute respiratory distress Francisco, CA, USA. 9Department of Cell and Developmental Biology, syndrome; ARRIVE: Animal research: reporting of in vivo experiments; Vanderbilt University, Nashville, TN, USA. 10Keenan Research Centre for ATS: American thoracic society; ESICM: European society of intensive care Biomedical Science, Hospital for Sick Children, Departments of Anesthesia medicine; FACTT: Fluids and catheters treatment trial; FiO2: Fraction of and Critical Care Medicine, St. Michael’s Hospital, Departments of Anesthesia, inspired oxygen; HARP-2: Hydroxymethylglutaryl-CoA reductase inhibition Physiology and Interdepartmental Division of Critical Care, University of with simvastatin in acute lung injury to reduce pulmonary dysfunction–2; Toronto, Toronto, Canada. 11Regenerative Medicine Institute at CÚRAM ICU: Intensive care unit; IL: Interleukin; PaO2: Partial pressure of arterial Centre for Research in Medical Devices, National University of Ireland Galway, oxygen; PEEP: Positive end-expiratory pressure; PRESS: Peer review of Galway, Ireland. 12Department of Intensive Care Medicine, Department of electronic search strategies; PRISMA: Preferred reporting items for systematic Respiratory Medicine, and Laboratory of Experimental Intensive Care and reviews and meta-analyses; PRISMA-P: Preferred reporting items for Anesthesiology, Amsterdam University Medical Center, Amsterdam, The systematic reviews and meta-analyses protocols; REDCAP: Research electronic Netherlands. 13Corporació Sanitaria Parc Tauli, CIBER de Enfermedades data capture; RAGE: Receptor for advanced glycation end-products; Respiratorias, Autonomous University of Barcelona, Barcelona, Spain. RCT: Randomized clinical trial; ROBINS-I: Risk of bias in nonrandomized 14 Laboratory of Pulmonary Investigation, Carlos Chagas Filho Institute of studies - of interventions; RR: Relative risk; SMD: Standardized mean Biophysics, Federal University of Rio de Janeiro, Rio de Janeiro, Brazil. differences; sTNFr-1: Soluble tumor necrosis factor receptor 1 15 Wellcome-Wolfson Institute for Experimental Medicine, Queens University Belfast and Regional Intensive Care Unit, Belfast Health and Social Care Trust, Acknowledgements Belfast, UK. 16Department of Anesthesiology and Critical Care, Sorbonne Not applicable University, GRC 29, AP-HP, DMU DREAM, Pitié-Salpêtrière Hospital, Paris, France. Patient and public involvement (PPI) statement Received: 23 January 2020 Accepted: 17 March 2020 This research was done without patient involvement. Patients were not invited to comment on the study design and were not consulted to develop patient relevant outcomes or interpret the results. Patients were not invited to contribute to the writing or editing of this document for readability or References accuracy. 1. Thompson BT, Chambers RC, Liu KD. Acute respiratory distress syndrome. N Engl J Med. 2017;377:562–72. 2. Bellani G, Laffey JG, Pham T, Fan E, Brochard L, Esteban A, et al. Authors’ contributions Epidemiology, patterns of care, and mortality for patients with acute AC, BP, DM, and MJ conceived the protocol design for this systematic respiratory distress syndrome in intensive care units in 50 countries. JAMA. review, and MJ is its guarantor. AC, BP, and MJ were responsible for initial 2016;315:788–800. drafting and manuscript revisions. AC, HB, NPD, MDC, RZ, JA, and MJ were 3. Definition Task Force ARDS, Ranieri VM, Rubenfeld GD, Thompson BT, responsible for the data collection strategy and preparation. CL and BP Ferguson ND. Caldwell E, et al. acute respiratory distress syndrome: the provided critical revisions and will be responsible for data management and Berlin Definition. JAMA. 2012;307:2526–33. statistical analysis. JMC, VS, LBW, CSC, JAB, JGL, NPJ, LDB, AA, PRMR, MAM, 4. Prescott HC, Calfee CS, Thompson BT, Angus DC, Liu VX. Toward smarter and DMA provided critical revisions in the design of the review protocol, lumping and smarter splitting: rethinking strategies for Sepsis and acute shared their expertise in translational biology, and provided oversight for respiratory distress syndrome clinical trial design. Am J Respir Crit Care Med. planned eligibility criteria and outcome measures. AC and MJ will oversee 2016;194:147–55. knowledge translation. All authors reviewed several drafts of the manuscript 5. Seymour CW, Kennedy JN, Wang S, Chang C-CH, Elliott CF, Xu Z, et al. and approved the final version. Derivation, Validation, and Potential Treatment Implications of Novel Clinical Phenotypes for Sepsis. JAMA. 2019;321(20):2003–17. Funding 6. Jabaudon M, Blondonnet R, Audard J, Fournet M, Godet T, Sapin V, et al. This research received no specific grant from any funding agency in the Recent directions in personalised acute respiratory distress syndrome public, commercial or not-for-profit sectors. medicine. Anaesth Crit Care Pain Med. 2018;37:251–8. Carla et al. Respiratory Research (2020) 21:81 Page 9 of 10 7. Sinha P, Calfee CS. Phenotypes in acute respiratory distress syndrome: 30. Jabaudon M, Boucher P, Imhoff E, Chabanne R, Faure J-S, Roszyk L, et al. moving towards precision medicine. Curr Opin Crit Care. 2019;25:12–20. Sevoflurane for sedation in acute respiratory distress syndrome. A 8. Beitler JR, Goligher EC, Schmidt M, Spieth PM, Zanella A, Martin-Loeches I, randomized controlled pilot study. Am J Respir Crit Care Med. 2017;195: et al. Personalized medicine for ARDS: the 2035 research agenda. Intensive 792–800. Care Med. 2016;42:756–67. 31. McGowan J, Sampson M, Salzwedel DM, Cogo E, Foerster V, Lefebvre C. 9. Calfee CS, Delucchi K, Parsons PE, Thompson BT, Ware LB, Matthay MA, et al. PRESS peer review of electronic search strategies: 2015 guideline statement. Subphenotypes in acute respiratory distress syndrome: latent class analysis of J Clin Epidemiol. 2016;75:40–6. data from two randomised controlled trials. Lancet Respir Med. 2014;2:611–20. 32. Hooijmans CR, Tillema A, Leenaars M, Ritskes-Hoitinga M. Enhancing search 10. Bos LD, Schouten LR, van Vught LA, Wiewel MA, Ong DSY, Cremer O, et al. efficiency by means of a search filter for finding all studies on animal Identification and validation of distinct biological phenotypes in patients with experimentation in PubMed. Lab Anim. 2010;44:170–5. acute respiratory distress syndrome by cluster analysis. Thorax. 2017;72:876–83. 33. de Vries RBM, Hooijmans CR, Tillema A, Leenaars M, Ritskes-Hoitinga M. A 11. Famous KR, Delucchi K, Ware LB, Kangelaris KN, Liu KD, Thompson BT, et al. Acute search filter for increasing the retrieval of animal studies in Embase. Lab respiratory distress syndrome subphenotypes respond differently to randomized Anim. 2011;45:268–70. fluid management strategy. Am J Respir Crit Care Med. 2017;195:331–8. 34. Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gøtzsche PC, Ioannidis JPA, et al. 12. Calfee CS, Delucchi KL, Sinha P, Matthay MA, Hackett J, Shankar-Hari M, The PRISMA statement for reporting systematic reviews and meta-analyses et al. Acute respiratory distress syndrome subphenotypes and differential of studies that evaluate health care interventions: explanation and response to simvastatin: secondary analysis of a randomised controlled trial. elaboration. J Clin Epidemiol. 2009;62:e1–34. Lancet Respir Med. 2018;6:691–8. 35. Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research 13. Matthay MA, Zemans RL, Zimmerman GA, Arabi YM, Beitler JR, Mercat A, electronic data capture (REDCap)--a metadata-driven methodology and et al. Acute respiratory distress syndrome. Nat Rev Dis Primers. 2019;5:18. workflow process for providing translational research informatics support. J 14. Matute-Bello G, Frevert CW, Martin TR. Animal models of acute lung injury. Biomed Inform. 2009;42:377–81. Am J Physiol Lung Cell Mol Physiol. 2008;295:L379–99. 36. Higgins JPT, Altman DG, Gøtzsche PC, Jüni P, Moher D, Oxman AD, et al. 15. Matute-Bello G, Downey G, Moore BB, Groshong SD, Matthay MA, Slutsky The Cochrane Collaboration’s tool for assessing risk of bias in randomised AS, et al. An official American Thoracic Society workshop report: features trials. BMJ. 2011;343:d5928. and measurements of experimental acute lung injury in animals. Am J 37. Sterne JA, Hernán MA, Reeves BC, Savović J, Berkman ND, Viswanathan M, Respir Cell Mol Biol. 2011;44:725–38. et al. ROBINS-I: a tool for assessing risk of bias in non-randomised studies of 16. Laffey JG, Kavanagh BP. Fifty years of research in ARDS. Insight into acute interventions. BMJ. 2016;355:i4919. respiratory distress syndrome. From models to patients. Am J Respir Crit 38. Henderson VC, Kimmelman J, Fergusson D, Grimshaw JM, Hackam DG. Care Med. 2017;196:18–28. Threats to validity in the design and conduct of preclinical efficacy studies: 17. Rocco PRM, Nieman GF. ARDS: what experimental models have taught us. a systematic review of guidelines for in vivo animal experiments. PLoS Med. Intensive Care Med. 2016;42:806–10. 2013;10:e1001489. 18. McAuley CDF, Hamid GF, Laffey UI, Abbott JG. McKenna J, et al. clinical 39. Kimmelman J, London AJ. Predicting harms and benefits in translational grade allogeneic human mesenchymal stem cells restore alveolar fluid trials: ethics, evidence, and uncertainty. PLoS Med. 2011;8:e1001010. clearance in human lungs rejected for transplantation. Am J Physiol Lung 40. Kilkenny C, Browne WJ, Cuthill IC, Emerson M, Altman DG. Improving Cell Mol Physiol. 2014;306(9):L809–15. bioscience research reporting: the ARRIVE guidelines for reporting animal 19. Park J, Kim S, Lim H, Liu A, Hu S, Lee J, et al. Therapeutic effects of human research. PLoS Biol. 2010;8:e1000412. mesenchymal stem cell microvesicles in an ex vivo perfused human lung 41. Muthen LK, Muthen B. Mplus Version 8 User’s Guide. Muthen & Muthen; injured with severe E coli pneumonia. Thorax. 2019;74:43–50. 2017. 20. Lee JW, Fang X, Gupta N, Serikov V, Matthay MA. Allogeneic human 42. DerSimonian R, Laird N. Meta-analysis in clinical trials revisited. Contemp mesenchymal stem cells for treatment of E. coli endotoxin-induced acute Clin Trials. 2015;45:139–45. lung injury in the ex vivo perfused human lung. Proc Natl Acad Sci U S A. 43. Friedrich JO, Adhikari NKJ, Beyene J. The ratio of means method as an 2009;106:16357–62. alternative to mean differences for analyzing continuous outcome variables 21. Ross JT, Nesseler N, Lee J-W, Ware LB, Matthay MA. The ex vivo human in meta-analysis: a simulation study. BMC Med Res Methodol. 2008;8:32. lung: research value for translational science. JCI Insight. 2019;4(11). 44. Chinn S. A simple method for converting an odds ratio to effect size for use 22. Matthay MA, Calfee CS, Zhuo H, Thompson BT, Wilson JG, Levitt JE, et al. in meta-analysis. Stat Med. 2000;19:3127–31. Treatment with allogeneic mesenchymal stromal cells for moderate to 45. Higgins JPT, Thompson SG. Quantifying heterogeneity in a meta-analysis. severe acute respiratory distress syndrome (START study): a randomised Stat Med. 2002;21:1539–58. phase 2a safety trial. Lancet Respir Med. 2019;7:154–62. 46. Meyer NJ, Calfee CS. Novel translational approaches to the search for 23. Matthay MA, Goolaerts A, Howard JP, Lee JW. Mesenchymal stem cells for precision therapies for acute respiratory distress syndrome. Lancet Respir acute lung injury: preclinical evidence. Crit Care Med. 2010;38:S569–73. Med. 2017;5:512–23. 24. Ware LB, Landeck M, Koyama T, Zhao Z, Singer J, Kern R, et al. A 47. Jabaudon M, Blondonnet R, Pereira B, Cartin-Ceba R, Lichtenstern C, Mauri T, randomized trial of the effects of nebulized albuterol on pulmonary edema et al. Plasma sRAGE is independently associated with increased mortality in in brain-dead organ donors. Am J Transplant. 2014;14:621–8. ARDS: a meta-analysis of individual patient data. Intensive Care Med. 2018; 25. National Heart, Lung, and Blood Institute Acute Respiratory Distress 44:1388–99. Syndrome (ARDS) Clinical Trials Network, Matthay MA, Brower RG, Carson S, 48. Sinha P, Delucchi KL, Thompson BT, McAuley DF, Matthay MA, Calfee CS, Douglas IS, Eisner M, et al. Randomized, placebo-controlled clinical trial of et al. Latent class analysis of ARDS subphenotypes: a secondary analysis of an aerosolized β2-agonist for treatment of acute lung injury. Am J Respir the statins for acutely injured lungs from sepsis (SAILS) study. Intensive Care Crit Care Med. 2011;184:561–8. Med. 2018;44:1859–69. 26. Blondonnet R, Audard J, Belville C, Clairefond G, Lutz J, Bouvier D, et al. 49. Bos LDJ, Scicluna BP, Ong DSY, Cremer O, van der Poll T, Schultz MJ. RAGE inhibition reduces acute lung injury in mice. Sci Rep. 2017;7:7208. Understanding heterogeneity in biologic phenotypes of acute respiratory 27. Audard J, Godet T, Blondonnet R, Joffredo J-B, Paquette B, Belville C, et al. distress syndrome by leukocyte expression profiles. Am J Respir Crit Care Inhibition of the receptor for advanced Glycation end-products in acute Med. 2019;200:42–50. respiratory distress syndrome: a randomised laboratory trial in piglets. Sci 50. Kitsios GD, Yang L, Manatakis DV, Nouraie M, Evankovich J, Bain W, et al. Rep. 2019;9:9227. Host-response subphenotypes offer prognostic enrichment in patients with 28. Reddy K, O’Kane C, McAuley D. Corticosteroids in acute respiratory distress or at risk for acute respiratory distress syndrome. Crit Care Med. 2019;47: syndrome: a step forward, but more evidence is needed. Lancet Respir Med. 1724–34. 2020;8(20):30048–5. https://doi.org/10.1016/S2213-2600(20)30048-5 [Epub 51. Viswan A, Ghosh P, Gupta D, Azim A, Sinha N. Distinct metabolic Endotype ahead of print]. mirroring acute respiratory distress syndrome (ARDS) subphenotype and its 29. Camprubí-Rimblas M, Tantinyà N, Bringué J, Guillamat-Prats R, Artigas A. heterogeneous biology. Sci Rep. 2019;9:2108. Anticoagulant therapy in acute respiratory distress syndrome. Ann Transl 52. Sarma A, Calfee CS, Ware LB. Biomarkers and precision medicine: state of Med. 2018;6:36. the art. Crit Care Clin. 2020;36:155–65. Carla et al. Respiratory Research (2020) 21:81 Page 10 of 10 53. Seymour CW, Kerti SJ, Lewis AJ, Kennedy J, Brant E, Griepentrog JE, et al. Murine sepsis phenotypes and differential treatment effects in a randomized trial of prompt antibiotics and fluids. Crit Care. 2019;23:384. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.