key: cord-0978888-hw0g8go3 authors: Sinha, Pratik; Bos, Lieuwe D. title: Pathophysiology of The Acute Respiratory Distress Syndrome: Insights from Clinical Studies date: 2021-05-26 journal: Crit Care Clin DOI: 10.1016/j.ccc.2021.05.005 sha: 4ab41889b537925d26136afd22f3f6ad548ab274 doc_id: 978888 cord_uid: hw0g8go3 Acute respiratory distress syndrome (ARDS) is a heterogeneous clinical syndrome that manifests secondary to numerous aetiological insults and consequently it is associated with a multitude of pathophysiological abnormalities. Despite over 50 years of experimental studies, translation of these benchside discoveries into effective biological therapies has been elusive. Recent advances in high throughput biological sampling, imaging and advances in data analytics has allowed studying ARDS in human subjects based on pragmatic studies. In this review, we present some of the key advances made in our knowledge of the pathophysiology of ARDS, based on histopathology, imaging, protein and transcriptomic biomarkers. Finally, we review the role of such human studies in understanding the pathophysiology of COVID-19 related ARDS. As the pandemic evolves, it is becoming increasingly clear that our knowledge of the pathophysiological drivers in both COVID-19 and non-COVID-19 ARDS is incomplete. Systematic and large-scale data collection that is multi-dimensional and includes samples from several anatomical compartments and over several timepoints is key to unlocking the great systems biology puzzle that is ARDS. regardless of the precipitating insult is clearly not valid, however, most ARDS clinical trials do not discriminate according to aetiology. Consequently, there has been a growing trend towards studying ARDS in human subjects in "real world" conditions based on pragmatic sample acquisition [9, 10] . Advances in novel biological measurements and data science methods has seen a rapid upsurge in translational and clinical studies in human subjects that has brought new insights into the pathophysiology of ARDS. Given the pace of innovation in both these disciplines, we may be entering a new era of learning in ARDS biology based on in vivo human subject studies. Moreover, such translational studies proffer paradigm changing approaches to experimental studies in ARDS, where the traditional linear bench to bedside approach is replace by a cyclical exchange of ideas from these research domains (Figure 1) . In this review, we summarise some of the recent advances made in our understanding of the pathophysiology of ARDS based on human studies. Further, we will review the pathophysiology of lung injury in coronavirus disease (COVID)-19, the disease manifest by SARS-CoV-2. The formalization of a clinical diagnosis for ARDS constitutes a pivotal moment in our understanding of its pathophysiology. In 1994, ARDS was given its first consensus diagnosis at the American-European Consensus Conference (AECC) [11] . In the absence of a tissue or biological diagnosis, investigators in the consensus panel set clinical criteria to diagnose patients with ARDS as acute onset of symptoms, PaO 2 /FiO 2 ≤ 300 mmHg would be classified as J o u r n a l P r e -p r o o f acute lung injury and ≤ 200 mmHg would be called ARDS, bilateral opacification on chest radiograph and a pulmonary occlusion pressure of ≤ 18 mmHg or no evidence of raised left atrial pressure. Since 2012, ARDS is clinically diagnosed using the Berlin definition ( Table 2) which iterated on the AECC definition by introducing three distinct categories of ARDS based on PaO 2 /FiO 2 : mild ≤ 300 mmHg, moderate ≤ 200 mmHg, and severe ≤ 100 mmHg [12] . Acuteness of the symptoms was time-bound to 7 days and a patient must be receiving 5 cmH 2 Therefore, there has been a growing trend towards seeking more uniform subgroups within which to study ARDS biology [13] . First described by Katzenstein and colleagues [14] , the term diffuse alveolar damage (DAD) refers to the histopathological findings of alveolar epithelial and endothelial cell injury with fluid and cellular exudate and presence of hyaline membranes and/or fibrosis. DAD was long been considered the hallmark histological finding in ARDS. Bachofen and Weibel differentiated these histopathological changes in ARDS temporally into three phases (1) exudative (early) phase characterized by interstitial oedema, capillary and neutrophilic infiltrates, (2) proliferative (subacute) phase characterize by proliferation of alveolar type II cells and fibroblast infiltration, (3) fibrotic (late) phase associated with collagen deposition, macrophage J o u r n a l P r e -p r o o f infiltrates and resolution of the exudative phase [15] . More recently, Thille and colleagues studied 159 patients that corroborated these histological phases of ARDS in the presence of DAD [16] , albeit, there is considerably greater overlap between the phases than previously described. It is interesting to note that during the development of the Berlin definitions, the investigators considered DAD a key morphological finding in ARDS and part of the conceptual framework the definition intended to capture. Yet, given the non-specificity of the Berlin definition and its predecessor, the AECC definition, it is likely that while DAD is being captured, so are many other pathological morphologies including those unrelated to ARDS. To that end, only approximately half the patients that meet the clinical criteria for ARDS have DAD on autopsy [17] [18] [19] [20] [21] . Even in open biopsy studies, DAD was observed in the same proportion of patients [22] [23] [24] . Consistent among these studies was that DAD was more prevalent in severe ARDS and associated with worse outcomes. In a meta-analysis of patients that met ARDS criteria and underwent open lung biopsies, Cardinal-Fernandez and colleagues found an array of heterogeneous morphologies in those meeting patients without DAD, with no single entity featuring in greater than 10% of the samples [25] . Among specimens without DAD, most were consistent with histological patterns of infective pneumonia. From these studies, it is difficult to ascertain whether DAD, a consistent finding in experimental animal models of ARDS [26] and in human autopsies pre-AECC/Berlin definition, was inaccurately described historically, or whether the clinical definitions of ARDS are poorly J o u r n a l P r e -p r o o f specific of acute inflammatory lung injury. Further, the clinical utility of histopathology studies is naturally limited due to either being performed at autopsy or necessitating invasive biopsies. Nonetheless, further autopsy studies are needed to better map cellular abnormalities at different phases of ARDS. Incorporating innovative methods to studying lung tissue, such as, next generation sequencing [27] and cryomicro-CT imaging [28] offer opportunities to gain novel insights in ARDS pathophysiology and should be considered in future investigations. The chest radiograph is included in the definition of ARDS to assess the presence of alveolar edema and bilateral opacification is used as a qualitative surrogate. A quantitative assessment of the amount of edema would reflect severity of ARDS. The Lung Injury Score (LIS) was an early attempt at quantification that integrated the number of affected quadrants with physiological parameters into a risk score [29] . LIS was used to enrich the study population in the CESAR trial, which tested veno-venous-ECMO to conventional ventilatory support in severe ARDS [30] . More recently, the Radiographic Assessment of Lung Edema (RALE) score has been developed to further quantify chest X-ray abnormalities in ARDS [31] . The RALE score is calculated by summing the products of the consolidation and density for each radiograph quadrant. The RALE gives a maximal score of 12 for each quadrant resulting in a maximum total score of 48. The consolidation score quantifies the extent of alveolar opacities in each quadrant :(none: 0 points; < 25%: 1 point; 25%-50%: 2 points; 50%-75%: 3 points; > 75%: 4 points), while alveolar J o u r n a l P r e -p r o o f opacification in each quadrant is scored up to 3 points (hazy: 1 point; moderate: 2 points; dense: 3 points) [31] . The RALE score correlated significantly with extravascular lung water in donor lungs and was found to predict survival at the time of ARDS diagnosis. Changing scores over time added to these predictions, where an increasing RALE score had a higher mortality than those with an improving score [32] . Use of the RALE score provides empirical evidence for a common observation in clinical practice, namely that patients with progressive infiltrative abnormalities have worse outcomes. In future trials, the RALE score may be used as a surrogate endpoint for therapeutic response or provide objective prognostic enrichment of patients with ARDS. Chest computer tomography (CT) provides considerable information additional to chest radiographs. CT is considered the gold standard tool for quantification gas-volumes and weight of consolidated lung tissue [33] . With this purpose, it has been used to monitor the effect of recruitment maneuvers on lung volume and re-aeration of consolidations [34] . Since the early days of chest CT, considerable heterogeneity in morphology has been observed in ARDS and for over twenty years investigators have sought methods to identify meaningful subgroups [35] . Three morphological patterns are differentiated: (1) a focal morphology with a basal-dorsal dominance of consolidations, (2) a patchy morphology with "islands" of consolidation or ground glass separated by spared areas throughout all lobes and (3) a diffuse morphology with similar involvement of all lobes without any clear gradient [36] . Patients with patchy and diffuse morphology are nowadays grouped together into a "non-focal" phenotype. Lungs with non-J o u r n a l P r e -p r o o f focal morphology are easier to recruit and less prone to overdistention compared to focal morphology [37] . In the LIVE study, patients were randomized to receive uniform lung protective mechanical ventilation or a lung-morphology driven ventilation. In the "personalized ventilation group" patients with a non-focal lung morphology received small tidal volume of 6 mL/kg predicted body weight (PBW), routine recruitment maneuvers and prone positioning was used as a rescue therapy. Patients with focal lung morphology received higher tidal volumes of 8 mL/kg PBW, lower PEEP strategy and prone positioning was mandatory [38] . The study showed no benefit of the personalized ventilation strategy in the intention to treat analysis. However, the morphological pattern misclassification by the treating physician was 21% and personalized intervention was associated with harmed in this group while the control group was not. Taken together, the results of this study provides a strong warning against premature classification of patients into subphenotypes because of the possibility of harm and the real-world challenges of CT interpretation. Lung ultrasound (LUS) is an attractive alternative to radiation dependent imaging techniques because images can be obtained at the bedside and provide a comprehensive and rapid overview of subpleural lung aeration. The global LUS score correlates well with extra-vascular lung water measured using invasive techniques [39] and can be used to estimate re-aeration of the lung after a recruitment maneuver [40] . Given the non-specificity of chest radiographs and J o u r n a l P r e -p r o o f complexity of CT imaging, an algorithmic approach based on LUS might be an attractive alternative, although this has yet to be systematically investigated. With the increased availability of chest CT scanning, our knowledge about lung aeration and their response to PEEP has increased considerably. Yet, impaired oxygenation secondary to functional shunt encountered in ARDS are insufficiently understood [41] . Perfusion remains the "dark side" of ventilation-perfusion matching due to a lack of tools for anatomical assessment of perfusion in critically ill patients. CT chest images acquired during intravenous contrast infusion have been used to estimate regional perfusion with a subsequent mathematical estimation of the match between ventilation (aeration) and perfusion. Dakin and colleagues used this approach and found that the amount of perfusion to consolidated lung areas (a surrogate for functional shunt) negatively correlated with PaO2/FiO2 [42] . Few other such studies, however, have been applied in critical care. Assessing and understanding functional perfusion abnormalities in relation to heterogeneity of lung aeration, and in response to ventilatory changes, represents a key unmet challenge towards better understanding ARDS pathogenesis. Although bronchoalveolar lavage fluid (BALF) is most proximal to the site of injury and likely the most relevant sample to study, the requirement of a bronchoscopy and inconsistencies in sample dilution have meant that BALF analysis is not routinely performed clinically and remains J o u r n a l P r e -p r o o f poorly studied in human subjects. Recent reviews have covered the role of biomarkers in BALF in the understanding the pathogenesis of ARDS including animal studies [13, 43] and a metaanalysis in human subjects [44] . In this section of the review, we will focus primarily on plasma biomarkers. A biological marker for ARDS is sorely lacking, however, finding such a biomarker is highly challenging. We know endothelial and epithelial cell injury are integral in ARDS pathogenesis and several biomarkers exist that are informative of injury to these cells. However, the extent to which each of these cells are injured is variable and dependent on the severity and mechanism of injury. Calfee and colleagues observed that circulating biomarkers of epithelial injury, such as surfactant protein-D (SP-D) and soluble Receptor for Advanced Glycation Endproducts (sRAGE), were higher in direct injury (e.g. pneumonia or aspiration) [45] . Whereas, angiopoietin-2 (ang-2), a marker of endothelial injury, was higher in indirect injury (e.g. sepsis). sRAGE levels in the plasma has been studied extensively in ARDS and elevated levels are associated with disease severity, adverse clinical outcomes and diffuse changes on CT scans of the lungs [46] [47] [48] [49] . While sRAGE is promising, its specificity to ARDS remains uncertain and has been implicated as a marker of severity in community acquired pneumonia [50] and in sepsis [51, 52] . Markers of endothelial injury are also elevated in ARDS and specifically sepsis-associated ARDS [53] . Elevated level of Ang-2 are known to be associated it with increased risk of developing ARDS [54] and associated with worse clinical outcomes [55] . Similarly, elevated levels of plasma von Willebrand factor, another marker of endothelial activation, was associated with worse outcomes in ARDS [56] . Biomarkers of coagulopathy/fibrinolysis and the extracellular matrix (ECM) are other components of the alveolar unit that have been studied in ARDS. Taking coagulopathy first, plasminogen activator inhibitor-1 and Protein C have both been associated with adverse clinical outcomes [57] . In the pediatric population, plasma matrix metalloproteinases (MMP)-8 and -9, markers of ECM injury, have been associated with prolonged ventilation in ARDS [58] and used to identify clusters with divergent clinical outcomes [59] . Inflammasome activity, as measure by interleukin-18 levels, is also known to be associated with adverse outcomes in ARDS [60, 61] . Despite several biomarkers of endothelial and epithelial injury known to be elevated in ARDS, these findings have yet to translate to meaningful therapies. In part, this is because the linkage of elevated biomarkers to function remains unestablished. The described pragmatic human translational studies are not the experimental domain to address mechanistic roles for these molecules and highlights a major limitation of such appraoches. Given ARDS is an acute inflammatory condition, it is unsurprising that pro-and antiinflammatory cytokines have been extensively studied in ARDS [62] . IL-1β, IL-1, IL-6, IL-8, IL-10 and soluble tumor necrosis factor (sTNFR)-1 have all been associated with clinical outcomes in ARDS. However, it is important to note that none of these biomarkers are specific to ARDS and are known to be elevated in other inflammatory conditions. Further, it is unclear whether elevated levels of these biomarkers are contributing to pathogenesis of ARDS or merely reflecting an increased burden of systemic inflammation. To maximize the informative potential of protein biomarkers, increasingly, investigators are using a combination of biomarkers to identify subgroups or clusters within ARDS populations using unbiased approaches. This genre of research, known as phenotyping, has become prominent in ARDS [63] and critical care research [64] . Work from our group has used a combination of protein biomarkers, vital signs, ventilatory variables, laboratory variables and demographics to identify unmeasured clusters using latent class analysis (LCA). LCA an unbiased probabilistic modeling algorithm that seeks to identify uniform subgroups in multivariate distributions [65] . Consistently, in independent secondary analyses of five randomized control trials (RCTs), we have identified two phenotypes of ARDS called the Hypo-and Hyperinflammatory phenotypes [66] [67] [68] [69] . The Hyperinflammatory phenotype is associated with higher levels of pro-inflammatory cytokines including IL-6, IL-8, sTNFR-1, and intracellular adhesion molecule-1. Additionally, the Hyperinflammatory phenotype is also associated with increased incidence of shock, lower Protein C levels and elevated markers of end-organ dysfunction including creatinine and bilirubin (Figure 2) . From a pathophysiological stand-point, J o u r n a l P r e -p r o o f proportions of patients with non-pulmonary sepsis were significantly higher in Hyperinflammatory phenotype, whereas, pulmonary infections were significantly higher in the Hypoinflammatory phenotype. Markers of endothelial activation (ang-2, vWF) were higher in the hyperinflammatory phenotype [67] , whereas the epithelial marker SP-D was lower [66] . Interestingly, sRAGE another epithelial marker was higher in the hyperinflammatory phenotype. Expectedly, mortality and ventilator days were significantly higher in the Hyperinflammatory phenotype in all analyses ( Table 3) . Further, divergent outcomes were observed in the phenotypes to randomized interventions in three of these trials to PEEP strategy [66] , fluid management strategy [67] and statin therapy [68] . The complexity of the LCA models, however, are a barrier to the identification of these phenotypes prospectively. To circumnavigate this, we developed models that either use a parsimonious set of biomarkers [70] or readily-available clinical data only [71] . Both approaches were able to classify phenotypes accurately and the divergent treatment responses were also observable using these clinically practical models. The models require prospective validation before they can be used in the clinical setting. Bos and colleagues used a similar panel of biomarkers (IL-1β, IL-6, IL-8, TNF-α, IL-10, IL-13, interferon gamma etc) to identify clusters in ARDS. Their approach differed to the abovementioned phenotyping scheme, in that, they restricted their predictor variables to protein biomarker and used hierarchical clustering. They too observed two clusters of ARDS which they termed "inflamed" and "uninflamed" [72] . As with the Hyperinflammatory phenotype, the "Inflamed" phenotype was associate with elevated pro-inflammatory cytokines and worse J o u r n a l P r e -p r o o f clinical outcomes. It remains unclear how much overlap there is between these different approaches of identifying clusters in ARDS. However, from a multitude of these studies, it seems apparent that concealed within ARDS are two biologically distinct subgroups that are primarily defined by their circulating inflammatory responses. Among the remaining questions, four require urgent consideration: 1) Are these phenotypes temporally stable and over what period of time? 2) Are these phenotypes specific to ARDS or generalizable to other inflammatory conditions? 3) Among the various approaches, which scheme should be used to uniformly identify the hyperinflammatory state in the clinical setting or clinical trial? 4) What are the optimal candidate interventions that could be evaluated in phenotype-specific trial? From a biological standpoint, it is unclear if the observed responses that define the phenotypes are always deleterious or whether they are part of a well conserved inflammatory response. Addressing these issues will mostly likely lead to successful therapies in ARDS. Several genomic approaches have been used to assess predisposition to ARDS. Genome-wide association studies may facilitate our understanding of ARDS pathogenesis by identifying genes that increase the likelihood of ARDS development. However, because ARDS is a complication from an underlying condition, these analyses require correction for the likelihood of common risk factors such as sepsis or pneumonia. An additional concern with these studies is the lack of J o u r n a l P r e -p r o o f reproducibility in independent datasets [73] . An important next step towards understanding the genuine implication of a genetic variant in the pathogenesis of ARDS are Mendelian randomization studies [74] . ANGPT2, the gene that encodes for ANG2 expression, was found to be related to the development of ARDS in a sepsis population through [75] . Importantly, this risk was mediated via an increase in plasma ANG2 concentration, suggesting a possible causal pathway. A similar approach was taken for sRAGE. Plasma sRAGE was strongly related to genetic variation and to the occurrence of ARDS in a sepsis cohort, suggesting that it acts as a causal intermediate in ARDS development [76] . Now that this approach has been taken for a marker of endothelial and epithelial injury, likely additional markers implicated as central in the pathogenesis of ARDS need evaluating through such studies. Transcriptomics analysis is the comprehensive assessment of messenger RNA from blood or tissue and provides insight in the complex interaction between genomics (providing the genetic potential) and exposures (resulting in the protein transcription of those genes). Therefore, it could potentially provide more information on the "state" a patient is in than genomic analysis alone. Gene expression of blood leukocytes has frequently been used to quantify the host response in critical illness in general and ARDS more specifically. In an analysis of multiple observational ARDS cohorts, Sweeney and colleagues found that 30 genes were associated with ARDS, however, after adjusting for severity of systemic inflammation, none of these genes were J o u r n a l P r e -p r o o f significant, leading the authors to conclude that plasma transcriptomics are unlikely to a meaningful tool in ARDS [77] . It is worth noting that this was a retrospective analysis of data extracted from public repositories and the ARDS diagnosis were uncertain and the populations included adults and children. Two approaches may provide further insight into the gene expression in ARDS: 1) focus more on gene expression in the organ of interest, the lung; 2) account for the biological heterogeneity observed in ARDS when analyzing blood gene expression profiles. A limited number of studies have focused on gene expression in pulmonary samples from ARDS. In a hallmark study, Morrell and colleagues simultaneously evaluated expression profiles of alveolar macrophages to those of peripheral blood monocytes (PBMs) in ARDS [78] . They observed that gene expression was profoundly different between these compartments and that enrichment of immune-inflammatory gene sets was associated with a favourable outcome in alveolar macrophages but an unfavourable outcome in PBMs, demonstrating distinct implication of inflammatory-responses that may be compartment-and cell-specific. The second approach was taken by Bos and colleagues in a post-hoc analysis of blood leukocyte expression obtained from patients with suspected sepsis and ARDS [79] . Patients were classified into two subphenotypes based on plasma biomarkers of inflammation, coagulation and endothelial injury as discussed in the section on Biomarkers in ARDS. Subsequent, expression profiles were compared between subphenotypes rather than with and without ARDS. The investigators reported that around 30% of genes were differentially expressed J o u r n a l P r e -p r o o f between the subphenotypes, with an enrichment of neutrophil related genes in the reactive (inflamed) subphenotype. Furthermore, the genes that were most up-regulated in the reactive subphenotype also discriminated between ARDS and a control group with sepsis but without ARDS. These data suggest that patients with the hyper-inflammatory subphenotype have a distinct gene expression profile related to neutrophil activation, oxidative phosphorylation and cholesterol metabolism. The advent of mass spectrometry (MS) and nuclear magnetic resonance (NMR) to study highthroughput metabolites has seen the emergence of the field metabolomics over the last two decades. Despite its growing use to study human biology, its application in ARDS remains in its infancy [80] . In a small pilot study, Stinger and colleagues studied plasma metabolites using NMR in sepsis-induced ARDS versus healthy volunteers [81] and metabolites pertaining to oxidant stress, energy homeostasis, apoptosis, and endothelial barrier function distinguished ARDS from healthy volunteers. Other investigators have subsequently studied differences between plasma metabolites in ARDS versus ventilated controls [82] , or healthy controls [83] with similar findings. MS has been used to study metabolites in oedema fluid or BALF in patients with ARDS. In comparison to controls, metabolites of oxidative stress (glutamate and proline) were elevated in ARDS [84, 85] . Differences in ARDS from controls were consistently observed, however, most of these described studies are limited by samples size (< 30). In the largest study of its kind, Viswan and colleagues took a different approach to see whether they can differentiate severity and aetiological sites (pulmonary vs extra-pulmonary) of ARDS using metabolic profile [86] . Both serum (n = 176) and BALF (n = 146) metabolic profiles showed good performance metric at differentiating ARDS severity, however, the ability to discriminate site of injury was poor. These findings were correspondent with an earlier work of Bos and colleagues who studied the discriminatory properties of exhaled breath metabolites in ARDS [87] . Lipidomics is another growing field in human biology and uses MS/NMR to study highthroughput quantification of lipids in biological compartments. Fatty acid-derived lipid mediators are critical in the regulation of the inflammatory response. Specifically, the role of lipid pro-resolving mediators in the resolution and homeostatic normalization of inflammation is being increasingly recognized [88, 89] . Lipidomics in critical illness is largely unexplored and perhaps represents a new frontier in our understanding of systems biology, particularly, given the ubiquity of these molecules both intra-and extracellularly. Future studies in human subject profiling the lipidome in ARDS and their functional role are eagerly anticipated. Until about a decade ago, the lung was considered sterile under normal conditions. Since, culture independent techniques for the detection and identification of microbiota has provided expansive insights into the lung microbiome in health and disease. The lung microbiome is J o u r n a l P r e -p r o o f shaped by three factors: (1) immigration of micro-organism into the lung, (2) elimination of micro-organisms through microbial killing and immigration via cough and mucocilliairy clearance, and (3) loco-regional growth circumstances that act as selective pressures on certain types of micro-organisms [90] . During intubation and invasive mechanical ventilation, these forces are disturbed significantly [91] . Therefore, it is unsurprising that duration of mechanical ventilation is one of the most important factors driving the change in lung microbiome in critically ill patients [92] . As lung injury occurs, additional nutrients become heterogeneously available in the lung and may further perpetuate regional growth differences and impose selective pressures towards specific micro-organisms [93, 94] . Simultaneously, specific bacteria seem to be enriched in the lung microbiome at the moment ARDS is diagnosed and invasive mechanical ventilation is initiated [95, 96] . Further, in these two independent studies, performed on two different continents, the same enrichment of gut bacteria was found to be related to ARDS and predicted unfavourable outcome [95, 96] . Taken together these findings suggest that: 1) changes in microbial composition in the lung may precede lung injury and plays a role in ARDS pathogenesis; 2) findings of microbiome disruption are agnostic to inter-individual and regional heterogeneity in microbial composition and antimicrobial practices. Future studies need to further clarify the causal relation between microbial dysbiosis and lung injury. Given the biosafety constraints of studying SARS-COV-2 in experimental models, findings at autopsy have been singularly informative in appreciating the pathophysiology of lung injury in COVID-19. Consistent among almost all studies that report autopsy finding of the lungs is diffuse alveolar damage [97] . Borczuk and colleagues in a multicenter study, reported autopsy findings in 68 deceased patients and observed DAD in most patients with virus detectable in alveolar type II cells and airway epithelia [98] . It has been postulated that as SARS-CoV-2 spike protein binds to angiotensin converting enzyme (ACE)-2 receptor to gain cellular entry and there is an abundance of these receptors on endothelial cells, COVID-19 is associated with J o u r n a l P r e -p r o o f increased endothelial activation and thromboembolic phenomena. To that that end, these investigators also observed thrombi present in large vessel of the lungs in the 42% of the cases. Elsewhere, in a case series of 80 patients, Edler and colleagues observed large vessel thrombi in the lungs in only 21% of the patients, however, they if they included deep vein thrombosis, the cumulative large vessel thrombi were 40% [99] . In a small study comparing COVID-19 to influenza at autopsy, the findings of thromboembolic phenomena were almost double in the former [100] and these findings were corroborated when rates were compared in COVID-19 compared to historical influenza data [101] . It is unclear whether the thromboembolic phenomena are due to direct invasion of the endothelial cells or a hypercoagulable state or both. The presence of virus in endothelial cells [102] and in organs outside of the respiratory tract [103] would suggest that direct viral pathogenicity is a plausible theory. Early reports on CT-images from patients with COVID-19 ARDS speculated that it was characterized by normal lung volumes with severe hypoxemia [104] . Subsequent studies were unable to show that lung volumes are preserved and showed no relation between compliance of the respiratory system and the extent of parenchymal involvement [105, 106] . In line with histopathological findings, perfusion defects have been consistently detected on lung imaging of COVID-19 ARDS (Figure 3) irrespective of the presence of pulmonary embolism and are likely reflective of the microthrombi [107] [108] [109] . A frequently described theory in COVID-19 is that a cytokine storm is the key driver of disease severity [110, 111] . Closer scrutiny of the described levels of pro-inflammatory cytokines such as IL-6 in COVID-19 would suggest that, while elevated above normal, they were much lower than those described in historical cohorts of ARDS [112] . In a systematic review, when proinflammatory cytokine levels in COVID-19 were compared with the Hyperinflammatory phenotype of ARDS, sepsis or cytokine release syndrome (post CAR-T therapy), the levels were significantly lower [113] . Interestingly, D-Dimer levels in severe COVID-19 were higher compared to historical critical care cohorts, suggesting a common theme of a hypercoagulable state. Other investigators have similarly observed attenuated IL-6 and IL-8 levels COVID-19 compared to non-COVID-19 ARDS [114, 115] . A prospective exploratory analysis of COVID-19 ARDS suggested that the prevalence of the Hyperinflammatory phenotype was between 11-21% compared to 30% observed in non-COVID-19 ARDS [116] . Together, these findings suggest that circulating inflammatory biomarkers may not be critical or unique in the pathophysiology of COVID-19 ARDS. Yet, mortality in COVID-19 ARDS is considerably higher. Rather than ruminating on the cytokine storm, a more intriguing question to ask is what biological phenomenon is driving injury in the lung and the observed excess mortality? As a unifying biological hypothesis, we speculate that COVID-19 ARDS is associated with a more immunosuppressive state, either systemically or locally in the lungs, which in turn leads to impaired viral clearance and the ensuing epithelial injury and hypercoagulable state that are consistently observed. The absence/attenuation of interferon J o u r n a l P r e -p r o o f responses have been observed in COVID-19 and are shown to be associated with adverse outcomes [117] . This hypothesis needs testing in both the lung and circulating compartments. In the context of the current understanding of ARDS with its clinical diagnosis, conceptually, it is perhaps easiest to comprehend its pathophysiology as a graded permutation of injuries to the four components of the alveolar unit: the epithelium, endothelium, extra-cellular matrix and coagulopathy of the microvasculature. The prototype injury of each anatomical domain is represented in Figure 4 . Clearly, these are not mutually exclusive injury types, however, the extent to which a domain principally drives the global lung injury is likely to be dependent on the original insult. Among these staggering numbers of patients with COVID-19, it is worth acknowledging that all cases are due to a single pathogen. This is noteworthy because critical care is accustomed to managing nebulous clinical syndromes with multiple aetiologies. Yet, from a biological stand-point, understanding a unifying pathophysiology in COVID-19 has been exceedingly challenging. It is then worth asking, what are the probabilities of making such a biological discovery in ARDS if we persist with its clinical diagnosis? Regardless, the steps needed to better understand the biology of the disease and the clinical syndrome are the same. Studies are needed where biological measurements are simultaneously made in the lungs and the circulation and over multiple timepoints using multidimensional data types. The heterogeneity subsumed within these diagnoses need to be broken down into biologically intuitive subgroups that are empirically derived. Finally, a central J o u r n a l P r e -p r o o f challenge facing the specialty is to harness these high-throughput data and translate it into a therapy that benefits patients with lung injury testing hypotheses in more experimental models. J o u r n a l P r e -p r o o f The American-European Consensus Conference on ARDS Definitions, mechanisms, relevant outcomes, and clinical trial coordination Acute respiratory distress syndrome: the Berlin Definition Phenotypes and personalized medicine in the acute respiratory distress syndrome Diffuse alveolar damage--the role of oxygen, shock, and related factors. A review Alterations of the gas exchange apparatus in adult respiratory insufficiency associated with septicemia Chronology of histological lesions in acute respiratory distress syndrome with diffuse alveolar damage: a prospective cohort study of clinical autopsies Comparison of clinical criteria for the acute respiratory distress syndrome with autopsy findings Accuracy of clinical diagnosis of acute respiratory distress syndrome in comparison with autopsy findings ARDS: a clinicopathological confrontation Discrepancy between clinical criteria for diagnosing acute respiratory distress syndrome secondary to community acquired pneumonia with autopsy findings of diffuse alveolar damage Comparison of the Berlin definition for acute respiratory distress syndrome with autopsy The role of open-lung biopsy in ARDS Open lung biopsy in nonresolving ARDS frequently identifies diffuse alveolar damage regardless of the severity stage and may have implications for patient management Histopathologic heterogeneity of acute respiratory distress syndrome revealed by surgical lung biopsy and its clinical implications The Presence of Diffuse Alveolar Damage on Open Lung Biopsy Is Associated With Mortality in Patients With Acute Respiratory Distress Syndrome: A Systematic Review and Meta-Analysis An official American Thoracic Society workshop report: features and measurements of experimental acute lung injury in animals Two distinct immunopathological profiles in autopsy lungs of COVID-19 Nondestructive cryomicro-CT imaging enables structural and molecular analysis of human lung tissue A Pathophysiologic Approach to Biomarkers in Acute Respiratory Distress Syndrome Lung fluid biomarkers for acute respiratory distress syndrome: a systematic review and meta-analysis Distinct molecular phenotypes of direct vs indirect ARDS in singlecenter and multicenter studies Plasma receptor for advanced glycation end products and clinical outcomes in acute lung injury Soluble Forms and Ligands of the Receptor for Advanced Glycation End-Products in Patients with Acute Respiratory Distress Syndrome: An Observational Prospective Study Soluble form of the receptor for advanced glycation end products is a marker of acute lung injury but not of severe sepsis in critically ill patients Elevated Plasma Levels of sRAGE Are Associated With Nonfocal CT-Based Lung Imaging in Patients With ARDS: A Prospective Multicenter Study Soluble RAGE as a severity marker in community acquired pneumonia associated sepsis Soluble receptor for advanced glycation end products predicts 28-day mortality in critically ill patients with sepsis Endothelial biomarkers in human sepsis: pathogenesis and prognosis for ARDS Plasma angiopoietin-2 predicts the onset of acute lung injury in critically ill patients Circulating angiopoietin-2 and the risk of mortality in patients with acute respiratory distress syndrome: a systematic review and meta-analysis of 10 prospective cohort studies Prognostic and pathogenetic value of combining clinical and biochemical indices in patients with acute lung injury Early elevation of matrix metalloproteinase-8 and -9 in pediatric ARDS is associated with an increased risk of prolonged mechanical ventilation Early Plasma Matrix Metalloproteinase Profiles. A Novel Pathway in Pediatric Acute Respiratory Distress Syndrome Inflammasome-regulated cytokines are critical mediators of acute lung injury Association of Elevated Plasma Interleukin-18 Level With Increased Mortality in a Clinical Trial of Statin Treatment for Acute Respiratory Distress Syndrome Biomarkers in acute lung injury: insights into the pathogenesis of acute lung injury Phenotypes in acute respiratory distress syndrome: moving towards precision medicine Subphenotypes in critical care: translation into clinical practice Practitioner's Guide to Latent Class Analysis: Methodological Considerations and Common Pitfalls Subphenotypes in acute respiratory distress syndrome: latent class analysis of data from two randomised controlled trials Acute Respiratory Distress Syndrome Subphenotypes Respond Differently to Randomized Fluid Management Strategy Acute respiratory distress syndrome subphenotypes and differential response to simvastatin: secondary analysis of a randomised controlled trial Latent class analysis of ARDS subphenotypes: a secondary analysis of the statins for acutely injured lungs from sepsis (SAILS) study Development and validation of parsimonious algorithms to classify acute respiratory distress syndrome phenotypes: a secondary analysis of randomised controlled trials Machine Learning Classifier Models Can Identify Acute Respiratory Distress Syndrome Phenotypes Using Readily Available Clinical Data Identification and validation of distinct biological phenotypes in patients with acute respiratory distress syndrome by cluster analysis Applying metabolomics to uncover novel biology in ARDS Metabolic consequences of sepsis-induced acute lung injury revealed by plasma (1)H-nuclear magnetic resonance quantitative metabolomics and computational analysis Metabolic profiling of human lung injury by H-1 high-resolution nuclear magnetic resonance spectroscopy of blood serum Explore potential plasma biomarkers of acute respiratory distress syndrome (ARDS) using GC-MS metabolomics analysis Untargeted LC-MS metabolomics of bronchoalveolar lavage fluid differentiates acute respiratory distress syndrome from health Profiling of ARDS pulmonary edema fluid identifies a metabolically distinct subset Distinct Metabolic Endotype Mirroring Acute Respiratory Distress Syndrome (ARDS) Subphenotype and its Heterogeneous Biology Exhaled breath metabolomics as a noninvasive diagnostic tool for acute respiratory distress syndrome Identification and signature profiles for pro-resolving and inflammatory lipid mediators in human tissue Pro-resolving lipid mediators are leads for resolution physiology Autopsy findings in COVID-19-related deaths: a literature review COVID-19 pulmonary pathology: a multi-institutional autopsy cohort from Italy and New York City Dying with SARS-CoV-2 infection-an autopsy study of the first consecutive Pulmonary Vascular Endothelialitis, Thrombosis, and Angiogenesis in Covid-19 Endothelial cell infection and endotheliitis in COVID-19 Histopathological findings and viral tropism in UK patients with severe fatal COVID-19: a post-mortem study The trinity of COVID-19: immunity, inflammation and intervention Is a "Cytokine Storm" Relevant to COVID-19? Cytokine elevation in severe and critical COVID-19: a rapid systematic review, meta-analysis, and comparison with other inflammatory syndromes Cytokine Levels in Critically Ill Patients With COVID-19 and Other Conditions Impaired type I interferon activity and inflammatory responses in severe COVID-19 patients Patterns of Care, and Mortality for Patients With Acute Respiratory Distress Syndrome in Intensive Care Units in 50 Countries Acute respiratory distress in adults Acute respiratory distress syndrome. Nat Rev Dis Primers Acute Respiratory Distress Syndrome Clinical trials in acute respiratory distress syndrome: challenges and opportunities Optimising experimental research in respiratory diseases: an ERS statement Should we shift the paradigm of preclinical models for ARDS therapies? Thorax Animal models of acute lung injury Why translational research matters: proceedings of the third international symposium on acute lung injury translational research (INSPIRES III) Novel translational approaches to the search for precision therapies for acute respiratory distress syndrome Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome Acute respiratory distress syndrome: the Berlin Definition Phenotypes and personalized medicine in the acute respiratory distress syndrome Diffuse alveolar damage--the role of oxygen, shock, and related factors. A review Alterations of the gas exchange apparatus in adult respiratory insufficiency associated with septicemia Chronology of histological lesions in acute respiratory distress syndrome with diffuse alveolar damage: a prospective cohort study of clinical autopsies Comparison of clinical criteria for the acute respiratory distress syndrome with autopsy findings Accuracy of clinical diagnosis of acute respiratory distress syndrome in comparison with autopsy findings ARDS: a clinicopathological confrontation Discrepancy between clinical criteria for diagnosing acute respiratory distress syndrome secondary to community acquired pneumonia with autopsy findings of diffuse alveolar damage Comparison of the Berlin definition for acute respiratory distress syndrome with autopsy The role of open-lung biopsy in ARDS Open lung biopsy in nonresolving ARDS frequently identifies diffuse alveolar damage regardless of the severity stage and may have implications for patient management Histopathologic heterogeneity of acute respiratory distress syndrome revealed by surgical lung biopsy and its clinical implications The Presence of Diffuse Alveolar Damage on Open Lung Biopsy Is Associated With Mortality in Patients With Acute Respiratory Distress Syndrome: A Systematic Review and Meta-Analysis An official American Thoracic Society workshop report: features and measurements of experimental acute lung injury in animals Two distinct immunopathological profiles in autopsy lungs of COVID-19 Nondestructive cryomicro-CT imaging enables structural and molecular analysis of human lung tissue An expanded definition of the adult respiratory distress syndrome Efficacy and economic assessment of conventional ventilatory support versus extracorporeal membrane oxygenation for severe adult respiratory failure (CESAR): a multicentre randomised controlled trial Severity scoring of lung oedema on the chest radiograph is associated with clinical outcomes in ARDS Early Changes Over Time in the Radiographic Assessment of Lung Edema Score Are Associated With Survival in ARDS What has computed tomography taught us about the acute respiratory distress syndrome? Lung recruitment in patients with the acute respiratory distress syndrome ARDS and the search for meaningful subgroups Regional distribution of gas and tissue in acute respiratory distress syndrome. I. Consequences for lung morphology. CT Scan ARDS Study Group Lung morphology predicts response to recruitment maneuver in patients with acute respiratory distress syndrome Personalised mechanical ventilation tailored to lung morphology versus low positive end-expiratory pressure for patients with acute respiratory distress syndrome in France (the LIVE study): a multicentre, single-blind, randomised controlled trial Clinical review: the role of ultrasound in estimating extravascular lung water Bedside ultrasound assessment of positive end-expiratory pressureinduced lung recruitment Anatomical and functional intrapulmonary shunt in acute respiratory distress syndrome Changes in lung composition and regional perfusion and tissue distribution in patients with ARDS A Pathophysiologic Approach to Biomarkers in Acute Respiratory Distress Syndrome Lung fluid biomarkers for acute respiratory distress syndrome: a systematic review and meta-analysis Distinct molecular phenotypes of direct vs indirect ARDS in singlecenter and multicenter studies Plasma receptor for advanced glycation end products and clinical outcomes in acute lung injury Soluble Forms and Ligands of the Receptor for Advanced Glycation End-Products in Patients with Acute Respiratory Distress Syndrome: An Observational Prospective Study Soluble form of the receptor for advanced glycation end products is a marker of acute lung injury but not of severe sepsis in critically ill patients Elevated Plasma Levels of sRAGE Are Associated With Nonfocal CT-Based Lung Imaging in Patients With ARDS: A Prospective Multicenter Study Soluble RAGE as a severity marker in community acquired pneumonia associated sepsis Soluble receptor for advanced glycation end products predicts 28-day mortality in critically ill patients with sepsis sRAGE is elevated in septic patients and associated with patients outcome Endothelial biomarkers in human sepsis: pathogenesis and prognosis for ARDS Plasma angiopoietin-2 predicts the onset of acute lung injury in critically ill patients Circulating angiopoietin-2 and the risk of mortality in patients with acute respiratory distress syndrome: a systematic review and meta-analysis of 10 prospective cohort studies Significance of von Willebrand factor in septic and nonseptic patients with acute lung injury Prognostic and pathogenetic value of combining clinical and biochemical indices in patients with acute lung injury Early elevation of matrix metalloproteinase-8 and -9 in pediatric ARDS is associated with an increased risk of prolonged mechanical ventilation Early Plasma Matrix Metalloproteinase Profiles. A Novel Pathway in Pediatric Acute Respiratory Distress Syndrome Inflammasome-regulated cytokines are critical mediators of acute lung injury Association of Elevated Plasma Interleukin-18 Level With Increased Mortality in a Clinical Trial of Statin Treatment for Acute Respiratory Distress Syndrome Biomarkers in acute lung injury: insights into the pathogenesis of acute lung injury Phenotypes in acute respiratory distress syndrome: moving towards precision medicine Subphenotypes in critical care: translation into clinical practice Practitioner's Guide to Latent Class Analysis: Methodological Considerations and Common Pitfalls Subphenotypes in acute respiratory distress syndrome: latent class analysis of data from two randomised controlled trials Acute Respiratory Distress Syndrome Subphenotypes Respond Differently to Randomized Fluid Management Strategy Acute respiratory distress syndrome subphenotypes and differential response to simvastatin: secondary analysis of a randomised controlled trial Latent class analysis of ARDS subphenotypes: a secondary analysis of the statins for acutely injured lungs from sepsis (SAILS) study Development and validation of parsimonious algorithms to classify acute respiratory distress syndrome phenotypes: a secondary analysis of randomised controlled trials Machine Learning Classifier Models Can Identify Acute Respiratory Distress Syndrome Phenotypes Using Readily Available Clinical Data Identification and validation of distinct biological phenotypes in patients with acute respiratory distress syndrome by cluster analysis Replicating genotype-phenotype associations Reading Mendelian randomisation studies: a guide, glossary, and checklist for clinicians Plasma angiopoietin-2 as a potential causal marker in sepsis-associated ARDS development: evidence from Mendelian randomization and mediation analysis Plasma sRAGE Acts as a Genetically Regulated Causal Intermediate in Sepsis-associated Acute Respiratory Distress Syndrome Multicohort Analysis of Whole-Blood Gene Expression Data Does Not Form a Robust Diagnostic for Acute Respiratory Distress Syndrome Peripheral and Alveolar Cell Transcriptional Programs Are Distinct in Acute Respiratory Distress Syndrome Understanding Heterogeneity in Biologic Phenotypes of Acute Respiratory Distress Syndrome by Leukocyte Expression Profiles Applying metabolomics to uncover novel biology in ARDS Metabolic consequences of sepsis-induced acute lung injury revealed by plasma (1)H-nuclear magnetic resonance quantitative metabolomics and computational analysis Metabolic profiling of human lung injury by H-1 high-resolution nuclear magnetic resonance spectroscopy of blood serum Explore potential plasma biomarkers of acute respiratory distress syndrome (ARDS) using GC-MS metabolomics analysis Untargeted LC-MS metabolomics of bronchoalveolar lavage fluid differentiates acute respiratory distress syndrome from health Profiling of ARDS pulmonary edema fluid identifies a metabolically distinct subset Distinct Metabolic Endotype Mirroring Acute Respiratory Distress Syndrome (ARDS) Subphenotype and its Heterogeneous Biology Exhaled breath metabolomics as a noninvasive diagnostic tool for acute respiratory distress syndrome Pro-resolving lipid mediators are leads for resolution physiology Identification and signature profiles for pro-resolving and inflammatory lipid mediators in human tissue The Microbiome and the Respiratory Tract The importance of airway and lung microbiome in the critically ill The dynamics of the pulmonary microbiome during mechanical ventilation in the intensive care unit and the association with occurrence of pneumonia A tale of two sites: how inflammation can reshape the microbiomes of the gut and lungs The Lung Microbiome and ARDS. It Is Time to Broaden the Model Lung Microbiota Is Related to Smoking Status and to Development of Acute Respiratory Distress Syndrome in Critically Ill Trauma Patients Lung Microbiota Predict Clinical Outcomes in Critically Ill Patients Autopsy findings in COVID-19-related deaths: a literature review COVID-19 pulmonary pathology: a multi-institutional autopsy cohort from Italy and New York City Dying with SARS-CoV-2 infection-an autopsy study of the first consecutive Pulmonary Vascular Endothelialitis, Thrombosis, and Angiogenesis in Covid-19 Lung Histopathology in Coronavirus Disease 2019 as Compared With Severe Acute Respiratory Sydrome and H1N1 Influenza: A Systematic Review Endothelial cell infection and endotheliitis in COVID-19 Histopathological findings and viral tropism in UK patients with severe fatal COVID-19: a post-mortem study COVID-19 Does Not Lead to a "Typical" Acute Respiratory Distress Syndrome Subphenotyping Acute Respiratory Distress Syndrome in Patients with COVID-19: Consequences for Ventilator Management Pathophysiology of COVID-19-associated acute respiratory distress syndrome: a multicentre prospective observational study Extensive pulmonary perfusion defects compatible with microthrombosis and thromboembolic disease in severe Covid-19 pneumonia COVID-19 pneumonia: microvascular disease revealed on pulmonary dual-energy computed tomography angiography Pulmonary Angiopathy in Severe COVID-19: Physiologic, Imaging, and Hematologic Observations The trinity of COVID-19: immunity, inflammation and intervention Is a "Cytokine Storm" Relevant to COVID-19? Cytokine elevation in severe and critical COVID-19: a rapid systematic review, meta-analysis, and comparison with other inflammatory syndromes Cytokine Levels in Critically Ill Patients With COVID-19 and Other Conditions Distinct inflammatory profiles distinguish COVID-19 from influenza with limited contributions from cytokine storm Prevalence of phenotypes of acute respiratory distress syndrome in critically ill patients with COVID-19: a prospective observational study Impaired type I interferon activity and inflammatory responses in severe COVID-19 patients