Latent class analysis of ARDS subphenotypes: a secondary analysis of the statins for acutely injured lungs from sepsis (SAILS) study

Abstract

Purpose: Using latent class analysis (LCA), we have consistently identified two distinct subphenotypes in four randomized controlled trial cohorts of ARDS. One subphenotype has hyper-inflammatory characteristics and is associated with worse clinical outcomes. Further, within 3 negative clinical trials, we observed differential treatment response by subphenotype to randomly assigned interventions. The main purpose of this study was to identify ARDS subphenotypes in a contemporary NHLBI Network trial of infection-associated ARDS (SAILS) using LCA and to test for differential treatment response to rosuvastatin therapy in the subphenotypes. Methods: LCA models were constructed using a combination of biomarker and clinical data at baseline in the SAILS study (n=745). LCA modeling was then repeated using an expanded set of clinical class-defining variables. Subphenotypes were tested for differential treatment response to rosuvastatin. Results: The 2-class LCA model best fit the population. 40% of the patients were classified as the “hyper-inflammatory” subphenotype. Including additional clinical variables in the LCA models did not identify new classes. Mortality at Day 60 and Day 90 was higher in the hyper-inflammatory subphenotype. No differences in outcome were observed between hyper-inflammatory patients randomized to rosuvastatin therapy versus placebo. Conclusions: Using LCA, a 2-subphenotype model best described the SAILS population. The subphenotypes have features consistent with those previously reported in 4 other cohorts. Addition of new class-defining variables in the LCA model did not yield additional subphenotypes. No treatment effect was observed with rosuvastatin. These findings further validate the presence of two subphenotypes and demonstrates their utility for patient stratification in ARDS.

Acute respiratory distress syndrome subphenotypes and therapy responsive traits among preclinical models: protocol for a systematic review and meta-analysis

“…Our review is timely since there is an increasing amount of research dedicated to the identification of ARDS subphenotypes, their underlying mechanisms, and their potential responsive treats [4,7,[9][10][11][12][48][49][50][51][52]. The results from this study should therefore be useful in order to generate hypotheses on whether some therapies (see Table 2) might differentially benefit to some preclinical subphenotypes, as compared to others.…”

ARDS Subphenotypes: Understanding a Heterogeneous Syndrome

“…In retrospective analyses, the hypo-and hyperinflammatory phenotypes discussed above have been observed to have differential treatment responses to several different interventions, including PEEP and fluid management strategies (Table 3) [27,29]. Subphenotypic differences in response to simvastatin were also observed in reanalysis of one clinical trial (HARP-2 trial) [40], but not in a similar reanalysis of a separate trial of rosuvastatin for ARDS (SAILS trial) [41]. While it is possible this discrepancy reflects differences in trial design (and the particular statin that was tested), it also highlights the uncertainty that remains when differential treatment response has been observed only retrospectively.…”

“…The use of metabolomics, transcriptomics, and genomics for ARDS phenotyping and predictive enrichment is in even earlier stages than proteomic phenotyping. Bos et al used the "uninflamed" and "reactive" subphenotypes they had previously identified based on plasma protein biomarkers to test whether there were differences in blood leukocyte gene expression between [27,29,40,41]. IL interleukin, bicarb bicarbonate, TNFr1 tumor necrosis factor receptor 1 groups, and found that approximately one-third of genes were differentially expressed.…”

Precision Medicine in Critical Illness: Sepsis and Acute Respiratory Distress Syndrome

“…In ARDS, this has been consistently demonstrated among clinical trial populations using an analytic method known as latent class analysis (LCA) to uncover potentially unobserved subpopulations while remaining agnostic to outcomes. In analyses considering clinical variables including vital signs, ventilator data, and laboratory values in addition to exploratory plasma biomarkers representing inflammation, vascular dysfunction, or alveolar injury, a latent class model consistently identified two classes of ARDS trial subjects that differed in their plasma expression of inflammatory biomarkers epitomized by interleukin (IL-) 8 or IL-6, and by their degree of systemic illness, characterized by low blood pressure and low serum bicarbonate [21][22][23][24].…”

“…Not only were subjects in the "hyperinflammatory" subphenotype group more likely to die, but the LCA group assignment (hyperinflammatory versus non-hyperinflammatory) exhibited significant statistical interaction with randomized treatment effects (Table 18.1) [21,22,24]. Thus, when the randomized interventions of higher positive end-expiratory pressure (PEEP), conservative fluid strategy, or simvastatin therapy were analyzed in groups stratified by LCA assignment, each therapy seemed to have a mortality benefit only observed in the hyperinflammatory group, with no signal for improvement in the nonhyperinflamed group [21,22,24]. In each trial, there was no evidence for heterogeneity in treatment effect by baseline severity of illness, as defined by the Acute Physiology and Chronic Health Evaluation (APACHE) score, nor by the severity of ARDS, as defined by the ratio of arterial partial pressure of oxygen to fraction of inspired oxygen (PaO 2 :FiO 2 ).…”