Abstract Sepsis is a frequently encountered critical care syndrome leading to 250, 000 deaths annually, making it one of the leading causes of mortality in the U.S. Despite over 75 randomized controlled trials (RCTs), no treatments have shown a survival benefit and currently there are no approved therapies for sepsis. The multitude of negative RCTs has been attributed to the clinical and biological heterogeneity subsumed within the non- specific clinical definition of sepsis. The failed therapies in the context of vast heterogeneity makes sepsis ideal for precision medicine-based approaches where targeted therapies are used in biologically-informed subgroups. Acute respiratory distress syndrome (ARDS) is another critical care syndrome with significant heterogeneity. In secondary analyses of five ARDS RCTs, we consistently identified two phenotypes, the hyperinflammatory and hypoinflammatory phenotypes, with divergent clinical outcomes, biomarker profiles and differential responses to therapies. Crucially, in an RCTs that only recruited sepsis-associated ARDS, the corresponding two phenotypes emerged again. Our preliminary data, in a separate cohort of 587 patients with sepsis, also identified these two phenotypes. In this proposal, we will test the hypothesis that the molecular phenotypes previously identified in ARDS are also evident in sepsis, and that they represent distinct biologic subtypes characterized by differences in circulating inflammatory responses. Using latent class analysis with a composite of clinical and biomarker data, we will seek phenotypes in four independent cohorts of sepsis (>4000 patients). We will develop clinically implementable models to classify the phenotypes using previously described algorithmic pipelines. Heterogeneous treatment effect in the phenotypes will be sought in CLOVERS, a multicenter RCT comparing fluid resuscitation strategies in severe sepsis. In a prospective cohort, we will study novel techniques such as next-generation sequencing, mass cytometry and functional immune responses in stimulated peripheral blood mononuclear cells to better understand the biological and immunological characteristics of the phenotypes and identify phenotype-specific treatable traits. We will also study the phenotypes longitudinally to evaluate their temporal stability. This proposal represents an independent niche of research for my group, which has the requisite experience and expertise to successfully deliver this program. The culmination of the program will potentially lead to several highly impactful discoveries with important implications for sepsis care. We anticipate identifying robust and reproducible phenotypes of sepsis in multiple cohorts, with phenotype-specific response to therapies. We will develop practical models that can identify phenotypes at the bedside. We will comprehensively map the biological and immunological profiles of the phenotypes to identify treatable traits that may enable precision-based...