Inflammatory Phenotypes In Severe Pneumonia: Clinical Evidence To Mouse Models For Precision Therapeutics

RATIONALE: Hyperinflammatory and hypoinflammatory phenotypes previously identified in sepsis and ARDS may enable precision therapies, but their clinical relevance and translational modeling in severe pneumonia remain incompletely characterized. OBJECTIVES: To examine biomarker-defined hyperinflammatory and hypoinflammatory phenotypes in critically ill patients with pneumonia (pulmonary sepsis), test whether the biomarkers that define these phenotypes identify subgroups and outcomes in a mouse model of bacterial pneumonia, and determine whether the mouse phenotypes respond differently to therapeutic interventions. METHODS: We performed latent class analysis (LCA) in a cohort of 548 ICU patients with pulmonary sepsis to identify inflammatory phenotypes and test the association of these phenotypes with clinical outcomes using validated classifier models. We developed a mouse model of pneumococcal pneumonia which recapitulates key aspects of these phenotypes and tested responses to dexamethasone and IL-6 receptor blockade. MEASUREMENTS AND MAIN RESULTS: Two phenotypes were identified in patients with pulmonary sepsis: a hyperinflammatory phenotype with more lung injury and increased mortality, and a hypoinflammatory phenotype with more favorable outcomes. Despite uniform pathogen exposure and baseline conditions, LCA identified two phenotypes in the mice with divergent trajectories (lung injury and mortality). Therapeutic benefit from anti-inflammatory interventions was observed exclusively in the more inflamed mice. CONCLUSIONS: By integrating clinical data from human studies and experimental findings in a clinically relevant mouse model, this study provides a translational framework for developing phenotype-targeted therapies for critical illness, with the goal of improving patient outcomes.