Sepsis Endotypes Defined by Lymphocyte Thresholds and Inflammation Inform Precision Immunomodulation

Abstract

Immunomodulatory therapies demonstrate variable efficacy in sepsis, suggesting biological heterogeneity inadequately captured by current stratification approaches. Although lymphopenia predicts mortality, functional thresholds and their interaction with inflammation remain poorly characterized. We investigated whether integrating lymphocyte status with systemic inflammation defines sepsis endotypes with differential treatment responsiveness. We retrospectively profiled 714 patients within 24 h using lymphocyte subsets and inflammatory biomarkers. Restricted cubic spline analysis revealed nonlinear associations between lymphocyte counts and mortality (p < 0.01), with steep risk increases at lower counts. Risk optimization identified critical thresholds at 374 cells/µL (total T cells), 340 cells/µL (CD4⁺), and 157 cells/µL (CD8⁺). Principal component analysis of inflammatory markers combined with lymphocyte stratification classified patients into four discrete endotypes with markedly divergent 28-day survival (55%–58% vs. 82–87%, p < 0.001). Patients with immunosuppressed/hypo-inflammatory endotype had higher survival among those who received corticosteroids (CD4⁺-depleted: 84.4% vs. 75.6%, p < 0.001; T-cell-depleted: 78.7% vs. 72.3%, p = 0.006), whereas hyperinflammatory endotypes showed no such association. Integration of publicly available single-cell (GSE167363) and bulk transcriptomics (GSE65682) datasets yielded a 15-gene T-cell dysfunction signature with external validation (CNP0004962, area under the curve [AUC] 0.76–0.85). These observational findings suggest that immune-inflammatory co-profiling identifies biologically distinct sepsis subgroups with differential treatment associations, generating testable hypotheses for prospective validation through endotype-guided trials.