Related Experiment Video
Updated: Aug 6, 2026
07:42
A Data-Driven Approach to Quantifying Immune States in Sepsis
Published on: February 7, 2025
Multi-cohort consensus clustering identifies three distinct transcriptomic endotypes in sepsis
Naixun Chi1, Siyu Mu1, Xin Jin1
1Hospital-acquired Infection Control Department, The Second Affiliated Hospital of Harbin Medical University, Harbin, China.
BMC Bioinformatics
|July 17, 2026
Summary
Sepsis endotyping reveals three transcriptomic patterns, but the C2 mortality advantage disappears when adjusting for baseline severity. Future studies must account for severity to accurately stratify sepsis patients.
Area of Science:
- Genomics
- Immunology
- Critical Care Medicine
Background:
- Sepsis is a complex syndrome with variable host immune responses.
- Existing transcriptomic endotyping methods lack comparability due to differences in gene selection and cohort composition.
- Distinguishing intrinsic prognostic signals from baseline severity in sepsis endotypes remains a challenge.
Purpose of the Study:
- To develop a standardized, reproducible transcriptomic endotyping framework for sepsis.
- To identify distinct sepsis endotypes based on gene expression patterns.
- To assess the prognostic value of identified endotypes, accounting for baseline severity.
Main Methods:
- Consensus clustering of 1002 sepsis patients from 5 discovery cohorts using 122 selected genes.
- Quantile normalization, ComBat batch correction, and Pearson correlation distance were applied.
- Endotype-mortality associations were evaluated using logistic regression and SOFA-adjusted models; generalizability was tested with a LASSO classifier.
Main Results:
- Three sepsis endotypes were identified: C1 (Immune Activation), C2 (Interferon Response), and C3 (Erythroid Dysregulation).
- Endotype C2 showed initially lower mortality, but this advantage was fully attenuated when adjusted for Sequential Organ Failure Assessment (SOFA) scores.
- A LASSO classifier demonstrated high sensitivity and generalizability across external cohorts, with endotype stability observed longitudinally.
Conclusions:
- Three recurring transcriptomic patterns in sepsis (immune activation, interferon response, erythroid/metabolic) were identified across cohorts.
- The prognostic significance of endotypes, particularly the C2 mortality advantage, appears largely explained by baseline severity rather than independent biology.
- Future sepsis endotyping research should prioritize systematic severity adjustment for accurate prognostic and treatment-stratification conclusions.