Quantifying immune dysregulation in pneumonia and sepsis with a parsimonious machine-learning model: a multicohort analysis across care settings and reanalysis of a hydrocortisone randomised controlled trial.

Summary

Using 35 biomarkers across CAP cohorts, the authors derived categorical (DIP1–3) and continuous (cDIP) immune dysregulation measures and distilled them to a 3-biomarker model (procalcitonin, soluble TREM-1, IL-6) with high accuracy. Increased dysregulation predicted mortality and secondary infections independent of clinical severity, and hydrocortisone conferred benefit only in severe dysregulation, with faster immune recovery.

Key Findings

• Unsupervised analysis organized 35 biomarkers into DIP1–3 stages and a continuous cDIP score capturing immune dysregulation.
• A parsimonious 3-biomarker model (procalcitonin, sTREM-1, IL-6) accurately reproduced DIP stage (91.2% accuracy) and cDIP (RMSE 0.056).
• Higher cDIP associated with increased mortality (OR 1.26 per 10% increase) and secondary infections, independent of clinical severity.
• Hydrocortisone reduced 30-day mortality only in severe dysregulation (DIP3; cDIP ≥0.63) and accelerated immune recovery.

Clinical Implications

Adopting a 3-analyte panel (procalcitonin, sTREM-1, IL-6) could stratify sepsis patients for steroids or other immunomodulators and monitor recovery, moving beyond severity-based enrollment.

Limitations

• Not a prospective interventional validation; treatment effects assessed post hoc
• Assay availability and turnaround may limit immediate bedside adoption
• Generalizability beyond CAP and included cohorts requires further testing

Future Directions

Prospective trials embedding DIP/cDIP-guided immunomodulation (e.g., steroids) with predefined thresholds, and integration into rapid point-of-care assays.