Identification and validation of robust hospital-acquired pneumonia subphenotypes associated with all-cause mortality: a multi-cohort derivation and validation.

Summary

Across four derivation cohorts and validation in an RCT dataset, two HAP subphenotypes were identified. Subphenotype 2 exhibited worse physiology, higher inflammation, microbiome dysbiosis, and higher 28-day mortality and treatment failure; antibiotic effect modification was observed for tedizolid in the RCT dataset.

Key Findings

• Two HAP subphenotypes were consistently identified across four cohorts using unsupervised clustering.
• Subphenotype 2 had lower temperature, worse PaO2/FiO2, higher inflammation, microbiome dysbiosis, and higher 28-day mortality and test-of-cure failure (p<0.01).
• In the VITAL RCT dataset, tedizolid treatment effect was modified by subphenotype (RR of failure 1.52 in subphenotype 1 vs 0.98 in subphenotype 2).
• A simplified machine learning classifier enabled practical subphenotype assignment and validated externally.

Clinical Implications

Subphenotype-based risk stratification could guide antibiotic selection and trial enrollment; patients in subphenotype 2 may benefit from intensified monitoring and tailored therapies.

Limitations

• Post hoc nature and potential cohort heterogeneity; phenotype assignment may vary across settings.
• Antibiotic effect modification was observational within a trial dataset and not randomized by subphenotype.

Future Directions

Prospective trials stratifying by HAP subphenotype to test tailored antibiotic regimens and adjunctive therapies; development of bedside tools incorporating minimal clinical variables +/- biomarkers.