Dynamic changes in peripheral blood lymphocyte trajectory predict the clinical outcomes of sepsis.

Summary

Using latent class mixed models on longitudinal lymphocyte counts, four trajectory phenotypes were identified and externally validated; the high-declining phenotype carried the highest 28-day mortality (~26%). A nomogram incorporating trajectory class and immune subsets provided individualized mortality risk estimates.

Key Findings

• Identified four longitudinal lymphocyte trajectory phenotypes (α high-declining, β stable-medium, γ high-increasing, δ stable-low) across retrospective (n=2,149) and validation (n=2,388) cohorts.
• The α (high-declining) phenotype had the highest severity and 28-day mortality (25.9%) in both cohorts.
• A multivariable model and nomogram using age, APACHE II, heart rate, NK cell count, infection source, and trajectory phenotype independently predicted 28-day mortality.

Clinical Implications

Incorporating serial lymphocyte trajectories and subsets can refine mortality risk assessment and identify candidates for immune-targeted therapies or immunoadjuvant trials.

Limitations

• Observational design limits causal inference and may be affected by measurement frequency and clinical practice variability
• Generalizability may vary across institutions and resource settings

Future Directions

Prospective interventional trials to test immunotherapy selection guided by trajectory phenotypes; integration with other omics (e.g., cytokines, transcriptomics) for multi-modal risk models.