IDENTIFYING A SEPSIS SUBPHENOTYPE CHARACTERIZED BY DYSREGULATED LIPOPROTEIN METABOLISM USING A SIMPLIFIED CLINICAL...

Area of Science:

• Sepsis pathophysiology
• Cholesterol metabolism
• Clinical prediction algorithms

Background:

• Sepsis patient heterogeneity is linked to cholesterol metabolism dysregulation.
• Two distinct sepsis subphenotypes exist: HYPO (low lipoproteins, high mortality) and NORMO (high lipoproteins, low mortality).
• A simplified clinical algorithm was developed for bedside recognition of these subphenotypes.

Purpose of the Study:

• To develop and validate a simplified clinical algorithm for bedside recognition of sepsis-related lipid dysregulation subphenotypes (HYPO vs. NORMO).
• To assess the predictive performance of the algorithm using clinical features.
• To evaluate the association between subphenotypes and 28-day mortality.

Main Methods:

• Analysis of data from four prospective internal studies and an external dataset of 281 patients.
• Development of a Random Forest classifier and logistic regression models using clinical features.
• Prediction of HYPO vs. NORMO subphenotypes and evaluation of model performance (AUC, sensitivity, specificity).

Main Results:

• Four clinical features (hepatic SOFA, cardiovascular SOFA, LDL-C, HDL-C) predicted subphenotypes with an AUC of 0.86.
• The algorithm achieved a sensitivity of 0.771 and specificity of 0.779.
• HYPO patients exhibited higher 28-day mortality (26% internal, 30% external) compared to NORMO patients (15% internal, 10% external).
• HYPO patients consistently had lower LDL-C, HDL-C, and total cholesterol than NORMO patients in both datasets.

Conclusions:

• A simplified clinical algorithm enables bedside identification of septic patients with lipid dysregulation subphenotypes.
• This tool may facilitate personalized management strategies for sepsis.
• External validation is recommended to confirm the algorithm's generalizability.