Rationale

Sepsis is a life-threatening, dysregulated response to infection. Lipid biomarkers including cholesterol are dynamically regulated during sepsis and predict short-term outcomes. In this study, we investigated the predictive ability of lipid biomarkers for physical function and long-term mortality after sepsis.

Methods

Prospective cohort study of sepsis patients admitted to a surgical intensive-care unit (ICU) within 24 h of sepsis bundle initiation. Samples were obtained at enrollment for lipid biomarkers. Multivariate regression models determined independent risk factors predictive of poor performance status (Zubrod score of 3/4/5) or survival at 1-year follow-up.

Measurements and main results

The study included 104 patients with surgical sepsis. Enrollment total cholesterol and high-density lipoprotein (HDL-C) levels were lower, and myeloperoxidase (MPO) levels were higher for patients with poor performance status at 1 year. A similar trend was seen in comparisons based on 1-year mortality, with HDL-C and ApoA-I levels being lower and MPO levels being higher in non-survivors. However, multivariable logistic regression only identified baseline Zubrod and initial SOFA score as significant independent predictors of poor performance status at 1 year. Multivariable Cox regression modeling for 1-year survival identified high Charlson comorbidity score, low ApoA-I levels, and longer vasopressor duration as predictors of mortality over 1-year post-sepsis.

Conclusions

In this surgical sepsis study, lipoproteins were not found to predict poor performance status at 1 year. ApoA-I levels, Charlson comorbidity scores, and duration of vasopressor use predicted 1 year survival. These data implicate cholesterol and lipoproteins as contributors to the underlying pathobiology of sepsis.