Admission lipid profile and outcomes in sepsis requiring early invasive source control

Lipid metabolism is profoundly altered in sepsis, but its prognostic utility is underused in clinical practice. We evaluated lipid profiles performed within 24-48 h of ICU admission in patients with sepsis or septic shock who had already undergone surgical or interventional source control and explored associations with organ support and hospital mortality.

We conducted a single-center retrospective observational cohort study at the intensive care unit of Hospital Universitari Sant Joan de Reus (2016-2023). We included adults meeting Sepsis-3 criteria who underwent surgical or interventional source control within 48 h of hospital arrival. Only patients with available lipid profiles were included, which may introduce selection bias. Hospital mortality was defined as the primary clinical outcome. Secondary outcomes included organ support requirements, ICU and hospital length of stay, correlations with SOFA severity scores, and inflammatory biomarkers. Multivariable logistic regression models were constructed as an exploratory analysis to investigate the independent associations between lipid parameters and hospital mortality.

Of 266 screened patients, 150 met the inclusion criteria, comprising 61.3% males, with a median age of 72.1 years (IQR, 61.7-79.8), and an overall hospital mortality rate of 18.7% (n = 28). Non-survivors exhibited significantly lower concentrations of high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), and total cholesterol levels, and higher triglyceride levels than survivors. In ROC analyses, HDL-C showed the highest accuracy (AUC 0.966, 95% CI 0.904-1.000), with an optimal cutoff of 15 mg/dL, sensitivity of 96.4%, and specificity of 98.3%. This was followed by total cholesterol (AUC 0.963, 95% CI 0.901-1.000) and LDL-C (AUC 0.920, 95% CI 0.851-0.989). In contrast, triglycerides demonstrated a more modest predictive performance (AUC 0.849, 95% CI 0.772-0.926). In multivariable analysis, lower HDL-C levels (OR 0.802, 95% CI 0.727-0.885; p < 0.001) and lower albumin levels (OR 0.241, 95% CI 0.066-0.879; p = 0.031) remained independently associated with hospital mortality in the final parsimonious model derived using backward stepwise selection from an initial candidate model including chronic kidney disease, CRRT requirement, vasopressor use, LDL-C, VLDL-C, and triglycerides.

Early hypolipidemia was independently associated with hospital mortality in septic patients undergoing timely source control. Post-source control lipid alterations may reflect residual immunometabolic vulnerability and warrant further evaluation as exploratory prognostic biomarkers. Nevertheless, given the retrospective observational design and limited sample size, these results should be interpreted as hypothesis-generating and require prospective external validation.

Not applicable.