Plasma lipid peroxidation as a non-invasive biomarker for distinguishing septic from non-septic postoperative critically ill patients.

BACKGROUND: Sepsis remains a major diagnostic challenge in critically ill postoperative patients, where infection must be distinguished from sterile inflammatory responses. This uncertainty often leads to empirical antibiotic use, contributing to antimicrobial overuse, resistance and drug-related toxicity. Current biomarkers such as C-reactive protein (CRP) and procalcitonin (PCT) have limited specificity in this setting. Lipid peroxidation (LPO), one component of the host redox response, may provide complementary diagnostic information beyond classical inflammatory biomarkers. METHODS: In this multicentre observational study, plasma samples from 279 adult postoperative intensive care patients were analysed to quantify LPO through malondialdehyde measurement. Patients were classified as septic (n = 131) or non-septic (n = 148) according to clinical criteria. Associations between LPO and infection were evaluated using multivariable logistic regression adjusted for age, sex and CRP. Diagnostic performance was assessed by receiver operating characteristic (ROC) analysis, nomograms, decision curve analysis and calibration plots. RESULTS: Plasma LPO levels were significantly lower in septic patients than in non-septic individuals. In multivariable analysis, lower LPO levels remained independently associated with infection (odds ratio 0.63, p < 0.001). Combining LPO with CRP and PCT improved diagnostic performance, achieving an area under the ROC curve of 0.902 and providing higher net clinical benefit than models excluding LPO. CONCLUSIONS: Plasma LPO is reduced in septic postoperative patients and independently associated with infection. Its integration into multimarker diagnostic models may improve infection identification, reduce diagnostic uncertainty and support more rational antibiotic use in critical care settings.