Comparative prognostic value of inflammation-based biomarkers for predicting in-hospital mortality in severe acute pancreatitis

BACKGROUND: Inflammation is central to the progression of severe acute pancreatitis (SAP), but the comparative prognostic value of inflammation-based biomarkers remains unclear. This study evaluated six indices-the systemic inflammatory response index (SIRI), neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), monocyte-to-lymphocyte ratio (MLR), glucose-to-lymphocyte ratio (GLR), and lactate-to-albumin ratio (LAR)-for predicting in-hospital mortality in SAP. METHODS: This retrospective study included 435 SAP patients admitted to a tertiary intensive care unit, from 2015 to 2024. Biomarkers were measured within 24 hour of admission. Logistic regression was used to identify independent predictors. Discriminatory performance was assessed with receiver operating characteristic (ROC) curves. Non-linear associations were examined using restricted cubic splines, and sensitivity analyses were performed using Youden index-based cutoffs. RESULTS: A total of 101 patients (23.2%) died during hospitalization. NLR, GLR, and LAR were independently associated with mortality, with LAR showing the strongest effect (adjusted odds ratio = 1.48). LAR demonstrated good discrimination (area under the curve [AUC] =0.710), comparable to the sequential organ failure assessment (SOFA) score (AUC = 0.741; P = 0.249). Adding LAR to SOFA resulted in a small but statistically significant improvement in discrimination (AUC = 0.753 vs. 0.741; P = 0.036), although the clinical impact of this incremental gain remains uncertain. Restricted cubic spline analyses showed approximately linear associations for all three biomarkers, and sensitivity analyses confirmed the robustness of these findings. CONCLUSION: LAR was the most informative inflammation-based biomarker for early mortality prediction in SAP, with performance comparable to the SOFA score. Its simplicity and predictive ability suggest that LAR may serve as a useful adjunct to existing scoring systems for early risk assessment.