Inflammatory biomarkers for early prediction of severe acute pancreatitis: A systematic review and meta-analysis

Acute pancreatitis (AP) can rapidly progress to severe disease, making accurate early risk stratification essential for timely treatment. This systematic review and meta-analysis evaluated the predictive performance of inflammatory biomarkers for AP severity. PubMed, Web of Science, the Cochrane Library, and Embase were searched from inception to November 5, 2025, for diagnostic accuracy studies in adults with AP. Study quality was assessed using the Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2) tool, and diagnostic performance was pooled using bivariate or conventional random-effects models. Overall, 88 studies comprising 12,318 participants and evaluating 17 biomarkers were included. Polymorphonuclear neutrophil elastase (PMN-elastase) showed the strongest overall performance, with an area under the curve (AUC) of 0.96 within 1 day after admission and sensitivity and specificity of 0.93 and 0.94, respectively, at admission. Interleukin-6 (IL-6) also demonstrated strong early predictive performance, with an AUC of 0.93 within 1 day and sensitivity and specificity of 0.96 and 0.80, respectively, within 2 days. C-reactive protein (CRP) performed better during serial monitoring than during early assessment, reaching an AUC of 0.92 on day 5 and a diagnostic odds ratio (DOR) of 177.22 on day 6. The CRP-to-albumin ratio, procalcitonin, interleukin-8, carboxypeptidase B activation peptide, and neopterin showed moderate-to-high predictive performance. By contrast, the neutrophil-to-lymphocyte ratio and white blood cell count showed modest accuracy, whereas the platelet-to-lymphocyte ratio, erythrocyte sedimentation rate, and interleukin-10 had limited predictive value. PMN-elastase and IL-6 are promising biomarkers for early assessment of AP severity, whereas CRP is more suitable for dynamic monitoring. However, substantial methodological and clinical heterogeneity precludes direct ranking of biomarkers and highlights the need for standardized, prospective studies evaluating multimarker models.