Comparison of clinical characteristics and predictive performance of severity scoring systems in hypertriglyceridemic versus non-hypertriglyceridemic acute pancreatitis: a retrospective cohort study
Abstract
Objective To compare early clinical and biochemical characteristics of hypertriglyceridemic acute pancreatitis (HTG-AP) and non-hypertriglyceridemic acute pancreatitis (NHTG-AP), and to evaluate the predictive performance of commonly used severity scoring systems for adverse in-hospital outcomes. Subjects and methods This single-center retrospective cohort study included 268 consecutive adults with acute pancreatitis admitted from January to December 2023. Patients were classified as HTG-AP (n = 128) or NHTG-AP (n = 140) according to predefined etiological criteria. Demographic data, laboratory results within 24 hours, and six scoring systems, including BISAP, MCTSI, modified Marshall score, Ranson criteria, qSOFA, and HAPS, were analyzed. The primary endpoint was a composite adverse outcome comprising persistent organ failure, pancreatic or peripancreatic necrosis requiring intervention, intensive care unit stay longer than 7 days, or in-hospital mortality. Predictors were identified using LASSO-selected multivariable logistic regression and assessed by temporal validation, receiver operating characteristic analysis, calibration, and bootstrap resampling. Results Compared with NHTG-AP, HTG-AP patients were younger, had higher body mass index and recurrence rates, showed greater systemic inflammation and atherogenic dyslipidemia, and had lower free thyroxine levels. Composite adverse outcomes were more frequent in HTG-AP than in NHTG-AP. BISAP and MCTSI were independent predictors of adverse outcomes. The combined BISAP-MCTSI model showed good discrimination in derivation and validation cohorts and outperformed Ranson, qSOFA, and HAPS after Bonferroni correction. Conclusion HTG-AP represents a distinct clinical phenotype with worse in-hospital outcomes. A combined BISAP-MCTSI model may improve early risk stratification and warrants prospective validation.