C-reactive protein-triglyceride-glucose index predicts acute pancreatitis severity: a retrospective cross-sectional study.

OBJECTIVE

Early prediction of acute pancreatitis (AP) severity is essential for timely intervention and improved outcomes. The C-reactive protein-triglyceride-glucose index (CTI) is a simple, routine laboratory-derived biomarker. It integrates signals of inflammation and insulin resistance (IR) and has demonstrated predictive value in cardiovascular and metabolic diseases. However, its association with AP severity remains unknown.

METHODS

This retrospective cross-sectional study included AP patients admitted between January 2019 and January 2026. CTI was calculated as 0.412 × ln (CRP, mg/L) + ln [triglycerides (mg/dL) × fasting glucose (mg/dL)/2]. We assessed the association between CTI and AP severity using Spearman's correlation, logistic regression, and receiver operating characteristic (ROC) curve analysis. Severity was graded according to both the revised Atlanta criteria (RAC) and the modified computed tomography severity index (MCTSI).

RESULTS

A total of 567 patients with AP were included in this study. The high CTI group had significantly more severe AP than the low CTI group under both the RAC and MCTSI grading systems (both P < 0.001). Spearman's rank correlation confirmed a positive monotonic association (P < 0.001). In multivariate models, CTI independently predicted severity under the RAC (odds ratio [OR] 1.180, 95% confidence interval [CI] 1.027-1.357, P = 0.019) and MCTSI (OR 1.436, 95% CI 1.237-1.667, P < 0.001). For MCTSI ≥ 8, the area under the ROC curve (AUC) reached 0.763 (95% CI 0.681-0.845, P < 0.001). The combined model of CTI + white blood cell (WBC) + lactate dehydrogenase (LDH) yielded the highest AUCs across all endpoints (range 0.697-0.795).

CONCLUSION

CTI independently predicts AP severity with moderate discriminative accuracy. Its routine availability and low cost support its potential as an adjunct for early risk stratification. Adding WBC and LDH further improved predictive performance, though the incremental benefit was modest.