Abstract
BACKGROUND: To assess whether glucose variability (GV) during acute pancreatitis (AP) predicts post-acute pancreatitis diabetes mellitus (PPDM-A), significantly affecting patient life quality. METHODS: This study was performed during 2016-2020 at Jinling Hospital, with a 3-year follow-up for each patient. Cox proportional hazards model was used to evaluate the association of GV with the possibility of developing PPDM-A. Dose-response relationships of GV with the three-year probability of PPDM-A were characterized based on a restricted cubic splines (RCS) model. GV was analyzed to predict the ability for PPDM-A by calculating area under the receiver operating characteristic curves (AUCs). RESULTS: PPDM-A rates rose from 16% at one year to 27.3% at three years post-AP. Multivariate Cox analysis indicated that the largest amplitude of glycemic excursions (LAGE) exhibited independent association with an increased PPDM-A risk within 3 years (HR = 1.21, 95% CI: 1.05-1.38, P <0.01). RCS results identified optimum LAGE threshold as 5.1, with significantly higher 3-year PPDM-A rates of abnormal LAGE group (LAGE ≥5.1 mmol/L) when compared with normal LAGE group (LAGE <5.1 mmol/L, P <0.001). AUCs for LAGE in predicting PPDM-A incidence in 12, 24, and 36 months were 0.883 (95% CI: 0.862-0.930), 0.916 (95% CI: 0.887-0.945), and 0.926 (95% CI: 0.895-0.948), respectively. CONCLUSIONS: LAGE in hospital stay accurately predicts PPDM-A. Further investigation plays an essential role in determining whether GV-targeting interventions can confer favorable clinical outcomes.