Drinking Patterns and CT/MRI Feature-Based Nomogram Models Can More Accurately Predict Acute Alcoholic Pancreatitis...

To develop and test a nomogram model based on drinking patterns and computed tomography and magnetic resonance imaging (CT/MRI) characteristics for recurrence of acute alcoholic pancreatitis (AAP) following its first onset.

Patients with initial AAP at the Affiliated Hospital of North Sichuan Medical College were retrospectively enrolled and categorized them into recurrence and non-recurrence groups. They were randomly assigned in a 7:3 ratio to form training and independent test sets, and their clinical and imaging data were collected and analyzed. Nomogram models were established to predict the recurrence of AAP.

Among 152 cases of initial AAP, 37 cases were recurrent and 115 were non-recurrent, with a mean (SD) age of 43.84 ± 10.28 years and 91.45% male participants. In the training set, 26 cases were recurrent and 80 were non-recurrent; in the independent test set, 11 cases were recurrent and 35 were non-recurrent. Multivariable logistic regression analysis showed that hyperlipidemia, pre-onset alcohol consumption, alcohol cessation, the Bedside Index for Severity in AP (BISAP) score, extrapancreatic inflammation on CT/MRI (EPIC/EPIM) score, and CT/MRI severity index (CTSI/MRSI) score were independent predictors of recurrence after initial AAP onset. Integrating these factors into a nomogram prediction model resulted in the area under the curve (AUC), sensitivity, and specificity values of 0.924 (95% CI 0.871-0.977), 0.885, and 0.838 for the training set and 0.843 (95% CI 0.714-0.972), 0.727, and 0.857 for the independent test set. The clinical model alone and the imaging model achieved AUCs of 0.799 (95% CI 0.703-0.896) and 0.827 (95% CI 0.742-0.912) in the training set and 0.800 (95% CI 0.646-0.954) and 0.686 (95% CI 0.492-0.880) in the independent test set, respectively.

Nomogram models based on drinking patterns and CT/MRI characteristics can more accurately predict AAP recurrence. The model serves as an effective tool for clinical prediction of AAP and helps clinicians in developing personalized prevention and treatment strategies.