Difficult cannulation during endoscopic retrograde cholangiopancreatography (ERCP) is a major determinant of complications, creating a clinical need for accurate pre-procedural risk stratification. This study aimed to identify MRI-based anatomical predictors of difficult ERCP cannulation, develop a predictive model for procedural difficulty, and support tailored prophylactic strategies for each level of difficulty.

A total of 121 patients who underwent contrast-enhanced MRI followed by ERCP at Hôtel-Dieu de France, Beirut, Lebanon, were included. Difficult cannulation was defined according to the criteria recommended in the ESGE Clinical Guideline on papillary cannulation and sphincterotomy techniques, requiring ≥1 of the following: more than five contacts with the papilla, cannulation time >5 minutes, or more than one unintended pancreatic duct cannulation or opacification. MRI-derived parameters included distal common bile duct (CBD) diameter, choledochoduodenal (CD) angle, and periampullary diverticulum (PAD) size. Variables and patients with excessive missingness were excluded, and multiple imputation was applied. Associations between anatomical features and ERCP difficulty were evaluated using univariable and multivariable logistic regression with LASSO regularization. ROC curves were used to determine optimal cutoff values.

A smaller distal CBD diameter was consistently associated with higher odds of difficult ERCP and remained an independent predictor across all models (OR = 0.872; 95% CI: 0.802–0.945; p < 0.001). A narrower CD angle was also significantly associated with difficult cannulation, mainly due to its strong association with unintended pancreatic duct cannulation or opacification (OR = 0.968; 95% CI: 0.950–0.985; p < 0.001). PAD size was not significantly associated (p > 0.4). ROC analysis identified a CBD diameter <5.5 mm as the optimal cutoff for predicting overall difficulty (AUC = 0.61; sensitivity 66.7%, specificity 52.2%). For predicting unintended pancreatic cannulations specifically, a CD angle <32.5° showed high specificity (89.9%) with an AUC of 0.65.

MRI-derived anatomical metrics, particularly distal CBD diameter and CD angle, serve as meaningful predictors of ERCP cannulation difficulty. These results align with previous findings by Lee et al., while highlighting the potential advantage of MRI’s superior soft-tissue contrast for pre-procedural risk assessment. Integrating MRI-based predictors into clinical workflows may enhance risk stratification and support more individualized prophylactic strategies ahead of ERCP.