Radiomic MRI model for predicting the development of worrisome features in branch-duct intraductal papillary mucinous...

Branch-duct intraductal papillary mucinous neoplasms (BD-IPMNs) are pancreatic cystic lesions originating from the pancreatic ducts, characterized by mucin production and progressive ductal dilation. They exhibit a wide spectrum of biological behavior, ranging from indolent lesions to entities with significant malignant potential. Although the 2024 Kyoto guidelines define worrisome features (WF) and high-risk stigmata (HRS) to support risk stratification and clinical management, predicting disease progression remains challenging. In this retrospective study, we investigated whether MRI-based radiomic analysis could identify, at the time of initial imaging, patients with BD-IPMNs who subsequently develop WF or HRS according to 2024 Kyoto guidelines. A total of 194 adult patients who underwent at least two MRI examinations between January 2011 and March 2025 were included, with a median follow-up of 53 months; progression was observed in 28.3% of patients, involving only some WF/HRS. Radiomic analysis included manual lesion segmentation, extraction of 107 features (shape, first- and second-order), and selection via LASSO within a weighted logistic regression framework to address class imbalance, using fivefold cross-validation; model performance was assessed with AUC and precision-recall metrics to account for skewed class distribution. After statistical analysis, nine shape-related features were found to be significant and a LASSO-based radiomic model, incorporating five features, was constructed. The model achieved an area under the curve (AUC) of 0.70 (95% CI 0.62-0.79). These results suggest that MRI-based radiomics may represent a valuable noninvasive tool for early risk stratification, predicting progression according to clinical-radiological criteria and potentially supporting personalized management of patients with BD-IPMNs. However, this study presents some limitations, including the retrospective design, the relatively small sample size and possible variability due to the use of multiple MRI scanners from different vendors. Prospective and multicentric studies with standardized imaging protocols are necessary to validate these findings and assess the added value of integrating radiomic data with clinical, histopathological and molecular information.