Rethinking imaging-based IPMN subtype classification: is mixed-type a necessary radiologic category?

To evaluate radiology-pathology concordance and interobserver agreement of intraductal papillary mucinous neoplasm (IPMN) subtype classification, and to determine whether categorical subtype classification adds value for malignancy risk stratification beyond main pancreatic duct (MPD) features.

In this single-center retrospective study, 144 consecutive patients who underwent surgical resection of an IPMN between 2005 and 2025 and had preoperative CT or MRI within 6 months of surgery were included. Images were independently reviewed by two blinded radiologists, with discrepancies adjudicated by a third. Adjudicated radiologic subtype classification was used for radiology-pathology concordance, logistic regression, and receiver operating characteristic (ROC) analyses; interobserver agreement was assessed using the independent reader interpretations. Logistic regression and ROC analyses identified predictors of malignancy.

Overall radiology-pathology concordance under the radiologic classification was 60.4% (87/144). Among pathologic mixed-type IPMNs, 60.3% (38/63) were classified as branch-duct IPMNs on imaging. Interobserver agreement was moderate for radiologic subtype (κ = 0.509) but excellent for MPD diameter (ICC = 0.909). On multivariable analysis, MPD diameter independently predicted malignancy (aOR, 1.31; 95% CI, 1.12-1.53; p = .001), whereas radiologic subtype was not. MPD diameter outperformed radiologic subtype for discrimination of malignancy (AUC, 0.737 vs. 0.624; p = .001). Adding radiologic subtype to MPD diameter provided no incremental discriminative value for malignancy (AUC, 0.739 vs. 0.737; p = .61).

Radiologic IPMN subtype classification demonstrated only modest concordance with pathology and moderate interobserver agreement. MPD diameter demonstrated higher interobserver agreement and superior discrimination for IPMN malignancy risk stratification. These findings suggest that the mixed-type IPMN subtype may not be necessary as a distinct category for malignancy risk stratification, and that greater emphasis on objective MPD measurement may provide a more reproducible and clinically informative approach.