Pathologic and Molecular Characterization of Branch Duct Intraductal Papillary Mucinous Neoplasm With Advanced Neoplasia Under-Recognized by Kyoto High-Risk Stigmata

Objectives: Kyoto high-risk stigmata (HRS) guide resection of branch-duct intraductal papillary mucinous neoplasms (BD-IPMNs) but miss a substantial proportion of high-grade dysplasia or invasive carcinoma (HGD/IC). We characterized the pathologic, molecular, and anatomic features of HGD/IC that Kyoto HRS fail to identify, and examined whether they are detectable before surgery by endoscopic ultrasound-guided needle-based confocal laser endomicroscopy (EUS-nCLE) and cyst-fluid molecular analysis. Methods: BD-IPMNs resected from 2015 to 2024 in two prospective trials were retrospectively analyzed. Dysplasia and IPMN-subtype were reassessed by blinded pancreaticobiliary pathologists with consensus and immunohistochemistry. Papillary architecture was classified as macropapillary versus non-macropapillary (micropapillary or flat). Diagnostic performance for HGD/IC was assessed against the consensus pathology reference standard. Results: Among 68 BD-IPMNs, 26 (38.2%) had HGD/IC, concentrated in non-macropapillary lesions (53.1%, 17/32, vs. 25.0%, 9/36; P =0.017), with the micropapillary pattern carrying the highest proportion (71.4%, 10/14). GNAS mutation was less frequent in non-macropapillary than macropapillary lesions (25.0% vs. 85.2%; P <0.001). Kyoto HRS were highly specific (95.2%) but only 53.8% sensitive; among 52 HRS-negative lesions, 23.1% (12/52) harbored HGD/IC, predominantly non-macropapillary, GNAS -negative, and located in the body or tail. Adding EUS-nCLE raised sensitivity from 52% to 80% (McNemar P =0.016) at 81% specificity, but nCLE was least sensitive in non-macropapillary lesions (43.8%), where absent GNAS provided a complementary preoperative marker. Conclusions: The HGD/IC missed by Kyoto HRS was characteristically micropapillary, GNAS -negative, and/or in the body or tail. Because architecture and subtype are histopathologic, preoperative recognition rests on combining Kyoto HRS, EUS-nCLE, and cyst-fluid molecular testing. Prospective external validation is warranted.