RESUMO
OBJECTIVE: The aim of this study is to determine the diagnostic accuracy of preoperative evaluation to detect main pancreatic duct involvement in pancreatic cystic lesions thus differentiating mixed intraductal papillary mucinous neoplasm (IPMN) from branch duct (BD)-IPMN. METHODS: The pathology database of pancreatic resections from 2000 to 2014 was reviewed. Main pancreatic duct-IPMNs and IPMNs with intracystic mass/nodules were excluded. The preoperative test characteristics were analyzed using surgical histopathology as the "gold standard." RESULTS: Sixty BD-IPMNs and 23 mixed-IPMNs were identified. Mixed-IPMNs were larger (mean [standard deviation], 4.14 [2.9] vs 2.74 [1.9] mm; P = 0.03) and demonstrated frequent high-grade dysplasia/adenocarcinoma (43% vs 12%, P = 0.004) than BD-IPMNs. Endoscopic ultrasound (EUS) (sensitivity, 80%; specificity, 78%; accuracy, 79%) had the best diagnostic accuracy, whereas magnetic resonance imaging (MRI) (sensitivity, 83%; specificity, 63%; accuracy, 68%) had the highest sensitivity for the diagnosis of mixed-IPMN. A combination of EUS and MRI reached maximum sensitivity but with decreased accuracy (sensitivity, 100%; specificity, 64%; accuracy, 67%). The area under the curve for receiver operation curve was 0.71 whereas the optimal cyst size to detect main duct involvement was 3 cm. CONCLUSIONS: For preoperative evaluation of pancreatic cystic lesions without evidence of intracystic nodules, a combination of MRI and EUS should be considered for improved detection of main duct involvement.