Differential diagnosis of benign or malignant intraductal papillary mucinous neoplasm of the pancreas by multidetector row helical computed tomography: evaluation of predictive factors by logistic regression analysis.

OBJECTIVE: The purpose of this study is to evaluate predictive factors for discriminating benign from malignant intraductal mucin-producing neoplasm (IPMN) of the pancreas on multidetector row computed tomography (MDCT).

MATERIALS AND METHODS: Fifty-three patients with IPMN underwent MDCT, and the imaging and pathological findings were evaluated. In patients with branch duct-type tumors, sex and age of the patient, location, shape, size and multiplicity of the cystic lesion, presence of mural nodule, and maximum diameter of main pancreatic duct (MPD) dilatation were evaluated by logistic regression analysis.

RESULTS: Tumors were classified as main duct-type (n = 7) and branch duct-type (n = 46). Among main duct-type tumors, all 7 lesions were diagnosed as malignant. Among 46 lesions of branch-type IPMN, 8 lesions were malignant, and 38 lesions were benign. On adjusted logistic regression analysis, combination factor of main duct dilatation and mural nodule or large cystic size had statistical significance for the risk of malignancy in branch duct-type IPMN.

CONCLUSIONS: Main duct-type IPMN is highly suggestive for malignancy. Combination factors of main ductal dilation and mural nodule, and main ductal dilation, and large cystic tumor size are thought to be predictive factors for malignant branch-type IPMN.