Background: The objective of the present study was to use clinical and laboratory data to develop a model for predicting the presence of carcinoma in patients with intraductal papillary mucinous neoplasm (IPMN).
Methods: Data were collected on 81 patients with IPMN who underwent a pancreatic resection between 1989 and 2008 at Aichi Cancer Center Hospital. Variables analyzed included age, gender, laboratory findings (serum amylase, carcinoembryonic antigen, and carbohydrate antigen 19-9 level), pancreatic juice cytology grade, and imaging studies. Factors associated with the presence of carcinoma were evaluated by univariate and multivariate logistic regression analysis.
Results: Among the 81 patients with IPMN, 34 (42%) had malignant tumors (noninvasive carcinoma in 22 and invasive carcinoma in 12), and 47 (58%) had adenoma. On multivariate analysis, existing carcinoma was associated with female gender, main pancreatic duct IPMN, nodule size, and pancreatic juice cytology grade. Based on these variables, a predictive nomogram was developed. The area under the receiver operating characteristic curve (AUC) for the model was 0.903. The sensitivity and specificity of the model were 97.1 and 68.1%, respectively, in the validation study, for which the predictive probability of >10% was used to indicate the presence of carcinoma.
Conclusions: The nomogram has high diagnostic predictability for carcinoma in patients with IPMN and for individual cancer probability. This instrument may help to identify patients who need a surgical procedure.