Intraductal papillary mucinous neoplasm (IPMN) of the pancreas has malignant potential. Predicting invasive IPMN has proven difficult and controversial. We tried to identify predictive factors for invasive IPMN.

Thirty six patients underwent resection for IPMN from February 2001 to July 2011. Clinicopathological features including demographic, imaging, microscopic, and serological findings were retrospectively reviewed. Receiver operating characteristic (ROC) curve analysis was used to analyze sensitivity and specificity of all possible cut-off values for the diameter of the main pancreatic duct and mass size predicting invasive IPMN. Student t-test, chi-square test, and logistic regression were used for univariate and multivariate analysis.

The mean age was 63.5±8.4 years. Males were more commonly affected (58.3% vs 41.7%). Pancreaticoduodenectomy was performed in 55.6% of patients, distal pancreatectomy in 36.1%, and central pancreatic resection in 8.3%. Non-invasive IPMNs were present in 80.6% (n=29), whereas invasive IPMNs were present in 19.4% (n=7). In univariate analysis, tumor location ( p =0.036), Kuroda classification ( p =0.048), mural nodule ( p =0.016), and main duct dilatation (≥8 mm) ( p =0.006) were statistically significant variables. ROC curve analysis showed that a value of 8 mm for the main duct dilatation and a value of 35 mm for the size of the mass lesion have 80% sensitivity and 75% specificity and 100% sensitivity and 82.6% specificity, respectively. However, in multivariate analysis, main ductal dilatation (≥8 mm) was identified to be the only independent factor for invasive IPMN ( p =0.049).

Main duct dilatation appears to be a useful indicator for predicting invasive IPMN.