Validation of the 2012 International Consensus Guidelines Using CT and MRI: Branch Duct and Main Duct Intraductal Papillary Mucinous Neoplasms of the Pancreas
We thank Dr Goh for his interest and valuable comments on our study, “Validation of the 2012 international consensus guidelines using CT and MRI: branch duct and main duct intraductal papillary mucinous neoplasms of the pancreas.” We are pleased that he raised the issue regarding the validation method of the 2012 guidelines.1 Actually, we intended to find the significant variables among “high-risk stigmata” and “worrisome features” suggested in the 2012 guidelines and evaluated diagnostic performance of these variables (Table 5 in original article).2 “Overall” values located in the last row of each type of intraductal papillary mucinous neoplasm (IPMN) in Table 5 indicate sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy of the 2012 guidelines using all significant variables we analyzed.
As Dr Goh pointed out, we fully agreed that evaluation of diagnostic values using all “high-risk stigmata” and “worrisome features” of the 2012 guidelines in our patient cohort is important. Therefore, we included the following results of additional analyses (new Tables 1 and 2). Table 1 shows the number of patients of benign and malignant IPMN categorized as high risk, worrisome features or lower risk according to the 2012 guidelines using computed tomography (CT) and magnetic resonance imaging (MRI). A large number of patients with benign main duct (MD)-IPMN are classified as high risk or worrisome risk group. It is probably due to the fact that main pancreatic duct size is included in both high-risk stigmata and worrisome features of the 2012 guidelines. In other words, the specificity of the 2012 guidelines in MD-IPMN would be lower than that in branch duct IPMN (BD-IPMN).
Table 2 demonstrates the diagnostic performance of CT and MRI for detecting malignant potential of IPMN according to the number of high-risk stigmata or worrisome features of the 2012 guidelines. We calculated the diagnostic performance of worrisome features in patients without any high-risk stigmata. In high-risk group of BD-IPMN, PPV and NPV were 42.9% and 95.7%, respectively, for both CT and MRI. In worrisome risk group of BD-IPMN, PPV and NPV were 6.3% and 96.7% for CT, respectively, and 4.8% and 96% for MRI, respectively. Using the criteria that at least 1 worrisome feature was present (sum of high-risk stigmata and worrisome feature (HR/Wo) ≥1), PPV and NPV in BD-IPMN were 23.3% and 96.7% for CT, respectively, and 20% and 96% for MRI, respectively. On the contrary, in high-risk group of MD-IPMN, PPV and NPV were 65.4% and 78.3% for CT, respectively, and 61% and 79.5% for MRI, respectively. In worrisome risk group of MD-IPMN, PPV and NPV were 22.2% and 80% for CT, respectively, and 18.2% and 66.7% for MRI, respectively. Using the criteria that at least 1 worrisome feature was present (HR/Wo≥1), PPV and NPV in MD-IPMN were 47.7% and 80% for CT, respectively, and 45.7% and 66.7% for MRI, respectively. These values are lower than those obtained using significant variables of the 2012 guidelines in the main text of our study (Table 5).2 In addition, the diagnostic performance is not significantly different between CT and MRI (P > 0.12). This comparable performance between CT and MRI is consistent with our result in the main text (Table 5).2
Finally, we accept Dr Goh's concern that we included only surgically resected IPMN. As we described in the limitations, malignancy rates could have been overestimated than true prevalence because many benign and low-risk IPMNs are not surgically resected.2,3 As diagnostic accuracy, PPV, and NPV are dependent on disease prevalence, these values should be carefully interpreted in our study.