Prediction of main pancreatic duct involvement in intraductal papillary mucinous neoplasms on magnetic resonance imaging

Purpose

To evaluate the measurement of main pancreatic duct (MPD) diameter on MRI for predicting MPD involvement in intraductal papillary mucinous neoplasms (IPMN).

Methods

This retrospective study included 595 patients with surgically confirmed IPMN who underwent preoperative MRI from 2015 to 2022. Three independent readers measured the maximum MPD diameter on two-dimensional axial and coronal T2-weighted imaging. Inter-plane and inter-reader agreements were assessed using the intraclass correlation coefficient (ICC). Multivariable logistic regression identified clinical and radiological factors associated with MPD involvement. Accuracy, sensitivity, and specificity of MPD diameter cutoffs, including the 5-mm threshold from the 2024 International Consensus Guidelines, were calculated.

Results

Of the 595 patients (mean age: 64.6 years ± 8.6, 394 men), 423 (71.1%) had IPMN with MPD involvement, whereas 172 (28.9%) did not have MPD involvement. The mean maximum MPD diameter was 7.9 ± 5.1 mm. Inter-plane agreement was excellent (ICC = 0.977–0.988), as was inter-reader agreement (ICC = 0.963). Only a large MPD diameter on MRI was independently associated with MPD involvement (odds ratio = 1.29 [95% confidence interval; 1.14–1.47], p <.01). Use of a maximum MPD diameter cutoff of ≥ 5 mm for MPD involvement yielded accuracy, sensitivity, and specificity of 76.0%, 79.2%, and 68.0%, respectively.

Conclusion

Despite excellent inter-plane and inter-reader agreement, the MRI-based prediction of MPD involvement in IPMN has limitations.