Abstract
PURPOSE
To assess the additional value of secretin-enhanced MRCP (SMRCP) over conventional MRCP in diagnosing divisum.
METHODS
Retrospective HIPAA-compliant and IRB-approved review found 140 patients with SMRCP and ERCP correlation within 6 months of each other. All studies were anonymized and the SMRCP images (SMRCP image set) were separated from 2D and 3D MRCP and axial and coronal T2-weighted images (conventional MRI image set). Each image set on each patient was assigned different and randomized case numbers. Two reviewers (R1 and R2) independently reviewed the image sets for divisum vs. no divisum, complete divisum vs. incomplete divisum, and the certainty of diagnosis (1 = definitely certain, 2 = moderately certain, and 3 = unsure). ERCP findings were taken as gold standard.
RESULTS
There was no difference in age and gender between the divisum (n = 97, with 13 incomplete divisum) and no divisum (n = 43) groups. In diagnosing divisum anatomy, the sensitivity was higher for SMRCP compared to conventional MRI for R1 (84.5 vs. 72.2, p = 0.02) but not R2 (89.7 vs. 84.4, p = 0.25). The specificity was higher in SMRCP image set compared to conventional MRI (R1: 88.1 vs. 76.2, p = 0.01; R2: 81.4 vs. 65.1, p < 0.001). The mean area under ROC curve was higher for SMRCP image set (R1: 0.86 vs. 0.74, p = 0.01; R2: 0.87 vs. 0.74, p = 0.01). The certainty of diagnosis was higher in SMRCP image set compared to conventional MRI (p = 0.02 for both reviewers). SMRCP was not found to be superior in distinguishing incomplete from complete divisum. The main reasons for erroneous SMRCP diagnosis were the presence of an ansa loop in the main duct and ductal strictures due to chronic pancreatitis.
CONCLUSION
Even though the reviewers had more sequences (axial and coronal) to evaluate in the non-secretin image set, there was some improvement in diagnosing divisum with SMRCP.
Collapse