OBJECTIVE: This study was designed to assess the diagnostic accuracy of standard nonsecretin-enhanced preprocedural magnetic resonance imaging/cholangiopancreatography (MRI/MRCP) in patients with and without pancreas divisum. METHODS: Patients undergoing MRI/MRCP followed by endoscopic retrograde cholangiopancreatography with between 2009 and 2016 were reviewed. The diagnostic accuracy of the MRI/MRCP was evaluated against the pancreatography. A subsequent independent blinded re-review performed by an expert abdominal radiologist was also evaluated. Multivariate binary logistic regression was performed to assess the impact of clinicopathologic factors on the diagnostic accuracy. RESULTS: A total of 189 patients were included in analysis. The sensitivity, specificity, positive predictive value, and negative predictive value of MRI/MRCP for pancreas divisum were 63%, 97%, 94%, and 82% initially and 81%, 91%, 91%, and 82% on the expert review. Motion artifact, the presence of pancreatic tumor, and pancreatic necrosis were not found to significantly impact the accuracy. A normal diameter pancreatic duct (P = 0.04) and complete divisum anatomy were correlated with improved accuracy (P = 0.001). CONCLUSIONS: Although expert review, normal duct diameter, and complete divisum are associated with increased sensitivity, pancreas divisum may be uncharacterized by preprocedural MRI in 19% to 37% of patients before the index endoscopic retrograde cholangiopancreatography.
OBJECTIVE: This study was designed to assess the diagnostic accuracy of standard nonsecretin-enhanced preprocedural magnetic resonance imaging/cholangiopancreatography (MRI/MRCP) in patients with and without pancreas divisum. METHODS:Patients undergoing MRI/MRCP followed by endoscopic retrograde cholangiopancreatography with between 2009 and 2016 were reviewed. The diagnostic accuracy of the MRI/MRCP was evaluated against the pancreatography. A subsequent independent blinded re-review performed by an expert abdominal radiologist was also evaluated. Multivariate binary logistic regression was performed to assess the impact of clinicopathologic factors on the diagnostic accuracy. RESULTS: A total of 189 patients were included in analysis. The sensitivity, specificity, positive predictive value, and negative predictive value of MRI/MRCP for pancreas divisum were 63%, 97%, 94%, and 82% initially and 81%, 91%, 91%, and 82% on the expert review. Motion artifact, the presence of pancreatic tumor, and pancreatic necrosis were not found to significantly impact the accuracy. A normal diameter pancreatic duct (P = 0.04) and complete divisum anatomy were correlated with improved accuracy (P = 0.001). CONCLUSIONS: Although expert review, normal duct diameter, and complete divisum are associated with increased sensitivity, pancreas divisum may be uncharacterized by preprocedural MRI in 19% to 37% of patients before the index endoscopic retrograde cholangiopancreatography.