Miss rate of pancreas divisum by magnetic resonance cholangiopancreatography in clinical practice

Objectives: Pancreas divisum is the most common congenital abnormality of the pancreatic anatomy. Magnetic resonance cholangiopancreatography has emerged as a noninvasive method of examining the pancreatic ductal anatomy. We aim to assess the sensitivity of MRCP for pancreas divisum. Methods: Patients with pancreas divisum at endoscopic retrograde cholangiopancreatography (ERCP) and who had prior MRCP between January 2001 and February 2006 were identified. Sensitivities were calculated for relevant subgroups with binomial 95% confidence intervals. Fisher exact P values were calculated. Results: Four hundred five patients had divisum at ERCP; of these, 111 (27%) had undergone MRCP before ERCP. Seventy-two (65%) patients had MRCP at outside centers. Twenty-three of the 72 MRCPs at referring centers correctly reported divisum. In contrast, the sensitivity was higher for the 24 MRCPs without secretin at our institution: 16 (67%). The sensitivity in the secretin-stimulated group was 67% (10/15). However, this was not different from that of MRCP without secretin. Of note, 9 (18%) of the 49 negative outside MRCPs had divisum suspected by the gastroenterologist reviewing the MRCP images before ERCP. Conclusions: Pancreas divisum seems to be often missed on MRCP, even when secretin is used. The absence of secretin, use of suboptimal magnetic resonance techniques, and inexperienced pancreatic MRCP examiners are all possible contributing factors.