Accuracy of magnetic resonance cholangiography compared to operative endoscopy in detecting biliary stones, a single center experience and review of literature.

AIM: To compare diagnostic sensitivity, specificity and accuracy of magnetic resonance cholangiopancreatography (MRCP) without contrast medium and endoscopic ultrasound (EUS)/endoscopic retrograde cholangiopancreatography (ERCP) for biliary calculi.

METHODS: From January 2012 to December 2013, two-hundred-sixty-three patients underwent MRCP at our institution, all MRCP procedure were performed with the same machinery. In two-hundred MRCP was done for pure hepatobiliary symptoms and these patients are the subjects of this study. Among these two-hundred patients, one-hundred-eleven (55.5%) underwent ERCP after MRCP. The retrospective study design consisted in the systematic revision of all images from MRCP and EUS/ERCP performed by two radiologist with a long experience in biliary imaging, an experienced endoscopist and a senior consultant in Hepatobiliopancreatic surgery. A false positive was defined an MRCP showing calculi with no findings at EUS/ERCP; a true positive was defined as a concordance between MRCP and EUS/ERCP findings; a false negative was defined as the absence of images suggesting calculi at MRCP with calculi localization/extraction at EUS/ERCP and a true negative was defined as a patient with no calculi at MRCP ad at least 6 mo of asymptomatic follow-up. Biliary tree dilatation was defined as a common bile duct diameter larger than 6 mm in a patient who had an in situ gallbladder. A third blinded radiologist who examined the MRCP and ERCP data reviewed misdiagnosed cases. Once obtained overall data on sensitivity, specificity, accuracy, positive predictive value (PPV) and negative predictive value (NPV) we divided patients in two groups composed of those having concordant MRCP and EUS/ERCP (Group A, 72 patients) and those having discordant MRCP and EUS/ERCP (Group B, 20 patients). Dataset comparisons had been made by the Student's t-test and χ (2) when appropriate.

RESULTS: Two-hundred patients (91 men, 109 women, mean age 67.6 years, and range 25-98 years) underwent MRCP. All patients attended regular follow-up for at least 6 mo. Morbidity and mortality related to MRCP were null. MRCP was the only exam performed in 89 patients because it did show only calculi into the gallbladder with no signs of the presence of calculi into the bile duct and symptoms resolved within a few days or after colecistectomy. The patients remained asymptomatic for at least 6 mo, and we assumed they were true negatives. One hundred eleven (53 men, 58 women, mean age 69 years, range 25-98 years) underwent ERCP following MRCP. We did not find any difference between the two groups in terms of race, age, and sex. The overall median interval between MRCP and ERCP was 9 d. In detecting biliary stones MRCP Sensitivity was 77.4%, Specificity 100% and Accuracy 80.5% with a PPV of 100% and NPV of 85%; EUS showed 95% sensitivity, 100% specificity, 95.5% accuracy with 100% PPV and 57.1% NPV. The association of EUS with ERCP performed at 100% in all the evaluated parameters. When comparing the two groups, we did not find any statistically significant difference regarding age, sex, and race. Similarly, we did not find any differences regarding the number of extracted stones: 116 stones in Group A (median 2, range 1 to 9) and 27 in Group B (median 2, range 1 to 4). When we compared the size of the extracted stones we found that the patients in Group B had significantly smaller stones: 14.16 ± 8.11 mm in Group A and 5.15 ± 2.09 mm in Group B; 95% confidence interval = 5.89-12.13, standard error = 1.577; P < 0.05. We also found that in Group B there was a significantly higher incidence of stones smaller than 5 mm: 36 in Group A and 18 in Group B, P < 0.05.

CONCLUSION: Major finding of the present study is that choledocholithiasis is still under-diagnosed in MRCP. Smaller stones (< 5 mm diameter) are hardly visualized on MRCP.