PWE-094 Patient selection for magnetic resonance cholangiopancreatography in management of common bile duct stones

Introduction Magnetic resonance cholangiopancreatography (MRCP) is highly sensitive and specific for detection of common bile duct (CBD) stones. The British Society of Gastroenterology guidelines on the management of suspected CBD stones recommend its use in patients with intermediate to high-risk as a non-invasive test to confirm the presence of ductal stones prior to therapy. Clinical, biochemical and ultrasound findings in isolation all suffer from low sensitivity, thus it is recommended that these variables be considered in combination. This study aims to identify the combination of clinical, biochemical or ultrasound findings that allow effective risk stratification in selecting patients for MRCP. Methods A retrospective analysis of 231 consecutive patients undergoing laparoscopic cholecystectomy was conducted. Clinical presentation, liver function tests (LFTs), and abdominal ultrasound findings were recorded, as well as findings from preoperative MRCP (if performed) and the eventual presence or absence of CBD stones at perioperative ERCP or on table cholangiogram. Multivariate logistic regression analysis was performed to investigate the relationship between clinical, biochemical and sonographic variables and the presence of CBD stone. Results 24 of 231 patients (10.4%) had CBD stones diagnosed intraoperatively or perioperatively. Raised LFTs (any of bilirubin, ALT or ALP) demonstrated significant odds ratio (OR) of 24.3 (95% CI 3.1 to 191.0), while clinical risk factors (history of jaundice, pancreatitis, cholangitis) or a dilated CBD on ultrasound showed increased but non-significant OR of 1.47 (95% CI 0.56 to 3.87) and 2.18 (95% CI 0.84 to 5.65), respectively. Combining LFTs and ultrasound findings to define a high-risk group, with abnormal LFT and/or abnormal CBD on ultrasound, yielded OR of 29.3 (95% CI 3.89 to 221.2). The combined high-risk group contained 114 patients (49%), with a sensitivity of 96% for presence of CBD stones, and positive predictive value of 20%. In the low risk group with normal LFTs and normal ultrasound, one out of 117 patients (0.85%) had a CBD stone. MRCP performed preoperatively in 74 (65%) of the high-risk group detected CBD stones in 14 patients (19%) with a negative predictive value of 98%. MRCP was performed on 10 (9%) of the low risk group: none showed CBD stones. Conclusion Patients with any abnormal LFT or abnormal CBD on ultrasound are at a significantly increased risk of CBD stones. Combining these biochemical and ultrasound features would enable efficient selection of high-risk patients to undergo MRCP prior to invasive therapy, and minimise the use of MRCP in low risk patients.