182 Predictive Markers for Common Bile Duct Stones: A Two-Centre Cohort Analysis of Clinical, Biochemical, and Radiological Markers

Abstract Background & Aim Prediction of common bile duct stones (CBDs) may optimise the cost-benefit of MRCP service through avoiding unnecessary scans and prioritising patients on its waiting list. In certain practices, intraoperative cholangiogram can be skipped in presence of a reliable predictive model for CBDs. This study aims at identifying independent clinical, biochemical and radiological predictors of CBDs. Method A retrospective analysis of a consecutive 300 patients from two NHS hospitals who had MRCP to assess for CBDs. Clinical (Cholangitis, pancreatitis), biochemical (initial and pre-MRCP liver function tests LFTs, trends of LFTs) and radiological (Ultrasound findings) variables were analysed. Univariate and multivariate logistic regression were used to identify significant and independent predictors. Our proposed multivariate model predicting CBDs was evaluated using Receiver Operating Characteristic (ROC) curve analysis. Results CBD stones were identified in 37% of patients. Univariate analysis identified significant predictors including age (P<0.001), Pre-MRCP bilirubin (P=0.001), Pre-MRCP ALP (P=0.003), Ultrasound findings (P=0.004), and LFTs trend (P<0.001). Multivariate analysis revealed statistically independent predictors as increasing age (OR 1.021 per year; P=0.006), abnormal ultrasound findings (OR 1.812; P=0.026), and persistent abnormal LFTs trends (OR 2.150; P=0.027). This multivariate model demonstrated an AUC of 0.718 (95% CI: 0.660-0.776, P<0.001), indicating an acceptable diagnostic performance for CBDs. Marginal predictors included male gender (OR 1.681, P=0.081) and Pre-MRCP bilirubin (OR 1.009, P=0.81). Conclusions Increasing age, abnormal ultrasound findings, and rising or persistent high LFTs trends are independent predictors of CBDs. This model can be used as a risk stratification