Sa1335 DYNAMIC LIVER TEST PATTERNS DO NOT PREDICT BILE DUCT STONES

Abstract

Numerous models have been developed to predict the presence of choledocholithiasis. Recent work has shown that these algorithms, which focus on a single set of liver tests and imaging findings, perform suboptimally. We aimed to develop a dynamic model using fluctuations in liver enzyme values combined with imaging and clinical history to more accurately forecast persistent bile duct stones. Consecutive unique hospitalized patients who received their first endoscopic retrograde cholangiopancreatography (ERCP) at Los Angeles County Medical Center between January 2010 and November 2016 for suspected bile duct stones were reviewed. We categorized the liver enzyme patterns prior to ERCP as increasing, decreasing, or alternating values. The primary outcome of this study was the proportion of patients with these liver enzyme patterns who had stones confirmed on ERCP. As a secondary outcome, we determined whether the American Society of Gastroenterology (ASGE) risk stratification algorithm was predictive of stones. Of the 604 hospitalized patients who underwent initial ERCP at our medical center, 82.3% were considered high risk, 17.4% intermediate risk, and 0.3% low risk for choledocholithiasis based on ASGE guidelines. Bile duct stones were confirmed in 410 (67.9%) patients. Detailed assessment of liver enzyme patterns yielded no consistent predictive dynamic patterns. Additionally, the ASGE risk stratification algorithm was not predictive of stones. For the 236 patients for whom magnetic resonance cholangiopancreatography (MRCP) was performed, this imaging modality was shown to have high predictive value for presence of stones on ERCP (OR 6.73 [95% confidence interval: 3.50-12.92]). Based on our exploration of predominantly intermediate and high risk patients, there is no dynamic pattern of liver enzyme tests that accurately predicts the presence of stones on ERCP. The ASGE risk stratification algorithm was also insensitive in this population. MRCP or endoscopic ultrasound may be considered to avoid unnecessary ERCP and associated complications.Table 1Multivariate associations between dynamic liver enzyme fluctuations and ERCP findings.Trends (with at least 20% difference):FrequencyStonesStones/SludgeOR95% CIOR95% CIT. BiliSingle value110 (18.2%)1.001.00Two values, increasing (or equal)148 (24.5%)0.590.34-1.030.480.23-1.03Two values, decreasing61 (10.1%)1.410.65-3.040.690.26-1.85Three values, increasing (or equal)160 (26.5%)0.590.35-1.020.580.27-1.24Three values, decreased then increased15 (2.5%)2.840.59-13.791.870.22-16.03Three values, increased then decreased24 (4.0%)0.990.37-2.640.780.22-2.78Three values, decreasing86 (14.2%)0.960.51-1.840.580.25-1.35ALKPSingle value110 (18.2%)1.001.00Two values, increasing (or equal)191 (31.6%)0.690.41-1.180.490.24-1.02Two values, decreasing19 (3.2%)1.870.50-7.001.930.23-16.09Three values, increasing (or equal)260 (43.1%)0.710.43-1.170.620.30-1.26Three values, decreased then increased2 (0.3%)NANANANAThree values, increased then decreased0 (0.0%)NANANANAThree values, decreasing22 (3.6%)1.560.53-4.580.590.17-1.99ASTSingle value112 (18.6%)1.001.00Two values, increasing (or equal)94 (15.6%)0.610.33-1.140.500.22-1.15Two values, decreasing123 (20.4%)0.780.43-1.400.490.23-1.08Three values, increasing (or equal)95 (15.8%)0.50*0.27-0.910.670.29-1.58Three values, decreased then increased17 (2.8%)0.450.15-1.330.19*0.06-0.61Three values, increased then decreased19 (3.2%)1.500.44-5.041.110.22-5.61Three values, decreasing143 (23.7%)0.840.47-1.480.700.32-1.54ALTSingle value110 (18.2%)1.001.00Two values, increasing (or equal)124 (20.5%)0.670.38-1.190.520.23-1.14Two values, decreasing86 (14.2%)0.870.46-1.660.550.24-1.30Three values, increasing (or equal)139 (23.0%)0.620.35-1.080.590.27-1.28Three values, decreased then increased7 (1.2%)0.220.05-1.060.12*0.02-0.60Three values, increased then decreased13 (2.2%)1.510.37-6.121.720.20-14.93Three values, decreasing125 (20.7%)0.980.54-1.760.700.31-1.57This table documents the frequency and multivariate associations of the six different liver enzyme fluctuation patterns with ERCP findings based on at least 20% difference among sequential enzyme values. Odds ratio and confidence interval of each pattern of total bilirubin, alkaline phosphatase, aspartate aminotransferase, and alanine aminotransferase were recorded. Abbreviations: ERCP = endoscopic retrograde cholangiopancreatography; T. Bili = total bilirubin; ALKP = alkaline phosphatase; AST = aspartate aminotransferase; ALT = alanine aminotransferase; OR = odds ratio; CI = confidence interval. NA: due to only 2 datapoints available. * = statistically significant with p < 0.05. Open table in a new tab Table 2Univariate associations between ASGE predictors and ERCP findings.FrequencyStonesStones/SludgeOR95% CIOR95% CI“Very strong” predictors:Bile duct stone seen on US326 (54.0%)1.340.95-1.891.210.78-1.88T. Bili >4 mg/dL203 (33.6%)0.880.61-1.260.920.58-1.47Clinical ascending cholangitis(excluded)NANANANA“Strong” predictors:CBD dilation >6 mm on US437 (72.5%)1.100.75-1.611.010.62-1.65T. Bili 1.8-4 mg/dL245 (40.6%)0.820.58-1.160.960.61-1.49“Moderate” predictors:Abnormal ALKP, AST, or ALT197 (32.6%)1.250.86-1.801.040.65-1.66Age >55 years106 (17.6%)0.670.43-1.030.810.47-1.41Clinical gallstone pancreatitis109 (18.1%)0.350.22-0.530.340.21-0.56Likelihood of choledocholithiasis per ASGE guidelines:High risk497 (82.3%)0.980.63-1.540.940.53-1.69Intermediate risk105 (17.4%)0.990.63-1.551.040.58-1.85Low risk2 (0.3%)NA*NA*NA*NA*This table records univariate association findings of ASGE risk categories and their individual predictors with ERCP findings, as well as their associated odds ratios and 95% confidence intervals. Abbreviations: ERCP = endoscopic retrograde cholangiopancreatography; T. Bili = total bilirubin; ALKP = alkaline phosphatase; AST = aspartate aminotransferase; ALT = alanine aminotransferase; CBD = common bile duct; US = ultrasound; ASGE = American Society of Gastrointestinal Endoscopy; OR = odds ratio; CI = confidence interval; NA = not available. High risk = any very strong predictor or both strong predictors; Low risk = no predictors present; Intermediate risk = all other patients. * Result is not available as the 2 low risk patients predicted the presence of stones perfectly. Open table in a new tab