Poor Performance of Noninvasive Predictors of Esophageal Varices During Primary Prophylaxis Surveillance in Biliary Atresia

Objective: Our objective was to analyze performance of noninvasive markers for significant esophageal varices in relation to outcomes of endoscopic surveillance and primary prophylaxis in biliary atresia (BA). Methods: This was a prospective follow-up study of a national cohort of BA patients born between 1989 and 2017, including 72 consecutive patients who underwent variceal surveillance endoscopies. The risk for developing significant varices (grade >= 2) and variceal bleeding was compared between successful (postoperative total bilirubin <= 34 mu mol/L) and failed portoenterostomy (PE) patients. AUROC analyses and Wilcoxon signed ranks test were used to assess accuracy of noninvasive measures to predict the presence of significant varices after successful PE. Results: In total, 72 patients underwent 471 endoscopies during 427 follow-up years. Among 45 successful PE patients (63%), varices appeared later [at median age 1.6 (0.7-14) vs. 0.8 (0.4-1.9) years] and bled less often [7% vs. 41%, p b 0.001 for both] than after failed PE. Liver biochemistry, stiffness, and predictive scores showed poor accuracy for the presence of significant varices. After failed PE, lowered plasma albumin concentration predicted varices with an AUROC of 0.69 (95% CI 0.52-0.85, p = 0.030). After successful PE the varices prediction rule with AUROC 0.72 (95% CI 0.64-0.79) was the most accurate predictor. Individual predictors showed no meaningful changes between the two consecutive endoscopies leading to discovery of varices. Conclusion: Accurate targeting of endoscopies based on noninvasive predictors remains difficult during primary variceal prophylaxis protocol in BA. The differing prognoses after successful and failed PE should be considered in variceal surveillance and future studies. Type of study: Diagnostic/prognosis study. (c) 2020 Elsevier Inc. All rights reserved.