Adult Liver Disease Prognostic Modelling for Long-term Outcomes in Biliary Atresia: An Observational Cohort Study.

Abstract

To assess the utility of prognostic scoring systems for adolescents with biliary atresia (BA) surviving with native liver, for predicting the subsequent requirement for liver transplantation (LT). Single-centre retrospective analysis of 397 BA patients who received Kasai Portoenterostomy (KP) 1980-1996 and survived with the native liver at 16 years. Laboratory and clinical variables at 16 years (timepoint 16 years) were used to calculate (i) LT allocation scores; Model for End-Stage Liver Disease [MELD/MELD-sodium (Na)], and UK End-Stage Liver Disease (UKELD); (ii) Mayo Primary Sclerosing Cholangitis risk score (MayoPSC) and (iii) a modified Paediatric End-Stage Liver Disease (PELD) score. Scores were compared between patients requiring LT after 16 years of age (LT > 16 years), and those who survived with native liver, at the latest follow-up. Additional subgroup analysis for patients with data available at 12 years (timepoint 12 years). MELD (area under the receiver operating characteristic [AUROC] 0.847) and UKELD (AUROC: 0.815) at 16 years of age predict the need for LT > 16 years. No advantage for MELD-Na over MELD was demonstrated. MELD >8.5 and UKELD >47 predicted LT > 16 years with 84% and 79% sensitivity and 73% and 73% specificity. PELD had a similar performance to MELD, but superiority to UKELD. MayoPSC revealed predictive accuracy for LT >16 years (AUROC 0.859), with a score of >0.87 predicting LT > 16 years with 85% sensitivity and 82% specificity. At timepoint 12 years, MELD and MayoPSC predicted LT >16 years. Change in MELD, PELD and MayoPSC between 12 and 16 years of age, was associated with LT >16 years. Adult LT allocation scores may help monitor progress in adolescent BA, but the omission of relevant risk factors limits their utility for listing in this cohort. A BA-specific prognostic score would improve the management of adolescent BA.