Adult Liver Disease Prognostic Modelling for Long‐term Outcomes in Biliary Atresia

OBJECTIVES: 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). METHODS: 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). RESULTS: 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. CONCLUSIONS: 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.