Prognostic markers in hepatitis A-related pediatric acute liver failure and validation of the Peds-hepatitis A virus prognostic model.

Abstract

Hepatitis A virus (HAV) is the commonest cause for pediatric acute liver failure (PALF) in India. The objective of the study was to identify the predictors of mortality and to evaluate the utility of Peds-HAV model in a cohort of non-LT HAV-PALF. The study included HAV-related PALF from two non-transplant centers. The predictors of outcome were identified by univariate analysis followed by Cox regression analysis. The prognostic accuracy of Peds-HAV model, King's College Hospital (KCH) criteria and pediatric end-stage liver disease score (PELD) were evaluated. As many as 140 children with PALF were included, of whom 96 (68.6%) children had HAV-PALF. On Cox regression analysis, international normalized ratio (INR) (p < 0.001), jaundice to encephalopathy (JE) interval (p < 0.001) and hepatic encephalopathy (HE) grade 3/4 (p = 0.01) were independent predictors of mortality. The mortality rates were 0% (0/42), 14.3% (3/21), 60% (9/15) and 94.4% (17/18) when none, 1, 2 or 3 criteria of the Peds-HAV were met, respectively. Peds-HAV model at a listing cut-off of ≥ 2 criteria predicted death with 89.7% sensitivity and 89.6% specificity. In contrast, KCH criteria had a lower sensitivity of 62.1%. PELD score had a sensitivity of 89.7% and specificity of 85.1% at a cut-off of 30. The overall prognostic accuracy of Peds-HAV model (89.6%) was higher than those of KCH (83.3%) and PELD (86.5%). INR, HE grade and JE interval were independent predictors of mortality. The study provides an external validation of Peds-HAV model as a prognostic score in HAV-PALF. Not applicable as this is a retrospective study.