Document resource
In paediatric acute liver failure (PALF), risk prediction models are unable to predict survival with native liver with certainty. Alpha fetoprotein (AFP) has been proposed a promising prognostic biomarker and has been evaluated in neonatal and adult ALF but has not been assessed in PALF. 1 2 We aimed to evaluate AFP as a prognostic marker for survival with native liver (SNL) in PALF (prothrombin time >20 not corrected with vitamin K and liver dysfunction) and compare this to the previously validated CHLA-acute liver failure (CHALF)3 which uses biochemical values to predict SNL versus need for Liver Transplant (LT) or mortality.This was a UK health research authority approved single centre retrospective cohort study. Patients were identified from the departmental PALF database from Jan 2022 to August 2024 and included if they had at least one AFP measurement during their admission. Initial AFP, maximal AFP and initial:maximal AFP ratios were recorded. CHALF score was calculated on admission or when sufficient biochemical values were available.AFP was measured in 19 of the 37 patients transferred to tertiary liver unit with PALF. Median (IQR) age was 3.5 (2.4–4.2) years and 9 were female. 6 children had a single AFP measurement and others had up to 10. 11 patients required LT and 1 died.Median first AFP was similar (SNL: 53(26–129), non-SNL: 50(16–155); p=0.87). Maximum AFP and median AFP ratio were higher in the SNL group (medians 583 vs 111 and 7.70 vs 3.97) but this was not statistically significant. Median CHALF score was significantly lower in the SNL group (35 vs 38, p = 0.02). First AFP, maximum AFP and AFP ratio performed poorly compared to the CHALF score (area under receiver operating characteristic curves (AUC) 0.53, 0.73 and 0.62 vs 0.81). Using univariate logistic regression, a coefficient for first AFP was calculated and combined into the CHALF score calculation yielding an AUC of 0.84.This analysis shows that the CHALF score performed better than AFP measurements when used in isolation. The inclusion of AFP into the existing CHALF score may improve its performance further. While maximal AFP and AFP ratio were higher in the SNL group, the differences were not statistically significant. However this cohort was too small to draw meaningful conclusions and data should be considered hypothesis generating only. Dynamic AFP trends have been demonstrated to be prognostic in adults and neonates. Incomplete and variability of repeated AFP measurements in this dataset were important limitations. Caution needs to be exercised as many laboratories in the UK use the wet enzymatic method for ammonia estimation which may give erroneously high levels in presence of high transaminases.4 The external quality assessment (EQA) programs have attempted to harmonize AFP estimation.5 Further work should concentrate on evaluating AFP in a larger PALF cohort, measuring repeat AFP at defined time points and assessing if AFP may be a useful addition to existing risk prediction models.Abstract OC55 Figure 1Receiver operating characteristic curves for initial AFP, AFP ratio, maximum AFP, CHALF and CHALF + AFP with area under the curve indicatedReferences Rolfes PS, Sundaram SS, Sokol RJ, Taylor SA. Establishing neonate-specific prognostic markers in acute liver failure: admission alpha fetoprotein and novel neonatal acute liver failure scores predict patient outcomes. J Pediatr. 2024;272:114080.Schiødt FV, Ostapowicz G, Murray N, et al. Alpha-fetoprotein and prognosis in acute liver failure. Liver Transpl. 2006;12(12):1776–1781.Ascher-Bartlett JM, Bangerth S, Jordan S, et al. CHALF score: a novel tool to rapidly risk stratify children in need of liver transplant evaluation during acute liver failure. Transplantation 2024;108(4):930–939.Herrera DJ, Hutchin T, Fullerton D, et al. Non-specific interference in the measurement of plasma ammonia: importance of using a sample blank. Ann Clin Biochem. 2010;47:81–3.van Rossum HH, Holdenrieder S, Ballieux BEPB, et al. Investigating the current harmonization status of tumor markers using global external quality assessment programs: a feasibility study. Clin Chem. 2024 Apr 3;70(4):669–679.