GE: Portuguese Journal of Gastroenterology (Jul 2023)

Prognostic Markers in Pediatric Acute Liver Failure

  • Andreia Filipa Nogueira,
  • Catarina Teixeira,
  • Carla Fernandes,
  • Rita Moinho,
  • Isabel Gonçalves,
  • Carla Regina Pinto,
  • Leonor Carvalho

DOI
https://doi.org/10.1159/000531269

Abstract

Read online

Introduction: Acute liver failure (ALF), although rare in children, is a complex progressive pathology, with multisystem involvement and high mortality. Isolated variables or those included in prognostic scores have been studied, to optimize organ allocation. However, its validation is challenging. This study aimed to assess the accuracy of several biomarkers and scores as predictors of prognosis in pediatric ALF (PALF). Methods: An observational study with retrospective data collection, including all cases of ALF, was defined according to the criteria of the Pediatric Acute Liver Failure Study Group, admitted to a pediatric intensive care unit (PICU) for 28 years. Two groups were defined: spontaneous recovery (SR) and non-SR (NSR) – submitted to liver transplantation (LT) or death at PICU discharge. Results: Fifty-nine patients were included, with a median age of 24 months, and 54% were female. The most frequent etiologies were metabolic (25.4%) and infectious (18.6%); 32.2% were undetermined. SR occurred in 21 patients (35.6%). In NSR group (N = 38, 64.4%), 25 required LT (42.4%) and 19 died (32.2%), 6 (15.7%) of whom after LT. The accuracy to predict NSR was acceptable for lactate at admission (AUC 0.72; 95% CI: 0.57–0.86; p = 0.006), ammonia peak (AUC 0.72; 95% CI: 0.58–0.86; p = 0.006), and INR peak (AUC 0.70; 95% CI: 0.56–0.85; p = 0.01). The cut-off value for lactate at admission was 1.95 mmol/L (sensitivity 78.4% and specificity 61.9%), ammonia peak was 64 μmol/L (sensitivity 100% and specificity 38.1%), and INR peak was 4.8 (sensitivity 61.1% and specificity 76.2%). Lactate on admission was shown to be an independent predictor of NSR on logistic regression model. Two prognostic scores had acceptable discrimination for NSR, LIU (AUC 0.73; 95% CI: 0.59–0.87; p = 0.004) and PRISM (AUC 0.71; 95% CI: 0.56–0.86; p = 0.03). In our study, the PALF delta score (PALF-ds) had lower discrimination capacity (AUC 0.63; 95% CI: 0.47–0.78; p = 0.11). Conclusions: The lactate at admission, an easily obtained parameter, had a similar capacity than the more complex scores, LIU and PRISM, to predict NSR. The prognostic value in our population of the promising dynamic score, PALF-ds, was lower than expected.

Keywords