Can Baseline Characteristics be Used to Predict Liver Disease Outcomes in Pediatric Nonalcoholic Fatty Liver Disease?

Abstract

Longitudinal studies on childhood predictors of nonalcoholic fatty liver disease (NAFLD) progression are lacking. The objective of this study was to determine whether baseline clinical or laboratory measures predict liver disease outcomes in a pediatric NAFLD cohort. A retrospective study of patients with presumed NAFLD was conducted using baseline and follow-up clinical and laboratory measures. Disease outcomes were defined using the mean serum alanine aminotransferase (ALT) levels from 24 to 36 months after the first visit. Logistic regression assessed the relationship between ALT progression/regression and predictor variables. Multivariable regression determined the best model for predicting the ALT outcome. Markov process modeling explored the likelihood for a patient to transition between ALT states. Of a total of 816 patients identified, 144 had sufficient data. Regression was seen in 26%, whereas 30% progressed. No baseline clinical or laboratory measurements had a significant effect on disease outcomes. Markov modeling demonstrated that subjects were more likely to either remain in their baseline ALT group or worsen rather than improve. Routinely obtained baseline clinical or laboratory measures cannot help risk-stratify youth with presumed NAFLD in terms of long-term outcomes. Close clinical, radiographic, and histologic evaluation of patients is warranted to determine those at risk of progression.