Metabolomic Predictors of Non-alcoholic Steatohepatitis and Advanced Fibrosis in Children

Abstract

Non-alcoholic fatty liver disease (NAFLD) is the leading cause of chronic liver disease in western countries both in children and adults. Metabolic dysregulation associated with gut microbial dysbiosis may influence disease progression from hepatic steatosis to inflammation and subsequent fibrosis. Using a multi-omics approach, we profiled the oral and fecal microbiome and plasma metabolites from 241 predominantly Latino children with non-alcoholic steatohepatitis (NASH), non-alcoholic fatty liver (NAFL), and controls. Children with more severe liver pathology were dysbiotic and had increased gene content associated with lipopolysaccharide biosynthesis and lipid, amino acid and carbohydrate metabolism. These changes were driven by increases in Bacteroides and concomitant decreases of Akkermansia, Anaerococcus, Corynebacterium, and Finegoldia. Non-targeted mass spectrometry revealed perturbations in one-carbon metabolism, mitochondrial dysfunction, and increased oxidative stress in children with steatohepatitis and fibrosis. Random forests modeling of plasma metabolites was highly predictive of non-alcoholic steatohepatitis (NASH) (97% accuracy) and hepatic fibrosis, steatosis and lobular inflammation (93.8% accuracy), and can differentiate steatohepatitis from simple steatosis (90.0% accuracy). Multi-omics predictive models for disease and histology findings revealed perturbations in one-carbon metabolism, mitochondrial dysfunction, and increased oxidative stress in children with steatohepatitis and fibrosis. These results highlight the promise of non-invasive biomarkers for the growing epidemic of fatty liver disease.

Keywords: NAFLD; NASH; microbiome; multiomics; pediatrics.