Lean NAFLD: A Distinct Entity Shaped by Differential Metabolic Adaptation.

BACKGROUND AND AIMS: Nonalcoholic fatty liver disease (NAFLD) affects a quarter of the adult population. A significant subset of patients are lean, but their underlying pathophysiology is not well understood. APPROACH AND RESULTS: We investigated the role of bile acids (BAs) and the gut microbiome in the pathogenesis of lean NAFLD. BA and fibroblast growth factor (FGF) 19 levels (a surrogate for intestinal farnesoid X receptor [FXR] activity), patatin-like phospholipase domain containing 3 (PNPLA3), and transmembrane 6 superfamily member 2 (TM6SF2) variants, and gut microbiota profiles in lean and nonlean NAFLD were investigated in a cohort of Caucasian patients with biopsy-proven NAFLD (n = 538), lean healthy controls (n = 30), and experimental murine models. Patients with lean NAFLD had a more favorable metabolic and histological profile compared with those with nonlean NAFLD (P < 0.05 for all). BA levels were significantly higher in NAFLD with advanced compared with earlier stages of liver fibrosis. Patients with lean NAFLD had higher serum secondary BA and FGF19 levels and reduced 7-alpha-hydroxy-4-cholesten-3-one (C4) levels (P < 0.05 for all). These differences were more profound in early compared with advanced stages of fibrosis (P < 0.05 for both). Lean patients demonstrated an altered gut microbiota profile. Similar findings were demonstrated in lean and nonlean murine models of NAFLD. Treating mice with an apical sodium-dependent BA transporter inhibitor (SC-435) resulted in marked increases in fgf15, a shift in the BA and microbiota profiles, and improved steatohepatitis in the lean model. CONCLUSIONS: Differences in metabolic adaptation between patients with lean and nonlean NAFLD, at least in part, explain the pathophysiology and provide options for therapy.

Core liver homeostatic co-expression networks are preserved but respond to perturbations in an organism- and disease-specific manner.

Findings about chronic complex diseases are difficult to extrapolate from animal models to humans. We reason that organs may have core network modules that are preserved between species and are predictably altered when homeostasis is disrupted. To test this idea, we perturbed hepatic homeostasis in mice by dietary challenge and compared the liver transcriptome with that in human fatty liver disease and liver cancer. Co-expression module preservation analysis pointed to alterations in immune responses and metabolism (core modules) in both human and mouse datasets. The extent of derailment in core modules was predictive of survival in the cancer genome atlas (TCGA) liver cancer dataset. We identified module eigengene quantitative trait loci (module-eQTL) for these predictive co-expression modules, targeting of which may resolve homeostatic perturbations and improve patient outcomes. The framework presented can be used to understand homeostasis at systems levels in pre-clinical models and in humans. A record of this paper's transparent peer review process is included in the supplemental information.