The transition from non-alcoholic fatty liver disease (NAFLD) to non-alcoholic steatohepatitis (NASH) is a critical and poorly reversible hepatic transition that results from the cumulative effects of environmental, dietary and metabolic exposures — collectively referred to as the hepatic exposome, only a selected subset of which was assessed in this study. Epigenetic reprogramming, encompassing DNA methylation, histone modification, and dysregulation of non-coding RNAs (ncRNAs), may serve as a molecular correlate of these exposures and offers potential value as a biomarker platform for discriminating between NAFLD and NASH disease states. The historical NAFLD/NASH terminology is retained throughout for consistency with the EASL 2016 diagnostic criteria used for cohort recruitment; these entities correspond to metabolic dysfunction-associated steatotic liver disease (MASLD) and steatohepatitis (MASH) under current nomenclature.

This experimental cross-sectional study aimed to characterize the epigenetic landscape of exposure-induced responses in the liver in 387 adult participants who were divided into healthy controls (n = 130), NAFLD patients (n = 128), and confirmed NASH patients (n = 129), and to assess the translational diagnostic and risk-stratification value of a multi-analyte epigenetic panel that includes expression of DNA methyltransferase (DNMT3A), ten-eleven translocation enzyme (TET2), global 5-methylcytosine (5-mC) levels, histone acetylation marks and circulating microRNAs (miR-21, miR-122, miR-155, miR-34a, miR-192).

Peripheral blood and liver biopsy samples were processed for bisulfite pyrosequencing, quantitative reverse-transcription PCR (qRT-PCR), chromatin immunoprecipitation sequencing (ChIP-seq) and Luminex-based miRNA profiling. Transient elastography, NAFLD activity score (NAS) and comprehensive metabolic profiling were performed as clinical phenotyping methods. One-way analysis of variance (ANOVA), Tukey's honestly significant difference (HSD) post-hoc testing, receiver operating characteristic (ROC) analysis, multivariable regression adjusted for age, BMI and diabetes status, and Spearman rank correlations were used.

Highly significant inter-group differences were found for all 10 epigenetic biomarkers (ANOVA p < 0.001), differences that remained significant after multivariable adjustment for age, BMI and diabetes status. DNMT3A expression was progressively upregulated from Healthy (1.00 ± 0.12) through NAFLD (1.48 ± 0.31) to NASH (2.17 ± 0.44; F(2,384) = 218.4, p < 0.001), while TET2 exhibited the inverse pattern (0.39 ± 0.13 in NASH). Circulating miR-122 showed a marked fold-change (7.43 ± 1.34 vs 1.00 in controls). The combined DNMT3A + miR-21 panel achieved an AUC of 0.912 (95% CI: 0.884–0.940, sensitivity 86.0%, specificity 88.4%) for discriminating NAFLD from NASH, outperforming ALT alone (AUC = 0.787); a simplified three-marker panel (DNMT3A + miR-21 + miR-122) achieved a comparable AUC of 0.927. The most comprehensive 10-biomarker epigenetic panel performed best, with an AUC of 0.941.

The hepatic exposome is associated with a distinct and quantifiable epigenetic signature, measurable in blood, that discriminates NAFLD from NASH disease states in this cohort. Because the design is cross-sectional and derived from a single, non-externally-validated cohort, these findings are exploratory and do not establish prediction of future NAFLD-to-NASH transition or a causal relationship; independent external validation is required before any consideration of clinical application.