Sex-dependent baseline expression and severity coupling define key transcriptomic differences across MASLD

Abstract Background Sex contributes to heterogeneity in metabolic dysfunction-associated steatotic liver disease (MASLD), but stage-specific transcriptomic remodeling remains incompletely defined. We aimed to characterize sex-stratified hepatic transcriptomic changes across control, simple steatosis, and metabolic dysfunction-associated steatohepatitis (MASH) and to test key signals in independent cohorts. Methods and results GSE89632 liver-biopsy transcriptomes ( n = 63; female control 13, female simple steatosis 6, female MASH 10, male control 11, male simple steatosis 14, male MASH 9) were analyzed with GEO2R using |log2FC| ≥ 1 and Benjamini-Hochberg-adjusted p ≤ 0.05. Shared and sex-specific signatures were defined by Venn comparisons and functionally annotated with Enrichr. A pathway-guided 19-gene panel was then tested by two-way ANOVA, correlation, and regression against NAFLD Activity Score (NAS) and hepatic arachidonic acid, followed by targeted validation of BCL2L11, PNPLA3, P4HA1, and LEPR in GSE130970 and GSE167523. Transcriptomic remodeling differed by sex and stage, with the greatest differential-expression burden in male MASH. Sex differences reflected shared directional responses with sex-specific membership rather than opposite-direction regulation. Male-biased interferon/antiviral signaling and female-biased chromatin/cell-cycle and extracellular-matrix programs emerged as dominant pathway themes. PNPLA3 showed sex-dependent severity coupling, whereas P4HA1 and LEPR showed higher female baseline expression across NAS. Targeted evaluation in two independent cohorts showed phenotype-dependent and partial concordance: LEPR showed the strongest support for a female-higher expression pattern in the NAS-based cohort, whereas PNPLA3 showed concordant sex-dependent disease association across different analytical frameworks; P4HA1 showed more limited support. Conclusions Sex-related transcriptomic heterogeneity in MASLD appears to be driven by pathway architecture, severity coupling, and baseline expression differences rather than opposite-direction transcriptional regulation.