Large-Scale Transcriptomic Meta-Analysis of Mouse Liver Responses to High-Fat Diet Identifies a Reproducible Core and Duration-Associated Variation

High-fat diet (HFD)-fed mice are widely used to investigate hepatic responses relevant to metabolic dysfunction-associated steatotic liver disease (MASLD), but reproducibility across experiments remains unclear. We performed a multilevel meta-analysis of 64 mouse liver RNA-sequencing GEO Series (GSEs) comprising 105 HFD-versus-control contrasts and compared the results with three human liver cohorts relevant to MASLD. Among 18,234 mouse genes, 6920 showed significant pooled HFD effects, whereas only 59 met all three study-defined reproducibility criteria: pooled FDR < 0.05, a 95% prediction interval excluding zero, and ≥80% of evaluable GSEs matching the pooled-effect direction (GSE-balanced directional consistency). These 59 genes constituted a high-confidence reproducible subset. Twenty-five Hallmark pathways showed significant pooled enrichment, and HFD duration was associated with 2744 gene-level effects and multiple Hallmark pathways. Mouse–human correspondence was weak, but among 10,658 one-to-one orthologs, mouse directional consistency was associated with the odds of a same-direction human effect (OR 1.075 per 10-percentage-point increase; 95% CI 1.035–1.115; p = 1.5 × 10−4). The hepatic HFD response therefore includes a small reproducible core within a broader duration-sensitive transcriptional response, while greater cross-experiment recurrence in mice was associated with modestly higher directional concordance in MASLD-relevant human liver cohorts, without constituting direct model validation.