Fatty liver disease affects roughly one in three adults globally, yet clinical outcomes vary enormously among individuals who appear metabolically similar. The emerging multi-omics picture of steatotic liver disease — integrating genomics, metabolomics, and proteomics — may finally explain why some patients progress to cirrhosis while others plateau, and could reshape how risk is assigned and treatment personalized.

Several common genetic variants anchor the genomic layer of this analysis. PNPLA3 and TM6SF2 variants confer elevated risk not only for hepatic steatosis but for progression to cirrhosis and hepatocellular carcinoma in both metabolic dysfunction-associated steatotic liver disease (MASLD) and alcohol-associated liver disease (ALD). HSD17B13 variants similarly modulate advanced-disease risk. Mechanistically, the cirrhosis-linked variants appear to converge on impaired hepatic lipid export rather than simply increased fat accumulation — a distinction with potential therapeutic implications. By contrast, GCKR variants associate strongly with steatosis but weakly with fibrotic progression, suggesting fat accumulation alone is insufficient to drive end-stage disease. At the metabolite level, characteristic disruptions in amino acids, bile acids, and lipid species differ between MASLD and ALD, reflecting distinct upstream drivers: insulin resistance and de novo lipogenesis in MASLD versus mitochondrial dysfunction in ALD. Plasma proteomics adds complementary signals tied to systemic inflammation and hepatic synthetic capacity.

This review situates these findings at a genuinely important inflection point. For years, MASLD risk stratification relied on imprecise surrogates like BMI and ALT. The identification of mechanism-specific genetic and metabolic signatures could enable earlier identification of the minority of patients destined for serious liver injury. However, a critical limitation pervades nearly all current multi-omics models: they are retrospective or post-hoc in design, meaning predictive performance in prospective clinical settings remains unvalidated. The field also lacks standardized biomarker panels and adequate representation across ethnicities, given that PNPLA3 variant frequencies differ substantially between populations. This is confirmatory and synthesizing work rather than paradigm-shifting, but it meaningfully advances the case for omics-guided liver disease management.