Dissecting the shared genetic architecture between nonalcoholic fatty liver disease and type 2 diabetes

Observational studies have reported a bidirectional correlation between nonalcoholic fatty liver disease (NAFLD) and type 2 diabetes (T2D), but the shared genetic basis between the two conditions remains unclear. Using genome-wide association study (GWAS) summary data from European-ancestry populations, we examined the cross-trait genetic correlation and identified genomic overlaps and shared risk loci. We employed a latent causal variable model and Mendelian randomization (MR) analysis to infer causal relationships. Colocalization analysis and conditional/conjunctional false discovery rate (condFDR/conjFDR) were used to identify genomic overlaps and shared risk loci. Two-step MR analysis was utilized to identify potential mediators. We observed a strong positive genomic correlation between NAFLD and T2D (rg = 0.652, P = 5.67 × 10-6) and identified tissue-specific transcriptomic correlations in the pancreas, liver, skeletal muscle, subcutaneous adipose, and blood. Genetic enrichment was observed in NAFLD conditional on associations with T2D and vice versa, indicating significant polygenic overlaps. We found robust evidence for the causal effect of NAFLD on T2D, particularly insulin-related T2D, rather than vice versa. Colocalization analysis identified shared genomic regions between NAFLD and T2D, including GCKR, FTO, MAU2-TM6SF2, and PNPLA3-SAMM50. High-density lipoprotein cholesterol and insulin were partly mediated the association between NAFLD and T2D. These findings unveil a close genetic link between NAFLD and T2D, shedding light on the biological mechanisms connecting NAFLD progression to T2D.