Mendelian Randomization Analysis Dissects the Relationship between NAFLD, T2D, and Obesity and Provides Implications to Precision Medicine

ABSTRACT Background Non-alcoholic fatty liver disease (NAFLD) is epidemiologically correlated with both type 2 diabetes (T2D) and obesity. However, the causal inter-relationships among the three diseases have not been completely investigated. Aim We aim to explore the causal relationships among the three diseases. Design and methods We performed a genome-wide association study (GWAS) on fatty liver disease in ∼400,000 UK BioBank samples. Using this data as well as the largest-to-date publicly available summary-level GWAS data, we performed a two-sample bidirectional Mendelian Randomization (MR) analysis. This analysis tested the causal inter-relationship between NAFLD, T2D, and obesity, as well as the association between genetically driven NAFLD (with two well-established SNPs at the PNPLA3 and TM6SF2 loci) and glycemic and lipidemic traits, respectively. Transgenic mice expressing the human PNPLA3 I148I (TghPNPLA3-I148I) and PNPLA3 I148M (TghPNPLA3-I148M) isoforms were used to further validate the causal effects. Results We found that genetically instrumented hepatic steatosis significantly increased the risk for T2D (OR=1.3, 95% CI: [1.2, 1.4], p =8.3e-14) but not the intermediate glycemic phenotypes at the Bonferroni-adjusted level of significance ( p <0.002). There was a moderate, but significant causal association between genetically driven hepatic steatosis and decreased risk for BMI (β=- 0.027 SD, 95%CI: [−0.043, −0.01], p =1.3e-4), but an increased risk for WHRadjBMI (Waist-Hip Ratio adjusted for BMI) (β=0.039 SD, 95%CI: [0.023, 0.054], p =8.2e-7), as well as a decreased level for total cholesterol (β=-0.084 SD, 95%CI [−0.13, −0.036], p =6.8e-4), but not triglycerides (β=0.02 SD, 95%CI [−0.023, 0.062], p =0.36). The reverse MR analyses suggested that genetically driven T2D (OR=1.1, 95% CI: [1.0, 1.2], p =1.7e-3), BMI (OR=2.3, 95% CI: [2.0, 2.7], p =1.4e-25) and WHRadjBMI (OR=1.5, 95% CI: [1.3, 1.8], p =1.1e-6) causally increase the NAFLD risk. In the animal study, as compared to the TghPNPLA3-I148I controls, the TghPNPLA3-I148M mice developed higher fasting glucose level and reduced glucose clearance. Meanwhile, the TghPNPLA3-I148M mice demonstrated a reduced body weight, increased central to peripheral fat ratio, decreased circulating total cholesterol as compared to the TghPNPLA3-I148I controls. Conclusion This large-scale bidirectional MR study suggests that lifelong, genetically driven NAFLD is a causal risk factor for T2D (hence potentially a “NAFLD-driven T2D” subtype) and central obesity (or “NAFLD-driven obesity” subtype), but protects against overall obesity; while genetically driven T2D, obesity, and central obesity also causally increase the risk of NAFLD, hence a “metabolic NAFLD”. This causal relationship revealed new insights into disease subtypes and provided novel hypotheses for precision treatment or prevention for the three diseases.