Abstract P199: Alcohol Consumption Modified the Effect of Leptin on Blood Lipids: A Mendelian Randomization Study

Abstract

Objective: The current study aimed to evaluate the relation between leptin and serum lipids, and to explore whether alcohol consumption modifies the effect of leptin on lipids. Method: We conducted a Mendelian randomization analysis to the baseline data for 3,780 participants of the Framingham Heart Study Generation III cohort. A genetic risk score (GRS) for leptin was generated by summing leptin increasing alleles of 3 single-nucleotide polymorphisms, weighted by their corresponding effect sizes on leptin reported by Kilpelainen et al. For those taking lipid lowering medications, total cholesterol (TC) was adjusted as TC/0.8, and low density lipoprotein (LDL) was imputed by replacing the adjusted TC into the Friedewald equation. Leptin and triglycerides (TG) were log transformed. Associations between the GRS and leptin, leptin and imputed lipids, and GRS and imputed lipids were assessed, respectively, by multivariate linear regression models, controlling for age and sex in basic model and additionally controlling for education, smoking, drinking, and physical activity in the fully adjusted model. Interactions between the GRS and alcohol drinking was evaluated by adding an interaction term, GRS*drinking, in the fully adjusted model. Sensitivity analyses were performed among those not taking lipid or glucose lowering medications. Result: In univariate analyses, the GRS was significantly associated with leptin (Beta=1.44, P =4.55E-4), but not with age ( P =0.33), sex ( P =0.62), education ( P =0.37), smoking ( P =0.59), drinking ( P =0.42), or physical activity ( P =0.80). Leptin was significantly associated with age and sex adjusted LDL (Beta=7.29, P =5.95E-38), high density lipoprotein (HDL) (Beta=-4.30, P =4.35E-62), TG (Beta=0.22, P =1.05E-119), and TC (Beta=8.06, P =1.72E-36). The GRS for leptin was associated with HDL (Beta=-16.72, P =0.01), but not LDL (Beta=2.64, P =0.85), TG (Beta=-0.35, P =0.16), or TC (Beta=-23.34, P =0.15), in partially adjusted model. In the fully adjusted model, association between the GRS and HDL was still significant (Beta=-16.70, P =0.01). When stratified by drinking status, the GRS for leptin was significantly associated with reduced LDL (Beta=-99.26, P =0.02), TG (Beta=-2.4, P =0.002), and TC (Beta=-156.10, P =0.001) among non-current drinkers, and reduced HDL (Beta=-19.98, P =0.005) among current drinkers, adjusting for age and sex. After further adjustment, these associations were still significant. In the fully adjusted model, significant interactions between the GRS and alcohol drinking were identified for LDL ( P =0.02), TG ( P =0.005), and TC ( P =0.008). Sensitivity analyses among those not taking lipid or glucose lowering medications revealed similar associations. Conclusion: Out study provided evidence for a causal relationship between leptin and lipids, and an interaction effect of alcohol drinking on leptin and lipids associations.