Abstract 10916: Mendelian Randomization Studies Do Not Support a Causal Effect of Plasma Lipids on Insulin Sensitivity

Abstract

Background: Dyslipidemia is strongly associated with decreased insulin sensitivity and type 2 diabetes. However, previous studies have provided conflicting evidence of whether dyslipidemia is a cause of impaired nsulin sensitivity.

Objective: To use the Mendelian randomization design to test the hypothesis that genetically-determined changes of triglycerides (TG) and high-density lipoprotein cholesterol (HDL-C) plasma concentrations causally influence insulin sensitivity.

Methods: We meta-analyzed four studies comprising 2,590 European non-diabetic individuals with available genotyping array data and lipid measurements, as well as direct measures of insulin sensitivity either from a hyperinsulinaemic euglycaemic clamp or an insulin suppression test. For each lipid trait, we created individual genetic risk scores weighted on effect sizes from previous literature. We used these genetic risk scores to test and quantify the causal associations between each lipid trait and insulin sensitivity in an instrumental variable framework.

Results: We observed strong associations of the genetic risk scores for the irrespective lipid trait (P-values<10^-29). Lipids were strongly associated with insulin sensitivity in standard linear regression (P-values<10^-6). However, there was no evidence of an effect of the genetically-determined lipids on insulin sensitivity. The estimated causal effect sizes of lipids on insulin sensitivity were close to null and, for TG and HDL-C, different to the effects observed in standard linear regression models (P_diff <10^-4). Figure 1 displays the effect of triglycerides on insulin sensitivity estimated from linear regression and instrumental variable analysis respectively (all units on SD-scale).

Conclusion: Our results provide evidence that dyslipidemia in non-diabetic subjects is not a major cause of decreased insulin sensitivity.