Using a two-sample Mendelian randomization design to investigate a possible causal effect of maternal lipid concentrations on offspring birth weight

Abstract

Background: The intrauterine environment is critical for fetal growth and development. However, observational associations between maternal gestational lipid concentrations and offspring birth weight (BW) have been inconsistent and ascertaining causality is challenging. Methods: We used a novel two-sample Mendelian randomization (MR) approach to estimate the causal effect of maternal gestational high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C) and triglyceride concentrations on offspring BW. Single nucleotide polymorphisms (SNPs) associated with serum HDL-C, LDL-C and triglyceride concentrations identified in the Global Lipids Genetics Consortium genome-wide association study meta-analysis (n = 188 577 European-ancestry individuals; sample 1) were selected as instrumental variables. The effects of these SNPs on offspring BW were estimated using a structural equation model in the UK Biobank and Early Growth Genetics consortium (n = 230 069 European-ancestry individuals; sample 2) that enabled partitioning of the genetic associations into maternal- (intrauterine) and fetal-specific effects. Results: We found no evidence for a causal effect of maternal gestational HDL-C, LDL-C or triglyceride concentrations on offspring BW [standard deviation change in BW per standard deviation higher in HDL-C = -0.005 (95% confidence interval: -0.039, 0.029), LDL-C = 0.014 (-0.017, 0.045), and triglycerides = 0.014 (-0.025, 0.052)]. Conclusions: Our findings suggest that maternal gestational HDL-C, LDL-C and triglyceride concentrations play a limited role in determining offspring BW. However, we cannot comment on the impact of these and other lipid fractions on fetal development more generally. Our study illustrates the power and flexibility of two-sample MR in assessing the causal effect of maternal environmental exposures on offspring outcomes.