Height, adiposity, and fat distribution differ in men and women and, in part, may explain sex differences in susceptibilities to complex diseases like cardiovascular disease. Genome-wide association studies (GWAS) of these traits have previously reported sexually dimorphic associations, yet studies have primarily been limited to interrogation of variants with genome wide significant main effects only. Because of these biological differences by sex and as there is growing interest in the study of gene-environment interactions in the context of GWAS in general, we conducted sex-specific meta-analyses of 9 phenotypes: height (HT), weight (WT), body mass index (BMI), waist circumference (WC), hip circumference (HIP), WC/HC ratio (WHR), WC adjusted for BMI (WCadjBMI), HC adjusted for BMI (HCadjBMI), and WHR adjusted for BMI (WHRadjBMI). In the discovery stage, we performed sex-specific meta-analyses of 46 GWAS, comprising 60,586 men and 73,137 women. Each study used an additive model to test up to ∼2.8M imputed SNPs for association with inverse-normal transformed phenotypes. From our first scan based on the sex-specific association P-values (P women , P men ) across all phenotypes, we selected 619 independent SNPs at a false discovery rate (FDR) of 5% to take forward to replication. We also conducted a second scan based on the P-value for sex difference (P sex-diff ) with better power to detect signals of opposite effect direction, yet we did not detect any signal at FDR of 5%. Follow-up of the 619 SNPs in up to 62,395 men and 74,657 women, many of which were genotyped on Metabochip, a custom Illumina iSelect array to which we submitted sex-specific SNPs, resulted in 205 loci with genome-wide significanct (P women or P men < 5x10 -8 ) p values in the combined discovery and follow-up analysis. For those 205 loci, we found 4 loci with significant (P sex-diff < 0.05/205) and 14 loci with suggestive (P sex-diff <0.05) evidence for sex-difference including known sexually dimorphic associations with anthropometric traits ( GRB14 , 1q41 , VEGFA , ADAMTS9), known anthropometric trait associations without any prior evidence for sexual-dimorphism ( 14q23.1, 3q21.3, 6q14.1, 4q12, 12q24.31, SEC16B, 17q21.32 , and 13q31.3 ), and novel sex-specific associations with anthropometric traits ( 5q11.2, 5q23.1, PPARG, 2q37.1, 17p11.2 , and 5q35.1 ). Interestingly, we found that our replicated loci for WHR/WC were enriched with markers with sex-differences, and that these genetic effects were uniformly stronger in women compared to men. Collectively, these results underscore the gain from sex-stratified GWAS in order to better pinpoint the genetics of complex traits and illustrate a sexually dimorphic genetic underpinning to some of these traits. Our results more globally emphasize the need to consider gene-environment interaction when searching for genes influencing risk to complex disease.
Read full abstract