Abstract 22: Influence of Smoking and Sex on Genetic Associations of Body Fat Distribution: The Giant (Genetic Investigation of ANthropometric Traits) Consortium

Visceral fat and smoking are important risk factors for CVD and other long-term poor health outcomes. Central adiposity is frequently assessed in population studies using waist-to-hip ratio adjusted for BMI (WHRadjBMI). Genome-wide association studies (GWAS) have discovered loci that influence WHRadjBMI, sometimes with sex-specific effects. However, we do not know whether and how smoking behavior influences the association between genetic variants and WHRadjBMI. Our study investigates the influence of smoking and genetic variants on WHRadjBMI, with particular consideration for differential effects by sex. Forty-one GIANT GWAS (108,397 participants) were used to investigate the influence of smoking on SNP WHRadjBMI associations using sex-stratified models considering SNP main effects adjusted for smoking status, and SNP main effects stratified by smoking status. Results from each study were meta-analyzed using an inverse-variance weighted fixed-effects model for men and women separately as well as combined. A total of 19 SNPs reached genome-wide significance (p<5x10-8) having adjusted the SNP effect for smoking across strata (women, men, and combined), including 11 previously reported WHRadjBMI loci. For known WHRadjBMI loci, the influence of smoking varied by sex. In women, four known WHRadjBMI SNPs (TBX15, DNM3, RSPO3, and ADAMTS9) had a greater effect in smokers vs. nonsmokers (TBX15: β(SE) = 0.0506 (0.014), p = 3.84x10-4) vs. β(SE) = 0.0382 (0.0072), p = 1.14x10-7). Three of these (TBX15, DNM3, RSPO3) have been previously associated with tobacco use disorder (TUD). In men, three known WHRadjBMI loci (ADAMTS9, DNM3, ZNRF3) had a greater effect among smokers vs nonsmokers (ZNRF3: β(SE) = 0.0365 (0.0148), p = 1.37x10-2) vs β(SE) = 0.0192 (0.0088), p = 12.87x10-2). Eight of our 19 top SNPs have not been previously associated with WHRadjBMI (within 500 kb). Of these eight loci, distinct patterns of association by smoking and sex were again noted. In addition, one locus (ABCA1) has been previously associated with TUD. In addition to identifying novel loci associated with body fat distribution, these findings underscore the importance of considering the effect of smoking when investigating the genetics of obesity related phenotypes and emphasize the need to consider sex separately when studying the effects of smoking on body fat distribution.