Genetic architecture of four smoking behaviors using partitioned h2SNP.

Abstract

Background and AimsSmoking is a leading cause of premature death. Although genome-wide association studies have identified many loci that influence smoking behaviors, much of the genetic variance in these traits remains unexplained. We sought to characterize the genetic architecture of four smoking behaviors through SNP-based heritability (h2SNP) analyses. DesignWe applied recently-developed partitioned h2SNP approaches to smoking behavior traits assessed in the UK Biobank. SettingUK Biobank. ParticipantsUK Biobank participants of European ancestry. The number of participants varied depending on the trait, from 54,792 to 323,068. MeasurementsSmoking initiation, age of initiation, cigarettes per day (CPD; count, log-transformed, binned, and dichotomized into heavy versus light), and smoking cessation. Imputed genome-wide SNPs. FindingsWe estimated h2SNP(SE)=0.18(0.01) for smoking initiation and 0.12(0.02) for smoking cessation, which were more than twice the previously reported estimates. Estimated age of initiation h2SNP=0.05(0.01) and binned CPD h2SNP=0.1(0.01) were similar to previous reports. These estimates remained substantially below published twin-based h2 of roughly 50%. CPD encoding strongly influenced estimates, with dichotomized CPD h2SNP=0.28. We found significant contributions of low-frequency variants and variants in low linkage-disequilibrium (LD) with surrounding genomic regions. Functional annotations related to LD, allele frequency, sequence conservation, and selective constraint also contributed significantly to the partitioned heritability. We found no evidence of dominance genetic variance for any trait. Conclusionh2SNP of these four specific smoking behaviors is modest overall. The patterns of partitioned h2SNP for these highly polygenic traits is consistent with negative selection. We found a predominant contribution of common variants, and our results suggest a role of low-frequency or rare variants, poorly tagged by surrounding regions. Deep sequencing of large samples and/or improved imputation will be required to fully assess the role of rare variants.