Analyses in single cohorts were adjusted for age, % of predicted FEV1 and principal components summarizing genetic ancestry. In total, meta-analysis included 1,164 current smokers (defined as “No” for smoking cessation) and 1,907 former smokers (defined as “Yes” for smoking cessation). Lambda inflation factors were between 0.995 and 1.000 for individual cohorts (table 1), and 0.998 for meta-analytic p values (see online supplementary figure 4 for a Q-Q plot of meta-analysis). No SNPs showed association with meta-analytic p<10−6 (table 3). Candidate SNP rs3025343 in the DBH locus showed nominally significant association with smoking cessation with the same direction of effect as seen in recent GWAS, yet with substantial heterogeneity between studies (table 3). SNP rs3025316, a proxy for rs3025343 (r2=0.94), was genotyped in all cohorts and confirmed this association (p= 0.002 (OR= 1.32 for the “T” effect allele; I2=18).
