of the variance in quantitative measures of smoking behavior. In a candidate gene approach, Saccone et al. [34] confirmed a well-recognized association between a non-synonymous, functional SNP (rs16969968) in chr 15q25 and nicotine dependence in both European Americans and African Americans in a case-control study of confirmed smokers, but other markers in this region showed more complex statistical evidence of association, as well as greater heterogeneity between racial groups. Gelernter et al. [5] looked at nicotine dependence in a GWAS using 2,114 European American and 2,602 African American smokers, although most genome-wide significant SNPs were only supported in one racial group. In part, this difference in evidence of association between smoking behavior phenotypes in GWAS or candidate gene studies may reflect background differences in linkage disequilibrium (LD) patterns between people of European and African ancestry, but also differences in marker allele frequencies and differences in sample sizes become important (where studies of African Americans are typically smaller). Hamidovic et al. [35] showed a quantitative assessment of smoking persistence (based on pack-years) also revealed different patterns of association with SNPs between European and African ancestry subjects from the ARIC study cohort. However, in an analysis of the MESA Lung cohort, Powell
