The results of analyses of all included SNPs

The results of analyses of all included SNPs on CPD in the WHI SHARe sample are presented in Table 2. In the models with CPD as their outcome in the WHI SHARe cohort, 1 SNP was found to be associated with CPD following Bonferroni adjustment), rs1051730 (adjusted p=0.027). For this SNP, each additional A allele increased subjects\' expected number of CPD by 0.081 cigarettes. In addition, 10 other SNPs were nominally significant for association with CPD but not statistically significant following Bonferroni correction. Given the low number of lung cancer cases in the WHI sample (n=86), we sought to conduct interaction analyses for the top 10 nominal SNPs for CPD from the WHI SHARe Lomustine in the multicenter lung cancer case–control study population described above. Genotype data were available for nine of these SNPs, with no data available for rs547843. The results for analyses of main effects of nominally significant WHI SHARe CPD SNPs on lung cancer in the multicenter case–control study are presented in Table 3. All nine SNPs were statistically significantly associated with lung cancer (Bonferroni-corrected P-values from 0.027 to 6.09×10). There were six SNPs that demonstrated nominally significant interactions with CPD for risk of incident lung cancer (Table 4), two of which approached statistical significance after Bonferroni correction: rs2036527[A], beta=−0.0171, p=0.0549; rs7180002, beta=−0.0205, p=0.0576. Fig. 3 illustrates the allele dose response relationships for each of these SNPs with incident lung cancer as estimated in the interaction models. The nature of each of these interactions was notable for a pattern suggesting – but not establishing, given that the interactions were not statistically significant after Bonferroni correction – stronger allele dose-responses for individuals who smoked fewer CPD. Odds ratios for lung cancer risk by average CPD are presented in Supplementary Materials.
Discussion To our knowledge, there have been at least four fine-mapping case control studies of lung cancer in African Americans that have examined the chromosome 15q25.1 locus and additional loci on chromosomes 5p15.33 and 6p22.1–21.31 (Walsh et al., 2013; Amos et al., 2010; Spitz et al., 2013; Hansen et al., 2010). One study of 1058 cases and 1314 controls from the Detroit area Surveillance, Epidemiology, and End Results (SEER) registry found that SNP rs1051730 on chromosome 15q25.1 was associated with lung cancer in African-American ever-smokers (Schwartz et al., 2009) — findings similar to a larger study of European-ancestry cases (n=1024) and controls (n=32,244) (Thorgeirsson et al., 2008). Another study identified multiple SNPs and a haplotype within the chromosome 15q25.1 region with lung cancer in 448 African-American lung cancer cases and 611 controls, which suggests that SNPs in this region affecting expression of the alpha 5 (CHRNA5), alpha 3 (CHRNA3) and beta 4 (CHRNB4) nicotinic acetylcholine receptor genes may be independently associated with lung cancer (Amos et al., 2010; Hansen et al., 2010). In a previously published meta-analysis using data from the WHI SHARe cohort (n=8208) and twelve other study groups forming the Study of Tobacco Use in Minority Populations (STOMP) Genetics Consortium (N=32,829 (David et al., 2012), rs2036527 was associated with CPD in African-Americans (David et al., 2012), which has subsequently been shown to portend increased potential benefit from smoking cessation treatment in treatment-seeking African-Americans smokers (Zhu et al., 2014). The s2036527 SNP is located in the 5′ distal enhancer region of the CHRNA5 gene, which forms a haplotype associated with increased CHRNA5 expression (Smith et al., 2011) and which has been associated with lung cancer in three fine-mapping studies of African Americans (Walsh et al., 2013; Amos et al., 2010; Hansen et al., 2010). However, mechanisms conferring increased risk for any of the highly correlated SNPs in this region are complex and may involve epigenetic effects on the CHRNB4 and TERT genes (Scherf et al., 2013) and decreased apoptosis in tumor cells resulting from over-expression of the PSMA4 gene, which has been suggested for rs2036527 (Liu et al., 2009).