ABSTRACT: Background Lung cancer is considered an archetypical environmentally induced disease because of the high risk from exposure to tobacco smoke. However, family studies clearly identified familial aggregation, beyond effects from familial correlations in smoking behavior. Family studies and genome wide association studies have identified selected variants influencing lung cancer risk but have been underpowered to provide a more comprehensive assessment of genetic architecture. Method We conducted a genome-wide association analysis using the Oncoarray, a genotyping platform that comprises over 530,000 genetic markers and designed to query variation influencing cancer risk. Here, we present the largest genome-wide scan of lung cancer susceptibility, in European-descent populations, comprising data derived from 29,863 patients and 55,586 controls queried for SNPs imputed from the 1000 Genomes Project. Analyses were adjusted for age, sex and the first three principal components for overall lung cancer and stratified by histology (adenocarcinoma, squamous carcinoma, and small cell carcinoma) and ever or never smoking. Results Results identified 24 loci influencing lung cancer risk for loci with minor allele frequencies of 0.5% or higher, of which 14 had not previously reached genome-wide significance levels. Odds ratio effect sizes ranged from 0.37 for the functional variants rs17879961 in CHEK2 to 2.12 for rs11571833 in BRCA2, [JF1] indicating the role of selected uncommon variants in strongly influencing lung cancer risk. Among the novel variants identified, 6 influenced overall lung cancer risk, 6 were specific to adenocarcinoma and 1 each [JF2] influenced never and ever smokers respectively. Aside from previously described variants in the CHRNA5 and CYP2A6 regions that influence cigarette consumption, all other loci showed significant heterogeneity among histologies. Array based heritability analysis also indicates no significant shared heritability between adenocarcinoma and squamous carcinoma, again pointing to striking etiological heterogeneity, other than that due to known smoking behavior related loci between these forms of lung cancer. Evaluation of eQTL results derived from normal lung tissue identified consistent cis-acting effects of the variant rs77468143 influencing expression of a little know gene SECISBP2L, rs6920364 influencing the extracellular ribonuclease RNASET2, and rs146729428 in the EPHX2, involved in inflammation. Several loci that influenced adenocarcinoma alone influenced telomere maintenance or cell cycle, while several of the squamous cancer-specific loci are involved in recombination repair. Conclusion These results indicate striking variation among histological subtypes of lung cancer. Further studies are ongoing to identify the impact that smoking has on lung cancer risk for these loci. These analyses provide a comprehensive assessment of genetic effects on lung cancer risk and will elucidate how these interact with smoking behavior. We would like to extend this study to include participants from Latin American countries because these were not yet well represented by our analyses.
