Abstract Cancer genetics has moved into a new era of discovering germ-line genetic variants that influence risk for specific cancers as a consequence of large-scale annotation of common genetic variation across the human genome. With the advent of fixed genotyping platforms and the investment in well-designed molecular epidemiology studies, the focus has shifted away from candidate gene studies towards large scale GWAS. In the process, we have seen clear evidence that substantial replication efforts are needed to conclusively identify new signals associated with cancer etiology. These studies seek to discover markers for disease, but rarely directly identify the variants biologically underlying the observed effect. Consequently, GWAS have primarily identified surrogate markers that necessitate follow-up fine mapping and laboratory analyses. So far, the published literature in cancer has reported approximately 100 novel regions of the genome associated with cancer etiology. GWAS have also demonstrated that the spectrum of cancers could have distinct genomic architectures, which reflects differential contributions of alleles with a range of frequencies and effect sizes. For nearly all regions, the estimated per allele effect sizes are less than 1.3 and the susceptibility alleles are common, with minor allele frequencies greater than 5%. The notable exception is for testicular cancer, which has a very strong tendency towards familial aggregation. In the GWAS of testicular cancer, common genetic markers on chromosome 12q21.32 were found to have per allele estimated risks of greater than 2.5; there is a plausible candidate gene in this region, KITLG. Still, because each common allele appears to confer a small effect, we are beginning to better appreciate the complex nature of common genetic variants that contribute to primary carcinogenesis. Furthermore, it is unlikely that there are many other common alleles with large effects for the common diseases, such as breast, colorectal, lung or prostate cancer because a sufficiently large enough number of cases and controls have been scanned. It is notable that the distribution of the number of regions identified by GWAS varies greatly by cancers, and does not appear to closely mimic hereditary estimates based on twins or family aggregation. For instance, GWAS have identified nearly 30 regions in the genome associated with prostate cancer, all of which are associated with early and aggressive cancers. On the other hand, for lung cancer, only three have been identified and one of them on chromosome 15q24/25 is also strongly associated with smoking behavior phenotypes. These observations raise the likelihood that for some cancers, a substantial environmental effect, such as smoking and lung cancer inhibits the ability to discover low penetrance, common alleles for these diseases. Because GWAS have been conducted in more than a dozen cancers, it is interesting to note that multiple cancers have mapped to two distinct regions, 8q24 and 5p15.33. The former, a ~600MB region centromeric to the MYC gene harbors a series of independent markers associated with breast, colorectal, prostate and urinary bladder cancer as well as chronic lymphocytic leukemia. Similarly, an unexpected spectrum of GWAS have pointed to common variants in a region of chromosome 5p15.33, which harbors the TERT-CLPTM1L locus; these include lung cancer (specifically adenocarcinoma), brain tumors, skin cancer (melanoma and basal cell carcinoma) and pancreatic cancer. This region also harbors less common mutations linked to acute myelogenous leukemia, dyskeratosis congenital (an inherited bone marrow failure syndrome) and pulmonary fibrosis. For each of the regions conclusively identified by GWAS, a series of follow-up studies are needed to fine map the most promising variants, preferably in different populations that can point investigators towards variants for studies designed to investigate plausibility for the association. Subsequently, laboratory investigation will be required to investigate each region that can lead to new insights into pathways and mechanisms underlying the association. To date, the reported GWAS have concentrated on regions associated with cancer etiology and not survival or other clinical outcomes. These suggests that common genetic variation may influence distinct processes important for carcinogenesis as opposed to those associated with disease progression. Though there has been considerable discussion of the promise of common variants as markers for disease risk, it is too early to apply the common variants detected in cancer GWAS to clinical and public health venues. Further studies are needed to map the markers and assess the relationship of multiple markers in sufficiently large studies. In addition, the analytical approach should account for the exposures to environmental and lifestyle choices that interact with germ-line genetic variants. The success of GWAS has opened new horizons for exploration and highlighted the complex genomic architecture of disease susceptibility. Citation Format: Stephen J. Chanock. Genome-wide association studies in cancer: What have we found and what next [abstract]. In: Proceedings of the AACR 101st Annual Meeting 2010; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr PL02-05
Read full abstract