Homozygosity for the G allele of rs6983267 at 8q24 increases colorectal cancer (CRC) risk approximately 1.5 fold. We report here that the risk allele G shows copy number increase during CRC development. Our computer algorithm, Enhancer Element Locator (EEL), identified an enhancer element that contains rs6983267. The element drove expression of a reporter gene in a pattern that is consistent with regulation by the key CRC pathway Wnt. rs6983267 affects a binding site for the Wnt-regulated transcription factor TCF4, with the risk allele G showing stronger binding in vitro and in vivo. Genome-wide ChIP assay revealed the element as the strongest TCF4 binding site within 1 Mb of MYC. An unambiguous correlation between rs6983267 genotype and MYC expression was not detected, and additional work is required to scrutinize all possible targets of the enhancer. Our work provides evidence that the common CRC predisposition associated with 8q24 arises from enhanced responsiveness to Wnt signaling.

Cohesin is present in almost all active enhancer regions, where it is associated with transcription factors. Cohesin frequently colocalizes with CTCF (CCCTC-binding factor), affecting genomic stability, expression and epigenetic homeostasis. Cohesin subunits are mutated in cancer, but CTCF/cohesin-binding sites (CBSs) in DNA have not been examined for mutations. Here we report frequent mutations at CBSs in cancers displaying a mutational signature where mutations in A•T base pairs predominate. Integration of whole-genome sequencing data from 213 colorectal cancer (CRC) samples and chromatin immunoprecipitation sequencing (ChIP-exo) data identified frequent point mutations at CBSs. In contrast, CRCs showing an ultramutator phenotype caused by defects in the exonuclease domain of DNA polymerase ɛ (POLE) displayed significantly fewer mutations at and adjacent to CBSs. Analysis of public data showed that multiple cancer types accumulate CBS mutations. CBSs are a major mutational hotspot in the noncoding cancer genome.

To identify colorectal cancer (CRC) susceptibility alleles, we conducted a genome-wide association study. In phase 1, we genotyped 550,163 tagSNPs in 940 familial colorectal tumor cases (627 CRC, 313 high-risk adenoma) and 965 controls. In phase 2, we genotyped 42,708 selected SNPs in 2,873 CRC cases and 2,871 controls. In phase 3, we evaluated 11 SNPs showing association at P < 10(-4) in a joint analysis of phases 1 and 2 in 4,287 CRC cases and 3,743 controls. Two SNPs were taken forward to phase 4 genotyping (10,731 CRC cases and 10,961 controls from eight centers). In addition to the previously reported 8q24, 15q13 and 18q21 CRC risk loci, we identified two previously unreported associations: rs10795668, located at 10p14 (P = 2.5 x 10(-13) overall; P = 6.9 x 10(-12) replication), and rs16892766, at 8q23.3 (P = 3.3 x 10(-18) overall; P = 9.6 x 10(-17) replication), which tags a plausible causative gene, EIF3H. These data provide further evidence for the 'common-disease common-variant' model of CRC predisposition.

Genome-wide association (GWA) studies have thus far identified 10 loci at which common variants influence the risk of developing colorectal cancer (CRC). To enhance power to identify additional loci, we conducted a meta-analysis of three GWA studies from the UK totalling 3,334 cases and 4,628 controls, followed by multiple validation analyses, involving a total of 18,095 CRC cases and 20,197 controls. We identified new associations at 4 CRC risk loci: 1q41 (rs6691170, OR=1.06, P=9.55x10-10; rs6687758, OR=1.09, P=2.27x10-9); 3q26.2 (rs10936599, OR=0.93, P=3.39x10-8); 12q13.13 (rs11169552, OR=0.92, P=1.89x10-10; rs7136702, OR=1.06, P=4.02=x10-8); and 20q13.33 (rs4925386, OR=0.93, P=1.89x10-10). As well as identifying multiple new CRC risk loci this analysis provides evidence that additional CRC-associated variants of similar effect size remain to be discovered.

Genome-wide association (GWA) studies have identified multiple loci at which common variants modestly influence the risk of developing colorectal cancer (CRC). To enhance power to identify additional loci with similar effect sizes, we conducted a meta-analysis of two GWA studies, comprising 13,315 individuals genotyped for 38,710 common tagging SNPs. We undertook replication testing in up to eight independent case-control series comprising 27,418 subjects. We identified four previously unreported CRC risk loci at 14q22.2 (rs4444235, BMP4; P = 8.1 × 10 −10 ), 16q22.1 (rs9929218, CDH1; P = 1.2 × 10 −8 ), 19q13.1 (rs10411210, RHPN2; P = 4.6 × 10 −9 ) and 20p12.3 (rs961253; P = 2.0 × 10 −10 ). These findings underscore the value of large sample series for discovery and follow-up of genetic variants contributing to the etiology of CRC.Whereas inherited susceptibility is responsible for ~35% of all CRC 1 , high-risk germline mutations in APC, the mismatch repair (MMR) genes, MUTYH (MYH), SMAD4, BMPR1A and STK11/LKB1 account for <6% of all cases 2 . Recent GWA studies have validated the hypothesis that part of the heritable risk is caused by common, low-risk variants, identifying CRC susceptibility loci mapping to 8q24 (rs6983267) 3, 4, 8q23.3 (rs16892766, EIF3H)5, 10p14 (rs10795668)5, 11q23 (rs3802842)6, 15q13 (rs4779584)7 and 18q21 (rs4939827, SMAD7) 6,8 .GWA studies are not contingent on prior information concerning candidate genes or pathways, and thereby have the ability to identify important variants in hitherto unstudied genes. However, the effect sizes of individual variants, the need for stringent thresholds for establishing statistical significance, and financial constraints on numbers of variants that can be followed up inevitably constrain study power. We recently published two separate GWA studies for CRC. To augment the power to detect additional CRC risk loci, we have conducted a meta-analysis of data from these studies and followed up the best supported associations in large sample sets. This analysis, in conjunction with a replication study using eight independent case-control series, has enabled us to identify four new loci predisposing to CRC. This brings to ten the number of independent loci conclusively associated with CRC risk, and provides additional insight into the genetic architecture of inherited susceptibility to CRC. RESULTS Meta-analysis of genome-wide association scansThe GWA studies were both conducted by centers in London and Edinburgh, and were both based on designs involving two-phase strategies and using samples from UK populations NIH Public Access Author ManuscriptNat Genet. Author manuscript; available in PMC 2010 March 11. Published in final edited form as:Nat Genet. London phase 2 was based on genotyping 2,873 CRC cases and 2,871 controls ascertained through the National Study of Colorectal Cancer Genetics (NSCCG), whereas Edinburgh phase 2 was based on genotyping 2,057 cases and 2,111 controls. For phase 2, the London and Edinburgh samples were genotyped for a common s...