There are a lot of loci floating around that have been shown to be associated with certain diseases, but whether or not those loci are merely linked to the disease or are actually causal is often unknown. Through genome-wide association studies, about 10 regions have thus far been associated with colorectal cancer.
At the Ontario Institute for Cancer Research, John McPherson and his colleagues are following up on those regions identified by the Assessment of Risk for Colorectal Tumors in Canada, or ARCTIC, project and family studies. The ARCTIC effort uncovered a locus in chromosomal region 8q24 associated with colorectal cancer. The locus was first found in a group of people from Ontario and confirmed in groups from Newfoundland, Scotland, and Seattle. Around the same time, a British group with researchers from the Institute of Cancer Research and Cancer Research UK identified another region, called SMAD7, with links to colorectal cancer. Of the 10 regions identified, McPherson says only the changes in SMAD7 appear to be associated with a protein. "But the rest of them, it's just an association region," he says. "So what we wanted to do was do some deep resequencing on those regions."
For that, they rounded up 120 people — 40 controls, 40 sporadic cases, and 40 probands. In addition to resequencing the 8q24 region from the ARCTIC study, McPherson and his colleagues looked at the sites implicated by familial and other studies — a little more than 3 megabases of sequence in total. They used a custom-designed array from NimbleGen that covered the regions under study and that generated more than 65 gigabases of paired-end reads.
McPherson is currently mining the resequencing data to find SNPs. So far, he's uncovered nearly 10,000 SNPs, half of which are novel. "The whole goal of this was to identify a new set of variants within the region, and then go back," he says. A few of the novel SNPs uncovered so far have caught the researchers' eyes: About 90 of them are in exons and, of those, about 50 affect protein structure; a handful even contain stop codons. However, McPherson cautions that it is still early days. "You just can't make a claim on that small a dataset at this point," he says. "It's just not a big enough population base to say for sure what you've discovered."
Although they have only resequenced 120 people, McPherson says they have a pool of 1,200 that they can tap into. "We can take the SNPs and put them through another 700, 800 people," he says. McPherson adds that they don't plan to resequence that larger group; rather, they want to see if the new variants are present in those populations. If a region of interest shows up, there's then a much stronger association, and perhaps a case for causality.
Currently, McPherson says, the odds ratios for any of the colorectal cancer markers are fairly low, as is typical of most loci from genome-wide association studies. "You find some associated alleles and it gives you an odds ratio of 1.2 or something," he says. "Maybe in conjunction with family history they are predictive, but alone they are probably not better than just family history at this point."
Eventually, McPherson hopes that resequencing and more follow-up studies will give a better idea of causality. "Having the actual causative alleles or knowing for sure that this gene is involved, I think that we can look for. It may not be a single causative SNP you'd test, but you could resequence the gene and get more definitive answers," he says.
As the answers are gathered, researchers such as McPherson still learn about basic underlying mechanisms of cancer, which could also prove fruitful. "That [knowledge] would hopefully lead down the road to more, better diagnostics or even treatment," McPherson says.