You might not be reading about Constantin Polychronakos if it weren’t for two major breakthroughs. One of the lead scientists on a type 2 diabetes gene-exploration project, he recently published results that got the community interested.
Polychronakos, a clinical physician who has spent the past 16 years studying diabetes, says that until recently the only way to tackle diabetes was with the usual candidate gene approach: look at large chromosomal regions known to be linked to diabetes in some way, and then try to home in on the particular genes actually implicated in the disease. “People have been doing candidate genes for the last at least 10 years, probably 20,” says Polychronakos. “All we’ve come up with are two good genes.” The dilemma was obvious: scientists and clinicians could only target suspected regions, leaving them fully aware — but unable to avoid this fact — that they were by definition missing any critical regions that hadn’t popped up in linkage studies already.
The problem, of course, was cost. Genome-wide scans were cost-prohibitive, and no other approach seemed viable. It was at this point that Polychronakos and his colleagues — a group of scientists spread across Canada, Britain, and France — submitted a funding proposal for yet another candidate gene project, and it was shortly after that that the two breakthroughs fell into his lap. One was the completion of the HapMap project, which gave researchers the ability to test just several hundred thousand tag SNPs to have a clear picture of what was going on with several million polymorphisms. Next, Polychronakos says, was the advance that microarray vendors figured out how to put enormous numbers of genotyping assays on a chip. Grant in hand, Polychronakos and his crew wound up going back to the funding agency and asking if they could switch from a candidate gene to a genome-wide approach; when that was approved, they opted to go with the Illumina Golden Gate technology for their genotyping work.
To fit their budget, Polychronakos and his team decided on a two-stage design for the project. First they’d study a group of 700 patients and 700 controls with extensive exploratory genotyping. “That’s what we could afford to genotype,” says Polychronakos, who would’ve been happy to genotype all 7,000 diabetes samples the team had access to if there had been money for it. “We compensated for this by choosing the phenotype very well,” he says. Diabetes has an especially messy phenotype with a number of factors. Polychronakos says his team decided to eliminate obesity as a factor, and wound up choosing factors that proved to be more highly correlative instead.
The first stage was the basis for the paper the team published earlier this year in Nature with lead author Robert Sladek and senior author Philippe Froguel. The initial study turned up a number of genes suspected to play a role in diabetes, including five genes that seem especially promising. Polychronakos says the paper represents “proof of principle and our best fast-tracked hits.”
But that’s certainly not the end of the story. Polychronakos says the group is already working on stage two of the project: taking the 5,000 most relevant and predictive SNPs turned up in the first round and testing those in a new cohort of diabetes patients. So far, Polychronakos says, the results are showing great risk association, which gives him hope that particular SNPs could be used to diagnose high-risk people early in life. “Type 2 diabetes is a preventable disease,” he says. This work could help “improve prevention. That’s the most immediate goal.”
Beyond predictive power, he notes, the study could turn up gene loci that are well suited for targeting with therapeutics — though that will be much further in the future, he believes.