Recently, British and Swedish researchers published what they call the first genome-wide association study for warfarin dosing with sufficient power to detect variants of modest effect. In 1,053 Swedish patients, they confirmed the well-known VKORC1 and CYP2C9 SNPs and uncovered a third SNP, CYP4F2, which they then corroborated in an additional cohort of 588 Swedish patients. GT's Ciara Curtin caught up with the Sanger Institute's Panos Deloukas to discuss the implications of his study and what future studies may determine. The team's paper was published in PLoS Genetics.
Genome Technology: Do you think that your study has identified all the common variants involved in warfarin response?
Panos Deloukas: In terms of common variants with major effects, I think it's fair to say that we have probably done a comprehensive search based on SNPs. That's not the case for copy number variants, which the array we used did not cover. We are probably only missing lower-frequency variants with minor or smaller effects. What is not covered at all is if there is a combination of rare variants that are contributing to the phenotype. In terms of common variations, I think we have exhausted that side.
GT: What role do you think rare variants will play in warfarin dosing, and how do you think they will be identified if people go after them?
PD: In terms of identifying further genetic variants, one will need to do a comprehensive search for structural variants. For the more rare variants, I think that the whole field is thinking that some of the new technologies in sequencing [are] the way forward.
What is, however, I think much more amenable [and] will be the first to be tried in terms of sequencing are the more rare kind of adverse events. As you know, warfarin is linked to bleeding, which is an adverse reaction. About 2 percent of patients develop this bleeding and I think [it makes] a lot of sense to go straight ahead with the sequencing of the samples, whether you do it in pools and compare pools of individuals that have the adverse reactions versus a pool of samples of treated patients that do not have it, [or not]. I think for this one, sequencing is the way forward.
GT: Do you think all warfarin patients should be genotyped, or just those with other risk factors?
PD: From what has emerged, if you combine the genetic factors that we and others have identified with non-genetic factors such as age, gender, and body weight, you can start [to get] a pretty good prediction of the dose required. I think what is also very important, but you need to set up the appropriate studies for what will be the best test to inform, is what the loading dose for an individual [should be]. This is a little bit different and more work will be needed there. I think genotyping combined with these non-genetic factors is a very useful tool to predict dose.
GT: Many of the variants uncovered were identified in people of Caucasian descent. Do you think that the genes you and others identified will have different effects, or that other genes will be involved, in people of other ethnicities?
PD: I think it's probably already known that the frequencies of the variants that are important in Caucasians, that they differ quite a lot in other ethnicities, especially in Asians, but also in African people. It will be essential that we get data from these populations. There is already an effort underway to do a genome scan in African-Americans. This [data] will need to be combined. I don't want to speculate whether there will be additional genes or not, but it cannot be excluded. It may be much more likely that we find different variants in the same gene being important.