Skip to main content
Premium Trial:

Request an Annual Quote

Multi-institutional GWAS IDs Variant to Improve Warfarin Dosing in People of African Descent


A genome-wide association study by researchers at several US and international institutions has found a novel variant — rs12777823 — that explains some of the variability in warfarin metabolism in African-American individuals and could potentially improve the effectiveness of pharmacogenomic dosing algorithms in this population.

The team published the finding, from a discovery study of 533 subjects and a replication study that included an additional 432 participants, last week in The Lancet.

According to the authors, previously identified variants in the VKORC1 and CYP2C9 genes explain up to 30 percent of the total variability in warfarin dose requirements among people of European origin, and slightly less in people of Asian origin. But in people of African descent, these markers have proved much less predictive.

VKORC1 and CYP2C9 genotypes are also the backbone of existing genetic dosing algorithms developed to help doctors more accurately estimate the correct starting dose for patients. For these algorithms to be useful in African-Americans, additional variants that predict dose requirements for those with African ancestry are required.

Julie Johnson of the Center for Pharmacogenomics at the University of Florida led the recent effort published in The Lancet. She told PGx Reporter there was a push within the International Warfarin Pharmacogenetics Consortium to make improvements for this population after the group published its warfarin dosing algorithm.

"There was a feeling that we were missing something, and that the best way to figure that out was to do a genome-wide association study," she said.

In the study researchers analyzed a total of 533 patients with self-reported African ancestry from two separate cohorts for the discovery phase of the study: one from the University of Alabama and one from various sites involved in the IWPC.

An additional 432 patients from the IWPC who were not recruited for the discovery cohort were then used to validate the predictive power of rs12777823.

According to Johnson, the two groups were initially doing separate studies but decided to combine efforts to give the study greater power. "We realized we'd have stronger findings if we put the data together," she said.

The team genotyped study participants from the UAB and IWPC cohorts and used stepwise conditional analysis to look for SNPs that could associated with higher or lower stable warfarin doses.

The most significant association, apart from VKORC1, was the SNP rs12777823, whose minor allele frequency was 25 percent in the discovery cohort, the researchers reported.

In the replication cohort, rs12777823 remained "significantly associated" with stable warfarin dose. While the group also identified 12 other SNPs in the discovery set that were "of interest," none of these remained associated with warfarin dose in the replication group.

Overall, the researchers found that the VKORC1 genotype remained the primary genetic determinant of dose requirement variability. Neither CYP2C9*2 nor CYP2C9*3 were significantly associated with dose variability, potentially due to their low frequency in the population. Only CYP2C9*11 showed a significant link in addition to rs12777823.

The group reported that rs12777823 appears to explain about 5 percent of the variability in warfarin dose requirements among those of African ancestry— about as much as CYP2C9*2 and CYP2C9*3 explain in individuals of European ancestry, and much more than the 1 to 2 percent those genotypes explain in African-Americans.

Age, weight, height, aspirin use, and amiodarone use were all also associated with different warfarin dose requirements in the group's analysis.

The researchers used modeling to demonstrate that adding rs12777823 to the IWPC's current warfarin dosing algorithm improved its ability to predict the correct therapeutic dose by a relative 21 percent.

According to the authors, the study results suggest that rs12777823 could be used to inform dose reductions in African-Americans, to improve on the ability of the IWPC's main algorithm in that population.

Johnson said that some members of the group now plan to integrate the SNP into clinical warfarin PGx dosing schemes. "At the University of Illinois, Chicago, they have an active warfarin pharmacogenetics program and they plan on adding it," she said.

She also said that Brian Gage, who was not involved in the study, but who manages the website — which incorporates both the IWPC algorithm and another algorithm he developed — plans to add information about rs12777823 to the site.

With a 25 percent minor allele frequency in the study cohorts, rs12777823 appears to be fairly common in African-Americans, according to the study authors.

"This [SNP] has a large enough effect that one, we were able to see and replicate it, and two, people recognize that in patients who carry that variant —which is a fair proportion — it's going to make a difference in how you dose them," Johnson said.

Interestingly, the SNP also is similarly common in European and Japanese populations, but without any significant association with warfarin dose according to previous GWAS, the researchers wrote in the study.

This may mean that the association between the SNP and warfarin sensitivity in African-Americans might not be due to rs12777823 itself but rather due to another variant in linkage disequilibrium with it in this population but not in others.

"Basically, this SNP is common in all the populations we've looked at," Johnson explained. "But we did not see an association with warfarin dose in non-Africans. What that implied to us is that this is not the functional variant, but this particular SNP is tagging some other functional variant in African ancestry individuals but it's not in the other two groups."

According to Johnson, some members of the study are planning to investigate this further by sequencing the region around the SNP.

"Work needs to be done to localize what the functional variant is, and it's also possible it's tagging a bunch of rare variants, in which case it might actually be unique to African-Americans," Johnson said.

Additionally, she said there is interest among some study authors in looking at how this SNP influences warfarin metabolism at the molecular level.

"We believe that this is affecting metabolism, and we think it's probably affecting metabolism via CYP2C9, but it's also possible that, because there are a bunch of CYP2C metabolism genes in a cluster there, the [functional variant] could be influencing that whole cluster."

Figuring that out could have implications for predicting metabolism and individualizing therapy for other drugs, Johnson said.

Also, identifying additional variants specific to African-Americans is an important goal and would undoubtedly improve dose prediction even further, the authors wrote.

There has been some debate over whether pharmacogenomic dosing algorithms — regardless of whether they perform well for African-Americans or not — are worth the additional cost they might add to clinical care.

Johnson said she believes this issue is temporary. "People don’t really dispute that you can probably use genetic information to start patients on a dose that is closer to their actual dose," she said. "But despite that, some people argue over whether that really matters; in other words, 'is the cost of the genotyping offset by its clinical benefit?'"

"But as we move to a point where genetic information is more available, we would be crazy not to use it."

Additionally, she noted that several trials expected to be completed this year may help demonstrate the current cost-benefit of PGx warfarin dosing, including the Clarification of Optimal Anticoagulation through Genetics, or COAG trial, which Johnson said is expected to publish results by this fall, as well as a European trial also expected to share data by the end of the year.

In another trial funded by the Centers for Medicare & Medicaid Services, Iverson Genetics is measuring whether more rapid genetic testing, with results returned within 24 hours, reduces the rate of adverse events more than when the drug is dosed without genetic testing data (PGx 3/9/2011).

"There have already been [studies] suggesting that there probably are substantial benefits [to PGx dosing], but not a randomized controlled trial," Johnson said.

"But if these next trials are positive, then the use of warfarin pharmacogenetic testing is probably going to go up substantially, in which case this new finding would allow us to do better in African-Americans."