Clinical and Economic Consequences of Pharmacogenetic-guided Dosing of Warfarin

Talitha I Verhoef; Tom Schalekamp; William K Redekop; Anthonius de Boer; Anke-Hilse Maitland-van der Zee


Expert Rev Pharmacoeconomics Outcomes Res. 2010;10(4):375-378. 

In This Article

Summary of Methods & Results

In their study Meckley et al. developed a decision analytic Markov model to perform an economic evaluation comparing genotype-guided dosing with standard anticoagulation care using a lifetime horizon.[1] The base-case scenario focused on 65-year-old patients with atrial fibrillation who were initiated on long-term treatment with warfarin. Patients were stratified by genotype into three different groups: the first group consisted of CYP2C9 wild-type/VKORC1 wild-type patients, the second of CYP2C9 wild-type/VKORC1 variant patients, and in the last group were the CYP2C9 variants. This last group was not stratified by VKORC1 genotype for group size reasons. All patients entered the Markov model in a healthy ('well') state and could move from this state to the states 'clot', 'bleed', 'sequelae' or 'death' in monthly cycles. The probabilities to experience a major bleeding or thromboembolic event were based on the time spent within, above or below therapeutic INR range. The time patients spent within, above and below therapeutic INR range was based on data from the COUMAGEN trial.[6] This trial provided data on the difference in time spent in therapeutic INR range between the genotype-guided dosing group and the standard dosing group, but the authors reanalyzed the COUMAGEN data in order to obtain additional information regarding the time spent above or below this range. The differences in time spent within the different INR ranges between the two dosing strategies were used in the model for the first month and reduced to zero in the sixth month of therapy. An increased bleeding risk of 2.26, independent of the effect of INR, was assumed from a meta-analysis for the CYP2C9 variant patients, which was subjected to sensitivity analysis.[7]

Utility (or quality-of-life) scores for the different health states were used to calculate the difference in quality-adjusted life-years (QALYs) between the standard and the genotype-guided dosing group. Genotyping itself was assumed to have no effect on the quality of life of the patient. The difference in costs between the two strategies was calculated and included only direct medical costs, since the authors applied a third party payer perspective. One-way sensitivity analyses were performed for all input parameters over prespecified ranges and several scenario analyses were conducted. The chance that the incremental cost–effectiveness ratio (ICER) would fall below a certain willingness-to-pay threshold (e.g., US$50,000 per QALY gained) was calculated in a probabilistic sensitivity analysis using Monte Carlo simulations.

Meckley et al. found that pharmacogenetics could reduce the time spent above therapeutic INR range in the CYP2C9 variant group by 15%, and the time spent below therapeutic range in the CYP2C9 wild-type/VKORC1 wild-type group by 8%.[1] In the third group (CYP2C9 wild-type/VKORC1 variants) there were no differences in time spent above or below therapeutic INR range between the two dosing strategies. In the base case analysis the incidence of bleedings was reduced by 0.17%, the incidence of thromboembolic events increased by 0.03% and incidence of death reduced by 0.13% in the pharmacogenetic-guided dosing group. These differences resulted in a QALY increase of 0.0027. As genotyping also led to an overall cost increase of US$162, the ICER was US$60,725 per QALY gained. When looking at the ICERs in the different genotype groups, pharmacogenetic-guided dosing was most cost effective in the group consisting of VKORC1 and CYP2C9 wild-type patients. In this group there was a decrease in the risk of bleedings, thromboembolic events and deaths. The ICER for this group was US$13,500 per QALY gained. For the patients with CYP2C9 variant alleles genotype-guided dosing was dominated by the standard dosing strategy, meaning that genotyping resulted in a decrease in QALYs and an increase in costs. This result arose because of an increase in the frequency of thromboembolic events in this group.

The uncertainty around the cost of a pharmacogenetic test, as investigated in the one-way sensitivity analysis, caused the largest part of the uncertainty around the cost–effectiveness ratio. In one of the scenario analyses, data from Caraco et al. were used instead of data from the COUMAGEN trial.[8] In this scenario, the genotyping strategy was the dominant strategy. Genotyping was also the dominant strategy when it was assumed that genotyping reduced the bleeding risk in CYP2C9 variant patients further. The probabilistic sensitivity analysis revealed that there was a 15% chance that the pharmacogenetic-guided dosing was the dominant strategy. In addition, it was estimated that there was a 46% chance that the true ICER was below US$50,000 per QALY gained and a 67% chance that it was below US$100,000 per QALY gained.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.