CYP2C9 and oral anticoagulation therapy with acenocoumarol and warfarin: similarities yet differences

Clinical Pharmacology & Therapeutics (2004) 75, 376–380; doi: 10.1016/j.clpt.2004.01.007 Interindividual variability in drug metabolism presents a major challenge to optimizing therapy for a particular patient. Much effort has gone into identifying the sources of such variability, and in recent years, attention has focused on possible genetic determinants that alter the expression or function of involved enzymes. With the cytochrome P450 (CYP) superfamily of enzymes, genetic polymorphisms such as those with CYP2C19 and CYP2D6 are well established, but their overall clinical importance is debatable except in limited instances. However, with CYP2C9, there is now considerable evidence that genetic variation may be of more practical significance and importance in optimizing drug therapy. This situation is primarily based on several retrospective studies with warfarin in which drug dosage and adverse events, during both induction and maintenance phases of anticoagulation, have been shown to be associated with the CYP2C9 genotype; the presence of CYP2C9*2 and, to a far greater extent, CYP2C9*3 alleles results in greater difficulty and problems in anticoagulation than in wild-type (CYP2C9*1/*1) homozygotes.1,2,3 Warfarin is mainly prescribed in North America and Asia, whereas acenocoumarol is more commonly used in several European countries. These 2 drugs are structurally related, have a mechanism of action similar to that of vitamin K antagonists, and exhibit a marked difference in dosage requirements (up to 10-fold) and unpredictable interindividual variability in anticoagulation response. Furthermore, their metabolism involves CYP2C9,4,5 from which it might be presumed that in the case of acenocoumarol the genetic variability in this enzyme would, as with warfarin, affect the drug's clinical effect. Indeed, 2 articles6,7 in this issue of the Journal demonstrate this; however, they also indicate that the genetic factor is less important than with warfarin despite the similarities between the 2 drugs. Morin et al6 describe an association between several genetic polymorphisms of the CYP2C9 gene and the immediate anticoagulation response, 24 hours after a single oral dose of acenocoumarol by healthy subjects, as compared with homozygous CYP2C9*1/*1 patients. In addition to coding region variants (CYP2C9*2, CYP2C9*3, and CYP2C9*5), single nucleotide polymorphisms (SNPs) in the −2.1-kb promoter region were also investigated. The haplotype containing the CYP2C9*3 allele was the only one having a statistically significant, albeit limited (14%), contribution to the overall interindividual variability of the anticoagulant response, as measured by a reduction in plasma factor VII activity. Schalekamp et al,7 in contrast, studied patients during their initial 3 to 6 months of anticoagulation therapy and concluded that the CYP2C9*3 allele was associated with a lower stabilization dose and a higher risk of overanticoagulation and that it took longer to reach a state of stabilization. In addition, they also noted that the international normalized ratio (INR) value measured 4 days after initiation of therapy was significantly but modestly higher in CYP2C9*3 carriers. In neither study was anticoagulation associated with the CYP2C9*2 allele. These CYP2C9 genotype–anticoagulant response data are different from those with warfarin, for which both the CYP2C9*2 and CYP2C9*3 alleles have much more pronounced effects.1,2,3 We will comment in brief on this and related issues. Like warfarin, acenocoumarol is administered as a racemic mixture of R- and S-enantiomers. However, there are substantial differences in the pharmacokinetic and pharmacodynamic characteristics between the 2 drugs. With acenocoumarol, the anticoagulation potencies of the R- and S-enantiomers are essentially comparable, as indicated by the area under the effect curve (AUEC) of prothrombin time normalized to the area under the plasma concentration–time curve (AUC) of the respective enantiomers (ie, AUEC/AUC) calculated on the basis of previously reported data.8,9 Furthermore, the mean oral clearance of S-acenocoumarol in homozygous wild-type (CYP2C9*1/*1) subjects is 13-fold greater than that of R-acenocoumarol (19.8 L/h versus 1.56 L/h), and its elimination half-life is far shorter (1.0 hours versus 8.8 hours).10 As a result, the overall anticoagulation response in individuals of this genotype is attributable largely to R-acenocoumarol. By contrast, the anticoagulation activity of the S-enantiomer of warfarin is 3 to 5 times greater than its antipode.11 However, its oral clearance in CYP2C9*1/*1 subjects is only 40% greater than that of R-warfarin (0.25 L/h versus 0.18 L/h),12 and the associated elimination half-life (32 hours versus 43 hours) is sufficiently long to produce an anticoagulant response throughout the usual once-daily dosing interval. Thus the intrinsically more potent S-warfarin is essentially responsible for the anticoagulant responses to racemic warfarin. CYP2C9 is almost exclusively involved in the metabolism of the S-enantiomers of both acenocoumarol and warfarin through 6- and 7-hydroxylations.4,5 On the other hand, other enzymes besides CYP2C9 (eg, CYP1A2, CYP2C19, and CYP3A4) mediate the metabolism of the R-enantiomers of both anticoagulants.4,5 Accordingly, the genetic polymorphisms of CYP2C9 would be predicted to have a greater influence on the metabolism and pharmacodynamic response of warfarin compared with acenocoumarol, and this would be more apparent with CYP2C9*3 than with CYP2C9*2 because of the considerably lower metabolic activity of the former.13 Indeed, patients with the CYP2C9*1/*3 genotype have an approximately 50% lower mean oral clearance of S-acenocoumarol than CYP2C9*1/*1 homozygotes (10.9 L/h versus 19.8 L/h),10 resulting in 2-fold higher steady-state S-acenocoumarol plasma concentrations in such patients relative to those of the homozygous wild-type genotype. The pharmacokinetics of R-acenocoumarol is only modestly affected in CYP2C9*1/*3 heterozygotes10; accordingly, the steady-state concentrations of this enantiomer would be minimally different from those in the CYP2C9*1/*1 genotype, and, therefore, its plasma concentrations would still far exceed (>6-fold) those of the S-enantiomer. Accordingly, the overall anticoagulation effect (INR) in such heterozygotes is attributable to the sum of the plasma concentrations of R- and S-acenocoumarol, but this would be similar or only modestly increased compared with wild-type homozygotes.10 The data of Morin et al6 are consistent with these pharmacokinetic considerations, in that subjects with the CYP2C9*1/*3 genotype showed only a 15% greater INR value and a 53% greater reduction in factor VII coagulant activity at 24 hours after a single oral dose of the drug. In the single CYP2C9*3/*3 homozygote, the increased effects were still quite modest, consistent with the still dominant role of R-acenocoumarol, because it has been estimated that the clearance of the S-enantiomer in such individuals is about 20% of that in homozygous wild-type subjects.10 Nevertheless, Schalekamp et al7 observed that, in patients treated with acenocoumarol for 3 to 6 months, the CYP2C9*3 allele was associated with greater difficulties in empirically determining an optimal anticoagulation dose than in wild-type patients and carriers of the CYP2C9*2 allele. In addition, the stabilized dose requirement for CYP2C9*3 patients was 20% less than in CYP2C9*1/*1 individuals. With warfarin, however, the reduced metabolic activity of the CYP2C9*2 and CYP2C9*3 variants also primarily affects only the S-enantiomer, which in this case is also the isomer responsible for almost all of the anticoagulation effect, especially in patients carrying these mutations. Thus a 50% lower clearance in CYP2C9*1/*3 patients is associated with a 2-fold reduction in warfarin's maintenance dose compared with that in wild-type homozygotes, and even in CYP2C9*2 carriers, the dose is lower.14,15 In CYP2C9*3/*3 homozygotes an even greater dose reduction (5-fold) is required.14,15 Morin et al6 also investigated SNPs in the −2.1-kb 5′-flanking region that putatively might affect CYP2C9 expression. As noted previously,16,17 a number of these were found to be in linkage disequilibrium with common functional SNPs in the coding region, and, therefore, haplotype associations rather than single SNP analyses are, in principle, probably more appropriate for genetic analysis. Morin et al note that just 4 major haplotypes accounted for 97% of the white population; 2 of them were linked with either CYP2C9*2 or CYP2C9*3, and in the case of the latter haplotype, the consequence of the impaired reduction in the encoded enzyme's activity far exceeded any effect on expression, if it was present. In fact, no 5′-flanking region variants through −2.1 kb that up-regulate CYP2C9 expression have yet been identified.16,17 Whether these exist in more distal regions remains to be determined, but such information would possibly provide insights into yet unresolved questions. Fig 1, for example, shows the relationship between the empirically established daily maintenance doses of warfarin and the unbound clearance of S-warfarin obtained in white patients (n = 47) and Japanese patients (n = 126) in our previous studies18,19,20 and in white patients (n = 93) reported by Scordo et al.14 In this figure, variability on the abscissa and ordinate represents that in the pharmacokinetics and pharmacodynamics of warfarin, respectively. Despite significant differences in target INR value (2 to 3 in white patients versus 1.5 to 2.5 in Japanese patients) and body size (76.5 ± 16.6 kg in white patients versus 56.7 ± 11.3 kg in Japanese patients), a significant relationship linked to CYP2C9 genotype exists. Given that the dosing rate reflects clearance at steady state and S-warfarin's clearance is predominantly CYP2C9-mediated, this is not too surprising; that is, patients who require higher doses of warfarin possess greater CYP2C9 activities. However, it is noteworthy that there is considerable variability within the overall pharmacokinetic and pharmacodynamic relationship and within any particular genotype. Similar interindividual variability in the warfarin dosage requirement among CYP2C9*1/*1 homozygous patients, who constitute the majority of all populations studied to date, has also been reported by Daly and King.15 Furthermore, Morin et al6 observed marked differences in acenocoumarol's short-term pharmacodynamic response. Because factors such as age, body size, sex, diet, and vitamin K status do not fully account for such apparently non–CYP2C9-associated variability,6,21,22 this strongly suggests that other, currently unknown environmental determinants or possibly unidentified genetic variants may be involved, especially in the 5′-flanking region or the transcriptional regulatory receptors of the gene.23 These considerations also apply to the significantly greater unbound clearance of S-warfarin in CYP2C9*1/*1 homozygotes of Japanese descent compared with white patients.20 In addition, the possibility of genetically determined variability in the various proteins involved in the anticoagulant effect, such as vitamin K–dependent coagulation factors,24 which may contribute to differences in warfarin and acenocoumarol dosage requirements, is a largely unexplored area with regard to interindividual variability in responsiveness. Relationships between unbound oral clearance (CLpo,u) for S-warfarin and daily doses of warfarin in white (Caucasian; a) and Japanese (b) patients of different genotypes: CYP2C9*1/*1 (open circles), CYP2C9*2 (*1/*2 or *2/*2) (gray circles), CYP2C9*1/*3 (gray triangles), CYP2C9*3/*3 or CYP2C9*2/*3 (black triangles), and CYP2C9*1/*11 (black diamond). Finally, the current reports,6,7 along with additional findings regarding warfarin, acenocoumarol, and other CYP2C9 substrates,25,26,27 clearly indicate that genotype contributes to variability in these drugs' overall pharmacodynamic responses. The critical question, however, is whether this factor is sufficiently determining that knowledge of an individual's CYP2C9 genotype before prescribing a drug dose would lead to improved optimization of therapy, including a reduced risk of bleeding episodes. Given the implied and possibly significant involvement of nongenetic determinants in warfarin's anticoagulant effects (Fig 1 and Verstuyft et al28), for example, the answer to this question is not self-evident. It is even less apparent for acenocoumarol. An appropriately designed, large, multicenter clinical trial will be required to test this hypothesis and could serve as a bellwether study for the future of genetically based, personalized drug therapy involving interindividual variability in drug metabolism. The design, funding, implementation, and evaluation of such a trial, including pharmacoeconomic issues, for drugs such as warfarin and acenocoumarol, whose patents will have expired, will be difficult; the development of alternative and safer anticoagulants such as direct thrombin inhibitors also complicates the issue. In the absence of such a study, there is the possibility that, as with CYP2C19 and CYP2D6, genetically determined variability in CYP2C9 activity may simply remain an interesting scientific phenomenon with limited clinical application except to explain observations in a limited number of situations. None of the authors has financial or personal relationships that could be perceived as conflicts of interest.