Download Building Individualized Medicine: Prevention of Adverse Reactions

0022-3565/07/3222-427–434$20.00
THE JOURNAL OF PHARMACOLOGY AND EXPERIMENTAL THERAPEUTICS
Copyright © 2007 by The American Society for Pharmacology and Experimental Therapeutics
JPET 322:427–434, 2007
Vol. 322, No. 2
117952/3231400
Printed in U.S.A.
Perspectives in Pharmacology
Building Individualized Medicine: Prevention of Adverse
Reactions to Warfarin Therapy
Evgeny Krynetskiy and Patrick McDonnell
Departments of Pharmaceutical Sciences (E.K.) and Pharmacy Practice (P.M.), Temple University School of Pharmacy,
Philadelphia, Pennsylvania
ABSTRACT
Warfarin is the most widely used oral anticoagulant in the world
for patients with venous thrombosis, pulmonary embolism,
chronic atrial fibrillation, and prosthetic heart valves. Approximately 30 genes contribute to therapeutic effects of warfarin,
and genetic polymorphisms in these genes may modulate its
anticoagulant activity. In contrast to monogenic pharmacogenetic traits, warfarin drug response is a polygenic trait, and
development of diagnostic tools predictive of adverse reactions
to warfarin requires a novel approach. A combination of two
strategies, biochemical isolation of allelic variants and linkage
disequilibrium association studies, was used to find an association between genetic polymorphisms in the candidate genes
The human genome is variable both within generations
(more than 7 million SNPs with a minor allele frequency at
least 5%) and between generations (175 mutations per diploid genome per generation) (Kruglyak and Nickerson, 2001).
Genetic variations make us unique in many senses, including
our response to drug therapy. Pharmacogenomics uses the
tools of human genetics to tailor medicinal treatment to an
individual’s genetic makeup. To this end, phenotypic manifestations (a therapeutic outcome or an adverse drug event,
ADE) are considered in relation to the underlying genetic
background of a patient.
In a pregenome era, these studies were based on biochemical isolation and characterization of drug-metabolizing enzymes and their genes, followed by sequence and functional
analysis of possible mutant variants. Characterization of mutations led to development of a genotyping assay that was
Article, publication date, and citation information can be found at
http://jpet.aspetjournals.org.
doi:10.1124/jpet.106.117952.
and warfarin response. A strong association was found between genetic polymorphisms in six genes, including VKORC1,
CYP2C9, PROC, EPHX1, GGCX, and ORM1, and interindividual
variability in the anticoagulant effect of warfarin; the strongest
predictors were VKORC1 and CYP2C9. Generation of single
nucleotide polymorphism (SNP)-based dense genetic maps
made it possible to identify haplotypes associated with drugresponse phenotypes. Discrimination between haplotypes associated with warfarin dose phenotypes can be achieved by a
limited set of informative polymorphisms (tag SNPs). The use of
tag SNPs in pharmacogenomic analysis provides a promising
tool for dissecting polygenic traits of drug response.
used to perform genetic analysis of patient DNA. Initial
achievements of pharmacogenomics were related to detection
of simple monogenic traits, e.g., abrogation of drug metabolism due to inactivation of a single enzyme. Characterization
of inactivating mutations in arylamine-N-acetyltransferase
2, cytochrome P450 2D6, and thiopurine S-methyltransferase provided molecular mechanisms of adverse drug
events caused by izoniazid, debrisoquine, and mercaptopurine (Gonzalez et al., 1988; Vatsis et al., 1991; Krynetski et
al., 1995). Biochemical analysis does not rely on statistical
evaluation of genetic markers and therefore works equally
well for common and rare genotypes. Such functional studies
require extensive knowledge of the biotransformation, transport, and physiological activity of the drug in question.
In contrast, linkage disequilibrium (LD) association studies rely on dense maps of human genome where nearly every
human gene and genomic region is marked by a sequence
variation. With development of massive parallel methods for
SNP detection, linkage analysis and association studies are
increasingly used to effectively characterize complex poly-
ABBREVIATIONS: SNP, single nucleotide polymorphism; LD, linkage disequilibrium; ADE, adverse drug event(s); INR, international normalized
ratio; VKOR, vitamin K epoxide reductase; ORM, orosomucoid.
427
Downloaded from jpet.aspetjournals.org at ASPET Journals on June 11, 2017
Received February 9, 2007; accepted May 10, 2007
428
Krynetskiy and McDonnell
Warfarin Therapy and Clinical Complications
An ADE is any undesired and harmful effect of a drug that
usually falls in one of two categories: 1) adverse drug reactions (ADR) defined as “any response to a drug that is noxious and unintended and occurs at doses normally used in
man for the prophylaxis, diagnosis, or therapy of diseases”
(World Health Organization, 1969); or 2) therapeutic failure,
i.e., the failure to achieve the desired therapeutic outcome,
with an event that can harm the patient.
The goal of warfarin therapy is to administer the lowest
possible dose of anticoagulant to prevent clot formation or
expansion while trying to avoid unintended ADE from overanticoagulation. Warfarin-related ADR result in unintended
hypoprothrombinemia, with or without frank hemorrhage.
ADE are common with warfarin, thus being a leading cause
of drug-induced hospital admissions (McDonnell and Jacobs,
2002). The incidence of these ADR ranges from 6 to 39% of
warfarin patients annually and is directly related to the
intensity of anticoagulation (Levine et al., 1995). By contrast,
therapeutic failure of warfarin treatment is an event that
produces unintended thromboembolism due to suboptimal
anticoagulation in these patients.
Warfarin has both anticoagulant and antithrombotic activity. The anticoagulant activity of warfarin depends on the
clearance of functional clotting factors from the systemic
circulation. The clearance is determined by half-lives of the
clotting factors, with earliest changes in prothrombin time
(24 to 26 h after warfarin administration) due to the clearance of factor VII (half-life of approximately 6 h). Early
changes in prothrombin times do not reflect the full antithrombotic effects of warfarin, which are achieved by the 5th
day in most patients after the functional clearance of prothrombin (half-life of approximately 50 h) (Hirsh et al., 1995).
Therefore, the maximal anticoagulant effect is achieved after
72 to 96 of exposure.
Pharmacogenomic interpatient differences in warfarin response are an important group of factors that need to be
considered for warfarin dosing. Mutations in genes involved
in the therapeutic effect of warfarin cause two distinct phenotypes—increased risk of hypoprothrombinemia and warfarin resistance. In many cases, genotype can be a critical
variable in determining the optimal dosing to achieve anticoagulant and antithrombotic effects (Voora et al., 2005).
Molecular Basis for Warfarin-Induced Adverse
Drug Reactions
The biochemical basis of warfarin action is reasonably well
understood (Fig. 1). The anticoagulant effect of warfarin is
achieved through inhibition of vitamin K epoxide reductase
(VKOR) activity, resulting in a decrease of the reduced form
of vitamin K. The deficiency of the reduced form of vitamin K
results in abrogation of post-translational modification of the
blood coagulation components, including clotting factors II,
VII, IX, and X; proteins C, S, and Z; and several other proteins, thus reducing their activity (Rost et al., 2005). Unintended hypoprothrombinemia in a genetically predisposed
subpopulation of patients most often originates from reduced
metabolic inactivation of warfarin by CYP2C9 or decreased
expression of the drug target (VKOR).
CYP2C9 is the major isoform of human liver cytochromes
P450 that modulates the physiological effect of warfarin. Low
rate of oxidative metabolism catalyzed by CYP2C9 hypomorphic alleles *2 and *3 results in slow metabolic inactivation of
warfarin, which in turn leads to undesired hypoprothrombinemia (Daly and King, 2003; Palkimas et al., 2003). The
decrease in CYP2C9 activity is most often caused by the
nonsynonymous SNPs present in CYP2C9*2 (rs1799853: T;
R144C) and CYP2C9*3 (rs1057910: C; I359L). The reduced
activity of CYP2C9*2 mutant (R144C) may reflect its diminished affinity for cytochrome P450 reductase, which interacts
with the positively charged residues on the surface of
Downloaded from jpet.aspetjournals.org at ASPET Journals on June 11, 2017
genic traits of drug response. The association studies make
use of statistical analysis of genetic markers (e.g., SNPs) in
large groups of patients stratified according to their dose
requirements. This approach uses genetic markers rather
than functional polymorphisms and therefore does not provide information about possible mechanisms behind alterations in drug response. Instead, it relies on the assumption
that a mutant allele is to be correlated with an allele of an
assayed SNP.
Ideally, both approaches should result in identification of
the same set of functionally important polymorphisms, which
in reality is rarely achieved. Whereas functional studies
proved to be efficient in analysis of single genes, association
studies hold great promise in identifying multiple alleles
contributing to polygenic traits. A clinically important example for these two strategies has been provided by recent
dissection of warfarin-pharmacogenomic determinants. Metabolism, transport, and pharmacological effects of warfarin
are determined by interplay of approximately 30 genes, and
genetic analysis of this pharmacological pathway is still in
progress (Wadelius and Pirmohamed, 2006).
Warfarin is the most widely used oral anticoagulant in the
world for patients with venous thrombosis, pulmonary embolism, chronic atrial fibrillation, and prosthetic heart valves
(Wadelius and Pirmohamed, 2006). The therapeutic index for
individual patients is narrow; therefore, patients are closely
monitored by international normalized ratio (INR) for prothrombin time. Clinical use of warfarin is complicated due to
variations in individual response, and prescribed doses may
vary 20-fold or more because dose requirements, the stability
of anticoagulation, and risk of bleeding are determined by a
complex combination of environmental and genetic factors.
Whereas nongenetic factors, such as intake of vitamin K,
disease, age, gender, concurrent medications, and body surface area, have been taken into consideration by clinicians,
the genetic analysis still remains an underused tool, and
clinicians must be further educated on how their patients can
benefit from timely recognition of pharmacogenetic variations (Krynetskiy and Evans, 2004).
Recent progress in elucidating pharmacogenetic variables
contributing to variance in warfarin dose made it practical to
implement genotyping of patients before beginning warfarin
therapy (Geisen et al., 2005; Voora et al., 2005). A rational
selection of genetic markers for genotyping poses a problem
for applied pharmacogenomics, and the choice of informative
polymorphisms predictive of altered warfarin pharmacokinetics or pharmacodynamics still remains to be a matter of
discussion. The current article aims to describe available
genotyping assays for evaluating patients’ genotype as a risk
factor in warfarin treatment-related complications.
Pharmacogenomics of Warfarin
429
CYP2C9. The mutation I359L, although conservative, leads
to a near-complete abrogation of CYP2C9 activity both in
vivo and in vitro. Structural analysis revealed that Ile359 lies
outside the active site in the vicinity of an important residue
Phe476 (Williams et al., 2003); the proximity of Leu359 to
Phe476 supposedly abrogates substrate binding.
Several variants of VKORC1 gene associated with decreased enzymatic activity and resistance to warfarin have
been described (Table 1). The carriers of these alleles either
require enhanced warfarin dose or they do not respond to
warfarin treatment at all (Rost et al., 2004, 2005). Although
the effect of some of these mutations on enzymatic activity
and warfarin-induced inhibition had been demonstrated in
vitro, significance of other mutations still remain to be confirmed (Harrington et al., 2005; Loebstein et al., 2006). On
the other hand, recently described coding polymorphism
113A⬎C (D38S) in Japanese population was linked to low
warfarin dose (Obayashi et al., 2006).
The presence of rs9923231:A in the promoter region of the
VKORC1 gene correlates with decreased VKORC1 transcription, presumably resulting in lower level of VKOR and therefore limiting the reduced vitamin K recycling. Lower doses of
warfarin are needed phenotypically to attain the same INR
in the group of patients with this genotype (D’Andrea et al.,
2005; Rieder et al., 2005; Yuan et al., 2005).
Although the three-dimensional structure of VKOR is currently unavailable, topology mapping and bioinformatics
analysis were used to identify transmembrane segments and
their cytoplasmic/exoplasmic orientation in this protein (Tie
et al., 2005). In parallel, site-mutagenesis of all cysteines and
two highly conserved residues (Ser57 and Arg98) helped to
define functionally important regions within VKOR (Rost et
TABLE 1
Genetic polymorphisms detected in ORF of VKORC1 gene
Position in ORF
Substitution
Effect
Dose
Reference
47G⬎C
85G⬎T
106G⬎T
112G⬎T
121G⬎T
129C⬎T
129C⬎T
134T⬎C
172A⬎G
196G⬎A
203A⬎G
292C⬎T
358C⬎T
358C⬎T
383T⬎G
383T⬎G
452G⬎A
C16S
V29L
D36Y
D38Y
A41S
C43C
C43C
V45A
R58G
V66M
H68R
R98W
L120L
L120L
L128R
L128R
R151Q
No change
Resistance
Resistance
No change
Resistance
N.A.
No effect
Resistance
Resistance
Resistance
No change
Inactive VKOR
N.A.
No effect
Resistance
Resistance
No effect
2.7 mg/day
100 mg/week
80 mg/week
5.3 mg/day
15.5 mg/day
N.A.
N.A.
No response
220 mg/week
25 mg/day
3.8 mg/day
N.A.
N.A.
N.A.
No response
45 mg/day
3.7 mg/day
Obayashi et al., 2006
Rost et al., 2004
Loebstein et al., 2006
D’Andrea et al., 2005
Rieder et al., 2005
Rost et al., 2004
D’Andrea et al., 2005
Rost et al., 2004
Rost et al., 2004
Harrington et al., 2005
Obayashi et al., 2006
Rost et al., 2004
Rost et al., 2004
D’Andrea et al., 2005
Rost et al., 2004
Bodin et al., 2005
D’Andrea et al., 2005
N.A., not available.
Downloaded from jpet.aspetjournals.org at ASPET Journals on June 11, 2017
Fig. 1. The vitamin K-dependent
␥-carboxylation system. In newly synthesized proteins, the ␥-glutamyl carboxylase (the product of GGCX gene)
converts glutamic residues (Glu) to
␥-carboxyglutamic residues (Gla). The
reduced cofactor vitamin K hydroquinone (vitamin KH2) is consumed in
this reaction; the resulting vitamin K
epoxide is recycled through catalysis
by VKOR (the product of VKORC1
gene) and vitamin K reductase. Warfarin inhibits VKOR, thus interrupting the vitamin K cycle. Adapted with
changes from Stafford (2005), and
chemical structures are given according to Stafford (2005).
430
Krynetskiy and McDonnell
Genotype as a Warfarin-Related Risk Factor
The role of genetic polymorphism in warfarin metabolism
was first demonstrated in 1994 when Rettie et al. (1994)
identified CYP2C9*2 as an isoform catalyzing (S)-7-hydroxylation of S-warfarin at low rate. This discovery was further
corroborated by genetic association between expression of
CYP2C9*2 and *3 and sensitivity to warfarin. Comparison of
daily warfarin doses in patients homozygous for the wildtype (*1) and hypomorphic alleles (*2 and *3) demonstrated
that reduced CYP2C9 activity was associated with lower
daily doses (Steward et al., 1997; Aithal et al., 1999).
This finding was confirmed multiple times in different
populations (for recent works, see Herman et al., 2005; Voora
et al., 2005; Loebstein et al., 2006; Obayashi et al., 2006;
Tham et al., 2006); several review articles summarized these
observations (Daly and King, 2003; Palkimas et al., 2003;
Wadelius and Pirmohamed, 2006). Although CYP2C9 is the
major catalyst responsible for biotransformation of warfarin
and a closely related acenocoumarol, CYP2C9-catalyzed metabolism of another warfarin analog, phenprocoumon, is less
important. Consequently, the effect of CYP2C9 polymorphism on phenprocoumon metabolism and anticoagulant response is moderate. Surprisingly, in a study performed in 201
patients, mostly of Swedish origin, an association between
the CYP2C9*2 polymorphism (rs1799853) and decrease in
warfarin dose was found to be not significant by univariate
analysis (Palkimas et al., 2003; Wadelius et al., 2006). The
hypomorphic allele CYP2C9*3 (rs1057910), which abolishes
CYP2C9-catalyzed metabolism of S-warfarin, was strongly
associated with warfarin dose, in accordance with previous
findings (Margaglione et al., 2000). Interestingly, the minor
allele of rs4917639 was in perfect LD (r2 ⫽ 1), with a composite minor allele formed by aggregating CYP2C9*2 and
CYP2C9*3 into a single allele, suggesting that the rs4917639
mutation occurred first, with rs1799853 (*2) and rs1057910
(*3) arising later independently on the same parent allele
(Wadelius et al., 2006).
Inhibition of vitamin K epoxide reductase is the major
mechanism of warfarin action, and the VKORC1 is the single
gene most strongly associated with warfarin dose (see below).
Existence of a mutation in an autosomal gene causing resistance to the hypoprothrombinemic effects of coumarin drugs
was first described in seven persons in three generations
(O’Reilly et al., 1964). The gene for vitamin K epoxide reductase (VKORC1) was cloned in 2004 by Rost et al. (2004) and
Li et al. (2004).
The role of VKORC1 genetic polymorphism in warfarin
therapy was demonstrated in a study on 147 patients where
an association between VKORC1 mutant alleles and warfarin dose requirement was found (D’Andrea et al., 2005). A
common polymorphism 1173C⬎T (rs9934438) was detected
in intron 1 of VKORC1 gene, which correlated with higher
warfarin dose required for carriers of VKORC1 1173CC genotype compared with 1173TT genotype. Statistical models
confirmed that the VKORC1 1173C⬎T polymorphisms, along
with CYP2C9 haplotypes, significantly correlated with different average doses of warfarin prescribed to attain the anticoagulation target. Because no functional role 1173C⬎T was
found in this study, this polymorphism was hypothesized to
be in LD with other allelic variants that modulate sensitivity
to warfarin therapy (D’Andrea et al., 2005).
This hypothesis was confirmed in a study on 104 Asian
patients receiving warfarin where a novel polymorphism
(⫺1639G⬎A) (rs9923231) in VKORC1 promoter region was
detected (Yuan et al., 2005). This polymorphism was in LD
with the VKORC1 1173C⬎T intronic polymorphism reported
by D’Andrea et al. (2005) and the affected transcription of
VKORC1 gene, thus confirming its functional significance.
In a retrospective study, the effect of VKORC1 haplotypes
on transcriptional regulation and warfarin dose was evaluated in 186 European-American patients receiving long-term
warfarin maintenance therapy (Rieder et al., 2005). Ten common noncoding SNPs in the VKORC1 gene with frequencies
greater than 5% had been identified, and seven of these SNPs
were significantly associated with warfarin dose (P ⬍ 0.001).
Among these seven SNPs, five SNPs (at positions 381, 3673,
6484, 6853, and 7566) strongly correlated with each other
(LD r2 ⱖ 0.9). Stepwise regression analysis identified these
five highly correlated SNPs to be predictive of approximately
25% variance in warfarin dose.
Five haplotypes with ⬎5% frequency were inferred from
ten common SNPs, and four of these haplotypes were independently associated with warfarin dose. Of these four haplotypes, two were associated with low warfarin dose requirement (2.9 –3 mg per day) and two with increased requirement
(5.5– 6.0 mg per day). A minimal set of four SNPs (861, 5808,
6853, and 9041) distinguished these groups. Patients were
further stratified according to CYP2C9 genotype. Stepwise
regression analysis showed that VKORC1 and CYP2C9 genotypes accounted for 21 and 6% of the variance in warfarin
dose correspondingly. Significant differences in frequency of
VKORC1 haplotypes were found among different ethnic
groups, which correlated with tendency to require low, intermediate, and high doses of warfarin for Asian, European, and
African populations (Rieder et al., 2005).
In a study of 200 blood donors from Germany, a complete
SNP map of the VKORC1 locus had been generated (Fig. 2)
Downloaded from jpet.aspetjournals.org at ASPET Journals on June 11, 2017
al., 2005). Cys51 and Ser57 located in the cytoplasmic extramembrane domain were absolutely necessary for VKOR
activity and presumably formed components of the active or
the binding center for the substrate vitamin K epoxide. Mutagenesis of another pair of cysteine residues located within
the C132XXC135 motif abrogated VKOR activity, suggesting
that this motif may serve as the redox center for VKOR (Tie
et al., 2005).
Interestingly, mutagenesis of Tyr139 resulted in warfarinresistant VKOR, providing a mechanistic explanation for
warfarin-resistant rat strains (Rost et al., 2004, 2005). The
Y139F variant is completely resistant to warfarin. Two
groups (Rost et al., 2004; Bodin et al., 2005) reported warfarin-resistant phenotype of heterozygotic patients with L128R
mutation in VKOR. Both Y139F and L128R are located
within the same transmembrane domain surrounding the
C132XXC135 motif, suggesting involvement of Tyr139 and
Leu128 in warfarin binding.
A possible role in warfarin dose requirement was suggested for a number of genes (e.g., EPHX1, ORM1/2, GGCX,
PROC) by association studies. The inclusive list of genes
coding for proteins involved in interaction with warfarin
pharmacological pathway can be found in a recent review
(Wadelius and Pirmohamed, 2006).
431
Pharmacogenomics of Warfarin
(Geisen et al., 2005). A total number of 28 SNPs were identified within the VKORC1 genomic sequence, including 1.8
kilobases of the 5⬘-flanking regions and 1.5 kilobases of the
3⬘-flanking regions, most of them in noncoding regions. Six of
these SNPs formed three haplotypes in complete LD, which
covers more than 99% of the genetic variability of VKORC1
in Europeans. A strong association was found between
VKORC1 haplotypes and phenotypes (increased warfarin resistance and increased warfarin sensitivity). For example,
among patients with increased coumarin sensitivity, 93%
was found to be homozygous carriers of the VKORC1*2 allele
(rs9923231:A). On the contrary, 86% of patients with partial
coumarin resistance were homozygous for the rs9923231:G
allele (non-VKORC1*2 haplotype). Interethnic differences in
VKORC1 haplotype frequencies also provided a good explanation for differences in coumarin requirements. The
VKORC1*2 haplotype frequencies in Han Chinese and Malay
correlate well with the reported low-dose warfarin requirement. In line with this observation, the warfarin-sensitive
haplotype VKORC1*2 is less frequent in Africans, which is in
perfect agreement with significantly higher average warfarin
requirement in African Americans. Thus, the VKORC1 haplotype distribution in three ethnic groups (Europeans, Africans, and Asians) is in complete concordance with the observed differences in warfarin response (Geisen et al., 2005).
In a comprehensive study, 29 genes in the warfarin pharmacological pathway were tested for approximately 900
SNPs (Wadelius et al., 2006). This association study made in
201 patients, mostly of North European descent, revealed
that the VKORC1 was the single gene most strongly associated with warfarin dose (Wadelius et al., 2006). Mapping of
the VKORC1 region on chromosome 16 revealed that three
SNPs (rs2359612, rs9934438, and rs9923231) were in strong
LD and accounted for approximately 30% of the warfarin
dose variance. A warfarin dose prediction model was prepared by combination of all genes nominally associated with
warfarin dose, and nongenetic factors (age, body weight, in-
teraction with other drugs, and indication for treatment).
Stepwise removal of variables with individual P values above
0 and low-explanatory value (R2) resulted in a model containing six genes, age, body weight, and drug interactions, which
explained 73% variance in warfarin dose. In the 29 genes
tested, three SNPs in VKORC1 and alleles CYP2C9*2 and
CYP2C9*3 were genetic factors most strongly associated
with warfarin dose (Table 2).
Because frequencies of haplotypes vary in ethnic groups,
some associations are difficult or impossible to find in certain
populations. In a study by Yuan et al. (2005), the 3730G⬎A
allele was found in LD with ⫺1639A⬎G allele and 1173C⬎T
allele in the Chinese population, both important predictors of
VKOR activity. However, this LD was not detected in the
Italian study (D’Andrea et al., 2005), probably because of the
difference in haplotypes. Several other polymorphisms, reasonably well characterized as contributing factors to warfarin dose requirement, were not found in association studies
TABLE 2
Multiple regression model of major determinants of the variance in
warfarin dose generated from association studies (adapted from
Wadelius et al., 2006)
Predictor
VKORC1
CYP2C9
SNP
rs9923231
rs1799853 (*2)
⫹
rs1057910 (*3)
P Value
⬍0.0001
⬍0.0001
Age
PROC
rs2069919
0.0029
0.0416
Body weight
EPHX1
rs4653436
0.0075
0.1016
Drug interaction
GGCX
ORM1
rs12714145
rs1687390
0.0878
0.0260
0.0571
VKORC1, vitamin K epoxide reductase complex subunit 1 gene; CYP2C9, cytochrome P450 2C9 gene; PROC, protein C gene; EPHX1, microsomal epoxide hydrolase 1 gene; GGCX, ␥-glutamyl carboxylase gene; ORM1, orosomucoid 1 gene or
␣-1-acid glycoprotein 1 gene.
Downloaded from jpet.aspetjournals.org at ASPET Journals on June 11, 2017
Fig. 2. Structure of VKORC1 haplotypes (adapted from Geisen et al., 2005). VKORC1*1 is a reference allele (wild type). Positions of genetic markers
(M2–M26) are given relative to nucleotides in the reference sequence (GenBank accession number AY587020) and corresponding SNPs (dbSNP:rs
numbers). The nucleotide changes relative to VKORC1*1 are shown as letters in the corresponding positions.
432
Krynetskiy and McDonnell
or were excluded due to monomorphic genotype and low
frequency of minor allele (Wadelius et al., 2006).
Other Polymorphisms Associated with
Warfarin Dose Requirement
Molecular Diagnostics of Enhanced Risk for
Warfarin-Induced ADR
After a strong association between genetic polymorphism
in VKORC1 gene and interindividual variability in the anti-
Search for Informative SNPs
Although acceptable for initial screening, routine analysis
of all known SNPs within the candidate genes may be impractical. In a recent study (Wadelius et al., 2006), approximately 900 SNPs in 29 genes have been genotyped. In the
setting where the causal SNPs (i.e., SNPs associated with
phenotype) remain unknown, a more rational approach is
selection of a subset of SNPs representing haplotypes (haplotype tagging SNPs, htSNPs) (Johnson et al., 2001). Besides
tagging common haplotypes, other algorithms for selecting
SNPs based on LD properties have been described (Goldstein
et al., 2003; Carlson et al., 2004). Among all possible SNPs
within a gene, a subset of tag SNPs can be selected for
genotyping analysis with minimal loss of information (Carlson et al., 2004). Selection of tag SNPs can significantly
reduce the volume of genotyping needed to find an association with a phenotype. Several programs for tag SNP selection are now available, including Tagger, HapBlock, and
WCLUSTAG (de Bakker et al., 2005; Zhang and Jin, 2003;
Sham et al., 2007). One of the limitations of this approach is
that tag SNPs identified in one population may not perform
well in another.
Downloaded from jpet.aspetjournals.org at ASPET Journals on June 11, 2017
Additional genes associated with warfarin dose requirement are indicated in Table 2: PROC (rs2069919), EPHX1
(rs4653436), GGCX (rs12714145), and ORM1 (rs1687390)
(Wadelius et al., 2006). Human ␣1-acid glycoprotein (or
ORM) is a major binding protein in blood plasma. ORM is
coded by two genes, ORM1 and ORM2, tightly linked on
chromosome 9. Whereas ORM2 is monomorphic in most populations, ORM1 is highly polymorphic (Li et al., 2002). ORM1
and ORM2 were significantly associated with warfarin dose
based on univariate regression analysis and multiple regression model (Wadelius et al., 2006). Warfarin enantiomers
were found to bind differently to ORM genetic variants, although clinical significance and warfarin-dosing requirements in carriers of genetic variants still remain to be
elucidated (Nakagawa et al., 2003; Hazai et al., 2006).
Genetic polymorphisms in genes coding for the components
of coagulation cascade account for altered sensitivity to warfarin (D’Ambrosio et al., 2004; Shikata et al., 2004). Of seven
vitamin K-dependent proteins (factors II, VII, IX, and X and
proteins S, C, and ␥-glutamyl carboxylase), genetic variations in factor II, factor VII, and ␥-glutamyl carboxylase were
associated with warfarin sensitivity (Shikata et al., 2004).
This study was performed in Japanese population where the
frequency of CYP2C9*3 is extremely low, thus giving opportunity to address factors contributing to the interpatient
variability other than altered metabolism. Polymorphisms in
promoter region of the gene of another vitamin K-dependent
protein, protein C, were significantly associated with warfarin dose (Aiach et al., 1999; Wadelius et al., 2006).
Several groups reported an association between warfarin
dose and genetic variants in ␥-glutamyl carboxylase gene
(GGCX) (Shikata et al., 2004; Chen et al., 2005; Loebstein et
al., 2005; Herman et al., 2006; Kimura et al., 2006; Wadelius
et al., 2006). ␥-Glutamyl carboxylase is a 758-amino acid
long, multi-pass, transmembrane domain protein in endoplasmic reticulum membrane involved in the ␥-carboxylation
of all vitamin K-dependent proteins, which act downstream
of VKOR in clotting cascade (Fig. 1). The binding site for the
substrate protein has been localized to the first 250 amino
acids, whereas the oxidation of reduced vitamin K to its
epoxide form occurs in the C-terminal region.
A highly polymorphic protein microsomal epoxide hydrolase (mEH, EPHX1) is hypothesized to be a component of the
vitamin K epoxide reduction system and therefore provides a
candidate for a association studies with warfarin dose requirement (Guenthner et al., 1998). Two association studies
indicate a possible role for EPHX1 in modulating warfarin
dose requirement, although this remains to be confirmed by
biochemical experiments (Loebstein et al., 2005; Wadelius et
al., 2006).
coagulant effect of warfarin was demonstrated, several studies further confirmed that genotype can be a predictor of
variance in warfarin dose (Geisen et al., 2005; Tham et al.,
2006; Wadelius et al., 2006). Inclusion of VKORC1 polymorphisms as co-variates to the prediction algorithm accounted
for more than 50% warfarin dose variability (Sconce et al.,
2005; Tham et al., 2006). In VKORC1 gene, of 28 SNPs
identified in a sample of 200 volunteers, six SNPs formed
three main haplotypes. There was a strong association between haplotypes distinguished by SNP (rs9923231) and coumarin dose phenotypes (P ⫽ 7.1 ⫻ 10⫺18). This SNP is in
strong LD with other SNPs, including rs9934438, rs2359612,
and rs7294 (Geisen et al., 2005).
Several algorithms have been proposed to achieve safer
anticoagulation based on demographic and environmental
factors, such as body weight, age, diet, and concomitant
drugs (Kovacs et al., 2003). Based on analysis of 369 patients,
an algorithm, including age, body surface area, CYP2C9 genotype, concomitant drug administration (amiodarone and
simvastatin), target INR, and race (Caucasian versus AfroAmerican), was developed that explained 39% variance in the
maintenance of warfarin dose (Gage et al., 2004). Prospective
validation of the dosing algorithm was performed that demonstrated feasibility of prediction of warfarin dosing based on
pharmacogenetic information in combination with nongenetic parameters. By using genotype-based dosing algorithm,
patients with a CYP2C9 variant achieved a stable warfarin
dose without excessive delay (Voora et al., 2005).
Given the low genetic diversity in VKORC1 and CYP2C9
haplotypes in an Asian population of Singapore, a simplified
model for warfarin daily dose requirement was developed
that included age, weight, CYP2C9*3 allele, and VKORC1
381CC and TT genotypes (Tham et al., 2006). This model was
validated in a separate cohort of 108 subjects and could
accurately predict warfarin dose requirements in Asian population.
Pharmacogenomics of Warfarin
Shifting the Paradigm toward Polygenic
Traits
Conclusions
Pharmacogenomics of warfarin is the next step in the direction of personalized medicine, in which polygenic rather
than monogenic traits are considered to make a rational
choice of the medication dose. Construction of dense genetic
maps based on single nucleotide polymorphisms made association studies an important instrument to dissect polygenic
traits of drug response. In combination with nongenetic factors, several polymorphic genes define warfarin dose-response phenotype. Of these genes, the strongest correlation is
between warfarin dose requirement and VKORC1 haplotypes
in combination with CYP2C9*2 and *3 genetic variants. Discrimination between haplotypes associated with warfarin
dose phenotypes can be achieved by a limited set of informative polymorphisms (tag SNPs). The use of tag SNPs in
pharmacogenomic analysis provides a promising tool for dissecting polygenic traits of drug response.
Acknowledgments
We are grateful to Dr. Natalia Krynetskaia for helpful discussion
and Dr. Anurag K. Mishra for thoughtful comments and preparation
of the Fig. 1.
References
Aiach M, Nicaud V, Henc-Gelas M, Gandrille S, Arnaud E, Amiral J, Guize L,
Fiessinger JN, and Emmerich J (1999) Complex association of protein C gene
promoter polymorphism with circulating protein C levels and thrombotic risk.
Arterioscler Thromb Vasc Biol 19:1573–1576.
Aithal GP, Day CP, Kesteven PJ, and Daly AK (1999) Association of polymorphisms
in the cytochrome P450 CYP2C9 with warfarin dose requirement and risk of
bleeding complications. Lancet 353:717–719.
Bodin L, Horellou MH, Flaujac C, Loriot MA, and Samama MM (2005) A vitamin K
epoxide reductase complex subunit-1 (VKORC1) mutation in a patient with vitamin K antagonist resistance. J Thromb Haemost 3:1533–1535.
Carlson CS, Eberle MA, Rieder MJ, Yi Q, Kruglyak L, and Nickerson DA (2004)
Selecting a maximally informative set of single-nucleotide polymorphisms for
association analyses using linkage disequilibrium. Am J Hum Genet 74:106 –120.
Chen LY, Eriksson N, Gwilliam R, Bentley D, Deloukas P, and Wadelius M (2005)
Gamma-glutamyl carboxylase (GGCX) microsatellite and warfarin dosing. Blood
106:3673–3674.
D’Ambrosio RL, D’Andrea G, Cappucci F, Chetta M, Di Perna P, Brancaccio V,
Grandone E, and Margaglione M (2004) Polymorphisms in factor II and factor VII
genes modulate oral anticoagulation with warfarin. Haematologica 89:1510 –1516.
D’Andrea G, D’Ambrosio RL, Di PP, Chetta M, Santacroce R, Brancaccio V, Grandone E, and Margaglione M (2005) A polymorphism in the VKORC1 gene is
associated with an interindividual variability in the dose-anticoagulant effect of
warfarin. Blood 105:645– 649.
Daly AK, and King BP (2003) Pharmacogenetics of oral anticoagulants. Pharmacogenetics 13:247–252.
de Bakker PI, Yelensky R, Pe’er I, Gabriel SB, Daly MJ, and Altshuler D (2005)
Efficiency and power in genetic association studies. Nat Genet 37:1217–1223.
Gage BF, Eby C, Milligan PE, Banet GA, Duncan JR, and McLeod HL (2004) Use of
pharmacogenetics and clinical factors to predict the maintenance dose of warfarin.
Thromb Haemost 91:87–94.
Geisen C, Watzka M, Sittinger K, Steffens M, Daugela L, Seifried E, Muller CR,
Wienker TF, and Oldenburg J (2005) VKORC1 haplotypes and their impact on the
inter-individual and inter-ethnical variability of oral anticoagulation. Thromb
Haemost 94:773–779.
Goldstein DB, Ahmadi KR, Weale ME, and Wood NW (2003) Genome scans and
candidate gene approaches in the study of common diseases and variable drug
responses. Trends Genet 19:615– 622.
Gonzalez FJ, Skoda RC, Kimura S, Umeno M, Zanger UM, Nebert DW, Gelboin HV,
Hardwick JP, and Meyer UA (1988) Characterization of the common genetic defect
in humans deficient in debrisoquine metabolism. Nature 331:442– 446.
Guenthner TM, Cai D, and Wallin R (1998) Co-purification of microsomal epoxide
hydrolase with the warfarin-sensitive vitamin K1 oxide reductase of the vitamin K
cycle. Biochem Pharmacol 55:169 –175.
Harrington DJ, Underwood S, Morse C, Shearer MJ, Tuddenham EG, and Mumford
AD (2005) Pharmacodynamic resistance to warfarin associated with a Val66Met
substitution in vitamin K epoxide reductase complex subunit 1. Thromb Haemost
93:23–26.
Hazai E, Visy J, Fitos I, Bikadi Z, and Simonyi M (2006) Selective binding of
coumarin enantiomers to human alpha1-acid glycoprotein genetic variants. Bioorg
Med Chem 14:1959 –1965.
Herman D, Locatelli I, Grabnar I, Peternel P, Stegnar M, Mrhar A, Breskvar K, and
Dolzan V (2005) Influence of CYP2C9 polymorphisms, demographic factors and
concomitant drug therapy on warfarin metabolism and maintenance dose. Pharmacogenomics J 5:193–202.
Herman D, Peternel P, Stegnar M, Breskvar K, and Dolzan V (2006) The influence
of sequence variations in factor VII, gamma-glutamyl carboxylase and vitamin K
epoxide reductase complex genes on warfarin dose requirement. Thromb Haemost
95:782–787.
Hirsh J, Dalen JE, Deykin D, Poller L and Bussey H (1995) Oral anticoagulants.
Mechanism of action, clinical effectiveness, and optimal therapeutic range. Chest
108:231S–246S.
Johnson GC, Esposito L, Barratt BJ, Smith AN, Heward J, Di GG, Ueda H, Cordell
HJ, Eaves IA, Dudbridge F, et al. (2001) Haplotype tagging for the identification
of common disease genes. Nat Genet 29:233–237.
Kimura R, Kokubo Y, Miyashita K, Otsubo R, Nagatsuka K, Otsuki T, Sakata T,
Nagura J, Okayama A, Minematsu K, et al. (2006) Polymorphisms in vitamin
K-dependent gamma-carboxylation-related genes influence interindividual variability in plasma protein C and protein S activities in the general population. Int
J Hematol 84:387–397.
Kovacs MJ, Rodger M, Anderson DR, Morrow B, Kells G, Kovacs J, Boyle E, and
Wells PS (2003) Comparison of 10-mg and 5-mg warfarin initiation nomograms
together with low-molecular-weight heparin for outpatient treatment of acute
venous thromboembolism. A randomized, double-blind, controlled trial. Ann Intern
Med 138:714 –719.
Kruglyak L and Nickerson DA (2001) Variation is the spice of life. Nat Genet
27:234 –236.
Krynetskiy EY and Evans WE (2004) Closing the gap between science and clinical
practice: the thiopurine S-methyltransferase polymorphism moves forward. Pharmacogenetics 14:395–396.
Krynetski EY, Schuetz JD, Galpin AJ, Pui CH, Relling MV, and Evans WE (1995) A
single point mutation leading to loss of catalytic activity in human thiopurine
S-methyltransferase. Proc Natl Acad Sci U S A 92:949 –953.
Levine MN, Hirsh J, Gent M, Turpie AG, Weitz J, Ginsberg J, Geerts W, LeClerc J,
Neemeh J, and Powers P (1995) Optimal duration of oral anticoagulant therapy: a
randomized trial comparing four weeks with three months of warfarin in patients
with proximal deep vein thrombosis. Thromb Haemost 74:606 – 611.
Li JH, Xu JQ, Cao XM, Ni L, Li Y, Zhuang YY, and Gong JB (2002) Influence of the
ORM1 phenotypes on serum unbound concentration and protein binding of quinidine. Clin Chim Acta 317:85–92.
Li T, Chang CY, Jin DY, Lin PJ, Khvorova A, and Stafford DW (2004) Identification
of the gene for vitamin K epoxide reductase. Nature 427:541–544.
Loebstein R, Dvoskin I, Halkin H, Vecsler M, Lubetsky A, Rechavi G, Amariglio N,
Cohen Y, Ken-Dror G, Almog S, et al. (2006) A coding VKORC1 Asp36Tyr polymorphism predisposes to warfarin resistance. Blood 109:2477–2480.
Loebstein R, Vecsler M, Kurnik D, Austerweil N, Gak E, Halkin H, and Almog S
(2005) Common genetic variants of microsomal epoxide hydrolase affect warfarin
dose requirements beyond the effect of cytochrome P450 2C9. Clin Pharmacol Ther
77:365–372.
Margaglione M, Colaizzo D, D’Andrea G, Brancaccio V, Ciampa A, Grandone E, and
Di MG (2000) Genetic modulation of oral anticoagulation with warfarin. Thromb
Haemost 84:775–778.
McDonnell PJ and Jacobs MR (2002) Hospital admissions resulting from preventable
adverse drug reactions. Ann Pharmacother 36:1331–1336.
Nakagawa T, Kishino S, Itoh S, Sugawara M, and Miyazaki K (2003) Differential
binding of disopyramide and warfarin enantiomers to human alpha(1)-acid glycoprotein variants. Br J Clin Pharmacol 56:664 – 669.
Downloaded from jpet.aspetjournals.org at ASPET Journals on June 11, 2017
Genetic analysis of variability in drug response is a rapidly
expanding area, mainly due to two factors. First, the knowledge of mechanisms of drug action and proteins involved in
metabolic activation/inactivation, transport, and drug-target
interactions in many cases is complete enough to identify
candidate genes. Second, the available technology provides
tools to rapidly detect genetic markers in specific loci associated with drug response. Together, these two factors make it
possible to revisit a huge collection of data pertinent to drug
interaction with human body. Genetic analysis has become
simple and fast enough to move it from scientific laboratories
to diagnostic medical centers; one can envision it to become
even more straightforward and automated to be a practical
instrument in pharmacies.
We already have a substantial set of pharmacogenomic
assays capable of predicting simple monogenic traits of altered metabolism. In a story with warfarin, up to six genes
have been associated with dose requirement. The efficient
use of this information will demand further genetic analysis
in populations with different racial backgrounds and novel
statistical instruments and models to accurately predict dose
requirement based on a combination of genetic and nongenetic factors.
433
434
Krynetskiy and McDonnell
AE (1997) Genetic association between sensitivity to warfarin and expression of
CYP2C9*3. Pharmacogenetics 7:361–367.
Tham LS, Goh BC, Nafziger A, Guo JY, Wang LZ, Soong R, and Lee SC (2006) A
warfarin-dosing model in Asians that uses single-nucleotide polymorphisms in
vitamin K epoxide reductase complex and cytochrome P450 2C9. Clin Pharmacol
Ther 80:346 –355.
Tie JK, Nicchitta C, von HG, and Stafford DW (2005) Membrane topology mapping
of vitamin K epoxide reductase by in vitro translation/cotranslocation. J Biol Chem
280:16410 –16416.
Vatsis KP, Martell KJ, and Weber WW (1991) Diverse point mutations in the human
gene for polymorphic N-acetyltransferase. Proc Natl Acad Sci U S A 88:6333–
6337.
Voora D, Eby C, Linder MW, Milligan PE, Bukaveckas BL, McLeod HL, Maloney W,
Clohisy J, Burnett RS, Grosso L, et al. (2005) Prospective dosing of warfarin based
on cytochrome P-450 2C9 genotype. Thromb Haemost 93:700 –705.
Wadelius M, Chen LY, Eriksson N, Bumpstead S, Ghori J, Wadelius C, Bentley D,
McGinnis R, and Deloukas P (2006) Association of warfarin dose with genes
involved in its action and metabolism. Hum Genet 121:23–34.
Wadelius M and Pirmohamed M (2006) Pharmacogenetics of warfarin: current
status and future challenges. Pharmacogenomics J 7:99 –111.
Williams PA, Cosme J, Ward A, Angove HC, Matak VD, and Jhoti H (2003) Crystal
structure of human cytochrome P450 2C9 with bound warfarin. Nature 424:464 –
468.
World Health Organization (1969) International Drug Monitoring. The role of the
hospital. WHO Technical Report Series 426:5–24.
Yuan HY, Chen JJ, Lee MT, Wung JC, Chen YF, Charng MJ, Lu MJ, Hung CR, Wei
CY, Chen CH, et al. (2005) A novel functional VKORC1 promoter polymorphism is
associated with inter-individual and inter-ethnic differences in warfarin sensitivity. Hum Mol Genet 14:1745–1751.
Zhang K and Jin L (2003) HaploBlockFinder: haplotype block analyses. Bioinformatics 19:1300 –1301.
Address correspondence to: Dr. Evgeny Krynetskiy, Dept. of Pharmaceutical Sciences, Temple University School of Pharmacy, 3307 North Broad
Street, Philadelphia, PA 19140. E-mail: [email protected]
Downloaded from jpet.aspetjournals.org at ASPET Journals on June 11, 2017
O’Reilly RA, Aggeler P, Hoag MS, Leong LS, and Kropatkin ML (1964) Hereditary
transmission of exceptional resistance to coumarin anticoagulant drugs. The first
reported kindred. N Engl J Med 271:809 – 815.
Obayashi K, Nakamura K, Kawana J, Ogata H, Hanada K, Kurabayashi M, Hasegawa A, Yamamoto K, and Horiuchi R (2006) VKORC1 gene variations are the
major contributors of variation in warfarin dose in Japanese patients. Clin Pharmacol Ther 80:169 –178.
Palkimas MP Jr., Skinner HM, Gandhi PJ, and Gardner AJ (2003) Polymorphism
induced sensitivity to warfarin: a review of the literature. J Thromb Thrombolysis
15:205–212.
Rettie AE, Wienkers LC, Gonzalez FJ, Trager WF, and Korzekwa KR (1994) Impaired (S)-warfarin metabolism catalysed by the R144C allelic variant of CYP2C9.
Pharmacogenetics 4:39 – 42.
Rieder MJ, Reiner AP, Gage BF, Nickerson DA, Eby CS, McLeod HL, Blough DK,
Thummel KE, Veenstra DL, and Rettie AE (2005) Effect of VKORC1 haplotypes on
transcriptional regulation and warfarin dose. N Engl J Med 352:2285–2293.
Rost S, Fregin A, Hunerberg M, Bevans CG, Muller CR, and Oldenburg J (2005)
Site-directed mutagenesis of coumarin-type anticoagulant-sensitive VKORC1: evidence that highly conserved amino acids define structural requirements for enzymatic activity and inhibition by warfarin. Thromb Haemost 94:780 –786.
Rost S, Fregin A, Ivaskevicius V, Conzelmann E, Hortnagel K, Pelz HJ, Lappegard
K, Seifried E, Scharrer I, Tuddenham EG, et al. (2004) Mutations in VKORC1
cause warfarin resistance and multiple coagulation factor deficiency type 2. Nature 427:537–541.
Sconce EA, Khan TI, Wynne HA, Avery P, Monkhouse L, King BP, Wood P, Kesteven
P, Daly AK, and Kamali F (2005) The impact of CYP2C9 and VKORC1 genetic
polymorphism and patient characteristics upon warfarin dose requirements: proposal for a new dosing regimen. Blood 106:2329 –2333.
Sham PC, Ao SI, Kwan JS, Kao P, Cheung F, Fong PY, and Ng MK (2007) Combining
functional and linkage disequilibrium information in the selection of tag SNPs.
Bioinformatics 23:129 –131.
Shikata E, Ieiri I, Ishiguro S, Aono H, Inoue K, Koide T, Ohgi S, and Otsubo K (2004)
Association of pharmacokinetic (CYP2C9) and pharmacodynamic (factors II, VII,
IX, and X; proteins S and C; and gamma-glutamyl carboxylase) gene variants with
warfarin sensitivity. Blood 103:2630 –2635.
Stafford DW (2005) The vitamin K cycle. J Thromb Haemost 3:1873–1878.
Steward DJ, Haining RL, Henne KR, Davis G, Rushmore TH, Trager WF, and Rettie