John T. Brandt, Lilly Corporate Center, Eli Lilly and Company, Indianapolis, IN 46285, USA. Tel.: +1 317 277 6478; fax: +1 317 276 4198; e-mail: firstname.lastname@example.org
Summary. Background: Thienopyridines are metabolized to active metabolites that irreversibly inhibit the platelet P2Y12 adenosine diphosphate receptor. The pharmacodynamic response to clopidogrel is more variable than the response to prasugrel, but the reasons for variation in response to clopidogrel are not well characterized.
Objective: To determine the relationship between genetic variation in cytochrome P450 (CYP) isoenzymes and the pharmacokinetic/pharmacodynamic response to prasugrel and clopidogrel.
Methods: Genotyping was performed for CYP1A2, CYP2B6, CYP2C19, CYP2C9, CYP3A4 and CYP3A5 on samples from healthy subjects participating in studies evaluating pharmacokinetic and pharmacodynamic responses to prasugrel (60 mg, n =71) or clopidogrel (300 mg, n =74).
Results: In subjects receiving clopidogrel, the presence of the CYP2C19*2 loss of function variant was significantly associated with lower exposure to clopidogrel active metabolite, as measured by the area under the concentration curve (AUC0–24; P =0.004) and maximal plasma concentration (Cmax; P =0.020), lower inhibition of platelet aggregation at 4 h (P =0.003) and poor-responder status (P =0.030). Similarly, CYP2C9 loss of function variants were significantly associated with lower AUC0–24 (P =0.043), lower Cmax (P =0.006), lower IPA (P =0.046) and poor-responder status (P =0.024). For prasugrel, there was no relationship observed between CYP2C19 or CYP2C9 loss of function genotypes and exposure to the active metabolite of prasugrel or pharmacodynamic response.
Conclusions: The common loss of function polymorphisms of CYP2C19 and CYP2C9 are associated with decreased exposure to the active metabolite of clopidogrel but not prasugrel. Decreased exposure to its active metabolite is associated with a diminished pharmacodynamic response to clopidogrel.
Thienopyridines are pro-drugs that are converted in vivo to active metabolites that react with and irreversibly inhibit the platelet P2Y12 adenosine diphosphate (ADP) receptor, which is involved in platelet activation and stabilization of the platelet aggregate [1–4]. Inhibition of platelet P2Y12 by clopidogrel, in combination with aspirin, has been shown to be effective in reducing the rate of major adverse cardiovascular events following acute coronary syndrome and is a routine component of the clinical management of patients with this syndrome [5–8].
Recent studies indicate that the pharmacodynamic response to clopidogrel is variable, with 20–40% of patients being classified as non-responders, poor-responders or resistant to clopidogrel because of low inhibition of ADP-induced platelet aggregation or activation [9–14]. A lower pharmacodynamic response has been linked to a higher relative risk of adverse cardiac events in several clinical studies, suggesting that a reduced pharmacodynamic response to clopidogrel is clinically relevant [15–20].
Studies comparing prasugrel, a novel thienopyridine currently in clinical development, to clopidogrel in healthy volunteers and patients with stable atherosclerosis have shown that prasugrel is more potent and achieves higher levels of inhibition of platelet aggregation (IPA) than seen with the approved 300 mg loading/75 mg maintenance dose regimen for clopidogrel [21–25]. In contrast, an in vitro study demonstrated that the active metabolites of prasugrel and clopidogrel are equipotent . Together, these findings suggest that differences in exposure to the active metabolite, either through differences in absorption or metabolism (or both), may account for the differences observed in pharmacodynamic response to prasugrel and clopidogrel.
The pathways leading to conversion of prasugrel and clopidogrel to their respective active metabolites differ. Prasugrel is rapidly hydrolyzed in vivo by esterases to a thiolactone, which is subsequently metabolized to the active metabolite in a single step by several CYP enzymes, mainly CYP3A4/5 and CYP2B6, and to a lesser extent by CYP2C19 and CYP2C9 . In contrast, two sequential CYP-dependent oxidative steps are required to convert clopidogrel to its active metabolite. The first step leads to formation of 2-oxo-clopidogrel, which is then metabolized to the active metabolite. In a competing metabolic reaction, esterases convert clopidogrel to inactive metabolites; this inactivation pathway accounts for an estimated 85% of a dose of clopidogrel . Enzymes involved in the metabolism of clopidogrel include CYP1A2, CYP2B6, CYP2C9, CYP2C19 and CYP3A4/5 [29–31].
We hypothesized that loss of function polymorphisms of CYP enzymes might contribute to decreased formation of the active metabolite of clopidogrel with a corresponding effect on the pharmacodynamic response. However, based on the in vitro metabolism and clinical data for prasugrel, such polymorphisms of CYP enzymes are not expected to affect the exposure to prasugrel active metabolite, and hence its pharmacodynamic response [27,32]. Therefore, we assessed the relationship between loss of function polymorphisms of various CYP enzymes and pharmacodynamic and pharmacokinetic responses to clopidogrel and prasugrel, as measured by IPA and exposure to the active metabolite of each drug.
Materials and methods
Subjects in this retrospective analysis participated in one of two studies conducted by the same research organization in the Netherlands examining the effect of loading doses of clopidogrel and prasugrel on platelet function. The two studies (internal study codes H7T-EW-TAAJ and H7T-EW-TAAK) had similar entry criteria, including that the individuals be healthy and free from concurrent medications, including aspirin. Individuals with baseline maximal platelet aggregation levels below 70% to 20 μm ADP were excluded. Both studies were approved by the local ethical review board and were conducted in a manner consistent with ethical principles based on the Declaration of Helsinki. Informed consent was obtained from all participants.
Material from 89 of 106 subjects participating in these studies was available for genetic analysis. Of these, 58 subjects participated in a cross-over design study during which they received prasugrel and clopidogrel on separate occasions; one subject received only prasugrel and one subject received only clopidogrel . The second study had a parallel arm design; 17 subjects receiving clopidogrel and 14 subjects receiving prasugrel were included in this analysis . Of the 89 total subjects, 71 received a 60 mg loading dose of prasugrel and 74 received a 300 mg loading dose of clopidogrel (Table 1). There were 77% males in each group. The ethnic background was predominantly Caucasian (n =80), with six of African descent, two Asian and one Mongolian–Caucasian.
Table 1. Demographic and baseline features of the study population
MPA baseline, maximal platelet aggregation at baseline before dosing with indicated drug.
Age, mean ± SD
41.5 ± 16.5
40.0 ± 16.9
MPA baseline, mean ± SD
87.2 ± 4.9
87.1 ± 4.9
Deoxyribonucleic acid (DNA) was isolated from peripheral blood cells and genotyping was completed using one of two methods. Exon specific polymerase chain reaction (PCR) amplification followed by standard restriction fragment length polymorphism analysis by gel electrophoresis was used for determination of CYP2C19 *2,*3, *4, *5, CYP2C9 *2, *3, CYP3A4 *1b,*2, *3 and CYP3A5 *2 (Cogenics, Morrisville, NC, USA). CYP1A2 *1C, *1F,*1K, *1L, CYP2B6 *11, *12 *14 and *15A, CYP2C9 *11, and CYP3A5 *3, *6, *8, *9 were determined with a research method at Eli Lilly and Co following PCR and separation by high-performance liquid chromatography.
CYP variants were classified based on the reported effect on function (Table 2; see http://www.cypalleles.ki.se). For the purpose of this analysis, we grouped CYP2C9 wt/wt and wt/*2 genotypes under the term ‘normal function’ and CYP2C9 *2/*2, wt/*3 and *2/*3 genotypes under the term ‘decreased function’ . For CYP2C19, the *2 allele was the only CYP2C19 decreased function allele observed in this study population .
Table 2. Detection of variant CYP alleles in study population (total n =89)
Allele frequency (%)
*Decreased function may represent reduced or absent activity. †Allele not detected. ‡wt/*1L can also be *1F/*1C. The genotype is uncertain as phase of the variants is unknown. §1L is not measured, but inferred, by a combination of *1F & *1C. ¶Insufficient data to determine genotype.
CYP2C19, n =89
CYP2C9, n =89
CYP3A4, n =89
CYP3A5, n =89
CYP1A2, n =86
CYP2B6, n =88
Pharmacodynamic and pharmacokinetic measurements
Pharmacodynamic measures The ex vivo platelet aggregation response to 5 and 20 μm ADP was measured predose and at 4 and 24 h postdose. Aggregation studies were performed within 2 h of sample acquisition using a BioData 4-channel aggregometer as previously described [25,32]. The pharmacodynamic response was measured as IPA, calculated as IPAt = [1–(MPAt/MPA0)] × 100%, where MPA0 is the maximal platelet aggregation before administration of study drug and MPAt is the MPA at time t.
A pharmacodynamic poor-responder was defined as a subject with less than 20% IPA to 20 μm ADP at either 4 or 24 h . Subjects not meeting this definition of poor-responder were classified as pharmacodynamic responders. The relationship between genotype and responder status was also explored using the following definitions of poor response: lowest quartile of response to clopidogrel 300 mg, change in MPA to 5 μm ADP <10 percentage points at 4 h, MPA >50% at 4 h and IPA <30% at 4 h [10,14,20,36].
Pharmacokinetics The active metabolites of prasugrel and clopidogrel were measured by validated liquid chromatography with tandem mass spectrometry methods, as previously described [25,32]. The pharmacokinetic parameters for the active metabolites were computed by standard non-compartmental methods using the log–linear trapezoidal method of WinNonlin Professional version 3.1 (Pharsight Corporation, Cary, NC, USA). Concentrations below the lower limit of quantification for the assays were excluded from analysis. The primary parameters for analysis were Cmax and AUC0–24.
The pharmacodynamic response to clopidogrel was related to the extent of exposure to the active metabolite of clopidogrel using a non-parametric Wilcoxon’s two-sample test of the AUC0–24 of the active metabolite of clopidogrel in the responder and poor-responder groups. In addition, the IPA response to clopidogrel was tested for association by a t-test of the standardized Spearman rank correlation coefficient.
The initial genetic analysis included chi-squared tests for departure from Hardy–Weinberg equilibrium. Each treatment was assessed separately for an association between exposure, as measured by Cmax and AUC0–24, and genetic variability in the CYP enzymes. The log-transformed AUC0–24 and Cmax values were tested for genotypic association by F-tests.
The relationship between responder status and CYP genotype was assessed for each treatment individually. Fisher’s exact test was used to assess the association of responder status with genotype for clopidogrel-treated subjects. The association between genotypes and response, as measured by quantitative IPA, was tested by an F-test in an anova model for both treatments. All analyses were completed using sas (version 9.1, SAS Institute Inc., Cary, NC, USA) and S-PLUS (version 7.0.6, Insightful Corporation, Seattle, WA, USA). A P-value < 0.05 was regarded as statistically significant.
Table 2 summarizes the observed frequencies of the CYP variants evaluated in this study population. No significant deviations from Hardy–Weinberg equilibrium (P >0.20) were observed for any of the genetic variants. Decreased function variants were seen for CYP2C19, CYP2C9 and CYP3A5.
Pharmacodynamic responses and pharmacokinetic parameters
Pharmacodynamics Baseline (predrug) maximal platelet aggregation did not differ between subjects receiving prasugrel or clopidogrel (Table 1). The pharmacodynamic response to clopidogrel 300 mg varied widely at both 4 and 24 h postdose; in contrast, IPA was greater with less variability following prasugrel 60 mg (Fig. 1). Thirty-six of 74 subjects (49%) had IPA less than 20% at 4 or 24 h following administration of clopidogrel 300 mg and were classified as pharmacodynamic poor-responders, while the remaining 38 subjects (51%) were classified as responders. All prasugrel-treated subjects had >20% IPA at 4 and 24 h following prasugrel 60 mg and were classified as pharmacodynamic responders.
The pharmacodynamic response to clopidogrel was related to the extent of exposure to the active metabolite of clopidogrel. Poor-responders had a significantly (P <0.001) lower exposure, as measured by AUC0–24, than did responders (Fig. 2A). In accord with the approach taken by other investigators, the subjects were also stratified into quartiles based on their pharmacodynamic response to clopidogrel [18–20]. Again, there was a statistically significant relationship between quartile of IPA response at 4 h and exposure to the active metabolite of clopidogrel, as measured by AUC0–24 (P <0.001; Fig. 2B), with increasing IPA associated with greater exposure.
There was a significant association between the CYP2C19*2 allele and poor-responder status to clopidogrel (P =0.030); 13 of 18 subjects (72.2%; Table 4) with the 2C19*2 allele were poor-responders. The presence of 2C19*2 was also associated with overall lower mean IPA at 4 h, AUC0–24 and Cmax (Table 3).
Table 4. Genotype by inhibition of platelet aggregation (IPA) responder status association for clopidogrel
Risk factor status
IPA responder status
Poor responder (%)
P-value for association
CYP2C9 *2/*2 or *3 carrier
P-value for association
Either CYP2C19*2 or CYP2C9 *2/*2 or *3
P-value for association
Table 3. Genotype association for pharmacokinetics of the active metabolites and pharmacodynamic measures
IPA at 4 h
IPA at 4 h
*Genotypic association using anova model following log normal adjustment to AUC0-24 and Cmax. AUC0–24, area under the plasma concentration curve over 0–24 h; Cmax, maximum plasma concentration; IPA at 4 h, inhibition of platelet aggregation 4 h after administration of drug.
wt/wt, mean ± SE
76.2 ± 17.9 ng h mL−1 (n =56)
58.4 ± 9.2 ng mL−1
39.1 ± 3.4%
544 ± 26.5 ng h mL−1 (n =54)
529 ± 35.3 ng mL−1
78.4 ± 1.2%
*2/wt, mean ± SE
41.5 ± 5.7 ng h mL−1(n =17)
35.3 ± 4.3 ng mL−1
20.3 ± 3.9%
504 ± 43.0 ng h mL−1 (n =16)
560 ± 51.8 ng mL−1
81.7 ± 2.3%
*2/*2 (n = 1), mean ± SE
26.9 ng h mL−1 (n =1)
27.9 ng mL−1
455 ng h mL−1 (n =1)
385 ng mL−1
Genotypic association significance (P value)*
Normal function, mean ± SE
74.3 ± 17.3 ng h mL−1 (n =58)
57.9 ± 8.9 ng mL−1
37.3 ± 3.1%
544 ± 25.4 ng h mL−1 (n =57)
561 ± 34.1 ng mL−1
79.7 ± 1.2%
Decreased function, mean ± SE
43.2 ± 5.6 ng h mL−1(n =16)
34.0 ± 5.4 ng mL−1
23.4 ± 6.7%
489 ± 45.5 ng h mL−1 (n =14)
421 ± 38.1 ng mL−1
77.4 ± 2.1%
Genotypic association significance (P value)*
The presence of CYP2C9 decreased function alleles was also significantly associated with decreased AUC0–24 (P =0.043), Cmax (P =0.006) and IPA at 4 h (P =0.046; Table 3) following administration of clopidogrel. In addition, an association between CYP2C9 genotypes with responder status to clopidogrel was observed, with 12 of 16 subjects (75%) with CYP2C9 decreased function genotypes classified as poor responders (P =0.024; Table 4).
The presence of CYP2C19*2 or CYP2C9 decreased function variants had no effect on the exposure to the active metabolite of prasugrel, and thus there was no effect on the IPA response to prasugrel (Table 3). There was a trend (P =0.055) towards a lower Cmax for prasugrel active metabolite in subjects with decreased function CYP2C9 genotypes (Table 3). In this analysis, no significant associations were observed between reduced function polymorphisms of CYP3A5 and pharmacodynamic response to clopidogrel or prasugrel or the pharmacokinetic measures, AUC0–24 and Cmax associated with either drug.
The possibility that the observed CYP2C9 or CYP2C19 effect was driven by concurrence of the variant genotypes was investigated. Only one subject had both CYP2C19*2 and CYP2C9 decreased function genotypes; thus, the findings for CYP2C19 and CYP2C9 appear to represent independent genetic effects for each CYP enzyme. Overall, the presence of either a CYP2C19*2 or CYP2C9 decreased function genotype was strongly associated with a pharmacodynamic poor-responder phenotype (P <0.001; Table 4). Indeed, 24 of 36 (66.7%) poor responders to clopidogrel had either a variant CYP2C19 or CYP2C9 genotype. A relationship between the presence of either CYP2C19*2 or CYP2C9 decreased function genotype and responder status to clopidogrel was noted with each of the other definitions of poor responder (Table 5).
Table 5. Relationship of poor-responder definition to the association of response to clopidogrel with genotype
Either CYP2C19*2 or CYP2C9 *2/*2 or *3
P-value for association
IPA at 4 h, inhibition of platelet aggregation 4 h after administration of drug; ADP, adenosine diphosphate; MPA at 4 h, maximal platelet aggregation 4 h after administration of drug.
One or more CYP-mediated oxidative steps are required to convert thienopyridines to their respective active metabolites. Consequently, variation in CYP function is a potential mechanism for decreased formation of the active metabolite of these agents. In this study, we found that subjects with genotypes of CYP2C19 and CYP2C9 associated with reduced function had decreased exposure to the active metabolite of clopidogrel but not prasugrel. CYP2C19*2 and the abnormal function CYP2C9 genotypes were also associated with a decreased pharmacodynamic response to clopidogrel. These observations with clopidogrel loading dose administration are consistent with the report from Hulot et al. , indicating that subjects with CYP2C19*2 had a reduced pharmacodynamic response to clopidogrel during daily maintenance dosing with 75 mg.
In contrast to the effects on clopidogrel, the CYP2C19 and CYP2C9 variant genotypes did not affect prasugrel pharmacodynamic and pharmacokinetic parameters. The differences in the metabolic pathways for these two molecules may account for this finding. In particular, prasugrel requires esterases to form its thiolactone followed by a single oxidative step that may be mediated by any one of several different CYPs . The lack of effect of CYP2C19 and CYP2C9 variant genotypes on the response to prasugrel suggests that for the metabolism of prasugrel, reduced activity in one CYP may be compensated for by residual activity of other CYPs.
We did not find an association between CYP3A5 genetic variants and the response to clopidogrel. However, emerging data point to a critical role for CYP3A4/5 in the metabolism of clopidogrel. Farid et al.  demonstrated that inhibition of CYP3A4 and CYP3A5 with ketoconazole was associated with a significant reduction in exposure to the clopidogrel active metabolite and a corresponding reduction in IPA. In contrast, inhibition of CYP3A4 and CYP3A5 by ketaconazole had no effect on prasugrel pharmacokinetics or pharmacodynamic response.
Suh et al.  reported that a loss of function genetic variant of CYP3A5 was associated with a diminished pharmacodynamic response to clopidogrel in the presence of the CYP3A4 inhibitor itraconazole. This polymorphism was associated with an increased risk of atherothrombotic events in patients treated with clopidogrel following percutaneous coronary intervention, suggesting that it is a clinically relevant determinant of response to clopidogrel . In line with the lack of effect seen in this report, decreased function of CYP3A5 had no effect on the response to clopidogrel in the absence of itraconazole. Angiollilo et al.  reported on an intronic sequence variant in CYP3A4 that was associated with the magnitude of platelet activation following administration of clopidogrel. Taken together, these data suggest that polymorphisms associated with decreased function of CYP2C9, CYP2C19 or CYP3A4/5 may each contribute to a decreased response to clopidogrel.
There are several aspects of this study that need to be considered. The reported polymorphisms for CYP2C9 have a variable effect on metabolic activity, with the effect being dependent on the enzyme substrate (drug) . We grouped the genotypes that have been characterized as moderately reduced and very low activity as ‘reduced activity’ genotypes in order to improve the power for assessment of this association. When grouped on this functional basis, there was a significant relationship between CYP2C9 function and exposure to clopidogrel. However, these findings should be confirmed in a larger population so that the individual genotypes can be adequately assessed.
We did not observe decreased function polymorphisms of CYP1A2, CYP2B6 and CYP3A4 in this population and thus could not assess the potential relationship between decreased activity of these CYPs and thienopyridine response. Consistent with the findings in this study, decreased function variants of these CYPs are uncommon and unlikely to be a major contributing factor to the variation in response to clopidogrel.
This was a retrospective analysis on a relatively small cohort of subjects from a relatively narrow geographic region. Consequently, the relationships between these genotypes and exposure to the active metabolite of clopidogrel and corresponding pharmacodynamic effect should be confirmed in larger cohorts including more diverse ethnic backgrounds. The subjects evaluated in this study were healthy and not on any other medication. Confirmation of these findings in patients receiving multiple medications, including aspirin and other CYP-metabolized drugs, as well as higher doses of clopidogrel, would be most helpful. Finally, it would be helpful to evaluate the relationship between these genetic variations in CYP function and clinical outcomes, analogous to the study of Suh et al. .
In summary, genetic variations in CYP2C19 and CYP2C9, which are associated with decreased functional activity, are associated with decreased exposure to the active metabolite of clopidogrel, but not prasugrel. These polymorphisms provide an explanation for many, but not all, cases of a poor pharmacodynamic response to clopidogrel. Whether these polymorphisms are also linked to clinical outcomes remains to be established.
J.T. Brandt, primary author, did the clinical study design, data review and interpretation. S.L. Close, secondary author, did the design of genomic analyses, data review and interpretation, review and editing. S.J. Iturria, primary statistician, did the statistical analysis, data interpretation, review and editing; C.D. Payne did the clinical study design and operations, data acquisition and interpretation, review and editing; N.A. Farid did the pharmacokinetic assays, pharmacokinetic data interpretation, review and editing; C.S. Ernest II did the pharmacokinetic analysis, pharmacokinetic data interpretation, review and editing; D.R. Lachno did the clinical study design and operations, design of genomic analyses, data acquisition and interpretation, review and editing; D. Salazar did the clinical study design, data interpretation, review and editing; K.J. Winters, senior author, did the clinical study design, data acquisition and interpretation, review and editing.
The authors gratefully acknowledge the input of L. Shen, N. Mukhopadhyay, B. Baker, J. Clemens and D. Marshall in the development of this manuscript and the editorial assistance of S. Sipowicz. The authors are indebted to the entire Prasugrel study team for their support during this project.
Disclosure of Conflict of Interests
J.T. Brandt, S.L. Close, S.J. Iturria, C.D. Payne, N.A. Farid, C.S. Ernest II, D.R. Lachno and K.J. Winters are employees and shareholders of Eli Lilly and Co. D. Salazar is an employee and shareholder of Daiichi-Sankyo, Inc.