Clopidogrel, in combination with aspirin, reduces the risk of cardiovascular events in patients with acute coronary syndromes (ACS) and in those undergoing percutaneous coronary intervention (PCI) (1- 3). However, the pharmacokinetics and pharmacodynamics of clopidogrel vary widely among individuals (4- 7). Single-center studies using a candidate gene approach have identified associations between reduced-function variants of the cytochrome 450 (CYP) 2C19 isozyme and on-treatment reactivity (OTR) while receiving clopidogrel (8- 13), and a genome-wide association study identified the reduced-function CYP2C19*2 allele as the major determinant of adenosine diphosphate–induced platelet aggregation in healthy Amish subjects treated with clopidogrel (14). These findings are consistent with a 2-step oxidative process required for transformation of clopidogrel to its labile active metabolite, in which CYP2C19 contributes substantially to each step (15). Candidate gene studies have also detected associations between other single nucleotide polymorphisms (SNPs) and clopidogrel pharmacodynamics and outcomes after PCI, in particular the 3435 C→T polymorphism of the ABCB1 gene that encodes the intestinal efflux transporter P-glycoprotein (16- 18), and the Q192R polymorphism of the PON1 gene that encodes the esterase paraoxonase-1 (19). The latter observation has led to a suggested alternative pathway of clopidogrel active metabolite formation that does not significantly involve CYP2C19, challenging the mechanistic basis underlying the association between reduced-function CYP2C19 allele carriage and clinical outcomes in ACS and PCI patients treated with clopidogrel (18,20- 24). However, the clinical impact of the ABCB1 and PON1 polymorphisms on clopidogrel efficacy has been inconsistent (25- 26), and therefore their effect on clopidogrel pharmacodynamics in patients undergoing PCI requires further validation.

A higher clopidogrel maintenance dose (MD) could potentially provide a more intense antiplatelet effect in patients with a genetic predisposition to a diminished response to standard-dose therapy by providing greater substrate for biotransformation into the active metabolite. However, the relationships between common genetic polymorphisms and clopidogrel pharmacodynamics using different dosing strategies have not been examined in the setting of a large clinical trial of patients undergoing PCI, where this antiplatelet drug is mostly used. The objective of the current study was to assess the genetic determinants of clopidogrel response variability during the acute and maintenance phases of therapy in a multicenter, randomized, double-blind trial of high-dose compared with standard-dose clopidogrel after PCI.

Patients were enrolled as part of a pre-specified genetic substudy of the GRAVITAS (Gauging Responsiveness with A VerifyNow assay–Impact on Thrombosis And Safety) trial. The study design, entry criteria, baseline characteristics, and primary results of this trial were described previously (27- 28). The GRAVITAS trial was a prospective, multinational, multicenter, double-blind, randomized, active-controlled trial. Patients with stable coronary artery disease or ACS who had undergone PCI with at least 1 drug-eluting stent and on a specified periprocedural clopidogrel dosing regimen underwent platelet function assessment with the VerifyNow P2Y12 test 12 to 24 h after PCI. Patients with high on-treatment platelet reactivity, defined as ≥230 P2Y12 reaction units (PRU), were randomly assigned by an interactive voice response system to treatment with high-dose clopidogrel (600 mg loading dose followed by a 150-mg MD) or standard-dose clopidogrel (no additional loading dose, 75-mg MD). A randomly selected cohort of patients with OTR <230 PRU were selected by the interactive voice response system to be followed in a blinded fashion and treated with a 75-mg MD of clopidogrel, and the remainder exited the study after platelet function testing per protocol. In those treated with study drug, platelet function was reassessed in a blinded fashion 30 ± 7 days and 180 ± 14 days after enrollment.

The genetic substudy was conducted at 48 participating sites, was approved by all institutional review boards, and written informed consent was obtained from all participants. All patients who were entered into the GRAVITAS trial interactive voice response system were eligible to be enrolled in the genetic substudy. A total of 1,170 patients consented and provided blood samples for genotyping either at the time of screening platelet function assessment 12 to 24 h after PCI or during follow-up (Figure 1). Patients were excluded from the analysis if they were not of Caucasian ancestry (self-identified) to avoid potential effects of population stratification.

Forty-three SNPs of 17 genes related to platelet reactivity and/or previously implicated for having an effect on clopidogrel responsiveness were genotyped using the MassARRAY platform (Sequenom, San Diego, California). Polymerase chain reaction assays and extension primers for these SNPs were designed using the MassARRAY design software, version 3.1. SNPs were genotyped using the iPLEX Gold assay (Sequenom, Inc., San Diego, California), based on multiplex polymerase chain reaction, followed by a single base primer extension reaction. The masses of the primer extension products correlating with genotype were determined using matrix-assisted laser desorption/ionization time-of-flight mass spectroscopy. Final genotypes were called using the MassARRAY Typer, version 4.0. Seven samples were removed due to low call rates (sample call cutoff = 90%). The mean call rate across the 43 SNPs was 99.6% (range, 93.3% to 100%). Hardy-Weinberg equilibrium was tested in a cohort of 611 patients who consented and had their genetic sample drawn at screening 12 to 24 h after PCI (i.e., patients who were and were not selected to be followed on study drug) because the distribution of OTR in this cohort should better reflect the general PCI population. Equilibrium was observed for 41 of the 43 SNPs, and therefore these were included in subsequent analyses (Table 1).

The primary pharmacodynamic endpoint was OTR. Pre-specified analyses included the relationship between SNPs and OTR at screening 12 to 24 h after PCI and at 30-day and 6-month follow-up; the change in OTR from screening to 30 days; and the rate of high OTR, defined as ≥230 PRU. The primary efficacy endpoint was a composite of death from cardiovascular causes, nonfatal myocardial infarction, or stent thrombosis. A clinical events committee blinded to treatment assignment and genotype and independent of the sponsor adjudicated all suspected primary efficacy end points, as previously described (28).

SNPs were tested for their association with OTR by linear regression using a codominant model after adjusting for clinical characteristics associated with OTR at an α = 0.05 level. These covariates were determined separately for the analyses of OTR at the time of randomization (12 to 24 h after PCI) and at 30 days; 6-month analyses used the same covariates as at 30 days. Bonferroni-adjusted p value cutoffs were used to determine statistical significance. Associations between the 41 genotyped SNPs and on-treatment platelet reactivity were thus evaluated at the α = 0.0012 level of significance (i.e., 0.05 ÷ 41). Multivariate generalized linear regression models were built for significant polymorphisms that included relevant clinical characteristics as fixed covariates. The specific portion of variance explained by genotype and clinical characteristics were determined by calculating the partial η² from this model. The population-attributable risk of persistently high reactivity was determined by subtracting the incidence of high reactivity at 30 days in noncarriers from the incidence in carriers and noncarriers combined. Categorical variables are reported as counts (percentages) and continuous variables as mean ± SD where appropriate. Because this was a substudy, a target sample size was not pre-specified. The sample size that was achieved provided 90% power to detect an effect size (f²) of 0.021 and 80% power to detect an effect size of 0.017 with α = 0.001. Analyses were performed with SAS version 9.2 (SAS Institute Inc., Cary, North Carolina). The sponsors had no role in the collection, management, analysis, or interpretation of the data or in the preparation, review, or approval of this paper.

A total 1,170 patients provided DNA samples over the course of the study, of which 1,028 were included in the analysis of OTR 12 to 24 h after PCI (Figure 1). Their mean age was 65.3 ± 10.5 years, 301 patients (29.3%) were women, 286 patients (27.8%) had diabetes mellitus, 118 patients (11.5%) presented with ACS, and 450 (44%) displayed normal OTR. A total of 741 patients were assigned to and followed on the study drug: 285 patients displayed high OTR and were randomly assigned to high-dose clopidogrel, 293 patients displayed high OTR and were randomly assigned to standard-dose clopdiogrel, and 157 patients displayed normal OTR and were treated with standard-dose clopidogrel. Repeat platelet function assessment on the study drug was available at 30 days in 702 patients (95%) and at 6 months in 672 patients (91%).

(Table 1) shows the relationships between the candidate SNPs and OTR 12 to 24 h after PCI adjusted for clinical covariates significantly associated with OTR according to univariate analysis (age, sex, body mass index, current smoking, creatinine clearance <60 ml/min, diabetes mellitus, history of congestive heart failure, hypertension, or hyperlipidemia). The CYP2C19*2 reduced-function allele was significantly associated with OTR (R² = 0.07, p = 2.2 × 10⁻¹⁵), whereas the PON1 Q192R, ABCB1 3435 C→T, and CYP2C19*17 alleles were not (R² = 0.002, p = 0.42; R² < 0.001, p = 0.97; and R² = 0.005, p = 0.08, respectively). There was a strong trend for an association between the CYP2B6*1B allele and OTR (R² = 0.013, p = 0.0017). No interaction was observed between the CYP2C19 and CYP2B6 genotypes (p = 0.11).

Carriers of reduced-function CYP2C19 alleles present in the study population (i.e., *2, *3, *4, *6, and *8) were pooled for further analyses. Compared with noncarriers, the risk of high OTR 12 to 24 h after PCI was 2.5-fold greater for carriers of 1 reduced-function CYP2C19 allele and approximately 4.5-fold greater for carriers of 2 reduced-function CYP2C19 alleles (Figure 2).

The CYP2C19*2 allele was significantly associated with OTR at 30 days (R² = 0.10, p = 1.3 × 10⁻¹⁷) and at 6 months (R² = 0.07, p = 1.9 × 10⁻¹¹). Carriage of PON1 Q192R and ABCB1 3435 C→T were not associated with OTR at 30 days or 6 months; an association between CYP2C19*17 carriage and reduced levels of OTR became apparent at 30 days (R² = 0.022, p = 0.0004), although the association did not reach the threshold of significance at 6 months (R² = 0.015, p = 0.01) (Table 1). Body mass index, diabetes mellitus, age, and current smoking also contributed significantly to the variance in OTR at both time points in the multivariate model incorporating the CYP2C19 genotype (Table 2).

(Figure 3) shows the change in OTR from randomization to 30 days according to treatment arm stratified by the CYP2C19 genotype. After adjustment for treatment assignment, characteristics significantly associated with the change in reactivity from randomization to 30 days on multivariate analysis were CYP2C19 genotype (η² = 0.054, p < 0.0001), age (η² = 0.02, p = 0.001), and body mass index (η² = 0.006, p = 0.032).

CYP2C19 reduced-function allele carriage was associated with a substantially greater risk of high OTR at 30 days and 6 months, particularly for carriers of 2 reduced-function alleles, irrespective of maintenance dosing strategy (Figure 4). Among the patients with high OTR after PCI who were randomly assigned to clopidogrel 150-mg MD, the portion of the risk of persistently high reactivity attributable to reduced-function allele carriage (population-attributable risk) was 5.2%. The population-attributable risk was 7.3% using a cut point of PRU >208 for persistently high reactivity. The CYP2C19 genotype was also significantly associated with a greater risk of persistently high reactivity during follow-up among the patients with high OTR randomly assigned to standard-dose clopidogrel (p = 0.0002) (Figure 4).

The primary clinical endpoint of death due to cardiovascular causes, nonfatal myocardial infarction, or stent thrombosis occurred in 14 patients (1.9%) during follow-up. Carriers of 2 reduced-function alleles were at significantly higher risk of the primary endpoint compared with noncarriers (1 event in 24 patients at-risk compared with 11 events in 466 patients at-risk; hazard ratio: 1.58; 95% confidence interval [CI]: 1.04 to 2.41; p = 0.03). There was no association between carriage of 1 reduced-function CYP2C19 allele and the primary endpoint (2 events in 244 patients at risk; hazard ratio: 1.07; 95% CI: 0.91 to 1.25; p = 0.42). There was no association between PON1 Q192R carriage and clinical outcomes (carriers of 1 allele compared with noncarriers: odds ratio: 0.97; 95% CI: 0.83 to 1.14; p = 0.73; carriers of 2 alleles compared with noncarriers: odds ratio: 0.90; 95% CI: 0.69 to 1.17; p = 0.44).

In this prospective substudy of a large, randomized, multicenter clinical trial, we tested the association between OTR and several gene variants that plausibly could be related to interindividual variability in adenosine diphosphate–induced platelet reactivity in patients treated with clopidogrel, including those encoding the CYP450 enzyme family, cell surface receptors, kinases, and other signaling molecules (29). This study provides strong evidence that: 1) CYP2C19 is a key genetic determinant of OTR while receiving clopidogrel early and late after PCI; 2) identification of the CYP2C19 genotype in patients with high OTR according to platelet function testing provides only limited information regarding the risk of persistently high reactivity with clopidogrel 150-mg MD; and 3) it is unlikely that PON1 or ABCB1 contributes significantly to clopidogrel response variability and, by extension, to clopidogrel metabolism.

Despite the clinical risk that appears to be imparted by the CYP2C19 genotype in PCI patients treated with clopidogrel (21- 22), the pharmacodynamic effect of CYP2C19 among this patient population has not been previously examined in a prospective, multicenter setting. Our finding of a very robust association between the CYP2C19 genotype and OTR is strengthened by several factors: the moderate-to-large effect size (R² between 0.065 and 0.10 during follow-up, adjusted odds ratios >2.5 for high OTR in carriers of at least 1 reduced-function allele); the magnitude of the p value (<1 × 10¹⁰); statistical correction for multiple comparisons; and the low likelihood of population stratification given the homogeneous ethnicity of the analysis cohort. We thereby avoided common pitfalls of candidate gene studies that can lead to false-positive reporting. We believe that these results provide definitive evidence supporting a pharmacodynamic basis for the reported association between CYP2C19 and cardiovascular outcomes. Our findings that, after adjustment for clinical and procedural characteristics, reduced-function CYP2C19 allele carriage explains approximately 10% of clopidogrel response variability in the long-term phase of therapy at 30 days and 7% at 6 months extend previous observations of a similar contribution in the acute setting (14,30), and supports the continued influence of CYP2C19 on clopidogrel pharmacodynamics over time.

In the GRAVITAS trial, high-dose clopidogrel compared with standard-dose clopidogrel did not reduce the incidence of cardiovascular events among patients with high OTR after PCI (28). An insufficient pharmacodynamic response may have contributed in part to the lack of clinical effectiveness observed with high-dose therapy (31). The current study demonstrates that among patients with high OTR according to platelet function testing after PCI, CYP2C19 is a key determinant of a persistently heightened level of OTR with a standard clopidogrel MD and a diminished antiplatelet effect with a high clopidogrel MD. However, in patients randomly assigned to the 150-mg MD of clopidogrel, the portion of the risk of persistently high OTR at 30 days attributable to reduced-function allele carriage was only 5% using the pre-specified definition of high reactivity (>230 PRU) and only 7% using a post hoc definition (>208 PRU) that has been associated with subsequent cardiovascular events (31).

The association that we observed between the CYP2C19*17 allele and an increased antiplatelet effect of clopidogrel was inconsistent, with a significant relationship noted at 30 days but not reaching the threshold for significance after PCI or at 6-month follow-up. Although some observational studies have shown increased platelet inhibition in carriers of the so-called gain-of-function CYP2C19*17 allele (32- 33), other studies have shown no effect (14,21,34- 36). The findings of clinical outcome studies are similarly inconsistent (18,25,32,37- 38). The comparative strength of the association of CYP2C19*2 in our cohort further supports the supposition that reduced-function alleles play the dominant genetic role in determining on-clopidogrel platelet reactivity (36), and, by extension, in influencing clinical outcomes.

We investigated the influence of polymorphisms of the ABCB1 gene, which encodes the P-glycoprotein efflux transporter believed to mediate the intestinal absorption of clopidogrel (16). Data regarding the effect of the ABCB1 3435 C→T polymorphism on clopidogrel pharmacodynamics is limited and inconsistent (14,17,35- 36). The impact of the ABCB1 genotype on clinical outcomes after PCI has also been discordant between studies (17- 18,25). In the present study, none of the common polymorphisms of ABCB1, including 3435 C→T, influenced OTR after PCI, at 30 days, or at 6 months, thereby challenging the biological plausibility of an effect of the ABCB1 genotype on clinical outcomes in clopidogrel-treated patients in the acute or maintenance phases.

We assessed the influence of PON1, which encodes the paroxonase-1 enzyme that has been proposed to mediate the rate-limiting step in clopidogrel metabolism (19). A case-cohort and replication study observed that the PON1 Q192R polymorphism was associated with lower paraoxonase-1 plasma activity, lower active metabolite levels, lower platelet inhibition, and a 10-fold greater risk of stent thrombosis in PCI patients receiving clopidogrel (19). We did not detect any influence of PON1 Q192R on the antiplatelet effect of clopidogrel, either in the acute or chronic phase of therapy after PCI. Therefore, according to our findings, there does not appear to be a pharmacodynamic basis for a clinical effect of PON1 on cardiovascular outcomes in clopidogrel-treated patients. A retrospective case-cohort study also did not confirm an effect of PON1 Q192R in patients with stent thrombosis (26), and a retrospective, post hoc single-center study of elective PCI patients also did not observe a relationship between the PON1 genotype and the effect of clopidogrel on residual platelet aggregation by light transmittance aggregometry (39).

On-treatment reactivity while receiving clopidogrel has been associated with clinical outcomes in a patient-level meta-analysis and in a pharmacodynamic analysis of the GRAVITAS trial (31,40). The findings of the current study have several implications regarding optimal antiplatelet therapy after PCI. The CYP2C19 genotype is predictive of high OTR in both the early and late phases after PCI. Carriers of 2 reduced-function CYP2C19 alleles are highly likely to display levels of OTR associated with adverse outcomes, and a 150-mg MD of clopidogrel had a particularly marginal pharmacodynamic effect in these patients. Therefore, in patients who are identified to be poor metabolizers by genotyping, the use of alternative P2Y₁₂ antagonists that are not influenced by CYP2C19 should be considered if an adequate pharmacodynamic effect is desired. In patients who have undergone platelet function testing and are thereby determined to be at higher risk of ischemic events, additional CYP2C19 genotyping provides little help in identifying who will adequately respond to a 150-mg MD of clopidogrel. Whether even larger doses of clopidogrel may provide an improved response in reduced-function CYP2C19 allele carriers is being examined in the ELEVATE–TIMI 56 (Escalating Clopidogrel by Involving a Genetic Strategy–Thrombolysis In Myocardial Infarction 56) trial (NCT01235351).

A potential limitation of the present study is that the GRAVITAS trial, by design, preferentially enrolled patients with higher levels of OTR. This selection bias potentially reduced the power of our association tests. Procedural characteristics and concomitant medications, including proton pump inhibitor use, were not available for the entire cohort at randomization, but were available and included in the adjusted analyses at follow-up. We did not examine the pharmacodynamic effect of CYP2C19 on different clopidogrel loading doses, although our findings are consistent with those reported by a single-center case-control study (41). The study was not powered to detect the influence of genotype on clinical events, and the significant association between carriage of 2 reduced-function CYP2C19 alleles and outcomes must be considered exploratory and hypothesis generating, although it is consistent with previous reports (18,21- 22).

The findings of this pharmacogenetic analysis of a multicenter, randomized clinical trial demonstrate a strong and consistent association between the CYP2C19 genotype and OTR while receiving clopidogrel after PCI, present during both the early and long-term phases after the procedure and irrespective of MD. In patients with high OTR identified by platelet function testing, the CYP2C19 genotype provides only limited incremental information regarding the risk of persistently high reactivity with a treatment strategy of clopidogrel 150 mg daily. Our findings do not support a pharmacodynamic basis for the influence of PON1 or ABCB1 on clinical outcomes after PCI in clopidogrel-treated patients.

The authors acknowledge Judy Sheard of the Scripps Translational Science Institute for her assistance in the management of this study.

A. Abbas, M. Amine, D. Angiolillo, R. Applegate, J. Aragon, H. Aronow, W. Batchelor, P. Berger, M. Buchbinder, L. Cannon, N. Chronos, D. Cohen, E. Fry, M. Fugit, R. Gammon, P Gordon, K. Garratt, A. Jacobs, M.Z. Jafar, M. Lucca, M. Lurie, E. Mahmud, A. Malik, T. Mann, S. Manoukian, S. Marshalko, T. McGarry, B. McLaurin, J.C. Merrit, R. Minutello, V. Misra, E.D. Nukta, C. O'Shaughnessy, S. Plante, D. Purdy, S. Puri, D. Rizik, M. Robbins, J. Saucedo, M. Schweiger, D. Spriggs, R. Stoler, P. Teirstein, M. Turco, R. Waksman, S. Ward, M. Warshofsky, G. Wong.