Reduced-Function CYP2C19 Genotype and Risk of Adverse Clinical Outcomes Among Patients Treated With Clopidogrel Predominantly for PCI: A Meta-Analysis

Abstract

Content

Clopidogrel, one of the most commonly prescribed medications, is a pro-drug requiring CYP450 biotransformation. Data suggest its pharmacologic effect varies based on CYP2C19 genotype, but there is uncertainty regarding the clinical risk imparted by specific genotypes.

Objective

In patients treated with clopidogrel, to define the risk of major adverse cardiovascular outcomes among carriers of one (∼26% prevalence in whites) and carriers of two (∼2% prevalence in whites) reduced-function CYP2C19 variants.

Data Sources and Study Selection

A literature search was conducted (January 2000-August 2010) of the MEDLINE, Cochrane, and EMBASE databases. Genetic studies were included where clopidogrel was initiated in predominantly invasively managed patients in a manner consistent with the current guideline recommendations and where clinical outcomes were ascertained.

Data Extraction

Investigators from nine studies evaluating CYP2C19 genotype and clinical outcomes in patients treated with clopidogrel contributed the relevant hazard ratios (HRs) and their 95% confidence intervals (CI) for specific cardiovascular outcomes by genotype.

Results

Among 9685 patients [91.3% of whom underwent percutaneous coronary intervention (PCI) and 54.5% of whom had an acute coronary syndrome (ACS)], 863 experienced the composite endpoint of cardiovascular death, myocardial infarction, or stroke; 84 patients had stent thrombosis among the 5894 evaluated for such. Overall, 71.5% were non-carriers, 26.3% had one, and 2.2% had two CYP2C19 reduced-function alleles. A significantly increased risk of the composite endpoint was evident in both carriers of one (HR 1.55, 95% CI 1.11-2.27, P=0.01) and two (HR 1.76, 95% CI 1.24-2.50, P=0.002) CYP2C19 reduced-function alleles. Similarly, there was a significantly increased risk of stent thrombosis in both carriers of one (HR 2.67, 95% CI 1.69-4.22, P<0.0001) and two (HR 3.97, 95% CI 1.75-9.02, P=0.001) CYP2C19 reduced-function alleles.

Conclusion

Among patients treated with clopidogrel for PCI, carriage of even one reduced-function CYP2C19 allele appears to be associated with a significantly increased risk of major adverse cardiovascular events, particularly stent thrombosis.

Introduction

Clopidogrel blocks the P2Y₁₂ ADP receptor on platelets and has been shown to reduce cardiovascular events in patients presenting with an acute coronary syndrome (ACS), particularly in those undergoing percutaneous coronary intervention (PCI).^{1, 2} However, there is a large degree of inter-individual variability in the pharmacodynamic response to clopidogrel.³ One source of the variability is the metabolism of clopidogrel, which is a pro-drug requiring biotransformation to generate its active metabolite. Cytochrome P-450 (CYP) isoenzymes, specifically CYP2C19,⁴ play a key role in clopidogrel metabolism and carriers of reduced-function genetic variants in CYP2C19 have lower active clopidogrel metabolite levels and diminished platelet inhibition.⁵

Based in part on a pharmacokinetic and pharmacodynamic study in 40 healthy subjects, the United States Food and Drug Administration (US FDA) announced a boxed warning on Plavix (clopidogrel) stating that the drug has a diminished effect in individuals based on their CYP2C19 genotype, specifically in those who harbor two CYP2C19 reduced-function alleles.^{6, 7} Yet, there is not consensus as to whether the diminished pharmacologic response translates into worse clinical outcomes and whether the proposed increased risk of adverse cardiovascular outcomes requires two CYP2C19 reduced-function alleles (present in approximately 2% of the white population) or can be seen with just one (present in approximately 26% of the white population).⁸

Individual clopidogrel pharmacogenetic studies have reported somewhat divergent results, and the confidence intervals corresponding to the hazard ratios for clinical events across different genotypes are sufficiently wide so as to not be able to address reliably the aforementioned issues. Moreover, to date, a number of studies have not generated data separately for carriers of one and for carriers of two reduced-function CYP2C19 alleles. Therefore, to define the risk of major adverse cardiovascular events in carriers of one and in carriers of two reduced-function CYP2C19 alleles, the investigators for each participating study agreed to perform a collaborative meta-analysis. In totality, we were able to examine the association of CYP2C19 genotype and clinical outcomes in 9685 patients who initiated guideline-recommended treatment with clopidogrel, predominantly for PCI.

Methods

Data Sources and Study Selection Criteria

A computerized literature search was conducted from January 2000 through August 2010 of the MEDLINE, Cochrane, and EMBASE databases by using search terms that included clopidogrel and CYP2C19. In addition, experts in the field were contacted and abstracts from major cardiology meetings were reviewed. The meta-analysis included studies (both cohort studies and clinical trials) where clopidogrel was initiated in predominantly invasively managed patients in a manner consistent with the current guideline recommendations.^{1, 2} Studies were excluded if they did not include clinical outcomes measurements.

A total of 31 studies were identified as potentially relevant and were screened for inclusion (Supplemental Table 1). Of these studies, 22 were subsequently excluded because they only provided pharmacodynamic or pharmacokinetic data (ie, no clinical outcomes data); data could not be extracted from what was presented in the main manuscript; or because they included patients from different clinical populations (eg, in whom treatment with clopidogrel was not guideline indicated or who were predominantly conservatively. Of the nine studies that were eligible and invited, all investigators agreed to provide study-level data and participate in a collaborative meta-analysis.^9-17 Study quality was assessed independently based on elements from the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist¹⁸ by two authors (JLM and MSS) with disagreements resolved by consensus.

Genotype Data, End Points, and Data Compilation

For each study, patients treated with clopidogrel were classified as carriers of zero (i.e., non-carriers or wildtype), one (i.e., heterozygotes), or two (e.g., homozygotes or compound heterozygotes) CYP2C19 (NCBI Genome build 37.1, NG_008384) reduced-function alleles. Of the reduced-function alleles, CYP2C19*2 is the most frequent variant (accounting for about 95% of the reduced-function allele carrier status).¹¹ In seven studies, patients were categorized on the basis of CYP2C19*2 (rs4244285) alone;^{9, 10, 12, 14-17} in one study, CYP2C19*2, *3 (rs4986893), *4 (rs28399504), and *5 (rs56337013);¹³ and in one study CYP2C19*2, *3, *4, *5, and *8 (rs41291556) (Supplemental Methods).¹¹

The investigators for each participating study provided the incidence of cardiovascular death, myocardial infarction, and ischemic stroke, as well as the composite of these end points across CYP2C19 genotypes in 9685 individuals. Investigators provided the hazard ratios (HRs) and 95% confidence intervals (CIs) for these end points for carriers of at least one, only one, or two CYP2C19 reduced-function alleles compared with non-carriers, with adjusted hazard ratios provided based on the investigators' determination of the need to do so in their main publication. Additionally, six of the nine studies evaluated stent thrombosis, and thus analogous data pertaining to the risk of definite or probable stent thrombosis, as defined by the Academic Research Consortium (ARC) criteria,¹⁹ were provided for 5894 subjects.^{10-12, 14-16} Outcomes were collected from 0 days to end of follow-up, as well as from 0 to 30 days and 31 days to end of follow-up. See Supplemental Methods for study-specific details. All of the provided data were verified by each of the participating investigators.

Statistical Analyses

We performed a meta-analysis combining HRs for each study using a random-effects model which considers both within-study and between-study variation (given the differences in study populations characteristics and clopidogrel dosing and therefore the potential differences in the pharmacogenetic association) with weighting based on inverse variance.²⁰ Results are presented as HRs with their 95% CIs. Heterogeneity of risk was diagnosed using the Q statistic and the degree assessed using the I² measure (which reflects the percentage of the total variability due to between-study heterogeneity versus within study variability). If heterogeneity was found, we then performed sensitivity analyses, serially excluding studies to determine the source. Additionally, sensitivity analyses were conducted examining for heterogeneity on the basis of ACS status and clopidogrel dosing. Comprehensive Meta Analysis version 2.2.048 was used for the analyses and a P value of <0.05 was set as the level of significance with no correction for multiple hypothesis testing given the inter-relatedness of the hypotheses.

Results

Overall, 9685 patients from nine studies contributed to the cardiovascular death, myocardial infarction, or stroke analysis (Table 1). The average age of patients was 64.2 years and 7204 patients (74.4%) were male. A total of 8847 (91.3%) patients underwent PCI and 5278 (54.5%) had an acute coronary syndrome. There were 6923 patients (71.5%) with no CYP2C19 reduced-function alleles (i.e., non-carriers or wildtype), 2544 (26.3%) with one CYP2C19 reduced-function allele (i.e., heterozygotes), and 218 (2.2%) with two CYP2C19 reduced-function alleles (e.g., homozygotes or compound heterozygotes). There were no significant differences in baseline characteristics across genotypes (Table 2). Six studies included stent thrombosis as an endpoint, and thus 5894 stented patients were included in the stent thrombosis analyses. In this subset, there were 4220 patients (71.6%) with no CYP2C19 reduced-function alleles, 1535 (26.0%) with one CYP2C19 reduced-function allele, and 139 (2.4%) with two CYP2C19 reduced-function alleles.

Table 1. Characteristics of the Studies Included in the Meta-analysis.

	CLARITY-TIMI 289	EXCELSIOR10	TRITON-TIMI 3811	AFIJI12	FAST-MI13	RECLOSE14	ISAR15	CLEAR-PLATELETS16	INTER MOUNTAIN17	Total
Baseline Characteristics
Age (yrs±SD)	59.9±10.5	66.4±9.1	60.1±11.1	40.1±5.1	66.2±13.7	68.3±11.0	66.5±10.2	64.2±11.5	63.0±11.5	64.2
Male	78.4% (178/227)	78.0% (622/797)	70.5% (1029/1459)	92.3% (239/259)	70.6% (1559/2208)	74.6% (576/772)	78.3% (1946/2485)	60.1% (137/228)	73.4% (918/1250)	74.4%
Diabetes	17.0% (38/223)	18.8% (150/797)	21.9% (319/1459)	10.4% (27/259)	31.7% (698/2202)	22.2% (171/772)	35.5% (881/2485)	37.7% (86/228)	28.3% (354/1250)	28.1%
Smoking	43.6% (99/227)	10.9% (87/797)	38.1% (556/1459)	56.0% (145/259)	31.4% (691/2202)	34.5% (266/772)	16.2% (402/2485)	25.4% (58/228)	17.6% (220/1250)	26.1%
White Race	85.0% (193/227)	100.0% (797/797)	97.8% (1427/1459)	78.0% (202/259)	n/a*	100.0% (772/772)	n/a**	61.4% (140/228)	100.0% (1250/1250)	95.8%
PCI	58.1% (132/227)	100.0% (797/797)	100.0% (1459/1459)	73.0% (189/259)	69.5% (1535/2208)	100.0% (772/772)	100.0% (2485/2485)	100.0% (228/228)	100.0% (1250/1250)	91.3%
ACS	100.0% (227/227)	0.0% (0/797)	100.0% (1459/1459)	100.0% (259/259)	53.2% (1174/2208)	70.3% (543/772)	34.0% (846/2485)	0.0% (0/228)	61.6% (770/1250)	54.5%
CYP2C19 Reduced-Function Alleles
None (n)	150	554	1064	186	1,573	525	1,805	160	906	6,923
One (n)	73	226	357	64	577	221	633	63	330	2,544
Two (n)	4	17	38	9	58	26	47	5	14	218
Study Information
Year Published	2008	2008	2009	2009	2009	2009	2009	2009	2009
Clopidogrel Dose (Loading), mg	300	600	300	600	300-600	600	600	300-600	300-600
Median (IQR) FU, days	30 (30-30)	180 (180-180)	445 (356-455)	391 (102-1095)	365 (365-365)	180 (180-180)	30 (30-30)	180 (180-365)	365 (365-365)
Max FU, days	30	180	457	2920	365	180	30	365	365
Complete Data FU	100.0%	99.1%	99.3%	100.0%	99.4%	100.0%	100.0%	100%	100%***

^*The Investigators for FAST-MI note that in the subjects in whom race was collected, virtually all were white.

^**The Investigators for ISAR note that the patients enrolled in the study were recruited from a single center in Germany with a near exclusively white population.

^***For mortality; completeness for composite endpoint not available. All studies were deemed of comparable quality as they met key elements of the STROBE checklist.¹⁸

ACS indicates acute coronary syndrome; FU, follow-up; PCI, percutaneous coronary intervention.

Table 2. Pooled Baseline Characteristics by CYP2C19 Genotype Status.

		CYP2C19 Reduced-Function Alleles
	Overall (N=9685)	None (N=6923)	One (N=2544)	Two (N=218)
Weighted Age (yrs)	64.2	64.1	64.6	63.7
Male Sex	7204 (74.4%)	5180 (74.8%)	1852 (72.8%)	172 (78.9%)
Diabetes	2724 (28.1%)	1926 (27.8%)	739 (29.0%)	58 (27.1%)
Current Smoker	2524 (26.1%)	1821 (26.3%)	648 (25.5%)	55 (25.2%)
ACS at Presentation	5278 (54.5%)	3820 (55.2%)	1339 (52.6%)	119 (54.6%)
PCI at Presentation	8847 (91.3%)	6336 (91.5%)	2316 (91.0%)	195 (89.4%)
White Race*	4781 (95.8%)	3399 (95.9%)	1277 (95.7%)	105 (92.9%)

Age presented as a weighted mean; the other data is presented as n (%). There were no significant differences for the categorical variables across CYP2C19 genotype. ACS indicates acute coronary syndrome; PCI, percutaneous coronary intervention.

^*Data on race (self-reported) was not captured uniformly in ISAR and FAST-MI; therefore the denominator for the overall, none, one, and two CYP2C19 reduced-function allele groups is 4992, 3545, 1334, and 113.

Cardiovascular Death, MI, or Stroke by CYP2C19 Genotype

Overall, 863 of the 9685 patients experienced the composite endpoint of cardiovascular death, myocardial infarction, or ischemic stroke. Carriers of one or two reduced-function CYP2C19 alleles versus non-carriers had a significantly increased risk of the composite endpoint (HR 1.57, 95% CI 1.13-2.16, P=0.006, Figure 1, Panel A). In terms of individual endpoints, 272 patients died of a cardiovascular cause, 575 had a non-fatal myocardial infarction, and 68 had a non-fatal stroke. There was directionally consistent risk for of all the components of the composite endpoint associated with carriage of one or two reduced-function CYP2C19 alleles. Specifically, for cardiovascular death the HR was 1.84 (95% CI 1.03-3.28, P=0.041), for non-fatal myocardial infarction the HR was 1.45 (95% CI 1.09-1.92, P=0.010), and for stroke the HR was 1.73 (95% CI 0.68-4.38, P=0.25).

Figure 1. Cardiovascular Death, Myocardial Infarction, or Ischemic Stroke by CYP2C19 Genotype.

Among patients treated with clopidogrel, hazard ratios are reported for cardiovascular death, myocardial infarction, or ischemic stroke among carriers of one or two (Panel A), one (Panel B), or two (Panel C) CYP2C19 reduced-function alleles versus non-carriers. Squares with horizontal lines represent the hazard ratio and corresponding 95% confidence intervals, and the size of each square reflects the statistical weight of the study in the meta-analysis. The diamond represents the 95% confidence for the overall hazard ratio. The number of events and the number of individuals at risk for events is presented for each study. In Panel C, studies that had no adverse cardiovascular events among carriers of two reduced-function CYP2C19 alleles could not be included in the analysis.

A. Carriers of One or Two CYP2C19 Reduced-Function Alleles versus Non-Carriers

B. Carriers of One CYP2C19 Reduced-Function Alleles versus Non-Carriers

C. Carriers of Two CYP2C19 Reduced-Function Alleles versus Non-Carriers

Risk in Carriers of One and Two Reduced-Function CYP2C19 Alleles

Compared with CYP2C19 non-carriers, there was a significantly increased risk of cardiovascular death, myocardial infarction, or stroke in the 26.3% of the overall study population who carried only one CYP2C19 reduction-function allele (HR 1.55, 95% CI 1.11-2.17, P=0.01, Figure 1, Panel B). Similarly, there was a significantly increased risk of cardiovascular death, myocardial infarction, or stroke in the 2.2% of the overall study population who carried two CYP2C19 reduction-function alleles (HR 1.76, 95% CI 1.24-2.50, P=0.002, Figure 1, Panel C).

Stent Thrombosis Outcomes by CYP2C19 Genotype

Overall, stent thrombosis occurred in 84 of the 5894 patients who had a stent implanted and were followed for stent thrombosis. Carriers of one or two reduced-function CYP2C19 alleles versus non-carriers had a significantly increased risk of stent thrombosis (HR 2.81, 95% CI 1.81-4.37, P<0.00001, Figure 2, Panel A). Analogous to the observations for cardiovascular death, myocardial infarction, or stroke, both carriers of only one CYP2C19 reduced-function allele (HR 2.67, 95% CI 1.69-4.22, P<0.0001, Figure 2, Panel B) and carriers of two alleles (HR 3.97, 95% CI 1.75-9.02, P=0.001, Figure 2, Panel C) were at significantly increased risk of stent thrombosis when compared with CYP2C19 non-carriers.

Figure 2. Stent Thrombosis by CYP2C19 Genotype.

Among patients treated with clopidogrel, hazard ratios are reported for stent thrombosis among carriers of one or two (Panel A), one (Panel B), or two (Panel C) CYP2C19 reduced-function alleles versus non-carriers. Squares with horizontal lines represent the hazard ratio and corresponding 95% confidence intervals, and the size of each square reflects the statistical weight of the study in the meta-analysis. The diamond represents the 95% confidence for the overall hazard ratio. The number of events and the number of individuals at risk for events is presented for each study. In Panel C, studies that had no stent thrombosis events among carriers of two reduced-function CYP2C19 alleles could not be included in the analysis.

A. Carriers of One or Two CYP2C19 Reduced-Function Alleles versus Non-Carriers

B. Carriers of One CYP2C19 Reduced-Function Alleles versus Non-Carriers

C. Carriers of Two CYP2C19 Reduced-Function Alleles versus Non-Carriers

Timing of Events

In landmark analyses, carriers of one or two reduced-function CYP2C19 alleles versus non-carriers had a HR of 1.36 (95% CI 1.11-1.65) for cardiovascular death, myocardial infarction, or stroke over the first 30 days and a HR of 1.61 (95% CI 0.88-2.95) from 31 days until the end of follow-up (Figure 3). For stent thrombosis, carriers of one or two reduced-function CYP2C19 alleles versus non-carriers had a HR of 2.94 (95% CI 1.75-4.94) over the first 30 days and a HR of 2.80 (95% CI 0.83-9.38) from 31 days until the end of follow-up (Figure 3).

Figure 3. Timing of Events for Cardiovascular Death, Myocardial Infarction, or Ischemic Stroke and Stent Thrombosis.

Among patients treated with clopidogrel, hazard ratios are reported for cardiovascular death, myocardial infarction, or ischemic stroke and for stent thrombosis among carriers of one or two CYP2C19 reduced-function alleles versus non-carriers. Diamonds with horizontal lines represent the hazard ratio and corresponding 95% confidence intervals across the different timepoints. Nine studies contributed to the endpoint of cardiovascular death, myocardial infarction, or stroke from 0 to 30 days, and six studies from 31 days to end of follow-up. Analogously, six studies contributed to the endpoint of stent thrombosis from 0 to 30 days, and three studies from 31 days to end of follow-up. The number of events and the number of individuals at risk for events is presented for each study. In the analysis, a patient could have had a non-fatal event during 0 to 30 days and a subsequent event after day 30.

Exploring Heterogeneity between Studies and Among Subgroups

There was evidence of heterogeneity for the endpoint of cardiovascular death, myocardial infarction, or stroke when comparing carriers of only one CYP2C19 reduced-function allele with non-carriers (Q=29.22, P<0.001 for heterogeneity; I²=73%). Exclusion of two studies, FAST-MI and AFIJI, resulted in resolution of heterogeneity (Q=4.63, P=0.59 for heterogeneity, I²=0%). The characteristics of patients in these two studies were similar to those in the other seven studies, and the HRs for FAST-MI and AFIJI fell on either side of the summary HR. After excluding these studies, the summary HR for carriers of only one CYP2C19 reduction-function allele was 1.42 (95% CI 1.19-1.69), which was similar to the HR calculated from analyzing all nine studies. There was no evidence of statistically significant heterogeneity for the endpoint of cardiovascular death, myocardial infarction, or stroke when comparing carriers of two CYP2C19 reduced-function alleles with non-carriers (Q=5.87, P=0.44 for heterogeneity), nor was there any significant heterogeneity in the stent thrombosis analyses (carriers of one CYP2C19 reduced function allele vs. non-carriers: Q=4.44, P=0.49 for heterogeneity; carriers of two CYP2C19 reduced function alleles vs. non-carriers: Q=5.77, P=0.22 for heterogeneity).

There was no evidence of heterogeneity for the endpoint of cardiovascular death, myocardial infarction, or stroke or for stent thrombosis across studies that had all, some, or no patients with ACS (Q=3.82, P=0.15 and Q=1.26, P=0.53, respectively) or across studies that used only 300 mg, 300 or 600 mg, or ≥600 mg of clopidogrel as the loading dose (Q=1.61, P=0.45 and Q=0.25, P=0.88, respectively).

Comment

By performing a collaborative meta-analysis with data by genotype, we found that among patients treated with clopidogrel predominantly for PCI, carriage of even one reduced-function CYP2C19 allele is associated with an increased risk of adverse cardiovascular events, particularly stent thrombosis. Thus, CYP2C19 genetic information identifies approximately 30% or more of the population who may be less likely to be protected from recurrent ischemic events after PCI despite treatment with standard doses of clopidogrel.

Guidelines about caring for patients with CYP2C19 polymorphisms are starting to be developed at national levels.^{6, 7, 21} For example, in March of 2010, the US FDA issued a warning suggesting that there can be a diminished effect of standard doses of clopidogrel in CYP2C19 poor metabolizers, who were defined as individuals with two reduced-function CYP2C19 alleles. The proportion of the population harboring two reduced-function CYP2C19 alleles is approximately 2% for whites, 4% for blacks, and 14% for Chinese.⁸ The FDA referenced a crossover study of 40 healthy subjects, who were treated with 300 mg of clopidogrel followed by 75 mg per day and with 600 mg followed by 150 mg per day. The study found that individuals with two reduced-function CYP2C19 alleles, as compared with carriers of one or none, exhibited substantially decreased active drug metabolite levels and inhibition of platelet aggregation. However, it should be noted that a number of other pharmacokinetic and pharmacodynamic studies have also found that individuals (including both healthy persons and, perhaps more germane, patients with coronary artery disease) who carry even one reduced-function CYP2C19 allele have a blunted pharmacologic response to treatment with clopidogrel, albeit less pronounced than the effect seen among carriers of two reduced-function alleles.^{10, 22-29} Thus, it is plausible that carriers of even one CYP2C19 reduced-function allele as compared with non-carriers would be at increased risk of adverse cardiovascular events in the setting of treatment with standard doses of clopidogrel.

The totality of the pharmacologic data support the findings of our meta-analysis on clinical outcomes, which suggest that patients undergoing PCI treated with standard doses of clopidogrel who have either one or two reduced-function CYP2C19 alleles are at increased risk for major adverse cardiovascular events. The observed HRs for adverse cardiovascular events of 1.55 (for carriers of one CYP2C19 reduced function allele vs. non-carriers) and 1.76 (for carriers of two CYP2C19 reduced function alleles vs. non-carriers) are plausible. The meta-analysis population predominantly underwent PCI, a setting in which dual antiplatelet therapy as compared with aspirin monotherapy results in risk reductions up to 75-85%.³⁰ In such a situation, even partial reductions in the antiplatelet effect of clopidogrel could translate into a several fold increase in the risk of major adverse cardiovascular outcomes. As would be expected, the point estimates for the hazard ratios were numerically higher in patients who carried two rather than one reduced-function alleles, but with overlapping confidence intervals. Moreover, the pharmacogenetic effect was more pronounced for the specific outcome of stent thrombosis than for the broader outcome of cardiovascular death, myocardial infarction, or stroke. This observation logically follows from the more pronounced risk reduction that has been documented with thienopyridines on the former versus the latter outcome.^{30, 31}

Three genetic studies not included in our meta-analysis warrant comment as their results highlight the influence of the clinical setting on the relationship between CYP2C19 genotype, clopidogrel, and clinical outcomes. Specifically, the most significant pharmacogenetic effect appears to be seen in patients treated with clopidogrel for PCI. In the genetic substudy from the PLATO trial approximately two-thirds of patients underwent PCI. Although the required data from this substudy could not be incorporated directly into our meta-analysis due to the manner in which their published data were presented (ie, only presenting results by carrier state rather than by number of alleles), at 30 days the rate of cardiovascular death, myocardial infarction, or stroke among patients treated with clopidogrel was 5.7% in carriers of 1 or 2 reduced-function CYP2C19 alleles and 3.8% in noncarriers (P=.028), which represents a 37% increased risk of events—similar to the current metaanalytic point estimate.³² In landmark analyses, the investigators did not observe any increased risk after 30 days, and the rate of cardiovascular death, myocardial infarction, or stroke by 12 months among patients treated with clopidogrel was 11.2% in carriers of a CYP2C19 reduced-function allele and 10.0% in non-carriers. In a sensitivity analysis adding data from the PLATO genetic substudy (estimating an HR through 12 months of 1.12 for carriers vs. non-carriers of a CYP2C19 reduced-function allele) to the other nine studies in the meta-analysis, the estimated risk among patients treated with clopidogrel of cardiovascular death, myocardial infarction, or stroke associated with carriage of a CYP2C19 reduced-function allele was largely unchanged (HR 1.43, 95% CI 1.11-1.84).

Patients in the genetic substudy from the CURE trial were conservatively managed with only 15.5% undergoing PCI with stenting. In this setting, the investigators observed no hazard associated with carriage of a CYP2C19 reduced-function allele among patients treated with clopidogrel (HR 0.86, 95% CI 0.63-1.17).³³ Notably, in contrast to the 75-85% risk reduction resulting from the addition of a thienopyridine in patients who undergo stenting,³⁰ in conservatively managed patients, treatment with clopidogrel is associated with only an approximately 20% reduction in cardiovascular death, MI, or stroke.³¹ Moreover, carriage of a CYP2C19 reduced-function allele does not completely negate the effects of clopidogrel, but rather is associated with active metabolite and platelet inhibition levels roughly 25-33% less than what is observed in non-carriers.¹¹ Taking these factors into account, one would expect carriage of a CYP2C19 reduced-function allele to confer only a 10-15% increase in risk in predominantly conservatively managed patients such as those in the genetic substudy of CURE, a value that falls within their observed 95% CI. Lastly, in the CHARISMA trial, clopidogrel was not given in a manner consistent with the current guideline recommendations, and notably not all patients had established coronary disease, of those with a prior myocardial infarction the median time from that myocardial infarction to inclusion in the study was approximately two years, and only 22% underwent prior PCI. In this setting, no risk was observed with carriage of CYP2C19 reduced-function alleles among patients treated with clopidogrel in the genetic substudy of the CHARISMA trial,³⁴ but as treatment with clopidogrel did not reduce adverse cardiovascular events in the overall population,³⁵ no pharmacogenetic interaction would be expected. In a further sensitivity analysis adding data from the CURE genetic substudy (see above) and the CHARISMA genetic substudy (estimating a HR of 1.25 for carriers vs. non-carriers of a CYP2C19 reduced-function allele) to the PLATO data and the other nine studies in the meta-analysis, there appears to remain a significant association between carriage of a CYP2C19 reduced-function allele and cardiovascular death, myocardial infarction, or stroke (HR 1.32, 95% CI 1.07-1.63) in the setting of treatment with clopidogrel.

The pharmacokinetic and pharmacodynamic literature supports the notion that the mechanism underlying the observed association between CYP2C19 reduced-function variants and clinical outcomes is reduced bioactivation of clopidogrel into its active metabolite. Nonetheless, it has been discussed whether variants in CYP2C19 could themselves be associated with an increase in adverse cardiovascular events, regardless of treatment with clopidogrel. To that end, in the placebo arm of CHARISMA a directionally higher hazard was observed among carriers of two reduced-function alleles versus non-carriers (HR 1.82, 95% CI 0.74-4.65).³⁴ In contrast, in the CURE genetic substudy, among subjects in the placebo arm, the rate of adverse cardiovascular events was numerically lower among carriers of a reduced-function CYP2C19 allele (11.6%) as compared with non-carriers (13.0%).³³ Similarly, in both CLARITY-TIMI 28 and CLEAR-PLATELETS there was no association between CYP2C19 genotype and adverse cardiovascular outcomes for patients not on clopidogrel.^{9, 16} Furthermore, among patients in TRITON-TIMI 38 treated with prasugrel (a third-generation thienopyridine that is metabolized differently than is clopidogrel) and among patients in PLATO treated with ticagrelor (a non-thienopyridine P2Y₁₂ ADP receptor blocker that does not undergo biotransformation to an active metabolite by CYP2C19), CYP2C19 genetic variants were not associated with an increased risk of clinical outcomes.^{32, 36} Lastly, in genome-wide association studies for incident myocardial infarction involving over 20,000 subjects, no association with CYP2C19 was found.³⁷

Tests are available to identify a patient's CYP2C19 genotype. Although these tests are not widely used at this time, some physicians have started to employ a strategy of CYP2C19 genotyping among subjects initiating treatment with clopidogrel.³⁸ Moving forward, point-of-care genotyping will likely be available, and this technology could further ease the implementation of CYP2C19 testing for interested clinicians and patients.³⁹ Point-of-care testing is also available for platelet function testing. The relationship between pharmacogenetic and pharmacodynamic testing continues to be explored. In one study, CYP2C19 polymorphisms appear to account for 12% of the variability in the effect of clopidogrel, as measured using ADP-induced platelet aggregation, and environmental factors account for <10% of the variability.¹⁶ Compounding the complexity of the relationship, light transmittance aggregometry itself has variable reproducibility.⁴⁰ The association seen between higher platelet reactivity and adverse cardiovascular outcomes is of comparable magnitude to the pharmacogenetic findings,⁴¹ and some, but not all studies have suggested that pharmacogenetic and platelet function testing offer independent predictive value.^{14, 16} Of note, it should be acknowledged that neither genotyping nor platelet function testing is a perfect discriminator of subsequent clinical outcomes, underscoring the complex, multifactorial nature of cardiovascular risk.⁴² Nonetheless, prospective trials of the clinical utility of incorporating genetics and platelet function testing into treatment decisions are underway.^{21, 43-45}

With respect to treatment options, there are some early data suggesting that increasing the dose of clopidogrel in carriers of a CYP2C19 reduced-function allele may improve the degree of platelet inhibition.^{7, 46-49} However, all of these studies have been small, with no study having more than 20 carriers of a CYP2C19 reduced-function allele, and with no study reporting clinical outcomes data. Larger studies that incorporate CYP2C19 genotyping and explore the influence of higher doses of clopidogrel on platelet function parameters, as well as clinical outcomes, will be useful in further assisting with therapeutic decisions. Understanding the ability to treat patients effectively with clopidogrel across CYP2C19 genotypes will be particularly important from a healthcare cost perspective, as the drug is already off patent in some countries and anticipated to go off patent in the US and others in the near future. Additionally, there are other antiplatelet agents that may serve as particularly attractive treatment options for patients with a genetically-impaired response to clopidogrel, such as either of the third generation P2Y₁₂ ADP receptor blockers prasugrel or ticagrelor,^{50, 51} neither of which appears to be influenced by polymorphisms in CYP2C19.^{32, 36} However, currently prasugrel is only approved by the FDA for use in ACS patients who are to be managed with PCI, and ticagrelor is not yet approved.

There are some limitations to these analyses. First, over 95% of the study population was white. Of note, though, the effects of CYP2C19 reduced-function alleles on platelet inhibition with clopidogrel appear to be consistent in white and non-white individuals.^{25, 26} Second, over 90% of the study population was treated with clopidogrel for a PCI, and, as discussed above, the impact of CYP2C19 genotype would be expected to be, and data suggest is, considerably less in patients who do not undergo PCI, where clopidogrel has more modest efficacy. Third, most of the studies included in the meta-analysis provided information only on CYP2C19*2. CYP2C19*2, though, is by far the most frequent variant, accounting for about 95% of the reduced-function allele carrier status and the lack of genotyping beyond the *2 allele would be expected to bias towards the null, since carriers of other CYP2C19 reduced-function alleles (e.g. CYP2C19*3) were included in the non-carriers for seven of the studies. Likewise, there are other genes, not included in this meta-analysis, which may influence the response to clopidogrel. Fourth, for logistical reasons, individual patient-level data could not be combined. Comparative studies, though, have demonstrated excellent quantitative agreement between summary data and patient-level data when the same datasets are used for both analyses and the same exposures and outcomes are used, as was the case for the present meta-analysis.⁵² Finally, the studies contributing to this meta-analysis were generally large clinical outcomes studies in which platelet function testing was not routinely performed. As such, the meta-analysis focuses on the relationship between genotype and outcomes; other studies have highlighted the predictive value of platelet function testing.⁵³

In conclusion, the findings of this collaborative meta-analysis demonstrate that common genetic variants are associated with almost one in three patients not receiving ideal protection from ischemic events when treated with standard doses of clopidogrel for PCI. Given how widely clopidogrel is used to treat patients with cardiovascular disease, determination of the optimal antiplatelet treatment doses or regimens for individual patients is needed in order to appropriately tailor therapy.