Hong-Guang Xie, Central Laboratory and Department of Clinical Pharmacology, Nanjing First Hospital, Nanjing Medical University, Nanjing 210006, China. Tel.: +86 25 52887034; fax: +86 25 52269924. E-mail: email@example.com
Correspondence: Yong-Fang Hu, Therapeutic Drug Monitoring and Clinical Toxicology Center, Department of Pharmacy, Peking University Third Hospital, Beijing 100191, China. Tel.: +86 10 82265740; fax: +86 10 82265740. E-mail: firstname.lastname@example.org
Summary. Background: A large number of clinical studies have documented that a loss-of-function variant CYP2C19*2 affects clinical profiles of clopidogrel (efficacy and safety). However, data on the impact of a gain-of-function variant CYP2C19*17 on the response to that drug seem to be less consistent. Objectives: To systematically summarize all available clinical data assessing the role of the CYP2C19*17 variant in patients taking clopidogrel. Methods: A literature search was conducted and a meta-analysis was performed for 11 eligible studies. The endpoints included the major adverse cardiovascular events (MACE, representing non-fatal myocardial infarction, stroke, revascularization, or death), bleeding events, mortality, stent thrombosis and high platelet reactivity (HPR). Results: Data from six clinical studies demonstrated that carriers of the CYP2C19*17 variant had a marked protection against recurrent cardiovascular events in patients with coronary artery disease compared with non-carriers, as measured by a 16% decrease in the incidence of MACE (10.0% vs. 11.9%; OR, 0.82; 95% CI, 0.72–0.94; P = 0.005). On the other hand, carriers had an increased risk of developing bleeding as expected (8.0% vs. 6.5%; OR, 1.25; 95% CI, 1.07–1.47; P = 0.006; four studies). Moreover, the presence of the CYP2C19*17 variant might lead to increased response to clopidogrel, as shown by a marked lower prevalence of HPR in carriers than in non-carriers (37.9% vs. 50.8%; OR, 0.60; 95% CI, 0.45–0.79; P = 0.0003; three studies). Conclusions: Carriers of the CYP2C19*17 variant have greater therapeutic responsiveness to clopidogrel than non-carriers, but they have an increased risk of developing bleeding as well.
Platelet activation and aggregation play a critical role in the pathogenesis of ischemic cardiovascular diseases, and therefore the antiplatelet drug clopidogrel (also known as Plavix) has been widely used for the prevention of their recurrence [1,2]. A large number of studies have demonstrated that clopidogrel must be converted to its active metabolite through the bioactivation pathways before it exerts antiplatelet effects [3–5], and that CYP2C19 is identified as one of the most important hepatic drug-metabolizing enzymes responsible for its metabolic activation [1,6–8]. Further studies suggest that expression of CYP2C19 protein and activity is highly polymorphic, with CYP2C19*2 and *3 being loss-of-function, but CYP2C19*17 being gain-of-function in the clinical settings [1,9–12]. Moreover, almost all meta-analyses have consistently documented that carriers of CYP2C19*2 have an impaired antiplatelet effect in clopidogrel-treated patients, resulting in an increased risk of the recurrence of major adverse cardiovascular events (MACE), stent thrombosis and even death [13–18]. Relative to a large number of clinical studies on the CYP2C19*2 variant, several studies have evaluated the effects of the CYP2C19*17 variant on adenosine diphosphate (ADP)-induced platelet aggregation and clinical outcomes (e.g. MACE, stent thrombosis, bleeding or death) in patients treated with clopidogrel [9,10,19–23]. However, they had less consistent results due to the small sample size and marked heterogeneity across them. For these reasons, a meta-analysis was conducted to clarify the real roles played by the CYP2C19*17 variant in clinical settings.
Literature search strategy
We performed a systematic search of PubMed (1966 to 30 August 2011), EMBase (1974 to 30 August 2011) and Web of Science (1899 to 30 August 2011) for all the studies showing the relationship between the CYP2C19*17 variant and platelet response or clinical outcomes in patients taking clopidogrel using the following search terms: clopidogrel ‘AND’ (CYP2C19, genetics, gene, SNP, or polymorphism). After that, the bibliographies listed in all identified original papers or review articles about the effects of CYP2C19 genetic variants were back-retrieved for the complete literature search. Finally, we checked out all abstracts from the selected scientific meetings: American Heart Association, American College of Cardiology, European Society of Cardiology, and Transcatheter Cardiovascular Therapeutics.
Our primary objective was to assess the specific contribution of the gain-of-function variant CYP2C19*17 to MACE in clopidogrel-treated patients with coronary artery disease (CAD). In addition, the prevalence of high platelet reactivity (HPR) was also evaluated as the surrogate endpoint to reflect cardiovascular risk.
Two researchers (Y. Li and H.L. Tang) independently assessed all retrieved articles for their eligibility for inclusion based on their study titles, abstracts, and full contexts. If the same patient population was used in more than one publication, only the final dataset was chosen to avoid data replication.
In terms of the fact that the datasets showing the effect of the CYP2C19*17 variant were very limited, we adopted the broad inclusion criteria. The studies were included if they were randomized or cohort studies in patients taking clopidogrel, and provided a complete dataset without any overlapping data showing the incidence of at least one of the selected endpoints stratified by the CYP2C19 genotype: MACE, stent thrombosis, HPR or poor platelet response, bleeding, or mortality. In addition, all the studies on altered platelet response to clopidogrel (as measured by platelet aggregation values after receiving a loading dose or the maintenance doses) were eligible if they adopted a clear cut-off value to classify patients into the two subgroups (good or poor response). Differences in the cut-off values applied were acceptable if they had sound reasons. If more than one method was used to measure platelet aggregation, only the one recognized by the expert consensus  was selected.
Two investigators (Y. Li and H.L. Tang) used a standardized form to write down information extracted from each of the selected studies, in which following information was noted: the number of patients, age, gender, genotype distribution, study design, follow-up period, treatment information and clinical outcomes. If data were not provided in tabular form, they were calculated from the corresponding paragraphs or supplementary appendix. Any disagreements were solved by further discussion or consultation.
The endpoints of our meta-analysis included the recurrence of MACE (defined as non-fatal myocardial infarction, stroke, revascularization or the composite of death), stent thrombosis, bleeding, mortality, and the possibility of developing HPR.
Not all the study results were summarized with the CYP2C19 genotypes stratified by CYP2C19*1/*1, *1/*17 and *17/*17, respectively. Therefore, we performed a dominant model to compare the CYP2C19*17 carriers (*17/*17 and *1/*17) with non-carriers (*1/*1). To simplify the case, all other CYP2C19 alleles, including CYP2C19*2, were assigned as the wild-type CYP2C19*1 for further analyses.
Data extracted from each individual study were pooled using Stata 10 to calculate the odds ratio (OR) as a parameter of drug efficacy. Individual and pooled results were illustrated by point estimates and 95% confidence intervals (CIs) with two-sided P-values using a fixed-effects model (Mantel–Haenzel method) or random-effects model (DerSimonian and Laird method). A P-value < 0.05 was considered statistically significant. In addition, the heterogeneity across all included studies was assessed by the Q-statistic, with a P-value < 0.05 or I2 > 50% being considered significant heterogeneity.
A total of 632 abstracts from the multiple databases were retrieved, of which 607 hits were excluded based on the title and content of each abstract. Of the remaining 25 articles retrieved for further assessments, we used our four-step inclusion criteria to judge their eligibility, after which 14 papers were excluded (Fig. 1). There were a total of 11 clinical studies that met our preset entry criteria for meta-analysis [9,10,19–21,25–30], of which eight showed clopidogrel-associated clinical adverse events in a total of 17 302 patients [9,10,19,20,25,26,28,29], and three explored platelet response to clopidogrel in 951 patients [21,27,30], as summarized in Table 1. All eligible studies were published from 2008 to 2011 and conducted in Europe or North America. The study populations were predominantly Caucasian patients with stable CAD or acute coronary syndromes (ACS), among whom the prevalence of CYP2C19*17 allele was estimated as 22%, consistent with that presented by a recent systematic report .
Table 1. Characteristics of the studies included in the meta-analysis
No. of patients with data (mean age, year)
Disease (PCI %)
Duration of follow-up (month)
Clinical adverse event
No. of each event
ACS, acute coronary syndromes; AF, atrial fibrillation; AMI, acute myocardial infarction; CAD, coronary artery disease; HPR, high platelet reactivity; LD, loading dose; MACE, major adverse cardiovascular events; MD, maintenance dose; MI, myocardial infarction; NA, not available; NSTE ACS, non ST-elevation ACS; PRI VASP, platelet reactivity index vasodilator stimulated phosphoprotein assay; RPA, residual platelet activity. *Patients with ST-elevation ACS were excluded. †Patients with atrial fibrillation and at least one additional risk factor for stroke who were not eligible for warfarin therapy. ‡Patients with symptomatic or a history of cardiovascular disease, cerebrovascular disease and peripheral artery disease.
Adverse clinical outcomes
LD 600 mg MD 75 mg day−1
TIMI major and minor bleeding
Post-hoc analysis of RCT
LD 300/600 mg MD 75 mg day−1
PLATO major bleeding
NSTE ACS (18.7)
Post-hoc analysis of RCT
MD 75 mg day−1
CURE major bleeding
MD 75 mg day−1
CURE major bleeding
LD 600 mg MD 75 mg day−1
Post-hoc analysis of prospective trial
LD 600 mg MD 75 mg day−1
TIMI major and minor bleeding
LD 600 mg MD 75 mg day−1
Mean LD 300 mg MD 75 mg day−1
CV event‡ (–)
Post-hoc analysis of RCT
MD 75 mg day−1
Severe or moderate GUSTO bleeding
MD 75 mg day−1
HPR (5 μm ADP-induced platelet aggregation > 46%)
NSTE ACS (100)
LD 600 mg
Non-responder (PRI VASP > 50%)
LD 600 mg
Poor responder (RPA cut-off value > 47%)
Pooled data from eight studies (12 416 patients in total) have demonstrated a moderate protective role of the CYP2C19*17 variant, as indicated by a 12% reduction in the recurrence of MACE in the whole cohort (carriers vs. non-carriers, 9.8% vs. 11.1%; OR, 0.86; 95% CI, 0.76–0.97; P = 0.01) in the presence of the CYP2C19*17 variant, as shown in Fig. 2 [10,19,20,25,28,29]. MACE were monitored in seven of eight studies throughout at least 12 months following the intervention [10,19,20,25,28,29], except for the study by Sibbing et al. , in which results were available only within 1 month. Therefore, a sensitivity analysis was performed to examine the presence of heterogeneity across the studies. Data indicated that carriers of the CYP2C19*17 allele had similar protection against MACE irrespective of the length of the follow-up period. When this effect was investigated in CAD patients in particular, among whom the urgency for antiplatelet therapy was more sound, the protective effect was even more profound (10.0% vs. 11.9%; OR, 0.82; 95% CI, 0.72–0.94; P = 0.005; six studies). There was no significant heterogeneity across all included studies (chi-square, P = 0.299 for heterogeneity).
Clinical bleeding events
There were 782 patients with bleeding, whose incidence was highly heterogeneous across the studies (12 228 patients in total). In the CAD patient subgroup, carriers of the CYP2C19*17 variants displayed a 23% increase in the risk of developing bleeding compared with non-carriers, as shown in Fig. 3 (8.0% vs. 6.5%; OR, 1.25; 95% CI, 1.07–1.47; P = 0.006; P = 0.024 for heterogeneity across four studies) [9,20,26,28]. The heterogeneity would be insignificant when the study by Campo et al.  was excluded (i.e. patients with bleeding events in the first 30 days were not included), because bleeding as a clinical safety endpoint occurred more often in the earlier time after PCI; however, the tendency to bleed among carriers of the CYP2C19*17 variant did not change much. More importantly, inconsistent classifications of bleeding in different studies would impair the reliability of the pooled analysis. Of five studies in which bleeding risk was reported, the generally accepted definition of TIMI (thrombolysis in myocardial infarction) major and minor bleeding was used by Sibbing et al.  and Camp et al. , whereas there was self-definition used in the PLATO trial , which is similar to the TIMI definition . However, the more conservative definition of bleeding was used in the CURE trial . When we ruled out the data from the CURE study, the potential risk of the CYP2C19*17 allele was even more profound (OR: 1.30; 95% CI: 1.09–1.55; P = 0.003). In contrast, non-carriers of the CYP2C19*17 variant in the CHARISMA and ACTIVE A studies suffered more from bleeding, but patients in these two studies were diagnosed as stable vascular disease or arterial fibrillation [20,25], suggesting that the nature of these diseases is different from CAD.
Death and stent thrombosis
The isolated incidences of death and stent thrombosis were another two endpoints involved. However, none of them had significant results (P = 0.38, and P = 0.73, respectively) although all patients enrolled had CAD (Fig. 4). The low incidence of such events and differences in follow-up time and patient’s characteristics between the studies may help explain these neutral results.
High platelet reactivity (HPR)
Unlike adverse clinical outcomes evaluated (e.g. MACE and bleeding), plasma clopidogrel active metabolite concentrations and on-treatment platelet reactivity are the direct pharmacodynamic indicators to show clopidogrel antiplatelet effects. For example, several studies have explored the relationship between the CYP2C19*17 variant and HPR, but they adopted different methodologies of quantification and delivered their results in different ways. In this meta-analysis, we tried to overcome these differences by connecting genotype with the possibility of poor clopidogrel responsiveness. In the three articles, such a relationship was expressed as the number of patients who were poor responders , non-responders  or had HPR , respectively. We combined the results from these three papers as a sum of the possibility for on-treatment HPR . As shown in Fig. 4, all selected studies indicated a parallel effect. The prevalence of on-treatment HPR was lower in the CYP2C19*17 carrier group than in the non-carrier group (37.9% vs. 50.8%; OR, 0.60; 95% CI, 0.45–0.79; P = 0.0003; n = 951). Although there was large heterogeneity across the methodologies of the platelet function assays and in the selected cut-offs for HPR, less heterogeneity in HPR was seen across the CYP2C19*17 carriers.
Our meta-analysis indicates that, compared with non-carriers of the CYP2C19*17 variant, carriers have a 16% reduction in the recurrence of the MACE in 9428 clopidogrel-treated CAD patients during 1 year of follow-up, but have increased risk of bleeding. As expected, the CYP2C19*17 carriers also have a lower on-treatment HPR than non-carriers. However, there are less significant effects on stent thrombosis and death in clinical settings.
Since identification of the CYP2C19*17 variant (−806C>T in the promoter region of the gene) as a gain-of-function variant , its clinical significance had not been explored until its presence was associated with increased efficacy and bleeding risk in patients treated with clopidogrel. Although some studies indicate that clopidogrel-treated CYP2C19*17 carriers, compared with non-carriers, have less residual platelet reactivity [21,22], a 22% reduction in MACE and a 37% reduction in revascularization in patients with acute myocardial infarction , and a significantly lower risk of experiencing recurrent ischemic cardiovascular events , other studies do not support such an increased efficacy [10,23,30]. Moreover, no association was observed between CYP2C19*17 and adverse composite cardiovascular events , or stent thrombosis . Some of the inconsistencies in the data on CYP2C19*17 may be due to small sample size, small effect size, study population heterogeneity, diverse methodologies for genotyping and platelet function testing, and partial linkage of CYP2C19*17 to CYP2C19*2 [1,23,35]. However, meta-analysis can be used to explore some sources of such heterogeneity across all the studies [36,37].
The occurrence of ischemic cardiovascular events is the result of increased platelet activation and aggregation as well as thrombosis formation. A previous study indicated that increased exposure to clopidogrel active metabolite in carriers of CYP2C19*17 would result in a greater inhibition of platelet aggregation than in non-carriers . Our pooled results further suggest that carriers of CYP2C19*17 have a lower on-treatment HPR than non-carriers, as expected, which is consistent with a significant inhibition of ADP-induced platelet aggregation in the CYP2C19*17 carriers compared with non-carriers in the CAD patients treated with clopidogrel [9,21,22]. Furthermore, multivariate linear regression analysis revealed that the presence of CYP2C19*17 was independently associated with increased platelet inhibition in clopidogrel-treated patients . In a recent meta-analysis, patients with HPR had an increased risk of developing myocardial infarction, stent thrombosis and death . Moreover, in our meta-analysis, a moderate reduction in the recurrence of MACE in carriers of CYP2C19*17 would be translated into the increased protection exerted by clopidogrel in these carriers compared with non-carriers [9,19,20], which is well explained, at least in part, by the lower HPR in carriers of CYP2C19*17 seen in our meta-analysis.
Enhanced antiplatelet effect may lead to bleeding (major or minor) in patients treated with clopidogrel. Therefore, it is not surprising that carriers of CYP2C19*17 have an increased risk of developing bleeding compared with non-carriers, and that the presence of CYP2C19*17 was also identified as an independent predictor of such bleeding in a multivariate linear regression analysis . After pooled analysis of all available data on CYP2C19*17, its carriage status is still a risk factor for bleeding. However, low platelet reactivity (i.e. enhanced response to clopidogrel) was not evaluated as an index in this meta-analysis to explore the association of CYP2C19*17 variant with low platelet reactivity based on the following reasons: (i) no widely accepted definition of low platelet reactivity, (ii) inconsistent methodologies used in the relevant literature to measure platelet reactivity, and (iii) only one study available showing that the CYP2C19*17 allele variant was associated with enhanced response as measured with platelet reactivity .
Our findings are inconsistent with those of a more recently published meta-analysis that explored the relationship between CYP2C19*17 carriage and the MACE associated with clopidogrel , in which only five studies (vs. eight studies in our meta-analysis) were pooled and produced a statistically insignificant association. After careful comparison, we found that such inconsistent conclusions between us and others  may be the result of different data extraction. For the same study , the MACE (including non-fatal myocardial infarction, stroke, revascularization or death as defined elsewhere [10,20]) were adopted as the endpoint in our meta-analysis, whereas just the events of myocardial infarction or death rather than all MACE events were included in the meta-analysis by Bauer et al. , suggesting that non-standardized use of the endpoint definitions may result in a bias of data selection. In addition, another similar meta-analysis (which evaluated the association of the CYP2C19*17 variant with clopidogrel efficacy and safety) had an incorrect category of the genotype and disease stratification , in which CYP2C19*2 carriers were classified incorrectly and patients with atrial fibrillation were included. Moreover, there are also several potential limitations worth discussion in the meta-analyses carried out by us and others. First, the number of studies on the genetic variant CYP2C19*17 is limited, and therefore publication bias would be difficult to avoid. In an attempt to reduce it, the studies would be stratified according to disease and duration of follow-up, and then sensitivity analysis would be performed as usual. In terms of the fact that meta-analysis itself could result in loss of some (if not all) valuable information on baseline data of individual patients after all the eligible data are pooled, the impact of confounding factors would be difficult to examine and adjust, and final conclusions could not be generalized accordingly. For example, a pooled analysis of the poor response to clopidogrel (i.e. HPR) would fail if baseline data of the individual patients were missing. Thus, a prospective, large-scaled, well-designed and multicenter clinical trial would be needed to further evaluate the role of CYP2C19*17 in clinical settings of diverse ethnicity.
In summary, our meta-analysis demonstrates that the presence of the CYP2C19*17 variant would increase not only clinical efficacy, but also bleeding risk in patients treated with clopidogrel. Therefore, genotyping of CYP2C19*17 before taking clopidogrel would benefit patients with ischemic cardiovascular diseases, in particular patients of European descent due to a higher allele frequency.
Y. Li performed the literature search, study selection, data extraction and statistical analysis, and wrote the manuscript draft. H.-L. Tang assisted in the literature search, study selection and data extraction. Y.-F. Hu guided the statistical analysis, and contributed to the interpretation of data and critical review of the articles for important intellectual content. H.-G. Xie had the idea for this article and contributed to the draft writing and final revision process. All of the authors approved the final version of the manuscript for publication.
This study is supported by the National Natural Science Foundation of China (No. 81072705) and by the State Key Basic Research and Development Plan (Grants 2011CB915504).
Disclosure of Conflict of Interests
The authors state that they have no conflict of interest.