Routine Assessment of On-Clopidogrel Platelet Reactivity and Gene Polymorphisms in Predicting Clinical Outcome Following Drug-Eluting Stent Implantation in Patients With Stable Coronary Artery Disease

Study population

All consecutive patients scheduled between January 7, 2008 and January 31, 2010 to undergo elective DES implantation at the Clinica Mediterranea (Naples, Italy) were assessed for their suitability for the study. Exclusion criteria were: either non–ST-segment or ST-segment elevation myocardial infarction; cardiogenic shock; allergy/intolerance to aspirin and/or clopidogrel; ongoing serious bleeding or bleeding diathesis; platelet count ≤75.000/mm³; planned or undelayable noncardiac surgery; previous percutaneous coronary intervention or coronary artery bypass grafting; severe liver disease (e.g., cirrhosis or portal hypertension); and life expectancy <1 year due to other medical conditions. All enrolled patients gave informed consent prior to the index procedure. Diabetes mellitus (DM) was diagnosed according to current guidelines (11). Chronic kidney disease (CKD) was defined as an estimated glomerular filtration rate <60 ml/min/1.73 m² (12). This study was approved by the local ethics committees.

Stenting procedure and pharmacological approach

At the discretion of the operator, either intraprocedural unfractionated heparin (goal-activated clotting time >250 s) or bivalirudin was used before the procedure. Tirofiban (a glycoprotein IIb/IIIa inhibitor) was administered according to the operator’s judgment. All patients received at least 1 DES. Either first-generation (Cypher Select [Cordis, Johnson & Johnson, Miami Lakes, Florida]; Taxus Libertè [Boston Scientific Corporation, Natick, Massachusetts]) or second-generation (Endeavor Resolute or Resolute Integrity [Medtronic CardioVascular, Santa Rosa, California]; Promus or Promus Element [Boston Scientific]; Nobori [Terumo, Tokyo, Japan]) DES were implanted. Patients received aspirin (100 mg daily) and clopidogrel (75 mg daily) before stent deployment. In patients not already on therapy, a clopidogrel loading dose of 600 mg was administered the day before the procedure.

Post-procedure management and follow-up

Dual antiplatelet therapy, consisting of the administration of aspirin (100 mg/day) and clopidogrel (75 mg/day) for at least 12 months, was prescribed to all patients. In patients deemed at high risk (including those with DM, CKD, multivessel CAD, chronic total occlusion, or bifurcation lesions), clopidogrel administration was recommended for more than 12 months. Clinical follow-up was performed via a visit to the outpatient clinic or by a telephone interview with the patient at 1 month and, thereafter, every 3 months until the 12th month. If clopidogrel was discontinued before the recommended period, the date of discontinuation was recorded.

Platelet reactivity assay

The inhibitory effect of clopidogrel on platelet aggregation was measured with the VerifyNow P2Y₁₂ test (Accumetrics Inc., San Diego, California) (13) within 24 h of DES deployment. Specifically, the test was performed in all instances on the morning following DES implantation, within 3 h of the last maintenance dose of clopidogrel. In patients receiving periprocedural tirofiban, the test was performed at least 48 h after drug discontinuation. OTR was assessed after, rather than before, stent implantation for the following reasons: 1) it was not the purpose of the present study to assess the relationship between OTR and periprocedural myocardial infarction; 2) knowledge of high OTR would have influenced the operator’s decision to use glycoprotein IIb/IIIa inhibitors and/or an additional loading dose of clopidogrel; and 3) it would have led to the direct exclusion of patients treated with balloon angioplasty rather than with stent implantation.

The VerifyNow P2Y₁₂ test is a rapid, point-of-care, cartridge-based, platelet-function assay designed to directly measure the blockade of drugs on the P2Y₁₂ receptor (13). The level of platelet aggregation is measured as an increase in light transmittance as activated platelets bind and aggregate fibrinogen-coated beads. A proprietary algorithm is used to report values in P2Y₁₂ reaction units (PRU). A PRU ≥230 reflects greater platelet reactivity and was used to define patients with high OTR (10). In a secondary analysis, a PRU >208 was also used to define patients with high OTR (14).

DNA extraction, selection of SNP, and genotyping

After DES deployment, 4 ml whole peripheral blood was obtained from patients, the genomic deoxyribonucleic acid (DNA) extracted from mononuclear cells with the DNA QIAamp Midi kit (Qiagen Inc., Valencia, California) according to the manufacturer’s instructions, and the DNA stored at –20°C until used. The quantity and quality of the genomic DNA were verified with a NanoDrop spectrophotometer (Thermo Fisher Scientific, Waltham, Massachusetts) before being assayed with the TaqMan OpenArray Genotyping System (Applied Biosystems, Carlsbad, California). The clopidogrel-metabolizing pathway SNPs used in the customized genotyping panel were: ABCB1 (rs1045642); CYP1A2 (rs762551); CYP2B6*9 (rs3745274); CYP2C8*3 (rs10509681); CYP2C9*2 (rs1799853); CYP2C19*2 (rs4244285); CYP2C19*3 (rs4986893); CYP2C19*4 (rs28399504); CYP2C19*5 (rs56337013); CYP2C19*17 (rs12248560); CYP3A4 (rs2242480, rs4986910); CYP3A5*3 (rs776746); P2RY12 (rs2046934); and PON1 (rs854560, rs662) (15,16). Sample processing was fully automated using the Freedom EVO150 robotic workstation (Tecan Group Ltd., Männedorf, Switzerland). Fluorescent signals were recorded by the OpenArray instrument, and specific cluster parameters were automatically obtained to precisely assign the relative genotypes. All samples were genotyped in duplicate to verify the results and avoid technical errors. All patients were stratified according to genotype effect on clopidogrel metabolism (extensive, intermediate, and poor metabolizers). For CYP2C19, patients were classified as: 1) extensive metabolizers, for individuals not carrying a loss-of-function variant (*1*1) or for carriers of at least 1 increased enzymatic-activity allele (*1*17 or *17*17); 2) intermediate metabolizers, for carriers of 1 loss-of-function allele (*1*2, *1*3, *1*4, or *1*5); 3) poor metabolizers, for carriers of 2 loss-of-function alleles (*2*2, *2*3, *2*4,*2*5*, *3*3, *3*4, *3*5*, *4*5, etc.); or 4) unknown metabolizers, for patients presenting with 1 loss-of-function allele plus a *17 allele.

Study endpoints

The primary endpoint of the study was on-treatment MACE, defined as cardiac death, nonfatal myocardial infarction, or stent thrombosis during clopidogrel treatment. Therefore, each patient was censored once off clopidogrel. Myocardial infarction during follow-up was defined as any typical change in cardiac biomarkers in conditions associated with cardiac ischemia, according to American College of Cardiology guidelines (17). Stent thrombosis was defined according to Academic Research Consortium criteria (18). An independent medical committee assigned and recorded all occurrences of MACE. MACE rate was stratified according to OTR and genotype.

Statistical analyses

The choice of sample size was based on the assumptions of: 1) an annual MACE rate >10%; and 2) a high OTR occurrence in 30% of patients. These assumptions were derived from the data available in the literature (14,19–21). Continuous selected variables are given as mean ± SD. Normality assumption was verified graphically (i.e., QQ plot) and was confirmed using the Shapiro-Wilk test. Categorical variables are expressed in percentages. Comparisons between groups were performed by Student t test or chi-square test. The percentage of successfully genotyped samples and the average genotyping success rate for each SNP (>85%) were calculated; Hardy-Weinberg equilibrium was determined using the chi-square goodness-of-fit test. Initially, to examine the effects of the selected SNP and of PRU ≥230, t test and 1-way analysis of variance were employed. Survival curves were generated using Kaplan-Meier product limit estimator. Unadjusted comparisons between predicted functional groups defined by PRU ≥230 versus <230 or by genotyping (e.g., CYP2C19-based metabolizer status) and the composite MACE outcomes were compared using log-rank test. The presence of at least 1 of the following criteria identified patients at high risk: DM; CKD; age >75 years; multivessel disease; and left ventricular ejection fraction (LVEF) <40% (22–30). Cox proportional hazards model was used to provide hazard ratios (HR) with 95% confidence intervals and adjustment for previously selected clinical predictors (selected a priori by the clinician according to the features of our patient population and to p < 0.01 at univariate analysis), including: age; LVEF; sex; DM; smoking; number of diseased coronary arteries; DES type; estimated glomerular filtration rate; chronic total occlusion; and CKD. To correct for multiple testing, 1,000 bootstrap iterations were computed. The area under the receiver-operating characteristic curve (AUC) was finally used to determine the discriminatory ability of the model. AUC is the probability that a randomly selected patient with the event had a predicted probability of event higher than that of a randomly selected patient without the event. A value of 1 denotes perfect discrimination, whereas a value of 0.5 is no better than chance. For all tests, p<0.05 was considered statistically significant. Statistical analyses were performed with the Stata program (version 11/SE, StataCorp LP, College Station, Texas).

Study population

A complete clinical follow-up was available for 1,432 of 1,461 (98%) patients. MACE at follow-up occurred in 114 (7.9%) patients. Specifically, cardiac death occurred in 14 (0.9%) patients, nonfatal myocardial infarction in 49 (3.4 %), and stent thrombosis in 51 (3.5%) patients (definite 1.0%, n = 14; probable 1.7%, n = 25; possible 0.8%, n = 12). Clinical, angiographic, and procedural features of patients with MACE (MACE group) and without MACE (event-free group) are summarized in Tables 1 and 2. The MACE group had higher percentages of individuals with LVEF <40%, CKD, DM, or chronic total occlusion than did the event-free group. The majority of patients (89.3%) in the 2 groups received second-generation DES. The median duration of dual antiplatelet therapy was 20 (range 15 to 26) months. Dual antiplatelet therapy lasted >1 year in 70% of the MACE group and in 69.2% of the event-free group (p = 0.87).

Response status and metabolizing status as MACE predictors within the whole population

As assessed by the VerifyNow P2Y₁₂ test, 40.5% (584 of 1,432) of patients had a PRU ≥230. No significant interactions were observed between response status and other drugs metabolized by cytochrome P450 (namely, statins, proton-pump inhibitors, and calcium-channel blockers). Durations of dual antiplatelet therapy were 20 (range 13.5 to 26.5) months in patients with PRU ≥230 and 19 (range 12 to 25) months in patients with PRU <230 (p = 0.15). Genotyping was successful in 1,432 patients (>85%). Intermediate metabolizers represented 27% of the total population, whereas poor metabolizers comprised 4.7%. In detail, the high-OTR group (i.e., with a PRU ≥230) was composed of 28.3% intermediate metabolizers and 6.7% poor metabolizers. On the contrary, the low-OTR group (PRU <230) was composed of 27.5% intermediate metabolizers and 3.5% poor metabolizers. The analysis of CYP2C19 genotype distribution revealed a higher percentage of poor metabolizers in the MACE group (Table 3). A significant association with absolute PRU values was observed in CYP2C19 metabolizer classes at the genotypic (1-way analysis of variance, p < 0.02) and allelic (1-tailed student t test, p < 0.001) levels. No significant differences were found between the MACE group and the event-free group when OTR and the other analyzed SNPs of the clopidogrel pathway were considered (Table 3).

Table 4 reports log-rank estimates for the phenotypic and genetic tests. Consistent associations at the genotypic and allelic levels were found only for CYP2C19 variants; in fact, log-rank test for MACE-free survival showed higher MACE rates in poor and intermediate metabolizers with respect to extensive metabolizers (Table 4, Figs. 1A and 1B). No significant association was observed between MACE and high OTR or the other tested SNPs (Table 4, Fig. 1C). No significant association was found when considering a different PRU cutoff (≥208) (Online Table 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Kaplan-Meier Curves for MACE-Free Survival

Patients were stratified based on CYP2C19 metabolizer status (A), allelic level (B), or on-treatment platelet reactivity (C). EM = extensive metabolizers; IM = intermediate metabolizers; MACE = major adverse cardiac events; PM = poor metabolizers; PRU = P2Y₁₂ reaction units.

Univariate and multivariate Cox proportional hazards models were used to provide corrected HRs for both the response status and the metabolizing status. The following independent variables were included in the Cox regression as potential factors affecting composite outcome: age; sex; smoking; DES type; DM; CKD; LVEF <40%; and multivessel disease. Treatments with statins, proton-pump inhibitors, and/or calcium-channel blockers were not included because the role as clinical predictors was not significant.

We found that MACE rate was higher in carriers of loss-of-function CYP2C19 alleles (patients classed as either intermediate or poor metabolizers) than in noncarriers (extensive metabolizer patients) (Table 5). The HR computed for the other analyzed SNP did not reveal significant differences for MACE occurrence (Online Tables 2 and 3). There was no significant difference in MACE rates when patients were stratified according to the PRU cutoff (Table 5).

To evaluate the prognostic ability of both the response status and the metabolizing status, multivariate logistic regression was undertaken first with clinical predictors alone and then including both genetic and OTR variables. AUC was 0.682 when considering only clinical predictors, slightly increased (0.701, p = 0.18) when genetic assay predictors were included, and unchanged (0.683, p = 0.51) with OTR assay predictor. Results identical to those of the whole population were found after excluding patients who received tirofiban (Online Tables 4 to 7).

Response status and metabolizing status as MACE predictors in subgroups at high risk

The predictive accuracies of the OTR and of CYP2C19-based metabolizer stratification were further assessed with the log-rank test in the high-risk subgroups of patients (i.e., those with DM, CKD, age >75 years, multivessel disease, or LVEF<40%). To obviate the low effect size caused by low frequency of composite cardiovascular events, CYP2C19-based categories were considered only in allelic status. The p values were calculated for each analyzed class (phenotypic and genetic). CYP2C19-based metabolizer stratification was confirmed to have prognostic value for MACE in all high-risk groups (Table 6, Figs. 2A and 2B). However, the OTR-based assay had prognostic value only for diabetic and CKD patients (Table 6, Figs. 3A and 3B). Similarly, univariate and multivariate Cox regression exhibited significant HR values for CYP2C19- and OTR-based assays in patients copresenting with DM or CKD (Table 7). Potential interaction effects evaluated in a subset population were not significant.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Kaplan-Meier Curves of MACE-Free Survival for CYP2C19 Metabolizer Groups

Survival in patients copresenting with diabetes mellitus (A) or chronic kidney disease (B). Abbreviations as in Figure 1.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Kaplan-Meier Curves of MACE-Free Survival Stratified for Clopidogrel Reactivity

Survival in diabetic patients (A) and chronic kidney disease patients (B) stratified into phenotypic groups based on the VerifyNow P2Y₁₂ test. Abbreviations as in Figure 1.

In diabetic patients, the AUCs were 0.642 with clinical predictors, 0.665 (p = 0.68) with metabolizer predictors, and 0.652 (p = 0.73) with the OTR. In CKD patients, the AUC was 0.717 for clinical predictors, 0.748 (p = 0.47) when including CYP2C19 genotyping variables, and 0.727 (p = 0.78) for the phenotypic predictors. The prognostic values of both response and metabolic status in diabetic patients according to the treatment (insulin versus no insulin), and in CKD patients according to the presence of DM, are reported as supplementary data (Online Table 8), as is the impact of second-generation DES (Online Tables 9 to 11).

Prognostication of MACE with genotyping

The present study strengthens the concept that the genetic status of CYP2C19 can be used to predict the occurrence of MACE after stenting in stable CAD patients (6,7,31–34). In agreement with Mega et al. (35), the results of the current study support the notion that carrying at least 1 loss-of-function CYP2C19 allele increases the risk of MACE. Indeed, intermediate and poor metabolizer categories of patients were found singularly or in a combined manner (allelic status) to be associated with a higher occurrence of MACE during clopidogrel therapy (7,35). CYP2C19 is a highly polymorphic gene that is critical for 2 essential oxidative steps of clopidogrel bioactivation (4). The clinical consequences of decreased clopidogrel activation due to deleterious effects of CYP2C19 variants have been extensively described (6,31–33). In addition, our study also agrees with others reporting that polymorphisms in ABCB1, CYP1A2, CYP2B6*9, CYP2C8, CYP2C9*2, CYP3A4, CYP3A5*3, P2RY12, or PON1 are not statistically associated with increased risk of MACE (6,36–38).

Prognostication of MACE with the VerifyNow P2Y₁₂ test

Our data extend and clarify the current knowledge on the potential clinical benefit of assessing OTR in patients scheduled for elective DES implantation. When dealing with stable CAD patients with an expected overall low MACE rate, the assessment of OTR (due to its modest sensitivity and specificity) is unlikely to have a prognostic value (14,39). On the contrary, when dealing with the high-risk population (including stable patients with DM and/or CKD, and patients with acute coronary syndromes), the assessment of OTR may have a relevant clinical benefit (40–43). This interpretation explains why, for example, in the ADAPT-DES (Assessment of Dual Antiplatelet Therapy with Drug-Eluting Stents) trial, the OTR (assessed with the VerifyNow P2Y₁₂ test) was a strong independent predictor of stent thrombosis at 30 days only in patients with acute coronary syndromes but not in patients with stable CAD (39). This result has been recently confirmed (44). The POPULAR (Do Platelet Function Assays Predict Clinical Outcomes in Clopidogrel-Pretreated Patients Undergoing Elective PCI) trial demonstrated a modest accuracy of 4 platelet reactivity tests (including the VerifyNow P2Y₁₂ test) in predicting clinical outcome (43). Possible explanations for the discrepancy between our study and the POPULAR trial are: 1) differences in the risk at baseline of the patient populations (as suggested by the different rate of DM, CKD, and bifurcation lesions); and 2) differences in the type of stent used in the 2 studies. Indeed, the extensive use of second-generation DES in our study may have had a relevant role in reducing the MACE rate at follow-up. As recently reported, when compared with first-generation DES, second-generation DES are associated with a 43% lower risk of definite stent thrombosis and with a 23% lower risk of death for up to 2 years (45). Finally, our interpretation may also explain the negative findings of some recent trials that aimed to demonstrate a clinically relevant effect of an antiplatelet therapy tailored according to the OTR result. Indeed, the strategies of increasing the clopidogrel maintenance dose (14) and of switching to prasugrel (46,47) in patients with high OTR failed to improve clinical outcome.

Study limitations

The results of the present study should be limited to the boundaries of the functional assay used (i.e., the VerifyNow P2Y₁₂ test) and cannot be directly extended to other clinical assays proposed to assess OTR (48,49). At present, there is no standardized procedure, due to significant variability among the methods and the absence of uniform cutoff values for identifying patients at higher risk. In addition, the optimal timing for platelet function testing (that is, whether on admission, just before or after surgery, or weeks after discharge) has not been assessed. Some evidence does exist on the superiority of measuring platelet function at 1 month after intervention (50), but other trials are required to better clarify the position of routine assessment of platelet assays in clinical practice. Lack of exact timing between the last clopidogrel maintenance dose and the VerifyNow test assessment may represent a further limitation. However, it is unlikely that a <3-h interval may have had an impact on the VerifyNow test result. The results in high-risk patients represent a subgroup analysis of a registry. Therefore, this finding should be tested in specifically designed randomized controlled trials. Moreover, the observed 7.9% MACE rate was lower than that hypothesized; this might make our study slightly underpowered. However, the observed MACE rate was in the range of the 95% confidence interval of the expected incidence (adjusted Wald interval: 3.6% to 13%). The baseline and procedural characteristics of our patient cohort may explain the high stent thrombosis rate. Compared with those in other studies, our cohort had greater incidences of DM, CKD, and bifurcation lesions. Finally, the cost-effectiveness of routine genotyping and residual platelet function testing procedures, as well as their therapeutic impact, were not studied and should be further investigated.