Enhanced Clopidogrel Responsiveness in Smokers
Smokers' Paradox Is Dependent on Cytochrome P450 CYP1A2 Status
Objective—Observational studies have reported enhanced response to clopidogrel in smokers (the smokers' paradox). We examined whether genetic variations in the cytochrome and drug transporter system are associated with the effect of smoking on clopidogrel response.
Methods and Results—Clopidogrel on-treatment platelet reactivity (OPR) was measured in 1431 consecutive patients who underwent coronary angiography. Gene samples were available and genotyping was successful in 1123 patients. Nine candidate single-nucleotide polymorphisms in 5 cytochrome genes and 1 drug transporter gene were assessed. The mean OPR of the entire population was 241.9±79.3 (P2Y12 reaction units). Two hundred forty-nine (17%) smokers had lower OPR compared with 1182 (83%) nonsmokers (227.6±76.0 versus 244.9±79.7, P=0.001). Among the 9 single-nucleotide polymorphisms, only CYP1A2 showed a genotype-dependent change in the effect of smoking on OPR. After adjustment for possible confounding factors, cigarette smoking was associated with a lower OPR by −19 P2Y12 reaction units (P=0.009) and lower risk for high OPR (odds ratio [OR], 0.48; 95% CI, 0.31 to 0.74) in the AA and CA genotypes but not in the CC genotype.
Conclusion—Enhanced clopidogrel response in smokers, known as the smokers' paradox, is not universal but was observed only in cytochrome P450 CYP1A2 (−163C>A) A-allele carriers, suggesting a genotype-dependent effect of smoking on clopidogrel responsiveness.
Clopidogrel is a prodrug that needs to be converted into an active metabolite via the hepatic cytochrome P450 enzyme system before it irreversibly binds to P2Y12 adenosine diphosphate receptors. Only 15% of clopidogrel is converted into active metabolites; the remaining 85% is inactivated by plasma esterases.1,2 High on-treatment platelet reactivity (HOPR) to clopidogrel is associated with increased occurrence ischemic events.3–5 An increase in activation of the hepatic cytochrome pathway would result in enhancement of clopidogrel responsiveness. Cigarette smoking has been shown to be associated with improved responsiveness to clopidogrel, resulting in decreased clopidogrel on-treatment platelet reactivity (OPR),6–8 and enhanced benefit among chronic clopidogrel users.9 Moreover, a recent analysis of the CHARISMA data showed that the effect of clopidogrel in reducing cardiovascular mortality is greatest among current smokers and is significantly decreased in nonsmokers.10
The mechanism of this so-called smokers' paradox is unknown. The cytochrome P450 1A2 enzyme, which is also involved in clopidogrel activation, has been postulated to be associated with smoking.11,12 Therefore, we sought to investigate the effect of cigarette smoking on OPR to clopidogrel and whether the genetic status of major cytochrome enzymes involved in clopidogrel metabolism may be involved in increased clopidogrel response in smokers.
The CROSS-VERIFY cohort (measuring clopidogrel resistance to assure safety after percutaneous coronary intervention using VerifyNow) is a prospective cohort including all patients undergoing coronary angiography or percutaneous coronary intervention (PCI) who agreed to measurement of clopidogrel OPR with the VerifyNow P2Y12 assay after clopidogrel therapy at Seoul National University Hospital (Seoul, Korea).
Written informed consent for study participation was obtained by each study patient before enrollment. The inclusion criterion for the present analysis was being enrolled in the CROSS-VERIFY cohort. Exclusion criteria were contraindication to aspirin, clopidogrel, or heparin; the use of intravenous glycoprotein IIb/IIIa inhibitor within 5 days before the platelet reactivity test; the concomitant use of cilostazol; uncontrolled malignancy; bleeding tendency; and ethnicity other than Korean heritage.
Smoking status was ascertained at the time of index procedure. Patients who were current smokers at the time of admission were considered smokers, whereas all other patients were considered nonsmokers. Nonsmoker status included never-smokers and former smokers.
The study complied with Declaration of Helsinki and was approved by the Institutional Review Board of Seoul National University Hospital.
Platelet Function Test
The inhibitory effect of clopidogrel on platelet reactivity was measured using the VerifyNow P2Y12 assay (Accumetrics Inc., San Diego, CA). Blood sample was obtained 12 to 24 hours after final dose of clopidogrel in patients who had been on 75 mg of clopidogrel for more than 7 days, and 12 to 24 hours after coronary angiography or PCI in patients who were loaded with clopidogrel before catheterization. A loading dose of 300 mg of clopidogrel was administered in patients who had been taking clopidogrel for less than 7 days; 600 mg was given to clopidogrel-naïve patients. Whole blood was anticoagulated in a sodium citrate bottle used exclusively for the VerifyNow P2Y12 assay. All patients took aspirin at 100 mg per day or 300 mg loading, if not taken previously.
The VerifyNow assay is a point-of-care test allowing for a rapid platelet reactivity measurement based on turbidimetric based optical detection system. The fibrinogen-coated microparticles aggregate in whole blood in proportion to the number of expressed platelet glycoprotein IIb/IIIa receptors. The P2Y12-mediated pathway was measured by optical signal change, and the results were reported in P2Y12 reaction units (PRU). Technical details and reliability of the assay have been reported previously.13,14 Higher PRU reflects higher P2Y12-mediated reactivity of platelets and thus HOPR, and lower PRU value reflects greater platelet inhibition by clopidogrel and thus low OPR. The coefficient of variation for the test was 7.5% in our institution.
Definition of HOPR
The median OPR is higher in Asians than in whites. Therefore, if we used the cutoff value of 235 PRU, which is usually used in whites for the definition of HOPR,7 the proportion of patients with HOPR would exceed 50% of the population. Therefore, we used a cutoff value of 275 PRU to define HOPR, which was derived from a separate study with 1-year survival analysis (Park KW, Park JJ, Kim HS, unpublished data, 2010). For generalizability, we also performed the same analysis using the cutoff value derived from whites of 235 as suggested by Price et al.7
We examined 9 single-nucleotide polymorphisms (SNPs) of 5 cytochrome genes and 1 transporter gene known to be involved in clopidogrel metabolism, ie, CYP1A2, CYP2B6, CYP2C19, CYP3A4, CYP3A5, and ABCB1, of 1123 patients who agreed with genetic analysis.15,16 The genotyping of CYP1A2*1F (−163C>A, rs762551), CYP2C19*2 (P227P, rs4244285), CYP2C19*3 (W212X, rs4986893), CYP3A4 (IVS10+12G/A, rs2242480), CYP3A5 (CYP3A5*3, rs776746), ABCB1 (C1236T, rs1128503), and ABCB1 (C3435T, rs1045642) was performed using the TaqMan fluorogenic 5′ nuclease assay (ABI, Foster City, CA). As for CYP2B6*6 (K262R, rs2279343) and CYP2C19*17 (−806C/T, rs12248560), the SNaPshot assay was performed according to the manufacturer's instructions (ABI Prism SNaPshot Multiplex kit). Analysis was carried out using Genemapper software (version 4.0, Applied Biosystems Inc, Foster City, CA). In brief, the final volume of polymerase chain reaction was 5 μL, containing 10 ng of genomic DNA and 2.5 μL of TaqMan Universal PCR Master Mix, with 0.13 μL of 40× Assay Mix (assay identifiers: C_8881221_40 for CYP1A2, C_25986767_70 for CYP2C19*2, C_27861809_10 for CYP2C19*3, C_26201900_30 for CYP3A4, C_26201809_30 for 3A5, C_7586662_10 for ABCB1 C1236T, and C_7586657_20 for ABCB1 C3435T). Thermal cycle conditions were as follows: 50°C for 2 minutes to activate the uracil N-glycosylase and to prevent carry-over contamination, 95°C for 10 minutes to activate the DNA polymerase, and 45 cycles of 95°C for 15 seconds and 60°C for 1 minute. All polymerase chain reactions were performed using 384-well plates by a Dual 384-Well GeneAmp PCR System 9700 (ABI), and the end point fluorescent readings were performed on an ABI Prism 7900 HT Sequence Detection System (ABI). Duplicate samples and negative controls were included to ensure accuracy of genotyping.
Data were presented as numbers and frequencies for categorical variables and as mean±SD for continuous variables. For comparison among groups, χ2 (or the Fisher exact test when any expected cell count was <5 for a 2×2 table) for categorical variables and the unpaired Student t test or 1-way analysis of variance for continuous variables were applied. The χ2 test for goodness of fit was used to verify agreement with Hardy-Weinberg equilibrium using the Fisher exact test. A general linear model univariate analysis was applied to quantify the smoking effect on clopidogrel responsiveness by entering patients' smoking status as factor and clinical factors as covariates. A multivariate logistic regression analysis was performed to determine the independent association of CYP1A2 gene polymorphisms with smoking status and HOPR. A 2-sided probability value less than 0.05 was considered statistically significant. Statistical tests were performed using SPSS, version 17 (SPSS Inc., Chicago, IL).
A total of 1549 patients were initially enrolled in the CROSS-VERIFY cohort. We excluded 1 white patient, 3 patients with prior use of glycoprotein IIb/IIIa inhibitor, and 114 patients with concomitant use of cilostazol, leaving 1431 patients available for further analysis. Nine hundred ten patients (63.6%) presented with stable angina, 420 patients (29.4%) with unstable angina, and 101 patients (7.1%) with acute myocardial infarction. Overall, 660 patients (46.1%) underwent coronary angiography only, whereas 771 patients (53.9%) received PCI.
Among them, 1123 patients agreed with genetic analysis. In 8 patients, the DNA amount was not large enough to perform all 9 SNPs analysis, so only 3 SNPs (CYP1A2, CYP2C19*2, and CYP3A5) could be screened. The remaining 1115 patients underwent genotyping of all 9 SNPs.
The OPR of the entire patient population (n=1439) was normally distributed, with a mean of 241.9±79.3 PRU. There were 249 cigarette smokers (17%), and they showed lower OPR compared with nonsmokers (227.6±76.0 versus 244.9±79.7, P=0.001), who made up 83% (n=1182) of the study population (Figure 1).
We compared the baseline characteristics between smokers and nonsmokers. Smokers were younger (59.7±10.1 versus 64.1±9.4 years, P<0.001), more likely to be male (93.2% versus 60.4%, P<0.001), and less likely to be hypertensive (54.4% versus 66.4%, P<0.001). Dyslipidemia and a previous history of PCI or coronary artery bypass graft were less frequent among smokers. Smokers had a lower mean body mass index (BMI) and lower high-density lipoprotein-cholesterol (HDL-C) levels but higher triglyceride and low-density lipoprotein cholesterol (LDL-C) levels. There was no difference in medication history between the 2 groups (Table 1).
Next, we evaluated the effect of various clinical variables on OPR (Table 2). Advanced age was defined as age ≥65 years, renal dysfunction as serum creatinine ≥1.5 mg/dL, and high BMI as BMI ≥25 (the mean BMI value of the study patients). Advanced age, female gender, hypertension, the use of calcium channel blockers (CCBs), and renal dysfunction were associated with significantly higher OPR, whereas cigarette smoking was associated with a significantly lower OPR. Previous history of PCI and coronary artery bypass graft, BMI, and hemoglobin were not associated with a change of clopidogrel responsiveness on univariate analysis. After adjustment for age (by decade), hypertension, smoking, CCB use, serum creatinine, LDL-C, and HDL-C, current smoking was associated with a decrease of OPR of −12.0 PRU (P=0.037) compared with nonsmokers (Table 3). Gender was not entered into the model because of significant interaction with smoking status (P<0.001).
The genotyping success rate varied between 97.8% and 99.7% depending on the SNP. Detailed information is shown in Table 4. The OPR of smokers and nonsmokers according to specific genotypes of the 9 SNPs are shown in Supplemental Table I, available online at http://atvb.ahajournals.org. For there to be a significant interaction between the effect of smoking and SNP, the difference observed between smokers and nonsmokers should show a trend (continuous increase or decrease in the effect of smoking) similar to the number of wild-type allele increases (a steady genotype-dependent effect). Among the 9 SNPs, only CYP1A2 (−163C>A) showed a genotype-dependent effect of smoking, suggesting significant interaction between smoking and CYP1A2 genotype (Supplemental Table I, Supplemental Figure I).
As for the CYP1A2 (−163C>A, rs762551) SNP, the observed genotype distribution was 41.5%, 43.6% and 14.9% for AA, CA, and CC genotypes, respectively. To quantify the effect of smoking according to genotype after adjustment for various clinical variables that significantly affect OPR, we performed a general linear model analysis with adjustment for age, hypertension, creatinine, LDL-C, HDL-C, and CCB use. Smoking was associated with a decrease in OPR of −20.0 PRU (P=0.053) and −20.3 PRU (P=0.048) in patients with the AA and CA genotypes, respectively, but not in those with the CC genotype (+6.5, P=0.673). Using the dominant model, cigarette smoking was associated with a significant decrease in OPR of −19.0 PRU (P=0.009) in patients carrying the A allele (either AA or CA genotype) (Table 3, Figure 2).
In subgroup analysis, the percentage of male smokers was 24.5% (186 of 758 patients), and that of female smokers was 3.6% (13 of 358 patients). The distribution of the CYP1A2 AA, CA, and CC genotypes in male smokers was 77, 80, and 29, respectively, and in female smokers, it was 4, 7, and 2, respectively (Supplemental Table II). Among male patients, the smokers with the CYP1A2 AA/CA genotype had significantly lower OPR compared with nonsmokers (216.7±76.0 versus 233.9±76.4, OPR difference=17, P=0.014), whereas there was no significant difference in those with the CYP1A2 CC genotype (231.1±76.0 versus 224.2±70.4, OPR-difference=−7, P=0.658). As for females, smokers with the CYP1A2 AA/CA genotype had a numerically lower OPR similar in magnitude to that seen in males compared with nonsmokers (239.7±85.9 versus 256.3±87.1, OPR difference=17, P=0.536), but because of the very small number of female smokers, this effect was not statistically significant. Similar to males, the mean OPR was similar between female smokers and nonsmokers in those with the CC genotype (256.0±62.2 versus 251.3±67.9, OPR difference=−5, P=0.923).
In a previous study by our group, we found that Koreans have a higher OPR compared with Westerners and that a cutoff value of 275 PRU best predicts thromboembolic events after PCI in Koreans (Park KW, Park JJ, Kim HS, unpublished data, 2010). This criterion was used to identify patients with high clopidogrel OPR. Using this cutoff value, smoking showed a protective effect only in A-allele carriers (OR, 0.39; 95% CI, 0.20 to 0.76 for AA genotype; OR, 0.54; 95% CI, 0.30 to 0.99 for CA genotype; OR, 0.48; 95% CI, 0.31 to 0.74 for AA/CA genotype). In patients with the CC genotype, the protective effect of smoking against HOPR was absent (OR, 2.38; 95% CI, 0.92 to 6.15) (Figure 3A), suggesting a genotype-specific manifestation of the smokers' paradox.
For generalizability and to confirm the genotype-dependent effect of smoking on OPR of clopidogrel using a second definition of HOPR, we performed an additional analysis using a cutoff value of 235 PRU, which was derived from whites.7 Similar to the results when using 275 PRU as the cutoff value, smoking showed a protective effect against HOPR only in A-allele carriers (OR, 0.64; 95% CI, 0.38 to 1.07 for AA genotype; OR, 0.54; 95% CI, 0.32 to 0.91 for CA genotype; OR, 0.60; 95% CI, 0.42 to 0.86 for AA/CA genotype) but not in those with the CC genotype (OR, 1.63; 95% CI, 0.68 to 3.97) (Figure 3B).
Response variability to clopidogrel is well known, and high clopidogrel OPR has been shown to be associated with increased risk for thromboembolic complications.5,17 Several clinical factors influence OPR to clopidogrel, and previous studies have repeatedly shown that smoking enhances the antiplatelet effect of clopidogrel.6,7 In the current study, we demonstrate that the enhanced antiplatelet response to clopidogrel in smokers, known as the smokers' paradox, is not a universal effect observed in all smokers but rather is dependent on genotype status, specifically the CYP1A2 (−163C>A) A allele. We showed that smoking significantly reduced mean OPR by approximately 20 PRU after adjustment for major clinical variables that affect OPR in only the AA and CA genotype, with no significant effects on the CC genotype. Furthermore, a significant reduction in the frequency of HOPR in smokers was only observed in the AA and CA genotypes.
Observations of Smokers' Paradox in Previous Studies
The observation that smoking enhanced response to clopidogrel and reduced clopidogrel OPR has been reported in several previous studies. Matetzky et al reported that smokers were less often clopidogrel resistant and that cigarette smoking seemed to enhance the clopidogrel antiplatelet effect.5 Saraff et al, in a subanalysis of the CREDO study, reported that the reduction in cardiovascular events in the aspirin plus clopidogrel-treated group compared with the aspirin only treated group is greater in smokers compared with nonsmokers.9 In addition, Berger et al showed that the effect of clopidogrel in reducing cardiovascular mortality is greatest among current smokers and that this benefit diminishes significantly in nonsmokers.10 Our results add to these previous observations that this effect of cigarette smoking on reduction of clopidogrel OPR is not universal but rather is dependent on genotype, specifically on the presence of the cytochrome P450 CYP1A2 (−163C>A) A allele.
CYP1A2 and Smokers' Paradox
Clopidogrel is a prodrug that needs to be converted into active metabolites. Only 15% of clopidogrel will be converted into active metabolites by enzymes, including CYP1A2, 2B6, 2C9, 2C19, and 3A4/5, whereas the remaining 85% is inactivated by plasma esterase.1,2 Increased activation of the hepatic cytochrome pathway could result in increased clopidogrel response. It has been proposed in a previous study that variations in CYP1A2 may result in increased CYP1A2 metabolic activity in smokers. Sachse et al demonstrated no significant differences in CYP1A2 metabolic activity between genotypes in nonsmokers but a 1.5-fold higher metabolic activity in subjects homozygous for the A allele compared with the other genotypes in smokers, suggesting genotype-dependent CYP1A2 activity among smokers.11 CYP1A2 comprises 10% of the liver CYP450 and is preliminarily regulated by the aromatic hydrocarbon receptor and induced through aromatic hydrocarbon receptor–medicated transactivation following ligand binding and nuclear translocation.18 In previous studies, in vivo measurement of human CYP1A2 activity showed mono-, bi-, and trimodal activity, suggesting polymorphic controls of the enzyme.19–21 In addition, sequencing of the CYP1A2 gene showed that a polymorphism in intron 1 could affect the inducibility of CYP1A2.22 It is possible that a selective genotype-dependent induction of the cytochrome P450 enzyme CYP1A2 by cigarette smoking could be an explanation for smokers' paradox, and this suggests that the phenomenon may not be universal but rather genotype specific, as seen in the present study.
In the present study, we found that cigarette smoking was associated with enhanced clopidogrel responsiveness only among A-allele carriers (AA or CA), whereas C-allele homozygotes remained unaffected: OPR was decreased in smokers with CYP1A2 with the A allele but not in those with a CC genotype. Furthermore, cigarette smoking was associated with a significant relative risk reduction for HOPR in A-allele carriers but not in noncarriers, which was confirmed using 2 different definitions of HOPR.
Applicability of the Results to Other Populations
Distribution of the cytochrome P450 CYP alleles varies widely across different populations. As for CYP1A2 (−163C>A), the distribution of AA, CA, and CC genotypes in our study population was 41.5%, 43.6%, 14.9%, suggesting that more than 80% of the population would be susceptible to the effect of smoking. In a Turkish population, the distribution was 25.8%, 50%, and 24.2%,23 and in a German population, it was 46%, 44%, and 10% for the AA, CA, and CC genotypes, respectively.11 Although the genotype distribution may vary slightly among different populations, these previous studies, along with our results, suggest that the effect of smoking on clopidogrel could affect a significant proportion of the population. A validation cohort would be optimal to confirm the applicability of the result to other populations.
This study has a few limitations. The first is the cross-sectional nature of the analysis. To completely elucidate the effect of smoking according to genotype status, the response to clopidogrel needs to be studied in the same person at 2 time points, once during active smoking and once after the discontinuation of smoking, after the effects of smoking have fully disappeared. The present study was not an intervention study but rather a cross-sectional observational study. Therefore, the results of the study should not be interpreted as causative but rather associative and hypothesis-generating at best.
Another limitation is the differences in the presence of clinical predictors of clopidogrel response variability. Although we incorporated statistical methods (analysis of covariance with general lineal model) to adjust for the effects of these differences, the only way to completely eliminate confounding of clinical predictors would be to study the effects of smoking cessation in the same cohort or to match for major clinical predictors when using a cross-sectional study design. Another limitation is the relatively low proportion of smokers (17%) in the current study population. In Korea, 42% of men and 5% of women are smokers (23% of the overall population). Many patients enrolled in the current study had already been in treatment in our cardiovascular clinic and had been encouraged to quit smoking, which explains the low rate of smokers at the time of study enrollment. In addition, although female smokers with the AA/CA genotype had lower OPR by 17 PRU compared with nonsmokers, which was a magnitude similar to that seen in male patients, the difference was not statistically significant. This was probably due to the very small number of female smokers included in the current study. In the general population, the proportion of female smokers is approximately 5% in Korea, and therefore, we do not think that our population is deviated from the general population in Korea.
Cigarette smoking was associated with significantly reduced OPR to clopidogrel. This phenomenon seems to be associated with the genotype of the cytochrome P450 CYP1A2 (−163C>A), suggesting that the cytochrome P450 system may be involved in the mechanism of the smokers' paradox response to clopidogrel.
Sources of Funding
This study was supported by a grant from the Clinical Research Center for Ischemic Heart Disease, Seoul, Republic of Korea (0412-CR02-0704-0001) and a grant from the Innovative Research Institute for Cell Therapy, Seoul National University Hospital (A062260), sponsored by the Ministry of Health, Welfare and Family, Republic of Korea.
- Received September 24, 2010.
- Accepted November 29, 2010.
- © 2011 American Heart Association, Inc.
- Kurihara A,
- Hagihara K,
- Kazui M,
- Ozeki T,
- Farid N,
- Ikeda T
- Geisler T,
- Langer H,
- Wydymus M,
- Gohring K,
- Zurn C,
- Bigalke B,
- Stellos K,
- May AE,
- Gawaz M
- Matetzky S,
- Shenkman B,
- Guetta V,
- Shechter M,
- Bienart R,
- Goldenberg I,
- Novikov I,
- Pres H,
- Savion N,
- Varon D,
- Hod H
- Saraff K,
- Steinhubl S,
- Hsu A,
- Topol E
- Berger J,
- Bhatt D,
- Steinhubl S,
- Shao M,
- Steg P,
- Montalescot G,
- Hacke W,
- Fox K,
- Lincoff A,
- Topol E
- Gurbel P,
- Bliden K,
- Hiatt B,
- O'Connor C
- MacLeod S,
- Tang Y,
- Yokoi T,
- Kamataki T,
- Doublin S,
- Lawson B,
- Massengill J,
- Kadlubar F,
- Lang N