Lack of Adverse Clopidogrel–Atorvastatin Clinical Interaction From Secondary Analysis of a Randomized, Placebo-Controlled Clopidogrel Trial
Background— Statins primarily metabolized by cytochrome P450 3A4 (CYP3A4) reportedly reduce clopidogrel’s metabolism to active metabolite, thus attenuating its inhibition of platelet aggregation ex vivo. However, the clinical impact of this interaction has not been evaluated.
Methods and Results— Clopidogrel for the Reduction of Events During Observation (CREDO) was a double-blind, placebo-controlled, randomized trial comparing pretreatment (300 mg) and 1-year (75 mg/d) clopidogrel therapy (clopidogrel) with no pretreatment and 1-month clopidogrel therapy (75 mg/d) (control) after a planned percutaneous coronary intervention. All patients received aspirin. The 1-year primary end point was a composite of death, myocardial infarction, and stroke. We performed a post hoc analysis to evaluate the clinical efficacy of concomitant clopidogrel and statin administration, categorizing baseline statin use to those predominantly CYP3A4-metabolized (atorvastatin, lovastatin, simvastatin, and cerivastatin) (CYP3A4-MET) or others (pravastatin and fluvastatin) (non-CYP3A4-MET). Of the 2116 patients enrolled, 1001 received a CYP3A4-MET and 158 a non-CYP3A4-MET statin. For the overall study population, the primary end point was significantly reduced in the clopidogrel group (8.5% versus 11.5%, RRR 26.9%; P=0.025). This clopidogrel benefit was similar with statin use, irrespective of treatment with a CYP3A4-MET (7.6% clopidogrel, 11.8% control, RRR 36.4%, 95% CI 3.9 to 57.9; P=0.03) or non-CYP3A4-MET statin (5.4% clopidogrel, 13.6% control, RRR 60.6%, 95% CI −23.9 to 87.4; P=0.11). Patients given atorvastatin or pravastatin had similar 1-year event rates. Additionally, concomitant therapy with statins had no impact on major or minor bleeding rates.
Conclusions— Although ex vivo testing has suggested a potential negative interaction when coadministering a CYP3A4-metabolized statin with clopidogrel, this was not clinically observed statistically in a post hoc analysis of a placebo-controlled study.
Received March 4, 2003; de novo received May 29, 2003; revision received July 7, 2003; accepted July 8, 2003.
Clopidogrel and statins are frequently coadministered in patients with ischemic heart disease or other atherothrombotic manifestations. However, a recent ex vivo platelet function study suggests that clopidogrel’s effectiveness in inhibiting platelet aggregation is attenuated by coadministration with atorvastatin. Atorvastatin is extensively metabolized by cytochrome P450 3A4 (CYP3A4) and might competitively inhibit the metabolic activation of clopidogrel in the liver.1 Other 3-hydroxy-3-methyglutaryl coenzyme A reductase inhibitors (statins) predominantly CYP3A4-metabolized (lovastatin, simvastatin, cerivastatin) are also predicted to have similar effects on clopidogrel’s metabolism. In contrast, pravastatin (not significantly metabolized by P450 isoenzymes) and fluvastatin (primarily CYP2C9-metabolized) should not affect clopidogrel’s antiplatelet activity.1,2 These laboratory findings engendered considerable concerns in the medical community. The CREDO (Clopidogrel for the Reduction of Events During Observation) trial3 conferred an opportunity to retrospectively evaluate potential clinical interaction with concomitant clopidogrel and statin therapy.
See p 910
The design, methods, and primary results of the CREDO trial have been described in detail.3 To summarize, 2116 patients with symptomatic coronary artery disease and objective evidence of ischemia who were undergoing elective percutaneous coronary intervention (PCI) or were deemed highly likely to undergo PCI were enrolled in 99 centers in North America from June 1999 through April 2001. Patients were randomized to receive a 300-mg clopidogrel loading dose (n=1053) or placebo (n=1063) 3 to 24 hours before PCI. After the procedure, all patients received clopidogrel 75 mg/d for 28 days. From day 29 through 12 months, the loading-dose group received clopidogrel 75 mg/d, whereas placebo was given to the control group. The 1-year primary end point was a composite of death, myocardial infarction (MI), and stroke in the intent-to-treat population. After hospital discharge, outpatient follow-up was performed on at 28 days, 6 months, and 1 year, with intervening telephone follow-up at 2 and 9 months.
Statin administration was nonrandomized and directed by the treating physicians. We recorded the statin choice at baseline, 2 days after randomization, and each outpatient and telephone follow-up. Dosages were not documented. Two statin classes were distinguished and the effects on the composite 28-day and 1-year death/MI/stroke events were evaluated: (1) predominantly CYP3A4 metabolized (atorvastatin, lovastatin, simvastatin, and cerivastatin) (CYP3A4-MET), and (2) not predominantly CYP3A4 metabolized (pravastatin and fluvastatin) (non-CYP3A4-MET). Major bleeding (intracranial hemorrhage or hemoglobin decrease >5 g/dL) and minor bleeding complications (>3 g/dL hemoglobin decrease when bleeding site was observed, or >4 g/dL when bleeding site not identified) were compared between groups. Additionally, clinical outcomes were separately evaluated in patients receiving atorvastatin or pravastatin. Baseline statin was utilized for our primary analyses; reanalyses using day 2 or 28 postrandomization statin use were also performed for 1-year events.
All evaluations were performed on the intent-to-treat population. Hypothesis tests were done using 2-sided tests at the 5% significance level. Baseline characteristics were compared with χ2 and Fisher’s exact tests for discrete variables, and Wilcoxon rank-sum for continuous variables. Logistic regression methods were used to test 28-day events, and Cox proportional hazard methods were used for 1-year events. Relative risk reductions (RRRs) with corresponding 95% confidence intervals (CI) are also provided. Log-rank tests were performed to test Kaplan-Meier estimates. The Breslow-Day test for homogeneity of odds ratios (ORs) was performed to evaluate the treatment effect of clopidogrel in the following groups: CYP3A4-MET versus non-CYP3A4-MET, and atorvastatin versus pravastatin.
CREDO enrolled 2116 patients, with 1172 (55.4%) receiving statins before randomization. Of these, 1001 received a CYP3A4-MET (564 atorvastatin) and 158 a non-CYP3A4-MET statin (142 pravastatin). Thirteen patients were listed as being on both types of statins and were excluded from our analyses. Baseline demographics were similar in the CYP3A4-MET and non-CYP3A4-MET groups (Table). The “no-statin” group had lower incidence of hyperlipidemia, prior MI, and hypertension, but more prior CABG compared with either statin groups (P<0.05). The number of patients on statin therapy increased after randomization: 1262 (59.6%) patients after 2 days, 1352 (63.9%) after 28 days, and 1262 (59.6%) at 1 year.
One-Year Composite Death/MI/Stroke
The main CREDO trial reported a 26.9% relative reduction in the primary composite 1-year events with clopidogrel compared with controls (8.6% versus 11.8%; P=0.02). This benefit was maintained with concomitant use of a statin (7.4% clopidogrel, 11.9% control, RRR 38.6%, 95% CI 9.8 to 58.2; P=0.01) regardless of the type: CYP3A4-MET (7.6% clopidogrel, 11.8% control, RRR 36.4%, 95% CI 3.9 to 57.9; P=0.03) or non-CYP3A4-MET (5.4% clopidogrel, 13.6% control, RRR 60.6%, 95% CI −23.9 to 87.4; P=0.11). Patients not treated with statins experienced a lesser degree of treatment benefit with randomization to clopidogrel (10.1% versus 11.6%, RRR 12.4%, 95% CI −29.0 to 41.0; P=0.51) (Figure, A). Direct comparisons of CYP3A4-MET and non-CYP3A4-MET groups revealed no difference in 1-year events for the overall CREDO population (9.8% versus 9.7% respectively; P=0.94), or those randomized to clopidogrel (n=557) (7.6% versus 5.4% respectively; P=0.51). Reanalysis of event rates based on statins taken at day 2 or 28 after randomization showed no difference (data not shown).
Day 28 Composite Death/MI/Stroke
Clopidogrel pretreatment did not reduce the 28-day death/MI/stroke event rates compared with placebo (5.5% clopidogrel, 6.9% control; P=0.21). This was not different among those not receiving a statin (6.4% clopidogrel, 7.1% control; P=0.69), or those given either statins: CYP3A4-MET (5.0% clopidogrel, 7.6% control; P=0.11), or non-CYP3A4-MET (4.0% clopidogrel, 2.4% control; P=0.59) (Figure, B). Direct comparisons of statin groups also showed comparable 28-day events for patients randomized to clopidogrel (4.0% CYP3A4-MET, 3.9% non-CYP3A4-MET; P=0.70). Pretreatment with study drug ≥6 hours before PCI did not affect the 28-day event rate between statin groups (6.6% CYP3A4-MET, 4.9% non-CYP3A4-MET; P=0.65).
Atorvastatin Versus Pravastatin
Concomitant atorvastatin (RRR 49.8%; P=0.02) or pravastatin (RRR 63.3%, P=0.13) did not influence clopidogrel’s superiority over controls at 1 year (Figure, A). The 28-day event rate was significantly lower with atorvastatin and clopidogrel pretreatment (3.4%) compared with no pretreatment (7.6%) (RRR 55.5%; P=0.041) (Figure, B). There was no difference in 28-day (3.4% atorvastatin, 4.6% pravastatin; P=0.63) or 1-year event rates (6.5% atorvastatin, 4.6% pravastatin; P=0.62) for those randomized to clopidogrel.
Both CYP3A4-MET and non-CYP3A4-MET groups had similar major (6.7% versus 7.6%, respectively; P=0.68) and minor (5.1% versus 5.7%, respectively; P=0.75) bleeding rates at 1 year. Likewise, there was no difference between atorvastatin and pravastatin.
This is the first study evaluating the clinical impact of concomitant clopidogrel and statin use. In this post hoc analysis, no statistical difference in clinical events was identified when clopidogrel was coadministered with a statin that was predominantly CYP3A4-metabolized or not. The recent publication by Lau et al1 raised concerns with point-of-care platelet function measurements showing attenuation of clopidogrel’s antiplatelet activity by atorvastatin in a dose-dependent manner. Although CYP3A4 is the purported predominant isoenzyme producing clopidogrel’s active metabolite,1,4 others have found conflicting results.4,5
We subdivided CREDO patients on a statin before randomization into two groups according to CYP3A4 metabolism. Neither group adversely affected clopidogrel’s efficacy relative to placebo for short- or long-term clinical ischemic events. Moreover, atorvastatin and pravastatin had no adverse effects with concomitant clopidogrel use. Although the odds ratio of 1-year events for patients randomized to clopidogrel appear better for non-CYP3A4-MET (OR=0.36) compared with CYP3A4-MET (OR=0.62) groups, this was not statistically significant with the Breslow-Day test, nor was it significantly different from those not on statins (OR=0.88).
The contradiction of Lau’s ex vivo data and CREDO’s clinical results elicits a few possible explanations. The sole platelet function assay used in Lau’s study was Plateletworks, which indirectly measured platelet aggregation. This bedside test uses a cell counter to measure objects exceeding threshold platelet size, which were recorded as platelet aggregation events. However, a control assay with conventional platelet aggregometry that is less susceptible to measuring artifacts was not performed. Another plausible explanation is insufficient statin dose in the CREDO population to achieve clinically relevant interaction with clopidogrel, although this seems unlikely as even 10 mg of atorvastatin significantly diminished clopidogrel’s platelet aggregation inhibition in Lau’s study. Alternatively, the benefit of clopidogrel in reducing inflammation via platelet activation inhibition (eg, reducing platelet CD40 ligand expression)6 may supersede its antiplatelet aggregation effect.
Our study has several limitations inherent with retrospective post hoc designs, precluding definitive conclusions. Because allocation into statin subgroups was nonrandomized, potential selection bias may exist. However, the baseline characteristics were similar between groups segregated by CYP3A4 metabolism. Additionally, clopidogrel treatment was randomized and blinded, thus any bias introduced by statin choice should be well balanced between clopidogrel and control groups. Although our study was underpowered for non-CYP3A4-MET and pravastatin groups (potential beta error), more importantly, we convincingly showed with adequate power that CYP3A4-MET or atorvastatin did not adversely affect clopidogrel’s efficacy. The lack of statin dosage notably restricts our ability to assess the dose-dependent effect of potential drug interaction. Finally, although our primary analyses were based exclusively on baseline statins, accommodation of statin changes by reanalyzing 1-year events with day 2 and 28 statin use did not alter our conclusions.
In summary, post hoc analysis of a placebo-controlled randomized trial suggests no adverse effect on death, MI, and/or stroke with clopidogrel and statin coadministration. At this juncture, there is no clear evidence that clinicians should choose statins on the basis of CYP3A4 metabolism when clopidogrel coadministration is required.
Dr Topol has served as a consultant to and received research support from Bristol Myers Squibb and Sanofi-Synthelabo. Dr Serebruany has received grants from Pfizer, Sanofi-Synthelabo, and Bristol Myers Squibb. Dr Serebruany has also served as a consultant to and holds stock in Pfizer and is named in a patent held by Pfizer.
This work was supported by Sanofi-Synthelabo/Bristol Myers Squibb Partnership.
Lau WC, Waskell LA, Watkins PB, et al. Atorvastatin reduces the ability of clopidogrel to inhibit platelet aggregation. Circulation. 2002; 107: 32–37.
Clarke TA, Waskell LA. The metabolism of clopidogrel is catalyzed by human cytochrome P450 3A and is inhibited by atorvastatin. Drug Metab Dispos. 2003; 31: 53–59.