Combined clopidogrel and proton pump inhibitor therapy is associated with higher cardiovascular event rates after percutaneous coronary intervention: a report from the BASKET trial

Authors


Raban V. Jeger MD, Cardiology, University Hospital Basel, Petersgraben 4, 4031 Basel, Switzerland.
(fax: +41 61 265 4598; e-mail: rjeger@uhbs.ch).

Abstract

Abstract.  Burkard T, Kaiser CA, Brunner-La Rocca H, Osswald S, Pfisterer ME, Jeger RV for the BASKET Investigators (Cardiology, University Hospital; and Cardiology, Maastricht University Hospital Center, Maastricht, the Netherlands). Combined clopidogrel and proton pump inhibitor therapy is associated with higher cardiovascular event rates after percutaneous coronary intervention: a report from the BASKET trial. J Intern Med 2012; 271: 257–263.

Objective.  To investigate whether there is an increased risk of cardiac events with a combined therapy of clopidogrel and proton pump inhibitors (PPIs) after percutaneous coronary intervention (PCI).

Design.  In the BAsel Stent Kosten Effektivitäts Trial (BASKET), all patients undergoing PCI received 6 months of clopidogrel and were analysed for the use of PPI therapy. Endpoints were major adverse cardiac events (MACE), myocardial infarction (MI), death and target vessel revascularization (TVR) after 36 months.

Results.  Of 801 patients with available discharge medication data, 109 (14%) received PPIs. Patients who received PPIs were older (66.5 ± 10.5 vs. 63.3 ± 11.3 years, = 0.006), more likely to be woman (80% vs. 69%, = 0.009) and have a history of diabetes (29.6% vs. 17.3%, = 0.002) or gastrointestinal ulcer disease (8.3% vs. 3.3%, = 0.015) and more often received nonsteroidal anti-inflammatory drugs (7.3% vs. 2.2%, = 0.003) and corticosteroids (11% vs. 3.6%, = 0.001) but not aspirin (91.7% vs. 97%, = 0.008) compared with those who did not receive PPIs. Patients who received PPI therapy had higher rates of MACE (30.3% vs. 20.8%, = 0.027) and MI (14.7% vs. 7.4%, = 0.01) but similar rates of death (9.2% vs. 7.4%, = 0.51) and TVR (20.2% vs. 15.3%, = 0.2) compared with those who did not. By multivariate analysis, diabetes (hazard ratio 1.83, 95% confidence interval 1.07–3.15) and PPI use (hazard ratio 1.88, 95% confidence interval 1.05–3.37) were the only independent risk factors for MI.

Conclusion.  In a real-world PCI population, the combination of PPIs and clopidogrel was associated with a doubling of MI rates after 3 years. Even after correction for confounding factors, concomitant PPI use remained an independent predictor of outcome emphasizing the clinical importance of this drug–drug interaction.

Introduction

Dual antiplatelet therapy (DAPT) consisting of aspirin (ASA) and clopidogrel is a cornerstone in the medical treatment of acute coronary syndromes and after percutaneous coronary intervention (PCI). Compared with ASA alone, the combination of ASA and clopidogrel was shown to significantly reduce the incidence of cardiovascular events after acute coronary syndromes [1], whereas restricted access to clopigogrel, ingestion delay or discontinuation was associated with worse cardiovascular outcomes after coronary stent implantation [2–4]. This became a major concern because DAPT was judged to be essential after stent implantation to prevent early and late in-stent thromboses [5, 6]. Clopidogrel is a thienopyridine that irreversibly inhibits the platelet P2Y12 adenosine diphosphate receptor. As a prodrug, it needs hepatic biotransformation into its active metabolite [7]. This activation is mediated by hepatic cytochrome P450 (CYP) isoenzymes, mainly cytochrome CYP2C19 [8]. Patients with a loss-of-function polymorphism of CYP2C19 have lower plasma levels of the active metabolite, reduced platelet inhibition under clopidogrel therapy and an increased risk of cardiovascular events [9, 10].

Proton pump inhibitors (PPIs) are widely used in hospitalized patients, mostly for prophylaxis of gastrointestinal bleeding [11]. According to current US guidelines, PPIs are indicated with DAPT even in the absence of gastrointestinal symptoms [12]. All PPIs are, at least partly, metabolized by CYP2C19 [13, 14]. Recent in vitro studies demonstrated that PPIs, particularly omeprazole, might alter the pharmacokinetics and decrease the antiplatelet effect of clopidogrel, which can be explained by inhibition of the CYP2C19-dependent activation of clopidogrel [15–17]. Two retrospective population-based studies showed increased cardiovascular event rates in patients treated with PPIs [18, 19], whereas other studies did not find any difference between outcomes [20]. Similarly, a prematurely stopped and, for cardiovascular events, underpowered prospective randomized trial showed no difference in cardiovascular outcome with respect to PPI therapy [21]. In view of the inconsistent findings, the randomized controlled data of the BAsel Stent Kosten Effektivitäts Trial (BASKET) [22] might add important information regarding the effect of concomitant use of clopidogrel and PPIs on outcomes in a cardiovascular high-risk population.

Methods

This is a retrospective analysis of the prospective randomized controlled BASKET data [22] with a 98% complete 3-year follow-up [23]. In brief, BASKET was an open-label trial to test the cost-effectiveness of two drug-eluting stents, the sirolimus-eluting Cypher® stent (Cordis Johnson & Johnson, Miami Lakes, FL, USA) and the paclitaxel-eluting Taxus® stent (Boston Scientific Corporation, Natick, MA, USA), compared with a bare metal stent, the Vision® stent (Guidant, Indianapolis, IN, USA), in an all-comer population undergoing PCI irrespective of the clinical indication. Patients were randomly assigned 2 : 1 to drug-eluting versus bare metal stents, with a 1 : 1 randomization to Cypher® versus Taxus® within the drug-eluting stent group. Exclusion criteria included a vessel diameter ≥4 mm (because the widest drug-eluting stent available during the study was 3.5 mm), in-stent restenosis and a lack of written informed consent. The study was approved by the local ethics committee. All events were adjudicated by an independent critical events committee.

Antiplatelet therapy included loading doses of ASA (250–500 mg intravenously) and clopidogrel (300 mg orally) and maintenance doses of ASA (100 mg daily to be continued indefinitely) and clopidogrel (75 mg daily for 6 months) in all stent groups. After PCI, long-term statin therapy was recommended for all patients. All endpoints of this study were clinically driven, and follow-up angiography was performed only if clinically indicated. The primary endpoint of this analysis was major adverse cardiac events (MACE), defined as a composite of cardiac death, nonfatal myocardial infarction and target vessel revascularization after 36 months. Myocardial infarction was defined as typical symptoms with an elevated level of cardiac enzymes (i.e. troponin I, troponin T or creatine phosphokinase) above the upper limit of normal or typical ST-segment changes in the electrocardiogram at the time of symptom development [24]. Secondary endpoints were the single components of MACE and definite, probable and possible stent thromboses according to the Academic Research Consortium criteria [25]. Discharge medication, particularly PPIs, was reviewed for all patients for this analysis. Patients with no discharge medication data were excluded from the analysis.

Statistical analysis

Data are presented as means ± standard deviation or as percentages. Categorical variables were compared using the χ2 test or Fisher’s exact test, as appropriate. Continuous variables were compared using the Student’s t test. Both forward and backward stepwise conditional Cox regression was used for multivariate analysis. Baseline differences between patients treated with or without PPIs were identified and incorporated in a multivariate Cox regression analysis if significantly different (< 0.05); variables incorporated included sex, age, weight, serum creatinine, diabetes mellitus, history of ulcer disease, nonsteroidal anti-inflammatory drugs at discharge, steroids at discharge and use of ASA. Kaplan–Meier curves were generated for analysis of the time-dependent occurrence of events using the log-rank test to detect differences. Results are expressed as hazard ratios with 95% confidence intervals. A two-sided < 0.05 was considered to be statistically significant. All calculations were performed with the use of a commercially available statistical package (SPSS 14.0, SPSS Inc., Chicago, IL, USA).

Results

A total of 826 patients were enrolled in BASKET, of whom six (1%) died during the initial hospitalization. Of the 820 remaining patients, 801 (98%) had available data regarding discharge medication. Overall, 109 (13%) patients were treated with PPIs, i.e. esomeprazole (n = 55, 51%), pantoprazole (n = 27, 25%), omeprazole (n = 19, 17%) and lansoprazole (n = 8, 7%) (Fig. 1).

Figure 1.

 Flow chart of the study population. PPI, proton pump inhibitor.

Baseline characteristics are shown in Table 1. Patients who received PPI medication were older, more likely to be woman and have a history of gastrointestinal ulcer disease and diabetes mellitus and had higher serum creatinine levels, compared with patients who did not receive PPI therapy. However, there were no differences in terms of angiographic characteristics between the two groups of patients. Those who received PPI treatment were more likely to have had a prescription for nonsteroidal anti-inflammatory drugs and steroids both at inclusion and discharge and less likely to have had a prescription for ASA at discharge.

Table 1.   Baseline characteristics
 No PPI use (n = 692)PPI use (n = 109)P-value
  1. PPI, proton pump inhibitor; SD, standard deviation; CAD, coronary artery disease; PCI, percutaneous coronary intervention; MI, myocardial infarction; CABG, coronary artery bypass graft; STEMI, ST-segment elevation MI; ACS, acute coronary syndrome; GP, glycoprotein; ASA, aspirin; NSAID, nonsteroidal anti-inflammatory drug.

Sex, male553 (79.9%)75 (68.8%)0.009
Age, years (mean ± SD)63.3 ± 11.366.5 ± 10.50.006
Creatinine, μmol L−1 (mean ± SD)94 ± 43112 ± 970.01
History of ulcer disease23 (3.3%)9 (8.3%)0.015
History of PCI115 (16.6%)15 (13.8%)0.45
History of MI193 (27.9%)24 (22.0%)0.2
History of CABG86 (12.4%)17 (15.6%)0.36
Diabetes mellitus119 (17.2%)32 (29.6%)0.002
Dyslipidaemia525 (75.9%)80 (73.4%)0.613
Arterial hypertension450 (65.0%)79 (72.5%)0.14
Current smoking206 (29.8%)27 (24.8%)0.3
Family history of CAD270 (39.0%)35 (32.1%)0.22
Indication for PCI
 STEMI152 (22.0%)23 (21.1%)0.22
 ACS241 (34.8%)47 (43.1%)
 Angina pectoris299 (43.2%)39 (35.8%)
Medications
 GP IIb/IIIa antagonist261 (37.7%)36 (33.0%)0.35
 Coumadins (at discharge)35 (5.1%)10 (9.2%)0.083
 ASA (at discharge)670 (96.8%)100 (91.7%)0.008
 Clopidogrel (at discharge)677 (97.8%)105 (96.3%)0.28
 NSAID (at baseline)23 (3.3%)11 (10.1%)0.001
 NSAID (at discharge)15 (2.2%)8 (7.3%)0.003
Steroids (at baseline)26 (3.8%)12 (9.2%)0.001
Steroids (at discharge)25 (3.6%)12 (9.2%)0.001
Angiography
 1-vessel disease224 (32.4%)30 (27.5%)0.31
 3-vessel disease246 (35.6%)45 (41.3%)0.25
 Type A lesion96 (13.9%)14 (12.8%)0.77
 Type B lesion554 (80.1%)88 (80.7%)0.87
 Type C lesion168 (24.3%)27 (24.8%)0.91
 Bare metal stent226 (32.7%)44 (40.4%)0.11
 2.5-mm stent diameter only192 (27.8%)32 (29.4%)0.73
 3.5-mm stent diameter only177 (25.6%)23 (21.1%)0.32
 Total stent length, mm (mean ± SD)34.0 ± 20.333.8 ± 20.10.9

Patients who received concomitant PPI medication had significantly higher rates of MACE and myocardial infarction compared with patients without PPI therapy, whereas there was no statistically significant difference between the groups with regard to death, stent thrombosis and target vessel revascularization, although event rates were numerically higher in the PPI group (Table 2, Fig. 2).

Table 2.   Clinical event rates
 No PPI use (n = 692) (%)PPI use (n = 109) (%)P-value
  1. PPI, proton pump inhibitor; MACE, major adverse cardiac events; definite, probable and possible stent thromboses according to the Academic Research Consortium criteria [25].

MACE20.830.30.027
Death7.49.20.51
Myocardial infarction7.414.70.01
Stent thrombosis (all)8.111.00.31
Stent thrombosis (definite/probable)6.28.30.42
Target vessel revascularization15.320.20.2
Figure 2.

 (a) Kaplan–Meier analysis for major adverse cardiac events (MACE); (b) Kaplan–Meier analysis for myocardial infarction. PPI, proton pump inhibitor.

After correction for confounding factors, only the use of PPIs and a history of diabetes mellitus were independent predictors of myocardial infarction (Table 3).

Table 3.   Multivariate analysis for myocardial infarction
 Hazard ratio95% confidence intervalP-value
  1. PPI, proton pump inhibitor.

PPI use1.881.05–3.370.034
Diabetes mellitus1.831.07–3.150.029

Discussion

The present analysis supports the hypothesis that concomitant use of clopidogrel and PPIs in high-risk cardiovascular patients undergoing PCI is associated with increased risk of cardiac events, specifically myocardial infarction. In general, we observed numerically higher event rates in the PPI group, although the study was underpowered to provide any firm conclusions.

Current US guidelines recommend PPI therapy in DAPT to reduce gastrointestinal complications [12]. By contrast, public statements from the US Food and Drug Administration (FDA) and the European Medicines Agency (EMEA) have issued statements regarding this drug–drug interaction and discouraged the concomitant use of omeprazole respectively omeprazole or esomeprazole and clopidogrel unless absolutely necessary.

Clopidogrel is a prodrug that is metabolized via the hepatic cytochrome P450 system using the CYP2C19 enzyme. Because of the similar route of metabolism, PPIs and clopidogrel exhibit in vitro interaction with an attenuation of the antiplatelet effect of clopidogrel in the presence of most PPIs [15–17]. In contrast to these consistent in vitro data, clinical data have been variable. Whereas most retrospective studies demonstrated an increased risk of cardiovascular events for patients concomitantly treated with clopidogrel and PPIs [18, 19, 26], the prospective randomized but underpowered and prematurely terminated Clopidogrel and the Optimization of GI EveNts Trial (COGENT) did not show any effect on outcome in a cardiovascular low-risk population, when using a fixed combination of omeprazole and clopidogrel (CGT-2168) that was specifically formulated to avoid pharmacokinetic interactions [21]. COGENT was initially designed to test a gastrointestinal composite endpoint but was halted prematurely when the sponsor lost financial support. With a very low cardiovascular event rate, it was underpowered for the cardiovascular endpoints; therefore, these results may not be applicable to a high-risk patient population treated with two standard drug formulations. Additionally, a subgroup analysis of the Trial to Assess Improvement in Therapeutic Outcomes by Optimizing Platelet Inhibition with Prasugrel–Thrombolysis in Myocardial Infarction (TRITON-TIMI) 38 showed similar rates of the composite endpoint, i.e., cardiovascular death, myocardial infarction or stroke, in patients treated with PPIs and prasugrel or clopidgrel [20]. Therefore, there is an inconsistency between in vitro and clinical data and even between different clinical studies. However, a recent meta-analysis showed a possible association between underlying cardiovascular risk and outcome in patients receiving concomitant PPI and clopidogrel treatment, with increased event rates in those at high cardiovascular risk [27]. In patients with an underlying annual cardiovascular risk of <10%, there was no increased risk of concomitant PPI and clopidogrel therapy (odds ratio 1.01, 95% confidence interval 0.88–1.16; = 0.84), whereas in patients with an underlying annual cardiovascular risk of ≥10%, the rate of MACE was elevated (odds ratio 1.49, 95% confidence interval 1.41–1.57; < 0.001) [27].

BASKET was a randomized controlled trial in an all-comer PCI population with no exclusion criteria, thus representing daily practice in a large interventional cardiology centre. Therefore, our data represent important evidence with regard to potential drug–drug interactions between PPIs and clopidogrel. We have shown that patients receiving PPI treatment had a higher baseline risk of cardiovascular events, but also that PPI use was an independent predictor of myocardial infarction when adjusted for baseline differences, with most events occurring early after PCI. Therefore, this analysis extends the hypothesis of a possible drug–drug interaction between PPI and clopidogrel to patients who are at high risk of recurrent cardiovascular events and especially stent thrombosis when the effect of DAPT is most needed. Our data corroborate current recommendations from the FDA and EMEA, although the EMEA issued only initially a class-effect warning and more recently a warning against omeprazole and esomeprazole, whereas the FDA issued a warning against omeprazole alone.

As a secondary analysis of previously published data, this study is subject to inherent limitations. However, BASKET was a randomized controlled trial that enrolled an all-comer PCI population and, therefore, represents daily practice in most hospitals. To account for potential confounders, differences in baseline characteristics were adjusted by multivariate analysis. Although the sample size was limited, the population was large enough to detect important differences in outcome between the groups. Outcome analyses between the different PPIs or between the different stents were not performed because numbers in each subgroup were too small to provide relevant results.

In conclusion, in this real-world PCI population, the combination of PPIs and clopidogrel was associated with a doubling of myocardial infarction rates at 3 years. Even after adjustment for confounding factors, PPI use remained an independent predictor of outcome emphasizing the clinical importance of this drug–drug interaction in our all-comer daily practice cardiovascular population with stent ‘off-label’ use in 66% of patients. This finding underscores the importance of the recommendations of the regulatory authorities to use PPI therapy with caution when treatment with clopidogrel is required.

Funding

This work was supported by the Basel Cardiovascular Research Foundation, Basel, Switzerland.

Conflicts of interest statement

No conflicts of interest to declare.

Ancillary