Pharmacodynamic Impacts of Proton Pump Inhibitors on the Efficacy of Clopidogrel In Vivo—A Systematic Review

Abstract

Background

There is considerable debate about whether concomitant use of proton pump inhibitors (PPIs) should be recommended for patients who are prescribed clopidogrel after acute coronary syndrome. Most pharmacokinetic and pharmacodynamic studies in vivo were conducted using small sample sizes and were single centered, resulting in conflicting data.

Hypothesis

PPIs may attenuate the antiplatelet effect of clopidogrel in vivo and lead to an increased risk of cardiovascular events.

Methods

PubMed, the Cochrane Library, Embase, Web of Science, and China Biology Medicine Disc were searched. Randomized controlled trials that compared pharmacodynamic impacts of a PPI on the efficacy of clopidogrel in vivo were included. Two independent reviewers evaluated study quality and extracted data for meta‐analysis.

Results

We identified 8 eligible studies. Compared to clopidogrel treatment alone, patients who received both a PPI and clopidogrel had less of a decrease in the platelet reactivity index (weighted mean difference [WMD]: 8.18; 95% confidence interval [CI]: 6.81–9.56; P < 0.00001), less adenosine 5′‐diphosphate–induced platelet aggregation inhibition (WMD: 7.28; 95% CI: 2.44–12.11; P = 0.003), higher P2Y12 reaction units (WMD: 40.58; 95% CI: 19.31–61.86; P = 0.0002), and higher risks of clopidogrel resistance (odds ratio [OR]: 2.49; 95% CI: 1.49–4.14; P = 0.0005). There were no significant differences, however, for the incidences of major adverse cardiovascular events between the 2 groups (OR: 1.07; 95% CI: 0.44–2.59; P = 0.88), and treatment with a PPI and clopidogrel significantly reduced the risk of adverse gastrointestinal events (OR: 0.16; 95% CI: 0.04–0.62; P = 0.008).

Conclusions

Concomitant use of a PPI with clopidogrel attenuated the antiplatelet effect of clopidogrel, but may be clinically unimportant because there were no clinical differences in the risk for major adverse cardiovascular events.

Additional Supporting Information may be found in the online version of this article.

The authors have no funding, financial relationships, or conflicts of interest to disclose.

Introduction

Clopidogrel is widely prescribed with aspirin for the prevention of atherothrombosis in patients with acute coronary syndrome (ACS). However, many clinical studies have reported that double antiplatelet therapy leads to the risk of adverse gastrointestinal events such as peptic ulcers or hemorrhages, which could compromise quality of life or even cause death. The American College of Cardiology Foundation/American College of Gastroenterology/American Heart Association (ACCF/ACG/AHA) 2008 expert consensus recommended that patients prescribed clopidogrel with aspirin should receive a proton pump inhibitor (PPI) to reduce gastrointestinal bleeding.1 Recent studies have raised concerns about the potential interaction between clopidogrel and PPIs. Clopidogrel is an inactive prodrug that is converted to an active thiol metabolite in a 2‐step process via hepatic cytochromes (CYPs), mainly CYP2C19, which inhibit adenosine 5′‐diphosphate (ADP)‐induced platelet aggregation by irreversibly blocking the platelet P2Y₁₂ receptor. However, CYP2C19 is also a principal enzyme in PPI metabolism. Reduced CYP2C19 function results in less PPI inactivation and increased acid inhibition. Mechanistic in vitro studies have demonstrated that PPIs may decrease metabolism of clopidogrel to its active metabolite by competitively inhibiting CYP2C19.2 Consistent with this finding, several observational studies have also shown a higher risk of cardiovascular events such as stent thrombosis and myocardial infarction (MI) in those prescribed both clopidogrel and a PPI. As a result, the US Food and Drug Administration and European Medicines Agency stated in 2009 and 2010 that “health‐care providers should re‐evaluate the need for starting or continuing treatment with a PPI in patients taking clopidogrel.”3 The ACCF/ACG/AHA consensus was also updated in 2010. The challenge for healthcare providers is to determine the risk/benefit balance for individual patients or subsets of the target population.4

Several pharmacokinetic and pharmacodynamic studies have been conducted in vivo to compare platelet function in clopidogrel patients treated with or without a PPI. Most of these studies, however, were conducted with small sample sizes and were single centered, resulting in conflicting data. To address this gap in knowledge and to provide a summary of the current available evidence, we conducted a systematic review to quantitatively evaluate the pharmacodynamic impacts of PPIs on the antiplatelet effect of clopidogrel.

Methods

Data Sources and Searches

Literature searches were performed in MEDLINE, Embase, ISI Web of Science, the Cochrane Database, and China Biology Medicine Disc (CBM) starting from the beginning of indexing to June 31, 2012 by 2 independent investigators (C.J. and L.J.J). We used the following words as search terms: clopidogrel, PPI, proton pump inhibitor, platelet, omeprazole, rabeprazole, lansoprazole, and esomeprazole. Additionally, studies in the reference lists of the identified articles were also hand searched. Only articles written in English or Chinese were included.

Study Selection

Inclusion criteria were: (1) Studies were in vivo randomized controlled trials (RCTs) using either a parallel group or crossover study design. (2) The subjects were patients with ACS who received clopidogrel as antiplatelet therapy, with or without aspirin. (3) The articles reported platelet function as at least 1 of the described end points. Exclusion criteria were: (1) The subjects included healthy volunteers or the subjects received GPIIb/IIIa antagonist with clopidogrel. (2) The subjects included patients with contraindications to the use of clopidogrel or PPIs. (3) The articles could not provide enough data for the meta‐analysis even after statistical computing.

Data Extraction and Quality Assessment

Platelet function was examined as end points using several measures, including the platelet Reactivity Index (PRI). Platelet reactivity was assessed by measuring platelet phosphorylated vasodilator‐stimulated phosphoprotein in whole blood. Platelet mean fluorescence intensity (MFI) was determined using a flow cytometer. Results were expressed as: PRI = ([MFI_PGE1−MFI_PGE1+ADP]/MFI_PGE1)× 100. Other measures used were ADP‐induced platelet aggregation inhibition; P2Y12 reaction units (PRU), whereby the P2Y12 VerifyNow system was used to measure platelet‐induced aggregation as an increase in light transmittance; and clopidogrel resistance. Patients were considered poor responders to clopidogrel if PRI was >50% or when a PRU value of at least 240 was defined.5, 6

All eligible studies were reviewed independently by 2 investigators (C.J. and L.J.J.), and relevant data were extracted for the further analysis. We then evaluated the quality and bias of the studies using the standards of the Cochrane Handbook version 5.1.0.

Data Synthesis and Analysis

We used Review Manager version 5.1 for Windows (The Nordic Cochrane Centre, The Cochrane Collaboration, Copenhagen, Denmark) to perform the meta‐analysis. Statistical heterogeneity of the studies was identified using a χ ² test and was quantified using an I ² test to assess its impact on the meta‐analysis. The studies were considered statistically heterogeneous if I ² > 50% and P ≤ 0.1. If heterogeneity existed, a random‐effects model was used to combine the outcomes of the separated studies. Otherwise, a fixed‐effects model was undertaken. Pooled analyses were conducted using the weighted mean difference (WMD) for continuous variables. WMD was calculated with 95% confidence intervals (CIs) using the inverse variance method. Odds ratio (OR) was estimated using the Mantel‐Haenszel method for dichotomous variables. A sensitivity analysis was also conducted to estimate the influence of each study on the overall results by repeating the meta‐analysis while omitting 1 study at a time. Funnel plots and statistical tests for funnel plot asymmetry were used to test publication bias.

Results

Search Results and Study Characteristics

The search strategy generated 252 citations: 76 from MEDLINE, 30 from the Web of Science, 45 from the Cochrane Database, 24 from Embase, and 77 from CBM. After screening the titles and abstracts, we excluded 229 articles due to duplications, language restrictions, unavailable full texts, non‐RCTs, or because they did not meet our criteria for other reasons. We reviewed the full text of the remaining 23 studies. A total of 8 studies7, 8, 9, 10, 11, 12, 13, 14 met our study criteria. Five of them were in English, and 3 were in Chinese. Two studies were placebo‐controlled trials, and 6 were blank‐controlled trials. The study flow diagram is shown in Figure 1. The characteristics and biases of all 8 studies are summarized in Table 1 and Table 2.

Figure 1.

Study flow diagram. Abbreviations: CBM, China Biology Medicine Disc; RCTs, randomized controlled trials.

Table 1.

Characteristics of Selected Studies

Study	Age, y	PPI Studied	Population	Study End Points
Fernando 20117	62 ± 11	E	PPI: 31; control: 31	PRI, PRU, clopidogrel nonresponders
Fontes‐Carvalho 20118	61.1 ± 11.9	O, P	PPI: 62; control: 31	PRU, clopidogrel nonresponders
Gilard 20089	PPI: 62.28 ± 15.04; control: 63.67 ± 12.22	O	PPI: 64; placebo: 60	PRI, clopidogrel nonresponders
Hsu 201110	PPI: 70.6 ± 11.5; control: 73.3 ± 10.7	E	PPI: 21; control: 21	ADP‐induced platelet aggregation inhibition, MACE, AGE
Zhang 2010.11	PPI: 57.1 ± 1.2; control: 57.8 ± 1.2	E	PPI: 33; control: 41	ADP‐induced platelet aggregation inhibition
Cai 201012	NR	O, P	PPI: 40; control: 20	ADP‐induced platelet aggregation inhibition, MACE, AGE
Guo 200913	PPI: 67.2 ± 11.2; control: 66.5 ± 12.9	O, E	PPI: 40; control: 20	PRI
Peng 201014	PPI: 59.82 ± 5.40; control: 62.35 ± 6.73	O, P	PPI: 60; control: 30	ADP‐induced platelet aggregation inhibition, MACE, AGE

Abbreviations: ADP, adenosine 5′‐diphosphate; AGE, adverse gastrointestinal events; E, esomeprazole; MACE, major adverse cardiovascular events; NR, not reported; O, omeprazole; P, pantoprazole; PPI, proton pump inhibitor; PRI, Platelet Reactivity Index; PRU, P2Y12 reaction units; R, rabeprazole.

Table 2.

Bias of Selected Studies

Study	Random Sequence Generation	Allocation Concealment	Blinding of Participants and Personnel	Blinding of Outcome Assessment	Incomplete Outcome Data	Selective Reporting
Fernando 20117	Low	Unclear	Low	Low	Low	Low
Fontes‐Carvalho 20118	Low	Unclear	Unclear	Unclear	Low	High
Gilard 20089	Unclear	Unclear	Low	Low	Low	Low
Hsu 201110	Low	Low	High	Low	High	Low
Zhang 201011	Unclear	Unclear	High	Unclear	High	High
Cai 201012	Low	Unclear	High	Unclear	High	High
Guo 200913	Unclear	High	High	Unclear	Low	Low
Peng 201014	Unclear	Unclear	High	Unclear	High	Low

Impacts of PPI on the Antiplatelet Effect of Clopidogrel

PRI

Three studies provided PRI results, and 362 patients from these studies were analyzed. Statistical heterogeneity was not found (P = 0.56, I ² = 0), so a fixed‐effects model (inverse variance method) was used. Compared to patients who received clopidogrel treatment alone, those who received both a PPI and clopidogrel had less of a decrease in PRI (WMD: 8.18; 95% CI: 6.81–9.56; P < 0.00001). This indicated that concomitant use of a PPI with clopidogrel may diminish the antiplatelet effect of clopidogrel.

ADP−Induced Platelet Aggregation Inhibition

Three studies included ADP‐induced platelet aggregation inhibition, and 192 patients from these studies were analyzed. Statistical heterogeneity was not found (P = 0.17, I ² = 43%), so a fixed‐effects model (inverse variance method) was used. The pooled analysis showed that patients who received both a PPI and clopidogrel had less ADP‐induced platelet aggregation inhibition (WMD: 7.28; 95% CI: 2.44–12.11; P = 0.003), which indicated that concomitant use of a PPI with clopidogrel may attenuate the antiplatelet effect of clopidogrel (Figure 2).

Figure 2.

Forest plot comparing adenosine 5′‐diphosphate–induced platelet aggregation inhibition in the group treated with clopidogrel and a proton pump inhibitor (PPI) to the group treated with clopidogrel alone. Abbreviations: CI, confidence interval; SD, standard deviation.

PRU

Two studies provided PRU values, and 155 patients from these studies were analyzed. There was statistical heterogeneity between the studies (P = 0.05, I ² = 74%), so a random‐effects model (inverse variance method) was used. Compared to patients who received clopidogrel treatment alone, those who received both a PPI and clopidogrel had higher PRU values (WMD: 40.58; 95% CI: 19.31–61.86; P = 0.0002).

Clopidogrel Resistance

Three studies included the risk of clopidogrel resistance, and 279 patients from these studies were analyzed. Statistical heterogeneity was not found (P = 0.16, I ² = 45%), so a fixed‐effects model (Mantel‐Haenszel method) was used. The risk of being a poor responder to clopidogrel when concomitantly treated with a PPI was significantly higher than being treated with clopidogrel alone (OR: 2.49; 95% CI: 1.49–4.14; P = 0.0005).

Major Adverse Cardiovascular Event

Clinical end points were analyzed in 3 RCTs for 314 patients. We did not find statistical heterogeneity across the studies (P = 0.71, I ² = 0), so a fixed‐effects model (Mantel‐Haenszel method) was used. Sixteen of 183 patients treated with clopidogrel and a PPI, and 9 of 132 patients treated with clopidogrel but without a PPI developed a major adverse cardiovascular event during follow‐up (OR: 1.07; 95% CI: 0.44–2.59; P = 0.88). Therefore, there was not enough evidence to indicate that concomitant use of a PPI with clopidogrel increased the risk of major adverse cardiovascular events (Figure 3).

Figure 3.

Forest plot comparing the risk for major adverse cardiovascular events in the group treated with clopidogrel and a proton pump inhibitor (PPI) to the group treated with clopidogrel alone. Abbreviations: CI, confidence interval, M‐H, Mantel‐Haeszel method.

Adverse Gastrointestinal Events

Adverse gastrointestinal events were reported in 2 articles. Statistical heterogeneity between the studies was not found (P = 0.43, I ² = 0), so a fixed‐effects model (Mantel‐Haenszel method) was used. The risk of developing a gastrointestinal event during follow‐up was significantly lower in the group receiving clopidogrel with a PPI than in the group receiving clopidogrel without a PPI (OR: 0.16; 95% CI: 0.04–0.62; P = 0.008).

Sensitivity Analysis

A sensitivity analysis was conducted to estimate the influence of each study on the overall results by repeating the meta‐analysis while omitting 1 study at a time. For the analysis of clopidogrel resistance, the overall odds ratio was adjusted to 1.58 (95% CI: 0.78–3.17) after omitting Gilard (2008).9 This indicated that the risk of being a poor responder to clopidogrel did not increase when concomitantly treated with a PPI. For the analysis of ADP‐induced platelet aggregation inhibition, the overall WMD was adjusted to 0.00 (95% CI: −9.10 to 9.09) after omitting Peng (2010).14 The rest of the sensitivity analysis did not show any significant differences from our original findings.

Publication Bias

The shape of the funnel plots for studies seemed symmetrical, indicating no statistical evidence of publication bias. However, the number of the published studies in our review was very limited, so publication bias could not be completely dismissed.

Discussion

The interaction of PPIs and clopidogrel has recently raised concerns about the safety of prophylactic PPI use for gastrointestinal events. Although nonstandard measurements have been universally accepted, analysis of platelet function in vitro with different assay methods demonstrated that PPIs could diminish the antiplatelet effect of clopidogrel. Our study showed that a similar attenuation also exists in vivo.

Several large retrospective observation studies reported that patients prescribed both clopidogrel and PPIs experienced a significant increase in cardiovascular events.15, 16 The results of RCTs, however, do not necessarily agree. Thus far, the Clopidogrel and the Optimization of Gastrointestinal Events (COGENT) study is the only large‐sample, randomized, double‐blind, placebo‐controlled study with clinical end points.17 A total of 3761 patients were assigned to either the omeprazole or placebo group and followed for an average of 106 days. There were no apparent cardiovascular interactions with clopidogrel (hazard ratio [HR]: 0.99, 95% CI: 0.68–1.44). Instead, among patients receiving aspirin and clopidogrel, prophylactic use of a PPI reduced the rate of upper gastrointestinal bleeding (HR: 0.34, 95% CI: 0.18–0.63). The results of the COGENT study are consistent with the clinical outcomes in our study; the attenuation of the antiplatelet effect that is caused by PPIs may not translate into a worse clinical prognosis.

Due to the inconsistency of these findings, the American College of Cardiology Foundation/American Heart Association/Society for Cardiovascular Angiography and Interventions guidelines for percutaneous coronary intervention stated in 2011 that “PPI should be used in patients with a history of prior gastrointestinal bleeding,” but “routine use of a PPI is not recommended for patients at low risk of gastrointestinal bleeding,” and “platelet function testing may be considered in patients at high risk for poor clinical outcomes.”18 Both the 2012 American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines/American College of Physicians/American Association for Thoracic Surgery/ Preventive Cardiovascular Nurses Association/ Society for Cardiovascular Angiography and Interventions/ Society of Thoracic Surgeons Guideline for the Diagnosis and Management of Patients with Stable Ischemic Heart Disease and the 2012 American College of Cardiology Foundation/American Heart Association update of the Guideline for the Management of Patients with Unstable Angina/Non–ST‐Elevation MI did not prohibit the use of PPI agents in appropriate clinical settings. However, they did highlight the potential risks and benefits of the combination therapy and mentioned that, as an alternative, pantoprazole or histamine H2 receptor antagonists could be sufficient in some patients.19, 20

The strengths of this systematic review stemmed from the comprehensive study search as well as the detailed data extraction and analysis. Because relevant in vivo studies were mostly small sampled and single centered, our study was the first to use meta‐analysis in such pharmacodynamic studies to increase the power and improve the precision of currently available evidence. We analyzed only RCTs because observational studies are usually influenced by bias, which may lead to underestimating or overestimating the true effect. Furthermore, objective assessments of the methodological quality of each study were conducted according to the standards established by the Cochrane Handbook version 5.1.0, to emphasize the risk of bias in the results. Instead of using a scale, the validity assessment was fully reported because scales have been shown to be unreliable and are less likely to be transparent to readers of the review. We included studies whose participants were patients with ACS and excluded those containing volunteers who were relatively young and healthy. This made the results of our study more relevant to real‐world settings.

The study had several limitations. First, unmeasured or residual confounding was likely to be present in these RCTs. Recent studies have shown that statins might also decrease the metabolism of clopidogrel to its active metabolite via CYP3A4. However, only 3 of the included studies reported the use of statins among groups.9, 11, 13 This led to residual bias in the outcomes, because the duration and dose of combination therapies using statins and clopidogrel may also be relevant to drug‐drug interactions and complicate the problem. The actual influence of a PPI on the efficacy of clopidogrel could not be distinguished from the potential influence of statins.

Additionally, the impact of PPIs on clopidogrel efficacy was considered a result of the competitive inhibition of cytochrome P450 2C19 isoenzyme. According to in vitro studies, pantoprazole and esomeprazole may be the weakest inhibitors of this isoenzymes,21 which means the drug interaction may not be a class effect. For the analysis of clopidogrel resistance, only Gilard et al9 used omeprazole, whereas esomeprazole and both omeprazole and pantoprazole were used in Fernando et al7 and Fontes‐Carvalho,8 respectively. The sensitivity analysis showed that the overall OR of being a poor responder to clopidogrel was adjusted from 2.49 (95% CI: 1.49–4.14) to 1.58 (95% CI: 0.78–3.17) after omitting Gilard et al.9 This also indicated that the drug interaction between a PPI and clopidogrel could not be assessed as simply a class effect, and that omeprazole may inhibit CYP2C19 more strongly when compared to either pantoprazole or esomeprazole.

Genetic polymorphisms of cytochrome P450 enzymes, especially CYP2C19, could also lead to individual responsiveness to clopidogrel. Carriers of a reduced‐function allele are more frequently resistant to clopidogrel. Mega et al reported that carriers of at least 1 CYP2C19 reduced‐function allele had a relative decrease of 32.4% in plasma exposure to the active clopidogrel metabolite and an absolute decrease in maximal platelet aggregation in response to clopidogrel.22 However, only 2 of all 8 studies examined the participants for CYP2C19 genotype.7, 10 Therefore, other factors besides PPIs that reduce the function of the CYP2C19 enzyme also require attention.

Finally, as in any meta‐analysis, the possibility of publication bias is of concern because small studies with null results tend not to be published.

Conclusion

This systematic review supported that the concomitant use of a PPI with clopidogrel may attenuate the pharmacodynamics of clopidogrel in vivo. This attenuation did not result in clinically significant adverse outcomes. However, evidence for protection from gastrointestinal events by concomitant use of a PPI with clopidogrel was more convincing. Future clinical studies should test patients for platelet function and possibly conduct CYP2C19 genotyping in addition to observing clinical outcomes. This would provide more precise evidence regarding the correlation between PPIs and clopidogrel. Studies with good experimental designs and large sample sizes are also still needed to confirm the results observed here.

Supporting information

Figure S1: Forest plot comparing platelet reactivity index variance in the group treated with clopidogrel and a proton pump inhibitor (PPI) to the group treated with clopidogrel alone. Abbreviations: CI, confidence interval; SD, standard deviation.

Figure S2: Forest plot comparing P₂Y₁₂ reaction units and values in the group treated with clopidogrel and a proton pump inhibitor (PPI) to the group treated with clopidogrelal one. Abbreviations: CI, confidence interval; SD, standard deviation.

Figure S3: Forest plot comparing clopidogrel resistance in the group treated with clopidogrel and a proton pump inhibitor (PPI) to the group treated with clopidogrel alone. Abbreviations: CI, confidence interval.

Figure S4: Forest plot comparing the risk of adverse gastrointestinal events in the group treated with clopidogrel and a proton pump inhibitor (PPI) to the group treated with clopidogrel alone. Abbreviations:CI, confidence interval.