Association of Clopidogrel Pretreatment With Mortality, Cardiovascular Events, and Major Bleeding Among Patients Undergoing Percutaneous Coronary Intervention:  A Systematic Review and Meta-analysis FREE

In addition to aspirin, clopidogrel has been shown to improve ischemic outcomes of patients with stable coronary artery disease following percutaneous coronary intervention (PCI)^1- 3 and of patients with acute coronary syndromes (ACS) who were either medically treated⁴ or who had undergone either revascularization by fibrinolysis or PCI.^5- 7 However, clopidogrel efficacy seems to be genetically altered in about 30% of white patients with suboptimal platelet inhibition, causing subsequent increased thrombotic event rates after coronary stenting.^8- 11

Quiz Ref IDAlthough the administration of a clopidogrel loading dose is necessary in the immediate peri-PCI setting, it is uncertain whether clopidogrel pretreatment (ie, treatment given in enoughtime before catheterization to be effective) is efficient when the coronary status is not known yet. Indeed, this pretreatment can either delay coronary artery bypass graft surgery or increase unnecessarily the risk of bleeding in patients who in the end do not need revascularization or who go to the operating room immediately after undergoing coronary angiogram. This issue has been assessed in various studies, with different clopidogrel-loading doses, timing, and clinical presentations, but none of these studies was powered for mortality. On this basis, the current European Society of Cardiology (ESC) and American College of Cardiology and American Heart Association (ACC/AHA) guidelines give a grade Ib to pretreatment in non–ST-elevation acute coronary syndrome (NSTE-ACS) scheduled for an invasive strategy.^12- 13

Previous meta-analyses have suggested a favorable risk-benefit ratio, but they examined specific clinical situations like primary PCI of ST-elevation myocardial infarction (STEMI)¹⁴ or focused on composite ischemic end points.¹⁵ No global appraisal of clopidogrel pretreatment in PCI is available. Although data have accumulated over the past 10 years, the question of a benefit with pretreatment on major clinical outcomes remains unanswered. We aimed to gather enough power from all the available data from randomized trials and registries involving patients with coronary artery disease (stable or with ACS) undergoing catheterization for potential revascularization to evaluate the association between clopidogrel pretreatment and clinical end points of mortality and major bleeding.

The main analysis included only randomized controlled trials (RCTs)^3,7,16- 18 and the PCI subgroups of RCTs for clopidogrel pretreatment vs no pretreatment^5- 6 that fulfilled the inclusion criteria, regardless of clinical presentation. Two confirmatory analyses were performed using the same method: (1) analyses of observational reports of RCTs^19- 20 and (2) analyses of observational studies^21- 26 (Table 1 and Table 2).

We conducted Cochrane Controlled Trials Registry and MEDLINE and EMBASE database searches for published articles from January 1980 through September 2012 using the following predefined search terms clopidogrel and pretreatment, or loading dose or preload or timing or upstream. Abstracts from selected major cardiology scientific meetings (AHA, ACC, ESC, and Transcatheter Cardiovascular Therapeutics) were reviewed. References from reviews and selected articles were also reviewed for potential relevant citations. We used no language restrictions. Studies were selected by 2 independent reviewers (A.B.A. and S.A.O.).

We restricted our analysis to trials that met all of the following inclusion criteria: (1) patients with coronary artery disease scheduled for catheterization, PCI, or both; (2) controlled comparison between clopidogrel pretreatment and no pretreatment (ie, placebo or no treatment) through random or nonrandom allocation; (3) data supplied on clopidogrel loading dose; (4) available data on at least mortality and bleeding. Randomized controlled trials, registries, and prespecified subgroups of studies reporting data on pretreatment with clopidogrel were considered for analysis. Full-text articles and meeting abstracts were included. Exclusion criteria were duplicate reports, the lack of control group, and ongoing studies. Data extraction and information on study design, clinical and safety outcomes were performed independently by 2 reviewers (A.B.A. and S.A.O.). Discrepancies were resolved by consensus.

Quiz Ref IDThe primary efficacy end point was mortality (by any cause). The primary safety end point was major bleeding (each study definition).

Secondary end points included the composite end point of major cardiac events but also the individual end points of myocardial infarction (MI), stroke, or urgent revascularization, as defined in each study. Cardiovascular (CV) death, minor bleeding, and stent thrombosis were also analyzed when available. Stent thrombosis was defined using the definite plus probable Academic Research Consortium definition.²⁷

Outcomes were based on the longest follow-up available for each study. Pretreatment was defined as the administration of clopidogrel before PCI or catheterization that included any loading dose of more than 300 mg or a maintenance dose of 75 mg or more when administrated for more than 5 days before PCI. End points, definitions, and study details are provided in Table 1.

The quality of RCTs included in the meta-analysis was assessed for descriptive purpose by the Jadad score for RCTs,²⁸ and the quality of nonrandomized studies by the Newcastle-OttawaScale for cohort studies (http://www.ohri.ca/programs/clinical_epidemiology/oxford.htm; see eTables 1 and 2).

To provide an estimation on the association between clopidogrel pretreatment and clinical outcomes, the results of studies were combined using a random model because we anticipated potential heterogeneity across the studies, especially between the observational studies. The results were confirmed by the Mantel-Haenszel fixed-effect model to avoid small studies being overly weighted. The main analysis was performed on RCTs and confirmation was sought through the analyses of observational data from RCTs and through observational studies.

Sensitivity analyses were performed after assessment of heterogeneity, on the following subgroups: (1) clinical presentation: elective PCI^{16- 19,21,26}; NSTE-ACS (defined as studies having a majority of patients with ACS involved in the study)^3,5,20,22,25; and STEMI^{6- 7,23- 24}; (2) clopidogrel doses: pretreatment <600 mg^{3,5- 6,19- 26,29- 30} vs ≥ 600 mg.^7,16- 18 Furthermore, to analyze more contemporary practice of pretreatment (typically a few hours before PCI), we conducted a sensitivity analysis that removed the 2 large studies—Clopidogrel in Unstable Angina to Prevent Recurrent Events (PCI CURE) and Clopidogrel as Adjunctive Reperfusion Therapy (PCI Clopidogrel as Adjunctive Reperfusion Therapy [CLARITY])—that had several days of pretreatment. To test the eventual effect of length of follow-up, we also performed a sensitivity analysis that replaced the 1-year by the 28-day follow-up results of the Clopidogrel for the Reduction of Events During Observation (CREDO) trial.

Because all studies included in the main analysis had different clinical presentations, loading doses, and timing of pretreatment, we examined the extent of heterogeneity between trials with the Cochran Q test; a P cut-off value of .10 was considered significant for heterogeneity. An I² test for heterogeneity between subgroups is reported in each figure.³¹

All probability values were 2 tailed and P = .05 was considered statistically significant. Odds ratios (ORs) with 95% CIs were calculated by use of RevMan software version 5.0 (The Cochrane Collaboration) and the language R using the meta-package.³²

Publication bias was evaluated by using funnel-plot graphs to check symmetrical distribution and convergence toward the pooled effect as the weight of the trials increased and by the fill and trim method.³³

Fifteen studies representing a total of 37 814 patients were included; of those, 5 RCTs focused solely on pretreatment,^3,7,16- 18 2 RCTs included PCI subgroups comparing pretreatment with no pretreatment and were included in the RCTs' primary analysis^5- 6 (8608 patients both); 2 prespecified nonrandomized subgroup analyses of pretreatment in RCTs^19- 20 (10 945 patients) were analyzed separately; and 6 registries^21- 26 (18 261 patients) were also analyzed separately (Figure 1). For the main analysis of RCTs, 4283 patients were pretreated (49.76%) and 4325 were not (50.24%). Seventy-five percent had ACS (n = 6510) including 25% of STEMI patients (n = 2158); the remaining patients had elective PCI (n = 2098). An observational study²³ that did not provide the loading dose of clopidogrel was excluded from the loading-dose subgroup analysis. Although another observational study²⁴ did not provide the exact timing of pretreatment, it was included because the delay time in treatment to procedure was considered sufficient.

For all studies, patients in the control groups were those who received a loading dose of clopidogrel within 2 hours of undergoing PCI^{7,16- 17,25,34} or immediately after PCI.^6,19- 24,26 The longest follow-up varied from the time patients were in the hospital to a maximum of 1 year, with a mean follow-up time of 192 days (7-365 days; Table 1). No heterogeneity existed between RCTs for major end points or for ischemic end points. Significant heterogeneity was found between observational analyses of RCTs (although not for death) and between observational studies for all end points.

In the RCT cohort of patients (n = 8608), clopidogrel pretreatment was not significantly associated with a reduction of all-cause mortality (absolute risk, 1.54% vs 1.97%; OR, 0.80; 95% CI, 0.57-1.11; P = .17). These results were consistent across the observational analyses of RCTs and the observational studies analyses (Figure 2). Cardiovascular death was available in 4 to 7 RCTs^{5- 6,16- 17}(n = 5129 patients), and the association between clopidogrel pretreatment and reduction of CV death was not present (absolute risk, 1.54% vs 1.97%; OR, 0.78; 95% CI, 0.44-1.39; P = .41). This end point was available for only 1 observational study,²¹ with consistent results (absolute risk, 1.44% vs 1.92%; OR, 0.80; 95% CI, 0.57-1.11; P = .17).

Clopidogrel pretreatment was not associated with a higher risk of major bleeding in the main analysis of RCTs (absolute risk, 3.57% vs 3.08%; OR, 1.18; 95% CI, 0.93-1.50; P = .18). Similar findings were obtained in observational analyses of RCTs and in the pooled analysis of observational studies (Figure 3).

Minor bleeding events were exploratory safety end points, available in 5 RCTs only.^{3,5- 6,17,34} There was a significant association between clopidogrel pretreatment and minor bleeding (absolute risk, 3.66% vs 2.74%; OR, 1.47; 95% CI, 1.02-2.14; P = .04); however, this was not confirmed in observational analyses of RCTs (OR, 1.01; 95% CI, 0.84-1.21; P = .91) or in observational studies (OR, 1.09; 95% CI, 0.80-1.47; P = .58) but data were available in only one¹⁹ and two^22,25 studies of these groups, respectively.

Major Coronary Event and Myocardial Infarction . In the main analysis, clopidogrel pretreatment was significantly associated with a reduction of major coronary events (absolute risk, 9.83% vs 12.35%; OR, 0.77; 95% CI, 0.66-0.89, P <.001; Figure 3) and MI (absolute risk, 4.53% vs 5.90%; OR, 0.75; 95% CI, 0.62-0.92, P = .004). These results were confirmed in the pooled analyses of observational studies but not in the analyses of observational studies of RCTs (Figure 3).

Other End Points . Stent thrombosis was exploratory because of its availability in only 1 RCT, in which the association was not significant with clopidogrel pretreatment vs no pretreatment (absolute risk, 0.98% vs 0.00%; OR, 5.07; 95% CI, 0.24-106.35; P = .30)¹⁷ and 4 observational studies,^21,24- 26 including 7497 patients, in which clopidogrel pretreatment was significantly associated with a reduction in probable and definite stent thrombosis (absolute risk, 0.78% vs 1.66%; OR, 0 .49; 95% CI, 0.31-0.78, P = .003).

Stroke was reported in 5 of the 7 RCTs^{3,6- 7,16,34} (n = 5541 patients). No significant association existed between clopidogrel pretreatment and the reduction in stroke (absolute risk, 0.54% vs 0.94%; OR, 0.59; 95% CI, 0.31-1.12, P = .11); data were available for only 1 observational study,²⁵ for which no association existed between pretreatment and reduction in stroke (absolute risk, 0.21% vs 0.00%; OR, 3.69; 95% CI, 0.15-90.90, P = .42).

Urgent revascularization was available in 5 RCTs^3,7,16- 18(n = 4087 patients) and 4 observational studies^{22- 23,25- 26} (n = 15 965 patients). There was no significant association between urgent revascularization and clopidogrel pretreatment in the primary analysis of the RCTs (absolute risk, 1.47% vs 1.61%; OR, 0.91; 95% CI, 0.56-1.49; P = .71). Results were consistent in the observational studies analysis (absolute risk, 5.34% vs 2.15%; OR, 0.98; 95% CI, 0.74-1.30; P = .89).

There were no available data on these end points from observational analyses of RCTs.

There was no heterogeneity between RCTs as assessed by the Cochran Q test, but significant heterogeneity was found in observational analyses of RCTs and observational studies for several end points.

Results for the prespecified RCT subsets of clinical presentation are summarized in eFigure 1. For all-cause mortality, a significant association existed between clopidogrel pretreatment and a reduction of death in the STEMI subgroup only (absolute risk, 1.28% vs 2.54%; OR, 0.50; 95% CI, 0.26-0.96; P = .04; number needed to treat, 79; eFigure 2). Clopidogrel pretreatment in STEMI was also significantly associated with a reduction in major coronary events (absolute risk, 3.56% vs 6.36%; OR, 0.54; 95% CI, 0.36-0.81; P = .003; number needed to treat, 36) as well as in patients with NSTE ACS (absolute risk, 13.91% vs 17.19%; OR, 0.78; 95% CI, 0.66-0.91; P = .002; number needed to treat, 30) but not in the lower-risk population of elective PCI (eFigure 1).

Safety outcomes did not differ across the 3 subgroups of clinical presentation by treatment (eTable 3). Similarly, little difference existed between the groups of doses (<600 vs ≥600 mg) on the various efficacy and safety end points (eTable 4). Because there was no significant heterogeneity between subgroups by clinical presentation and clopidogrel loading dose, these results are only exploratory and should not been considered as definitive.

The other sensitivity analyses on pretreatment delay (excluding PCI CURE and PCI CLARITY) and duration of follow-up (using the 28-day results of CREDO) showed results similar to those of the main analysis.³

Quiz Ref IDAlthough data have accumulated over the past 10 years about the efficacy and safety of clopidogrel pretreatment, the association of this strategy with a decrease in all-cause death remains uncertain because no large RCT has addressed this issue. The current meta-analysis collected information in a population of more than 37 000 patients and found no significant association between clopidogrel pretreatment and survival nor between clopidogrel pretreatment and major bleeding. This meta-analysis demonstrated, however, a significant association between clopidogrel pretreatment and the reduction of major coronary events or MIs in the primary RCTs analyses combining all types of patients, with fully consistent results obtained from observational analyses of RCTs and observational studies. Although no significant heterogeneity existed for clinical presentation, the higher-risk STEMI population appeared to gain the most benefit from pretreatment. In contrast, patients undergoing elective PCI had no apparent benefit from clopidogrel pretreatment, questioning the need of such a systematic strategy at least in low-risk patients.

Clopidogrel pretreatment has been largely accepted and applied in accordance with the ACC/AHA and ESC guidelines,^12- 13,35 which are based on evidence from the results of 3 studies performed in the early 2000s. First, in the PCI-CURE substudy,⁵ clopidogrel pretreatment (300 mg loading dose with a median of 10 days before catheterization) was associated with a 30% reduction of the composite end point of CV death, MI, and urgent target vessel revascularization at 30 days, without significant difference in major bleeding. Although all-cause mortality was not reported in the study, CV death was not different between groups and reduction in major coronary events was driven by the decrease in periprocedural MI.

Second, the CREDO trial³ demonstrated an interaction between the timing of pretreatment and the protection from adverse CV events in stable patients undergoing scheduled angioplasty. Indeed, although not powered to show a difference in ischemic end points, the group of patients loaded with 300 mg of clopidogrel at least 6 hours before PCI experienced a 38.6% decrease in major coronary events that reached significance among those pretreated at least 15 hours before the procedure.^3,36

Third, the CLARITY-PCI substudy⁶ found a significant decrease in major coronary events, with no increased risk of bleeding in patients undergoing secondary PCI after fibrinolysis associated with clopidogrel pretreatment. Additional evidence comes from smaller studies and a meta-analysis that used different doses of clopidogrel for elective PCI,^37- 38 NSTEMI,^39- 40 and STEMI,^39,41 and from the CURRENT-OASIS 7 subgroup analysis of patients with ACS undergoing PCI.⁴² In the latter, all patients were pretreated with 600 mg followed by 150 mg of a maintenance dose for 7 days and 75 mg daily thereafter, which significantly reduced by 14% the combined primary end point of CV death, MI, and stroke compared with the standard 300-mg loading dose and 75-mg maintenance dose; there was no excess of thrombosis in MI major bleeding.⁴³

With this accumulation of data, administration of clopidogrel before PCI became the rule with a class I recommendation. The ESC recommends pretreatment with a 300-mg loading dose for more than 6 hours before elective PCI (or 600 mg >2 hours before; class Ic),⁴² and a 600-mg loading dose as soon as possible for primary PCI for patients with STEMI (class Ic).⁴² The low level of evidence reflects the absence of indisputable evidence. In the more recent ESC guidelines, the higher loading dose of 600 mg for PCI for patients with NSTE-ACS has a class Ib recommendation when ticagrelor or prasugrel is not an option.¹² The ACC/AHA PCI guidelines for patients undergoing elective PCI and patients with NSTE-ACS or STEMI who are scheduled for PCI also recommends pretreatment with a 300- to 600-mg loading dose (preferably 600 mg) as early as possible before PCI (class Ib recommendation).^13,44

Several issues remain unresolved with regard to the evidence supporting the recommendations. First, many of the patients included had their PCIs postponed for up to several days after pretreatment and therefore do not accurately reflect contemporary practice of early revascularization, often performed within hours of first medical contact. Our meta-analysis includes the most recent studies including registries and may better reflect current real-world practice. The concordance between RCTs and observational studies supports our findings.

Second, it is unclear whether all patients or just certain types of presentation should be pretreated, although the recommendations are quite broad in favor of pretreatment. Our data suggest a benefit limited to the patients at higher risk, STEMI especially, with a significant association between clopidogrel pretreatment and the reduction in hard outcomes. However, because of the absence of significant heterogeneity by condition, these results are only informative and would need additional exploration in specific trials.

Third, none of the studies or previous meta-analyses was adequately powered to show a benefit on all-cause mortality. The results of our meta-analysis, which is powered for mortality, do not support clopidogrel pretreatment on this basis in the overall PCI population.

Quiz Ref IDThe reasons clopidogrel pretreatment does not decrease mortality in the stable patients could not be ascertained but several hypotheses may be generated. First, the benefit of this treatment may be related to baseline platelet reactivity, which is lower in more stable patients.^45- 46 Second, clopidogrel, both for loading and maintenance doses, is associated with moderate level of inhibition and a wide variability in response that may explain the absence of effect on mortality.⁴⁷

Third, the benefit expected from pretreatment during the waiting period for catheterization of patients with NSTE-ACS may not be detectable with clopidogrel, which has a slow onset of action, a delay of action critical in the current era when the speed at which patients transfer to the catheterization laboratory is faster than it was 10 years ago. Progress has been made in time to catheterization, in PCI procedures, and in new antiplatelet strategies, which challenges the concept of pretreatment. Some studies have shown that delays to revascularization can be substantially shortened in NSTE-ACS^48- 49 while studies like ACCOAST (A Comparison of Prasugrel at PCI or Time of Diagnosis of Non-ST Elevation Myocardial Infarction, NCT01015287) and ATLANTIC (A 30 Day Study to Evaluate Efficacy and Safety of Pre-hospital vs In-hospital Initiation of Ticagrelor Therapy in STEMI Patients Planned for Percutaneous Coronary Intervention [PCI],NCT01347580) are evaluating, in the modern era of PCI, the concept of pretreatment with new P2Y₁₂ inhibitors in NSTE-ACS and STEMI, respectively. These studies are needed to understand better the role of pretreatment not evaluated in the large pivotal trials that tested the new P2Y₁₂ inhibitors prasugrel (no pretreatment) and ticagrelor (systematic pretreatment).

Fourth, most of the serious events, in particular death, do not occur in the catheterization laboratory but are delayed by days or weeks when clopidogrel is actually effective and death then relates to other factors of poor prognosis.

Fifth, one-third of elective coronary angiograms are normal,⁵⁰ which does not justify clopidogrel pretreatment that exposes patients to bleeding risk with no benefit to be expected on ischemic events or mortality.¹⁸

Finally, some patients with NSTE-ACS could be reoriented toward urgent coronary artery bypass revascularization, during which clopidogrel exposure may double the rate of reoperation because of bleeding, thus increasing transfusions and deaths.⁵¹ Bleeding in general is a driver of mortality, and the present meta-analysis suggests a significant excess of major bleeding in lower-risk patients.

We acknowledge several limitations of our study that, as in all systematic reviews, may have been influenced by several forms of bias; we tried to minimize selection bias by using a predefined search strategy, with independent selection and data extraction by 2 independent reviewers, without any language restriction and including registries and RCTs as well as their sub-analyses. Furthermore, funnel plots or the fill and trim method did not suggest any publication bias. This meta-analysis, however, was not performed on individual data. For the initial selection of studies, data abstractors were not blinded to authors, affiliations, and journals. Observational data have limitations inherent to the nature of the data collected with potential serious biases. The pooled analyses of these studies have been reported separately and should be seen only as supportive of the main analysis performed on RCTs.

Despite regrouping a large number of patients, this meta-analysis of randomized trials only may have had an insufficient statistical power to definitely exclude an effect on mortality. Because the CIs are wide, and even if an OR of 0.80 was found, it is not reasonably possible to exclude that the pretreatment efficacy may be greater (at best a reduction of mortality of 43%) or smaller (at worse an increase in mortality of 11%). Bleeding could furthermore have mitigated the mortality benefit, but the information on fatal bleeding is lacking in most publications.

Because of varied designs, end point definitions, patient presentations, clopidogrel loading doses, timing of loading before PCI, other drug regimens and durations of follow-up, we have conducted several confirmatory subset analyses along with the main analysis on RCTs that was not hampered by any heterogeneity between trials. The fact that observational studies analyses are fully consistent with the RCTs analysis support our findings.

Furthermore, because we cannot completely exclude a potential interaction between clopidogrel pretreatment and an unknown factor of heterogeneity, and although no multilevel meta-analysis was done, we performed multiple sensitivity analyses that do not suggest any difference from the main analysis when considering many of these covariate variables. It is thus unlikely that such unidentified interactions could affect the primary end point of the study. In all cases, most subgroup comparisons must be viewed as exploratory and not as definitive, and specific research on these subgroups should be performed to better understand the effect of pretreatment.

Quiz Ref IDIn conclusion, among patients scheduled for PCI, clopidogrel pretreatment was not associated with a lower risk of overall mortality, but was associated with a significantly lower risk of a major coronary event. Although a pretreatment strategy has been recommended for years in patients undergoing PCI, this study shows the limits of the available evidence, with no significant benefit on hard outcomes. The value of pretreatment, including with new antiplatelet agents, needs to be assessed in large prospective studies.