Author + information
- Received August 2, 2002
- Revision received September 19, 2002
- Accepted September 26, 2002
- Published online January 1, 2003.
- ↵*Reprint requests and correspondence:
Dr. John P. A. Ioannidis, Chairman, Department of Hygiene and Epidemiology, University of Ioannina School of Medicine, Ioannina 45110, Greece.
Objectives We sought to evaluate the impact of intravenous antagonists of the platelet IIb/IIIa receptor on the survival of patients undergoing percutaneous coronary interventions (PCIs).
Background Several trials have shown that intravenous antagonists of the platelet glycoprotein (GP) IIb/IIIa receptor reduce the incidence of myocardial infarction (MI) and composite cardiac outcomes (death, MI, or revascularization) in patients undergoing PCI. However, individual studies have not had adequate power to examine differences in mortality.
Methods We performed a meta-analysis of 19 randomized, placebo-controlled trials (20 comparisons, n = 20,137). Death was the primary outcome. Secondary outcomes included MI, composite cardiac outcomes, and major bleeding.
Results Mortality was significantly reduced at 30 days (risk ratio [RR] 0.69 [95% confidence interval [CI] 0.53 to 0.90]), at six months (RR 0.79 [95% CI 0.64 to 0.97]), and including longer follow-up (RR 0.79 [95% CI 0.66 to 0.94]), with no significant between-study heterogeneity. The relative risk reduction was largely similar in trials of patients with or without acute myocardial infarction (AMI), in trials continuing or discontinuing heparin after the procedure, and in trials using stents or another PCI as the intended primary procedure. Myocardial infarction and composite outcomes were significantly reduced (p < 0.001 for all) at 30 days and six months. Major bleeding was significantly increased only in trials where heparin infusion was continued after the procedure (RR 1.70 [95% CI 1.36 to 2.14]), although there was no excess bleeding when heparin was discontinued (RR 1.02 [95% CI 0.85 to 1.24]).
Conclusions In patients undergoing PCI, GP IIb/IIIa receptor antagonists confer a significant and sustained decrease (20% to 30%) in the risk of death.
Platelet glycoprotein (GP) IIb/IIIa receptor antagonists significantly reduce the incidence of myocardial infarction (MI) and the need for revascularization among patients who undergo percutaneous coronary interventions (PCIs) (1–8). However, previous trials (1–6)and meta-analyses (7,8)could not conclusively determine whether these drugs also significantly reduce mortality, per se. Studies of patients with acute coronary syndromes where PCIs were performed only selectively and at the physician’s discretion show no significant survival benefit (9), except for diabetics (10). It remains unclear whether platelet GP IIb/IIIa receptor antagonists can reduce mortality in patients committed to undergo PCI.
Several additional pertinent trials have been conducted recently, especially in acute myocardial infarction (AMI) patients (11–15). Therefore, we performed a comprehensive meta-analysis of 20 randomized comparisons to clarify the effect of intravenous GP IIb/IIIa receptor antagonists on mortality in patients undergoing PCI with various clinical backgrounds.
Eligibility and search strategy
The meta-analysis included randomized trials comparing any type of intravenous GP IIb/IIIa receptor antagonist (e.g., abciximab, eptifibatide, tirofiban, lamifiban) against placebo or no treatment in patient populations where all subjects underwent PCI (elective, urgent/emergent). Eligible PCIs included percutaneous transluminal coronary angioplasty (PTCA), stent placement, and atherectomy (directional, rotational, or excimer laser). We only targeted comparisons where PCI choices were the same in the compared arms. We allowed different dosing of concomitant heparin. We excluded trials where only selected patients eventually underwent PCI, because the decision for PCI subsequent to randomization would be influenced by the clinical course.
We identified eligible trials in MEDLINE, EMBASE, and the Cochrane Controlled Trials Registry (last search April 2002) using the key words: abciximab, eptifibatide, tirofiban, and lamifiban, combined with key words suggested for identifying controlled trials. Bibliographies of articles and reviews were screened.
The primary outcome was death at 30 days and 6 months. We also examined available data on longer follow-up (one to three years). Secondary outcomes included MI and a composite of major adverse cardiac events (death, MI, or revascularization) at 30 days and 6 months. Safety outcomes included major bleeding by the Thrombolysis in Myocardial Infarction (TIMI) criteria, or, alternatively, severe/moderate bleeding by the Global Utilization of Streptokinase and TPA for Occluded arteries (GUSTO) criteria, and intracranial hemorrhage.
We systematically extracted information on study design and reported trial quality (16), participant characteristics, and outcomes, as mentioned. In trials with more than two arms, arms using different dosing schedules were merged to avoid duplicating the control arm. Two investigators (Drs. Karvouni and Katritsis) extracted data independently, discussed discrepancies, and reached a consensus with a third independent investigator (Dr. Ioannidis).
Intention-to-treat analyses use risk ratios (RRs) (relative risk reductions = 1 − RR). Odds ratios were similar (data not shown). Risk differences are also provided for death and major bleeding. Heterogeneity was assessed with the chi-squared–based Q statistic (significant for p < 0.10). We used both fixed and random effects models. In the absence of between-study heterogeneity, these models coincide; otherwise, random effects provide wider confidence intervals (CIs) (17).
Subgroup analyses were performed for mortality according to: 1) patient population (AMI vs. non-AMI); 2) intended PCI; 3) type of GP IIb/IIIa receptor antagonist; and 4) heparin use post-PCI. Heterogeneity between subgroups was assessed with the Q statistic.
Bias diagnostics evaluated whether results of small versus large studies (18)and early versus late trials (19)differed. No bias was detected (data not shown). Analyses were conducted using Meta-Analyst (Joseph Lau, Boston, Massachusetts) and SPSS (SPSS Inc., Chicago, Illinois). The p values are two-tailed.
Nineteen trials (1–6,11–15,20–36)with 20 pertinent comparisons (n = 20,137) were eligible (Table 1). One trial (14)had four arms, resulting in two pertinent comparisons (stent placement vs. PTCA). All but four trials (12,15,25,26)were multicenter. Most trials used abciximab. Four trials used eptifibatide and two trials used tirofiban. Stent placement was the initial procedure in 7 comparisons (12–14,20,23,26,29), and PTCA or atherectomy in 13. Five studies (six comparisons) (11–15)targeted AMI patients exclusively. Acute myocardial infarction patients comprised also about one-third of the Evaluation of IIb/IIIa platelet receptor antagonist 7E3 in Preventing Ischemic Complications (EPIC) and Randomized Efficacy Study of Tirofiban for Outcomes and REstenosis (RESTORE) patients (1,35).
One trial (15)used alternate randomization, whereas all others were truly randomized, although the details on the randomization were provided infrequently (1,14,28,29). Allocation concealment was explicitly described in eight trials (1,4,20,23,24,26,28,29), and withdrawals in nine trials (1,4,6,12,20,23,26,29,34). All but two trials (12,26)were double-masked.
Overall death rates in treatment versus control arms were 105/11,676 (0.90%) versus 116/8,461 (1.37%) at 30 days, 172/8,686 (1.98%) versus 176/6,965 (2.53%) at six months, and 252/78,686 (2.90%) versus 234/6,965 (3.36%) at longer follow-up (Fig. 1). Glycoprotein IIb/IIIa receptor antagonists conferred a significant 31% relative reduction in 30-day mortality (95% CI 10% to 47%, p = 0.006 for fixed effects; p = 0.008 for random effects; no between-study heterogeneity). A relative risk reduction of 21% was maintained at six months (95% CI 3% to 36%, p = 0.028 for fixed effects; 95% CI 0% to 37%, p = 0.048 for random effects) and when longer follow-up was considered (21%; 95% CI 6% to 34%, p = 0.008 for fixed effects; 95% CI 3% to 36%, p = 0.023 for random effects). The respective absolute decrease was 0.31% (95% CI 0.05% to 0.57%) at 30 days, 0.46% (95% CI 0.05% to 0.88%) at six months, and 0.59% (95% CI 0.10% to 1.07%) including long-term follow-up by fixed effects. Random effects were similar (data not shown). No single trial showed statistically significant mortality reductions, but most showed trends favoring the experimental arms (Fig. 1).
The benefit was similar in patients with and without AMI (Table 2). No clear effect was observed in two trials, including both AMI and non-AMI patients, but the CIs were large. Overall 30-day mortality rates in the control arms were 3.1% (95% CI 2.4% to 4.1%) in AMI trials, 1.1% (95% CI 0.7% to 1.8%) in trials with both AMI and non-AMI patients, and 0.9% (95% CI 0.6% to 1.2%) in non-AMI trials. The six-month mortality rates were 4.8% (CI 3.8% to 6.0%), 2.2% (CI 1.6% to 3.0%), and 1.6% (CI 1.2% to 2.1%), respectively. The survival benefit was similar in trials where stent placement was the primary procedure and in trials continuing versus discontinuing heparin. No mortality reduction was observed with tirofiban at either time point, but the CIs were large (Table 2).
Glycoprotein IIb/IIIa receptor antagonists significantly decreased (p < 0.001) MI and composite outcomes both at 30 days and 6 months (Table 3). There was a modest increase in the risk of major bleeding (random effects risk difference 0.68%, 95% CI −0.05% to 1.40%), with significant between-study heterogeneity. The findings were not different in subgroups defined by AMI, type of PCI, and type of agent (data not shown). The heterogeneity disappeared when we separated trials based on whether heparin had been continued or not after the procedure. Excess bleeding was observed only in trials continuing heparin (RR 1.70 [95% CI 1.36 to 2.14] vs. RR 1.02 [95% CI 0.85 to 1.24] in those stopping heparin). The risk of hemorrhagic stroke was not increased (Table 3).
Intravenous GP IIb/IIIa receptor antagonists confer a definite survival benefit among patients who undergo PCI. A 20% to 30% relative risk reduction is clinically important and complements our insight about the previously well-documented effects of these agents in reducing the risk of MI and composite cardiac end points (7). The survival benefit is clearly observed at 30 days and persists at long-term follow-up.
We observed no significant differences in the magnitude of the treatment effect in subgroup analyses according to clinical setting (AMI or non-AMI), type of intervention (stent vs. other), and timing of heparin administration. No survival benefit was documented with tirofiban. It is unknown whether this might be due to the relatively low tirofiban dose employed. Both abciximab and eptifibatide showed considerable survival benefits, the latter missing formal statistical significance, probably due to smaller numbers of randomized subjects.
In the overall meta-analysis, the number of patients needed to treat to save one life at 30 days, six months, and long-term follow-up was 320, 220, and 170, respectively. However, AMI patients are at higher risk than non-AMI patients for death and other adverse cardiac events. Therefore, the absolute benefit (risk difference) may actually be larger in high-risk AMI patients than in low-risk patients undergoing elective PCI, where the mortality risk is very low. The absolute benefit may be small and cost-effectiveness may be questionable in patients at very low risk.
Although hemorrhagic stroke was not a concern, we documented an excess of major bleeding with significant between-study heterogeneity. Heterogeneity may reflect whether heparin is continued or not after PCI. Heparin is well known to increase the bleeding risk. When heparin was discontinued after the procedure, there was no excess major bleeding, and the therapeutic effect remained undiminished. Subgroup analyses should be interpreted cautiously, but the data suggest that major bleeding does not countermatch the survival benefits and may not be a problem if heparin is promptly discontinued.
There has been no demonstrable overall mortality reduction in trials of acute coronary syndromes where only selected patients undergo PCI (7,10), with the exception of diabetics (10). Actually the largest mortality reduction in these trials occurred for diabetics who also underwent PCI eventually (10). The survival benefit of intravenous GP IIb/IIIa receptor antagonists may be linked to the PCI, per se, and the consequences of the procedure-related intimal damage and revascularization. In this meta-analysis, no separate data were available to address whether the beneficial effect on survival was different in diabetics versus nondiabetics. However, the proportion of diabetics was limited and unlikely to explain all of the observed mortality reduction. Separate subgroup analyses focusing on high-risk subjects (diabetics, subjects with high troponin levels) are warranted, but subgroup differences should be interpreted cautiously.
This meta-analysis provides adequate evidence for using intravenous GP IIb/IIIa receptor antagonists during PCIs to reduce short- and long-term mortality, but data should not be generalized to patients in whom no PCI is contemplated.
- acute myocardial infarction
- confidence interval
- myocardial infarction
- percutaneous coronary intervention
- percutaneous transluminal coronary angioplasty
- risk ratio
- Received August 2, 2002.
- Revision received September 19, 2002.
- Accepted September 26, 2002.
- American College of Cardiology Foundation
- The EPILOG Investigators
- Lincoff A.M.,
- Tcheng J.E.,
- Califf R.M.,
- et al.
- Kong D.F.,
- Califf R.M.,
- Miller D.P.,
- et al.
- Roffi M.,
- Chew D.P.,
- Mukherjee D.,
- et al.
- Brener S.J.,
- Barr L.A.,
- Burchenal J.E.,
- et al.
- Neumann F.J.,
- Kastrati A.,
- Schmitt C.,
- et al.
- ↵Jadad AR. Meta-analysis of randomised clinical trials in pain relief (thesis). Oxford University, 1993
- Petitti D.B.
- The ERASER Investigators
- Simoons M.L.,
- de Boer M.J.,
- van den Brand M.J.,
- et al.
- Tcheng J.E.,
- Harrington R.A.,
- Kottke-Marchant K.,
- et al.
- The RESTORE Investigators
- Gibson C.M.,
- Goel M.,
- Cohen D.J.,
- et al.
- Kereiakes D.J.,
- Kleiman N.S.,
- Ambrose J.,
- et al.