Early Percutaneous Coronary Intervention, Platelet Inhibition With Eptifibatide, and Clinical Outcomes in Patients With Acute Coronary Syndromes
Background—Platelet glycoprotein (GP) IIb/IIIa antagonists prevent the composite end point of death or myocardial infarction (MI) in patients with acute coronary syndromes. There is uncertainty about whether this effect is confined to patients who have percutaneous coronary interventions (PCIs) and whether PCIs further prevent death or MI in patients already treated with GP IIb/IIIa antagonists.
Methods and Results—PURSUIT patients were treated with the GP IIb/IIIa antagonist eptifibatide or placebo; PCIs were performed according to physician practices. In 2253 of 9641 patients (23.4%), PCI was performed by 30 days. Early (<72 hours) PCI was performed in 1228 (12.7%). In 34 placebo patients (5.5%) and 10 treated with eptifibatide (1.7%) (P=0.001), MI preceded early PCI. In patients censored for PCI across the 30-day period, there was a significant reduction in the primary composite end point in eptifibatide patients (P=0.035). Eptifibatide reduced 30-day events in patients who had early PCI (11.6% versus 16.7%, P=0.01) and in patients who did not (14.6% versus 15.6%, P=0.23). After adjustment for PCI propensity, there was no evidence that eptifibatide treatment effect differed between patients with or without early PCI (P for interaction=0.634). PCI was not associated with a reduction of the primary composite end point but was associated with a reduced (nonspecified) composite of death or Q-wave MI. This association disappeared after adjustment for propensity for early PCI.
Conclusions—Eptifibatide reduced the composite rates of death or MI in PCI patients and those managed conservatively.
Acute coronary syndromes (ACS) are characterized by obstructed coronary flow, platelet aggregation, and often by subsequent thrombosis at the site of disrupted atherosclerotic plaque. Management has consisted largely of attempts to modulate coagulation with aspirin and heparin,1 2 3 and more recently, by suppressing platelet aggregation with glycoprotein (GP) IIb/IIIa antagonists.4 5 6 Another approach includes early revascularization of a culprit coronary artery to reduce underlying stenosis and lower the likelihood and sequelae of subsequent occlusion. However, percutaneous coronary intervention (PCI) in this setting may increase the risk of vascular thrombosis and myocardial infarction (MI) compared with elective intervention.7 8 In 1 randomized trial, early aggressive PCI in patients with ACS did not reduce the incidence of death or MI in the first year9 ; another study suggested that early PCI in patients with recent myocardial necrosis may actually increase MI and mortality.10
Antagonists of GP IIb/IIIa given during and after PCI can reduce the likelihood of periprocedural MI 2 3 4 5 6 11 12 13 14 15 and long-term mortality.16 Thus, it is appealing to treat ACS patients with a GP IIb/IIIa antagonist as part of a strategy that includes aggressive early intervention (during GP IIb/IIIa inhibition). We reviewed a large trial of the GP IIb/IIIa antagonist eptifibatide (Integrilin) in ACS patients to determine the effects in those who had PCI and in those who were managed conservatively and whether early PCI as part of a strategy that included GP IIb/IIIa antagonism could reduce the risk of death or MI compared with delayed management.
In the PURSUIT4 trial, patients presented with ≤24 hours of ischemic chest discomfort and had either ST-segment or T-wave deviation or elevation of the creatine kinase MB fraction without sustained (>30 minutes) ST elevation. Patients were randomized to either placebo or intravenous eptifibatide as a 180-μg/kg bolus followed by a 1.3- or 2-μg · kg−1 · min−1 infusion for a maximum of 72 hours. After review of the first 2000 patients by the data and safety monitoring board, the lower infusion rate was stopped, in accord with the study protocol. In patients who had PCI during the first 72 hours (early PCI), study drug infusion was to continue through the intervention and a minimum of 24 hours afterward to 96 hours maximum. All other treatment decisions, including angiography and PCI, were left to the treating physician. When PCI was planned, investigators were encouraged (but not required) to perform it within 72 hours of randomization. The present analysis is restricted to patients treated with the higher dose versus placebo.
The primary end point was the composite of death or MI within 30 days. Death or MI among patients who had PCI within the first 72 hours was a secondary end point, because the timing would capture those who had intervention while taking the study drug. The hospital and study records of patients suspected of having new MIs were reviewed and classified by a blinded adjudication committee.4 Investigators were also asked to report whether the patient had had an MI.
Statistical comparisons of the rates of death or MI for interventional subgroups were made by logistic regression modeling.
To estimate the difference in 30-day outcomes for the treatment arms during the period when patients were managed medically only, we did a censored analysis using Cox regression survival modeling. For each patient, we calculated the time until death, MI, or last follow-up (within 30 days). For those who had PCI, follow-up was censored at intervention; thus, survival information only during the period of purely medical treatment was used.
We estimated the relation between early (≤72 hours) PCI and event-free survival with Cox regression modeling techniques using time-dependent covariates. Although these techniques do not prove causality, we have referred to association with a reduced event rate as “protective” and association with an increased event rate as “detrimental.” We also made models for the main effect of randomized treatment and the main effect and interaction terms of randomized treatment with PCI.
Adjustment for Selection Bias
We developed a propensity score for the likelihood of PCI17 18 19 by identifying baseline factors to predict PCI. The PCI probability was determined for each patient by use of a logistic model. The time-dependent models were applied again with the inclusion of the propensity score, which allowed us to estimate the association of PCI or eptifibatide with clinical outcomes and the interaction of the 2 after adjustment for PCI selection bias.
We also used a second approach to adjust for PCI selection bias. The proportion of patients who had early PCI (“PCI use”) was calculated for each center. We then calculated the rates and odds ratios for death or MI within each tertile of PCI use for all patients in the 2 treatment arms. Because eptifibatide treatment was randomly allocated within each center, we could estimate the treatment effect for sites likely to use early PCI versus those unlikely to.
Patients Who Had PCI
Of 9641 patients in the 2 primary arms of PURSUIT, 5937 (62%) had coronary angiography; angiographic data are available in 5546 (58%). Single-vessel disease was present in 1615 (29%), 2-vessel disease in 1445 (26%), 3-vessel or left main disease in 1764 (32%), and minimal or no coronary artery disease in 722 (13%). The median time from enrollment to index PCI was 67 hours (25th, 75th percentiles: 23, 138 hours) (Figure 1⇓). PCI was within 72 hours in 1228 patients (12.7%) and by 30 days in 2253 (24%). Compared with patients who did not have early PCI, patients in the early PCI group were younger, heavier, taller, and more likely to be smokers but less likely to be female or to have ST-segment depression, rales, or previous infarction (Table 1⇓).
In eptifibatide patients, 282 PCIs (26%) were urgent, compared with 328 (28%) in placebo patients (P=0.25). Intracoronary stents were placed in 51% of PCIs in eptifibatide patients and 50% of PCIs in placebo patients (P=0.81).
Across the 30 days when patients were censored for PCI, there was a significant reduction in the primary composite end point in eptifibatide patients (P=0.035, Table 2⇓). The reduction approached significance when patients who had PCI or bypass surgery (CABG) were censored (P=0.058).
Propensity to Perform PCI
The propensity score was highly predictive of the likelihood of PCI (χ2=837) (Figure 2⇓). Younger patients with less comorbidity and patients in North America were more likely to have PCI. The propensity score includes an interaction between sex and treatment. PCI use was not different for men and women assigned to eptifibatide. However, among placebo patients, men were more likely to undergo PCI (OR 1.32, 95% CI 1.09–1.59).
Patients Who Had Early PCI
Among the 1228 early PCI patients, 1105 (91%) were taking the study drug at the time of the intervention. Study drug was continued a median of 26 hours after initiation of PCI. Abciximab was used in 116 (9.5%) of the patients who had early PCI: 40 (34%) were assigned to eptifibatide and 76 (66%) to placebo. Of 622 placebo-treated patients who had early PCI, 156 (25%) had recurrent ischemia and 34 (5%) suffered an MI before the procedure, compared with 112 (18%) and 10 (1.7%), respectively, of 606 eptifibatide patients.
Kaplan-Meier survival estimates for patients who had early PCI are shown in Figure 3⇓ and Table 3⇓. Eptifibatide-treated patients who had early PCI experienced a significant (P<0.01) decrease in events during 30-day follow-up. Among patients randomized to placebo, the rates of death or MI were higher in patients who had early PCI than in those managed conservatively or by a delayed invasive strategy. In eptifibatide-treated patients, these rates were similar regardless of the strategy chosen, indicating that eptifibatide blunted the early hazard associated with PCI. Eptifibatide reduced the composite of death or MI before and after early PCI (Figure 4⇓). This reduction was present when the analysis was restricted to events after PCI in patients who received intracoronary stents (Table 4⇓).
Eptifibatide Effect in Early PCI Versus Delayed Invasive or Conservative Strategy
In PURSUIT, the PCI use analysis showed that the eptifibatide effect was greatest in the tertiles of centers with the highest likelihood of early PCI (Table 5⇓). Unadjusted event rates showed a greater effect for eptifibatide in early PCI patients than in those treated conservatively or with a delayed invasive strategy (51 versus 11 events prevented/1000 patients treated). However, after adjustment for differing clinical characteristics using the propensity score, this difference disappeared (20 versus 18 events prevented/1000 patients treated, P for interaction=0.634).
Early PCI and Clinical Outcome
Table 6⇓ shows time-dependent covariate analyses for the associations between PCI and the 30-day composite of death or MI. Early PCI had little relation to the likelihood of death or MI (hazard ratio 0.889, P=0.205) either before or after statistical adjustment. Similar results were seen for the end point of death or investigator-determined MI. However, early PCI was associated with a lower composite rate of death or Q-wave MI (hazard ratio=0.493, P<0.001; Figure 5⇓). After adjustment for treatment and propensity, the association with early PCI remained but was of borderline significance (hazard ratio=0.669, P=0.044). For each of the 3 end points, the eptifibatide benefit was still significant after adjustment for the propensity score and PCI use. The hazard ratio for death or MI was 0.75 (P=0.008), for death or Q-wave infarction 0.782 (P=0.031), and for death or investigator-determined infarction 0.77 (P<0.001). The interaction between PCI and randomized eptifibatide treatment was also sampled after each statistical adjustment and did not reach significance, indicating that the effect of eptifibatide was not limited to either group.
Management of patients with ACS is evolving as revascularization and antiplatelet strategies are applied more aggressively. Many observers view the strategies as competitive, given the ability of many hospitals to perform revascularization rapidly in patients presenting with ACS. This study indicates that the strategies are more likely complementary. Eptifibatide reduced the number of deaths and MIs before and after PCI in stented and nonstented patients and in the large number of patients who did not undergo a PCI.
Eptifibatide in Patients Managed Invasively
Although eptifibatide reduced the overall rates of death or MI in patients managed aggressively or conservatively, the reduction appeared to be greater in patients who had early PCI. When events preceding PCI were excluded, there was a 31% relative reduction in events at 30 days in patients who underwent PCI within 72 hours; the reduction was 6% in patients who had not had PCI during this period. A similar effect was seen when centers were grouped according to their propensity to perform PCI within the first 72 hours (use analysis). Conversely, baseline patient characteristics were quite different between the management groups, and correction of the event rates using the propensity score indicated that the adjusted benefit of eptifibatide did not differ between patients who did or did not have early PCI.
Unfortunately, no methods resolve these discrepancies precisely. Although treatment with eptifibatide or placebo was randomly assigned, PCI is inevitably determined by both preexisting patient characteristics (involving demographics, coronary anatomy, and physician biases) and postrandomization clinical events. We and others have observed that in patients with unstable syndromes, the characteristics differ in those who had or did not have PCI.20 21 22 Attempts to provide clinically meaningful approaches to this issue require adjustment for the characteristics that affect an individual patient’s propensity to undergo PCI. The propensity score has been validated as a useful instrument to test therapies not randomly assigned.23 Adjustment according to the propensity score indicated no difference in the effects of eptifibatide between patients who had PCI and those managed medically. Thus, selecting ACS patients for PCI may also select a group likely to benefit from a GP IIb/IIIa antagonist.
Effect of PCI on Outcome
The concept of early PCI in ACS patients is theoretically attractive for many reasons, including the hope that increasing the vascular luminal diameter may prevent subsequent ischemia and infarctions. However, before the era of GP IIb/IIIa antagonism and aggressive PCI with stent placement, 2 randomized trials9 10 and a recent registry study24 failed to confirm this. Other trials indicate that an intravenous GP IIb/IIIa antagonist is beneficial during PCI. Very recently, the FRISC-2 investigators showed a reduction in death or MI in patients assigned to early revascularization; how this balance would be affected by early treatment with a GP IIb/IIIa antagonist is not known.25
The association of PCI with an early increase in the rate of enzymatically detected MI11 24 may obscure conclusions about the ultimate benefit of PCIs. Using Kaplan-Meier estimates of event rates, we saw that the composite of death or MI at 30 days was more common among patients who had PCI. However, the analysis also suggested that early PCI may be associated with a reduction in a composite of death or Q-wave infarction. Although not drawn from a randomized cohort, the data raise the possibility that PCI is associated with an early hazard and with a reduction in later, more significant events. Interestingly, this hazard was present primarily in placebo patients and was blunted by eptifibatide.
Selection bias is the major study limitation. Although patients were randomized to placebo or eptifibatide, selection for PCI was an operator-dependent decision, often based on events after randomization. In addition, the decision for early PCI is likely to accompany a variety of other practice patterns, which may confound the outcomes of GP IIb/IIIa antagonist treatment. Thus, seeing these findings as hypothesis-generating is important. The use analysis partly circumvented these problems by comparing treatments in each tertile according to the likelihood of PCI at a given center, although other medical practices may also vary between centers in different tertiles. Factorial randomization between eptifibatide and placebo and between early invasive and conservative management would have provided a clearer answer to the interaction between the treatment modes. However, such a design would help assess the interaction between GP IIb/IIIa inhibition and a revascularization strategy rather than a particular procedure (such as PCI), because the latter depends primarily on anatomic substrates rather than randomized treatment assignment.
Limitations include the difficulty in diagnosing periprocedural MI in patients who have early PCI while cardiac enzymes from enrollment are still elevated. The 24-hour infusion required after PCI may result in a further advantage for eptifibatide treatment. Finally, this analysis does not account for CABG, although the censored survival analysis suggests a reduction in events preceding bypass similar to that seen preceding PCI.
Our findings offer important insights into the interaction between PCI and GP IIb/IIIa antagonism in ACS patients. In many centers, such patients undergo early angiography and anatomically directed PCI to reduce the degree of narrowing associated with a culprit lesion. Our findings suggest that the 2 approaches are not competitive and may, in fact, be complementary, because eptifibatide treatment prevented events both before and after PCI. Therefore, even in the patient being considered for early PCI, very early treatment with a GP IIb/IIIa antagonist most likely forms a useful part of a larger comprehensive strategy.
This study was supported in part by a grant from COR Therapeutics, Inc, South San Francisco, Calif, and Key Schering Plough, Inc, Kenilworth, NJ. The authors are grateful to John Daniel, A.J. Mayhew, and Susan Nichols for assistance in preparing the manuscript.
Reprint requests to Neal S. Kleiman, MD, 6565 Fannin, MS F-1090, Houston, TX 77030.
Guest Editor for this article was David P. Faxon, MD, USC School of Medicine, Los Angeles, Calif.
Dr Lorenz is an employee of Cor Therapeutics, Inc, the manufacturer of Integrilin, the form of eptifibatide used in this study.
- Received March 9, 1999.
- Revision received September 9, 1999.
- Accepted September 23, 1999.
- Copyright © 2000 by American Heart Association
Braunwald E, Jones RH, Mark DB, Brown J, Brown L, Cheitlin MD, Concannon CA, Cowan M, Edwards C, Fuster V, Goldman L, Green LA, Grines CL, Lytle BW, McCauley KM, Mushlin AI, Rose GC, Smith EEI, Swain JA, Topol EJ, Willerson JT, Agency for Health Care Policy and Research. Diagnosing and managing unstable angina. Circulation.. 1994;90:613–622.
de Feyter PJ, Suryapranata H, Serruys PW, Beatt K, van Domburg R, van den Brand M, Tijssen JJ, Azar AJ, Hugenholtz PG. Coronary angioplasty for unstable angina: immediate and late results in 200 consecutive patients with identification of risk factors for unfavorable early and late outcome. J Am Coll Cardiol. 1988;12:324–333.
Ellis SG, Roubin GS, King SB III, Douglas JS Jr, Weintraub WS, Thomas RG, Cox WR. Angiographic and clinical predictors of acute closure after native vessel coronary angioplasty. Circulation. 1988;77:372–379.
The TIMI IIIB Investigators. Effects of tissue plasminogen activator and a comparison of early invasive and conservative strategies in unstable angina and non–Q-wave myocardial infarction: results of the TIMI IIIB Trial. Circulation. 1994;89:1545–1556.
Boden WE, O’Rourke RA, Crawford MH, Blaustein AS, Deedwania PC, Zoble RG, Wexler LF, Kleiger RE, Pepine CJ, Ferry DR, Chow BK, Lavori PW, for the Veterans Affairs Non-Q-Wave Infarction Strategies in Hospital (VANQWISH) Trial Investigators. Outcomes in patients with acute non-Q-wave myocardial infarction randomly assigned to an invasive as compared with a conservative management strategy. N Engl J Med. 1998;338:1785–1792.
The RESTORE Investigators. Effects of platelet glycoprotein IIb/IIIa blockade with tirofiban on adverse cardiac events in patients with unstable angina or acute myocardial infarction undergoing coronary angioplasty. Circulation. 1997;96:1445–1453.
Lincoff AM, Califf RM, Anderson KM, Weisman HF, Aguirre FV, Kleiman NS, Harrington RA, Topol EJ, for the EPIC Investigators. Evidence for prevention of death and myocardial infarction with platelet membrane glycoprotein IIb/IIIa receptor blockade by abciximab (c7E3 Fab) among patients with unstable angina undergoing percutaneous coronary revascularization. J Am Coll Cardiol. 1997;30:149–156.
Topol EJ, Ferguson JJ, Weisman HF, Tcheng JE, Ellis SG, Kleiman NS, Ivanhoe RJ, Wang AL, Miller DP, Anderson KM, Califf RM, for the EPIC Investigator Group. Long-term protection from myocardial ischemic events in a randomized trial of brief integrin beta3 blockade with percutaneous coronary intervention. JAMA. 1997;278:479–484.
Rosenbaum PR. Discussing hidden bias in observational studies. Ann Intern Med. 1991;115:901–905.
Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika. 1983;70:41–55.
Berger PB, Holmes DR Jr, Stebbins AL, Bates ER, Califf RM, Topol EJ. Impact of an aggressive invasive catheterization and revascularization strategy on mortality in patients with cardiogenic shock in the GUSTO-I trial. Circulation. 1997;96:122–127.
Stone PH, Thompson B, Anderson HV, Kronenberg MW, Gibson RS, Rogers WJ, Diver DJ, Theroux P, Warnica JW, Nasmith JB, Kells C, Kleiman N, McCabe CH, Schactman M, Knatterud GL, Braunwald E. Influence of race, sex, and age on management of unstable angina and non-Q-wave myocardial infarction: the TIMI III registry. JAMA. 1996;275:1104–1112.
Connors AF Jr, Speroff T, Dawson NV, Thomas C, Harrell FE Jr, Wagner D, Desbiens N, Goldman L, Wu AW, Califf RM, Fulkerson WJ Jr, Vidaillet H, Broste S, Bellamy P, Lynn J, Knaus WA, for the SUPPORT Investigators. The effectiveness of right heart catheterization in the initial care of critically ill patients. JAMA. 1996;276:889–897.
Abdelmeguid AE, Topol EJ, Whitlow PL, Sapp SK, Ellis SG. Significance of mild transient release of creatine kinase-MB fraction after percutaneous coronary interventions. Circulation. 1996;94:1528–1536.
Yusuf S, Flather M, Pogue J, Hunt D, Varigos J, Piegas L, Avezum A, Anderson J, Keltai M, Budaj A, Fox K, Ceremuzynski L, OASIS (Organisation to Assess Strategies for Ischaemic Syndromes) Registry Investigators. Variations between countries in invasive cardiac procedures and outcomes in patients with suspected unstable angina or myocardial infarction without initial ST elevation. Lancet.. 1998;352:507–514.