Randomized Trial of Aspirin, Sibrafiban, or Both for Secondary Prevention After Acute Coronary Syndromes
Background—The first Sibrafiban Versus Aspirin to Yield Maximum Protection From Ischemic Heart Events Post–Acute Coronary Syndromes (SYMPHONY) trial showed no benefit of 2 doses of sibrafiban over aspirin for secondary prevention after acute coronary syndromes. In 2nd SYMPHONY, we compared low-dose sibrafiban plus aspirin (LDS+A), high-dose sibrafiban (HDS), and aspirin alone.
Methods and Results—When the first SYMPHONY results became known, enrollment in 2nd SYMPHONY was stopped prematurely at 6671 patients who had been treated for a median of 90 days. The primary end point of death, myocardial (re)infarction (MI), or severe recurrent ischemia did not differ significantly between aspirin (9.3%) and LDS+A (9.2%; OR, 0.98; 95% CI, 0.80 to 1.20) or HDS (10.5%; OR, 1.14; 95% CI, 0.9 to 1.39) patients. Secondary end points did not differ significantly between aspirin and LDS+A patients. Death or MI occurred significantly more often with HDS (OR, 1.43; 95% CI, 1.14 to 1.80), as did mortality alone (OR, 1.83; 95% CI, 1.17 to 2.88) and MI (OR, 1.32; 95% CI, 1.03 to 1.69). Major bleeding was significantly more frequent in LDS+A patients (5.7%) versus aspirin alone (4.0%) but not in HDS patients (4.6%).
Conclusions—Combining aspirin with LDS did not improve outcomes after acute coronary syndromes and caused more bleeding compared with aspirin alone. There was a trend toward increased mortality in this group and a significant increase in the high-dose arm.
Aspirin given for ≥30 days reduces secondary vascular events by 25% to 30% after acute coronary syndromes (ACS),1 but the rate of death or myocardial (re)infarction (MI) remains high. Intravenous glycoprotein (GP) IIb/IIIa antagonists have been effective in reducing death or MI in patients hospitalized for ACS.2 The first Sibrafiban Versus Aspirin to Yield Maximum Protection From Ischemic Heart Events Post–Acute Coronary Syndromes (SYMPHONY) study tested whether sibrafiban, a potent oral GP IIb/IIIa inhibitor, could extend the benefit of intravenous agents into secondary prevention.3 The study showed no significant differences between treatment with aspirin compared with either of 2 sibrafiban regimens on the 90-day composite of death, MI, or severe recurrent ischemia,4 but its design did not include aspirin background therapy with sibrafiban. Although aspirin is a relatively weak antiplatelet agent, its other properties, including its antiinflammatory effect and irreversible inhibition of the thromboxane A2 pathway to platelet activation, may be important for secondary prevention after ACS.
The 2nd SYMPHONY study was designed to assess whether longer treatment (12 to 18 months) with low-dose sibrafiban given with background aspirin therapy (LDS+A) or high-dose sibrafiban alone (HDS) would be more effective for secondary prevention than aspirin alone. When the negative results of the first SYMPHONY became known, enrollment in 2nd SYMPHONY was terminated. Data collection and follow-up through the end of the study period were completed for enrolled patients, however. This article gives the primary efficacy and safety results.
Patients within 7 days of an ACS who were stable for ≥12 hours without signs of active ischemia, hemodynamic instability, or Killip class >II were eligible. Inclusion and exclusion criteria were those of the first SYMPHONY.3 4 The ethics committee or institutional review board of each site approved the protocol, and patients gave written, informed consent to participate before randomization.
Randomization and Study Treatment
The randomization scheme was generated by use of the same specifications as for the first SYMPHONY.4 Drug kits appeared identical except for kit numbers. Sibrafiban and aspirin tablets and their placebo tablets were color coded by type of kit: HDS, LDS+A, or aspirin. Each kit contained 2 types of enteric-coated tablets of identical size and shape but different colors (red or white); 1 of each was taken twice daily. Only a statistician, the programmers of the interactive voice-response system, the pharmacy staff, and the sponsor’s drug-safety group could access the randomization and drug kit identity codes. For urgent clinical need for unblinding, Helpline personnel at randomization centers could access treatment codes.
Sibrafiban doses (3.0, 4.5, or 6.0 mg) were based on weight and serum creatinine with the use of a model developed in phase II work.3 5 To facilitate blinding by paralleling the need for twice-daily sibrafiban dosing, aspirin (80 mg) was given twice daily in the aspirin and LDS+A arms. Except on the day of or after percutaneous coronary intervention (PCI), open-label aspirin use was prohibited, as were other oral antiplatelet agents. Patients undergoing coronary stenting were assigned in a double-blind fashion to receive ticlopidine 250 mg (aspirin group) or ticlopidine placebo (either sibrafiban group) twice daily for 2 to 4 weeks. The Data and Safety Monitoring Committee followed as a safety end point the rate of stent thrombosis in the experimental arms. All other management was at the discretion of treating physicians.
Planned enrollment was 8400 patients who were to be treated for ≥12 months after the last patient was enrolled or a maximum of 18 months. The sponsor stopped enrollment on August 6, 1999, after 6671 patients had been randomized at 716 sites in 35 countries. On August 24, 1999, sites were notified in writing of study termination and were to contact patients to arrange a final visit within 1 month. Sites also were instructed to notify their review boards or ethics committees of the first SYMPHONY results, early termination of 2nd SYMPHONY, and plans to convert from blinded study drugs to open-label therapy. Patients continued study treatment until the final visit, when they were switched to open-label aspirin (or other antiplatelet therapy if the patient’s primary physician recommended); the first dose was given in clinic. Patients receiving study stent medication were to receive open-label therapy afterward, as deemed appropriate by the primary cardiologist.
Clinical follow-up and data collection were completed through the study termination date for those who had finished study drug therapy or through study drug discontinuation at the final visit for those still taking study drug at study termination. For patients terminating treatment solely because the study was stopped, data for death, MI, and severe recurrent ischemia were obtained 1 month after treatment was stopped.
The primary end point was the time to all-cause mortality, MI, or severe recurrent ischemia, whichever occurred first. A blinded committee adjudicated severe recurrent ischemia (>20 minutes of chest pain leading to urgent revascularization) and MI (creatine kinase-MB above the upper limit of normal after an ischemic event, >3 times the upper limit of normal after PCI, >5 times the upper limit of normal after bypass surgery, or new, significant Q waves on ECG). The primary safety end points were major or minor bleeding3 and their composite occurring <7 days after study drug termination; tolerability was assessed as the rate of study drug discontinuation for bleeding.
The study was designed to test each sibrafiban regimen versus aspirin separately. The sample size was an estimated 2520 patients per arm to detect a 20% relative risk reduction with sibrafiban compared with aspirin with 80% power at α=0.025. Calculations assumed constant enrollment over 10 months; 12-month treatment duration after the last patient was enrolled (maximum, 18 months); expected rates of the primary end point in the aspirin arm of 12%, 15%, and 18% after 12, 17, and 22 months, respectively; and that for each sibrafiban arm, the hazard over time for the composite end point would be proportional to that of the aspirin arm. Target enrollment was 2800 patients per arm to allow for dropouts and slow accrual. Early study termination, resulting in shorter and variable follow-up and reduced sample size, substantially reduced the final power.
The primary analysis population was all patients randomized who received ≥1 dose of study drug. The primary analysis was by randomized treatment assignment. Kaplan-Meier curves were generated for the primary and secondary end points and compared by use of the log-rank test. All tests were 2-tailed. No corrections were made for multiple tests on secondary end points. ORs with 95% CIs were used to describe comparisons of the incidence of events by treatment.
Of 6671 patients randomized, 17 patients in the aspirin arm, 8 in the LDS+A arm, and 9 in the HDS arm received no study drug. The remaining 6637 patients composed the primary analysis population. Baseline characteristics did not differ by treatment (Table 1⇓).
For 74% of patients, the qualifying event was acute MI (63% with ST-segment elevation). The median time from qualifying event to randomization was 89 hours (interquartile range [IQR], 61 to 131 hours); from qualifying event to treatment start, the median time was 94 hours (IQR, 63 to 132 hours). The median durations of therapy and follow-up were 90 and 95 days (IQR, 36 to 139 and 65 to 158 days), respectively, and did not differ by treatment (Table 2⇓).
The primary end point did not differ significantly by treatment (log-rank P=0.89 for LDS+A and P=0.28 for HDS, each versus aspirin alone) (Figure 1⇓ and Table 3⇓). Compared with aspirin, there was a nonsignificant 31% increase in mortality (Figure 2⇓) and a 12% increase in the composite of death or MI with LDS+A. Patients given HDS had significantly higher rates of death, MI, and their composite compared with aspirin-treated patients. Furthermore, the rates of rehospitalization were significantly higher in both sibrafiban arms compared with the aspirin arm. Table 4⇓ presents the results of a blinded assessment of cause of the 121 study deaths.
Because of large variations in treatment and follow-up duration, we evaluated the rates of primary and secondary efficacy end points among patients in the lowest and highest quartiles of these variables. Within the quartile with the longest expected follow-up, mortality was slightly lower in the LDS+A arm (2.0%) compared with the aspirin arm (2.5%), but results for all other end points mirrored the overall patterns. Regardless of follow-up duration, the HDS arm consistently showed an excess of death, MI, and their composite.
Overall, event rates in patients remaining on treatment (“as-treated/on-treatment”) were lower in all arms than rates in the intention-to-treat analysis. The rates of the primary end point in this analysis tended to be nonsignificantly lower for the LDS+A (6.9%) and HDS (7.5%) groups than for the aspirin group (7.7%). For the comparison of HDS with aspirin, the differences in the rates of death (0.6% versus 0.3%) and MI (5.4% versus 4.6%) were smaller and no longer significant.
Treatment effects did not differ among major clinical subgroups. Although treatment effects differed somewhat by regional stratum, tests for heterogeneity of effect by region were nonsignificant.
Overall, 29.5% of patients underwent revascularization, either bypass surgery or PCI, between the qualifying event and randomization. Another 24.8% underwent revascularization after beginning treatment (7.7% had bypass surgery and 16.9% had PCI, not mutually exclusive). Stents were used in >87% of all PCIs (HDS, n=766; LDS+A, n=812; aspirin control, n=803). Stent thrombosis was significantly more common among patients receiving HDS compared with aspirin (6.4% versus 4.0%, P=0.031) but was similar between the LDS+A and aspirin groups (3.9% versus 4.0%, P=0.972). Despite the high rate of stent thrombosis in the HDS arm, removing stented patients from the primary efficacy analysis did not measurably alter the pattern of results. Among nonstented patients, death, MI, or severe recurrent ischemia occurred in 6.8% of aspirin control patients compared with 8.0% and 8.4% of LDS+A and HDS patients, respectively.
Of the 5235 patients who stopped study drug use because of trial termination, 1-month follow-up information was collected from 4591 (87.7% of expected). The median length of follow-up after study drug termination was similar by treatment. Rates of death, MI, severe recurrent ischemia, and their composites did not differ significantly among the groups, nor did their timing (Table 5⇓).
Except for major bleeding in the HDS arm, major bleeding, minor bleeding, and their composite were significantly more frequent among sibrafiban-treated patients (Table 6⇓). Most bleeding was minor and mucocutaneous or puncture site–related, but LDS+A patients had a significantly higher transfusion rate that paralleled their significantly higher rate of major bleeding. Study drug discontinuation for bleeding was significantly higher in both sibrafiban arms compared with the aspirin arm. The rates of thrombocytopenia and severe thrombocytopenia (platelet count <20 000/mL) were low and did not differ by treatment. The frequency of stroke was low with no significant difference in stroke subtypes by treatment.
In 2nd SYMPHONY, addition of aspirin to LDS+A did not reduce the composite incidence of death, MI, or severe recurrent ischemia relative to aspirin alone. Indeed, this group showed a trend toward increased mortality and the composite of death or MI, and the HDS arm had statistically significant increases in mortality, MI, and their composite compared with aspirin alone. In both sibrafiban arms, especially the LDS+A group, bleeding was significantly more frequent than with aspirin, as was all-cause rehospitalization.
In the first SYMPHONY, sibrafiban-treated patients received no background aspirin therapy. Through its antiinflammatory effects and by providing some ongoing antiplatelet effect over longer periods (especially when sibrafiban therapy was interrupted or stopped), aspirin might have provided additional benefit to those assigned sibrafiban. Outcomes did not differ, however, with the addition of aspirin to LDS compared with aspirin alone over 90 days of treatment (median), suggesting no additive benefit in short-term therapy. Because the early termination of 2nd SYMPHONY resulted in highly varied treatment durations, whether prolonged combination treatment might have produced benefit is unknown. Analysis of patients in the upper versus lower quartile of expected treatment duration (≥5 versus ∼2 months) showed no evidence that longer treatment would have significantly affected the pattern of results. Although patients remaining on treatment had slightly lower event rates, which were favorable for the sibrafiban arms compared with aspirin control, the implication of these findings is unclear and warrants further investigation.
Although overall mortality has been low, phase III trials of oral GP IIb/IIIa antagonists have reported higher mortality among patients receiving these agents compared with aspirin.4 6 7 In 2 trials, Orbofiban in Patients With Unstable Coronary Syndromes (OPUS)6 and 2nd SYMPHONY, the difference was statistically significant. This excess mortality has raised concern about a prothrombotic effect of these drugs. The excess thrombotic-event deaths in orbofiban-treated patients in OPUS support this hypothesis. In a blinded cause-of-death analysis in 2nd SYMPHONY, however, rates of ischemic cardiac and other causes of deaths did not differ between the LDS+A and aspirin control arms, consistent with their comparable rates of nonfatal MI. The HDS group had slightly fewer deaths caused by ischemic cardiac events but had more sudden cardiac deaths.
The dichotomy between fewer ischemic cardiac events and more sudden cardiac deaths in the HDS group might be reconciled by autopsy studies, which show evidence of coronary thrombus in ≥74% of sudden cardiac deaths.8 9 10 Thus, the excess sudden deaths in the HDS arm of 2nd SYMPHONY may represent evidence of thrombosis. If sudden death and ischemic cardiac events are combined, there are 8 more possible thrombotic events causing death in the HDS arm compared with the aspirin arm (24 versus 16; OR, 1.5; 95% CI, 0.8 to 2.9). With the significantly higher rate of nonfatal MI, this could represent a prothrombotic tendency of this dose. ECG data were unavailable to investigate other explanations for excess sudden deaths, such as QT interval prolongation.
Recent investigation with several GP IIb/IIIa antagonists has produced many potential explanations for their apparent toxicity. Although not all have been specifically shown for sibrafiban, some agents may paradoxically activate platelets, enhancing fibrinogen binding, activating other platelet-surface complexes, or potentiating α-granule or lysosomal release.11 12 13 14 15 16 17 These drugs also may affect inflammation, increasingly recognized as important in atherosclerosis and ACS.18 Patients with ACS given certain GP IIb/IIIa antagonists have shown significantly increased neutrophil-platelet coaggregates and more sustained elevations in leukocyte adhesion molecules.11 19
Important pharmacokinetic and pharmacodynamic properties of the agents also may contribute to their failure.11 16 17 In contrast to intravenous agents, used at doses that generate sustained, high levels of receptor occupancy and inhibition of platelet aggregation, the oral agents used long term are not tolerated at such doses. Furthermore, because of their short terminal half-lives, plasma levels of available oral drugs vary considerably during a dosing interval. Particularly with drugs that reversibly bind the receptor with low binding affinities, rapid “off rates,” or both, these fluctuations in plasma concentration may leave the activated receptor unoccupied or unprotected and susceptible to fibrinogen binding, possibly potentiating a prothrombotic state. Both aspirin and clopidogrel, effective in secondary prevention,1 20 irreversibly inhibit pathways to platelet activation (thromboxane A2 and ADP, respectively), potentially avoiding these pitfalls despite less potent platelet inhibition. Concomitant use of these drugs in ACS patients and the administration of aspirin to all 2nd SYMPHONY patients at the final visit could explain the absence of events clustered as intravenous infusions end or on cessation of sibrafiban at the termination of our study.
Newer agents with higher receptor affinity, slower receptor off rates, and longer terminal half-lives may represent an improvement over the first generation of drugs. Preliminary data from a phase II study of roxifiban (one such agent) in coronary artery disease presented at the 49th Scientific Sessions of the American College of Cardiology are promising but predominantly for the end point of ischemia.21 Severe recurrent ischemia also was reduced in both SYMPHONY studies, as was the urgent revascularization end point in OPUS.4 6 In all 3 studies, however, the end points of death and MI favored the aspirin control arms. How recurrent ischemia has been favorably affected and how death or MI is adversely affected are unknown.
Another important consideration for the failure of the oral agents tested is an incomplete understanding of possible differences in the pathophysiology of ACS, for which intravenous agents are effective, compared with subacute or chronic coronary disease, which may render patients less susceptible to the benefits of GP IIb/IIIa antagonism. Increased bleeding in the SYMPHONY trials also may be important to their failure, particularly because it led to treatment interruptions. Further evaluation of the potential association between bleeding and efficacy is ongoing. At the bleeding rates observed in 2nd SYMPHONY, however, the practical use of agents such as sibrafiban could be limited even if efficacy results were favorable.
Careful study of data from completed studies and of potential genetic associations with response to these agents may help target populations for future investigation. At present, however, the oral GP IIb/IIIa antagonists, alone or with aspirin, have not favorably influenced outcomes after ACS and, although overall mortality has been low in all studies, generally have increased mortality. Newer agents with different biological properties are appealing in concept. Careful investigation of the basic mechanisms of these agents on platelet receptor physiology and thrombosis, the complex pathophysiology of acute and chronic coronary disease, and the role of inflammation is imperative to their development.
A complete list of investigators and sites can be found in the online Appendix (www.circulationaha.org). As primary investigator, Dr L. Kristin Newby assumes full responsibility for the overall content and integrity of the manuscript.
- Received September 13, 2000.
- Revision received December 13, 2000.
- Accepted December 19, 2000.
- Copyright © 2001 by American Heart Association
Antiplatelet Trialists Collaboration. Collaborative overview of randomised trials of antiplatelet therapy, I: prevention of death, myocardial infarction, and stroke by prolonged antiplatelet therapy in various categories of patients. BMJ. 1994;308:81–106.
Kong DF, Califf RM, Miller DP, et al. Clinical outcomes of therapeutic agents that block the platelet glycoprotein IIb/IIIa integrin in ischemic heart disease. Circulation. 1998;98:2829–2835.
Cannon CP, McCabe C, Borzak S, et al. Randomized trial of an oral platelet glycoprotein IIb/IIIa antagonist, sibrafiban, in patients after an acute coronary syndrome: results of the TIMI-12 trial. Circulation. 1998;97:340–349.
Cannon CP, McCabe CH, Wilcox RG, et al. Oral glycoprotein IIb/IIIa inhibition with Orbofiban in Patients With Unstable Coronary Syndromes (OPUS-TIMI 16) trial. Circulation. 2000;102:149–156.
Rentrop KP. Thrombi in acute coronary syndromes: revisited and revised. Circulation. 2000;101:1619–1626.
Falk E. Unstable angina with fatal outcome: dynamic coronary thrombosis leading to infarction and/or sudden death: autopsy evidence of recurrent mural thrombosis with peripheral embolization culminating in total vascular occlusion. Circulation. 1985;71:699–708.
Peter K, Schwarz M, Ylanne J, et al. Induction of fibrinogen binding and platelet aggregation as a potential intrinsic property of various glycoprotein IIb/IIIa (αIIbβ3) inhibitors. Blood. 1998;92:3240–3249.
Schneider DJ, Taatjes DJ, Sobel BE. Paradoxical inhibition of fibrinogen binding and potentiation of α-granule release by specific types of inhibitors of glycoprotein IIb-IIIa. Cardiovasc Res. 2000;45:437–446.
Scarborough RM, Kleiman NS, Phillips DR. Platelet glycoprotein IIb/IIIa antagonists: what are the relevant issues concerning their pharmacology and clinical use? Circulation. 1999;100:437–444.
Mousa SA, Khurana S, Forsythe MS. Comparative in vitro efficacy of different platelet glycoprotein IIb/IIIa antagonists on platelet-mediated clot strength induced by tissue factor with use of thromboelastography. Arterioscler Thromb Vasc Biol. 2000;20:1162–1167.
Langer A, Goodman SG, Reilly TM, et al. Second generation oral IIb/IIIa receptor blockade: phase II experience with roxifiban. J Am Coll Cardiol. 2000;351:308A. Abstract.