Randomized COMparison of Platelet Inhibition With Abciximab, TiRofiban and Eptifibatide During Percutaneous Coronary Intervention in Acute Coronary Syndromes
The COMPARE Trial
Background— The relative anti-aggregatory effects of currently prescribed platelet glycoprotein IIb/IIIa receptor antagonists during and after percutaneous coronary intervention for acute coronary syndromes have not been established.
Methods and Results— We randomized 70 acute coronary syndrome patients undergoing percutaneous coronary intervention to receive abciximab, eptifibatide, or tirofiban at doses used in the Evaluation of Platelet IIb/IIIa Inhibitor for STENTing (EPISTENT), Platelet glycoprotein IIb/IIIa in Unstable angina Receptor Suppression Using Integrilin Therapy (PURSUIT), and Platelet Receptor Inhibition in ischemic Syndrome Management in Patients Limited by Unstable Signs and symptoms (PRISM-PLUS)/Randomized Efficacy Study of Tirofiban for Outcomes and Restenosis (RESTORE) trials, respectively. Platelet aggregation (PA) in response to 20 μmol/L of adenosine diphosphate was measured with turbidimetric aggregometry in both D-phenylalanyl-l-prolyl-l-arginine chloromethylketone and citrate-anticoagulated blood early (15 and 30 minutes) and late (4, 12, and 18 to 24 hours) after drug initiation. At 15 and 30 minutes, PA was significantly less inhibited by the tirofiban-RESTORE regimen compared with abciximab (P=0.028) and eptifibatide regimens (P=0.0001). The abciximab regimen, however, showed increasingly varied anti-aggregatory effects during continued infusion for ≥4 hours. Citrate exaggerated ex vivo platelet inhibition after eptifibatide and tirofiban, but had the opposite effect on abciximab. Of all regimens evaluated, the eptifibatide regimen inhibited PA most consistently throughout both the early and late periods.
Conclusions— Currently recommended drug regimens to inhibit the platelet glycoprotein IIb/IIIa receptor have distinct pharmacodynamic profiles that might affect their relative efficacy in acute coronary syndromes and percutaneous coronary intervention.
Received January 29, 2002; revision received June 26, 2002; accepted June 26, 2002.
Pharmacological antagonism of the platelet glycoprotein (GP) IIb/IIIa receptor reduces ischemic events in patients with acute coronary syndromes (ACS) undergoing percutaneous coronary intervention (PCI).1–5⇓⇓⇓⇓ Currently, the chimeric human mouse monoclonal antibody fragment abciximab (ReoPro, Centocor, Inc), the cyclic heptapeptide eptifibatide (Integrilin, COR Therapeutics, Inc), and the nonpeptide tyrosine derivative tirofiban (Aggrastat, Merck & Co) are GP IIb/IIIa antagonists approved for such use. Because differences between these drugs exist in chemical structure, pharmacokinetics,6 and the parameters underlying dose-finding studies, current clinical regimens may not have identical pharmacodynamic profiles. For example, tirofiban has been dosed to achieve >90% inhibition of platelet aggregation (PA) in response to 5 μmol/L adenosine diphosphate (ADP) in citrate anticoagulated blood.7 In contrast, eptifibatide and abciximab have been dosed to inhibit PA in response to stronger agonists, such as 20 μmol/L ADP or thrombin receptor agonist peptide.8–10⇓⇓ Furthermore, whether the calcium-chelating effect of citrate alters abciximab or tirofiban pharmacodynamics, as has been observed with eptifibatide,11 remains unclear.
In the Comparison Of Measurements of Platelet aggregation with Aggrastat, Reopro, and Eptifibatide (COMPARE) trial, we directly compared the pharmacodynamics of abciximab, eptifibatide, and 2 tirofiban regimens. Specifically, we measured PA during and after PCI in ACS patients given the drug regimens used in the Evaluation of Platelet IIb/IIIa Inhibitor for STENTing (EPISTENT) trial,12 the Platelet glycoprotein IIb/IIIa in Unstable angina Receptor Suppression Using Integrilin Therapy (PURSUIT) trial,5 the Platelet Receptor Inhibition in ischemic Syndrome Management in Patients Limited by Unstable Signs and symptoms (PRISM-PLUS) trial,4 and the Randomized Efficacy Study of Tirofiban for Outcomes and Restenosis (RESTORE) trial.13
Between June 1999 and August 2000, 73 patients were enrolled in a single-center, open-label randomized study performed at Duke University Medical Center. The Institutional Review Board approved the protocol, and patients gave informed consent before enrollment. Patients with ACS (unstable angina or non-ST-elevation myocardial infarction), for whom PCI was planned were eligible. Exclusion criteria included prior treatment with any platelet GP IIb/IIIa antagonist within 30 days and any relative or absolute contraindication to platelet GP IIb/IIIa inhibition.
Initially, 48 patients were randomized to abciximab, eptifibatide, or tirofiban in doses from the EPISTENT,12 PURSUIT,5 and PRISM-PLUS4 trials, respectively. Abciximab 0.25 mg/kg was given 10 minutes before PCI, followed by 0.125 μg · kg−1 · min−1 (10 μg/min maximum) for 12 hours; eptifibatide 180 μg/kg was given 10 minutes before PCI, followed by 2 μg · kg−1 · min−1 for 18 hours; and tirofiban 0.4 μg · kg−1 · min−1 was given for 30 minutes, starting 10 minutes before PCI and followed by 0.10 μg · kg−1 · min−1 for 18 hours.
After enrollment of these 48 patients, the Duke Institutional Review Board approved an extension of the protocol to assess the tirofiban dose used in the RESTORE trial,13 which was tirofiban 10 μg/kg given 10 minutes before PCI, followed by 0.15 μg · kg−1 · min−1 for 18 hours. For this phase, another 25 patients were randomized in a 3:1:1 manner to receive either this tirofiban regimen, eptifibatide, or abciximab. Because abciximab and eptifibatide doses remained constant throughout, their pharmacodynamic data from the 2 phases were combined.
PCI and Concomitant Medications
Coronary stent implantation was performed according to institutional standards. Heparin was given intravenously to achieve an activated clotting time (ACT) ≥200 seconds for abciximab and of 250 to 300 seconds for eptifibatide and tirofiban. The study drug was started at least 10 minutes before activation of the first device (ie, first balloon inflation). Postprocedural heparin was discouraged. All patients received aspirin (325 mg) before and after PCI. Clopidogrel was given ≥4 hours before PCI (300 mg) and daily thereafter for 30 days (75 mg).
Assessment of Platelet Function
Blood samples were collected separately in citrate and D-phenylalanyl-l-prolyl-l-arginine chloromethylketone (PPACK)-anticoagulated plastic tubes at 6 prespecified times; immediately before study-drug bolus (baseline), at 15 and 30 minutes (during PCI), and 4, 12, and 18 to 24 hours thereafter. In abciximab-treated patients, 12-hour samples were taken immediately before the infusion was stopped.
Platelet-rich plasma (PRP) and platelet-poor plasma (PPP) were prepared by centrifugation at 135 g for 15 minutes at room temperature. PRP samples with platelet counts >250 000/μL were adjusted to ≈250 000/μL by standard dilution with autologous PPP. After calibration with PRP (0% aggregation) and PPP samples (100% aggregation), we measured ex vivo PA in response to 20 μmol/L ADP in PRP using optical light-transmission aggregometry (PaP 4, BioData Corporation).14
Baseline characteristics were summarized using descriptive statistics. To normalize for baseline (predrug) heterogeneity, PA was expressed as a percentage of baseline. Two-tailed Wilcoxon rank-sum tests were used to compare median PA between drug regimens, and to compare each drug’s profile with citrate and PPACK anticoagulants. Because citrate enhances eptifibatide’s antiplatelet effects,11 aggregation in PPACK anticoagulated samples was primarily used for inter-drug comparisons. Assuming a standard deviation of 10%, we estimated that 14 patients per group would provide ≥90% statistical power (1-β ≥0.80, α=0.05) to detect an 8% absolute difference in PA.
Using the MIXED procedure in SAS, we also developed a random-effects, linear-regression model15 that characterized PA during each major time interval (0 to 15 minutes, 15 minutes to 4 hours, and 4 hours to infusion end). Separate models were developed to summarize profiles in citrate and PPACK. These models were used to predict and compare the change in PA (slope) that occurred from the beginning to end of each time interval. The major advantages of this statistical approach over Wilcoxon tests are that it accounts for the inherent variation in the timing of blood sampling, allows PA profiles to be represented as smooth functions over time, and accounts for the variability in baseline PA and for any intra-subject correlation over time.16,17⇓ Finally, to compare the pharmacodynamic variability within each drug regimen, we used F tests to compare standard deviations of PA at each time-point.
Of the 73 patients enrolled, 22 were randomized to abciximab, 20 to eptifibatide, 16 to tirofiban-PRISM-PLUS, and 15 to tirofiban-RESTORE. One patient from each of the abciximab, eptifibatide, and tirofiban-PRISM-PLUS groups dropped out because of hemodynamic instability, bleeding, and patient refusal of blood sampling, respectively. The clinical and laboratory characteristics of the remaining 70 patients were similar between groups (Table 1). By protocol, the duration of abciximab therapy was shorter (12 hours) than that of other regimens (16 to 19 hours). The mean interval between drug discontinuation and final blood sampling at 18 to 24 hours was 9.8±1.8 hours for abciximab, 3.6±3.9 hours for eptifibatide, 6.1±6.8 hours for tirofiban-PRISM-PLUS, and 2.9±1.0 hours for tirofiban-RESTORE. Major adverse clinical events were infrequent. Two in-hospital deaths occurred, each after a large myocardial infarction in a patient from the abciximab and tirofiban-PRISM-PLUS groups, respectively. Only one patient (tirofiban-PRISM-PLUS) required urgent revascularization, and no strokes occurred. Major bleeding occurred in 1 patient (tirofiban-PRISM-PLUS).
Early Platelet Aggregation
Early PA in response to 20 μmol/L ADP is summarized in Figure 1 and Table 2. Baseline PA varied considerably within each regimen but not between regimens (Figure 1). PA was significantly less inhibited at 15 and 30 minutes with the tirofiban-RESTORE regimen compared with abciximab and eptifibatide (Figure 1, Table 2). At 15 minutes, fewer patients achieved ≥80% inhibition of PA with the tirofiban-RESTORE regimen compared with abciximab or eptifibatide regimens (Figure 3a). At this time, there were no significant differences between the tirofiban-PRISM-PLUS, eptifibatide, and abciximab regimens. At 30 minutes, however, PA was inhibited more with the tirofiban-PRISM-PLUS regimen than with the tirofiban-RESTORE regimen (Table 2).
Late Platelet Aggregation
With continued infusion of the tirofiban-RESTORE regimen, PA inhibition increased and became more uniform (Figure 2, Table 2). At 4 and 12 hours, both tirofiban regimens and eptifibatide markedly inhibited PA. With continued abciximab infusion, however, inhibition diminished and became more variable (Figure 2, Table 2 and Table 4). At 12 hours, eptifibatide inhibited PA more than abciximab (Table 2). Although there were no significant differences in PA between abciximab and each of the 2 tirofiban regimens at 4 and 12 hours (Table 2), abciximab showed significantly more variability during this time than both tirofiban regimens and eptifibatide (Table 4). Furthermore, fewer abciximab-treated patients maintained ≥80% inhibition at 12 hours than with eptifibatide or tirofiban-RESTORE (Figure 3B).
Linear Regression Analysis of Platelet Inhibition
To summarize the mean change in PA (slope) from the beginning to the end of each time interval, random-effects linear-regression models were fitted to the data (Table 3). Consistent with the direct comparisons described above, the predicted reduction in PA (slope) from baseline to 15 minutes was significantly less with the tirofiban-RESTORE regimen than with abciximab (Table 3). With continued infusion between 15 minutes and 4 hours, however, incrementally more inhibition of PA occurred with the tirofiban-RESTORE regimen (slope, −3.2) than with abciximab, for which there was recovery of PA (abciximab slope, +0.16; P=0.0003 versus tirofiban-RESTORE). With linear regression, the other drug regimens did not differ significantly from abciximab during any of the 3 major intervals.
Citrate Versus PPACK Anticoagulation
For eptifibatide and tirofiban, ex vivo PA was more inhibited in citrated specimens than in PPACK specimens (Table 5). In contrast, platelet inhibition after abciximab was greater in PPACK specimens compared with citrate.
The COMPARE trial provides a direct, randomized pharmacodynamic comparison of current abciximab, eptifibatide, and tirofiban regimens used to treat patients with ACS undergoing PCI. Although all regimens provided effective platelet inhibition, the tirofiban-RESTORE regimen produced less inhibition at 15 to 30 minutes compared with abciximab or eptifibatide. With continued infusion of the tirofiban-RESTORE regimen, platelet inhibition increased to levels comparable to those achieved by abciximab and eptifibatide. In contrast, although the abciximab regimen consistently inhibited PA early on, more recovery of PA occurred with continued infusion (4 to 12 hours). Of the 4 regimens evaluated, the eptifibatide regimen provided the most consistent platelet inhibition throughout infusion.
We did not anticipate that the tirofiban-RESTORE regimen would produce less inhibition of PA during PCI. This suggests that the 10 μg/kg tirofiban bolus may have less effect on PA than abciximab or eptifibatide loading strategies. There was also a statistically insignificant trend suggesting less inhibition early after the tirofiban-PRISM-PLUS regimen compared with abciximab and eptifibatide. Because our study was not statistically powered to detect subtle inter-drug differences, this lack of statistical significance may have been due to β error. The observation that at 30 minutes the tirofiban-PRISM-PLUS regimen produced significantly more platelet inhibition than the tirofiban-RESTORE regimen was also unexpected. Although this raises the possibility that a 10 μg/kg tirofiban bolus followed by a slower continuous infusion rate (0.15 μg · kg−1 · min−1) may produce less platelet inhibition than continuously loading tirofiban at a faster infusion rate (0.4 μg · kg−1 · min−1), the lack of randomization between these 2 regimens in this study renders this assertion inconclusive.
Experimental conditions used in prior dose-finding studies provide a potential explanation for the differences in the pharmacodynamic profiles observed. Tirofiban dosing has been on the basis of inhibiting PA in response to a weak agonist (5 μmol/L ADP), whereas current eptifibatide and abciximab regimens have been dosed to inhibit PA in response to more potent agonists (20 μmol/L ADP and thrombin).7–10⇓⇓⇓ The use of citrated blood in determining tirofiban dosing may have also resulted in under-dosing relative to eptifibatide and abciximab, as our results and others18 suggest that the calcium-chelating effects of citrate exaggerate ex vivo platelet inhibition with small molecule, low-affinity GP IIb/IIIa antagonists (eptifibatide and tirofiban). Unlike tirofiban, current eptifibatide regimens have already accounted for this phenomenon.9–11⇓⇓ If corroborated by other studies, the unexpected observation in our study that platelet inhibition with abciximab was greater in PPACK than in citrate raises the possibility that calcium concentrations might also affect this drug’s pharmacodynamics.
We observed that the intensity and predictability of platelet inhibition decreased 4 to 12 hours after abciximab infusion, whereas tirofiban and eptifibatide regimens provided consistent inhibition during this interval. Others have reported variable recovery of PA with continued abciximab infusion.19,20⇓ One explanation for this may relate to the pharmacological differences between abciximab and the small molecules. Plasma concentrations of free drug are much lower during abciximab infusion than with tirofiban or eptifibatide. These lower concentrations of unbound abciximab may be inadequate to inhibit the release of stored GP IIb/IIIa receptors occurring with further platelet activation.21 In the present study, as well as another,19 diabetic status and other factors (patient weight, platelet count, or hematocrit) did not account for abciximab’s pharmacodynamic variability.
Few other studies have directly compared the pharmacodynamics of platelet GP IIb/IIIa antagonists. Using turbidimetric aggregometry, Kereiakes et al20 also reported less inhibition of PA in response to 20 μmol/L ADP after tirofiban-PRISM-PLUS dosing compared with the eptifibatide and abciximab regimens. These inter-drug differences, however, were not evident using the Accumetrics rapid platelet-function assay (RPFA, Accumetrics). Tirofiban-RESTORE was not assessed in that study. Also using RPFA, Steinhubl et al22 reported no differences between the tirofiban-RESTORE, eptifibatide, and abciximab regimens 10 minutes after drug bolus. A possible explanation for the discrepancy between that study and the present study is that turbidimetric aggregometry may be more sensitive than RPFA in detecting subtle pharmacodynamic differences, as suggested by the results of Kereiakes et al.20 The only other randomized comparison of these regimens showed comparable platelet inhibition at 2 hours (using RPFA),23 results not inconsistent with ours.
The pharmacodynamic observations from the COMPARE trial provide a scientific rationale for the paradoxes observed in recent clinical trials. The lesser inhibition noted during PCI with the tirofiban-RESTORE regimen may explain the excess periprocedural thrombotic events reported with this regimen compared with abciximab in the do Tirofiban (Aggrastat) And ReoPro Give similar Efficacy outcomes Trial (TARGET).24 The potent inhibition achieved several hours after continued infusion of the tirofiban-PRISM-PLUS regimen in our study is consistent with this regimen’s demonstrated clinical efficacy in ACS patients when PCI is delayed until 4 to 48 hours after drug initiation, as observed in the PRISM-PLUS4 and Treat angina with Aggrastat and determine Costs of Therapy with Invasive or Conservative Strategies (TACTICS)-TIMI-1825 trials. The variable recovery of PA with continued abciximab infusion may account for its lack of efficacy when infused continuously for 24 to 48 hours in the Global Utilization of Streptokinase and tPA for Occluded arteries (GUSTO)-IV ACS study,26 despite overwhelming efficacy during PCI.1,12⇓ Taken together, the observations from the COMPARE trial and prior studies3,12,13,23,24,26⇓⇓⇓⇓⇓ suggest that potent platelet inhibition during PCI may be a key determinant of drug efficacy in this setting, whereas more prolonged inhibition may be necessary when GP IIb/IIIa antagonists are used as medical therapy for ACS (without PCI). This hypothesis is also supported by the fact that the vast majority of PCI-related thrombotic events occur during or soon after the procedure27 and relate to the degree of inhibition achieved during PCI,23 whereas ischemic events after ACS accrue more gradually.
This study has several limitations. The use of stronger platelet agonists, such as thrombin or collagen, might influence pharmacodynamics. We predict, however, that this would have only exaggerated the observed differences between regimens. We did not evaluate the pharmacodynamics of the double-bolus eptifibatide regimen used in the Enhanced Suppression of the Platelet IIb/IIIa Receptor with Integrilin Therapy (ESPRIT) trial;28 however, this regimen would be expected to provide even more potent inhibition than the PURSUIT regimen evaluated. We administered the 10 μg/kg tirofiban-RESTORE bolus over a slightly longer interval (10 minutes) than in RESTORE or TARGET (3 minutes).13,24⇓ The 15 and 30-minute samples, however, should still have reflected the impact of this bolus. Finally, we cannot exclude the influence of unknown confounders on these pharmacodynamics.
In summary, turbidimetric aggregometry detected significant differences in the pharmacodynamics of current platelet GP IIb/IIIa receptor antagonist regimens. A 10 μg/kg tirofiban bolus inhibited PA less than abciximab and eptifibatide regimens, and continued abciximab infusion was associated with variable recovery of PA. How these pharmacodynamic differences affect the relative efficacy of these regimens in ACS and PCI requires further investigation.
Supported by unrestricted grants from COR Therapeutics and Merck & Co. Drs Califf, Harrington, and Tcheng have received research support from COR Therapeutics, Merck & Co, or both. Dr Gretler is employed by COR Therapeutics, and Dr Dibattiste is employed by Merck & Co.
We would like to thank Anne N. Wagner for her assistance in preparing this manuscript.
- ↵Boersma E, Akkerhuis KM, Theroux P, et al. Platelet glycoprotein IIb/IIIa receptor inhibition in non-ST-elevation acute coronary syndromes: early benefit during medical treatment only, with additional protection during percutaneous coronary intervention. Circulation. 1999; 100: 2045–2048.
- ↵Kleiman NS, Lincoff AM, Flaker GC, et al. Early percutaneous coronary intervention, platelet inhibition with eptifibatide, and clinical outcomes in patients with acute coronary syndromes. Circulation. 2000; 101: 751–757.
- ↵Nurden AT, Poujol C, Durrieu-Jais C, et al. Platelet glycoprotein IIb/IIIa inhibitors: basic and clinical aspects. Arterioscler Thromb Vasc Biol. 1999; 19: 2835–2840.
- ↵Tcheng JE, Ellis SG, George BS, et al. Pharmacodynamics of chimeric glycoprotein IIb/IIIa integrin antibody Fab-7E3 in high risk coronary angioplasty. Circulation. 1994; 90: 1756–1764.
- ↵Tardiff BE, Jennings LK, Harrington RA, et al. Pharmacodynamics and pharmacokinetics of eptifibatide in patients with acute coronary syndromes: prospective analysis from PURSUIT. Circulation. 2001; 104: 399–405.
- ↵Gilchrist IC, O’Shea C, Kosoglou T, et al. Pharmacodynamics and pharmacokinetics of higher-dose, double-bolus eptifibatide in percutaneous coronary intervention. Circulation. 2001; 104: 406–411.
- ↵Phillips DR, Teng W, Arfsten A, et al. Effect of Ca2+ on GP IIb-IIIa interactions with Integrilin: enhanced GP IIb-IIIa binding and inhibition of platelet aggregation by reductions in the concentration of ionized calcium in plasma anticoagulated with citrate. Circulation. 1997; 96: 1488–1494.
- ↵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.
- ↵Born GV, Cross MJ. The aggregation of blood platelets. J Physiol. 1963; 163: 178–195.
- ↵Crowder MJ, Hand DJ. Analysis of Repeated Measures. London, UK: Chapman and Hall; 1990.
- ↵Diggle PJ, Liang KY, Zeger SL. Analysis of Longitudinal Data. Oxford, UK: Clarendon Press; 1994.
- ↵Mousa SA, Bozarth JM, Forsythe MS, et al. Differential antiplatelet efficacy for various GPIIb/IIIa antagonists: role of plasma calcium levels. Cardiovasc Res. 2000; 47: 819–826.
- ↵Steinhubl SR, Kottke-Marchant K, Moliterno DJ, et al. Attainment and maintenance of platelet inhibition through standard dosing of abciximab in diabetic and nondiabetic patients undergoing percutaneous coronary intervention. Circulation. 1999; 100: 1977–1982.
- ↵Steinhubl SR, Talley JD, Braden GA, et al. Point-of-care measured platelet inhibition correlates with a reduced risk of an adverse cardiac event after percutaneous coronary intervention: results of the GOLD (AU-Assessing Ultegra) multicenter study. Circulation. 2001; 103: 2572–2578.
- ↵Tcheng JE, O’Shea JC, Cohen EA, et al. Novel dosing regimen of eptifibatide in planned coronary stent implantation (ESPRIT): a randomized, placebo-controlled trial. Lancet. 2001; 356: 2037–2045.