The CELEcor Blinded RAndomized trial in sTE-elevation myocardial infarction (CELEBRATE) trial was an international, double-blind, placebo-controlled, randomised, Phase III, superiority trial in patients with ST-elevation MI (STEMI) and intended percutaneous coronary intervention (PCI). The trial compared zalunfiban administered at first medical contact at a single dose of either 0.110 mg/kg or 0.130 mg/kg subcutaneously versus placebo in a 1:1:1 ratio, on top of standard therapy, including a loading dose of aspirin 500 mg IV, ticagrelor (180 mg orally) and unfractionated heparin (UFH; 5,000 IU IV).1 The primary endpoint was a composite, ranked, hierarchical (7 points to worst down to 1 point to best) clinical scale of all-cause death, stroke, recurrent MI, acute stent thrombosis, new onset or rehospitalisation for heart failure, larger infarct size or none of the above, over 30 days of follow-up. The primary safety endpoint was severe or life-threatening bleeding according to the Global Use of Strategies to Open Occluded Coronary Arteries (GUSTO) criteria.2 Exclusion criteria included cardiogenic shock, out-of-hospital cardiac arrest, current use of oral anticoagulation, haemodialysis, recent major bleeding, surgery or a history of stroke, and obesity, with a body weight of >130 kg.
There were 2,467 participants (853 on zalunfiban 0.11 mg/kg, 818 on zalunfiban 0.13 mg/kg, and 796 on placebo) within 4 hours from symptom onset. The primary efficacy endpoint was significantly reduced by zalunfiban versus placebo, with an adjusted OR of 0.79 (95% CI [0.65–0.98], p=0.028). This difference was mainly driven by a 0.8% absolute reduction in 24-hour stent thrombosis (ranked 4) and by a 3.5% absolute reduction in ‘no major adverse events’ (ranked 1), while the remaining points were largely superimposable between the two groups. GUSTO severe bleeding was non-significant (0.4% higher in those receiving zalunfiban [1.2% versus 0.8%; p=0.40]), while GUSTO mild-to-moderate bleeding was significant, with almost a threefold increase in the zalunfiban group (6.4% versus 2.5%, p<0.001). Similarly, the Bleeding Academic Research Consortium (BARC) 2–5 type bleeding was nearly triple in the zalunfiban group, largely driven by an increase in BARC 2 type. Thrombocytopenia (defined as platelet count <150.000/μl) occurred in 8.1% and 6.3% of the zalunfiban and placebo groups, respectively.
The CELEBRATE trial addresses a commonly debated issue, i.e. whether a fast and highly effective platelet inhibition at first medical contact can improve outcomes. A recent report from the American College of Cardiology concluded that STEMI is still associated with high out-of-hospital cardiovascular mortality, cardiogenic shock on arrival (10%) and cardiac arrest before intervention (5.1%), resulting in 4–7% mortality.3 Another study reported that 2.1% of patients with STEMI die soon after first medical contact and before reaching the hospital.4 Thus, early action remains an unmet need, but trials so far have been negative.
The rationale of CELEBRATE was that an expedited, subcutaneous administration of a drug inhibiting the final common pathway of platelet aggregation might block thrombus progression and facilitate reperfusion. Older glycoprotein IIb/IIIa receptor inhibitors (GPIs), such as the antibody abciximab and the small molecule tirofiban, while improving the outcomes of reperfusion, required continuous IV infusion, had a very long half-life (days in the case of abciximab), and a potential for triggering platelet activation and clumping. Those pharmacokinetic and pharmacodynamic features were associated with a high incidence of major bleeding, drug-induced thrombocytopenia and problematic overlapping with loading doses of the other antiplatelet agents before PCI.5 Therefore, GPIs (eptifibatide and tirofiban) are recommended only for bailout or peri-procedural complications, with limited use.6
Zalunfiban is a novel, small molecule that stabilises platelet glycoprotein IIb/IIIa receptor (GPR) in its inactive form, hampering its transition into a fibrinogen-binding conformation. Zalunfiban has more favourable pharmacokinetics and pharmacodynamics than earlier GPIs. Stabilising the GPR in the inactive conformation may reduce the risk of GPI-induced thrombocytopenia, a rare (<1%) but feared side effect of earlier GPIs.7 Moreover, the half-life of zalunfiban is short (Tmax 15 minutes, terminal half-life 3 hours), which may avoid significant overlap with the loading doses of aspirin and P2Y12 inhibitors, while weaning of GPR inhibition may reduce bleeding risk during PCI. Another practical advantage of zalunfiban is the subcutaneous injection, based on high solubility in small volumes (1–2 ml), allowing a rapid administration by first-aid personnel.
However, the CELEBRATE trial has several limitations. First, it was originally designed as a dose-finding, Phase IIb study with surrogate endpoints, i.e. the restoration of coronary artery blood flow at initial angiography and the resolution of ST-segment deviation 1 hour after PCI in STEMI patients treated in the ambulance. After consultation with regulatory authorities, the trial was converted into a Phase III trial, with adaptation of the primary surrogate endpoint into a composite, ranked, hierarchical, 7-point clinical scale and the two zalunfiban doses were merged for the primary analysis.
Second, although the pooling of zalunfiban 0.11 and 0.13 mg/kg doses was prespecified to ensure sufficient power to assess efficacy, the background of the highest dose appears undocumented. After initial safety studies in human volunteers, the Phase IIa dose-finding study included 27 STEMI patients across three cohorts receiving escalating, weight-adjusted 0.075 (n=8), 0.090 (n=9) or 0.110 mg/kg (n=10) subcutaneous zalunfiban.8 Zalunfiban showed a dose-dependent inhibition of platelet aggregation (≥77% inhibition with the iso-TRAP channel, which corresponds to 80% inhibition of light transmission aggregometry stimulated by 20 µM ADP) in patients at the highest dose of 0.110 mg/kg, 15 minutes after subcutaneous administration.8 Thus, an entirely new dose of zalunfiban was added to a Phase III study protocol, which is uncommon in Phase III trial design.
How did the additional dose of 0.130 mg/kg make its way into the Phase III trial? Was this dose ever tested? No rationale or data are given for this additional dose. One could speculate that the doses tested in the Phase IIa trial were not as effective in inhibiting platelets as expected. The original sample size calculation for the original Phase IIb trial was based on an expected effect size for the thrombolysis in MI frame count and ST-segment resolution, resulting in a far smaller number of patients needed to be included in this trial (1,236) than eventually proposed for the final Phase III CELEBRATE trial (2,499).
Third, the primary outcomes of CELEBRATE include endpoints not fully dependent on platelets (all-cause mortality, early heart failure, “none of the above”) and merge efficacy and safety endpoints (ischaemic and haemorrhagic stroke), over a very short follow-up (30 days). As a matter of fact, the major driver of the benefit was ‘none of the above’, while mortality and MI did not differ. Moreover, it seems that only the secondary outcome of post-PCI angiography infarct size was determined objectively, was measured after 24 hours and assessed by a central laboratory. All other outcomes (accident and emergencies, bleeding, injection reaction, and mortality) were based on information gathered during follow-up telephone calls 30 days and 1 year after hospitalisation and, in cases of stroke diagnosis, an additional telephone call was made after 90 days to record stroke disability. Therefore, the 30-day outcomes and those at 1 year might be less reliable because they are based on telephone calls.
Fourth, concluding that the safety of zalunfiban and placebo is similar does not seem to be supported by trial data. Based on a small sample size (~800 patients per arm) followed for 30 days, the small absolute number of events in 30 days (six events in the placebo arm) cannot allow any reliable conclusion. Moreover, two elements should be considered: GUSTO moderate and BARC-2 events were significantly increased by zalunfiban, which is consistent with its antiplatelet action. In addition, the point estimate of the risk of major GUSTO bleeding (OR 1.6) and the upper limit of the 95% CI interval (4.89) cannot exclude harm. Thus, zalunfiban was associated with 3.5% more mild or moderate bleeding, with a number needed to harm of 28, while the absolute risk reduction was overall 3.5%, corresponding to a close number needed to treat of 29.
In the discussion, the authors claim that zalunfiban was “associated with lower rates of stent thrombosis, new-onset heart failure or heart failure requiring hospitalisation, or high-sensitivity troponin T release greater than 30 times the upper limit of normal”, all known to be associated with mortality. However, the data presented in Table 3 for each major adverse clinical endpoint shows only a statistically significant reduced OR for acute stent thrombosis <24 hours post-PCI (0.77, 95% CI [0.58–1.00]). Altogether, it appears that the efficacy gain is marginally based on biochemical markers, without any convincing impact on hard clinical outcomes except for acute in-stent thrombosis.
What is the overall added clinical value of the addition of zalunfiban to standard treatment? At present, several questions remain unanswered by the CELEBRATE trial. However, this trial may open the avenue for new studies on new GPIs with improved design, stronger hypotheses, clearer efficacy and safety endpoints and, consequently, larger sample size. Given the remaining complications of STEMI, despite rapid and aggressive antithrombotic management, a gain in myocardial salvage can be worthwhile if it is cost-effective and adequately demonstrated.