The effect of cangrelor and access site on ischaemic and bleeding events: insights from CHAMPION PHOENIX

Abstract Aims To assess whether the use of the femoral or radial approach for percutaneous coronary intervention (PCI) interacted with the efficacy and safety of cangrelor, an intravenous P2Y12 inhibitor, in CHAMPION PHOENIX. Methods and results A total of 11 145 patients were randomly assigned in a double-dummy, double-blind manner either to a cangrelor bolus and 2-h infusion or to clopidogrel at the time of PCI. The primary endpoint, a composite of death, myocardial infarction, ischaemia-driven revascularization, or stent thrombosis, and the primary safety endpoint, Global Use of Strategies to Open Occluded Coronary Arteries (GUSTO) defined severe bleeding, were evaluated at 48 h. Of the patients undergoing PCI and receiving study drug treatment, a total of 8064 (74%) and 2855 (26%) patients underwent femoral or radial PCI, respectively. Among the femoral cohort, the primary endpoint rate was 4.8% with cangrelor vs. 6.0% with clopidogrel (odds ratio, OR [95% confidence interval, CI] = 0.79 [0.65–0.96]); among the radial cohort, the primary endpoint was 4.4% with cangrelor vs. 5.7% with clopidogrel (OR [95% CI] = 0.76 [0.54–1.06]), P-interaction 0.83. The rate of GUSTO severe bleeding in the femoral cohort was 0.2% with cangrelor vs. 0.1% with clopidogrel (OR [95% CI] = 1.73 [0.51–5.93]). Among the radial cohort, the rate of GUSTO severe bleeding was 0.1% with cangrelor vs. 0.1% with clopidogrel (OR [95% CI] = 1.02 [0.14–7.28]), P-interaction 0.65. The evaluation of safety endpoints with the more sensitive ACUITY-defined bleeding found major bleeding in the femoral cohort to be 5.2% with cangrelor vs. 3.1% with clopidogrel (OR [95% CI] = 1.69 [1.35–2.12]); among the radial cohort the rate of ACUITY major bleeding was 1.5% with cangrelor vs. 0.7% with clopidogrel (OR [95% CI] = 2.17 [1.02–4.62], P-interaction 0.54). Conclusion In CHAMPION PHOENIX, cangrelor reduced ischaemic events with no significant increase in GUSTO-defined severe bleeding. The absolute rates of bleeding, regardless of the definition, tended to be lower when PCI was performed via the radial artery. Clinical trial registration http://www.clinicaltrials.gov identifier: NCT01156571.


Introduction
The association between bleeding and increased morbidity and mortality after percutaneous coronary intervention (PCI) has prompted the implementation of bleeding reduction strategies at the time of PCI. 1,2 According to clinical trial data, compared with femoral access, the radial artery approach for PCI has been demonstrated to have similar efficacy with the benefit of reduced major bleeding and access site complications. 3 -6 Owing to a favourable safety profile, several guidelines endorse the radial artery as the preferred PCI access site. 7,8 Cangrelor is an intravenous P2Y 12 receptor antagonist with an immediate (within 2 min) onset of action and a half-life of 3-6 min, allowing platelet function to return to baseline within 60 min of infusion cessation. In the Cangrelor vs. Standard Therapy to Achieve Optimal Management of Platelet Inhibition (CHAMPION) PHOENIX trial, the intravenous P2Y 12 receptor antagonist, cangrelor, reduced the rate of ischaemic events at 48 h in patients undergoing percutaneous revascularization without a significant increase in severe bleeding or transfusions. 9 In this pre-specified secondary analysis, we explore the efficacy and safety of cangrelor according to access site (femoral vs. radial) in the CHAMPION PHOENIX trial.

Patient population
CHAMPION PHOENIX was a double-blind, double-dummy, placebocontrolled trial of cangrelor in patients undergoing PCI. Both the design and primary findings have been published previously. 9,10 Briefly, 11 145 patients undergoing either elective or urgent PCI and receiving guideline-recommended therapy were randomized after angiography to receive a bolus (30 mg/kg) and infusion (4 mg/kg/min for a minimum of 2 h or the duration of the procedure whichever was longer) of cangrelor or a 600 or 300 mg loading dose of clopidogrel. The timing (before or after PCI) and dose of clopidogrel were at the discretion of the site investigator. The access approach for PCI was determined by the site investigator and did not require institutional review board (IRB) approval. At the end of the infusion, patients then received either 600 mg of clopidogrel (cangrelor group) or matching placebo (clopidogrel group; Figure 1).

Endpoints
The primary efficacy endpoint was a composite of death (by any cause), myocardial infarction (MI), ischaemia-driven revascularization (IDR), or stent thrombosis in the first 48 h following randomization. Criteria for MI within 48 h post-PCI were defined as an elevation in creatine kinasemyocardial band (CK-MB) greater than three times the upper limit of normal or by a combination of CK-MB elevation in addition to ischaemic symptoms, angiographic evidence, and/or ECG changes. The key secondary endpoint was the incidence of stent thrombosis at 48 h, which was defined according to the Academic Research Consortium criteria or as intraprocedural stent thrombosis. 11 All events of death, MI, IDR, and stent thrombosis were adjudicated. The primary safety endpoint was severe non-coronary artery bypass grafting (CABG) bleeding according to the Global Use of Strategies to Open Occluded Coronary Arteries (GUSTO) criteria. Thrombolysis In Myocardial Infarction (TIMI) and Acute Catheterization and Urgent Intervention Triage strategY (ACUITY) bleeding definitions were also evaluated. Bleeding endpoints, based on prespecified criteria, were derived from investigator-reported data using a computer algorithm. Bleeding endpoints were not adjudicated.

Statistical analysis
This was a pre-specified analysis outlined in the CHAMPION PHOENIX study protocol, which was approved by local ethics committees. Independent verification of these analyses by Harvard Clinical Research Institute did not require specific IRB approval. Efficacy analyses were performed from the modified intention-to-treat (mITT) population, defined as subjects undergoing PCI and receiving study drug. Bleeding analyses were performed from the safety population, defined as subjects receiving study drug. Baseline characteristics were summarized by vascular access (femoral vs. radial) and treatment (cangrelor vs. The effect of cangrelor and access site on ischaemic and bleeding events clopidogrel); and were compared using analysis of variance for continuous variables and the x 2 test for categorical variables. Treatment comparisons within access site (cangrelor vs. clopidogrel) were not adjusted for baseline characteristics, were based on the x 2 test, and are presented as odds ratios with 95% confidence intervals (CIs). The number needed to treat (NNT) was derived within the femoral and radial subgroups as the inverse of the difference between the cangrelor and clopidogrel event rates. The interaction between treatment and PCI access site was tested using the Breslow-Day method. Time-to-event curves for the primary efficacy endpoint at 48 h were constructed using the Kaplan -Meier method and compared using the log-rank test.
Multivariable logistic regression modelling was performed using propensity scores to assess the effect of PCI access site (femoral or radial), using femoral approach as the reference, with study treatment in the model and the propensity score based on the following potential variables: diagnosis at presentation, region (US/non-US), smoking status, hyperlipidaemia, previous MI, prior PCI, prior CABG, peripheral artery disease, planned clopidogrel loading dose (300 or 600 mg), and anticoagulant (bivalirudin or heparin). Statistical analyses were conducted using the SAS software, version 9.3 (SAS Institute, Cary, North Carolina, USA).

Patients
Out of 11 145 patients randomized in CHAMPION PHOENIX, 10 942 patients comprised the mITT population who received study treatment and underwent PCI. Of these patients, 8064 (74%) underwent PCI via the femoral artery and 2855 (26%) via the radial approach. Three patients in the radial cohort did not complete the 48-h post-PCI follow-up and were excluded from the efficacy endpoint analyses; 23 patients underwent brachial PCI and were not included in this analysis.

Cangrelor vs. clopidogrel
Baseline characteristics according to access site (femoral vs. radial) and randomized treatment (cangrelor vs. clopidogrel) are depicted in Table 1. In the femoral cohort, subjects randomized to cangrelor had higher rates of peripheral artery disease (8.2 vs. 6.7%, P ¼ 0.01). In the radial cohort, subjects randomized to cangrelor had a lower median weight (84 vs. 85 kg, P ¼ 0.008).
In both access cohorts, there were no significant differences in the rates of GUSTO severe bleeding, TIMI major bleeding, or blood transfusions in patients treated with cangrelor compared with clopidogrel. In the femoral cohort, the rate of ACUITY major bleeding was 5.2% with cangrelor vs. 3 Table 3. The effects of cangrelor on bleeding in the overall study population and according to access site are shown in Figure 3B.  In the femoral cohort, the rate of GUSTO severe/moderate bleeding was 0.5% compared with 0.2% in the radial cohort

Discussion
Intravenous adenosine diphosphate (ADP) receptor blockade with cangrelor, when compared with clopidogrel, reduced the primary composite outcome of death, MI, IDR, or stent thrombosis at 48 h after randomization regardless of PCI access site. Among the femoral access subjects, cangrelor compared with clopidogrel reduced the odds of the primary composite outcome by 21%; within the radial cohort, there was a consistent 24% reduction in the odds of the primary composite outcome. Although the interaction tests for access site did not reach statistical significance, when the primary composite endpoint was evaluated according to access site, cangrelor demonstrated a reduction in odds of ischaemic events among patients undergoing femoral PCI. In patients undergoing PCI via the radial artery, cangrelor demonstrated a non-significant trend towards fewer ischaemic events, likely due to lack of statistical power because of a smaller sample size. In the femoral cohort, cangrelor's benefit with respect to ischaemic events was driven by a reduction in MI and stent thrombosis, consistent with the overall CHAMPION PHOENIX results. The radial cohort, however, only experienced a reduction in MI. The lack of benefit regarding stent thrombosis in the radial cohort is conceivably due to a low event frequency within this particular subgroup: 95 total (1.2%) in the femoral vs. 25 total (0.9%) in the radial. In both the femoral and radial groups, cangrelor compared with clopidogrel was not associated with a significant increase in the pre-specified GUSTO-defined severe bleeding, the primary safety  endpoint, or in blood transfusions; though more sensitive definitions such as ACUITY-defined bleeding did show increased rates of bleeding with cangrelor in both the femoral and radial cohorts. It is important to note that the absolute rates of GUSTO severe/moderate bleeding and blood transfusions were approximately two to three times higher with the femoral approach compared with the radial, in both the cangrelor and clopidogrel arms of the study. These findings are consistent with prior studies. The analysis of all femoral vs. radial randomized PCI clinical trials has found radial access, when compared with femoral, to be associated with a 42% reduction in non-CABG bleeding. 5 Contemporary studies have demonstrated that up to 70% of bleeding events occurring in the PCI setting can be attributed to complications arising at the site of vascular access. 12 The present study suggests that in a large contemporary international trial, the radial approach for PCI has the potential to play a key role in reducing periprocedural bleeding in a wide variety of PCI patients. It is postulated that a reduction in access site bleeding may in turn lead to fewer subsequent adverse events. 13 For example, the aforementioned 42% reduction in major non-CABG bleeding associated with the radial approach for PCI was paralleled with an aggregate reduction of major adverse cardiac events (death, MI, or stroke) of 14%, P ¼ 0.005. 5 After multivariable analysis the present study finds the odds of periprocedural bleeding 30-66% lower, depending on bleeding definition, when PCI was performed via the radial artery compared with the femoral approach. However, the favourable bleeding profile associated with the radial artery approach to PCI in CHAMPION PHOENIX did not translate into a reduction in the primary efficacy endpoint at 48 h.
There are certain limitations to this analysis. First, similar to the overall CHAMPION PHOENIX trial, there is the potential for the benefit of cangrelor to be attenuated in the setting of more prolonged pretreatment with clopidogrel or with the use or ticagrelor or prasugrel. Secondly, bleeding endpoints were not adjudicated. Thirdly, the treatment by radial vs. femoral access was not randomized and even with the adjusted analyses, there may be residual confounding. Lastly, CHAMPION PHOENIX was not powered to test the interaction between treatment and PCI access site; therefore, all interaction terms should be interpreted with caution.
In CHAMPION PHOENIX, intravenous ADP-receptor inhibition with cangrelor reduced ischaemic events with no significant increase in severe bleeding or blood transfusions regardless of PCI access site. Compared with the femoral approach, rates of bleeding complications appeared to be lower with radial access for PCI in both randomized arms of the study. Institute received funding from The Medicines Company for these analyses.