Bridging Antiplatelet Therapy With Cangrelor in Patients Undergoing Cardiac Surgery:  A Randomized Controlled Trial FREE

Dual antiplatelet therapy with aspirin and an oral P2Y₁₂ receptor inhibitor is the standard of care to prevent the short- and long-term risk of recurrent atherothrombotic events in high-risk settings, such as patients with an acute coronary syndrome (ACS) and those undergoing percutaneous coronary intervention (PCI).^1- 4 However, the ischemic benefit associated with more intense platelet blockade in these high-risk settings occurs at the expense of an increased risk of bleeding complications. Given that the risk of bleeding is significantly increased among patients undergoing surgical procedures, in particular coronary artery bypass grafting (CABG) surgery, discontinuation of antiplatelet therapy for a time frame that allows recovery of platelet function is warranted.^1- 4

However, premature discontinuation of antiplatelet therapy in these settings has been associated with an increase in ischemic complications.^5- 8 These concerns are magnified among patients treated with drug-eluting stents in whom thrombotic occlusions occurring as a consequence of antiplatelet treatment discontinuation, in particular P2Y₁₂ receptor blockers, are associated with substantial morbidity and mortality.^9- 12 These findings underscore the need to define strategies of platelet inhibition that allow to safely “bridge” patients to their surgical procedure with minimum risk of ischemic events or bleeding complications.

Cangrelor, a nonthienopyridine adenosine triphosphate analogue, is an intravenous (IV) antagonist of the P2Y₁₂ receptor characterized by rapid, potent, predictable, and reversible platelet inhibition with rapid offset of effect.¹³ Therefore, this compound has desirable pharmacodynamic properties to be considered for bridging patients to surgery in whom discontinuation of antiplatelet therapy, particularly a P2Y₁₂ receptor inhibitor, can lead to adverse consequences (eg, stent thrombosis) while preserving normal hemostasis at the time of surgery.¹⁴ In the current trial we hypothesized that cangrelor may be a safe and effective drug to bridge patients from irreversible platelet P2Y₁₂ inhibitors to open heart surgery.

The trial was designed and led by an executive committee that was chaired by a BRIDGE lead investigator (E.J.T.) and included academic investigators in conjunction with the sponsor, The Medicines Company. Pertinent national regulatory authorities and ethics committee at participating centers approved the protocol. All patients provided written informed consent.

The BRIDGE trial consisted of 2 independent stages. Eligibility criteria were the same for both stages. Stage 1, conducted between January 2009 and August 2009 and represented an open-label phase of the study, aimed to identify the dose of cangrelor that would achieve a desired antiplatelet effect after thienopyridine discontinuation. Specifically, cangrelor IV infusion was to be administered to cohorts of 5 patients at a time in a step-wise fashion at predetermined doses (0.5 μg/kg, 0.75 μg/kg, 1.0 μg/kg, and 1.5 μg/kg per minute) until percent platelet inhibition as measured by VerifyNow P2Y12 (Accumetrics) was greater than 60% in 80% of daily samples or a dose of 2.0 μg/kg per minute was reached.

Stage 2 was a prospective, randomized, double-blind, placebo-controlled phase of the study enrolling patients independent of stage 1 conducted between October 2009 and April 2011. The aim was to assess whether a cangrelor IV infusion (at the dose determined in stage I) would maintain levels of platelet reactivity of less than 240 P2Y₁₂ Reaction Units (PRUs) throughout the preoperative period as measured by the VerifyNow P2Y12 assay. This level approximated the levels of platelet reactivity expected to be maintained if a thienopyridine had not been discontinued.^15- 16

Eligible patients were randomly assigned in a 1:1 ratio by an interactive voice-response system (IVRS) to receive cangrelor infusion plus standard of care or placebo infusion plus standard of care, using the cangrelor dose determined in stage 1. The fixed, blocked, permuted randomization schedule was generated with a block size of 4 and was stratified according to the anticipated delay until CABG surgery (≤3 days or >3 days).

The study drug was provided in blinded kits and allocated by IVRS. Study drug infusion was initiated after thienopyridine therapy was discontinued and was continued throughout the preoperative period up to 1 to 6 hours before surgical incision. The study drug was not administered during or after CABG surgery. It was recommended that patients wait 5 days after discontinuation of ticlopidine and clopidogrel, and 7 days after prasugrel, before undergoing surgery in accordance with practice guidelines.^1- 4 However, the timing of surgery was left to the discretion of the investigator with a minimum of 48 hours and an allowed maximum of 7 days of study drug infusion. Aspirin therapy was maintained at a dosing regimen as per routine local practice.

Safety analyses were carried out using data from patients who had received at least 1 dose of the study drug. A dose confirmation analysis and safety data review was conducted by an independent and unblinded data safety monitoring board (DSMB) based on the prespecified plan. According to prespecified DSMB charter, the DSMB reviewed the dosing data for the first 24 patients randomized into stage 2 on April 20, 2010. To allow for a better assessment of the optimal dose further unblinded review was requested by the DSMB. This review confirmed the dose; however, given the unblinded review, the DSMB and executive committee recommended exclusion of these patients from primary efficacy analysis.

Patients at least 18 years of age planned to undergo nonemergency CABG surgery were eligible to be enrolled. All patients had to have received a thienopyridine (at least 500 mg ticlopidine, 75 mg of clopidogrel, or 10 mg of prasugrel) within at least 72 hours prior to randomization either for the treatment of an ACS or for long-term preventive therapy following coronary stent implantation, drug-eluting stents, or bare metal stents. CABG surgery, either on-pump or off-pump, had to occur no sooner than 48 hours but no longer than 7 days from randomization, with patients hospitalized until planned CABG surgery. Exclusion criteria are described in the eAppendix).

Platelet function was assessed with the point-of-care platelet function test according to manufacturer instructions before, during, and after study drug infusion.^15- 16 Blood sampling for platelet function testing during study drug infusion was performed daily; it was recommended that blood sampling be drawn daily at the same time. Study drug infusion was discontinued 1 to 6 hours prior to CABG surgery and the last sample for platelet function testing had to be within 12 hours before infusion stop. Blood sampling after study drug stop had to be just before surgical incision.

This test has been previously described in detail.^15- 16 In brief, this test measures adenosine diphosphate–induced platelet agglutination as an increase in light transmittance and uses a proprietary algorithm to report values in PRU. A higher PRU result reflects greater P2Y₁₂-mediated reactivity. A second activator, isothrombin-receptor–activating peptide, is incorporated into a second channel of the assay device and provides an estimated inhibition (percent VerifyNow inhibition) without a prethienopyridine sample by reporting the ratio of the results of the adenosine diphosphate and isothrombin-receptor–activating peptide channels. Specialized software developed for the trial encrypted the platelet function results to maintain double blinding.

The primary efficacy end point of stage 1 was maintenance of platelet inhibition in at least 80% of patient samples above 60% as determined by point-of-care platelet function testing measured during the study drug infusion prior to surgery. The primary efficacy end point of stage 2 was the proportion of patients with platelet reactivity of less than 240 PRU for all samples assessed during study drug infusion prior to surgery. Additional efficacy end points included the percentage of total patient samples that maintained more than 60% platelet inhibition during the study drug infusion; the percentage of total patient samples that maintained platelet reactivity of less than 240 PRU during the study drug infusion; the percentage of patients who maintained platelet reactivity of less than 240 PRU in their last serum sample taken while receiving treatment sample before surgery; and the percentage of patients in whom all platelet reactivity evaluations during study drug infusion prior to surgery were less than or equal to their baseline platelet reactivity prior to receipt of the study drug.

The main safety end point (secondary end point) of the trial was excessive CABG surgery–related bleeding, as defined by the occurrence of 1 or more of the following 3 components during the CABG procedure through hospital discharge: surgical reexploration, 24-hour chest tube output of more than 1.5 L, or packed red blood cell transfusion of more than 4 units. In addition to the protocol-defined end point of excessive CABG surgery–related bleeding, the Bleeding Academic Research Consortium (BARC)–defined CABG surgery–related bleeding¹⁷ was assessed, ie, the occurrence of 1 or more of the following during the CABG surgical procedure through hospital discharge: fatal bleeding; perioperative intracranial bleeding within 48 hours; reoperation following closure of sternotomy for the purpose of controlling bleeding; transfusion of 5 or more units of whole blood or packed red blood cells within a 48-hour period; and chest tube output of at least 2 L within a 24-hour period.¹⁷ Preoperative bleeding defined according to the Thrombolysis in Myocardial Infarction (TIMI), Global Use of Strategies To Open coronary arteries (GUSTO), and Acute Catheterization and Urgent Intervention Triage Strategy (ACUITY) definitions,^18- 20 as well all blood product transfusions up to 7 days after CABG surgery or discharge, whichever was sooner, were also collected.

Additional safety observations included the incidence of combined ischemic end point of death, myocardial infarction, stroke, or need for urgent revascularization from the time of randomization until discontinuation of the study drug and within 30 days following CABG surgery. Incidence of other adverse events and serious adverse events were recorded up to 7 days after CABG surgery or discharge, whichever occurred sooner (see eAppendix for Protocol Definitions for Efficacy and Safety Assessment available at http://www.jama.com). All safety and ischemic events were site reported and nonadjudicated.

We estimated that, assuming 30% of placebo-treated patients and at least 60% of cangrelor-treated patients would reach the primary end point, a sample size of 106 patients (53 for each group) would provide 90% power to detect a statistically significant difference between the treatment groups at a 2-sided α of .05 using a χ² test. The sample-size calculation was performed with PASS 2008 software(Power Analysis and Sample Size, NCSS). We selected a sample size of up to 100 patients per treatment group to further evaluate the safety of cangrelor before and after CABG surgery.

Two populations were defined and used in the analysis and presentation of the data. The safety population included patients who received any study drug, and patients were classified according to the actual treatment received. The intention-to-treat (ITT) population was defined as all patients randomized during the trial's second stage and who received the study drug except for the first 24 patients whose data were unblinded for the DSMB review dose confirmation. Treatment classification for ITT analysis was based on the randomized treatment. Platelet function results were considered valid if at least 1 sample was confirmed as having been drawn while the patient was receiving the infusion and if the sample was analyzed within the manufacturer's recommended time window. Per-protocol, missing data were not imputed; therefore, the efficacy analysis involved only patients in the ITT population with valid PRU data. Sensitivity analyses including a per-protocol analysis (excluding major deviations) and an analysis with safety was also performed.

Baseline difference of quantitative factors, such as duration of infusion, was compared using the Wilcoxon test. The primary efficacy end point, the percentage of patients who maintained a PRU of less than 240 during study drug infusion prior to surgery, was analyzed using logistic regression adjusted for the expected days to surgery (either ≤3 days or >3 days) with the ITT population. The χ² test was performed for other efficacy end points measured in proportions. Analysis of variance was used to compare quantitative end points, such as the PRU value, between the treatment groups. No missing data imputation or multiple comparison α adjustment was applied. SAS version 9.2 (SAS Institute Inc) for Windows was used for all statistical analysis.

A total of 11 patients from 4 sites in the United States participated in the open-label dose-finding phase of the trial (stage 1). Of these, 10 patients had evaluable data for the purpose of dose considerations; 5 patients completed the first cohort at a dose of 0.5 μg/kg per minute; and 5 completed the second at a dose of 0.75 μg/kg per minute (eTable 1). The cangrelor dose of 0.75 μg/kg per minute met the efficacy end point of maintenance of platelet inhibition in at least 80% of patient samples higher than 60% (94.4% [17 of 18]; 95% CI, 83.9%-100%), and was considered for the randomized, double-blind, placebo-controlled phase of the trial. This dose was confirmed by designated unblinded review of data of the first 24 patients randomized to stage 2.

A total of 210 patients from 34 global sites (eAppendix, listed by investigator and country) requiring bridging from oral thienopyridine therapy to CABG surgery were randomly assigned in the blinded strata (stage 2) to either cangrelor (n = 106) or placebo (n = 104). One patient in the cangrelor group and 2 in the placebo group did not receive study medication; 1 patient in the placebo group was erroneously given an active medication kit. Therefore, a total of 106 treated with cangrelor and 101 patients treated with placebo represented the safety population of this study. The first 24 patients (12 per group) were excluded from the efficacy analyses because of the unblinding requirement made by the DSM. Therefore, the efficacy-ITT population was composed of 93 patients in the cangrelor group and 90 in the placebo group. Only 1 patient (0.9%), randomly assigned to placebo, was lost to 30-day follow-up. Patient disposition is illustrated in Figure 1.

The treatment groups were well balanced with regard to baseline clinical and demographic characteristics (Table 1). In the cangrelor group, 15.1% of patients had presented with an ST-elevation MI and 32.1% with a non-STEMI, whereas, 11.9% of patients in the placebo group had presented with an ST-elevation MI and 44.5% with a non-STEMI; the remaining patients were enrolled in a nonacute setting. The CABG surgery was performed in 102 patients (96%) in the cangrelor group and 96 (95%) in the placebo group.

The median time from thienopyridine discontinuation to study drug infusion was 29.1 hours (interquartile range [IQR], 11-38 hours) in the cangrelor group and 29.5 hours (IQR, 14-39 hours) in the placebo group (Wilcoxon, P = .80; eTable 2). The median duration of infusion was 2.8 days (IQR, 2.5-3.8 days) for cangrelor vs 3.4 days (IQR, 2.6-4.7 days) for placebo (Wilcoxon, P = .046). The median time from discontinuation of study drug infusion to surgical incision was 3.2 hours (IQR, 2-5 hours) in the cangrelor group and 3.2 hours (IQR, 2-5 hours) in the placebo group (Wilcoxon, P = .82).

Valid PRU results were not available for 15 patients (9 in cangrelor and 6 in placebo groups) in the planned ITT analysis population during the infusion period. The median number of valid PRU values was 3.0 in both groups (eTable 3). At baseline, prior to infusion of study medication, the levels of platelet reactivity (P = .82) and percentage of patients with platelet reactivity lower than 240 PRU (P = .19) did not differ between groups (Table 2).

The primary efficacy end point of percentage of patients with platelet reactivity less than 240 PRU throughout the entire infusion of study drug was significantly higher in the cangrelor group (98.8% [83 of 84]; 95% CI, 96.5%-100%) than in the placebo group (19.0% [16 of 84]; 95% CI, 10.7%-27.4%; P < .001) The crude relative risk (RR) was 5.2 (95% CI, 3.3-8.1) and the RR adjusted for expected days in surgery was 5.2 (95% CI, 3.3-8.0). These outcomes were yielded independent of prior thienopyridine dose and time of discontinuation (RR, 4.3; 95% CI, 2.8-6.6) and were consistent with the sensitivity analysis (95.9% [93 of 97] vs 20.0% [19 of 95]; RR, 4.8; 95% CI, 3.2-7.2). Adjusting for both expected days to surgery and duration of infusion did not alter the effect of cangrelor (RR, 5.1; 95% CI, 3.3-8.0).

The percentage of overall samples displaying platelet reactivity less than 240 PRU, patients with all samples with baseline PRU value, total patient samples that maintained higher than 60% platelet inhibition during study drug infusion were all greater with cangrelor than with placebo (all P < .001; Table 2). Accordingly, the last sample taken during study drug infusion showed lower PRU values and a greater percentage of patients with platelet reactivity lower than 240 PRU with cangrelor (all P < .001; Table 2). Following discontinuation of the study medication infusion prior to surgical incision, PRU levels (P = .21) and the percentage of patients with platelet reactivity lower than 240 PRU (P = .31) were similar between groups (Table 2). The distribution of platelet reactivity during the overall study time course is illustrated in Figure 2.

Study-defined excessive CABG surgery–related bleeding occurred in a total of 22 patients and was not significantly different between patients randomly assigned to receive cangrelor (11.8% [12 of 102]) or placebo (10.4% [10 of 96]) (RR,1.1; 95% CI, 0.5-2.5; P = .76). There were no differences in BARC-defined CABG surgery–related bleeding occurring during the surgical procedure until patient discharge (RR, 0.9; 95% CI, 0.4-2.2; P = .89; Table 3). Pre-CABG–surgery major-bleeding events were rare and not different between treatment groups; minor bleeding events were numerically more frequent with cangrelor (Table 3).

Minor bleeding during study drug infusion was mostly attributed to ecchymosis at the site of venipuncture (eTable 4 on ACUITY minor bleeding available at http://www.jama.com).

Ischemic end points were low, 2.8% (3 of 106) and 4.0% (4 of 101) in cangrelor and placebo, respectively, prior to surgery (eTable 5). Adverse events occurred similarly in both groups. Incidence of dyspnea was low; 1.9% with cangrelor vs 1.0% with placebo (3 events total). Incidence of postbaseline clinically significant laboratory tests in hematology and serum chemistry was low and similar in both groups (eTables 6 and 7).

We compared the antiplatelet efficacy of cangrelor with placebo in patients undergoing CABG surgery who were being treated with a thienopyridine for secondary prevention of recurrent events following an ACS or coronary stent implantation. Cangrelor at an infusion dose of 0.75 μg/kg per minute consistently achieved and maintained platelet inhibition at levels known to be associated with a low risk of thrombotic events compared with placebo. Bridging with a prolonged infusion of cangrelor did not increase major bleeding prior to surgery, as defined according to several established classifications, although minor bleeding was numerically higher.

The rapid recovery of platelet function after discontinuing cangrelor infusion is shown by the similar levels of platelet inhibition compared with placebo prior to CABG surgery and is consistent with the short half-life of cangrelor (3-6 minutes). There was no excess in CABG surgery–related bleeding with cangrelor. In addition, there was no increased incidence of nonbleeding adverse events (including dyspnea) or laboratory abnormalities despite extended dosing. These observations support the hypothesis that intravenous cangrelor is a feasible management strategy, providing prolonged platelet P2Y₁₂ inhibition in patients who must wait for cardiac surgery after thienopyridine discontinuation.

The early benefits associated with oral P2Y₁₂ receptor inhibition, in particular clopidogrel, have made its upstream use the standard of care in ACS patients.^{1- 4,21- 22} These cardiovascular benefits have also been observed in patients requiring CABG surgery, not only with clopidogrel but also with the more potent novel P2Y₁₂ receptor blockers prasugrel and ticagrelor.^23- 25 Thienopyridines irreversibly block the P2Y₁₂ receptor for the life-span of the platelet; therefore, the trade-off of this treatment strategy is the increased risk of bleeding complications in patients requiring surgery if they have been exposed to thienopyridine therapy within the prior 5 to 7 days. Similar safety concerns exists for ticagrelor, an oral nonthienopyridine P2Y₁₂–inhibiting agent, which despite its reversible effects has an offset of action that is relatively slow, requiring a wash-out period of 5 to 7 days prior to surgery.^25- 26

Moreover, consistent with reports assessing platelet reactivity while taking clopidogrel,²⁷ even after discontinuation of thienopyridine therapy, there was a broad variability in platelet reactivity. In the placebo group, there were a considerable number of patients with a last PRU lower than 240, indicating that not only a large number of patients may not be adequately protected when stopping thienopyridine therapy for up to a week and that large number of patients may be exposed to an increased risk of perioperative bleeding due to ongoing platelet inhibition at the time of CABG surgery.²⁸ A strategy to maintain adequate platelet inhibition until the time of surgery, while avoiding the complications of both coronary thrombosis and surgical bleeding, is currently lacking.

This concern has occurred because premature discontinuation of thienopyridine treatment is associated with an increased risk of stent thrombosis that often leads to myocardial infarction and death.^9- 12 Cessation of the thienopyridine for nearly a week before surgery, with patients not hospitalized or monitored, but carrying an excess risk of major ischemic events, has been a troubling and not infrequent problem for clinicians, because it is estimated that approximately 5% of patients will require some type of surgery within the first 12 months after stent implant or an ACS diagnosis.¹⁴ Furthermore, in retrospective reviews it has been estimated that 1% to 2% and, prospectively, as many as 12.5% of patients develop recurrent complications during this waiting period^23,29—complications that might either further delay definitive surgical therapy, prompt emergency interventions in unfavorable situations, or force alternative treatment strategies that might result in less than ideal outcomes, while potentially increasing the complexity, length, and cost of hospitalization.

Various approaches using currently available intravenous antithrombotic drugs, such as heparin and glycoprotein IIb/IIIa (Gp IIb/IIIa) inhibitors, have been proposed for bridging strategies.¹⁴ However, these are associated with important drawbacks. Anticoagulants do not reduce the incidence of stent thrombosis,³⁰ and heparin can enhance platelet reactivity.³¹ Glycoprotein IIb/IIIa inhibitors, in particular the small molecules tirofiban and eptifibatide, present some of the advantage of cangrelor including potency, rapid onset of action, and consistent platelet inhibitory effects.³² However, small molecule Gp IIb/IIIa inhibitors have a slower offset of action requiring 4 to 6 hours to return to baseline platelet function, which is achieved within 1 hour of receiving cangrelor.^13,32 In addition, these agents are used at dosing regimens recommended for ACS treatment and dose-finding studies targeting lower levels of platelet inhibition that would minimize bleeding complications, known to be increased with prolonged Gp IIb/IIIa inhibiting therapy,³³ particularly while bridging patients to surgery are lacking.

The overall favorable safety profile with prolonged infusion (up to 7 days) of cangrelor at a dose shown to be below thresholds associated with thrombotic risk but not excessive to minimize bleeding as identified in this trial is reassuring. Ultimately, cangrelor may represent a more natural bridging strategy because it selectively targets the P2Y₁₂ receptor. Although the drug is not yet commercially available, it has been extensively studied in large-scale trials of patients with coronary artery disease.^34- 35

This study had several limitations. First, this trial only enrolled patients undergoing open heart surgery and did not assess the strategy for common noncardiac surgical procedures, such as orthopedic or gastrointestinal tract operations. However, different noncardiac operations have variable bleeding risk, and their inclusion could have limited the ability to adequately assess the safety of cangrelor. Indeed, the selection of a homogenous surgical cohort with high-bleeding risk such as CABG surgery limited such confounding. Second, the present study was not powered to assess if more platelet blockade with cangrelor prior to CABG surgery would reduce the risk of ischemic events compared with placebo. Similarly, the trial was not powered to assess differences in bleeding event rates, although the similarity of the results is reassuring. Third, it may be argued that in stage 1 platelet inhibition was used to assess the antiplatelet efficacy of cangrelor, whereas stage 2 considered platelet reactivity while receiving treatment. However, this was due to an evolving understanding of platelet function testing in which studies have shown that levels of platelet reactivity while receiving treatment have better prognostic implications than percentage inhibition.³⁶ Whether similar outcomes are achieved with cangrelor in patients with prior exposure to the oral nonthienopyridine ticagrelor cannot be extrapolated from our analysis.

In this trial, cangrelor achieved and maintained target levels of platelet inhibition known to be associated with a low risk of thrombotic events compared with placebo, without a significant excess in bleeding complications. Our data support the hypothesis that intravenous cangrelor is a feasible management strategy in patients waiting for cardiac surgery who require prolonged platelet P2Y₁₂ inhibition after thienopyridine discontinuation.