Author + information
- Received March 30, 2012
- Revision received July 2, 2012
- Accepted July 2, 2012
- Published online October 23, 2012.
- Christoph Varenhorst, MD, PhD⁎,⁎ (, )
- Ulrica Alström, MD, PhD†,
- Benjamin M. Scirica, MD, MPH‡,
- Charles W. Hogue, MD§,
- Nils Åsenblad, Msc, PhLic⁎,
- Robert F. Storey, MD∥,
- Ph. Gabriel Steg, MD¶,#,
- Jay Horrow, MD⁎⁎,
- Kenneth W. Mahaffey, MD††,
- Richard C. Becker, MD††,
- Stefan James, MD, PhD⁎,
- Christopher P. Cannon, MD‡,
- Gunnar Brandrup-Wognsen, MD, PhD‡‡,
- Lars Wallentin, MD, PhD⁎ and
- Claes Held, MD, PhD⁎
Reprint requests and correspondence:
Dr. Christoph Varenhorst, Uppsala Clinical Research Center/MTC, Dag Hammarskjölds väg 14B, 1tr, MTC, Science Park, 752 37 Uppsala, Sweden
Objectives This study investigated the differences in specific causes of post–coronary artery bypass graft surgery (CABG) deaths in the PLATO (Platelet Inhibition and Patient Outcomes) trial.
Background In the PLATO trial, patients assigned to ticagrelor compared with clopidogrel and who underwent CABG had significantly lower total and cardiovascular mortality.
Methods In the 1,261 patients with CABG performed within 7 days after stopping study drug, reviewers blinded to treatment assignment classified causes of death into subcategories of vascular and nonvascular, and specifically identified bleeding or infection events that either caused or subsequently contributed to death.
Results Numerically more vascular deaths occurred in the clopidogrel versus the ticagrelor group related to myocardial infarction (14 vs. 10), heart failure (9 vs. 6), arrhythmia or sudden death (9 vs. 3), and bleeding, including hemorrhagic stroke (7 vs. 2). Clopidogrel was also associated with an excess of nonvascular deaths related to infection (8 vs. 2). Among factors directly causing or contributing to death, bleeding and infections were more common in the clopidogrel group compared with the ticagrelor group (infections: 16 vs. 6, p < 0.05, and bleeding: 27 vs. 9, p < 0.01, for clopidogrel and ticagrelor, respectively).
Conclusions The mortality reduction with ticagrelor versus clopidogrel following CABG in the PLATO trial was associated with fewer deaths from cardiovascular, bleeding, and infection complications. (Platelet Inhibition and Patient Outcomes [PLATO]; NCT00391872)
The PLATO (Platelet Inhibition and Patient Outcomes) trial (NCT00391872) demonstrated that ticagrelor was superior to clopidogrel for the prevention of cardiovascular (CV) events, including myocardial infarction, CV death, and stent thrombosis, and numerically for total death in patients with acute coronary syndromes (ACS) (1). In the subset of patients who underwent coronary artery bypass graft surgery (CABG) after randomization and received study drug within 7 days of surgery (n = 1,261), ticagrelor was associated with a reduction in total and CV mortality (2). In this subpopulation, a 50% relative reduction in total and CV mortality was observed in patients randomized to ticagrelor versus clopidogrel.
The underlying direct causes of and contributing factors to mortality in the CABG-treated patients and the ability to explain differences in outcomes between the treatment arms have not been fully understood in the trial. There was no difference in baseline characteristics and timing of CABG between patients allocated to ticagrelor or clopidogrel (2). Several mechanisms linked to the mode of action of ticagrelor, including its reversibility, more rapid offset of platelet inhibitory effect, and attenuation of red blood cell adenosine re-uptake have been hypothesized (3). In support of a drug-related effect, the greatest mortality difference was observed when the last dose of study drug was taken 2 to 4 days before surgery (2). The ticagrelor U.S. label and the European Society of Cardiology non–ST-segment elevation ACS guidelines recommend stopping ticagrelor 5 days before CABG when possible.
The aim of this study was to investigate the causes of CABG-related death in the PLATO trial by further classifying, in a blinded fashion, all CABG-related deaths according to direct and indirectly related causes.
Patients and study design
In the PLATO trial, ticagrelor was compared with clopidogrel for the prevention of CV events in ACS. Patients were randomly assigned to receive ticagrelor or clopidogrel, administered in a double-blind, double-dummy fashion. The trial recruited 18,624 patients within 24 h of onset of ACS, with or without ST-segment elevation, whether intended for interventional or medical treatment. Randomized treatment continued for 6 to 12 months. The primary composite endpoint was myocardial infarction, stroke, and vascular death up to and including 12 months after enrollment. The overall study design, population, and outcomes have been published previously (1,4). Ten percent (n = 1,899) of the patients underwent CABG post-randomization. Of these, 1,261 patients underwent CABG within 7 days after the last intake of the study drug (Fig. 1). The protocol recommended ticagrelor or its matching placebo to be withheld for 24 to 72 h, and clopidogrel or its matching placebo for 5 days prior to surgery.
In the present ancillary study, all post-CABG deaths in patients undergoing surgery within 7 days of the last intake of the study drug (n = 87) were identified and further independently categorized by 2 reviewers blinded to treatment assignment. Disagreements were grouped into major and minor disagreements. All major disagreements were resolved in meetings as a consensus decision with at least 3 reviewers present of whom 2 were not the original reviewers. Minor disagreements were resolved by 2 reviewers.
The PLATO event adjudication committee originally classified each death as either vascular or nonvascular, and designated a time of death: these designations remained unchanged. Reviewers for this study further subclassified vascular deaths as sudden death; acute myocardial infarction; arrhythmia; heart failure/cardiogenic shock; cerebrovascular event (ischemic stroke, hemorrhagic stroke, ischemic stroke with hemorrhagic conversion, or intracranial hemorrhage); pulmonary embolism; peripheral arterial disease; hemorrhage (e.g., death in which a bleeding event directly led to death within 7 days, subcategorized into perioperative CABG-related bleeding, gastrointestinal bleeding, and other bleeding); or other vascular death. Nonvascular deaths were classified as primary infection/sepsis (surgical site infection, urinary tract infection, blood stream infection, respiratory tract infection, gastrointestinal tract infection, CV system infection, other); pulmonary causes; renal causes; gastrointestinal causes; multiple organ dysfunction syndrome; malignancy; accident/trauma; suicide; or other nonvascular death.
In addition to reviewing the primary cause of death, bleeding or infections that could have contributed to death, were identified. Bleeding contributing to death was defined as an event not considered as the primary cause of death, but a clinically overt bleeding event that was part of a causal chain of events leading to death within 30 days of the bleed. The bleeding events related to death were subclassified into intra/post-operative CABG-related bleeding; intracranial hemorrhage; gastrointestinal bleeding; or other bleeding. If infection contributed to, but was not considered as the primary cause of death, it was subclassified into surgical site infection; urinary tract infection; bloodstream infection; respiratory tract infection; gastrointestinal system infection; CV system infection; and other infection.
Reviewers determined and agreed to the list of all classification and subclassification categories before commencing their blinded review of the clinical information in the PLATO database.
The analyses comprised all patients undergoing CABG with last intake of the study drug within 7 days before surgery, with further analyses of the subgroup of patients with post-CABG death. Baseline characteristics and procedures in hospital or at discharge were summarized for all post-CABG deaths and the total CABG group using frequencies for categorical variables and medians with interquartile intervals for continuous variables. Subcategorized vascular and nonvascular deaths were described by treatment group using frequency tables. The Fisher exact test was used to compare group frequencies. The time to death due to infection (primary cause of death or contributing to death) and death due to bleeding (primary cause or contributing) were analyzed with Cox proportional hazards model with the treatment group as an independent variable. Cumulative incidence curves were obtained using the Kaplan-Meier technique. The proportional hazard assumptions were checked by log-cumulative hazard plots (data not shown). Hazard ratios (HRs) and 95% confidence intervals (CIs) were calculated using Cox models. A p value <0.05 was regarded as statistically significant. All analyses were done with SAS software (version 9.2, SAS Institute, Cary, North Carolina).
Baseline characteristics and treatments and evaluations of the patients undergoing CABG within 7 days of study drug cessation (n = 1,261) and of the subset of patients who died post-CABG (n = 87) are shown in Tables 1 and 2,⇓ and Online Table 1.
Vascular and nonvascular deaths subcategorized
There were significantly more vascular deaths in the clopidogrel group compared with the ticagrelor group. For clopidogrel, there were numerically more vascular deaths attributed to myocardial infarction, heart failure, sudden death/arrhythmia, and hemorrhagic stroke/bleeding compared with ticagrelor, but none of these differences were statistically significant (Table 3). For nonvascular deaths, there were numerically, but not statistically significant, more infections as the primary cause of death in clopidogrel-treated patients compared with ticagrelor-treated patients (Table 3). The treatment effect on total and CV mortality was consistent in most subgroups in the 1,261-patient population with only a significant interaction for sex (Online Tables 2 and 3).
For 27 patients randomized to clopidogrel and 9 randomized to ticagrelor (p < 0.01), bleeding contributed to death or was directly related to death (Fig. 2A,Table 4). Similarly, infections being the direct or contributing cause of death were more common in patients randomized to clopidogrel than to ticagrelor (16 vs. 6, p < 0.05) (Fig. 2B, Table 4).
Time to death after CABG
Overall, bleeding as the primary or contributing cause of death occurred more frequently within the first month after CABG (Fig. 3A). The estimated 12-month rates of bleeding directly causing or contributing to death were 4.6% for clopidogrel versus 1.4% for ticagrelor, corresponding to a HR of 0.33 (95% CI: 0.15 to 0.70).
At 12 months post-CABG, the cumulative incidence of combination of deaths contributing to or being primarily caused by infection was 2.9% for patients randomized to clopidogrel and 1% for ticagrelor (HR: 0.36; 95 % CI: 0.14 to 0.93) (Fig. 2B). The deaths related to infections were evenly distributed over time during the first-month, 1- to 3-month, and more than 3-month time intervals among the ticagrelor-treated patients (Fig. 3B).
The main finding of this study is that the observed lower mortality in post-CABG patients assigned to ticagrelor compared with clopidogrel was explained, not only by lower rates of CV deaths, but also by a reduction in bleeding and infection events leading to death. In this study, bleeding as the direct cause or as a contributing factor to death was one-third as frequent in patients treated with ticagrelor compared with clopidogrel (HR: 0.33; 95% CI: 0.15 to 0.70). These findings are seemingly in contrast to those of the initial analyses of the PLATO CABG-population, where no significant differences in bleeding overall between the 2 treatment arms were observed, despite using several different bleeding definitions (2). However, this study focuses, in addition to the previous analyses, on bleeding events that were contributing to death and those in which bleeding was the primary cause of death. Ticagrelor was associated with a lower bleeding rate compared with clopidogrel. The more rapid offset of the platelet inhibitory effect after ticagrelor discontinuation may be one reasonable explanation for this association and a contributing factor explaining the difference. This interpretation fits with the observation that the greatest difference in mortality was found when surgery was performed 2 to 4 days after the last intake of the study drug. A longer time gap between drug discontinuation and surgery might require bridging with short-acting antithrombotic agents such as low-molecular-weight heparins or glycoprotein IIb/IIIa inhibitors.
Clopidogrel is an irreversible platelet inhibitor, and normal platelet activity is not regained until 5 to 10 days after discontinuation. Cessation of clopidogrel treatment is therefore recommended at least 5 days before CABG surgery in order to limit the risk of bleeding (5,6). Ticagrelor, on the other hand, is a direct-acting, reversibly binding P2Y12 inhibitor with a faster offset of platelet inhibition after drug discontinuation compared with clopidogrel (7,8). Moderate-to-poor responders to clopidogrel have a similar time to recovery of platelet function compared with ticagrelor-treated patients after cessation of treatment (9). Thus, approximately a third of the patients with high levels of platelet inhibition on clopidogrel need 7 or more days to recover near-normal platelet reactivity, explaining why some clopidogrel-treated patients may be particularly vulnerable to serious hemorrhagic complications to CABG surgery, especially if performed early.
The ticagrelor U.S. label and the European Society of Cardiology guidelines on non–ST-segment elevation ACS recommend stopping ticagrelor 5 days before CABG when possible. The results of the PLATO CABG substudy (2) and the present analyses suggest that the interval might be shortened to 2 to 3 days. However, further studies are required to define the optimal time point of stopping ticagrelor before CABG surgery.
The PLATO trial enrolled a broad selection of patients with ACS within 24 h after symptom onset and before any coronary angiography was performed. Therefore, the study included a relatively large cohort of patients who needed urgent CABG and thereby being exposed to clopidogrel or ticagrelor within 7 days of CABG. Numerous observational studies have documented an increased bleeding rate when clopidogrel is used within 5 days before surgery (10–13), and therefore, withholding clopidogrel at least 5 days before surgery is endorsed by current guidelines (5,6,14). The risk–benefit of continued dual antiplatelet treatment until the time of CABG likely varies based on the clinical setting and the urgency of the surgery. A meta-analysis by Nijjer et al. (15) suggests that ACS patients requiring urgent CABG should undergo this surgery without a delay for a clopidogrel-free period. On the other hand, this proposal does not take into consideration the variable levels of platelet inhibition in patients taking clopidogrel and the potentially greater bleeding risk in patients with adequate levels of inhibition. In the PLATO trial, the protocol recommended ticagrelor/placebo to be withheld for 24 to 72 h before operation, compared with 5 days for clopidogrel/placebo. Furthermore, half of the patients randomized to ticagrelor received open-label clopidogrel pre-randomization. Therefore, in the real-life setting, the risk for ischemic and bleeding events when using ticagrelor instead of clopidogrel could be lower before as well as after the CABG procedure.
For patients with ACS needing CABG surgery, these results suggest a preference for ticagrelor over clopidogrel to reduce ischemic events, long-term complications of bleeding, and most importantly, mortality. Currently, also, prasugrel is approved as an alternative to clopidogrel in ACS patients treated with PCI procedures (16). Prasugrel, like clopidogrel, is a thienopyridine and an irreversible platelet inhibitor. Therefore, pre-treatment levels of platelet activity are not achieved until 7 to 10 days after cessation of treatment (17), similar to the recovery time for high responders to clopidogrel (9). In the TRITON–TIMI 38 (Trial to Assess Improvement in Therapeutic Outcomes by Optimizing Platelet Inhibition With Prasugrel–Thrombolysis In Myocardial Infarction 38) study, CABG-related bleeding events were 4-fold higher in patients treated with prasugrel compared with clopidogrel (CABG-related TIMI major bleeding 13.4% vs. 3.2% for prasugrel and clopidogrel, respectively), although this did not translate into higher mortality in the prasugrel group (18).
Infection-related CABG deaths were evenly distributed over the first 3 months after surgery. It is well known that infections contribute to an adverse outcome after CABG surgery (19). Platelets are pivotal contributors to the immune response in infection, and an up-regulated platelet activation in septic patients leads to organ hypoperfusion, pathological tissue injury, and poor patient outcomes (20). Despite growing evidence to suggest a pathophysiological role for platelets in inflammatory conditions, there is no, or very little, evidence to support a targeted antiplatelet treatment in afflicted patients. In contrast to pre-clinical results, suggesting a potential benefit of antiplatelet treatment in infection (21), a retrospective study has suggested an increased risk of infection in clopidogrel-treated patients undergoing CABG compared with those treated with aspirin alone (22). The mechanisms by which clopidogrel may increase the risk of infection-related mortality are not known, although clopidogrel treatment has been associated with aplastic anemia, thrombocytopenia, and neutropenia (23). The higher rate of infections with clopidogrel may also be directly related to more bleeding, as patients who bleed are more likely to have prolonged drainage tube placements and length of hospitalization. The impact of antiplatelet agents, such as clopidogrel and ticagrelor, on circulating platelets, markers of inflammation, and clinical outcomes during systemic infection, needs to be further clarified.
The analysis was a retrospective investigation of a post-randomization–identified cohort with post hoc–blinded classification of specific causes of death by subcategorizing the vascular and nonvascular deaths identified in the main study. The results should therefore be regarded as exploratory and hypothesis generating. However, we sought to limit bias by a process in which reviewers identified and agreed on the different categories prior to examining case records and performed all reviews blinded to study treatment.
In ACS patients receiving dual antiplatelet therapy within 7 days of CABG, ticagrelor, compared with clopidogrel, was associated with fewer deaths related to cardiovascular, bleeding, and infection events. The cause-of-death pattern was similar between treatment groups. Further studies are needed to investigate the underlying mechanisms responsible for the better survival after CABG in patients with ACS treated with ticagrelor.
For supplementary tables, please see the online version of this article.
This study was supported by AstraZeneca. Dr. Varenhorst has received a research grant from AstraZeneca to perform this research; and is a member of the Speakers' Bureaus for AstraZeneca, Eli Lilly & Company, and The Medicines Company. Dr. Scirica has received research grants from Merck, AstraZeneca, Johnson & Johnson, Bayer, Daiichi Sankyo, Bristol-Myers Squibb, Gilead Sciences, and Novartis; and is a consultant to Gilead Sciences, Arena, and Lexicon. Dr. Hogue has received research support from Covidien; and is a consultant to Covidien, Merck, and Ornim. Dr. Storey is a consultant to AstraZeneca, Accumetrics, Merck, Novartis, Eli Lilly & Company, Daiichi Sankyo, Eisai, Iroko, The Medicines Company, Sanofi Aventis, Bristol-Myers Squibb, and Medscape. Dr. Steg is a consultant to AstraZeneca, Amarin, Amgen, Astellas, Bayer, Boehringer-Ingelheim, Bristol-Myers Squibb, Eli Lilly & Company, Daiichi Sankyo, Eisai, GlaxoSmithKline, Medtronic, Merck, Novartis, Otsuka, Pfizer, Roche, The Medicines Company, Sanofi Aventis, and Servier; is a member of the Speakers' Bureaus for AstraZeneca, Bayer, Bristol-Myers Squibb, Pfizer, The Medicines Company, Sanofi Aventis, and Servier; and has received Data Safety Monitoring Board support from Ablynx. Dr. Horrow is the executive director of AstraZeneca LP, and has equity ownership in the company. Dr. Mahaffey has received grant support from AstraZeneca, Bayer, Boehringer Ingelheim, Bristol-Myers Squibb, Daiichi Sankyo, Eli Lilly & Company, GlaxoSmithKline, Johnson & Johnson, Merck, Novartis, Pozen, Regado Biotechnologies, Sanofi Aventis, Schering Plough, and The Medicines Company; and is a consultant to AstraZeneca, Bayer, Boehringer Ingelheim, Bristol-Myers Squibb, Daiichi Sankyo, Eli Lilly & Company, GlaxoSmithKline, Johnson & Johnson, Merck, Pfizer, Polymedix, and Sanofi Aventis. Dr. Becker is a consultant to and receives support from AstraZeneca, Bayer, Bristol-Myers Squibb, Daiichi Sankyo, Johnson & Johnson, The Medicines Company, Merck, and Regado. Dr. James has received institutional grants from AstraZeneca, Eli Lilly & Company, and MSD; and honoraria from AstraZeneca, Eli Lilly & Company, MSD, Merck, and The Medicines Company. Dr. Cannon has received grant support from Accumetrics, AstraZeneca, Essentialis, GlaxoSmithKline, Merck, Regeneron, Sanofi Aventis, and Takeda; is a member of the Advisory Boards for Alnylam, Bristol-Myers Squibb, and Pfizer; has received funds from these companies, which have been donated to charity; and is a clinical advisor to Automated Medical Systems with equity in the company. Dr. Wallentin is a consultant to Merck/Schering Plough, Regado Biosciences, Evolva, Portola, C.S.L. Behring, Athera Biotechnologies, Boehringer-Ingelheim, AstraZeneca, GlaxoSmithKline, and Brisol-Myers Squibb/Pfizer; has received research grants from AstraZeneca, Merck/Schering-Plough, Boehringer-Ingelheim, Bristol-Myers Squibb/Pfizer, and GlaxoSmithKline; and lecture fees from AstraZeneca, Boehringer-Ingelheim, Bristol-Myers Squibb/Pfizer, GlaxoSmithKline, and Schering-Plough/Merck. Dr. Held is on the Advisory Boards for AstraZeneca, Roche, and Pfizer; and has received institutional grants from Schering-Plough/Merck, GlaxoSmithKline, Bristol-Myers Squibb, and Roche. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- acute coronary syndromes
- coronary artery bypass graft surgery
- confidence interval
- hazard ratio
- Thrombolysis In Myocardial Infarction
- Received March 30, 2012.
- Revision received July 2, 2012.
- Accepted July 2, 2012.
- American College of Cardiology Foundation
- Held C.,
- Asenblad N.,
- Bassand J.P.,
- et al.
- Schneider D.J.
- James S.,
- Akerblom A.,
- Cannon C.P.,
- et al.
- Eagle K.A.,
- Guyton R.A.,
- Davidoff R.,
- et al.
- Dunning J.,
- Versteegh M.,
- Fabbri A.,
- et al.
- Husted S.,
- Emanuelsson H.,
- Heptinstall S.,
- Sandset P.M.,
- Wickens M.,
- Peters G.
- Gurbel P.A.,
- Bliden K.P.,
- Butler K.,
- et al.
- Hongo R.H.,
- Ley J.,
- Dick S.E.,
- Yee R.R.
- Berger J.S.,
- Frye C.B.,
- Harshaw Q.,
- Edwards F.H.,
- Steinhubl S.R.,
- Becker R.C.
- Nijjer S.S.,
- Watson G.,
- Athanasiou T.,
- Malik I.S.
- Kushner F.G.,
- Hand M.,
- Smith S.C. Jr..,
- et al.