Author Affiliations: Health Outcomes and Clinical Epidemiology Section, Department of Quantitative Health Sciences, Cleveland Clinic, Cleveland, Ohio.
Nonplatelet mechanisms may have a role in these events; therefore, other drugs have been tested. Warfarin sodium has been used among patients with ACS in combination with aspirin vs aspirin alone, showing a significant decrease in recurrent events but also a significant increase in major bleeding events with the combined treatment.2
In contrast to warfarin, novel oral anticoagulants (NOACs), such as direct thrombin inhibitors and anti-Xa, do not require monitoring and have minimal food and drug interactions. Moreover, compared with warfarin NOACs have demonstrated significant reductions in thrombotic events and lower rates of major bleeding events in patients having atrial fibrillation and in patients undergoing knee and hip surgery.3 In recent years, several phase 2 and phase 3 trials of NOACs for the treatment of patients with ACS have been published, with most of the patients receiving dual antiplatelet treatment. Although recent detailed review articles have described the pros and cons of the use of NOACs in ACS,4,5 the available evidence has not been quantitatively summarized.
In this issue of the Archives, Komócsi et al6 publish meta-analysis results of 7 heterogeneous trials in which the use of NOACs was evaluated among patients with ACS. Of note, 5 were dose-finding trials, in which the primary outcome was safety, and 2 were large phase 3 trials that included most of the patients analyzed in the meta-analysis. Direct thrombin inhibitors were administered within 14 days of the ACS, and anti-Xa was given within 7 days of the ACS. Patients in 6 of 7 trials received both aspirin and a thienopyridine. Exclusion criteria and the median follow-up times were also heterogeneous among the trials. The 5 dose-finding trials evaluated several NOAC dosages. The composite efficacy outcomes were heterogeneous across the trials and included stroke, overall mortality, recurrent ischemia, cardiovascular mortality, and myocardial infarction (MI).
The authors analyzed efficacy and safety outcomes in the meta-analysis, including a net clinical benefit outcome that combined a composite of major ischemic events (MI, death, ischemic stroke, or severe recurrent ischemia) and thrombolysis in myocardial infarction (TIMI)–defined major bleeding. Komócsi et al found a significant 3-fold increase in the risk for major bleeding associated with the use of NOACs, a significant 2-fold increase in the risk for any bleeding, slightly decreased (though significant) risks for composite ischemic outcomes and for definitive and probable stent thrombosis, and a nonsignificant decreased risk for overall mortality. This was translated into a net clinical benefit that was not significantly different compared with control group findings. Notably, the use of NOACs was associated with significant lower risk for MI. For most of the outcomes, there was no evidence or a low degree of statistical heterogeneity of effects. Sensitivity analyses using fixed-effects models and excluding a trial of ximelagatran provided conclusions similar to those of the main analysis. Subgroup analyses by trial phase, premature termination, type of NOAC, and proportion of use of dual antiplatelet therapy gave similar conclusions as well. Finally, the evaluation of different drug dosages showed no effect modification. Therefore, the conclusions of the meta-analysis seem to be robust.
Several issues deserve further discussion. Phase 2 dose-finding trials are specifically designed to evaluate the safety of drugs, and detail and effort to evaluate secondary ischemic outcomes may be lower than those for primary outcomes. Therefore, potential risk exists for nondifferential misclassification of secondary ischemic outcomes, a type of information bias that will direct the odds ratio toward the null (ie, to an odds ratio of 1). This bias may have diluted the beneficial effects of NOACs in the study by Komócsi et al.6 However, the overall results for beneficial and harmful effects in the meta-analysis are mostly driven by data from the 2 phase 3 trials (Apixaban for the Prevention of Acute Ischemic and Safety Events–2 [APPRAISE-2] and Anti-Xa Therapy to Lower Cardiovascular Events in Addition to Standard Therapy in Subjects With Acute Coronary Syndrome–Thrombolysis in Myocardial Infarction 51 [ATLAS ACS–TIMI 51]), in which the ascertainment of mortality, ischemic events, and safety outcomes was appropriate. These 2 trials gave contradictory effects for efficacy outcomes. The APPRAISE-2 trial included older and sicker populations than the ATLAS ACS–TIMI 51 trial, and higher rates of the primary efficacy outcomes were seen in the former trial. Therefore, more of these events in the APPRAISE-2 trial may have corresponded to a different pathogenesis that was less related to thrombotic events and less responsive to NOACs.5 Also, the APPRAISE-2 trial was terminated prematurely, without a definitive conclusion on efficacy.
In the meta-analysis by Komócsi et al,6 the composite of ischemic events included death instead of cardiovascular death. This may seem counterintuitive because different causes of death do not follow the same pathogenesis of ischemic events. However, there are other causes of cardiovascular death that are not ischemic. Probably, the best term to describe the composite outcome of benefit would have been composite efficacy outcome, to resemble the nomenclature provided by the individual trials. Indeed, the Oral Ximelagatran for Secondary Prophylaxis After Myocardial Infarction (ESTEEM), Study Evaluating Safety, Tolerability and Efficacy of YM150 in Subjects With Acute Coronary Syndromes (RUBY-1), and Rivaroxaban in Combination With Aspirin Alone or With Aspirin and a Thienopyridine in Patients With Acute Coronary Syndromes–Thrombolysis in Myocardial Infarction 46 (ATLAS ACS–TIMI 46) trials reported overall death as part of the composite efficacy outcomes. The use of composite outcomes has several advantages and disadvantages compared with the use of individual outcomes.7 In particular, the effect of the NOACs on the composite efficacy outcome may be masked by the nonsignificant effect on overall death. The effect on the composite efficacy outcome would have been more significant without considering death. However, death is such a strong, hard outcome that its inclusion in evaluating efficacy outweighs its weak to moderate relationship with ischemia.
Absolute risk measures are important to place relative effects into context. Komócsi et al6 provided only relative measures of effects in their meta-analysis. A 3-fold increase in the relative risk for TIMI major bleeding seems a huge increase. However, the absolute risk increase is only 0.9% (95% CI, 0.7%-1.1%) (1.3% for NOACs vs 0.4% for control). Conversely, the significant risk reduction in the composite ischemic outcome by 14.0% is translated into an absolute risk decrease of −1.3% (95% CI, −1.9% to −0.7%) (6.5% for NOACs vs 7.8% for control), and the significant risk reduction in MI is translated into an absolute risk decrease of −0.8% (95% CI, −1.2% to −0.3%) (3.7% for NOACs vs 4.5% for control). The absolute risk difference for the net clinical benefit, as shown in Figure 4 of the meta-analysis, was −0.5% (95% CI, 1.0%-0.0%) (7.8% for NOACs and 8.3% for control), and the nonsignificant decreased risk for overall death is translated into an absolute risk difference of −0.5% (95% CI, −0.9% to −0.1%) (2.5% for NOACs vs 3.0% for control). The benefit is largely canceled by the harm; therefore, the routine use of NOACs among patients with ACS is unwarranted.
Would NOACs be useful for specific populations of patients with ACS? The meta-analysis by Komócsi et al6 explored subgroups by percentage of use of dual antiplatelet therapy and showed no significant differences across subpopulations. Trials included in the meta-analysis were not performed among patients with specific types of ACS; therefore, it is unknown whether the effects of NOACs differ among patients with unstable angina, ST-elevation MI, and non–ST-elevation MI. Also, no data are available to date on the use of NOACs in patients with ACS taking prasugrel or ticagrelor, undergoing percutaneous coronary interventions, or having indications for anticoagulation (eg, cancer, mitral stenosis, mechanical prosthetic valves, or prior stroke without atrial fibrillation). Trials are needed to evaluate the use of NOACs in these specific populations with ACS.
Correspondence: Dr Hernandez, Health Outcomes and Clinical Epidemiology Section, Department of Quantitative Health Sciences, Cleveland Clinic, 9500 Euclid Ave, Mailbox JJN3-01, Cleveland, OH 44195 (email@example.com).
Published Online: September 24, 2012. doi:10.1001/2013.jamainternmed.293
Conflict of Interest Disclosures: None reported.
Hernandez AV. No Place for Novel Oral Anticoagulants in Current Treatment of Acute Coronary SyndromesComment on “Use of New-Generation Oral Anticoagulant Agents in Patients Receiving Antiplatelet Therapy After an Acute Coronary Syndrome”. Arch Intern Med. 2012;172(20):1546-1547. doi:10.1001/2013.jamainternmed.293