Author + information
- Stuart J. Connolly, MD∗ ()
- Population Health Research Institute, McMaster University, Hamilton Health Sciences, Hamilton, Ontario, Canada
- ↵∗Reprint requests and correspondence:
Dr. Stuart J. Connolly, Department of Medicine, Population Health Research Institute, Hamilton Health Sciences, 30 Birge Street, Hamilton, Ontario L8L 0A6, Canada.
Gómez-Outes et al. (1) have performed a meta-analysis of the large pivotal trials comparing the novel oral anticoagulants (NOAC) to warfarin. Their unique contribution to a deeper understanding of the results of these important studies is to specifically focus on the causes of death. Their paper provides insight into a number of important issues in the management of patients with atrial fibrillation at risk for stroke, including the high rates of both sudden/unwitnessed and other heart failure deaths. However, this editorial focuses on the main finding of the meta-analysis, which is a mortality reduction with the NOACs and its apparent mechanism: reduction in death due to bleeding. The pernicious effects of anticoagulant-related bleeding are often underappreciated by the medical community, and the Gómez-Outes et al. paper (1) nicely draws our attention to the beneficial effect of minimizing these bleeding events.
As was already reported previously (2), this meta-analysis indicates that, as a group, NOACs reduce all-cause mortality in patients with atrial fibrillation compared with warfarin. The treatment effect is modest but highly statistically significant, and with no important heterogeneity between studies (hazard ratio: 0.91; 95% confidence interval [CI]: 0.87 to 0.96; p < 0.001). On the basis of analysis of death by cause, the main driver of the observed mortality reduction is not a reduction in ischemic stroke. Ischemic stroke, which was not significantly reduced by NOACs, only represented 6% of all deaths. The types of death that are clearly reduced by the NOACs are those due to bleeding, which in this analysis are called “vascular bleeding deaths.” These include all intracranial bleeding deaths and other extracranial bleeding deaths. These are reduced by about 50% in patients treated with NOACs compared with those treated with warfarin. Thus, the top-line interpretation of the meta-analysis is that NOACs reduce mortality compared with warfarin, primarily through a reduction in fatal bleeding, especially intracranial. However, this simple interpretation runs aground when the absolute numbers of deaths prevented are analyzed. The reduction in bleeding deaths only explains about one-third of the mortality reduction because only about 6% of deaths were categorized as vascular bleeding deaths. Therefore, even if these are reduced by 50%, this reduction explains less than one-half of the observed 9% mortality reduction.
The clear reduction in bleeding with the NOACs compared with warfarin likely has effects beyond the immediately fatal bleeding events that are highlighted in this meta-analysis. Several analyses of the NOAC trials suggest that there are pernicious long-term consequences of anticoagulant-related major bleeding that are not captured by the immediate bleeding-related case fatality rate. A recent analysis from the RE-LY (Randomized Evaluation of Long Term Anticoagulant Therapy) and ACTIVE (Atrial Fibrillation Clopidogrel Trial With Irbesartan for Prevention of Vascular Events) trials (3) computed the relative risk of subsequent death after stroke, myocardial infarction, and major bleeding events occurring during the studies, comparing patients with the event to those without and adjusting for differences in baseline clinical characteristics. Not surprisingly, patients who had either ischemic stroke or myocardial infarction during the study had a 7- to 8-fold increased risk of death during subsequent follow-up. The risk of death after intracerebral hemorrhage was even higher, with a hazard ratio for death of 27. More surprisingly, the elevated risk of death after extracranial bleeding was 5-fold higher than for patients without such bleeding. This occurred despite the observation that these types of bleeds are seldom immediately fatal. Another recent analysis sheds further light on this topic. The ARISTOTLE (Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation) investigators (4) analyzed the sequelae of acute major bleeding and observed the adjusted 30-day rate of death for patients after intracranial bleeding was 121-fold higher compared with those without intracranial bleeding, whereas for non-intracranial bleeding, it was 12-fold higher. Part of the reason for this excess in mortality is likely related to the very high rates of discontinuation of anticoagulation following the bleeding, along with low rates of re-starting. In the ARISTOTLE analysis, by 30 days, 90% of patients remained off anticoagulants after intracranial bleeding and 50% remained off after gastrointestinal bleeding. This study also observed 12-fold higher rates of myocardial infarction or stroke in patients during the 30-day period after an acute major bleed, a finding that is consistent with the discontinuation of anticoagulant therapy in high-risk patients who are in a potentially prothrombotic state after an acute major bleeding episode. Although discontinuation of anticoagulation after bleeding is a highly plausible mechanism to explain the long-term negative impact on mortality, there are likely other factors involved. These could include adverse effects of transfusion, ongoing anemia, subclinical organ damage incurred during acute bleeding episodes, and a prothrombotic state induced by acute major bleeding.
The importance of minimizing anticoagulation-related bleeding is demonstrated by the intriguing results of the ENGAGE AF–TIMI 48 (Effective Anticoagulation With Factor Xa Next Generation in Atrial Fibrillation–Thrombolysis In Myocardial Infarction 48) trial (5), which compared 2 different dosing regimens of edoxaban, 60/30 mg and 30/15 mg, in comparison to warfarin in a randomized trial, including >7,000 high-risk patients in each of the 3 treatment arms. Both dosing strategies of edoxaban met the study-defined primary outcome of noninferiority to warfarin for stroke or systemic embolism. Of particular interest to the present discussion on benefits of minimizing anticoagulation-related major bleeding was the observation that, although the low-dose edoxaban regimen was less effective than warfarin against ischemic stroke (hazard ratio: 1.41; 95% CI: 1.19 to 1.67; p < 0.00), it clearly did reduce mortality compared with warfarin (hazard ratio: 0.87; 95% CI: 0.79 to 0.96; p = 0.006). This mortality reduction appears to be almost entirely due to a large (>50%) reductions in all types of bleeding, both intracranial and extracranial, and including gastrointestinal bleeding.
Recent trials of antithrombotic therapy in other clinical areas also suggest a long-term mortality hazard from major bleeding, or by corollary, an important benefit from minimization of bleeding. The widely cited WOEST (What is the Optimal antiplatElet & Anticoagulant Therapy in Patients With Oral Anticoagulation and Coronary StenTing) trial (6) compared single antiplatelet therapy to dual antiplatelet therapy in warfarin-treated patients undergoing percutaneous coronary intervention and reported a mortality reduction that is most likely due to the large reduction in bleeding of all types observed.
In conclusion, this meta-analysis assessing causes of death points to the important mortality benefit of the NOACs over warfarin, and partially explains this through quite large relative reductions in immediately fatal bleeding. The longer-term adverse consequences of major bleeding episodes have been quite clearly seen in other analyses of the NOAC trials and in other studies. The mechanisms by which these effects occur remain largely obscure and are an area of potentially rewarding research for the future. In the meantime, these findings remind us that we should pay particular attention to protecting our patients from bleeding when we use oral anticoagulants or other antithrombotic agents. The common idea that strokes and myocardial infarctions leave permanent damage, but patients recover without long-term consequences from major bleeding, is an intuition that is not supported by current data.
↵∗ Editorials published in the Journal of the American College of Cardiology reflect the views of the authors and do not necessarily represent the views of JACC or the American College of Cardiology.
Dr. Connolly has reported that he has received consulting and lecture fees, and his hospital and research institute has received funding for research, from Bristol-Myers Squibb, Pfizer, Bayer, Portola, Janssen, Boehringer Ingelheim, and Daiichi-Sankyo.
- American College of Cardiology Foundation
- Gómez-Outes A.,
- Lagunar-Ruíz J.,
- Terleira-Fernández A.-I.,
- Calvo-Rojas G.,
- Suárez-Gea M.L.,
- Vargas-Castrillón E.
- Eikelboom J.W.,
- Connolly S.J.,
- Hart R.G.,
- et al.
- Held C.,
- Hylek E.M.,
- Alexander J.H.,
- et al.