Author + information
- Robert J. Applegate, MD⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Robert J. Applegate, Section of Cardiology, Wake Forest School of Medicine, Medical Center Boulevard, Winston-Salem, North Carolina 27157-1045
The landmark observation by DeWood et al. (1) that acute myocardial infarctions result from in situ coronary thrombosis ushered in a new era of therapy for patients with acute coronary syndromes. Since that time, a variety of pharmacologic treatments have been developed to inhibit the coagulation cascade, targeting almost all cellular and protein constituents of the thrombotic process. Not surprisingly, strategies aimed at thrombolysis, clot dissolution, and prevention of its recurrence within the coronary tree may have similar effects in other vascular beds. The unintended consequences of these antithrombotic therapies at noncoronary artery sites may result in significant bleeding that in some cases, such as an intracranial hemorrhage, may be fatal. Although bleeding as a “side effect” of these therapies was recognized early on, attempts to minimize bleeding was tempered by the recognition that antithrombotic therapies prevented significant recurrent ischemia and were often lifesaving. Specific examples include the use of glycoprotein IIb/IIIa inhibitors in patients with acute coronary syndromes and percutaneous coronary intervention (PCI) (initially balloon angioplasty) (2) and the early intensive anticoagulant strategies to prevent acute stent thrombosis in the first generation of bare-metal stents (3). Thus, during the early days of management of acute coronary syndromes, the benefits of thrombotic therapy, particularly during PCI, outweighed the risk of bleeding.
Nonetheless, over the past 2 decades, multiple strategies have evolved to minimize the extent and severity of bleeding accompanying the use of antithrombotic therapies. The use of fibrin-specific instead of nonfibrin-specific thrombolytics for ST-segment elevation myocardial infarction (4) and the recognition of the importance of the use of lower intensity anticoagulation with heparin in conjunction with intravenous glycoprotein IIb/IIIa inhibitors for the treatment of acute coronary syndromes (5) are just 2 examples of efforts to lower rates of bleeding while maintaining the efficacy of the antithrombotic therapy. During PCI, the use of lower intensity heparin dosing in conjunction with glycoprotein IIb/IIIa inhibitors has reduced rates of periprocedural bleeding. Innovations in PCI technology, specifically the development and refinement of stent therapy, have reduced the incidence of acute stent thrombosis and the accompanying need for intense periprocedural acute antithrombotic therapies. Additionally, the introduction of bivalirudin has provided an alternative to heparin as the anticoagulant of choice during PCI with associated lower rates of bleeding. Finally, a better understanding of optimal stent use has allowed the use of dual antiplatelet therapy instead of warfarin for the prevention of stent thrombosis. Thus, the need for intensive anticoagulation therapy at the time of PCI has decreased. Because of all these factors and as a result of more predictable PCI outcomes with very low rates of periprocedural ischemic events, the bleeding risk associated with PCI has come under close scrutiny.
There is mounting clinical evidence that periprocedural bleeding is an independent predictor of post-procedure adverse outcomes, including death. The adverse consequences of major bleeding are most readily apparent during the hospital course of these patients. Moreover, the reason for the adverse nature of periprocedural bleeding is easily understood given that the intensity of the antithrombotic therapy occurs at this time, when there has been a transient breach of the venous and/or arterial systems. Additionally, however, it has been increasingly recognized that the adverse consequences of periprocedural bleeding extends beyond the hospital, for at least as long as 6 months. Why the impact of periprocedural bleeding should persist beyond the hospitalization and even 30 days is not readily apparent because the effectiveness of most of the antithrombotic therapies would have long worn off.
The mechanisms postulated to be responsible for the association of major bleeding and adverse outcomes can best be considered by looking at the periprocedural period in 2 distinct phases. During the early phase, the time period surrounding the index hospitalization for PCI and as long as 30 days thereafter, 4 mechanisms have been proposed to explain the adverse consequences of periprocedural bleeding. Frank exsanguination (e.g., retroperitoneal hemorrhage) is a well-recognized cause of bleeding-related mortality. Also in this time period, the withdrawal of important antithrombotic therapies can plausibly affect and lead to ischemic consequences such as stent thrombosis, which is often associated with significant morbidity and mortality. Treatment of major bleeding frequently includes administration of transfusions, and it has been suggested that transfusions themselves are independently associated with adverse outcomes. This could be a particular problem if transfusions are administered by protocol (i.e., to maintain hemoglobin >10 g/dl). Finally, the stimulation of prothrombotic forces such as platelet aggregation in response to bleeding has been put forth as potential explanation for an increase in adverse (ischemic) events in those with major bleeding compared with those without major bleeding.
The second phase that needs to be considered is the time period beginning 30 days after the index procedure. During this time period, the mechanism(s) responsible for an adverse impact of bleeding experienced at an index hospitalization are not as clearly understood. There are several questions that bear on this discussion that remain to be clearly addressed: 1) Is the adverse association of periprocedural bleeding and late morbidity and mortality similar for the access site compared with nonaccess site of bleeding? 2) Can we demonstrate a decrease in medication use in this later period for those who have adverse morbidity and mortality and periprocedural bleeding? 3) Does periprocedural bleeding simply identify a subgroup of patients who are at higher risk of post-procedure adverse events? 4) Finally, how long does periprocedural bleeding affect long-term adverse outcomes?
There are published data that address some but not all of these questions. Rao et al. (6) analyzed data from 4 randomized clinical trials evaluating the clinical efficacy of glycoprotein IIb/IIIa inhibitors in patients with acute coronary syndromes, including GUSTO IIB (Global Utilization of Streptokinase and Tissue-Plasminogen Activator for Occluded Coronary Arteries), PURSUIT (Platelet Glycoprotein IIb/IIIa in Unstable Angina, Receptor Suppression Using Integrilin Therapy), and PARAGON A and B (Platelet IIb/IIIa Antagonism for the Reduction of Acute coronary syndrome events in a Global Utilization Network). In this post-hoc evaluation, they found substantially worse in-hospital and 30-day outcomes in those with any bleeding, but the relationship was most notable in those with severe bleeding compared with those who did not have any evidence of bleeding. Importantly, the adverse impact of bleeding on the odds ratio of both 30-day and 6-month mortality occurred with both procedure-related bleeds, as well as nonprocedure-related bleeds, suggesting that the access site itself was not solely responsible for the impact of bleeding on mortality. Moreover, these observations were adjusted for multiple covariates that were associated with adverse 30-day and 6-month outcomes themselves, and bleeding remained an independent predictor of mortality. Eikelboom et al. (7) examined information from the OASIS (Organization to Assess Ischemic Syndromes and the Clopidogrel in Unstable Angina to Prevent Recurrent Ischemic Events) Registry, OASIS II, and the CURE (Clopidogrel in Unstable angina to prevent recurrent Events) trials evaluating the clinical efficacy of thienopyridines in patients with acute coronary syndromes. They observed a similar association between the severity of bleeding and mortality after 30 days in these trials as did Rao et al. (6). In this particular evaluation, major and life-threatening but not minor bleeding was associated with worse outcomes at 6 months. These observations persisted despite adjustment for known baseline predictors of worse outcomes over these time periods. Multiple other studies have subsequently confirmed these initial important observations and have stimulated continued interest in better understanding the relationship between initial major bleeding and adverse long-term outcomes.
In this issue of the Journal, Suh et al. (8) provide important new information on the long-term consequences of in-hospital major bleeding from the HORIZONS-AMI (Harmonizing Outcomes With Revascularization and Stents in Acute Myocardial Infarction) trial. In this randomized clinical trial of primary PCI for ST-segment elevation myocardial infarction, there was a 2 × 2 randomization comparing bivalirudin with heparin plus glycoprotein inhibitor use as well as bare-metal with paclitaxel-eluting stent use. The overall incidence of major bleeding in the trial during the index hospitalization was 6.5%. Similar to previous reports, patients who experienced major bleeding were more often older, female, and diabetic. This post-hoc analysis provides 2 important pieces of information not previously available in an evaluation of the long-term consequences of periprocedural bleeding: landmark analysis of mortality from index hospitalization to 3 years and medication use from index hospitalization to 3 years. Not unexpectedly, the hazard ratio for mortality in those with periprocedural bleeding was the highest in the first 30 days. However, the adverse hazard ratio persisted during each of the 2 ensuing landmark intervals, with only slight attenuation compared with the earliest period: from 1 to 3 years, the hazard ratio of death was 3.67 (range, 2.0 to 6.52) in those patients with major in-hospital bleeding compared with those without this complication. This same adverse association of periprocedural bleeding and clinical outcomes was observed for all major adverse cardiac events, including myocardial infarction, ischemia-driven target vascular revascularization, stroke, and stent thrombosis (although not statistically significant). Interestingly, and perhaps counterintuitively, aspirin use at 3 years was similar in both groups at 95%, and thienopyridine use was similar at all time points except at 3 years when there was a paradoxically higher use of thienopyridines in those who had major bleeding compared with those who did not have major bleeding. Also, of note, beta-blocker and statin use was lower in those with major bleeding compared with those who did not have major bleeding at discharge, although there are no data reflecting use at 3 years.
What can we learn from this post-hoc analysis of the HORIZONS-AMI trial from Suh et al. (8)? First, these data extend the association between in-hospital major bleeding and mortality to 3 years. Second, this association appears to be independent of use of medications such as aspirin and thienopyridines, although it is unclear what the role of the decreased administration of beta-blocker and statin therapy would have in these events at 3 years. These observations provide important answers to some of the questions about the mechanisms and duration of the adverse association of bleeding and mortality. However, if the adverse consequences of in-hospital major bleeding are not related to the withdrawal of lifesaving medications (i.e., dual antiplatelet therapy), then what is the mechanism responsible for increased late mortality in these patients?
The association between major in-hospital bleeding and increased mortality in their analysis persisted after adjustment for baseline covariates. Despite this, it is concerning that in this study, as well as multiple other studies on the subject, major bleeding occurred more often in patients with baseline characteristics known to be associated with worse outcomes over this time period (e.g., older age, diabetic, female). Is it simply that bleeding is more likely to occur in those who are at risk of late adverse cardiac events despite our best efforts to attempt to adjust for these confounders? Suh et al. (8) acknowledge that their observations should stimulate further investigation of the long-term consequences of periprocedural bleeding. Until we have a clearer understanding of the mechanisms of this association, we will need to rely on bleeding avoidance strategies such as gentler antithrombotic pharmacology and access techniques that limit bleeding, to prevent the potentially morbid cascade that seems to follow (9).
Dr. Applegate is on the advisory board for Abbott Vascular; received research grants from Abbott Vascular, St. Jude Medical, and Terumo Corporation; and is a consultant for Abbott Vascular and St. Jude Medical.
↵⁎ 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.
- American College of Cardiology Foundation
- Eikelboom J.W.,
- Mehta S.R.,
- Anand S.S.,
- Xie C.,
- Fox K.A.,
- Yusuf S.
- Suh J.-W.,
- Mehran R.,
- Claessen B.E.,
- et al.
- Dauerman H.L.,
- Rao S.V.,
- Resnic F.S.,
- Applegate R.J.