Introduction
Clinical outcomes after acute coronary syndrome (ACS) have significantly improved over time. A significant proportion of the reduction in mortality has been the result of development and adoption of potent anticoagulant and antiplatelet therapies coupled with invasive risk stratification in high-risk patients (see Chapter 13 ). In addition, the use of secondary prevention strategies to control risk factors has further improved long-term outcomes (see Chapter 34 ). Although antithrombotic medications reduce the risk for recurrent ischemic events, they increase the risk for bleeding and need for blood transfusion. Paradoxically, bleeding events are associated with a subsequent higher risk for myocardial infarction (MI), stroke, stent thrombosis, and death. Randomized trials of “bleeding avoidance strategies” (BAS) in patients with ACS have demonstrated a reduction in adverse events, including mortality. Because of the availability of BAS, use of management strategies that reduce ischemia while minimizing bleeding have the potential to further improve outcomes. This chapter summarizes the epidemiology and management of bleeding events in ACS and MI.
Epidemiology of Bleeding
Reported Incidence of Bleeding: Influence of Definitions
The reported incidence of major bleeding events among patients with ACS varies greatly depending on the definition used, the fidelity with which the events are identified, the types of antithrombotic therapies used, and whether invasive risk stratification with revascularization is pursued. Before discussing the incidence of bleeding in ACS, it is important to review how bleeding is defined. Several schemas exist, but a few have been consistently used to identify bleeding events across randomized trials or registries that include ACS patients. Bleeding events as defined by several of these classification systems have also been correlated with other adverse events such as recurrent MI, stroke, stent thrombosis, and death (see section on Bleeding and Outcomes ).
Most bleeding definitions are made up of specific data elements that broadly fall into three categories: changes in hemoglobin or hematocrit; clinical events (e.g., gastrointestinal bleeding or intracranial hemorrhage); and consequences (e.g., blood transfusion or fatal bleeding). The extremes of bleeding definitions could focus solely on either changes in hemoglobin or on clinical events, and either could “cast the net” too widely—that is, identify events that would not be considered bleeding events by most clinicians or too narrowly by identifying only the most severe bleeding events and missing other less severe, but clinically important, events. Most definitions include combinations of all three categories. Table 29-1 displays commonly used bleeding definitions across randomized trials and registries.
Trial/Definition | Patient Population | Intervention | Bleeding Definition |
---|---|---|---|
TIMI | ACS, PCI | N/A | Major
|
Minor
| |||
GUSTO | ACS, PCI | N/A | Severe or life-threatening
|
SYNERGY | NSTE-ACS | Enoxaparin vs. heparin | TIMI and GUSTO |
PURSUIT | NSTE-ACS | Eptifibatide/heparin vs. heparin | TIMI and GUSTO |
CURE | NSTE-ACS | Aspirin vs. aspirin + clopidogrel | Major bleeding
|
GUSTO IIb | NSTE-ACS | Hirudin vs. heparin | GUSTO |
CURRENT-OASIS 7 | NSTE-ACS | High-dose aspirin vs. low-dose aspirin; high-dose clopidogrel vs. standard dose clopidogrel | Severe bleeding
|
OASIS-5 | NSTE-ACS | Fondaparinux vs. enoxaparin | Major bleeding
|
ACUITY and HORIZONS-MI | NSTE-ACS | Bivalirudin alone vs. heparin or enoxaparin + GP IIb/IIIa vs. bivalirudin + GP IIb/IIIa | Major bleeding
|
ACTION-GWTG | ACS | N/A |
|
CRUSADE | NSTE-ACS | N/A | Major bleeding
|
GRACE | NSTE-ACS |
| |
TRITON TIMI 38 | NSTE-ACS | TIMI | |
PLATO | NSTE-ACS | Major life-threatening
|
Established Definitions
Thrombolysis in Myocardial Infarction and Global Use of Strategies to Open Occluded Arteries Definitions
Historically, the most commonly used definitions were the Thrombolysis in Myocardial Infarction (TIMI) and Global Use of Strategies to Open Occluded arteries (GUSTO) bleeding scales. The TIMI definition was developed in the context of fibrinolytic therapy for ST-elevation MI (STEMI) and was generally based on decreases in hemoglobin. The categorization for TIMI bleeding originally consisted of minimal, minor, and major bleeding, defined according to the degree of hemoglobin change. Because intracranial hemorrhage is the most feared complication of fibrinolytic therapy, it was included as part of the TIMI major bleeding definition. Since its original development, the TIMI bleeding definition has evolved to include a qualifier that bases the required decrease in hemoglobin on whether a clinically overt bleeding event has occurred ( Table 29-1 ). The GUSTO definition was also developed in the context of fibrinolytic therapy for STEMI, and was based solely on clinical events. GUSTO bleeding is categorized as mild, moderate, or severe, and again, intracranial hemorrhage is considered severe according to the GUSTO definition. Other definitions listed in Table 29-1 combine elements of the TIMI and GUSTO definitions and add others (e.g., access site hematoma).
The existence of multiple bleeding definitions contributes to the variations in the reported rates of bleeding. When two or more definitions are used in the same study, the assessment of the safety of a management strategy can be confusing. One example of this is the SYNERGY trial, which compared enoxaparin with unfractionated heparin in 9978 ACS patients without persistent ST-segment elevation who were slated to undergo early invasive risk stratification. The primary efficacy endpoint was 30-day death or MI; the primary safety endpoint was bleeding defined according to both the TIMI and GUSTO scales. There was no significant difference in 30-day death or MI between the two strategies; however, the bleeding data showed disparate findings. There was no significant difference in GUSTO severe bleeding or transfusion rates (i.e., GUSTO moderate bleeding), but there was a significant excess of TIMI major bleeding among patients assigned to enoxaparin. These results challenged any firm conclusion about the relative safety of enoxaparin over unfractionated heparin and underscored the influence of definition on reported bleeding rates.
Capture of Bleeding Events
Another factor that may affect the incidence of bleeding is the methods for detecting events. Clinical trials often use independent clinical events committees that review source documents (medical charts) to detect adverse events. This minimizes the risk of bias at the site level. In contrast, registries generally do not use adjudication and rely on site-identified events. This is particularly true of registries in which the primary purpose is quality improvement rather than comparing the effectiveness of one treatment with another. As such, the rates of bleeding are often higher in clinical trials than in registries, despite the inclusion of higher risk patients in the latter. For example, the rate of postprocedural major bleeding among patients who underwent primary percutaneous coronary intervention (PCI) for STEMI assigned to unfractionated heparin plus glycoprotein IIb/IIIa inhibitors in the HORIZONS-MI trial was 8.4%, whereas the rate of bleeding among patients who underwent primary PCI in the National Cardiovascular Data Registry’s Cath-PCI registry was approximately 4% to 5%. Certainly, this difference is driven in part by the differences in definitions, but it is also likely caused by variations in the way the events were identified and reported.
Bleeding Academic Research Consortium Definition
To overcome some of the challenges posed by different bleeding definitions, several groups have attempted to develop a standardized approach to classifying and reporting bleeding events. Rather than developing a new definition, the Academic Research Consortium (ARC) proposed a list of standardized data elements that should be recorded in the case report form for a clinical trial of antithrombotic therapy (or the data collection form of a registry of ACS). These data elements can then either be reported separately or combined to reconstruct other bleeding definitions. The ARC, which originally consisted of members from four academic research organizations that designed and executed pivotal trials of drug-eluting stents, developed a set of standardized definitions for endpoints used in coronary stent trials. Periprocedural bleeding was not considered in their initial deliberations; however, subsequent iterations of ARC have developed definitions for bleeding endpoints for ACS trials, transcatheter valve trials, and peripheral arterial disease trials. The Bleeding Academic Research Consortium (BARC) definition of bleeding for use in ACS trials is shown in Table 29-2 .
|
Bleeding Incidence
After considering the differences in definitions across studies, the estimated incidence of bleeding during acute management of ACS is between 1% and 12%; any attempt to report a single number as the rate of ACS-related bleeding complications is likely to be misleading. Moreover, the incidence of bleeding in ACS patients may change with evolution of therapy over time and may be decreasing. One can use either clinical trials or registries to estimate the incidence of bleeding. In the SYNERGY trial, which included ACS patients who underwent an early invasive strategy, the rate of major or severe bleeding ranged between 2.7% (GUSTO severe) to 9.1% (TIMI major) among patients assigned to enoxaparin. Similarly, in the PLATO trial, the rates of TIMI major bleeding and protocol-defined PLATO major bleeding were 7.9% and 11.6%, respectively, among patients assigned to ticagrelor. Another source of information on bleeding complications is registry data. The ACTION–Get With the Guidelines registry is a nationally representative quality improvement registry for ACS that tracks and reports in-hospital bleeding complications (see Table 29-1 for definitions). The incidence of bleeding in this registry is 10.8%. Clinicians should be cognizant of the factors that influence the reporting of bleeding rates, and either use one definition to survey their own practice, or participate in an ACS registry like the ACTION-Get With the Guidelines registry to receive bleeding data on their patients benchmarked against a registry average.
Risk Factors for Bleeding Complications
The presence of numerous clinical trial and registry databases has created a deep knowledge base regarding risk factors and risk indicators associated with hemorrhagic complications in ACS. A major risk for bleeding is the use of invasive procedures, such as cardiac catheterization, percutaneous coronary intervention, and coronary artery bypass grafting (CABG), which compounds the risk for bleeding in patients treated with potent antithrombotic medications. The widespread adoption of these procedures has led to the broad categorization of bleeding into either “access-site related bleeding” and “nonaccess-site related bleeding.” This categorization also facilitates the grouping of BAS, which can be directed at the vascular access site or systemically (see the section on Bleeding Avoidance Strategies). The proportion of bleeding related to the vascular access site appears to vary with the patient’s clinical presentation. In the context of ACS, patients with non ST-elevation (NSTE)-ACS have a higher proportion of nonaccess-site bleeding. This observation is borne out in registry data and in randomized clinical trials. For example, in the National Cardiovascular Data Registry’s Cath-PCI registry, of the bleeding events that occurred in NSTE-ACS patients who underwent PCI, two-thirds of the bleeds were not related to the access site. Similarly, in the RIVAL trial, which compared radial with femoral access in 7021 patients with ACS (>70% of whom had NSTE-ACS), two-thirds of major bleeding was unrelated to the vascular access site. In contrast, patients with STEMI not only have a higher rate of bleeding, but a larger proportion of these patients (∼50%) have bleeding related to the vascular access site. These differences are likely related to the much higher use of PCI and its attendant intense antithrombotic therapy among patients with STEMI.
Risk Scores for Bleeding
In addition to clinical presentation, published risk models for bleeding have identified a variety of demographic and clinical features that are associated with a higher risk for bleeding complications. Because most bleeding events in ACS occur early, either during the index hospitalization or within 30 days, most of these models have used in-hospital or 30-day bleeding as the primary outcome. A commonly used bleeding prediction model is one derived from the ACUITY and HORIZONS-MI clinical trials, which examined a bivalirudin strategy in patients with NSTE-ACS or STEMI, respectively, who underwent primary PCI. Both trials used the same definition of bleeding, and the combined trials consisted of 17,421 patients. The overall rate of non-CABG–related major bleeding was 7.4%. Baseline covariates independently associated with bleeding included age, female sex, elevated serum creatinine, elevated white blood cell count, pre-existing anemia, NSTEMI, and STEMI. One treatment covariate—the use of heparin plus a glycoprotein IIb/IIIa inhibitor—was also a significant predictor. The model has reasonable discrimination, with a c-index of 0.74. The goodness-of-fit was evaluated by comparing predicted with observed bleeding. Finally, an integer risk score was developed for use at the bedside ( Figure 29-1 ). Another model was derived from the CRUSADE registry, which is used for risk adjustment in an ACS quality improvement registry. This model included only NSTE-ACS patients and did not include treatment variables, so that site-level bleeding outcomes could be benchmarked and allow for process changes to reduce bleeding risk at centers with high bleeding rates. Using 71,277 patients as the development data set and 17,857 patients as the validation data set, the investigators identified eight independent predictors of major bleeding (defined according to the registry; see Table 29-1 ). These risk indicators included baseline hematocrit, baseline creatinine clearance, baseline heart rate, female sex, heart failure at presentation, baseline systolic blood pressure, previous vascular disease, and diabetes mellitus. The c-indices in the development and validation cohorts were 0.72 and 0.71, respectively. Similar to the ACUITY bedside risk score, the investigators developed the “CRUSADE” bleeding risk score (available at http://www.crusadebleedingscore.org ), which showed an increase in the predicted probability as the score increased.