Author + information
- aCardiovascular Research Foundation, New York, New York
- bDepartment of Medicine, Division of Cardiology, Columbia University, New York, New York
- ↵∗Reprint requests and correspondence:
Dr. Ori Ben-Yehuda, Cardiovascular Research Foundation, 111 East 59th Street, New York, New York 10022.
Percutaneous coronary intervention and antithrombotic drugs reduce the risk of ischemic events and improve prognosis for patients with acute coronary syndromes (1,2). However, this benefit comes at the expense of increased bleeding (2). Both in-hospital and post-discharge bleeding are associated with increased risk of death (3,4). In fact, observational studies have suggested that bleeding complications are as dangerous as a recurrent myocardial infarction (5). Bleeding is therefore a central safety outcome in cardiovascular clinical trials.
Assessing bleeding is not straightforward, however, and involves multiple factors: absolute amount of blood loss, rate of blood loss, hemodilution effects of fluids, effect of transfusion, as well as varying hemodynamic and ischemic effects of the bleeding episode itself, which depend to a large extent on the patient’s underlying comorbidities and baseline hemoglobin levels. An elderly patient with residual coronary disease, for example, may have a markedly different response to an acute drop in hemoglobin compared with a young patient without significant residual coronary disease. How bleeding is assessed and treated may also vary across geographies and individual practitioners.
Unfortunately, bleeding definitions in pivotal contemporary multicenter clinical trials have not been consistent (2,6,7). Because these definitions differ considerably, it is difficult to compare results across trials and studies. It is also possible to reach different conclusions regarding the bleeding risk or the prognostic impact of bleeding within a trial depending on which bleeding definition is used. Examples include analyses from the SYNERGY (Superior Yield of the New Strategy of Enoxaparin, Revascularization and Glycoprotein IIb/IIIa Inhibitors) and CURE (Clopidogrel in Unstable Angina to Prevent Recurrent Events) trials in which significant between-group differences in bleeding risk were observed using one bleeding score, but not using others (8,9).
Consistent use of a meaningful bleeding classification in cardiovascular trials would allow more reliable comparisons of different antithrombotic drugs across trials. It would also facilitate detection and verification of factors that differentially affect a patient’s bleeding risk and risk of ischemic complications and/or that differentially affect the prognostic impact of bleeding and ischemia. Ultimately, patients could benefit from tailored antithrombotic therapy, which optimizes their individual risk/benefit ratio. An optimal bleeding score would not only be applicable to large groups as in a clinical trial but would also be applicable to an individual patient and help define transfusion thresholds.
In an attempt to create a contemporary, post-fibrinolytic era, standardized definition of bleeding, the Bleeding Academic Research Consortium (BARC), a group composed of representatives of academic research organizations, the U.S. Food and Drug Administration and industry, and other thought leaders in cardiovascular disease, convened and proposed a new hierarchically graded classification in 2011 (7). In contrast to the traditional Thrombolysis In Myocardial Infarction (TIMI) bleeding criteria, which are based largely on laboratory tests and the Global Use of Strategies to Open Occluded Arteries (GUSTO) bleeding criteria, which are based mainly on the clinical consequences of the bleed, BARC criteria incorporate both clinical and laboratory elements. By considering the laboratory elements that constitute the criteria for TIMI bleeding within their clinical context, the BARC definition seeks to better reflect the severity of the bleed. Another difference is that BARC uses an ordinal numeric system rather than qualitative terms such as “major” or “minor” to grade the severity of a bleed (7).
A major inherent limitation of the BARC bleeding score is that unlike other non-bleeding risk scores such as the Framingham Risk Score, it was not derived from an analysis of any specific database to identify independent risk factors, but was rather built on consensus, modification, and extension of other bleeding scores such as the TIMI bleeding score. The TIMI thresholds of drop in hemoglobin (3 to <5 g/dl and ≥5 g/dl) were thus adopted to define BARC 3a and 3b, respectively. These designations are by definition somewhat arbitrary—is the effect of a 2.9 g/dl drop in hemoglobin any different from a 3 g/dl drop? Is it appropriate to have a score that is categorical for a continuous variable? Moreover, given its empiric derivation, the BARC score requires validation, a point highlighted by the authors of the score themselves in the original publication (7).
In this issue of the Journal, Vranckx et al. (10) report that BARC scores independently predicted long-term mortality in the randomized multicenter TRACER (Thrombin Receptor Antagonist for Clinical Event Reduction in Acute Coronary Syndrome) trial. The TRACER trial enrolled 12,944 patients, of whom 1,975 had at least 1 BARC grade 2 or greater bleed and 652 died. A sufficient number of deaths were distributed to each BARC category to allow the authors to construct robust statistical models. The authors further report that the mortality risk increased gradually with increasing BARC grade and that the association between bleeding and mortality appeared to be independent of whether the patients bled in-hospital or after discharge. These findings are consistent with previous reports that both in-hospital and post-discharge bleeds are independently associated with mortality (3,4), and provide the largest single trial assessment of the BARC bleeding score (Table 1).
The study also compared the BARC criteria to the TIMI and GUSTO criteria. BARC ≥3 bleeding captured considerably more bleeds than either TIMI minor/major or GUSTO moderate/severe. When added to a multivariate model for predicting all-cause death that included baseline characteristics, BARC≥3 improved the model at least as much as any combination of TIMI or GUSTO scores. A bleeding score that captures more events while retaining a similar association with mortality risk may be advantageous in clinical practice because it identifies more patients at risk. It is also of benefit when planning a clinical trial because the statistical power of trial is determined by the number of events (e.g., bleeds) that occur.
In their analysis, Vranckx et al. (10) demonstrate that increasing BARC scores up to 3c (BARC 4 is a separate category, CABG-related bleeding, and does not imply worse bleeding per se) do indeed translate to increased mortality risk. But it should also be recognized that the BARC score is less useful as a clinical tool in patient care than as a clinical trial tool. Many of its elements are determined, not only by the bleed itself (e.g., degree of drop in hemoglobin), but by the response to the bleed (e.g., act of blood transfusion or medical evaluation). The score does not guide the clinician as to whether to initiate an evaluation or to transfuse blood. In addition, the score does not take into account the patient’s underlying hemoglobin, previous bleeding episodes and their effect, or other comorbidities.
Vranckx et al. (10) should be congratulated for undertaking a complex and thorough analysis that provides additional compelling data on the utility of the BARC bleeding scale. As the scale is prospectively included in future acute coronary syndromes trials, further validation can be anticipated, as well as validation of BARC modifications such as the VARC (Valve Academic Research Consortium) bleeding score (11).
↵∗ 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.
Both authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- 2016 American College of Cardiology Foundation
- Suh J.W.,
- Mehran R.,
- Claessen B.E.,
- et al.
- Genereux P.,
- Giustino G.,
- Witzenbichler B.,
- et al.
- Mehran R.,
- Pocock S.J.,
- Stone G.W.,
- et al.
- Mehran R.,
- Rao S.V.,
- Bhatt D.L.,
- et al.
- Ferguson J.J.,
- Califf R.M.,
- Antman E.M.,
- et al.
- Vranckx P.,
- White H.D.,
- Huang Z.,
- et al.
- Ndrepepa G.,
- Schuster T.,
- Hadamitzky M.,
- et al.
- Kikkert W.J.,
- van Geloven N.,
- van der Laan M.H.,
- et al.