There has been a significant evolution in the management of patients with acute coronary syndromes (ACS), largely driven by advances in pharmacological therapy. Increased implementation of evidence-based antithrombotics and antiplatelet agents has led to improved patient outcomes.1,2 Additionally, management algorithms favouring early invasive intervention in high-risk patients are gaining support.3,4 Despite improvements in endpoints such as death, myocardial infarction (MI) and recurrent ischaemia, these therapies are not without their risks, namely bleeding and blood transfusion. Recent studies suggest that bleeding is independently associated with an increased risk of both short- and longer-term adverse events (including MI, stroke and death) among patients with ACS.5-9 Furthermore, the safety of blood transfusion in this population has also been called into question,10-12 suggesting that management strategies that maintain an adequate anticoagulant effect to reduce ischaemia, while at the same time minimising the risk of bleeding, may further improve ACS outcomes.
Aetiology and Risk Factors for Bleeding in ACS
Rates of major bleeding in clinical studies of ACS have ranged from 0.8% to 11.5%5,6,13,14 and, according to some registries, it is the most common non-cardiac complication of therapy in this patient population.5,13 Clinical risk factors for in-hospital bleeding have been identified in analyses of several contemporary ACS trials and registry databases. Older age, female sex, lower body weight and chronic kidney disease have been consistently associated with an increased risk for bleeding complications. In addition, the use of anticoagulants, fibrinolytics and invasive procedures (e.g., cardiac catheterisation with or without intervention, intra-aortic counterpulsation) were also predictors of in-hospital bleeding. The most common sites of bleeding were vascular access sites and the gastrointestinal tract, followed by less common retroperitoneal bleeds and genito-urinary haemorrhage.5-7
There are also data to suggest that inappropriate dosing of anticoagulant medications is associated with bleeding in patients with ACS. In the Can Rapid Risk Stratification of Unstable Angina Patients Suppress Adverse Outcomes with Early Implementation of the ACC/AHA Guidelines (CRUSADE) registry, for instance, over 42% of all patients with non-ST-segment elevation (NSTE) ACS received at least one antithrombotic drug in excess.13
Not only were many of these medications overdosed, but a large number of these dosing errors occurred in vulnerable patients with risk factors for bleeding. In particular, dose adjustment of renally excreted anticoagulants, such as enoxaparin and eptifibatide, in patients with chronic kidney disease is critically important to reduce the risk of over-dosage and subsequent bleeding.15
Defining Clinical Bleeding Events
One of the more difficult barriers to overcome in clinical studies addressing bleeding in ACS has been the development of a consensus classification for bleeding severity. For example, the Global Use of Strategies to Open Occluded Coronary Arteries16 (GUSTO) investigators classified bleeding, using clinical variables, as either mild, moderate or severe/life-threatening. GUSTO severe or life-threatening bleeding is defined as either intracranial haemorrhage or bleeding resulting in haemodynamic compromise necessitating intervention. GUSTO moderate bleeding is defined as bleeding requiring transfusion, but not resulting in haemodynamic compromise. All other bleeding events are classified as GUSTO mild.
Alternatively, the Thrombolysis in Myocardial Infarction (TIMI)17 investigators utilised a more laboratory-based classification scheme when defining bleeding events. TIMI major bleeding is defined as intracranial haemorrhage or bleeding with a haemoglobin decrease of >5 g/dl or haematocrit decrease of >15%. The definition of TIMI minor bleeding is dependent on whether or not there is an identifiable source of blood loss. If a bleeding site is found, then TIMI minor bleeding is defined as a haemoglobin decrease of >3 g/dl or a haematocrit decrease of >10%; if no site is found, then it is defined as a haemoglobin decrease of >4 g/dl or haematocrit decrease of >12%. Finally, TIMI minimal bleeding is defined as any clinically overt sign of haemorrhage that is associated with a haemoglobin decrease of <3 g/dl or a haematocrit decrease of <9%.
While both of these definitions were developed in the context of thrombolytic therapy for ST-segment elevation MI (STEMI), they have been used extensively to classify bleeding events in trials and registries of non-ST-segment elevation (NSTE) ACS. However, no standard bleeding definition exists and other clinical trials in ACS have utilised a combination of the TIMI and GUSTO classifications or their own unique definitions.18.19Table 1 displays several of the commonly employed bleeding classification schemes used in clinical trials and registries of ACS.
Bleeding and Prognosis in ACS
Rather than merely a nuisance of therapy, bleeding in the setting of ACS has consistently been shown to be associated with adverse outcomes. Although the exact mechanisms underlying this risk association are unclear, it is likely that several factors contribute to the increased mortality seen in these studies. First, bleeding complications likely lead to cessation or reversal of anticoagulation that may precipitate rebound ischaemia and/or recurrent infarction. Second, several human and animal models of bleeding and haemorrhagic shock have shown upregulation of inflammatory cytokines, vasoactive substances and modulators of the neurohormonal system, all of which help to maintain vital organ perfusion, but all of which have also been linked to adverse cardiovascular outcomes.20-22 Third, chronic bleeding may lead to anaemia, which is independently associated with increased mortality in patients with coronary artery disease (see below). Finally, administration of blood transfusion may also serve to increase the risk of ischaemic events and subsequent mortality (see below).
While the pathophysiology linking bleeding with adverse cardiovascular events in ACS has been poorly defined, the association between bleeding and prognosis has been demonstrated in several clinical studies. Major bleeding in the large, multicentre, prospective Global Registry of Acute Coronary Events (GRACE)5 registry was found to occur in 3.9% of all patients with ACS. In-hospital mortality rates were significantly higher in patients with major bleeding than in those without major bleeding (18.6% versus 5.1%, p<0.001), regardless of clinical presentation, and major bleeding was independently associated with an increased risk of hospital death by multivariate analysis.
A subsequent analysis of bleeding events from the Canadian Acute Coronary Registries identified almost 6,000 patients with NSTE ACS who developed in-hospital major bleeding.6 These patients not only had significantly higher rates of in-hospital mortality, MI and stroke, but one-year outcomes of death and MI were also higher in these individuals. On multivariate analysis, major bleeding proved to be a powerful predictor of in-hospital and one-year mortality, MI and death or MI.
Using pooled data from four contemporary ACS trials, Rao et al.7 found that 27.6% of all study participants with available bleeding data had at least one bleeding event during their index hospitalisation. These investigators divided bleeding into mild, moderate and severe categories and found that survival decreased significantly as bleeding severity increased. Additionally, after adjusting for potential confounders, there appeared to be a stepwise increase in the hazard for 30-day mortality, 30-day mortality or MI, and six-month mortality based upon bleeding severity. Interestingly, even patients who developed 'mild' bleeding were found to have significant increases in the adjusted hazard for death and death or MI, suggesting any degree of bleeding in the setting of ACS may have negative prognostic implications.
More recently, Eikelboom et al.9 reviewed bleeding events from the Organization to Assess Ischemic Syndromes (OASIS) Registry,23-25 OASIS-226 and the Clopidogrel in Unstable Angina (CURE)18 trial. In an analysis of 34,126 patients with ACS, they found that major bleeding was associated with a five-fold increase in the risk of death. Similar to previous studies,7 there appeared to be a dose relationship between the severity of bleeding and death, with increasing bleeding severity associated with an increasing risk of death. Furthermore, major bleeding was associated with an increased risk of recurrent ischaemic events, including MI and stroke, even after adjustment for baseline characteristics and propensity for major bleed.
In another study by Rao and colleagues,8 the relative prognostic value of the GUSTO and TIMI bleeding classifications was assessed in 15,858 patients with NSTE ACS. These investigators found that both bleeding scales identified patients at increased risk for adverse outcomes. However, after adjusting for blood transfusion, the stepwise increase in risk with worsening GUSTO bleeding persisted, while the risk with worsening TIMI bleeding was no longer statistically significant. When bleeding definitions were included in the same risk model, only GUSTO bleeding was associated with worse prognosis. Furthermore, when exploring interaction terms between these two scales, it was found that TIMI bleeding did not affect risk among patients who already met criteria for GUSTO bleeding; the converse, however, was not true, suggesting that perhaps a clinically-based assessment of bleeding may be more important for prognosis than a laboratory-guided approach.
Pharmacological Strategies to Reduce Bleeding
Given the prognostic implications of bleeding suggested by these previous studies, more recent clinical trials of pharmacologic agents for the treatment of ACS have made a more concerted effort to both evaluate and target bleeding complications. OASIS-5 randomised over 20,000 NSTE ACS patients to fondaparinux (2.5 mg/day), a direct thrombin inhibitor, or enoxaparin (1.0 mg/kg twice daily).27 The primary end-point of death, MI or refractory ischaemia at day nine was similar between the two study groups, meeting the pre-specified criteria for non-inferiority; however, the rates of major bleed, minor bleed and any bleed were significantly lower in the fondaparinux-treated patients. The finding of increased guiding catheter thrombosis in those receiving fondaparinux in both OASIS-5 and more recently in OASIS-6,28 a study of fondaparinux in patients with STEMI, has raised concern for the routine use of this novel agent.
The evaluation of bivalirudin, another direct thrombin inhibitor thought to have lower bleeding potential, has been studied in patients undergoing percutaneous coronary intervention (PCI). The Randomized Evaluation in PCI Linking Angiomax to Reduced Clinical Events (REPLACE)-2 trial was a randomised, double-blind, active-controlled trial conducted among over 6,000 patients undergoing urgent or elective PCI.18 Patients in this study were randomly assigned to receive intravenous bivalirudin, with provisional glycoprotein IIb/IIIa inhibition, or heparin with planned glycoprotein IIb/IIIa inhibition. The bivalirudin arm of this study was found to be statistically 'not inferior' to the heparin plus planned glycoprotein IIb/IIIa blockade strategy with regard to suppression of acute ischaemic end-points, and was indeed associated with fewer bleeding events.
A retrospective review of the REPLACE-2 data, looking specifically at ACS patients, found a similar rate of death, MI or urgent revascularisation at 30 days compared with a stable cohort of patients, but a higher rate of revascularisation at six months.29 Additionally, a trend for lower major bleeding in the bivalirudin-treated patients was also observed. Results from a more definitive, prospective study of bivalirudin in moderate- to high-risk ACS patients should help to define the role for this pharmacologic agent in future clinical practice.30
Anaemia as a Cardiovascular Risk Factor
In and of itself, anaemia is a relatively common finding in patients with ACS, and may be a marker for chronic illness or the result of bleeding complications from anticoagulant therapy, invasive procedures or both. Already determined to be an independent risk factor for adverse cardiovascular outcomes in community cohorts31 and in patients undergoing PCI,32 a recent review of nearly 40,000 ACS clinical trial participants found a similar, highly significant and independent association between low haemoglobin concentration and a variety of adverse cardiovascular events.33 Individuals with anaemia were more likely to be older, female, and to have worse renal function compared with ACS patients without anaemia. Additionally, those patients with low haemoglobin concentrations were less likely to undergo early revascularisation. The pathophysiology underlying the association between anaemia and adverse cardiovascular outcomes is largely speculative, but likely involves mismatch between myocardial oxygen supply and demand. It is also plausible that changes in physician practice, such as utilising a less invasive management strategy or withholding evidence-based therapeutics that have greater bleeding potential, may also contribute to this clinical association.
Transfusion Therapy in Patients with ACS
Because anaemia and bleeding events are often seen in patients with ACS, blood transfusion therapy is commonly employed. Clinical studies suggest that approximately 10-15% of patients underwent transfusion during their index hospitalisation.4,10 Transfusion therapy was more common in older patients, women, those with a lower body mass index (BMI) and those with multiple medical co-morbidities, with renal insufficiency being most predictive.4,10 In the critical care literature, appropriate use of transfusion therapy remains controversial.34 More recent analyses of transfusion practice in patients with cardiovascular disease will likely fuel this controversy.
Studies of clinical outcomes in cardiac patients receiving blood transfusions have shown disparate results. A trial comparing a restrictive transfusion strategy with a more liberal strategy of transfusion to maintain a haemoglobin level of 10 mg/dl in critically ill patients found no clinical benefit to the more aggressive application of transfusion therapy.35 A post hoc analysis of this randomised trial, limited to patients with cardiovascular disease, further supported those findings.9 On the other hand, a community-based study of elderly patients with acute MI found that blood transfusion had a beneficial effect on survival in individuals with a baseline haematocrit level of 30% or less.36
In ACS trials, however, transfusion therapy seems to be associated with worse clinical outcomes. Pooled data from Platelet IIb/IIIa Antagonism for the Reduction of Acute Coronary Syndrome Events in a Global Organization Network (PARAGON) B, GUSTO IIb, and Platelet Glycoprotein IIb/IIIa in Unstable Angina: Receptor Suppression Using Integrilin Therapy (PURSUIT) showed a higher hazard ratio for death and the composite of death and MI with transfusion therapy in ACS patients. Furthermore, analysis of this data looking specifically at the association between transfusion and nadir haematocrit suggested that a haematocrit level as low as 25% may be tolerated without transfusion therapy in an otherwise stable patient with ischaemic heart disease.10
Recently corroborating these findings, an adjusted analysis of the CRUSADE data revealed that patients who received blood transfusion had longer hospital stays, along with higher absolute rates of death and death or MI.4
Conclusions
Despite advances in invasive and non-invasive therapeutics, which improve outcomes in ACS, bleeding and transfusion remain potential risks. Studies have consistently shown that advanced age, female gender, lower body mass and renal insufficiency are associated with an increased risk for both. Though the exact mechanism of harm has not been clearly delineated, bleeding and blood transfusion are associated with worse clinical outcomes among patients with ACS. Future studies in this patient population will need to address appropriate dosing strategies for pharmacologic agents to reduce the risk of bleeding complications, while still maintaining an effective anticoagulant effect. Additionally, randomised trials of transfusion strategies are needed to determine the appropriate use of blood transfusion in patients with ischaemic heart disease.