A recent article published in the NEJM by Steg et al demonstrates that it is not the efficacy of an intervention that determines the success or failure of a trial, but rather the definition of its endpoints (1). In contrast to the majority of trials examining novel anticoagulants for ACS, Steg et al chose to focus on the bleeding risk caused by their drug, bivalirudin, when compared to heparin or enoxaparin. Simultaneously downplaying the importance of recurrent MI, normally a primary outcome in these types of trials, here was demoted to a secondary consideration. This was not done out of an overpowering ethical concern for transparency, but rather based off of the knowledge that highlighting bleeding risk as the primary outcome measure would lead to the most positive study. In other words they stacked their deck.
This trial, entitled “The European Ambulance Acute Coronary Syndrome Angiography Trial”, given the unfortunate nickname EUROMAX, compares heparin or enoxaparin to bivalirudin when started in prehospital patients undergoing a ST-elevation MI. In this randomized open label trial they found that patients given bivalirudin experienced less bleeding, but slightly more recurrent MIs and early stent restenosis. Neither the decreased bleeding or increased rate of infarction was serious enough to affect mortality. Since the initial trials performed on bivalirudin there has been a consistent decrease in the incidence of bleeding with a concurrent increase in myocardial infarction and stent restenosis (2,3,4). The authors were obviously aware of this. So much so that halfway through the trial they modified their primary endpoint from death, major bleeding and myocardial infarction to include just death and major bleeding. They justify this methodologic faux pas by stating their enrollment was slower than anticipated and they feared their study was no longer powered to detect a difference in their initial endpoint. In other words, in a smaller-than-expected cohort the authors worried the decrease in irrelevant bleeding would not statistically overpower the increase in irrelevant myocardial infarctions, leading to a statistically negative trial. So what initially appeared to be an intriguingly genuine comparison of anticoagulants for ACS, ends up being yet another attempt of a drug company sponsored trial to manipulate the various surrogate endpoints in their favor. This places us, as physicians, in a clinical dilemma, which do we prefer more bleeding or more recurrent MIs? It is, of course, never this simple. The most important question is how clinically relevant is the degree of bleeding the authors claim their drug prevents?
Since we began examining ACS interventions to inhibit clot formation and promote clot degredation, the need for a standardized method of evaluating bleeding severity has been paramount. Obviously the simplest way is to examine if the bleeding caused by such intervention increases patient’s mortality, thereby eliminating any benefits its anticoagulant properties may have provided. Some argue that this is not good enough. More subtle tools are needed to identify bleeding that may have some clinical impact but would go unnoticed in trials too underpowered to detect any change in mortality. Thus clinical scores were invented. These scores were intended to categorize bleeding into levels of severity. In ACS research the most well-known are the TIMI and GUSTO bleeding criteria( Table 1). Both of these scores were invented by their respective study groups (TIMI and GUSTO) during the early thrombolytic ACS trials. It is important to note that both these criteria were not so much derived as they were assembled. The specific categories of each criteria were based on the investigator's determination of major vs. minor bleeding. No high quality prospective studies were conducted initially to determine what effects these categories had on clinical outcomes, but they seemed reasonable and were for the most part accepted. The TIMI criteria are far more dependent on changes in lab values whereas the GUSTO consists purely of clinical characteristics (5). In 2005 Rao et al published a study in JACC examining both the TIMI and GUSTO criteria and how well they predicted mortality (6). Using the PERSUIT and PARAGON cohorts the authors found the GUSTO criteria to be far more predictive of 30-day mortality than the TIMI bleeding criteria. The changes in lab values may or may not necessarily have patient oriented manifestations, but clinical presentation is far more telling in terms of severity of bleeds.
Since the invention of the TIMI and GUSTO models, various hybrid criteria have been used by different investigators to highlight or conceal the rate of bleeding caused by the intervention in question. Recently, in hopes of creating a more uniform tool, a proposition for a standardized definition of bleeding in cardiovascular trials was published in Circulation(5). This definition was assembled by a group which called themselves the Bleeding Academic Research Consortium, and thus the BARC criteria was born…
The BARC criteria(see Table 2) consists of 6 categories of bleeding. Type 0 defined as no bleeding and Type 5 defined as fatal bleeding. Types 1-4 range between these two extremes with incremental increases in severity. It seems the BARC criteria is constructed from the components of both the TIMI and GUSTO systems. Interestingly the more severe bleeding defined by this criteria (3b and above) is primarily made up of criteria similar to GUSTO and the less severe types (3a and below) are more TIMI-like in nature. The rule was prospectively validated in a study published in October 2013 in Heart. In this trial the investigators enrolled over 1,800 STEMI patients and examined how the BARC, TIMI, and GUSTO criteria predicted 30-day mortality (7). Both GUSTO and BARC outperformed the TIMI criteria in predicting 30-day outcomes. Not surprising, the categories in each of these criteria that predicted clinically significantly bleeding (GUSTO major bleeding and BARC type 3b and above) were operationally identical.
The authors of the EUROMAX trial chose to use their own definition of bleeding built to accentuate the increased risk of bleeding caused by heparin and enoxaparin. Their definition of major bleeding is far more inclusive and incorporates what would be considered minor bleeding using any of the previous 3 criteria (1). That being said, the absolute difference in major bleeding between the bivalirudin and control group was 3.4%. If you examine the same cohort using the TIMI criteria, the difference in major bleeding is only 0.8%. Using the GUSTO criteria, the difference in life threatening bleeding drops to 0.3%. It is obvious that there is more bleeding in the control group when compared to those given bivalirudin, but given that there is no clinical difference in TIMI or GUSTO major bleeding, the clinical significance of this excess bleeding is questionable.
This is not to say that bivalirudin is an inferior treatment to heparin or enoxaparin in patients with ST-elevation. Without knowledge of the clinical relevance of the excess reinfarctions and stent restenosis it is difficult to assess. There was in fact another intervention previously compared to heparin and LMWHs with a very similar risk profile to bivalirudin. This intervention, like bivalirudin, in clinical trials showed a decreased risk of bleeding and slight increase in the rate of reinfarction and restenosis. This miracle drug was called placebo. In the initial heparin and LMWH trials comparing its use in ACS to placebo no mortality benefit was found, with a small decrease in reinfarction at 7 days (which was gone by 30 days) and an approximately 3% increase in the risk of bleeding (8,9,10,11). It is obviously difficult to compare groups between trials, but given the initial data on heparin’s performance in ACS it begs the questions whether bivalirudin is anything more than an expensive placebo?
It is not truly important which bleeding criteria we choose or if we choose one at all. What is important is that we demand an increasing degree of transparency in future trials examining novel medications for ACS. Other than aspirin and thrombolytics, we have yet to find a medication that has a consistent mortality benefit. Every other intervention examined has hovered around statistical equivalency. It is these types of trials that are most easily manipulated to show statistical significance while providing no clinical benefit. In such cases it is important to ask just how pertinent these benefits are to our patients. Before we go adding more treatments to our already bloated ACS algorithm with must demand evidence of true clinical relevance.
1. Steg et al. Bivalirudin started during emergency transport for primary PCI. N Engl J Med 2013
2. Stone GW, Witzenbichler B, Guagliumi G, et al. Bivalirudin during primary PCI in acute myocardial infarction. N Engl J Med 2008;358:2218-2230
3. Lincoff, et al. Bivalirudin and provisional glycoprotein IIb/IIIa blockade compared with heparin and planned glycoprotein IIb/IIIa blockade during percutaneous coronary intervention: REPLACE-2 randomized trial. JAMA: the Journal of the American Medical Association 289 (7): 853–863
4. Stone GW, McLaurin BT, Cox DA, et al. Bivalirudin for patients with acute coronary syndromes. N Engl J Med 2006
5. Mehran et al. Standardized Bleeding Definitions for Cardiovascular Clinical Trials: A Consensus Report From the Bleeding Academic Research Consortiums. Circulation. 2011;123:2736-2747
6. Rao et al. A Comparison of the Clinical Impact of Bleeding Measured by Two Different Classifications Among Patients With Acute Coronary Syndromes. JACC 2006 Feb 21;47(4):809-16
9. Matic et al. Prognostic implications of bleeding measured by Bleeding Academic Redearch Consortium categories in patients undergoing PCIs. Heart 2013-304564
10. The RISC Group. Risk of myocardial infarction and death during treatment with low dose aspirin and intravenous heparin in men with unstable coronary artery disease. Lancet. 1990 Oct 6;336(8719):827-30.
11. FRISC study group. Low-molecular-weight heparin during instability in coronary artery disease, Fragmin during Instability in Coronary Artery Disease. Lancet. 1996 Mar 2;347(9001):561-8.
12. Magee K et al. Heparin versus placebo for acute coronary syndromes. Cochrane Database Syst Rev. 2008 Apr 16;(2)
13. Newman et al. http://www.thennt.com/blog/2011/04/not-so-fast-heparin/