Author + information
- ↵⁎Reprint requests and correspondence:
Dr. Harold L. Dauerman, Cardiac Unit, McClure 1, Fletcher Allen Health Care, 111 Colchester Avenue, Burlington, Vermont 05401.
In this issue of the Journal, Ndrepepa et al. (1) present a meta-analysis of 4 ISAR (Intracoronary Stenting and Antithrombotic Regimen) percutaneous coronary intervention (PCI) trials suggesting that bleeding complications and myocardial infarction (MI) are equivalent predictors of 1-year mortality after PCI. Among 5,384 ISAR patients with a broad spectrum of coronary syndromes and risk, bleeding in the first 30 days after PCI conferred a nearly 3-fold increased risk of dying at 1 year. The end points of 30-day bleeding, MI, and urgent revascularization had a comparable ability to predict 1-year death. These results are congruent with analyses of the REPLACE-2 (Randomized Evaluation in PCI Linking Angiomax to Reduced Clinical Events) trial comparing PCI using bivalirudin versus heparin and glycoprotein IIb/IIIa inhibitor (GPI); patients with major bleeding had a similar relative risk of 1-year mortality (2.66, 95% confidence interval 1.44 to 4.92) as did those with MI (2.46, 95% confidence interval 1.44 to 4.20) (2). On the basis of this observation, should we change our clinical trials of PCI pharmacology to place a traditional safety end point (bleeding) equivalently into the primary efficacy end point (death, MI, and urgent revascularization)?
Early Bleeding, Late Death
The complication of retroperitoneal bleeding is a clinically obvious connection between bleeding and hospital death (3). The more general association between major bleeding complications and early mortality among patients with acute coronary syndromes and PCI has also been previously described (4,5). But, the current meta-analysis provides 3 insights that challenge our current trial designs and pharmacology choices. First, bleeding events in the first 30 days confer an increased risk of death beyond 30 days; the curves continue to diverge from 30 days (7% mortality) to 1 year (14% mortality), even though there might be no further bleeding events beyond 30 days. Second, the increased mortality risk is associated with both Thrombolysis In Myocardial Infarction (TIMI) major and minor bleeding. Third, the discriminatory potential (c statistic) of MI, urgent revascularization, and bleeding for 1-year mortality are identical (0.78).
Three pathophysiological mechanisms might explain the association among PCI, bleeding, and death. 1) Bleeding might be a marker for a systemic inflammatory process that increases the risk of subsequent thrombosis. This hypothesis is supported by a recent substudy of the ACUITY (Acute Catheterization and Urgent Intervention Triage strategY) trial, suggesting that patients with bleeding complications have a heightened inflammatory state (6). 2) Early bleeding might lead to early cessation of dual antiplatelet therapy, which has been associated with increased cardiovascular events (7). 3) Bleeding leads to blood transfusions that might themselves lead to increased cardiovascular risk (8). All of these potential mechanisms might be supported by the observation that patients who bleed are at higher risk for ischemic events (4,5).
But, how can we understand the ISAR finding that minor bleeding is a significant predictor of death? Not all bleeding seems equivalent, just as not all periprocedural infarctions are clinically similar (9). The issue is the definition of TIMI minor and major bleeding. A TIMI minor bleed in the ISAR trials (clinically overt hemorrhage associated with a fall in hematocrit <15%) could include a patient with a large retroperitoneal bleed. This is not the same as a minor bleed as defined in later TIMI trials or in multiple other schemas (10–12). Given this rather expansive definition of “minor,” the ISAR trial analysis is not telling us that a small hematoma is equivalent to MI and urgent revascularization in predicting 1-year death; rather, it is telling us that there are some serious and misnomered bleeds conferring increased risk both early after the bleed and then throughout the first year.
The Quadruple End Point and Trials of Periprocedural MI
Should we accept the ISAR proposal and move from the triangle (death, MI, urgent revascularization) to the square (death, MI, urgent revascularization, and bleeding) as the basis for defining efficacy of future PCI pharmacology agents (Fig. 1)? Some would argue that such a change would prevent approval of effective agents, owing to masking of the primary goal (prevention of infarction) by secondary bleeding concerns (13). What would happen if we applied the ISAR findings to an important triple end point PCI pharmacology trial? The ESPRIT (Enhanced Suppression of the Platelet IIb/IIIa Receptor with Integrilin Therapy) trial compared aspirin, thienopyridine, eptifibatide, and unfractionated heparin with aspirin, thienopyridine, and heparin alone among patients undergoing elective stenting (14). The primary end point of this trial was death, MI, and urgent revascularization, with very low rates of death and Q-wave acute MI in both treatment groups. The benefit of eptifibatide for the primary 30-day end point was robust (10.5% vs. 6.6%, p = 0.0015) and led to wide use of eptifibatide in cardiac catheterization laboratories.
But, what would happen if the primary end point had been expanded to the square: the ESPRIT primary end point + major bleeding complications? Major bleeding complications were significantly increased in the eptifibatide versus placebo arms (1.3% vs. 0.4%, p = 0.027). If one were to crudely add the major bleeding events to the primary end point in each arm, the reduction in the 30-day primary end point is lessened (10.9% vs 7.9%) but still remains significant (p = 0.02 by chi-square). Notably, this crude addition might be an overestimate of the total number of quadruple end point events in ESPRIT: patients who bleed might also be patients with MI after PCI and thus we might be double counting events with simple addition (4,5). But even in this worst case scenario, the adoption of the ISAR suggested a quadruple end point would not mask the anti-ischemic potential of currently used adjunctive therapy.
The Quadruple End Point and Trials of Stent Thrombosis
What would happen if we applied the quadruple end point to a trial investigating PCI pharmacology aimed at preventing the devastating complication of stent thrombosis? STARS (Stent Anticoagulation Restenosis Study) randomized patients undergoing bare-metal stenting to aspirin, aspirin and warfarin, or aspirin and ticlopidine (15). The dramatic reduction in stent thrombosis with dual antiplatelet therapy compared with aspirin alone (0.5% vs. 2.9%, p = 0.005) is the basis for the current standard of care in PCI pharmacology. In STARS, major bleeding complications were 3-fold higher in the dual versus single antiplatelet therapy group (5.5% vs. 1.8%, p < 0.001). If one were to use crude addition to assemble a quadruple end point for STARS (primary end point + bleeding), the trend would no longer favor dual antiplatelet therapy as the standard of care after stenting (5.4% event rate for aspirin vs. 6.0% for dual antiplatelet therapy, p = NS).
This example lends caution to the use of the quadruple end point outside of the ISAR-style trial designs that have clearly linked major bleeding to adverse outcomes. The ISAR trials investigated the utility of pharmacology regimens designed to reduce primarily asymptomatic periprocedural MI. STARS did not include asymptomatic periprocedural infarctions in the primary end point. Thus, STARS is a different kind of PCI pharmacology trial in which bleeding complications are not yet demonstrated to be comparable to the type of infarction (stent thrombosis, Q-wave, or ST-segment elevation) being primarily prevented. In future PCI pharmacology trials with the primary goal of preventing post-PCI death, Q-wave MI, or stent thrombosis, the quadruple end point may not necessarily apply.
The Future of Net Clinical Benefit
This proposed change to a quadruple end point for many PCI pharmacology trials is contingent on a new consensus terminology for bleeding (10–12). Major bleeding needs a single, predictive definition across trials. Minor bleeding should reflect a situation that would not impact mortality. This rationale for uniformity of bleeding definitions is analogous to the need for a uniform definition of ischemic events across PCI trials (16). At this point, the quadruple end point should be applied only to populations similar to those in the ISAR trial analysis; for example, the 1-year impact of major bleeding complications as opposed to recurrent infarction is unknown in higher-risk patient groups (i.e., patients with cardiogenic shock) (17).
The ISAR proposal of balancing bleeding and ischemic concerns in a composite end point estimating the “net clinical benefit” of a drug might seem new and controversial (18). But, the concept is at least 15 years old: for example, the landmark GUSTO-1 (Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries) trial compared different fibrinolytic strategies and had a primary end point of mortality (19); the secondary end point, however, was “net benefit, defined as a reduction in mortality and disabling stroke.” Thus, the ISAR group’s proposal of expanding the primary goal of our PCI pharmacology trials from the triangle to the square is not entirely revolutionary; rather, it continues our emphasis on developing adjunctive pharmacology that derives benefit from being both safe and effective.
The author would like to thank Dr. David J. Schneider and Dr. Robert A. Harrington for their helpful comments.
↵1 Dr. Dauerman is on the steering committee for the CHAMPION PCI (cangrelor) trial and a consultant to The Medicines Company.
↵⁎ Editorials published in the Journal of 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
- Ndrepepa G.,
- Berger P.B.,
- Mehilli J.,
- et al.
- Eikelboom J.W.,
- Mehta S.R.,
- Anand S.S.,
- Xie C.,
- Fox K.A.,
- Yusuf S.
- Manoukian S.V.,
- Feit F.,
- Mehran R.,
- et al.
- Jozic J.,
- Steinhubl S.R.,
- Dangas G.D.,
- Hooper W.C.,
- Stone G.W.
- Jeremias A.,
- Baim D.S.,
- Ho K.K.,
- et al.
- Steinhubl S.R.,
- Kastrati A.,
- Berger P.B.
- Rao S.V.,
- O’Grady K.,
- Pieper K.S.,
- et al.
- Cutlip D.E.,
- Windecker S.,
- Mehran R.,
- et al.