OUP user menu

New paradigms of care for STEMI focusing on mortality and attributable death analysis: what do device and drug trials teach us?

Roxana Mehran , Gregg W. Stone
DOI: http://dx.doi.org/10.1093/eurheartj/sup011 C4-C8 First published online: 22 May 2009

Abstract

In patients with acute coronary syndromes, ST-segment elevation myocardial infarction (STEMI), and in those undergoing percutaneous coronary intervention (PCI), major bleeding has been shown to be a powerful independent determinant of mortality, at least as important as MI or re-infarction. Treatment with bivalirudin compared with heparin and a GP IIb/IIIa inhibitor results in a significant reduction in 30-day major bleeding, thrombocytopaenia and transfusions, with similar rates of ischaemic events in high-risk patients with STEMI undergoing primary PCI. Furthermore, bivalirudin monotherapy resulted in significant 31 and 43% reductions in rates of all-cause and cardiac mortality (absolute 1.4 and 1.7% reductions) at 1 year, with non-significantly different rates of re-infarction and stent thrombosis. The prevention of haemorrhagic complications after primary PCI in STEMI results in improved early and late survival. Optimal drug selection and technique to minimize bleeding are essential to enhance outcomes for patients undergoing interventional therapies.

KEYWORDS
  • STEMI
  • Mortality
  • Attributable death
  • Myocardial infarction
  • Antithrombin

Introduction

Compared with pharmacologic reperfusion alone, primary percutaneous coronary intervention (PCI) decreases the rate of recurrent ischaemic events,1,2 and the addition of glycoprotein (GP) IIb/IIIa inhibitors to heparin during primary PCI has shown to reduce mortality and morbidity in two meta-analyses.3,4 However, the protective ischaemic benefit of GP IIb/IIIa inhibitors is offset by an increased risk of bleeding complications and thrombocytopaenia, both of which have been strongly associated with short- and long-term mortality.48 In the large-scale REPLACE-2 and ACUITY randomized trials of patients with unstable angina and moderate- and high-risk non-ST-segment elevation myocardial infarction (MI) acute coronary syndromes (ACS), the direct thrombin inhibitor bivalirudin has demonstrated similar protection from ischaemia as heparin plus a GP IIb/IIIa inhibitor, but significantly reduced bleeding complications.9,10 Results of the pivotal HORIZONS-AMI trial in patients with ST-segment elevation myocardial infarction (STEMI) demonstrated a remarkable reduction in bleeding along with a significant decrease in 30 day and 1-year mortality with bivalirudin alone compared with heparin plus routine use of GP IIb/IIIa inhibitors.11,12 These findings suggest that it is time to focus on the factors that attribute to the risk of mortality in patients with ACS.

The ACUITY trial

The ACUITY trial randomized 13 819 patients with moderate and high-risk non-STEMI ACS to receive either heparin (unfractionated or enoxaparin) with routine GP IIb/IIIa inhibitors, bivalirudin plus routine GP IIb/IIIa inhibitors, or bivalirudin alone (with provisional GP IIb/IIIa inhibitors used in 9.1% of patients overall and in 6.5% of patients for procedural complications).10 All patients underwent angiography with subsequent triage to medical management, PCI, or CABG. Thirty-day primary endpoints were composite ischaemia (death, MI, and unplanned revascularization), non-CABG major bleeding, and net adverse clinical events (NACE; composite ischaemia+non-CABG major bleeding). Composite ischaemia and mortality were assessed up to 1 year.

Impact of major bleeding and myocardial infarction on mortality

In order to make balanced decisions regarding the optimal treatment of patients with STEMI, it is necessary to understand what factors impact long-term mortality. An analysis of the ACUITY trial investigated the relationship between a non-CABG major bleeding event and MI occurring within the first 30 days of randomization and the risk of subsequent mortality at 1 year.13 Patients with no MI or major bleeding had a very low risk of 1 year mortality (3.4%), whereas the risk increased to 8.6% for a patient with an MI only and 12.5% for patient with a major bleeding event (Figure 1). If a patient had an MI and major bleeding within 30 days, the risk of death at 1 year was 28.9%. Interestingly, the mortality rates at 30 days were similar for patients with either an MI or major bleeding; however, as time went on, the risk of mortality at 1 year was higher in patients with major bleeding than those who had an MI.13

Figure 1

Impact of myocardial infarction, major bleeding (non-CABG related), both or neither in the first 30 days on risk of death over 1 year in the ACUITY trial.

The impact of major bleeding and MI in the first 30 days on the risk of death over 1 year was evaluated using a Cox model adjusted for 36 baseline predictors, with MI, and major bleeding entered as time-updated covariates.14 The results indicated that MI was associated with a higher hazard ratio (HR) than major bleeding (HR = 2.51 vs. HR = 2.00, respectively), though major bleeding after transfusion had the highest hazard (HR = 3.93). Despite these findings, the number of attributable deaths was greater in patients who had major bleeding compared with those with an MI (Figure 2). This may explain why the estimated risk of mortality at 1 year was greater in patients who experienced major bleeding than those who had an MI.

Figure 2

Influence of major bleeding and myocardial infarction in the first 30 days on the risk of death over 1 year in the ACUITY trial.

The influence of MI, major bleeding, and transfusion in the first 30 days on the risk of death over 1 year was also analysed based upon the timing of the occurrence (Day 0–1, Days 2–7, Days 8–30, and Days 31+) in order to understand when it is most important to protect a patient from an MI or major bleeding. The results indicated that the risk of death from an MI decreases as time passes from an HR of 17.6 on Day 0–1 to an HR of 1.4 from Days 31+. In contrast, there was a consistently high HR for major bleeding at all time points (5.5, 5.8, 5.6, and 2.4, respectively), suggesting that it is important to protect patients from major bleeding events at all times. Furthermore, the number of deaths attributable to an MI remained fairly consistent over time, while those attributed to a major bleeding event increased as time passed, from 9 deaths on Day 0–1 to 18, 24, and 42 deaths during Days 2–7, Days 8–30, and Days 31+, respectively.14

Impact of antithrombin therapy on outcomes

With regard to antithrombin treatment, bivalirudin alone significantly reduced the 30 day major bleeding rates compared with heparin plus a GP IIb/IIIa inhibitor (3.0 vs. 5.7%, P < 0.001), and importantly, there was no difference in the rate of MI (5.4 vs. 4.9%, P = 0.35).10 Since the impact of bleeding on long-term mortality is so important, and bivalirudin monotherapy reduced bleeding, then one would expect to see a mortality benefit at 1 year in these patients. But in fact, there was no significant difference in rates of 1-year mortality between the three treatment groups; however, Kaplan–Meier estimated rates of 1-year mortality were numerically lowest in the bivalirudin alone arm (3.8%), intermediate in the bivalirudin plus GP IIb/IIIa inhibitor arm (4.2%), and highest in the heparin plus GP IIb/IIIa inhibitor arm (4.4%).15

One potential explanation for the lack of a mortality benefit is that the trial was conducted in patients with ACS, who were not very high risk, and it is possible that the effect would be evidenced in a higher risk population. A logistic regression analysis evaluating the impact of 36 potential variables on the risk of 30 day MI indentified increased baseline cardiac biomarkers (CKMB/troponin), family history of coronary artery disease, increasing age, baseline ST-segment deviation, and prior MI to be significantly associated. Randomization to bivalirudin with or without a GP IIb/IIIa inhibitor was not significantly associated with a risk of MI.14 The findings indicate that it is not randomization to a GP IIb/IIIa inhibitor that might have protected a patient, or randomization to bivalirudin alone that may have harmed a patient, but essentially the baseline characteristics of the patient that predict the risk of 30 day MI.

With regard to 30 day major bleeding, baseline characteristics were also found to be significantly associated with an increased risk. In the logistic regression model, variables predictive of major bleeding included female gender, anaemia, increasing age, decreased creatinine clearance, increased white blood cell count, prior PCI, prior stroke, and baseline ST-segment deviation. However, in contrast to MI model results, randomization to bivalirudin alone was found to be an important predictor of decreased bleeding, though randomization to bivalirudin plus a GP IIb/IIIa inhibitor was not significant.16

The HORIZONS-AMI trial

The HORIZONS-AMI trial randomized 3602 patients with STEMI undergoing primary PCI to therapy with unfractionated (UFH) plus GP IIb/IIIa inhibition or bivalirudin monotherapy.11 Angiography was performed followed by triage to PCI, CABG, or medical management. After patency was restored, eligible patients were further randomized in a 3:1 fashion to receive either a paclitaxel-eluting stent (TAXUS, Boston Scientific) or a bare metal stent (Express2, Boston Scientific). The two 30-day primary endpoints were non-CABG major bleeding and NACE [defined as a combination of major adverse cardiac events (MACE: death, re-infarction, target vessel revascularization for ischaemia, and stroke) and major bleeding].

Impact of major bleeding and myocardial infarction on mortality

An adjusted Cox model with MACE components and major bleeding as time-updated covariates relating 30 day events to 30 day mortality found that, as was the case in non-STEMI patients in ACUITY, the two most important events predictive of 30 day mortality were re-infarction and major bleeding. Re-infarction was extremely hazardous, more than twice that of major bleeding, but as in the ACUITY trial, there were more than twice as many deaths attributable to major bleeding compared to those with MI.17 Furthermore, all definitions of bleeding (protocol-defined, TIMI, and GUSTO), as well as thrombocytopaenia, were important and attributed to death (Table 1).18 The data suggest that even though a more common event such as major bleeding is less hazardous, it can attribute to more deaths in high-risk patients presenting with acute MI.

View this table:
Table 1

Time-updated covariate adjusted Cox model relating 30 day bleeding events and thrombocytopaenia to 30 day mortality in the ACUITY trial

VariableHR (95% CI)Attributable deathsaP-value
Protocol-defined
 Major bleeding (non-CABG)4.52 (2.77, 7.40)20.3 (16.6, 22.5)<0.001
 Major bleeding (all)5.71 (3.63, 8.99)28.9 (25.3, 31.1)<0.001
 Transfusion4.18 (2.30, 7.59)12.2 (9.0, 13.9)<0.001
TIMI
 Major7.98 (4.92, 12.95)22.7 (20.7, 24.0)<0.001
 Minor1.79 (0.81, 3.96)3.1 (1.6, 5.2)0.15
 Major or minor5.45 (3.44, 8.62)26.9 (23.4, 29.2)<0.001
GUSTO
 Severe11.98 (5.90, 24.34)8.2 (7.5, 8.6)<0.001
 Severe/moderate3.72 (2.28, 6.06)19.0 (14.6, 21.7)<0.01
Thrombocytopaenia
 <100 000 cells/mm33.54 (2.01, 6.24)10.8 (7.5, 12.6)<0.001
 <50 000 cells/mm34.84 (2.13, 10.97)5.6 (3.7, 6.4)<0.001
 <20 000 cells/mm32.38 (0.53, 10.63)1.2 (−1.8, 1.8)0.26
  • aEighty-eight deaths in 3550 patients.

Impact of antithrombin therapy on outcomes

With regard to antithrombin therapy, no difference was observed in 30 day re-infarction (1.8 vs. 1.8%, P = 0.90), target vessel revascularization (2.6 vs. 1.9%, P = 0.18), or stroke (0.7 vs. 0.6%, P = 0.68) with bivalirudin vs. heparin plus GP IIb/IIIa inhibition, results that persisted at 1 year.11,12 However, bivalirudin monotherapy significantly reduced protocol-defined major bleeding (4.9 vs. 8.3%, P < 0.001) and thrombocytopaenia (1.8 vs. 3.9%, P = 0.0002), as well as all bleeding endpoints, regardless of the definition used (Table 2).11

View this table:
Table 2

Thirty day bleeding endpoints by randomized antithrombin therapy in the HORIZONS-AMI trial

UFH+GP IIb/IIIa inhibitors (n = 1802)Bivalirudin (n = 1800)P-value
Protocol-defined
 Major bleeding (non-CABG)8.3%4.9%<0.001
 Major bleeding (all)10.8%6.8%<0.001
 Transfusion3.5%2.1%0.009
TIMI
 Major5.0%3.1%0.002
 Minor4.6%2.8%0.006
 Major or minor9.6%5.9%<0.0001
GUSTO
 Life-threatening or severe0.6%0.4%0.49
 Moderate5.0%3.1%0.002
 Life-threatening, severe, or moderate5.6%3.5%0.002

At 30 days, bivalirudin monotherapy resulted in significantly lower rates of all-cause [2.1 vs. 3.1%, HR (95% CI) = 0.66 (0.44, 1.00), P = 0.048]. Cardiac-related mortality was also significantly reduced with bivalirudin [1.8 vs. 2.9%, HR (95% CI) = 0.62 (0.40–0.96), P = 0.03] compared with heparin plus GP IIb/IIIa inhibition, whereas there was no difference in non-cardiac-related mortality (Figure 3).11 This benefit was maintained at 1 year, with significantly reduced rates of all-cause mortality with bivalirudin [3.4 vs. 4.8%, HR (95% CI) = 0.69 (0.50, 0.97), P = 0.029]. In fact, the absolute reduction was even greater than at 30 days, with 14 patients saved per 1000 from all-cause mortality, and 17 patients per 1000 saved from cardiac mortality when treated with bivalirudin (Figure 4).12 The reduction in cardiac-related mortality with bivalirudin vs. UFH plus a GP IIb/IIIa inhibitor also persisted at 1 year [2.1 vs. 3.8%, HR (95% CI) = 0.57 (0.38, 0.84), P = 0.005]. For the first time, it was shown that a reduction in bleeding could reduce mortality, since the only difference between the bivalirudin monotherapy and heparin and GP IIb/IIIa inhibitor arm was a reduction in bleeding and thrombocytopaenia. There are many factors that may explain how major bleeding impacts mortality, one being that when a patient bleeds, anticoagulation, and anti-ischaemic medications are discontinued, therefore placing the patient at higher risk for ischaemic events and possibly subsequent mortality.

Figure 3

Thirty day cardiac and non-cardiac mortality in the HORIZONS-AMI trial (Adapted from Stone GW, Witzenbichler B, Guagliumi G, et al. Bivalirudin during primary PCI in acute myocardial infarction. N Engl J Med 2008;358:2218–2230. Copyright © 2008 Massachusetts Medical Society. All rights reserved.).

Figure 4

One year all-cause mortality in the HORIZONS-AMI trial.

There was no significant difference in the overall rate of Academic Research Consortium (ARC)-defined definite or probable stent thrombosis with bivalirudin vs. heparin plus GP IIb/IIIa inhibition (2.5 vs. 1.9%, P = 0.30); however, there were more episodes of acute stent thrombosis (≤24 h) in patients who received bivalirudin monotherapy (1.3 vs. 0.3%, P < 0.001).11 It is interesting that despite the increased rate of acute stent thrombosis, cardiac mortality was still reduced, a finding that may be explained by the fact that although stent thrombosis was highly hazardous compared with major bleeding, there was three times more attributable death to major bleeding compared with the rare event of stent thrombosis.17 At 1 year, there was no difference in definite or probable stent thrombosis in the bivalirudin vs. heparin group (3.5 vs. 3.2%, P = 0.59).

Impact of paclitaxel-eluting vs. bare metal stent

There was no interaction between antithrombin therapy and stent randomization with respect to the 30 day pharmacology endpoints including NACE (P for interaction = 0.95), MACE (P for interaction = 0.89), and major bleeding (P for interaction = 1.0).12 There was also no interaction between drug and stent randomization with respect to the 1-year stent endpoints including ischaemic target lesion revascularization (P for interaction = 0.17), safety MACE, defined as death, re-infarction, stroke, or stent thrombosis (P for interaction = 0.89), and binary restenosis (P for interaction = 0.18).12 Rates of all-cause 1-year mortality were lowest with bivalirudin, regardless of the stent used (bivalirudin with Taxus, 2.6%; bivalirudin with Express, 3.0%; heparin plus GP IIb/IIIa inhibitor with Taxus, 4.0%; heparin plus GP IIb/IIIa inhibitor with Express, 4.6%; P for interaction = 0.75).12

Limitations

Both ACUITY and HORIZONS were open-label trials; however, potential bias was mitigated by high protocol procedure compliance and the use of blinded clinical event adjudication committees and core laboratories. Furthermore, the studies were underpowered for mortality, and therefore all findings should be considered hypothesis generating.

Conclusions

Major bleeding is a powerful independent determinant of mortality, at least as important as MI or re-infarction, in patients with ACS, STEMI, and in those undergoing PCI. In high-risk patients with STEMI undergoing primary PCI, treatment with bivalirudin compared with heparin and a GP IIb/IIIa inhibitor results in a significant reduction in 30 day major bleeding, thrombocytopaenia and transfusions, with similar rates of ischaemic events. At 1 year, bivalirudin monotherapy resulted in significant 31 and 43% reductions in rates of all-cause and cardiac mortality (absolute 1.4 and 1.7% reductions), with non-significantly different rates of re-infarction and stent thrombosis.

The prevention of haemorrhagic complications after primary PCI in STEMI results in improved early and late survival. Optimal drug selection and technique to minimize bleeding are essential to enhance outcomes for patients undergoing interventional therapies.

Conflict of interest: Clinical Research Support from The Medicines Company, and Boston Scientific; Educational Research Support from The Medicines Company, Boston Scientific, Abbott, Medtronic and Cordis.

References

View Abstract