OUP user menu

Profile of bleeding and ischaemic complications with bivalirudin and unfractionated heparin after percutaneous coronary intervention

Raisuke Iijima, Gjin Ndrepepa, Julinda Mehilli, Robert A. Byrne, Stefanie Schulz, Franz-Josef Neumann, Gert Richardt, Peter B. Berger, Albert Schömig, Adnan Kastrati
DOI: http://dx.doi.org/10.1093/eurheartj/ehn586 290-296 First published online: 15 January 2009


Aims The aim of this study was to identify a subset of patients at high risk of bleeding or myocardial infarction from a percutaneous coronary intervention and to investigate whether such high-risk subsets derive preferential benefit from heparin or bivalirudin.

Methods and results This study included 4570 patients with coronary artery disease enrolled in the Intracoronary Stenting and Antithrombotic Regimen: Rapid Early Action for Coronary Treatment trial and randomized to receive bivalirudin or heparin. Primary outcomes were in-hospital incidence of major bleeding and 30-day incidence of myocardial infarction. Major bleeding, myocardial infarction, and bleeding plus myocardial infarction occurred in 140, 204, and 34 patients, respectively. Older age, female sex, lower body weight, low cholesterol, multi-lesion intervention, complex lesions, and heparin therapy were independent correlates of increased risk of bleeding. Multi-lesion intervention, unstable angina, and lower body weight correlated independently with increased risks of myocardial infarction. Compared with heparin, bivalirudin was associated with a reduction in major bleeding (3.1 vs. 4.6%, P = 0.008), but mostly in low-risk patients. A reduction in the bleeding risk inversely correlated with an increase in the risk of myocardial infarction with bivalirudin (R = −0.61).

Conclusion Bivalirudin and unfractionated heparin have a differential effect on risk of bleeding and myocardial infarction across various subsets of patients.

  • Bivalirudin
  • Bleeding
  • Coronary artery disease
  • Heparin
  • Myocardial infarction
  • Percutaneous coronary intervention


The widespread use of percutaneous coronary intervention (PCI) in patients with coronary artery disease has mandated the application of aggressive adjunctive anti-thrombotic and anticoagulant therapies to reduce thrombotic complications.1,2 Although such therapy has reduced ischaemic complications, most notably procedural infarction, it has increased the risk of bleeding and the requirement for blood transfusion.3,4 Recent findings suggest that both myocardial infarction and bleeding occurring after a PCI procedure have comparable prognostic impact on mortality at 1 year.5 Patients who may be at particularly increased risk for myocardial infarction or major bleeding after PCI have been identified.69 There is a general perception that patients at highest risk for a given adverse event benefit mostly from a therapy that is effective in reducing that event.

Unfractionated heparin and bivalirudin are the most commonly used anticoagulants during PCI procedures. The results of the Intracoronary Stenting and Antithrombotic Regimen: Rapid Early Action for Coronary Treatment (ISAR-REACT) 3 trial revealed that bivalirudin does not improve net clinical benefit (the 30-day quadruple endpoint), compared with unfractionated heparin in troponin-negative patients with stable and unstable angina undergoing PCI ≥2 h after a 600 mg loading dose of clopidogrel.10 In that trial, bivalirudin and unfractionated heparin had opposite effects on the individual components of the composite endpoint. Patients treated with bivalirudin had significantly less bleeding, and those treated with heparin tended to have fewer myocardial infarctions than with bivalirudin.10 Thus identification of subgroups of patients more prone to benefit from bivalirudin or heparin used during PCI procedures would be of great clinical value.

The objectives of this study were to identify a subset of patients with a particularly high risk of periprocedural bleeding or myocardial infarction from a PCI and to investigate whether such high-risk subsets derive preferential benefit from heparin or bivalirudin.


Patients and protocol

In the ISAR-REACT 3 trial, 4570 patients undergoing PCI who were at low-to-intermediate risk were enrolled between September 2005 and January 2008. Details of the trial design, patient eligibility, and laboratory measurements have been reported previously.10 In brief, all patients were pre-treated with 325–500 mg aspirin and a 600 mg loading dose of clopidogrel at least 2 h before the PCI procedure. Before the guide wire had crossed the lesion, patients were randomized to receive either a 0.75 mg/kg bolus of bivalirudin followed by an infusion of 1.75 mg/kg/h for the duration of the procedure (n = 2289 patients) or a 140 U/kg bolus of unfractionated heparin followed by a placebo infusion for the duration of the procedure (n = 2281 patients). Double blinding was achieved using identically appearing vials in the two study groups. Post-procedural therapy included daily aspirin, 80–325 mg indefinitely, clopidogrel, 75–150 mg/day until discharge but no longer than 3 days followed by 75 mg/day for at least 1 month (bare-metal stents) or 6 months (drug-eluting stents) and other cardiac medications prescribed by the patient's physician.

Electrocardiographic examinations and laboratory measurements including cardiac enzymes, haemoglobin, and platelet count were performed every 8 h for the first 24 h after the PCI procedure and daily afterwards, until discharge.

Definitions and follow-up

The primary outcomes of this analysis were the in-hospital incidence of major bleeding and the 30-day incidence of myocardial infarction. Major bleeding was defined according to the criteria used in the Randomized Evaluation in PCI Linking Angiomax to Reduced Clinical Events (REPLACE-2) trial: intracranial, intraocular, or retroperitoneal haemorrhage, clinically overt blood loss resulting in a decrease in haemoglobin of >3 g/dL, any decrease in haemoglobin of >4 g/dL, or transfusion of ≥2 U of packed red blood cells or whole blood.3 The definition of myocardial infarction included the development of pathological Q waves (≥30 ms in duration and ≥0.1 mV in depth) in two or more contiguous precordial leads or two or more adjacent limb leads, or elevation of creatine kinase (CK) MB isoenzyme (or total CK if CK-MB not available) two or more times the upper limit of normal. Follow-up was obtained in all patients at 30 days by phone; patients who had cardiac complaints underwent a complete clinical, electrocardiographic, and laboratory evaluation.

Statistical analysis

The data are presented as mean ± SD or counts (%). Categorical data were compared with the χ2 test or Fisher's exact test when cell values were less than 5. Continuous data were compared using a two-tailed unpaired t-test. Multiple logistic regression analysis was used to identify independent risk factors associated with bleeding or myocardial infarction. Adjustment was performed for clinically relevant variables considered to impact the study outcomes: age, sex, diabetes, cholesterol level, arterial hypertension, smoking, type of angina, number of treated lesions, patient's weight, previous myocardial infarction, previous coronary artery bypass surgery, complex lesions, and treatment type. Age and patient's weight were entered as continuous variables because they fulfilled the linearity assumption. This was proven by the use of cubic spline function,11 which showed that the relation between these continuous variables and the outcome(s) of interest did not deviate significantly from the linear form. The same model was used to check for heterogeneity of treatment differences across the levels of a baseline variable by assessing the interaction between assigned treatment and baseline variable with respect to the endpoint of interest. The analysis was performed with the S-plus statistical package (S-PLUS, Insightful Corp., Seattle, WA, USA). A two-sided P-value <0.05 was considered to indicate statistical significance.


Patients were divided into four groups: patients who had major bleeding (n = 140), myocardial infarction (n = 204), major bleeding plus myocardial infarction (n = 34) and those with neither major bleeding nor myocardial infarction (n = 4192).

Baseline characteristics

Baseline characteristics of patients in the four groups are shown in Table 1. There were significant differences among patients of four groups with regard to age, sex, proportion of patients with hypercholesterolaemia, severity of coronary artery disease, intervention in multiple lesions, and body weight.

View this table:
Table 1

Baseline characteristics of the patients

CharacteristicNo major bleeding or myocardial infarction (n = 4192)Major bleeding (n = 140)Myocardial infarction (n = 204)Major bleeding and myocardial infarction (n = 34)P-value
Age (years)66.7 ± 10.071.6 ± 9.266.7 ± 10.072.2 ± 9.9<0.001
Women, no. (%)944 (22.5)59 (42.1)60 (29.4)12 (35.3)<0.001
Diabetes, no. (%)1134 (27.1)47 (33.6)64 (31.4)9 (26.5)0.206
 Insulin treated, no. (%)333 (7.9)15 (10.7)15 (0.7)4 (11.7)0.536
 Oral medication treated, no. (%)623 (14.9)21 (15.0)29 (14.2)3 (8.8)0.792
Arterial hypertension, no. (%)3734 (89.1)131 (93.6)181 (88.7)32 (94.1)0.289
Hypercholesterolaemia, no. (%)3355 (80.0)91 (65.0)172 (84.3)27 (79.4)<0.001
Current smoking, no. (%)662 (14.8)16 (11.4)25 (12.3)2 (5.9)0.231
 Unstable, no. (%)625 (14.9)25 (17.9)43 (21.1)10 (29.4)
 Stable, no. (%)3567 (85.1)115 (82.1)161 (79.0)24 (70.6)
No. of diseased coronary vessels0.024
 One vessel, no. (%)853 (20.4)30 (21.4)25 (12.3)3 (8.8)
 Two vessels, no. (%)1192 (28.4)31 (22.1)59 (28.9)9 (26.5)
 Three vessels, no. (%)2147 (51.2)79 (56.4)120 (58.8)22 (64.7)
Intervention in multiple lesions, no. (%)1889 (45.1)83 (59.3)128 (62.7)27 (79.4)<0.001
Prior myocardial infarction, no. (%)1314 (31.4)41 (29.3)56 (27.5)12 (35.3)0.598
Prior coronary artery bypass surgery, no. (%)485 (11.6)16 (11.4)24 (11.8)9 (26.5)0.063
Body weight (kg)81.8 ± 14.174.9 ± 15.480.4 ± 14.869.5 ± 11.2<0.001
Serum creatinine (mg/dL)0.96 ± 0.260.96 ± 0.290.96 ± 0.230.98 ± 0.260.947
  • Data are presented as number of patients (%) or mean ± SD.

Angiographic and procedural (lesion-based) characteristics are shown in Table 2. Significant differences among patients of four groups were observed only with regard to proportion of patients with complex lesions.

View this table:
Table 2

Angiographic lesion-based procedural characteristics

CharacteristicNo major bleeding or myocardial infarction (n = 6948)Major bleeding (n = 284)Myocardial infarction (n = 433)Major bleeding and myocardial infarction (n = 90)P-value
Left ventricular ejection fraction (%)57.8 ± 10.757.1 ± 10.658.3 ± 10.057.1 ± 7.90.591
Treated vessel, no. (%)0.133
 Left anterior descending2732 (39.3)123 (43.3)182 (42.0)38 (42.2)
 Right coronary artery2040 (29.4)79 (27.8)118 (27.3)21 (23.3)
 Left circumflex1790 (25.8)66 (23.2)115 (26.6)19 (21.2)
 Left main artery260 (3.7)11 (3.9)13 (3.0)9 (10.0)
 Bypass graft126 (1.8)5 (1.8)5 (1.1)3 (3.3)
Complex (type B2/C) lesions, no. (%)4636 (66.7)222 (78.2)312 (72.1)76 (84.4)<0.001
Chronic occlusion, no. (%)465 (6.7)19 (6.7)42 (9.7)8 (8.9)0.096
Type of interventions, no. (%)0.099
 Drug-eluting stent6097 (87.8)247 (87.0)380 (87.8)75 (83.3)
 Bare-metal stent385 (5.5)12 (4.2)33 (7.6)6 (6.7)
 Balloon angioplasty466 (6.7)25 (8.8)20 (4.6)9 (10.0)
  • Data are presented as number of lesion (%) or mean ± SD.

Predictors of major bleeding and myocardial infarction

Multiple logistic regression model was used to identify the independent correlates of major bleeding and myocardial infarction. In these models, we included all study patients. Patients with both complications, major bleeding and myocardial infarction, were included in either model. The results are shown in Table 3. Age, female gender, low body weight, low cholesterol, number of treated lesions, complex lesions, and treatment with heparin were independent correlates of bleeding. The multivariable analysis identified intervention in multiple lesions, unstable angina, and a low body weight as independent correlates of increased risk of myocardial infarction at 30 days (Table 3).

View this table:
Table 3

Independent predictors of major bleeding and myocardial infarction at 30 days

CharacteristicAdjusted odds ratio (95% confidence intervals)P-value
Predictors of major bleeding at 30 days
 Age (for a 10-year increase in age)1.37 (1.14–1.64)<0.001
 Females1.47 (1.02–2.14)0.039
 Body weight (for a 10 kg decrease in weight)1.37 (1.19–1.57)<0.001
 Low cholesterol1.85 (1.32–2.63)<0.001
 Number of treated lesions1.99 (1.44–2.76)<0.001
 Complex lesion1.68 (1.16–2.43)0.006
 Bivalirudin treatment0.64 (0.47–0.88)0.006
Predictor of myocardial infarction at 30 days
 Number of treated lesions2.20 (1.66–2.90)<0.001
 Unstable angina1.38 (1.01–1.89)0.021
 Body weight (for a 10 kg decrease in weight)1.14 (1.02–1.27)0.021

Impact of bivalirudin and heparin on major bleeding and myocardial infarction

We analysed the relationship between the safety and efficacy of bivalirudin vs. heparin for the seven independent correlates of major bleeding or myocardial infarction: age, sex, body weight, cholesterol levels (elevated vs. normal level), multi-lesion intervention (vs. single lesion intervention), angina type (unstable vs. stable), and complexity of lesions (complex vs. simple). For both these outcomes, there was no significant interaction between treatment effect and variables defining the various subsets.

Major bleeding occurred in 70 of 2289 patients (3.1%) who received bivalirudin vs. 104 of 2281 patients (4.6%) who received unfractionated heparin [odds ratio (OR) = 0.66, 95% confidence interval (CI) 0.46–0.90, P = 0.008]. Bivalirudin was superior to heparin with regard to major bleeding in patients ≤75 years of age (2.1 vs. 3.6%), men (2.3 vs. 3.6%), patients with body weight >70 kg (1.9 vs. 3.4%), normal cholesterol level (4.3 vs. 7.6%), single lesion intervention (1.7 vs. 3.5%), complex lesions (3.7 vs. 5.4%), and stable angina (3.0 vs. 4.3%; Figure 1). Thus, bivalirudin significantly reduced the risk of bleeding mostly in subsets of patients considered at low risk for bleeding. Bivalirudin was not helpful in reducing the risk of bleeding in subsets of patients at a higher risk for this event (Figure 1).

Figure 1

Treatment effect of bivalirudin vs. unfractionated heparin in various subgroups. Left panel: treatment effect on major bleeding. Right panel: treatment effect on myocardial infarction. There was no significant interaction between treatment and any of the subgroups regarding the risk of bleeding (P ≥ 0.226 for all interactions) or myocardial infarction (P ≥ 0.06 for all interactions). The cut-off values of 75 years for age and 70 kg for body weight are close to the values defining the upper and lower quartiles, respectively.

By 30 days, myocardial infarction occurred in 128 of 2289 patients (5.6%) in the bivalirudin group vs. 110 of 2281 patients (4.8%) in the heparin group (OR = 1.17, 95% CI 0.90–1.52, P = 0.241). Although a trend towards more myocardial infarctions among patients treated with bivalirudin vs. those treated with unfractionated heparin was observed in the majority of subgroups, statistical significance was achieved only in patients with a body weight >70 kg (5.4 vs. 4.0%; Figure 1).

The relationship between the risk of bleeding and the risk of myocardial infarction with bivalirudin in all 14 subsets of patients is shown in Figure 2. There is an inverse relationship between the risk of bleeding and myocardial infarction (R = −0.61; P = 0.02); subsets that had the greatest reduction in risk of bleeding with bivalirudin had the greatest increase in the risk of myocardial infarction with this drug as well.

Figure 2

Relation between the risk of bleeding and the risk of myocardial infarction with bivalirudin. MI, myocardial infarction.


The main findings of this analysis are: (i) subsets of patients at higher risk for post-procedural complications can be readily identified; (ii) although bivalirudin reduced the risk of bleeding in the entire study population, it did not decrease bleeding in those subsets of patients considered at high risk for this complication; (iii) there was an inverse relationship between the impact of bivalirudin on bleeding and myocardial infarction.

Bivalirudin was associated with a lower risk of major bleeding, following PCI in the overall study population. With regard to the predisposing factors for bleeding such as female sex, our data concur with a previous study of patients treated with either heparin plus glycoprotein IIb/IIIa inhibitors or bivalirudin.12 In our analysis, two angiographic factors—intervention in multiple lesions and complex lesions—were also identified as independent correlates of major bleeding. Intervention in multiple lesions was also an independent correlate of myocardial infarction. These findings might be explained by the longer procedural time and more complicated procedures required in patients with more extensive and complex coronary atherosclerosis, which would be expected to increase the risk of both bleeding and procedural infarction.13,14

Bivalirudin is a direct thrombin inhibitor that has been studied as an alternative to unfractionated heparin in patients with stable angina, non-ST-elevation acute coronary syndromes, and ST-elevated myocardial infarction undergoing PCI.3,15,16 Recent randomized trials suggest that bivalirudin is at least as effective as unfractionated heparin in preventing ischaemic complications in patients undergoing PCI and is associated with less major bleeding.17 In most randomized trials, however, bivalirudin has been compared with heparin plus routine use of glycoprotein IIb/IIIa inhibitors.3,15,16 The ISAR-REACT 3 trial, which compared bivalirudin and unfractionated heparin without adjunct use of glycoprotein IIb/IIIa inhibitors, allowed a better comparison of safety and efficacy profiles of these anticoagulants used during PCI procedures.10 Although the primary quadruple endpoint did not favour either bivalirudin or heparin, there were trends and significant differences that went in different directions for these two anticoagulants in terms of individual components of the primary quadruple endpoint.

Intuitively, one might think that bivalirudin therapy may be more beneficial in patients with risk factors for bleeding. Indeed, previous studies have reported that bivalirudin reduces the risk of bleeding in women and patients with renal insufficiency.12,18 In contrast, the present analysis did not demonstrate the superiority of bivalirudin over heparin in reducing the risk of bleeding in subgroups at highest risk of bleeding. Our novel finding was that bivalirudin was most advantageous in reducing the risk of bleeding in patients at low risk, such as younger patients, men, patients with greater body weight, and those with single lesion intervention. The use of bivalirudin in these subgroups was associated with a trend towards an increased risk of myocardial infarction. This risk–efficacy profile of bivalirudin action could be instrumental to explanation of an apparent paradox reported in the Acute Catheterization and Urgent Intervention Triage Strategy (ACUITY) trial.19 In that trial, although bivalirudin significantly reduced the incidence of bleeding, which was shown in this study (and others) to be an independent correlate of mortality,3,5 bivalirudin failed to reduce 1-year mortality.19 Two findings in the present analysis could shed light on that observation. First, although bleeding in high- and low-risk subsets correlates with prognosis—specifically mortality at 30 days and 1 year—a reduction in the frequency of bleeding in low-risk groups, as demonstrated in the present study, may not lead to a detectable reduction in mortality in patients at such low risk of death. Secondly, the beneficial effects of bivalirudin in reducing the risk of bleeding may be offset by its lesser ability to prevent myocardial infarction, another important correlate of mortality.5

In an analysis of the overall ISAR-REACT 3 study population and in subgroup analyses, heparin was associated with a trend towards a lower 30-day incidence of myocardial infarction than bivalirudin. In particular, we found that in patients of >70 kg of body weight, bivalirudin was associated with a higher risk of myocardial infarction (5.4 vs. 4.0%). Interestingly, in the same subgroup of patients, bivalirudin did significantly reduce major bleeding (1.9 vs. 3.4%). Therefore, the advantages of bivalirudin in reducing bleeding in patients of greater body weight might be offset by a greater risk of myocardial infarction in such patients. Body weight is an important determinant of the degree of anticoagulant effect when both bivalirudin and unfractionated heparin are administered, and current guidelines recommend a weight-adjusted initial bolus for both drugs.1 These data suggest that the optimal dosing of both these anticoagulants may not yet be known.

The most important limitation of the present study is that this analysis was a post hoc analysis of the ISAR-REACT 3 trial and, therefore, it is subject to limitations inherent to all such analyses. Such analyses should be interpreted with caution because of the potential impact of multiple testing. In spite of these limitations, the present analysis advances the understanding of ischaemic and bleeding profiles after treatment with bivalirudin or unfractionated heparin following 600 mg loading dose of clopidogrel in patients undergoing elective PCI.

This study showed that bivalirudin and unfractionated heparin have differential effects on the risk of bleeding, and the risk of myocardial infarction is various subsets of patients. Although the subsets of patients at higher risk for bleeding can be easily identified, bivalirudin does not seem to offer any specific advantage in these high-risk patients; rather, it appears to reduce bleeding mostly, when compared with heparin alone, in patients at low risk of bleeding. There appears to be an inverse relation; the greater the reduction of bleeding risk with bivalirudin, the lower the protection against myocardial infarction with this drug.


Supported in part by Nycomed Pharma, Unterschleißheim, Germany, and by a grant from Deutsches Herzzentrum, Munich, Germany (KKF 1.1-05, 984323).

Conflict of interest: A.K. reports receiving lecture fees from Bristol-Myers Squibb, Cordis, Lilly, Medtronic, and Sanofi–Aventis; R.A.B., being a research fellow of the Irish Board for Training in Cardiovascular Medicine, sponsored by A. Menarini Pharmaceuticals; G.R. reports receiving lecture fees from Boston Scientific and Cordis and grant support from Bristol-Myers Squibb; and P.B.B., serving as a consultant to Daiichi Sankyo–Lilly and PlaCor. No other potential conflict of interest relevant to this article was reported.


  • ClinicalTrials.gov Identifier: NCT00262054


View Abstract