OUP user menu

Risk of intracranial haemorrhage with combined fibrinolytic and glycoprotein IIb/IIIa inhibitor therapy in acute myocardial infarction

S. Savonitto, P.W. Armstrong, A.M. Lincoff, G. Jia, C.A. Sila, J. Booth, P. Terrosu, C. Cavallini, H.D. White, D. Ardissino, R.M. Califf, E.J. Topol
DOI: http://dx.doi.org/10.1016/j.ehj.2003.07.004 1807-1814 First published online: 2 October 2003


Background Intracranial haemorrhage is an important limitation to pharmacologic reperfusion therapy for acute myocardial infarction. The combination of a glycoprotein IIb/IIIa inhibitor, half-dose plasminogen activator and reduced-dose heparin has been evaluated as an alternative to standard fibrinolytic therapy in this setting.

Methods and results We evaluated the relation between univariate and multivariate predictors of intracranial haemorrhage and the effect of treatment with either reteplase alone (10U bolus twice, 30min apart) with standard-dose heparin (5000U bolus followed by an infusion of 1000Uh−1for patients ≥80kg and 800Uh−1for those <80kg) or combination therapy with abciximab (0.25mg/kg bolus and 0.125μg/kg/min for 12h) and half-dose reteplase (two boluses of 5U 30min apart) with reduced-dose heparin (60Ukg−1bolus, maximum 5000U, followed by an infusion of 7Ukg−1h−1) in the 16 588 patients randomized in the GUSTO V trial. Overall, the incidence of intracranial haemorrhage was similar in the two groups (0.6% vs 0.6%; OR 1.05, 95% CI 0.71, 1.56). The median (25th–75th) time from drug administration to intracranial haemorrhage was 5.5 (3.4–11) hours with combination therapy and 9.2 (5.9–22) hours with reteplase (P=0.048). Among the multivariable predictors of intracranial haemorrhage, only age showed a significant interaction with treatment effect (age per treatment interaction chi-square 4.60, P=0.032) with a lower risk of combination therapy for younger patients and a higher risk for the elderly.

Conclusions Although no additional risk of intracranial haemorrhage has been observed with combination therapy in the whole population, a significant age pertreatment interaction exists, with a lower risk with combination therapy in younger patients, and a higher risk in the elderly.

  • Fibrinolysis
  • Stroke
  • Myocardial infarction
  • Heparin
  • Abciximab

1 Introduction

Fibrinolytic therapy has been clearly shown to reduce mortality in acute myocardial infarction (MI)1,2and is the most widely available reperfusion therapy.3However, its net clinical benefit, defined as death and disabling stroke, is modest due to its suboptimal clot-lysing efficacy4and the associated risk of intracranial haemorrhage (ICH).5,6An increase in the risk of bleeding, particularly intracranial, has so far limited attempts to improve the efficacy of pharmacological reperfusion therapy by modifying the plasminogen activator (alteplase and its derivatives)7,8or the antithrombin agent (heparin or a direct thrombin inhibitor);9,10however, limited data exist on the effect of potentiating the antiplatelet component of fibrinolytic therapy.

Angiographic studies have shown that a higher reperfusion efficacy can be achieved, both in the epicardial artery and in the myocardium, by combining a glycoprotein (GP) IIb/IIIa blocker, with reduced doses of both heparin and the plasminogen activator.11–15This finding was accompanied by the hope that an increased efficacy could be obtained without necessarily paying the toll of an increased risk of ICH16for three reasons: first, treatment with GP IIb/IIIa blockers in the absence of fibrinolytics amongst several thousands of patients in controlled trials of percutaneous interventions and non-ST-elevation acute coronary syndromes has not increased the risk of ICH compared to placebo;17–20second, reducing the heparin dosage with fibrinolysis has beenaccompanied by lower rates of ICH;21third, a lower dose of plasminogen activator has been associated with lower bleeding risk.22,23

The GUSTO V trial randomized 16 588 patients to either abciximab, in combination with reduced-dose heparin and half-dose reteplase, or standard fibrinolytic therapy with reteplase and unfractionated heparin.24The study was powered to detect a significant difference in mortality at 30 days between the two treatments, but also provides an opportunity for evaluating the risk of stroke associated with this novel reperfusive approach to MI.

2 Methods

2.1 Patient population

The study design, randomization procedures and inclusion/exclusion criteria have been previously published in detail.24Patients with an acute ST elevation MI with onset <6h prior to enrolment were eligible if they were not candidates for immediate catheter-based reperfusion and had no contraindications for thrombolysis. Of note for the present study, patients <18-year-old were excluded, but there was no upper age limit for inclusion; patients with blood pressure >180mmHg systolic or >110mmHg diastolic recorded at any time after symptom onset or stroke within the past 2 years were excluded.

2.2 Drug regimens

Patients were randomized to receive either one of two open label study drug kits that consisted of: (1) reteplase at standard dosing of 10U bolus twice, 30min apart, or (2) the combination of a standard abciximab infusion of 0.25mg/kg bolus and 0.125μg/kg/min (maximum of 10μg/min) for 12h and half dose reteplase (two boluses of 5U 30min apart). Unfractionated heparin was to be administered for ≥24h at an infusion rate adjusted by a nomogram to achieve an aPTT between 50 and 70s. The initial dosing for heparin varied according to the study drug assignment: for reteplase alone it was a 5000U bolus followed by a 1000U/h infusion patients ⩾80kg or 800U/h in patients <80kg; in the combined arm patients received a 60U/kg heparin bolus (maximum 5000U) followed by an infusion of 7U/kg/h. The first aPTT measurement was to be performed 6h after bolus administration, and then at 6-h intervals according to the nomogram. If any therapeutic heparin bolus ≥60U/kg had been given within 1h of randomization or the patient had been on therapeutic heparin infusion prior to randomization, no study heparin bolus was to be administered; the heparin infusion had to be set at 7U/kg/hr and an aPTT be drawn prior to administration of reteplase. As soon as the result of the aPTT test was available, the rate of heparin infusion was to be adjusted according to the nomogram. Aspirin was administered as 150–325mg orally or 250–500mg intravenously at the time of randomization and continued at 75–325mg orally daily for the remainder of the study period. Administration of non-study abciximab or another GP IIb/IIIa inhibitor concomitantly with the abciximab study drug in the combination therapy arm was strictly prohibited, whereas it was left to the discretion of the investigator under all other circumstances. Heparin dosing during early PCI was left to the discretion of the investigator for patients receiving no abciximab, whereas specific guidelines were provided for those receiving abciximab: for these patients, heparin dosing guidelines differed according to whether the patient was receiving the drug per protocol (during the first 12h after randomization in the combination arm of the study) or as an adjuvant therapy under all other circumstances. In both cases, the guidelines assumed that the patient had been maintained on a heparin infusion since randomisation, and the target ACT during intervention was 200s to be maintained using weight adjusted boluses (maximum 3000U).

2.3 Definition of stroke and adjudication of events

A stroke was defined clinically as an abrupt neurologic deterioration resulting in death or lasting for >24h. All of the patients with suspected stroke were to undergo complete evaluation including either brain computed tomography or magneticresonance imaging: clinical notes, discharge summaries, neurological or neurosurgical consultation notes, autopsy reports, and brain imaging studies were collected. A Clinical Event Committee, masked to actual therapy assignment, reviewed all of the documents, with classification of whether a stroke had occurred, and the type of stroke. The definition of an ICH included intraparenchymal, subdural, epidural, intraventricular or subarachnoid haemorrhage, and haemorrhagic transformation of a bland cerebral infarct. Non-haemorrhagic stroke was categorized if there was clinical evidence of a stroke and no evidence of intracranial haemorrhage on brain imaging studies or at autopsy. The uncertain classification was used if there was clinical evidence of a stroke and no brain imaging or autopsy data were available to determine the type of stroke.

The disability status of surviving stroke patients was assessed at discharge and between day 30 and 45 by a physician unaware of treatment assignment, using the modified Rankin scale: for the purpose of the present study, any grade ≥2 as a consequence of a cerebrovascular event occurred after randomization was defined as a disabling stroke.

2.4 Statistical analysis

All of the randomized patients were included in analyses. For univariate analyses, Pearson's chi-square test was used for categorical variables; for continuous variables, the t-test was used for those normally distributed and the Mann–Whitney–Wilcoxon rank sum test otherwise. A multiple logistic regression model of ICH was developed to estimate adjusted odds ratios for randomized treatment and other significant factors. Stepwise variable selection method was used in building the logistic regression. Since the vast majority of the events happened within a few hours after randomization and are likely to have influenced subsequent therapy, only prerandomization concomitant treatments were entered in the logistic regression model. For the continuous variables, if an effect was not linear, linear spline transformations were used. All two-way interactions between the significant variables included in the model were explored and the only significant interaction between age and treatment was incorporated into the final model. For different ages, the treatment effect on ICH was calculated using the parameter estimates and their variance and covariance from the logistic regression model (Fig. 1). For all statistical analyses, a P value of Math0.05 derived from Ward's chi-square statistics was used as the level for statistical significance and calculations were performed using SAS version 8 software.

Fig. 1

Relationship of age and relative risk of intracranial haemorrhage with abciximab plus reteplase (combo) compared to reteplase only. Point estimates and 95% confidence intervals from regression analysis. The raw average ICH rate at different age levels was estimated using univariate logistic regression; then, the ICH event rate by treatment is derived by the odds ratio for ICH from the multivariate regression model.

3 Results

Of the 366 cases with suspected stroke reviewed by the Clinical Event Committee (2.2% of the randomized patients) 154 were confirmed as having had a stroke in hospital, 73 with reteplase (0.9%) and 81 with reteplase and abciximab (1.0%, odds ratio 1.10, 95% CI 0.80–1.51, P=0.37). As reported in Table 1, 101 of the strokes were haemorrhagic, 42 non-haemorrhagic and 11 of unknown type, without differences between treatments. The mode of confirmation of ICH was a CT scan in 96 cases, magnetic resonance in two and autopsy in three cases.

View this table:
Table 1

Incidence of stroke by treatment

TotalReteplaseReteplase and abciximabOdds ratio (95% CI)P-value
Reviewed by CEC366 (2.2)171 (2.1)195 (2.3)
No event174 (1.0)81 (1.0)93 (1.1)
Insufficient documentation38 (0.2)17 (0.2)21 (0.3)
Confirmed event154 (0.9)73 (0.9)81 (1.0)1.10 (0.80–1.51)0.55
Intracranial haemorrhage101 (0.6)49 (0.6)52 (0.6)1.05 (0.71–1.56)0.79
Intraparenchymal haemorrhage85 (0.5)43 (0.5)42 (0.5)
Other intracranial bleedingb16 (0.1)6 (0.1)10 (0.1)
Cerebral infarction with haemorrhagic transformation0 (0.0)0 (0.0)0 (0.0)
Non-haemorrhagic stroke42 (0.3)21 (0.3)21 (0.3)
Stroke, type uncertain11 (0.1)3 (0.0)8 (0.1)
30-day mortality
Patients with ICH65/101 (64)27/49 (55)38/52 (73)2.05 (0.82–5.17)0.127a
Patients with non-ICH stroke22/53 (42)10/24 (42)12/29 (41)
Patients with no stroke869/16 434 (5.3)451/8187 (5.5)418/8247 (5.1)
Overall combined event rate
Death+nonfatal stroke1023/16 588 (6.17)524/8260 (6.34)499/8328 (5.99)0.346
Death+nonfatal ICH992/16 588 (5.98)510/8260 (6.17)482/8328 (5.79)0.294
Death+disabling stroke993/16 588 (5.99)508/8260 (6.15)485/8328 (5.83)0.376
  • CEC stands for Clinical Event Committee.

  • a P value adjusted for other independent predictors of mortality in the ICH population (age, systolic blood pressure and heart rate): note that patients with ICH were older in the abciximab plus reteplase (median 73 years, interquartile range 68–79) group than in reteplase group (median 68 years, interquartile range 62–75).

  • b Includes subdural, subarachnoid, epidural and intraventricular bleeding.

The median (25th–75th) time from drug administration to ICH was 9.2 (5.9–22) hours with reteplase and 5.5 (3.4–11) hours with abciximab plus reteplase (P=0.048). The median time from drug administration to non-haemorrhagic stroke was 34 (4.7–79) hours with reteplase and 100 (39–131) hours with abciximab and reteplase (P=0.05).

Thirty-day mortality in patients with ICH was not significantly different between treatments (Table 1). Non-fatal, disabling strokes occurred in 0.3% of reteplase-treated and 0.2% of abciximab-plus-reteplase-treated patients (P=0.37). The composites of mortality and non-fatal stroke or mortality and non-fatal ICH were not different between groups (Table 1). No differences between treatments were observed at 30–45 days in terms of disability measured according the modified Rankin scale.

3.1 Risk factors for intracranial haemorrhage

Among the baseline demographic and clinical characteristics, age, weight, gender, treated hypertension and current smoker were associated with ICH at univariate analysis. Age, gender and baseline diastolic blood pressure remained highly significant predictors in the logistic regression model. Among the concomitant medications, treatment with digitalis or nitrates before randomization was associated with significantly higher ICH rates at univariate analysis. However only digoxin was associated with a significantly higher risk in the logistic regression model.

The administration of heparin 1 to 4h prior to randomization (a median dose of 5000U to 783 patients in the whole population) was a significant predictor of ICH at multivariate analysis (ICH rate was 0.55% among the patients who did not receive heparin and 1.21% among those who received any heparin prior to randomization, P=0.041). A second heparin bolus was administered to 98 of the 783 patients who had received pre-randomization heparin: none of these patients had an ICH, whereas nine out of the 685 not receiving a second bolus did (P=0.254). No interaction between pre-randomization heparin and randomized treatment was observed in terms of subsequent ICH risk. The weight-adjusted heparin dose was a significant correlate in the univariate, but not the multivariate analysis. Neither the heparin bolus dose, nor its infusion rate post randomization, were significantly different between patients with or without ICH. The median aPTT peak value was slightly lower in patients with ICH compared to those without ICH (71s [41–110], vs 76s [57–111], P=0.044), and significantly more patients with ICH had peak aPTT values lower than the target of 50 to 70s (33.7% vs 15.7%, P<0.001). The median value of peak aPTT was significantly lower with abciximab and reteplase compared to reteplase only (66s [52 to 94] vs 88s [64 to 121], P<0.001). The median time from drug administration to peak aPTT was significantly longer with abciximab and reteplase compared to reteplase only (18.7h [6.7–33.4] vs 7.1h [5.9–21.5], p<0.001).

3.2 Risk of intracranial haemorrhage and treatment effect

At multivariate analysis, a significant interaction between age and randomized treatment was observed (Table 2). As shown in Fig. 1, the direction of the risk of ICH according to randomized treatment changes with age, with a lower risk for abciximab and half-dose reteplase in younger age and a higher risk in the elderly. Besides age, no other variable showed any interaction with randomized treatment for the risk of ICH.

View this table:
Table 2

Independent baseline demographic, clinical and pharmacologic predictors of intracranial haemorrhagea

VariableOdds ratioOdds ratio CIChi squareP-value
Age in 10 years1.5241.183, 1.96310.6560.001
Interaction: age×treatment1.4671.020, 2.1094.2700.039
Female gender1.8571.226, 2.8128.5380.004
Digitalis 48h pre-randomization2.2101.035, 4.7184.2010.040
Heparin 1–4h pre-randomization2.1051.052, 4.2124.4280.035
Baseline diastolic BP (<90mmHg)1.0371.016, 1.05912.3320.001
Baseline diastolic BP (≥90mmHg)0.9560.902, 1.0132.3590.125
  • C statistics for the model=0.758; chi square overall model=94.304; P-value <0.001.n=16 548/16 588; number of events events 101/101.

  • a BP=blood pressure is considered as a continuous variable modeled using a two-piece linear spline: the risk of ICH increases significantly as diastolic blood pressure increases until 90mmHg; after 90mmHg there is a non-significant effect of the diastolic BP on the risk if ICH.

4 Discussion

The mechanism of ICH associated with fibrinolytic therapy for acute MI remains unclear. Radiologic and pathologic studies incriminate cerebral amyloid angiopathy, a small vessel degenerative disease mostly associated with longstanding hypertension and aging, as the most frequent anatomical abnormality underlying this complication.25–28Large clinical studies of fibrinolytic therapy have pointed in the same direction, identifying old age, hypertension and prior stroke as the most powerful independent predictors of ICH.5,6,29–33Of the three components of the contemporary pharmacologic reperfusion therapy, the plasminogen activator1,5,629–33and the thrombin antagonist9,10,21have been clearly shown to increase the risk of ICH in a dose-dependent manner. However, antiplatelet agents have not been associated with intracranial bleeding, either when used alone34–36or in combination with heparin17–20or a fibrinolytic agent.37Therefore, a pharmacologic reperfusion regimen with increased antiplatelet potency and lower antithrombin and plasminogen activator dosages, which in angiographic trials had produced a high rate of coronary patency,11–15seemed very attractive also from the standpoint of potential safety.16,38Phase two studies with abciximab, half-dose alteplase or reteplase and various doses of unfractionated heparin showed a higher rate of early patency in the infarct-related artery compared to traditional fibrinolytic therapy, and comparable bleeding risk,11–13but lacked enough power to evaluate the risk of ICH. A more robust estimate of this risk is derived from the GUSTO V trial, which enrolled almost 16 600 patients having as primary endpoint the evaluation of the 30-day mortality.

The risk of intracranial haemorrhage in GUSTO V was 0.6%, one of the lowest rates observed in clinical trials of thrombolytic therapy since the era of streptokinase, perhaps due to careful exclusion of patients with previous stroke or uncontrolled hypertension. No differences in overall stroke rates were observed between reteplase monotherapy and combination therapy. However, advanced age, the most powerful independent predictor of stroke, was not an exclusion criterion for the study, thus allowing detection of a significant interaction between age and randomised treatment with a lower risk of ICH in younger patients and a higher risk in the elderly randomised to reteplase plus abciximab. The lower risk observed in younger patients is concordant with the expected benefit of reducing the dosage of heparin and the plasminogen activator, and leads to hypothesize that the pathogenesis of iathrogenic ICH in young patients differs from that of the elderly.

There may be several reasons for the finding of an interaction between age and treatment. First, there may be an aspecific response of the elderly to a more aggressive and faster fibrinolytic regimen, as has been shown with double bolus r-tPA in the COBALT study7or with lanoteplase (which causes an acute increase in aPTT) in the InTIME-II study.8The faster recanalization of the infarct-related artery in comparison with standard regimens11–13has been clearly observed in angiographic studies of patients with acute MI, and is one of the expected benefits of this treatment. However, in the present study, this regimen also resulted in a significantly shorter time to ICH, with an increased risk in elderly patients.

Whether this greater lytic effect depends on abciximab or on an amplification of the heparin effect is unclear and cannot be clarified by the GUSTO V data, as the first aPTT was to be measured six hours after the heparin bolus. Since the heparin bolus dose and initial infusion rates were lower in the abciximab group, the time to peak aPTT was significantly longer and the peak aPTT significantly lower with combination therapy compared with reteplase only. However, the aPTT likely reflects only a partial view of the coagulation system activation encountered in vivo, particularly after multiple pharmacological manipulations with combination fibrinolytic therapies. A finding suggesting a causative role of heparin for ICH is the higher risk observed in patients who had received heparin within the 4h prior to randomization: however, this observation must be taken with caution, since it was not a pre-specified hypothesis and prior heparin was not a randomized treatment. Nevertheless, the experience with abciximab in the catheterization laboratory has clearly demonstrated a lower bleeding risk using lower heparin doses,39suggesting a potentiation of the bleeding risk when abciximab and heparin are used together. In the TIMI 14 study, a heparin dose lower than that selected for GUSTO V (30 unitskg−1bolus and 4 unitskg−1h−1infusion) resulted in a major haemorrhage rate of 1%, lower than the 7% observed with combination therapy and regular dose heparin.11The higher heparin dosage used in GUSTO V was selected, however, because the very low dose tested in TIMI 14 was associated with a slightly lower TIMI 3 perfusion rate at 90min. Overall, there is increasing evidence that reducing the heparin dose, and particularly the bolus dose, is associated with lower risk of major bleeding and ICH.21,40A recent additional concern regarding the administration of a low molecular weight heparin bolus dose in association with a plasminogen activator comes from the preliminary results of the ASSENT-III PLUS study41in which a 3000U intravenous bolus of enoxaparin, followed by a 100U/kg subcutaneous dose, in conjunction with full dose tenecteplase was associated with an excess risk of ICH, particularly in patients older than 75 years. Based on these considerations, current recommendations point to reducing the heparin bolus dose particularly in the elderly and low weight patients.42

4.1 Limitations of the study

Besides the limitations of our analysis of the heparin-derived risk based on the aPTT values discussed above, it should be noted that our multivariate analysis focused on a pre-specified secondary endpoint of the GUSTO V trial: whereas the power of the study was calculated on a 7% 30-day mortality rate, in the present analysis we are dealing with a 0.6% ICH rate, with consequent limitations. This limitation may play a role in overlooking other possible risk factors or drug interactions. Although we were able to identify a lower relative risk of ICH in younger patients treated with combination therapy, the absolute overall risk in this population is low thus translating in a small absolute difference; on the other hand, even a small increase in relative risk in the elderly may translate in a sizably higher number of ICHs.

4.2 Clinical implications

The overall finding of no difference in the risk of ICH between reteplase alone and abciximab plus half-dose reteplase in GUSTO V must be interpreted in the light of a significant interaction between age and treatment effect. Reducing the dosage of reteplase and heparin by combining abciximab resulted in a lower risk of ICH in younger patients up to the age of 60–65 years. This finding may be important for patients with large infarcts when primary angioplasty is not readily available, and particularly in view of current efforts to identify the optimal treatment for ‘facilitated’ angioplasty, for whom combination therapy with a GP IIb/IIIa blocker and reduced dosages of a plasminogen activator and heparin may be a promising option. On the other hand, elderly patients are at higher risk of ICH when treated with abciximab and reteplase compared to reteplase alone, and derive no mortality benefit:24these data are consistent with other trials showing no benefit and increased risk of ICH in the elderly with any attempt to increase thrombolytic efficacy.10,43

In the fast growing elderly population, mortality for acute myocardial infarction remains high and improvement of reperfusion therapy in this age group has to be set as a priority target for clinical research. Meta-analyses from randomized trials have shown a modest but persistent benefit from fibrinolytic therapy compared to control in this population,1,44but they have been challenged by the conflicting information derived from large observational databases.45,46Promising data on primary angioplasty in the elderly47need to be confirmed by larger prospective studies, and the wide application of catheter-based reperfusion therapy for the treatment of myocardial infarction in these patients will certainly represent a major logistic challenge.


The data reported in the present manuscript were presented at the XXIV Congress of the European Society of Cardiology, Berlin 31 August – 4 September 2002. The GUSTO V trial was sponsored by Eli Lilly and Centocor.


View Abstract