OUP user menu

Comparative efficacy of thrombolytics in acute myocardial infarction: a systematic review

Y. Dundar , R. Hill , R. Dickson , T. Walley
DOI: http://dx.doi.org/10.1093/qjmed/hcg016 103-113 First published online: 1 February 2003


Background: The comparative clinical effectiveness of new (reteplase, tenecteplase) vs. older (alteplase, streptokinase) thrombolytic agents in the treatment of acute myocardial infarction is uncertain.

Aim: To examine 30–35 day mortality and major adverse effects of thrombolytic agents in the treatment of acute myocardial infarction.

Design: Systematic review of randomized controlled trials comparing the clinical efficacy of included drug regimens.

Methods: We searched MEDLINE, EMBASE, Science Citation Index/Web of Science from 1980 to December 2001, and the Cochrane Library (2001, Issue 4). Reference lists of included studies and a number of medical journals were hand searched. Randomized controlled trials that compared any two of the included drugs provided to patients in the early stages of acute myocardial infarction, were included. Outcome measures included: mortality, bleeding, stroke, reinfarction, allergy and anaphylaxis.

Results: We found 14 studies, total study population 142 907. For available comparisons (all alteplase vs. streptokinase, reteplase vs. streptokinase or alteplase, tenecteplase vs. alteplase), meta‐analysis showed no significant differences in mortality at 30–35 days. The GUSTO‐I study showed an apparent benefit of accelerated alteplase over streptokinase, but its inclusion or exclusion made little difference. Total stroke and haemorrhagic stroke rates were lower for streptokinase than for all alteplase combined (total stroke, OR 1.29, 95%CI 1.13–1.46; haemorrhagic stroke OR 1.83, 95%CI 1.14–2.93).

Discussion: All thrombolytic drugs appear to be of similar efficacy in reducing mortality, and the apparent benefits of accelerated alteplase in GUSTO‐I are consistent with this. Whether accelerated alteplase is sufficiently different from other regimens of administering alteplase to be excluded from a meta‐analysis, and whether more weight should be placed on a meta‐analysis than on a single trial, are matters for debate.


The benefits of thrombolytic therapy in patients with acute myocardial infarction are well established, in the meta‐analyses by Yusuf et al.1 and by the Fibrinolytic Therapy Trialists (FTT) Collaborative Group who showed that thrombolytic therapy decreases mortality at 35 days by 1.9%.2 Later trials compared the effectiveness of a variety of agents, but especially streptokinase and alteplase.

The ‘first generation’ thrombolytics had clinical disadvantages such as low specificity for fibrin, increased risk of allergic reactions (in particular with streptokinase) and short half‐life. Newer thrombolytic agents such as reteplase and tenecteplase have been developed with potential advantages that include: prolonged half‐life, increased fibrin specificity and increased resistance to inhibition by plasminogen activators. However, these laboratory‐measured advantages may not translate into measurable clinical benefits. For instance, the new thrombolytic drug lanoteplase was withdrawn from development as a result of in an increased incidence of intracranial haemorrhage.3,,4

We therefore conducted a systematic review to examine the comparative effectiveness of older and newer agents used for early thrombolysis, so as to allow recommendations to be made to service users. If there were appropriate head‐to‐head comparisons between all thrombolytic drugs, then drawing conclusions from such a review would be simple. However, such direct comparisons do not exist, and therefore indirect comparisons were required to inform clinical guidance. This required a two‐stage process: the first evaluating and comparing the evidence from clinical trials, and the second an indirect and therefore more speculative comparison where interpretation of data was more difficult.

This paper presents the results of the systematic review, the first of these two processes; the second is described in the accompanying commentary.5


The review was restricted to those drugs currently available in the UK, i.e. streptokinase, alteplase, reteplase and tenecteplase. Two other drugs, anistreplase and urokinase, are licensed in the UK but are not available for commercial reasons.


The search strategy covered the period from 1980 to December 2001, and included the following electronic databases: MEDLINE, EMBASE, Science Citation Index/Web of Science and The Cochrane Library (2001, Issue 4). Search terms used were ‘myocardial infarction’, ‘heart infarction’ and ‘thrombolysis' combined with drug terms (e.g. alteplase (t‐PA) reteplase, streptokinase and tenecteplase). In addition, reference lists of included studies were examined and a number of medical journals were hand‐searched to identify other potentially relevant papers.

Selection of studies

Studies included in the review were randomized controlled trials that compared the effectiveness of any two of the studied thrombolytic agents used in patients experiencing acute myocardial infarction. The review focused on clinical outcomes, and included mortality, bleeding, stroke, re‐infarction, allergy and anaphylaxis.

Quality assessment

Two reviewers assessed the methodological quality of included studies independently using the criteria based on the NHS Centre for Reviews and Dissemination (CRD) Report No. 4.6 Components of study quality included in this tool are: study randomization, baseline comparability, eligibility criteria, blinding, number of and reasons for withdrawals, and whether an intention‐to‐treat analysis was used.

Data extraction

Data were independently extracted by one reviewer and checked by a second.


Meta‐analysis used RevMan 4.1.1 (Cochrane collaboration). Treatment effects are presented using odds ratios (OR) with corresponding 95%CIs, using a random effects model. Analysis of data included mortality (30–35 day), stroke (total and haemorrhagic), major bleed and reinfarction.


A total of 162 references were identified, of which 20 studies (reported in 50 publications) met the inclusion criteria. Of these, four were dose‐ranging trials and two were comparisons of the same drug. The review therefore includes 14 studies,7–,21 involving a total study population of 142 907 participants. Two studies (ISG and GISSI‐2) provided combined data and this combination of data was maintained in the review. All the studies were conducted in hospital settings.

Study characteristics

Studies are summarized in Table 1 and ranged in size from 122 to 41 299 patients. Ten studies compared alteplase and streptokinase. Of these, one7 used accelerated administration of alteplase, which includes a bolus dose, followed by infusion over 90 min. One study compared alteplase to tenecteplase, two studies compared accelerated alteplase with reteplase, and one study compared streptokinase with reteplase. No study provided a direct comparison of reteplase and tenecteplase.

View this table:
Table 1 

Characteristics of studies


A formal meta‐analysis was performed for those comparisons where there was more than one relevant trial. The inclusion of GUSTO I in such a meta‐analysis is controversial, since it is argued the accelerated regimen is sufficiently different from the older regimen used in most trials. Therefore, data comparing streptokinase and alteplase are presented in two analyses, one including and one excluding the GUSTO I study. There is no statistical evidence for heterogeneity between studies in either analysis. An inability to confidently extract data related to bleeding events from GUSTO I precluded its inclusion in the analysis.

The main results of the meta‐analysis were as follows:

All alteplase vs. streptokinase (Figure 1)

No difference in mortality or reinfarction. Total stroke and haemorrhagic stroke rates were lower in streptokinase group.

Figure 1. 

Meta‐analyses: all alteplase (including accelerated and non‐accelerated alteplase regiments) vs. streptokinase. t‐PA, alteplase; SK, streptokinase.

Alteplase excluding accelerated alteplase vs. streptokinase (Figure 2)

No difference in mortality. In the streptokinase group, there was a lower incidence of total stroke and haemorrhagic stroke. Major bleed and reinfarction rates were lower in the alteplase group.

Figure 2. 

Meta‐analyses: alteplase excluding accelerated alteplase (no GUSTO I) vs. streptokinase. t‐PA, alteplase; SK, streptokinase.

Accelerated alteplase vs. reteplase (Figure 3)

No differences in mortality, total stroke, haemorrhagic stroke, major bleeds or reinfarction. Data on haemorrhagic stroke, major bleeds or reinfarction were only available for GUSTO‐III, but this accounts for 98% of the patients studied.

For the following two comparisons, there was only one study reported: these are summarized in Table 2.

Figure 3. 

Meta‐analyses: accelerated alteplase vs. reteplase. t‐PA, alteplase; r‐PA, reteplase.

View this table:
Table 2 

Meta‐analyses: single study comparisons a. Accelerated alteplase/tenecteplase—ASSENT‐2

View this table:


Accelerated alteplase vs. tenecteplase

No differences in mortality, total stroke, haemorrhagic stroke or reinfarction. Fewer major bleeds with tenecteplase.

Reteplase vs. streptokinase

No differences in mortality, total stroke, major bleeds. There was a lower incidence of haemorrhagic strokes in the streptokinase group.

Adverse events

There are substantial differences in the definition of bleeding events reported in the studies. Therefore, as would be expected, the reported rates for a ‘major bleed’ varied between 0%17 and 18% for streptokinase.8 We found a slightly higher risk of major bleed associated with the use of streptokinase than with alteplase (Figure 1).

There was a significantly higher risk of stroke, largely accounted for by an increase in the incidence of haemorrhagic stroke, associated with the use of alteplase compared to streptokinase. This difference was statistically significant in both meta‐analyses (Figures 1 and 2) and in GUSTO I (alteplase 0.72%, streptokinase 0.52%). Reteplase was also associated with an increased tendency to stroke and a significant increase in haemorrhagic stroke when compared to streptokinase (Table 2). No differences were found between alteplase and reteplase in GUSTO III or between alteplase and tenecteplase in ASSENT‐2.

When compared with other thrombolytic drugs, streptokinase was associated with a higher incidence of allergic reactions, which also included anaphylaxis.

There was no significant difference between any of the drugs with regard to reinfarction rates.


Six studies7,11–13,18,19,,21 conducted subgroup analyses of mortality at 30–35 days, which were based on the three most common subgroups of patients including age, infarct location and time from symptom onset.

There were no consistent differences with respect to drugs in these subgroups. There were some apparent differences between reteplase and alteplase in GUSTO III (better mortality benefit in late‐treated patients with alteplase) and between tenecteplase and alteplase in ASSENT‐2 (absolute difference of 2% in 30‐day mortality favouring tenecteplase in patients treated within 4 h of symptom onset).


There has been no previous systematic comparison of thrombolytic agents, although a Cochrane review is underway (Bijsterveld NR, personal communication, 2002). Organisations such as the American College of Chest Physicians (ACCP)22 and the American Heart Association23 provide regular updates of guidelines for care in this area. Their method of reviewing evidence has evolved, but their methods of identification, quality assessment and data extraction and analysis do not follow the current gold standard as set out by such groups as the Cochrane Collaboration.

The conclusions from our review may seem simple—i.e. that no one drug is more effective than any other—but are potentially confusing and require interpretation. In this Discussion section, we address the controversies surrounding the use of meta‐analysis in this area, and the weaknesses of our work. The accompanying Commentary5 uses the results of this review to expand the evidence‐based comparisons into areas where direct head‐to‐head comparisons of treatments are not available. It also deals with issues related to what constitutes equivalence in thrombolytic therapy.

Meta‐analysis is often conducted as part of systematic reviewing, but has been controversial in cardiology. An example often used by the proponents of meta‐analysis is that unequivocal evidence of the effectiveness of thrombolysis for acute myocardial infarction was available from research by 1971.24,,25 However it took almost 20 years before the treatment was approved by the FDA. Even then, implementation into clinical practice was driven not by meta‐analysis but by the publication of two large trials, GISSI26 and ISIS‐2.27

Conversely, the opponents of meta‐analysis point out its flawed use in areas such as use of magnesium in acute myocardial infarction. In this instance, a meta‐analysis supported its use,28 but a single large trial reversed this conclusion.29 In retrospect, the meta‐analysis was flawed, probably by failure to identify all relevant studies; this in turn arose from publication bias affecting unfavourable trials,30 at a time when the limitations of the technique were less well understood. Thus the role of meta‐analysis versus a single large trial has been heatedly debated.

Cardiology, in contrast to other areas of medicine, has an established tradition of very large well‐focused, clinical trials designed to provide definitive answers to questions.31 Such trials have other advantages—their size allows precision in defining the size of any benefit, and to some extent guarantees generalizability. Meta‐analysis may therefore seem less important or useful. This is reflected in our review where the meta‐analyses, with the exception of the comparison of alteplase and streptokinase, are all dominated by one large trial. It might be argued that accelerated alteplase is quite different in effect from standard alteplase, and should not be included in this meta‐analysis which dilutes the evidence of its particular benefits: this is considered more fully in the accompanying Commentary.5

The benefits and limitations of meta‐analysis have been extensively discussed.32,,33 This review highlights the limitations in two areas. The first is that the primary focus is placed on the impact of treatment on mortality, well reported in all trials, and may exclude the assessment of adverse events such as stroke and major bleeding events, which are variably reported. Second, a meta‐analysis may mislead if it groups disparate trials together inappropriately. Conversely, an excessive focus on a single trial such as GUSTO‐I may mislead if it excludes other relevant studies and does not consider the totality of the evidence available.34 This is well illustrated in our results, where the GUSTO‐I study shows an apparent benefit of accelerated alteplase over streptokinase, but the meta‐analysis shows no benefit and no heterogeneity (although the ability to detect heterogenetity in such analyses is limited).

In conclusion, this systematic review suggests that all thrombolytic drugs are of similar efficacy in reducing mortality, and that the apparent benefits of accelerated alteplase in one trial are still consistent with this. There are differences in rates of stroke, favouring streptokinase over newer drugs. The key questions remaining are whether accelerated alteplase is sufficiently different from other regimens of administering alteplase to be excluded from a meta‐analysis, and whether we should place more weight on a meta‐analysis rather than on a single trial. This is a matter for judgment and debate, and the implications of these are considered in the accompanying Commentary.5


The review described in this article is based on research published as a Health Technology Assessment (HTA) Monograph commissioned by the (UK) NHS R&D HTA programme. The views expressed in this article are those of the authors and not necessarily those of the HTA Programme, National Institute of Clinical Excellence or the Department of Health.

The authors are pleased to acknowledge the contributions of the members of the HTA review team: A Boland, A Bagust, A Haycox, R Mujica Mota and our Review Advisory Panel.


  • Address correspondence to Professor T. Walley, Department of Pharmacology and Therapeutics, University of Liverpool, 70 Pembroke Place, Liverpool L69 3GF. e‐mail: twalley{at}liv.ac.uk


View Abstract