4 Evidence

The Appraisal Committee reviewed the evidence from a number of sources (Appendix B).

4.1 Clinical effectiveness

4.1.1 In-hospital thrombolysis Fourteen randomised controlled trials (RCTs) comparing thrombolytic drugs were included in the review. Overall the studies were considered to be of excellent quality. In total, the trials involved over 142,000 patients, and five of the trials included over 10,000 patients each. The trials had similar inclusion criteria in terms of age (usually <70 or <75 years), ECG changes, duration of symptoms, and presentation within 6 hours of symptom onset. Five of the trials included between 12% and 26% of patients aged over 70–75 years. Women were under-represented in all of the studies. Primary endpoints included 30-day mortality, 90-minute artery patency/flow rates and left ventricular function. Secondary endpoints included bleeding, stroke, congestive heart failure, reinfarction, allergy and anaphylaxis. The results of the trials were also pooled in a meta-analysis. No direct trial comparisons between tenecteplase and streptokinase or between tenecteplase and reteplase have been undertaken, and only cautious conclusions can be drawn from the indirect comparisons that can be deduced from other studies.

Streptokinase Two placebo-controlled trials were instrumental in establishing the efficacy of streptokinase in reducing mortality. The GISSI trial (published in 1986) included 11,712 patients, and the ISIS-2 trial (published in 1988) included 17,187 patients. In the GISSI study, 21-day mortality was 10.7% in patients treated with streptokinase and 13% in those treated with placebo. This represents a statistically significant absolute reduction of 2.3% (risk ratio 0.81; 95% confidence ratio [CI] 0.72 to 0.9). In the ISIS-2 study, vascular mortality at 5 weeks was 9.2% in patients treated with streptokinase and 12% in those treated with placebo. This represents a statistically significant absolute reduction of 2.8%. These benefits were independent of those of early aspirin treatment.

Alteplase A meta-analysis of eight comparisons of standard alteplase with streptokinase found no significant difference between the two drugs in terms of mortality up to 35 days (odds ratio 1.0; 95% CI 0.94 to 1.06). A statistically significant difference in reinfarction rates in favour of alteplase was found (odds ratio 0.86; 95% CI 0.77 to 0.95). However, alteplase was associated with a statistically significant higher risk of stroke (odds ratio 1.37; 95% CI 1.16 to 1.62), due to a doubling in the risk of haemorrhagic stroke (odds ratio 2.13; 95% CI 1.04 to 4.36). However, streptokinase was associated with a statistically significant higher risk of major bleeds (other than stroke) than alteplase (odds ratio 0.81; 95% CI 0.68 to 0.97). The categorisation and reporting of major bleeding varied between the trials and so it is difficult to judge the clinical significance of these findings. The studies included in this meta-analysis used the standard alteplase administration regimen, whereas the GUSTO-I trial used the accelerated regimen and is the only trial to have demonstrated superiority between different thrombolytic agents. The GUSTO-I trial included over 40,000 patients. It found an odds ratio of 0.85 (95% CI 0.78 to 0.94) for 30-day mortality for accelerated alteplase compared with streptokinase, and an absolute reduction in mortality at 30 days of 1.0% (6.3% versus 7.3%; 95% CI 0.4% to 1.6%) in favour of accelerated alteplase. However, this benefit was balanced by a statistically significantly higher incidence of haemorrhagic stroke (odds ratio 1.42; 95% CI 1.05 to 1.91). Using a combined outcome measure of mortality and disabling stroke, the absolute advantage of accelerated alteplase over streptokinase was lower (0.9%; p = 0.006). Rates of bleeds (moderate or worse), allergic reaction, anaphylaxis, congestive heart failure, and sustained hypotension were statistically significantly lower in the group treated with accelerated alteplase. A further meta-analysis of nine comparisons of alteplase with streptokinase, including the findings of GUSTO-I (i.e. accelerated alteplase), found no significant difference between the two drugs in terms of mortality up to 35 days (odds ratio 0.94; 95% CI 0.85 to 1.04).

Reteplase Reteplase has also been compared with streptokinase in a study involving 5986 patients (the INJECT study). This study found an absolute difference of 0.5% (95% CI –1.98% to 0.96%) in 35-day mortality in favour of reteplase (not statistically significant). If it is accepted that a 1% difference in mortality is the limit of equivalence in thrombolytic therapy, this suggests that it is unlikely that reteplase is inferior to streptokinase. An alternative interpretation is that in terms of overall effects on mortality and disabling stroke reteplase may be inferior to streptokinase, as the trial also found a statistically significantly lower risk of haemorrhagic stroke (odds ratio 2.1; 95% CI 1.02 to 4.31) in the streptokinase group. However, the trial also found that the rates of heart failure (23.6% vs 26.3%, p<0.05) and allergic reactions (1.1% vs 1.8%, p<0.05) were statistically significantly lower in the reteplase group. Reteplase has also been compared with accelerated alteplase in one relatively small (n = 324) study that examined intermediate angiographic endpoints of coronary vessel patency (RAPID-2), and one larger study that examined patient-focused endpoints (GUSTO-III, n = 15,059). GUSTO-III was designed to test the clinical superiority of reteplase over accelerated alteplase, following the findings in RAPID-2 of better coronary artery patency with reteplase. However, GUSTO-III found no statistically significant difference between the two drugs, in terms of survival or adverse effects. The mortality rate at 30 days was 7.5% in the reteplase group and 7.2% in the accelerated alteplase group: an absolute risk reduction of 0.23% in favour of accelerated alteplase (95% CI –1.10% to 0.66%). Given the confidence limits, reteplase cannot be considered as equivalent to accelerated alteplase.

Tenecteplase ASSENT-2, an equivalence trial of over 16,000 patients compared tenecteplase and accelerated alteplase. The study found that 30-day mortality was almost the same in the tenecteplase group (6.2%) and the accelerated alteplase (6.2%) group. The absolute difference of 0.03% in favour of accelerated alteplase was not statistically significant (95% CI -0.55% to 0.61%). Given the confidence limits, tenecteplase and accelerated alteplase can be considered equivalent in terms of mortality. However, there was a small but statistically significant reduction in the incidence of bleeding with tenecteplase (26.4% compared with 28.9% in the accelerated alteplase group), resulting in fewer blood transfusions in the tenecteplase group (4.3% of patients compared with 5.5% in the accelerated alteplase group). Also, the rate of heart failure was statistically significantly lower in the tenecteplase group than in the accelerated alteplase group (6.1% vs 7.0%, p = 0.026).

Subgroups None of the trials discussed was designed to investigate clinical subgroups, such as by age or site of infarct (anterior, inferior). It was concluded that there was no convincing evidence of relative differences in the effectiveness of the available drugs in subgroups. The greater absolute benefit found in patients with anterior infarcts in GUSTO-I may simply be a reflection of the higher baseline risk in this group. The greater relative benefit in patients aged under 75 years was not reflected in their level of absolute risk reduction. None of the differences between the subgroups appeared to be statistically significant by interaction.

Summary In summary, given the evidence on clinical effectiveness, it can be concluded that, in the hospital setting, in terms of mortality:

  • standard alteplase is as effective as streptokinase

  • reteplase is at least as effective as streptokinase, and

  • tenecteplase is as effective as accelerated alteplase. If accelerated alteplase is believed to be superior to streptokinase, then indirectly tenecteplase would also be considered to be superior to streptokinase. Conclusions regarding the equivalence of reteplase compared with accelerated alteplase depend on the interpretation of GUSTO-III. Furthermore, if reteplase is considered to be equivalent to accelerated alteplase, then this indirectly implies that reteplase is as effective as tenecteplase. Important differences in major adverse events between the thrombolytic agents are also apparent. The newer drugs are associated with a higher risk of haemorrhagic stroke compared with streptokinase, but there are no apparent differences in the frequency of haemorrhagic stroke between accelerated alteplase and reteplase (GUSTO-III), or between accelerated alteplase and tenecteplase (ASSENT-2). However, compared with streptokinase, the newer drugs may also be associated with a lower incidence of congestive heart failure. In addition, allergic reactions are more common with streptokinase than with the other drugs, and major bleeds (leading to transfusions) may also be more common with streptokinase, although the evidence on this is not consistent across the trials. There is also some evidence that tenecteplase may be associated with lower rates of major bleeds and heart failure than accelerated alteplase.

4.1.2 Pre-hospital thrombolysis No RCTs were found comparing the different thrombolytic drugs in pre-hospital settings. However, nine RCTs and a systematic review investigating the feasibility, safety and efficacy of pre-hospital administration of thrombolysis compared with hospital administration were considered in the context of the appraisal. A number of other papers reporting non-randomised studies and audits of pre-hospital thrombolysis were also considered in relation to practical and implementation issues. The RCTs were small, except for one that included over 5000 patients (EMIP). They were undertaken in a mixture of urban and/or rural settings in Israel, continental Europe, Canada, the USA, Northern Ireland, and Scotland. A variety of thrombolytic drugs were studied – four studies used alteplase, four used streptokinase-type drugs, and one used urokinase (which is not available in the UK). Only the USA study (MITI) involved paramedics administering the thrombolytic (after remote consultation with a physician). In all but one of the other studies, a hospital physician attended the patient and administered the drug. In the rural Scottish trial (GREAT) a general practitioner undertook assessment and treatment. The RCTs found that, on average, pre-hospital thrombolysis was administered 58 minutes earlier than hospital thrombolysis; the differences ranged from 33 minutes in the MITI study to 130 minutes in the GREAT study. Individually, the trials failed to show statistically significant reductions in in-hospital mortality, although findings in all of the studies favoured pre-hospital administration. However, a meta-analysis of six of the trials found a statistically significant absolute reduction in mortality of 1.6% (95% CI 0.2% to 3%), and a relative risk reduction of 17% (95% CI 2% to 30%, p = 0.03) favouring pre-hospital administration of thrombolysis. This analysis is heavily influenced by the results of the GREAT study (in which thrombolysis was administered by general practitioners in rural Scotland) and therefore does not directly relate to the potential for paramedic based pre-hospital thrombolysis. A number of observational studies examining pre-hospital thrombolysis were considered, although these generally only provide further insight into feasibility and safety. They include studies of administration of anistreplase (a streptokinase-like drug that is no longer available in the UK) by paramedics or general practitioners in a Dutch city, reteplase administered by ambulance-based nurses in Sweden, reteplase administered by paramedics in the USA, anistreplase administered in a rural Italian emergency room, and two reports of a small number of cases of reteplase administered by paramedics with hospital telemetry support in England.

4.2 Cost effectiveness

4.2.1 In-hospital thrombolysis The Assessment Group's literature review found eight published articles on the cost-effectiveness of thrombolytic agents that met the inclusion criteria for the review of cost effectiveness. All compared streptokinase and alteplase (standard and accelerated) in a hospital setting. Three of the articles reported different aspects of the same cost-effectiveness model. Most studies reported incremental costs per life-year gained, and three also reported incremental cost per quality-adjusted life year (QALY). Most of the studies were based on the effectiveness results of GUSTO-I, in which data on resource use were collected only for USA centres. Consequently the analyses undertaken in Canada, Ireland and France had to attempt to translate these to settings in other countries. In general, the studies had the following limitations: costs and benefits were not measured in the same populations; comparator treatments were often inadequately described; and the derivation of utility values was inadequately explained. None of the studies undertook costing at a patient level and, while in general similar cost categories were included, only some of the studies included the longer-term costs of stroke and heart failure. Some of the studies included consideration of adverse events, including stroke, reinfarction, major bleeds, anaphylaxis, and congestive heart failure. The analyses undertaken following GUSTO-I, which found a survival advantage for accelerated alteplase at 30 days, showed the drug to be cost effective compared with streptokinase within the context of the clinical trial in the US healthcare system. In all of the studies, sensitivity analyses found that assumptions regarding mortality differences and costs were important, and so any conclusions drawn are heavily dependent on the interpretation of the effectiveness findings of GUSTO-I. In particular, the economic analysis undertaken in the USA alongside GUSTO-I modelled lifetime costs and benefits, and reported an incremental cost per life-year gained of $32,678 and an incremental cost per QALY of $36,402 for accelerated alteplase compared with streptokinase. The subgroup analyses found that accelerated alteplase became more cost effective in patients with higher absolute mortality risk – for example, $13,410 per life-year gained in patients older than 75 years with anterior myocardial infarction. However, the analysis requires extremely cautious interpretation given a number of issues, including uncertainties over the interpretation of GUSTO-I (in general and in subgroups), application of US data on resource use, and the assumption that costs did not differ significantly between treatment groups. Overall, there is little relevant published evidence on the economics of thrombolytics in a UK setting, and none examining the currently available bolus drugs. However, two cost-effectiveness models were submitted by manufacturers. It is logical to assume that the earlier the administration the greater the reduction in damage to the heart. However, while precise assumptions about the survival/time-to-treatment curve affect the benefit results in any modelling, it is unlikely that any one drug has a large advantage over any other with regard to timing of administration. The two manufacturers' models are similar in structure and scope, although they differ in terms of method and level of detail. Roche's model examines costs up to 30 days, assumes all four drugs have equivalent efficacy, and has less detailed costing. In contrast the Boehringer Ingelheim model includes costing up to 10 years, includes long-term costing for individual complications (such as congestive heart failure and stroke), and incorporates differential survival and complication outcomes for the drugs and more detailed estimation of utilities. Both models incorporate a range of different assumptions regarding adverse events. The models also incorporate adjustment for the timing of administration, including time-savings in pre-hospital settings in which only bolus drugs are compared. The Roche model essentially represents a cost-minimisation analysis, and finds reteplase slightly less costly than accelerated alteplase or tenecteplase in hospital. The Boehringer Ingelheim model assumes better survival and a lower incidence of post-infarct congestive heart failure (streptokinase 15.4%, accelerated alteplase 13.5%, reteplase 13.5%, tenecteplase 11.8%) for tenecteplase. These assumptions, together with 10-year discounted costs, lead to a finding that tenecteplase dominates accelerated alteplase and reteplase in hospital (that is, it is of lower cost and greater effectiveness). The Assessment Group adjusted key parameters, tested sensitivities and presented revised results using the manufacturers' models. The sensitivity analysis examined the parameter values submitted by the manufacturers for the following: 30-day mortality, strokes, major bleeds, reinfarctions and congestive heart failure. The Assessment Group used the adjusted models to compare the three newer drugs with streptokinase. For each comparison, the additional benefit (using QALYs) of the newer thrombolytic was small, while the additional cost was substantial. The cost differences between the newer drugs are relatively small. The most reliable finding is that streptokinase is by far the cheapest drug and although it is a little less effective (in terms of discounted QALYs), it is the most cost effective. Using the adjusted manufacturers' models, the incremental costs per QALY reported for the three drugs compared with streptokinase were: accelerated alteplase, £7219 (adjusted Boehringer Ingelheim model) and £7878 (adjusted Roche model); reteplase, £7893 and £10,247; and tenecteplase, £8321 and £9509. However, these cost–utility rankings of the three drugs relative to streptokinase are sensitive to changes in assumptions in the models, and so are not conclusive.

4.2.2 Pre-hospital thrombolysis No published articles examining the cost effectiveness of different thrombolytic drugs in pre-hospital settings were found. There is a published economic analysis of the GREAT study comparing cost effectiveness of pre-hospital and in-hospital thrombolysis (that is, the cost effectiveness of the different drugs), which found that pre-hospital delivery had an incremental cost per life saved of £3890. The sensitivity analysis found that the cost per life saved could increase to £88,000. It should also be borne in mind that the benefits found in the GREAT trial were larger than those found in other studies, the economic analysis was not undertaken alongside the trial, the interventions were not described in detail, and the model of rural Scottish general practitioner care is unlikely to be applicable throughout the NHS in England and Wales. In the pre-hospital setting, Roche's model assumes that reteplase and tenecteplase have equivalent efficacy and that reteplase is slightly cheaper. The Boehringer Ingelheim model finds that tenecteplase dominates reteplase in pre-hospital settings (that is, it has a lower cost and greater effectiveness). Building on the conclusions about in-hospital cost effectiveness, and since the general pre-hospital delivery costs for the two suitable bolus drugs (reteplase and tenecteplase) would be the same, the relative cost effectiveness of the drugs in pre-hospital settings is likely to be similar to that in hospital (assuming equal effectiveness of both drugs in each setting). On this basis it was concluded that it was not possible to distinguish between reteplase and tenecteplase on grounds of cost effectiveness in pre-hospital settings.

4.3 Consideration of the evidence

4.3.1 In-hospital thrombolysis The Committee noted the debate over the applicability of the findings of GUSTO-I beyond the North American centres (where most of the benefit of alteplase over streptokinase was found). The Committee considered that the efficacy of accelerated alteplase should not be determined solely from the results of the GUSTO-I trial. Despite concerns over the interpretation of GUSTO-I, the Committee concluded that it was likely that the newer thrombolytic agents are more effective than streptokinase in terms of 30-day mortality. The Committee was aware of the documented higher rates of stroke associated with the newer agents and carefully considered the views of clinical experts on this issue. The Committee considered that differences in the benefit of one thromboloytic agent over another are less clear if the combination of mortality and stroke endpoints are taken into account, particularly for subgroups at higher risk of haemorrhagic stroke. Furthermore, when considering the combination of mortality and stroke endpoints, it could be argued that the differences in overall benefit are less clear, particularly for subgroups at higher risk of developing haemorrhagic stroke. In taking the view that the use of streptokinase is cost effective, the Committee concluded that, although the acquisition cost of each of the newer drugs is substantially higher than that of streptokinase, the available economic evidence demonstrates that the newer drugs have an acceptable incremental cost-effectiveness ratio when compared with streptokinase. Given that streptokinase is associated with a lower risk of stroke and is a cost-effective drug, the Committee also considered it appropriate that all of the available thrombolytic drugs should be considered as options for use in care pathways for AMI. Local organisational and clinical policy considerations, such as proximity of CCU facilities and A&E staffing, will also have an impact on decisions regarding the appropriate use of each of the drugs in hospital. Because the drugs will have different risk–benefit profiles for individual patients, the Committee concluded that the decision about which of the available drugs to use should be made after balancing the likelihood of the benefits and risks (for example, stroke) to which the different drugs would expose the individual. The Committee took into account the potential importance of the methods of administration of the different thrombolytics and their effect on door-to-needle times. However, the impact of this factor on reducing myocardial damage and important clinical outcomes was very dependent on the overall pain-to-needle time. Thus, a saving of a few minutes in the door-to-needle time was likely to have a much greater impact on these endpoints where the pain-to-needle time was 1 hour compared with the situation where the pain-to-needle time was 6 hours.

4.3.2 Pre-hospital thrombolysis The Committee noted that while there is observational evidence to support pre-hospital thrombolysis, applying the results to the current NHS context is difficult, in that a minority used currently available bolus drugs, most are not paramedic based, and none was reliably generalisable to England and Wales. In the absence of comparative evidence on thrombolytics in pre-hospital settings, the Committee considered that the findings of trials comparing different thrombolytic drugs in hospital could still reasonably be applied to pre-hospital settings, with consideration of the additional relevant factors including safety and applicability examined in the pre-hospital studies outlined above. On the basis of advice from experts that only the bolus drugs were appropriate for pre-hospital administration given the practical difficulties explained in section 3.2.6, and given that no high-quality evidence was available to differentiate reteplase and tenecteplase in terms of clinical effectiveness or cost effectiveness in pre-hospital settings, the Committee considered that either reteplase or tenecteplase could be used in these settings, provided that the necessary infrastructure and training is provided to fully establish an appropriate model of pre-hospital thrombolytic administration. Given the risks associated with thrombolytic drugs and the fact that pre-hospital administration is an emerging practice in England and Wales, the Committee considered it important to ensure high-quality training and supervision of staff involved in the administration of thrombolysis. It was also considered important that clinicians and organisations delivering pre-hospital thrombolysis should develop clear clinical protocols for the use of thrombolytic drugs, such as those developed by the JRCALC, and adopt robust clinical governance arrangements to monitor the use of and outcomes associated with these drugs.