Thrombolysis for Acute Ischemic Stroke
Graeme J. Hankey MD, FRACP, FRCP Section Editor:
Recombinant tissue plasminogen activator (rtPA) is licensed for thrombolytic therapy in ischemic stroke if administered within 3 hours of stroke onset in selected patients (to be extended to 4.5 hours in some countries). However, many patients who could benefit from rtPA are not offered it and there is evidence that rtPA could benefit many patients who do not meet current license criteria. Furthermore, the patient characteristics that might assist in stratifying risk and benefit and the latest time at which thrombolysis may be effective remain unknown.
The objectives of this study were to assess the safety and efficacy of thrombolytic agents used in acute treatment of ischemic stroke, factors that might influence risk or benefit, and estimate if current data can identify the latest time window for treatment.
We used the Cochrane Stroke Group Trials Register, MEDLINE, and EMBASE; contact with trialists, and hand-searching of pertinent journals (all to October 2008).
Selection criteria were randomized trials of any thrombolytic agent compared with a control in patients with definite ischemic stroke.
Data Collection and Analysis
Two reviewers applied the inclusion criteria, extracted the data (published and unpublished), and assessed trial quality. We calculated ORs and 95% CIs for all main outcomes using fixed effects methods (additionally by random effects methods where significant between-trial heterogeneity was present), calculated the effect per 1000 patients treated, and performed sensitivity analyses on key variables.
Twenty-six trials with data on 7152 patients testing urokinase, streptokinase, rtPA, recombinant prourokinase, or desmoteplase were included.1 Fifty-six percent of all data came from trials testing rtPA (11 trials, 3977 patients). Four trials used intra-arterial administration (all non-rtPA); the rest used intravenous treatment. Most data come from trials that started treatment up to 6 hours after stroke onset. Among more recent trials, 3 treated up to 9 and 1 up to 24 hours after stroke onset. Twenty-three trials used plain CT scanning for brain imaging, 3 trials used a version of the “mismatch concept” to include/exclude patients, and 1 further trial collected data on mismatch. Very few patients (0.5% of all data, approximately 1% of rtPA data) were aged >80 years. There were imbalances in key prognostic variables and several did not have complete blinding of outcome assessment. The results for all thrombolytic drugs were similar,1 but we focus on rtPA data (Table).
Among trials testing recombinant tissue plasminogen activator (rtPA), there was a significant increase in symptomatic and fatal intracranial hemorrhage (ICH) and a nonsignificant increase in total early and late deaths (ie, within the first 7 to 10 days and by 3 months, respectively; Table). Fatal ICH accounted for most of the nonsignificant excess of total early deaths, because removing the fatal ICH from the total early deaths showed that rtPA significantly reduced the odds of early death from non-ICH causes. Furthermore, among patients who survived the first 7 to 10 days, there was no subsequent excess of deaths with rtPA (Table).
rtPA significantly reduced the proportion of patients with modified Rankin Scale (mRS) score 3 to 6 at 3 months after stroke (Table). In absolute numbers, for every 1000 patients treated during the first 6 hours after stroke, 60 (95% CI, 30 to 90) fewer patients were dead or dependent at late follow-up. However, there was significant heterogeneity of treatment effect (all drugs, I2 38%, P=0.04; rtPA trials, I2 62%, P=0.007). A random effects analysis produced slightly less positive, although still statistically significant, results. Among survivors (mRS score 3 to 5), rtPA significantly reduced dependency with no heterogeneity (Table).
Defining “poor outcome” at 3 months as mRS score 2 to 6 gave a more positive effect on functional outcome than for mRS score 3 to 6 that did not reach significance (note this analysis is restricted to just those trials that provided both outcomes; Table). However, using mRS score of 2 to 6 was associated with significant heterogeneity, whereas mRS score of 3 to 6 was not, indicating that mRS score of 3 to 6 may provide not just a more cautious, but also a more robust estimate of rtPA effect.
Several factors could modify the effect of thrombolysis and explain the heterogeneity. These indirect comparisons should be treated with caution and regarded as hypothesis-generating rather than reliable evidence (note full details in Wardlaw et al1). For example, rtPA given within 4 hours of stroke appeared more effective in reducing death or dependency, but there was still benefit between 3 and 6 hours (Table). Patients with moderate stroke (National Institutes of Health Stroke Scale score approximately 10 to 15) may get most benefit from thrombolysis and patients with severe stroke (National Institute of Neurological Diseases and Stroke score approximately 18 to 30) the least benefit, but these analyses are based on small samples. Data were insufficient to examine the effect of visible infarct signs on CT. Results of studies selecting patients using MR diffusion/perfusion imaging mismatch were similar to results of studies that selected patients on the basis of plain CT, although direct comparisons were difficult because there were other differences than imaging between the studies.
Overall, among the selected populations of patients included in the trials to date (99% of whom were aged <80 years), rtPA significantly reduced the proportion of patients with poor outcomes after stroke (and conversely increased the proportion of patients with good outcomes). This overall benefit was apparent despite a nonsignificant increase in deaths, mostly attributable to ICH. There were insufficient data to determine the risk–benefit ratio for different categories of patients.
Applicability of Findings to Clinical Practice
Use of rtPA within the existing license is supported by the available evidence, but data are insufficient to determine risks and benefits in clinically important subgroups of patients, especially those aged >80 years, but also by vascular risk factors and medical history, brain scan appearances, stroke subtype, or the latest time for benefit.
More data are required to determine the magnitude of overall benefit in clinically important subgroups of patients, especially those aged >80 years, but also by time to treatment; grades of stroke severity; subtypes of stroke (eg, lacunar, cortical); comorbidities such as diabetes, hypertension, and prior stroke; aspirin use before stroke; and findings on different kinds of brain imaging. To answer these questions reliably, and in particular to be able to tailor treatment to the individual and increasingly more complex patient with stroke, more data are needed from new randomized controlled trials.
Sources of Funding
Supported by the Scottish Funding Council Scottish Imaging Network, A Platform for Scientific Excellence (SINAPSE) Collaboration (J.M.W.); Swedish Heart–Lung Fund, AFA, and The Marianne and Marcus Wallenberg Foundation (V.M.); and Helse Sør-Øst RHF, Oslo, Norway (E.B.).
- Received December 12, 2009.
- Accepted December 16, 2009.