SYSTEMATIC REVIEW OF EVIDENCE ON THROMBOLYTIC THERAPY FOR ACUTE ISCHEMIC STROKE

Background Recent trials of thrombolytic therapy in acute ischaemic st roke have given apparently conflicting results. Only one trial, the Na tional Institute of Neurological Disorders and Stroke trial of tissue plasminogen activator (tPA), suggested that thrombolysis was definitel y beneficial, To make sense of these results, we have done a systemati c review of all available randomised trials of thrombolysis in acute i schaemic stroke. Methods From all available completed randomised trial s of thrombolytic therapy compared with control in acute ischaemic str oke (with prerandomisation CT), we checked tabular data on deaths duri ng roughly the first 2 weeks, deaths from all causes and functional ou tcome (disability) at the end of the trial follow-up period, and early symptomatic and fatal intracranial haemorrhages. Findings 12 trials i ncluded 3435 patients, of whom 694 (20%) were dead and 1001 (39%) of 2 567 were functionally dependent at the end of follow-up (duration of f ollow-up varied between trials, but the longest was 6 months). 214 (6% ) of the 3435 patients had early symptomatic or fatal intracranial hae morrhages. Thrombolytic therapy was associated with a significant exce ss of early deaths (91 per 1000 patients treated [95% CI 54-134]), and total deaths (37 per 1000 [20-83]), but there was nevertheless a sign ificant reduction in the number of patients in the combined outcome of dead or dependent (65 fewer per 1000 patients treated [28-107]). Ther e was a substantial and significant excess of symptomatic and fatal in tracranial haemorrhages with thrombolysis-which was similar in all rec ent trials-of about 70 extra symptomatic intracranial haemorrhages per 1000 patients treated (of which 51 per 1000 were fatal), In the cohor t of patients randomised within 3 h of stroke, there was a significant reduction in the number of patients who were dead or dependent al the end of follow-up (141 fewer dead or dependent per 1000 patients treat ed [75-206] and a non-significant increase in the number dead (nine pe r 1000 treated [-39 to 70]). There was significant heterogeneity betwe en the trials for total deaths at the end of follow-up, which may be p artly explained by differences in the use of antithrombotic drugs with in the first 24 h of thrombolysis; the variation in severity of stroke s included: the time window to thrombolytic treatment; and the dose of thrombolytic drug used. There were no direct comparisons of tPA with streptokinase or urokinase: much of the poor outcome in the streptokin ase-treated patients might be explained by the inclusion of more sever e strokes, greater use of antithrombotic drugs, higher doses, and the longer time to treatment compared with the trials that used tPA. Inter pretation Thrombolysis requires further testing in large randomised tr ials because the risks seem substantial, and the benefit uncertain. Th e time window for effective treatment remains unclear. There is no obj ective evidence to suggest that tPA is safer than streptokinase; the a pparent hazards and benefits may be similar when differences in trial design and baseline variables are accounted for.