Intravenous rt-PA versus Endovascular Therapy for Acute Ischemic Stroke Pitchaiah Mandava, MD, PhD, MSEE, Jose I. Suarez, MD, and Thomas A. Kent, MD Corresponding author Thomas A. Kent, MD Department of Neurology, Baylor College of Medicine/MEDVAMC, 2002 Holcombe Blvd (127), Houston TX 77030, USA. E-mail: tkent@bcm.tmc.edu Current Atherosclerosis Reports 2008, 10: 332 – 338 Current Medicine Group LLC ISSN 1523-3804 Copyright © 2008 by Current Medicine Group LLC The influence of baseline stroke severity on outcome makes comparisons between nonrandomized studies of intravenous and intra-arterial (IA) therapy problematic. Using pooled data from the placebo arms of random- ized trials in acute ischemic stroke, we derived predictive functions for outcome. We then compared the outcomes from published trials to these functions. Net benefit was calculated by comparison of the individual study with the predicted outcome based on the therapeutic time window. Similar net benefit for IA therapy and intrave- nous therapy was found at 3 hours and 6 hours; a slight advantage for IA therapy was mitigated by an increase in mortality at 6 hours and by publication bias. No net ben- efit for IA therapy was shown in the time window greater than 6 hours. Conclusive evidence for the superiority of either therapy awaits prospective randomized trials. Introduction Intravenous recombinant tissue plasminogen activator (rt-PA) within 3 hours of onset remains the mainstay of therapy for acute ischemic stroke [1 ], but considerable effort is under way to improve outcomes for patients treated with intravenous rt-PA and to expand the time window so that more patients can receive treatment [2–4]. Neuroprotectant approaches have yet to demonstrate eficacy in phase III trials [5–8], and the approaches most widely reported use therapies directed at removing the thrombus [9 ,10]. Both intravenous and intra-arterial (IA) catheter-based therapies are being pursued [11 ]. These trials are reported mostly in the form of individual case series with comparison to historical controls. The few randomized trials of IA therapy [ 12 ,13] did not compare outcomes to best medical therapy. In our view, the primary dif iculty in assessing these reports is that outcome from stroke is highly dependent on baseline severity and, to a lesser extent, on age [14– 16,17•]. This phenomenon inluences not just comparison with historical controls but also small trials in which randomization is quite dif icult to achieve [ 18]. Similarly, meta-analyses or systematic analyses will be confounded by differences in populations. We sought to overcome these dif iculties by develop- ing a method by which individual studies are compared with published natural history models based on com- parable baseline characteristics [17•]. In the case of IA therapy, we found wide variability in outcomes for reported case series, which, when pooled, yielded little net beneit in terms of functional outcome or mortality. Though acknowledging numerous limitations of this method [17•], we were able to identify factors that were associated with the best performers, including use of lower doses of thrombolytic agents and treating patients with more severe stroke. In this report, we develop a new model based on a larger range of more recently reported acute stroke trials and examine the outcomes of trials of both intravenous and IA therapy based on reported therapeutic time windows. Methods We developed a predictive model for outcomes from stroke based on the results of individual control arms of randomized trials and the baseline severity and age of the population that was reported. This analysis was per- formed in three steps. Selection of studies To select intravenous thrombolytic studies, MEDLINE databases were searched for the words “Acute Ischemic Stroke” and “Acute Ischaemic Stroke.” To compare with endovascular therapies, we selected double-blind, ran- domized controlled trials using intravenous thrombolytics that reported treatment time window, median baseline National Institutes of Health Stroke Scale (NIHSS) score, mean age, 3-month mortality, and modi ied Rankin Score (mRS) outcomes.