Predictors of hemorrhagic transformation in patients receiving intra-arterial thrombolysis

Background and Purpose-Hemorrhagic transformation (HT) is a major complication of intra-arterial (IA) thrombolytic therapy. Identifying significant predictors of hemorrhage after thrombolysis would be useful in guiding patient selection for IA treatment. Methods-Data were collected retrospectively on consecutive patients with acute focal cerebral ischemia within the anterior or posterior circulation who were treated with combined intravenous (IV)-IA or pure IA thrombolysis over an 8-year period at the UCLA Medical Center. Results-Eighty-nine patients were treated. Median baseline National Institutes of Health Stroke Scale (NIHSS) score was 16, and mean age was 69 years. Twenty-six patients received IA tissue plasminogen activator (tPA) only, 22 received IV-IA tPA, and 41 received IA urokinase only. Asymptomatic HT occurred in 29 patients (33%), minor symptomatic HT (1- to 3-point worsening in NIHSS score) occurred in 10 patients (11%), and major symptomatic HT ( greater than or equal to4-point worsening in NIHSS score) occurred in 6 patients (7%). The rate of any HT was similar in patients treated with pure IA thrombolysis (39%) versus combined IV-IA thrombolysis (41%). In pure IA cases, the rate of any HT was 50% with tPA versus 32% with urokinase (P=0.2). Eighty-six percent of the patients with HT versus 39% of the patients without HT were dead or disabled (modified Rankin score >2) at day 7 (P<0.0001). On multivariate analysis, independent predictors of any HT were higher NIHSS score, longer time to recanalization, lower platelet count, and higher glucose level. A model using these variables correctly predicted HT with positive predictive value 70% and overall accuracy 78%. Conclusions-In this large series of IA thrombolysis, rates of HT were similar to those demonstrated in prior series and clinical trials. Higher NIHSS score, longer time to recanalization, lower platelet count, and higher glucose level were independent predictors of any HT.