Abstract 53: The Malignant MRI profile: Implications for Endovascular Therapy

Abstract

Background: The Malignant MRI profile identifies stroke patients with poor outcomes and an increased incidence of parenchymal hematoma following iv thrombolysis; outcomes following endovascular reperfusion therapy have not been described. Methods: The NIH funded DEFUSE 2 trial enrolled consecutive acute stroke patients in whom endovascular therapy was anticipated. An MRI scan was obtained immediately prior to intra-arterial reperfusion therapy, then repeated following the procedure and on day 5. Perfusion-weighted (PWI) and diffusion-weighted imaging (DWI) maps were created and lesion volumes estimated with an automated software program (RAPID). In the DEFUSE 2, the Malignant profile was pre-specified as a DWI lesion ≥70mL and/or a PWI lesion based on Tmax>10sec threshold (Tmax>10s) ≥100mL. A receiver operating characteristic (ROC) curve analysis was performed to identify Tmax>10s and DWI volumes that predicted poor outcome (defined as a mRS 5-6 at 30 days) with a high specificity. Patients with an M1 or ICA occlusion who did not undergo endovascular therapy based on local site criteria were also included in the ROC analysis. Results: We report a preliminary analysis of the DEFUSE 2 database (full data will be presented at the meeting). One hundred and one patients were triaged to the cath lab for endovascular therapy. Of the 83 patients who had adequate data available for this analysis, 9 (11%) met the predefined criteria for the Malignant profile. 56% of the Malignant patients had poor outcome compared with 30% of the non-Malignant cases (p=0.14). Malignant patients had an increased risk of parenchymal hematoma (PH1 or PH2): 44% vs. 14% (p=0.04). Only 1 of the Malignant patients achieved a mRS of 0-2 at 30 days. Early reperfusion was obtained in 6 of the 9 Malignant patients but was not associated with an increase in favorable clinical outcome or a decrease in the risk of poor outcome. Fifteen patients with an M1 or ICA occlusion did not undergo endovascular therapy based on local site criteria. Ten of these patients had the Malignant profile and 8 of these 10 had a poor outcome. ROC curve analysis identified a DWI lesion of 112 mL and a Tmax>10s lesion of 116 mL as optimal thresholds to predict poor outcome; both achieved a specificity of 98% and sensitivities of 27% and 24% respectively. 81% (13/16) of the Malignant patients identified by the DWI and/or Tmax>10s optimal thresholds had poor outcome. Conclusion: Patients with large baseline DWI and/or large severe PWI lesions are likely to have poor outcomes with or without endovascular reperfusion therapy. Automated imaging software can identify these patients rapidly.