Evaluation of the clinical-diffusion and perfusion-diffusion mismatch models in DEFUSE.

BACKGROUND AND PURPOSE The perfusion-diffusion mismatch (PDM) model has been proposed as a tool to select acute stroke patients who are most likely to benefit from reperfusion therapy. The clinical-diffusion mismatch (CDM) model is an alternative method that is technically less challenging because it does not require perfusion-weighted imaging. This study is an evaluation of these 2 models in the DEFUSE dataset. METHODS DEFUSE is an open-label multicenter study in which acute stroke patients were treated with intravenous tPA between 3 and 6 hours after symptoms onset and an MRI was obtained before and 3 to 6 hours after treatment. Presence of PDM and CDM was determined for each patient. RESULTS Based on conventional predefined mismatch criteria, PDM was present in 54% of the DEFUSE population and CDM in 62%. There was no agreement beyond chance between the 2 mismatch models (kappa 0.07). The presence of PDM was associated with an increased chance of favorable clinical response after reperfusion (OR, 5.4; P=0.039). Reperfusion was not associated with a significant increase in the rate of favorable clinical response in patients with CDM (OR, 2.2; P=0.34). Using optimized mismatch criteria, determined retrospectively based on DEFUSE data, the OR for favorable clinical response was 70 (P=0.001) for PDM and 5.1 (P=0.066) for CDM. CONCLUSIONS The PDM model appears to be more accurate than the CDM model for selecting patients who are likely to benefit from reperfusion therapy in the 3- to 6-hour time window.