Background: Diffusion-perfusion mismatch (perfusion-weighted imaging [PWI] abnormality minus diffusion-weighted imaging [DWI] abnormality) can identify candidates for acute stroke intervention, but PWI is often not obtainable. The authors hypothesized that language tests can predict volume of hypoperfusion, and thus mismatch, in acute left hemisphere stroke, and that the estimated mismatch can predict potential for early recovery of language.
Methods: A consecutive series of 81 patients with acute left hemisphere ischemic stroke underwent language testing within 1 day of MRI scans. Volumes of abnormality on PWI and DWI were measured without knowledge of language scores. Using tests that correlated well with PWI abnormality (oral naming and repetition), the authors computed an estimated PWI abnormality (ePWI) for each patient from a linear regression model and derived a diffusion-clinical percent mismatch ([ePWI-DWI/ePWI] x 100). The authors then tested the hypothesis that patients with > or =20% diffusion-clinical mismatch have a greater chance of short-term improvement in language by examining scores of the 23 patients with repeat testing within 1 week.
Results: Within-group comparisons: patients with > or =20% diffusion-clinical mismatch showed improvement in total language score within 1 week (Wilcoxon signed rank: p < 0.02), whereas patients without mismatch did not. Across-group comparison: patients with > or =20% mismatch showed more short-term improvement in language scores than those without mismatch (Mann-Whitney U: p < 0.05).
Conclusions: Tests of oral naming or repetition can be used in patients with acute left hemisphere stroke to estimate perfusion-weighted imaging abnormality and compute a diffusion-clinical mismatch that may predict potential for short-term language improvement.