Background and purpose: In acute stroke, diffusion-weighted imaging (DWI) lesions are commonly considered markers of irreversible ischemia yet can occasionally reverse. However, the extent and clinical correlates of DWI reversal in thrombolyzed patients remain unclear. We assessed the extent of reversible acute DWI lesions (RADs) and their relationships with clinical outcome in patients thrombolyzed≤4.5 hours from onset.
Methods: Data were retrospectively analyzed. RAD was defined as an acute DWI lesion not part of a 24-hour DWI lesion as determined voxelwise. Associations with an early neurological improvement (early neurological improvement=ΔNational Institutes of Health Stroke Scale≥8 or National Institutes of Health Stroke Scale≤2 at 24 hours) or an excellent outcome (modified Rankin Scale≤1) were assessed in multivariate analyses.
Results: One hundred seventy-six patients were included. The median (interquartile range) time to treatment from onset was 150 minutes (120-194). Eighty-nine patients (50%) exhibited visually-detectable RAD irrespective of its extent. Over the whole population, the median percentage and volume of RAD were 11% (4-36) and 2.4 mL (0.5-8). Subtracting RAD from initial DWI altered perfusion-weighted imaging-DWI classification in 5 of 100 patients (shift from "no mismatch" to "mismatch" profile in all). Percent RAD was significantly greater in patients treated≤3 hours (P=0.049), without proximal occlusion (P=0.003), and in 24-hour recanalizers (P<0.001). Early neurological improvement was independently associated with percent RAD. This association increased with percent RAD split in quartiles in a "dose-dependent" manner (P for trend=0.01). Excellent outcome was independently associated with percent RAD (P for trend<0.001).
Conclusions: DWI reversal was often sizeable in patients treated≤4.5 hours. It was strongly associated with, albeit not necessarily causal for, early neurological improvement.