BACKGROUND AND PURPOSE: We sought to study the role of MRI-derived cerebral blood flow (CBF) measurements for the prediction of lesion development in acute stroke patients. METHODS: Thirty-two patients were treated with tissue plasminogen activator. Diffusion-weighted (DWI) and perfusion-weighted MRI, T2-weighted imaging, and MR angiography were performed before treatment (2.8+/-0.9 hours after symptom onset) and on follow-up (days 1 and 7). CBF thresholds (12 and 22 mL/100 g per minute) were applied to bolus tracking MRI maps to determine predictive cutoff levels. RESULTS: In 21 patients (group A), the terminal lesion volume (T2-weighted imaging) was larger than the initial DWI lesion volume (89+/-93 versus 21+/-38 mL). In 11 patients (group B), the terminal lesion volume was smaller than the initial DWI lesion volume (7+/-27 versus 15+/-29 mL). The initial DWI lesion volume did not differ between both groups. The presence of a tissue volume > or =50 mL with a CBF value <or =12 mL/100 g per minute was predictive for lesion enlargement to day 7 in T2-weighted imaging (positive predictive value, 0.80). CONCLUSIONS: The presence of a tissue volume > or =50 mL with a CBF value <or =12 mL/100 g per minute predicts further lesion growth in hyperacute stroke patients. MRI-derived CBF values, with all their present limitations, are valuable in early estimation of prognosis of stroke patients.
