Abstract 3595: Perfusion imaging predicts Outcome in TIA and Minor Stroke. A Prospective Derivation-Validation Study

Abstract

Background: Patients presenting with transient or minor ischemic symptoms (TIA/MIS) are at risk for early deterioration. Identification of those at highest risk for progression may justify more aggressive acute reperfusion treatments. We tested the hypothesis that baseline perfusion (PWI)- diffusion (DWI) mismatch predicts clinical deterioration and infarct growth on follow-up imaging in this population. Methods: Patients with TIA/MIS (NIH Stroke Scale ≤ 3) were prospectively enrolled and imaged within 24 hours of symptom onset as part of two sequential prospective imaging studies. All patients had clinical follow-up. Baseline DWI and PWI (tmax+4s delay) and follow-up FLAIR infarct volumes (day 30 (derivation), day 90 (validation) cohort) were measured. Mismatch volumes were calculated as (Tmax+4s delay) - DWI lesion volume. Primary outcome was infarct growth on FLAIR imaging which was defined a priori as growth of at least 2.5 ml. Secondary outcome was clinical progression. Results: 137 patients were included in the derivation and 281 patients in the validation cohorts. The rates of DWI (54% vs 56%, p= 0.67) and PWI lesions (42% vs 34.5%, p=0.16) at baseline were similar between the cohorts. The median time between symptom onset and baseline imaging was significantly shorter in the derivation (9.2 h, IQR=9.4) relative to the validation sets (15.1h, IQR=12.5 p<0.001). More patients had follow-up imaging in the derivation (87%) compared to the validation (76%) cohort (p=0.021). Primary and secondary outcome occurred in 18.5% and 9.5% in the derivation and 5.5% and 4.6% in the validation cohort. In the derivation cohort, baseline mismatch volumes adjusting for age, sex and time from symptom onset to MRI significantly predicted radiographic progression (OR=1.06 [1.03-1.09], p<0.001). The optimal threshold for maximizing sensitivity (Sen) and specificity (Spec) in predicting infarct growth occurred at a mismatch volume of 10ml; which correctly predicted infarct expansion with 82% (Sen) and 91%(Spec) (Area under the curve (AUC)= 0.89 [0.80-0.98]). In the validation cohort, this threshold was highly predictive of radiological progression (p=0.011, McNemars test). Linear regression showed that for every 10ml of mismatch, there would be 2.5ml infarct growth on day 30 FLAIR [R=0.80, p<0.001] (derivation set) and 1.1 ml of growth on day 90 FLAIR (R=0.22, p<0.001) (validation set). Baseline mismatch showed a high discriminative value in predicting clinical deterioration in the derivation (AUC =0.81 [0.67-0.96]) and moderate value in the validation cohort (AUC=0.66 [0.46, 0.85]). Conclusion: In a population of patients with minor stroke and TIA, early MR perfusion-diffusion mismatch predicts infarct growth and clinical progression. These findings suggest that there may be a group of patients with minor symptoms in whom reperfusion strategies may be beneficial.