Introduction: The decrease in efficacy of reperfusion/recanalization interventions over time may be due in part to decreases in the presence a salvageable ischemic penumbra, which can be imaged as the perfusion-diffusion mismatch on MRI. Recent clinical trial evidence supporting the efficacy of endovascular interventions at later time windows suggest that a small core with a high NIHSS may also serve as a marker of the ischemic penumbra. We investigated the relationship of time, the site of arterial occlusion and NIHSS to the probability of a diffusion-perfusion mismatch.
Method: The LESION study was a prospectively collected cohort of 1,092 consecutive subjects, NIHSS>3, scanned within 24 hours of symptom onset of time last known well. Target diffusion-perfusion mismatch was qualitatively assessed by a method previously demonstrated as equivalent to volumetric methods or clinical-diffusion core mismatch. Endovascular target (EVT) was defined as an arterial occlusion of intracranial ICA or MCA-M1.
Results: Within the first 6 hrs, 26.8% (174 of 475 patients) had an EVT occlusion compared to 19.4% (62 of 258) between 6-24 hrs. Of all EVT occlusions, 26.3% occurred later than 6 hrs. Images with an EVT occlusion had a target mismatch 77.1% of the time, compared to 39.5% for occlusions in other arterial branches (p < 0.001). While mismatch was present more often between 0-6 hrs compared to 6-24 hrs (OR = 1.706; p < 0.001), mismatch was still present in 40.1% of images collected after 6 hrs (vs 53.3% 0-6 hrs). Overall, the presence of mismatch declined over time (OR = 0.934; p<0.001). Multivariate modeling of target mismatch was best fit by including: time (OR= 0.943; p< 0.001), NIHSS (OR= 1.037; p< 0.005), EVT occlusion (OR= 9.450; p< 0.001), and an interaction between NIHSS and endovascular target (OR= 0.950; p=0.026).
Discussion: Overall, within the first 24 hrs three quarters of patients with EVT occlusions have target diffusion-perfusion mismatch. Approximately one quarter of all EVT occlusions occur between 6-24 hrs. Multivariate models including time from last known well, NIHSS on presentation, and the presence of EVT occlusion best predicted the presence of a target diffusion-perfusion mismatch.