Abstract 2550: Early 3 Hour Imaging Demonstrates Substantial Diffusion Reversal with Lower Infarction Risk Compared to Tissue without Diffusion Reversal

Abstract

Background: Literature suggests that diffusion reversal--determined by visually inspecting diffusion-weighted images which reverse on subsequent imaging--is uncommon. However, these studies performed the initial scan 3 to 6 hrs from onset which may miss lesions that have already undergone reversal. We measured apparent diffusion coefficient (ADC) reversal between 2 scans obtained at 3 and 6 hrs from onset comparing infarct probability in reversed regions to brain regions with persistent ADC lesions. Methods: Ischemic stroke patients underwent 3 MRI’s: <4.5 hrs (tp1), at 6 hrs (tp2), and at 1 mo (tp3) after onset. Co-registered maps measured ADC lesions (tp1, tp2) and FLAIR infarct (tp3). Diffusion lesions were manually outlined as hypointense regions on ADC. A population-derived ADC threshold to distinguish abnormal from normal ADC was calculated based on the maximum ADC value capturing >90% of voxels within the outlined lesions. ADC values less than or greater than this threshold were defined as abnormal and normal, respectively. ADC “reversal” was defined as voxels with abnormal tp1 ADC and normal tp2 ADC, whereas ADC “non-reversal” was defined as abnormal ADC on both tp1 and tp2. Final infarcts were outlined as hyperintense regions on tp3 FLAIR. Infarct probability (% of the tp1 ADC lesion which infarcted on tp3) was compared between reversed and non-reversed tissue using a Mann-Whitney test. Leukoariosis or CSF voxels due to infarct atrophy were excluded from the infarct analysis. Results: 39 patients were prospectively scanned at 2.8hr (tp1), 6.4hr (tp2), and 1mo (tp3) after stroke onset (NIHSS=14, 74% received tPA). The population-derived ADC threshold to distinguish between abnormal and normal was 71 × 10 -5 mm 2 /s. Median ADC lesion volume at tp1 was 34ml [15, 70] and at tp2 was 34ml [14, 65]. Median ADC reversal volume was 6.6ml [5, 17], accounting for 34% [11, 44] of the tp1 ADC lesion volume (ADC non-reversal volume was 21ml [6, 50]). 31 of 39 (79%) and 20 of 39 (51%) patients had reversal volumes >10% and >33% of their initial ADC lesion, respectively. Infarct probability was lower (57%) in ADC reversal tissue compared to tissue with non-reversal (82%, p<0.0001) ( Figure ). To confirm that the derived ADC threshold distinguished abnormal tissue on tp1 from normal tissue on tp2, ADC tp1 and tp2 values were compared and differed (66 vs. 77, p<0.0001). Conclusion: While ADC closely approximates infarct core, during early ischemia, a substantial proportion of the initial ADC lesion may reverse. Brain tissue with this MR signature carries a significantly lower risk of infarction than tissue without ADC reversal.