The role of white matter disconnection in stroke as a predictor of clinical outcome after mechanical thrombectomy.

​Rationale: Mechanical thrombectomy is a promising approach to acute treatment in large vessel occlusion (LVO) ischemic stroke. This technique has shown to be safe and effective when performed both in early and late-window trials. Several clinical and mainly volumetric, radiological features are used as prognostic factors for functional outcome and patient eligibility. However, emerging evidence supports the idea that lesion topography is strongly associated with prognosis and functional brain recovery. The aim of this study is to examine the role of clinical (i.e. mRS and NIHSS) vs. standard volume-based lesion (i.e. ASPECTS, core, penumbra and final lesion volume) vs. topological radiological (i.e. white matter structural disconnection) features, in patients eligible for acute mechanical thrombectomy. Materials and methods: We selected a group of patients (n=50) who underwent acute mechanical thrombectomy over 47 months, from January 2018 to November 2021, and occurred at the Stroke Unit and Clinica Neurologica of the Hospital of Padova. They were studied with the modified Rankin Scale (mRS) and the National Institutes of Health Stroke Scale (NIHSS) both at admission (pre) and at discharge (post), then, again with the mRS at 90 days from the acute event. The lesions were manually segmented on structural MRI and CT scans using the program ITK-SNAP. Four models were performed through a linear regression analysis. Specifically, we computed a baseline clinical model (M1) based on demographics, pre-stroke mRS and admission NIHSS. Then we added commonly used (standard) radiological parameters of lesion or perfusion damage (core, penumbra, ASPECTS) (M2). Therefore, we added information about the white matter structural disconnection to clinical variables (M3). Finally, we tested a baseline clinical +early recovery model (M4), which included age, pre-mRS, admission NIHSS, and post-mRS. The lesions were normalized in atlas space and displayed to study their distribution and structural disconnections (SDC). Results: The mean baseline mRS was 0.48±0.90, while the mean 90-day mRS was 2.18±1.81. The linear regression analysis showed a significant positive correlation between 90-day mRS and clinical variables (pre-mRS, NIHSS at presentation, post-mRS), while radiological variables (ASPECTS, core, and penumbra volume) did not seem to be associated with functional outcome. The results of the ANOVA analysis showed that, between the four models tested, M4 (including age, pre-mRS, NIHSS at presentation, and post-mRS as independent variables) was the one providing the highest adjusted R-squared [Adj.R-squared=0.614] and explained 62% of the variance in outcome prediction. At a voxel-wise level, we found a significant positive correlation between brain recovery (Delta 90-day mRS-pre-mRS) and damage, affecting predominantly the left corticospinal tract and the corresponding structural white matter disconnection (SDC), which also extended to the cingulum and, bilaterally, to the callosal commissure. Conclusion: In our sample, acute clinical status represents the most valuable prognostic factor. Interestingly, while radiological (i.e. volumetric and semi-quantitative) features, such as ASPECTS, core, and penumbra volume, did not show any significant correlation with 90-day mRS, structural white matter disconnection and lesion topography, in particular of the left corticospinal tract, were associated with a poorer recovery after endovascular treatment. These results could have important future implications in pre-treatment patients’ selection and in post-treatment post stroke rehabilitation.