Brain dysconnectome: a potential biomarker for functional outcome after mechanical thrombectomy

Mechanical thrombectomy (MT) is a safe and effective procedure that has improved the prognosis of patients with large vessel obstruction (LVO) stroke. Even though we select patients based on various clinical and imaging criteria, more than half of those undergoing this procedure remain with severe disability. Several factors, both pre-treatment (pre-stroke mRS, NIHSS, event-recanalization time, ASPECTS, core/ penumbra volumes) and post-treatment (TICI, NIHSS, final infarct volume, complications) have been identified as predictors of outcome. Among these factors, the volume of the lesion and the size of the hypo-perfused region of the brain are used to make individual decisions about treatment. In retrospective studies, the volume of the lesion weakly correlates with clinical outcomes, while lesion location is more predictive. In addition measures of large-scale (brain networks) disruption have been developed based on the concept of structural and functional disconnection, i.e., the ensemble of structural and functional connections that are directly or indirectly damaged by the focal injury. These disconnection measures have been shown to be strongly predictive of acute impairment and recovery of function. Here we plan to measure in a group of patients with LVO who underwent MT the relationship between outcome (3 month-mRS) and the location of the lesion or its effect on structural and functional networks. To examine whether the outcome is more related to the vascular distribution stroke or their effects on structural-functional brain networks, we mapped the lesions onto a vascular atlas, a gray matter functional regions’ atlas, or a white matter structural connections’ atlas. Then using multi-variate statistical models, we tested which atlas was more predictive of clinical outcome. The same analysis was then applied to structural and functional disconnection patterns. A total of n=66 patients underwent MT at the Neurology Unit of Padua hospital from January 2019 to June 2022. They were examined with the mRS and the NIHSS at admission and discharge. The mRS was also administered at three months post-stroke for measuring clinical outcomes. The location and volume of the lesions were measured from CT, and FLAIR MRI scans performed after MT and manually segmented using the software ITK-SNAP. We computed voxel-wise maps of structural and functional disconnections that were significantly related to functional outcomes. We also investigated the relationship between lesion location computed on three different atlases (vascular, functional grey matter, and structural white matter atlas) and 3-month mRS. The mean pre-event mRS was 0.5±0.9, post-MT mRS was 3.1±1.9, at three-month mRS was 2.5±2.1. A voxel-wise analysis of the functional disconnection showed a significant involvement of the sensory-motor network (SMN) (R2= 0.340), the visual network (VIS)(R2= 0.379), and the dorsal attention network (DAN)(R2=0.318). The voxel-wise structural disconnection analysis localized sensorimotor pathways and long-range association pathways. The prediction of lesion topography on clinical outcome was more robust for the functional atlas (R2=0.382), followed by the structural atlas (R2=0.338), while the vascular atlas provided the lowest prediction (R2=0.146). Structural disconnection performed better than functional disconnection in predicting outcomes (respectively R2=0.339 and R2=0.205). Stroke lesion topography is a strong prognostic factor of outcome at three months when computed on an atlas of functional and structural networks, as compared to a vascular-based atlas. These findings indicate that pre-treatment evaluations for MT shall take into consideration the network structure of the brain and less its vascular supply. Structural disconnection measures are of high prognostic value. Future studies will define the prognostic value of a network-based atlas in a pre-treatment setting.