1 Neural Architecture, Behaviour and Connectivity Laboratory, Department of Psychology, Concordia University, Montreal, QC, Canada. (M.K., C.S.).
2 School of Health, Concordia University, Montreal, QC, Canada. (M.K., C.S.).
3 Department of Neurology, Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany (L.G., A.V., B.S., C.S.).
4 Department of Neurology and Neurosurgery, McGill University, Montreal, QC, Canada (Y.I.-M.).

Background: Stroke is a leading cause of death and disability, with proprioceptive impairments affecting up to 64% of survivors. These impairments hinder sensorimotor recovery, significantly impacting poststroke quality of life. Proprioception depends on an integrated brain network but remains underexplored due to limitations in clinical assessments, hindering links between stroke-related damage and functional deficits. We combined quantitative proprioceptive measurements (arm position matching task) with connectome-based lesion-symptom mapping to identify white matter (WM) disconnection patterns underlying proprioceptive deficits in chronic sensorimotor stroke while controlling for motor impairment.

Methods: In this single-center observational study (Leipzig, Germany, 2015-2018), we investigated relationships between WM disconnection and proprioceptive deficits in chronic stroke survivors with paretic arm function using connectome-based lesion-symptom mapping and kinematic assessments. Lesions were manually delineated, and proprioception was quantified using the arm position matching task on the KINARM Exoskeleton. Patient-specific voxelwise WM disconnection maps were generated using the tractography-based lesion assessment standard, quantifying disconnection relative to a healthy WM connectome (n=1001; females=556; age=22-37 years). Proprioceptive scores were regressed against disconnection maps using voxelwise linear regressions (familywise error-corrected, controlled for age and sex). A secondary analysis included motor performance (visually guided reaching task) as a covariate to isolate proprioceptive-specific effects.

Results: Of 42 patients, 39 had valid arm position matching data, and 38 had valid visually guided reaching data included in the analyses (females=13; age=35-81 years). Arm position matching task scores were significantly associated with WM disconnection (d=0.58-1; P<0.005 familywise error; t=3.64-6.86) in a wide range of tracts previously implicated in proprioceptive function and beyond. Crucially, these associations persisted when controlling for motor performance using visually guided reaching task scores (d=0.44-0.93; P<0.05 familywise error; t=2.69-5.72).

Conclusions: We provide evidence that proprioceptive impairments in chronic stroke may arise from network-wide WM disconnection in key tracts mediating proprioceptive function. Our findings highlight the benefits of connectome-based lesion-symptom mapping for assessing stroke-related proprioceptive deficits and offer a framework for network-informed assessments of functional impairments that could guide targeted therapies poststroke.