Abstract TP325: White Matter Disconnectivity as a Biomarker of Aiming Errors in Spatial Neglect

Introduction: Spatial neglect (SN) is asymmetric action, imagery, or orientation after brain lesion, causing disability. People with SN are half as likely to return home, have longer hospitalizations, one-third the community mobility, and require triple the caregiver supervision, yet >50% remain undiagnosed. Motor-intentional “Aiming” SN is particularly under-recognized, despite responsiveness to targeted interventions such as prism adaptation therapy. We aimed to identify white matter (WM) tract disconnectivity patterns that predict Aiming spatial errors to detect high-risk patients. Hypothesis: Disconnectivity involving superior colliculus, frontal-basal ganglia circuits, cortical frontal networks, and anterior/mid-corpus callosum would best predict Aiming SN. Methods: Eighteen right-brain stroke survivors (mean age 62.7±11.6, 9 male, 2.2±1.3 months post-stroke) with SN were included across three sites (NCT05983185). MRI lesion maps were overlaid on HCP1065 tractography atlas to calculate the percentage of each WM tract affected. Tracts lesioned in <5 participants were excluded. Penalized regression (LASSO, ridge; 5-fold cross-validation) in SPSS identified predictors of motor exploratory Catherine Bergego Scale, adjusting for NIHSS, age, and sex. Results: LASSO selected only right fornix disconnectivity and total number of tracts affected as primary predictors. Ridge regression confirmed these variables and additionally identified moderate associations with right middle longitudinal fasciculus, frontal aslant, corticopontine (parietal), and corticospinal (anterior) tracts. Several tracts, including right SLF3 and optic radiation, showed negative coefficients, suggesting differential involvement across SN subtypes. Both models exhibited overfitting in holdout data due to small sample size, highlighting need for validation. Conclusion: The number of WM tracts affected was linked to degree of SN aiming errors, as was right fornix disconnectivity, the latter aligned with fornix’s role in visuospatial memory and integrative navigation. Ridge analysis suggested broad right-hemisphere frontal-parietal and brainstem network contributions. Negative coefficients may indicate either suppression effects or tracts correlated with perceptual “Where” SN rather than “Aiming” SN. Results require validation in a larger cohort and suggest that mapping WM disconnectivity may offer a neuroimaging biomarker for early identification and targeted rehabilitation of Aiming SN.