White matter fractional anisotropy predicts Parkinson's disease medication changes with deep brain stimulation.

Abstract

Diffusion-based tractography has been used to predict deep brain stimulation (DBS) outcomes in Parkinson's disease (PD), but the predictive value of fractional anisotropy (FA) maps, which can be generated with far less computation, remains underexplored. We retrospectively analyzed preoperative diffusion MRI from 143 patients with PD who underwent DBS between 2013 and 2023. Whole-brain and tract-specific FA measures were extracted and evaluated against the pre-to-postoperative change in levodopa equivalent daily dose (ΔLEDD). Nested regression models compared clinical predictors alone with models that added whole-brain FA or stability-selected tract-specific FA. Tract-specific FA explained an additional 2.3-3.4% of ΔLEDD variance (p = 0.038), whereas whole-brain FA did not (p = 0.497); effects were in the same direction, though smaller and no longer significant, after adjustment for gold-standard motor response to dopaminergic medications. The uncinate fasciculus and cingulum of the cingulate gyrus were most consistently selected. Preoperative tract-specific FA may complement established clinical predictors of DBS outcome in PD.