Objective: To determine whether pre-operative brain morphometry, combined with baseline neuropsychological performance, can predict post-operative cognitive performance in PD patients undergoing DBS using a sparse predictive modeling framework.

Background: Cognitive decline is a recognized risk of deep brain stimulation (DBS) for Parkinson’s disease (PD), yet preoperative prediction of individual cognitive trajectories remains limited. Baseline neuropsychological performance and demographic factors provide some prognostic value, but structural neuroimaging data remains underutilized as potential predictors.

Method: Twelve patients with PD undergoing DBS (7 subthalamic nucleus, 5 globus pallidus internus) completed comprehensive neuropsychological testing before and after surgery, along with pre-operative structural MRI. FreeSurfer-derived subcortical volumes and Desikan-Killiany cortical thickness measures were normalized and combined with baseline neuropsychological subscale scores as candidate predictors of post-operative cognitive performance. Relaxed lasso regression (lasso feature selection followed by ordinary least squares refit) was evaluated using leave-one-out cross-validation, with feature selection performed independently within each fold to prevent information leakage.

Results: Baseline neuropsychological scores alone predicted post-operative global cognition with moderate accuracy (LOO R² = 0.52, r = 0.81, p = 0.001). Adding structural brain measures improved prediction (LOO R² = 0.75, r = 0.87, p < 0.001, RMSE = 4.12). The combined model consistently retained four predictors: baseline global cognitive score (selected in 100% of folds), primary motor cortex thickness (100%), baseline language score (92%), and cuneus volume (83%).

Conclusion: In this small, single-site cohort, a sparse set of pre-operative structural brain features, particularly primary motor cortex thickness and cuneus volume, added predictive value for post-operative cognitive performance beyond baseline neuropsychological scores alone. These preliminary findings suggest that readily obtainable MRI measures may help identify patients at greater risk for poor cognitive performance after DBS, though validation in larger, independent samples is essential before clinical application.

« Back to 2026 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/structural-brain-measures-predict-cognitive-performance-after-deep-brain-stimulation-for-parkinsons-disease/