Preoperative Functional Connectivity Predicts Antiparkinson Drug Change after Deep Brain Stimulation

Area of Science:

• Neuroimaging
• Movement Disorders
• Biomarker Discovery

Background:

• Deep brain stimulation (DBS) offers symptomatic relief for Parkinson's disease (PD) motor deficits, yet outcomes vary significantly among patients.
• Current clinical predictors of DBS success are unreliable, explaining only a small fraction of outcome variability.
• There is a critical need for reliable biomarkers, such as those derived from functional magnetic resonance imaging (fMRI), to improve prediction accuracy for DBS outcomes.

Purpose of the Study:

• To investigate the utility of preoperative resting-state fMRI-derived motor network connectivity as a biomarker for predicting DBS treatment outcomes in PD.
• To determine if fMRI-based connectivity measures can enhance the predictive power of existing clinical factors.
• To assess the specificity of fMRI features in differentiating PD from Huntington's disease and healthy controls.

Main Methods:

• Retrospective analysis of resting-state fMRI data from 120 PD patients undergoing DBS targeting the subthalamic nucleus or globus pallidus.
• Computation of motor network connectivity and extraction of clinical predictors (age, sex, target, hemisphere, medication response).
• Regression and classification analyses to evaluate the predictive value of fMRI features alone and in combination with clinical factors, and to assess diagnostic specificity.

Main Results:

• fMRI features, combined with clinical predictors, explained significantly more outcome variance (adjusted R²=0.36) compared to clinical predictors alone (adjusted R²=0.13).
• Five specific cortico-basal ganglia-cerebellar network connectivity pairs predicted DBS outcomes and differentiated PD from controls and Huntington's disease with 67% and 73% accuracy, respectively.
• Exploratory analysis showed that intracerebellar connectivity dramatically improved PD classification accuracy (≥98%), despite being a less stable predictor of DBS outcomes.

Conclusions:

• Patient-specific motor network connectivity derived from fMRI significantly enhances the prediction of DBS outcomes in Parkinson's disease.
• These findings suggest that fMRI-based biomarkers can aid in the detection and monitoring of basal ganglia disorders.
• Future research should focus on validating and expanding these models for multi-parameter MR prediction to develop clinical decision support tools for DBS therapy.