Objective: We aim to investigate the association between structural connectivity and its influence on the freezing of gait (FOG) after globus pallidus internus (GPi) deep brain stimulation (DBS) in Parkinson’s disease (PD).

Background: Freezing of gait is one of the most debilitating symptoms of PD, increasing the risk of fracture and falls. The clinical outcomes of FOG after DBS exhibit variability, and the response to GPi DBS may be influenced by individual differences in white matter structures and axonal modulation by DBS.

Method: The study population included patients with PD who underwent GPi DBS from 2015 to 2020. We investigated the brain structural connectivity using diffusion tensor images and probabilistic tractography from pre-operative magnetic resonance imaging. We compared the structural connectivity between each brain region and volume of tissue activated(VTA) in patients with improved FOG and in those without after DBS stimulation. Additionally, it was investigated whether structural connectivity could predict the DBS stimulation effect using machine learning models.

Results: We included 58 patients with PD with GPi DBS, and 45 patients had FOG before DBS. Median age at surgery was 62.0 years. The patients without improved FOG showed higher connectivity between VTA and right inferior parietal cortex (median 0.024 vs 0.010, P=0.009), somatosensory cortex (median 0.005 vs 0.002, P=0.05), and rostral medial frontal cortex (median 0.027 vs 0.008, P=0.03). The patients with improved FOG showed higher connectivity with left lateral visual cortex (median 0.007 vs 0.025, P=0.04) and left ventral inferior parietal cortex (median 0.008 vs 0.034, P=0.06). The prediction of FOG improvement using structural connectivity after DBS surgery showed an accuracy of 0.72 (sensitivity 0.86, specificity 0.47) using logistic regression model and accuracy of 0.74 (sensitivity 0.89, specificity 0.47) using support vector machines.

Conclusion: We found that higher connectivity between the GPi target and the right parietal and somatosensory cortex is associated with less improvement in FOG, while left parietal and visual cortex is associated with improvement in FOG after GPi DBS in PD. Moreover, the prediction of FOG improvement using structural connectivity showed moderate accuracy in machine learning models, indicating potential for preoperative connectivity patterns to guide therapeutic expectations.

« Back to 2024 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/connectivity-based-analysis-of-the-impact-of-gpi-deep-brain-stimulation-on-the-freezing-of-gait-in-parkinsons-disease/