Objective: To explore the feasibility and efficacy of using features extracted from chronic LFP recordings to predict the optimal stimulation contact and amplitude for patients with PD receiving globus pallidus internus (GPi) DBS.

Background: GPi DBS is an effective treatment for advanced PD. However, selecting the optimal stimulation parameters is a laborious, trial-and-error process. Recently, an algorithm was developed to predict the optimal DBS settings based on pallidal LFPs [1]. However, the real-life performance of the algorithm has not been examined in a prospective trial.

Method: The algorithm surveys LFPs between the alpha and low gamma frequency ranges; input features include initial power, max change in power, rate of power changes, and frequency of the max changes.

Fifteen advanced PD patients receiving GPi DBS via a commercially available LFP sensing-enabled device will be enrolled. At the first-month programming visit, resting LFPs and LFPs in response to increasing amplitudes of therapeutic stimulation will be recorded. Each patient will receive 2 new settings programmed by the DBS provider as usual and 1 algorithm-predicted setting based on the LFP recordings; the order of the settings will be randomized. Patients will be instructed to stay on each new setting for at least 1 week if tolerating. Patients will be followed monthly and their best setting at the subsequent programming visits will be noted.

Results: To date, 3 patients (2 male, 1 female; 1 left, 2 right; 2 first implant, 1 second implant) have been enrolled. At the 2-month programming visit, 1 patient preferred the algorithm-selected contact; 1 reported equally good outcomes with both the algorithm-selected contact and a different contact; the other preferred a different contact. The frequencies where the maximum change of power occurred for each patient were all within the beta frequency range. The contacts selected by the algorithm were also the contacts where the highest beta peak powers were recorded. While on the algorithm-predicted settings, 2 out of 3 patients increased the amplitude above the algorithm prediction. At the 4-month visit, the contact configurations remained unchanged for all 3 patients.

Conclusion: The preliminary data suggest it may be feasible to use an LFP-based algorithm to predict the optimal stimulation contact for PD patients undergoing GPi DBS.

References: [1] Cagle JN, Johnson KA, Almeida L, Wong JK, Ramirez-Zamora A, Okun MS, Foote KD, de Hemptinne C. Brain Recording Analysis and Visualization Online (BRAVO): An open-source visualization tool for deep brain stimulation data. Brain Stimul. 2023 May-Jun;16(3):793-797. doi: 10.1016/j.brs.2023.04.018. Epub 2023 Apr 24. PMID: 37100201.

« Back to 2024 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/automated-selection-of-deep-brain-stimulation-dbs-parameters-for-parkinsons-disease-pd-using-pallidal-local-field-potentials-lfp-a-prospective-pilot-study/