Combining Multimodal Biomarkers to Guide Deep Brain Stimulation Programming in Parkinson Disease

Area of Science:

• Neuroscience
• Biomedical Engineering
• Clinical Neurology

Background:

• Deep brain stimulation (DBS) programming is a manual, time-intensive process.
• Optimizing DBS lead contact selection is crucial for effective Parkinson disease treatment.
• Subthalamic nucleus (STN) local field potentials (LFPs) show potential for guiding DBS programming.

Purpose of the Study:

• To evaluate an algorithmic approach for selecting optimal DBS stimulation contacts.
• To investigate the predictive value of combined STN LFP features and imaging markers.
• To predict key clinical DBS parameters: efficacy, therapeutic window, and side-effect threshold.

Main Methods:

• Recorded STN LFPs from multicontact DBS leads in 27 hemispheres at rest and during movement.
• Analyzed features across multiple frequency bands (alpha, beta, gamma, HFO).
• Incorporated anatomical stimulation sweet spots and combined electrophysiological and imaging markers.

Main Results:

• Both resting and movement-state LFP features improved prediction accuracy.
• Resting gamma, modulated beta, and movement-modulated HFO were highly predictive.
• The algorithm achieved nearly 90% accuracy in identifying optimal contacts, outperforming single-marker approaches.
• Combining electrophysiological and imaging markers further enhanced prediction.

Conclusions:

• Algorithmic selection of LFP and imaging markers offers an efficient DBS programming strategy.
• This approach can improve clinical efficiency and patient outcomes in DBS therapy.
• LFP-guided DBS programming represents a significant advancement in Parkinson disease management.