Long‐Term Stability of Spatial Distribution and Peak Dynamics of Subthalamic Beta Power in Parkinson's Disease Patients

J. Behnke, Robert L. Peach, Jeroen Habets, Johannes L. Busch, Jonathan Kaplan, Jan Roediger, Varvara Mathiopoulou, Lucia K. Feldmann, Moritz Gerster, Juliette Vivien, Gerd‐Helge Schneider, Katharina Faust, Patricia Krause, Andrea A. KühnMovement 2025 · DOI

TL;DR

This study investigates the long-term stability of subthalamic beta oscillations, a key biomarker for Parkinson's Disease (PD), in patients receiving deep brain stimulation (DBS). The findings suggest that beta power is a stable chronic biomarker, particularly high-beta peaks, making it reliable for guiding DBS programming and adaptive stimulation even years after surgery.

What Problem Does This Paper Address?

Parkinson's Disease (PD) is characterized by motor symptoms like bradykinesia and rigidity, which are strongly linked to abnormal oscillatory activity in the subthalamic nucleus (STN), specifically in the beta frequency band (13-35 Hz). Beta power has emerged as a crucial biomarker, showing promise for optimizing DBS by providing a feedback signal for adaptive stimulation and guiding the selection of electrode contacts. However, for these applications to be clinically effective and reliable, understanding the long-term stability of this beta activity over months and years after surgery is essential. Previous research has largely focused on the acute postoperative period, leaving a gap in our knowledge about how beta power and its characteristics evolve over extended durations. This paper aims to fill that gap by analyzing the longitudinal dynamics of beta peak parameters and their spatial distribution along DBS electrodes in PD patients.

Methods

The researchers recorded local field potentials (LFPs) from 12 bipolar channels per hemisphere in 33 PD patients across multiple sessions: at baseline (0 months post-surgery), and at 3, 12, and 18-44 months post-surgery. They analyzed the periodic beta band power (13-35 Hz), focusing on peak frequency (CF) and peak power (area under the curve ± 3 Hz around the peak CF). To assess spatial stability, they examined the distribution of beta power across the electrode contacts. They employed a spectral parameterization software (specparam) to isolate periodic components and identify peak characteristics. For statistical analysis, they used paired Wilcoxon signed-rank tests, Friedman tests with post-hoc Wilcoxon tests, and repeated-measures ANOVAs, with Holm correction for multiple comparisons.

Datasets

The study utilized longitudinal LFP data from 33 PD patients who underwent bilateral subthalamic DBS. The data was collected across four time points post-surgery: 0 months (within 12 days of lead implantation), 3 months, 12 months, and >18 months (ranging from 18 to 44 months post-surgery). Subgroup analyses were performed based on the consistency of recording sessions across these time points.

Key Insights and Ideas

  • Beta Power as a Stable Biomarker: The study strongly supports the idea that beta power, especially in the high-beta range, is a stable and reliable biomarker for PD patients over the long term, even years after DBS implantation.
  • Early Postoperative Fluctuations: Significant changes in beta power and peak frequency are observed primarily in the early postoperative period (up to 3 months), likely due to the "stun effect" or microlesional impact of lead implantation. This period is characterized by increased beta power and greater variability.
  • High-Beta Peaks are More Reliable: High-beta peaks (21-35 Hz) demonstrated greater consistency in detection and stability in their frequency over time compared to low-beta peaks (13-20 Hz). This makes them a potentially more robust target for adaptive DBS programming.
  • Spatial Stability is Key for Programming: The spatial distribution of beta power along the electrode contacts, particularly for ring electrodes, shows increasing stability over time. This is crucial for beta-guided contact selection, ensuring that the chosen contacts consistently reflect the underlying brain activity.
  • Active Contacts Show Higher Beta Power: Critically, contacts actively used for therapeutic stimulation consistently exhibited higher normalized beta power than inactive contacts, reinforcing the utility of beta power for guiding contact selection in DBS programming.

Key Results

  • Beta Power Increase: Beta peak power significantly increased between 0 and 3 months post-surgery (P < 0.0001) and then stabilized.
  • Peak Frequency Shifts: While highest beta peaks showed no significant directional shifts on average, about 50% of them shifted by more than 5 Hz, particularly in the early period. Low-beta peaks showed more variability and disappearance than high-beta peaks, especially in later sessions.
  • Spatial Stability: The ranked order of beta power across bipolar channels showed positive correlations over time, with lowest correlations observed in the early postoperative period. Segmented contacts with maximal beta power showed increasing stability along the vertical axis (up to 82% remaining in their level) and directional stability (64-71% remaining in the same direction).
  • Active vs. Inactive Contacts: Normalized beta power was significantly higher in active stimulation contacts compared to inactive ones across all follow-up sessions (P < 0.0001). Contacts with maximal beta power were chosen for therapeutic use in 45-69% of hemispheres.

Limitations and Future Work

Limitations: * Short Recording Windows: The BSSU mode limited recordings to 20-second windows, capturing only a snapshot of beta oscillations and potentially missing diurnal fluctuations. * Bipolar LFPs Only: The study primarily used bipolar LFPs, with monopolar estimations based on weighting methods, which have limitations, especially for ring contacts. Future advancements enabling monopolar power derivation would be beneficial. * Unclear Stabilization Timing: The exact timing of stabilization after surgery remains unclear due to the absence of intermediate measurements. * Limited Late Sessions: Analysis of late sessions (>18 months) was limited by smaller patient numbers in some subgroups.

Future Work: * Longer-Term Monitoring: Further research in larger patient cohorts and with longer follow-up periods is needed to better understand the factors influencing beta oscillations, such as medication fluctuations or long-term DBS effects. * Multimodal Programming: Exploring multimodal DBS programming that combines biomarkers from LFPs with imaging data could offer a more efficient approach. * Investigating Low-Beta Variability: Further investigation is needed to understand why low-beta peaks show higher variability and disappearance, and whether this relates to medication or DBS effects. * Directional Stimulation Efficacy: The implications of horizontal long-term fluctuations in directional contacts on therapeutic efficacy require further investigation.

Summary updated

Abstract

Our findings indicate that beta power is a stable chronic biomarker usable for beta-guided programming. For adaptive stimulation, high-beta peaks might be more reliable over time. Greater stability of beta power, center frequency, and spatial distribution beyond an initial stabilization period suggests that the microlesional effect significantly impacts neuronal oscillations, which should be considered in routine clinical practice when using beta activity for automated programming algorithms. © 2025 The Author(s). Movement Disorders published by Wiley Periodicals LLC on behalf of International Parkinson and Movement Disorder Society.
BibTeX
@inproceedings{behnke2025longterm,
  author = {J. Behnke and Robert L. Peach and Jeroen Habets and Johannes L. Busch and Jonathan Kaplan and Jan Roediger and Varvara Mathiopoulou and Lucia K. Feldmann and Moritz Gerster and Juliette Vivien and Gerd‐Helge Schneider and Katharina Faust and Patricia Krause and Andrea A. Kühn},
  title = {{Long‐Term Stability of Spatial Distribution and Peak Dynamics of Subthalamic Beta Power in Parkinson's Disease Patients}},
  booktitle = {Movement},
  year = {2025},
  doi = {10.1002/mds.30169},
}
Figure