Intraoperative Characterization of Subthalamic Nucleus-to-Cortex Evoked Potentials in Parkinson's Disease Deep Brain Stimulation

Abstract

Deep brain stimulation (DBS) of the subthalamic nucleus (STN) is a clinically effective tool for treating medically refractory Parkinson's disease (PD), but its neural mechanisms remain debated. Previous work has demonstrated that STN DBS results in evoked potentials (EPs) in the primary motor cortex (M1), suggesting that modulation of cortical physiology may be involved in its therapeutic effects. Due to technical challenges presented by high-amplitude DBS artifacts, these EPs are often measured in response to low-frequency stimulation, which is generally ineffective at PD symptom management. This study aims to characterize STN-to-cortex EPs seen during clinically relevant high-frequency STN DBS for PD. Intraoperatively, we applied STN DBS to 6 PD patients while recording electrocorticography (ECoG) from an electrode strip over the ipsilateral central sulcus. Using recently published techniques, we removed large stimulation artifacts to enable quantification of STN-to-cortex EPs. Two cortical EPs were observed - one synchronized with DBS onset and persisting during ongoing stimulation, and one immediately following DBS offset, here termed the "start" and the "end" EPs respectively. The start EP is, to our knowledge, the first long-latency cortical EP reported during ongoing high-frequency DBS. The start and end EPs differ in magnitude (p < 0.05) and latency (p < 0.001), and the end, but not the start, EP magnitude has a significant relationship (p < 0.001, adjusted for random effects of subject) to ongoing high gamma (80-150 Hz) power during the EP. These contrasts may suggest mechanistic or circuit differences in EP production during the two time periods. This represents a potential framework for relating DBS clinical efficacy to the effects of a variety of stimulation parameters on EPs.

Keywords: Parkinson’s disease; deep brain stimulation; electrocorticography; evoked potential; high-frequency stimulation; subthalamic nucleus.

FIGURE 1

Electrode placement. ECoG (A) and DBS (B) electrode locations shown for the 6 subjects in MNI space. For ECoG strips, electrode 1 was the most posterior and electrode 8 was the most anterior. The 4 gray bands on each DBS contact (B) represent the contacts, with electrode 0 the deepest and electrode 3 the most superficial (Medtronic naming conventions). GPe, globus pallidus external segment; GPi, globus pallidus internal segment; STN, subthalamic nucleus.

FIGURE 2

Stimulation and evoked potential measurement. (A) Monophasic stimulation was delivered in a bipolar configuration to DBS electrodes (purple dots). Stimulation occurred in 5 s bursts at 4 amplitudes (purple bars), though only EPs evoked by 3 V stimuli were analyzed. Signals were recorded at cortical electrodes (orange dots). (B) Raw trial (orange) shows stimulation artifacts, which were removed by an unsupervised dictionary-based learning algorithm (black). (C) The average of 30 trials (top trace, ±SEM, z-scored) was used to identify EPs. A long baseline period (blue) prior to stimulation onset (purple vertical line) and the 100 ms windows immediately after stimulation onset (t = 0 s, green) and offset (t = 0.5 s, red) were the regions of interest. The peak-to-trough amplitude was computed for each period (vertical red and green lines), as well as the latency to peak and trough components for the two EPs (horizontal red and green dashed lines, C-i). The RMS amplitude was extracted for these time periods for each z-scored individual trial (example trial shown in C-ii). (D) The peak-to-trough of the average trace and the median RMS of all trials (separate medians for the start and end EPs) were highly correlated across subjects in a non-parametric test (generalized linear model in black with confidence intervals in gray, r² and p from Spearman correlation).

FIGURE 3

STN DBS evokes start and end EPs. Z-scored start (A) and end (B) EPs (average of 30 trials) are shown for each electrode (columns, contacts were in different BAs for each subject – see Figure 1), subject (rows), and DBS stimulation electrode pair (line type). EPs in pink had a statistically significantly larger magnitude than baseline deviations (p < 0.05, FDR corrected by subject). The one trace in yellow was also statistically different than baseline, but it had a lower magnitude. Median RMS (C) and latencies to negative and positive deflections on the average traces (D) were compared between the start and end EPs; significant (p < 0.05, FDR corrected) differences indicated by stars. (E) The average of all significant start (green) and end (red) EPs.

FIGURE 4

Effects of brodman Area on EPs. The median RMS (A), percent of EPs that differed significantly from baseline amplitudes (B), latency to negative deflection (C), and latency to positive deflection (D) were compared between start and end EPs with recording electrodes grouped by Brodman area across subjects. Brodmann areas with inconsistent coverage across subjects were pooled and shown for comparison but not included in statistical analysis. Black stars indicate significant (p < 0.05, FDR corrected) differences between start and end EP measures within each BA. Red stars indicate significant (p ≤ 0.05, FDR corrected) post hoc comparisons between end EPs in different regions after a Kruskal-Wallis test revealed dependence of both latency measures on BA. The mean of all start (green) and end (red) EPs for each electrode in each BA are shown in (E).

FIGURE 5

Relationship between EP magnitude and gamma power. A mixed linear model was used to assess the relationship between EP magnitude low (30–80 Hz; A,B) and high (80–130 Hz; C,D) gamma power during the start (A,C) and end (B,D) EP intervals. The model controlled for random effects of subject, adjusting for intercept and slope. Each plot shows the model’s predicted fits for each subject as well as the overall model in black. Overall model and slope are reported along with 95% confidence intervals.