Subthalamic beta dynamics mirror Parkinsonian bradykinesia months after neurostimulator implantation

Abstract

Background: Exaggerated oscillatory activity in the beta frequency band in the subthalamic nucleus has been suggested to be related to bradykinesia in Parkinson's disease (PD). However, studies seeking correlations between such activity in the local field potential and motor performance have been limited to the immediate postoperative period, which may be confounded by a stun effect that leads to the temporary alleviation of PD deficits.

Methods: Local field potentials were recorded simultaneously with motor performance in PD patients several months after neurostimulator implantation. This was enabled by the chronic implantation of a pulse generator with the capacity to record and transmit local field potentials from deep brain stimulation electrodes. Specifically, we investigated oscillatory beta power dynamics and objective measures of bradykinesia during an upper limb alternating pronation and supination task in 9 patients.

Results: Although beta power was suppressed during continuously repeated movements, this suppression progressively diminished over time in tandem with a progressive decrement in the frequency and amplitude of movements. The relationship between changes within local field potentials and movement parameters was significant across patients, and not present for theta/alpha frequencies (5-12 Hz). Change in movement frequency furthermore related to beta power dynamics within patients.

Conclusions: Changes in beta power are linked to changes in movement performance and the sequence effect of bradykinesia months after neurostimulator implantation. These findings provide further evidence that beta power may serve as a biomarker for bradykinesia and provide a suitable substrate for feedback control in chronic adaptive deep brain stimulation. © 2017 The Authors. Movement Disorders published by Wiley Periodicals, Inc. on behalf of International Parkinson and Movement Disorder Society.

Keywords: Parkinson's disease; bradykinesia; deep brain stimulation; local field potential.

Figure 1

Suppression of oscillatory activity in a continuous alternating pronation and supination (rotometer) task months after neurostimulator implantation. (A) Schematic drawing of rotometer‐task paradigm. (B) Movement reactivity spectral data (calculated by subtracting rest from movement power spectral data and normalized by dividing by rest power spectral data taken from in between the 30‐second movement blocks) averaged over all 3 blocks of movement and then across the 9 participants. Shadow indicates standard error of mean. Averaging was performed irrespective of motor performance.

Figure 2

Changes in motor performance. Mean activity over the 2 10‐second windows (W1, window 1; W2, window 2) per block are displayed for each individual. Values were normalized by individual mean subtraction (see Methods). Below each window means + standard error of mean are given for the respective window. Analysis of variance revealed significant effect of window, no effect of block or window‐block interaction (see Results).

Figure 3

Example trace of beta power dynamics alongside motor impairment. Gray boxes indicate 10‐second windows, the means of which were used in further analysis. (A) Movement trace of a PD patient performing continuous and alternating pronation and supination movements for 30 seconds. Raw movement trace shown was detrended to allow better assessment of movement amplitude. (B) Trace of individual beta power (patient‐specific beta peak during movement performance ± 5 Hz) smoothed using an overlapping, sliding average window to capture the general trend in beta activity over time. Smoothing was applied for visualization purposes only. Unlike Figure 1, beta power is not normalized by rest power.

Figure 4

Relationship between change in individual beta power and change in movement characteristics across and within participants. The latter were frequency (A and B) and amplitude (C). Individual beta power was defined as patient‐specific beta peak during movement performance ± 5 Hz. All 3 blocks of performance by each respective individual are displayed. Circled data points were taken from ipsilateral hemispheres. Across the participant analysis (A and C), red = 1st 30‐second block of movement; blue = 2nd 30‐second block of movement; khaki = 3rd 30‐second block of movement. Within‐participant analysis (B): each color accounts for 1 individual. Rho values taken from Spearman's correlation for each individual were Fisher‐rho‐to‐z‐transformed and compared against 0 using a 1‐sample, 2‐tailed t‐test to arrive at the P value shown.