Coupling between beta and high-frequency activity in the human subthalamic nucleus may be a pathophysiological mechanism in Parkinson's disease

Abstract

In Parkinson's disease (PD), the oscillatory activity recorded from the basal ganglia shows dopamine-dependent changes. In the "off" parkinsonian motor state, there is prominent activity in the beta band (12-30 Hz) that is mostly attenuated after dopaminergic therapy ("on" medication state). The on state is also characterized by activity in the gamma (60-80 Hz) and high-frequency (300 Hz) bands that is modulated by movement. We recorded local field potentials from a group of 15 PD patients (three females) treated with bilateral deep brain stimulation of the subthalamic nucleus, using a high sampling rate (2 kHz) and filters suitable to study high-frequency activity (0.3-1000 Hz). We observed high-frequency oscillations (HFOs) in both the off and on motor states. In the off state, the amplitude of the HFOs was coupled to the phase of the abnormal beta activity. The beta-coupled HFOs showed little or even negative movement-related changes in amplitude. Moreover, the degree of movement-related modulation of the HFOs correlated negatively with the rigidity/bradykinesia scores. In the on motor state, the HFOs were liberated from this beta coupling, and they displayed marked movement-related amplitude modulation. Cross-frequency interactions between the phase of slow activities and the amplitude of fast frequencies have been attributed an important role in information processing in cortical structures. Our findings suggest that nonlinear coupling between frequencies may not only be a physiological mechanism (as shown previously) but also that it may participate in the pathophysiology of parkinsonism.

Figures

Figure 1.

Axial T1 section and parasagittal MRI reconstruction showing the correct placement of the DBS in the STN of a representative patient (patient 15).

Figure 2.

Subthalamic beta and HFOs in the off and on motor states have different characteristics. A, Temporal evolution of the power spectrum of patient 1 (left STN, dorsal contacts 2–3) in both motor states (off and on) during rest periods. The bottom images represent the 3–100 Hz frequency range and the top images represent the 100–400 Hz frequency range. Power values are shown in a logarithmic scale, adjusted for each frequency range. HFOs in the 250–350 Hz range can be observed in both motor states. The thin horizontal lines observed at 50, 150, 250, and 350 Hz are artifacts (at the mains frequency and its odd harmonics). B, Mean power spectrum (logarithmic scale) in the dorsal contact pairs (0–400 Hz), in both motor states (n = 22). The artifacts at mains frequency and its harmonics have been removed and replaced by interpolated values. Most differences between the on and off plots are centered in the beta, gamma, and HFO ranges. The off low-beta peak attenuates in the on state, whereas the high beta activity does not change. The small gamma differences observed are attributable to three subjects. A peak in the HFO range can be observed in both states, although at different frequencies. C, Topographical distribution of the HFOs in the whole group in both motor states (off and on). D, Comparison of the mean relative power values of low- and high-beta peaks in both motor states (off and on) in different contact pairs (*p < 0.05). All contact pairs were included in the analysis; 2–3 means dorsal contact pairs, 1–2 means intermediate contact pairs, and 0–1 means ventral contact pairs.

Figure 3.

A, Filtered LFP from a sample patient (patient 9) in the off motor state highlighting the phase reversal of the low-beta activity (right) and the HFOs (left) in the dorsal STN. B, Phase difference histogram (360 bins, 300 s segment) from the same patient showing the phase reversal around contact 2 in the HFOs. The mean phase difference between contact pairs 2–3 and 1–2 is ∼180° (π radians). The horizontal line indicates the mean level in an ideally random distribution. C, Phase difference histogram from the same patient and contact pairs showing the phase reversal in the low-beta activity around electrode 2 (same parameters as in B).

Figure 4.

HFOs in the off motor state (but not in the on motor state) display amplitude modulation at the frequency of the low-beta power peak. A, Temporal evolution of the power spectrum from patient 11 (right STN, dorsal contacts 2–3) during the off-to-on transition (marked with a black arrow). Low-beta and HFO bands show a similar temporal pattern during the transition. B, Power spectrum of the envelope of the HFOs (variations in amplitude, as shown in the small insert) from the same patient and contacts (right STN, dorsal contacts 2–3) at rest in both motor states (off and on). A marked low-beta peak can be observed in the off motor state (marked with “x”), whereas in the on state there is a smaller and more diffuse high-beta increment (marked with “o”). C, Relationship between the frequency of both beta bands in the raw signal power spectrum (blue dots for low beta and green dots for high beta) and the frequency of the peak (characteristic scale) observed in the power spectrum of the HFO envelope, in the off (left) and on (right) states. In the off state, the highly significant linear correlation observed with the low-beta peak contrasts with the lack of correlation with the high-beta peak. In the on state, a smaller (but significant) correlation is observed in the high-beta range (there is no low-beta power peak in this motor state).

Figure 5.

Cross-frequency coupling between the phase of the low-beta activity and the amplitude of the HFOs at rest is evident in the off motor state. A, Phase to amplitude representations of the MI at rest in a representative patient (patient 10, left STN, contacts 2–3, 1–2, and 0–1), both in the off (top images) and on state (bottom images). All plots use the same scale. Please note that cross-frequency coupling is 2 orders of magnitude higher in the off state. B, Topographical distribution of MI values from the macroelectrode contact pairs in both motor states (off and on) showing the predominance in dorsal contact pairs in the off state (*p < 0.05). C, Two additional representations of the beta–HFO coupling in both motor states (patient 10, left STN, contact pair 2–3). Top, Average of 500 ms segments of the raw signal centered at the peaks of the HFO. The amplitude of the HFOs is greater during the valley and the ascending curve of the beta oscillations in the off state. This relationship is lost in the on state. Bottom, Mean normalized power of the 100–400 Hz range time-locked to the phase minimums of the beta activity.

Figure 6.

Movement-related changes in HFO power and beta–HFO coupling in both motor states. A, Average of the single-sweep Stockwell transforms of a sample patient (P14, left STN, intermediate contacts 1–2) in both motor states (off and on) during movement. Time 0 indicates the beginning of movement. A perimovement power increase in the HFO band (∼340 Hz) can only be observed in the on state. Please note the logarithmic scale in the y-axis. B, Linear representation from the whole group of patients of the normalized movement-related power changes in the HFO range in both motor states (off and on). Again, a perimovement power increase can only be observed in the on state. C, Bar diagram showing the relative movement-related power changes in all the nuclei studied in both motor states (off and on). In the on state, most nuclei show movement-related power increases, whereas in the off state the changes are mostly minimal or absent. D, Phase to amplitude representations of the MI values in premovement (Pre-) and perimovement periods (Mov-) from patient 14 (left STN, intermediate contacts 1–2). A perimovement decrease is observed in both motor states. However, off perimovement values are still much higher than on premovement values. E, Mean MI values premovement and perimovement in both motor states (off and on, *p < 0.05) showing that, regardless of the relative perimovement decrease, off values are still higher than on values.

Figure 7.

Regression plot displaying the relationship between the movement-related changes in HFO power and the rigidity/bradykinesia scores.