Cortical Dynamics Within and Between Parietal and Motor Cortex in Essential Tremor

Abstract

Background:

Evidence from functional imaging in essential tremor suggests that activity within parietal and motor cortices may be associated with worsening of tremor at increased visual feedback.

Objectives:

Examine how cortical oscillations within these regions and the connectivity between these regions is associated with worsening of tremor in essential tremor in response to high visual feedback.

Method:

The study included 24 essential tremor participants and 17 controls. We measured cortical activity and tremor magnitude at low and high feedback conditions. Cortical activity was measured using high-density electroencephalogram and isolated using source localization.

Results:

Changes in power across feedback in the 4–12 Hz and 12–30 Hz bands were reduced within the contralateral motor cortex of essential tremor patients compared to controls. The 12–30 Hz bidirectional connectivity between the parietal and contralateral motor cortex was decreased in essential tremor patients. Worsening of tremor from low to high visual feedback was associated with 4–12 Hz activity in contralateral motor cortex. The greatest separation between groups was found when using the difference of the contralateral motor cortex activity at high and low feedback, rather than either feedback condition alone.

Conclusion:

Our findings provide new evidence that tremor in essential tremor is associated with reduced power across feedback in the motor cortex and reduced connectivity between the parietal and motor cortices. Combined with previous work on the cerebellar-thalamocortical motor circuit, our findings suggest that the network level disturbances associated with essential tremor extend to the cortico-cortical pathway between the parietal cortex and motor cortex.

Essential tremor (ET) is the most common adult-onset movement disorder,¹ characterized by postural and kinetic tremor of the upper limbs² and sometimes in the head, lower limbs, face, and trunk.^3,4 In addition to tremor, several studies have reported a wide range of behavioral and cognitive changes in ET.^3,5–8 Multiple neurophysiological and neuroimaging studies examining ET subjects have reported changes in the motor cortex, thalamus, and cerebellum,^9–13 which together form the cerebellar-thalamo-motor cortical pathway. More recently, ET has also been associated with changes in functional activity in regions beyond the cerebellar-thalamo-motor cortical pathway, including the parietal cortex.¹⁴ Despite our improved understanding of the key areas that are affected in ET, we know less about how cortical neuronal oscillations and connectivity within and between the parietal cortex and other cortical areas relate to the magnitude of tremor in ET.

Previous electroencephalography (EEG)^15–20 and magnetoencephalography (MEG)^21,22 studies involving healthy subjects have consistently shown that movement preparation or execution modulates beta-band (12–30 Hz) activity in parietal and motor cortex, leading to a sharp decrease in beta-band power, which increases back to a normal level once the movement is completed.²³ Recent electrocorticography (EcOG)-based²⁴ and local field potential–based²⁵ studies examining movement disorders, such as ET, Parkinson’s disease (PD), and dystonia, have demonstrated disease-specific changes in modulation of beta-band power, although this observation is not observed in all patients. In addition, studies using EEG and electromyography (EMG) have demonstrated cortico-muscular coherence in the tremor band (4–12 Hz)^26,27 and in the beta-band²⁸ in ET subjects.

Based on the results from neuroimaging and neuropathology,^{12,13,29–40} ET has been predominantly seen as a disorder associated with changes within the cerebello-thalamo-cortical network. A recent functional MRI (fMRI) study¹⁴ found changes in activation within the dorsal visual processing stream within parietal and motor cortex, and it was found that ET subjects’ tremor worsened acutely with increases in visual feedback.¹⁴ Also, visually sensitive tremor amplitude was predicted by the fMRI signal in the parietal and motor cortices, along with the cerebellum.¹⁴ Here, we focus on examining cortical neuronal oscillations and directionally specific connectivity within and between parietal and motor cortices in ET patients with different levels of tremor severity using an experimental motor control paradigm consisting of low and high visual feedback. We test three hypotheses: (1) increasing visual feedback level will differentially influence 4–12 Hz and 12–30 Hz cortical oscillations in the parietal and motor cortex in ET and control subjects; (2) ET will have reduced directionally specific connectivity between parietal and motor cortex; and (3) the magnitude of tremor in ET will be related to 4–12 Hz and/or 12–30 Hz cortical oscillations. The findings of this study provide new insight into the electrophysiological features within and between parietal and motor cortex that relate to tremor severity in ET.

Patients and Methods

Subjects

The study included 24 ET and 17 controls matched for age and sex (Table 1). ET was diagnosed by movement disorder specialists based on established criteria.⁴¹ Tremor was assessed using the Fahn, Tolosa, Marin Tremor Rating Scale (FTM-TRS),⁴² and cognition was measured using the Montreal Cognitive Assessment (MoCA)⁴³ (Table 1). All procedures were approved by the institutional review board in accord with the Declaration of Helsinki.

TABLE 1.

Demographics and ET characteristics

	Group
Measure	ET	Control
Sample size	24	17
Sex	(12M/12F)	(8M/9F)
Age, years (mean/SD)	(66.04/11.56)	(62.94/8.64)
MoCA (mean/SD)	(27.56/3.02)	(27.50/1.21)
FTM (mean/SD)	(37/18.40)a	(0.50/0.96)
Disease duration, years (mean/SD)	(24.28/20.68)	—

^aIndicates P < 0.0001.

Experiment Paradigm

The Supporting Information provides further details for all methods. Subjects sat upright and viewed a monitor displaying their force. Force was measured from the thumb and index finger, and other digits were flexed. The monitor displayed a force bar and target bar. Subjects matched the force bar to the target bar by producing 15% of their maximum voluntary contraction (MVC) force. The paradigm consisted of two feedback levels: low and high. Feedback was controlled by manipulating the visual angle while maintaining distance from the monitor constant. Visual angles for low and high visual feedback were 0.039 and 6.9 degrees.⁴⁴ High visual feedback provided subjects with a greater level of information regarding their force production.

The experiment included five blocks of 10 trials at each visual feedback level. Each trial was 5 seconds, and the resting period between consecutive trials was 10 seconds. Low and high visual feedback blocks were alternated and pseudo-randomized across subjects. During rest, the target bar was white and force bar was red. When the red bar turned green, subjects produced pinch grip force to match the green force bar with the target. Subjects were required to maintain this force for 5 seconds, followed by rest. During each trial, force and EEG were recorded and time-synchronized.

Data Acquisition

Subjects produced pinch grip force against two load cells attached to the grip apparatus located in front of their body midline.⁴⁵ Each load cell was amplified using a Coulbourn amplifier and sampled at 2,000 Hz using a 16-bit A/D converter. A Biosemi EEG system with 128 Ag-AgCl Active Two electrodes was used. EEG signals were amplified at the source with an output impedance of <1 ohm and digitized at 2,048 Hz. A MotionMonitor system time-aligned the force and EEG data and resampled the signals at 2,000 Hz.

Force Data Processing

Three measures were examined: mean force noted as % maximum force, root mean square error (RMSE) with respect to the target force, and force tremor estimated as the sum of power of force between 4–12 Hz.¹⁴ All measures were computed over a 2.5-second window after 1.5 seconds from force onset. For each measure, the homogeneity of group was first examined using Levene’s test at each visual feedback level. If the between-group variances were found to be significantly different in at least one of the two conditions, nonparametric tests were used to examine the group and condition effects. The Mann-Whitney U test was used for comparing group effects, and the Wilcoxon signed-rank test was used for testing condition effects for each group. Significance was set at α = 0.05, and P values obtained from the group and condition tests were false discovery rate (FDR) corrected.⁴⁶

EEG Data Processing

For each trial, epoch was defined as a time window beginning 1.5 seconds before force onset and ending 4 seconds after force onset. Trials with early (<1 second) or late (>1 second) force onset relative to a trigger that signaled the beginning of the task period were discarded. EEG channels were bandpass filtered between 1 and 70 Hz, and channels with artifact were rejected and interpolated. Trials with more than 10 rejected channels or showing large absolute values at multiple channels based on EEGLAB’s joint-probability artifact detector⁴⁷ were deleted. EEG signals were rereferenced to the global average of all channels, and line-noise at 60 Hz was removed.⁴⁸ Each channel was down-sampled to 250 Hz. The average remaining trials at low and high visual feedback conditions for controls were: 46.88 (standard deviation [SD] = 2.45) and 47.76 (SD = 2.31); and for ETs, the remaining trials were 46.83 (SD = 3.25) and 46.63 (SD = 3.03). For each participant, epochs were concatenated across conditions, and independent components (ICs) and IC weights were computed using EEGLAB’s runica procedure. For each IC, an equivalent dipole was obtained using EEGLAB’s DIPFIT function. Dipoles were excluded if they were outside the Montreal Neurological Institute brain, or if their activity did not resemble a dipolar distribution (residual variance > 10%).²⁰ Measure projection analysis (MPA) was conducted to group subject-specific dipoles across groups and conditions into a spatially distinct set of domains,⁴⁹ based on their distances and their underlying brain activities measured as event-related spectral perturbation (ERSP). ERSP represents a time-frequency plot that is normalized by the mean baseline spectrum.⁵⁰

ERSP Analysis

For each domain per subject per condition, using dipoles that contributed to the domain, a mean ERSP matrix was developed by averaging the dipole ERSPs estimated using a wavelet based procedure. The dimension of each mean ERSP matrix was 200 cells along the time-dimension, representing 0.832 seconds before and 3.328 seconds after force onset and 100 cells along the frequency-dimension representing 1 to 60 Hz. At each cell, analysis of variance (ANOVA) was conducted to detect group, condition, and interaction effects, and the results were FDR corrected at P < 0.05.⁴⁶ For the domains showing interaction effects, using the interaction mask, a mean ERSP value was established for all subjects at each condition at frequency bands: 4–12 and 12–30 Hz. ERSP values were compared between groups, at low visual feedback, at high visual feedback, and at high minus low visual feedback based on the area under the receiver operator characteristics (ROC) curve (AUC). The P values of these six AUC curves were FDR corrected.

Baseline Spectral Analysis

For each domain per subject per condition, using –1.5 to 0 second time-segment of the source ICs of the dipoles that contributed to the domain, an average source signal representing the baseline period was first developed, and the power in the frequency bands 4–12 and 12–30 Hz were estimated and compared across groups and conditions using ANOVA at P < 0.05, FDR corrected.

Domain-Pair Connectivity Strength Analysis

For each participant per condition, the directed information transfer between a pair of dipoles in 4–12 and 12–30 Hz was first estimated using direct directed transfer function evaluated using SIFT toolbox-based vector autoregressive modeling.^51,52 Then, to compute the weighted-connection strength from domain-i to domain-j, in-brain dipoles were first assigned to either domain-i or domain-j based on the domain center that was closer, with the maximum distance set to 6 cm. For a given frequency band, directed weighted-connectivity strength from domain-i to domain-j was defined as the sum of all weights of connections from domain-i dipoles to domain-j dipoles in the corresponding frequency band. Connectivity strength was examined for group, condition, and interaction effects using ANOVA in 4–12 and 12–30 Hz. Significance level was set at α = 0.05, and P values were FDR corrected.⁴⁶

Multilinear Regression Analysis

A stepwise multilinear regression model based on Akaike information criterion was developed to determine predictors of tremor for the ET group. Predictors included those ERSP and directionally specific connectivity measures that showed significant between-group differences.

Results

Behavioral Analysis

As expected, ET subjects had higher FTM-TRS than controls (P < 0.001; Table 1). There was no significant difference in MoCA scores between groups (P = 0.935; Table 1).

Force Analysis

When comparing mean force as %MVC, no between-group effects were detected at either condition (low condition: P = 0.791; high condition: P = 0.791). %MVC was reduced at high visual feedback as compared to low visual feedback for both groups (P_s < 0.001; Fig. 1C). For RMSE (Fig. 1D), no between-group effects were detected at either condition (low condition: P = 0.791; high condition: P = 0.224), but RMSE was reduced at high visual feedback compared to low visual feedback for both groups (P < 0.001). ET subjects had significantly more force tremor than controls at both feedback conditions (P_s < 0.003). Only for the ET group was a condition effect detected (P < 0.034), such that at high visual feedback the force tremor was larger than at low visual feedback. In summary, ET showed higher force tremor than controls for both low and high visual feedback, and tremor worsened at high visual feedback for ET (Supporting Information Tables S2, S3, and S4).

FIG. 1.

The pinch grip device, force data, and kinematic analysis. (A) An illustration of the pinch grip force apparatus used in this study. (B) The force produced by a representative healthy control participant at low and high visual feedback conditions; at low visual feedback, there was a greater error than at high visual feedback condition. (C) The mean force amplitude measured as %MVC, (D) the mean force error measured as RMSE, and (E) the mean force tremor measured as the sum of power of the force signal in the 4–12 Hz frequency band, at low and high visual feedback conditions for each group, are shown as bar graphs.

EEG Analysis

MPA yielded three domains: left motor and premotor area (Domain-LMOT); right superior parietal area (Domain-PAR); and right motor and premotor area (Domain-RMOT). An ERSP analysis per domain was performed to examine group, condition, and interaction effects in the 4–12 Hz and 12–30 Hz frequency bands post–force onset.

Domain-LMOT

The domain included left precentral gyrus, left superior frontal, left middle frontal gyrus, right superior frontal gyrus, and right precentral gyrus (Fig. 2A; Table 2). Figure 2B illustrates the average ERSP activity for each group and condition at Domain-LMOT. For both groups and both conditions, there was reduced power in the beta-band during the task period relative to the baseline. Figure 2C shows that the change in power within 4–30 Hz was not different between groups, but that both condition and interaction effects were significant. Between 4–12 Hz (Condition effect; Fig. 2C), high gain led to an increase in power (yellow patch), whereas a decrease in power was evident in the 12–30 Hz beta-band (blue patch). In examining the interaction effect in Figure 2C, we quantified the ERSP in the yellow/red clusters shown in Figure 2C separately for each frequency band. Figure 2D shows that at high gain condition, in the 4–12 Hz bands (P = 0.026) and the 12–30 Hz bands (P = 0.047), controls showed a greater reduction in power relative to the baseline than the ET subjects; at low feedback, no between-group differences were found in either bands (4–12 Hz: P = 0.725; and 12–30 Hz: P = 0.402). Baseline spectral analysis of the average IC source signal representing the domain revealed no group, condition, or interaction effects (Supporting Information Tables S5 and S6).

FIG. 2.

ERSP analysis at Domain-LMOT. (A) The figure illustrates Domain-LMOT (the green blob) along with its constituent dipoles represented as small spheres; the domain was localized to the left motor cortex. The dipoles are colored based on their contribution to the domain; the red dipoles contribute the most. The dipoles are more densely packed toward the posterior end of the domain than the anterior end, and therefore in the axial and the sagittal projection, a red patch can be seen only in the posterior end. (B) The figure illustrates the mean ERSP plots for each pair of group and condition. The x-axis represents 0.8 seconds of baseline duration preceding the force onset (0 seconds), and 3 seconds duration post–force onset. The y-axis represents frequency of cortical oscillation in log scale, and the black rectangular window represents 4–30 Hz. The blue patch within the rectangular window represents a decrease in beta-band power, and the yellow patch at force onset represents an increase in power relative to the baseline. (C) The effects detected using repeated-measures ANOVA are presented; the green regions indicate nonsignificant effects. For this domain, only condition and interaction effects were detected. In the condition effect plot, the colors represent the mean high visual feedback minus mean low visual feedback ERSP values (collapsed across groups) at the time-frequency points where significant condition effects after FDR correction were detected. In the interaction effect plot, the colors represent the [(mean ET ERSP at high visual feedback – mean ET ERSP at low visual feedback) – (mean control ERSP at high visual feedback – mean control ERSP at low visual feedback)] values at the time-frequencies points where significant interaction effects after FDR correction were detected. (D) The mean ERSP values in 4–12 and 12–30 Hz for each group at each condition within the interaction mask are illustrated; the controls showed a greater modulation of ERSP values across feedback levels than the ET subjects.

TABLE 2.

Domains detected using measure projection analysis

Anatomical Regions (Probability)
Domain-LMOT: Left premotor/supplementary and primary motor/primary somatosensory:
Left precentral gyrus (0.30)
Left superior frontal gyrus (0.29)
Left middle frontal gyrus (0.24)
Right superior frontal gyrus (0.06)
Right precentral gyrus (0.06)
Domain-PAR: Right somatosensory association:
Right superior parietal gyrus (0.40)
Right precuneus (0.13)
Right postcentral gyrus (0.10)
Left precuneus (0.09)
Right angular gyrus (0.08)
Left postcentral gyrus (0.05)
Left precentral gyrus (0.05)
Domain-RMOT: Right premotor/supplementary motor/primary motor:
Right middle frontal gyrus (0.46)
Right superior frontal gyrus (0.31)
Right precentral gyrus (0.22)

Domain-PAR

The domain predominantly overlapped with parietal cortex (Supporting Information Fig. S1A; Table 2). Supporting Information Figure S1B illustrates the average ERSP activity for each pair of group and condition at Domain-PAR, and in Supporting Information Figure S1C, group, condition, and interaction effects in 4–30 Hz ERSP activity are presented. No group or interaction effects were detected, and only a condition effect was found (Supporting Information Fig. S1C). Baseline spectral analysis revealed no group, condition, and interaction effects (Supporting Information Tables S5 and S6).

Domain-RMOT

The domain predominantly overlapped on the right motor cortex (Supporting Information Fig. S2A; Table 2). Supporting Information Figure S2B illustrates the average ERSP activity for each pair of group and condition at Domain-RMOT, and in Supporting Information Figure S2C, group, condition, and interaction effects in 4–30 Hz ERSP activity are presented. No group or interaction effects were detected, and only a condition effect was found. Baseline spectral analysis revealed no group, condition, and interaction effects (Supporting Information Tables S5 and S6).

ROC Analysis

Because and interaction effect was detected at Domain-LMOT, we examined the effectiveness of ERSP measures within the interaction mask at low and high feedback conditions and at high minus low feedback conditions to separate the groups based on AUC values. The analysis was conducted for each frequency band and gain level separately (Supporting Information Section S5.3).

Magnitude of the mean difference between the high and the low visual feedback levels was consistently smaller for the ET group relative to controls (Fig. 3C,F). AUC values at low visual feedback varied from 0.60 (P = 0.347) to 0.65 (P = 0.120; Fig. 3G), and at high visual feedback, AUC was 0.74 (P_s = 0.03) for both 4–12 Hz and 12–30 Hz bands (Fig. 3H). AUC values at high minus low visual feedback varied from 0.90 to 0.91 (P_s < 0.003), thus suggesting that the ERSP difference between high and low visual feedback levels was more effective in separating groups than the ERSP values at individual feedback levels (Fig. 3I; Supporting Information Table S7). In summary, the change in ERSP values across feedback levels in 4–12 Hz and 12–30 Hz frequency bands of ET subjects was different than controls.

FIG. 3.

ROC analysis at Domain-LMOT. The mean ERSP values for controls (red bar) and ET subjects (white bar; A) at low visual feedback, (B) at high visual feedback, and (C) at high minus low visual feedback in the frequency band, 4 to 12 Hz, are illustrated. The mean ERSP values for controls (blue bar) and ET subjects (white bar; D) at low visual feedback, (E) at high visual feedback, and (F) at high minus low visual feedback in the frequency band, 12–30 Hz, are illustrated. The results of AUC analysis (G) for low visual feedback, (H) for high visual feedback, and (I) for high minus low visual feedback are presented. In these figures, the red curve represents the AUC analysis for 4–12 Hz, and the blue curve represents the AUC analysis for 12–30 Hz. Based on the AUC analysis, the ERSP difference between high and low visual feedback level was found to be more effective in separating the groups than the ERSP values at individual feedback levels.

Connectivity Strength Analysis

The ET group showed decreased directionally specific connectivity relative to controls from Domain-LMOT to Domain-PAR and vice versa at both feedback levels in the 12–30 Hz frequency band (P_s = 0.024; Fig. 4A,B). Mean connectivity strength at high visual feedback was reduced compared with low visual feedback for both groups (LMOT-PAR, P = 0.011; PAR-LMOT, P = 0.012). For the remaining pairs of domains, in 12–30 Hz, only a condition effect was detected (Fig. 4C–F; Supporting Information Table S8). Analysis between all pairs of domains in the 4–12 Hz frequency band only yielded condition effects (Supporting Information Table S7), with high visual feedback showing weaker connectivity strength than low visual feedback.

FIG. 4.

Connectivity strength analysis between domain-pairs in 12–30 Hz. The mean connectivity strength in 12–30 Hz frequency band (A) from Domain-LMOT to Domain-PAR, (B) from Domain-PAR to Domain-LMOT, (C) from Domain-PAR to Domain-RMOT, (D) from Domain-RMOT to Domain-PAR, (E) from Domain-LMOT to Domain-RMOT, and (F) from Domain-RMOT to Domain-LMOT, for each group at each visual feedback level, are shown as bar graphs. (G) Only between Domain-LMOT and Domain-PAR were group and condition effects found. For the remaining pairs of domains, only condition effects were detected.

Multilinear Regression Analysis

Multilinear regression analysis focused on the ET group. The following independent variables were examined: the subject-specific mean ERSP difference between high and low visual feedback levels established using time-frequency points that showed significant interaction effect in (1) 4–12 Hz and (2) 12–30 Hz at Domain-LMOT, (3) subject-specific mean difference in directionally specific connectivity strengths at high and low visual feedback levels between Domain-LMOT and Domain-PAR, and (4) age. The dependent variable was the difference in force tremor at high and low visual feedback levels. Because interaction effects were detected in 4–12 Hz and 12–30 Hz frequency bands during ERSP analysis, and group effect was found in bidirectional connectivity in 12–30 Hz between domains PAR and LMOT, we chose these variables for the multilinear regression analysis.

The stepwise procedure reduced the model to the ERSP difference between high and low visual feedback levels in 4–12 Hz for LMOT. The reduced model was found to be significant (P = 0.0469). Based on the model, the change in 4–12 Hz ERSP activity in contralateral motor cortex was found to be positively correlated with the change in force tremor across the feedback levels. That is, if the ERSP activity in 4–12 Hz showed a greater power at high visual feedback than at low visual feedback, then a greater increase in force tremor was observed from low to high visual feedback conditions as well.

Discussion

The goal of the study was to examine how cortical oscillations and directionally specific connectivity relate to visually amplified tremor in ET. High visual feedback significantly exacerbated tremor severity in ET and provided an opportunity to examine changes in cortical oscillations and directionally specific connectivity associated with visually amplified tremor. Our experiments led to three important observations. First, we found that by altering the visual feedback level from low to high gain during force production, there was reduced ERSP activity in 4–12 Hz (tremor band) and 12–30 Hz (beta-band) in the contralateral motor cortex in both groups, and the change was reduced in ET subjects as compared to controls. Second, for the ET group, bidirectional connectivity in beta-band between superior parietal cortex and the contralateral motor cortex was significantly lower than controls. Third, we found that the worsening of tremor from low to high visual feedback was correlated with the ERSP in the 4–12 Hz band of the contralateral motor cortex. Our findings provide new evidence to support the hypothesis that, in ET, there is a deficit in motor cortical function, such that in addition to the previous work indicating changes in the functional pathway between cerebellum and motor cortex in ET,^29,53 there also exists altered cortical oscillations and connectivity between parietal cortex and motor cortex.

A consistent finding across many motor control studies that dates back to the work of Woodworth is that providing visual feedback improves motor performance.⁵⁴ The classic observation that enhancing visual feedback improves motor performance can be seen in the current data as well, because the force accuracy improved for both groups with high visual feedback compared with low visual feedback. A paradox though is that in patients with ET, providing high visual feedback also leads to more tremor,^14,55 and this was confirmed using the current paradigm. A key question is why is this occurring for ET, and what role does the pathway linking parietal and motor cortex play in this observation?

During the visually guided force task, we detected a reduction in beta-band power in both groups at each visual feedback level within bilateral motor cortex and in ipsilateral superior parietal cortex. However, within the contralateral motor cortex, the control group exhibited a larger reduction in power at high visual feedback compared with low visual feedback in the 4–12 Hz and 12–30 Hz bands, whereas the cortical oscillations in the ET group did not respond to enhanced visual feedback (see Fig. 2D). In the regression model, the change in ERSP within the 4–12 Hz band of the contralateral motor cortex predicted the change in tremor during force control. We also observed using the ROC analysis that the differentiation between ET and controls had the highest AUC when using the change in ERSP in the 4–12 Hz and 12–30 Hz bands from low to high visual feedback, rather than using the ERSP at either low or high visual feedback alone. This suggests that how patients with ET respond to visual feedback may be a contributing feature of ET pathophysiology.

Previous electrophysiological studies examining EcOG^24,25,56 in movement disorder patients have reported disease-specific alteration of beta-band power in sensorimotor cortex. For example, an EcOG study found a significant reduction in the magnitude of the movement-related beta-band power in primary motor and sensory cortices in dystonia patients as compared to PD and ET.²⁵ Similarly, a separate EcOG study reported a greater reduction in beta-band power in sensorimotor cortex of PD subjects than ET subjects during movement preparation.²⁴ Our findings showed greater levels of ERSP in the beta-band of motor cortex in ET relative to controls at high visual gain, thus suggesting that there may be a deficit for ET patients in modulating large populations of neurons in the performance of visually guided force control.

We also found reduced bidirectional connectivity between the superior parietal cortex and contralateral motor cortex in ET. Using fMRI, studies have reported changes in activation within primary visual cortex, inferior/superior parietal cortex, primary motor cortex, dorsal premotor cortex, and supplementary motor area.^14,57 In a recent fMRI study¹⁴ consisting of a cohort of ET subjects that overlapped with ET subjects examined in the current work, probabilistic tractography analyses revealed no structural changes within the cerebellum to motor cortex pathway and within the dorsal processing stream linking parietal and motor cortex of ET subjects when compared to controls. The current findings, showing altered connectivity between parietal cortex and motor cortex, complement another EEG study examining coherence between EMG and EEG signals between cerebellum and motor cortex.⁵³ This is not to say, though, that the reduced connectivity between parietal and motor cortex is driving the tremor or generating tremor. It could be that when visual feedback exacerbates tremor at the limb, this leads to greater oscillations in the limb and thus greater firing rate from muscle spindle afferents that project to the parietal and sensorimotor cortex. As a result, it is possible that the current EEG findings capture this effect in parietal and sensorimotor cortex. A recent study found that cooling the limb reduced ET amplitude,⁵⁸ and an additional study found that stimulating nerves in the wrist reduced essential tremor amplitude.⁵⁹ Thus, feedback from peripheral receptors may be important, although not measured directly in the current study.

It is important to place the current findings, which focused on the motor cortex and parietal cortex, in the context of other studies in the literature. When examining pathology of ET, most of the focus has been on the cerebellum, with key evidence pointing to swelling of Purkinje cells²⁹ and decreased climbing fiber/Purkinje cell synaptic density, which also relates to tremor severity.⁶⁰ These studies are comparing postmortem tissue to tremor measured clinically. It is also the case that studies have identified that during neurosurgery, neurons in the ventral intermediate nucleus of the thalamus show coherence with tremor measured using EMG.⁶¹ In addition, the current study and other recent fMRI and EEG studies point to additional regions of the sensorimotor network that relate to essential tremor severity. Although our analysis approach revealed changes in the cortex, it is possible that our observations in the cortex may have been driven by ET-related changes in subcortical circuits, including the thalamus and cerebellum. Nevertheless, the collective picture emerging from the current EEG study and previous EEG, MEG, fMRI, and postmortem studies^14,53,62 is that tremor in ET is associated with a widespread network with multiple nodes, rather than one key region of the brain, and this perspective has important implications for pharmacological and other nonpharmacological treatments for ET because multiple targets may exist.