Sensory Tricks in Cervical Dystonia Correlate with Enhanced Brain Activity During Motor Preparation

Abstract

Introduction:

Although sensory tricks are well known as the maneuvers that temporarily relieve dystonic symptoms in patients with cervical dystonia (CD), the underlying neurophysiological mechanisms remain unclear. We aimed to investigate brain potentials related to sensory tricks in patients with CD.

Methods:

Thirteen patients with CD and 13 age-matched healthy volunteers participated. The experiment consisted of three conditions (moving the neck, moving an arm, and performing sensory tricks) presented in different blocks in random order in a contingent negative variation (CNV) paradigm. Warning and trigger stimuli (S1 and S2) were presented to the participants, who were instructed to prepare to perform the specific task for each condition after S1, and then to perform the task after S2. Early and late components of the CNV were measured.

Results:

The late CNVs in patients with CD were significantly larger than those in healthy participants in Fz, FCz, Cz, and C3 electrodes. Only in patients with CD, the late CNVs were significantly greater for the ‘sensory tricks’ condition compared to the ‘move neck’ condition in Fz and C3 electrodes.

Conclusion:

The late CNV is increased during sensory tricks in patients with CD, suggesting that sensory tricks may affect mechanisms related to the motor preparatory phase in the premotor and primary motor areas. Sensory tricks may normalize impaired motor preparation in dystonia, leading to improved dystonic symptoms.

1. Introduction

Sensory tricks are maneuvers that temporarily relieve dystonic symptoms. The sensory trick phenomenon is a cardinal feature of cervical dystonia (CD) reflecting an improvement in sensorimotor function[1]. Neurophysiological studies have mostly focused on elucidating the mechanism of dystonia in patients with CD[2], while only few studies have investigated the physiological mechanisms of sensory tricks in patients with CD[3, 4].

Neurophysiological mechanisms underlying sensory tricks can be probed using EEG. For instance, movement-related potentials (MRPs) could explain the neurophysiological processes occurring before, during, and after the application of a sensory trick. In particular, the contingent negative variation (CNV) -- a slow brain potential produced in an experimental paradigm in which two stimuli (warning and imperative) are provided [5]-- can gauge neuronal activity during the preparatory phase of the motor action. Previous studies have shown that the CNV amplitudes of CD patients are significantly lower than those of normal controls during head rotation, suggesting that CD is a disorder of movement preparation [6]. However, no studies examined these brain potentials in relation to sensory tricks in CD patients.

Here, we examined the CNV in patients with CD in a forewarned task paradigm with two sequential cues, wherein the patients either implemented sensory tricks to relieve their dystonic symptoms or performed other similar motor tasks. We hypothesized that the CNV would be impaired in patients with CD during voluntary head rotation and the CNV would be normalized during sensory tricks.

2. Methods

2.1. Participants

Thirteen CD patients (mean age 59.4, range 31–78) and 13 age-matched healthy volunteers (mean age 58.8, range 43–82) participated. Healthy volunteers were individually matched to the patients and were asked to mimic dystonic symptoms of matched patient. All CD patients demonstrated effective sensory tricks to improve their symptoms. They used the same manner of touching the face with one hand as the type of sensory trick. At the time of the experiments, at least 11 weeks had elapsed from their last botulinum toxin injection. This experiment was approved by the National Institutes of Health Institutional Review Board. Before the experiment, all participants gave written informed consent for this protocol.

2.2. EEG Recording

EEGs were recorded from 32 channel surface electrodes mounted on a cap (Braincap, Brain products, Germany) using the international 10–20 system referenced to Pz. EEG data, converted to the digitally linked earlobe reference, were acquired using BrainAmp (Brain products, Germany) at a sampling rate of 5 kHz and bandpass filtered from 0.05 to 100 Hz.

2.3. Study Protocols

The experiment consisted of three conditions (moving the neck, moving an arm, and performing sensory tricks) presented in different blocks in random order in a CNV paradigm. For measuring CNV, stimulus 1 (S1) was presented in white letters for 2 s, indicating the task to be performed, followed by stimulus 2 (S2) in green letters displayed for 10 s, which was the go command to perform the task (Figure 1). Detailed experimental protocol is described in the Supplementary material.

Figure 1.

Experimental paradigm for measuring contingent negative variation. Stimulus 1 was seen on the screen in white letters for 2 seconds and stimulus 2 was given in green letters for 10 seconds.

2.4. EEG Preprocessing

Gross movement artifacts that were considered unrelated to the tasks were first removed by visual inspection. Then, supervised machine learning algorithm (MARA; Multiple Artifact Rejection Algorithm[7, 8]) was implemented to identify and remove artifacts based on six features from the spatial, spectral, and temporal domains. After artifact rejection, the continuous EEG data were epoched into one-second segments starting at the beginning of each trial. The EEG data were analyzed for the following three periods:

1. Period 1 (−1750 ms < t < − 1000 ms): early CNV (eCNV) period (after S1).

2. Period 2 (−1000 ms < t < − 250 ms): late CNV (lCNV) period (before S2).

3. Period 3 (0 ms < t < 1000 ms): execution period (after S2).

2.5. MRPs

The eCNV was calculated as the mean amplitude of EEG activity over Period 1 and the lCNV as the mean EEG activity during Period 2. Brain potentials during the movement execution period were also calculated (Period 3). The baseline was calculated by averaging the amplitude during the 500 ms before S1 (baseline: −2500 ms < t < −2000 ms) (See Supplementary Figure 1 for examples of CNVs).

2.7. Statistical Analysis

The data were analyzed for the selected electrodes: Fz, FCz, Cz, and C3 and P3 in the contralateral cortex to the arm using sensory tricks. Mixed two-factor multivariate analysis of variance (MANOVA) was performed on the CNVs from the five electrodes with the group (between-subject) and task (within-subject) as independent variables (IBM SPSS Statistics, version 24, IBM Corp., Armonk, New York, USA). Normality of all model residuals was tested and confirmed using the Shapiro–Wilk test. A univariate analysis of variance was performed on the variables that were found statistically significant by MANOVA. Post-hoc comparisons between the task conditions were performed using univariate analyses of variance if the task effects or group-task interactions were found to be significant by MANOVA. The p-values were corrected for multiple comparisons with Bonferroni’s adjustment.

3. Results

We observed that ‘sensory tricks’ that involved touching the face were effective in correcting the dystonic posture in all patients, but the ‘touch shoulder' trick was not effective in any patient. All patients could maintain their neck posture straight in ‘move neck’ condition. Control subjects correctly mimicked the movements of the patients for all conditions.

Across all five electrodes examined (Fz, FCz, Cz, C3, P3), neither the group nor the task condition were found to have significant effects on eCNV values (p > 0.05) (Figure 2A). In contrast, the lCNV values of CD patients were significantly greater than those of healthy participants across the tasks, especially at FCz and Fz (FCz: p = 0.01; Fz: p = 0.02), while the difference was not significant at C3 and P3 (C3: p = 0.06; P3: p = 0.07) (Figure 2B). A significant group×task interaction was only observed in the lCNV (Figure 2B) due to the between-group difference during sensory tricks but not in the other tasks (group×task interaction-Fz: p = 0.04; C₃: p = 0.09). Post-hoc analyses showed that the lCNV values of the CD patients during ‘sensory tricks’ were significantly greater than those in ‘move neck’ condition (sensory trick vs. move neck in patients - Cz: p = 0.02; Fz: p = 0.02; FCz: p = 0.01; C3: p = 0.01).

Figure 2.

Comparison of early and late contingent negative variation (CNV), and movement related potential (MRP) between groups according to the tasks performed. Early CNVs were comparable between groups for each task (A). Late CNV was larger over Cz, FCz, and Fz in CD patients than in healthy controls. Late CNV was larger over Cz, Fz, and C3 during sensory tricks than during movement of the neck to the neutral position (B). MRP during movement execution was larger over Fz, FCz and C3 in CD patients than in healthy control (C). *p<.05.

Significant between-group and between-task differences were also observed during movement execution, albeit to a smaller degree (group - Fz: p = 0.04, FCz: p = 0.04, Cz: p = 0.05, C3: p = 0.04; task - Cz: p = 0.05) (Figure 2C).

4. Discussion

We evaluated the physiological mechanisms of sensory tricks in CD patients with the CNV. The main findings from the CNV analysis were: 1) as effect of group, lCNV negativity was larger in cervical dystonia patients compared with normal controls over Cz, FCz, and Fz, corresponding to the premotor area, 2) as an effect of task, lCNV negativity was larger over Cz, Fz, and C3, corresponding to the premotor and primary motor areas, when patients used sensory tricks than when they intentionally moved the neck to the neutral position, 3) As for lCNV, MRP differed between groups during the motor execution period, 4) No significant difference in the lCNV amplitudes were found among the three target tasks in healthy subjects, and 5) There was no significant difference in eCNV among groups or tasks.

The most significant finding of this study was that lCNV negativity increased when CD patients used sensory tricks compared with when they intentionally moved the neck to the neutral position, in both the premotor area (Cz and Fz) and the primary motor cortex (C3). As there was no between-task difference in CNV of control subjects, we can exclude the possibility that the CNV amplitude was affected by movement type [9]. The larger CNV amplitude during sensory tricks indicates that the anticipatory brain mechanism in the premovement phase is more active when performing sensory tricks. The increased anticipatory mechanism related to sensory tricks was prominent in the premotor and primary motor areas, but was not related to primary sensory areas, suggesting that sensory tricks may directly act on the motor systems, including motor preparation, programming and execution, rather than influence the primary sensory area. Indeed, dystonia has been suggested as a disorder of motor programming[10]. In addition, abnormal posture in patients with cervical dystonia is first normalized by voluntary activity of the antagonist muscle or pressure on the face, and then sensory input stabilizes the position according to the previous study[4]. Considering that sensorimotor integration is an important pathophysiological problem in patients with dystonia[1], and sensory tricks are strong evidence of sensorimotor function, the increased CNV during sensory tricks suggests that sensorimotor integration may be augmented during the motor preparation process [11]. Taken together, these results show that sensory tricks contribute to a temporary normalization of abnormal movement execution by raising the activity concerned with motor preparation and expectation. Accordingly, the nature of sensory tricks does not need to be always “sensory” because the contribution of sensory system may not be critical and so, the term “alleviating maneuver” putting more emphasis on motor aspect may be more appropriate[1].

It is not clear how such increased motor preparation activity induced by sensory tricks achieved a temporary normalization of volitional movement. Previous studies have suggested that changes in the central sensory transmission process during the motor preparation period may lead to the incorrect choice of motor commands in dystonia patients[10, 12]. Thus, sensory tricks may play a role in normalizing the gating of somatosensory input via the modulation of earlier brain potentials related to the motor expectation and preparation periods, represented in this study by the lCNV, leading to improvement in dystonic symptoms. The relationship between CNV and sensory evoked potentials during the reaction time motor task paradigm needs to be clarified in future studies.

A previous study showed that the lCNV was task-specifically lower in CD patients than in controls [6]. The authors suggested that the task-specific CNV amplitude loss in CD patients leads to failure of appropriate motor preparation, resulting in co-contraction of the competing muscles. Similarly, CNV amplitude was found to decrease in patients with writer’s cramp making finger movements[13]. In our study, however, lCNV in CD patients was larger than that in normal controls during neck movement. The difference in experimental paradigms may explain such discrepancy. In our protocol, the CD patients were asked to move their neck from the dystonic position to the straight-ahead position, while normal controls were asked to mimic a dystonic posture like that of the matched patient, and then to turn their neck to straight-ahead. The movement patterns of the two experiments are are opposite in terms of movement intention. In the earlier experiment, the starting point of the neck was straight ahead, even though the neck muscles would have to be actively fighting the dystonia. Moreover, the normal controls were asked to contract their neck muscles in the starting position to mimic dystonia. Since the CNV depends on intention, preparation and expectation for movement, such differences in experimental paradigm could certainly influence the results.

eCNV, which has been known to less directly related to motor tasks than lCNV, did not show any significant effect in any of the analyses related to sensory tricks[9]. The difference in the brain potentials in the movement execution period was similar to the difference in lCNV between the groups, thereby implying that movement preparation and execution share the same mechanisms.

There are several limitations to our study. First, the small sample size (a few “trends” are observed likely owing to little study power); second, a group of CD without sensory tricks would have been another good control group; third, limiting the methods to one single neurophysiologic technique (i.e., CNV) does not probably give enough information to fully understand the complex mechanism underpinning sensory tricks.

Highlights.

• Movement related potentials measured by contingent negative variation are enhanced during sensory tricks in patients with cervical dystonia.

• Sensory tricks may normalize impaired motor preparation in dystonia, leading to improved dystonic symptoms.

• Movement related potentials related to sensory tricks act independently of brain oscillatory activities.