Childhood dystonia is defined as “a movement disorder in which involuntary sustained or intermittent muscle contractions cause twisting and repetitive movements, abnormal postures, or both” [1, 2]. Muscle activity in dystonia typically exhibits both overflow into task-unrelated muscles as well as greater variability within task-related muscles than in healthy subjects [3–5]. Increased variability may represent an inability to suppress motor noise [6], resulting in the superposition of unwanted motion components on the desired movement. This may lead to more variable and less efficient motor outcomes, which are typical in dystonia [7–9]. Even if movement abnormalities in dystonia are obvious on observation-based examinations, objective measures to characterize dystonia and to gain insights into its pathophysiology are still strongly needed.

During movement, it is difficult to separate voluntary from unwanted variable components of activity in muscles. Here, we apply a Fourier analysis to electromyographic (EMG) and kinematic data acquired during the performance of a continuous figure-eight writing task in children with dystonia and age-matched healthy controls. This new method exploits the frequency-domain features of cyclic motor tasks to discriminate between task-correlated and task-uncorrelated components within task-related muscles. We refer to “task-correlated” components to indicate muscle activity showing the same frequency content as the kinematics of the motor task. We define “task-uncorrelated” components as muscle components occurring at frequencies unrelated to the cyclic figure-eight movements, which include both fixed and variable elements that do not contribute to the achievement of the desired motor task.

We hypothesize that movement abnormalities in dystonia, reputedly related to basal ganglia dysfunction, are partially due to the inability to suppress involuntary variable activity unrelated to the task. Therefore, we expect children with dystonia to present significantly greater task-uncorrelated components of muscle activity compared to control subjects.

Results are presented as medians and lower and upper quartiles.

EMG results

Significant effects was found for the independent factor Group [p = 0.049], with no significant effect of the within-subject factor Muscle and no significant interaction. The task-correlation index was higher for controls [Control: 0.453 (0.259-0.603); Dystonia: 0.321 (0.226-0.442)], indicating that, for healthy children, 45 % of muscle activity was correlated with the task, whereas only 32 % was correlated for children with dystonia (Fig. 3). The finding supported the prediction of a greater magnitude of task-uncorrelated components in the EMGs of subjects with dystonia.

Kinematic results

No significant difference between the two groups was found either for the accuracy error [Control: 0.255 (0.234-0.298); Dystonia: 0.335 (0.282-0.563) cm], or for the speed [Control: 18.341 (11.282-20.382) cm/s; Dystonia: 10.351 (6.101-14.187) cm/s] of the final outcome. A large range of movement speeds was reported for both children with dystonia [6.075-27.892 cm/s] and age-matched control children [8.071-33.040 cm/s]. This is not surprising since the task did not impose any time constraint, while participants were asked to pay particular attention to accuracy while following the trace provided. A difference between the two groups was found when examining AccErr/Speed, which was significantly increased in children with dystonia after controlling for the effect of age [Control: 0.017 (0.014-0.019) cm/(cm/s); Dystonia: 0.034 (0.026-0.051) cm/(cm/s), p = 0.002]. At the same execution speed, children with dystonia performed, on average, a movement leading to a less accurate outcome, meaning that the speed-accuracy trade-off is altered in dystonia. Spatial variability was significantly increased in children with dystonia for tablet, elbow, and acromion trajectories [Tablet: Control: 0.3612 (0.354-0.433) cm; Dystonia: 0.806 (0.670-0.876) cm, p = 0.001; Elbow: Control: 0.101 (0.096-0.123) cm; Dystonia: 0.333 (0.178-0.575) cm, p = 0.003; Acromion: Control: 0.317 (0.262-0.358) cm; Dystonia: 0.728 (0703-1.127) cm, p = 0.003]. Children with dystonia presented significantly increased temporal variability compared to healthy controls [Control: 0.352 (0.328-0.756) s; Dystonia: 1.603 (1.125-2.739) s, p = 0.030]. Overall, the motor performance of children with dystonia was characterized by increased variability both in temporal and spatial terms (Fig. 4). When investigating statistical dependence between the task-correlation index and the computed kinematic parameters, significant results emerged for accuracy error [rho (ρ) = -0738; p = 0.001] and spatial variability of tablet [ρ = -0621; p = 0.012]. The negative ρ values show that the task-correlation index decreases when spatial variability and inaccuracy of the end-effector increase. No significant linear regression was reported between the task-correlation index and either the BAD_arm or the BAD_total scores for children with dystonia. The lack of correlation between the task-correlation index and the BAD Scale score may be due to the small sample size (7 children with dystonia) and to the limited range of the BAD Scale (4-point scale for each segment). However, it is likely that the measure here proposed captures a different type of movement abnormality in dystonia, mostly task-dependent, as supported by the correlation with the kinematic parameters.

Our results show that, compared to age-matched healthy controls, the motor performance of children with dystonia is characterized by increased task-uncorrelated muscle components. This finding is consistent with the hypothesis that, in dystonia, the ability to appropriately suppress variable and uncorrelated muscle activity is impaired. The method here presented devises an innovative approach to quantify abnormal sub-components of muscle activity by applying a simple and well-known analysis to movement outcome and relative muscular control signals. The comparison of these task-uncorrelated muscle components between children with dystonia and healthy subjects allows us to gain insights into the underlying mechanisms responsible for the clinical phenomenology of dystonia. The proposed method is rather conservative, indeed it may underestimate the abnormalities in dystonia because it ignores involuntary activity that occurs at task-relevant frequencies. Muscle activity may reflect a combination of i) voluntary activity of muscles needed to accomplish the task, ii) involuntary overflow of activity into other muscles, iii) involuntary variable activity uncorrelated to the task. Our method extracts and quantifies the latter, while it is not designed to measure overflow of voluntary activity from task-relevant muscles, since voluntary activity is task-correlated. Nevertheless, this method provides a promising new technique for identifying one component of abnormal activity in specific muscles during movement, and it documents superimposed task-uncorrelated patterns of activity in dystonia.

The effectiveness of the proposed method as a measure to quantify abnormalities in dystonia is further validated by the significant linear correlation with kinematic parameters such as the spatial variability and the accuracy error of the final outcome. The task-correlation index addresses the need for a quantitative and objective assessment of movement abnormalities in dystonia during the execution of specific motor tasks. On the other hand, the BAD Scale was developed to assess, through observation-based examinations, varied functional limitations due to dystonia in persons with Cerebral Palsy and Traumatic Brain Injury, and it is not related to any specific task [10]. Thus, the measure we propose aims at integrating the BAD Scale severity score, rather than replacing it.

Results show that the motor performance achieved by children with dystonia is characterized by an overall lower quality, with high variability and an altered trade-off between speed and accuracy. Children with dystonia exhibited increased spatial variability at all levels of the upper body. Indeed, tablet, elbow, and acromion trajectories were less repeatable across trials. In addition, the analysis of temporal variability revealed a behavior significantly less consistent in time for the group with dystonia. The finding is consistent with a previous study from our team [6], that reported evidence of increased motor variability in dystonia due to the inability to remove noisy components. When analyzing the speed of movement, a wide range of speeds was found across the subjects of both groups, and no significant difference between children with and without dystonia was reported. The result is not surprising, since the task did not impose any time constraint. However, a significant difference between the two populations emerged in the ratio between accuracy error and speed of movement, which was higher for subjects with dystonia. In agreement with previous studies [7, 15], the result shows that the nature of the relationship between speed and accuracy is altered in dystonia. Since the signal-dependent noise that affects the sensorimotor system inherently imposes a trade-off between speed and accuracy of movement [18], the observed altered trade-off supports the hypothesis of increased signal-dependent noise in dystonia [15]. In other words, optimal control theories [19] assert that, when performing upper-limb tasks, subjects minimize the variance of the arm position for a specified duration or the movement duration for a specified positional variance. This criterion is met in a less effective way in dystonia, since the neural control signals are corrupted by a noise whose variance is more sensitive to the size of the signals themselves than in healthy systems [15, 20].

Both primary and secondary dystonia are associated with dysfunction or injury to the basal ganglia [4, 21, 22]. Neurophysiological studies support the theory that, when voluntary movement is generated, the basal ganglia are responsible for the focused selection of the desired motor pattern and for the inhibition of undesired and competing movements [23–25]. Our finding is consistent with the hypothesis that abnormalities arising from the basal ganglia in dystonia may be related to the inability to remove unwanted potential movements or elements of movement. As a consequence, movement in childhood dystonia may represent the superposition of unwanted and variable motion components on the desired movement. Nevertheless, it is worth noting that, in many cases, the injury is not limited to basal ganglia, and damages to other brain areas like cerebellum, brainstem, or sensory cortex can be the cause of dystonia. In addition, the uncorrelated muscle components detected during the execution of the figure-eight task may capture activity other than involuntary muscle noise arising from underlying central motor commands problem. For instance, the increased task-uncorrelated components found in dystonia may also partially consist of activity reflecting non-efficiency of movement, such as corrective activity to counter possible errors generated while tracing the figure-eight, or activity due to variations in grip postures. For these reasons, the nature and the neural substrate related to these unwanted noisy elements would need further investigations, perhaps involving functional imaging.

The main weakness of the current study is represented by the small sample size. To increase the robustness and the generalizability of our results, further assessment of the value of these measurement techniques will require a larger sample of subjects, with a balance between boys and girls in each group. Such future studies should unravel the issue of whether the speed-accuracy trade-off may be influenced by the difference in the composition of the two groups, since it is known that boys tend to exhibit increased variability of movement [26]. Moreover, a larger sample size would allow us to take into account the effect of age as a covariate and to further investigate the relationship between our outcome measures and the BAD Scale score. Finally, a larger sample of subjects may be recruited to test the reliability and the sensitivity of the proposed index as a tool that may be used in clinical practice to detect, at the individual level, increased magnitude of task-uncorrelated components of muscle activity.

The present work represents a proof-of-concept of a potentially useful new method to objectively measure movement abnormalities in childhood dystonia in terms of deviations from normal motor control principles, thus offering the possibility to characterize the phenomenology of dystonia and gaining insights into its pathophysiology. Our simple and innovative approach applies spectral analysis to motor outcomes and related muscle patterns during the performance of a continuous figure-eight writing task to easily distinguish muscle activity correlated with the task from uncorrelated involuntary components.

The results reported here show that, compared to age-matched healthy controls, children with dystonia present increased task-uncorrelated components in multiple muscles, and increased spatial and temporal variability. These findings are consistent with the hypothesis that, in dystonia, the ability to appropriately suppress unwanted and irrelevant muscle activity is impaired, resulting in the superposition of unwanted motion components on the desired movement. As a consequence, we observed more variable and less efficient motor outcomes in children with dystonia compared to age-matched healthy controls.

The investigative tool presented here has the potential to inform the design of neurorehabilitation therapies for dystonia. The study’s finding fits with the idea that dystonia is a central motor control problem, in which appropriate and inappropriate motor signals can combine in the same muscle. Such a finding helps to speculate on the mechanisms underlying the phenomenology of dystonia and, as a consequence, may be useful to drive the design of novel techniques that will intervene by modifying central signals, such as neural prostheses, Deep Brain Stimulation, Transcranial Magnetic Stimulation, or Peripheral Nerve Stimulation intended to selectively modify central connections.