Normal Motor Adaptation in Cervical Dystonia: A Fundamental Cerebellar Computation is Intact

Work in animal models strongly supports a causal cerebellar contribution in the genesis of dystonia. For example, in murine animal models, a dystonic-like condition can be provoked (excitation) or eliminated (inhibition or cerebellectomy) by modulating the activity of the cerebellum [17] and genetically modified animal models are increasingly sophisticated in their ability to probe and implicate the cerebellum [18].

In humans, clinical data suggests that structural and degenerative disorders of the cerebellum can cause secondary forms of dystonia [19]. However, secondary dystonia, by definition, has differences to primary dystonia in which no gross structural pathology is observed. In addition, such disorders rarely selectively involve only the cerebellum, and pathology within the cerebellum may evoke compensatory change in a multitude of interconnected regions. Therefore, parallels drawn between this clinical data and primary dystonia should be tentative. In primary dystonia, subtle structural and functional abnormalities of the cerebellum and its communicating tracts have been consistently demonstrated with a range of imaging techniques (for review see [20‐22]). Some electrophysiological data also point to cerebellar abnormalities in dystonia. Cerebellar brain inhibition (CBI) of the motor cortex was reduced in a small number of patients with focal hand dystonia [23]. In addition, the ability to acquire eye blink conditioning was reduced in cervical and focal hand dystonia, suggesting impaired cerebellar function [24]. Interestingly, this deficit in conditioning could be improved by further practice or inhibitory cerebellar stimulation, suggesting that cerebellar ‘dysfunction’ in dystonia is a dynamic process [25]. To date, deficits in CBI have not been replicated in a larger study nor investigated in other subgroups of dystonia and we have recently shown that eye blink conditioning is intact in the genetic dystonias DYT1 and DYT6 (submitted). Another line of investigation examining cerebellar function in dystonia has been the study of motor tasks which require intact cerebellar function. In DYT1 dystonia, both manifesting and non-manifesting subjects are impaired in sequence learning in which the sequential order targets is learnt [26, 27] and functional imaging demonstrated overactivity of the left cerebellar cortex (whilst subjects moved the right arm) [27, 28]. However, sequence learning recruits many brain regions including the basal ganglia, and the overactivation of the contralateral cerebellar hemisphere to hand movement (cerebellar control is ipsilateral) makes the functional significance of these findings in dystonia difficult to elucidate. Recently, it has been shown that sequence learning is normal in cervical dystonia [29].

Motor adaptation is a task commonly used, across species, to directly examine cerebellar function [14, 30]. An environmental perturbation introduces a movement error requiring subjects to adapt their performance of a task. The sensory prediction error (how the actual sensory movement outcome differed from the predicted sensory movement outcome) is used to update subsequent motor performance, with this type of learning being strongly dependent on the cerebellum [5]. Interestingly, the cerebellum has not only been linked to the formation of forward models which predict the sensory outcomes of motor commands; it may be that the cerebellum has a role in forming cognitive predictions for non-motor cerebellar functions such as language [31, 32]. This argument is supported by the highly conserved structure of the cerebellar microanatomical architecture, which is thought to imply that the computational qualities of cerebellar cortex remain constant [32, 33].

Patients with myoclonus dystonia (caused by mutations of the SGCE gene, DYT11) have been shown to have impaired saccadic adaptation [34]. It is difficult to dissociate in this condition whether the predominant phenotype of myoclonus dominates the neuroanatomical findings or whether the milder dystonia also has a role. In primary dystonia, motor learning/adaptation has been examined in focal hand dystonia using a joystick task [35]. Each trial had a different visuomotor perturbation, and a different position of the target and subjects were asked to correct their movement during each trial. No impairment in motor learning was demonstrated but there was impaired retention. However, contrary to the author’s conclusions, this suggests a change in the ability of the motor cortex to retain the new memory rather than a cerebellar deficit [14, 35]. Our data in cervical dystonia builds on previous work that we performed with a more simplistic visuomotor adaptation task [29]. This current study differs in that we used a purpose built robot which required larger more complex movements recruiting proximal arm and shoulder muscles. We also used a shooting paradigm which does not allow for online correction and modelled data in a manner which we believe optimally assesses for differences in adaptation. The forcefield condition is more relevant to dystonia in which subtle proprioceptive deficits have been described [36]. Furthermore, visuomotor and forcefield adaptation examine distinct (and common) regions of cerebellar function. Within the anterior lobe of the cerebellum, which contains one of the two body representations within the cerebellum, lobules IV and V are thought to be more important for the forcefield task and lobule VI is more important for visuomotor adaptation [8]. Regions in the posterolateral cerebellum (crus I and II) are thought to be required for both tasks [22]. This analysis of the two perturbations with a large number of patients leads us to confidently conclude that motor adaptation is normal in cervical dystonia.

How do our results link in with the growing body of evidence which implicates the cerebellum in the pathophysiology of dystonia? Certainly for cervical dystonia, if there is cerebellar dysfunction, the nature and extent of cerebellar dysfunction remain to be established. Normal cerebellar adaptation in cervical dystonia is in contrast to deficits in eye blink conditioning within the same subtype of dystonia [24, 25]. Both paradigms are well-characterised paradigms in their assessment of cerebellar function. Perhaps the deficit in eye blink conditioning, with its greater reliance on millisecond timing, signifies that timing is the specific cerebellar deficit in dystonia? This viewpoint links well with abnormalities in temporal discrimination and other timing tasks that have been found within the millisecond range in focal dystonia [37‐39]. One counter to this argument is the observation that all movement parameters were normal in the current study, and each parameter has a millisecond timing requirement. Furthermore, a deficit in millisecond timing capabilities could potentially impair the generation of sensory prediction errors, which would include time as one of their dimensions.

The normal performance in these adaptation tasks that required use of both visual and proprioceptive input was of interest. Although visual processing is normal in cervical dystonia, previous studies have described deficits in proprioceptive tasks dystonic subjects are less sensitive at detecting passive movements of the fingers [40] and arms are abnormal in their perception of the vibration induced illusion of movement (which is induced by stimulating muscle spindles with a vibration stimulus) [36, 41, 42]. How can performance in our tasks be normal in the face of such obvious deficits? One possibility is that tests of proprioceptive sensation are mostly static tasks whereas ours were dynamic, involving sensation during active movement. Furthermore, the psychophysical tasks described above require sensory processing and decision making at many levels of the nervous system and some of these are likely to be distinct to networks involved in implicit motor tasks. For example, higher order/consciously regulated elements of decision making could have a greater influence on psychophysical tasks.

The question of whether movement in the asymptomatic arm of patients with cervical dystonia is entirely normal perhaps remains to be definitively answered with future experimental work. Some have described abnormalities in kinematic variables recorded during reaching studies similar to the task used in this article (movement time was not matched between groups and thus some of this data is difficult to interpret [40]) and electrophysiologically, abnormalities in inhibition have been demonstrated at many levels of the nervous system concerned with the control of the arm musculature (e.g., abnormal reciprocal inhibition of forearm muscles in cervical dystonia [43]). However, other studies in including ours suggest near normal motor performance [29]. Conservation of motor skill in the arms is the norm with most patients with cervical dystonia, and we argue that this is perhaps against a global movement deficit in the focal dystonias.

Our conclusions for dystonic tremor are more tentative. We did not find evidence to support a relationship between the ability to adapt and the severity of dystonic tremor. Secondly, splitting subjects into whether or not they had tremor did not reveal a group difference in rates of adaptation. The pathophysiology of dystonic tremor is poorly understood but many primary tremor models are thought to involve the cerebellothalamocortical network. Certainly, in patients with essential tremor, there seems to be multimodal evidence for pathological involvement of the cerebellum (structural imaging [44], functional imaging [45], eye movement analysis [46], deficits in eye blink conditioning [47] and motor adaptation [12].) Here, we have performed one of the first studies to examine the role of the cerebellum in the generation of dystonic head tremor and have not yet found a clear interaction. Our findings support studies that suggest different mechanisms between essential and dystonic tremor. For example, in essential tremor, the second agonist burst during ballistic movements is delayed and this finding is often ascribed to a lack of cerebellar prediction [48]. This delay in timing is not observed in patients with dystonic tremor [49].

A final implication of our results is that the preservation of adaptation, a type of motor learning, may have potential therapeutic implications. Adaptation could be used to reduce errors in dystonic movements, and this could translate into advances in physical therapy for dystonia [50].

A limitation of our study is the possibility that our task was insensitive to a deficit in adaptation. Perhaps errors were too large in our task to detect cerebellar dysfunction within a biologically relevant range. Against this is the observation that patients with cerebellar damage had an equal difficulty with small and large perturbation errors [51]. Furthermore, based on our mean and variance from the visuomotor adaptation condition (effect size 0.097) and assuming a power level of 0.8, we would need approximately 2,700 subjects in total in order to achieve a significant result. Therefore, we do not believe our null results are due to a lack of power. Another perhaps unavoidable limitation is that patients were receiving botulinum toxin injections (the mainstay of treatment for cervical dystonia). We tested patients when maximally symptomatic prior to injections but the long-term influence of botulinum injections on results cannot not be fully assessed in this or other studies that have used an identical approach.