Little Brain, Big Expectations

The cerebellum has been implicated in the mechanisms of several movement disorders. With the recent reports of successful modulation of its functioning, this highly connected structure has emerged as a promising way to provide symptomatic relief not yet obtained by usual treatments. Here we review the most relevant papers published to date, the limitations and gaps in literature, discuss why several papers have failed in showing efficacy, and present a new way of stimulating the cerebellum. References for this critique review were identified by searches on PubMed for the terms “Parkinson’s disease”, “ataxia”, “dystonia”, “tremor”, and “dyskinesias” in combination with the type of stimulation and the stimulation site. Studies conducted thus far have shed light on the potential of cerebellar neuromodulation for attenuating symptoms in patients with some forms of isolated and combined dystonia, dyskinesia in Parkinson’s disease, and neurodegenerative ataxia. However, there is still a high heterogeneity of results and uncertainty about the possibility of maintaining long-term benefits. Because of the complicated architecture of the cerebellum, the modulation techniques employed may have to focus on targeting the activity of the cerebellar nuclei rather than the cerebellar cortex. Measures of cerebellar activity may reduce the variability in outcomes.


Introduction
Current neuromodulation techniques to treat Parkinson's disease (PD), essential tremor, and isolated dystonia are mainly based on targeting deep basal ganglia nuclei. Despite well-defined benefits of such intervention, some symptoms, such as gait and balance impairments in PD, and complex syndromes, such as combined dystonia and cerebellar ataxia, are only marginally influenced by basal ganglia-based approaches, fueling the quest for novel targets to improve long-term control of these so far ill-controlled symptoms.
Traditionally, the study of the basal ganglia and thalamus have been used to map movement disorders into specific subcortical regions [1]. However, many neurologic symptoms correspond more closely to networks of connected distant regions [2]. Likewise, targeting other nodes of the movement circuitry could influence functionally and structurally interconnected regions, leading to new treatment targets for complex neurological syndromes [3].
In this scenario, the connectivity power of the cerebellum has motivated the study of its modulation among many teams worldwide, and it has been so far explored in a range of well-conducted preclinical and clinical studies [4,5]. The appeal of the cerebellum for neuromodulation strategies is easy to

A Window to Connect the Whole Brain
There is growing evidence that the ideal area for neuromodulation is rather heterogenous within the same "anatomical" target, and influencing the activity of subregions within the same target may provide different clinical results based on the distinct, functionally related networks [2]. For example, parkinsonian patients respond better to subthalamic deep brain stimulation (STN DBS) when the stimulation site is functionally connected to the supplementary motor area [2], while tics in patients with Gilles de la Tourette syndrome are better controlled when the frontal middle gyrus and cingulate are more intensely connected with thalamic stimulation [6]. Cerebellar modulation opens the possibility of modulating the dentato-thalamic pathway and the activities of distant areas, such as the prefrontal, parietal and temporal lobes, and basal ganglia, due to its largely cortical and subcortical connections [5] (Figure 1).
In primates, deep cerebellar nuclei exert a primarily facilitatory effect on excitability in the contralateral primary motor cortex (M1) through dentothalamocortical projections [7]. In healthy individuals, a transcranial magnetic stimulation (TMS) pulse delivered to the cerebellum a few milliseconds before a TMS pulse is administered to the contralateral M1 results in M1 inhibition, revealed by decreased motor-evoked potential amplitude responses (cerebellar brain inhibition) [8]. This is thought to occur due to disruption of the tonic cerebellar facilitatory output to the contralateral M1 under physiologic conditions [3,8]. This normal balance is perturbed by disease (i.e., degenerative ataxia, cerebellar stroke, and dystonia) [3,5,8], and may affect the physiologic interhemispheric inhibition (how both M1s interact with one another) ( Figure 1). For example, abnormal asymmetry in cortical excitability between the right and left hemispheres has been related to the motor impairment seen in cerebellar ataxia [7,8], which was normalized after cerebellar stimulation, improving the symptoms. This network connectivity allows for the construction of models to explain how the modulation of a normal or diseased cerebellum can restore the function of a dysfunctional network due to neurodegeneration or lesions to one of its hubs [3].

Figure 1.
There is an intracortical inhibition between both M1 cortices that is related to maintaining the integrity of axial and limbs movements. The modulation of dentate nucleus activity through tDCS, TMS, or DBS could restore the changes in M1 cortical excitability that are present in some syndromes, such as degenerative ataxia, cerebellar stroke, and dystonia. Additionally, the recent disynaptic connection from the cerebellum to the striatum opens up the possibility of directly modulating aberrant electricity activity in the basal ganglia seen in a range of movement disorders. M1: primary motor cortex; tDCS: transcranial direct current stimulation; TMS: transcranial magnetic stimulation; DBS: deep brain stimulation (adapted from França et al. [9]).

Why Target the Cerebellum in Movement Disorders?
Neuroanatomical studies using transneuronal virus tracers in monkeys have demonstrated that substantial interactions exist between the basal ganglia and the cerebellum [10]. Probabilistic diffusion tractography has confirmed that dentato-thalamo-striato-pallidal and subthalamocerebellar connections also exist in the human brain [11]. Consequently, abnormal cerebellar output could alter activity in the basal ganglia and drive aberrant electricity activity, causing or worsening movement disorders [12]. Furthermore, basal ganglia activity may influence the cerebellum via projections of the subthalamic nucleus to pontine nuclei, which then project to the cerebellum, demonstrating bidirectional connections between these structures [12]. Functional perturbation in these connections may underlie the pathophysiology of dystonia, PD, and spinocerebellar ataxia [3].
It has been shown, for example, that abnormal bursts of cerebellar electroencephalographic activity are correlated with dystonic postures [13]. Notably, disruption of the disynaptic connections between the cerebellum and basal ganglia have been shown to alleviate dystonia in a mouse model [13]. Furthermore, studies of patients with genetic isolated dystonia DYT-TOR1A (formerly known as DYT1) have shown that patients exhibit specific changes in cerebellar connectivity compared with controls and unaffected mutation carriers [14]. Because the non-responder rate of globus pallidus There is an intracortical inhibition between both M1 cortices that is related to maintaining the integrity of axial and limbs movements. The modulation of dentate nucleus activity through tDCS, TMS, or DBS could restore the changes in M1 cortical excitability that are present in some syndromes, such as degenerative ataxia, cerebellar stroke, and dystonia. Additionally, the recent disynaptic connection from the cerebellum to the striatum opens up the possibility of directly modulating aberrant electricity activity in the basal ganglia seen in a range of movement disorders. M1: primary motor cortex; tDCS: transcranial direct current stimulation; TMS: transcranial magnetic stimulation; DBS: deep brain stimulation (adapted from França et al. [9]).

Why Target the Cerebellum in Movement Disorders?
Neuroanatomical studies using transneuronal virus tracers in monkeys have demonstrated that substantial interactions exist between the basal ganglia and the cerebellum [10]. Probabilistic diffusion tractography has confirmed that dentato-thalamo-striato-pallidal and subthalamo-cerebellar connections also exist in the human brain [11]. Consequently, abnormal cerebellar output could alter activity in the basal ganglia and drive aberrant electricity activity, causing or worsening movement disorders [12]. Furthermore, basal ganglia activity may influence the cerebellum via projections of the subthalamic nucleus to pontine nuclei, which then project to the cerebellum, demonstrating bidirectional connections between these structures [12]. Functional perturbation in these connections may underlie the pathophysiology of dystonia, PD, and spinocerebellar ataxia [3].
It has been shown, for example, that abnormal bursts of cerebellar electroencephalographic activity are correlated with dystonic postures [13]. Notably, disruption of the disynaptic connections between the cerebellum and basal ganglia have been shown to alleviate dystonia in a mouse model [13]. Furthermore, studies of patients with genetic isolated dystonia DYT-TOR1A (formerly known as DYT1) have shown that patients exhibit specific changes in cerebellar connectivity compared with controls and unaffected mutation carriers [14]. Because the non-responder rate of globus pallidus internus DBS in isolated dystonia can reach 25% in clinical trials [15], and patients with combined dystonia, such as cerebral palsy, are typically poor responders to pallidal stimulation [15], novel primary targets for dystonia or rescue treatments must be explored.
In PD, cerebellar brain inhibition is reduced, suggesting that cerebellar function or transmission along the cerebellothalamocortical pathway is compromised [16]. Additionally, PD patients have deficient short-latency and long-lasting cerebellar-thalamocortical inhibitory interactions [3]. Previous TMS studies for tremor have suggested that the cerebello-thalamo-cortical circuit may play a pivotal role in the pathogenesis of parkinsonian tremor, and neuroimaging studies have found hyperactivity in the cerebellum in PD [3,5].
Besides its widespread connections, unlike the deeply located basal ganglia and brainstem targets already tested for DBS, the cerebellum can be preoperatively and non-invasively modulated. Thus far, except for the preoperative use of levodopa challenge prior to surgery in PD, there are no other consistent ways of preoperatively predicting surgery outcomes.

What Recent Positive Studies Have Revealed
Cerebellar stimulation could alleviate some aspects of dystonia, especially those related to posture, as has been recently shown in rodents [17]. There is also evidence from clinical studies that TMS of the cerebellum may alleviate symptoms in cervical dystonic patients (Table 1) [12]. Cerebellar anodal transcranial direct current stimulation (tDCS) improved handwriting and circle-drawing tasks in patients with writing dystonia [18]. Another study demonstrated that bilateral deep anterior cerebellar stimulation in patients with secondary dystonia reduces both dystonic symptoms and spasticity [19]. More recently, a patient with generalized fixed dystonia, having failed bilateral pallidotomy, presented significant benefits after high-frequency bilateral superior cerebellar peduncles and dentate nuclei DBS, highlighting that cerebellar DBS may be a new option for fixed dystonia, refractory to classical DBS approaches [20]. In PD, cerebellar continuous theta burst stimulation has been found to change local intracortical circuits in the primary motor cortex and reduce levodopa-induced dyskinesias [21]. Single-blind, sham-controlled To date, most trials involving ataxic patients have focused on degenerative ataxias. Studies have identified temporary and long-lasting (3 months) functional improvement after cerebellar tDCS in patients with ataxia [3,5,37]. Recently, we have demonstrated in a clinical trial that cerebellar TMS using a deep coil improved ataxia in patients with spinocerebellar ataxia type 3 (SCA3), multiple-system atrophy, and post-lesion ataxia (post-stroke or neurosurgery) [9].
Regarding invasive stimulation, low-frequency DBS of the dentate nucleus has been applied in a rat model of neurogenerative ataxia [4]. A frequency of 30 Hz improved motor symptoms, such as ataxia and tremor, and high-frequency stimulation worsened incoordination. This study is probably the most significant in suggesting that the "hot spot" for stimulation would be located at the dentate nucleus. The authors found that the dorsal part of the nucleus was the most effective target for stimulation. In humans, two case reports demonstrated improvement in ataxia after cerebellar DBS in SCA3 and post-lesion ataxia [45][46][47].
Overall, studies conducted thus far, despite having methodological flaws, have shed light on the possibility of relieving symptoms in patients with some forms of dystonia, dyskinesia in PD, and neurodegenerative ataxia.

Playing Devil's Advocate
The recent inclusion of cerebellar stimulation as an option to treat refractory cerebellar ataxia is likely due to the absence of any safer, better treatment option, along with non-invasive stimulation being safe in these settings. However, despite some good outcomes of cerebellar modulation in treating movement disorders in general, there is still a high heterogeneity of parameters employed in the available studies. The best stimulation paradigms and the best profiles of responders are still coupled with uncertainties about the possibility of maintaining long-term benefits [5], which makes it still difficult to currently advise the cerebellum as a new target. Although neurodegenerative ataxia remains orphaned of disease-modifying therapies, current results from cerebellar neuromodulation approaches may suffer from publication bias of positive results and small sample sizes, besides suboptimal blinding. Also, most studies have focused on stimulating still-imprecise areas within the cerebellar cortices, using tDCS or figure-of-eight TMS (i.e., superficial stimulations), with the goal of having an indirect effect on cerebellar-cortex connections [5]. There is currently a lack of information about the specific effects of cerebellar-cortex stimulation on various groups of neurons (e.g., Purkinje neurons, inhibitory interneurons of the cerebellar cortex, and granule cells) and afferent pathways (e.g., mossy fibers and climbing fibers) [5]. Because the cerebellum has a highly convoluted and completely different cytoarchitecture than the neocortex, generalizations of current density and geometry obtained from neocortical stimulation by TMS and tDCS are at least over-optimistic. This lack of specificity makes us rethink whether we are applying the stimulus at the right spot. Because of the complicated architecture of the cerebellum, the focus perhaps should shift from modulating the cerebellar cortex to targeting its output nuclei. This strategy could increase the stimulation's efficiency and reduce variability in the outcomes of cortical stimulation. On an organizational level, the fibers from the cerebellar nuclei directly regulate movement commands in the spinal cord and brainstem, increase motor signals in the cerebral cortex, and modulate signals for adaptive learning via connections to the inferior olive. Direct stimulus to the dentate nucleus via a double-cone coil TMS (which allows for the stimulation of deep structures) [8] and DBS could be more precise, resetting abnormal firing oscillations or enhancing cerebellar output activity, depending on the parameters [4].
Several studies using cerebellar tDCS have compared both anodal and cathodal stimulation with a sham condition. Varying results have been obtained. Most of the studies report a different effect for anodal and cathodal tDCS. Some studies [48] have reported increased cerebellar brain inhibition following anodal stimulation applied over the cerebellar cortex. On the other hand, cathodal stimulation has reduced cerebellar brain inhibition. Two studies found the opposite effect [49,50]. Other studies did not find any effect after either stimulation type [24,51]. Additionally, many studies evaluating the effects of cerebellar cortical stimulation have been negative for motor outcomes in PD [23], essential tremor [42], and dystonia [27] or these studies found considerable side effects [22]. A recent, randomized, sham-controlled study failed to show the efficacy of figure-of-eight TMS over the cerebellum in 22 essential tremor patients [44]. Again, the absence of Magnetic Resonance Imaging-navigated systems and the superficial TMS stimulation applied bring doubt upon which regions of this overpopulated brain area we are stimulating.

So, What Is Next?
It is still unknown exactly what type of activity we are triggering when we stimulate the dentate nucleus. There are probable antidromic effects within the cerebellar cortex, but it would be interesting to test whether there are different responses within the thalamus and other downstream targets, depending on the topography stimulated. If this is true, one must consider the possibility that direct dentate nucleus stimulation could have variable effects, according to which specific regions are recruited [5]. Evidence suggests that the hot spot of modulation is likely located in more dorsal parts of the dentate nucleus, the presumed motor domain [4]. The study of the volume of tissue activated through DBS contacts can represent a powerful research platform to study connectomics from distributed brain networks in the "human connectome" [2].
Additionally, knowledge about modifications in the cerebellum circuitry in each disease, both neuropathological and functional, should help practitioners make decisions about the ideal type of stimuli to apply over the cerebellum. Such work is necessary before proceeding to multicenter clinical trials. Measures of cerebellar activity using functional and Positron Emission Tomography studies and cortical excitability may help with this issue.
Whether the "little brain" will be a primary or a rescue/adjunctive therapy in movement disorders remains an open question. It could perhaps be an alternative target for patients for whom the risk of surgery is high. Substantial changes in clinical practice are often tied to apprehension, but remarkable benefits may arise from innovations. Funding: No specific funding was received for this work.

Conflicts of Interest:
The authors declare no conflict of interest.