Use of non-invasive stimulation in movement disorders: a critical review

Abstract Background: Noninvasive stimulation has been widely used in the past 30 years to study and treat a large number of neurological diseases, including movement disorders. Objective: In this critical review, we illustrate the rationale for use of these techniques in movement disorders and summarize the best medical evidence based on the main clinical trials performed to date. Methods: A nationally representative group of experts performed a comprehensive review of the literature in order to analyze the key clinical decision-making factors driving transcranial magnetic stimulation (TMS) and transcranial direct current stimulation (tDCS) in movement disorders. Classes of evidence and recommendations were described for each disease. Results: Despite unavoidable heterogeneities and low effect size , TMS is likely to be effective for treating motor symptoms and depression in Parkinson’s disease (PD). The efficacy in other movement disorders is unclear. TMS is possibly effective for focal hand dystonia, essential tremor and cerebellar ataxia. Additionally, it is likely to be ineffective in reducing tics in Tourette syndrome. Lastly, tDCS is likely to be effective in improving gait in PD. Conclusions: There is encouraging evidence for the use of noninvasive stimulation on a subset of symptoms in selected movement disorders, although the means to optimize protocols for improving positive outcomes in routine clinical practice remain undetermined. Similarly, the best stimulation paradigms and responder profile need to be investigated in large clinical trials with established therapeutic and assessment paradigms that could also allow genuine long-term benefits to be determined.

transcranial magnetic stimulation (TMS) and transcranial direct current stimulation (tDCS) in movement disorders. Classes of evidence and recommendations were described for each disease. Results: Despite unavoidable heterogeneities and low effect size, TMS is likely to be effective for treating motor symptoms and depression in Parkinson's disease (PD). The efficacy in other movement disorders is unclear. TMS is possibly effective for focal hand dystonia, essential tremor and cerebellar ataxia. Additionally, it is likely to be ineffective in reducing tics in Tourette syndrome. Lastly, tDCS is likely to be effective in improving gait in PD. Conclusions: There is encouraging evidence for the use of noninvasive stimulation on a subset of symptoms in selected movement disorders, although the means to optimize protocols for improving positive outcomes in routine clinical practice remain undetermined. Similarly, the best stimulation paradigms and responder profile need to be investigated in large clinical trials with established therapeutic and assessment paradigms that could also allow genuine long-term benefits to be determined.
During TMS administration, a short-lasting magnetic field is created in the coil, which induces a relatively focal electrical field in the subjacent cortex ( Figure 1). This electrical field can cause depolarization or hyperpolarization of the voltage-gated ion-channels in the cell membranes, which leads to modifications in neuronal activation thresholds 2 . The TMS technique has distinct variables, such as stimulus amplitude, pulse waveform, pulse duration and the diameter and shape of the coil 3 . The interaction of TMS pulse and brain tissue is complex, depending on cortical morphology, tissue conductivity, design of the TMS coil and the current running through it. In general, low (≤ 1 Hz) or high (≥ 5 Hz) frequency rTMS is used to, respectively, decrease or increase cortical brain excitability 3 . Repetitive TMS (rTMS) refers to the application of trains of repeated magnetic pulses delivered to the scalp. Repetitive stimuli can lead to a build-up of effects that might enhance the therapeutic benefits gained from a single application through reduction of the action potential threshold, response enhancement and neuroplasticity due to generation of multiple action potential 3 . In order to produce cumulative effects, besides rTMS trains, several sessions are usually needed, which in most cases are given daily for 5, 10 or more days 1,4 . An alternative use of rTMS has been developed, modelled on theta burst stimulation (TBS) in animals, a technique that comprises short, repeated bursts of TMS pulse at 50 Hz. This is a potentially useful option in movement disorder treatment, and it is discussed later in this paper 3 . TMS is now FDA-approved, as a treatment for patients with depression (unipolar or bipolar), migraine and obsessive-compulsive disorder.
Transcranial direct current stimulation (tDCS) modulates spontaneous neuronal activity through an electrical current applied through a pair of electrodes placed on the scalp. tDCS does not facilitate massive synchronized discharge of action potentials, as TMS does, but it induces polarity-specific changes of resting membrane potential 5 . Cortical excitability is diminished through cathodal stimulation (neuronal hyperpolarization), whilst anodal stimulation increases excitability by means of depolarizing neurons. Although tDCS has been studied less in relation to movement disorders, this review will highlight the most relevant studies that have applied these techniques in clinical trials.

why use noninvasive stimulation in movement disorders?
Despite the fact that the basal ganglia, the core of movement disorder pathophysiology, cannot be reached directly by tDCS or TMS, stimulation over an appropriate cortical region that is part of the basal ganglia circuitry could influence activity within these loops and potentially produce clinical benefit 4 . Thus, the therapeutic background mechanisms for the use of these techniques have both local properties (over well-established cortical areas associated with movement disorders) and remote properties. For example, stimulation of the supplementary motor area (SMA) has been associated with improvement of freezing of gait (FoG) in PD 6 . Because the SMA is a pivotal area for gait initiation and is activated less in PD with FoG than in PD without FoG, its direct stimulation makes sense and has been specifically studied 7 .
The distant, indirect effects of stimulation have been recently studied using noninvasive and invasive techniques (deep brain stimulation, DBS) in movement disorders. For example, parkinsonian patients respond better to subthalamic nucleus (STN) DBS when the stimulation site is functionally connected to the SMA (mainly dorsal STN projections) 8 , while tics in patients with Tourette syndrome (TS) are better controlled when thalamic stimulation is more densely connected with the cingulate and frontal middle gyrus 9 . TMS of the human primary motor cortex (M1) induces release of dopamine in the ipsilateral putamen, as detected by [11C]raclopride PET 10 . Additionally, magnetic stimulation of the pre-SMA significantly affects intrinsic connectivity between the pre-SMA and the striatum observed in functional MRI paradigms 11 . Besides metabolic and functional influences on basal ganglia, motor TMS suppresses pathological beta frequency oscillations in the STN in PD patients 12 . Taken together, the local and remote effects of cortical stimulation would be able to reset abnormal activity in one or more nodes of the altered distributed network and ultimately improve the symptoms of the underlying movement disorder. In the following sections, we analyze the main clinical studies on this matter to date.

MeTHOdS
A nationally representative group of experts performed a comprehensive analysis of the literature in order to develop a critical review on key clinical decision-making factors driving noninvasive modulation in movement disorders.
The methodological quality of the trials reviewed was classified in accordance with the PEDro scale, which includes ten criteria (internal validity with eight criteria and statistical analysis with two criteria; Supplementary File 1) 13 . Classes of evidence and recommendations were described in accordance with the American Academy of Neurology's classification for rating therapeutic studies, whenever possible (Supplementary File 2 and 3) 14 . Furthermore, the aim of the present review was to analyze the relevance and applicability of the evidence to the targeted patient group and the likelihood of a clinical impact through the intervention.
Studies were identified by searching electronic databases and scanning the reference lists of articles. Only articles in English were included, through systematically searching the Medline, Embase and Google Scholar databases. The last search was run on July 19, 2020. The reference and citation lists of relevant studies were manually screened for potential eligible articles. We searched for the terms "Parkinson's disease", "dystonia", "tremor", "ataxia" and "Tourette syndrome", in combination with terms describing the type of stimulation (TMS and tDCS). We included published reports of clinical trials that examined the clinical effects after neuromodulation interventions (classes I, II, III and IV of evidence). For the sake of conciseness, we did not include studies already present in or preceding the meta-analysis and systematic reviews that we reviewed.

Parkinson's disease
Noninvasive stimulation has been studied in PD for almost three decades, and this has resulted in a much larger amount of clinical exploration than in relation to the other diseases discussed in this review. In PD, our search revealed 85 trials using TMS and 37 using tDCS. Given that many recent metaanalyses and systematic reviews are available, only recent non-included trials were selected for individual review, totaling nine studies (Table 1). PEDro scores ranged from 5 to 9 (mean = 8.0 ± 1.4), with six high-quality studies (PEDro 8 or 9).

Transcranial magnetic stimulation
The motor effects of rTMS have been subjected to several meta-analyses. The most recent of these included 28 randomized controlled or crossover trials up to July 2019, with 25 trials evaluating motor symptoms in a total of 787 patients 15 . Only five rTMS studies were published subsequently, and these are discussed individually. All forms of rTMS stimulation were found to ameliorate motor symptoms in a statistically significant manner, even within subgroup analyses. Using UPDRS-III as a standard motor outcome, a quantitative analysis was possible. This showed distinct effect magnitudes, measured using the standard mean difference (SMD). Per stimulation site, the primary motor cortex (M1) had the highest SMD compared with placebo, with 2.22; followed by the dorsal lateral prefrontal cortex (DLPFC) with 1.42; and the SMA and a combination of M1 + DLPFC, both with reported SMD of 1.27 15 . High-frequency rTMS was more effective than low-frequency rTMS. The improvement was greater under off-medication conditions (SMD 2.98) than under on-medication conditions (SMD 1.51) 15 . This is mostly in agreement with a previous 2018 meta-analysis on 23 trials, in which significant motor benefit from high-frequency M1 and SMA stimulation was reported, with greater effect from the former, while non-motor effects were demonstrated using low-frequency rTMS or stimulation over the DLPFC 16 .
Recent trials are in agreement about the motor efficacy of the technique, as measured by means of UPDRS-III, including two positive trials with 20 Hz rTMS on M1 17 , one with 10 Hz on the SMA 18 , and two comparative trials. The first of these comparative trials showed longer-lasting efficacy of 20 Hz rTMS on M1, compared with 1 Hz 19 , while the other showed improvement with 5 Hz only on the pre-SMA, compared with M1 20 .
Dysphagia was evaluated through a single trial, which showed class II evidence of limited improvement after highfrequency rTMS sessions (consisting of 10 daily sessions targeting M1, followed by monthly booster sessions). The improvement of dysphagia was correlated with overall motor improvement 21 . Cerebellum rTMS has been tested as a target for motor improvement in PD, but with disappointing results. Single-session cerebellar 1 Hz rTMS in 20 patients resulted in minor improvement of upper-limb mobility, but worsening of fine motor movements 22 .
Regarding non-motor symptoms, the most recent metaanalysis with rTMS, on depression in the setting of PD, included eight trials comprising a total of 331 patients 15 . Like non-PD depression patients, left DLPFC stimulation was found to be effective for the improvement of depression scores (SMD 1.64), which was consistent with a previous meta-analysis 23 . Lastly, a recent meta-analysis found that rTMS on the M1 or PFC did not have any significant effect on cognition 24 , while minor cognitive improvement was observed in a later study using 20 Hz rTMS on bilateral M1, in PD patients with dementia 17 . No study on TMS and PD has described any serious side effects.

Interpretation
High-frequency rTMS stimulation over M1 is currently the most supported form of rTMS for motor symptoms in PD and is likely to be effective. Other sites and parameters have shown smaller margins of effect so far. High-frequency rTMS DLPFC has already been approved by the FDA for treating major depression in non-PD patients and is likely to be effective for depression in PD patients, thus mirroring what has been established for non-PD patients in this setting.
Gait cadence and duration of turn for sitting also improved in the active group over four weeks of follow-up.  who all showed improvement in dyskinesias after either a single or five sessions 22 . However, cerebellar cTBS did not show any significant effect on tremor 22 . cTBS over M1 had no significant motor effect although when applied over the SMA, a mild yet significant motor improvement after a single session was detected 25 . Intermittent TBS (iTBS) over the premotor cortex (PMC) resulted in no improvement in motor or mood scores 26 . For cognition, single and multiple-session iTBS over the DLPFC failed to elicit any improvement 27 . No study on TBS and PD has described any serious side effects.

Interpretation
Cerebellar cTBS is likely to be effective for controlling levodopa-induced dyskinesias, but there is currently insufficient evidence to support the use of TBS to treat any other motor or non-motor symptoms or signs.

Transcranial direct current stimulation
The most recent systematic review for tDCS and motor symptoms included 29 sham-controlled trials up to July 2019 with a total of 550 patients. Despite the number of studies, the study designs were highly heterogeneous. Both anodal and cathodal stimulation over M1, DLPFC and SMA showed inconsistent and contradictory results, with the sole exception of few distinct multitarget approaches with positive findings for locomotion 28 . A more restricted systematic review analyzed only upper-limb outcomes from 10 studies. The results showed UPDRS improvement after unilateral or bilateral M1 tDCS, especially for bradykinesia. However, these results did not allow the authors to recommend any specific protocol, due to the diverse paradigms found in the studies 29 .
The effect of tDCS on gait was specifically evaluated through a meta-analysis involving 18 sham-controlled trials with a total of 325 PD patients. The results showed that there was an overall positive but small effect size 30 . Again, the diversity of trial designs limited the interpretation, with single or multiple stimulation targets among PMC, M1, PFC or SMA varying in each trial. However, the positive effect size was homogeneous throughout the studies, and the individual effect size variability was low. Six studies that assessed longterm effects failed to show significance 30 . A recent crossover trial with anodal tDCS over the cerebellum reported improvement in dyskinesia during the active phase, with no other detectable motor improvement 22 .

Interpretation
Current tDCS application for PD is significantly hampered by significant heterogeneity in the population characteristics, PD stages, stimulation parameters, trial designs and outcome measurements. Thus, deriving a consensus on any set of stimulation procedures is extremely difficult. No recommendation is currently possible for motor symptoms other than gait, for which the method is likely to be effective but with a limited effect size (its use as an add-on therapy is discussed below).

dystonia
Dystonia is a syndrome characterized primarily by excessive muscle contractions leading to abnormal postures and involuntary twisting movements 31 . As it represents a heterogenous group of disorders, in this review we will focus on isolated dystonia, given that certain pathophysiological mechanisms have been consistently identified and may have implications for widening the therapeutic strategies, such as noninvasive stimulation 31 .
Three general abnormalities appear to underlie network pathophysiology: loss of inhibition at different levels of the central nervous system leading to overflow phenomena, altered sensory integration and maladaptive synapse plasticity 32 .
The current treatment of dystonia includes selected drugs for specific cases, such as dopa-responsive dystonia, botulinum toxin injections for focal dystonia and surgery, notably DBS. However, new insights into the pathophysiology of dystonia and positive results from invasive neuromodulation trials have shown that this disorder should be seen as a consequence of network malfunction. In dystonia, there is an increased tendency to form aberrant input and output connections, potentially leading to abnormal improper connectivity and ultimately impairment of motor control 1 . As such, dystonia could be potentially responsive to noninvasive neuromodulation strategies.
Here, we selected studies on adults with three forms of focal dystonia: hand dystonia, cranio-cervical dystonia (CCD) and blepharospasm.

Hand dystonia
Fifteen clinical trials have evaluated the effects of noninvasive stimulation for hand dystonia. These trials consisted of a total of 176 patients with different clinical presentations: writer's cramp (96), musician's dystonia (70), focal hand dystonia (9) and dystonic tremor (1). These studies were heterogenous with regard to methodology, neuromodulation technique and parameters and outcome measurements. Their main findings are summarized in Table 2. In terms of methodological quality, PEDro scores ranged from 3 to 8 (6.5 ± 1.5). Three studies had scores < 6 and were considered to be low quality [33][34][35] . For the remaining ten studies, seven were classified as moderate and five as high quality with PEDro scores of 6/7 and 8 respectively [36][37][38][39][40] . In fact, most of the trials presented class III or IV evidence and had an exploratory approach to the technique and its clinical outcome.
Most studies have targeted the motor cortex (PMC in six; M1 in five; SMA in one). Regarding the neuromodulation technique, seven studies applied rTMS ( four singleblinded, one double-blinded, two open), seven used tDCS (all double-blinded), and one used cTBS (one single-blinded). The majority of these studies were performed in a single session and had short follow-up periods.
Overall, a limited number of studies reported clinical improvement, especially with short-term evaluation. An open-label study of rTMS on M1 showed that this method can reinforce deficient intracortical inhibition and may temporarily improve handwriting 34 . These results were reinforced by two other single-blinded trials 41,42 and one double-blinded trial 43 , in which rTMS of the PMC led to temporary improvement in handwriting. Another target with good response was S1, which induced improvement in writer's cramp, measured by a 20 min writer's task 44 .
Regarding tDCS, only one double-blinded cathodic stimulation reported a good outcome, with an improvement in the effectiveness of rehabilitation in task-specific (musician's) dystonia 38 . A current of 2 mA was used, applied to the SMA bilaterally during the first 20 min of a 1 hour sensorimotor rehabilitation program over 10 sessions. In this trial, neuromodulation caused an increase in the effectiveness of rehabilitation of musician's dystonia.

Interpretation
The evidence presented to date does not provide a high level of recommendation for the use of noninvasive modulation in hand dystonia. So far, studies have presented heterogeneous designs and quality, although interesting positive outcomes have been reported. The most promising target seems to be the PMC. Low-frequency rTMS over the PMC is possibly effective in improving handwriting temporally in FHD (studies with class III evidence). For other types of hand dystonia, including musician's dystonia, there is insufficient evidence to support the use of rTMS or tDCS.

Cranio-cervical dystonia (CCd)
Regarding CCD, only three clinical trials, consisting of a total of 38 patients, have evaluated the therapeutic effects of noninvasive stimulation (Table 3). PEDro scores ranged from 4 (one low-quality study) to 9 (mean, 7.25 ± 1.8), while most had scores of up to 7, which was considered high quality although still presenting class III evidence.
These studies presented heterogeneous techniques and targets: (1) two tTMS in motor cortical areas; and (2) one cTBS in the cerebellum. Koch et al. 45 performed a doubleblind sham-controlled trial of five sessions per week for two weeks among 18 CD patients with cTBS of the cerebellum bilaterally. They showed that there was a small, but significant improvement in TWRTS in comparison with baseline scores, at the first evaluation after the end of stimulation period (mean 33.6 ± 4.2 versus 38.8 ± 4.1; p = 0.008). Another study showed a mild improvement through targeting dPMC and M1 46 . Although a limited number of case series have reported improvements of CCD scores through tDCS, no formal clinical trial has been performed to date.
Only one double-blinded, sham control study (class III evidence) with noninvasive stimulation has been conducted in relation to blepharospasm 47 . The intervention applied was rTMS to the anterior cingulate cortex, and this led to significant improvements of all clinical outcomes at the end of the active stimulations and at 1 h afterwards.

Interpretation
There is insufficient evidence to support or refute the effectiveness of noninvasive stimulation in CCD and blepharospasm, mainly because the samples of patients studied so far have been small, and the overall quality of these studies has been low. Even so, the studies performed have explored targets (such as the cerebellum and the motor cortex) that can be potentially effective for CCD, and larger studies are warranted.
Only one study provided class II evidence; three studies provided class IV, and the remaining five provided class III evidence. PEDro scores ranged from 6 to 10 (7.4 ± 1.5). Only one study satisfied all the PEDro score criteria 48 , and six studies scored 6 to 7 [49][50][51][52][53]56 . Therefore, we considered that the majority of these studies were of moderate/high quality.
The length of the follow-up was quite variable, ranging from ≤ 1 hour after the end of stimulation ( five studies) to different numbers of weeks after completion of the study protocol ( four studies), with a maximum follow-up of 12 weeks 55 . Obviously, this lack of uniformity limits the potential for accurate comparison of the data. In the longest follow-up, the improvement in tremor after 15 sessions of 1 Hz rTMS continued to be present until after eight weeks, but not after twelve weeks 55 . Three studies showed clinical improvements through validated tremor scales after noninvasive stimulation, over periods ranging from 5 min to 12     weeks afterwards 48,51,55 . All three used 1 Hz rTMS: two over the cerebellum and one over the motor cortex 48,51,55 . Seven studies also analyzed tremor characteristics using accelerometric records and, after considering the data from three additional studies, demonstrated positive short-term results ranging from immediate to 40 min after the procedure 49,50,53 . Two used cTBS over the motor cortex, and one used cerebellar 1 Hz rTMS 53 . No cTBS study reported any improvement in clinical tremor scales, although two had positive accelerometric outcome data 50,53 . Among the studies with the highest PEDro scores (9-10) 48,54,55 , two were positive and one was negative. Four studies (three reporting positive accelerometric results) did not mention the presence or absence of side effects, although we believe that side effects are particularly important in clinical trials 49,50,53,55 . Out of the remaining studies, only two reported mild side effects, including one with positive clinical results 48,52 .

Interpretation
Through gathering the available data, low-frequency rTMS over the cerebellum was seen to be possibly effective in improving tremor in ET (studies with class II and III evidence). The outcomes from tDCS studies do not support any conclusions and, thus, tDCS cannot be recommended 52 . Overall, the heterogeneity of the reports and outcomes is still an obstacle in systematically implementing noninvasive neuromodulation for ET in clinical practice.

Cerebellar ataxia
Eleven clinical trials included patients with cerebellar ataxia (n = 238) and evaluated clinical changes after noninvasive stimulation (Table 5). Different ataxia etiologies were included: eight studies included patients with heredodegenerative ataxias (n = 155) 22,57,58 , four included patients with sporadic adult-onset ataxia (n = 14) 58,59 , three included patients with cerebellar-type multiple-system atrophy (n = 15) 22,58 , two included patients with cerebellar ataxia due to stroke (n = 38) 60,61 , two included immune-mediated ataxias (n = 10) 22,62 and one included cases of ataxia in the setting of cerebral palsy (n = 6) 63 . Overall, the population included tended to be heterogeneous throughout the studies, and was limited to a single diagnosis in only four studies (multiple sclerosis in one, stroke in two and cerebral palsy in one) 22,62 .
Two studies provided class IV evidence, while seven provided class III, and two provided class II. PEDro scores ranged from 4 to 9 (6.7 ± 1.9), considering all the studies. Studies using tDCS had higher PEDro scores (TMS mean = 6; and tDCS mean = 7.6).
Almost all the studies used the cerebellum as the target; only one study targeted the motor cortex (M1) 62 . Regarding the neuromodulation technique, four studies used singlepulse TMS (two of them were double-blind) 57,64-66 , three used rTMS (two double-blind) 60-62 and five used anodal tDCS (three double-blind) 59,63,[67][68][69] . The chosen stimulation frequency ranged from 1 to 50 Hz, and the number of sessions ranged from one to ten. All five tDCS studies used anodal stimulation, with currents ranging from 1 to 2 mA and numbers of sessions from one to ten. No study involved maintenance sessions, and none measured long-term effects, although it could be argued that three months of follow-up would constitute long-term in noninvasive neuromodulation.
Only one study, using one session of 1 mA anodal tDCS, reported no clinical improvement after neuromodulation 67 ; the remainder described positive results. Many studies assessed ataxia exclusively through axial measures of gait and balance, and only six studies applied validated ataxia scales 57,59,60,66,68,69 . Both studies with the highest PEDro scores (9 and 10) used anodal tDCS and reported clinical improvement: one with validated scales and the other with gait evaluation. Out of the three studies with PEDro scores of 8, two used anodal tDCS and one used single-pulse TMS 22,57,58 ; all of these reported positive outcomes with validated scales.
Reporting of adverse effects was far from ideal, since half of the studies did not mention them 59,60,62,66,67,69 . Among the remaining studies, five reported none 57,61,64,65,68 , while one reported only mild transient side effects after anodal tDCS 63 .

Interpretation
Overall, the studies on cerebellar ataxia were of moderate quality, and all but one reported short-term positive results. As such, there is insufficient evidence so far to support the use of rTMS to treat ataxia. Anodal cerebellar tDCS is possibly effective in treating degenerative cerebellar ataxia, but multicenter trials and metanalyses are needed. Considering that cerebellar ataxia is currently an orphan symptom, and that these procedures are noninvasive and safe, off-label treatment could be attempted in selected cases.

Tourette syndrome (TS)
The pathological mechanism underlying TS is a matter of debate, but some studies have proposed that the primary motor (M1?) cortex and the SMA are hyperexcitable in these patients 70 . These findings have led to the hypothesis that normalizing the excitability of these areas would be a potential treatment for tics. Accordingly, low-frequency rTMS applied over bilateral SMA was found to improve tic severity for up to six months in two open-label trials 71,72 . On the other hand, two randomized controlled trials using the same target did not show any significant improvement in tics after TMS 73,74 , although in one of them an additional three-week open-label active treatment ( for those patients initially randomized to active rTMS) resulted in an overall 29.7% reduction in tic severity, compared with baseline 74 . Additionally, active TMS induced significant inhibition in the primary motor cortex network. Other targets such as motor and pre-motor cortex were found to have effect on tics in TS patients 75,76 (Table 6).    Predictors of efficacy of TMS for TS have been recently discussed, and it has been pointed out that rTMS is more effective in TS with comorbidities (such as obsessive-compulsive disorder and attention deficit hyperactivity disorder), compared with "purer" TS forms 70 . In accordance with the mechanism for TMS, a higher degree of brain electrical activity in individuals with TS and comorbidities ( for example, ADHD has been associated with excess theta activity) might constitute a variable that is predictive of good response 70 .
Studies using tDCS in TS are scarce. A recent sham-controlled trial showed that a single session of tDCS over SMA was not effective, with regard to time conditions (pre/post) and stimulation conditions (cathodal/sham), in ten patients 77 .
The mean total PEDro score for the reports on these trials was 6.1 (3.1 ± 3). Three studies were considered to present moderate/high quality (> 5 points).

Interpretation
Repetitive TMS over the SMA is likely to be ineffective in reducing tics in TS. The lack of available data regarding other targets and the use of tDCS do not allow any conclusions to be reached regarding these techniques. Two points about the studies need to be addressed. First, the amount of stimulation given (rTMS or tDCS) seems to be an important consideration, since multiple sessions appear to be more effective than a single session. Likewise, a downstream effect from TMS has been demonstrated after repetitive stimulation over the SMA 74 . Second, the high phenotypic variation and comorbidities might have contributed to the disparity of the study outcomes. Besides clinical characteristics, the baseline electrical brain activity has been associated with different TMS responses 70 . How rTMS and tDCS affect this interplay between clinical and electrophysiological patients' characteristics should be addressed in further studies.

discussion and future directions
In this review, we have summarized the evidence available regarding noninvasive stimulation therapies for managing the main movement disorders. We found that TMS is likely to be effective for motor symptoms, dyskinesia and depression in PD. Additionally, TMS is possibly effective in improving handwriting in FHD, tremor in ET and cerebellar ataxia and is likely to be ineffective in reducing tics in TS. tDCS is likely to be effective in improving gait in PD.
Despite good results in several studies and the robust background rationale for applying noninvasive stimulation in movement disorders, it remains to be determined how to optimize rTMS and tDCS protocols to give them relevance in routine clinical practice. The best stimulation paradigms and the best responder profiles are still coupled with uncertainties regarding the long-term genuine benefits 4 . Two major factors are determinant. First, there has been large heterogeneity among the protocols used, with different cortical targets and numbers of sessions, which complicates the generalization of the current findings. In comparison, most studies relating to depression have focused on a single recommended site and protocol, producing consistent results. Second, there is still a lack of large-scale trials that could assess the placebo effects of stimulation and the real effects of stimulation in large samples of patients.

noninvasive stimulation as an add-on therapy
In our view, because noninvasive stimulation has a low effect size, it should be seen as a "supporting actor, " and not as the "leading role". This does not mean that it is less important than rehabilitation or medications, but it thus forms a complementary option, as in the case of botulinum toxin in dystonia or physiotherapy in FoG and ataxia, for example. In this era of personalized medicine, it is of paramount importance to note that noninvasive stimulation will probably serve a subset of patients with a subset of clinical characteristics and specific brain activity abnormalities or network disruption, but not all patients. Additionally, it could be a therapeutic tool for patients for whom surgical approaches are contraindicated, given the safety and lower rate of side effects shown by noninvasive stimulation. Another promising approach is the use of rTMS prior to a training intervention (e.g. prior to physiotherapy if treating gait in PD) or even during a given task in the case of tDCS (e.g. while performing physiotherapy for writer's cramp). In these cases, the rationale is to strengthen the effectiveness of synaptic connections and recruit fibers required to improve performance during a given task 7 .

new stimulation paradigms are needed
The classical targets explored so far (M1, supplementary area and prefrontal cortex) need to be investigated in multicenter clinical trials, and new paradigms need to be explored with combined or new targets. As discussed earlier, because of the great connectivity of the cerebellum, this has emerged as a potential hot spot for TMS and tDCS 22 . Along this line, because the spinal cord acts as a 'highway' of neural information and remotely regulates cortical areas relating to the pathophysiology of movement disorders, its modulation is theoretically interesting. A novel approach involving stimulation of the spinal cord noninvasively through a transcutaneous electrical or magnetic field applied to treat spasticity and urinary incontinence has recently emerged 7 . Its applicability is likely to catch the neurology community's attention over the coming years.
Lastly, network dysfunctions in movement disorders are not straightforward and involve multiple supraspinal circuits that are both distinct and overlapping 6,8 . The presence of different underlying circuitries may explain the variability in cortical targets that have been pursued in the literature. An approach involving multiple/combined cortical targets, as recently explored in FoG, may be a key step forward. In that investigation, stimulation of the dorsolateral prefrontal cortex simultaneously with M1 induced greater benefit for diminishing FoG severity, compared with M1 stimulation alone 7 .
Noninvasive stimulation approaches have the potential to be successfully applied in clinical practice, either as an adjuvant therapy or as the protagonist in patients with contraindications for other treatments. Large-scale clinical trials focusing on targets previously attempted, from which positive results were obtained, are needed. The best scenario for this developing therapy will probably involve optimization of stimulation protocols, together with new paradigms looking to recruit fibers and neurons during rTMS and tDCS sessions to enhance rehabilitation effects.

aCKnOwLedGeMenTS
We would like to thank Paula Starck for designing Figure 1.

SUPPLeMenTaRY MaTeRiaL
The following material is available online for this article: S1. PEDro scale -Internal Validity (8 Criteria). S2. American Academy of Neurology Classification of Evidence for the Rating of a Therapeutic Study and Recommendations.
S3. Recommendations based on study classes.