|Home | About | Journals | Submit | Contact Us | Français|
Repetitive transcranial magnetic stimulation (rTMS) and transcranial direct current stimulation (tDCS) are promising noninvasive cortical stimulation methods for adjunctive treatment of movement disorders. They avoid surgical risks and provide theoretical advantages of specific neural circuit neuromodulation. Neuromodulatory effects depend on extrinsic stimulation factors (cortical target, frequency, intensity, duration, number of sessions), intrinsic patient factors (disease process, individual variability and symptoms, state of medication treatment), and outcome measures. Most studies to date have shown beneficial effects of rTMS or tDCS on clinical symptoms in Parkinson’s disease (PD) and support the notion of spatial specificity to the effects on motor and non-motor symptoms. However, stimulation parameters have varied widely and some studies are poorly controlled. Studies of rTMS or tDCS in dystonia have provided abundant data on physiology, but few on clinical effects. Multiple mechanisms likely contribute to the clinical effects of rTMS and tDCS in movement disorders, including normalization of cortical excitability, rebalancing of distributed neural network activity, and induction of dopamine release. However, it remains unclear how to individually adjust rTMS or tDCS factors for the most beneficial effects on symptoms of PD or dystonia. Nonetheless the noninvasive nature, minimal side effects, positive effects in preliminary clinical studies, and increasing evidence for rational mechanisms make rTMS and tDCS attractive for ongoing investigation.
Although traditional neurologic nosology broadly classifies movement disorders, such as Parkinson’s disease (PD) and dystonia, as basal ganglia disorders, it is well-recognized that such a singular localization does not explain the breadth of phenomenology or pathophysiology of these disorders.1 Further, the traditional nosology also fails to capture the impact of disease at the individual level. For example in PD patients, expression of symptoms of slowness of movement, unsteady gait, tremor, forgetfulness, depressed mood, or attentional problems and disability vary greatly from individual to individual. This individual variability is partly due to individual differences (genetic and otherwise), and is also partly due to variability in the mechanisms or extent of injury and the capacity of the organism to cope with it. Thus, identification of neural dysfunction within given patients, and individualizing therapy, may provide a more direct and powerful therapeutic target, than a given diagnostic label. Advances in the last four decades have emphasized diverse concepts such as depletion of neurotransmitters (e.g., dopamine), altered network loops between the basal ganglia and cortical targets, and abnormal cortical plasticity as contributing to the pathophysiology of movement disorders. From these concepts, a wide range of current treatment options including medications, botulinum toxin, and deep brain stimulation (DBS) have been developed.2,3
In spite of these advances, limitations in current therapies remain. Dopamine replacement medications are an effective cornerstone of current medical management of PD, particularly for motor symptoms. However, dopamine-resistant symptoms, such as freezing of gait, cognitive deficits, depression, dementia, and hallucinations, have become increasingly recognized as prevalent and as contributing disproportionately to morbidity.4 In addition, long-term treatment with dopaminergic medications may result in problematic motor fluctuations. While DBS procedures in PD can treat medication-induced motor fluctuations in selected patients, there has been increasing recognition of cognitive and mood side effects of DBS in addition to risks attendant with invasive surgical options.5 In dystonia, abnormal movements are treated by combinations of rehabilitation therapy, medications, botulinum toxin and, occasionally, DBS — all of which have significant rates of treatment failure.6
Within this context, noninvasive neuromodulation methods such as repetitive transcranial magnetic stimulation (rTMS) and transcranial direct current stimulation (tDCS) have the potential to fulfill adjunctive treatment roles by a rational and selective modulation of symptoms and their underlying neuropathophysiology on an individual basis.7 Because rTMS and tDCS are applied noninvasively over the scalp, these neuromodulatory techniques avoid complications associated with DBS surgery and the side effects of systemic medications. Repetitive TMS or tDCS theoretically can be applied over selected cortical regions in order to modulate the particular cortical-subcortical network that is linked with a given subset of symptoms. In contrast, DBS, particularly in the subthalamic nucleus, stimulates small subcortical regions where multiple subcortical-cortical loops converge which can result in unintended effects across multiple symptom domains. Further, rTMS and tDCS can modulate or shape cortical excitability, a phenomena which appears to underlie adaptive and maladaptive plasticity, which may be particularly relevant to dystonia where maladaptive plasticity is considered to play a prominent pathophysiologic role.8
In this review, we discuss the rationale and mechanisms for these noninvasive neuromodulation methods and summarize recent trials examining the clinical efficacy of rTMS and tDCS for PD and dystonia as exemplar movement disorders.
Repetitive transcranial magnetic stimulation (rTMS) refers to application of trains of repeated magnetic pulses delivered to the scalp. Passing a brief time-varying current through an insulated coil held parallel to the scalp surface generates a magnetic field perpendicular to the coil. This magnetic field, in turn, induces a weak eddy current within the underlying cerebral cortex. While circular-shaped TMS coils induce a relatively nonfocal circular band of stimulation in the brain, focal figure-8 coils can target stimulation with a functional spatial resolution of 0.5-1 cm.9
Transcranial direct current stimulation (tDCS) refers to application of a constant direct current through a pair of surface electrodes affixed to the scalp. The low-intensity current, 1-2 milliamperes (mA), flows from cathode to anode. Although the scalp possesses high impedance, sufficient intracranial current flows to produce changes in membrane resting thresholds within the cortex beneath each electrode. This results in an increase in activity under the anode and a decreased activity beneath the cathode.
Contemporary interest in studying rTMS or tDCS as potential treatments for patient groups has been partly driven by observations that both methods can produce measurable effects that transiently outlast the duration of stimulation by at least 10-90 minutes.10-12 These effects are considered markers of lasting neuromodulatory effects of each method. Because TMS discharges can perturb ongoing location- and task-specific cortical function and induce twitches of contralateral muscles (over motor cortex), application of rTMS is also considered a neurostimulatory method. In contrast, tDCS produces only a slight tingle as the current is turned on, but has little other discernable effect during application and is considered a predominantly neuromodulatory method.
Development of clinically relevant applications for these neuromodulatory methods depends on reliably inducing lasting effects, understanding their mechanisms, and determining causal clinical correlations in patient groups. Lasting neuromodulatory effects of rTMS or tDCS depend on both extrinsic and intrinsic factors as well as the outcome measures used.13 Extrinsic factors include intensity and frequency of rTMS, the number of pulses per session, the number of rTMS sessions, the interval between consecutive sessions, the targeted cortical region, the orientation of the coil and the coil design. Intrinsic factors include the disease state being treated, the functional state of the brain during treatment, and interindividual variability. Studies of rTMS and tDCS to date have been mostly small studies that necessarily test only a subset of these intrinsic and extrinsic factors. Examples of outcomes used in rTMS and tDCS studies include measures of local cortical excitability, neuroimaging measures of function, and clinical outcomes. While testing for clinical outcomes represent the ideal outcome measure for neuromodulation studies, the other outcome measures provide valuable information about mechanisms for rTMS or tDCS and quantitative measures with which to test effects of different intrinsic and extrinsic factors.
Physiologic effects of rTMS depend on a large number of factors: coil geometry, stimulation site, intensity, frequency, and duration of stimulation, and number of stimulation sessions. Circular TMS coils induce cortical currents that span at least the diameter of the coil and thus are less specific than more focal figure-8 coils. The latter also provide the ability to target specific cortical regions.
Cortical excitability, defined as the responsiveness of the brain to stimulation, can be modulated systematically by rTMS and tDCS. Different rTMS frequencies produce bidirectional changes in healthy subjects with low-frequencies (1 Hz or less) suppressing excitability and high-frequencies (faster than 5 Hz) facilitating cortical excitability.14,15 These excitability changes are transient, lasting 10-30 min with longer duration effects generally linked to higher number of pulses and at higher intensities.11,16,17 Depending on location of the cathode and anode, tDCS also produces differential lasting effects on cortical excitability.12 Anodal stimulation increases cortical excitability and cathodal stimulation decreases it.18-20 Furthermore, lasting effects of tDCS on cortical excitability tend to be greater with higher current intensities for longer durations.19 Converging evidence from pharamacologic and neurophysiologic studies supports the existence of modulation of synaptic strength in local cortical inhibitory and excitatory circuitry, similar to mechanisms of long-term potentiation (LTP) and long-term depression (LTD), as putative mechanisms for these bidirectional frequency effects of rTMS and tDCS.15,21,22
In order to be clinically useful, rTMS and tDCS must induce changes that last longer than a few minutes to hours. Achieving this goal could require repeated sessions of rTMS or tDCS, similar to the model used for electroconvulsive therapy (ECT) for depression. Two daily sessions of rTMS over dorsal premotor cortex (PMd) in neurologically normal subjects can show a cumulative facilitation in M1 excitability on the second day.23 Cortical systems of neurological patients have adapted to dysfunctional circuits, and this adaptation may render them more sensitive to effects of TMS or tDCS than in normal subjects, resulting in greater or longer-lasting responses.24,25 In support of this possibility, while sustained changes in excitability were seen 1 week after rTMS over PMd in PD patients, no such changes were seen in a control group.25 Sustained cumulative clinical effects across repeated sessions have been reported,26-28 but the mechanisms of this effect require further investigation.
Effects of rTMS are modulated by intrinsic factors that reflect interindividual differences. While high versus low frequency rTMS produce reliable bi-directional modulation of excitability as a group, individuals show substantial variability in their response to rTMS, with some subjects showing facilitation of cortical excitability following 1 Hz rTMS and others showing excitability suppression following 10 Hz rTMS.29 Recent studies have suggested a role for genetic polymorphisms in this intersubject variability in responses to rTMS.30
While some intrinsic factors cannot be altered, other individual factors can be manipulated. Among PD patients, effects of rTMS differ depending on whether the patients are in a medication ON or OFF state.31,32 Similarly, among dystonia patients, excitability may differ depending on the context of a voluntary action or on the degree to which botulinum toxin treatment has also influenced cortical excitability.33 Principles of activity- and practice-dependent plasticity would also predict that combinations of rehabilitation therapies with rTMS or tDCS might have promise in facilitating beneficial neuromodulation,34 though few studies have yet been conducted to test this possibility.
The most common outcomes used to quantify neuromodulation effects following rTMS and tDCS are those of cortical excitability. Studies investigating the effects of different extrinsic rTMS factors (frequency, duration, and intensity) generally report their results in terms of modulation of cortical excitability. Since cortical excitability is often altered in patients with PD or dystonia, excitability not only can be used a marker for the disorder of interest, but also can be used as an objective measure for testing extrinsic and intrinsic factors for combinations that would help restore abnormal excitability.
Various measures of excitability, such as motor thresholds (MT), motor-evoked potentials (MEP), cortical silent period (CSP) durations, and paired-pulse intracortical inhibition (ICI) and facilitation (ICF), are defined based on different paradigms of single and paired-pulse TMS over M1 while measuring responses in a target muscle using surface electromyography. Each cortical excitability measure tests different aspects of excitability.35 For example, MT has been related to resting membrane excitability,36 CSP durations to inhibition from GABA-B receptors,37 and ICI to inhibition from GABA-A receptors.38 As such, effects of rTMS and tDCS can be related to differential effects on various mechanistic influences on cortical excitability.
Effects of rTMS and tDCS on sites functionally connected to, but distant from, the site of stimulation can be assessed using a variety of neuroimaging outcome measures.39,40 Functional imaging has revealed abnormalities of cortical activation or blood flow (rCBF) within cortical and subcortical regions in movement disorder patients, suggesting that adaptive (or maladaptive) changes in cortical activity occur in these patients in response to the particular disease state.41 Studies with neuroimaging outcomes which show that rTMS or tDCS that can restore normal patterns of brain activity can provide further support as to which factors may be most relevant for clinical application of these neuromodulatory methods.
Measurements with H2O PET indicated that 10 Hz rTMS directed over the left middle prefrontal gyrus induced rapid rCBF changes not only under the stimulation site, but also in the functionally connected anterior cingulate gyrus.39 Similarly, with flurodeoxyglucose (FDG) PET, following subthreshold 5 Hz rTMS over primary motor cortex (M1), glucose metabolism was increased at the stimulation site and in both distant contralateral M1 and supplementary motor area (SMA).42 These distant effects on a task-specific network can be frequency dependent. Using H2O PET, 1 Hz rTMS induced a reduction, while 5 Hz rTMS showed facilitation, of connectivity between a network of several task-related motor regions.43
In contrast to neuroimaging of regional brain activation or rCBF, radioligand PET or SPECT can image neurotransmitter changes within brain regions distant from the stimulation site and provide additional insight into mechanisms of neuromodulation. Of particular interest to parkinsonian disorders, 10 Hz rTMS over frontal cortex can induce focal dopamine release in subcortical basal ganglia structures and may provide another potential mechanism for beneficial effects of rTMS in PD patients. Consistent with the functional somatotopy of striatal-cortical loops, rTMS of dorsolateral prefrontal cortex (DLPFC) induced dopamine release in the caudate44 and M1 induced dopamine release in the ventrolateral putamen.45
For adjunctive management of movement disorder patients, rTMS and tDCS aim to provide symptomatic benefit for patients with a minimum of adverse effects. Achieving these goals requires clinically relevant outcomes that are specific to each disorder and will be discussed in separate sections below for PD and dystonia.
In contrast to subcortical continuous stimulation in DBS, two particular advantages of the noninvasive cortical neuromodulation approach are the temporal and spatial specificity of each stimulation session. Temporal specificity refers to the ability to deliver each rTMS or tDCS session during particular individual contexts, and provide control of some intrinsic factors. For example, future studies may help to clarify the differential effects of neuromodulation during ON or OFF medication state in PD patients, during botulinum toxin therapy in dystonia patients, or during rehabilitation concurrently with tDCS, thereby allowing a determination of the optimal time for rTMS or tDCS application. The spatial specificity of neuromodulation arises from the ability to stimulate cortical regions within brain networks that are relatively specific for separate groups of symptoms. Examples in PD patients include the targeting of dyskinesia symptoms by rTMS over SMA46 or distinguishing mood from motor effects by rTMS over cortical regions involved with each particular domain.13
Finally, study designs must account for placebo effects for all outcome measures, and particularly for clinical rating scales. Since placebo effects are prominent in studies of investigational devices, and especially in PD patients, clinical studies of rTMS and tDCS require explicit attention to sham neuromodulatory conditions.47,48 The auditory click and scalp sensations of TMS make a true sham-TMS condition a unique challenge.49 Because the low current in tDCS cannot be felt after being turned on, sham tDCS conditions are indistinguishable from real tDCS conditions.50
Both rTMS and tDCS are generally safe, noninvasive procedures with minimal adverse effects. Of most significant concern has been the possibility of inadvertent seizures due to rTMS. Since rare seizures have been associated with prolonged trains of fast-frequency rTMS at high intensities, rTMS stimulation parameters in human subjects are limited by safety guidelines to minimize this risk.51,52 Since publication of these guidelines, experience with rTMS which include fast-frequencies, different patient populations, and use over non-motor regions has largely confirmed the safety of rTMS under these guidelines.53 To date, the literature contains several hundred patients with movement disorders who have been studied with rTMS with no reports of accidental seizures. Among movement disorder patients, no study has shown worsening of clinical rating scores, and only one study over SMA has reported a subclinical worsening spiral drawing.54 TDCS is well tolerated with little recognized risk for seizures. Concerns focus on limiting current flow to less than 2 mA and ensuring adequate electrode size to minimize current density on the scalp.
Regardless of this safety record, caution is required with neuromodulation. Careful monitoring for adverse effects remains warranted, particularly for protocols developed using novel stimulation methods (e.g., theta burst),55 multiple session rTMS or tDCS, or context-related stimulations that aim to have longer lasting or greater modulation of plasticity.
PD is characterized by the involvement of multiple neuro-anatomical pathways,56 with some of the best-recognized clinical features resulting from degeneration of dopaminergic neurons in the substantia nigra resulting in functional dopamine depletion in the striatum. The degeneration of dopaminergic nigrostriatal pathways results in deafferentation of functional targets in the cortex that likely contributes to the pathophysiology of motor and non-motor disturbances in patients with PD. Cortical consequences of this deafferentation include alterations in cortical excitability57 and task-specific network activity.58 As such, factors investigated for rTMS and tDCS neuromodulation in PD include those which induce dopamine release and which can normalize abnormal cortical excitability or network activity.
The finding that cortical rTMS can induce release of subcortical dopamine44,45 has raised interest in this phenomena as a potential mechanism for clinical benefits from rTMS in PD. In PD patients, Strafella and colleagues59 showed that 10 Hz rTMS over M1 can release dopamine in mild hemi-parkinsonian PD patients, and that the release is greater in the more affected hemisphere. This study was not designed to examine clinical benefit, nor did it include control subjects. A subsequent study demonstrated that sham rTMS in moderate PD patients also showed subcortical dopamine release,60 leading to uncertainties as to the significance of dopamine release by rTMS.
Two multisession rTMS studies have also assessed dopamine release hypothesis. A significant reduction of CSF homovanillic acid (HVA) was reported in PD patients who had received 3-4 months of weekly sessions of 0.2 Hz rTMS.61 Since HVA is a dopamine metabolite, this effect was interpreted as inhibiting the dopamine system (despite the observation that PD symptoms improved), a finding at odds with a dopamine release hypothesis. However, with no parallel control group and lack of correlation between CSF HVA levels and PD severity,62 these CSF results are of uncertain clinical significance. Khedr and colleagues63 recently reported an increase in serum dopamine levels immediately following 6 days of daily 25 Hz rTMS sessions over M1 which correlated with motor UPDRS scores. However, the degree to which serum dopamine levels correlate with striatal dopaminergic function is unclear, and again this study was without a control group. Better controlled studies are needed to investigate the validity and clinical significance of the rTMS dopamine release hypothesis.
From a variety of neuroimaging studies, common patterns of cortical activation in PD patients have emerged. In general, decreased activity has been reported around the supplementary motor area (SMA) (often including pre-SMA) and dorsolateral prefrontal cortex (DLPFC) with increased activity in parietal and lateral premotor areas in PD patients.64,65 While the hypoactive brain areas are usually interpreted as a “primary” dysfunction associated with parkinsonian symptoms, the hyperactivity has been interpreted as a neural correlate of adaptive plasticity within the motor system to compensate for the defective cortico-basal ganglia-thalamocortical circuitry.65 Overactivity of premotor–parietal circuits in PD has been thought to represent a compensatory mechanism for deficient activation of impaired striato-mesial–frontal projections.66 Symptomatic therapy with either levodopa64 or subthalamic nucleus (STN) stimulation67 can partially reverse these abnormalities.
These studies suggest a variety of cortical targets for neuromodulation. For example, one might speculate that “facilitatory” rTMS could be used to enhance activity (and excitability) in hypoactive cortical areas and thereby improve clinical functions. Further, neuromodulation can focus on certain cortical nodes within functionally segregated striato-cortical circuits to further target selected symptoms. For example, a motor-putamen loop might be targeted by rTMS over M1 for modulation of motor symptoms while the prefrontal-caudate loop might be targeted by rTMS over DLPFC for modulation of depression symptoms.
Studies of cortical excitability determinants in PD indicate an elevated resting excitability with activity-associated impairment of facilitation. The elevated resting excitability has been reported with larger MEPs, shorter CSP durations, and lower short-latency intracortical inhibition (ICI) in PD patients compared to normal subjects.68,69 During movement preparation, normal MEP facilitation during pre-movement periods is reduced in PD patients.69,70 The slow recruitment of motor cortex (M1) excitability prior to voluntary movement often is interpreted as a primary correlate for bradykinesia while elevated resting excitability represents a compensatory response making it easier to recruit activity from a resting state.66 However, the cortical disinhibition during active contraction suggested by shorter cortical silent period in PD does not follow this model and may also be associated with the compensatory response to bradykinesia.71 Like fMRI network activations, these excitability abnormalities also partially normalize with dopaminergic72 or DBS therapy.73
Clinical correlations to excitability changes in PD are relatively sparse and probably depend on individual symptoms and degree of cortical compensation. Enhanced MEP size at rest has been linked to rigidity in PD,68 subliminal motor thresholds to bradykinesia,74 and cortical silent period durations to UPDRS scores on the more affected side in early PD,71 but many studies do not report such correlations. Even so, assessment of normalization of selected excitability measures remains potentially useful, though not yet validated, as a marker for clinical improvement after rTMS.
Among studies that have applied rTMS in PD patients with clinical outcomes, extrinsic repetitive rTMS factors vary widely. This variation is exemplified by Figure 1, which plots the distribution of published rTMS studies with clinical outcomes across three extrinsic factors: site of stimulation, frequencies tested, and treatment durations. While treatment durations vary between studies, both single session and multisession studies have suggested benefit across a range of cortical sites with most multisession studies following a model of daily sessions over 7- 14 days. Study designs, patient populations, duration of follow-up, and outcome measures are highly variable. For clinical outcome measures, all but one study75 used the Unified Parkinson’s Disease Rating Scale (UPDRS) as clinical outcome measures. For particular symptoms, additional rating scales for depression76-79 or dyskinesias46,80 were employed. Many early studies were not placebo- or sham-controlled. Despite these differences, all but four studies54,81-83 reported some benefit in clinical ratings of PD symptoms. While this may, in part, be due to publication bias, a recent meta-analysis reveals an overall beneficial effect of rTMS, including a subanalysis restricted to sham-controlled studies.84
A circular coil centered over the vertex stimulates a wide variety of cortical areas and does not lend well to interpretation of a mechanism or rationale for treatment. These studies tend to showed poor reproducibility with no benefit reported in controlled studies across low and high frequencies.81,82 In most later studies, figure-8 coils were used to provide greater selectivity for spatial location and from which to interpret potential mechanisms. The most common sites studied are M1 and DLPFC, both of which correspond to cortical targets of motor and prefrontal basal ganglia-thalamic loops, respectively, while SMA and PMd have been explored in few studies.
The primary motor cortex (M1) is a key cortical target for the motor cortical-subcortical loop. The standard basal ganglia circuit model implicates impaired basal ganglia-thalamo-cortical drive as a cause for motor PD symptoms58 and, as such, the primary motor cortex is a common site for cortical neuromodulation to facilitate deficient thalamo-cortical drive. Repetitive TMS studies over motor cortex are summarized in Table 1.
The first encouraging report of rTMS in PD patients used subthreshold (low intensity) 5 Hz rTMS over M1 and found improvement in simple and choice reaction time and pegboard task performance.85 These results were subsequently not replicated in a less severely affected PD group.86 However, interpretation of these studies was limited by the fact that these studies examined task function during rTMS. Since TMS stimulation over M1 might well be predicted to impair ongoing fine motor control, the lasting effects following rTMS trains could more theoretically lead to motor benefits.87 This off-line use of rTMS has become the predominant model for investigating possible benefits of rTMS in patients and constitute the focus of this review.
In a series of single-session studies, Siebner and colleagues88 illustrated lasting effects following high-frequency rTMS over the motor cortex on several outcome measures that suggested benefit to motor symptoms in PD. Repetitive TMS at 5 Hz over the hand region of M1 showed normalization of cortical silent periods88 with improvement in clinical measures of contralateral bradykinesia89 and measures of an aiming movement without loss of accuracy.90 Lefaucheur and colleagues91 confirmed benefits of M1 rTMS with both low (0.5 Hz) and high (10 Hz) frequency rTMS. While 10 Hz rTMS reduced contralateral bradykinesia, 0.5 Hz reduced bilateral rigidity and improved gait speed. Both frequencies normalized silent periods, but while 0.5 Hz rTMS normalized ICI, 10 Hz enhanced ICF. Thus, theories of mechanisms for benefits of M1 rTMS must account for modulation of both excitatory and inhibitory circuits. Bornke and colleagues92 showed that single-sessions of 10 Hz rTMS could improve measures comparably to that of levodopa. Taken together, these studies provide preliminary evidence that a single session of either rTMS over the M1 might improve motor symptoms in PD compared to sham stimulation with more studies employing high-frequencies to enhance abnormal impaired thalamo-cortical drive in PD.
All published multisession studies over M1 have been conducted with high-frequency rTMS and have included a sham-TMS control. Khedr and colleagues27 applied suprathreshold 5 Hz rTMS over arm and leg M1 areas daily for 10 days in unmedicated PD patients. Results showed a cumulative improvement in UPDRS motor subscale scores and walking speed in the rTMS treatment group that persisted 1 month after rTMS compared to a sham-rTMS group. In a follow-up study, early and late stage PD patients were studied with both 10 Hz and 25 Hz daily rTMS over 6 days.93 Results showed cumulative improvement in UPDRS, gait, and tapping speed in groups receiving rTMS compared to control-site rTMS over occipital cortex. The greatest benefits were seen with 25 Hz rTMS in the early PD patients and benefits persisted for up to 1 month post-treatment. Of interest, additional open-label rTMS treatments (daily sessions for 3 days) at monthly intervals appeared to counteract transient wearing-off of benefits from previous rTMS treatments.
The dorsolateral prefrontal cortex (DLPFC) is a key cortical target for the prefrontal cortical-subcortical loop, a circuit that is involved with attention, working memory, and mood regulation. In addition, a majority of studies applying rTMS as treatment for depression have employed high-frequencies to left DLPFC with focal figure-8 coils. All DLPFC rTMS studies have been multisession studies, and several have focused on depressed PD patients. Repetitive TMS studies over prefrontal cortex are summarized in Table 2.
A few early DLPFC studies also used circular coils centered over prefrontal regions.76,94,95 Although these showed clinical benefit, they used nonfocal coils with very low frequencies (0.2 Hz) which, in normal control subjects, do not been produce long-lasting excitability changes.15
In an open-label study of depressed PD patients, 10 daily sessions of 10 Hz rTMS over left DLPFC suggested benefits on both depression scales and UPDRS motor scores when OFF medications.79 In a sham-TMS and active medication treatment control study of depressed PD patients, Fregni and colleagues78 showed that a 10-day course of 15 Hz rTMS over left DLPFC (plus placebo medication) had an equivalent efficacy on depression rating scales as fluoxetine antidepressant (plus sham rTMS) which persisted for 8 weeks.
The effects of rTMS on DLPFC may be specific for depression rather than for motor symptoms. A recent study reported that 10 days of 10 Hz rTMS over DLPFC in non-depressed PD patients did not have a significant effect on motor measures of finger tapping, reaching and gait beyond motor practice.96 Similarly, 10 sessions of 15 Hz rTMS over DLPFC showed no effect on quantitative measures of speech volume and intensity compared to sham rTMS, but open-label single sessions of 5 Hz rTMS over mouth area M1 showed motor speech improvement in voice intensity and fundamental frequency.97
Two imaging studies have investigated the mechanisms of rTMS DLPFC modulation in PD on the cortical activity. Fregni and colleagues28 contrasted rTMS with fluoxetine treatment while investigating the pathophysiological basis for outcomes using SPECT measures of rCBF. PD patients with depression showed reduced baseline rCBF in areas involved in mood regulation including left prefrontal cortex and posterior cingulate. While both rTMS and fluoxetine improved depression equivalently, rTMS differentially increased rCBF in bilateral prefrontal cortex while the fluoxetine group showed increased rCBF in the occipital lobe. An fMRI study, also addressing the neural correlates of depression treatment in PD with multiple sessions of 5 Hz rTMS over left DLPFC versus fluoxetine, found that rTMS produced increased activation in left DLPFC and anterior cingulate with decreased activation in right DLPFC, right fusiform gyrus and cerebellum. Fluoxetine showed an increased activation in right medial and lateral premotor regions, but no changes around either DLPFC.98 Differences between these imaging findings likely reflect differences in rTMS factors and the emotional face observation paradigm used for activation in the fMRI study versus resting rCBF in the SPECT study. However, both studies support effects of rTMS over the DLPFC circuits, not seen in the fluoxetine groups, and suggest that the local normalization of hypoactivity of DLPFC in depressed PD patients may partly underlie rTMS associated improvements in mood.
In a placebo-controlled, multi-session study involving both motor cortex and DLPFC, Lomarev and colleagues26 showed improvement in timed motor tasks and UPDRS scores over 8 sessions over 4 weeks with 25 Hz rTMS over M1 and DLPFC. Improvement in upper extremity bradykinesia was correlated with increase in MEP size after each session of rTMS. However, since MEP size did not increase when compared before and after all eight sessions of rTMS, the mechanism for cumulative benefits of rTMS cannot be explained solely by a long-lasting facilitation of cortical excitability. Nevertheless, the cumulative clinical improvement in gait and upper extremity bradykinesia lasted at least 1 month after the rTMS course.
While the use of several cortical targets simultaneously confounds the ability to determine the topographic effects of high frequency rTMS over DLPFC or M1, this protocol illustrates a potential use of rTMS to simultaneously modulate different striato-cortical loops. As such, mood or cognitive outcome measures would have been of interest in this protocol. Future combined studies could consider modulation of different striato-cortical loops with individually tuned factors for each circuit or set of symptoms.
Imaging data has for the most part supported underactivity of the SMA (and pre-SMA) as playing an important role in PD bradykinesia,64 although some task-related paradigms do show overactivation of SMA.99 However, its location within the interhemispheric fissure makes it a difficult noninvasive cortical target. Further, the proximity of the anterior pre-SMA (related to prefrontal circuits) and SMA proper (related to motor circuits) makes selective topographic targeting of rTMS more difficult.
As an example of targeting specific symptom sets in PD, several studies have investigated rTMS over SMA for the modulation of dyskinesias in PD patients (Table 3). Dyskinesias are abnormal medication-induced involuntary movements which develop in many advanced PD patients and which may limit the ability for medications to remain optimally effective. Functional neuroimaging has demonstrated overactivation of SMA in patients with dyskinetic PD.100,101
Consistent with bi-directional frequency modulation of SMA activity, 1 Hz rTMS over SMA was able to transiently reduce drug-induced dyskinesias while 5 Hz rTMS was associated with a nonsignificant increase of dyskinesias.46 However, the transient benefits of rTMS on dyskinesias did not appear to be enhanced when 1 Hz rTMS was applied across daily sessions.80
In these SMA studies, no adverse effects or motor deterioration was observed in rTMS conditions. An early study, using 10 Hz rTMS over SMA in PD patients, showed a subclinical slowing of reaction time and impairment of spiral drawing, which may have been related to the high intensities and frequencies over an area of the brain with convergent connections to both prefrontal and motor circuits.54
As a cortical target, two studies have investigated effects of rTMS over PMd, but neither with clinical behavior as a primary outcome (Table 4). In a study in early, untreated PD patients, 1 Hz rTMS over PMd was applied while lasting effects on excitability was assessed with single-pulse TMS over M1.25 Notably, effects on paired-pulse intracortical excitability following premotor rTMS persisted 1 week after rTMS which was not seen in healthy control subjects.25 Nonspecific improvements in motor performance and UPDRS were attributed to training, a placebo effect, or both. In the other study, 5 Hz rTMS over PMd was shown to modulate M1 excitability in normal subjects, but not in PD patients OFF medication. Dopaminergic medication restored the ability for 5 Hz PMd rTMS to facilitate M1 excitability in PD patients.
To date, there have been only two publications on tDCS and PD. In the 2006 study published by Fregni and colleagues,102 the motor effects of single-session tDCS of the primary motor cortex (M1) and DLPFC in Parkinson’s patients were studied in the OFF state. This study showed that anodal tDCS of M1 results in a significant motor function enhancement in PD as indexed by simple reaction time and motor scores of UPDRS, and was compared with sham stimulation. These effects were specific for tDCS polarity and site of stimulation as cathodal stimulation of M1 and anodal stimulation of the DLPFC induced small effects that were not significantly different from sham stimulation. In addition, tDCS effects were associated with a polarity-dependent effect on corticospinal motor excitability in PD patients: whereas anodal stimulation results in a robust increase of corticospinal excitability, cathodal stimulation slightly decreases it. Interestingly the increase in the M1 excitability after anodal tDCS was marginally correlated with motor function improvement. This result might be seen as paradoxical when compared with rTMS studies as a potential mechanism for rTMS effects is the normalization of enhanced resting cortical excitability; however, it should be noted that this tDCS study only evaluated one particular measure of cortico-spinal excitability, MEP amplitude and area, and it is possible that anodal tDCS might have modulated intracortical excitability in the same direction as rTMS as shown in normal subjects.103
In another tDCS study by the same group, working memory was assessed in PD.104 The results of this study showed a significant improvement in working memory as indexed by task accuracy after active anodal tDCS of the left DLPFC with 2 mA. The other conditions of stimulation: sham tDCS, anodal tDCS of left DLPFC with 1 mA or anodal tDCS of M1 did not result in a significant task performance change; suggesting not only site-specificity but also dose-specificity (1 mA versus 2 mA).
Although tDCS also seems to induce beneficial effects in PD, further studies are needed to replicate these results, to determine duration of benefits, and to assess effects under different contexts (e.g., ON versus OFF medications).
Dystonia refers to a syndrome of sustained involuntary muscle contractions usually producing twisting and repetitive movements or abnormal postures.105 Dystonia that occurs in the absence of other neurological abnormalities and without an identified brain lesion or cause (other than a genetic mutation) can be considered primary dystonia. Primary dystonia is considered to arise from a functional disturbance of basal ganglia circuits which results in abnormal motor commands.106 This abnormal motor output contributes to inappropriate activation of muscle groups that interfere with voluntary motor control and produce twisted postures of limbs, trunk, or neck. Generalized dystonia can involve the whole body. In contrast, focal dystonia involves abnormal muscle contractions restricted to one part of the body and can be surprisingly context-specific. Writer’s cramp is a particular example where dystonic postures arise in the affected hand only when writing, but not when using hand muscles for other purposes.
Despite dystonia being a widely heterogeneous group of disorders, dystonia pathophysiology, and applications of rTMS and tDCS, mostly have focused on upper extremity limb dystonia. Hand dystonia is often easily activated experimentally and its cortical representation within the contralateral primary motor cortex is easily accessible with noninvasive cortical stimulation. Physiologically, the abnormal motor outflow in dystonia corresponds to excess and nonselective muscle activation. Sensory and sensorimotor integration abnormalities suggest that dystonia pathophysiology is not restricted to the motor system.107 These abnormalities may be considered consequences of excess maladaptive plasticity. In this view, the sensorimotor system in dystonia patients is abnormally sensitive to external stimuli and then may generate inappropriate and nonspecific sensorimotor associations which interfere with context-specific motor actions.8 The potential for rTMS and tDCS to downregulate excess plasticity by appropriate cortical neuromodulation provides a theoretical foundation to normalize or correct abnormal dystonic physiology for symptomatic benefit.
Neuroimaging studies have shown consistent abnormalities in limb dystonia patients distributed within the subcortical and cortical motor system involving bilateral dorsal premotor cortices (PMd), primary sensory and motor cortices (S1, M1), and supplementary motor area (SMA).108-111 However, as expected with the heterogeneity of dystonia, the degree of activity increase or decrease within cortical regions, varies among studies depending on patient variability and differences in imaging protocol and task-activation conditions. In patients with writer’s cramp, excess activity is commonly described in left sensorimotor (SM1), premotor cortex (PMd), SMA, and cerebellum with dystonia-inducing tasks.109-112 In contrast, relative reductions have also been described in left sensorimotor cortex and PMd during nonspecific and specific activation tasks.113,114
Although variable, the predominant pattern in dystonia is one of context-related excess activation of primary sensory, motor, and premotor cortices. These overactivation patterns are often interpreted as loss of specificity in muscle activation, excess maladaptive plasticity in the motor cortex, or increased difficulty of task. These patterns also suggest M1 and PMd as targets for downregulation by extrinsic rTMS or tDCS neuromodulation.
Dystonia is associated with loss of inhibition at multiple levels of the neuroaxis including spinal cord, brainstem, and cortex. The excessive and inappropriate muscle activation patterns seen in patients with focal dystonia reflect disinhibition of cortical-subcortical motor circuits which may be a consequence of abnormalities of sensorimotor integration and maladaptive plasticity.
Among writer’s cramp patients, silent periods are shorter and short-interval intracortical inhibition (ICI) is reduced compared to control subjects.115-117 This disinhibition can be task-specific within the same hand muscle.118 Normal surround inhibition of muscles near the specific muscle intended for movement is impaired in hand dystonia patients.119 Treatment with botulinum toxin may transiently normalize intracortical inhibition in association with clinical benefit.33 Although disinhibition predominates, context-specific impairment in pre-movement excitability facilitation has also been described, perhaps explaining bradykinesia of dystonic movement in some patients.120
Consistent with altered sensorimotor integration, the normal inhibition of excitability following a peripheral electrical stimulus becomes facilitated in patients with dystonia.121 Repeated pairs of peripheral electrical stimulation synchronized with cortical TMS pulses, that normally produce robust facilitation in excitability, resulted in further exaggerated excitability in patients with dystonia.122 Similarly, while muscle vibration normally facilitates cortical excitability to that muscle, excitability was suppressed in patients with focal dystonia.123 These abnormal responses suggest maladaptive plasticity may be a fundamental deficit in symptomatic dystonia.124
Downregulation of these excess plastic responses in a disorder where disinhibition predominates can be a challenge. Low-frequency rTMS or cathodal tDCS, which typically reduce excitability in normal subjects, may not necessarily have the same effect in dystonic patients. However, several studies have shown the potential for normalization of excitability or of abnormal network patterns with low frequency rTMS over M1 or PMd. A few preliminary studies and case reports have also shown some success using with these factors (Table 5). Generalization from these studies, and future directions, must take into account the heterogeneity of dystonia and lack of standardized protocols for assessing outcomes. For example, the two controlled studies focused on writer’s cramp and analysis of handwriting117,125 while the two case reports suggested different patterns of benefit (pain126 and neck dystonia127) on patients with different dystonia diagnoses.
Consistent with maladaptive plasticity in dystonia, motor cortex excitability often responds inappropriately, often disinhibition, following a train of rTMS. After several trains of suprathreshold 1 Hz rTMS over M1, cortical excitability was suppressed significantly in control subjects, but facilitated significantly in writer’s cramp patients.128 With much lower thresholds, 1 Hz rTMS over M1 did not alter any several measures of excitability in focal hand dystonia patients, even though excitability in control subjects was reduced.129 Brief trains of up to 20 pulses of suprathreshold 1 Hz did not change alter excitability, but 5 Hz resulted in an exaggerated, and longer-lived, excitability facilitation compared to control subjects.130
Cathodal tDCS suppresses excitability in control subjects, but tends to increase excitability in dystonia patients. Further, in control subjects, 1 Hz rTMS over M1 following priming by excitatory anodal tDCS, also uniformly reduced excitability. In contrast, dystonia patients had no consistent effect with rTMS regardless of preceding anodal or cathodal tDCS.131 These results were interpreted as a disruption of normal homeostatic regulation of excitability in response to external tDCS and rTMS stimulation. This study also suggests that these rTMS and tDCS factors may not easily restore normal mechanisms of plasticity.
Regardless, one study has suggested some clinical benefit in dystonia patients following low-frequency rTMS over M1 in association with normalization of this abnormal disinhibition.117 Among writer’s cramp patients, single-sessions of subthreshold 1 Hz rTMS over M1 normalized intracortical inhibition and prolonged silent periods. Patients showed a benefit in mean writing pressure with several patients showing clear improvements in handwriting. The normalization of excitability by rTMS in this study are in contrast with other studies which show either facilitation or no change in excitability in dystonia patients following 1 Hz rTMS.128,129 Variation in intensity and excitability measures, and study design may account for these differences, but replication of these findings would be of interest.
The dorsal premotor cortex (PMd) has dense reciprocal connections with both M1 and SMA within the abnormal dystonic cortical network. Similar to M1, rTMS over PMd can modulate M1 excitability bi-directionally with low-frequencies suppressing and high-frequencies enhancing excitability.132 However, rTMS effects over PMd can induce lasting effects on cortical excitability, often to greater degree than with rTMS over M1 itself.133 With PET scanning, subthreshold 1 Hz rTMS over left PMd resulted in reduced rCBF in left sensorimotor cortex (SM1), left PMd, SMA, and cerebellum in both normal controls and patients with focal dystonia.24 However, the decrease was significantly greater among patients in bilateral PMd, SMA, and precuneus. Findings support the use of 1 Hz rTMS over PMd to effect widespread inhibitory effects throughout the motor network in dystonia patients. Interestingly, 1 Hz rTMS over PMd also normalizes spinal reflexes that are abnormal in patients with DYT1 generalized dystonia.134 As such, effects of rTMS over PMd extend down the neuroaxis and provide evidence that modulation at this cortical site can normalize sensorimotor integration at multiple neuroanatomic levels.
Testing low intensity 0.2 Hz rTMS over M1, SMA, and PMd in separate sessions within writer’s cramp patients, Murase and colleagues125 found that only stimulation over the PMd site prolonged silent periods and improved handwriting. Two open case study reports have reported effects of 5 daily sessions of 1 Hz left PMd rTMS. Following rTMS, three patients with severe, generalized, secondary dystonia showed reduced painful axial spasms, but less consistent reductions of abnormal movements or disability.126 In the other case, a patient with primary dystonia affecting neck and limb showed improvement of neck, but not limb, dystonia symptoms after rTMS sessions.127 These studies provide encouraging data in support of the use of multisession rTMS over PMd to modulate dystonia. However, the topographic nature of rTMS modulation over the left PMd and the relationship between rTMS and pain or disability in these patients remain to be clarified.
In conclusion, rTMS and tDCS are promising non-invasive cortical stimulation tools which may provide a future option for adjunctive therapy in PD, dystonia, and other movement disorders. While the studies with clinical outcomes reviewed here show the potential for benefit, many are small studies that span a large range of intrinsic and extrinsic factors. Currently it remains unclear whether rTMS or tDCS have beneficial effects on PD or dystonia and neither modulatory technique has established protocols to assure the predictable long-term effects on clinical outcome measures needed to establish a clinically relevant role in treatment. Future controlled studies, by focusing on selected extrinsic and intrinsic factors, can provide evidence for potentially spatially-specific and symptom-specific roles for rTMS and tDCS. Studies which include functional neuroimaging or cortical excitability outcome measures will add to accumulating evidence for rational mechanisms of rTMS and tDCS effects.
This work was supported by NIH grant K24 RR018875 and a BBVA Translational Research Chair to A.P.L., the Harvard Medical School Scholars in Clinical Sciences Program (NIH K30 HL004095-03) to F.F., NIH grant (K23 NS045764) to A.D.W.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.