Spinal Cord Stimulation for Gait Disorders in Parkinson ’ s Disease and Atypical Parkinsonism: A Systematic Review of Preclinical and Clinical Data

Background: Falls in extrapyramidal disorders, particularly Parkinson ’ s disease (PD), multisystem atrophy (MSA), and progressive supranuclear palsy (PSP), are key milestones affecting patients ’ quality of life, incurring increased morbidity/mortality and high healthcare costs. Unfortunately, gait and balance in parkinsonisms respond poorly to currently available treatments. A seren-dipitous observation of improved gait and balance in patients with PD receiving spinal cord stimulation (SCS) for back pain kindled an interest in using SCS to treat gait disorders in parkinsonisms. Objectives: We reviewed preclinical and clinical studies of SCS to treat gait dysfunction in parkinsonisms, covering its putative mechanisms and ef ﬁ cacies. Materials and Methods: Preclinical studies in animal models of PD and clinical studies in patients with PD, PSP, and MSA who received SCS for gait disorders were included. The main outcome assessed was clinical improvement in gait, together with outcome measures used and possible mechanism of actions. Results: We identi ﬁ ed 500 references, and 45 met the selection criteria and have been included in this study for analysis. Despite positive results in animal models, the outcomes in human studies are inconsistent. Conclusions: The lack of blind and statistically powered studies, the heterogeneity in patient selection and study outcomes, and the poor understanding of the underlying mechanisms of action of SCS are some of the limiting factors in the ﬁ eld. Addressing these limitations will allow us to draw more reliable conclusions on the effects of SCS on gait and balance in extrapyramidal disorders.


INTRODUCTION
Parkinson's disease (PD) is the second most common neurodegenerative disorder, affecting an estimated 10 million people worldwide, with an ever-increasing prevalence in the older population. 1 The pathological hallmark of PD is the loss of dopaminergic projection neurons in the substantia nigra, manifesting as bradykinesia, tremor, and rigidity. 2 These features respond to dopamine replacement therapy, particularly in early-stage PD. As PD progresses, more debilitating features such as gait dysfunction and postural instability occur, which can lead to falls. 3 Falls are a major milestone in PD progression, resulting in loss of independence, worsened quality of life, and markedly increased mortality. 4 The socioeconomic costs of falls in PD are high, with 80% of spending on PD care arising from acute admissions, primarily falls related. 5 Critically, gait and balance dysfunction both respond poorly to dopaminergic drugs and advanced therapy, including subthalamic nucleus (STN) deep brain stimulation (DBS). 6 1 In atypical parkinsonism, such as progressive supranulear palsy (PSP) and multiple system atrophy (MSA), falls and postural instability occur early in the disease course and are critical symptoms for the diagnosis. 7,8 Spinal cord stimulation (SCS) is an established therapy to treat persistent spinal pain syndrome with radicular pain, complex regional pain syndrome, peripheral neuropathy, phantom limb pain, angina, and ischemic limb pain. 9 In humans, a positive effect on locomotion was serendipitously observed in patients with PD implanted with SCS for back pain. 10 Subsequently, a small number of patients with PD (without back pain) received SCS for freezing of gait (FoG), with encouraging results and few reported adverse events. 11 In this review, we will first discuss SCS results in animal models of PD, according to motor symptoms improvement. Next, we will focus on human studies targeting gait dysfunction in PD and other movement disorders, highlighting differences in study design, patient selection, and stimulation parameters. Finally, we will briefly discuss some possible mechanisms, mainly based on results from animal studies. These considerations will provide useful information for the appropriate testing of SCS as a potential treatment in advanced PD.

Literature Review
This systematic review was carried out in the PubMed/MEDLINE data base, using the terms "Spinal Cord Stimulation AND Movement Disorders," "Spinal Cord Stimulation AND Parkinson's Disease," "Spinal Cord Stimulation and Freezing of Gait," "Spinal Cord Stimulation and Multiple System Atrophy," and "Spinal Cord Stimulation and Progressive Supranuclear Palsy." We searched for original articles and case reports, written in English, and published from January 2009 to November 2022. Furthermore, six recently published reviews on SCS in PD were screened for more articles.

Physical Parameters of SCS
A detailed description of the biophysics of SCS is beyond the scope of this review, and more specific studies on this topic can be found in the literature. 12,13 To allow the reader a comprehensive understanding of the studies on SCS for gait in PD, we summarized some basic details of SCS parameters and waveforms in Figure 1.

RESULTS
A total of 500 unique records were screened, and 51 full texts were considered eligible for this review, including six review articles. These were screened for further original reports and then discarded. The screening excluded 419 studies because the research topic was not considered appropriate, given neither SCS nor PD of gait dysfunction was mentioned. One article was discarded because no full text was available.
Finally, 45 studies were selected. We subdivided these into animal and human studies. Because all animal studies focused on PD, we subdivided them according to the main clinical effect investigated, namely, acute effects on motor symptoms, the duration of the effect on motor features (long-lasting and cumulative effects on motor symptoms), and a summary of the observed neurophysiological findings relevant to understanding mechanisms. The human studies were classified according to the underlying disease and the indication for the SCS implant (Tables 1-3).

Animal Models Clinical Data
Seven studies have been published so far on the effects of SCS in animal models of PD. Six of seven focused on nonprimate animal models, with only one study in adult 6-OHDA parkinsonian marmoset. SCS electrodes have been implanted in the upper thoracic spinal cord in four studies [36][37][38][39] and in the lumbar spinal cord at L2 level in one study, 40 whereas in two studies, a monopolar electrode was placed at C1-C2 level. 41,42 SCS was delivered Figure 1. Biophysics of SCS. We provide a visual representation of the three most used paradigms in SCS studies. In tonic stimulation, electric pulses are delivered at a consistent frequency, pulse width, and amplitude. In burst stimulation, electric pulses are closely spaced and delivered at high frequency within a burst, and bursts are repeated at a certain frequency. In high-frequency (HF) stimulation, electric pulses are delivered at frequencies up to 10 kHz. [Color figure can be viewed at www.neuromodulationjournal.org]   -T4  T5-T7  T6-T8  T6-T9  T7-T9  T8-T9  T8-T10  T9-T10  T9-T11  T10-       with different parameters, and details for each study are presented in Table 4. SCS was delivered in the long term in one recent study, 42 with intermittent daily or weekly application of electric fields usually provided. Neurophysiological recordings were obtained in three of six studies, focusing on changes in brain activity rather than local effects on spinal cord. 36,37,40 Outcome measures can be subdivided into acute and longlasting effects on motor features. Possible neuroprotective effects have been suggested, but these results are not discussed in this review. 39,41,42 Acute Effects on Motor Features In 2009, Fuentes et al showed that SCS at a frequency of 300 Hz markedly increased the total amount of locomotion and bradykinesia in a rat model of PD, both with and without levodopa supplementation. 36 Similar changes also were seen in a 6-OHDA parkinsonian marmoset model, 37 over a wider range of stimulation frequencies (4-300 Hz).
In an experiment to evaluate possible neuroprotective effects of SCS in a model of 6-OHDA lesioned rats, Shinko et al investigated the effect of three different SCS frequencies (2,50,200 Hz) and found that 50 Hz was effective in improving motor performance. 41 Recently, the same group reported a marked improvement in motor activity after 50 Hz SCS was continuously delivered for eight and 24 hours. 42 In 2016, Brys et al applied SCS in unilateral substantia nigra AAV6-lesioned rats. The authors used a stimulation intensity approximately 25% less than that reported from previous studies. 36,37 SCS was able to alleviate the motor asymmetry in the stimulated group. 38 Finally, Zhong et al assessed the effectiveness of stepping control of the lumbar spine epidural stimulation in 6-OHDA unilateral lesioned rats. Two different SCS protocols were compared: a tonic stimulation at 40 or 333 Hz with a pulse width of 0.2 milliseconds and a train stimulation (one train per second, train duration 250 milliseconds, frequency 40 Hz, pulse width 0.2 milliseconds). Both SCS paradigms significantly improved stepping initiation and duration at two and four weeks after the unilateral nigrostriatal lesion. Train stimulation at four weeks after lesion was superior to tonic stimulation in ensuring a higher quality of stepping. 40

Long-Lasting and Cumulative Effects on Motor Symptoms
Long-lasting effects of SCS on motor symptoms in animals are less well characterized than are the acute effects. The previously mentioned studies showed that the effects of stimulation are almost immediate and last while the stimulation is maintained, waning a few seconds after the stimulator is switched off. Indeed, there are no reports of sustained motor improvement after the stimulation is switched off.
In contrast, Yadav et al 39 showed a cumulative effect of SCS in adult 6-OHDA lesioned rats, with a progressive improvement in posture and in some gait parameters at the end of a six-week observational period. An early and progressive recovery in body weight also was observed in the 6-OHDA + SCS rats, suggesting that SCS could improve the global wellness of the nigrostriatal lesioned rats. Interestingly, a linear correlation was found between body weight and motor functions, suggesting that (1) rats with the greater weight loss have severe motor symptoms and (2) the weight recovery correlates with the motor improvement. 39

Neurophysiological Data
Neurophysiological changes due to SCS have been reported only in animal studies. SCS effects on brain oscillatory activity have been extensively studied, whereas local effects on spinal cord circuitry have been inferred from electromyographic (EMG) recordings.
Fuentes et al recorded local field potentials (LFP) from the dorsolateral striatum and primary motor cortex (M1) of dopamine depleted mice. LFP recordings showed an immediate and prolonged shift in spectral power, mainly toward high frequencies, with a reduction in low-frequency oscillations during SCS. 36 LFP and single-unit activity recordings also have been recorded in a primate model of PD. 37 For this purpose, animals were implanted with cerebral microelectrodes in both hemispheres, targeting several structures, including M1, primary somatosensory cortex, putamen, subthalamic nucleus, globus pallidus, and thalamus. A suppression of LFP oscillations in all parts of the cortico-basal ganglia-thalamic loop was observed during SCS, regardless of the stimulation paradigm. Interestingly, SCS treatment resembled the effects of the levodopa treatment on LFPs, with a shift toward a healthy brain activity pattern. Taken together, these experiments showed that SCS probably acts through the disruption of an excessively synchronized neuronal oscillatory activity, and a decrease in low-frequency oscillations probably has a permissive role for the initiation of locomotion. 36,37 Local effects on spinal cord circuitry have been hypothesized by Zhong et al. EMG data showed a difficulty in initiating stepping in a running wheel in 6-OHDA-lesioned rats, a deficit that seems to be restored by SCS. 40 These data are in line with results from SCS studies on spinal cord injury in animals, in which the neuromodulation of spinal cord circuitry in the absence of any supraspinal control seems to be effective in regaining stepping. 43

SCS in Patients With Parkinson's Disease
The first reports of SCS benefit for axial symptoms were based on a serendipitous observation in patients with PD receiving SCS for intractable pain. Subsequent open-label studies comprise patients with PD with postural instability and gait disorders (PIGD) who received SCS implantation at either cervical or thoracic levels. Stimulation protocols were varied (tonic vs burst stimulation, high vs low frequency, long-vs short-pulse width). We focused our discussion on studies designed to assess an improvement in gait dysfunction. A detailed breakdown of patients with PD who had a concomitant improvement in gait dysfunction after having received an SCS implant for pain is presented in Table 1. In Tables 2  and 3, a detailed breakdown is presented of SCS settings, waveforms, manufacturer, clinical scales, and laboratory measures used in studies on gait dysfunction in patients with PD.

Cervical SCS
After the findings of a beneficial effect of SCS on gait and motor function in rodents, 36 Thevathasan et al investigated the effects of high cervical SCS in two patients with PD, one of whom had a severe gait problem. No effects on UPDRS-III, the hand-arm movement test, time foot-tapping score, and time to walk up 10 meters test were observed. 22 Thoracic SCS In 2017, Pinto de Souza et al investigated the safety and efficacy of SCS in four patients with PD with significant gait dysfunction and who were previously treated with STN-DBS. 23              number of steps in the 20-meter walking test, in timed up and go (TUG) test, and in the overall stride length. These patients also showed decreased anticipatory postural adjustments (APA) and FoG duration. 24 This is the first open-label study that showed a possible beneficial effect of SCS on gait and balance in patients with PD not affected by a pain disorder.
Subsequently, in an open-label, nonrandomized study, Samotus et al observed an improvement in gait, balance, and motor symptoms in five patients with PD with PIGD. 11 In particular, UPDRS-III showed a significant improvement of bradykinesia, whereas gait analysis highlighted improvements in stride velocity, step length, spatiotemporal asymmetry, and FoG episodes. Improvements were sustained for the entire six-month period. 11 Four of five patients showed a sustained improvement over a three-year observational period. A long-lasting reduction in the number of FoG episodes was observed, whereas the remaining gait parameters showed an overall stability compared with baseline. 27 SCS parameters reveal positive effects for long pulse widths and lower frequencies, in contrast to a previous study in which a combination of high frequency and short pulse width was shown as effective.
Hubsch et al reported short-and long-term improvement in the stand-walk-sit test, UPDRS-III, Freezing of Gait Questionnaire (FOGQ), and the Parkinson's Disease Questionnaire (PDQ-39) after SCS, with three of five patients who retained the initial SCS parameters at two years and with all patients satisfied with the effects of the stimulation. 25 In an open-label pilot study, Prasad et al failed to show any improvement in UPDRS-III and FOGQ in the acute phase and after a one-year period of long-term stimulation. 26 Compared with previous studies, the combination high frequency/low pulse width or low frequency/long pulse width was set up for long-term stimulation in only one patient, whereas all the other patients were stimulated with a low frequency/short pulse width combination.
Finally, Zhou and Bao described one patient with PD with FoG who underwent thoracic tonic SCS 30 and had a dramatic improvement in the lower limb rigidity and tremor and in bradykinesia as assessed by the UPDRS-III. Improvement in gait speed, balance, and FoG also were noted.

Atypical Parkinsonism
No studies have been designed to assess the efficacy of SCS on gait and other motor symptoms in animal models of atypical parkinsonism.
In humans, eight studies have been published so far, including a total of 28 patients. Three articles described the effect of SCS in six patients affected by a tauopathy, 28,29,31 whereas four articles described eight patients with MSA. [32][33][34][35] In addition, 14 patients reported by Mazzone et al in 2019 were affected by an atypical parkinsonism or a vascular parkinsonism, whereas only four were diagnosed with idiopathic PD. 19 However, the high heterogeneity between the two groups (age, disease duration, levodopa baseline dosage, and SCS parameters and waveforms) made it difficult to ascertain any possible differential effect of SCS in idiopathic PD versus atypical parkinsonism, and between different stimulation modalities. 19

SCS in Patients With Tauopathies
Rohani et al described two patients with primary progressive FoG who showed a progressive and long-lasting improvement in FoG after surgery. 31 No details about the SCS parameters were reported. Samotus et al recently reported a positive effect on spatiotemporal gait measures, FoG episodes, and motor symptoms in three patients affected by PSP. 28 Using a combination of low frequency/high pulse width, they showed an improvement in step length, stride velocity, swing time, and single support time, and a reduction in mean stride width, stance, and double support gait phases. UPDRS-III improved in two patients and worsened in one. Finally, the effect on duration of 360 • turning was inconsistent between subjects, but an improvement of FoG frequency and duration was noticed. Improvement in spatiotemporal gait measures, FoG episodes, and UPDRS-III also was observed in one of two patients with cortico-basal syndrome and FoG. 29

SCS in Patients With MSA
Zhang et al reported a patient affected by MSA with parkinsonian features, who had an improvement in FoG at six months from surgery. 32 Interestingly, this patient showed an improvement in their 18F-fluorodeoxyglucose positron emission tomography frontal-parietal hypometabolism six months after SCS. 32 More recently, Wang et al 33 described five patients affected by MSA treated with cervical tonic SCS aiming at improving speech and gait parameters. Gait performance was measured using a TUG test, and stride length also was obtained. Speech was assessed using a dedicated device, and speech subsystems of articulation, phonation, prosody, resonance, and respiration were obtained. A multiparametric dysphonia severity index also was calculated. Finally, the UPDRS-III and the Parkinson's Disease Questionnaire 8 (PDQ-8) were obtained. All the outcome measures were collected at baseline and at three and six months. Despite no effects on gait and motor symptoms being observed, an improvement in speech intelligence as measured by speech item of the UPDRS-III and voice quality was described. 33 Conversely, Li et al 34 described a patient with MSA-parkinsonian type (MSA-P) who underwent a concomitant STN-DBS and thoracic SCS to treat motor symptoms. UPDRS-III, Unified Multiple Systems Atrophy Rating Scale Part I, II, IV, NFOGQ, Gait and Fall questionnaire, PDQ-39, and TUG were assessed at baseline and at four and eight months postoperatively. The combined stimulation consistently led to a better improvement than did either DBS or SCS stimulation alone for all the clinical outcomes. 34 Despite dysautonomia being a cardinal feature of MSA, none of the seven patients reported in the previous studies was investigated for possible effects of SCS on dysautonomia. Given the degeneration of catecholaminergic neurons in the rostral ventrolateral medulla and of sympathetic preganglionic neurons in the thoracic spinal cord is responsible for dysautonomia in patients with MSA, thoracic SCS may be considered to target those symptoms. To this extent, Squair et al 35 treated with thoracic SCS a patient with MSA and severe postural hypotension. Thoracic spinal cord and dorsal root ganglia exit from T9 to T12 were targeted because this was previously shown to be a highly autonomic reactive area in the spinal cord. A significant reduction in the number and severity of postural hypotension episodes led to a dramatic improvement in syncopal episodes and total amount of gait distance. Although the patient underwent both autonomic rehabilitation and SCS, gait was improved only when the SCS was active, suggesting a critical role for SCS in mediating the clinical improvement. 35

DISCUSSION
Clinical Effects SCS appears to be effective in improving locomotor activities and in alleviating some motor symptoms in different animal models of PD. The most effective SCS paradigm seems to be a high-frequency (300 Hz) stimulation delivered by high thoracic (T2/T3) SCS. 36,37,39 However, low frequencies have been effective in improving spontaneous locomotion and duration of locomotion periods. 37,40,41 Most of the animal models have been stimulated for a few seconds/minutes daily or weekly, whereas few data are available on continuous long-term stimulation. 42 Limited data are available on new stimulation paradigms, such as burst stimulation and highfrequency stimulation, in both animal models and human studies, although some preliminary reports in patients with PD did not show any difference between tonic and burst stimulation as assessed by the UPDRS-III. 19 Human studies have reported an improvement in the gait of most patients, regardless of the absence or presence of pain, though the overall effects are heterogeneous. The effects seem to be more consistent for low thoracic SCS (T8-10). A total of 66 patients with PD have been described, and 23 of them had SCS surgery for PIGD without pain. Of these 23 patients, 15 showed an improvement in their gait and in some cardinal features of PD, such as bradykinesia. Four patients have been followed up for three years, and results are reported. 27 Only eight patients with PD did not show any improvement in gait and motor symptoms. 22,26 A positive effect on pain and motor features has been described in all the 43 patients with PD with pain treated with SCS, though the improvement in pain could confound the interpretation of gait improvement.
So far, researchers have explored different combinations of frequency/pulse width, but the two most used combinations are medium frequency/short pulse width (such as 300 Hz/90 ms) and low frequency/long pulse width (such as ≤130 Hz/≥240 ms). No differences in stimulation intensity can be observed, with values set according to the paresthesia threshold. Interestingly, only two of eight patients who failed to show any clinically relevant effect were treated with a frequency/pulse width combination of high frequency/low pulse width or low frequency/long pulse width; the remaining six patients were stimulated with a low frequency/short pulse width combination or with a high frequency/long pulse width. 22,26 Clinical data on burst SCS are extremely preliminary, but results seem promising. 18 Interestingly, SCS is emerging both as a salvage therapy for patients with DBS experiencing DBS-refractory axial symptoms 10,15,17,23 and even potentially as a first choice treatment for patients with PD with mainly axial and gait disorders. 11 A possible effect of SCS on axial symptoms would be desirable, given axial symptoms such as camptocormia are usually less responsive to the classic dopaminergic replacement treatment than are tremor and bradykinesia.
In atypical parkinsonism, SCS has shown promising effects in patients with tauopathies, with sustained improvements specifically in FoG, together with a broader improvement in other motor symptoms as assessed by the UPDRS-III. In patients with MSA, the effect of cervical and thoracic SCS has been described on autonomic dysfunctions in two patients, whereas an improvement in speech has been reported in five patients with cervical SCS. An overall effect on gait and motor function has been described in only one patient who also has received STN-DBS therapy. A total of six patients were labeled as having "atypical parkinsonism" but without further characterization, and therefore, we were not able to infer any specific information on them. 19 Finally, eight patients with vascular parkinsonism have been described as having an acute and sustained improvement in motor features, assessed by the UPDRS-III, and in cadence, step, and stride length, assessed by laboratory gait analysis. 19 Outcome Measures Parkinson's Disease We analyzed the outcome measures used in the double-blind, cross-over, randomized studies of SCS in patients with PD with PIGD and without pain. The most used outcome measure was the UPDRS-III, which has been assessed in all studies, both as a total score and for subitems analysis, with particular interest in the postural and instability items. Clinical effects on FoG have been investigated using the FOGQ and the NFOGQ, but this has only been assessed in six of nine studies. 11,[23][24][25][26][27] Quality of life has been assessed by means of PDQ-8 in two studies, 11,27 whereas PDQ-39 was used in four studies. [23][24][25][26] Clinical assessment of balance has been investigated in five studies, using the Berg Balance scale, 23,24 the activities-specific balance confidence (ABC) scale, 11,27 and the Tinetti balance scale. 26 No studies have assessed the impact of SCS on nonmotor symptoms. Number of falls has never been used as an outcome measure.
Gait was assessed clinically using the 10-meter walking test in one study, 22 the 20-meter walking test with and without dual task in two studies, 23,24 and TUG with and without dual tasking in two studies. 23,24 The TUG also has been used in different case reports/ case series. However, it has been assessed on different distances, and this renders any comparison difficult. Laboratory assessment of gait was performed in six studies, and gait analyses encompass step length, stride width, stride velocity, step time, stance and swing time, gait speed, cadence, and single/double support time. 11,19,23,24,26,27 None of the reported studies on patients with PD assessed changes in static postural control by means of sway recording. Laboratory tests of dynamic balance and postural control were performed in only one study, focusing on time and amplitude of APA. 24

Atypical Parkinsonism
There was less heterogeneity in clinical outcome measures in studies of patients with a tauopathy and SCS, given five of seven cases have been described by the same group. UPDRS-III, FOGQ, and ABC have been used to assess patients at baseline and at three, six, and 12 months after surgery. 28,29 Laboratory assessment of gait was performed in six patients, and gait analysis encompasses step length, stride width, stride velocity, step time, stance and swing time, gait speed, cadence, and single/double support time. 28,29,31 Number of freezing episodes and duration of FoG also were assessed in five patients. 28,29 Patients with MSA have been assessed with the UPDRS-III regarding their motor symptoms. FOGQ and NFOGQ were used in two cases, 32,34 whereas balance was clinically assessed by means of the Gait and Fall questionnaire in two cases. 32,34 Speech was formally assessed in only one study. 33 Dysautonomia has been assessed clinically and with laboratory measures in only one subject. The Unified Multiple System Atrophy Rating scale, specifically developed by the Movement Disorder Society for patients with MSA, has been used only for one patient, and the subscale on dysautonomia has not been collected. 34

Mechanisms of Action
In pain, SCS has been hypothesized to activate the dorsal column, leading to the modulation of several cortical structures, such as somatosensory cortex, prefrontal cortex, cingulate cortex, and thalamus. 9 In addition, changes in local neurotransmitters release, direct modulation of spinal circuitry, and axon regeneration properties have been observed in animal and human studies with spinal cord injury and pain. [44][45][46] Similar modulation of cortical regions is likely in patients with PD; however, the mechanisms of action of SCS in patients with PD with gait problems remain unclear.
Most of the SCS studies in human PD so far focused on the possible clinical benefits for gait and balance, rather than mechanisms of action. It has been widely accepted that neurophysiological changes in brain and spinal cord activity observed during SCS in animal models, including nonhuman primates, can be observed in humans as well. 47 However, results obtained in animals are not necessarily transferable to humans, owing to various factors (different stimulation parameters, waveform and duration, electrode positions and size, and comparative anatomy). 39 Despite similar patterns of neural activity being thought to drive locomotion in humans and quadruped animals, there is evidence showing that a functional reorganization of the supraspinal (cortical and subcortical) and spinal motor systems may be needed to allow balance and gait in bipeds. Therefore, more mechanistic studies are required to assess ways SCS modulates the postural and gait network. [47][48][49] Interestingly, animal models and patients with PD with SCS showed a generalized clinical improvement that exceeds the pure effects on gait and balance. In particular, bradykinesia has shown the greatest improvement in animal models. 36,37 The mechanisms underlying this broad clinical effect of SCS are not clear, and a univocal physiological mechanism that can justify such a generalized improvement is unlikely.
Data from animal studies support the hypothesis that SCS could exert its activities locally on spinal circuitry and remotely on oscillatory brain activities. According to the selected stimulation parameters and waveforms, SCS could act on different structures of the locomotor functional network, 50 causing a combined modulatory effect on local spinal cord circuitry and on ascending projections to cortical structures, finally indirectly modulating specific cortico-subcortical motor circuitry. 40 Animal models support this synergistic mechanism of action: Fuentes et al first suggested that SCS improves locomotion in a short-term and long-term PD animal model by activating the dorsal column-medial lemniscal pathway, and ultimately modulating low-frequency oscillations in basal ganglia and cerebral cortex. 36 Interestingly, the activation of the ascending pathways may not only desynchronize abnormal cortico-striatal oscillations, 37 but it also may indirectly modulate the supplementary motor area and the cholinergic PPN through the activation of extralemniscal pathways to the brainstem and the thalamus. 36,37 This hypothesis also has been supported by results from a posturography study by De Lima-Pardini, in which defective APA are modulated by SCS. 24 Defective APA cause difficulty in gait initiation, and circuits involving supplementary motor area are thought to be concerned. 51 Because no effects from SCS are observed on reactive posture control, these results, together with the effects on oscillatory brain activity, suggest that SCS might act on circuitry that requires a high cortical influence. 24 In contrast, Zhong et al hypothesized, according to EMG data, that SCS could increase the ability of spinal circuitry in processing the proprioceptive input specific to a given task, in addition to having a supraspinal effect on the sensorimotor regions of both basal ganglia and cerebral cortex. 40 Such an effect on cortical areas seems confirmed by what Zhang et al observed in their patient affected by MSA. 32 After the publication of negative findings of SCS for FoG in one study, 26 a possible role for habituation (a process that leads to decreased responsiveness to a stimulus) in explaining the failure to improve FoG has been raised. 52 This was first observed by Mazzone et al, 19 who reported that a mean of 17.6 modifications of SCS parameters were required every three months to sustain the clinical effects in six patients (three with idiopathic PD and three with atypical parkinsonism) receiving tonic stimulation. Such an effect was not present in the 15 patients treated with burst stimulation. 19 Similarly, Cury et al described a patient, previously on continuous tonic SCS, who improved under a "cycling" tonic stimulation pattern, in which the SCS was turned ON and OFF every 15 minutes. 52 However, no more data are available in the literature to substantiate this mechanism.

Limitations
The trials performed so far have left several unanswered questions about SCS that must be addressed before this procedure can be used more widely in PD. Firstly, almost all the published studies are unblinded and carried out in small cohorts of patients with PD. This is partially due to some limitations in blinding procedures, caused by the paresthesia sensation evoked by SCS. New SCS waveform will allow us to design fully blinded studies in a larger cohort of patients. Secondly, although these studies have shown that overall, SCS seems to have a beneficial effect on gait in PD, they also have shown a heterogeneous outcome because some patients had a poor response to treatment in the short term, or a lack of consistency of the effect over few months. Identifying potential good responders to SCS before the surgery would be an extremely useful advance in this field, and trials in the potential use of transspinal magnetic stimulation are ongoing. Thirdly, patient selection and gait characterization in these studies were limited, and this lack of clinical phenotyping could have been responsible for the heterogeneous outcome of these studies. Clear inclusion criteria and more homogeneous outcomes are needed, especially to evaluate SCS effects in a real-world scenario, where gait disorders can cause severe impairment and life-threatening events such as falls. Fourthly, mechanisms of actions of SCS have never been assessed in these studies, and only one article investigated the effect of SCS on APA to explain the improvement in FoG. This aspect is relevant when designing new controlled trials, given each SCS waveform may lead to a different spinal and supraspinal effect, which may lead to an improvement in some clinical symptoms. In addition, a better knowledge of the ways SCS exerts its effect not only may contribute to improve patient selection and clinical SPINAL CORD STIMULATION IN PARKINSONISM www.neuromodulationjournal.org outcome but also may lead to better knowledge of gait disorders in humans.

CONCLUSIONS
SCS is emerging as a possible treatment for gait disorders in PD. Results in animal models are promising, whereas evidence in humans is limited and inconsistent thus far. More mechanistic studies, followed by prospective double-blind clinical trials in a large cohort of patients who are well characterized, are required to further elucidate its efficacy.