Transcranial Direct Current Stimulation for Parkinson's Disease: A Systematic Review and Meta-Analysis

Background: Parkinson's disease is a common neurodegenerative disorder with motor and non-motor symptoms. Recently, as adjuvant therapy, transcranial direct current stimulation (tDCS) has been shown to improve the motor and non-motor function of patients with Parkinson's disease (PD). This systematic review aimed to evaluate the existing evidence for the efficacy of tDCS for PD. We included English databases (PubMed, the Cochrane Library, Embase, and Web of Science) and Chinese databases [Wanfang database, China National Knowledge Infrastructure (CNKI), China Science and Technology Journal Database (VIP), and China Biology Medicine (CBM)] without restricting the year of publication. Twenty-one tDCS studies, with a total of 736 participants, were included in the analysis. Two independent researchers extracted the data and characteristics of each study. There was a significant pooled effect size (−1.29; 95% CI = −1.60, −0.98; p < 0.00001; I2 = 0%) in the Unified PD Rating Scale (UPDRS) I and the Montreal cognitive assessment (SMD = 0.87, 95% CI = 0.50 to 1.24; p < 0.00001; I2 = 0%). The poor effect size was observed in the UPDRS III scores (SMD = −0.13; 95% CI = −0.64, 0.38; p = 0.61; I2 = 77%), and similar results were observed for the timed up and go (TUG) test, Berg balance scale, and gait assessment. The results of this meta-analysis showed that there was insufficient evidence that tDCS improves the motor function of patients with PD. However, tDCS seemed to improve their cognitive performance. Further multicenter research with a larger sample size is needed. In addition, future research should focus on determining the tDCS parameters that are most beneficial to the functional recovery of patients with PD.


INTRODUCTION
Parkinson's disease is a common neurodegenerative disease in the elderly and its characteristic pathological changes are the progressive degeneration of the dopaminergic neurons in the substantia nigra and the significant decrease in the dopamine secretion of the striatum (Berg et al., 2014;Beretta et al., 2020a). Parkinson's disease (PD) is uncommon before 50 years of age in men and women, and the prevalence, morbidity, and death rates associated with this condition increase with age. In addition, men are more susceptible to develop this disease than women, and the incidence is 1.4 times higher than that in women. By the age of 60, the prevalence rate is ∼0.5%, while the prevalence rises to ∼3.9% in elderly individuals between 85 and 89 years of age (Dorsey et al., 2018). In China, the prevalence of PD among people over 65 years of age is ∼1.7% (Zhang et al., 2005). The cardinal symptoms of PD include motor-and non-motor-related features. The motor symptoms include bradykinesia, rigidity, and static tremor, as well as postural and gait disorder (Jankovic, 2008). The common disabled non-motor symptoms in patients with PD mainly include emotional and cognitive disorders (Ransmayr, 2015). These symptoms can lead to dysfunctions such as balance disorders, cognitive impairment, and dysphagia, which reduce the ability for self-care in daily life and may even lead to death, increasing the economic burden on the family and society.
The treatment of dysfunctions in patients with PD requires comprehensive therapy and multidisciplinary participation, including movement therapy, dopamine replacement therapy, and the combined use of anticholinergic agents and deep brain stimulation (Goodwill et al., 2017). However, the effects of the medication may diminish over time (Jankovic and Stacy, 2007). These may include movement symptoms and fluctuations (Jankovic, 2008) and obsessive behaviors (Raja and Bentivoglio, 2012), as well as an increased risk of developing dementia (Gray et al., 2015). Adaptive deep brain stimulation has shown great potential in the treatment of PD, but its applicability in cognitive and other psychiatric disorders remains uncertain (Beudel and Brown, 2016;Guidetti et al., 2021). Therefore, alternative interventions should be explored.
In recent years, a growing number of researchers have paid more and more attention to the study of non-invasive brain stimulation techniques on the function of PD, such as transcranial direct current stimulation (tDCS). tDCS has two electrodes, an anode, and a cathode, which provide constant direct currents on the scalp. It has been shown to induce changes in the resting membrane potential of the cerebral cortex and change the excitability of neurons (Nonnekes et al., 2014). The anode increases the excitability of the cortical tissues, and the cathode decreases the excitability (Broeder et al., 2015;Lefaucheur et al., 2017), which may induce the release of neurotransmitters and increase the extracellular dopamine levels, as has been demonstrated in animal models (Tanaka et al., 2013). This may facilitate signal transduction in brain tissue. In addition, studies have shown that tDCS on the cognitive regions of the cerebral cortex could improve cortical excitability, and affect cognitive networks (Miniussi et al., 2013). tDCS has been suggested to improve cognitive ability (Boggio et al., 2006;Doruk et al., 2014;Biundo et al., 2015) and verbal fluency (Pereira et al., 2013), and cerebellar tDCS can activate specific neural networks and strengthen the regulation of behavioral responses associated with emotion-related stimuli (Ruggiero et al., 2021). However, studies have shown that between tDCS and sham intervention, there was no difference in the movement performance, reaction time, and self-assessment mobility in the Unified PD Rating Scale (UPDRS; Benninger et al., 2010).
Although tDCS shows great potential in treating PD, the results of the existing studies on the treatment of PD with tDCS were inconsistent, so it is necessary to systematically review the existing studies. This systematic review and meta-analysis aimed to summarize the available evidence to evaluate the clinical efficacy of tDCS in the treatment of PD.

METHODS
The systematic review and meta-analysis were conducted according to the preferred report items of systematic review and meta-analysis (Page et al., 2021).

Study Types
Only relevant randomized controlled trials (RCTs) were included to investigate the efficacy of tDCS in PD treatment. Comments, case reports, quasi-RCTs, animal experiments, or non-RCTs were excluded.

Participants
According to diagnostic criteria (National Collaborating Centre for Chronic, 2006), participants diagnosed with PD were included in this review. There were no restrictions on age, gender, or race.

Intervention
The studies included tDCS intervention alone or in combination with any other interventions, including sham stimulation, Western medical treatment, or rehabilitation. Except for the tDCS intervention in the experimental trial, the interventions in the experimental and comparison trials are the same.

Outcomes
The primary outcomes were the motor and non-motor function assessments, including the UPDRS, timed up and go test (TUG), Berg balance scale (BBS), gait assessment, mini-mental state examination (MMSE), Montreal cognitive assessment (MoCA), PD-cognitive rating scale (PD-CRS), and PD Quality of Life Questionnaire-39 (PDQ-39).

Search Strategy
The search was performed by restricting the language to English and Chinese without restricting the year of publication. Two of the authors (Xiang Liu and Huiyu Liu) created a search strategy, which was approved by all authors. To identify all potentially relevant studies, we searched the following databases: English databases: PubMed, the Cochrane Library, Embase, and Web of Science. Chinese databases: Wanfang database, China National Knowledge Infrastructure (CNKI), China Science and Technology Journal Database (VIP), and China Biology Medicine (CBM). All searches were conducted in June 2021 and covered the databases from their inception. The search terms were (a) "Parkinson's" or "Parkinson's disease" or "PD" or "idiopathic Parkinson's disease" or "IPD" or "primary Parkinson's syndrome"; (b) "transcranial direct current stimulation" or "tDCS"; and (c) "randomized controlled trials" or "RCTs" or "controlled clinical trial" or "randomized" or "randomly" or "trial."

Study Selection
The researchers (Xiang Liu and Huiyu Liu) scanned the title, abstract, and keywords of the articles found in the electronic search and excluded the irrelevant articles. Then, we obtained the full text of the included articles; subsequently, the full text of the potential studies was evaluated to determine their acceptability. All the differences and opinions were resolved through a discussion between the two researchers, and the research was selected according to the inclusion criteria. A third reviewer (Wang Pu or Youliang Wen) was consulted to resolve differences in data extraction.

Data Extraction
We prepared a data extraction table that included the authors, publication date, and the characteristics (numbers, gender, age, and PD durations, and others), experimental group, control group, stimulation area, intensity, duration, frequency, and period of treatment, outcomes, and adverse events. It was defined that, in the normal use of qualified tDCS, any deleterious event occurring that results in human harm and is not related to the anticipated effects of tDCS.

Assessment of the Risk of Bias in the Included Studies
The Cochrane bias risk assessment tool (Higgins et al., 2011) was used to assess the methodological quality. Two reviewers (Xiang Liu and Huiyu Liu) independently assessed the risk of bias of each included study according to the following characteristics: random sequence generation, allocation concealment, blinding, completeness of outcome data, selective outcome reporting, and other biases. The risks were classified into three levels: low risk of bias, unclear risk of bias, and high risk of bias. Any difference was resolved through discussions, and if consensus was not reached, the third reviewer (Pu Wang or Youliang Wen) was consulted.

Statistical Analyses
The review Manage 5.3 software (Cochrane, London, United Kingdom) was used to conduct the statistical analyses of the overall and subgroup treatment effects of tDCS intervention in PD. For continuous data, the mean difference (MD) or standardized mean difference (SMD) was used for analysis. The dichotomous data analysis was calculated using the risk ratio (RR), and for both, 95% CIs was assessed using the Z-test.   The heterogeneity between each group was tested using the I² test and Cochran's Q statistic (Zintzaras and Ioannidis, 2005). Less than 25% of the I² values indicate low heterogeneity, 25-50% indicate moderate heterogeneity, and >50% indicate high heterogeneity. A random-effects model was applied if high heterogeneity was observed (p < 0.05 or I² > 50%); otherwise, a fixed-effects model was applied (p > 0.05 or I² < 50%). When significant heterogeneity was present, subgroup analysis and sensitivity analysis were performed on the data, and if the source of heterogeneity could not be determined, descriptive analysis was performed. Figure 1 shows the flow diagram for the selection of the included studies.

Study Classification
We searched out a total of 357 records from the relevant databases. First, 217 duplicated records were excluded, and 140 remained. Then, we read the articles carefully. In addition, 119 articles were excluded for the following reasons: review or metaanalysis (n = 20), not related to tDCS (n = 15), related to other disease (n = 26), not RCTs (n = 16), not a full-text article (n = 5), animal experiment (n = 9), and no original data (n = 28). Finally, 21 studies were eventually included in this review.

Characteristics of the Included Studies
The participants and the methodological characteristics of the included studies are shown in Tables 1, 2. A total of 21 articles (all in all, 736 patients with PD) were included in this systematic review, of which 66.7% were by foreign authors and the remaining 33.3% were reported in a Chinese database. Twelve out of the 21 studies reported the drug status of participants, while the remaining nine did not mention the medication status.  Among the 21 included studies, all received active tDCS or sham tDCS alone, or in combination with other treatments. The stimulation area of the brain includes the dorsolateral prefrontal cortex (DLPFC), the primary motor cortex (M1), prefrontal cortex (PFC), premotor cortex (PMC), and central zero (Cz) position. Among these areas, the DLPFC was the most frequently stimulated site, with 13 studies out of 21, followed by M1 (four studies), PFC, PMC, and Cz position. Finally, 17 studies applied repeated sessions of tDCS protocols, three studies used a single session of tDCS, and one study did not report.

Methodological Quality
The details of the risk of bias of all the included studies are shown in Figures 2, 3. All the articles used the random method to generate sequences. In the allocation concealment, only one study was high risk and nine studies were low risk. For the blinding of the outcome assessment, three retrieved studies were high risk and nine were unclear. During the evaluation process, we found that two articles mentioned patients who dropped out but were not included in the analysis, which may increase the risk of bias. In addition, all the studies clearly described selective reporting.

Time Up and go Test (TUG) and Berg Balance Scale (BBS)
The TUG test was used as an evaluation standard. Nine studies employed the TUG test. The meta-analysis showed an insignificant pooled effect size (−0.12; 95% CI = −0.43, 0.19; p = 0.46; I² = 0%) (Figure 7). Three studies used the BBS. The result was non-significant as that of the TUG, and the pooled effect size was SMD = −0.03; 95% CI = −0.45 to 0.39; p = 0.88) (Figure 7).

Adverse Events
Adverse events were reported in five studies. Yang and He (2017) reported that both the experimental group and the control group experienced adverse events, including insomnia, dizziness, postural hypotension, and constipation, with an incidence rate of 20.21%. In Criminger et al. (2018), one patient developed a headache. In Schabrun et al. (2016), one participant experienced a strong tingling and a momentary flash of light over the area of one electrode, and the sensations lasted for about 5 s. Yotnuengnit et al. (2018) mentioned that during the intervention period, two participants experienced a burning sensation on their forehead skin. In Benninger et al. (2010), one subject had a small number of first-degree burns. However, none of these adverse events resulted in serious consequences in any of the included studies.

DISCUSSION
The present systematic review of clinical studies of tDCS in PD aimed to evaluate the efficacy of tDCS as a clinical therapy for PD based on the existing evidence. In this review, the high heterogeneity of stimulation parameters does not allow us to firmly conclude that tDCS improves cognitive performance. In addition, it is uncertain whether tDCS can improve motor function in patients with PD. Further studies with larger sample sizes are needed to explore this possibility more thoroughly. Below, we discuss the effect of tDCS on motor and nonmotor functions.
The 21 studies included were consistent in terms of the tDCS areas. In all studies, electrodes were placed to target the brain regions associated with motor or cognitive function. The studies aimed at improving cognition in the DLPFC or PFC, and the studies aimed to improve motor function in patients with PD employed electrodes at the M1 or premotor cortex. Fourteen studies used the DLPFC as the stimulation area. The DLPFC is considered to be the central region of executive functions in humans, and patients with damage to this region may show cognitive difficulties in organizing behavioral responses, extracting memory, and generating motor programs. The metaanalysis showed that tDCS could have a significant therapeutic effect on cognitive performance. Although the sample size was relatively small, the results of this meta-analysis extend the previous findings that the tDCS protocol may improve cognitive function in PD (Doruk et al., 2014;Biundo et al., 2015). This may be related to the cortical excitability enhanced by anode tDCS (Broeder et al., 2015). Transcranial direct current stimulation, as a non-invasive method of brain stimulation, can promote cerebral cortical blood flow (Cosmo et al., 2015). Polardependent shifts are capable of generating resting membrane potentials at the neuronal level (Broeder et al., 2015). In relation to this, the stimulation of the cortex by tDCS may promote the neural connectivity between the cortical and subcortical network, and improve neuroplasticity in patients with PD (Bindman et al., 1964;Nitsche and Paulus, 2000). Can anode tDCS improve motor function in PD? The results of our meta-analysis showed that tDCS did not significantly improve UPDRS III, TUG, gait, or balance. We attempted to determine the reason for this from the tDCS parameters. For the effect of tDCS on the UPDRS III of PD, we performed a subgroup analysis of the stimulation duration, intensity, and session, and the same results were observed. We also compared the stimulation area of tDCS, the number of subjects included, the severity of the disease, and the status of drug treatment between the studies, and found large differences. Bueno et al. (2019) showed that a single DLPFC stimulation was insufficient to promote gait changes in PD patients, and they argued that the assessment tools used in the study do not represent the gold standard and may not be sensitive enough to detect changes. Therefore, we believe that many factors contribute to the tDCS induced improvement of motor functions in patients with PD. However, some studies have demonstrated the effectiveness of tDCS in improving the motor function of patients with PD. In Qiao and Yan (2018), the anode stimulation of the left DLPFC in terms of the step width resulted in a significant difference between tDCS and sham intervention, suggesting that tDCS intervention can improve bradykinesia in patients with early PD to a certain extent. In previous studies (Schabrun et al., 2016), the anode stimulation of tDCS was found to improve the cadence of PD. Wu and Wu (2016) and Yang and He (2017) also showed that the stimulation of the left DLPFC with anode tDCS can improve the scores of the UPDRS III in PD. In one case (Kaski et al., 2014a) report of a patient with moderate PD, while dancing the tango, the peak velocity of the trunk was significantly greater than the sham stimulation upon stimulating anode tDCS the primary and premotor cortex. This may be because tDCS intervention triggers the motor and prefrontal cortex regions, which may lead to the release of dopamine in patients with PD and promote the improvement of motor functions (Voon et al., 2009;Manenti et al., 2016;Lee et al., 2019). Another possible mechanism is that cerebellar tDCS improves sensorimotor functions through either cerebellobrain inhibition or dentate disinhibition (Ferrucci et al., 2016;Kimpel et al., 2020). Hence, there was insufficient evidence to demonstrate the effectiveness of tDCS in improving the motor functions of patients with PD. More sensitive tests, such as a three-dimensional gait analysis system or electromyography, are needed to assess motor functions in the future. This is consistent with the conclusions drawn from a review by Nardone et al. (2020). Future research should identify the optimal stimulation targets for motor function in patients with PD.
In this review, 17 studies used 2 mA stimulus intensity, and none compared the effects of different tDCS intensities on PD. Why choose 2 mA stimulation intensity? Possibly because some studies have shown that 2 mA stimulation increases cortical excitability more than 1 mA stimulation (Shekhawat et al., 2013;Murray et al., 2015). In Beretta et al. (2020b), the anode tDCS stimulation of M1 improved the ability of patients with PD to respond to external interference, and 2 mA showed more improvement than 1 mA. However, other studies have found that, for anode tDCS, regardless of the current intensity of 1 or 2 mA, the amplitudes of the motor-evoked potentials did not change significantly (Tremblay et al., 2016;Jamil et al., 2017). In addition, in our meta-analysis, the results suggested that the intensity of stimulation does not affect UPDRS III; nevertheless, high heterogeneity was noted. We conducted a subgroup analysis, which showed that a current stimulation of <2 mA had a more positive effect than the stimulation of 2 mA on the UPDRS III, which is inconsistent with the clinical observations. This may be due to the small sample size and the higher risk of bias in the included articles. Therefore, further research is needed to understand whether the intensity of tDCS is an important factor affecting the results. None of the included studies used stimulation of more than 2 mA. However, in Agboada et al. (2019), an increase in excitability was observed from the lower to higher current intensities (1 vs. 3 mA). Therefore, further studies on the tolerance and efficacy of this method should be conducted in the future.
The stimulation duration of the tDCS protocol differed among the included studies from 7 to 30 min. This large difference may explain the obvious differences in effects. Regarding the stimulation duration, although most studies used tDCS for 20 min, positive results for shorter stimulation durations have also been reported. Kaski et al. (2014b) found that placing the anode tDCS over the primary motor cortex for 15 min combined with physical activity had significant benefits for gait speed and balance. Nitsche and Paulus (2001) reported that 5-13 min of anode tDCS on the motor cortex leads to increased motorevoked potentials. Another study reported that increasing the tDCS duration would be necessary to increase the magnitude or duration of plasticity (Tremblay et al., 2016). Agboada et al. (2019) reported that there was no significant correlation between therapeutic effect and stimulation durations. In this review, the association between stimulation duration and efficacy in patients with PD was not determined; therefore, it is necessary to conduct a more comprehensive evaluation of the effect of the duration of tDCS on PD treatment.
In most studies in this review, repeated sessions were conducted, but some studies have shown that a single tDCS on the left DLPFC in patients with PD can improve balance and functional activity (Lattari et al., 2017). A meta-analysis showed that compared with mono-target stimulations, multitarget stimulation showed significant improvements in mobility, balance, gait velocity, and fall reduction (Orrù et al., 2019). Whether single or repetitive stimulation is more effective is still uncertain, and the effect of tDCS sessions on the treatment of PD needs to be further evaluated. Compared with the therapeutic effects of tDCS on single stimulation, multiple stimulations in PD functional rehabilitation will likely become a hot research topic in the future. The PDQ-39 and UPDRS II were used to assess the quality of life of the subjects. In Manenti et al. (2018), no significant difference in the PDQ-39 scores was observed between the experimental and control groups. This may be because daily activities are complex motor manifestations that require a combination of motor and cognitive abilities, as well as other functional capabilities.
In the analysis of the included studies, several limitations affect the results. First, the quality of the studies was relatively moderate. Second, the sample size was small, making it difficult to generalize the results. Third, there was high heterogeneity in the research design of the included articles and tDCS protocol, which makes it difficult to determine the most suitable protocol for improving the clinical symptoms of PD. Finally, only Chinese and English literature were included, and the risk of article selection may affect the comprehensiveness of the results. Therefore, in the future, more multi-level and multi-center authoritative controlled studies with large samples are needed. The determination and unification of the tDCS treatment parameters, the establishment of the sham stimulation group, and determination of the curative effect and treatment mechanism, all need to be explored in more detail, and treatment standards and norms should be issued to guide clinical practice.

CONCLUSION
The results showed that tDCS appears to improve cognitive performance, but there is insufficient evidence to demonstrate that tDCS is effective in the treatment of the motor function of patients with PD. Further multicenter research with larger sample sizes is needed. Future research should focus on determining the tDCS parameters that are most beneficial to the functional recovery of patients with PD.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author/s.

AUTHOR CONTRIBUTIONS
PW, XG, and YW: conceptualization and writingreview and editing. XL, HL, ZL, JR, JW, PW, XG, and YW: data curation. XL, HL, ZL, JR, JW, PW, and YW: formal analysis, methodology, and writing-original draft. PW and XG: investigation. HL, PW, and XG: project administration. XL, HL, ZL, and JR: software. All authors contributed to the article and approved the submitted version.