Transcranial direct current stimulation combined with physical or cognitive training in people with Parkinson’s disease: a systematic review

Background Pharmacologic therapy is the primary treatment used to manage Parkinson’s disease (PD) symptoms. However, it becomes less effective with time and some symptoms do not respond to medication. Complementary interventions are therefore required for PD. Recent studies have implemented transcranial direct current stimulation (tDCS) in combination with other modalities of interventions, such as physical and cognitive training. Although the combination of tDCS with physical and cognitive training seems promising, the existing studies present mixed results. Therefore, a systematic review of the literature is necessary. Aims This systematic review aims to (i) assess the clinical effects of tDCS when applied in combination with physical or cognitive therapies in people with PD and; (ii) analyze how specific details of the intervention protocols may relate to findings. Methods The search strategy detailed the technique of stimulation, population and combined interventions (i.e. cognitive and/or physical training). Only controlled studies were included. Results Seventeen of an initial yield of 408 studies satisfied the criteria. Studies involved small sample sizes. tDCS protocols and characteristics of combined interventions varied. The reviewed studies suggest that synergistic effects may be obtained for cognition, upper limb function, gait/mobility and posture when tDCS is combined with cognitive and/or motor interventions in PD. Conclusion The reported results encourage further research to better understand the therapeutic utility of tDCS and to inform optimal clinical use in PD. Future studies in this field should focus on determining optimal stimulation parameters and intervention characteristics for maximal benefits in people with PD.


Introduction
Parkinson's disease (PD) is a neurodegenerative disorder mainly characterized by the progressive loss of dopaminergic neurons in the substantia nigra pars compacta. This leads to excessive GABAergic (inhibitory) signaling from the output nuclei of the basal ganglia to the thalamus and other subcortical structures [1,2]. In turn, the thalamus sends reduced excitatory signaling to many cortical areas, leading to a broad cortical dysfunction in PD [1,2], which includes sensorimotor and cognitive areas. PD is traditionally described as a movement disorder, including symptoms such as bradykinesia, resting tremor, rigidity, postural instability and gait impairments [3,4]. Mood disorders (e.g. anxiety and depressive symptoms) and cognitive impairments (e.g. executive function, memory, etc.) are also common and disabling in PD [5,6]. Dopaminergic medication is the primary treatment used to manage PD symptoms. However, it becomes less effective and side effects emerge with time, such as dyskinesia, motor fluctuations and hallucination [7,8]. Additionally, cognitive impairments and postural instability do not respond to dopaminergic medication [9,10]. Complementary interventions are therefore required for PD.
A growing body of evidence suggests that transcranial direct current stimulation (tDCS), a low-cost method of non-invasive brain stimulation, could potentially become a clinical tool for PD in the near future [11][12][13][14]. tDCS directs, through scalp electrodes, a constant low amplitude electric current (generally between 1 and 2 mA), which has been shown to modulate excitability in both cortical [15] and subcortical brain areas [16,17]. Anodal tDCS leads to increased neuronal excitability whereas cathodal tDCS leads to reduced neuronal excitability [15,18]. tDCS can also modulate oxygen supply to cortical and subcortical areas [19] and neuronal synapsis strength [20], triggering plasticity processes. Of particular interest to its application in PD, anodal tDCS increases extracellular dopamine levels in the striatum [21] and inhibits GABAergic neurons [22,23]. Recent systematic reviews confirmed that, as a stand-alone intervention, tDCS promotes benefits on motor function and to a lesser extent on cognition in people with PD [12,24].
Taking advantage of tDCS portability, researchers have implemented tDCS in combination with other modalities of complementary interventions, such as physical (i.e. exercise, physiotherapy, etc.) [25] and cognitive training [26,27]. The idea is that such combinations would promote greater, synergistic effects than the interventions applied separately [28,29]. In this context, tDCS can be applied concurrently or as a priming technique. It has been argued that such applications may reinforce longterm potentiation-like processes [30], promoting greater retention of benefits from combined therapy [31]. Although the combination of tDCS with physical and cognitive training seems promising, the existing studies present mixed results. Therefore, a systematic review of the literature is necessary. This systematic review assessed: (i) the clinical effects of tDCS when applied in combination with physical or cognitive therapies in people with PD and; (ii) how specific details of the intervention protocols may relate to findings.

Search strategy
Two of the authors (VSB and NRC) created a search strategy, which was approved by all authors, to identify all potentially relevant studies. Table 1 shows the terms and synonyms used to search papers in the databases in the title, abstract or keywords. The search strategy included three fields (connected with "AND") with independent search terms. Terms in the same field were connected with the conjunction "OR". The first search field focused on the population (i.e. Parkinson's disease). The second search field comprised types of nonpharmacologic treatments (i.e. motor/physical therapies, rehabilitation, cognitive therapies, and exercises). The third search field focused on tDCS. The search terms were combined and explored with the medical subject headings (MeSH) in different databases (Pubmed, Scopus, Embase, Web of Science and PsycNET).

Study selection
The flow chart with information regarding the different phases of the search and screening process is shown in Fig. 1 [32], and two because they did not have control group or sham condition [26,27]. Table 2 shows the extracted data regarding the methodological aspects of the studies included in the present systematic review.

Participants
The sample size varied from 1 to 53 participants, with a mean age between 56.67 and 79 years. The mean of the UPDRS motor section (part III) score ranged from 10.9 to 47.7. The disease duration ranged from 4.4 to 9.4 years and Levodopa Equivalent Dose (LED) ranged from 251.56 to 912 mg/day.

Study design
Seven studies had cross-over design [25,28,34,37,[39][40][41] and ten were controlled trials with parallel arms. Three studies did not mention information regarding how or if experiments were blinded [37,40,42], three studies consisted of single-blind experiments [28,34,36] and eleven studies consisted of double-blind experiments. It is worth mentioning that all the included studies randomly assigned participants into groups/conditions.

Duration of the stimulation (active and sham)
Stimulation time for active protocols varied between 7.5 and 30 min, with nine studies using 20-min sessions [28, 33, 34, 36-39, 42, 44]. Most of the studies used sham protocols in which the current was delivered during the initial period of the session (8 to 60 s) and then turned off; ten studies reported that the current was delivered during the initial 30 s [25,28,31,[35][36][37][38][39][40][41]. Two studies reported that the current was turned on once again for the last 10 s of the session [29,43].

Gait and mobility
Two out of eight studies reported synergistic effects on gait and mobility at post-test, including increases in gait speed and stride length and improvement in the turn phase of TUG [25,41]. The other six studies assessing gait reported similar findings for both active and sham tDCS at post-test [29,31,35,37,44,45]. Two of out five studies observed synergistic effects on gait and mobility at follow-up assessment [31,35]. The other three studies assessing gait and mobility reported similar findings for both active and sham tDCS at follow-up test [29,44,45].

Postural control
Two out of four studies reported synergistic effects on postural control at post-test, including reduced time taken to recover balance following retropulsion and increased trunk peak velocity during tango [25,41]. The other two studies assessing postural control reported similar findings for both active and sham tDCS at posttest and follow-up test [29,35].

Upper limb movements and motor symptoms
The four studies assessing upper limb function observed synergistic effects. Broeder

Neuropsychiatric symptoms, fatigue, sleep disorders and quality of life
Neuropsychiatric symptoms (i.e. depression [29]), sleep disorders [38] and/or quality of life [29,35,42] were less frequently assessed. Synergistic effects on neuropsychiatric symptoms (at post-test and follow-up) [43] and fatigue (at post-test) [38] were reported by only one study.

Discussion
This review examined 17 studies that assessed the effects of tDCS when applied in combination with physical or cognitive therapies in people with PD. In summary, the included studies had appropriate design    [33] * Active tDCS reduced performance in the attention/executive skills and delayed memory index when compared to sham tDCS at the post-test.
* Active tDCS tends to improve performance in the immediate memory index compared to the sham group at the follow-up test.
* No significant UPDRS-III motor changes were observed between groups at 4 and 16-week follow-up tests.
* No effects of tDCS were found for the amplitude, variability, and speed of the strokes outside the freezing episodes.
* Patients who reported freezing episodes in daily life (n = 6) showed a beneficial effect of tDCS on stroke characteristics.
• No adverse events of tDCS were reported.
Broeder (2019) [34] * Active tDCS improved writing during the tDCS protocol, at the post-test and at follow-up compared to sham. * Active tDCS increased writing amplitude at follow-up period compared to post period. * Active tDCS enhanced cortical excitability compared to sham at the post-test.
* Active tDCS enhanced cortical excitability compared to sham at the post-test.
• No adverse events of tDCS were reported.
Costa-Ribeiro (2016) [31] * Both groups improved functional mobility either in on or off medication condition compared with baseline.
* However, for both medication conditions, these gains were maintained only in the tDCS+ gait training at follow-up test.
* In the tDCS + gait training, enhancement of cortical excitability was observed at post-intervention and 1-month follow-up (both only for the "on" phase).
*For all outcome measures, no significant differences were found between groups.
* The improvement in velocity and TUG were maintained at the follow-up test only for patients in the Cueing gait training + tDCS group.
• No adverse events were reported by any of the participants.
Criminger (2018) [36] *No differences were observed for TUG between conditions. *Increased DTC in the TUG motor (gait) after a tDCSbike session when compared to tDCSwii.
*Increased DTC in the TUG cognitive (cognitive) after a tDCSbike session when compared to tDCSwii.
*Increased DTC in the TUG cognitive (gait) after a tDCSwii session when compared to tDCSbike.
• 1 participant was excluded from the initial sample (n = 18) after presenting headache in the first session.
* All groups decreased Hmax/Mmax ratio and intracortical facilitation at post-test.
* Sham tDCS + treadmill and treadmill groups decreased reciprocal Ia inhibition at post-test when compared to pre-test.
* tDCS + physical training decreased the walking time and the time taken to regain stability following the retropulsion stimulus when compared with tDCS.
* tDCS + physical training improved the turn phase of TUG.
* Sham + physical training decreased walking time and increased stride length but these were comparatively less than with tDCS + physical training.
* Dance + tDCS increased for the 90% velocity range and total trunk velocity area.
Lawrence (2018) [42] * Standard CT improved memory at follow-up test, quality of life, and activities of daily life at post-test. However, decreased visuospatial ability at follow-up test. * CG had no improvements.
* Physical therapy + tDCS increased PD-CRS total, frontal-subcortical scores and verbal fluency at post, and stabilized the effect at follow-up test.
* Physical therapy + tDCS group decreased the time necessary for completing TMT-B at post-test.
* Both groups improved the Standing Stork, Four Square Step, and Sit, and Reach Tests at post-test, with improvements maintained at follow-up test for the Standing Stork, Four Square Step tests.
* Both groups improved TUG performance at follow-up test.
Manenti (2018) [43] * Both groups improved language, attentional and executive functions at post and follow-up periods. * Both groups increased phonemic fluency at post-test and semantic fluency at follow-up. * tDCS + computerized cognitive training showed lower depressive symptoms and greater phonemic fluency when compared to Sham + computerized cognitive training at post-test and follow-up.
• No adverse events were reported by any of the participants.
Schabrun (2016) [44] * Both groups improved gait velocity, cadence, step length and double support time in gait dual-tasks and bradykinesia at postand follow-up tests.
* Both groups improved functional mobility during TUG with words at post and follow-up tests.
* Active tDCS + dual-task gait training improved the number of correct responses during TUG with counting and TUG with words at post-test, with a trend to maintain this performance in TUG with words at follow-up test.
* There were no differences between groups for reaction time and attention.
(i.e. cross-over or parallel arms) and control (i.e. sham tDCS). However, most included studies involved small sample sizes (n < 24), which makes the results difficult to generalize to the full range of people with PD. Also, tDCS protocols varied for stimulation time (from 7.5 and 30 min) and number of sessions (from 1 to 16 sessions), making comparisons and definitive conclusion regarding potential synergistic effects challenging. The most consistent synergistic effects were reported for cognition [29,33,40,[42][43][44] and upper limb function [28,34,39,40]. Although findings related to other aspects of PD were inconsistent, synergistic effects were also reported for gait and postural control [25,31,35,41]. The large heterogeneity in stimulation parameters and combined interventions may explain the large variation of findings that have been reported by the reviewed studies. In addition, it is not possible to determine if clinical and demographic characteristics of the studied individuals influenced the observed variability on results.
Study types are chronic (C) or acute (A). * Others include Quality of Life, fatigue, sleep disorders.   • Adverse effects (occurrence)

Post Assessment Follow-up Assessment
• One participant experienced strong tingling over the site of one electrode and a momentary flash of light in his eyes. The sensations lasted approximately 5 s. The participant ceased training that day but continued on subsequent days with no other events, and no other symptoms.
Yotnuengnit (2018) [45] * All groups improved gait velocity and step time at post-test and at 2nd and 6th week follow-up. * Physical therapy group increased cadence at 2nd and 6th week follow-up tests.
* tDCS and sham + physical therapy improved UPDRS II in all tests and the tDCS + physical therapy improved at the post and 2 weeks follow tests. * All groups improved UPDRS III at post and 2nd week follow-up tests.

Methodological aspects
The included studies were consistent with regards to tDCS polarity and site of stimulation. In all studies, the anode electrode was placed in order to target brain area related to motor or cognitive functions. Studies aiming to improve cognition targeted DLPFC while those aiming to improve motor aspects of PD targeted M1, SMA and/or PMC. As PD is characterized by reduced dopaminergic signaling by the substantia nigra pars compacta and the consequent increased GABAergic signaling from the basal ganglia to other encephalic regions, it makes sense to use anodal tDCS for rehabilitation in PD. Anodal tDCS has been shown to increase extracellular dopamine levels in the striatum [21] and inhibit GABAergic neurons [22,23]. However, it is somewhat surprising that only one of the reviewed studies used cathodal tDCS [40]. Ishikuro et al. [40] applied the cathodal tDCS over the occipital area. A growing body of evidence supports the hypothesis that the functional interhemispheric imbalance contributes to the clinical motor deficits in PD. For example, PD is associated with asymmetry in M1 excitability [47,48], with the moreaffected hemisphere showing decreased excitability in comparison to the less-affected one. Thus, cathodal tDCS applied to the less-affected hemisphere (as well as anodal tDCS on the more-affected hemisphere) may also benefit patients with PD by leading to a more balanced interhemispheric activity. Cosentino et al. (2017) [49] observed that anodal tDCS of the more-affected M1 and cathodal tDCS of the less-affected M1 were able to induce polarity-specific changes in cortical excitability, leading to a more balanced interhemispheric excitability. These authors also observed that motor performances of both hands improved after both stimulation protocols [49]. Additional studies investigating the effects of cathodal tDCS in PD is required, especially when applied in combination with physical or cognitive interventions.
Another consistent aspect of the intervention protocols of the reviewed studies refers to current intensity. Thirteen studies used 2 mA and none compared the effects of different tDCS intensities when applied in combination with other interventions. The choice for 2 mA may be justified by the fact that some neurophysiological studies have shown greater increase in cortical excitability after 2 mA tDCS when compared to 1 mA [50,51]. Also, longer lasting effects have been associated with greater current intensities [15]. However, other studies have found no differences in cortical excitability when comparing 2 mA and 1 mA [52][53][54]. Further research is required to understand if current intensity is a moderator factor when tDCS is applied in combination with other complementary interventions. Tolerability and safety of current intensities greater than 2 mA are still to be investigated in this kind of interventions in PD.

Reported findings
Consistent synergistic effects were reported for cognition when tDCS was applied in combination with other modalities of complementary interventions. All six studies reporting synergistic effects on cognition used multiple sessions (4 to 16 sessions). Current intensity included 1.0, 1.5 or 2.0 mA. The reported synergistic effects were consistent with the area targeted by tDCS. Four studies targeted the DLPFC and one study target the frontal polar area, cortical regions known to be involved in executive function and working memory [29,33,42]. Interestingly, one study that stimulated M1 observed increased number of correct responses provided while performing the TUG test under dual-task condition (i.e., concomitant cognitive task) [44]. Authors argued that individuals with PD improved their ability to dual task when walking due to improved movement automaticity after the intervention. It is possible that anodal tDCS on M1 improved the efficiency of the direct locomotor pathway (i.e. neuronal commands are transmitted directly from M1 to the spinal cord), leading to a more automatic control [55,56]. Then, the attentional and executive resources previously required for the control of movements could be reallocated to the performance of the concomitant cognitive task, which led to better performance of such task.
Upper limb function and motor symptoms (as assessed by UPDRS-III) respond differently to combined interventions involving tDCS. Consistent synergistic effects were reported for upper limb function [28,34,39,40]. Since methods of the studies reporting the synergistic effects on upper limb function varied, it is difficult to establish associations with results. Overall, synergistic effects were observed after a single or multiple sessions, with tDCS targeting DLPFC or M1. On the other hand, motor symptoms as assessed by UPDRS-III seem to not benefit from the addition of tDCS on physical/cognitive interventions. Only one (out of five) study observed synergistic effects on motor symptoms [40]. UPDRS-III may miss subtle motor improvements and, therefore, we suggest future studies to use more objective assessments of clinical motor symptoms (e.g., inertial sensors, electromyography, etc.).
Although inconsistent, synergistic effects were also reported for gait and mobility when tDCS was combined with other complementary interventions in people with PD. Two out of eight studies reported synergistic effects of tDCS on gait and mobility at post-test [25,41]. These two studies involved immediate physical interventions (i.e. gait training and tango dance) with stimulation targeting bilateral M1 and PMC. Studies targeting one cortical area (i.e. unilateral M1 or prefrontal cortex -PFC) observed similar results for both active and sham tDCS. Given the multiple cortical regions involved in gait and the bilateral representation of these regions [55,57], it is possible that bilateral stimulation of multiple cortical areas is required to provide synergistic immediate effects (i.e. at post-test) to gait in people with PD. Also, two studies observed synergistic effects on gait and mobility at follow-up assessment [31,35]. Costa-Ribeiro et al. [31,35] observed that tDCS (applied before the physical intervention) prolonged the effects of cued gait training on functional mobility and that this benefit is independent of dopaminergic medication. The synergistic effects on mobility at follow-up assessment could be explained by the changes in cortical excitability. While both sham and active tDCS groups increased cortical excitability and improved mobility at post assessment, only the gait training plus active tDCS group maintained the benefits at the follow-up assessment [31]. Thus, there seems to be a positive relationship between increased cortical excitability and improvement in mobility in patients with PD [11,31].
Synergistic effects on postural control were also reported when tDCS was combined with physical interventions in people with PD. Four studies assessed postural control, but only two studies presented synergistic effects of active tDCS combined with gait training and tango dance [25,41]. These studies involved a single session of intervention and stimulated motor areas (i.e. M1 and PMC) bilaterally. The other two studies that reported no synergistic effects at the post and follow-up assessments involved 10 sessions of physical interventions (i.e. physical therapy and cued gait training) and stimulated M1 and PFC unilaterally (contralateral to the most affected body side). Postural control involves several cortical areas (i.e. PFC, PMC, SMA, M1, and primary sensory cortex -S1) in both hemispheres and the cerebellum [2,58]. Therefore, it is possible that bilateral stimulation of multiple areas is necessary for the synergistic immediate effects to postural control in patients with PD. Besides, it should be noted that the synergistic effects were evidenced by studies that performed the stimulation during the physical intervention. Physical training has been shown to normalize M1 excitability in people with PD [59], while tDCS may decrease the threshold for these changes to occur, facilitating longlasting effects [11,25].
Limitations apply while interpreting the current findings. This review is limited by the small number of papers identified (n = 17) in the literature and the varied protocols tested in the included studies. This limits our interpretations and makes definitive conclusion regarding potential synergistic effects challenging. Additionally, although we acknowledge the contributions of openlabel studies to the developments in this emerging area of research, we opted to exclude open-label studies from our review due to the inherent methodological flaws of such design. Despite these limitations, this review provides a useful synthesis of the existing studies on the combination of tDCS with physical/cognitive interventions in PD, which may guide the development of the field towards a more robust body of evidence.

Conclusions
Although the reviewed studies used appropriate design and control, they involved a limited number of participants, which may imply underpowered analysis. Thus, large-scale studies are needed. Despite this major flaw, the reported results of tDCS interventions combined with cognitive and/or motor interventions encourage further research to better understand its therapeutic utility and to inform optimal clinical use in PD. The reviewed studies suggest that synergistic effects may be obtained for cognition, upper limb function gait/mobility and posture when tDCS is combined with cognitive and/ or motor interventions in PD. Future studies in this field should focus on determining optimal stimulation parameters and intervention characteristics for maximal benefits in people with PD. Research on identifying potential predictors of response to tDCS-based interventions (i.e. tDCS combined with cognitive and/or motor interventions) in people with PD should also be conducted. Availability of data and materials All primary data were extracted from the referenced sources. The datasets used during the current study are available from the corresponding author on reasonable request.
Ethics approval and consent to participate Not applicable.

Consent for publication
Not applicable.