The Effect of Uni-Hemispheric Dual-Site Anodal tDCS on Brain Metabolic Changes in Stroke Patients: A Randomized Clinical Trial

Uni-hemispheric concurrent dual-site anodal transcranial direct current stimulation (UHCDS a-tDCS) of the primary motor cortex (M1) and the dorsolateral prefrontal cortex (DLPFC) may enhance the efficacy of a-tDCS after stroke. However, the cellular and molecular mechanisms underlying its beneficial effects have not been defined. We aimed to investigate the effect of a-tDCSM1-DLPFC on brain metabolite concentrations (N-acetyl aspartate (NAA), choline (Cho)) in stroke patients using magnetic resonance spectroscopy (MRS). In this double-blind, sham-controlled, randomized clinical trial (RCT), 18 patients with a first chronic stroke in the territory of the middle cerebral artery trunk were recruited. Patients were allocated to one of the following two groups: (1) Experimental 1, who received five consecutive sessions of a-tDCSM1-DLPFC M1 (active)-DLPFC (active). (2) Experimental 2, who received five consecutive sessions of a-tDCSM1-DLPFC M1 (active)-DLPFC (sham). MRS assessments were performed before and 24 h after the last intervention. Results showed that after five sessions of a-tDCSM1-DLPFC, there were no significant changes in NAA and Cho levels between groups (Cohen’s d = 1.4, Cohen’s d = 0.93). Thus, dual site a-tDCSM1-DLPFC did not affect brain metabolites compared to single site a-tDCS M1.


Introduction
Stroke is the second leading cause of death worldwide [1]. More than 50% of survivors suffer from chronic disability [2]. Motor impairment is the most common physical complication. However, improving motor function in stroke patients remains a challenge [3]. Recently, neurorehabilitation has progressed towards direct brain stimulation, and studies have suggested that brain modulation may have beneficial effects on motor training [4]. Non-invasive brain stimulation (NIBS) aims to transcranially modulate the excitability of specific brain areas [5]. Transcranial direct current stimulation (tDCS) is a form of NIBS that delivers low-intensity direct current through the scalp and facilitates cell plasticity by acting on the neuronal network [6][7][8]. A recent meta-analysis demonstrated the efficacy of tDCS for motor recovery in stroke patients [9].
Changing the parameters of tDCS to achieve the maximum effect is clinically important. One of the most important parameters is electrode placement. Studies have shown that stimulation of brain areas functionally connected to the primary motor cortex (M 1 ) Brain Sci. 2023, 13, 1100 2 of 10 increases corticospinal excitability (CSE) [10]. A related method, called uni-hemispheric concurrent dual-site a-tDCS (UHCDS a-tDCS) stimulates two functionally connected brain regions simultaneously [11]. We chose M 1 and the dorsolateral prefrontal cortex (DLPFC). The DLPFC is largely responsible for attention, executive function, and working memory [12]. There is evidence of a strong link between executive function and the prefrontal cortex [13]. It is possible that DLPFC stimulation in addition to M 1 has an additive effect on motor recovery via functional connectivity to M 1 , which is thought to be stronger than M 1 stimulation alone.
Neuroimaging evidence suggests that changes in neuronal and glial metabolism may play an important role in both functional decline and recovery of brain function. Proton magnetic resonance spectroscopy (H-MRS) can detect changes in the metabolic levels of neurotransmitters such as N-acetyl aspartate (NAA), choline (Cho), and creatine (Cr), and can provide a good picture of the metabolic state of damaged tissue [14].
N-acetyl aspartate (NAA) is used as a non-invasive marker of neurological health. Stroke survivors have shown decreased levels of brain NAA [15], suggesting a loss of neurons.
NAA deficiency is associated with reduced levels of ATP, acetyl CoA and other metabolites involved in energy metabolism [11]. The researchers found that the recovery of NAA levels was only observed in conjunction with the regeneration of ATP [15]. Cr, found in neurons and glial cells, plays an important role in maintaining the high levels of energy required to maintain membrane potentials [11]. Cho and its metabolites can affect functions such as maintaining the structural integrity of cell membranes and transmembrane signaling [12,13].
Hone-Blanchet et al. showed that anodal tDCS to the left DLPFC and cathodal tDCS to the right DLPFC in healthy subjects had rapid excitatory effects during stimulation and increased the amount of NAA in the left DLPFC [16]. Carlson et al. reported decreases in glutamate/glutamine and Cr after cathodal tDCS compared to sham tDCS [17].
The present study aims to extend the previous MRS research with metabolites in stroke patients. The aim of this study is to investigate whether the addition of DLPFC stimulation to M 1 (UHCDS a-tDCS M1-DLPFC ) can alter brain metabolite concentrations. We hypothesized that the levels of brain metabolites such as NAA, creatine, and choline would change significantly after UHCDS a-tDCS M1-DLPFC treatment compared to baseline levels.

Participants and Study Design
Eighteen patients with a first chronic stroke (>6 months post-stroke) in the MCA territory were enrolled in this double-blind, randomized clinical trial. The study sample was recruited from 533 patients who were admitted to Pars Hospital with a diagnosis of stroke between 20 June 2021 and 20 July 2022, diagnosed by a physiotherapist and a neurologist based on the admission criteria.
Ischemic stroke was confirmed clinically and by neuroimaging. Patients had no history of chronic neurological or cardiac disease and were not taking any medication that could alter their cognitive state. The severity of wrist flexor Spasticity was 1 or higher on the Modified Modified Ashworth Scale (MMAS). They were able to communicate verbally with the therapist. They did not have severe cognitive and memory impairment according to the Persian version of the Mini-Mental State Examination (MMSE) (MMAS ≥ 23). Figure 1 shows the study procedure.
Patients were assured that they could withdraw from the study at any time. All patients gave written informed consent to participate in the study. The study was approved by the Ethics Committee of the University of Social Welfare and Rehabilitation (IR: USWR.REC.1400.185). Patients were assured that they could withdraw from the study at any time. All p tients gave written informed consent to participate in the study. The study was approve by the Ethics Committee of the University of Social Welfare and Rehabilitation (I USWR.REC.1400.185).

Randomization
The assessor and the participants were kept blinded to the group allocation. Ra domization was carried out using the Randomization.com website (accessed on 20, Marc 2023). The patients were randomized into two groups: Experimental 1 and Experiment 2, using a computer-generated randomization block. (1) Experimental 1 received five co secutive sessions of a-tDCS M1-DLPFC M1 (active)-DLPFC (active). (2) Experimental 2 r ceived five consecutive sessions of a-tDCS M1-DLPFC M1 (active)-DLPFC (sham). All p tients were assessed by MRS before and 24 h after five consecutive sessions of tDCS inte vention. All patients completed the intervention period and there were no dropouts. Fi ure 1 demonstrates the CONSORT flow diagram depicting the phases of enrollment, i tervention allocation, follow-up, and data analysis in this two-group parallel randomize trial ( Figure 1).

Randomization
The assessor and the participants were kept blinded to the group allocation. Randomization was carried out using the Randomization.com website (accessed on 20 March 2023). The patients were randomized into two groups: Experimental 1 and Experimental 2, using a computer-generated randomization block. (1) Experimental 1 received five consecutive sessions of a-tDCS M 1 -DLPFC M 1 (active)-DLPFC (active). (2) Experimental 2 received five consecutive sessions of a-tDCS M 1 -DLPFC M 1 (active)-DLPFC (sham). All patients were assessed by MRS before and 24 h after five consecutive sessions of tDCS intervention. All patients completed the intervention period and there were no dropouts. Figure 1 demonstrates the CONSORT flow diagram depicting the phases of enrollment, intervention allocation, follow-up, and data analysis in this two-group parallel randomized trial ( Figure 1).

H-MRS Protocol
MRS data were acquired using a Siemens 1.5 T scanner (Erlangen, Germany) with an eight-channel receive-only head coil. A conventional 3-dimensional brain image (sagittal T1 MPRAGE, TR/TE = 1800/3.5, field of view (FOV) = 256 × 256 × 160 mm 3 , resolution = 1 × 1 × 1 mm 3 ) was acquired for all patients before the MRS sequence as a reference image for volume of interest (VOI) positioning. For single-voxel spectroscopy (SVS), MRS was acquired using a point-resolved spectroscopy (PRESS) sequence. Two 2 × 2 × 2 cm 3 voxels were located in the primary motor cortex (M 1 ), dorsolateral prefrontal cortex (DLPFC). Voxels were carefully placed to avoid contact with subcutaneous fat, skull, vasculature, arachnoid space, and cerebrospinal fluid. Manual shimming was performed on all acquisitions. Parameters were set to TR/TE = 1500/135 and NEX = 128. Six saturation bands were placed around the VOI to suppress external volume signals. The average duration of each H-MRS acquisition was 10 ± 2 min (5 min for each region) with no complications.

MRS Data Processing
Data were pre-processed by applying a water removal algorithm to the reference offset of 4.65 ppm to remove residual water signals. SVS raw data were fitted using TARQUIN (Gerg Reynolds and Martin Wilson, version 4.3.10). The predefined data set of NAA, Cho, and Cr target metabolites was selected for peak fitting and metabolite concentration. The metabolite ratios of NAA/Cr and Cho/Cr were calculated by dividing the metabolite values in the same spectrum for the M 1 region.

Transcranial Direct Current Stimulation
Two single-channel tDCS devices delivered direct current stimulation through two saline-soaked electrodes. Electrode placement was determined using the international 10-20 system of electroencephalography. In both groups, the active electrodes were placed on M 1 (C3/C4) and DLPFC (F3/F4) according to the involved hemisphere, and the reference electrodes were placed on the supraorbital area of the uninvolved side ( Figure 2) [10]. According to the previous research [18], a constant current of 1 mA was applied for 20 min. In the sham group "experimental 2", the stimulation was switched off after 30 s only in the DLPFC region. The standard 5 × 7 cm 2 electrode was used as the reference electrode. To localize the excitability of the motor cortex and increase the excitability of the corticospinal tract, an active electrode of 4 × 4 cm 2 was applied to the M 1 and DLPFC regions [10,19].
placed around the VOI to suppress external volume signals. The average duration of each H-MRS acquisition was 10 ± 2 min (5 min for each region) with no complications.

MRS Data Processing
Data were pre-processed by applying a water removal algorithm to the reference offset of 4.65 ppm to remove residual water signals. SVS raw data were fitted using TAR-QUIN (Gerg Reynolds and Martin Wilson, version 4.3.10). The predefined data set of NAA, Cho, and Cr target metabolites was selected for peak fitting and metabolite concentration. The metabolite ratios of NAA/Cr and Cho/Cr were calculated by dividing the metabolite values in the same spectrum for the M1 region.

Transcranial Direct Current Stimulation
Two single-channel tDCS devices delivered direct current stimulation through two saline-soaked electrodes. Electrode placement was determined using the international 10-20 system of electroencephalography. In both groups, the active electrodes were placed on M1 (C3/C4) and DLPFC (F3/F4) according to the involved hemisphere, and the reference electrodes were placed on the supraorbital area of the uninvolved side ( Figure 2) [10]. According to the previous research [18], a constant current of 1 mA was applied for 20 min. In the sham group "experimental 2", the stimulation was switched off after 30 s only in the DLPFC region. The standard 5 × 7 cm 2 electrode was used as the reference electrode. To localize the excitability of the motor cortex and increase the excitability of the corticospinal tract, an active electrode of 4 × 4 cm 2 was applied to the M1 and DLPFC regions [10,19]. Ref. [20] Schematic illustration of electrode montage in experimental 1: UHCDS a-tDCSM1-DLPFC and experimental 2: UHCDS a-tDCSM1-DLPFC (M1active-DLPFC sham); The reference electrodes were placed over the contralateral supraorbital area in two conditions. In both groups, the active electrodes were positioned over M1 and dorsolateral prefrontal cortex (DLPFC). Ref. [20] Schematic illustration of electrode montage in experimental 1: UHCDS a-tDCS M1-DLPFC and experimental 2: UHCDS a-tDCS M1-DLPFC (M 1active -DLPFC sham ); The reference electrodes were placed over the contralateral supraorbital area in two conditions. In both groups, the active electrodes were positioned over M 1 and dorsolateral prefrontal cortex (DLPFC).

Measurement of Metabolites
MRS is an objective, non-invasive technique to detect and quantify changes in certain biochemical compounds such as NAA, Cr, and Cho in brain tissue. MRS data were collected from M 1 for all patients.

Experimental Procedures
The study procedures consisted of three steps: baseline assessment, intervention period, and post-intervention period. In the first step, MRS data were collected from patients in both groups at baseline. In the next step, all patients received five sessions of tDCS according to the group allocation. The stimulation dose was selected based on a previously published study. In the [10,18] post-intervention period, patients underwent MRS 24 h after the last tDCS session (Figure 3).

Experimental Procedures
The study procedures consisted of three steps: baseline assessment, intervention period, and post-intervention period. In the first step, MRS data were collected from patients in both groups at baseline. In the next step, all patients received five sessions of tDCS according to the group allocation.
The stimulation dose was selected based on a previously published study. In the [10,18] post-intervention period, patients underwent MRS 24 h after the last tDCS session (Figure 3).

Outcome Measures and Data Analysis
The primary outcome was the concentration of brain metabolites (NAA, Cr, Cho) and the metabolite ratio (NAA/Cr, Cho/Cr) in M1 tested by H-MRS. Metabolite levels on local brain H-MRS are often reported as ratios rather than absolute concentrations. The most common denominator is the Cr level, which is thought to be stable under normal conditions as well as under some pathological conditions [21]. Therefore, we examined NAA/Cr and Cho/Cr. Data analysis was performed using SPSS software version 26 (IBM SPSS Statistics for Windows, version 26, IBM Corp, Armonk, NY, USA). Continuous variables were summarized as mean ± standard deviation. The Shapiro-Wilk test was used to determine the normal distribution of quantitative data. The test results indicated that the MRS data were not normally distributed. Non-parametric Mann-Whitney U test and Wilcoxon signed rank test were used to compare MRS data between/within groups.
Group differences were examined by ANCOVA controlling for baseline metabolite. p < 0.05 was considered statistically significant. The sample size was calculated using G*Power software (version 3:1, Heinrich-Heine-University) based on the effect size (d = 2.0) derived from the Rayen study (power of 0.90 and α = 0.05). We compensated for 20% of the dropouts.

Outcome Measures and Data Analysis
The primary outcome was the concentration of brain metabolites (NAA, Cr, Cho) and the metabolite ratio (NAA/Cr, Cho/Cr) in M 1 tested by H-MRS. Metabolite levels on local brain H-MRS are often reported as ratios rather than absolute concentrations. The most common denominator is the Cr level, which is thought to be stable under normal conditions as well as under some pathological conditions [21]. Therefore, we examined NAA/Cr and Cho/Cr. Data analysis was performed using SPSS software version 26 (IBM SPSS Statistics for Windows, version 26, IBM Corp, Armonk, NY, USA). Continuous variables were summarized as mean ± standard deviation. The Shapiro-Wilk test was used to determine the normal distribution of quantitative data. The test results indicated that the MRS data were not normally distributed. Non-parametric Mann-Whitney U test and Wilcoxon signed rank test were used to compare MRS data between/within groups.
Group differences were examined by ANCOVA controlling for baseline metabolite. p < 0.05 was considered statistically significant. The sample size was calculated using G*Power software (version 3:1, Heinrich-Heine-University) based on the effect size (d = 2.0) derived from the Rayen study (power of 0.90 and α = 0.05). We compensated for 20% of the dropouts.

Results
Eighteen stroke patients (10 female, 8 male) with a mean age of 60.94 ± 6.92 years were enrolled. The mean time since stroke onset was 34.28 ± 8.91 weeks. Table 1 shows that there were no statistical differences between the two study groups in terms of demographic characteristics, comorbidities, and spasticity level. This study assessed the mean NAA, Cr, Cho, NAA/Cr, and NAA/Cho between/within the two groups at baseline and after intervention in M 1 .

Between-Group Comparison
The results showed significantly higher NAA and Cho concentrations in M 1 after the intervention (p = 0.040, p = 0.050 respectively), with large effect sizes for NAA and Cho, 1.41 and 0.93 respectively. Metabolite ratio results showed a non-significant difference in NAA/Cr and Cho/Cr after intervention (p = 0.113, p = 0.387).

Comparison within Groups
The result showed significant changes in NAA, Cr, and Cho in group Experimental 2 (p = 0.008), and the concentration of metabolites was increased. In group Experimental 1 there were significant differences in Cr. Cr concentration was decreased ( Table 2). For changes in metabolite ratios, there was a significant difference in NAA/Cr in both groups. However, changes in Cho/Cr (p = 0.004) were only observed in group Experimental 2 ( Figure 4). Table 2. Differences between baseline and post-intervention of brain metabolites in the M 1 between two groups.

Discussion
To our knowledge, this is the first study investigating changes in brain metabolites after uni-hemispheric concurrent dual-site a-tDCS in chronic stroke patients.

Discussion
To our knowledge, this is the first study investigating changes in brain metabolites after uni-hemispheric concurrent dual-site a-tDCS in chronic stroke patients.
The main findings of the results were significantly higher NAA, Cr, and Cho concentration in the M 1 , in the group single-site a-tDCS M1 compared to a-tDCS M1-DLPFC , as measured by 1.5 T MR spectroscopy.
Previous literature has investigated bi-hemispheric single-site tDCS in healthy subjects [16,21,22] and children with spastic cerebral palsy (CP) [23,24]. Studies have shown that a-tDCS increases the levels of NAA and Cho [16,24]. Our study was also consistent with the previous study, and the group Experimental 2 that received the single-site stimulation had a significant increase in metabolites after the intervention. Hone-Blanchet et al. [16] showed that the online effect of a single session of a-tDCS on the DLPFC increased the amount of NAA. Auvichayapat et al. [24] reported an increase in Cr, Cho, and NAA after tDCS in the basal ganglia of CP patients. N-acetyl aspartate is usually considered a neuronal marker because it is only found in mature neurons.
Researchers have found an association between low levels of brain NAA concentration and poor motor function in patients after stroke, and increased levels of NAA were also predictive of recovery [25]. Glodzik-Sobanska et al. showed that an increase in NAA in stroke patients was associated with neurological improvement [24]. Perhaps an increase in NAA after a-tDCS is due to an increase in neuronal excitability leading to long-term potentiation, such as plasticity. However, the study of metabolites in dual-site stimulation has not been investigated. Previous fMRI studies have shown that dual-site stimulation increases corticospinal excitability up to twofold [26,27].
Our results also showed an increase in Cho concentration in both groups, particularly significant in Experimental group 2. This finding is consistent with Auvichayapat et al. [24] Cho is a membrane marker and its metabolites play an important role in a variety of mechanisms, such as maintaining the structural integrity of the cell membrane, methyl metabolism, and transmembrane signaling. In this case, choline repletion may affect neuronal connections and facilitate neuroplasticity in the adult CNS.
The lesser increase of NAA and Cho in the group receiving dual-site a-tDCS of both the DLPFC and M 1 region could be explained by the concept of homeostasis-that is, the ability of the human brain to regulate changes in synaptic plasticity to avoid drastic changes in its function. Homeostasis maintains stable function against changes in the activity of the number and strength of synapses. Homeostatic plasticity is increasingly recognized as a regulator of neural change within physiological limits [24]. In this context, researchers emphasize homeostatic plasticity as a tool to prevent the instability of the neural network that occurs in neurorehabilitation. Thus, dual-site stimulation could not induce further changes by overshooting the physiological range.

Limitations
The limitations of this study should also be noted. Firstly, changes in brain metabolites were measured only 24 h after the last stimulation session, and at longer follow-up times or immediately after the intervention. Therefore, we were not able to investigate immediate and long-term effects. Secondly, a single voxel MRS was used with a 1.5 T MRI, which may have limited the collection of data from multiple brain regions simultaneously. It could be suggested that further studies use a multi-voxel 3 or 7 T MRI system to measure stimulation effects in multiple brain regions, and to investigate other metabolites. Thirdly, the tDCS intervention consisted of five consecutive days of 20 min tDCS applications, may not be sufficient to alter brain metabolites. Finally, chronic stroke patients were included in the current study, so it is suggested that future studies investigate the changes in metabolites in subacute patients and examine the levels of metabolites in both hemispheres.

Conclusions
This study aimed to investigate the effect of adding transcranial direct stimulation of the DLPFC to M 1 stimulation on changes in brain metabolites in the M 1 region. The results showed that there are no significant changes in the amount of brain metabolites after UHCDS a-tDCS M1-DLPFC compared to a-tDCS M 1 .