Effect of rTMS on Parkinson's Cognitive Deficit: A Systematic Review and Meta-analysis CURRENT STATUS: UNDER

Background: To evaluate the effects and optimal parameters of repetitive transcranial magnetic stimulation (rTMS) on cognition improvement of Parkinson's disease (PD) patients with cognitive deficit and to estimate which cognitive function may obtain more benefit from rTMS. Method: The articles dealing with rTMS on cognitive deficit of PD patients were retrieved from the databases until December 2018. Two researchers selected research papers, evaluated their quality, extracted data, and cross-checked them according to the inclusion and exclusion criteria. The standardized mean differences (SMDs) with 95 % confidence interval (CI) of cognitive outcome for different parameters, scales, and cognitive functions were estimated. Results: Fourteen studies involving 163 subjects were included in the study. A significant effect size was observed for the mini-mental state examination (MMSE) global cognitive outcome. Further subtests for different cognitive domains demonstrated prominent effect for the executive function. The significant effect sizes were respectively found with multiple sessions of high-frequency rTMS over frontal cortex, especially with over dorsolateral prefrontal cortex (DLPFC), based on subgroup analysis of the executive function. All of the other cognitive domains including working memory, attention, and language ability did not obtain significant effects. Conclusions: Multiple sessions of high frequency rTMS over the DLPFC may have positive effect on executive function in PD patients with cognitive deficit. Further well designed studies with large sample sizes are needed to verify our results and ascertain the long-term effects of rTMS.

3 impairment (MCI) at the onset of the disease. As a recent review [4] has shown. An average of 27.0 % of non-demented patients in PD patients suffered cognitive decline. Nearly 80.0 % of patients eventually develop dementia in the later stages of the disease, and dementia is an important risk factor associated with PD mortality [5]. The forms of cognitive deficit in PD patients vary and including executive dysfunction, visual spatial disorder, memory decline, and language dysfunction.
Among them, executive function impairment was the most prominent. In a clinical practice, multicenter collaborative study, [6] the study found executive dysfunction accounted for 10.1% in PD with cognitive deficit. The underlying pathogenesis and mechanism of cognitive deficit is still unclear, but it is closely related to the complex neuropathological abnormalities of PD [7]. For instance, in PD patients, neurotransmitters in the brain are changed, dopaminergic neurons in the substantia nigra are lost, striatum dopamine is depleted, and the cortical-subcortical dopamine loop between basal ganglia and frontal lobe is significantly damaged. Beyond that, the atrophy of the hippocampus and frontal cortex as well as the precipitation of abnormal proteins may be some of the factors that cause cognitive deficit. To date, clinical research for cognitive therapy of PD patients is insufficient.
Traditional treatment has been physical and cognitive exercises, pharmacotherapy, and cognitive therapy. The above methods have no significantly deterministic clinical efficacy, and some of which have a series of side effects such as nausea, vomiting and aggravating symptoms of exercise [8]. So many researchers are looking for newer and better practical treatments.
In recent years, non-invasive brain stimulation techniques, such as repetitive transcranial magnetic stimulation (rTMS), are used as a potential therapeutic tool and are recommended for the treatment of PD patient. rTMS mainly induces currents in local areas of the cerebral cortex through rapidly changing magnetic fields to depolarize the central nervous synapse cells and creating the synaptic terminal nerve activity, thus causing a series of brain metabolic changes in neuronal potential activity and other physiological functional responses. As rTMS stimulation is administered to the brain, in general, the brain regulates its functions in response to external stimuli. Long-term external stimuli can change the structure of the brain's nerves reconfiguring it for adaptation to the outside world. This is called as neuroplasticity. At this time, the brain's synaptic structure and neural network changes accordingly, eventually changing the corresponding behavior and learning ability as well as playing a certain role in the treatment of damaged brain. The cognitive after effect of rTMS has been reported by many authors in different patient populations including dementia patients, PD patients, and even in healthy controls [9 -12]. However, due to the lack of understanding on the mechanism of sustained repair of cortical excitability caused by stimulation, and the variability of the within-subject and between-subject induced by rTMS, there is no consensus on rTMS parameters or overall efficacy.
One study [13] showed that efficacy of rTMS was closely linked to the stimulation site, pulse number, stimulation intensity, frequency, and the number of treatment sessions.
The debates about the efficacy of rTMS treatment on cognitive function of PD patients has continued. Five reviews/meta-analyses were published to summarize the clinical research results in recent years. Two reviews were published by Anderkova et al. In 2014, their study [14] preliminarily summarized the impact of rTMS on cognitive impairment in PD patients, AD patients, and MCI patients from a clinical perspective, which contained its after-effects and inter-individual variability in their magnitude, discrepancies in stimulation protocols and study designs, varied selection of the specific stimulated areas and control procedures, and neuropsychological methods for assessment of after-effects. In 2017, their second study [15] reviewed the therapeutic potential of non-invasive stimulation and included both rTMS and transcranial direct current stimulation (tDCS) on depression and NMS in PD patients. The results were quite preliminary but hopeful, showing rTMS had some positive effects on depressive symptoms and cognitive impairment in PD patients. Also in 2017, Dinkelbach's review [16] summarized the importance of rTMS stimulation sites, considering dorsolateral prefrontal lobe (DLPFC) as a crossroads of depression and cognitive therapy. Randver's review, [17] in 2018, collected available literature on rTMS in the treatment of NMS of Parkinson's disease (i.e., emotional and cognitive disorders) associated with DLPFC which showed high-frequency prefrontal rTMS was beneficial for PD-related depression, but the availability of reducing PD-related cognitive impairment has remained uncertain. The above review had preliminary results, but still had obvious disputes and was without specific data support. Also in 2017, Lawrence et al. [18] performed a detailed analysis on rTMS treatment for cognitive domains in patients with PD, which only included three articles, and 5 showed a negative outcome. In 2018, though, Cohen et al, [19] and Buard et al. [20] both published studies about the effects of rTMS treatment on cognitive function of PD patients. Therefore, the purpose of this study was to provide an objective and comprehensive analysis that whether rTMS treatment was effective on cognitive deficit of PD patients which cognitive domain obtained more from rTMS stimulation and which rTMS parameters are the most appropriate.

Method
The meta-analysis was conducted based on the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement.

Search strategy
The search criteria for published related clinical studies was publications before December, 2018 and were searched from the following databases: PubMed, Cochrane Library, Embase, ScienceDirect, and Web of science. In order to collect the literature comprehensively, a wide range of terms were used: "rTMS" or "TMS" or "magnetic stimulation" or 'repetitive transcranial magnetic stimulation', "cognitive" or "cognition" or "MCI" or "mild-cognitive impairment" or "neurodegenerative", and "Parkinson" or "PD".

Inclusion and exclusion criteria
The included research studies strictly meet the following inclusion criteria: application of rTMS, involvement of PD patients, measurement of the cognitive domain, published in English. Studies were excluded if there was no data available for the size of the effect size. In order to include all relevant articles more broadly, the design of the test was not limited. Randomized controlled trials (RCT), crossover trials, and self-controlled trials were included in the search.

Study quality
The research publication evaluation process was carried out by the first and second authors of the study, and the evaluation criteria was based on the improved evaluation scale of Moher et al.It mainly included the following evaluation items; 1) whether the experimental design was randomized; 2) whether the blind method was adopted and the type of the blind method was recorded in detail; 3) whether there was an exit in the middle of the experiment, and if so whether the study record the 6 number of exits in detail; 4) whether the detailed basic information of the subjects was included; 5) whether the experiment was a comparison between the control group and the experimental group; and 6) whether any adverse reactions were reported, and if any, the number of adverse reactions, the type, and severity of adverse reactions were described in the article.

Data extraction
Two experienced reviewers independently evaluated the studies based on the inclusion/exclusion criteria above. If any disagreement occurred, it was resolved through discussion and consultation with third-party reviewers. The detailed basic information was extracted which included the first author, release year, intervention method, the course of the disease, number of subjects, treatment site, and rTMS parameters (Table 1). In order to reduce the heterogeneity produced by different experimental designs including RCT, crossover trials, and self-controlled trials, we only extracted the mean and standard deviations (SD) of the cognitive scale data before and after rTMS treatment. If the standard deviation (SEM) of the mean was provided, it was convert to a standard deviation (SD) by using the formula of SD = SEM×. The response time and accuracy of the different cognitive tasks was recorded to combine the results to reflect the therapeutic effect of rTMS.

Statistical Analysis
The meta-analysis was conducted by using RevMan 5.3 software provided by Cochrane collaboration (London, UK). The standardized mean difference (SMD) and its 95 % confidence interval were selected to display the combined results. A Q test was used to analyze the heterogeneity among the studies.
When there were statistics among the studies (I 2 ≤ 50 %), the fixed effect model was used for metaanalysis. On the contrary, meta-analysis was conducted with randomized effects model. In addition, by examining the data extraction method and the raw data included in the study, by analyzing the clinical intervention measures and experimental design, and by using sensitivity analysis as well as other methods to find the cause of heterogeneity. We used the inverted funnel chart to assess the presence of possible publication bias. Comparison of outcome variables used a P < 0.05 value for statistical significance.

Characteristics of the included research literature
The computer retrieved 208 articles. After reading the title and abstract as well as excluding duplicate documents, 14 documents were finally left that met the criteria. The detailed article screening process is shown in Figure 1. A total of six parallel design experiments, five cross-design experiments, and three self-control design experiments were included in this study. Seven articles were evaluated for overall cognitive function, 13 articles for different cognitive domains (e.g., executive function, working memory, attention, and language function). All articles contained the data immediately after the treatment, and only three articles had long-term follow-up data.

Characteristics of the patients with PD
A total of 249 subjects were included in the 14 studies, 70 subjects of which also suffered from depression, and 163 subjects of which included in this study. Three articles included a total of 46 patients with idiopathic PD. Most studies had no detailed information on the patient's motor syndrome such as tremor, Bradykinesia, or posture gait abnormalities. A few studies have reported this in detail but nothing existed on clinical heterogeneity on the patient's motor symptoms was included in the studies. Most patients had cognitive deficit, including six patients who suffered from dementia. The ages of all patients were over 60 years old, and about 10 of them were over 70 years old. The mean disease duration of almost all patients was more than five years, and even the mean disease duration of individual patients was more than 10 years which indicated most patients had a high risk of cognitive impairment. Most patients had stable medications for a period of time before treatment and during treatment, but the studies lacked detailed information on the specific medications.

Document quality evaluation result
Randomized allocation was used in 11 studies, but detailed distribution method was not mentioned in the other three studies. Seven articles were double-blinded, four studies were single-blinded, and three studies did not report blind-related information which was defined as an unclear blinded method. Four studies documented the dropout of patients in the middle of the study, and the remaining ten studies did not report any dropouts if they occurred or not. All articles contained complete patient information such as age, duration of illness, and education level. Adverse reactions were reported in eight studies, two of which definitely reported the number, and the rest did not report the number. No adverse reactions were reported in the remaining six studies. Mild headache was the main side effect. Individual patients had symptoms such as fear, headache, tinnitus, dizziness, hearing loss, fainting, nausea, but no serious side effects (Table 2).

Different scale subgroup results
The

Discussion
In this study, we quantitatively tested the efficacy of rTMS on cognitive deficit of PD patients. On the whole, the results of publish works showed the effect of rTMS, mainly in specific tasks MMSE, which had a significant performance, suggesting the effect of rTMS on patients was associated to taskspecific cognitive improvement. Moreover, the stratified results showed the high frequency rTMS stimulation over the DLPFC for multiple sessions had a significant performance on executive function of PD patients, but in other cognitive domains, no positive performance was found.
In 2017, the meta-analysis [18] showed that rTMS treatment had no significant effects on cognitive recovery in PD patients. The main reason was that only three articles in the study contained a limited number of subjects. In our study, 14 studies including more subjects and showed that rTMS treatment played a positive role in the improvement of cognitive deficit which was similar to previous review of Anderkova et al. [15] in 2017. Moreover, the review of Dinkelbach et al. [16]confirmed rTMS over prefrontal areas can lead to an enhancement of cognitive functions that are predominantly impaired in PD. In fact, the efficacy of rTMS on cognition was reported in other mental illness meta-analysis, For example, in both dementia [21] and schizophrenia [9], studies have shown rTMS was a feasible goal-directed behavioral adjustments. The damage of executive functions reflects the damage to the frontal lobes of the brain, particularly the DLPFC, which ultimately leads to the degradation of the nigrostriatal dopamine pathway and the midbrain pathway [24]. Low dopaminergic status, such as before dopaminergic therapy, after the removal of levodopa or other dopaminergic drugs, and sudden drug reduction, can cause disorders in executive function which result in reduced flexibility of mental activity [25]. The causes of cognitive impairment in non-demented PD generally include changes in dopaminergic and cholinergic neurotransmitters, neuropathological changes in the limbic system, cortex and other systems, Lewy bodies, neurofibrillary tangles, and cerebrovascular diseases [26].
One previous study [27] found that rTMS stimulated the frontal cortex to regulate the dopamine system causing an increase in dopamine release in the basal ganglia which in turn improves the executive function of PD patients. Beyond that, executive function, as a process of higher cognitive function, is usually closely related to the cooperation of multiple brain regions. It is probable that stimulation of the DLPFC not only impacts cortical excitability but also within the stimulated cortex that has been engaged in the cognitive task which leads to excitability changes of the whole circuitry.
That is, the associative basal ganglia-thalamo-cortical loop is interconnected with the stimulated area [28,29]. function is the most common impaired domain followed by impaired memory function [33,34].
Another PD-MCI multi-center study showed the similar result that executive function disorders accounted for a large proportion of PD patients with cognitive impairment [6]. Second, the duration of the subjects included in our study was more than five years or even longer. Many patients may be in moderate cognitive impairment, and the effect of rTMS is not obvious compared to the MCI. For example, during the progression of dementia, there is a large amount of neurofibrillary tangles and amyloid deposits in the brain which are accompanied by a decline in axonal transport function and loss of neurons [35]. This may affect the rTMS treatment outcome. Therefore, although the structural anatomy of working memory is in the dorsolateral prefrontal cortex, the effect is not significant. For language function, from the past research, the anatomical structure of language function was mainly located in the lower part of the frontal gyrus [36,37]. Some imaging studies [38] have even found that language dysfunction is related to temporal lobe and language hemispheres. In addition, the brain 16 regions related to attention are mainly located in parietal and temporal lobes. In the past studies, there have been a lot of similar reports, [39,40] and the damage of the parietal structure also leads to visual neglect. [41] Hilgetag et al. [42] also found that the patient's visual attention was improved by stimulating the lateral parietal lobe through rTMS. However, most of the stimuli sites included in their study were located in the prefrontal lobe which may not lead to a corresponding improvement in language and attention function after stimulation. Of course, the completion of any cognitive task is not the result of a single brain region but the product of multiple brain regions. However, the application of accurate rTMS positioning is still a key task in improving different cognitive functions.
There are some shortcomings in this study. First, although 14 articles were included, the published studies showed a great variability in simulation parameters, study protocol designs, and outcome measures. Second, some of the PD patients even suffered from moderate or major depression, and the intricate intertwined disorder was an important factor influencing the outcome. Third, there are many scales for the measurement of cognitive impairment, but there is no uniform standard in the world. The poor directivity and specificity of some scales may lead to some deviations in the results.
Finally, a large number of articles have no data after follow-up, so that we could not analyze the longterm effects of rTMS treatment. In later studies, we hope to have some large sample of experiments involving long-term follow-up effect after treatment to increase the reliability that rTMS performs on cognitive impairment.

Conclusion
This study shows rTMS therapy may have a promising effective way of treatment on the cognitive impairment of PD patients. In particular, executive dysfunction of PD patients who had benefit with high-frequency rTMS stimulation located in the DLPFC for multiple sessions. In the future, we hope that there will be more experimental design which is rigorous, and have large sample experiments to support our results. Not applicable.

Consent for publication
Not applicable.

Conflict of Interest Statement
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. All authors declared no competing interests.

Availability of data and materials
Data generated during this study are included in this published article.

Funding
Not applicable.

Supplementary Files
This is a list of supplementary files associated with this preprint. Click to download.