Effect of Deep Brain Stimulation on Swallowing Function: A Systematic Review

The effect of deep brain stimulation (DBS) on swallowing function in movement disorders is unclear. Here, we systematically reviewed this topic by searching keywords following PICOS strategy of problem (swallowing or swallow or dysphagia or aspiration) and intervention (deep brain stimulation, or DBS) in the PubMed and Web of Science in English in April 2020, with comparators [subthalamic nucleus (STN), globus pallidus interna (GPi), ventralis intermedius, (ViM), post-subthalamic area, or caudal zona incerta (PSA/cZi); ON/OFF DBS state/settings, ON/OFF medication state, Parkinson's disease (PD), dystonia, tremor], outcomes (swallowing function measures, subjective/objective) and study types (good quality original studies) in mind. We found that STN DBS at usual high-frequency stimulation could have beneficial effect (more so on subjective measures and/or OFF medication), no effect, or detrimental effect (more so on objective measures and/or ON medication) on swallowing function in patients with PD, while low-frequency stimulation (LFS) could have beneficial effect on swallowing function in patients with freezing of gait. GPi DBS could have a beneficial effect (regardless of medication state and outcome measures) or no effect, but no detrimental effect, on swallowing function in PD. GPi DBS also has beneficial effects on swallowing function in majority of the studies on Meige syndrome but not in other diseases with dystonia. PSA/cZi DBS rarely has detrimental effect on swallowing functions in patients with PD or tremor. There is limited information on ViM to assess. Information on swallowing function by DBS remains limited. Well-designed studies and direct comparison of targets are further needed.


INTRODUCTION
Dysphagia, or impaired swallow function, is one of the two major causes of mortalities in Parkinson's disease (PD) (along with falls related to the loss of balance). Dysphagia usually does not respond well to dopaminergic medication treatment (1,2). Although deep brain stimulation (DBS) has significant beneficial effects in PD patients with motor fluctuation, dyskinesia, or medication refractory tremor (3)(4)(5)(6)(7), it has less benefits in axial symptoms of balance, speech, and swallowing function. Some studies even raise concerns about worsening of the axial symptom after DBS, particularly with long-term DBS at the usual high-frequency stimulation (HFS) (8)(9)(10)(11)(12)(13), while axial symptoms have been found to predict the mortality of PD patients with STN DBS (14). Low-frequency stimulation (LFS) has been reported to have beneficial effect on axial symptoms in patients with freezing of gait (FOG) at usual HFS (15)(16)(17)(18). Most common DBS targets to treat PD are STN (subthalamic nucleus) or GPi (globus pallidus interna) (3)(4)(5)(6)(7). They both have a similar effect on motor function of PD, but different effects in non-motor symptoms, such as cognitive function and depression, with different extents in medication reduction after the surgery as well (5,19). GPi also seems to have a better outcome on axial symptoms, particularly after more than 2-year stimulation compared to STN (12).
The effect of DBS on swallowing function has not been well-studied across various movement disorders and targets. There was a retrospective study on the effect of unilateral STN vs. unilateral GPi on swallowing function in PD patients, which demonstrated a better swallowing function in penetrationaspiration (PA) scores on the videofluoroscopic swallow study (VFSS) in GPi compared to STN at medication OFF status, although there was a difference in baseline swallowing function between these two groups (20). LFS of STN was found to have beneficial effect on dysphagia compared to HFS in patients with FOG refractory to usual HFS of STN (16,17). DBS targeting the post-subthalamic area and caudal zona incerta (PSA/cZi) was thought to be associated with fewer side effects compared to ventralis intermedius (ViM) or STN (21), including the swallowing function (22)(23)(24). GPi DBS has also been used to treat various dystonia (25)(26)(27)(28), including Meige syndrome (29)(30)(31)(32), and its effect on the swallowing function is also of interest to review compared to that in PD.
Besides diseases and targets, ON/OFF DBS state and stimulation frequencies, ON/OFF medication state, outcome measures for swallowing function (subjective questionnaires or scales vs. objective assessments, such as VFSS), and study designs (randomized double blind vs. open label retrospective or prospective) could also affect the swallowing function.
There was only one review article specifically focusing on the effect of DBS on swallowing function comparing different targets in the literature, mainly on unilateral GPi to STN DBS in patients with PD (33), which was published about 7 years ago. Therefore, it is necessary to have a comprehensive review with updated information on the effect of DBS on swallowing function covering various targets and movement disorders to reflect recent advances in the field, which will help guide our clinical practice in applying DBS for movement disorders.

METHODS
We systematically searched the PubMed and the Web of Science in April 2020 for all available publications in English by keywords following PICOS concepts: problem = (dysphagia or swallowing or swallow or aspiration) and intervention = (DBS or deep brain stimulation) to include all pertinent articles, with comparators [subthalamic nucleus (STN), globus pallidus interna (GPi), ventralis intermedius (ViM), post-subthalamic area or caudal zona incerta (PSA/cZi), ON/OFF DBS state/settings (ON/OFF) medication state; Parkinson's disease (PD), dystonia, tremor], outcomes (swallowing function measures, subjective/objective) and study types (good quality original studies) in mind during the search. We followed PRISMA guideline for systematic review, and the flow chart of the literature search and selection process of the review is depicted in Figure 1 (34,35). A total of 145 publications were found from PubMed and 169 from Web of Science. After removing the duplicate entries, screening was performed to narrow down to 177 articles by excluding reviews, comments, viewpoints, author responses, letters, book chapters, single case reports with insufficient information, and meeting abstracts. Then the full texts were assessed, and we removed studies without clear outcome measures on swallowing function by DBS. We finally identified 32 unique articles. We included DBS studies targeting STN, GPi, ViM, or PSA/cZi on patients with PD, various dystonia (including Meige syndrome), and essential tremor (ET), and compared swallowing function measures (subjective vs. objective) at ON/OFF DBS state under different settings (including stimulation frequencies), or postoperative to pre-operative baseline, at ON/OFF medication state. Basic demographics and types of study designs (retrospective vs. prospective, open vs. blind) were also taken into consideration in assessments. Table 1 , 22 articles  were on PD patients, with 19 targeting STN, 3 targeting GPi,  and 3 targeting PSA/cZi, as some studies were targeting more than one target. There were six articles on Meige syndrome and five on non-Meige dystonia or dyskinesia (including primary generalized dystonia, segmental dystonia, and cerebral palsy), all targeting GPi. There was only one article on ET targeting PSA and none on ET targeting ViM on swallowing function. The majority of the studies used HFS of 125-210 Hz, but two studies used LFS of 60 Hz (16,17). The assessments included subjective measures, such as swallowing questionnaires or scales, and objective measures, such as VFSS and fiberoptic endoscopic evaluation of swallowing (FEES).

Each pertinent publication is listed in detail in
We summarized the result as below, based on the diseases and targets. in 10 PD patients assessed by anamnesis, functional oral intake scale, and clinical swallowing function (38). STN (20). Kraus reported that at least three patients developed worsening dysphagia or new dysphagia after bilateral STN DBS in a group of 27 PD patients during a mean of 30 months follow-up, based on the assessment for adverse effect, with unclear medication state though (8). Add-on stimulation of substantia nigra reticular (SNr) to STN did not have beneficial effect (41). Worsening of the dysphagia could be related to the suboptimal placement of the DBS electrodes or suboptimal programming in some cases, as turning off or reprogramming of the DBS made the swallowing symptoms better or go away in these cases (42,43).
Some studies even reported beneficial effects on the swallowing function but mostly at OFF medication status, on subjective measures, or at LFS. Ciucci    Significantly benefit in dystonia on neurostim than sham stim at 3 months. No improvement in swallow and speech after 6 months neurostim.

Unknown
GPi DBS is a safe and efficacious long-term treatment for dystonia with sustained effects on motor impairment and disability, accompanied by a robust improvement in mood and quality of life. The most common stim-related side effects were dysarthria (n = 4), swallowing difficulties (n = 1) and bradykinesia (n = 2), which were all partially reversible with adjustment of stimulation settings.

PD With PSA/cZi DBS
The swallowing function of eight PD patients with bilateral cZi DBS was assessed before and after DBS by FEES and questionnaire (22). There was no clear-cut effect of DBS at 6 and 12 months on any of the swallowing parameters except for the pre-swallow spillage, which was slightly worse in the ON stimulation state 12 months after DBS, although the medication was cut down by one-third post-operatively. Sundstedt (24). They found that cZi DBS did not have a negative impact on swallowing function, with no changes on PA scores, pharyngeal residual or premature spillage, although the medication was cut down by one-third post-operatively.

Dystonia and Meige Syndrome With GPi DBS
Bilateral GPi DBS has been shown to improve the swallowing function in majority of the studies in patients with Meige syndrome, as demonstrated by improved Burke-Fahn-Masden Dystonia Rating Scale (BFMDRS) speech and swallowing scores in 12 patients who followed up to 38 months on average (29), in 11 patients who followed up for 23 months on average (30), in 6 patients who followed up to 60 months (31), and in 40 patients who followed up at 6, 12, and 24 months after surgery (32). There was one study by Limotai et al. in six patients with Meige syndrome, with one unilateral and five bilateral GPi, evaluated 6 and 12 months after DSB for Unified Dystonia Rating Scale (UDRS) and BFMDR speech and swallowing function, but they did not find improvement in speech and swallowing function in this cohort (50). Bilateral GPi also has been used in patients with 11 non-Meige dystonia patients and 9 Meige syndrome patients (51), with significantly improved swallowing and speech scores in BFMDR up to 36 months after the DBS. Bilateral GPi also has been used in primary generalized dystonia and segmental dystonia patients (25)(26)(27), and dyskinetic cerebral palsy patients (28), but no changes or just slightly worsening in speech and swallowing function after DBS compared to baseline were reported.

ET With ViM or PSA/cZi DBS
There is no specifically designed study on the evaluation of dysphagia in ET by ViM or PSA/cZi DBS, although transient mild dysphagia after the DBS implantation surgery was reported, which usually resolved within several weeks (21,52).

DISCUSSIONS
The majority of the studies were open label, retrospective or prospective, small-size studies, with subjective and/or objective assessments of swallowing function, at ON DBS compared to OFF DBS and ON/OFF dopaminergic medication state. There were only a few prospective randomized double blind studies (16,17,39), a few on comparing different targets (20,39), and a few on comparing different frequency stimulations (16,17). Most studies used bilateral targets although some were unilateral or mixed targets, as bilateral DBS is more likely to affect the axial symptoms, including dysphagia. Some of them were not fairly compared, as there were reduced dopaminergic medications post-operatively. Although the medications probably would not have a major impact on the objective swallowing functions (1, 2), beneficial effect of dopaminergic medication was also reported in a small proportion of patients (53). Taking dopaminergic medications could also affect the subjective measure with overall improvement of the parkinsonism. Therefore, it probably could explain why some studies showed improved swallowing function at subjective measures but not objective measures at ON medication state, and why the beneficial effect of DBS is more appreciated at OFF medication state or less appreciated at ON medication state. We found that STN DBS at usual HFS could have beneficial effect (more so on subjective measures of scales, questionnaires, or swallowing item in UPDRS-II, and/or OFF medication state), no effect, or detrimental effect (more so on objective measures of VFSS or FEES, and/or ON medication state) on swallowing function in patients with PD. The effect of LFS stimulation on FOG has been consistently reported positively by many studies, as summarized in a review article (18). However, there have been only a few studies addressing its effect on dysphagia. Two studies of randomized double blinded crossover prospective studies in the short-and long-term effects did find significant benefit of LFS on acute and shortterm studies (16,17), but not the long-term benefits (17), although the long-term effect remains unclear given the small sample size and sub-clinical dysphagia in participants, which could limit the power to detect the potential difference. These studies were conducted at ON medication state in bilateral STN DBS patients with refractory FOG to HFS; hence, the beneficial effect should not necessarily be generalized to the whole PD population.
GPi DBS seems more likely to improve the swallowing function or process compared to the STN DBS, more so at OFF medication state (20,39). In contrast to STN DBS, GPi DBS does not have detrimental effect on swallowing function or process at ON medication state (20,39). Even though the non-matched baseline swallowing function in the two groups, and the retrospective and non-randomized design in assigning the targets could all affect the interpretation of the favorable PA scores in unilateral GPi compared to STN DBS (20), similar results were also obtained in a randomized, doubleblind study comparing the effect of bilateral GPi to bilateral STN DBS on jaw velocity (39), suggesting that GPi DBS is probably more favorable than STN DBS in overall swallowing function for PD patients, particularly at OFF medication state. Although there is no benefit of STN or GPi DBS on swallowing function in PD compared to the best medication therapy (at ON medication state), STN DBS could even be harmful at ON medication state, based on limited studies available so far.
Targeting GPi seemed to have positive results on Meige syndrome in the majority of the studies (29)(30)(31)(32). One of the possibilities behind the benefit is the direct effect on the pharyngeal and laryngeal dystonia by GPi, which could help to improve dysphagia symptoms. There was no study on using STN in Meige syndrome and other dystonia on dysphagia. Hence, it is not certain if targeting STN would have similar benefit, as STN has also been found to be beneficial to dystonia in PD (54). There is no beneficial effect of GPi DBS on dysphagia in patients with primary generalized dystonia, segmental dystonia, and dyskinesic cerebral palsy patients, and there rarely is worsening effect either (25)(26)(27)(28).
The PSA and cZi are relatively new targets. They have the potential to provide more efficient stimulations but fewer side effects due to their anatomic characteristics, with the fibers from both the basal ganglia and cerebellar merging together at the PSA/cZi area, and studies so far found that PSA/cZi DBS rarely has a detrimental effect on swallowing functions in patients with PD or tremor (21,55). There has been limited information on the effect of ViM DBS on swallowing function to assess so far.
In summary, we found that STN DBS at usual HFS could have beneficial effect (more so on subjective measures and/or OFF medication state), no effect, or detrimental effect (more so on objective measures and/or ON medication state) on swallowing function in patients with PD, while LFS of STN could have beneficial effect on swallowing functions in PD patients with FOG refractory to HFS. GPi DBS could have a beneficial effect (regardless of medication state, and subjective or objective measures), or no effect (more so at ON medication state), but no detrimental effect (in contrast to STN DBS, even at ON medication state) on swallowing function in PD, suggesting that GPi DBS could be probably more favorable than STN DBS in overall swallowing function for PD patients, particularly at OFF medication state. GPi DBS also has beneficial effects on swallowing function in the majority of the studies on Meige syndrome but no beneficial effect on swallowing function in other dystonia. Stimulation of PSA/cZi rarely has detrimental effect on swallowing functions. The effect of ViM on swallowing function in ET patients is too limited to assess. Overall, most of them are retrospective, open label, small-size studies, with medication reduction post-operatively. There are only a few randomized, double blind studies, a few on direct comparisons among targets or between stimulation frequencies. The overall evidence levels of these studies are low, ranging from IV to III. Information on swallowing function by DBS remains limited. Well-designed studies and direct comparison of targets and stimulating parameters are further needed to gain more insights on the effect of DBS on swallowing function in movement disorders.