Home-based transcutaneous auricular vagus nerve stimulation (taVNS) improves motor and non-motor symptoms by improving autonomic and brain functions in patients with Parkinson's disease: A randomized clinical trial

Abstract

Home-based transcutaneous auricular vagus nerve stimulation (taVNS) holds therapeutic potential for neurological disorders, yet its application in Parkinson's disease (PD) remains underexplored. In this single-blinded, placebo-controlled randomized clinical trial, PD patients received either home-based taVNS with specific stimulation parameters or sham stimulation for three weeks. TaVNS significantly improved motor symptoms, reflected as reduced MDS-UPDRS Ⅲ scores, and alleviated non-motor symptoms including quality of life and sleep disturbances compared with sham stimulation. Neuroimaging revealed that taVNS decreased glutamate levels in the striatum and thalamus, increased Regional Homogeneity values in the rolandic operculum, and enhanced fractional anisotropy in the left hippocampal cingulum and right inferior longitudinal fasciculus. Serum acetylcholine levels were elevated following taVNS and correlated with motor improvement. No serious adverse events occurred. These findings suggest that taVNS with specific parameters effectively alleviates motor and non-motor symptoms in PD, possibly through modulation of brain networks and vagal activity.

Trial registration

Chinese Clinical Trial Registry: ChiCTR230007082.

Introduction

Parkinson's disease (PD), a progressive neurodegenerative disorder, is projected to affect over 12 million people by 2040, underscoring the urgency for efficacious management strategies [1]. PD is primarily marked by motor symptoms, such as tremors, bradykinesia, and rigidity. However, its impact extends to non-motor symptoms, such as cognitive impairment, depression, sleep disorders, and autonomic dysfunction [2]. Both symptom types substantially diminish patients' quality of life, necessitating comprehensive management for improved outcomes.

Traditional treatments like Levodopa and Deep Brain Stimulation (DBS) face limitations. Levodopa, while effective in alleviating motor symptoms, often results in long-term complications such as dyskinesia and fluctuating efficacy [3]. DBS, useful for motor fluctuations or dyskinesias, is invasive and involves surgical risks, sometimes causing hypomania or mania, rendering it unsuitable for all patients [4].

Transcutaneous auricular vagus nerve stimulation (taVNS) emerges as a promising non-invasive therapy for various neurological disorders. Anatomical studies reveal that the auricular branch of the vagus nerve (ABVN) is the sole branch reaching the body surface [5]. Predominantly located in the concha, particularly the cymba conchae and cavum conchae, the cymba conchae is exclusively supplied by the ABVN [6]. taVNS involves applying electrical current to the ABVN's cutaneous receptive field in the outer ear [7]. Research indicates that taVNS has applications in neurologic and psychiatric disorders [8], improving sleep quality in primary insomnia [9] and alleviating moderate depression symptoms [10].

Recent studies have further investigated the therapeutic potential of taVNS in PD. Zhang et al. [11] demonstrated that a 7-day course of taVNS improved multiple gait parameters and modulated sensorimotor cortical activity in PD patients. Lench et al. [12] documented the feasibility and safety of multi-session taVNS in PD, although no significant differences were observed in MDS-UPDRS scores or cognitive outcomes. In a double-blind crossover study involving patients with implanted DBS systems, Marano et al. [13] observed that left-sided taVNS significantly attenuated β-band power in the contralateral subthalamic nucleus, concomitant with improved Timed Up and Go (TUG) performance—manifested as increased gait speed and reduced variability. Importantly, the taVNS-induced reduction in β power correlated positively with gait improvement [13]. Complementing these findings, van Midden et al. [14] utilized wearable motion sensors in a randomized within-subject trial, reporting frequency-specific gait enhancements: taVNS at 100 Hz enhanced stride length and arm swing velocity, while stimulation at 25 Hz improved gait speed and reduced turning duration compared to sham stimulation.

Emerging evidence also suggests potential benefits for non-motor symptoms. Zaehle et al. [15] highlighted taVNS's capacity to modulate the locus coeruleus–norepinephrine system, which is implicated in cognitive and attentional deficits. Supporting this, Zhang et al. [16] reported that a two-week taVNS intervention significantly alleviated anxiety symptoms, evidenced by reduced Hamilton Anxiety Rating Scale (HAMA) scores and increased prefrontal activation during verbal fluency tasks.

Previous research has established that taVNS activates the auricular branch of the vagus nerve, facilitating afferent signal transmission to the nucleus tractus solitarius (NTS) and locus coeruleus (LC) within the brainstem [17]. These signals propagate through direct and indirect ascending projections originating from the NTS to multiple cortical and subcortical regions, including the parabrachial area, hypothalamus, amygdala, anterior cingulate cortex, anterior insula, and nucleus accumbens [18]. Through these pathways, taVNS can modulate several neurotransmitter systems, including norepinephrine [19], dopamine [20], serotonin [21], and GABA [22]. PD is known to involve inflammatory processes [23], and vagus nerve stimulation has been shown to enhance cholinergic anti-inflammatory activity and reduce systemic inflammation [24]. Grounded in these mechanistic insights, the present study quantified serum acetylcholine levels to elucidate potential pathways underlying the effects of taVNS.

Magnetic resonance imaging (MRI) offers an objective and non-invasive method to detect alterations in brain biomarkers post-taVNS treatment. Multimodal MRI can provide various data across multiple dimensions. Studies using proton magnetic resonance spectroscopy (¹H MRS) have indicated that patients with early PD exhibit abnormal glutamate metabolism [25]. This phenomenon may be associated with damage to the nigrostriatal dopamine system. Neuroinflammation in the PD brain reduces glutamate reuptake by astrocytes, leading to significant glutamate accumulation and excitotoxicity [26]. Elevated glutamate levels have been implicated in the pathogenesis of PD. Glutamate chemical exchange saturation transfer (GluCEST) imaging is a novel technique that enables non-invasive assessment of brain glutamate levels from a metabolic perspective [27]. Furthermore, GluCEST has demonstrated higher sensitivity compared to ¹H MRS. For instance, diffusion tensor imaging (DTI) quantifies brain microstructures by identifying water molecule diffusion [28]. Previous DTI studies in PD have reported widespread white matter microstructural damage in patients, manifested as reduced fractional anisotropy (FA) values in the substantia nigra [29]. Blood oxygen level-dependent functional MRI (BOLD-fMRI) assesses neuronal activity by observing blood flow changes related to neuronal function [30]. Regional Homogeneity (ReHo) indicates synchronization of neuronal activity in brain regions [31]. Combined analysis of multimodal data provides in-depth insights into brain structure, function, and metabolism, supporting evaluation of therapeutic effects.

Despite some promising findings reported in the above-mentioned studies, there are several issues involved in the clinical implementation and dissemination of taVNS for PD. First, the majority of the reported taVNS studies in PD was performed in clinical setting and patients had to come to hospital for the treatment, resulting in increased expenses and reduced subject compliance as well as reduced efficacy due to infrequent administration of the therapy. Second, mechanisms involved in taVNS for PD are largely unknown.

The aims of this study were to explore clinical effects of home-based taVNS using a special set of parameters and a wearable device on both motor and non-motor symptoms, and to investigate the mechanisms of taVNS involving autonomic function and functional brain networks using MRI, DTI and GluCEST.

Materials and methods

Protocol approvals, registrations, and patient consents

This study was approved by the Ethical Committee of the Binzhou Medical University Hospital (Protocol: 2020-006-02) and conducted in accordance with the principles of the Declaration of Helsinki and applicable laws and regulations. It was registered at the Chinese Clinical Trial Registry (unique identifier: ChiCTR2300070828). Written informed consent was obtained from all participants before the initiation of the study.

Patients and study design

This was a single-center, single-blinded randomized controlled trial conducted at Binzhou Medical University Hospital, China, from April 2023 to January 2024. Inclusion criteria were 1) aged 18 or older, 2) diagnosed with PD according to the UK PD Society Brain Bank PD Diagnostic Criteria and with Hoehn and Yahr staging (H–Y) of 1–3, and 3) willing to sign the consent form. Exclusion criteria included 1) abrupt-onset of Parkinsonism, 2) visual hallucinations, 3) psychiatric symptoms, 4) ear trauma or pain, anatomical ear damage, 5) history of major cardiovascular disease, brain surgery or neurogenic orthostatic hypotension, 6) use of anticholinergic medications, 7) allergies to surface electrodes, 8) pregnancy or lactation, 9) substantial medical knowledge that might unblind the treatment.

The sample size estimation was based on a previous study [12], aiming to detect a 3.25-point reduction on the Movement Disorder Society-Unified Parkinson's Disease Rating Scale Part Ⅲ (MDS-UPDRS Ⅲ) [32] with a pooled standard deviation of 2.0. With 90 % power and an alpha of 0.05, a total of 36 patients were required (18 for each arm of the study). Additional participants were recruited to account for withdrawals.

The study protocol is detailed in Fig. 1. Sixty patients with PD were assessed for eligibility, with 49 eligible patients randomized a 1:1 to either taVNS (n = 25) or sham (n = 24) using a computer-generated randomization table. Participants maintained stable PD medication doses for at least 30 days before and during the trial. The principal investigator was aware of group assignments but participants were blinded. Data analysis was performed by independent evaluators who were blinded with the treatment assignment. Baseline assessments included blood sample collection, evaluation of motor and non-motor symptoms using clinical scales in the OFF state, and MRI imaging. During the three-week trial, participants self-administered treatment at home. Post-treatment, the same assessments were repeated, along with documentation of adverse events.

Fig. 1.

Study flowchart.

taVNS and sham stimulation

The bilateral cymba conchae was selected for taVNS, while the sham point was positioned at the elbow about 15 cm upper from the arm wrist, based on prior studies with no placebo effects in functional gastrointestinal diseases [33,34]. The inner arm was chosen to minimize the possibility of inadvertently stimulating auricular vagal nerve fibers. To ensure blinding, all participants were informed that different stimulation locations might be used in the study and that sensory experiences may vary across individuals, thereby reducing the likelihood that participants could identify their group assignment.

Stimulation was delivered via a match-size wearable digital stimulator (SNM-FDC01, Ningbo MedKinetic, Ningbo, China). A special set of parameters was used for both taVNS and sham stimulation: a frequency of 5 Hz, pulse width of 500 μs, 10-s on-time, 60-s off-time, and amplitude of 1.5–10 mA (The stimulation intensity was individually adjusted to each participant's maximal tolerable level and increased until a clear but non-painful tingling sensation was perceived). This set of frequency, pulse width, and on-time was previously optimized in a rodent model for treating inflammation using vagal nerve stimulation [35]. We chose this set of parameters because neuroinflammation has been consistently reported to play a major role in PD [36]. All equipment was calibrated pre-experiment to ensure accurate parameter delivery. Stimulation sessions lasted 30 min, twice daily (8 a.m. and 8 p.m.), lasting for 3 weeks (Fig. 2).

Fig. 2.

Methods of taVNS. (A) taVNS electrodes and stimulator used in this trial. (B) taVNS stimulation location and electrodes (silicon electrodes). (C) Sham stimulation location and electrodes (silicon electrodes). (D) Stimulation parameters and paradigm.

Prior to the initiation of treatment, researchers provided detailed training to both participants and their caregivers on correct taVNS operation and daily treatment requirements. To ensure that the stimulation was applied twice a day, patients were provided with printed forms for self-monitoring. Specifically, each patient received three forms—one per week—and marked a checkmark for each completed session to ensure all applications were done properly. All participants were instructed to record the time and duration of each stimulation session in a daily treatment diary throughout the three-week intervention period. In addition, research staff contacted participants once every seven days by telephone to verify adherence, and reinforce the twice-daily stimulation schedule.

Outcome measurements

Motor symptoms evaluation

The primary outcome was the assessment of motor symptoms in PD patients. The MDS-UPDRS Ⅲ was utilized to evaluate motor symptoms, comprising 18 examination items scored based on the observed severity, ranging from 0 [normal] to 4 [severe] [37]. Assessments were conducted at baseline and at the end of the 3-week treatment. Importantly, the final assessment occurred at least 8 h after the last stimulation session to minimize any acute effects of the stimulation. The MDS-UPDRS III evaluations were performed by neurologists specializing in PD who had received formal training in the standardized MDS-UPDRS scoring procedure and were selected by the principal investigator. Although the principal investigator was aware of group assignments, motor assessment was conducted by raters who were blinded to treatment allocation.

Non-motor symptoms in PD

Secondary outcomes focused on non-motor symptoms in PD patients, evaluated using various assessment scales:

• (i)Parkinson's Disease Questionnaire-39 (PDQ-39): This scale includes 39 items assessing quality of life in PD patients, with scores ranging from 0 (no symptoms) to 4 (most severe), where higher scores indicate lower quality of life [38].
• (ii)Scales for Outcomes in Parkinson's Disease-Autonomic (SCOPA-AUT): This scale evaluates autonomic nervous system function via 25 items, with scores ranging from 0 (no symptoms) to 3 (severe), where higher scores indicate more severe dysfunction [39].
• (iii)Patient Assessment of Constipation symptom (PAC-SYM): Consisting of 25 items, this scale assesses the severity of constipation, scoring from 0 (absence of symptoms) to 4 (most severe), where lower scores indicate a lower symptom burden [40].
• (iv)Hamilton Anxiety Rating Scale (HAMA): This scale evaluates anxiety levels through 14 items, scored from 0 (absence of symptoms) to 4 (most severe), where higher scores reflect greater anxiety [41].
• (v)Hamilton Depression Rating Scale (HAMD): This scale evaluates depression levels and includes 24 items, scored from 0 (absence of symptoms) to 4 (most severe) or 0 (absence of symptoms) to 2 (clearly present), with higher scores indicating more severe depression [42].
• (vi)Parkinson's Disease Sleep Scale-2 (PDSS-2): This 15-item scale assesses participants' sleep quality, with scores from 0 (absence of symptoms) to 4 (most severe), where higher scores indicate poorer sleep quality [43].
• (vii)Fatigue Scale-14 (FS-14): This scale, comprising 14 items, evaluates fatigue levels with total scores ranging from 0 to 14, where higher scores indicate more severe fatigue [44].
• (viii)Montreal Cognitive Assessment (MOCA): This scale assesses cognitive function with total scores ranging from 0 to 30, where lower scores indicate poorer cognitive function [45].

These assessments were conducted at baseline and at the conclusion of the 3-week treatment.

MRI data acquisition

Brain functions were assessed utilizing GluCEST, BOLD-fMRI, and DTI. PD typically begins unilaterally and progresses to involve both sides, leading to asymmetry in symptoms [46]. Therefore, it was essential to analyze both the onset side and the contralateral sides. Both taVNS and sham stimulation groups underwent T₁ MPRAGE (magnetization prepared rapid gradient echo), GluCEST, BOLD-fMRI, and DTI scans before treatment and after three weeks of intervention.

MRI data were acquired using a 3-T whole-body MAGNETOM Skyra system (Siemens Healthineers, Erlangen, Germany) equipped with a 20-channel phased-array head-neck coil. Three-dimensional (3D) T₁-weighted images were obtained using a MPRAGE sequence (repetition time (TR) = 1680 msec, echo time (TE) = 2.29 msec, flip angle (FA) = 90°, field of view (FOV) = 240 × 240 mm², matrix size = 256 × 256, slice thickness = 0.94 mm and voxel volume = 1 mm³).

GluCEST was used to noninvasively visualize and quantify the relative concentration and spatial distribution of glutamate in the brain. For GluCEST imaging, data were acquired using a two-dimensional turbo spin echo (TSE) sequence, with parameters set as follows: FOV = 21.2 × 18.6 cm², matrix size = 182 × 128, turbo factor = 128, section thickness = 5 mm, TR = 4000 msec, TE = 7.5 msec, and radiofrequency (RF) saturation power = 3 μT. The Z-spectrum encompassed 54 frequency offsets, ranging from −6 to +6 ppm, interleaving positive and negative offsets. A single slice encompassing the maximal area of the bilateral striatum was selected for GluCEST imaging.

DTI was used to visualize and analyze the microstructure of white matter in the brain, including the orientation, integrity and connectivity of white matter tracts. The DTI parameters included FOV = 220 × 220 mm², matrix size = 128 × 128, section thickness = 4 mm, TR = 3700 msec, and TE = 92 msec.

BOLD-fMRI parameters included FOV = 200 × 200 mm², matrix size = 128 × 128, section thickness = 4 mm, TR = 2000 msec, and TE = 30 msec.

All MRI images were reviewed by two experienced radiologists (with 10 and 11 years of experience, respectively) who were blinded with the treatment assignment. T₁-weighted images with structural damage and those exhibiting motion artifacts were excluded from analysis.

MRI image processing

GluCEST images were processed using custom-written scripts in MATLAB (MathWorks, Natick, MA, USA, 2018a). Initially, the GluCEST z-spectra were obtained by normalizing images at multiple frequency offsets using an unsaturated reference image. Subsequently, z-spectrum data underwent voxelwise fitting with a 12th-order polynomial to correct for spatial B0 inhomogeneity [47,48]. The relative concentration of glutamate was calculated using the magnetization transfer ratio asymmetry (MTRasym) analysis. The formula for MTRasym is defined as follows: MTR_{asym (Δω)} = (S_(Δω) - S_(-Δω))/S₀, where Δω denotes the saturation frequency offset, S_(Δω) signifies the saturated signal at Δω, and S₀ represents the unsaturated signal. The GluCEST signal and MTR_asym spectrum for each voxel were derived from the subtraction within the z-spectrum range of −6 PPM to 6 PPM [49]. Ultimately, the MTR_asym values we calculated were reclassified as the relative glutamate concentrations within the striatum and thalamus ROIs.

BOLD-fMRI image processing was conducted using the Resting-State fMRI Data Analysis Toolkit (RESTplus). Briefly, the first 10 time points from each subject's series were discarded to eliminate initial noise. This process was followed by realignment and correction for head motion (exclusion criteria: head movement >0.5 mm or angular rotation >0.5° in any direction), co-registration of T1 and EPI images, spatial normalization to a standard Montreal Neurological Institute template, and voxel resampling to a voxel size of 3 × 3 × 3 mm³. The statistical significance threshold was set at p < 0.001, combined with a voxel size >15, corresponding to a corrected p < 0.05, as determined using the AlphaSim tool in RESTplus software.

DTI data was assessed using the Tract-Based Spatial Statistics (TBSS), a powerful technique for analyzing DTI data, particularly when comparing white matter integrity and organization across multiple subjects or groups. We used the FSL (FMRIB Software Library, https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/) software developed by Oxford University. For skeletonization and region of interest (ROI) extraction from fractional anisotropy (FA) images, we followed the ENIGMA-TBSS protocols (http://enigma.ini.usc.edu/wpcontent/uploads/2014/01/ENIGMA_TBSS_protocol_USC.pdf).

Serum acetylcholine measurement

Treatment-induced changes in serum levels of acetylcholine (a primary neurotransmitter of the parasympathetic nervous system) were assessed using a commercial acetylcholine assay kit. Blood samples were collected at baseline and after-treatment from both the taVNS and sham stimulation groups. To ensure consistency, sample collection for each patient was conducted at the same time points during the baseline period and at the end of the 3-week treatment. For instance, if blood was collected at 11 a.m. during the baseline, it was also collected around 11 a.m. at the end of the treatment period. This protocol aimed to minimize the influence of daily fluctuations in peripheral blood acetylcholine levels [50]. Serum acetylcholine levels were measured using commercial kits (A105-2-1) from Nanjing Jiancheng Bioengineering Institute (Nanjing, Jiangsu, China), with absorbance readings taken at 550 nm using an ELISA reader.

Adverse events

Adverse events were assessed according to the Common Terminology Criteria for Adverse Events (CTCAE) Version 5.0. Each adverse event was graded based on severity, with classifications as follows: grade 1 (mild), grade 2 (moderate), grade 3 (serious but not immediately life-threatening), grade 4 (life-threatening requiring urgent intervention), and grade 5 (death associated with an adverse event). Participants were encouraged to report any adverse event by telephone throughout the study. In cases of severe adverse reactions, treatment was to be immediately discontinued, and professional medical assistance was provided. All reported adverse events were documented.

Statistical analysis

Quantitative variables are presented as mean ± standard deviation (SD). Independent sample t-tests and Mann-Whitney tests were employed to assess differences between the taVNS and sham stimulation groups. Categorical variables were compared using Pearson's chi-square test. Paired t-tests and Wilcoxon signed-rank tests were utilized for pre-and post-treatment comparisons. Pearson correlation analysis explored the relationship between changes in acetylcholine levels and MDS-UPDRS Ⅲ scores in the taVNS group before and after treatment. Fisher's exact test was used to assess the rates of adverse events between the taVNS and sham stimulation groups. Data from participants who dropped out of the study were excluded from all statistical analyses, except for those involving adverse event reporting. Normality of data was assessed using the Shapiro-Wilk test. All statistical analyses were performed using IBM SPSS Statistics 26 and GraphPad Prism 9. A two-sided p-value <0.05 was considered statistically significant.

Results

Patient demographics and baseline data

A total of 49 patients were enrolled in the study, with 25 assigned to the taVNS group and 24 to the sham stimulation group. Four patients in each group dropped out of the study. Those who discontinued participation were not included in the baseline and endpoint outcome analyses. Ultimately, 41 participants completed the entire clinical trial process. Among them, 21 individuals (16 males, mean age 64.00 ± 10.85 years) received taVNS, while 20 individuals (13 males, mean age 62.70 ± 12.23 years) received sham stimulation. Table 1 summarizes the demographic and clinical characteristics of the patients. Both groups were matched regarding age, gender, disease duration, and H–Y staging. At baseline, there were no statistically significant differences in motor and non-motor symptoms, including MDS-UPDRS III scores, PDQ-39, PDSS-2, SCOPA-AUT, HAMA, HAMD, PAC-SYM, FS-14, or MOCA.

Table 1.

Demographic data and disease characteristics at baseline.

Demographic and baseline variables	taVNS (n = 21) , mean ± SD	Sham (n = 20) , mean ± SD	Group Comparisons (P value)
Age in years	64.00 ± 10.85	62.70 ± 12.23	0.72
Male sex, count (%)	16, 76.19	13, 65.00	0.431
Years since diagnosis	4.62 ± 2.50	3.95 ± 2.31	0.309
H and Y score	2.14 ± 0.48	2.10 ± 0.45	0.758
MDS-UPDRS Ⅲ scores (OFF-state)	42.00 ± 14.48	39.35 ± 15.93	0.58
PDQ-39	31.86 ± 22.78	31.80 ± 20.41	0.993
PDSS-2	14.71 ± 9.84	14.30 ± 9.2	0.89
SCOPA-CUT	12.33 ± 7.28	11.00 ± 7.28	0.561
HAMA	8.86 ± 5.78	7.70 ± 5.37	0.511
HAMD	11.14 ± 8.7	8.20 ± 6.52	0.229
PAC-SYM	8.76 ± 8.73	8.45 ± 8.26	0.948
FS-14	7.14 ± 3.53	6.80 ± 3.64	0.761
MOCA	20.71 ± 4.16	21.45 ± 3.07	0.525

MDS-UPDRS III: Movement Disorder Society-Unified Parkinson's Disease Rating Scale Part III; PDQ-39; Parkinson's Disease Questionnaire-39; PDSS-2; Parkinson's Disease Sleep Scale-2; SCOPA-AUT; Scales for Outcomes in Parkinson's Disease-Autonomic; HAMA; Hamilton Anxiety Rating Scale; HAMD; Hamilton Depression Rating Scale; PAC-SYM; Patient Assessment of Constipation symptom; FS-14; Fatigue Scale-14; MOCA; Montreal Cognitive Assessment.

taVNS ameliorated motor and non-motor symptoms

taVNS significantly reduced motor symptoms assessed by the MDS-UPDRS Ⅲ, compared to the sham group (p < 0.0001; Fig. 3A). Within the taVNS group, post-treatment scores decreased by 6.10 ± 4.29 points relative to baseline (p < 0.0001; Fig. 3B). In contrast, sham stimulation elicited no significant change in MDS-UPDRS Ⅲ scores (p = 0.1013, Fig. 3C).

Fig. 3.

Alterations in MDS-UPDRS Ⅲ scores after taVNS. (A) Between-group comparison of changes in MDS-UPDRS Ⅲ (pre-to-post treatment) for taVNS versus sham. (B) MDS-UPDRS Ⅲ scores pre and post taVNS treatment. (C) MDS-UPDRS Ⅲ scores pre and post sham stimulation. ∗∗∗∗: p < 0.0001.

For non-motor outcomes, evaluated using the PDQ-39 (quality of life), PDSS-2 (sleep disturbances), SCOPA-AUT (autonomic dysfunction), HAMA anxiety), HAMD (depression), PAC-SYM (constipation severity), FS-14 (fatigue), and MOCA (cognitive impairment). The taVNS group exhibited significantly greater reduction in PDQ-39 (p < 0.001, Fig. 4A), PDSS-2 (p < 0.05, Fig. 4B), and SCOPA-AUT (p < 0.01, Fig. 4C) scores compared with the sham group. However, no significant between-group differences were detected for HAMA (p = 0.6763, Fig. 4D), HAMD (p = 0.2391, Fig. 4E), PAC-SYM (p = 0.2598, Fig. 4F), FS-14 (p = 0.9566, Fig. 4G), or MOCA (p = 0.4203, Fig. 4H) scores.

Fig. 4.

Alterations in non-motor symptoms scale scores after taVNS. (A–H) Between-group comparison of changes in PDQ-39, PDSS-2, SCOPA-AUT, HAMA, HAMD, PAC-SYM, FS-14, and MOCA scores pre-to-post treatment for taVNS versus sham stimulation. (I– P) PDQ-39, PDSS-2, SCOPA-AUT, HAMA, HAMD, PAC-SYM, FS-14, and MOCA scale scores pre and post taVNS treatment. (Q–X) PDQ-39, PDSS-2, SCOPA-AUT, HAMA, HAMD, PAC-SYM, FS-14, and MOCA scale scores pre and post sham stimulation. ∗, p < 0.05; p < 0.01; ∗∗∗, p < 0.001.

Intragroup analysis revealed significant decreases in PDQ-39 (p < 0.001, Fig. 4I), PDSS-2 (p < 0.001, Fig. 4J), SCOPA-AUT (p < 0.01, Fig. 4K), HAMA (p < 0.01, Fig. 4L), HAMD (p < 0.05, Fig. 4M) and PAC-SYM scores (p < 0.001, Fig. 4N) scores in the taVNS cohort from baseline to post-treatment, with no notable change in FS-14 (p = 0.0706, Fig. 4O) or MOCA (p = 0.0693, Fig. 4P) scores. For the sham group, significant improvements were confined to PDSS-2 (p < 0.05, Fig. 4R), HAMA (p < 0.01, Fig. 4T), and PAC-SYM (p < 0.05, Fig. 4V) scores, whereas PDQ-39 (p = 0.6712, Fig. 4Q), SCOPA-AUT (p > 0.9999, Fig. 4S), HAMD (p = 0.1875, Fig. 4U), FS-14 (p = 0.0828, Fig. 4W), and MOCA(p = 0.0896, Fig. 4X) scores remained unchanged.

taVNS-induced increases in serum acetylcholine and correlation with motor symptoms

To evaluate the effects of taVNS on the vagus nerve, blood samples were collected from participants for acetylcholine analysis after three weeks of treatment. The results indicated that taVNS significantly increased serum acetylcholine levels compared with pre-treatment levels (p < 0.001, Fig. 5A). In contrast, no statistically significant difference in serum acetylcholine levels was observed in the sham stimulation group between pre -and post-treatment (p = 0.7841, Fig. 5B). Following taVNS treatment, the elevation in acetylcholine levels was significantly greater compared to the sham stimulation group (p < 0.01, Fig. 5C).

Fig. 5.

Alterations in serum acetylcholine levels and correlation analysis after taVNS. (A–B) Serum acetylcholine level in taVNS and sham stimulation groups. (C) The treatment-induced changes in acetylcholine level between taVNS and sham stimulation. (D) Correlation of taVNS-induced improvement in motor symptoms with taVNS-induced release of acetylcholine in PD patients. ∗∗∗: p < 0.001, (pre-treatment vs. post-treatment); ∗∗: p < 0.01, (pre-treatment vs. post-treatment).

In the taVNS group, we investigated the correlation between the treatment-induced changes in the MDS-UPDRS Ⅲ score and serum acetylcholine levels. The reduction in motor symptoms resulting from taVNS treatment was significantly correlated with the increase in serum acetylcholine levels as determined by Pearson correlation analysis (r = 0.4559, p = 0.0378; Fig. 5D). In addition, we examined the correlations between changes in serum acetylcholine and changes in non-motor symptoms resulting from taVNS treatment, including PAC-SYM, PDQ-39, PDSS-2, SCOPA-AUT, FS-14, HAMA, HAMD, and MOCA scores. No significant correlations were observed for any of these outcomes (Supplementary Table 1).

taVNS induced changes in glutamate levels in the striatum and thalamus

Fig. 6A displays GluCEST images of representative regions in the striatum and thalamus. In these images, increased glutamate levels correspond to a redder appearance. Intergroup comparisons indicated statistically significant decreases in glutamate levels following taVNS treatment, as opposed to sham stimulation, in both the striatum and thalamus on the onset side (p < 0.001 and p < 0.05, respectively), as well as in the striatum on the contralateral side (p < 0.01). However, no significant differences were found in glutamate levels within the thalamus (Fig. 6B). taVNS significantly decreased glutamate levels in the striatum (p < 0.001 and p < 0.01, respectively) and thalamus (p < 0.01 and p < 0.05, respectively) on both the onset and contralateral sides, compared with pre-treatment levels (Fig. 6C). No significant difference was observed in glutamate levels in the sham stimulation group between pre- and post-stimulation (Fig. 6D).

Fig. 6.

Analysis of glutamate levels changes pre-and post-treatment in the taVNS and sham stimulation groups. GluCEST images in participants, respectively. (A) a-b, ROI of the TaVNS striatum and thalamus. c-d ROI of the sham stimulation group striatum and thalamus. (B) The alteration of glutamate levels changes post-and pre-treatment on the onset side and contralateral side between the taVNS and sham stimulation groups. (C) The alteration in glutamate levels on the bilateral striatum and thalamus in the taVNS. (D) The alteration in glutamate levels on the bilateral striatum and thalamus in the sham stimulation groups. ∗, p < 0.05; ∗∗, p < 0.01; ∗∗∗, p < 0.001.

taVNS induced changes in BOLD signals in the brain

As illustrated in Fig. 7, an increase in regional homogeneity (ReHo) values in the rolandic operculum was observed following taVNS compared to the sham stimulation group (FWE corrected, p < 0.001). The elevated region is represented in red in the figure. No changes in ReHo values were detected in the sham stimulation group.

Fig. 7.

Analysis of BOLD signals changes post-treatment in the taVNS. BOLD images in participants, respectively. The red points represent regions of the brain that have increased Reho values.

taVNS induced changes in FA values in the brain

Post-taVNS whole-brain DTI analysis showed significantly increased FA (a biomarker of white matter integrity) values in the left hippocampal cingulum (CGH) (p < 0.05) and the right inferior longitudinal fasciculus (ILF) (p < 0.05). No significant changes were noted in other brain regions (Fig. 8, Table 2). Additionally, no changes in FA values were observed in the sham stimulation group.

Fig. 8.

Analysis of DTI changes post-treatment in the taVNS. White matter regions (red points) with significantly increasing FA values in the left CGH and the right ILF. (a–b) Sagittal and coronal images of right ILF. (c–d) Sagittal and coronal images of left CGH. Green indicates the skeleton of white matter fiber bundle, and red indicates the area with increased FA values.

Table 2.

Brain regions with increased FA values post-treatment of taVNS.

Brain regions	Cluster size	Cluster centroid (MNI)			p value
		X	Y	Z
Left hippocampal cingulum	60	−24	−23	−22	0.031a
Right inferior longitudinal fasciculus	15	44	−2	−23	0.039a

^ap < 0.05 (pre-treatment vs. post-treatment).

Adverse events

In the taVNS group (n = 25 at baseline), five participants reported adverse events, including nausea (n = 2, grade 1), headache (n = 1, grade 2), diarrhea (n = 1, grade 1), and pain in the cymba conchae area of the skin (n = 1, grade 1) In the sham stimulation group (n = 24 at baseline), one participant reported pain at the sham point (grade 1). There was no significant difference in the rates of adverse events between the taVNS and the sham stimulation groups (Table 3). In the taVNS group, two patients withdrew due to adverse events, specifically headache and diarrhea. In the sham group, there were no withdrawals due to skin pain. All reported adverse events in this study were classified as mild to moderate, with no severe or serious adverse events associated with the treatment.

Table 3.

Individual adverse events and severity.

Adverse Event	taVNS	Sham Stimulation	CTCAE Grade	p
	Number (Percent)
Nausea	2 (8.00 %)	0 (0 %)	1	0.49
Headache	1 (4.00 %)	0 (0 %)	2	>0.9999
Diarrhea	1 (4.00 %)	0 (0 %)	1	>0.9999
Pain at the stimulated skin site	1 (4.00 %)	1 (4.16 %)	1	>0.9999
Total	5 (20.00 %)	1 (4.16 %)	/	0.189

Discussion

This prospective, randomized, single-blind, placebo-controlled clinical trial demonstrated the beneficial effects of 3 weeks of taVNS on both motor and non-motor symptoms in PD patients. Mechanistically, taVNS decreased glutamate levels in the striatum and thalamus and increased ReHo values in rolandic operculum. TaVNS therapy induced significantly higher FA values in the left CGH and right ILF. In addition, the taVNS-induced improvement in motor symptoms was correlated with the enhanced release of systemic acetylcholine, suggesting an autonomic mechanism. Moreover, taVNS was found safe with no major adverse events during the 3-week treatment.

To the best of our knowledge, this was the first home-based taVNS study demonstrating improvements in both motor and non-motor symptoms in patients with PD. The single-blind design, combined with the use of independent evaluators and blinded participants, was instrumental in minimizing experimenter bias in clinical trials. Additionally, we employed various objective measures, including neuroimaging parameters and serum biomarkers, to rigorously evaluate treatment efficacy.

Previous studies have shown that paired associative stimulation combined with earlobe “sham VNS” can increase motor-evoked potential amplitudes [51], and other research has demonstrated that earlobe stimulation can activate brain regions similar to those activated by VNS [52]. These findings suggest that the earlobe may not be an ideal sham site in taVNS experiments. To minimize the potential influence of stimulating auricular vagal fibers, we therefore selected the inner arm as the sham stimulation location.

In intragroup comparisons, taVNS resulted in significantly improved scores for motor symptoms as assessed by the MDS-UPDRS Ⅲ, quality of life as measured by the PDQ-39, sleep quality as indicated by the PDSS-2, autonomic function via the SCOPA-AUT, anxiety levels using the HAMA, and constipation severity with the PAC-SYM. Compared to sham stimulation, taVNS showed significant enhancements in motor symptoms, quality of life, sleep quality, and autonomic function.

A previous clinical trial involving ten electrical stimulation sessions on the tragus over two weeks reported a slight decrease in MDS-UPDRS Ⅲ scores (2.00 ± 5.60), though this did not reach statistical significance [12]. Another study found that electrical stimulation at auricular acupuncture points associated with the vagus nerve improved cognitive performance in patients with mild cognitive impairment [53]. In contrast to these earlier studies, our research involved electrical stimulation performed via the cymba concha, which is fully 100 % innervated by vagal afferent nerves, with therapy administered twice daily at home by the participants.

Although several studies have examined the potential therapeutic effects of VNS or taVNS in PD and have reported findings related to motor function, mood regulation, and neurophysiological measures, the overall results remain inconsistent [[11], [12], [13], [14], [15], [16]]. Many taVNS trials have been limited by short intervention durations, substantial variability in stimulation parameters, heterogeneity in stimulation sites, and the use of diverse clinical endpoints. These methodological differences likely contribute to the variability observed across studies. A consistent observation across existing research is that stimulation delivered to auricular regions with denser vagal afferent innervation is more likely to induce measurable neurophysiological effects. The observed improvements with taVNS in this study could likely be attributed to several factors: 1) twice daily treatment instead of a few times weekly; 2) targeted stimulation of the cymba concha or vagal afferent nerve; 3) a more precise selection of stimulation parameters that were optimized for treating inflammation [35]. However, when compared to sham stimulation, taVNS did not exhibit significant effects on anxiety, depression, constipation, fatigue, or cognitive behaviors. This lack of effect might be attributed to the relatively mild nature of these symptoms and small sample size as well as the choice of stimulation parameters. For instance, taVNS utilizing the same device but different parameters (25 Hz, 2s-on and 3s-off) has been shown to improve constipation in patients with irritable bowel syndrome [33].

There remains ongoing debate regarding the minimal clinically important difference (MCID) for the MDS-UPDRS III. Sánchez-Ferro et al. determined that in early PD patients, the MCID for MDS-UPDRS Ⅲ was 5, based on evaluations of the response to dopaminergic agents in 31 newly diagnosed patients over six months [54]. Conversely, Horvath et al. reported the MCID as 3.25, derived from paired investigations of MDS-UPDRS scores and Clinician-reported Global Impression Improvement scales in 260 PD patients [32]. In the present study, taVNS demonstrated a 6.10 point reduction in the MDS-UPDRS III scores compared to baseline, suggesting a clinically meaningful improvement.

The primary enhancements observed in motor symptoms included changes of 1.71 points for bradykinesia (MDS-UPDRS items 3.4–3.8, 3.14), 2.05 points for tremor (MDS-UPDRS items 3.15–3.18), and 1.67 points for rigidity (MDS-UPDRS item 3.3). In a prior phase 3 study of rasagiline monotherapy in Japanese patients with early PD, a reduction of 4.47 points in MDS-UPDRS III scores was reported after 26 weeks, with changes of 1.69 for bradykinesia, 1.20 for tremor, and 1.00 for rigidity [55]. Additionally, a trial of globus pallidus revealed an average improvement of 6.0 points in the MDS-UPDRS III score in the off-medication state at three months post-treatment, although specific subscore changes were not detailed [56]. This suggests that taVNS may facilitate a more pronounced clinically meaningful improvement in a shorter time frame, particularly in addressing bradykinesia, tremor, and rigidity.

In our study, we observed a significant decrease in glutamate levels after taVNS therapy compared to the sham stimulation treatment as assessed via GluCEST analysis. Concurrently, patients' motor function improved steadily as glutamate levels declined, suggesting that taVNS may mitigate glutamate excitotoxicity. The association between decreased glutamate levels in the striatum and thalamus and improved motor function suggests a potential mechanism through which taVNS could alleviate motor symptoms in PD by modulating central glutamatergic transmission. Importantly, no prior study has applied GluCEST to investigate glutamate metabolism in the brains of PD patients.

Existing BOLD-fMRI studies on taVNS are limited and primarily focus on acute changes that occur during concurrent taVNS administration and BOLD-fMRI acquisition [57]. In this study, we investigated changes in brain function following 3 weeks of taVNS treatment. A previous BOLD-fMRI study on altered brain function in PD reported significant changes in ReHo values across both motor and non-motor areas [58]. In contrast, we found a significant increase in ReHo values in the Rolandic operculum after 3 weeks of stimulation. The Rolandic operculum plays a complex role in sensory, motor, autonomic, and cognitive processing [59]. It is also involved in the gustatory and visceral sensory systems, functioning alongside the cingulate-operculum network, which is connected to the gastrointestinal tract via the brain-gut axis [60]. The observed increase in ReHo values in this region indicates enhanced neuronal activity, suggesting greater coherence and centrality of activity. This specific enhancement likely reflects a targeted neuromodulatory effect of taVNS, the mechanism of which may involve the restoration or optimization of functional integration within cortical networks that are prone to disruption in PD.

CGH serves as a crucial cortico-limbic relay region, originating from the presubiculum of the hippocampus. It is instrumental in memory consolidation, spatial navigation, and closely associated with the progression of neurodegenerative diseases [61]. The ILF is a large white matter tract that facilitates bidirectional connectivity between the occipital cortex and temporal lobe [62]. FA values are considered indices of white matter integrity; alterations in this metric can be attributed to neuroinflammation, myelin loss, and other factors [63]. In our study, taVNS significantly increased FA values in the left CGH and ILF compared to the sham-stimulated group. These findings suggest that taVNS may exert a neuroprotective effect on the CGH and ILF in patients with PD. We utilized different imaging sequences to analyze changes in brain function following 3 weeks of taVNS treatment; however, the relationship between taVNS-induced alterations in the Rolandic operculum, CGH, and ILF remains unclear.

Stimulation of the vagus nerve has been shown to not only alleviate peripheral inflammation but also reduce central nervous system inflammation through the α7 nicotinic acetylcholine receptor [64]. In a rat model of PD, vagus nerve stimulation improved motor function, increased the expression of tyrosine hydroxylase (TH), and decreased the expression of glial fibrillary acidic protein (GFAP) and ionized calcium binding adaptor molecule 1 (Iba-1), both of which are markers of neuroinflammation [65]. In our study, the serum acetylcholine level was significantly higher in the taVNS group compared to the sham group, indicating enhanced vagus nerve activity. Interestingly, Pearson correlation analysis revealed a link between the taVNS-induced improvement in motor symptoms and the increase in serum acetylcholine levels. This association suggests that the cholinergic anti-inflammatory pathway may contribute to the therapeutic effects of taVNS in PD, potentially by reducing neuroinflammation and promoting dopaminergic neuron survival. These findings extend prior taVNS research demonstrating autonomic and anti-inflammatory benefits in disorders of gut–brain interaction [33,34].

Although taVNS has rarely been used for treating PD, it has been applied in various other clinical contexts with an excellent overall safety profile. A recent meta-analysis on taVNS reported an incidence of adverse events at 12.8 per 100,000 person-minutes-days of stimulation, with the most commonly reported events including ear pain, headache, and tingling [66,67]. In the present study, we documented a total of six adverse reactions, including nausea, headache, diarrhea, and skin pain. All reported adverse reactions were mild to moderate, with no serious events occurring.

There were several limitations in the present study. First, the follow-up period was relatively short. The trial was designed to assess the short-term efficacy of home-based taVNS rather than its long-term effects, and the current findings therefore only support potential short-term improvements. Longer follow-up studies are needed to determine whether these benefits persist. Second, the study was conducted with a relatively small sample size, and larger double-blind trials will be required to more rigorously establish the clinical efficacy of taVNS in PD. Third, stimulation intensity was individually adjusted to ensure tolerability but was not systematically recorded, limiting assessment of potential dose-response relationships. Fourth, although our multimodal MRI and serum analyses suggested potential mechanisms involving autonomic regulation, glutamatergic modulation, and functional brain changes, GluCEST, BOLD-fMRI measures, and serum ACh are not validated biomarkers of therapeutic efficacy in PD and should be interpreted as exploratory outcomes. Finally, taVNS may influence neurotransmitter systems beyond acetylcholine, including norepinephrine, dopamine, serotonin, and GABA [[19], [20], [21], [22]], which are relevant to PD Future studies should further investigate these pathways and include biomarker validation and mechanistic experiments.

In conclusion, twice daily taVNS with the special set of parameters administered at home via a wearable device is well-tolerated and alleviates both motor and non-motor symptoms in patients with PD, possibly mediated by enhanced vagal activity and improvements in brain microstructure and functions. Pivotal clinical studies are warranted to establish taVNS as a non-invasive adjunctive therapy for PD in the future.

Author contributions

All authors have been involved with this study since inception.

Study design: Rui Wang, Xianglin Li, Hongcai Wang and Jiande DZ Chen.

Conduction of clinical trial: Rui Wang, Miaomiao Liu, Quanyuan Liu, Yifei You, Xu Li, Yan Chen, Yuwei Liu, and Jing Wang.

Data analysis; Rui Wang, Miaomiao Liu, Xianzhi Wang, Zhijie Yin, Xianglin Li, and Hongcai Wang.

Manuscript preparation and data interpretation: Rui Wang, Miaomiao Liu, Hongcai Wang, and Jiande DZ Chen.

Data availability

Data may be shared at the request of any qualified investigator for the purposes of replicating the procedures and results, provided that no patient information is disclosed.

Financial disclosures

All authors report no biomedical financial interests or potential conflicts of interest.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Abbreviations

ABVN

auricular branch of the vagus nerve

BOLD-fMRI

Blood oxygen-level dependent functional MRI

CGH

hippocampal cingulum

DBS

deep brain stimulation

DTI

Diffusion Tensor imaging

FA

fractional anisotropy

GluCEST

glutamate chemical exchange saturation transfer

¹H MRS

proton magnetic resonance spectroscopy

H–Y

Hoehn and Yahr staging

ILF

inferior longitudinal fasciculus

MDS-UPDRS III

Movement Disorder Society-Unified Parkinson's Disease Rating Scale Part III

MRI

magnetic resonance imaging

PD

Parkinson's disease

taVNS

transcutaneous auricular vagus nerve stimulation

Appendix A. Supplementary data

The following is the Supplementary data to this article: