Subthalamic deep brain stimulation in isolated generalised or segmental dystonia (RELAX Study): a multicentre, randomised, double-blind, controlled trial

Published 1 week ago• 26 minute read

The safety and effectiveness of deep brain stimulation of the subthalamic nucleus (STN-DBS) for the treatment of dystonia lack high-level evidence-based medical support. This study aimed to clarify the efficacy and safety of STN-DBS and perform a post hoc analysis comparing it with DBS of the internal globus pallidus (GPi-DBS).

This multicentre, randomised, double-blind, controlled trial included 67 patients aged 6–60 years old diagnosed with genetic or idiopathic isolated generalised or segmental dystonia. They were enrolled from seven hospitals in China and randomly assigned to undergo GPi-DBS or STN-DBS. After surgery, they were randomised to receive either neurostimulation or sham stimulation for 3 months. At the 3-month follow-up, neurostimulation was also initiated in the sham stimulation group, and all patients were followed up for more than 3 years after treatment. The primary outcome was the Burke-Fahn-Marsden Dystonia Rating Scale movement (BFMDRS-M) score.

In the STN group, the neurostimulation subgroup exhibited significant improvement (p<0.001), which is also superior to the sham stimulation subgroup (p=0.028) at 3-month follow-up. At the 6-month and >3-year follow-ups, all patients receiving STN-DBS showed a significant improvement in BFMDRS-M scores (p<0.001). Further post hoc analysis revealed that both STN-DBS and GPi-DBS could produce similar therapeutic effects on motor symptoms (P_{6 months}=0.865, P_{>3 years}=0.905). There were no ongoing serious adverse events throughout the study.

For isolated generalised and segmental dystonia patients, the STN is a selectable DBS target with ensured safety and efficacy. STN-DBS and GPi-DBS may achieve comparable therapeutic effects on motor symptoms.

Data are available on reasonable request. All authors will honour legitimate requests for our clinical trial data from qualified researchers. Data will be shared with external researchers whose proposed use of the data has been approved. Before data are released, the researcher(s) must sign a data sharing agreement, after which the deidentified and anonymised datasets can be accessed from the corresponding authors.

http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

WHAT IS ALREADY KNOWN ON THIS TOPIC

WHAT THIS STUDY ADDS

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

Dystonia is a movement disorder characterised by abnormal movements and/or postures. In 2013, a new classification system reclassified dystonia, replacing primary dystonia with genetic or idiopathic isolated dystonia.1 For patients with generalised or segmental dystonia, drug therapy is often unsatisfactory, and botulinum toxin injection can only be used in restricted areas. Deep brain stimulation (DBS) has been shown to effectively treat dystonia that is refractory to medication.2 Several high-level evidence-based medical studies have demonstrated that DBS of the internal globus pallidus (GPi-DBS) improved motor symptoms and quality of life in primary generalised and/or segmental dystonia, with long-term safety and efficacy.3–6 However, this target is associated with high stimulation current, difficult target localisation and minimal efficacy in some dystonia patients.4 7

Other stimulation targets within the basal ganglia–cerebello-thalamo-cortical circuit have also been attempted, most notably in the subthalamic nucleus (STN).8 In advanced Parkinson’s disease (PD), a randomised controlled trial (RCT) showed no differences in primary outcomes between STN-DBS and GPi-DBS.9 Several retrospective or single-arm prospective single-centre studies have demonstrated the safety and efficacy of STN-DBS in isolated dystonia.10–12 However, the ability to draw conclusions from these studies is also limited by lack of blinding, placebo effects and small sample sizes.10–12 Therefore, we conducted a multicentre, randomised, double-blind, sham-controlled trial aimed at clarifying the efficacy and safety of STN-DBS in isolated generalised or segmental dystonia and performed a post hoc analysis comparing it with GPi-DBS.

This study was a randomised, double-blind, sham-controlled trial conducted at seven medical centres in China, with all patients receiving neurostimulation exceeding 3 years. Main inclusion criteria were clinically diagnosed genetic or idiopathic isolated generalised or segmental dystonia; age 6–60 years; unsatisfactory response to optimal pharmacological treatment. Main exclusion criteria were cognitive impairment; severe depression or other serious mental illness; history of traumatic brain injury, tumour or severe cerebrovascular disease; severe brain atrophy; pregnancy; other medical comorbidities that cannot tolerate surgery. The detailed inclusion and exclusion criteria were shown in the study protocol.

Patients were screened for eligibility and videotaped with the protocol by Comella et al13 at baseline (within 3 weeks before surgery). Baseline assessment included clinical evaluation of dystonia using the Burke-Fahn-Marsden Dystonia Rating Scale (BFMDRS), the primary outcome indicator. The secondary outcome indicators were the Medical Outcomes Study 36-item Short-Form General Health Survey (SF-36), the Beck Depression Inventory (BDI), the Mini–Mental State Examination (MMSE), the time up and go test (TUG) and a Visual Analogue Scale (VAS) (range 0–10). Among them, higher BFMDRS score, VAS, BDI and TUG indicate worse condition, while higher SF-36 and MMSE indicate better condition. Before surgery, patients were randomly assigned to receive either GPi-DBS or STN-DBS using the Interactive Web Response System.

Quadripolar electrodes (L301, length 1.5 mm, spacing 0.5 mm, PINS, Beijing, China) were implanted bilaterally into the STN. Quadripolar electrodes (L302, length 1.5 mm, spacing 1.5 mm, PINS, Beijing, China) were implanted bilaterally into the GPi. In the GPi group, the electrodes were implanted bilaterally in the posteroventrolateral portion; the initial implantation target was 19–21 mm lateral to, 1–3 mm anterior to and 2–4 mm below the midcommissural point. In the STN group, the initial target was 11–13 mm lateral to, 2–4 mm posterior to and 3–5 mm below the mid-commissural point. Individualised T2WI MRI images and intraoperative microelectrode recordings were used for precise localisation of the implantation site. Intraoperative stimulation was done in patients undergoing local anaesthesia. The electrodes were connected to a fully implanted neurostimulator (G104, PINS) by the extension kit (E202, PINS) during the same surgery. Post-treatment CT imaging was performed and fused with preoperative MRI scans to confirm electrode localisation. Patients were randomised in a 1:1 ratio to receive either neurostimulation (group A) or sham stimulation (group B) within the first 3 months. Randomisation was conducted by trial investigators at Peking University Clinical Research Institute.

At 2–4 weeks post-treatment, thresholds inducing acute adverse effects were tested for all patients. In the stimulation group (group A), neurostimulation parameter setting was based on post-treatment imaging data and experience of programmable physicians. In the sham stimulation group, the neurostimulator was turned off after the threshold testing was completed.

At 3 months post-treatment, patients were reassessed using the baseline assessment tools, and neurostimulation was initiated in the sham stimulation group. The same assessments were conducted at 6 months and 3 years after surgery. Group assignments remained blinded to patients until the 6-month post-treatment assessment. Two independent dystonia experts rated the BFMDRS movement (BFMDRS-M) score while watching video recordings, and the average of these two scores was used as the final score. Evaluators were unaware of the group assignments and assessment time until 3 years after surgery. Clinical follow-up and programming could be arranged at any time requested by the treating physician or the patient.

Adverse events were classified as new symptoms or worsening of pre-existing symptoms. Serious adverse events were defined as death, disability, life-threatening conditions, hospital admission, prolonged hospitalisation or events requiring intervention to prevent these outcomes.

The primary outcome was BFMDRS-M. The secondary outcomes included improvement rate of BFMDRS-M, BFMDRS disability score (BFMDRS-D), SF-36, BDI, MMSE, TUG and VAS for dystonia and pain severity.

This study focused on the difference in the change of BFMDRS-M score 3 months post-treatment between the neurostimulation group and the sham stimulation group. According to the study by Kupsch et al5 and Cao et al,14 this study was designed with an 80% test power (1−β), a bilateral test level of 5% (α) and allowing for an overall dropout rate of 10%. Seventy patients are needed, including 32 patients for the STN-DBS group and 38 patients for the GPi-DBS group. A p<0.05 in the intergroup comparison of the main effectiveness parameters was considered statistically significant, indicating the trial results were qualified.

Statistical analyses were performed by statisticians from Peking University Clinical Research Institute using the SAS statistical package, V.9.4 (SAS Institute). Outcomes at 3 months, 6 months and 3 years post-treatment were compared with baseline using the paired t-test or Wilcoxon signed-rank test for matched pairs, as appropriate. Differences between stimulation or target groups were analysed with t-tests or Mann-Whitney U tests. The normality of continuous variables was assessed using the Shapiro-Wilk test. Normally distributed data were presented as mean±SD, and between-group comparisons were performed using the independent samples t-test. These results were reported as the mean difference and its 95% CIs. For skewed data, the median and IQR were used for descriptive statistics, and between-group comparisons were conducted using the Wilcoxon rank-sum test. These results were reported as the Hodges-Lehmann estimated median difference and its 95% CI. Tables uniformly display the mean, SD, mean difference and their corresponding 95% CI. Categorical data were compared using the χ² test or Fisher’s exact test. All statistical tests were two-tailed, and a p≤0.05 was considered statistically significant. The results were analysed without imputation for missing values.

The primary outcome BFMDRS-M score underwent superiority testing. Under α=0.025 (one-sided test) and β=0.2, the change of BFMDRS-M score 3 months post-treatment was assessed in group A and group B. The absolute difference in score changes between the two groups was calculated. If the lower limit of its 95% CI was greater than 0, it was considered to establish superiority.

This trial is registered with ClinicalTrials.gov, number NCT 03017586.

The funder of the study had no role in study design, data collection, data analysis, data interpretation or writing of the report.

The study enrolled 72 patients diagnosed with isolated dystonia from 7 medical centres in China between January 2018 and December 2019. After further screening, 67 patients were randomly assigned to different groups and were followed up for over 3 years (figure 1). 35 participants were in the neurostimulation group (group A: STN n=17, GPi n=18) and 32 participants in the sham stimulation group (group B: STN n=13, GPi n=19). At the 3-month postoperative follow-up, assessment data for one patient were missing because the patient refused to be videotaped, thus rendering the BFMDRS-M score unattainable. The remaining patients all have complete follow-up data. The baseline clinical characteristics and classification of the enrolled patients are summarised in table 1.

Figure 1

Enrolment of patients and treatment assignment. *Assessment data for one patient were missing. The patient refused to be videotaped, thus rendering the BFMDRS-M score unattainable. BFMDRS-M, Burke-Fahn-Marsden Dystonia Rating Scale movement; GPi, internal globus pallidus; MMSE, Mini–Mental State Examination;SAE, serious adverse events; STN, subthalamic nucleus.

Table 1

Baseline characteristics

Of the 67 enrolled patients, 57 were diagnosed with generalised dystonia and 10 with segmental dystonia. Of these, 37 patients had idiopathic dystonia, and 30 patients had inherited dystonia, with 18 patients carrying TOR1A gene mutations (DYT1 dystonia) (online supplemental table 1). Before DBS surgery, 77.61% (52/67) of the patients were receiving oral medication or botulinum toxin type A, but the therapeutic effect was not satisfactory. Specific drug use has been described in table 1.

At the 3-month post-treatment follow-up, researchers compared the BFMDRS-M scores of 30 patients treated with STN-DBS (figure 2). The results showed improvements in both group A (mean±SD: 17.76±16.03; median: 13.75, IQR (7.5, 22.75)) and group B (mean±SD: 6.12±8.97; median: 4, IQR (−0.25, 11.31)), but a significant difference in the extent of improvement was observed, with group A showing more satisfactory progress (mean difference: 11.64, 95% CI (1.12, 22.16); median difference 8.5, 95% CI (1.25, 18.25); W=151.5, p=0.028), particularly in the treatment effects. on Facial (mean difference 1.75, 95% CI (0.43, 3.07); t=2.72, p=0.012) and axial dystonia (mean difference: 3.57, 95% CI (0.04, 7.1); median difference 0.75, 95% CI (0.25, 5.75); W=153, p=0.023). At this follow-up point, patients who received GPi-DBS also showed a similar improvement trend in BFMDRS-M scores. The mean improvement in group A was 16.90 (SD: 16.01; median: 10.62, IQR (5.5, 28.81)), while in group B, it was 1 (SD: 8.5; median: 1, IQR (0, 4.88)), with a significant difference between the two groups (mean difference: 15.4, 95% CI (6.66, 24.14); median difference 10.25, 95% CI (4, 25.5); W=290.5, p<0.001). At the 6-month post-treatment follow-up, group A had been continuously stimulated for 6 months, while group B had only been stimulated for 3 months. We found no significant difference in movement score improvement between the two groups in 30 patients treated with STN-DBS compared with baseline assessments (mean±SD: 22.26±15.57 points vs 19.94±17.88 points, mean difference 2.32, 95% CI (−10.2, 14.85), t=0.38, p=0.707). However, among the 35 patients treated with GPi-DBS, the improvement was more significant in Group A (mean difference: 12.22, 95% CI (2.27, 22.17); median difference: 10.13, 95% CI (0.25, 19.25), W=236, p=0.048). Specific data are presented in table 2.

Figure 2

Changes of BFMDRS-M and BFMDRS-D scores in different stages. (A, B) Changes in the neurostimulation group and sham stimulation group of STN-DBS were compared at the 3 months and the 6 months follow-up. (C, D) Changes in STN-DBS and GPi-DBS at the 6 months and the >3 years follow-up were compared by post hoc analysis. The bars in this figure represent the mean and the whiskers the SEM. BFMDRS-D, Burke-Fahn-Marsden Dystonia Rating Scale Disability Score; BFMDRS-M, Burke-Fahn-Marsden Dystonia Rating Scale movement; DBS, deep brain stimulation; GPi, internal globus pallidus; STN, subthalamic nucleus.

Table 2

BFMDRS movement scores of the 3 months and 6 months follow-up

Compared with baseline, at the 3-month post-treatment follow-up,group A showed significant improvements in the SF-36 physical health and mental health subscales, while group B showed no significant changes. Similarly, group A showed significant improvements in the BDI and VAS scores, while group B showed no significant changes.

At the 6-month post-treatment follow-up, both group A and group B showed significant improvements in the SF-36 total score and VAS scores compared with baseline, but there was no significant difference between the two groups in the extent of improvement. Group A’s depression level, as assessed by the BDI score, continued to improve significantly (mean difference: 5.59; median difference: 5.50, 95% CI (2.00, 9.50); W=120.5, p=0.007), while group B showed no significant change (mean difference: 2.0; median difference 2.42, 95% CI (−3.00, 8.00); W=51, p=0.344). This trend was consistent with the findings at the 3-month follow-up. The detailed data for the above secondary outcome indicators are recorded in table 3.

Table 3

Secondary outcome measures of the 3 months and 6 months follow-up

Both the primary and secondary outcome indicators confirmed the short-term efficacy of STN as a treatment target for isolated dystonia. The improvements in primary and secondary outcomes for patients in the STN-DBS and GPi-DBS groups at the 3-month follow-up are summarised in tables 2 and 3 and online supplemental table 2.

Follow-up assessments were conducted at >3 years postoperatively for all 67 enrolled patients. Significant improvements were observed in the BFMDRS-M scores (total and subscores, table 4) (mean difference: 27.44; median difference 24.88, 95% CI (17.75, 35.00); W=462, p<0.001) and the BFMDRS-D scores (online supplemental table 3) for all patients receiving STN-DBS treatment compared with baseline (mean difference: 4.87; median difference 5.25, 95% CI (4.00, 6.50); W=400.5, p<0.001). Specifically, the total BFMDRS-M score improved by 27.44±21.23 points (improvement rate of 63.37%±30.35%) at the >3 years follow-up. Patients in the GPi-DBS group, serving as a control, also exhibited significant improvements during long-term follow-up. As for the response rate, we defined patients with a BFMDRS-M score improvement rate of less than 25% as non-responders and found that among the 67 enrolled patients, 29 had a positive response at the 3-month follow-up (43.28%), 53 at the 6-month follow-up (79.10%) and 60 at the >3 years follow-up (89.55%) (online supplemental tables 4 and 5).

Table 4

The 6 months and >3 years follow-up assessments (STN vs GPi)

Long-term treatment with STN-DBS led to varying degrees of improvement across other secondary outcomes. At the >3 years follow-up, the SF-36 total score and patient-rated dystonia severity (VAS) continued to show significant improvements from baseline. Although there was a trend towards improvement in the mental health component of the SF-36, it was not significantly different from baseline (mean difference −7.86, 95% CI (−16.63, 0.90); t=−1.84, p=0.077). No significant changes were observed in cognitive function (MMSE). Similar to the STN-DBS group, patients in the GPi-DBS group showed significant improvements in most secondary outcomes at both follow-ups, except for pain severity and cognitive function. The specific data assessed at each time point are detailed in table 4.

This study validated the safety and efficacy of STN-DBS while also enrolling patients for GPi-DBS treatment in a randomised, double-blind manner, conducting a post hoc analysis on the safety and efficacy of 30 STN-DBS and 37 GPi-DBS treated patients (figure 2). In terms of improvement rates, there was no significant difference between the STN and GPi groups in the improvement of BFMDRS-M scores (table 4) (6 months follow-up: mean difference: 1.17, 95% CI (−14.04, 16.39); median difference: 0.45, 95% CI (−13.98, 15.24); W=560.5, p=0.945; >3 years follow-up: mean difference: 1.2, 95% CI (−15.96,18.37); median difference: 0.03, 95% CI (−12.91, 12.23); W=555.5, p=0.995). Furthermore, significant improvements were observed in the BFMDRS-D scores for both STN and GPi groups at the long-term follow-up, with no difference between groups (online supplemental table 3, >3 years follow-up: mean difference: 0.79, 95% CI (−1.17, 2.74); median difference 0.5, 95% CI (−1.5, 2); W=568, p=0.869).

All 67 enrolled patients showed improvements in the SF-36 total score at both 6 month and >3 years follow-ups, with no significant difference between the STN and GPi groups at the 6 months follow-up (mean difference: 0.04, 95% CI (−11.93, 12.02); t=0.007, p=0.994). Analysis of patient depression status revealed improvements at the 6 months follow-up in both the STN and GPi groups without intergroup differences, but at the>3 years follow-up, the improvement in the BDI scale was not significant in the STN group (mean difference: 2.13; median difference 2.66, 95% CI (−1.50, 6.50); W=257.5, p=0.214). In terms of cognition, there was no difference between the groups at the >3 years follow-up (mean difference: 0.37, 95% CI (−0.77, 1.5); median difference 0, 95% CI (−1, 1); W=577, p=0.628). Detailed data of these intergroup comparisons are also presented in table 4.

As an important indicator, we also took note of the stimulation parameters of the pulse generator. We found that GPi-DBS requires higher stimulation currents to reach a stable state compared with STN-DBS (GPi 3.17±0.84 V, STN 2.44±0.69 V, online supplemental table 6), indicating a substantially longer battery life for the former group. This finding aligns with previous RCTs conducted on PD.15

The researchers analysed the adverse events that occurred at various stages of the trial. Throughout the entire trial, there were no device-related adverse events. Two serious adverse events occurred, specifically subcutaneous infection and cerebrospinal fluid leakage, both of which were completely resolved after timely treatment. Adverse events such as fever, infections of various systems, headache and digestive system symptoms totalled 26 instances, with the majority (20 instances, 76.92%) occurring within the first 2 weeks postoperation and were alleviated with treatment.

Adverse events related to the disease/stimulation amounted to 19 instances, with 16 instances (84.21%) occurring between 2 weeks and 6 months postoperation. The most common were dyskinesia (four instances), paresthesia (three instances) and gait disorders (three instances). There was no significant difference in the overall occurrence of adverse events between the STN and GPi groups. Emotional regulation disorders, paresthesia, dysarthria, dysphagia, gait disorder and dyskinesia persisted after standardised programming, although they were less severe.

This study represents a multicentre, double-blind RCT that, for the first time globally, provides high-level evidence-based medical evidence for the treatment of isolated dystonia with STN-DBS. It also marks the first instance of a prospective trial comparing STN-DBS and GPi-DBS using post hoc analysis. The study found that at the 3-month follow-up, the neurostimulation group for STN-DBS experienced a decrease in BMFDRS-M score, with an improvement rate of 37.39%. At the 6-month follow-up, the BMFDRS-M score improvement rate was 54.24%, and at the >3 years follow-up, 63.37%. These results not only demonstrate the efficacy of the treatment but also suggest a trend that patients’ benefits might gradually increase with the extension of treatment duration. During the same period, 37 patients undergoing GPi-DBS treatment were randomly enrolled as a control group. The findings indicate that at the 3-year follow-up, the BFMDRS-M score in the GPi group decreased by 62.17%, showing no significant difference from the STN-DBS group.

The efficacy of GPi as a target for DBS treatment in isolated dystonia has been confirmed by several RCTs.3–6 Additionally, recent 10-year follow-up results show that approximately 67.5% of patients exhibited significant and stable symptom improvement.16 However, high-level evidence supporting the safety and effectiveness of STN-DBS remains scarce. Ostrem et al10 conducted a single-arm prospective study on STN-DBS, preliminarily proving the efficacy of the STN target. Subsequent research has primarily focused on comparing the STN and GPi targets. A prospective, randomised, double-blind crossover trial implanted electrodes in both the STN and GPi of patients, and after short-term and long-term self-comparisons, found that stimulating both targets did not significantly differ in improving patients’ long-term quality of life.11 17 However, the prospective studies included a small number of patients and did not undergo rigorous randomisation. Recent retrospective studies comparing the treatment effects of STN-DBS and GPi-DBS also suggested that both could achieve similar improvements in symptoms of movement disorders. Different targets may have varying therapeutic focuses. For example, GPi-DBS may be more effective for axial symptoms, while STN-DBS is more effective for appendicular symptoms.12 However, STN’s effectiveness still lacks evidence from multicentre RCTs and this study fills this gap by not only confirming the long-term safety and effectiveness of STN-DBS in treating isolated dystonia but also comparing the two targets post hoc, concluding that both targets offer similar relief from patients’ motor disorders. This provides a very important basis for the application of STN-DBS in isolated dystonia.

In the selection of DBS treatment targets for patients with isolated dystonia, clinicians undertake a comprehensive consideration. Beyond the most crucial aspect of improving motor disorder symptoms, this includes clinical indices such as cognitive function, mood and quality of life. Previous studies have shown that STN-DBS provides satisfying long-term improvements in motor symptoms, especially for facial and upper limb symptoms.18 19 Our study findings also indicate that at the 6-month follow-up, the STN group showed greater improvement in facial symptom scores than the GPi group (p=0.028), and at the 3-year follow-up, improvements in speech and swallowing disorders were better (p=0.016), aligning with previous research. Also, we analysed the symptoms by dividing them into cranial, cervical, trunk and limb regions and found that stimulation of the STN may lead to a better long-term symptom improvement in the cranial region compared with GPi stimulation, with (online supplemental table 7). However, these results may also be due to baseline differences between the two groups in these scoring items, necessitating further exploration. Previous studies have also suggested that STN-DBS may have the advantage of a faster onset of effect.12 However, since the primary outcome of this study focused on the overall therapeutic efficacy, we did not assess patients’ treatment effects at the 1-month follow-up. At the 3-month follow-up, we did not observe any differences in response dynamics. Therefore, we are currently unable to draw conclusions on this issue, and further research may be needed to clarify it in the future.

It was found that stimulation of STN may have the potential to alleviate pain symptoms.20 In studies focusing on PD, STN-DBS was found to be beneficial for patients’ mood.21 Our article also conducted a post hoc analysis on patients’ pain (VAS) and mood (BDI). The results showed that neither STN-DBS nor GPi-DBS had a significant effect on pain improvement. However, regarding the treatment effect on mood, we found that both STN-DBS and GPi-DBS improved mood at the 6-month follow-up (STN p=0.008, GPi p<0.001). At the >3 years follow-up, the improvement in mood with STN-DBS was not significant compared with baseline, whereas GPi-DBS showed a significant difference (STN p=0.214, GPi p=0.02). Studies in PD have found that cognitive function may deteriorate after STN-DBS treatment, particularly affecting verbal fluency,15 22 23 but this adverse effect is significantly less severe in dystonia.24 Our study results suggest that at both 6 months and 3 years follow-ups, no adverse effects on cognitive function were observed due to DBS treatment. Although our results are not entirely consistent with previous studies, obtaining definitive conclusions will require more specifically designed research in the future.

We analysed the medication usage of patients after entering the clinical trial. The analysis indicated that with the extension of the follow-up period, the use of medication gradually decreased in both the STN and GPi groups, with no significant differences observed between the groups (online supplemental table 8). However, due to the variety of medications involved and the difficulty in making comparisons between different drugs because of the lack of standardised dosage conversions, it is challenging to conduct a meaningful comparison. We hope that future research will better clarify this issue.

In terms of treatment-related adverse effects, although previous literature has reported a device removal rate of up to 11.4% for DBS surgery,25 no device-related adverse events occurred during our study. We believe this may be related to our strict requirements for the surgeons participating from each centre. To ensure the smooth progress of the trial and maximise patient benefit, we invited the most experienced functional neurosurgeons (with more than 10 years of neurosurgical experience) from each centre to perform the surgeries. Additionally, we strictly managed and cared for patients during the perioperative period, which may have contributed to preventing device-related complications. Most surgery-related adverse effects occurred within 2 weeks postoperation and were alleviated with timely symptomatic treatment. Like GPi-DBS, long-term stimulation of the STN target can cause disorders in articulation, gait and sensation. However, the number of cases with stimulation-related adverse reactions was limited, and the severity was mild (table 5).

A limitation of our study is that it was not designed for non-inferiority testing between STN and GPi. This is primarily because the low incidence of isolated generalised and segmental dystonia, coupled with the scarcity of prospective studies targeting STN, made it challenging to establish an ideal non-inferiority margin for relevant testing. Nevertheless, we conducted patient screening, enrolment and trials in a strictly randomised manner, designed a rigorous trial to assess the safety and efficacy of STN-DBS and conducted long-term follow-ups on patients who received STN-DBS and GPi-DBS, comparing them through post hoc analysis. Recently, the concept of the minimal clinically important difference (MCID) has been explored in the scoring of treatment effects for movement disorders.26 27 After the initiation of our study, Pintér et al28 validated the MCID for the BFMDRS score in 2020, suggesting that an improvement rate of 16.3% in the BFMDRS-M is considered clinically significant for patients with genetic and idiopathic segmental and generalised dystonia. Based on this, we compared the improvement rates of BFMDRS-M for STN-DBS and GPi-DBS at 6 months and >3 years follow-up in our study, finding the difference in improvement rates between the two groups to be 1.17% (95% CI (−14.04%, 16.39%)) and 1.2% (95% CI (−15.96%, 18.37%)), respectively. These results suggest that the likelihood of STN-DBS’s improvement rate being inferior to GPi-DBS’s to a degree that reaches the MCID is very small, both at 6 months and at >3 years follow-up.

For patients with isolated segmental and generalised dystonia, the therapeutic effect of GPi-DBS is supported by high-level evidence-based medicine, but the application of the STN target remains controversial. Our study follows the clinical trial design of the GPi target, using the same inclusion criteria, and for the first time globally confirms the long-term safety and efficacy of the STN target in the treatment of isolated segmental and generalised dystonia. Additionally, our post hoc analysis of the treatment effects of STN-DBS and GPi-DBS found similar degrees of symptom relief for movement disorders between the two targets, providing significant evidence for the selection of DBS treatment targets for isolated dystonia.

Not applicable.

This study involves human participants. The study was approved by the ethics committee of each participating centre. Ethics Committee: Ethics Committee of Drug Clinical Trials, Chinese Academy of Medical Sciences and Peking Union Hospital; Ethics Approval Number: HS2017044; Type: Meeting Review; Date: 20170517. Ethics Committee: Ethics Committee of Sir Run Run Shaw Hospital, School of Medicine, Zhejiang University; Ethics Approval Number: 20170905-8; Type: Meeting Review; Date: 20170912. Ethics Committee: Ethics Committee of Tongji Hospital of Shanghai; Ethics Approval Number: 2017-439; Type: Meeting Review; Date: 20171122. Ethics Committee: Clinical Trial Ethics Committee of Huazhong University of Science and Technology; Ethics Approval Number: (2017) Ethics Approval No. (260); Type: Meeting Review; Date: 20171025. Ethics Committee: Clinical Trial Ethics Sub-committee of West China Hospital, Sichuan University; Ethics Approval Number: 2018 Clinical Trial (Device) Approval (3) No; Type: Meeting Review; Date: 20180117. Ethics Committee: Medical Ethics Committee of Nanjing Brain Hospital; Ethics Approval Number: 2017-009-01; Type: Meeting Review; Date: 20171129. Ethics Committee: Drug Clinical Trial Ethics Committee of Qilu Hospital of Shandong University; Ethics Approval Number: 2019013; Type: Meeting Review; Date: 20190316. Participants gave informed consent to participate in the study before taking part.

We thank Beijing Pins Medical Co., the sponsors and collaborators, for supporting the study. We thank the patients and their families for making the study possible and all the investigators and doctors from seven medical centres in China.