Article Text

Presurgical structural imaging and clinical outcome in combined bed nucleus of the stria terminalis–nucleus accumbens deep brain stimulation for treatment-resistant depression
  1. Fengting Wang1,
  2. Lulin Dai1,2,
  3. Tao Wang1,
  4. Yingying Zhang1,3,
  5. Yuhan Wang1,
  6. Yijie Zhao3,4,
  7. Yixin Pan1,
  8. Liuguan Bian1,
  9. Dianyou Li1,
  10. Shikun Zhan1,
  11. Yijie Lai1,
  12. Valerie Voon3,5 and
  13. Bomin Sun1
  1. 1 Department of Neurosurgery, Shanghai Jiao Tong University Medical School Affiliated Ruijin Hospital, Shanghai, China
  2. 2 Department of Psychiatry, Zhejiang University School of Medicine Sir Run Run Shaw Hospital, Hangzhou, China
  3. 3 Fudan University Institute of Science and Technology for Brain-inspired Intelligence, Shanghai, China
  4. 4 Clinical Research Center for Mental Disorders, Shanghai Pudong New Area Mental Health Center, Shanghai, China
  5. 5 Department of Psychiatry, University of Cambridge, Cambridge, UK
  1. Correspondence to Professor Bomin Sun; sbm11224{at}; Professor Valerie Voon; vv247{at}; Dr Yijie Lai; laiyijie0901{at}


Background Structural imaging holds great potential for precise targeting and stimulation for deep brain stimulation (DBS). The anatomical information it provides may serve as potential biomarkers for predicting the efficacy of DBS in treatment-resistant depression (TRD).

Aims The primary aim is to identify preoperative imaging biomarkers that correlate with the efficacy of DBS in patients with TRD.

Methods Preoperative imaging parameters were estimated and correlated with the 6-month clinical outcome of patients with TRD receiving combined bed nucleus of the stria terminalis (BNST)–nucleus accumbens (NAc) DBS. White matter (WM) properties were extracted and compared between the response/non-response and remission/non-remission groups. Structural connectome was constructed and analysed using graph theory. Distances of the volume of activated tissue (VAT) to the main modulating tracts were also estimated to evaluate the correlations.

Results Differences in fibre bundle properties of tracts, including superior thalamic radiation and reticulospinal tract, were observed between the remission and non-remission groups. Distance of the centre of the VAT to tracts connecting the ventral tegmental area and the anterior limb of internal capsule on the left side varied between the remission and non-remission groups (p=0.010, t=3.07). The normalised clustering coefficient (γ) and the small-world property (σ) in graph analysis correlated with the symptom improvement after the correction of age.

Conclusions Presurgical structural alterations in WM tracts connecting the frontal area with subcortical regions, as well as the distance of the VAT to the modulating tracts, may influence the clinical outcome of BNST–NAc DBS. These findings provide potential imaging biomarkers for the DBS treatment for patients with TRD.

  • Depressive Disorder, Treatment-Resistant
  • Neuropsychiatry

Data availability statement

Data are available upon reasonable request.

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:

Statistics from

Request Permissions

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.


  • Structural alterations are shown to have an impact on deep brain stimulation (DBS) efficacy.


  • This is the first study demonstrating presurgical structural features that influence the outcome of patients with treatment-resistant depression (TRD) receiving combined bed nucleus of the stria terminalis (BNST)–nucleus accumbens (NAc) DBS. White matter integrity of tracts connecting the frontal area with subcortical regions and the distance of the volume of activated tissue to the ventral tegmental area-anterior limb of internal capsule tract can play a role in DBS modulation in patients with TRD.


  • This study provided potential imaging biomarkers for outcome prediction of combined BNST–NAc DBS and helped in understanding the underlying mechanisms of DBS modulation.


Major depressive disorder is a global public health concern and the leading cause of disability, with over 30% of patients experiencing treatment resistance.1 Patients with treatment-resistant depression (TRD) are refractory to conventional psychotherapeutic strategies, while deep brain stimulation (DBS) may exert long-lasting and better efficacy.2 However, various clinical trials with different DBS targets have reported substantial variabilities in individual response to DBS, with some patients achieving complete remission after DBS while others exhibiting poor response, highlighting the need for outcome prediction in patient management.2 DBS targets, including ventral capsule/ventral striatum, ventral anterior limb of the internal capsule, bed nucleus of the stria terminalis (BNST) and nucleus accumbens (NAc), are anatomically close to each other and interconnected by dense fibre tracts, the stimulation of which is reported to modulate overlapping yet differentiated pathways that may share a cooperative functionality.3 We hypothesised that combined stimulation of these targets may synergistically exert the therapeutic effect, and therefore targeted the BNST–NAc region.

Structural imaging is a prerequisite step in functional neurosurgery. It provides anatomical information required for targeting and can illustrate chronic treatment-related structural alterations. Differences in fibre properties have been shown to correlate with DBS outcomes in conditions such as Parkinson’s disease and obsessive–compulsive disorder.4 5 In the treatment of TRD, frontal volumes and frontal white matter (WM) structure were reported to correlate with the DBS outcome of the superolateral medial forebrain bundle (slMFB).6 A large profile of WM tracts, especially tracts connecting the medial frontal area, was associated with DBS response in patients who received subcallosal cingulate cortex DBS.7 Nevertheless, these results have not been replicated. The prediction of DBS outcome using preoperative neuroimaging features remains unexplored in the combined BNST–NAc DBS for TRD.

Major depressive disorder is hypothesised to be a disconnection syndrome in which connectivity between hub regions in brain circuits is impaired, leading to an imbalance of emotional circuitry.8 In this study, we evaluated preoperative structural brain connectivity, including the main fibre tracts connecting cortical and subcortical regions, as well as the brain connectome using graph analysis in patients with TRD. Our goal is to identify preoperative imaging biomarkers that predict the efficacy of DBS in patients with TRD.



Patients with TRD who were considered for neurosurgery at Ruijin Hospital from April 2021 to December 2022 were recruited for the imaging study. Among the 23 patients who underwent surgery, 8 patients without diffusion-weighted imaging (DWI) results were excluded from the analysis. One patient was excluded due to poor imaging quality. A total of 14 patients were enrolled in the analysis (figure 1). Detailed inclusion and exclusion criteria for patients with TRD eligible for neurosurgery are described in online supplemental file 1.

Supplemental material

Figure 1

Flowchart of patient screening and inclusion. DWI, diffusion-weighted imaging; HAMD, Hamilton Depression Rating Scale; ICD, International Classification of Diseases; MDD, major depressive disorder.

Surgical procedure and clinical assessment

Electrodes were implanted bilaterally in the dual target region of BNST and NAc using stereotactic frame-based magnetic resonance technique using SR1202-S 8-contact leads (8 contacts of 1.5 mm with a spacing of 0.5 mm; SceneRay, Suzhou, China).9 The details of the procedure have been previously described.10

Clinical outcomes were measured using the 17-item Hamilton Depression Rating Scale (HAMD-17),11 the Montgomery-Asberg Depression Rating Scale 2012 and the 14-item Hamilton Anxiety Rating Scale (HAMA-14).13 The assessments were conducted presurgically and postsurgically at each programming session. Other examiner-rating or self-reported scales were also collected at baseline, including the Depression and Somatic Symptoms Scale (DSSS),14 the 16-item Quick Inventory of Depressive Symptomatology self-report,15 the Sheehan Disability Scale (SDS),16 the Medical Outcomes Study 36-Item Short Form Health Survey (SF-36),17 the Barratt Impulsiveness Scale-11 (BIS-11)18 and the Dimensional Anhedonia Rating Scale (DARS).19 The criteria for response and remission were ≥50% reduction of HAMD-17 Score from baseline and HAMD Score ≤7, respectively.

Postoperative programming scheme has been described in a previous study.10 An initial programming session 1 week after surgery was launched to assess the efficacy and safety of each electrode contact. Experienced psychiatrists and neurosurgeons tested each contact individually through careful parameter titration with monopolar stimulation. Amplitude was gradually increased in 0.5 V steps from 2.0 to 6.0 V with fixed frequency (130 Hz) and pulse width (90 μs) for 30–60 s, except when any acute adverse effects were observed. If little clinical improvement was achieved, we tested voltages greater than 6.0 V and adjustments in pulse width and frequency. Induction of beneficial effects or adverse effects was documented. Follow-up programming was based on the subject’s experience feedback and scale evaluation.

Image acquisition

Preoperative imaging was performed on a 3T MRI system (Magnetom Prisma, Siemens, Erlangen, Germany). Axial three-dimensional T1-weighted Turbo-Field-Echo sequence was acquired with the following specifications: TR/TE 3000/2.56 ms, voxel size 0.8×0.8×0.8 mm3, field of view (FOV) 256×256 mm2. DWI images were acquired using a single-shot, spin-echo echo planar imaging sequence with 30 diffusion directions of b=1500 s/mm2, 60 diffusion directions of b=3000 s/mm2 and 11 interspersed scans of b=0 s/mm2 along AP direction. Scans with b=0 s/mm2 along PA direction were also acquired for the correction of susceptibility-induced distortions. The other specifications include TR/TE 6800/75 ms, voxel size 1.5×1.5×1.5 mm3 and FOV 192×192 mm2.

Electrode reconstruction and volume of activated tissue (VAT) estimation

DBS electrodes were localised using Lead-DBS ( and the PaCER algorithm. In brief, postoperative CT images were first linearly coregistered to preoperative MRI and normalised into ICBM 2009b NLIN asymmetric space using the SyN approach implemented in Advanced Normalisation Tools (ANTs). DBS electrodes were then localised using Lead-DBS and warped into the MNI space using the PaCER algorithm after visual review and refinement of the coregistration and normalisation.20

To construct a volume conductor model of the DBS electrode and surrounding tissue, a tetrahedral volume mesh was generated based on the surface mesh of DBS electrodes and subcortical nuclei using the Iso2Mesh toolbox ( Conductivities of 0.33 S/m and 0.14 S/m were assigned to grey matter (GM) and WM, respectively. Based on the volume conductor model, the potential distribution was simulated using the integration of the FieldTrip-SimBio pipeline into Lead-DBS (; In case of monopolar stimulation, the surface of the volume mesh served as the anode. Subsequently, the gradient of the potential distribution was calculated by derivation of the finite element method solution. The gradient was thresholded for magnitudes above a commonly used value of 0.2 V/mm to define the VAT.

Tractography and microstructural measures

Motion, eddy current and susceptibility-induced distortion corrections were processed using FSL V.6.0 (FMRIB Software Library).21 The multishell diffusion data were then reconstructed in DSI studio using generalised q-sampling imaging22 with a diffusion sampling length ratio of 1.25. The normalised quantitative anisotropy (NQA), fraction anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD) and radial diffusivity (RD) maps were derived for each participant. A deterministic fibre tracking algorithm23 was used with a step size of 0.5 mm, an NQA threshold of 0.02, a change threshold of 20% and an angular threshold of 60°. Tracks with length shorter than 30 or longer than 300 mm were discarded. A total of 1 000 000 seeds were placed at the whole brain and at the VAT estimated in the previous step registered to the native space by DSI studio. Automated fibre tracking was conducted to recognise and cluster the tracts based on the HCP tractography atlas.24 Mean fibre features of tracts were then extracted for each subject for statistical analysis. Percentage of fibre tracts traversing the VAT was calculated by dividing the number of specific fibre tracts by the total number of fibre tracts recognised in tracts traversing the VAT.

Tract-based spatial statistics (TBSS) was used to conduct the voxel-wise statistical analysis.25 All subjects’ FA data were aligned to an FMRIB58-FA_1 mm template in MNI space in a non-linear registration calculation method. The mean FA image of all subjects was created. Voxels with FA value threshold of 0.2 were abstracted to establish average FA images and average FA skeleton images. The standard space FA images of each subject were projected onto their average FA skeleton images and generated individual FA skeleton images, respectively. The NQA, AD, RD and MD skeleton images were also calculated based on FA skeleton images. Finally, a randomise tool was used to analyse the whole-brain non-parametric statistical values between groups, and a total of 5000 permutations were performed for each contrast. The p value images, which were corrected by threshold-free cluster enhancement, were thresholded at 0.05. Age was used as a covariate.

Distance estimation

Distances of the centre of the VAT to the anterior thalamic radiation and tracts connecting the ventral tegmental area (VTA) to the anterior limb of internal capsule (ALIC) as a reference of slMFB were calculated.26 27 The left/right anterior thalamic radiation was built using the automated fibre tracking algorithm in DSI studio. The VTA-ALIC tract was built using the left/right VTA and the corresponding left/right ALIC as regions of interest (ROIs). The tracking parameters were kept the same as the whole brain tractography. The VTA was extracted from automated anatomical labelling atlas 3 and the ALIC was extracted from the JHU ICBM-DTI-81 White-Matter atlas. The VAT built in MNI space was transformed into the patient-specific coordinate system as the tracts. Each tract contained several fibres. The minimum perpendicular distance from the centre of gravity of the VAT to the fibres was calculated per tract. The median was then used as the distance to the centre of the tract for each VAT. Only the distance to the tracts on the same side was analysed.

Connectome construction and graph analysis

MRtrix V.3 software was used for the construction of the structural connectome. Preprocessing of diffusion images included denoising, removal of Gibbs ringing artefacts and the use of the FSL toolbox21 to correct for distortions induced by susceptibility off-resonance fields, eddy currents and head motion. Fibre orientation densities were generated by multishell, multitissue constrained spherical deconvolution method and were subsequently normalised to obtain quantitative measures of density. T1-weighted images of each subject were registered to the mean b=0 image after preprocessing of DWI images. Anatomical images were coregistered with the standard space using ANTs to transform the AAL90 template to subject space. Brain tissue information was estimated from all T1-weighted images to obtain image segmentations of GM, sub GM, WM and cerebral spinal fluid using FSL. A total of 10 million probabilistic streamlines were generated using Second-order Integration over Fibre Orientation Distributions and Anatomically Constrained Tractography framework.28 The initial tractogram was generated with 10 million streamlines using the default settings and the Spherical-deconvolution Informed Filtering of Tractograms (SIFT) methodology was applied to filter the streamlines to 1 million. The tcksift2 method was adopted to compute the streamline weights that were designed to reduce known biases in tractography data.29 Finally, the SIFT2-weighted connections divided by the sum of the GM volume of two node regions were calculated to generate the undirected 90×90 weighted matrices.

The matrices were analysed by the Graph Theory Network Analysis toolbox.30 Binary connection matrices were generalised with a sparse threshold of 30%. Node metrics included betweenness centrality, degree centrality, nodal efficiency, nodal clustering coefficient, nodal local efficiency and nodal shortest path. Global metrics included network global efficiency, local efficiency and small-world indexes including clustering coefficient, characteristic path length, normalised clustering coefficient (γ), normalised shortest path length and small-world property (σ). The false discovery rate (FDR) correction was used for the comparison of multiple brain regions between the two groups in node parameters. Age was used as a covariate in the analysis.

Statistical analysis

Data were first tested for distribution using the Kolmogorov-Smirnov test, and then parametric (paired two-sample t-test or Pearson’s correlation coefficient) or non-parametric (paired Wilcoxon rank-sum test or Spearman’s correlation coefficient) statistic was used to assess potential differences or correlation of continuous variables. χ2 fisher exact test was used to compare categorical variables between groups. Variables included age, gender, employment status, marriage status, years of education, disease duration as well as multiple psychiatric scales scored at baseline. Pearson/Spearman’s partial correlation coefficients were calculated in the correlation analysis of fibre bundle properties (FA, MD, AD, RD, NQA) with HAMD-17 improvement considering age as a covariate. The FDR correction was used in graph analysis. Two-tailed p value <0.05 was considered significant. Statistical data were presented as mean (SD).


Baseline characteristics and clinical outcomes

Among the 14 patients included in the study, the mean improvement in HAMD-17 scores was 41.4% (SD=30.4%) at the 6-month follow-up. Three patients achieved remission, defined as a reduction in HAMD-17 scores to ≤7. Additionally, five patients (5/14) experienced a >50% improvement in HAMD-17 scores, indicating clinical response (table 1). The average number of times for parameter adjustments within 6 months was 8.0 (SD=7.3).

Table 1

Characteristics of included patients and the clinical scales at baseline and 6-month follow-up

When correlating baseline characteristics with HAMD-17 improvements, associations were found in the depression severity measured by HAMD-17, MARDS and self-reported DSSS scale, anxiety measured by HAMA-14, social life impairments measured by SDS, attention impulsiveness measured by BIS-11, vitality measured by SF-36 and anhedonia measured by DARS scale. Significant differences were also observed in the items of depression, anxiety, anhedonia, attention impulsiveness severity and social life impairments between the remission and non-remission groups (online supplemental tables 1 and 2).

WM features

In the TBSS analysis, no significant clusters of FA, MD, AD, RD and NQA differences were observed between the response/non-response and remission/non-remission groups. In the analysis of fibre bundle properties, the FA, MD and RD values of left superior thalamic radiation were found to be significantly different between groups, with the FA value being statistically higher and the MD and RD values statistically lower in the remission group (FA: p=0.008, t=3.19, MD: p=0.012, t=2.94, RD: p=0.013, t=2.93, figure 2). The FA and RD values of the right reticulospinal tract (FA: p=0.009, t=3.10, RD: p=0.038, t=2.33) and the right parahippocampal parietal cingulum (FA: p=0.036, t=2.36, RD: p=0.038, t=2.33) were different between the remission and non-remission groups, with the FA value being significantly higher and the RD value significantly lower in the remission group. In the correlation analysis, the FA, MD and RD values of left superior thalamic radiation correlated with the HAMD-17 improvement after the correction of age (FA: p=0.003, r=0.75, MD: p=0.010, r=–0.67, RD: p=0.006, r=–0.72). No combined changes in fibre bundle properties were observed between the response and non-response groups.

Figure 2

Tracts with significant difference in fibre bundle properties and the component of tracts traversing the volume of activated tissue. (A) Tracts with significant differences in fibre bundle properties between the remission and non-remission groups. (B) Percentage of fibre tracts that traverse the volume of activated tissue. VAT, volume of activated tissue.

In the analysis of fibre tracts traversing the individual VAT, anterior thalamic radiation is the major component in the fibre tracts (figure 2). Significant differences in the percentage of fibre tracts going through the VAT were found in the left superior thalamic radiation between the remission and non-remission groups (p=0.033, t=2.44) and in the right corticopontine tract frontal between the response and non-response groups (p=0.036, t=2.36). The FA, MD and RD values of the left anterior thalamic radiation (FA: p=0.014, t=2.86, MD: p=0.022, t=2.62, RD: p=0.018, r=2.75) and the left superior thalamic radiation (FA: p=0.013, t=2.95, MD: p=0.011, t=3.04, RD: p=0.008, t=3.25), as well as the FA and RD values of the right anterior thalamic radiation (FA: p=0.041, t=2.30, RD: p=0.016, t=2.79) were found to be different between the remission and non-remission groups (online supplemental tables 3–9). Differences in the FA and RD values of right anterior thalamic radiation were also observed between the response and non-response groups (FA: p=0.011, t=3.00, RD: p=0.012, t=2.97).

Distance estimation

We estimated the distance of the centre of each VAT to the centre of corresponding anterior thalamic radiation and tracts connecting VTA and ALIC (VTA-ALIC) on the same side (figure 3). No significant correlations were found between the distance of the centre of the VAT to any fibre tracts evaluated and the HAMD-17 improvements. The MNI X coordinate on the left negatively correlated with the HAMD-17 improvements (p=0.003, r=–0.74). When comparing the remission and non-remission groups, a significant difference in the distance to the left VTA-ALIC tract was observed (p=0.010, t=3.07). A tendency of difference in the distance to the right VTA-ALIC tract (p=0.054, t=2.14) was also observed. Centres of the VAT in the remission group had a significantly closer distance to the left VTA-ALIC tracts than the non-remission group (figure 3, table 2). Significant differences were also observed in the left and right MNI X coordinates between the remission and non-remission groups. No significant difference in the distance was observed between the response and non-response groups.

Table 2

Coordinates of the volume of activated tissue and their distance to specific fibre tracts

Figure 3

Tracts and their relative positions with VAT and the difference of indexes in graph analysis. (A) Illustration of the VTA-ALIC tract and the anterior thalamic radiation. (B) The relative position of the volume of activated tissue and the specific fibre tracts. (C) Distance to the left VTA-ALIC tracts and the left anterior thalamic radiation between the remission and non-remission groups. (D) Differences in the normalised clustering coefficient (γ) and the small-world property (σ) between the remission and non-remission groups. ALIC, anterior limb of internal capsule; VAT, volume of activated tissue; VTA, ventral tegmental area.

Graph analysis

No significant difference was observed in the edges and nodes in the structural connectome between the response/non-response and remission/non-remission groups. Nevertheless, global metrics including the normalised clustering coefficient (γ) and the small-world property (σ) were both statistically higher in the remission group compared to the non-remission group (γ: p=0.048, t=2.22, σ: p=0.044, t=2.27, figure 3). The difference was also observed between the response and non-response groups (γ: p=0.007, t=3.33, σ: p=0.008, t=3.25). The γ and σ also correlated with the symptom improvement after the correction of age (γ: p=0.006, r=0.72, σ: p=0.011, r=0.68).


Main findings

This study correlated baseline characteristics and presurgical structural imaging to identify potential biomarkers associated with the individual outcome of the combined BNST–NAc DBS. Patients with more severe depression symptoms, including anhedonia, lack of vitality, increased attention impulsiveness and anxiety, tend to benefit less from the DBS treatment. Among the analysis of fibre bundle properties and graph indexes, fibre features in tracts including superior thalamic radiation and reticulospinal tract, graph indexes including the small world property and the normalised clustering coefficient in the structural connectome were identified to correlate with the clinical outcome. Besides, we found that distance of the centre of the VAT estimated by stimulation parameters to the tract connecting the VTA-ALIC varied between the remission and non-remission groups. The findings from structural imaging indicated the influence of the individual structural alterations on DBS modulation, providing clues for the prediction of individual outcomes after DBS surgery in patients with TRD.

Our study replicated previous findings that WM integrity of fibre tracts connecting the frontal cortex with subcortical regions affected the efficacy of DBS.9 The associations were observed in fibre bundles including thalamic radiation and the reticulospinal tract. Decreased FA and increased MD and RD values were shown in the tracts in the non-remission group. The fibre bundle properties were strongly associated with the demyelination of the fibre tracts.31 Physiology studies have found that poorly myelinated fibres fail to transmit consecutive antidromic spikes at the start of DBS, suggesting antidromic excitation is strongly influenced by the myeloarchitecture.32 In a diffusion tensor imaging-functional MRI study, increased integrity of WM in the ALIC region was positively correlated with reward-related activation in the NAc, indicating that the microstructural integrity of fibre tracts may influence the intensity of reward-related responsiveness of the ventral striatum.33 Changed fibre properties may restrain information transfer in the brain network, thereby impairing the modulatory effect of DBS. Differences in the brain network were indeed observed in graph analysis. Although our intention to find potential edges that correlate with clinical improvements yielded no significant findings, the small world property and the normalised clustering coefficient did correlate with HAMD-17 improvement, which demonstrated that network efficiency of the brain and its balance with local specialisations had an impact on the effect of DBS surgery.

The anterior thalamic radiation and slMFB were reported to be the main fibre tracts that mediated the effect of DBS in the ALIC and nearby regions.26 Previous studies showed that stimulation in the ALIC closer to both tracts correlated with symptom alleviation.27 Our target, compared with the previous one, was located more ventrally and medially, reaching more areas in the NAc and BNST nuclei. Nevertheless, according to the results of our study, a relatively lateral position of the VAT centre on the left was associated with better clinical outcomes. This can be explained, at least partially, by the modulation of fibre bundles, as our tracking shows that a more lateral position of the VAT tends to activate more fibres connecting VTA-ALIC, which are located more laterally than the anterior thalamic radiation in the level of stimulation in the brain.34 Notably, in the previous study, the significant correlation was not observed until the duration of chronic stimulation after the optimisation of stimulation parameters was used as a covariate, suggesting the duration of stimulation may be an influential factor.27 A longitudinal analysis considering its effect may help further clarify the issue.


Various factors may influence the clinical outcome of patients with TRD, including the duration of the stimulation, the adjustment of the parameters and the placebo effect. The restriction of deterministic fibre tracking with automated recognition of tracts based on atlas, the single-centre experience as well as the limited number of patients can influence the generalisability of our analysis. A double-blinded randomised clinical trial is currently underway to rule out the placebo effect. Fibre tracking with more focus on specific tracts and ROIs may provide further insights into the mechanisms of DBS modulation.


Presurgical structure alterations in patients with TRD can influence the outcome of the combined BNST–NAc DBS. Fibre tracts connecting the frontal area with subcortical regions, the small world property and the normalised clustering coefficient of the brain network are correlated with the DBS outcome. Distance to the VTA-ALIC tract also influences the modulatory effects. These findings shed light on the regulatory mechanisms of DBS and can inform presurgical planning for the treatment of TRD. Future studies incorporating electrophysiology analysis may further elucidate the complexity of DBS modulation in patients with TRD.

Supplemental material

Data availability statement

Data are available upon reasonable request.

Ethics statements

Patient consent for publication

Ethics approval

This study involves human participants. All procedures were approved by the ethics committee of Ruijin Hospital Shanghai Jiao Tong University School of Medicine (approval number: 202152, supplementary file 2). Participants gave informed consent to participate in the study before taking part.


Fengting Wang obtained her bachelor degree of science from School of Pharmacy, Shanghai Jiao Tong University and her MD degree in neurosurgery from Shanghai Jiao Tong University School of Medicine, China in June, 2023. She has been a resident doctor and postdoc in the department of functional neurosurgery in Ruijin Hospital, Shanghai, China, since August 2023. Her main research interests include predictions of efficacy of deep brain stimulation (DBS) in psychiatric and movement disorders as well as the modulating mechanisms of DBS and MR-guided focused ultrasound. Her work in NPJ Parkinson’s disease recapitulated brain morphometric findings that predict the outcome of DBS in Parkinson’s disease. Her collaborating work in Journal of Neurology, Neurosurgery and Psychiatry delineated a frontothalamic pathway and connectivity profile responsive for clinical outcome of DBS for treatment-resistant depression.

Embedded Image

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Contributors YL, VV and BS contributed to the conception and design of the study. FW, LD, TW, YZhang, WY, YZhao, YP, LB, DL, SZ and YL contributed to acquisition, postprocessing and analysis of the data. FW, YL, VV and BS drafted the manuscript and prepared the figures. All authors approved the final version of the manuscript. BS accepted the full responsibility of the overall content.

  • Funding This deep brain stimulation intervention was supported by an unrestricted, investigator-initiated research grant by Scenery (BS), which provided the devices used. The project was sponsored by SJTU Trans-med Awards Research (2019015 to BS) and Shanghai Clinical Research Centre for Mental Health (19MC191100 to BS). BS was also sponsored by the National Natural Science Foundation of China (81771482). VV was supported by the Guangci Professorship Programme of Ruijin Hospital (N/A) and a Medical Research Council Senior Clinical Fellowship (MR/P008747/1). YL was sponsored by the National Natural Science Foundation of China (82101546) and the Shanghai Sailing Program (21YF1426700). The funding sources were not involved in the design and conduct of the study.

  • Competing interests BS and DL receive occasional fees from Scenery for educational purposes. No other conflicts were reported.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.