This is a multicentre, double-blind and sham-controlled randomised trial among participants with TRD. Eligible participants will be recruited from seven hospitals across five provinces in China, with the Changping Laboratory serving as the trial sponsor and overseeing comprehensive quality control and data management. Participants will be randomised in a 2:1 ratio into active and sham interventions. iTBS interventions will last 21 consecutive days (3 weeks), and participants in both groups will follow the same schedule for post-treatment follow-ups at weeks 6, 12 and 27 (6 months post-treatment). The study flow chart provides an overview of the study design and procedures (figure 1).

The sample size is calculated using the two-sample test model within PASS 16.0.4. Parameters are set to achieve 90% power and an inspection level of α=0.05. The effect size estimated by the Montgomery–Åsberg Depression Rating Scale (MADRS)23 for the difference between the active and sham groups is 0.40.8 Considering a dropout rate of 15%, the necessary sample size is 354 participants. We aim to enrol 360 participants.

Participants diagnosed with TRD will be recruited. Diagnosis will be confirmed using the Structured Clinical Interview for the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition,24 and treatment resistance will be determined via the Maudsley Staging Method (MSM).25 Detailed inclusion/exclusion criteria are provided in table 1.

A pre-screening will be conducted over the telephone or in person using a simple questionnaire to determine initial eligibility. Potential participants will then undergo a screening visit for further eligibility assessment. Trained research staff will verify each inclusion/exclusion criterion to confirm study eligibility.

In both treatment arms, the MT20A and P50 magnetic stimulator paired with a real-time MRI-guided neuronavigation system (Neural Galaxy Inc., Beijing, CN) and equipped with a figure-of-eight coil will be used. Each participant will receive three daily sessions of 1800 pulses per session for 21 days, completing the intervention within 28 days with no more than a 3-day gap. Stimulation intensity is set at 100% resting motor threshold (RMT) with 50 min intervals between sessions. If 100% RMT is intolerable based on participants’ feedback, the intensity may initially be reduced to no less than 80% RMT, with attempts to increase to 100% after participants adapt to the intervention. RMT will be tested weekly. Active and sham treatments will be administered identically in all aspects, except for the stimulation intensity, which is non-therapeutic in the sham condition.

A treatment record sheet will be filled in daily during the intervention. Each centre will have dedicated TMS operators who receive uniform training and undergo treatment accuracy assessments conducted by Changping Laboratory. Operators who do not meet accuracy standards will receive additional training and supervision until they meet the required standards. They will be instructed to minimise communication with patients and avoid discussing possible group assignments with them.

During the intervention period, participants will continue their baseline medications, with the dosage and type remaining unchanged unless clinically necessary. If a patient experiences severe insomnia symptoms, intermittent use of sleep medication is allowed, limited to no more than five consecutive days. Approved options include zolpidem (≤10 mg), zopiclone (≤7.5 mg), eszopiclone (≤3 mg) or zaleplon (≤10 mg), with total usage not exceeding 2 weeks during the trial period. No additional psychotropic medications or physical treatments will be permitted during the intervention period. During the follow-up period, medication adjustments will be allowed under physician guidance. Detailed records of concurrent medication use will be maintained throughout the study.

At the screening visit, participants will complete a questionnaire interview covering sociodemographic information (age, sex and education), clinical evaluations (history of diseases, duration of current episodes and treatment history), medication use (antidepressants, mood stabilisers and psychiatric medications) and physical examinations (blood pressure, height, weight and pulse rate).

Depression severity and the primary outcome measure will be assessed using the MADRS. To provide a more comprehensive evaluation of depressive symptoms, the 17-item Hamilton Depression Scale (HAMD)26 will be also included as a secondary outcome measure. This dual approach allows for sensitivity to both core mood and somatic/anxiety symptoms. Additional secondary outcomes will include the Clinical Global Impression-severity (CGI-S), Global Assessment Function (GAF) and cognitive functioning. The MADRS and CGI-S will be evaluated at all visits: at baseline and at weeks 1, 2, 3, 6, 12 and 27. All evaluators of clinician-rated scales will undergo multi-centre consistency training organised by the coordinating centre (Beijing Anding Hospital). Additionally, to explore the neural mechanism underlying heterogeneous depression, various psychological scales will be assessed using the Snaith–Hamilton Pleasure Scale (SHAPS),27 Hamilton Anxiety Scale (HAMA),28 Beck Scale for Suicide Ideation (SSI),29 Insomnia Severity Index (ISI),30 Pittsburgh Sleep Quality Index (PSQI),31 Rumination Response Scale (RRS)32 and Patient Health Questionnaire (PHQ)33 at baseline and immediately after treatment. Patients will fill out blind assessment forms to indicate the group they believe they were assigned to after all interventions are completed.

Cognitive impairment will be assessed using two different tools at different centres, measured at baseline and immediately after treatment. The Chinese version of the MATRICS Consensus Cognitive Battery consisting of nine tests34 35 will be used at six clinical centres. Additionally, the Chinese Brief Cognitive Test, consisting of four tests, will be used at one centre, as it became available in China in 2023 after the study initiation at the first six centres.36

To evaluate the long-term antidepressant effects and relapse rates, a follow-up at 6 months post-treatment will be conducted with individuals who completed the 21-day therapy. These sessions will involve gathering updates on the patient’s current condition and assessing their scores on the MADRS. table 2 displays the schedule and assessments at screening, baseline, intervention and follow-up. Medication modifications, new treatment additions and adverse events (AEs) will be recorded when they occur.

To identify the individualised target and investigate the neural underpinnings of the heterogeneity of depression, participants will have structural and resting-state functional MRI (fMRI) scans at baseline and post-treatment. MRI scans will be acquired at each clinical centre using a 3.0 T MRI scanner and employing an imaging sequence specified by Changping Laboratory. Minor parameter adjustments will be made as needed to accommodate variations in scanners. The detailed parameters of the seven centres are listed in table 3.

The voxel size for structural MRI will be 1 × 1 × 1 mm, and for fMRI scans, standard T2*-weighted echo-planar imaging (repetition time = 3000 ms, echo time = 30 ms, flip angle = 90°, 3 × 3 × 3 mm voxels, 5 runs, 360 s/run) will be used. Throughout the 30 min resting-state scans, participants will be instructed to avoid repetitive thoughts, keep their eyes closed and keep their heads still. MRI data will be collected within 7 days before or after the intervention, depending on the availability of the MRI scanners and participants. In instances where the image quality is considered insufficient at pretreatment, patients will be asked to undergo a rescan to ensure a total of at least 24 min (4 runs) resting-state scans of satisfactory quality for accurate target identification. However, post-treatment scans will not require a rescan if the image quality is lacking to prioritise the patient’s rights.

MRI data collected at baseline and post-treatment will be preprocessed centrally at Changping Laboratory, following procedures described in our previous studies.37 38 Specifically, structural images will be used to reconstruct the cortical surface. Resting-state fMRI data will undergo the following preprocessing steps: 1) removing the first two frames; 2) slice timing and motion correction; 3) global mean signal normalisation across runs; 4) linear detrending and band-pass temporal filtering (0.01–0.08 Hz); 5) regression of covariates, including motion parameters, average signals from the whole-brain, white matter, cerebrospinal fluid and their temporal derivatives; and 6) registration to the surface fsaverage6 template and smoothing with a 6 mm full-width half-maximum kernel in the surface space.

Treatment target identification using MRI data collected at baseline will be analysed centrally at Changping Laboratory. The precision target for each participant is defined in the left DLPFC and guided by FC. Specifically, the cerebral cortex will be divided into 92 functional sectors using an iterative parcellation algorithm as previously reported.22 A region in the left DLPFC and FPN will be selected as the target parcel, and a vertex within the target parcel on the cerebral gyrus will be selected as the stimulation target. A target sheet containing the patient’s ID, age, MADRS score, target coordinates and location on the brain will be forwarded to TMS operators at each clinical centre.

Randomisation procedures will be employed following baseline data collection (including clinical assessment and MRI scan). Participants will be randomly assigned in a 2:1 ratio to the active or sham intervention group before the start of the first treatment session. Stratified block randomisation with variable block sizes will be employed, stratified by study centres and symptom severity defined by MADRS score (MADRS≥34 and MADRS<34). The random allocation sequence will be generated using Clinflash (Clinflash Healthcare Technology). On participant eligibility, an independent unblinded person will obtain a random number corresponding to the participant’s information.

Participants and their relatives, outcomes evaluators, care providers and investigators will be blinded to the intervention assignments. The active and sham coils are located on separate equipment or within a double-sided coil where the active and sham sides appear the same. To ensure allocation concealment, two electrodes (Skintact RT-34, Leonhard Lang GmbH, Austria) will be placed on participants’ forehead: one centred on the nose and the other placed about 1–5 cm laterally, with both located approximately 1 cm above the eyebrows. Focal electrical stimulation will be delivered at the participant’s maximum tolerable intensity to mimic the cutaneous sensation produced by active magnetic stimulation.39 40 The equipment or coil designated as active or sham is labelled discreetly ‘A’ and ‘B’. The actual identity of ‘active’ or ‘sham’ corresponding to ‘A’ or ‘B’ is known only to independent unblinded personnel. Following the random allocation of treatment groups, this unblinded person will send the group information to the treatment operator, who remains blind to the active or sham assignment. The operator will be instructed to use A or B of the coil or equipment for treatment. Usage of A or B will be recorded by the operators on the treatment record sheet after each session to ensure correct usage.

Patients will be instructed not to discuss their therapeutic experiences with others. Clinical evaluators will work independently from operators and will not have access to any treatment-related information. Everyone involved will be required not to ask and discuss the treatment group. Unblinding must be restricted and allowed only when necessary for participant management, particularly in situations involving serious AEs (SAEs). After all interventions, patients will complete a blind assessment form to indicate the group assignment they believe they were in.

The primary outcome is the change in MADRS immediately following the 21-day intervention. The secondary outcomes are the response rate defined by MADRS reduction ≥50% and remission rate defined by MADRS <11; changes observed in HAMD, CGI-S and GAF; cognition change; AEs; and dropout rate.

All scale evaluators must have a minimum of 3 years of professional experience in mood disorders or related fields. They will undergo training and pass an examination on clinical scale consistency assessment organised by the coordinating centre (Beijing Anding Hospital). Similarly, all TMS operators will participate in technical training and pass an examination on operating TMS equipment organised by Changping Laboratory. Furthermore, all researchers involved in this study will attend technical training organised by Changping Laboratory, covering aspects including the study protocol, informed consent, case report form and quality control.

Each centre and Changping Laboratory have quality control teams. Each site conducts comprehensive self-examinations of the entire clinical trial process. Additionally, on enrolment of the first two participants and approximately every five participants thereafter, the sponsor dispatches a highly trained and dedicated clinical research associate for on-site monitoring. This ensures strict adherence to the study protocol and standards throughout the implementation phase.

All data, including information from the screening period and post-enrolment, will be documented in paper case report forms (CRFs). The patient’s personal information will not exist in the CRF; instead, a unified ID number will be used for identification. On completion of the trial, the CRF will be forwarded to Changping Laboratory. Two persons will independently input the data from the CRF into a database established using Epidata version 3.1. The two datasets will be compared and verified, with any discrepancies resolved to produce the final validated dataset. In cases where patients deviate from the protocol and are potentially subject to exclusion from statistical analysis, investigators from Changping Laboratory and each clinical centre will discuss and jointly determine whether these participants should be excluded from the analysis.

Unless otherwise specified, all hypothesis testing will be two-sided with a significance level of 0.05. Statistical analyses will be conducted by two independent statisticians using SAS 9.4 and MATLAB R2020a (The MathWorks, Inc., 2020).

The first participant was enrolled on 24 May 2023 and is expected to finish in December 2025.

None.

The study will be conducted in compliance with this protocol (version: V006, dated 6 September 2023) and the Declaration of Helsinki. Protocol modifications are communicated to all investigators and updated in the trial registry. Ethical approval has been obtained from each centre’s ethics committee, with the first ethical approval granted by the Ethics Committee of Beijing Anding Hospital on 30 December 2022(Approval 2022206FS-2). The study was registered on 17 January 2023 (ChiCTR2300067671).

Written informed consent will be obtained from all participants before enrolment. A sample consent form is shown in the online supplemental material. Participants will receive approximately 200 RMB for each visit as compensation for their time and transportation costs related to trial follow-ups. Payment methods and schedules are determined by each centre. Participants will be informed of their right to receive active intervention after trial completion if they were initially assigned to the sham group.

The tolerability and safety of the iTBS during the intervention will be recorded by TMS operators. However, operators are instructed to minimise communication during the intervention and will not actively inquire about discomfort. Reports of discomfort will be documented only if participants initiated the feedback. If any participant is identified as being at risk to themselves (such as suicidal ideation) or showing signs of developing an SAE, they will be referred to appropriate clinical services.

Given the well-established safety profile of iTBS, an independent Data Monitoring Committee is not required for this study. Safety oversight will instead be conducted by the principal investigator at each site, with all AEs documented and reported to the relevant ethics committees for review.

The findings from this trial will be shared via academic conferences and published in international peer-reviewed journals by Changping Laboratory following data clearance, trial locking and statistical analysis by professional teams. Each centre is restricted from distributing interim or incomplete results before the official results are presented.

Consent obtained directly from patient(s)

This study involves human participants and was approved by 1. The Ethics Committee of Beijing Anding Hospital, Capital Medical University, ID: 2022206FS-2; 2. The Medical Ethics Committee of Peking University Sixth Hospital (Institute of Mental Health), ID: 2022-57; 3. The Biomedical Ethics Review Committee of West China Hospital, Sichuan University, ID: 1767; 4. The Medical Ethics Committee of Henan Provincial People’s Hospital, ID: 2022-163; 5. The Biomedical Ethics Committee of Anhui Medical University, ID: 83220417; 6. The Medical Ethics Committee of Tianjin Anding Hospital (Tianjin Mental Health Center), ID: 2022-37; 7. The Medical Ethics Committee of the Fifth People’s Hospital of Luoyang, ID: LYWY-KT-2023004 Participants gave informed consent to participate in the study before taking part.