Notwithstanding the cumulative evidence on the safety and efficacy of transcranial magnetic stimulation in depression care, the non-response rate to transcranial magnetic stimulation (TMS) amongst treatment-resistant depression has remained substantial despite the health care cost and time incurred. There remains a compelling clinical need to find valid biomarkers to inform personalized treatment. Using supervised machine learning on 4 combined features of neuroimaging markers, our group recently reported excellent prediction for clinical response in 70 patients receiving TMS to left dorsolateral prefrontal cortex for medication-resistant major depression in 2015-18 (Phase 1 study).The clinical utility of these potential neuroimaging biomarkers is still uncertain without further validation of the trained model in an independent clinical cohort.

1. Sampling Clinical participants will be recruited from the Transcranial Magnetic Stimulation Day Care Centre of Tai Po Hospital based in the New Territories East Service Cluster under the public sector. Written informed consent will be obtained from all participants according to the Declaration of Helsinki. Each subject will be paid HKD$800 to subsidize their travel.

2. Pre-treatment assessment

   1. Baseline clinical assessment (one week before the first TMS session) Apart from collecting the basic demographic profile of the research participant (age, education level, gender), the research psychiatrist will administer MADRS, DSM-IV, SCID-I/II to ascertain current/ lifetime Axis I and II psychiatric diagnosis, age at onset of MDD and number of major depressive episodes, number of failed antidepressants and clinical global impression scale (CGI) of the current episode. The psychiatrist will also review the past and current significant medical history from the Hospital Authority's Clinical Management System supplemented by participant's self-report. Participant will complete Beck Depression Inventory that has been validated in local Chinese dialect. Handedness is assessed with the Edinburgh Handedness Inventory.
   2. Pre-treatment MRI of Brain Pre-treatment MRI will be acquired up to two weeks before treatment on a 3.0 T Philips Achieva Medical Scanner with an eight-channel SENSE head coil (Philips Healthcare, The Netherlands) at the Prince of Wales Hospital, New Territories East Service Cluster of Hospital Authority of Hong Kong. The first scan is a high-resolution T1-weighted structural scan covering the whole brain acquired with the following parameters: repetition time = 7.54 ms, echo time = 3.53 ms, flip angle = 8⁰, 1.1 x 1.1 x 0.6 mm voxels, number of slices = 285, slice orientation = sagittal, slice thickness = 1.2 mm, Field of View = 250 mm3, and matrix size = 240 x 240. This scan will be used to register with the resting-state fMRI data, and for segmentation into grey matter, white matter and cerebrospinal fluid, and normalization to template space. Following the T1-weighted sequence, a 12-minute resting-state fMRI scan will be collected with the following protocol: repetition time = 2050ms, echo time = 25ms, flip angle = 90⁰, 3.2 mm3 voxels, slice thickness = 3.2 mm, Field of View = 205 mm², and matrix size = 64 x 64. Participants will be instructed to look at a fixation cross during the scan session.

3. TMS device, stimulation target localization and stimulation parameters MagVita X100 will be used to deliver intermittent theta burst stimulation (iTBS) to left DLPFC in the Transcranial Magnetic Stimulation Day Care Centre of Tai Po Hospital. The treatment facility is fully funded and operated by the Hospital Authority as part of the drug-resistant protocol accessible to all patients receiving specialist psychiatric service in the public sector and was established since 2021. All treatment sessions are delivered by qualified operators (psychiatrists or nurses) as part of the core clinical service. The treatment protocol adopts F3 beam target with MagVita X100 following the manufacturer's guide that utilizes a formulaic estimation of F3 location system. The US-FDA approved 3-minute intermittent theta-burst protocol (i-TBS) 18 comprises of 18 cycles of 10 bursts. Each burst is triplet of pulses discharged at 50 hz and the burst frequency is 5Hz. Between two cycles of bursts is 8-second inter-train rest. The device output is set at 120% above the resting motor threshold determined on the observable motor excitability on M1 motor cortex corresponding to the right-hand abductor pollicis brevis (at least 5 out of 10 trials reaching 50 microvolt read from MEG in response to single pulse TMS). The resting motor threshold is determined only once every week. The full course includes 20 daily sessions over four weeks.

4. Follow-up assessments are scheduled at the end of week 2, week 4, week 6, week 8, and week 12 to delineate the trajectory of depressive symptoms over a course of 12 weeks. Primary outcome measures are the scores on MADRS and CGI at the end of week 4 and week 12. Secondary outcome measure is score on BDI at the end of week 4 and week 12. Patients are classified as short-term responders where clinical response is defined as CGI= 2 and >/=50% reduction of MADRS score from baseline at the end of week 4. For those who fulfill clinical response where CGI= 2 and there is >/=50% reduction of MADRS score from baseline at the end of week 12 are classified as long-term responders. Patients are classified as short-term remitters where clinical remission is defined as CGI= 1 and MADRS score<7, respectively at the end of week 4. For those who fulfill clinical remission where CGI= 1 and MADRS score is <7 at the end of week 12 are classified as long-term remitters.

5. Serum BDNF level would be evaluated at baseline (within 2 weeks before treatment), at the end of week 2, week 4 and week 12 as a neurochemical biomarker of neuroplastic response. Blood samples will be collected in anticoagulant-free tubes and kept at room temperature for 1 hour before further blood sample processing. The serum collected will be used for serum marker assay. The blood samples collected in anticoagulant-free tubes would be centrifuged at 3500g, at 4°C for 10 minutes within 3 hours of blood collection to obtain serum and stored at -70°C until assay. Commercial ELISA kit will be used to assay the levels of BDNF. Measurement will be performed accordingly to manufacturer's instructions. Samples and standards will be run in duplicate and BDNF concentrations will be calculated using the standard curve. BDNF (Val66Met) will be genotyped to explore its relationship with neuroplastic change and clinical response in this study. In brief, DNA will be extracted by commercial DNA extraction kit from blood samples according to manufacturer's instruction. It will then be genotyped by Taqman genotyping assay.

Inclusion Criteria:

• right-handed
• meet the Diagnostic and Statistical Manual of Mental Disorder, Fourth Edition (DSM-IV) criteria for major depressive disorder
• at least moderate episode or with a score of >20 on Montgomery-asberg Depression Rating Scale (MADRS) and >18 on Hamilton Depression Rating Scale(HDRS) 17-item;
• has failed to respond adequately to at least one full course (>6 weeks) of antidepressant medication or medication intolerant.

Exclusion Criteria:

• significant head trauma
• active abuse of alcohol or illegal substances
• current psychotic symptoms
• suicide ideation/recent suicide attempts
• other DSM-IV Axis I and II psychiatric diagnosis
• neurological disorders and contraindications to fMRI (e.g. pace makers, metal implants, pregnancy) or rTMS, or having undergone electroconvulsive therapy in the preceding year.

As they are all clinical subjects, their data should be strictly confidential.