Real-World NeuroModulate: Investigating Non-Invasive Brain Stimulation for Neuropsychiatric Disorders at TUM

This study investigates the neurobiological effects of standard intermittent theta burst stimulation (iTBS) applied to the left dorsolateral prefrontal cortex (DLPFC) in patients with a depressive syndrome undergoing routine clinical TMS treatment. The study follows a prospective, naturalistic longitudinal design and focuses on the association between iTBS-induced clinical change and repeated multimodal magnetic resonance imaging (MRI) measurements. The primary scientific objective is the identification of imaging-correlates of treatment response, as well as the identification of biomarkers that allow for early prediction of treatment response and later relapse following iTBS.

Patients who receive a clinical indication for TMS as part of routine psychiatric care undergo an acute treatment course consisting of 20 iTBS sessions administered over four consecutive weeks with five sessions per week. The treatment protocol is fixed to standard intermittent theta burst stimulation applied over the left DLPFC. The clinical decision to initiate TMS is made independently of study participation.

The severity of depressive symptoms is assessed at baseline (prior to the first session), as well as after two weeks and after four weeks of treatment, using both self-report and clinician-rated scales (HDRS and MADRS, as well as the BDI). Response is defined as at least a 50% reduction in symptoms, operationalized as a minimum 50% decrease in the respective scale scores. The MADRS is used to classify participants as responders versus non-responders. In addition, sociodemographic data, personality characteristics (degree of sensation-seeking personality), and history of traumatization (CTQ) are collected at baseline.

Prior to treatment initiation, the individual resting motor threshold (rMT) is determined by stimulation of the primary motor cortex, using either visual detection of motor evoked responses or surface electromyography. The stimulation intensity for iTBS is set at 120 % of the individual rMT. In cases where this intensity is not tolerated, e. g. due to pain, the intensity will be reduced to 80% of the individual rMT.

The iTBS protocol consists of bursts of three pulses at 50 Hz repeated at intervals of 200 ms (5 Hz). Each train lasts 2 seconds and is followed by an 8-second inter-train interval. One iTBS session has a total duration of 3 minutes and 9 seconds and comprises 600 pulses. One single iTBS session is delivered per treatment day throughout the four-week acute stimulation phase. No accelerated, high-intensity, or multi-session-per-day protocols are applied.

The stimulation target, the left DLPFC, is localized by a standardized scalp-based localization approach using the BeamF3 algorithm.

Multimodal MRI is acquired at four predefined time points: prior to the start of iTBS (baseline), after 10 stimulation sessions (mid-treatment), after completion of all 20 stimulation sessions (post-treatment), and at a three-month follow-up after the end of the acute stimulation phase. All MRI scans are performed on 3 Tesla clinical MRI systems. The total scan duration per session is approximately 60 minutes. No intravenous contrast agents are administered.

The MRI protocol includes structural, diffusion, perfusion, relaxometry and functional MR imaging. The initial scan also includes a T2\* and fluid-attenuated inversion recovery (FLAIR) sequence to detect potential clinically relevant pathologies.

Structural MRI comprises high-resolution T1-weighted sequences for anatomical reference, cortical thickness and volumetric analyses, and neuronavigation.

Multi-modal relaxometry will be performed by acquiring 6-echo proton density (PD), magnetization transfer saturation (MTsat) and T1 scans, combined with B1 field mapping. From these scans, quantitative tissue maps for R1, R2\*, PD and MTsat will be generated using the hMRI toolbox.

Diffusion-weighted imaging (DWI) will be performed to characterize white matter microstructure and structural connectivity. The employed sequence consists of 60 gradient directions at b=2000 s/mm², 30 gradient directions a b=711 s/mm² and 6 interleaved b0 volumes. From these scans, we will generate both microstructural parameter maps using the Neurite Orientation Dispersion and Density Imaging (NODDI) framework as well as diffusion kurtosis imaging (DKI). To investigate connectivity, we will employ constrained spherical deconvolution to generate whole-brain tractography. We will use commonly employed cortical parcellations such as the Destrieux atlas and the Human Connectome Project Multi-Modal Parcellation (HCP-MMP) to generate connectivity matrices. These will then be analyzed using network-based statistics. Furthermore, we will directly analyze fibre orientation distribution functions in a fixel-based approach.

Perfusion imaging is obtained using arterial spin labeling (ASL). This non-invasive technique allows quantification of regional cerebral blood flow (CBF) without the use of contrast agents and enables assessment of iTBS-related changes in cerebral perfusion across cortical and subcortical regions. Specifically, we are employing a multi-delay, hadamard-encoded 3D pseudo-continuous ASL sequence with voxel size of 3 mm isotropic. This sequence further enables measuring the T2 signal progression of labelled protons. From the imaging data, we will generate quantitative CBF maps and arterial transit time maps. We will further use modeling of T2 signal progression to generate blood-brain barrier integrity maps.

Functional MRI is acquired using blood-oxygenation-level-dependent (BOLD) contrast during resting-state conditions. Participants are instructed to remain still, stay awake, and not engage in any specific task. Resting-state fMRI is used to quantify intrinsic functional connectivity within large-scale brain networks, with a particular focus on prefrontal-limbic circuits, default mode network connectivity, salience network dynamics, and frontoparietal control networks. We will generate respective tissue maps quantifying the degree of voxel-wise correlation with network activity for each network. Additionally, we will generate regional homogeneity (ReHo) and amplitude of low frequency fluctuations (ALFF) maps. We will further generate functional connectome matrices using similar parcellations as specified for the DWI approach.

This longitudinal multimodal imaging approach allows between- and within-subject characterization of neurobiological trajectories across the acute iTBS phase and into the follow-up period. By combining repeated structural, diffusion, perfusion, and functional MRI, the study aims to identify neurobiological signatures of iTBS response, mechanisms of action at the network level, and potential predictors of treatment outcomes.

We plan to enroll 100 patients. An interim analysis is planned after at least 60 patients enrolled.

• Difference in connectome trajectory between Responders and Non-Responders across multi-modal connectomes. [Time frame: From enrollment to the end of treatment (4 weeks).]
• Prediction of Responder Status using Structural and Functional Networks at Baseline [Time frame: Data generated at enrollment.]
• Prediction of Responder Status using Structural and Functional Networks at Baseline and Mid-Therapy [Time frame: Enrollment to two weeks after therapy start.]
• Prediction of Responder Status using Tissue Parameter Maps at Baseline [Time frame: Data generated at enrollment.]
• Prediction of Responder Status using Tissue Parameter Maps at Baseline and Mid-Therapy [Time frame: Enrollment to two weeks after therapy start.]