This was an open-label, randomized controlled study. Participants were randomized in a 1:1:1 fashion to receive dual-target, high-dose TMS (DHT), dual-target, conventional-dose TMS (DCT), and single-target, conventional-dose TMS (SCT). SCT was designated as the control group with protocol reference to evidence-based guidelines, while DHT and DCT served as the experimental groups.
DHT refers to the left primary motor cortex of the lower leg (M1) and dorsolateral prefrontal cortex (DLPFC) receiving 9000 pulses/day of intermittent TBS (iTBS) for 5 consecutive days. DCT refers to the left M1 and DLPFC receiving 1800 pulses/day of iTBS for 5 consecutive days. The order of stimulation for the two targets was randomized across participants and the order of stimulation within participants remained unchanged throughout the session. SCT refers to the left M1 receiving 1800 pulses/day of iTBS treatment for 5 consecutive days.
Motor and cognitive measures were performed ("on medication" state) 1 day before TMS (baseline), 1 day after completion of TMS (post), and 1 month after completion of TMS (follow-up). The primary outcome of this study was the change in FOGQ scores from baseline to follow-up. Secondary outcomes included changes in the Unified Parkinson's Disease Rating Scale Part III (UPDRS III), 5 m Timed Up-and-Go test (TUG), FOG-provoking test (Standing-Start 180° Turn Test, SS-180), Stroop color-naming (SCN), Stroop word-reading (SWR), Stroop color-word (SCW), and Color Trails Test interference index (CTTII) from baseline to follow-up.
Evaluate the patient's brain function and structural changes by collecting Rest-EEG，transcranial magnetic stimulation-EEG, near-infrared functional brain imaging (fNIRS）and fMRI.
TMS was performed using a Magstim Rapid2 transcranial magnetic stimulator (Magstim Company, Whitland, UK) with a 70-mm air-cooled figure-of-eight coil. All stimulations were guided by the participant's anatomical image (1 × 1 × 1 mm3) and a frameless neuronavigation system (Brainsight; Rogue Research, Montreal, QC, Canada).
Stimulation intensity: Because each participant had different sensitivity to TMS, the individual Resting Motor Threshold (RMT) was measured before intervention. Specifically, surface electrodes (Ag/AgCl) were attached to both ends of the right abductor pollicis muscle (the ground wire was connected to the ulnar styloid process) and the left finger motor cortex was stimulated one time. The motor-evoked potentials were recorded from the hand muscles. The RMT was considered when the evoked potential \> 50 μV occurred in \> 5 of 10 consecutive stimulations. The stimulation intensity during intervention was 80% of the RMT in the current study.
Stimulation sequence: Each iTBS sequence in the high-dose stimulation sequence released 900 pulses at a time with a pulse cluster repeated every 200 ms at a frequency of 5 Hz. Each pulse cluster contained three pulses with a frequency of 50 Hz, stimulation time of 2 s, and interval of 8 s. A total of 10 iTBS sessions was performed each day with an interval of 40 min and the daily stimulation dose was 9000 pulses. Each iTBS sequence in the conventional-dose stimulation sequence released 600 pulses at a time with a pulse cluster repeated every 200 ms at a frequency of 5 Hz. Each pulse cluster contained three pulses with a frequency of 50 Hz, stimulation time of 2 s, and interval of 8 s. A total of three iTBS sessions were performed each day with an interval of 40 min and the daily stimulation dose was 1800 pulses.
Stimulation targets: There were two stimulation targets in this study that were used to intervene in the motor and cognitive cortices. The stimulation target of the motor cortex is located in the lower leg of the left primary motor cortex in the Montreal Neurological Institute space (coordinates: -10, -24, 75 \[https:/ /afni.nimh.nih.gov/MNI\_Atlas\]) based on structural MRI. The stimulation target of the cognitive cortex was the left dorsolateral prefrontal cortex, which has the strongest functional connectivity with the executive control network (see Supplementary materials), based on fMRI. Finally, targets were transformed into the native space for each participant by applying an inverse matrix produced during brain structure and function segmentation using SPM12 (www.fil.ion.ucl.ac.uk/spm) and TMStarget software.
