Effect of cerebrospinal dual-site magnetic stimulation on freezing of gait in Parkinson’s disease

Freezing of gait (FOG), particularly when unresponsive to levodopa, is a challenging phenotype in Parkinson’s disease (PD). The widely cited Interference model posits that FOG arises from overload and impaired coordination among cognitive, limbic, and motor cortical–basal ganglia circuits. Modulating dysfunctional networks to rebalance connectivity between motor cortex, striatum, and subthalamic nucleus may alleviate FOG. Repetitive transcranial magnetic stimulation (rTMS) can modulate cortico-subcortical network connectivity, and the primary motor cortex representation of the lower limbs (M1-LL) is a plausible target, though prior efficacy has been inconsistent. Impaired corticomotor inhibition (CI) of the tibialis anterior (TA) muscle has been linked to FOG, and low-frequency rTMS over M1 can enhance corticospinal inhibition. The study first hypothesized that cumulative 1 Hz rTMS over bilateral M1-LL would ameliorate FOG in PD. A second hypothesis was that combining rTMS with transcutaneous magnetic spinal cord stimulation (SCS) would produce synergistic benefits beyond rTMS alone, given prior pilot data showing invasive and noninvasive SCS can improve FOG, potentially via modulation of supplementary motor area and gait initiation networks.

Prior work has produced mixed results for rTMS in FOG and gait in PD, with variability in targets and frequencies (e.g., M1, SMA; low- vs high-frequency). Pilot studies of epidural SCS (thoracic levels) in small PD cohorts demonstrated improvements in gait and FOG, including in levodopa-unresponsive cases. Case reports and small series suggested benefits of SCS on FOG and postural control. Noninvasive, transcutaneous magnetic SCS at mid-to-lower thoracic levels has recently shown feasibility and symptomatic improvements without severe adverse effects. Electrophysiological studies indicate altered corticomotor inhibition in PD with FOG, and low-frequency rTMS can increase cortical inhibition (reduced MEP amplitude, prolonged cortical silent period). These findings motivated testing cumulative low-frequency rTMS over bilateral M1-LL and its combination with transcutaneous magnetic SCS.

Design: Randomized, double-blind, sham-controlled, parallel-group clinical trial. Participants: 68 PD patients with levodopa-unresponsive FOG were screened; 57 were randomized. Inclusion: age ≥40, able to walk ≥30 m independently, stable antiparkinsonian medication for ≥4 weeks, FOG confirmed by FOG-Q and clinical observation during walking/turning/doorway tasks; levodopa-unresponsive FOG defined as FOG present in both medication ON and OFF states. Exclusion: TMS contraindications, severe dyskinesia/tremor, MMSE < 24, psychiatric disorders, interfering medications, other gait disorders. Ethics/registration: Approved by the Research Ethics Committee of the First Affiliated Hospital of Nanjing Medical University; registered NCT05174929. Randomization and blinding: Participants randomized into three groups using permuted blocks by an independent team member; all participants TMS-naïve and blinded to intervention; assessors blinded. Groups and interventions (10 sessions over 2 weeks; daily on weekdays, medication ON state): - DS (dual-site): 1 Hz rTMS over bilateral M1-LL plus transcutaneous magnetic SCS (T10–T12). - SS (single-site): 1 Hz rTMS over bilateral M1-LL plus sham SCS. - NS (no-site sham): sham rTMS plus sham SCS. rTMS parameters: Double-cone coil (Neurosoft, Russia); target localized by optimal MEP in target muscle; interfered side first then less-affected side; 800 pulses at 1 Hz per hemisphere (13 min 20 s), 120% resting motor threshold (RMT). Sham rTMS used a disconnected coil with another coil to mimic sound; earplugs worn by all. SCS parameters (DS only): 1 Hz, 1500 pulses, figure-of-eight coil over T10–T12 (lumbar enlargement level); intensity set to visible muscle twitch; ~25 minutes; sham SCS with coil at 90°. Assessments: Baseline, 1 day post-intervention (Post), and 1 month post-intervention (Post1m). Primary outcome: FOG-Q. Secondary outcomes: UPDRS-III, gait metrics (gait speed, stride length, stride time variability, double support time/percentage). Adverse events collected after each session. Electrophysiology: Corticospinal/cortical excitability recorded from more-affected side: APB and TA. Measures: RMT, MEP amplitude at set intensities (AMP), cortical silent period (CSP), short-interval intracortical inhibition (SICI; ISI 4 ms), intracortical facilitation (ICF; ISI 15 ms), short-latency afferent inhibition (SAI; median nerve conditioning, ISIs around N20 latency). SICI/ICF with conditioning at 80% RMT, test at 130% RMT; 10 trials per ISI. CSP at 150% RMT during 20% tonic contraction. EMG bandwidth 20–2000 Hz, sampling 5 kHz. Statistics: Normality by histograms and Shapiro–Wilk. Baseline comparisons by one-way ANOVA or Kruskal–Wallis; gender by chi-square. Repeated-measures two-way ANOVAs for TIME (Baseline, Post, Post1m) × GROUP (DS, SS, NS). For electrophysiology, repeated-measures ANCOVAs with LEDD as covariate. Significant interactions followed by Bonferroni-corrected post hocs; FDR correction applied for multiple tests. Correlations between changes in electrophysiology and behavior explored with LEDD as regressor. Significance p < 0.05. Analyses in SPSS 27.

Participants: 57 randomized (approx. 19 per group), with comparable demographics, baseline clinical, gait, and electrophysiological characteristics across groups. Safety: No serious adverse events. Five subjects reported transient headaches during initial rTMS session; four completed the study without recurrence; one discontinued further treatment due to apprehension. Adverse event distribution: DS n=3, SS n=2. FOG (primary outcome): Significant effects on FOG-Q—TIME (F ≈ 26.148, p < 0.001, partial η² = 0.506), GROUP (F ≈ 4.343, p = 0.018, partial η² = 0.143), and TIME × GROUP interaction (p < 0.001, partial η² = 0.542). Both DS and SS improved at Post and Post1m versus Baseline (all p < 0.01). DS showed superior mitigation of FOG compared to SS and NS. Motor symptoms (UPDRS-III): DS and SS improved versus Baseline at Post and Post1m, with narrative reductions of ~5–6 points sustained at 1 month; NS showed no improvement. Gait outcomes: DS and SS decreased double support time at Post (DS p = 0.002; SS p = 0.016); NS increased double support time at Post (p = 0.008). DS was superior to NS in improving gait speed, stride length, and double support time (all p < 0.05). Overall, SS showed some gait improvements, but DS exhibited superior performance across gait metrics. Electrophysiology: No significant pre–post changes in TA corticospinal excitability. For APB, significant TIME × GROUP interaction for SICI (SICIAPB: F = 12.340, p < 0.01, partial η² = 0.330). Both DS and SS increased SICIAPB relative to Baseline (p < 0.01). SS improved SICIAPB versus sham (p = 0.005), and DS produced greater SICIAPB recovery than SS (p = 0.048) and NS (p < 0.001). No significant changes in AMP, CSP, ICF, or SAI of APB. Correlations: In DS, greater SICIAPB improvement correlated with greater increases in gait speed (r = -0.738, p = 0.004) and stride length (r = -0.740, p = 0.007), indicating tighter coupling of gait gains with restored cortical inhibitory function.

Low-frequency (1 Hz) rTMS over bilateral M1-LL safely improved FOG, motor function, and gait in PD patients with levodopa-unresponsive FOG, with benefits persisting for at least one month. Adding transcutaneous magnetic SCS (dual-site cerebrospinal protocol) enhanced clinical efficacy beyond cortical stimulation alone, particularly for FOG and gait metrics. Electrophysiological findings showed increased intracortical inhibition (greater SICI in APB) after active interventions, most pronounced with dual-site stimulation, and the magnitude of inhibitory restoration correlated with gains in gait speed and stride length. These results support the concept that targeting distributed motor networks and rebalancing cortical–subcortical and possibly spinal circuitry can alleviate FOG, aligning with models emphasizing interference among cognitive–limbic–motor loops. The synergy of rTMS with magnetic SCS may engage supplementary motor area and brainstem/spinal locomotor circuits, contributing to improved gait initiation and stability.

Cumulative low-frequency rTMS over bilateral M1-LL improves levodopa-unresponsive FOG, motor symptoms, and gait in PD. Combining rTMS with transcutaneous magnetic SCS confers superior benefits compared with rTMS alone, with associated enhancement of cortical inhibitory mechanisms. The cerebrospinal dual-site regimen emerges as a promising therapeutic strategy for levodopa-unresponsive FOG and gait dysfunction in PD. Future work should refine targeting and dosing, include dedicated SCS-only arms, and leverage advanced neuroimaging to map network-level mechanisms.

- No neuronavigation was used for rTMS targeting; EMG-based localization is less precise, potentially reducing focality and reproducibility. - Sham design may have been suboptimal in sensory mimicry and blinding efficacy, though all participants were TMS-naïve. - No dedicated transcutaneous magnetic SCS-only arm, limiting isolation of SCS effects. - Use of a double-cone coil reduces focality and may stimulate regions beyond bilateral M1-LL. - Auditory confounds from TMS clicks were mitigated with earplugs/sham, but adaptable auditory control could better minimize sound-related effects. - Stimulation delivered via the TMS manufacturer's dashboard; automation could improve objectivity and reproducibility. - Lack of advanced imaging (e.g., fMRI, MRS) to validate network connectivity and neurotransmitter-related changes; limited systematic EMG response recordings. - Some outcome table values appear internally inconsistent in the manuscript, warranting cautious interpretation of absolute numeric changes.