A total of 32 patients [25 males (78%); mean (SD) age, 54.14 (9.02) years] were enrolled from Sichuan University West China Hospital Rehabilitation Medicine (Chengdu, Sichuan Province, China) between September 2019 and September 2020. Inclusion criteria were as follows: (1) age between 18 and 80 years (Koch et al., 2018); (2) first-ever unilateral ischemic or hemorrhagic stroke confirmed by computed tomography or magnetic resonance imaging; (3) subacute stroke survivors (stroke onset ranged from 2 weeks to 6 months) (Chien et al., 2020; Soulard et al., 2020); (4) having affected elbow flexors and wrist flexors spasticity with the modified Ashworth scale (MAS) score between 1+ and 3 (Barros Galvao et al., 2014); and (5) absence of cognitive impairment that is determined by the mini-mental state examination score is over 27 (Sun et al., 2014). Exclusion criteria included the following: (1) coexisting other neurological diseases; (2) cerebellar or brain stem injury; (3) used anti-spasticity drugs or injected botulinum toxin type A within 3 months before enrollment; (4) severe general impairment or concomitant diseases; and (5) contraindications for rTMS (e.g., history of seizures, intracranial metallic implants, cardiac pacemakers, and pregnancy).

This randomized, double-blind, sham-controlled trial was designed to explore the safety and the short-term efficacy of cerebellar iTBS on upper limb spasticity after subacute stroke. Eligible participants were randomly allocated in a 1:1 ratio to either cerebellar iTBS group or sham stimulation group. The randomization sequences were generated based on the table of random digits and were concealed in opaque numbered envelopes, which were opened in numerical order by a neutral non-involved researcher. All outcome measures were evaluated at baseline (T0) and after 10 sessions of intervention (T1). Each evaluation was performed by a clinician or by a physical therapist who was blinded to the experimental condition of the patient. Patients themselves were also unaware of the group assignment.

The sample size was estimated using G^*power of 3.1.9.2 (Faul et al., 2007), with the following parameters: effect size (d) = 1.35, α = 0.05 (two tails), power (1–β) = 90%, and allocation ratio n2/n1 = 1. The effect size was determined based on the result of our pilot study. After calculation, the necessary sample size of n = 26 was obtained. Considering the compliance of subjects, a total of 32 patients were required to allow for a 20% dropout rate.

This study was approved by the local institutional biomedical Ethics Committee on September 30, 2019, and complied with the Declaration of Helsinki. All patients provided written informed consent before the experiment. The trial was then registered on October 13, 2019, in the Chinese Clinical Trial Registry (registration number: ChiCTR1900026516).

Each patient received 1 session of cerebellar iTBS or sham stimulation daily, always before conventional physical therapy, for a total of 10 sessions. The overall intervention periods were 5 days/week for 2 consecutive weeks. All the patients did not use anti-spasticity drugs or injected botulinum toxin type A throughout the whole trial.

Professional physical therapists monitored adverse effects throughout the trial. Data management and analyses were performed using GraphPad Prism 7.00 (GraphPad Software, Inc., La Jolla, CA, USA). Before undergoing statistical analyses, the normal distribution of data was evaluated by the Shapiro–Wilk test. Continuous variables, ordinal variables, and categorical variables were, respectively, presented as mean (±standard deviation, SD), medians (interquartile range, IQR), and number (percentage, %). The level of significance was set at α = 0.05. Descriptive analyses were conducted to show the demographic and clinical characteristics of subjects. Fisher's exact test and unpaired t-test were used to evaluate the differences between the groups in the distribution of the characteristics of the subject at baseline.

For the MAS, the scores 0, 1, 1+, 2, 3, and 4 were converted to 0, 1, 2, 3, 4, and 5. The intervention efficacy within the group and between groups was analyzed with the Wilcoxon matched-pairs signed-rank test and the Mann–Whitney U test, respectively. For other outcome measures, a paired t-test of raw data was performed to evaluate the treatment effects within groups. Additionally, an unpaired t-test of changes between T0 and T1 was conducted to analyze the difference between groups.

After the screening based on the inclusion and exclusion criteria, a total of 32 out of 404 patients recruited were identified as being eligible for the trial. The whole procedure was well-tolerated, and no adverse events were reported in either group. During the intervention period, three subjects withdrew because two subjects were discharged in advance and one subject was transferred to another hospital that is near home (Figure 1). At baseline, no significant between-group differences were found in age, sex, time since stroke, types of stroke, paretic side, and the severity of stroke assessed by NIHSS (Table 1).

Table 2 lists the descriptive data for all outcome measures in both groups at T0 and T1. Figure 2 provides the statistical analysis results between groups.

Our study systematically assessed elbow flexors and wrist flexors spasticity by clinical, electrophysiological, and biomechanical measurements. Both clinical and biomechanical measurements showed significant improvements after cerebellar iTBS when compared with sham stimulation.

From a clinical point of view, we found that cerebellar iTBS coupled with conventional physical therapy decreased MAS scores of elbow flexors and wrist flexors from 3 to 2 points, passing from a marked increase level (2) to a slight increase level (1+) (Bohannon and Smith, 1987). In addition, the result of the difference between groups showed that the median changes of MAS scores of both elbow flexors and wrist flexors decreased one level in the cerebellar iTBS group when compared with the sham stimulation group, revealing that the efficacy of the cerebellar iTBS combined with conventional physical therapy was significantly better than that of the conventional physical therapy alone. Notably, the above changes reached the minimal clinically important difference that indicates a clinical significance was detected (Chen et al., 2019). Consistent with the result of MAS, MTS was also significantly improved by the 2 weeks of cerebellar iTBS coupled with conventional physical therapy, with the average angle differences between R2 and R1 of elbow flexors and wrist flexors reduced 43.00 degrees and 51.93 degrees, respectively. The range of motion at the upper limb is associated with cosmesis, hygiene, and active movement capabilities (Malhotra et al., 2011). Therefore, the increase in the range of motion at the upper limb can remove some participation restrictions and improve the quality of life in individuals with stroke.

From an electrophysiological point of view, spasticity is associated with the over hyperexcitability of spinal alpha motor neurons. The H-reflex is commonly used to quantify the level of spinal alpha motor neuron excitability (Pizzi et al., 2005). The H_max/M_max ratio is proposed as a measure of the percentage of motoneurons activated by eliciting the monosynaptic H-reflex compared with those directly activated (Okuyama et al., 2018), which presents high reliability and good sensitivity in detecting changes in spasticity (Levin and Hui-Chan, 1993). In this study, no significant improvements in the H_max/M_max ratio were found between groups. It may be attributed to the high variability of H-reflex in measuring the median nerve (Kim et al., 2015). Another possible reason is that the stimulation intensity of iTBS (80% aMT) was lower than the intensity of conventional rTMS (100% rMT) to induce electrophysiological changes in PSS improvement. Kim et al. also reported similar findings that iTBS over the affected motor cortex did not affect the H-reflex evoked in flexor carpi radialis (Kim et al., 2015).

From a biomechanical point of view, shear wave ultrasound elastography is feasible in muscle hardness assessment that may offer a better quantification of spasticity compared with clinical and electrophysiological measurements (Gao et al., 2018; Chen et al., 2019). In contrast to normal muscle cells, spastic muscle cells from stroke had shorter resting sarcomeres and increased elastic moduli, indicating that muscle stiffness can reflect the disease-related alterations in tissue properties (Wu et al., 2017). As shear waves travel faster in stiffer tissues, greater SWV and echo intensity were detected in spastic muscles (Lee et al., 2015). Our data demonstrated that the average SWV values of both biceps brachii and flexor carpi radialis were decreased in both groups after the intervention, and the cerebellar iTBS coupled with conventional physical therapy significantly decreased more SWV than sham stimulation. It implied that the therapeutic effect of cerebellar iTBS on changing muscle tissue properties was relatively obvious.

In agreement with relevant reported research published in 2014 (Barros Galvao et al., 2014), it was found that physiotherapy combined with additional low-frequency rTMS on unaffected primary motor cortex was more effective than physiotherapy alone in reducing upper limb spasticity in patients with chronic stroke (Barros Galvao et al., 2014). Furthermore, the benefits of iTBS on the affected motor cortex as an effective intervention to improve PSS had been identified by Kim et al. (2015). Their research found that a single session of iTBS contributed to a transient improvement in upper limb spasticity after stroke, which proved that iTBS seems to be an effective adjuvant to manage upper limb spasticity. Consistent with the above-reported studies, our study revealed significant improvements in upper limb spasticity after cerebellar iTBS in individuals with subacute stroke.

It is important to point out that the mean age of the subjects in the cerebellar iTBS group was 5.94 years older than that in the sham stimulation group, although it did not reach statistical significance. Age is a significant predictor of upper limb spasticity after stroke, with an odds ratio of 0.01 (Tedesco Triccas et al., 2019). The increase in aging individuals could have an impact on a higher incidence of upper limb spasticity. Besides, age is also a critical factor for predicting stroke outcome, with a negative correlation between age and the score of the motor component of the Functional Independence Measure and an active correlation between age and the length of hospital stay (Koyama et al., 2020). In our study, a significant decrease in upper limb spasticity was detected in the older cerebellar iTBS group patients.

The mechanism of iTBS over the ipsilesional lateral cerebellum reducing upper limb spasticity is still unclear. iTBS consists of high-frequency stimulation bursts, and the stimulus pattern is based on the natural theta rhythm occurring in the hippocampus of the brain, which can strongly modulate the neural activity of the cerebellum (Klomjai et al., 2015; Koch et al., 2019). An animal study reported that there were two modes of synaptic plasticity in the cerebellum, long-term depression (LTD) and long-term potentiation (LTP) (Jörntell and Hansel, 2006). Cerebellar iTBS can induce plasticity changes in the cerebellum of stroke patients (Koch et al., 2018). Besides, the molecular evidence also suggested that high-frequency rTMS could induce neural plasticity in the cerebellum associated with LTD (Lee et al., 2014). iTBS over the ipsilesional lateral cerebellum can increase the activation of Purkinje cells; the inhibitory synaptic connections between Purkinje cells and deep cerebellar nuclei are enhanced so that the neural plasticity in the cerebellum has been regulated (Fernandez et al., 2018). Besides, cerebellum iTBS can influence the activities of spinal neurons involved in the muscle tone adjustments in two ways: the inhibitory synaptic connections between Purkinje cells and dentate nucleus reduce the tonic excitatory effect of the dentate nucleus over the contralateral cerebral cortex through the ventrolateral nucleus of the thalamus; alternatively, interposed nuclei (globose and emboliform) and fastigial nucleus directly affect both the medial and lateral descending motor systems to reduce muscle spasticity (Teixeira et al., 2015). The relevant descending pathways include the corticospinal, reticulospinal, vestibulospinal, rubrospinal, and tectospinal tracts (Matsugi and Okada, 2020). Considering the results of our study, no significant difference in corticospinal excitability assessments (MEP latency and amplitude, and CMCT) was observed between groups. Therefore, the cerebellar iTBS improved upper limb spasticity may attribute to promoting the functional connection between the cerebellum and other brain areas rather than concerting with the cerebral cortex.

Physiological evidence shows that the cerebellum can interfere in the muscle tone adjustment by regulating neuronal discharge in different brain stem nuclei, primarily the reticular formation, red nucleus, and vestibular nucleus. Besides, the vestibular nucleus is involved in the alpha motor neuron activation, while the reticular formation and red nucleus are involved in the gamma motor neuron activation (D'Angelo, 2018). Chothia et al. found that anodal cerebellar direct current stimulation regulated the reticular spinal tract and rubrospinal tract to affect the motor neurons of the spinal cord (Chothia et al., 2016).

Other than motor areas, cerebellar iTBS also affects non-motor areas. Casula et al. found that the induction of cerebellar plasticity by iTBS was also associated with relevant changes in the neural activity of the posterior parietal cortex (Casula et al., 2016). The posterior parietal cortex participates in the perception and processing of action-related information and encodes the more abstract aspects of sensorimotor control processes, which is involved in the upper limb rehabilitation in patients with PSS (Veverka et al., 2019).

In our study, increases in BI scores were shown in both groups after interventions. These effects are likely due to the course of coupled 2 weeks of daily conventional physical therapy, independently from the cerebellar iTBS treatment. Conventional physical therapy has been confirmed as an effective way for the recovery of function after stroke (McDonnell and Stinear, 2017), whereas iTBS is an adjuvant tool to reinforce the therapeutic effect of conventional physical therapy.

We acknowledge that some limitations still existed in this study. First, the lack of follow-up did not allow us to explore the long-term efficacy of cerebellar iTBS. Second, the potential mechanisms of cerebellar iTBS need to explore further to confirm our hypothesis. Therefore, high-quality randomized controlled trials with larger sample sizes to investigate the long-term efficacy and potential mechanisms of cerebellar iTBS are recommended for future studies.