Introduction

Upper limb spasticity (ULS) is an incapacitating post-stroke motor deficit since it significantly impacts activities of daily living due to functional loss, joint stiffness, and pain [1, 2]. The neurophysiological mechanisms for post-stroke ULS development are still poorly understood. Spasticity results of an increase in the excitability of the stretch reflex circuit which is regulated by excitatory and inhibitory descending signals of supra-spinal origin [3,4,5,6]. Since the motor cortex has an inhibitory control over spinal cord through the corticospinal tract [7, 8], it is acknowledged that cortical injuries could lead to loss of such control and an increase in spinal cord excitability, resulting in spasticity [9, 10]. Non-invasive brain stimulation therapies as repetitive transcranial magnetic stimulation (rTMS) could be used to increase the cortical excitability, resulting in a normalization of the motor cortex inhibitory control over spinal excitability [11, 12].

Previous studies have examined the efficacy of rTMS coupled with training in reducing ULS. Kakuda et al. [12] reported that 22 sessions of 1-Hz rTMS combined with occupational therapy reduced spasticity and improved motor function of stroke-affected upper limb. Similarly, Galvão et al. [13] noted a reduction of post-stroke ULS after ten sessions of 1-Hz rTMS coupled with physical therapy (PT). Conversely, Etoh et al. [14] showed no improvement of ULS after ten sessions of 1-Hz rTMS combined with 40 min of repetitive exercises. Additionally, despite a small improvement in upper limb measures, including spasticity, Rose et al. [15] concluded that 16 sessions of 1-Hz rTMS followed by 1 h of functional tasks did not lead to additional improvements in upper limb function of chronic stroke patients. Insights into how rTMS when combined with motor training acts on cortical and spinal excitability could be a key-factor for a successful implementation of rTMS as adjuvant therapy in neurorehabilitation.

Changes of the cortical and spinal activity that might underlie anti-spastic effects of rTMS are still poorly understood. To the best of our knowledge, only two studies [16, 17] have investigated the effects of 1-Hz rTMS over post-stroke spasticity through electrophysiological measures as Hmax/Mmax ratio (ratio between the maximum amplitude of the H-wave and the maximum amplitude of the M-wave). However, none of them coupled rTMS with training and both of them evaluated spasticity of lower limb.

Our study is rationally based on the theory of interhemispheric competition after stroke [18]. Healthy individuals have a balanced interaction between hemispheres. After a brain damage, this interaction becomes unequal due to a reduced activity of intracortical neuronal circuits of the affected hemisphere added to an excessive interhemispheric inhibition of the unaffected hemisphere. Considering such model, suppressing of the unaffected hemisphere excitability would reduce the interhemispheric inhibition and lead to an enhancement of descending inhibitory input through the corticospinal tract, followed by a decrease of the moto neuron excitability, resulting into a reduction of spasticity. We explored whether 1-Hz rTMS over the unaffected hemisphere of post-stroke patients associated with PT can modulate cortical as well as spinal cord excitability. In addition, the effects of such association over the ULS of patients was daily assessed during ten sessions.

Methods

Study design

A randomized, double-blind, sham-controlled trial with 4 weeks of follow-up was performed. Patients were randomly assigned into two groups: (i) experimental—rTMS and PT or (ii) control—sham rTMS and PT. A researcher not involved in the study performed the allocation sequence using a web-based computer software (www.randomization.com).

All patients and researchers who were involved in PT and evaluations were blinded to group allocations; only the researcher who administered rTMS was aware of the treatment group. This study was registered at ClinicalTrials.Gov (NCT01875536), was performed in accordance with the Declaration of Helsinki, and was approved by the local research ethics committee.

Participants

Participants were recruited through advertisements in a local newspaper, university website, and by screening outpatient clinics in local hospitals. Inclusion criteria were (i) ischemic/hemorrhagic stroke diagnosed by a neurologist and confirmed through CT or MRI, (ii) stroke onset ≥ 6 months and < 10 years, (iii) age between 40 and 75 years, (iv) wrist muscle tone score at modified Ashworth scale (MAS) [19] between 1 + and 2, and (v) absence of cognitive impairments, as determined by the Mini-Mental State Examination [20] (score > 20). Exclusion criteria were the following: (i) had clinical evidence of multiple brain lesions attested by physical examination and complementary imaging as CT or MRI, (ii) used antispastic drugs within the 6 months before enrollment, (iii) were pregnant, (iv) were using metallic implants or cardiac pacemakers, or (v) had a history of seizure or cerebral aneurysm. All patients gave written informed consent before starting the experiment.

Outcome measures

Interviews and clinical evaluations were carried out with all participants for collection of clinical data by an experienced blinded assessor.

Electrophysiological measures

Motor cortex excitability was determined by the intensity of the magnetic stimulator output (MSO), expressed as percentage of maximal stimulator output inducing motor-evoked potential (MEP) amplitude of 1 mV elicited by single-pulse TMS. TMS was performed using a magnetic stimulator (NEUROSOFT-Neuro-MS, Russian), connected to a 70-mm figure-8 coil positioned tangentially to the scalp at 45° from the midsagittal line over the primary motor cortex of the unaffected hemisphere. The coil was positioned at patient’s hotspot—the site where stimulation resulted consistently in the largest MEPs. Surface electromyography (EMG) recording was made from the first dorsal interosseous (FDI) muscle with Ag-AgCl surface electrodes. Responses were amplified and bandpass-filtered at 2 kHz. All data were digitized at an analog-to-digital rate of 5 kHz and further relayed into a laboratory computer using the Neuro-MEP-Micro software (Neurosoft Company, Russian). Motor cortex excitability was recorded at baseline and at each treatment session. Before each assessment, intensity was adjusted to elicit, on average, baseline MEPs of 1 mV peak-to-peak amplitude. The coil position was marked with a waterproof pen to guarantee identical position during the whole course of the study.

Excitability of the spinal cord was assessed through the variance in amplitude of the Hoffmann reflex (H-reflex). H-reflex measures excitability of motor neurons and that of the spinal cord indirectly and is used to evaluate spasticity [21]. H-reflex amplitude was normalized to the maximum muscle response using the Hmax/Mmax ratio. Conceptually, the Hmax/Mmax ratio reflects the proportion of motor neurons that are reflexly activated [22]. Hmax expresses the maximum amplitude, peak-to-peak, of the H-wave, and Mmax indicates that of the M-wave. Both indices were measured on the paretic side through peripheral electrical stimulation (rectangular pulses, 1-ms duration; frequency 0.2 Hz) of the median nerve and recorded by surface electromyography.

Electrodes were positioned as stated by Palmieri et al. [23]. Electromyographic data were collected at a sampling frequency of 2500 Hz, amplified, and bandpass-filtered (10–1000 Hz). H-reflex was measured at baseline, after ten treatment sessions (i.e., post-intervention) and at the 4-week follow-up.

Clinical measures

Before each session, Modified Ashworth scale (MAS) was recorded to determine spasticity of wrist flexor muscles in affected hand. This instrument is a 6-point scale that scores the average resistance to passive movements for each joint [24].

Interventions

Repetitive transcranial magnetic stimulation

rTMS of the motor cortex (1-Hz, 1500 pulses with an intensity of 90% of the rest motor threshold-MT [8]) was administered over the hotspot of FDI muscle of the unaffected hemisphere, three times per week—always before PT intervention—for a total of ten sessions. MT was defined as the lowest TMS intensity required to produce a MEP amplitude > 50 μV in at least six of ten trials. For each patient, coil position during rTMS sessions was defined by the place where MT was recorded.

In the sham rTMS, a coil disconnected from the stimulator was held over the scalp, while a second coil connected to the stimulator was positioned behind the patient’s head without touching the scalp. Patients were only exposed to acoustic stimulation and were not aware which of the coils were touching the scalp.

Physiotherapy

Thirty minutes of the PT program was applied immediately after stimulation with activities being adjusted according to patient’s functional capacity and focused primarily on upper limb rehabilitation. Specifically, the program had five objectives: (i) improve patient’s flexibility (stretching exercises of the flexors of wrist/fingers, biceps brachii, and pectoralis major and mobilization of the cervical muscles); (ii) improve strength (exercises of proprioceptive neuromuscular facilitation); (iii) improve coordination and balance; and (iv) improve patient’s mobility during transfers activities. All activities were based on recommendations of clinical practice guidelines for stroke patients [25].

Statistical analysis and data processing

Descriptive statistic was used to present demographic and clinical characteristics of participants. Groups were compared at baseline characteristics using an independent t test or chi-square test (χ2).

All data met the criterion for normal distribution (i.e., Kolmogorov-Smirnov test, p > 0.05). For cortical excitability, the MSO mean difference between each treatment session and first session was analyzed via repeated measures analysis of variance (rm-ANOVA), with Group (2 levels: experimental and control) as between-subjects factor, and Time (10 levels: from the first to the tenth session of treatment) as within-subjects factor. For spinal cord excitability, the difference in percentage of variance in the Hmax/Mmax ratio at post-intervention and at the 4-week follow-up was analyzed via rm-ANOVA, with Group (2 levels: experimental and control) as between-subjects factor, and Time (2 levels: post-intervention and 4-week follow-up) as within-subjects factor. Sphericity assumption was tested by Mauchly’s Test and adjustments were applied using a Greenhouse-Geisser correction. Pair-wise comparisons were performed for within- and between-subject analysis (Bonferroni corrections were applied).

To facilitate data analysis, MAS scores 1 +, 2, and 3 were assigned numerical values 2, 3, and 4, respectively. Since MAS data did not meet the criterion of normality, the difference in MAS scores between each treatment session and the first session was calculated and evaluated using Friedman test. Intra-group analysis was performed by Wilcoxon test, while comparisons between groups were analyzed by Mann-Whitney test. All analyses were performed using SPSS for Windows, Version 20 (Armonk, NY: IBM Corp). Statistical significance was set at p < 0.05.

Results

Participant characteristics and flow of the trial

As shown in Fig. 1, out of a total of 148 patients who were screened for eligibility, 20 met study criteria and were randomized to the experimental (n = 10) and control group (n = 10). No difference was found between group at baseline (Table 1). No adverse events were reported by any of the participants.

Fig. 1
figure 1

Inclusion flow diagram

Full size image
Table 1 Clinical and demographic characteristics of participants
Full size table

Electrophysiological measures

Figure 2 shows differences on cortical excitability; rm-ANOVA revealed an interaction effect between time and groups (F(9, 162) = 1.944; p = 0.049), while no significant main effect of time (F(9, 162) = 1.164; p = 0.33) and group (F(1, 18) = 3.212; p = 0.09) were found. Pair-wise comparisons using the Bonferroni corrections revealed a significant difference between groups at post-intervention (p = 0.044) pointing out to an enhancement of cortical excitability (i.e., lower values of MSO in the experimental group). MSO showed a decreased trend over the ten sessions in the experimental group. Pair-wise comparisons with Bonferroni corrections showed that this reduction was significant at the tenth session (p = 0.028).

Fig. 2
figure 2

Difference of output stimulator (%) between the first therapeutic session and each treatment session of experimental group (dashed line, circle) and control group (solid line, square). Data are shown as mean and standard error. *Represents significant difference between groups. Filled symbols represent significant difference from the first therapeutic session

Full size image

Figure 3 shows differences in the spinal excitability measure throughout sessions. Two patients in each group were excluded from analysis due to poor reflex response. rm-ANOVA revealed a main group effect (F(1, 11) = 6088; p = 0.031). Bonferroni adjustment comparison showed significant difference on the Hmax/Mmax ratio between groups at 4-week follow-up (p = 0.021).

Fig. 3
figure 3

Variation (%) of Hmax/Mmax ratio from baseline in experimental group (dashed line, circle) and control group (solid line, square). Data are shown as mean and standard error. *Represents significant difference between groups

Full size image

Clinical measure of spasticity

Figure 4 depicts changes in MAS scores throughout sessions. Friedman test revealed a significant difference in MAS scores between groups over time (χ2(9) = 24.35; p = 0.004). Pairwise comparisons with Wilcoxon signed-rank test showed a decrease of MAS scores from the third session (p < 0.05) when compared to baseline in the experimental group, pointing out to a decrease of spasticity. No significant difference was observed in the control group. Mann-Whitney test revealed differences between groups from the sixth session (p < 0.05).

Fig. 4
figure 4

Variation of Modified Ashworth scale score (MAS) from baseline of the first session before each therapeutic session in experimental group (dashed line, circle) and control group (solid line, square). Data are shown as mean and standard error. *Represents significant difference between groups. Filled symbols represent significant difference from the first therapeutic session

Full size image

Discussion

To the best of our knowledge, this is the first double-blind RCT to investigate the effects of inhibitory rTMS coupled with PT on the cortical and spinal excitability, and as well as on the degree of ULS in stroke patients, for latter, tracking the effect daily during sessions. Our results suggest that rTMS coupled with PT decreases the level of ULS (beginning from the sixth treatment session) and the spinal cord excitability. Further, this association increases the cortical excitability in the unaffected hemisphere.

Effects of rTMS coupled with PT on cortical and spinal excitability

Based on the theory of interhemispheric competition [18], we have hypothesized that 1-Hz rTMS would decrease the excitability of unaffected motor cortex and restore the balance of transcallosal inhibitory circuits between the hemispheres, relieving the inhibition of the unaffected hemisphere and increasing the cortical activity of the affected hemisphere, as previously demonstrated [26]. However, we found that stimulation associated with PT increased the excitability of the unaffected hemisphere, as shown by the decrease of MSO in the experimental group. A possible explanation is that PT after effect interfered with prior inhibitory effect of 1-Hz rTMS reversing the suppressive effect of rTMS in facilitation. Supporting this view, studies in animals that have shown that physiological activity after induction of LTP/LTD can reduce, abolish, or reverse changes in synaptic plasticity [27,28,29]. This phenomenon has been also showed in human when two techniques of non-invasive brain stimulation were combined [30, 31]. Moreover, Huang et al. [32] have coupled inhibitory rTMS with motor training and observed that voluntary muscle contractions immediately after cortical stimulation increased the cortical excitability, reversing the suppressive effect of rTMS in facilitation.

Interestingly, increase in cortical excitability of the unaffected hemisphere was followed by decrease of Hmax/Mmax amplitude ration and spasticity. Indeed, the unaffected hemisphere seems to play a pivotal role in post-stroke rehabilitation [33]. In line with our results, earlier functional imaging studies have reported that an increased activity of the unaffected hemisphere might be related to the recovery of stroke patients [34, 35]. Also, Lotze et al. [36] reported that an inhibition of unaffected hemisphere areas can impair motor performance of the affected hand. Taken together, these findings suggest that the aim to strictly balance the interaction between hemispheres may be oversimplified [37]. Thus, other more complex models that take into account the lesion location and size, chronicity, and prior synaptic history should be investigated in order to advance in the developing of rTMS use as part of routine clinical practice.

Effects of rTMS coupled with PT on spasticity

Previously, we have shown that ten sessions of PT combined with inhibitory rTMS applied to the unaffected hemisphere are more effective than PT alone in reducing ULS in patients with chronic stroke [13]. Now, our results demonstrated that ULS significantly declines after the sixth session of treatment with rTMS and PT. This finding has an important clinical implication, suggesting that a minimum of six rTMS sessions are necessary in order to mitigate spasticity. Other studies have also reported antispastic effects of rTMS in neurological patients [38], but none has tracked the effects over time during treatment.

Since it is known that spasticity results from a hyperexcitability of the stretch reflex [39], we expected a reduction of spinal excitability along with MAS score decrease after treatment. Indeed, we noted such reduction throughout sessions only in the experimental group. Considering the decline in spasticity occurred from the sixth treatment session and a significant decrease of spinal excitability at follow-up, we suggest (i) the loss of muscular tone in response to passive stretching is not influenced merely by the decrease of the spinal excitability or (ii) the Hmax/Mmax ratio may lack clinical relevance when approaching spasticity. Indeed, the weak relationship between Hmax/Mmax ratio and clinical measurement for spasticity has been reported [40].

This study has some limitations. Firstly, some questions can be raised if an inclusion of patients with disease time ranging from 6 months to 10 years is not a time frame extremely wide. The time of development of paresis, as well as the amount of initial injury, can alter the final result of PT. Usually, better outcomes can be achieved when patients have high levels of adaptive neuroplasticity—such as at the first years after the stroke. Even though possible, the wide time frame used as inclusion criteria does not appear to have negatively impacted our results since 75% of patients (15 out of 20) had 5 years or less of disease. Additionally, age can be a key-factor for motor rehabilitation. Younger patients tend to respond in a better way than older ones. Once again, we used as inclusion criteria age between 40 and 75 years. It seems reasonable to assume that have in the same sample young and elderly patients could be a confounding factor, but 75% of our sample had 60 years or more. So, it can be assumed that they were more or less at the same motor capacity level.

Conclusion

We have demonstrated that 1-Hz rTMS over the unaffected motor cortex combined with PT decreased spinal cord excitability and ULS in chronic post-stroke patients. Besides evidence supporting the idea that rTMS combined with motor therapy may to be used in the management of spasticity, our study brings novel information on which manner this interaction occurs. This knowledge can be a key-factor for a successful implementation of rTMS as adjuvant therapy in stroke rehabilitation. Further studies with a larger number of patients may support the employment of rTMS in the PT arsenal.