Testing spasticity mechanisms in chronic stroke before and after intervention with contralesional motor cortex 1 Hz rTMS and physiotherapy

Fifty-four chronic stroke patients (38 males, 16 females) with an average age of 58 ± 12 years participated in this study. All patients had unilateral hemiparesis due to stroke at least 6 months prior to inclusion (46 ± 42 months). The study was conducted at the TMS outpatient clinic, Department of Neurology, Tübingen University Hospital in Germany between September 2019 and March 2022.

Before enrolling in the study, a neurologist thoroughly examined the patients and determined their suitability for receiving rTMS intervention based on the clinical safety guidelines put forward by Rossi et al. [65]. The criteria for receiving rTMS included: (1) no seizures, or a seizure-free period of at least 4 months prior to the commencement of the therapy; (2) for patients with intracranial implants (e.g., aneurysm clips, shunts, stimulators), a minimum distance of 8 cm between the TMS coil and the implant. Because this study is primarily concerned with examining the physiological changes which relate to rTMS intervention, patients were included in the study even when they did not exhibit any spasticity at the wrist. Patients who suffered from severe contractures in the wrist joint which impeded the placement of the hand in the orthosis (n = 3) were not measured using the hand-held dynamometer. Detailed patient characteristics are provided in Table 1. After the clinical screening, anatomical T1-weighted magnetic resonance imaging (MRI) was acquired for MRI-guided TMS neuronavigation.

To determine optimal stimulation sites and coil orientation during the therapy, we first mapped the area surrounding the precentral gyrus of the motor cortex in the non-lesioned hemisphere utilizing the patient MRI images integrated within the neuronavigation system (Nexstim, Helsinki). Biphasic single pulse TMS (with a posterior anterior initial phase) was used to identify the motor hotspot for the abductor pollicis brevis (APB) and first dorsal interosseous (FDI) muscles. Surface EMG recordings were acquired using bipolar adhesive electrodes (Ambu® Neuroline 720, Copenhagen) placed on the bellies of APB and FDI muscles, with a reference electrode on the ulnar styloid process. TMS was delivered using a 72 mm figure-of-eight coil figure-of-eight coil (Nexstim Focal Coil, Nexstim, Helsinki) placed perpendicular to the motor cortex. Surface EMG recordings were acquired using bipolar adhesive electrodes (Ambu® Neuroline 720, Copenhagen) placed on the bellies of APB and FDI muscles, with a reference electrode on the ulnar styloid process. TMS was delivered using a 72 mm figure-of-eight coil figure-of-eight coil (Nexstim) placed perpendicular to the motor cortex. The motor hot spot was defined as the location which elicited the largest motor-evoked potentials (MEPs) in APB or FDI. The resting motor threshold (RMT) was defined as the lowest stimulation intensity which triggered an MEP with a peak-to-peak amplitude of at least 50 µV in at least 5 out of 10 consecutive stimulations delivered to the motor hot spot. The muscle with lower RMT served as the reference for the therapy. The neuronavigation system with head and TMS coil trackers, along with a stereotactic camera enabled precise registration of the TMS coil's location relative to the patient's head and specific brain area.

During therapeutic rTMS, subjects were seated in an electronically adjustable reclining chair with relaxed arms. The navigated brain stimulation system delivered 1200 biphasic pulses at 1 Hz to the motor hot spot in the contralesional motor cortex at a stimulation intensity of 120% of the reference muscle's RMT utilizing the same coil location and orientation as registered during the mapping session. The use of the neuronavigation system enabled the consistent targeting of the initial hot spot throughout the intervention. Each subject received 3 rTMS sessions per week for a duration of 4–6 weeks (~ 15 sessions). A single rTMS session lasted ≈30 min including 10 min of preparation and 20 min of rTMS.

In order to enhance the effects of rTMS for motor rehabilitation purposes, patients received 50 min of personalized exercise-based physiotherapy immediately following the end of the rTMS session. The training focused on the arm and hand function including mobility exercises, strength training, object manipulation and fine motor training.

A battery of assessments took place before the start and at the end of the intervention. These included (a) clinical evaluation of spasticity, motor impairment and motor function, as well as patient-centered outcome measures; (b) objective evaluation of spasticity using a hand-held dynamometer; (c) electrophysiological measurement of three spinal circuit mechanisms. All measurements were performed by a single experimenter who is an experienced researcher and physiotherapist (first author).

The clinical tests included the Fugl-Meyer Assessment scale-upper extremity (FMA-UE), the Wolf motor function test (WMFT), and the modified Ashworth scale (MAS).

The FMA-UE [66] is a quantitative measure of upper limb impairment in stroke patients. It includes 33 performance-based items that are rated on a 3-point ordinal scale (0 cannot perform, 1 can perform partially, and 2 can perform fully) with a maximum total score of 66 points.

The Wolf Motor Function test (WMFT) [67] includes 15 timed tasks for the evaluation of upper limb motor function. Both the duration (in seconds) required to perform each task as well as the quality of the movement are recorded. Movement quality is measured on an ordinal scale between 0 and 5 (0: no movement attempt, 1: movement attempt failed, 2: minimal synergy, 3: significant synergy, 4: almost normal, 5: movement is normal). The total time score is the sum of the time required for each task (when the task is not completed within 120 s, the performance time is recorded as 120 s), and the quality score is the sum of quality scores for all the tasks, with a maximum score of 75.

To estimate the MAS score in the wrist extensors, the examiner moved the wrist joint from full flexion to full extension and simultaneously estimated the perceived resistance according to a 6-point ordinal scale (0: no increase in muscle tone, 1: slight increase in muscle tone at the end of the ROM, 1+: slight increase in muscle tone throughout less than half of the ROM, 2: marked increase in muscle tone in most of the ROM, 3: passive movement difficult, 4: affected parts rigid). To get a total MAS score for the upper limb, we repeated the same measurement procedure for multiple joints and added the individual MAS scores of 14 movements in these joints (flexion, abduction, internal rotation and external rotation; of the shoulder; flexion and extension of the elbow; pronation and supination of the forearm; flexion and extension of the wrist; flexion and extension of the fingers; and adduction and abduction of the thumb). The score 1 + was transformed into 1.5 to allow mathematical addition of the scores.

Three questionnaires were used to capture the subjective experience of the patients: (1) visual analogue scale (VAS) where patients reported the degree of spasticity they perceived in their arm on a scale that ranged between 0 and 100, where 0 is no spasticity at all and 100 is the highest spasticity imaginable; (2) Motor Activity Log (MAL-30), a structured interview where patients described the frequency with which they used the affected limb to perform each of 30 different activities of daily living [68, 69]. The scores range between 0 and 5 where 0 is never and 5 is exactly as much as before the injury. The reason why the patient did not use the affected limb for a particular task was also noted. Whenever the reason was related to side dominance (e.g., writing with the left affected side when the patient is right-handed) or when the activity is irrelevant to the patient’s life (combing hair when the patient is bold), the question was discarded and the highest score for that question (5 points) was subtracted from the total score. The final score was then expressed as a ratio; (3) Disability rating scale [70, 71] a scale that is concerned with the effect of spasticity on activities of daily living. Patients were asked to report the difficulty they experienced in performing six activities on a scale from zero to five (0 no difficulty and 5 cannot perform the task). The activities included: palm hygiene, entering the spastic arm in a sleeve, washing the underarm, washing the elbow, cutting fingernails and doing exercises with the arm.

The assessment of resistance to wrist extension was done using the Portable Spasticity Assessment Device (PSAD) (Movotec, Denmark). The PSAD is a hand-held dynamometer which encloses multiple sensors including dynamometers, accelerometers, and a gyroscope to enable the simultaneous acquisition measurement of force, joint movement, and muscle activity [29, 33‐37]. The PSAD has been validated for measurement of passive and reflex-mediated stiffness in the ankle joint plantar flexors in previous studies. We have demonstrated validity and reliability of the device for quantifying the reflex-mediated stiffness in the wrist flexors in chronic stroke patients [38] (Fig. 1).

To assess the passive and reflex-mediated components of the resistance to wrist extension, the experimenter moved the wrist joint throughout the available range of motion by applying force through the handle of the device at slow and fast velocities. Slow stretches (< 20°/s) were intended to measure the passive stiffness in the absence of a stretch reflex. Reflex-mediated stiffness, on the other hand, is measured by moving the joint with a sufficiently high velocity (> 300°/s) to trigger a stretch reflex in the wrist flexors [32, 72]. Throughout the assessment, the forearm of the patient was firmly fixed by the experimenter to ensure that the applied stretching force was translated into a rotational torque at the level of the wrist joint alone.

EMG signal was sampled with a frequency of 1 kHz from Flexor Carpi Radialis (FCR) and Extensor Carpi Radialis (ECR) muscles using bipolar surface adhesive electrodes (Covidien-Kendall, 24 mm) placed over the belly of the corresponding muscles with an inter-electrode distance of 2 cm (Fig. 1A). The ground electrode was placed on the ulnar styloid process. Before the placement of the EMG electrodes, the skin was prepared using abrasive gel to improve the contact with the electrodes.

The PSAD was wirelessly connected to a data acquisition software which provided online visual feedback on the velocity of the stretch and EMG activity in FCR and ECR muscles (Fig. 1B). Patients were encouraged to relax completely and to avoid actively helping or resisting the movement. When visible EMG activity was detected before the stretch or during slow stretches, the measurement was rejected and repeated after encouraging and helping the patient to relax. Each measurement session was composed of six slow and six fast stretches. All acquired data including position, acceleration, angular velocity, EMG data and forces were saved for offline analysis.

To measure the excitability of the post-activation depression, reciprocal inhibition and presynaptic inhibition, patients sat in an electronically-adjustable chair with their back upright, head straight and eyes open. The assessed arm was supported using a height-adjustable table. The shoulder was relaxed in a position of ~ 45° abduction and forward flexion, elbow supported by the table and slightly flexed and the forearm pronated. Since the electrophysiological measurement immediately followed the PSAD measurement, the same EMG electrodes were used to record the activity in FCR and ECR muscles. Bittium NeurOne hardware and software (Bittium NeurOne Tesla, Kuopio, Finland) were used to sample EMG data at a frequency of 10 kHz. A Tesla low pass filter, integrated within the device, was utilized with a lower cutoff frequency of 1500 Hz.

The Hoffmann reflex (H reflex) in the upper limb was measured in the FCR by stimulating the median nerve at the elbow [73]. Nerve stimulation was delivered through rectangular shocks of 1 ms duration using metal electrodes (brass) (1.5 cm radius), which were covered by saline-soaked sponges and placed over the median nerve in the cubital fossa. The optimal placement of the stimulation electrodes was ensured by checking the H and M wave responses in the EMG. To qualify as an H reflex, the response had to appear at a latency of 16–21 ms [74] and to decrease after reaching its peak amplitude when the stimulus intensity was further increased. During the paired-pulse paradigms, the radial nerve was stimulated at the spiral groove ~ 10 cm above the radial epicondyle using identical metal electrodes fixed by Velcro straps. A customized Matlab program was used to trigger two direct current stimulators (DS07 Digitimer, United Kingdom) according to the desired stimulation paradigm.

To determine the optimal stimulation intensity, we stimulated the median nerve with varying stimulus intensities ranging between 5 and 90 mA presented in a random order (2 repetitions per intensity, total stimulations = 68) with an interstimulus interval (ISI) of 6 s. The peak-to-peak amplitudes of H reflexes and M waves were calculated for each intensity and the recruitment curve plotted (Fig. 2D). The desired stimulation intensity was adjusted to produce an H reflex with a peak-to-peak amplitude of Hmax/2 (Fig. 2D). Care was taken to select an intensity with an H reflex on the ascending phase of the input–output (recruitment) curve. Supramaximal stimulus intensity was then used to record the maximum M response (Mmax).

Post-activation depression was measured by varying the inter-stimulus interval (ISI) of consecutive stimuli. The median nerve was stimulated with two different ISIs: 2 and 8 s (Fig. 2A). Blocks of trains of stimuli (10 stimuli per block), with an ISI of 8 or 2 s alternated until 30 reflexes at each interval were obtained (3 blocks of 10 stimuli per ISI). These stimulation blocks were intercepted by 3–4 stimuli with an ISI of 6 s, used to obtain unconditioned H reflexes of the amplitude H max/2. The outcome measure of this assessment is the calculated ratio of the H reflex amplitude evoked every 2 s to that of the H reflex evoked every 8 s: (H2/H8) (Fig. 2E).

A single electrical stimulus at motor threshold to the radial nerve at the spiral groove produces three phases of inhibition in the wrist flexor H reflex. The first occurs at a conditioning-test interval of − 1 to + 3 ms and represents the disynaptic reciprocal inhibition of flexor motoneurons by afferent input from the antagonist extensor muscle (Fig. 2B) [75]. The second phase also called D1 [76] occurs at a conditioning test interval of 5–40 ms and is produced by large diameter group I extensor muscle afferents which act on the H reflex pathway at a premotorneuronal site [54] (Fig. 2C). The third phase (D2) is much longer lasting (50–1000 ms) and is postulated to be transcortical [77].

The mechanism of the D1 inhibition which modulates the release of transmitters from Ia-MN presynaptically, was classically thought to be mediated by GABA_A receptor-induced depolarization of the afferent terminals (PAD) with a subsequent reduction of transmitter release [78]. However, recent studies have profoundly reshaped this interpretation. GABA_A receptors are predominantly located on the internodes of primary afferents, generating a “primary afferent depolarization” that secures the propagation of the action potential, reducing the likelihood of propagation failure. Consequently, synaptic transmission to the motoneurons is actually amplified rather than diminished [79, 80]. Conversely, at the Ia afferent terminals, GABA_B receptors (G-protein-coupled receptors) are present, which inactivate voltage-gated Ca2 + channels, resulting in a reduction of transmitter release, i.e., presynaptic inhibition [79]. These recent findings are likely to have implications for the electrophysiological measurement of presynaptic inhibition. In this study, however, we replicated the experimental protocol established by Lamy et al. [53].

The degree of attenuation of the H reflex in the FCR after stimulation of the Ia afferents in the antagonist muscle (ECR) was measured with a double stimulation paradigm. The median and radial nerves were stimulated with an ISI of 0 ms for reciprocal inhibition. The placement of the radial nerve stimulation electrodes ~ 10 cm proximal to those stimulating the median nerve allows the activity of the radial Ia afferents to reach the spinal cord 1 ms before that of the median Ia afferents. This 1 ms advanced arrival enables the radial nerve to condition the response of the FCR MN by activating an inhibitory interneuron mono-synaptically (synaptic delay ~ 1 ms). For measuring the presynaptic inhibition, the conditioning pulse to the radial nerve preceded the test pulse by 13 ms. The radial nerve was stimulated with an intensity that was at or just below the ECR motor threshold. Single stimuli of the median nerve (unconditioned H reflexes) were randomly alternated with double stimuli (conditioned H reflexes) every 6 s until 30 stimuli of each condition were collected. The outcome measure of this assessment was the amplitude of the conditioned H reflex expressed as percentage of its unconditioned amplitude.

Table 1 shows the characteristics of all the patients recruited for the study. Thirty-four patients had left, and 20 patients had right chronic hemiparesis. On average, time-since-stroke was 46 ± 42 months. Our patient group was heterogeneous in terms of impairment and motor function with an FMA-UE score (mean ± SD) of 26 ± 17 (range, 4–59). The MAS also varied with an average total MAS of 12 ± 7.3/70 (range, 1–35), and an average MAS wrist extension of 1.5 ± 1.3 (range, 0–4). The patient’s own impression about how much spasticity they experienced, using the visual analogue scale had a mean score of 45 ± 26% (range, 0–100), while the motor activity log (MAL), which describes the use of the affected upper limb in performing activities of daily living showed that most subjects relied almost entirely on their non-affected side for most of the activities of daily living with a MAL score of 11 ± 16% (range, 0–69). Initially, 54 patients were enrolled in the study but only 51 patients completed both pre- and post-intervention sessions.

Considering the variability in characteristics among stroke patients in our study sample, we conducted additional subgroup analyses to investigate how these patient characteristics might have influenced the main findings. These analyses can be found in Additional file 1.

The mean values and standard deviations of the clinical scores before and after the intervention, as well as the results of the statistical analysis for each outcome parameter are listed in Table 2. The analysis showed a significant improvement in the FMA-UE: (F (1,51.1) = 36.3, p < 0.001), as well as in the WMFT, both in the time (F (1,26.2) = 9.8, p = 0.004) and function (F (1,26.1) = 12.2, p = 0.002) components, but not the grip strength (F (1,25.1) = 0.088, p = 0.77). Note that only those patients with sufficient residual arm ability underwent the WMFT in order to avoid causing frustration for those patients who are unable to initiate any of the functional tasks which constitute the WMFT.

Spasticity estimated using the modified Ashworth scale (MAS) was significantly reduced, both when examining the wrist extension movement alone (F (1,49.9) = 10, p = 0.003) or when considering the total score which includes multiple arm joints and movements (F (1,50.3) = 10.33, p = 0.002).

The analysis showed no significant change in the visual analogue scale of spasticity or the disability rating scale after the intervention. On the other hand, the MAL, a more robust measure of the use of the affected upper limb in everyday life, showed a statistically significant response to the intervention with an almost 5% average increased use of the affected upper limb in activities of daily living.

The results of the linear mixed model analysis showed that the intervention had an exclusive effect on the active component of resistance to passive stretch measured using the hand-held dynamometer (Fig. 3). The stretch reflex-mediated torque was significantly reduced after the intervention F (1,32.5) = 5.7, p = 0.023 while the passive stiffness component did not change F (1,31.6) = 0.5, p = 0.49.

In all three analyses (post-activation depression, reciprocal inhibition and presynaptic inhibition), the model with the best fit was the simplest with a single fixed repeated effect: SESSION (with two levels: pre and post). Modeling the amplitude of the unconditioned H reflex as a random variable in the linear mixed effects model did not improve its fit. The analysis showed no effect of the intervention on the amount of the post-activation depression (F (1,46.39) = 1.67, p = 0.20), presynaptic inhibition (F (1,39.9) = 1.08, p = 0.30), or reciprocal inhibition (F (1,43.93) = 2.4, p = 0.13) (Fig. 4A–C).

Table 3 shows the results of the Spearman correlation analyses between post minus pre intervention changes in the FMA-UE and the stretch reflex torque with post minus pre intervention changes in the excitability of the spinal mechanisms. Only the change in FMA-UE score and the change in post-activation depression were significantly correlated. This correlation indicated that patients who had a greater increase in the FMA-UE score had a greater decrease in post-activation depression (Fig. 5).

The application of low-frequency rTMS over the motor cortex in the contralesional hemisphere decreases the excitability of the stimulated hemisphere, which leads to an indirect increase in the excitability of the non-stimulated ipsilesional hemisphere by reducing the interhemispheric inhibition from the stimulated to the non-stimulated motor cortex [81‐84]. This is the mechanism thought to underlie rTMS-induced neuroplastic changes used to promote the efficiency of rehabilitation after stroke [85‐87]. In this study, our patient population experienced a significant reduction in motor disability and an improvement in motor function evident by a significant increase in both the FMA-UE and WMFT scores. The increase in the WMFT scores included both the “time” and “function” components, which is important, because the functional score of the WMFT describes the quality of the movement including the fluidity and the influence of synergy on its performance.

The improvement in motor function was reflected in an increased use of the affected hand and arm in activities of daily living, measured using the MAL. This is probably the most important aspect of any rehabilitation intervention, as the increased use of the arm indicates that the intervention-related changes are clinically meaningful. Moreover, the ability to apply the arm and hand in everyday life facilitates the continuation of the rehabilitative process after the end of the therapy as a high repetition of exercise movements is necessary for motor improvement [88].

There is accumulating literature that indicates a positive effect of low-frequency rTMS on spasticity evaluated using the MAS for reviews, see [19, 89‐91]. The results of our clinical data analysis also showed a significant reduction in the total MAS score and the MAS score for wrist extension. These findings add little to our understanding of the pathophysiology of spasticity and the possible effects that different interventions may have on it. This is due to significant mismatches between the clinical scale and the definition of spasticity. First, the “velocity dependency” aspect of the response to the stretch cannot be ascertained using the MAS. During the performance of the test, the examiner moves the joint and simultaneously estimates the perceived resistance without precisely controlling the velocity of the movement. Second, the “stretch reflex” component of resistance to joint stretch is almost certainly indistinguishable by the clinician from increased resistance caused by shortening of the muscle–tendon complex and changes in their microstructure, i.e., muscle contracture [25‐29].

To the best of our knowledge, this is the first study to explore the effect of low-frequency rTMS and physiotherapy intervention on spasticity measured objectively using a device which enables the discrimination between the reflex-mediated and passive stiffness components of the response to muscle stretch. Our findings indicated that the intervention had an exclusive effect on the stretch reflex torque (Fig. 3), confirming that contralesional low-frequency rTMS indeed has a positive effect on reducing spasticity in chronic stroke.

The outcome that the passive resistance to wrist stretch remains unaffected by the intervention, aligns with expectations, considering that rTMS primarily influences the neurogenic component of resistance. This occurs through alterations in the excitability of corticospinal neurons, which in turn affect spinal interneurons and modulate their activity [18, 61, 63, 92, 93]. Given the novelty of the application of the objective measurement tool in chronic stroke, and the limitation of the study design, these results need to be replicated in a randomized sham-controlled paradigm.

We expected that a reduction in spasticity would be accompanied by changes in the spinal mechanisms, which seem most likely related to the development of spasticity. Indeed, previous studies demonstrated that rTMS can change the excitability of the spinal circuitry [16‐18, 92]. For example, Perez et al. (2005) found that a short train of high-frequency rTMS (20 pulses, at an intensity of 1.2 × MEP threshold and a frequency of 5 Hz) over the leg motor cortical area in healthy subjects increased presynaptic inhibition of Ia afferents which project onto soleus motoneurons. Meunier and Pierrot-Deseilligny (1998) also reported that TMS provided an increase in the radial-induced inhibition of the wrist flexor H-reflex which suggests an increase in presynaptic inhibition. In our findings, none of the examined spinal mechanisms showed a significant change after the intervention.

There are multiple possible explanations for this finding, the first is related to the time scale of the effect of rTMS on the spinal circuits. Previous studies compared the activity before and after short interventions with a maximum intervention-measurement delay of 30 min [16‐18, 94]. The effects of rTMS on the spinal circuits may be short-lived as reported by Perez et al. [18] who found that the size of the H reflex returned to its normal size at a stimulus-test interval of 3 s. It is important to consider that in the aforementioned studies, the stimulation intervention was also short, unlike the multi-session rehabilitative intervention that our patients underwent. It is possible that short-term effects on the spinal circuits themselves trigger secondary, and longer lasting changes in other mechanisms related to the firing threshold of motoneurons [95, 96].

Another possibility concerns the actual relevance of the examined spinal mechanisms to the pathophysiology of spasticity. There seems to be agreement in the literature that the reduction in post-activation depression is at least one of the primary mechanisms underlying spasticity [97‐99]. It has been consistently found to be reduced on the affected but not the unaffected side in spastic patients after stroke [50‐53] and to be related to the degree of spasticity measured with the Ashworth scale in the upper limb [53]. Substantial evidence is also available for the reduction of reciprocal and presynaptic inhibition in the upper limb in the stroke population [54‐56], but the contribution of these mechanisms to the development of spasticity is uncertain [44, 46, 53]. It is not unlikely that intervention-related changes might have taken place in other spinal circuits, ones that we have not measured. A likely candidate is the one that underlies the inhibition in the third phase of radially induced FCR inhibition. This inhibition, with a latency of 100 to 300 ms, is caused by long-loop inhibitory connections to supraspinal centers that receive input from the premotor cortex. In a similar study, in healthy subjects, Huang et al. [16] found that continuous theta-burst stimulation had no effect on the first and second phases of inhibition between forearm extensor and flexor muscles. However, they did find a reduction in the amount of inhibition in the third phase. Similar findings were reported by Huang et al. [17] in patients with dystonia. They suggested that if the third phase of radially induced inhibition of FCR is due to activity in a long-loop spinal– brainstem–spinal pathway, then the effect on this phase may be due to changes in activity of premotor-brainstem connections.

One must also keep in mind that the inherent variability in the H reflex measures may have masked underlying changes, a concern that was considered by Nielsen et al. [46] in their review, as an obstacle in the way of progressing our understanding of the pathophysiology of spasticity. Even within the healthy population, disynaptic reciprocal inhibition, for example, was found to be absent in many subjects, although they are not spastic [100]. H reflex measures are even more challenging in the upper limb than in the lower limb, with a large cross-talk between the flexor and extensor EMG channels due to their physical proximity. In addition to the variability in the physiological measurements themselves, our patients represent a heterogeneous population with different lesion sizes and locations. The subsequent adaptations in the spinal networks as a response to the primary lesion may vary considerably between the individuals, and so may the adaptation in these networks to the intervention.

A noteworthy observation arises when examining the association between changes in motor disability and the excitability of spinal circuitry. While the significant correlation between the reduction in motor impairment and post-activation depression following intervention (Fig. 5) does not establish causality, it prompts further investigation into the underlying physiological mechanisms driving motor recovery.

After an injury to the central nervous system, multiple mechanisms are triggered, which aim to amplify the signal arriving at the motoneuron from the residual intact pathways [41]. Examples of these processes include collateral sprouting and the formation of new synapses from intact axons onto partially denervated motoneurons [101‐104] systematic changes in gene expression and the upregulation of the essential transmitter receptors [105]; and the downregulation of the potassium-chloride cotransporter [106]. All of these factors contribute to an increased excitability below the lesion (thus certainly contributing to the active stretch reflexes in spasticity), but they may also facilitate the recovery of voluntary movement (with specific motor patterns) even when the descending pathways mediating this information are weakened by the lesion.

Indeed, motor recovery and the development and ultimate disappearance of spasticity are interconnected and go through different stages during post stroke complete recovery [43, 107‐109]. The direction (facilitation vs. inhibition) in which the excitability of a certain spinal circuit or mechanism needs to develop in order to allow movement to take place probably depends on many factors. It is perhaps naive to assume that increased excitability of the corticospinal tract as a result of the rTMS intervention should translate into global reduction in the hyperexcitability of the involved spinal circuitry. In order to enable motor recovery during the rehabilitative process, it is plausible that the nervous system might still need to facilitate, even further, some of the processes which may amplify the descending corticospinal input.

Whether spasticity per se needs to be considered a target for therapeutic interventions is, to say the least, disputable [110‐114]. There is evidence that mechanisms other than spasticity underlie the spastic movement disorder including the impaired ability to voluntarily activate muscles, i.e., paresis, muscle atrophy [115] and co-activation of antagonist muscles [113, 114, 116‐118]. Nonetheless, interventions targeted towards motor rehabilitation, including motor cortex rTMS, which simultaneously improve motor function and reduce spasticity by improving motor control and (probably) increasing ipsilesional corticospinal tract excitability [81‐84] might mimic natural recovery, in which spasticity does decrease as a function of improvement in motor control [107] and could eventually disappear in the case of full recovery [109].