Focal dystonia is a movement disorder that can affect any body part and severely impair a person's ability to function in their daily life. In focal hand dystonia (FHD) there is an involuntary activation of agonist and antagonist muscles in the hand and forearm. In task specific hand dystonia, the dystonia manifests when a person attempts to execute specific tasks such as writing, typing, or playing a musical instrument. Most often these tasks are related to a repetitive action, often done in the context of an occupation, thus, impairing the ability to work and reducing quality of life. The individual with FHD otherwise has a normal neurological exam and normal function of the hand with other tasks. Although the pathophysiology of FHD remains unclear, and may differ among the different types of dystonias, there is considerable evidence for abnormal inhibition in primary motor cortex (M1), premotor cortex (PMC), spinal cord, and brainstem (Siebner et al., 1999; Butefisch et al., 2005; Hallett, 2006, 2011; Quartarone et al., 2014). For example, it has been shown that participants with task specific dystonia have disturbances in task dependent inhibition in M1 (Butefisch et al., 2005) and intracortical inhibition (Kimberley et al., 2009). While dystonia is mainly a motor problem, mild abnormalities of sensation have been reported in patients with dystonia, even in body parts not affected by the dystonia. There is also evidence to suggest that the problem may be related to faulty sensorimotor integration (for review, Abbruzzese and Berardelli, 2003; Quartarone et al., 2014). Another abnormality in FHD is that of plasticity. It has been demonstrated that decreases in M1 inhibition facilitate the induction of plasticity (Hess et al., 1996; Di Lazzaro et al., 2006), thus, the abnormally increased plasticity in dystonia could be related to the abnormally decreased motor cortical inhibition that is found in patients with dystonia (Hallett, 2011).

The evidence for abnormal inhibition in patients with dystonia is derived from experiments using transcranial magnetic stimulation (TMS) to M1 as a testing tool to measure the level of intracortical inhibition in the GABAnergic system (Ridding et al., 1995; Butefisch et al., 2005). Most related to this proposal, the cortical silent period (CSP) is attributed to both spinal and cortical inhibition and is a test of GABA_B activity (Chen et al., 1999; Kimberley et al., 2009). CSP is the duration of EMG quiescence following a TMS pulse to M1 that is superimposed upon a low level of voluntary muscle contract in the muscle of interest (Kimberley et al., 2009).

While TMS provides a means to measure these abnormalities in M1 excitability, when applied repetitively (repetitive TMS, rTMS) it can also be used as a tool to modulate excitability of the stimulated neuronal tissue. When applied at low frequencies, it results in increased inhibition in the stimulated area of the brain that persists for more than 20 min post intervention (Chen et al., 2003). This rTMS protocol has been applied in persons with FHD in a single session and a temporary reduction in dystonic posturing and increased inhibition was recorded with a maximal effect when stimulation was applied to the PMC (Murase et al., 2005). These findings have been extended by applying low-frequency (1 Hz) rTMS to the PMC in persons with FHD for 5 days, resulting in improved handwriting velocity and changes in brain excitability (Borich et al., 2009). Additional work examined whether rTMS applied during performance of an active, but non-dystonic, hand “writing” task would enhance the beneficial effects (Kimberley et al., 2013). The multiple sessions of rTMS strengthened intracortical inhibition causing a prolongation of CSP after 3 days of intervention and pen force was reduced at day 1 and 5; notably, 68% of patients self-reported as “responders” after 5 days of intervention, suggesting potential beneficial effect of the inhibitory neuromodulation (Kimberley et al., 2013). Together, evidence suggests that rTMS induced PMC inhibition is an encouraging approach for neuromodulation therapy in people with FHD. Neuromodulation alone may not be adequate to produce clinically meaningful changes, however.

Beyond botulinum toxin, which has a variable and temporary benefit, there are few opportunities for treatment for people with FHD. Rehabilitation interventions based upon the hypothesis of aberrant learning and maladaptive plasticity have had some modest success [for review (Altenmuller and Jabusch, 2010)]. Sensorimotor training is one of these interventions that has been shown to modify brain activation and produce some functional improvements (Byl and McKenzie, 2000; Zeuner et al., 2002, 2003; Byl et al., 2003, 2009). The objective of the training is to redefine spatial and temporal processing capacities through guided activities that emphasize different aspects of sensory feedback (e.g., somatosensation, proprioception, kinesthesia) in order to improve sensorimotor integration and restore task-specific skills. A significant limitation of this training is the duration of treatment, however, which ranges from 8 to 12 weeks and requires several hours of therapy/day. Additionally, although improvements have been shown to occur in some cases, participants do not achieve premorbid levels of function and it is not universally beneficial.

The purpose of this study was to investigate whether the pairing PMC rTMS with sensorimotor retraining is more effective in improving dystonia and abnormal inhibition when compared to PMC rTMS with a control intervention. Combining neuromodulation treatment by means of rTMS or transcranial direct current stimulation with motor training is an approach that has been used in the rehabilitation of motor stroke (Hao et al., 2013) and builds on the evidence that plastic changes are either enhanced or blocked depending on the rTMS protocol (Butefisch et al., 2004). Considering the evidence for abnormally enhanced plasticity in people with FHD, executing a sensorimotor training in a more normal neurophysiological state (i.e., improved M1 inhibition) may enhance the effect of the training. Given the previous work that demonstrated increased inhibition in people with FHD following rTMS to PMC (Borich et al., 2009), as well as positive clinical changes from a prolonged intervention protocol, we hypothesized that rTMS applied to the PMC followed by sensorimotor training (SMR) would result in reduced dystonia symptoms and increased M1 inhibition, compared to rTMS alone.

The results suggest rTMS paired with sensorimotor training is not superior to rTMS with control. The findings further suggest a potential longitudinal benefit associated with two bouts of rTMS separated by 1 month, as indicated by subjective rating improvements.

The inhibitory rTMS applied here was intended to promote a favorable inhibited cortical state with rehabilitation applied during the improved neural environment following the neuromodulation. SMR was selected as a likely beneficial treatment because evidence of clinical improvement has been observed with this intervention (Byl et al., 2009). Yet, this therapy is very time intensive and may be challenging for patients to undertake, thus adjunct therapies that may enhance the effects of SMR would be advantageous. The SMR treatment performed here was brief in comparison to previous work, however which could have contributed to the lack of superior beneficial effect observed. Interestingly, despite the short duration of training, there was a significant improvement in dynamic sensory discrimination as measured by the BCB and two point static and dynamic discrimination testing on some fingers. This suggests that even a small amount of training may be able to improve this ability and it may be associated with perceived symptom improvement.

The control intervention was selected to deliver equal dose of intervention time and match investigator attention provided to the participant, but was not intended to be a specific therapy for FHD, as there is no evidence that general stretching and massage should improve FHD. Active digit and hand movement and strengthening were avoided as a control intervention, as it has been shown that there was no difference in improvement of FHD with either a task specific training or a more generic motor intervention (Zeuner et al., 2008). The evidence here is unable to determine if the control therapy gave additional benefit beyond what would be delivered with rTMS alone, as there was no group that received only rTMS.

It is interesting to note that there was no effect of increased inhibition after rTMS as measured by an increased CSP in the group analysis. Our single subject analysis suggests variable responses between subjects. It is important to note, however, that our measurements were taken at least 24 h after conclusion of the rTMS. It has been previously reported, that the increased inhibitory effect of rTMS may only be transiently observable and not persist beyond immediately post intervention (Kimberley et al., 2015). It is unknown if a more comprehensive TMS assessment may have revealed other changes.

Another consideration is that the training practice began immediately following the neuromodulation which may have influenced the potency of training effects, as there is evidence to support that a preceding bout of neuromodulation may alter the effect of motor training (Jung and Ziemann, 2009). Further, it is unknown what the ideal time frame is for rTMS effects to optimally manifest or whether treatment tasks should precede or follow the neuromodulation for the most effective capitalization of neuroplasticity. It is noteworthy to consider that the vast majority of literature characterizing the effects of non-invasive brain stimulation are based on a healthy response, yet it has been shown that people with dystonia have an abnormal plastic response to both behavioral and neuromodulatory perturbations (Weise et al., 2011; Meunier et al., 2012; Belvisi et al., 2013). This deficient homeostatic plasticity may then augment the cortical response secondary to rTMS combined with behavior training, unlike in a healthy individual. Future work should recognize that the ideal protocol for learning in a healthy individual may not clearly translate to an intervention in dystonia. Inclusion of a healthy control group may be advantageous to compare responses to neuromodulation interventions and further our understanding of how rTMS in healthy adults may support or differ in regards to persons with dystonia.

The paramount limitation to this study was a small number of participants. The small sample may have underpowered the study and as a result, real change between groups went undetected. The high variance seen in our single subject analysis is not unusual as a current conundrum in neuromodulation is the variable response between subjects to many different types of neuromodulation (López-Alonso et al., 2014). Notably, individual participants did display significant changes in TMS outcomes as determined by the single subject analysis, yet when averaged as a group, the effects were nullified. Of note, however, is that the physiologic TMS measures did not agree with the clinical measures. This has finding has been supported in other work as well (Pomeroy et al., 2007; Kimberley et al., 2013; Sadnicka et al., 2014).

Additionally, self-rated improvements did not agree with other measures. This may indicate that the outcome measures used are not sensitive enough to capture the therapeutic effects; however, a placebo effect must also be considered. The likelihood of placebo effect is weakened though when one considers that the beneficial effect had a moderate effect size. A disagreement between objective measurement and self-rated improvement has been reported in musician's dystonia where patients tended to rate their improvement higher than an objective keyboard assessment task (Van Vugt et al., 2014). Van Vugt et al. (2014) speculated that this was due to patient self-selection of tasks during their day that less severely triggered dystonic symptoms, leading to an overestimation of ability. A similar phenomenon may be at play in these results.

Clinical studies in dystonia are often characterized by small samples and a heterogeneity of responses require careful consideration in the study design (Galpern et al., 2011). It is interesting that self-reported change continued to improve despite ongoing intervention. This finding suggests that the choice of a cross-over design in FHD or rTMS trials may not be ideal. This design was selected to maximize recruitment in a rare disease, reduce between subject variability, and give all participants the chance to experience the active intervention. Due to variability in the disorder on a day-to-day basis and variability in baseline TMS responses, a two-group design would better control for a potential carryover effect. Accordingly, factors that predict a participants' response should be further evaluated to investigate the role of non-invasive brain stimulation in those with focal dystonia and multicenter trials should be considered to achieve higher recruitment.

This design did not assess for potential benefit from the SMR alone with sham rTMS, nor a sham-sham effect (sham rTMS + control intervention). Given the current state of the literature that reports highly variable response to SMR training, and only after extensive practice, we felt it was unlikely that 30 min of training alone would be sufficient to demonstrate benefit, however this could be additionally investigated. Also, the effect of rTMS to PMC vs. sham rTMS has been studied with the same parameters as used in this trial, thus, if large physiologic or behavioral changes were observed a historical comparison could have been made (Murase et al., 2005; Borich et al., 2009; Kimberley et al., 2013, 2015). Finally, there is sufficient literature that suggests that in some people with FHD, inhibitory rTMS to the PMC may have beneficial effects, but other potential targets to stimulation should be explored.

The results suggest rTMS paired with sensorimotor training is not superior to rTMS paired with control therapy, but there are potentially beneficial therapeutic effects of rTMS combined with non-specific rehabilitation in the treatment of FHD according to self-perceived measures. The majority of participants (6 of 9, 67%) reported improvement in self-rated function; however, two subjects experienced no change and one subjects reported worsening of symptoms. The variability of responses suggests rTMS interventions may not be effective in all participants with symptoms of dystonia and the optimal intervention and timing of synergist intervention remains to be determined.