Individuals with moderate multiple sclerosis (MS) present with an array of symptoms, with walking impairment being among the most common. Gait speed, cadence, stride length and time spent on double-limb support are frequently affected, and correlate with reduced independence and productivity, impacting overall quality of life [1]. It is a major factor in estimates of disease progression [2–4], and approximately 40% of MS patients will need some form of walking assistance within 15 years of disease onset [5]. Consequently, interventions targeting gait disturbance in patients with MS are in demand. Currently these include rehabilitation therapy and pharmacological management.

While therapeutic exercise has long been believed to increase symptoms, more recent evidence has demonstrated that exercise and increased physical activity are beneficial for people with MS, and are becoming more commonly included as part of treatment interventions [6–10]. The optimal type or mode of exercise, intensity, frequency, duration and maintenance of exercise training to manage the progression of balance and gait dysfunction in MS has, however, yet to be established through randomized trials [3].

Rehabilitation therapy includes occupational, physical and speech therapies as part of the comprehensive management of symptoms [11, 12]. Compensatory strategies such as energy conservation, use of adaptive equipment, and environmental adaptations are often utilized in these interventions [13–20]. Medical devices and physical modalities may also be used in multidisciplinary interventions [21]. A recent Cochrane review of multidisciplinary rehabilitation for adults with MS did not find strong evidence of interventions that resulted long-term improvements [12].

Pharmacological management may slow disease progress or reduce motor symptoms. For example, Ampyra (Acorda Therapeutics) was approved after it was shown to be effective in improving walking speed by 20% when compared to a placebo group that improved 8% [2]. This outcome, however, was observed in only 35% of those tested, and many participants reported significant side effects. Additionally, there is little evidence that these drugs will prevent the mobility disability of persons with 2^nd stage MS (Kurtzke Expanded Disability Status Scale (EDSS) score of 4.0 or greater) [1, 3, 22, 23].

The medical devices and physical modalities utilized in treatment for people with MS vary. Functional electrical stimulation, as delivered through electrodes attached to the skin, has been shown to be effective in improving gait, as long as the electrodes are attached and the stimulation unit is on [24–29]. Other contemporary forms of neurostimulation (aimed at induced neuromodulation) are invasive, expensive and have the potential for adverse effects. For example, deep brain stimulation and vagus nerve stimulation, which use implanted pacemaker-like electrical devices, are indicated for decreasing tremors in MS, but carry surgical risks [30–33]. These therapies have not been widely attempted in MS rehabilitation. Neurostimulation that directly stimulates the peripheral or central nervous systems to improve motor impairments is still being investigated. Non-invasive neurostimulation that has been tried in people with MS are typically large clinic-based devices that employ powerful electromagnetic stimulation of the brain’s cortex (transcranial magnetic stimulation) [34, 35], or electrodes that pass electric current through the skull (transcranial direct current brain stimulation) [36, 37]. Intermittent theta burst stimulation of the motor cortex has been demonstrated to reduce spasticity in people with MS [38], and when combined with exercise therapy, to decrease fatigue [35]. Success with these interventions has, however, been limited. Consequently, there continues to be a need for alternative approaches that have the potential to help improve mobility in people with MS.

Previous studies have shown that the tongue can be used as an effective interface for sending electrical signals to the central nervous system [39–45], for example sensory substitution in balance-impaired or blind individuals [43, 46–50]. Individuals with primary vestibular disorders who trained using electrical stimulation through the tongue coupled to head-position information demonstrated balance improvements that were sustained for weeks beyond the final stimulation session [46, 51].

Using a different methodology that delivers tongue stimulation which, like the present study, is devoid of information linked to head position or any other exogenous variable, our recent functional MRI results indicated that the improvements that occurred with the neuromodulation training (using electrical stimulation on the tongue) are likely related to modulation of neural activity within structures of the brain that control balance and movement [44, 45, 52].

Twenty people with MS participated in this study. The subjects were randomized into active and control groups as they enrolled. The groups were similar across age and EDSS scores (see Table 1). There was a difference between the groups for mean number of years with a diagnosis of MS (p = 0.01) and for baseline DGI, although the latter did not reach statistical significance. All subjects completed the 14-week intervention and the 5 data collection points except for 2 subjects in the control group who were unavailable for data collection point #3 due to travel.

Subjects in the Active group achieved both statistically significant (p < 0.05) and clinically significant (DGI change of at least 4 points [13]) improvements in gait by the 6-week test point and these improvements continued through the 14-week test point (Figure 3 and Table 3). Subjects in the control group did not achieve a statistically-significant improvement in gait at any test points, although the apparent improvement at Week 10 would be considered clinically significant.

When the pairwise differences in DGI change from baseline (Table 3) are compared for the two groups, the active group shows a statistically-significant greater improvement than the control group at 10 weeks (p = 0.027) and 14 weeks (p < 0.001), using the independent t-test. (The Corresponding p values for the Wilcoxon test were 0.034 and 0.002).

All subjects reported an increase in salivation at the outset of the study due to the presence of the device in the mouth. With instruction, each was able to develop swallowing strategies to manage saliva volume while maintaining the device in their mouth. Five subjects also reported mild headaches and temporo-mandibular joint pain during the first few days of participation. These symptoms were mitigated by the end of the 2-week in-lab training period by instructing them to not bite the array, relax their jaw, press the array firmly with the tongue, and breathe uniformly. Two RRMS subjects in the Active group experienced relapses requiring suspension of their participation, one for a period of 5 weeks, the others for 3 months. When they were able to resume training at the same level they had achieved prior to the relapse they were reintegrated into the study without adverse effect. Three subjects (2 Control, one Active group) experienced minor illnesses that required suspension of training for approximately 2 weeks each. Each was able to resume training without complication, and their testing schedules adjusted accordingly. Finally, several subject experienced fatigue after completion of the morning training session. They were encouraged to practice relaxation exercises to help them prepare for performing the afternoon sessions. This proved to be a useful skill for them to develop in managing their energy levels so that they could consistently complete their daily participation in the study.

This study investigated whether individuals with multiple sclerosis could improve their gait using CN-NINM intervention, a training program that incorporates exercises combined with noninvasive electrical stimulation of the tongue. The results show that subjects in the active group had greater improvements in their gait relative to the control group, and the improvements were clinically and statistically significant.

The sample size of this study was small (10 subjects per group), which may limit the external validity of the results. The sample size estimation for this study was based on the robust results of our earlier non-controlled proof of concept studies.

In considering sources of the induced improvement, our previous studies have demonstrated through fMRI that neuromodulation training using electrical stimulation of the tongue, combined with movement exercises in balance-impaired individuals, induces activity of the cerebellum and brainstem nuclei, structures of the brain that process balance and movement [44, 45, 52]. Additionally, Mori, et al. showed that using transcranial magnetic stimulation combined with exercise therapy resulted in a reduction in spasticity and fatigue [35]. These findings, combined with known afferent neural pathways from the tongue, suggest that tongue stimulation preferentially predisposes certain cerebellum and brainstem nuclei to beneficial neuroplastic effects resulting from movement exercises dependent on these nuclei, with the end result of improved movement control.

We hypothesize that CN-NINM induces neuroplasticity by noninvasive stimulation of two major cranial nerves: trigeminal, CN-V, and facial, CN-VII. This stimulation excites a flow of action potentials (AP’s) to the brainstem (pons varolli and medulla) and cerebellum via the lingual branch of the cranial nerve (CN-Vc), and chorda tympani branch of CN-VII. This effect of the stimulation extends to the corresponding nuclei of the brainstem – at least in the sensory and spinal nuclei of trigeminal nuclear complex and the caudal part of the nucleus tractus solitarius [44, 45, 52]. We postulate that the intensive activation of these structures initiates a sequential cascade of changes in neighboring and/or connected nuclei by direct collateral connections, brainstem interneuron circuitry and/or passive transmission of biochemical compounds in the intercellular space. There is evidence from related research that has observed changes in both neurotransmitter and other neuroactive compounds in response to chronic stimulation. Each AP in the trigeminal nuclei and brainstem releases up to 23 biologically active compounds including neurotransmitters that affect synaptic transmission, and the brainstem has the highest glia-to-neuron density anywhere in the CNS (50:1) [61]. The full function of the glial network is unknown, although it is generally agreed that it plays a vital role in the regulation of neural behavior through management of the chemical environment at the synaptic gap. Consequently, we believe the stimulation directly activates not only the neuronal network by electrical impulses (AP’s) but also the glial network by neurochemical impact. The net effect of this upregulation of neuroactive compounds is to potentiate the networks involved within the CNS and set the stage for sustained focal and global changes in brain behavior [62–68].

Improvement is also common in patients receiving a new intervention with a therapist [7, 69]. A better indication of efficacy is if the patient improves when independently training at home. Here we see that both groups continued to improve during the at-home phase of the study. In a related study, Di Fabio, et al., found that patients with progressive MS who continued with an extended home outpatient rehabilitation over the course of 1 year experienced a lower rate of decline in physical function when compared with subjects who did not exercise [70, 71]. Our study did not compare subjects who had performed the extended exercise protocol with any who had exercised and stopped. Consequently, to investigate this phenomenon, the natural progression of this study would be to repeat it using a longer intervention period.

In our study, though all subjects appeared to demonstrate improvements initially, only the active group continued to improve over the length of the study. It is likely that the early improvement in both groups was due to the intense involvement of the subjects with the trainers for the first 2 weeks of the study. It can been seen in Table 3 that improvements in performance for the active group continued to accumulate as subjects trained at home after the initial 2-week training phase. Subjects who trained using exercise only without stimulation (control group) continued to improve for the first month at home and then exhibited a plateau or even a decrease in performance. This provides preliminary evidence that the intervention is effective when performed independently at home.

We observed a significant difference between groups for number of years with MS, with those in the active group having had MS for a longer period of time. Because the progression of the disease varies from person to person, however, we did not feel that chronicity (years with MS) was as important as symptom presentation. We based our inclusion criteria on symptom presentation, specifically gait dysfunction, and EDSS score. We also observed that the mean initial (baseline) DGI score was lower in the active group (although this did not approach statistical significance), raising the concern that this may bias the results. Because of this apparent difference, we chose to analyze data using a simple DGI difference from baseline, rather than a % change as suggested by other groups which would have unduly favored the active group [60]. Similar analysis of % DGI change yielded identical conclusions.

We did not investigate whether the subtype of MS had an effect on the reported outcomes. It is acknowledged that because subject assignment was purely random there is some possibility that the disease subtype could affect sensitivity to the intervention and therefore the results. We note, however, that these sub-classifications relate primarily to the progression of the disease and not to their particular state during participation in the study. Furthermore, we excluded all candidates that had any changes in symptoms or medication in the 3 months prior to participation to ensure that their presentation was as stable as possible. Nonetheless, it would be beneficial to repeat this study using only MS patients of one particular type, such as PPMS or SPMS to determine if the rate or magnitude of change in performance differs as a function of the disease sub-classification.

The remaining question that may be posed is whether CN-NINM training could be practically deployed in a rehabilitation setting. The in-lab training for this study was admittedly time intensive (approximately 2 to 3 hours per day, per subject), involving far more time for therapy than a typical clinical setting would allow. This was an intentional departure from most therapeutic models for rehabilitation, derived from our prior experience with treating vestibular disorders [43, 46]. Given the rigors of the intervention, and the nature of this neurodegenerative disease, this intensive in-lab phase was designed to ensure that subjects understood and could reliably perform the training program before using the device at home. The results suggest that this new paradigm for home-health rehabilitation, particularly for disorders previously deemed untreatable, is efficacious. Additional studies are necessary to determine if an abbreviated intervention would be as effective as the model presented here.

The results of this pilot RCT demonstrate that non-invasive electrotactile stimulation, when combined with targeted physical therapy exercises, can significantly reduce clinical symptoms of gait dysfunction in multiple sclerosis. This complements a growing body of evidence demonstrating that neuromodulation combining electrical stimulation with exercise therapy has the potential to have a positive effect on motor control for people with neurological conditions. The results also demonstrate that the CN-NINM intervention shows promise for development as a clinical tool for improvement in gait in people with MS. Additionally, because the tongue stimulation device is portable it allows people to train at home, affording an efficacious therapeutic model not previously reported. This leads us to speculate whether people with other types of neurological conditions would benefit from this type of intervention to improve motor control. We suggest that further studies are warranted to investigate the breadth of applicability that this form of therapeutic intervention may have for meaningful neurorehabilitation.

All authors are affiliated with the University of Wisconsin, Madison.

MT teaches in the Department of Biomedical Engineering at the University of Wisconsin, Madison. He is a co-inventor of the PoNS device and is a contributor to the development of the CN-NINM training method. YD is a co-inventor of the PoNS device and is a major contributor to the development of the CN-NINM training method. KK is an expert in electrotactile stimulation and a co-inventor of the PoNS device. KR is an Occupational Therapist with an interest in human subjects rehabilitation research. AS is a psychiatrist who specializes in cognitive function and emotional processing. KS is a Physical Therapist with an interest in the application of neuromodulation for rehabilitation and is a contributor to the development of the CN-NINM training method.