Safety and Efficacy of At-Home Robotic Locomotion Therapy in Individuals with Chronic Incomplete Spinal Cord Injury: A Prospective, Pre-Post Intervention, Proof-of-Concept Study

Background The compact Motorized orthosis for home rehabilitation of Gait (MoreGait) was developed for continuation of locomotion training at home. MoreGait generates afferent stimuli of walking with the user in a semi-supine position and provides feedback about deviations from the reference walking pattern. Objective Prospective, pre-post intervention, proof-of-concept study to test the feasibility of an unsupervised home-based application of five MoreGait prototypes in subjects with incomplete spinal cord injury (iSCI). Methods Twenty-five (5 tetraplegia, 20 paraplegia) participants with chronic (mean time since injury: 5.8 ± 5.4 (standard deviation, SD) years) sensorimotor iSCI (7 ASIA Impairment Scale (AIS) C, 18 AIS D; Walking Index for Spinal Cord Injury (WISCI II): Interquartile range 9 to 16) completed the training (45 minutes / day, at least 4 days / week, 8 weeks). Baseline status was documented 4 and 2 weeks before and at training onset. Training effects were assessed after 4 and 8 weeks of therapy. Results After therapy, 9 of 25 study participants improved with respect to the dependency on walking aids assessed by the WISCI II. For all individuals, the short-distance walking velocity measured by the 10-Meter Walk Test showed significant improvements compared to baseline (100%) for both self-selected (Mean 139.4% ± 35.5% (SD)) and maximum (Mean 143.1% ± 40.6% (SD)) speed conditions as well as the endurance estimated with the six-minute walk test (Mean 166.6% ± 72.1% (SD)). One device-related adverse event (pressure sore on the big toe) occurred in over 800 training sessions. Conclusions Home-based robotic locomotion training with MoreGait is feasible and safe. The magnitude of functional improvements achieved by MoreGait in individuals with iSCI is well within the range of complex locomotion robots used in hospitals. Thus, unsupervised MoreGait training potentially represents an option to prolong effective training aiming at recovery of locomotor function beyond in-patient rehabilitation. Trial Registration German Clinical Trials Register (DKRS) DRKS00005587


Summary
The improvement of partially or totally lost motor functions is the primary goal of the rehabilitation of patients with neurological impairments after spinal cord lesions or stroke. The restoration of the ambulatory function is the primary focus of the therapeutic efforts, because of its high relevance for reintegration into social and professional life. At the current state physiotherapeutic approaches based on motor learning principles are mainly contributing to the success of motor rehabilitation. The physiotherapeutic method investigated best in stroke survivors and individuals with incomplete spinal cord injury (SCI) is locomotion training on a motorized treadmill. Due to the fact that particularly in the very acute stage patients are not sufficiently able to initiate stepping movements voluntarily, the leg movements are supported by locomotion robots, an active tilt-table or manually by therapists. These large-scale devices are mainly used in specialized centers because of their investment and maintenance costs and need for qualified personnel for operation. With robotic devices relevant improvements in the ambulatory function of individuals with incomplete SCI were shown. However, it became apparent that sustained, long-term improvements induced by locomotion training can only be maintained if training is applied over a long period of time. Thus, it is important to have a low-cost locomotion therapy device, with which the skills achieved with large-scale devices in clinics can be maintained or even further improved in the home environment. Most recent research results show, that gait phase related, cyclic loading and stimulation of the foot sole seems to be most relevant parameter for success of the therapy. The aim of this research project funded by the German Federal Ministry of Research and Education within its innovation award program is to develop a modular, flexible, to its technical minimum reduced concept for locomotion therapy at home. After several functional demonstrators of this concept are provided it is intended to perform a clinical pilot study in individuals with motor incomplete SCI (ASIA Impairment Scale C and D) to test the safety and efficacy of these de vices. The study is planned as a single center, longitudinal study with assessments at baseline, therapy and follow-up intervals. Within 8 weeks of therapy the safety and efficacy of the daily application of the device will be tested. The patients will be randomly assigned to two groups, which will train with two versions of the device differing in the kind of foot sole stimulation. For assessment of the training outcome standardized neurological scores quantifying the motor and sensory status as well as functional scores for quantification of the gait function will be applied. The assessments will be performed 4 weeks, 2 weeks and directly at therapy start (all baseline), 4 and 8 weeks after therapy start and 3 months after therapy end (follow-up). With this project basic scientific knowledge about the possibilities for stimulation of the human spinal pattern generator will be generated and directly transferred into novel training methods and devices for automated locomotion therapy.

Scientific background
The improvement of partially or totally lost motor functions is the primary goal of the rehabilitation of patients with neurological impairments after spinal cord lesions or stroke. The restoration of the ability to walk is the primary focus of the therapeutic efforts, because of its high relevance for reintegration into social and professional life. Also in other patient groups such as orthopaedic or geriatric patients, who are immobilized over a substantial period of time [1], an early and intensive mobilization plays an important therapeutic and functional role. At the current state mainly physiotherapeutic approaches based on motor learning principles are contributing to the success of motor rehabilitation. The physiotherapeutic method investigated best in individuals with incomplete spinal cord injury (SCI) [2, 3, 4] and stroke survivors [5] is locomotion training on a motorized treadmill. Due to the fact that particularly in the very acute stage patients are not sufficiently able to initiate stepping movements voluntarily, these movements are supported by locomotion training robots (own work), an active tilt-table (own work) or manually by therapists. The automated locomotion therapy should be used preferably, because of the high physical workload of the therapist during manual assist, the associated limitations in the duration of the training sessions and the non-reproducible walking pattern during the training.
Most of available literature about locomotion studies reports of relevant positive adaptations on physiological as well as motor functions [6]. Dietz and colleagues [2] proved that individuals with a chronic motor incomplete SCI after several months of locomotion training improved to a degree that they were able to initiate step movements on their own. The results of a study performed by Wernig and colleagues [3] showed that a treadmill training over several weeks lead to an independent ambulation in ¾ of the study participants with incomplete SCI. Beside the improvements in motor functions also positive therapeutic effects in terms of stabilisation of the cardiovascular system (own work) and higher bone density [7] were reported. Although body weight supported treadmill training has evolved to a standard therapy in the rehabilitation of patients with neurological gait disorders over the last two decades, its neurological and biochemical mechanisms are still not well understood. It is assumed that a reorganisation of nerval processes and structures [4] occurs. In particular, it is speculated that a spinal pattern generator (CPG) exists, which is reactivated by the task-oriented treadmill training [6,8]. The basic functional mechanisms of this CPG neuronal network, which is fundamentally involved in the generation of cyclic movement patterns, was described on the basis of comprehensive animal experiments. While the CPG receives in the physiological condition its input from the brain, it can be reactivated in the pathological condition by repeated and intensive afferent input from the periphery. It was shown, that a physiological, gait-phase related loading of the foot sole plays a crucial role in CPG activation [9]. Moreover, the results from a couple of studies indicate that the application of mechanical vibratory stimuli represent an intensive trigger for activation of different human sensor systems [10]. However, at the current sate of technology no device exists, which combines the basic principles of gait-phase related activation of foot pressure afferents with the possibility of generating vibratory stimuli. All established therapies that target on improvement of the ambulatory function are based on the principles of motor learning. A key principle of motor learning is that a higher number of repetitions of a movement task result in a higher therapeutic effect. Because of the trend towards shorter times of primary rehabilitation in specialized centers, a high intensity of locomotion therapy can only be achieved with an at-home training. However, currently there is no device existing for continuation of an automated robotic locomotion training at-home.

Aims of the study
The primary aim of this research project funded by the German Federal Ministry of Research and Education as a key experiment is the development and provision of a couple of prototypes of a novel robotic locomotion training device for at-home use. The aim of the clinical pilot study, which is performed in the framework of this project, is the evaluation of the safety and usability of the novel therapeutic device. Additionally, first results about the efficacy of the training should be obtained regarding to improvements of the ambulatory function of individuals with chronic motor incomplete SCI. Standardized neurological scores for objective quantification of the integrity of the motor and sensory spinal tracts (ASIA) and assessments for the spasticity status (Modified Asworth Scale) together with functional test focusing on walking function will be applied. The assessments will be performed 4 weeks before, 2 weeks before and directly at therapy start (all baseline), 4 and 8 weeks after therapy start and 3 months after therapy end (follow-up).

Primary outcome measures
The primary outcome measures are consisting of all tests for direct assessment of the walking function i.e., the changes of the 10 Meter Timed Walk Test, the TUG Test, the 6 Minute Walk Test and the WISCI II between baseline and end of therapy will be analyzed.

Secondary outcome measures
The secondary outcome variables consist of the motor and sensory scores of the International Standards for Neurological Classification of Spinal Cord Injury released by the ASIA (American Spinal Injury Association) for the quantification of the neurological changes and the modified Ashworth-Scale for the assessment of spasticity between baseline and therapy end. Furthermore, the subjective evaluation of the walking function and of the quality of life with the Münchner Lebensqualitäts -Dimensionen Liste (MLDL) will be analyzed. The analysis of the data assessed at a follow-up visit 3 months after therapy end will help to answer the question, if trained skills can be maintained or will even further improve.

General description
The investigational class IIa locomotion training device consists of two active orthoses (1) for moving the knee and ankle joints, two stimulative shoes (2), an adjustable seat (3), a box with electrical and mechanical components (4) and a human-machine interface (5). The movements of the active orthosis are generated by artificial, pneumatic muscles (6) in an antagonistic configuration. The air-mass flow of each muscle is controlled by dedicated proportional valves (7). In contrast to rigid electrical spindle drives the pneumatic muscles have the advantage of being inherently soft. This together with orthotic fittings with soft inlays (8) will ensure a safe movement of the users' legs. The stimulative shoe generates mechanical afferent stimuli on the foot sole for activation of the spinal pattern generator. To provide the possibility for implementation of different forces and stimulation patterns, each of the mediolateral bars of the stimulative shoe can be activated separately by individual short stroke cylinders (9), which are controlled by fast switching valves. A seat with adjustable height and inclination allows for a variation of the range of motion of the hip joint during the training. The component box (11) mounted under the seat contains a compressor unit providing the pressurized air needed for operation of the artificial muscles. A real-time controller (12) implements a model-based, robust and safe control of the device's movements. A hardware-based, redundant watchdog (13) continuously monitors the correct operation of the real-time controller and immediately stops the operation of the device in case a malfunction is detected. The human-machine interface visualises the training activity to the user (biofeedback), saves the measurement and training data and lets the user interactively operate the device.
Main features of the MotionTherapy@Home therapy device  Movements comparable to physiological walking (can be increased up to normal gait speed of 1.0 -1.2 m/s)  Apparatus for user-defined configuration of the mechanical stimulation of the foot sole (in anterior-posterior direction) called stimulative shoe  Tiltable seat backrest  Redundant, hardware-based safety monitoring

Effects (therapeutic, diagnostic)
The primary therapeutic goal of the novel locomotion trainer is the improvement of the ambulatory function of individuals with incomplete SCI. With the continuous, intensive generation of stimuli equivalent to those of physiological walking a relevant impact on the goal-directed reorganisation of spinal as well as supraspinal neural structures is expected. Additionally, the training-induced neuroplasticity will likely result in improvements of gait speed, distance and endurance and also of gait quality (symmetry, dependency on walking aids). Beside the direct functional effects also positive effects on the secondary consequences of SCI are expected. These more general therapeutic effects will only qualitatively be assessed. A more quantitative documentation is planned for future studies. In detail, the following positive therapeutic effects are expected:  Antispatic effect -due to passive joint mobilisation and activation of the spinal pattern generator  Prophylaxis of contractures-due to passive joint mobilisation with appropriate velocity  Reduction of edema -lymph drainage due to passive muscle mobilisation and activation of the spinal pattern generator  Stabilisation / activation of the cardiovascular system-shift of blood volume from legs into trunk due to activation of venous muscle pump  Eupepsia -passive mobilisation of intestines and activation of the autonomous nervous system

Undesired effects, other risks, stress for study participants
 Severe osteoporosis increases the risk for fractures. In the targeted study population of individuals with motor incomplete SCI a severe osteoporosis is very unlikely because of the preserved ability to stand and carry weight on the lower extremities. Persons with a known history of osteoporosis will not be included in the study. Worldwide, to our knowledge only one incident of a fracture occurred during a therapy with the gait robot "Lokomat", which was caused by a handling error.  In case of improper operation of the device by the end user there is an increased risk for skin squeezing, skin abrasion or skin burns of the body regions, where the orthotic fittings support the lower extremities. To minimize these risks patients will be in detail instructed in the proper use and handling of the device.  In principle, the device with its orthotic fittings bears the inherent risk of causing pressure sores. Although this risk is low in the included study participants with motor incomplete SCI due to their moderate to low muscular atrophy, a damage of the skin can occur particularly at the edges of the fittings. Based on the experiences with the routine application of locomotion robots in our hospital from the technical side all known measures e.g., use of soft materials for the orthotic fittings and avoidance of sharp edges, were undertaken to achieve a high level of safety and comfort for the user.  A triggered spasticity may lead to an increased active torque of the pneumatic drives, which may result in an injury of muscles or tendons. During the training the angles of all joints are continuously monitored and the device will stop automatically in case a predefined torque threshold is exceeded. Due to the implemented drive concept based on pneumatic actuators the joint torque is inherently limited to a safe level by the maximal pressure of the compressed air.  The other risks of the therapy correspond to those of regular physiotherapeutic interventions.

Type of study
The study is designed as a prospective, single center longitudinal pilot study with two therapy groups. Both groups receive therapy following the same schedule (baseline-, therapy and follow-up interval), but with two different types of foot sole stimulation (simulation physiological foot loading pattern vs. unspecific, gait-phase related vibrational stimuli). Safety and efficacy evaluation is to be made within a sample of 30 individuals with chronic, motor incomplete SCI (ASIA Impairment Scale C or D = limited ambulation possible). As only individuals with chronic (date of injury > 1 year) SCI are included, in whom spontaneous neurological recovery does not happen, a control group is not necessary. Recruiting patients within the time after trauma, in which spontaneous recovery occurs, would drastically enlarge sample size due to this group's high heterogeneity. Patients with spinal cord lesions due to tumors are excluded because of negative effects on walking function caused by the degenerative course of the disease.
The study design essentially matches the procedures of our study with the motor-driven gait orthosis "Lokomat" conducted 3 years ago, allowing for a direct comparison of the results of both studies.

Randomization procedure
Randomization procedure: The assignment to therapy group 1 (specific foot sole stimulation) and therapy group 2 (unspecific foot sole stimulation) will be randomized. The randomization will be balanced before beginning of the study into to therapy groups (0=unspecific, 1=specific) by a computer-generated list of random numbers with a block size of 2x15 patients.

Concealment:
The random number list is generated before study start by the statistician of the Stiftung Orthopädische Universitätsklinik. Due to the consecutive patient recruitment a further stratification is not necessary.
Procedure: Each patient who fulfils the inclusion and exclusion criteria will be immediately reported to the statistician. The statistician tells after having being informed about the data of every consecutively included patient via phone or personally the group assignment on the randomization list in a descending order and notes the name and birth date of the patient in the list. The first patient will receive the predefined assignment of number 1, the second patient the assignment of number 2, and so on. Vice versa, the person involved in the project makes a note of the assignment. All assignments will be documented in a logbook, which is provided on request only after study end. This procedure ensures that the concealment of the group assignment defined by the rest of the randomization list is guaranteed. A drop-out patient will be replaced by the next consecutively included patient with retaining his/her group assignment. The therapy period lasts 8 weeks with patients training 30-45 min. daily at least 4 and maximally 6 times per week using the novel therapy device.

Experimental procedure
 At the first therapy session patients are made familiar with the machine's functionalities and provided with a manual of instruction.  After that, a therapist adapts the orthotic fittings, the distance between the leg orthoses, the range of motion and gait speed individually to the patient's needs. After parameters are set up correctly, a 30-min. test training is performed, allowing the patient to familiarize with the device plus providing an opportunity to the therapist to correct errors.  After the test training the device is made available to the patient for an 8 weeks period for therapy at home. Where applicable, the device can be taken home by the patient straightaway. Otherwise it is delivered by a forwarder or personally.  The therapy sessions will follow the following scheme: Before a therapy session the device has to be switched on using the power switch located at the side of the seat. After this the device initiates a short self-test, checking the core components for correct functionality. On successful test conclusion, the patient transfers to the seat and fixates his legs in the designated orthotic fittings for thigh and shank, using hook and loop fasteners. The user interface allows changing the orthosis position in a way that a comfortable transfer is enabled. Finally the leg straps have to be closed concentrically over the legs and the tongue of the stimulative show needs to be manually fixated.
As soon as both limbs are safely fixed, the training can begin. To proceed with the training, a hand button has to be constantly pushed. With the hand button pushed down and the start button on the screen activated, training starts at slowest speed. The hand button enhances personal safety, as the training is instantly stopped when the button is released. The same effect is achieved when pushing the emergency button located at the side of the seating surface.  In case the therapy is interrupted because the hand button is released accidentally, training can be resumed within 30 seconds by choosing the option "continue with training". Otherwise, an emergency shut-down is initiated and training restart requires a complete reboot of the device (turn main power switch off and on again). Training speed can be adjusted in the user interface.  During training, the user interface displays data which inform the patient about his active muscle force during training. For instance, a "smiley" indicates a correct active participation of the patient. The user interface also informs about elapsed training time, distance covered and device errors.  Essential therapy data is saved on a mobile data storage device and is to be taken to study visits.

Screening assessments
During the baseline interval (4 and 2 weeks before study onset) a subset of study assessments will be performed. These are the following:  ASIA motor and sensory score, ASIA impairment scale [17]  Modified Ashworth-Scale [18]  WISCI II (Walking index for spinal cord injury) [19,20]  10 Meter Timed Walk Test including maximal walking speed, Timed "up and go" (TUG) test [21,22]  6 Minute Walk Test [23] 10.4 Invasive assessments None

Non-invasive assessments
For the assessment standardized neurological scores documenting the state of neurological innervation (ASIA) and spasticity (Ashworth) are used along with functional scores quantifying walking ability. In detail, the following clinical scores are collected:  ASIA motor (voluntary muscle force) and sensory score (sharp-dull discrimination, light touch perception), ASIA Impairment Scale (classification of overall neurological status defined by motor and sensory scores [17])  Modified Ashworth-Scale (assessment of joint resistance in lower extremities = categorization of spasticity) [18]  WISCI II = walking index for spinal cord injury II (classification of dependency on walking aids) [19,20]  10 Meter Timed Walk Test (time the patient needs to cover a distance of 10 m with comfortable or fast walking speed), Timed "up and go" (TUG) test (time the patient needs to stand up, walk a distance of 6 m and sit down again) [21,22]  6 Minute Walk Test (distance walked in 6 min) [23]