Hybrid Assistive Limb Intervention in a Patient with Late Neurological Deterioration after Thoracic Myelopathy Surgery due to Ossification of the Ligamentum Flavum

Purpose We evaluated improvements in gait after using the Hybrid Assistive Limb (HAL®) exoskeleton robot in a patient with late-onset neurological deterioration of lower extremity function after undergoing thoracic spine surgery for a myelopathy due to ossification of the ligamentum flavum. Case Presentation A 70-year-old man participated in ten 20 min sessions of HAL intervention, twice weekly for five weeks. The effects of each HAL session were evaluated based on changes in performance on the 10 m walk test (10 MWT), lower limb kinematics quantified from motion capture, and the activation ratio of the gastrocnemius, measured before and after the intervention. Muscle activity was recorded using surface electromyography and synchronized to measured kinematics. The HAL intervention improved gait speed and step length, with an increase in the hip flexion angle during the swing phase and a decrease in the activation ratio of the gastrocnemius. The modified Ashworth scale improved from 1+ to 1 and International Standards for Neurological and Functional Classification of Spinal Cord Injury motor scores from 34 to 49. Conclusion Intervention using the HAL exoskeleton robot may be an effective method to improve functional ambulation in patients with chronic spinal disorders.


Introduction
Robotic devices have been shown to be bene cial during the physical rehabilitation of patients with neurological disorders, including those with spinal cord injuries [1,2]. e Hybrid Assistive Limb (HAL) is an exoskeleton robot that assists the voluntary control of knee and hip joint motion bilaterally, by detecting signals from force-pressure sensors in shoes and weak bioelectric signals on the surface of the skin of associated muscles (Figure 1). Power units of the hip and knee joints of the exoskeleton suit include angular sensors and actuators for cybernic voluntary control (CVC) and cybernic autonomous control (CAC) [3]. HAL has unique operation system in using user's neuromuscular activities, compared to other rehabilitation robots such as Lokomat [4] (Hocoma, Switzerland), LOPES [5], or ReWalk (Robotics, Israel) [6], which are powered exoskeleton robots with angular sensors in the joints and pelvis and foot forcepressure sensors. E ective gait training outcomes using HAL have been reported in patients with chronic stroke [7][8][9] and spinal cord injury [9][10][11][12], as well as in patients following surgery for spinal myelopathy of the thoracic spine due to ossi cation of the posterior longitudinal ligament (OLF) [13][14][15]. We previously reported the e ect for recovery of muscle activity during motion using HAL for chronic spinal cord injury [12,16].
Gait dysfunction is the most important and common postoperative complication following surgery in patients with spinal myelopathy [13,17,18]; delayed-onset paralysis is a serious complication of the procedure. Lower limbs paralysis of the postoperative patients causes decreased walking distance, limitation of balance, and increased risk of falling. Such patients need rehabilitation to reduce the risk of falling, by an e cient use of their residual muscle activities, to maintain walking distance, and to keep their activities of daily living. In our case report, we describe improvement in gait obtained after a therapeutic intervention using the HAL robot for a patient who developed late-onset deterioration of neurological function after spinal surgery for OLF of the thoracic spine. Our intervention was conducted in accordance with the Declaration of Helsinki, with approval from the Ethics Committee of the Tsukuba University Faculty of Medicine (approval no.: H26-22). e patient provided informed consent for participation and publication, including the use of accompanying images.

Patient, Clinical Findings, and Timeline.
A 70-year-old man presented with progressive gait dysfunction following spinal surgery for a thoracic OLF. e progression of his gait dysfunction is summarized in Figure 2. e patient had undergone a T9-T12 laminectomy after being diagnosed with a thoracic OLF, 3 years after his rst visit. Five years after the operation, he underwent a laminectomy of L2-L3 and L5 for a lumbar OLF, which was associated with lower limb pain and gait disturbance. Pain and gait function improved after the second spinal surgery. However, 7 years after the second operation, he experienced a renewed deterioration in his gait function, which was associated with limited balance ability, slower walking speed, and frequent falls, that restricted him to walking indoors only. He also experienced a loss of hand dexterity and sensation, with di culty performing ne motor skills, such as fumbling with buttons. e patient was diagnosed with OLF again 11 years after the second operation and underwent posterior decompression with instrumented fusion of C3-C7. After the third surgery, hyperesthesia and motor impairment of his upper limbs improved, although the gait disturbances persisted. With 40 min of physical therapy, twice weekly, over several months following the third surgery, the patient regained the capacity to walk continuously for 10 min. However, his gait gradually deteriorated again, 1 year after his third surgery. He presented to our institution with di culty walking indoors due to limited balance. inking of his situation as being chronic stage, we considered that motor learning to reduce the risk of falling was necessary rather than improvement of muscle strength. For this purpose, the decision to use the HAL for gait improvement was made, because of our experiences of the e cacy of HAL in motor learning [12,16].

Diagnostic Assessment.
e T2-weighted magnetic resonance spinal image in Figure 3, obtained 2 years prior to the HAL intervention and 15 years after the T9-T12 laminectomy, shows a high-intensity change persisting in the spinal cord. From this nding, the patient was diagnosed with late-onset neurological deterioration due to spinal atrophy at the site of prior decompression of the thoracic spinal cord.

HAL Intervention.
e patient completed ten 20 min gait sessions using HAL during 5 consecutive weeks. Each session included 20-minute overground walk with HAL and 10 MWT without HAL before and after the session. All sessions were performed with a therapist, two assistants, and an engineer, under the supervision of a physician in case of emergency. e therapist and two assistants attached and detached the HAL exoskeleton suit, with the engineer implementing the gait analysis. For safety reasons, the Allin-One Walking Trainer (Ropox, Denmark), which includes a harness, was used for all sessions to prevent falls ( Figure 4). In this case, the CVC mode was used as it supports a patient's voluntary motion by providing an assistive torque to each joint according to voluntary muscle activity. e CVC mode allows the operator to adjust the degree of physical support provided to the patient and to gradually reduce this support as session progresses.

Follow-Up.
e 10 m walk test (10 MWT) was used to measure the primary outcome of the study and e ectiveness of the intervention. e 10 MWT was performed without HAL or any assistance, but under supervision to ensure safety. e number of steps and time required to complete it were measured. Video recording and motion capture (MX with 16 T20S cameras; Vicon, UK) were used for the gait analysis, using the VICON Plug-in Gait marker placement for the lower limbs. Motion data were collected at 100 Hz, and the sagittal plane joint angles were computed using the Plug-in Gait model.
Step length and cadence were also quanti ed from the kinematic data, using heel strike as the reference time point.
Activity of the gastrocnemius muscle was recorded, bilaterally, using a Trigno Lab wireless EMG system (Delsys, United States). e EMG data was collected at 2000 Hz in synchronization with the motion data and band-pass ltered at 30-400 Hz (using custom scripts, MATLAB, MathWorks, United States). e patient had a drop foot during swing phase, because of excessive and out-of-phase activation of the ankle plantar exors. erefore, the phase-dependent muscle activation of the plantar exors was evaluated using the ratio of the integrated activity of the gastrocnemius during the swing phase to the activity of the muscle over the entire step cycle (swing and stance phase combined). e kinematic data for seven step cycles were extracted from each 10 MWT trial. For each of the extracted cycles, the maximum and minimum angles at the hip, knee, and ankle joints were extracted to calculate the joint range of motion. e kinematic pro le of each lower limb joint for each extracted step cycle was normalized to the duration of the step cycle and averaged across cycles.
We also evaluated the modi ed Ashworth scale (mAs) of the gastrocnemius muscle [19]    Spinal Cord Injury (ISNCSCI) motor scores motor score as a measure of lower limb motor function.
All outcome variables were measured before and after the HAL intervention (PRE-and POST-10 sessions). Simple linear regression was used to examine the relationship between session number and changes in gait speed, cadence, and step length during the 10 MWT. All analyses were performed using JMP ® 13.0.0 (SAS Institute Inc., Cary, NC, USA); a p value < 0.05 was considered signi cant.  Figures 6-9. After the HAL intervention, the mean ± SD maximum hip angle exion during the swing phase increased from 23.4 ± 1.3°to 31.3 ± 1.0°on the left side and from 28.1 ± 1.6°to 35.3 ± 2.0°on the right side (Figure 6(a)). Similarly, maximum exion decreased at the knee from 57.2 ± 1.5°to 53.3 ± 0.6°on the left side and from 63.6 ± 1.9°t o 57.3 ± 2.4°on the right side (Figure 7(a)), with a further decrease in ankle plantar exion, from 5.3 ± 1.8°to 3.6 ± 1.2°o n the right side (Figure 8(a)). Surface EMG pro les of the gastrocnemius are shown in Figure 8(b). e muscle activation ratio of the gastrocnemius decreased from 30 ± 10% to 14 ± 5% (Figure 8(c)), while toe clearance increased from 59 ± 6 to 62 ± 7 mm (Figure 8(d)).
Step length increased from 477 ± 31 mm to 510 ± 23 mm on the left side and from 417 ± 43 to 464 ± 26 mm on the right side (Figure 9). e gastrocnemius mAs score improved from 1+ to 1. e lower limb ISNCSCI motor score improved from 34 to 49, with a noteworthy increase from 3 to 5 for left hip exion and ankle plantar exion, bilaterally. No adverse events associated with the HAL intervention were identi ed.

Discussion and Conclusions
In this case report, we describe improvement in the gait function of a 70-year-old patient with chronic spinal disorder following a gait intervention using the HAL exoskeleton robot. e delayed-onset paralysis in our patient was a possible symptom of spinal atrophy [20,21], which can be associated with serious gait dysfunction that should not be ignored [13,17,18]. Surgical decompression can provide an e ective treatment for spinal myelopathy. Typically, following decompression, neurological recovery of the upper limb function precedes recovery of lower limb function, with sphincter function being the last to recover and often being incomplete [22,23]. Rehabilitation is usually needed to improve gait function postoperatively. In a systematic review of gait rehabilitation using the HAL, Wall et al. [11] reported the feasibility of the HAL exoskeleton suit to improve gait function in patients with lower extremity paresis in hospital and rehabilitation settings. Cruciger et al. [24] reported improved gait function using locomotion training with the HAL exoskeleton in patients with chronic neuropathic pain and those with chronic spinal cord injury.
HAL is a wearable robot suit to assist voluntary control of the knee and hip joint motion by detecting very weak bioelectric signals from associated musculature. However, other robotic devices, such as the Lokomat [4] and the LOPES [5], and powered exoskeleton devices, such as ReWalk [6], do not include sensors to detect the neuromuscular activation of wearers. erefore, the HAL is unique in its use of innate neuromuscular control, compared with other rehabilitation robots, which can enhance motor learning [12,16]. e HAL intervention we implemented improved speci c components of gait (speed and step length), which subsequently improved ISNCSCI subscores of lower limb motor function, from 34 to 49 m. Moreover, the improvement in mAs score of the gastrocnemius indicates that the motion assist provided by the HAL during the sessions may be e ective in reducing plantar exor spasticity. Our previous study showed a similar reduction of spasticity after a HAL intervention in a patient with complete quadriplegia [25].
is decrease in spasticity may have contributed to improved muscle control and, therefore, strengthening of lower limb muscles. Improvements in the distance of independent walking and decrease in spasticity were considered as fundamental to the improvement in ISNCSCI scores. To our knowledge, this is the rst report to have used gait analysis including kinematic and muscle activation data to quantify the e ects of gait training using the HAL exoskeleton robot for a patient with a spinal cord disorder. e greatest concern during locomotion for our patient was the risk of falling due to insu cient hip exion and foot clearance over the swing phase of gait. After the HAL intervention, we identi ed a decrease in gastrocnemius activity  during the swing phase, with an increase in activity over the stance phase. Plausible reason for the improved toe clearance might have been decreased gastrocnemius spasticity over the swing phase, which needs further investigation. A previous study reported a signi cant decrease in ankle plantar exor power in elderly individuals, compared with young adults [26,27], with strengthening of the ankle plantar exors being e ective in improving stability during gait and preventing falls [28]. In our patient, a decrease in the range of plantar exion of the right ankle during gait, compared with normal values, persisted even after the HAL intervention. However, manual muscle testing of the gastrocnemius improved from a score of "3/5" to "5/5", bilaterally. We consider that this improvement in ankle plantar exor strength would reduce the likelihood of falling. e increase in the exion angle of the hip over the swing phase resulted in an increase in step length. is increase in the range of hip exion, in combination with an increase in gastrocnemius activity during stance phase and a decrease during swing, as well as an increase in toe clearance, reduced the likelihood of falling, while increasing step length and improving gait function for daily locomotion. erefore, our HAL intervention resulted in considerable functional improvement for our patient. is way, the gait improvement observed in the current case after HAL intervention indicates a new possibility of HAL: modi cation of gait to achieve improvement of daily activities in patients with limited balance ability because of neurological disorder, making comparison to e ectiveness of gait training in the rather severe cases reported in previous reports [12][13][14][15]25]. When evaluating the clinical applicability of our ndings, it is important to note that this is a single case report and, therefore, a clinical trial is needed to evaluate the e ectiveness of the HAL exoskeletal training in improving gait in patients with di erent conditions. is would be important to clearly establish guidelines for the use of HAL exoskeletal training in rehabilitation. Speci cally, for patients with spinal cord injuries and disorders, case-control studies would be important to compare the e ectiveness of the HAL intervention compared to conventional physical therapy or interventions using other robotic systems.
In conclusion, using a HAL-based intervention improved the gait of a patient presenting with late onset of neurological deterioration after surgery for thoracic myelopathy caused by OLF, including improvements of the ISNCSCI scores of lower limb motor function. Based on our experience, we propose that gait training using the HAL exoskeleton robot be used to provide a safe and feasible intervention to improve functional gait in patients with chronic spinal cord disorders. Step length before and after the HAL intervention.

Ethical Approval
is intervention was conducted in accordance with the Declaration of Helsinki, with approval from the Ethics Committee of the Tsukuba University Faculty of Medicine (approval no.: H26-22). Consent e patient provided informed consent for participation and for publication, including the use of accompanying images. Disclosure e current study was designed by the authors. Cyberdyne was not directly involved in the study design, the collection, analysis, or interpretation of data, writing the report, or the decision to submit the paper for publication. No commercial party having a direct nancial interest in the results of the research reported in this case report will confer a direct nancial bene t to the authors (Masakazu Taketomi, Yukiyo Shimizu, Hideki Kadone, Shigeki Kubota, Tetsuya Abe, Aiki Marushima, Tomoyuki Ueno, Yasushi Hada, and Masashi Yamazaki) or on any organization with which the authors are associated.