Reweighting of the sensory inputs for postural control in patients with cervical spondylotic myelopathy after surgery

Cervical spondylotic myelopathy (CSM) is a degenerative disease of the cervical spine. CSM is caused by progressive degeneration of the cervical spine and changes to the surrounding tissues [1]. Adjacent tissues and abnormal cervical motions may narrow the cervical canal and compress the spinal cord. The compression of the spinal cord can cause various neurological symptoms, such as paresthesia, muscle weakness, muscle spasm, and gait disturbance [2]. Moreover, patients with CSM exhibited postural instability compared with healthy adults [3]. Postural instability may result from impairment in proprioception and the integration of the somatosensory system, which is controlled by the neural tract in the spinal cord [3, 4]. Postural control is controlled by the sensorimotor system, which is associated with the visual, vestibular, cerebellar, and proprioceptive systems [5] (Fig. 1). When any sensory system is impaired, other sensory systems may over-activate to compensate for the insufficiency of the impaired sensory system [6]. However, understanding regarding the compensatory sensory weighting in the postural control of patients with CSM is limited.

Spinal decompression surgery is a recommended treatment aimed to remove the compressed structure and enlarge the space of spinal canal [7]. Surgical decision-making is critical for clinicians and patients. Moreover, various symptoms and signs may affect the prognosis of the surgery. Previous studies have provided inconsistent results regarding the effectiveness of surgery on the gait performance of patients with CSM [8, 9]. Furthermore, no study has discussed postural control in patients with CSM after surgery. After surgery, the proprioception in patients with CSM may recover [10]. Therefore, the sensory system may be reweighted because the demand of compensation is reduced. However, the sensory-reweighting mechanism in patients with CSM after surgery remains unknown.

To evaluate postural steadiness quantitatively, the center of pressure (COP) is often used and measured by a force platform [11]. COP displacement reflects how the body moves to maintain balance stability, and COP can be used for evaluation of the balance performance in patients with spine disorder [12]. Time-domain measures of the COP, such as mean distance and velocity, are the common methods for evaluating the postural steadiness. However, time-domain measures reflect only the final integration of sensorimotor systems without indicating the underlying mechanism. Furthermore, balance performance may not be reflected by time-domain measures when people use various control strategies, such as a rigid pattern which might be associated with the fear of falling [13]. Therefore, a new approach is required to analyze the COP signal for understanding the integration of sensorimotor systems in postural control.

Time–frequency analysis of COP was suggested for investigating the postural control of the human body [14]. The wavelet transform, which is a type of time–frequency analysis, was applied to describe the characteristics of body sway and displacement [14]. The input COP displacements were considered linear combination of several wavelets [14]. In a previous study, the time–frequency approach was used to investigate the impact of age and injuries on the balance mechanism [15, 16]. However, no study has applied the time–frequency analysis to investigate the changes in the postural control mechanism of patients with CSM after surgery.

In time-frequency analysis, each frequency band is considered an afferent of each subcomponent of the sensory system [17]. The correspondence of sensory systems and frequency bands is illustrated in Fig. 2 [18‐26]. Since the frequency band of each sensory system may be overlapped, the classifications of frequency band were various in the previous studies. In the early studies of time–frequency analysis, the authors classified the sensory system into two or three frequency bands [27, 28]. In recent studies, authors have classified the sensory system into four frequency bands [29‐31]. Therefore, we analyzed the COP data in four distinct frequency band: moderate-frequency (1.56 to 6.25 Hz), low-frequency (0.39 to 1.56 Hz), very-low-frequency (0.10 to 0.39 Hz), and ultralow-frequency (less than 0.10 Hz) [29‐31]. This new approach may provide additional information regarding the postural control in patients with CSM during quiet standing. It may also be used as an assessment tool by clinicians for surgery decision-making.

As indicated by the aforementioned literature review, the sensory weighting mechanism in patients with CSM is unclear. Therefore, the purpose of this study was to investigate the difference of sensory weighting in the postural control of patients with CSM and healthy adults. We hypothesized that the weighting of the sensory systems in the postural control of patients with CSM may be different from that in healthy adults. We also hypothesized that the sensory systems in postural control would be reweighted after surgery.

This paper describes a prospective longitudinal study with 6-months follow-up observing the balance performance and functional outcome in individuals with CSM after surgery. The Institutional Review Board of National Taiwan University Hospital approved this study according to the Declaration of Helsinki. The study was registered at ClinicalTrial.​gov (NCT03396055). The flow diagram of this study is illustrated in Fig. 3. The experiment procedure was explained to the participants at the Department of Surgery in the National Taiwan University Hospital.

In this study, 61 adults were recruited from National Taiwan University Hospital, and 48 adults were consequently analyzed (CSM group: n = 24, healthy control group: n = 24). The demographics and functional outcomes of the participants are described in Table 1. No significant difference was observed in the sex, age, height, or weight of the CSM group and control groups.

This study investigated the difference of sensory weighting in the postural control of healthy adults and patients with CSM after surgery. The results of this study supported our hypotheses that the postural control of patients with CSM is less efficient than that of healthy adults before surgery. The low efficiency in the postural control of the patients with CSM was caused by insufficient proprioceptive inputs, as indicated by the decreased percentage of the moderate-frequency band. After surgery, postural control improved marginally, with less compensation from the visual and vestibular inputs compared with the baseline values. However, the patients with CSM still exhibited inefficient postural control.

After surgery, the functional outcome indicated by the mJOA-LEF and Nurick scale improved; however, the functional outcome indicated by the JOACMEQ-LEF score did not. The mJOA results were consistent with the results of previous studies, in which the mJOA and Nurick scale have been demonstrated to improve 3 and 6 months after surgery in patients with CSM [42, 43]. Moreover, a previous study that used the functional independence measure for assessing locomotion in patients with CSM reported a positive surgical effect on locomotion [44]. In this study, the mJOA and Nurick scale provided evidence of the positive effect of surgery on walking ability.

To consider the effect of post-surgical rehabilitation, 4 patients with CSM in this study reported that they had treated with physical therapy between 3 months and 6 months after surgery. Only one of these 4 patients tended to improve between 3 months and 6 months after surgery. The types of physical therapy they received after surgery included cervical range of motion exercise, cervical traction and modalities intervention. These interventions may improve the range of motion of cervical joint, but may not directly affect the balance performance of patients with CSM. Therefore, the improvement of balance performance was mostly contributed from the surgery in this study.

In this study, the patients with CSM exhibited insufficient muscular activity and a spinal reflexive loop from the lower extremities. The patients with CSM required the visual, vestibular, and cerebellar systems to be over-activated to compensate for the aforementioned insufficiency. Similar results with large COP displacements have been obtained for patients with CSM [38, 45]. According to our results, the proprioception input decreased with compensation from other sensory systems; this is in agreement with previous research on postural control in patients with neck pain [31]. The decreased recruitment of the muscular proprioceptive system was countered by the increasing usage of the vestibular system in patients with neck pain [31].

The unsteady balance performance may be caused by the blockage of information transmission along the lateral corticospinal tract and dorsal spinocerebellar tracts, which are responsible for movement and proprioceptive feedback, respectively. Poor joint proprioception leads to poor estimation of body position [4], and a prolonged conduction time of the corticospinal tract may cause delayed body adjustment [46].

Under the EC condition, the patients with CSM required increased sensorimotor inputs to cope with the postural unsteadiness, as indicated by the increased total energy content. The aforementioned result is consistent with a study that investigated the influence of vision on postural control [41]. Although our study indicated the increased demand of sensory feedback in patients with CSM, we did not observe differences between the two groups in the weighting of each sensory system. The possible reason for this observation is that standing with EC condition may be too challenging for CSM patients, as indicated by the increased weighting on proprioceptive system.

Under the EO condition, the postural sway decreased 3 months after surgery in patients with CSM. Moreover, a marginal increase in the weighting of proprioceptive systems and a decrease in the weighting of visual and vestibular systems were observed in the patients with CSM after surgery. The discrepancy of compensation from the visual and vestibular systems between the patients and healthy adults diminished 6 months after surgery. This result may be an indirect evidence of the improvement in the proprioception of the lower extremities [10]. This study is the first to investigate the postural control in patients with CSM after surgery.

Under the EC condition, the postural sway of the patients with CSM decreased after surgery; however, the sway remained greater than that of the healthy adults. In a previous study, impairment in the lower extremity function commonly persisted after surgery [47], which may affect postural control in patients with CSM. Long-term compression of the spinal cord may induce cell apoptosis of the nerves in the spinal cord, causing an irreversible neurological deficit and slow recovery of the spinal cord [48]. Furthermore, metabolic changes in the cortex may decrease the recovery in motor and sensory functions for patients with CSM [49]. Therefore, the postoperative training must be used for providing new stimulation to the neurons of the spinal cord and enhancing cortical plasticity.

In addition to the neural tract, proprioception from the musculoskeletal system may be altered because of degenerative changes or the surgical realignment of the cervical spine. Effective postural control requires coordination among multiple joints [50, 51]. Structural changes in joint position and alignment may affect joint coordination when controlling posture for adaptation [41, 52]. In a previous study has shown that an adaptive strategy of postural control was developed for maintaining balance in patients with spinal stenosis after surgery [52]. Therefore, changes in cervical alignment caused by internal factors, such as degeneration, or external factors, such as surgery, may contribute to proprioceptive insufficiencies around the neck [30]. Furthermore, changes in cervical alignment may affect head position and body orientation, resulting in poor balance [53, 54].

Time–frequency analysis has been used to analyze the COP signal for postural control in pathological diseases [29, 30, 41]. Our team was the first to use Gabor transform wavelet analysis for analyzing the COP time–frequency domain [39]. In contrast to discrete wavelet analysis, which has been performed in many previous studies, Gabor transform wavelet analysis can be used to decompose the COP signal and obtain the instantaneous frequency and local characteristics of the original signal with a fluctuating frequency [39]. The Gabor transform can effectively extract noise from the signal. Therefore, the energy content obtained with the Gabor transform is an accurate approximation of the real energy of the signal. Furthermore, the energy content of the COP signal in each frequency band can be represented by a spectrogram with warm colors for high energy and cold colors for low energy [14]. The color mapping of each frequency band may be used as bio-feedback for balance training in future studies.

Discrete wavelet analysis has been used in patients with adolescent idiopathic scoliosis. In discrete wavelet analysis, the energy content is decomposed into three distinct frequency bands, which represent the somatosensory, vestibular, and visual systems [41]. In this study, we decomposed the energy content into four distinct frequency bands, which represent the proprioceptive, cerebellar, vestibular, and visual systems. This quadripartite classification can provide a comprehensive understanding of the sensory system in various challenging tasks, including visual disturbance and multitasking.

The energy content of the moderate-frequency band of this study was higher than observed by Sim et al. [41], whereas the energy content of the ultralow-frequency band of this study was lower than that observed by Sim et al. These discrepancies may arise from the age difference of the participants in the two studies. The participants in our study were older adults; however, the participants in the study of Sim et al. were adolescents. In older adults, increased muscle activity may compensate for insufficient visual feedback [55]. This compensation strategy may have caused the increased energy content in the moderate-frequency band and decreased percentage of the ultralow-frequency band of our study. The difference may also result from the different methods of time–frequency analysis used. The Gabor transform can accurately provide the energy content of each frequency band with specific cutoff points, whereas discrete wavelet analysis can decompose the signal of each frequency band only with approximate cutoff points.

In this study, the underlying mechanism of postural control in patients with CSM was investigated using a modern mathematical method. This method can simultaneously consider the time and frequency domains in the signal process. Moreover, the effect of surgery on the postural control in patients with CSM was investigated after 6 months. The compensatory changes in the neurological system of the patients with CSM were also observed after surgery. Our study indicated that simple stabilometry can be used in clinical settings for assessing the sensory weighting in the postural control of patient populations. In future studies, a large sample size can be used, and patients can be divided into subgroups based on the severity of CSM. Postural assessment can be used as a diagnostic tool to assist in surgical decision-making. Furthermore, the compensatory strategy existed even after surgery, which implies that postoperative balance training is required for patients with CSM to prevent falling and secondary injury. In future studies, simulations of falling and task-specific training, such as perturbation-based training, can be applied to patients with CSM to accelerate the reweighting of postural control systems and stimulate neuromuscular modulation [56, 57].

Our study has several limitations. First, static balance was not preaccessed to identify the eligible patients. The Nurick scale was used in this study to select potential patients with balance disturbance. However, the Nurick scale addresses ambulation problems rather than balance disturbance [36]. Therefore, the participants included in this study may have had minor balance issues. Second, the patients had disparate symptoms and disease durations, which caused a large standard deviation in the CSM group. The onset of CSM symptoms was unclear or vaguely reported by the patients in a subjective manner. Recruiting patients with a similar duration and severity of CSM was difficult. However, we performed the first assessment within 1 month before surgery to ensure that the baseline was consistent for all the patients. Lastly, we did not conduct long-term follow-ups 1 or 2 years after surgery. Although this study has several limitations, it provides results of a scientific investigation of the postural control and effect of surgery in patients with CSM. In future studies, a larger sample size should be used and long-term outcomes should be investigated.

Before surgery, the patients with CSM may have had inefficient postural control with compensatory sensory weighting, as indicated by the decreased weighting on proprioception and increased weighting on the other three studied sensory inputs. After surgery, postural control improved, with low compensation from the visual and vestibular inputs in the patients with CSM. However, the improvement was insufficient because the weighting on proprioception was lower in the patients with CSM than in the healthy adults. The COP energy content analyzed through the Gabor wavelet transform can enable further understanding of human postural control. In future studies, long-term follow-up after surgery should be conducted. Furthermore, patients with CSM may require balance training and posture education after surgery to prevent falling and secondary cervical injury.