Clinical Factors for Predicting Neurophysiological Improvements Upon Extension and Flexion in Cervical Spondylotic Myelopathy Patients: A Retrospective Cohort Study

Zhengran Yu Sun Yat-sen University First A liated Hospital Jiacheng Chen Sun Yat-sen University First A liated Hospital Xing Cheng Sun Yat-sen University First A liated Hospital Zhiyuan Zou Sun Yat-sen University First A liated Hospital Dingxiang Xie Sun Yat-sen University First A liated Hospital Yuguang Chen Sun Yat-sen University First A liated Hospital Xuenong Zou Sun Yat-sen University First A liated Hospital Xinsheng Peng (  pengxs66@yahoo.com ) Sun Yat-sen University First A liated Hospital https://orcid.org/0000-0002-6192-5635


Introduction
The static and dynamic narrowing of the cervical canal is one of the most important factors that causes cervical spondylotic myelopathy (CSM). [1,2] Cervical motions could rapidly alter (improve or worsen) cervical and referred symptoms, depending on the direction of end-range positioning or of repetitive end-range testing of cervical spine movements. [3] Somatosensory evoked potential (SSEP) is an objective neurophysiological test widely used in evaluating neurological functions and predicting prognosis in CSM. [4] In our previous studies [5,6], we found most patients had dynamic SSEP (DSSEP) deteriorated upon extension and exion, as other studies suggested [7]. For those patients, their DSSEP N13 amplitude ratios correlated with preoperative mJOA scores and postoperative recovery rates, as well as compressive degrees of the spinal cord in axial MRI and cervical segmental instability in dynamic X-ray. [6] More severely compressed spinal cord and unstable canal structures rendered smaller DSSEP N13 amplitude ratios, i.e. worse dynamic neurological compensatory capacities. [6] However, there were some CSM patients presented signi cantly improved DSSEP amplitudes and/or latencies during cervical spine exion and/or extension, whose DSSEP N13 amplitude ratios were obviously unmatched with their MRI compression degrees. [5,6] These patients also frequently presented symptomatic alleviations at extension or exion. The reasons why patients presented abnormal DSSEP upon extension and/or exion positions are still not clear.
It has been reported that cervical extension shortens and widens the cervical cord, but diminishes canal spaces and worsens the medullary compression visible on dynamic MRI. [8,9] On the other hand, although the spinal canal diameter could increase in exion, the stretching forces applied to the spinal cord could also increase longitudinal tension, and cause ventral spinal cord compression against osteophytes and discs -worsening an eventual ventral compression at this position [10]. Based on our previous nding on the correlations between DSSEPs and radiographic results, we expanded our sample size and separated extension-or exion-DSSEP improved patients away from others, in order to investigate the commonalities of CSM patients with improved neurological function upon extension and exion respectively, and to determine prediction criteria for extension-or exion-DSSEP improvements with clinical and neutral-position imaging.

Patient cohort
This single-center retrospective study included consecutive CSM patients at our department with preoperative DSSEP and MRI tests between 2015 and 2019.
The Human Ethics Committee of the First A liated Hospital of Sun Yat-sen University approved the trial (Ethics number: [2020]151), and each patient provided informed consent.
Patients with a previous surgical or trauma history, spinal tumor, or peripheral neurological disease were excluded. The demographic data collected included sex, age and critical comorbidities. Thirty-eight subjects overlapped with our prior report [6], which evaluated correlations between amplitude ratios of DSSEP and MRI measurements. The CSM disease duration (i.e., time from the onset of CSM-related neurological signs) and modi ed Japanese Orthopedic Association (mJOA) score [11] for each patient at the time of DSSEP tests were recorded. Other clinical signs, including gait impairment, upper limb weakness and atrophy, and the Hoffmann sign, were also documented.
Realization and measurement of DSSEP An electrophysiological monitoring system (Nicolet Endeavor CR) was used to elicit and record the DSSEPs. Median and ulnar nerve DSSEPs were examined using established methods described in our previous study. [6,12] Recording electrodes were placed over the spinous process of the 2nd cervical vertebra (C2S), the contralateral parietal cortex (Cc) and forehead reference site (Fz) regions of the scalp, and Erb's points ipsilateral (EPi) and contralateral (EPc) to the stimulation. [13] The DSSEP waves for each recording montage, labeled EPi-EPc, C2S-EPc, and Cc-Fz, were recorded as N9, N13, and N20, respectively. We adopted N9 as the standard reference channel.
When N9 was unidenti able or poorly reproducible, the existence of peripheral nerve pathology was suspected. The DSSEPs were measured upon a neutral neck position rst. Patients were then tested with neck positions at approximately 35° exion, followed by approximately 20° extension of the cervical spine, using a device for elevating the head and neck with minimal discomfort to the subject (Fig. 1). To con rm the reproducibility of the DSSEPs, each measurement was carried out at least three times by a spine surgeon and two electrophysiologists.
We compared the same patient's median nerve SSEP upon extension or exion with those in the neutral position. A DSSEP improvement upon extension or exion was de ned as a shortened N13 or N20 latency exceeding 2.5 SD of that at neutral position (which were 1.78 ms for N13 and 2.01 ms for N20 in this study); or increased N13 or N20 amplitude exceeding 20% compared with the patient's SSEP in the neutral position. We de ned an immeasurable SSEP as a waveform that could not be identi ed by averaging over 500 sweeps. Any measurable SSEP waveform would be considered a DSSEP improvement compared with an immeasurable SSEP waveform. Patients with improved DSSEP upon extension or exion was classi ed into the extension-improved (EI) or exion-improved (FI) groups respectively. Otherwise, the patients were classi ed into the extension-nonimproved (EN) or exionnonimproved (FN) group.
Imaging methods and analysis protocol All MR examinations were performed with a 3.0-T MR imager (Siemens Trio) with the patients lying in the supine position on a spine-array coil. The authors evaluated compressed spinal cords using standard imaging sequences.
Qualitative MRI features on sagittal T2-weighted sequences included the presence of cervical ligamentum avum hypertrophy (LFH) and cord intramedullary T2WI signal hyperintensity (IHI). LFH is de ned as a thickened ligamentum avum compared with the thickness of adjacent segments, with a loss of epidural fat tissue and dural sac compression. [14] Quantitative MRI features on T2-weighted sequences included the compression ratio(CR), which was obtained via dividing sagittal diameter by transverse diameter of the cord at the most compressed site [6], the number of stenotic segments, cervical stenosis grade (classi cation of Mühle et al. [15], Grade 0 to 3) and disc degeneration grade (classi cation of Miyazaki et al. [16], Grade 1 to 5) of the most compressed segment.
The cervical alignment types were measured on plain cervical lateral radiographs and categorized into one of the following four groups according to the modi ed Toyama method [17] (Fig. 2).
These measurements were carried out by 3 independent investigators.

Statistical analysis
Age, CSM disease duration and mJOA score differences between the EI/FI and EN/FN groups were calculated with Student's T-test. The differences in DSSEP changes upon exion, cervical alignment types, ligamentum avum hypertrophy and intramedullary intensity between the EI/FI and EN/FN groups were calculated with chi-square tests.
The numbers of stenotic segments, Mühle stenosis grade and disc degeneration stage between these groups were compared with the Kruskal-Wallis method.
With regard to clinical and MR imaging criteria that predict DSSEP improvement upon extension or exion, the following criteria were evaluated: disease duration no more than 6 months, no more than 2 stenotic levels, straight or sigmoid cervical alignment, and Mühle grade 3 stenosis.
Bivariate analysis was used to evaluate the relationship between DSSEP improvement upon extension or exion and a set of clinical and radiographic criteria of interest. Variables with p < 0.2 in the bivariate analysis were entered into a forward stepwise multivariate logistic regression model. Model t was assessed with the Omnibus tests of model coe cients and Hosmer-Lemeshow goodness of t test. A signi cant value for the Omnibus chi-square test indicates a credible improvement of the new model over the baseline model, and a non-signi cant value for the Hosmer-Lemeshow chi-square test suggests an absence of biased t. After the nal logistic model was performed, the probability of DSSEP improvement upon extension or exion was calculated. The statistical software R (R version 3.6.1) was used for statistical analysis.

Demographic and clinical results in each group
Forty-nine CSM patients (55.8 ± 11.3 years; 28 men) were included in this study. Nine (18.4%) patients had DSSEPs improved upon extension, and eleven (22.4%) upon exion. Obviously, the EI group's extension DSSEP N13 amplitude ratios were signi cantly higher than the EN group (T-test, p < 0.001), but the two groups' exion DSSEP N13 amplitude ratios did not vary. The FI group had signi cantly higher exion DSSEP N13 amplitude ratios(T-test, p < 0.001), but its extension DSSEP N13 amplitude ratios did not change compared with the FN group. The EI group patients had signi cantly shorter disease duration (T-test, p = 0.024) than patients in EN group. No differences were found between the EI and EN groups with respect to sex, age, mJOA score, gait impairment, upper limb weakness, or positive Hoffmann signs. No statistically signi cant differences were found for any demographic or clinical data between the FI and FN groups. (Table 1)

Discussion
To the best of our knowledge, this is the rst cohort study to report neurophysiological improvement upon dynamic positions among CSM patients, and to identify clinical and radiographic factors related to neurological improvements upon cervical extension and exion.
Currently, prolonged extension and exion are commonly recognized as deleterious activities for CSM patients. [18] Cervical extensions make the ligamentum avum bulging inward, decrease the dorsal subarachnoid space [19] and increase Mühle stenosis grade. [20] On the other hand, cervical exions increase the longitudinal strain of the cord and induce compression against ventral spondylotic bar. [9,21,22] This was in line with our previous [6] and current nding that most patients had decreased DSSEP at both extension and exion. However, we also found 9 (18.4%) patients had signi cant DSSEP improvement upon extension, and 11 (22.4%) had signi cant improvement upon exion in the current study, demonstrating that the extended and exed position could relieve patients' neurological de cits in some cases. Interestingly, many patients in the EI group reported their preference of activities requiring neck extension, such as badminton and some types of gymnastics, whereas patients in the FI group usually felt more comfortable at exion, suggesting the consistency between symptomatic and DSSEP changes. Age, duration of symptoms and baseline mJOA score were reported to be signi cant predictors for CSM outcomes. [23] In this study, although patients in the EI groups exhibited no difference in age, sex, mJOA scores, or several other clinical signs and symptoms compared with those in the EN group, they had signi cantly shorter disease durations. This can be explained by that newly emerged spinal cord impingements are generally more easily reversible than those of patients suffering from long-standing compression at dynamic neck positions.
Regarding the radiographic characteristics, the CR were similar between the EI and EN groups, or the FI and FN groups, indicating the correlations between the percent change of DSSEP amplitude ratios and the spinal cord compression degrees described in our previous study [6] were not suitable for these DSSEP-improved patients. We found the number of involved segments and cervical alignment types were signi cantly different between the EI and EN groups in this study. Further Logistic regression analysis found that an involved-segment number ≤ 2 was a signi cant criterion for predicting DSSEP improvement upon extension. Dynamic MRI studies revealed that for patients with multiple involved segments, many of the segments that were not signi cantly compressed in the neutral position could narrow greatly upon extension. [2,9,24] Thus, patients with multilevel stenosis would suffer more serious neurological deterioration upon extension, probably due to signi cantly less compensative space resulting from multiple segmental pincer effects, as our dynamic MRI for a patient with deteriorated exion DSSEP shows (Fig. 3). On the contrary, fewer segments usually cause focal and limited compression and leave more compensatory space, making the patient's neurological de cits more easily relieved upon extension. EI patients also tended to have straight or sigmoid cervical alignments, which is another signi cant criterion for predicting DSSEP improvement upon extension. For CSM patients with these two alignment types, their cervical cords were usually tightly longitudinally stretched and suffered from focal anterior compression, such as protruding discs or osteophytes from focal kyphosis in the neutral position. During extension, their cervical cords could be longitudinally relaxed and draped backward, thus ameliorating the stretching tension and the anterior compression to some extent, [9] as our dynamic MRI for a patient with improved extension DSSEP shows (Fig. 4). Lordotic patients will not experience such bene ts because their cords are already longitudinally relaxed in their neutral positions. [2] Kyphotic patients experienced much more severe potential ligamentum avum bulges and pincer effects upon extension, [25][26][27] which could offset the bene ts of decreased longitudinal tension.
In addition, disease duration ≤ 6 months is a signi cant predictive criterion for DSSEP improvement upon exion.
The rationale is the same as that mentioned above, i.e., that new spinal cord impingements were more restorative and reversible once the compression status changed. Besides, the Mühle stenosis grade in the FI group was signi cantly greater (Kruskal-Wallis, p < 0.05), and the grading of Mühle grade 3 is a signi cant criterion for predicting DSSEP improvement upon exion. Mounting evidence has shown that CSM patients could have expanded cervical canal, even with cord decompression on exion MRI. [8,9,24,28] The diameter of the dorsal subarachnoid space at each level from C2 to C7 could increase up to 89% in exion. [19] According to those dynamic MRI results, the severely compressed Muhle grade 3 patients at neutral MRI could probably enjoy more bene ts from spinal canal enlargement upon exion, and thus are more likely to present DSSEP improvement upon exion.

Conclusions
In conclusion, our data indicate that the extended and exed neck positions could relieve CSM patients' electrophysiological de cits in some cases. Statistically, we found that involved-segment number ≤ 2 and straight or sigmoid cervical alignment are signi cant in predicting the improvement of DSSEPs upon extension, while the

Consent for publication
Each patient provided informed consent for this publication.

Availability of data and materials
The datasets generated during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests. Authors' contributions ZY and JC were involved in analysis and interpretation of data, as well as in drafting and revising the manuscript.
XC revised it critically for important intellectual content. XP and XZ made substantial contributions to conception and design of the study, and gave nal approval of the version to be published. ZZ, DX, and YC were involved in acquisition of data. All authors read and approved the nal manuscript. Cervical alignment types (modi ed Toyama method [17]): A line connecting the mid-points of the inferior margin of C2 and the superior margin of C7 was constructed. Lordotic group: all centroids are anterior to the line and the distance between at least one centroid and the line is 2 mm or more; Straight group: the distance between the line and each centroid is less than 2 mm; Sigmoid group: some centroids are anterior to and some posterior to the line and the distance between the line and at least one centroid is 2 mm or more; Kyphotic group: all the centroids are posterior to the line and the distance between at least one centroid and the line is 2 mm or more. Figure 3 Dynamic MR images of a patient with deteriorated DSSEP at extension and unchanged DSSEP at exion. From left to right were cervical exion, neutral and extension positions. Upon a neutral position, the patient had lordotic cervical alignment, as well as protruded C5/6 and herniated C3/4 and C4/5 segments. The Mühle stenosis grade of this patient was Grade 2. Upon exion, the spinal cord compression did not change signi cantly. Upon extension, all three segments narrowed signi cantly and thus exacerbated the spinal cord compression.