Predictors of symptomatic myelopathy in degenerative cervical spinal cord compression

Abstract Objectives To update a previously established list of predictors for neurological cervical cord dysfunction in nonmyelopathic degenerative cervical cord compression (NMDCCC). Material and Methods A prospective observational follow‐up study was performed in a cohort of 112 consecutive NMDCCC subjects (55 women and 57 men; median age 59 years, range 40–79 years), either asymptomatic (40 subjects) or presenting with cervical radiculopathy or cervical pain (72 subjects), who had completed a follow‐up of at least 2 years (median duration 3 years). Development of clinical signs of degenerative cervical myelopathy (DCM) as the main outcome was monitored and correlated with a large number of demographic, clinical, electrophysiological, and MRI parameters including diffusion tensor imaging characteristics (DTI) established at entry. Results Clinical evidence of the first signs and symptoms of DCM were found in 15 patients (13.4%). Development of DCM was associated with several parameters, including the clinical (radiculopathy, prolonged gait and run‐time), electrophysiological (SEP, MEP and EMG signs of cervical cord dysfunction), and MRI (anteroposterior diameter of the cervical cord and cervical canal, cross‐sectional area, compression ratio, type of compression, T2 hyperintensity). DTI parameters showed no significant predictive power. Multivariate analysis showed that radiculopathy, cross‐sectional area (CSA) ≤ 70.1 mm2, and compression ratio (CR) ≤ 0.4 were the only independent significant predictors for progression into symptomatic myelopathy. Conclusions In addition to previously described independent predictors of DCM development (radiculopathy and electrophysiological dysfunction of cervical cord), MRI parameters, namely CSA and CR, should also be considered as significant predictors for development of DCM.


| INTRODUCTION
Degenerative cervical cord compression detected by imaging methods, mostly magnetic resonance imaging (MRI), is a prerequisite for the clinical diagnosis of degenerative cervical myelopathy (DCM).
This overarching term is preferred to describe the various degenerative conditions of the cervical spine that cause myelopathy, including most frequent cervical spondylotic myelopathy, but also degenerative disc disease and ossification of the posterior longitudinal ligament and of the ligamentum flavum (Nouri, Tetreault, Singh, Karadimas, & Fehlings, 2015). There is a considerable body of current research related to various aspects of DCM, including prognostic factors (Tetreault, Karpova, & Fehlings, 2015;Tetreault, Nouri, Singh, Fawcett, & Fehlings, 2014). In recent years, studies have demonstrated that asymptomatic degenerative cervical cord compression detected on MRI (Boden et al., 1990;Matsumoto et al., 1998;Teresi et al., 1987) may be of a prevalence that exceeds that of symptomatic myelopathy (Bednarik et al., 2004(Bednarik et al., , 2008Bednařík et al., 1998;Kovalova et al., 2016;Wilson et al., 2013). Knowledge of the prevalence, as well as the frequency, of myelopathy development, and of risk factors influencing this progression, however, is sparse (Wilson et al., 2013). Such knowledge would be of crucial importance to the practical management of asymptomatic degenerative cervical cord compression, and bear upon the important issue of indications for preventive surgical decompression.
In other studies, we have established that the presence of symptomatic cervical radiculopathy and central conduction deficit in the cervical cord, disclosed by electrophysiological methodssomatosensory (SEP) and/or motor-evoked potentials (MEP)were independent predictors for the development of symptomatic DCM (Bednarik et al., 2004(Bednarik et al., , 2008Bednařík et al., 1998). These results tally, in part, with those of an international survey undertaken by the spine care community (Wilson et al., 2013) that identified the presence of radiculopathy together with MRI evidence of intramedullar T2 hyperintensity as important factors influencing the decision to perform preventive decompressive surgery in nonmyelopathic patients with degenerative cervical cord compression.
Our previous study (Bednarik et al., 2008), although extensive, had several limitations. Most importantly, the patients, although lacking any clear myelopathic symptoms and/or signs (i.e., "nonmyelopathic"), were in fact not completely asymptomatic, as our cohort was recruited from consecutive patients referred for radiculopathy and/or cervical axial pain. The term "asymptomatic" degenerative cervical cord compression should be reserved for completely asymptomatic cases, while nonmyelopathic subjects with or without signs/ symptoms of radiculopathy or cervical pain should be referred to in terms of "nonmyelopathic degenerative cervical cord compression" (NMDCCC). As one of two alternative criteria for MRI-detected cervical cord compression, we used compression ratio (CR) < 0.4 that might preclude less severe diffuse compression to be included into the study.
Further, spinal cord T2 hyperintensity is considered an important risk factor by the spine care community (Wilson et al., 2013), and diffusion tensor imaging (DTI) parameters have shown the capacity to differentiate cervical myelopathy patients not only from normal individuals (Chen et al., 2016;Guan et al., 2015;Lee et al., 2015) but also from nonmyelopathic cervical cord compression cases (Kerkovsky et al., 2012), and further to correlate with severity of myelopathy (Rajasekaran et al., 2014), the segments of the cervical cord involved (Suetomi et al., 2016), and to predict postsurgical outcome (Arima et al., 2015). A re-evaluation of the predictive model describing the risk of progression of NMDCCC to symptomatic myelopathy (Bednarik et al., 2008) was thus indicated, in a sample also including completely asymptomatic subjects with less severe stages of degenerative cervical cord compression and with the use of DTI parameters to validate the previous model.

| MATERIAL & METHODS
The sample size calculation, about 120 patients, was based on an anticipated frequency of DCM development of about 18% over the course of 3 years (derived from the previous study, Bednarik et al., 2008) and the number of evaluated predictors (20).
The study sample here consisted of a cohort of consecutive subjects who had been referred to the Department of Neurology between January 2012 and December 2013 with clinical signs and symptoms of cervical radiculopathy, moderate-to-severe chronic or intermittent axial cervical pain, and volunteers in whom MRI signs of degenerative cervical cord compression had previously been detected during an epidemiological study focusing on the prevalence of degenerative cervical cord compression in the population of the province of South Moravia (Kovalova et al., 2016). The inclusion of volunteers from the epidemiological study, complying with the criteria for the current study into prospective evaluation, had been planned beforehand.
All subjects in the study had to comply with the following inclusion criteria: • MR signs of degenerative compression of the cervical spinal cord with or without concomitant change in signal intensity from the cervical cord on T2/T1 images (see "Imaging" below) • Absence of any current myelopathic clinical signs and symptoms that could probably be attributed to cervical cord involvement, from the following list. Originally, 137 NMDCCC subjects were included into the prospective evaluation. Twenty-five subjects were lost during follow-up and the follow-up of at least 2 years was completed by a group of 112 subjects (55 women and 57 men; median age 59 years, range 34-79 years): 72 subjects had nonmyelopathic signs or symptoms probably related to degenerative changes of the cervical spine (namely axial pain and/ or symptoms or signs of upper extremity monoradiculopathy), while 40 subjects were completely asymptomatic. The whole study cohort was a completely new sample, and none of these subjects had been included in a previously published prospective study on this topic (Bednarik et al., 2004).
Ethical approval for the study was granted by the Ethical Committee of the University Hospital, Brno.

| Clinical evaluation
A detailed clinical examination was carried out at the beginning of the study and every 6 months thereafter. Patients were instructed about possible signs and symptoms that might indicate newly developed DCM and encouraged to arrange a consultation with a neurologist from the study group if they suspected a progression to myelopathy.
A standardized, timed 10-m walk and run (as quickly as possible) was evaluated, in terms of time taken and number of steps required.
The primary end-point of the study was the detection of development of symptomatic DCM based on the occurrence of at least one symptom and one sign (from the list used as exclusionary criteria-see above), which were probably attributed to degenerative cervical cord compression, were not present at the beginning of the follow-up and had no other topical or etiological explanation.
Clinical evaluation focused on the determination of development of symptomatic myelopathy (as primary outcome) was performed by neurology specialists experienced in the diagnosis and practical management of myelopathic cases (ZK, ZKJ, MN) and the final decision on meeting the outcome, that is, development of symptomatic DCM, was approved by ZK, a senior neurologist with a long-term experience in clinical studies on cervical myelopathy.

| Imaging
Plain anteroposterior, oblique, and lateral radiograms were obtained in all patients. Their Torg-Pavlov ratio (TPR) at C5 level was calculated from lateral radiograms as the anteroposterior diameter of the spinal canal divided by the anteroposterior diameter of the vertebral body. All subjects underwent MRI examination of the cervical spine on a 1.5 T MR device with a 16-channel head and neck coil.
The standardized imaging protocol included conventional pulse sequences in sagittal-T1, T2 and short-tau inversion recovery (STIR) and axial planes (gradient-echo T2) for the purpose of morphological evaluation and a DTI sequence in the axial plane coherently covering five segments of the cervical spine from levels C2/3 to C6/7. The DTI scans were acquired at a slice thickness of 4 mm, with the same geometry settings as those employed for the axial T2 images. The clinical status of patients/volunteers was blinded to the neuroradiologists who examined the cervical spine MRIs. The MRI of every subject was evaluated by two neuroradiologists, who agreed on the assessment of the compression in the majority of cases. Where disagreement existed-seldom-the final decision was based on a cooperative decision.
The imaging criterion for cervical cord compression was defined as a change in spinal cord contour or shape at the level of an intervertebral disc on axial or sagittal MRI scan compared to that at midpoint level of neighboring vertebrae.
Spinal cord compression was further graded as: • Impingement, that is focal concave, usually anterior, defect of spinal

| Electrophysiological evaluation
Short-latency SEPs from the median and the tibial nerves were elicited at the beginning of the study by electrical stimulation of mixed nerves at the wrist and the ankle. Similarly, MEPs were elicited by means of transcranial and root magnetic stimulation and recorded from abductor digiti minimi and abductor hallucis muscles on both sides. Details on the methodology of electrophysiological examination and evaluation of results with definition of central conduction abnormality attributable to possible cervical spinal cord lesion are described in previous publications (Bednarik et al., 2004(Bednarik et al., , 2008Bednařík et al., 1998).
Motor and sensory conduction studies were performed on six motor nerves (median, ulnar, and tibial nerves bilaterally) and four sensory (ulnar and sural nerves bilaterally) using conventional techniques.
Needle EMG from four muscles (deltoid, biceps brachii, triceps brachii, and first dorsal interosseous) was performed bilaterally with assessment of spontaneous activity, motor unit potential parameters, and interference patterns. EMG signs of acute motor axonal neuropathy in one myotome (C5-Th1) corresponding with radicular signs and symptoms were classified as radicular. EMG signs of acute, subacute, or chronic motor axonal neuropathy, established in more than one myotome (C5-Th1) unilaterally or bilaterally, were classified as signs of anterior horn cell lesion resulting from degenerative cervical myelopathy.
The following variables were recorded at the entry examination and their association with the predefined end-points (i.e., development of clinically symptomatic DCM and time taken for it) were analyzed.

| Statistical analysis
Standard univariate statistical techniques were used to test differences between the chosen subgroups of patients and association between the parameters examined: Fisher's exact test for binary outcomes (or its extension-Fisher-Freeman-Halton exact test for contingency tables larger than 2 × 2) and Mann-Whitney U test for continuous variables.
The power of parameters to discriminate between NMDCCC subjects who developed symptomatic DCM and those who remained asymptomatic was evaluated by receiver operating curve (ROC) Finally, multivariate model-adjusted logistic regression-was used to seek independent predictors for the development of symptomatic DCM. The variables were selected using a forward step-wise selection algorithm.

| RESULTS
Clinical evidence of the first signs and symptoms of DCM within the entire follow-up period was found in 15 patients (13.4%): the DCM+ subgroup. DCM developed in seven cases (6.3%) during the first 12 months of the follow-up period. The frequency of myelopathic symptoms and signs in our cohort are summarized in

| DISCUSSION
This contribution reports the results of a validation study on the pre- In a previous study, with a cohort of 199 NMDCCC individuals followed for 48 months, the authors documented the predictive value of cervical radiculopathy and electrophysiological signs of cervical cord dysfunction detected with SEP and MEP. This cohort, however, included nonmyelopathic but not completely asymptomatic cases, referred to a neurologist for radiculopathy or cervical pain. In this study, 37.5% of nonmyelopathic subjects had the least severe type of compression (type I) and 20.5% the most severe, type IIb.

Gate ataxia 3
Flaccid paresis of upper extremity (plurisegmental) 3 Spastic paresis of lower extremity, spastic gate 2 The main limitation of this study is the low number of outcome events in relation to the high number of potential predictors, which weakened the statistical evaluation. In contrast to radiculopathy, which proved a significant predictor in both the current and the previous study (Bednarik et al., 2008)   EMG signs of myelopathy 7 (6.3%) 3 (20.0%) 4 (4.1%) .049 Abnormal MEP 10 (8.9%) 5 (33.3%) 5 (5.2%) .004 Abnormal SEP 17 (15.2%) 6 (40.0%) 11 (11.3%) .011 Torg-Pavlov ratio 0.9 (0.5; 1.5) 0.9 (0.6; 1.2) 0.9 (0.5; 1.5  The used MRI criterion for cervical cord compression based on subjective evaluation of a spinal cord contour or shape might be considered controversial. In our previous studies on that topic (Bednarik et al., 2004(Bednarik et al., , 2008, we used the presence of impingement (i.e., focal change of contour) and/or CR < 0.4 as MRI criteria for cervical cord compression. However, using these criteria might have prevented less severe circular compressions from inclusion into the study and the compression ratio from showing off its predictive value.
We addressed the issue of an optimal quantitative imaging criterion for cervical cord compression in a recent cross-sectional study of a large cohort of randomly recruited individuals (Kovalova et al., 2016).
We used the same qualitative criterion (a change in spinal cord contour) as a gold standard and validated several quantitative MRI parameters for their sensitivity and specificity to discriminate between nonmyelopathic compression and no compression. An anteroposterior diameter of the cervical spinal canal of <9.9 mm was associated with the highest probability of MRI-detected nonmyelopathic cervical cord compression T A B L E 3 Discrimination power of parameters to distinguish between NMDCCC subjects who developed symptomatic DCM (n = 15) and those that remained asymptomatic (n = 97) T A B L E 4 Predictive power of parameters to distinguish between NMDCCC subjects who developed symptomatic DCM (n = 15) and those that remained asymptomatic (n = 97) using univariate analysis Prolonged gait or run proved to be able to discriminate/predict those patients with higher risk of developing symptomatic myelopathy. Lower statistical power of our study due to low number of outcome events in relation to the high number of potential predictors might be the reason why these functional tests, as well as some other predictors, did not prove to be an independent predictors using multivariate analysis. They are, however, promising and worthy further evaluation.
The degenerative compression is certainly a continuum with increased severity of compression and concomitant dysfunction/impairment of spinal cord. As it is not possible to differentiate reliably between symptomatic and nonmyelopathic cervical cord compression cases exclusively on clinical grounds, this limitation could lead to some confusion in terminology. One might speculate whether patients with MRI signs of cervical cord compression and abnormal conduction across spinal cord tracts proved by SEPs or MEP, those with MRI intramedullar signal changes, or those with prolonged time on quantified walk are really nonmyelopathic. Nevertheless, the current concept of symptomatic DCM is based on the presence of clear clinical symptoms and signs, and those "abnormal" or "subclinical" parameters increasing the risk for development of symptomatic myelopathy might define a subgroup of degenerative compressions that might be labeled as highrisk NMDCCC or "presymptomatic myelopathy." In conclusion, previously and recently identified predictors of DCM development in NMDCCC individuals could help the decision-making process for preventive surgical decompression and, more importantly, in defining a subgroup of NMDCCC individuals at higher risk of DCM, among whom a randomized trial evaluating the benefit of such decompression would be justifiable.
T A B L E 4 (Continued) T A B L E 5 Predictive power of parameters to distinguish between NMDCC subjects who developed symptomatic DCM (n = 15) and those that remained asymptomatic (n = 97): multivariate model based on step-wise analysis of data CR, compression ratio; CSA, cross-sectional spinal cord area; EMG, electromyography; NMDCCC, nonmyelopathic degenerative cervical cord compression. Significant independent predictors are expressed in bold type.