Clinical Patterns of Cervical Spondylotic Myelopathy

Clinical myelopathy typically appears in late adulthood in the setting of progressive degenerative changes, including cervical disc degeneration, osteophytic spur and transverse bar formation, posterior longitudinal ligament calcification, ligamentum flavum thickening, and osteoarthritic facet hypertrophy.^1–3 Progressive encroachment on the spinal canal by ventral and dorsal anatomic structures may first lead to spinal cord compression—compression that occurs only transiently during physiologic cervical range of motion. The appearance of clinical signs and symptoms arising from this condition has been described as “dynamic stenosis.” With progressive narrowing of the spinal canal, dynamic compression may eventually evolve into static compression of the enclosed spinal cord and the appearance of classic CSM.

Retrospective observational studies indicate that development of CSM is more common in patients with underlying congenital stenosis of the spinal canal. A sagittal spinal canal diameter of less than 12 mm is strongly associated with signs and symptoms of myelopathy, whereas a diameter greater than 16 mm confers a low risk.^4–8

Histopathology of Cervical Spondylotic Myelopathy

The theory that ischemic injury is the pathophysiologic basis of CSM originates in early histologic studies of cervical myelopathy, which revealed several changes consistent with ischemic tissue damage. These include cystic cavitation, gliosis, anterior horn cell dropout, and prominent involvement of the central gray matter, as well as wallerian degeneration of the posterior columns and corticospinal tracts.^2,9–11 In these studies, the most severe histologic changes were observed at the level of ventral spondylotic bars, with the most visible histologic changes occurring in the lateral funiculi of the spinal cord, particularly the corticospinal tracts. The anterior columns and dorsal region of the dorsal columns appeared to demonstrate the least extent of injury-related change.

Attempts have been made to correlate the severity of histopathologic findings with the range of clinical findings in patients with CSM. In general, less severe myelopathy has been associated with changes confined largely to the lateral funiculi, whereas more severe cases appear to be associated with involvement of the medial gray area and ventral aspect of the dorsal columns, as well as gliosis and anterior horn cell dropout. In cases of severe CSM there is extensive wallerian degeneration, proceeding proximally and distally from the site of spinal cord compression.

Spinal Cord Ischemia and Cervical Spondylotic Myelopathy

The anatomic basis for the ischemic insult proposed in CSM has been attributed to various mechanisms, including compression of radicular feeders in the neuroforamina, compromise of venous drainage by ventral spondylotic bars, and compression of the anterior spinal artery, as well as its ventral branches.^12,13 Several animal studies support the concept of a potential role for compressive ischemia in the pathogenesis of CSM.^14–16

Cadaver studies have demonstrated that flattening of the cervical spinal cord is associated with elongation of the laterally directed terminal branches of the central arteries arising from the anterior spinal artery, as well as elongation of the penetrating branches of the lateral pial plexus (corona radiata). It is hypothesized that attenuation of these transversely directed arteries results in decreased arterial blood flow to the corticospinal tracts. Shortening of the ventral-dorsal dimension of the spinal cord, however, results in widening of the arteries directed in the ventral-dorsal direction and relative preservation of blood flow to the anterior columns. These findings might explain the relative vulnerability to injury of the laterally positioned corticospinal tracts, compared with the anterior columns.¹⁷

Recent clinical studies strongly suggest that compression and ischemia alone do not fully explain the pathogenesis of CSM. Despite observational studies associating CSM with various anatomic factors, such as the presence of decreased ventral-dorsal spinal canal diameter, subluxation, and dorsal osteophytes, at least one study has demonstrated that these factors hold no significant predictive value in terms of identifying which patients are at risk for clinical progression of their myelopathy.¹⁸ Several other studies have also failed to identify an association between the degree of spinal stenosis and spinal cord compression and clinical prognosis.^7,12,19

Moreover, surgical decompression that results in expansion of the spinal canal and relief of compressive pressures does not consistently alter the natural history of CSM.²⁰ Ebersold et al.²¹ performed a retrospective review of 100 patients with CSM undergoing surgical decompression, with an average 7-year follow-up, and concluded that decompression alone resulted in no clear, long-term improvement. Two thirds of patients experienced initial clinical improvement, but half of these demonstrated subsequent clinical deterioration. At final follow-up, only a third of the original group were improved, leading the authors to conclude that long-term outcome was not predicated on the presence or severity of spinal cord compression and ischemia, but on other, “nonvascular” factors.

Biomechanical Factors and Cervical Spondylotic Myelopathy

There is a growing body of evidence indicating that abnormal or excessive motion of the cervical spine is strongly associated with clinical progression of CSM. In a retrospective clinical review, Adams and Logue¹² demonstrated a cervical flexion-extension arc in excess of 40 degrees was the most significant variable in predicting poor clinical outcome in patients with CSM. Similar retrospective studies have been performed by Barnes and Saunders,¹⁸ as well as by Yonenobu et al.,¹⁹ in which patients with a flexion-extension arc of greater than 60 degrees after laminectomy were at increased risk for development of progressive myelopathy.

In contrast to the relatively poor results after simple decompression for CSM, several studies demonstrate excellent clinical results associated with the elimination of abnormal cervical motion. Using a simple neck brace to restrict cervical motion often leads to improvement in patients with cervical myelopathy from disc protrusions.²² The largest series of patients undergoing ventral decompression and fusion for CSM demonstrated an 86% improvement rate, with no significant deterioration.²³ Most recently, Uchida et al.²⁴ discovered that among patients with CSM who had kyphotic deformity in excess of 10 degrees, correction of sagittal alignment of the vertebrae significantly improved neurologic outcomes. Uchida et al. state that “ kyphotic alignment may contribute to cervical myelopathy,” that longitudinal distraction is a factor in progressive spinal cord dysfunction, and that the pathophysiologic mechanism is similar to that of tethered cord syndrome.²⁴ Overall, surgical fusion through a variety of approaches has been associated with favorable clinical results, including ventral decompression and fusion without instrumentation²¹ or with ventral plating, ^25–29 and dorsal decompression with instrumented fusion.^30–33

The significant clinical recovery experienced by most myelopathic patients after decompression and fusion indicates that neurologic deficits resulting from cervical myelopathy are recoverable.^{23,25,26,29–31} Moreover, the rapid improvement experienced by many patients after surgery suggests that these patients do not have irreversible, ischemic histologic changes demonstrated in many early pathologic studies. In contrast, failure of some patients to improve clinically after decompression and fusion may be a result of irreversible spinal cord injury. Histologic examination of spinal cord tissue from these patients may reveal severe ischemic injury.²

Pathophysiology of Deformative Stress Injury of the Cervical Spinal Cord

The significance of spinal stenosis and spinal cord compression in early CSM may not be the generation of local ischemia, but rather the creation of a tethering effect, which results in production of local, potentially injurious, tissue strain and shear forces. The concept that increased cervical mobility, coupled with kyphotic deformity, results in spinal cord elongation and increased axial strain forces is well documented.^{12,13,17,18,24,34–41} Several studies have demonstrated the adverse effects of even low-grade mechanical stretching on neural tissues. During normal motion, large axial strains occur in the cervical spinal cord.⁴² The white matter of the spinal cord can be viewed as an axial array of parallel fibers, with individual fibers demonstrating variable levels of crimping. As a whole the cord is initially compliant to stretch, but it becomes progressively stiffer as the fibers straighten and begin to bear tensile load.³⁵ Rapid occurrence of these strains can exceed the material properties of the tissue, leading to tissue disruption and transient or permanent neurologic injury. The degree of injury appears to be related to the peak strain of the tissue and the loading rate.⁴³

Cadaver studies suggest that even physiologic flexion of the cervical spine leads to stretching and the production of strain forces in the neuraxis.¹⁷ Flexion of the spinal column has been found to result in significant elongation of the spinal canal, with concomitant stretching of the spinal cord. During physiologic flexion of the head and trunk in rhesus monkeys, net movement of the spinal cord occurs from the upper spine downward to the level of C4-5, whereas net movement of the spinal cord occurs upward below this level.³⁴ Net movement occurs to a greater extent below C4-5, with 1.6 mm of movement at C1 and 6 mm of movement at T3. The amount of spinal cord stretch occurring at each level is proportional to the degree of flexion at the adjacent intervertebral disc space. Thus, forces that are generated in the spinal cord upon flexion can be visualized with neutral and flexion MRIs of the cervical spine. Flexion of the neck results in significant elongation of the enclosed spinal cord (Fig. 10-1). The increase in length (l) over the original length of the same section of the spinal cord (l_o) provides the strain (ε), thus:

FIGURE 10-1 Strain within the cord on regular flexion. A, The red line represents a hypothetical white matter tract measured from the base of the C7 level to the pontomedullary line. B, The same tract is shown in flexion. The indicated portion of the tract increases in length from 94 mm to 116 mm, representing a strain ε of approximately 0.24.

At the lower cervical and upper thoracic spine, where the amount of flexion tends to be greatest, local spinal cord strain can reach 24%. Thus, the strain produced at the cervicothoracic junction can exceed 0.2, the strain level at which the giant squid axon ceases to function.⁴³ This phenomenon might explain the clinical observation that signs are often localized to levels apparently remote from the level of stenosis (e.g., hand intrinsic muscle wasting with high cervical stenosis).

In the absence of a compressive pathologic process, the natural elongation of the spinal cord that occurs with neck flexion and hyperextension is distributed over the entire length of the spinal cord. However, with tethering of the spinal cord, as a result of local compression, the axial strain cannot be distributed throughout the cord and is instead limited to the segment of cord between the distracting force and the tethering point. Local spinal cord degenerative changes are frequently identified adjacent to thickened dentate ligaments, which suggests that localization of injurious mechanical forces at these levels may be associated with the tethering effect of the ligaments.^36,44 A biomechanical study of the material properties of the dura mater indicates that elastic behavior is uniform throughout the length of the spinal canal; however, strain forces are significantly greater in the cervical region than in either the thoracic or lumbar region.⁴⁵

The tethering action of the dentate ligaments may be responsible for accentuating the effect of tensile spinal cord stress and exacerbating local tissue injury. Moreover, it has been suggested that dorsal displacement of the spinal cord, as a result of the presence of ventral spondylotic bars, may lead to stretching of the dentate ligaments and tethering of the cervical cord through the ventrolaterally positioned nerve root sleeves. Repetitive and persistent microtrauma to these nerve root sleeves may lead to the progressive thickening that has been observed with age.⁴⁴ Therefore, axial tension generated in the spinal cord during physiologic motion may be amplified at certain levels, as a result of two separate factors—overall spinal canal lengthening and the local tethering effects of the dentate ligaments.

Several investigators have attributed delayed, progressive cervical myelopathy to a combination of underlying structural kyphosis and abnormal or excessive cervical motion.^12,13,24,38 Dynamic lengthening of the cervical spinal cord that occurs during neck flexion is magnified in patients with cervical kyphosis. Conversely, kinematic MRI studies have demonstrated that lengthening of the spinal cord also occurs during neck extension in some patients with fixed kyphotic deformity of the cervical spine. In the setting of static spinal cord compression and superimposed instability, cervical extension can also lead to aggravation of the cord impingement and significant upper cervical spinal cord elongation.⁴⁶

Mathematical Models of Spinal Cord Stretch Injury

Numerous mathematical models for spinal cord stretch injury have been developed. Levine³⁶ represented the spinal cord as a simplified solid material with uniform elastic properties to predict the three-dimensional stresses experienced during physiologic motion and in spondylosis. According to this model, flattening of the cord is not a result of ventral-dorsal compression, but rather the consequence of laterally directed tension arising from the dentate ligaments, which tighten in flexion. This model, with a ventral spondylotic bar and tethering dentate ligaments, predicts maximal stresses in the lateral funiculi. The model provides a possible explanation for the characteristic histologic findings in CSM, in which there is relative sparing of the anterior and posterior funiculi. It also explains why histopathologic changes are found over a relatively extended segment of spinal cord tissue, as opposed to being limited to the point of compression. However, the importance of the dentate ligaments in the etiology of CSM is brought into question by the inconsistent results of sectioning these ligaments at the time of surgery.⁴⁷

Breig³⁸ also developed a mechanical model to explain some of the apparent inconsistencies found in histologic studies of CSM. For instance, in addressing the question of why some chronic, ventral compression injuries result in predominantly dorsal cord injury, cadaver models demonstrated that a compression force applied ventrally to the spinal cord in the presence of stenosis creates a pincer mechanism, resulting in increased axial tension in the cord and fissuring opposite the side of compression. In this model the spinal cord is represented as a viscoelastic cylinder that, when compressed from the sides, exhibits net tissue creep to the free ends of the cylinder. As a result, tension forces are created perpendicular to the plane of compression. With mild compressive deformation of the spinal cord, elastic stretch of the axis cylinders occurs. However, when the ventral-dorsal diameter of the spinal cord is reduced by 20% to 30%, axial tension forces exceed the material properties of the tissue and result in tearing of axial fibers. The stress field produced by this pincer mechanism is multidirectional, and secondary shearing forces are also created. This model explains how ventral compression of the spinal cord in the presence of stenosis might result in stretch and shear injury to myelin and neural elements.

Finite Element Models of Spinal Cord Stretch Injury

More recently, researchers have produced mathematical models of the cord using finite element analysis, a method adapted from materials science and fluid mechanics. Finite element analysis reduces a continuous structure into discrete, finite “brick” elements. This allows the approximation of partial differential equations by a linear system of ordinary differential equations, which can then be solved by numerical methods with the appropriate boundary conditions.⁴⁸ In this particular case, the equations concern mechanical strain (stretch), “out of plane” loading (shear due to transverse compression, such as from a retroflexed odontoid process), and material properties such as Young’s modulus of elasticity or Poisson’s ratio. Ichihara et al.⁴⁰ used finite element analysis to simulate the cervical spinal cord under compression and showed different amounts of stress at a given strain rate were to be expected owing to the differing material properties of gray and white matter. Kato et al.⁴¹ showed that the addition of a small amount of flexion to a model with static compression significantly increased predicted stresses, with the majority of stresses in the anterior and posterior horns. Henderson et al.⁴⁹ demonstrated that increased deformative stresses in the corticospinal tracts, as predicted by the finite element analysis, were strongly correlated with neurologic deficits in a cohort of children with cervical and medullary symptoms. Elevated stress levels due to strain occurred during normal neck flexion in the spinal cord at the C1 level of one patient (MRIs from this patient are shown in Fig. 10-1); the addition of compression (shear) from a retroflexed odontoid process generates much higher stress levels with the same degree of flexion (Fig. 10-2).

FIGURE 10-2 Finite element analysis of a portion of the cervical spine of the patient whose MRIs are shown in Figure 10-1. A, Sagittal view demonstrating the stresses on flexion. B, Sagittal view demonstrating more severe stresses on addition of local compression due to retroflexed odontoid with same degree of flexion as in A. C, Axial view at C1 of A. D, Axial view at C1 of B.

Spinal Cord Tethering and Shear Injury

Studies involving the tethered spinal cord syndrome may also contribute to a better understanding of the pathogenesis of CSM. Stretch injury is now widely accepted as the principal cause of myelopathy in tethered cord syndrome. The symptoms and clinical findings of pain, numbness, weakness, pes cavus, scoliosis, and bowel and bladder dysfunction have all been attributed to stretching injury of the spinal cord.^50–56 The degree or amount of traction on the conus medullaris determines the age of onset of symptoms. Extensive tethering and severe stretching of the conus medullaris results in neurologic disturbances in infancy, whereas a lesser degree of tethering often remains subclinical until adulthood, when symptoms may become manifest in the setting of an acute event (i.e., hyperflexion injury) or chronic process (e.g., development of ventral disc or bone protrusions).⁵⁷ Although clinical manifestations of tethered cord syndrome are more commonly referable to the lumbosacral spinal cord, many neurologic findings are referable to the cervical cord. For example, long tract involvement in tethered cord syndrome may lead to hand numbness and poor coordination, as well as upper extremity hyper-reflexia and even speech difficulties. Quadriparesis has also been reported.⁵⁸