Association of Cauda Equina Compression on Magnetic Resonance Images and Clinical Signs in Dogs With Degenerative Lumbosacral Stenosis

Introduction

The lumbosacral canal is bordered dorsally by the vertebral lamina, the articular facets of the lumbosacral joints, and the ligamentum flavum; laterally by the pedicles; and ventrally by the dorsal longitudinal ligament, the annulus fibrosus, and the vertebral bodies.¹ The cauda equina lies within the lumbosacral canal and is composed of the seventh lumbar (L₇), the sacral, and the caudal nerve roots.² Disease of these roots causes deficits of sciatic, pudendal, pelvic, perineal, and caudal rectal nerve function.²³ Clinical signs include lumbosacral pain, paresis, lameness, urinary and fecal incontinence, and dysesthesias (abnormalities of skin sensation).^4–9

Disease of the cauda equina may be due to idiopathic stenosis, discospondylitis, trauma, neoplasia, inflammatory disease, vascular compromise, and congenital abnormalities.^13410–15 Degenerative lumbosacral stenosis (DLSS) is reportedly the most common disease of the cauda equina in large-breed dogs.⁵⁶¹⁶ It is characterized by lumbosacral intervertebral disk protrusion, subluxation of the facet joints, thickening of the joint capsule of the articular facets, and hypertrophy of the ligamentum flavum.¹⁶ Lumbosacral transitional vertebrae and osteochondrosis of the sacral endplate may predispose animals to DLSS.¹⁷¹⁸ Dogs predisposed to DLSS include males, the German shepherd dog breed, and those that are highly active or working.^5–9

Various diagnostic investigations have been used to evaluate the lumbosacral spine including radiography, myelography, epidurography, diskography, vertebral venous venography, electromyography, and computed tomography.^1119–30 Magnetic resonance imaging (MRI) is an accurate means of evaluating the lumbar spine in both dogs and humans.^31–34 The purposes of this paper were to report the results of clinical and MRI examinations in a series of dogs with DLSS and to test the hypothesis that severity of clinical signs is directly related to the severity of cauda equina compression.

Materials and Methods

The medical records of 46 client-owned dogs presented to the Veterinary Hospital of the University of Pennsylvania between 1997 and 2000, that had received an MRI examination of the lumbosacral area, were reviewed. Excluded from the study were animals with clinical or imaging findings inconsistent with disease of the cauda equina; those with neoplastic, infectious, vascular, or traumatic disease (n=3); those having undergone an incomplete MRI examination (n=8); or those whose MRI examination of the lumbosacral area was of nondiagnostic quality (n=8). These exclusion criteria eliminated 19 animals from the study, leaving 27 animals with a diagnosis of DLSS. Breed, age, sex, time from onset of clinical signs to presentation, progression of clinical signs, results of other diagnostic evaluations, and results of clinical and neurological examinations were reviewed.

For the MRI study, dogs were given general anesthesia and placed in dorsal recumbency in a frog-legged position. Imaging was performed using a 1.5 Tesla superconducting magnet.^a T1-weighted transverse images were acquired with a repetition time (TR) between 350 and 700 milliseconds and an echo time (TE) between 8 and 10 milliseconds. T2-weighted sagittal and transverse images were acquired with a TR between 3,000 and 6,000 milliseconds and a TE between 88 and 126 milliseconds. Matrix sizes of 256 × 128 and 256 × 192 were used. Slice thickness varied from 2 to 4 mm, with 3-mm slices being used in the majority of cases; 0, 0.5 mm, or 1 mm was skipped between slices. No fat suppression was used in most cases. In all but one case, a quadrature spinal receiver surface coil was used. For the only small dog in the study, an extremity (i.e., birdcage) coil was used. In 10 cases, the plane of section for the transverse images was angled parallel to the long axis of the intervertebral disk space; in the remainder, it was angled perpendicular to the long axis of the spine. The caudal thoracic, entire lumbar, and sacral spine areas were included on the sagittal images in most cases.

Dogs were placed into one of three categories based on their clinical signs [Table 1].⁹³⁵ Two measurements were made on MRI images to quantify cauda equina compression. On T1-weighted sagittal images, the distance from dorsal to ventral margins of the epidural fat at the site of maximal lumbosacral compression and at the midbody of the sixth lumbar (L₆) vertebra were measured. The lumbosacral measurement was divided by that at the midbody of the L₆ vertebra to derive a sagittal compression ratio (SR) that attempted to account for variations in dog breed and size [Figure 1]. A lower value indicated a greater amount of compression. On transverse T1- or T2-weighted images (dependent on image quality), the cross-sectional area of the lumbosacral canal (at point of maximal compression) was measured from digitized images on a computer screen using a computer software package^b [Figure 2]. All images had calibration marks, eliminating any magnification errors in the transfer of images into the digital format. Area measurements were made using the following landmarks: dorsally and ventrally, the margin of the epidural fat; laterally, along a line drawn from the cranial edge of the articular facet parallel to the long axis of the spine. The transverse cross-sectional area (mm²) at the midbody of the L₆ vertebra was similarly measured by tracing the margins of the epidural fat. The lumbosacral canal area was divided by the area at the midbody of the L₆ vertebra to give a second cross-sectional ratio (CSR) of compression, calculated from transverse images [Figure 2]. Compression at other lumbar or thoracic spinal levels was also recorded.

For comparison, four normal dogs had MRI examinations and had compression ratios calculated.

Analysis of variance was used to compare each clinical classification with both compression ratios. To test for linearity in the clinical classification system, a linear contrast was used. A Student’s t-test was used to evaluate association between severity of compression and the presence of degenerate lumbosacral disks on T2-weighted sagittal MRI images, decreased magnitude of sciatic reflexes unilaterally or bilaterally, urinary or fecal incontinence, and presence of spondylosis deformans. All statistical analyses were run based on the entire population of DLSS affected dogs (n=27) as well as on the subpopulation of dogs showing exclusively lumbosacral compression on MRI, with no other compressive lesions evident (n=14). A probability of P<0.05 was considered statistically significant.

Results

Twenty-seven dogs met the inclusion criteria for the study. Affected dogs included Labrador retrievers (n=6), large mixed-breed dogs (n=6), golden retrievers (n=3), German shepherd dogs (n=2), and one each of 10 other breeds. Fifteen males and 12 females were in the group (male/female ratio, 5:4). Mean weight was 35 kg, with a range of 11.6 to 68 kg. Only one dog weighed <24 kg. Mean age at presentation was 6.7 years, with a range of 1.4 to 14.5 years [Table 2]. Mean time of onset to presentation at the hospital was 4.9 months, with a range of 0 to 24 months. The nature of disease progression varied. Sixteen of 22 dogs, in which progression was recorded, had a slow progression of signs. In three dogs, signs were episodic; in two other dogs, signs were acute in onset and rapidly progressive; and in one dog, signs were said to be “stable.” The most common clinical signs were lumbosacral pain, lameness, and paresis [Table 3]. Almost half of the dogs had some degree of neurological deficits [Table 3].

Electromyography (EMG) was performed in five dogs; in four cases, EMG was consistent with denervation of muscles innervated by the sciatic and pudendal nerves. In one case, the EMG was normal. Eight of 16 dogs that had lateral lumbar radiographs had evidence of spondylosis at the lumbosacral junction. Six of 17 dogs that had ventrodorsal radiographs of the pelvis taken had radiographic evidence of hip dysplasia. Three dogs had vertebral malformations: malalignment of the L₆ to L₇ intervertebral junction in case nos. 17 and 22, and sacralization of the L₇ vertebra in case no. 1. Myelography was performed in two dogs, and in both cases contrast agent in the thecal sac extended to the cranial margin of the sacrum. Case no. 16 had incomplete filling over the lumbosacral space (no flexion/extension views), and in case no. 17 (despite good contrast filling), no abnormalities were seen even with flexion-extension views. The most common findings on MRI were ventral compression, nerve root compression, intervertebral disk degeneration, and facet joint osteophytosis [Table 4]. Three of six cases that had asymmetric nerve root compression visible on axial MRI also had consistent lateralizing neurological signs.

Mean SR of normal dogs was 0.87 (range, 0.72 to 0.97), and mean CSR was 0.81 (range, 0.67 to 1). Mean SRs for the mild, moderate, and severely affected DLSS groups were 0.54 (range, 0.23 to 0.81), 0.5 (range, 0.18 to 0.9), and 0.62 (range, 0.29 to 0.93), respectively. Mean CSRs for mild, moderate, and severely affected dogs were 0.67 (range, 0.29 to 1.27), 0.28 (range, 0.12 to 0.37), and 0.82 (range, 0.7 to 0.97), respectively. In six cases, the CSR could not be accurately measured due to lack of clear lumbosacral canal margins on transverse MRI. The compression ratios of normal dogs and those with DLSS are shown in Table 2 and represented graphically in Figure 3.

Tests for linearity did not reveal a linear relationship between the clinical classification system and either ratio (P=0.53 and P=0.31 for SR and CSR, respectively). Student’s t-tests showed no significant association between severity of compression (in either ratio) and presence of spondylosis deformans (P=0.3 and P=0.7 for SR and CSR, respectively), presence of urinary and fecal incontinence (P=0.4 and P=0.1, respectively), decrease in magnitude of sciatic reflexes (P=0.9 and P=0.3, respectively), and the presence of a degenerative lumbosacral disk on sagittal T2-weighted MRI (P=0.06 and P=0.6, respectively). Association between presence of a degenerate lumbosacral disk and SR showed a trend toward significance at P=0.06.

In addition to cauda equina compression, there was evidence of spinal cord compression in 13 (48%) cases. Seven dogs had compression at the L₆ to L₇ intervertebral space, and five cases had compression at the fifth lumbar (L₅) to L₆ intervertebral space; one dog had compression at the fourth lumbar (L₄) to L₅ intervertebral space. In four of these cases, there was also evidence of at least one compressive lesion between the 11th thoracic (T₁₁) to the 12th thoracic (T₁₂) intervertebral space and the third lumbar (L₃) to L₄ intervertebral space.

Tests for linearity also showed no significant association between either SR (P=0.8) or CSR (P=0.9) and the clinical severity group when the population of dogs with an exclusively lumbosacral lesion was analyzed (n=14). Of these dogs, nine were in the mild clinical severity category, four were moderately affected, and one was in the severe group. A significant difference was shown between dogs with a degenerate lumbosacral disk and those without, using the SR (P=0.03).

Discussion

The most important finding of the present study is the lack of correlation between severity of clinical signs and the severity of compression as determined by MRI [Figure 3]. In some cases, urinary or fecal incontinence, or both, was present in dogs with minimal compression. One dog (case no. 3) presented with urinary and fecal incontinence, hind-limb paresis, and severe caudal lumbar pain. The transverse T1-weighted MRI at the lumbosacral space demonstrated almost no compression of either the spinal canal or the nerve roots in the intervertebral foramina. Conversely, some dogs exhibited pain without neurological deficits and had a lumbosacral canal that was severely compressed. These findings conflict with an experimental study in dogs that demonstrated progressive and predictable neurological deterioration in direct proportion to increasing degrees of mechanically induced cauda equina compression.³⁶ In the experimental study, compression of the lumbosacral space of ≤25% caused no neurological deficits. Fifty percent compression caused paresis, and 75% compression caused urinary and fecal incontinence.³⁶ The reasons that clinical signs of naturally occurring DLSS do not correlate well with cauda equina compression (as shown by MRI in this study) are unknown. The authors hypothesize that clinical signs arise only partly from compression of nerve roots or only arise due to compression of nerve roots in certain dogs. The effect of duration of clinical signs and degree of compression is not known. If dogs with DLSS were examined earlier in their clinical course, the relationship between degree of compression and severity of clinical signs may have been different. Substantial epidural fat is present in the lumbosacral canal, and this may allow for a certain “anatomical reserve” of compression to occur before neurological deficits are seen. This may account for those dogs that appear to have severe compression with no neurological deficit and may explain why urinary and fecal incontinence is uncommon in dogs with DLSS. Furthermore, compressive lesions in asymptomatic patients have been described in both dogs and humans.¹¹³⁴ An exception may be at the intervertebral foramen where the L₇ nerve root exits through a relatively narrow channel, where minimal compression may cause severe neurological deficit and pain.

The fact that there are dogs that demonstrate severe signs of cauda equina dysfunction despite minimal compression suggests that there may be other pathophysiological processes occurring. Cauda equina neuritis is a recognized condition in horses.³⁷ The etiology remains unclear, but it is characterized by nonsuppurative inflammation of the mainly extradural components of the cauda equina.³⁷ It has only been described in two dogs that presented with signs similar to DLSS.¹⁵ Both dogs had normal survey spinal radiographs and myelograms. Both were euthanized and found to have marked interstitial and perivascular infiltration with mononuclear cells. Axonal degeneration was the predominant finding, although one also had evidence of demyelination. Both dogs had EMG evidence of denervation as well as decreased sciatic nerve conduction velocities. Cauda equina neuritis could represent an underlying etiology in dogs that do not have cauda equina compression but do have signs of lumbosacral disease.

Age, weight, and gender distributions in this study were similar to other reports of DLSS.^5–79 The breed distribution was different from other studies, possibly reflecting the distribution of breed in the local area.⁹ Lumbosacral pain, pelvic-limb lameness, and paresis were the most common clinical signs in this study. In agreement with other reports, half (n=13, 48%) of the dogs had some form of neurological deficit (e.g., paresis, reflex abnormalities, incontinence).⁵⁶ Asymmetric lesions were suggested in over half of the cases by physical examination. Other authors have made other observations.^9313338 Diminished sciatic nerve reflexes, caused by compression of the lower motor neurons to the pelvic limbs, were not consistent findings. It has been suggested that urinary incontinence precedes fecal incontinence, although this was not the case in this study where fecal incontinence was more common.¹⁴ Self-mutilation, although present in other reported cases of cauda equina disease, was not noted in any dogs in this study.⁵⁶

Dorsal compression was evident in eight of 27 (30%) cases on sagittal MRI. This finding was not noted by other authors.³³ Seven of the eight cases with dorsal compression also had evidence of facet joint osteoarthritis on transverse images. It is suggested that joint capsule thickening, osteophyte formation, and hypertrophy of the ligamentum flavum cause dorsal compression, although which specific structures were causing compression could not be defined. It was not possible to directly visualize the ligamentum flavum on MRI in this study, consistent with other reports.³³ Ventral compression was present in almost all dogs and was caused by intervertebral disk protrusion in all cases. Nerve root compression was present in 17 of 25 (68%) cases and was asymmetric in six of those cases.

Disk degeneration is characterized by dehydration of mainly the nucleus pulposus and annulus fibrosus and results in a decreased signal intensity on T2-weighted MRI.³³ Though DLSS has been defined as a Hansen type II fibroid disk degeneration leading to disk protrusion and compression of neural structures, in 11 of 27 (41%) cases in the authors’ study the lumbosacral disk had normal signal intensity on T2-weighted sagittal images.¹⁶ It has been reported in humans that only about half of the extruded or protruded disks appear degenerate.³⁴ Either disk degeneration is not a prerequisite for disk protrusion or extrusion, or perhaps T2-weighted images are not sufficiently sensitive to detect the earlier stages of degeneration.³⁴

All diagnostic techniques for imaging the lumbosacral spine have inherent weaknesses, and MRI is no exception. Positioning may result in differences in the degree of infolding of a hypertrophied ligamentum flavum.⁵¹⁶³³ Compression caused by thickening or infolding of the ligamentum flavum during hyperextension may have been missed on MRI studies performed with dogs in a relatively neutral (i.e., frog-legged) position. Studies have quantified influence of positioning by flexion/extension radiography, myelography, and computed tomography in patients with compressive lesions.²¹²³³⁹ Many dogs and humans have more severe compression at the lumbosacral space in extension.²³³⁹ This is supported by the observation that pain is often relieved in humans by leaning forward and thereby flexing the spine, and similarly by the observation that dogs with DLSS will often stand with their backs flexed in an attempt to relieve pain.³⁹⁴⁰ However, in both normal dogs and humans, significant changes in spinal canal mensuration between flexion and extension and between compression and distraction occur.^21223941 Thus, static studies may underestimate pathological compression, and flexion/extension studies could overestimate the significance of some normal findings.²³

Lumbosacral instability has commonly been quoted as a possible factor in the etiology of DLSS based on the results of flexion/extension studies.²¹²²⁴¹ No imaging modality can assess the functional weight-bearing relationships of anatomical structures that could result in significant dynamic compression. Spondylosis deformans, lumbosacral subluxation, and the lumbosacral angle have anecdotally been implicated as radiographic evidence of instability. However, none has been shown to have any diagnostic specificity for DLSS.⁴¹ In this study, dogs with lumbosacral spondylosis deformans did not have significantly more cauda equina compression than those without. Studies of relative lumbosacral flexion and extension in dogs with DLSS have shown conflicting results.²¹²² Further work remains to be done to quantify the significance of instability in the etiology of DLSS.

There are limitations to the present study. Small sample size may have introduced type II errors (no difference found between groups, when in fact there was one) into all statistical analyses. Over half of the cases presented were in the “mild” clinical category. With smaller numbers in the more severe categories, there was less chance of finding a statistical correlation. The effect of duration of clinical signs and degree of compression is not known. It may be that if dogs were examined earlier in their clinical course, their degree of neural structure compression or the severity of their clinical signs may have been different. The effect of other compressive lesions elsewhere in the spine cannot be accurately quantified. In all cases, subjective assessment of the MRI lesions, in conjunction with knowledge of the neurological examination, suggested that the other compressive lesions were not responsible for the clinical signs. It should be noted that disk compression at the L₅ to L₆ and the L₆ to L₇ intervertebral spaces may also result in signs of cauda equina disease. However, when dogs with exclusively lumbosacral compression were considered, there was also no significant correlation found between degree of compression and severity of clinical signs. Necropsy was not performed in any case; therefore, compression-induced changes in the nerve roots were not evaluated. The index of compression that was used may not account for variation in the relationship between the dimensions of the vertebral canal at the level of the L₆ vertebra and the lumbosacral junction. Variation in vertebral canal dimensions was limited by the use of standardized positioning. The use of different imaging angles for axial lumbosacral images could lead to changes in the dimensions of the spinal canal. The calculations represented the most sensitive way of quantifying compression available and, therefore, have greater validity than any other method.

Lack of an obvious relationship between clinical findings and compression at the lumbosacral space has also been reported in humans.³⁵ Furthermore, there appears to be little correlation between the severity of MRI compression and postoperative outcome in both dogs and humans.⁴²⁴³ It should be emphasized, however, that MRI is invaluable in accurately identifying the underlying disease process as well as giving the surgeon in-depth preoperative knowledge of both site and extent of the lesion.

Conclusion

The degree of cauda equina compression was well visualized by MRI, but there was poor correlation between the severity of clinical signs and severity of compression. This can be only partially explained by the large anatomical reserve capacity of the lumbosacral vertebral canal, as several dogs in the study exhibited severe clinical signs despite minimal compression. It is proposed that other, as yet unidentified, pathophysiological mechanisms may be involved. It is further proposed that the diagnosis of DLSS or the decision to carry out decompressive surgery should not be made without reference to the clinical signs.

Acknowledgment

The authors thank Dr. Deena Tiches for her contribution of cases to this study.