CONTRAST ENHANCEMENT OF EXTRADURAL COMPRESSIVE MATERIAL ON MAGNETIC RESONANCE IMAGING

Gadolinium‐enhancement of compressive extradural material is detected occasionally with magnetic resonance (MR) imaging in dogs. Our goal was to characterize contrast enhancement of extradural compressive material associated with intervertebral disc herniation, and to evaluate the association between enhancement and histopathologic findings and the onset of clinical signs. Ninety‐three dogs with a total of 99 lesions diagnosed as intervertebral disc herniation on MR imaging were assessed. Images were evaluated for lesion location, type of herniation, degree of compression, intramedullary T2‐weighted (T2W) intensities, and contrast enhancement. In 23 dogs, surgically removed compressive material was evaluated histopathologically for hemorrhage, inflammation, neovascularization, fibroplasia, fibrosis, mineralization, necrosis, and chronicity. Contrast enhancement of extradural compressive material, meninges, and both the compressive materials and meninges was present in 51.5%, 39.4%, and 17.2% of lesions, respectively. Extradural enhancement occurred more frequently in extrusions than protrusions (P=0.001). Meningeal enhancement and more severe neurologic deficits were significantly associated with a shorter duration of clinical signs (P=0.04 and 0.01, respectively). Intramedullary T2W hyperintensities, present with 44.4% of lesions, were associated with more severe neurologic deficits (P=0.001). Lesions with extradural enhancement were more often considered subacute to chronic in duration and more frequently associated with hemorrhage compared with nonenhancing material; however, no statistically significant association was established between contrast enhancement and histopathologic findings. Contrast enhancement of extradural compressive material and the meninges was found to be common with intervertebral disc herniation, and should not be interpreted as a specific sign of a mass lesion such as neoplasia.     

Magnetic resonance imaging enhancement of intervertebral disc disease in 30 dogs following chemical fat saturation

Objective : To describe the patterns of enhancement of extradural intervertebral disc on chemically fat saturated gadolinium‐enhanced magnetic resonance images and to investigate the clinical and pathological associations with enhancement. Methods : Medical records and magnetic resonance images were reviewed from 30 dogs with histopathologically confirmed disc disease and enhancement on a T1‐weighted postcontrast fat saturated sequence. Results : Median duration of neurological signs was 4 days and the most common grade of severity was II, seen in 46·6% of dogs. Homogeneous, heterogeneous and peripheral patterns of disc enhancement were described, with peripheral enhancement most commonly identified (57% of dogs). There were no clinical or pathological differences between the dogs with each of the patterns. The mean signal intensity of a region of interest within the extruded disc material and contrast‐to‐noise ratio of the disc material were significantly higher on postcontrast T1‐weighted fat saturated images (P=<0·0001 each). Clinical Significance : The use of fat saturated gadolinium‐enhanced magnetic resonance imaging can detect enhancement of extradural disc material. Patterns of enhancement are not associated with the clinical presentation or pathological features.     

Presence of residual material following mini-hemilaminectomy in dogs with thoracolumbar intervertebral disc extrusion

Presence of residual material following mini-hemilaminectomy in dogs (n = 9) with spontaneous thoracolumbar intervertebral disc extrusion was prospectively investigated. Volume of extruded disc material within the vertebral canal and the proportion of residual material were determined using pre- and post-operative magnetic resonance imaging. The degree of spinal cord compression, proportion of extradural material considered hemorrhage, and invasion of the articular facets were also determined. Residual material was identified in 44% of the mini-hemilaminectomies. The median percentage of residual material calculated was 7.7% of the preoperative volume. No observed effect of surgical side or site, proportion of extradural material considered hemorrhage, patient weight, and duration of clinical signs was detected.     

Thoracolumbar disc extrusion associated with extensive epidural haemorrhage: a retrospective study of 23 dogs

OBJECTIVES: To describe the clinical features and outcome in dogs suffering from thoracolumbar disc extrusion associated with extensive epidural haemorrhage (DEEH) and treated with extensive hemilaminectomy (from three to seven vertebrae). METHODS: The records of 23 dogs with surgically confirmed DEEH were reviewed retrospectively. RESULTS: All cases were characterised by rapid progression to severe neurological dysfunction (grade III, V and VI). Myelography was performed in 21 cases and showed an absence (16 cases) or attenuation (five cases) of contrast medium column along three to seven vertebrae. In two dogs, magnetic resonance imaging was accurate in confirming extradural compression due to disc material and haemorrhage, determining the extent of compression and side of the lesion. All cases were treated surgically with extensive hemilaminectomy involving all the compressed spinal segments. Twenty-one dogs (91 per cent) recovered and regained ambulatory function. Two dogs, without deep pain perception before surgery, did not improve. A two-year follow-up history was available for 15 dogs. Disc extrusion recurred in two dogs (9 per cent), two and 20 months after surgery. CLINICAL SIGNIFICANCE: Extensive hemilaminectomy can adequately decompress the spinal cord after DEEH and may produce a recovery and recurrence rate similar to thoracolumbar disc extrusion not complicated by extensive epidural haemorrhage.     

Cervical vertebral malformation-malarticulation syndrome in the Bernese mountain dog: clinical and magnetic resonance imaging features

O bjectives : To describe the clinical and magnetic resonance imaging features of cervical vertebral malformation-malarticulation in Bernese mountain dogs.
M ethods : Seven Bernese mountain dogs (four males and three females) were diagnosed with cervical vertebral malformation-malarticulation by magnetic resonance imaging. The following data were evaluated retrospectively: (1) abnormalities of the cervical vertebral column and spinal cord, (2) spinal cord compression, (3) intervertebral disc degeneration and herniation, (4) severity of clinical signs pretreatment and after treatment, (5) type of treatment and (6) outcome.
R esults : Spin echo T1-weighted and T2-weighted images disclosed multi-level, extradural compressive spinal cord lesions (ventral, dorsolateral or both) spanning from intervertebral disc spaces C3-4 to C6-7. In all seven dogs, T2-weighted images disclosed one or more intramedullary hyperintensities associated with extradural spinal cord compression. Surgery was performed in five dogs. Two dogs were managed medically. The prognosis for surgical or conservative management in Bernese mountain dogs was similar to cervical vertebral malformation-malarticulation in other breeds.
C linical S ignificance : Cervical vertebral malformation-malarticulation is an important differential diagnosis for young to middle-aged Bernese mountain dogs with a C1-5 or C6-T2 neuroanatomic localisation. Dorsolateral spinal cord compression associated with articular process hypertrophy was the most common feature of cervical vertebral malformation-malarticulation in the seven Bernese mountain dogs evaluated.     

THE COMPUTED TOMOGRAPHIC APPEARANCE OF ACUTE THORACOLUMBAR INTERVERTEBRAL DISC HERNIATIONS IN DOGS

The appearance of herniated intervertebral disc material in the thoracolumbar vertebral canal was evaluated in 23 dogs using computed tomography (CT). The images were then compared with the myelographic and surgical findings. The normal spinal cord, outlined by epidural fat over intervertebral disc spaces, was of intermediate attenuation on transverse CT images. Herniated disc material was identified in all animals as a heterogeneous hyperattenuating extradural mass. The attenuation of the disc material increased with the degree of mineralization. In seven dogs, the herniated material was only slightly more attenuating than the spinal cord. In these dogs, small fragments of mineralized disc material and significant hemorrhage were found in the epidural space at surgery. In dogs with a long standing history of disc herniations, disc material identified in the vertebral canal had a more hyperattenuating and homogeneous appearance than recently herniated disc material. We conclude that mineralized, herniated disc material and hemorrhage can be identified quickly and safely in dogs using CT.     

Magnetic resonance imaging features of spinal epidural empyema in five dogs.

Spinal epidural empyema is defined an accumulation of purulent material in the epidural space of the vertebral canal. Spinal epidural empyema should be considered as a differential diagnosis in dogs with pyrexia, spinal pain, and rapidly progressing myelopathy. Magnetic resonance (MR) imaging is the imaging test of choice in humans. Here, we describe the MR imaging features of five dogs with confirmed spinal epidural empyema. The epidural lesions appeared as high or mixed signal masses in T2-weighted (T2W) images. Increased signal within the spinal cord gray matter at the site of the lesion was detected in T2W images in all dogs. Two patterns of enhancement were detected on postcontrast T1-weighted (T1W) images. Mild to moderate peripheral enhancement was seen in three dogs and a diffuse pattern of enhancement was seen in one. Discospondylitis was identified in three dogs on T1W postcontrast images. Decompressive spinal surgery was performed in all dogs. Bacteria isolated from the abnormal epidural tissue were Enterobacter cloacae, coagulase-positive Staphylococci, Pasteurella multocida, and Escherichia coli. In one dog bacteria were not isolated. These MR imaging features, along with appropriate clinical signs, can allow prompt diagnosis and appropriate treatment planning.     

Ventral intraspinal cysts associated with the intervertebral disc: magnetic resonance imaging observations in seven dogs

Objective— To report clinical and diagnostic imaging features, and outcome after surgical treatment of ventral intraspinal cysts in dogs.
Study Design— Retrospective study.
Animals— Dogs (n=7) with ventral intraspinal cysts.
Methods— Clinical signs, magnetic resonance imaging (MRI) findings and surgical findings of 7 dogs and histologic findings (1 dog) with intraspinal cysts associated with the intervertebral disc were reviewed.
Results— Ventral intraspinal cyst is characterized by: (1) clinical signs indistinguishable from those of typical disc herniation; (2) an extradural, round to oval, mass lesion with low T1 and high T2 signal intensity on MRI, compatible with a liquid-containing cyst; (3) cyst is in close proximity to the intervertebral disc; and (4) MRI signs of disc degeneration. Although the exact cause is unknown, underlying minor disc injury may predispose to cyst formation.
Conclusion— Intraspinal cysts have clinical signs identical to those of disc herniation. Given the close proximity of the cyst to the corresponding disc and the similarity of MRI findings to discal cysts in humans, we propose the term "canine discal cyst" to describe this observation.
Clinical Relevance— Discal cysts should be considered in the differential choices for cystic extradural compressing lesions.     

MAGNETIC RESONANCE IMAGING FEATURES OF EXTRADURAL HEMATOMAS ASSOCIATED WITH INTER VERTEBRAL DISC HERNIATION IN A DOG

Myelography and magnetic resonance imaging (MRI) were performed on a 4-year-old neutered female Rottweiler with bilateral pelvic limb paresis. On the myelogram, there was extradural spinal cord compression at the level of the T11-12 intervertebral disc. Inadvertent placement of epidural contrast medium also allowed identification of a 1-cm circular filling defect in the epidural space dorsal to the compressed spinal cord. MRI showed partial loss of the nucleus pulposus signal of the T11-12 disc, a focal signal void within the vertebral canal at T11 compatible with a free disc fragment, and extradural masses compressing the spinal cord at T10-11 and T11-12. Hemorrhage within the masses was confirmed on T2*-weighted images. A mixture of hematoma and mineralized disc material was found at surgery, and there was no histopathologic evidence of neoplasia. In this article, the appearance on MRI of hemorrhage associated with intervertebral disc herniation is discussed.     

SUSCEPTIBILITY ARTIFACTS ON T2*‐WEIGHTED MAGNETIC RESONANCE IMAGING OF THE CANINE AND FELINE SPINE

The T2*‐weighted gradient recalled echo sequence is a sensitive means to detect blood degradation products. While not a routine sequence in magnetic resonance imaging of the spine in small animals, it can provide additional valuable information in select cases. The goal of this retrospective, cross‐sectional study was to describe findings when acquiring this sequence during magnetic resonance imaging examination of the spine in small animals. The University of Tennessee's veterinary radiology database was searched for dogs and cats that underwent magnetic resonance imaging for suspect spinal disease in which a T2*‐weighted gradient recalled echo sequence was acquired and susceptibility artifact was identified. The following information was recorded: signalment, clinical signs, location and appearance of susceptibility artifact, and final diagnosis. Thirty‐nine cases were included in the study. Extradural susceptibility artifacts were observed in cases of intervertebral disc herniation with or without associated hemorrhage (n = 28), extradural hemorrhage associated with spinal trauma (n = 2), hemophilia (n = 1), and in a cystic extradural mass (n = 1). Remaining lesions displaying susceptibility artifact were intramedullary and included presumptive acute noncompressive nucleus pulposus extrusion (n = 2), hematoma (n = 1), hemangiosarcoma metastasis (n = 1), intramedullary disc extrusion (n = 1), presumptive meningomyelitis (n = 1), and a mass of undetermined etiology (n = 1). Inclusion of a T2*‐weighted gradient recalled echo sequence may be helpful in spinal magnetic resonance imaging when standard imaging sequences are ambiguous or intramedullary lesions are observed.     

MAGNETIC RESONANCE IMAGING CHARACTERISTICS OF ASCENDING HEMORRHAGIC MYELOMALACIA IN A DOG

A 4-year-old neutered female terrier-cross was evaluated for an acute onset of paraplegia. Utilizing magnetic resonance (MR) imaging, the cause of the neurologic deficits was determined to be a lumbar intervertebral disc extrusion. The MR study additionally demonstrated parenchymal hyperintensity on T2-weighted images and similarly located diffuse hypointensity on gradient echo images, cranial and caudal to the compressive extradural lesion. Hemorrhagic myelomalacia was suspected based on these MR characteristics, which was subsequently confirmed surgically and histopathologically.     

MAGNETIC RESONANCE IMAGING FEATURES OF DISCOSPONDYLITIS IN DOGS

The diagnosis of discospondylitis is based mainly on diagnostic imaging and laboratory results. Herein, we describe the magnetic resonance imaging (MRI) findings in 13 dogs with confirmed discospondylitis. In total there were 17 sites of discospondylitis. Eleven (81.1%) of the dogs had spinal pain for >3 weeks and a variable degree of neurologic signs. Two dogs had spinal pain and ataxia for 4 days. Radiographs were available in nine of the dogs. In MR images there was always involvement of two adjacent vertebral endplates and the associated disk. The involved endplates and adjacent marrow were T1‐hypointense with hyperintensity in short tau inversion recovery (STIR) images in all dogs, and all dogs also had contrast enhancement of endplates and paravertebral tissues. The intervertebral disks were hyperintense in T2W and STIR images and characterized by contrast enhancement in 15 sites (88.2%). Endplate erosion was present in 15 sites (88.2%) and was associated with T2‐hypointense bone marrow adjacent to it. In two sites (11.8%) endplate erosion was not MR images or radiographically. The vertebral bone marrow in these sites was T2‐hyperintense. Epidural extension was conspicuous in postcontrast images at 15 sites (88.2%). Spinal cord compression was present at 15 sites (88.2%), and all affected dogs had neurologic signs. Subluxation was present in two sites (11.8%). MRI shows characteristic features of discospondylitis, and it allows the recognition of the exact location and extension (to the epidural space and paravertebral soft tissues) of the infection. Furthermore, MRI increases lesion conspicuity in early discospondylitis that may not be visualized by radiography.     

Clinical signs, magnetic resonance imaging features, and outcome after surgical and medical treatment of otogenic intracranial infection in 11 cats and 4 dogs

Brainstem dysfunction resulting from central extension of infection is a life-threatening complication of otitis media/interna (OMI) that has been described infrequently in dogs and cats. We review the clinical signs of disease, diagnostic findings, and results of surgical and medical treatments of brainstem disease attributable to otogenic intracranial infection in cats and dogs. Eleven cats and 4 dogs were examined because of acute, subacute, or chronic clinical signs of brain disease including central vestibular signs, altered mentation, abnormal posture/gait, cranial nerve deficits, and seizures. Results of a minimal database (CBC, serum biochemical panel, urinalysis, thoracic radiographs, and abdominal ultrasonographic images or radiographs) were within reference intervals in all animals. Magnetic resonance (MR) images of the head were acquired for all animals, and cisternal cerebrospinal fluid (CSF) from 9 of 11 cats and 3 of 4 dogs was examined. Surgical exploration and ventral bulla osteotomy were done for 12 of 15 animals, followed by 1–3 months of antibiotic therapy; the remaining animals were euthanized before treatment. In all animals, MR imaging was effective in characterizing the location and extent of the pathologic changes intracranially as well as within middle/inner ear structures. Results of CSF analysis were characteristic of bacterial infection in most of the animals with acute or subacute disease. Since long-term outcome in all treated animals was very good to excellent, it was concluded that dogs and cats with intracranial disease secondary to extension of otitis media/interna have a good-to-excellent prognosis when the condition was diagnosed and was treated by surgical exploration and appropriate antibiotic therapy.     

COMPARISON OF COMPUTED TOMOGRAPHIC AND PATHOLOGIC FINDINGS IN 17 DOGS WITH PRIMARY ADRENAL NEOPLASIA

The CT appearance of canine adrenal masses has been reported, but associations between imaging features and pathologic features of these lesions have not been investigated in detail. The purpose of this study was to test associations between different types of adrenal neoplasia and their CT and pathologic features. A retrospective cross‐sectional study was performed and inclusion criteria were histologic diagnosis of primary adrenal neoplasia, contrast‐enhanced CT examination of the abdomen and surgical resection of the mass or necropsy examination. For all included dogs, CT images and histopathologic specimens were reviewed independently by two veterinary radiologists and a veterinary pathologist, respectively. Seventeen dogs met inclusion criteria. Diagnoses were adenocarcinoma in nine (53%) dogs, pheochromocytoma in five (29%) dogs, and adenoma in three (18%) dogs. Pheochromocytoma was associated with CT signs of vascular invasion (likelihood ratio = 4.8, 95% CI = 1.3–18.3, P = 0.03) and macroscopic vascular invasion (likelihood ratio = 9.6, 95% CI = 1.4–65.9, P = 0.02). There was excellent agreement between signs of vascular invasion in CT images and vascular invasion at surgery or necropsy (kappa = 0.86, P = 0.001). A peripheral contrast‐enhancing rim in delayed postcontrast CT images was associated with fibrous encapsulation of the tumor (kappa = 0.53, P = 0.05), and a heterogeneous pattern of contrast distribution in delayed postcontrast CT images was associated with adrenal hemorrhage or infarction on histological examination (kappa = 0.45, P = 0.05). Findings indicated that CT enabled assessment of adrenal neoplasia features that reflected their biological behavior and pathological findings, however overlapping characteristics between tumor types limited the potential for reliably distinguishing them based on CT alone.     

MRI features can help to confirm a diagnosis of progressive myelomalacia,but may not be accurate in dogs lacking characteristic clinical signs at the time of imaging

Progressive myelomalacia (PMM) is a fatal sequela of acute thoracolumbar intervertebral disc extrusion in dogs, with unpredictable onset in the days after the inciting injury. No single reliable diagnostic test is currently available. Magnetic resonance imaging (MRI) features such as T2-weighted spinal cord hyperintensity and loss of subarachnoid signal in a half-Fourier single-shot turbo spin echo (HASTE) sequence have been associated with PMM, but are sometimes present in other dogs with severe deficits. Magnetic resonance imaging findings in 22 dogs with a clinical or histopathologic diagnosis of PMM and 38 deep pain-negative paraplegic dogs were compared in a retrospective case-control study. Length of T2-weighted hyperintense spinal cord change and HASTE signal loss were significantly associated with clinically evident PMM (P = .0019 and P = .0085), however, there were no significant differences between groups when analysis was restricted to dogs not yet showing clinical signs of PMM. The PMM group also had significantly shorter compressive lesions than the control group (P = 0.026), suggesting a possible role of more severe focal pressure at the extrusion site. A segment of total loss of contrast enhancement in the venous sinuses and meninges, a feature not previously described, was more common in the PMM group and the difference approached significance (P = 0.054). Findings show that MRI features can support the diagnosis in dogs with clinical evidence of PMM, and absence of these features supports absence of PMM at time of imaging. However, their absence does not reliably differentiate dogs with imminent progressive myelomalacia from other dogs with severe deficits following intervertebral disc extrusion.     

CERVICAL VERTEBRAL FUSION AND CONCURRENT INTERVERTEBRAL DISC EXTRUSION IN FOUR DOGS

Cervical vertebral fusion was noted radiographically in four dogs presented for signs of cervical spinal cord compression. Ventral extradural spinal cord compression was seen on myelography at intervertebral disc spaces adjacent to the fused vertebrae in two dogs and at a site removed in two dogs. At surgery, no intervertebral disc space was found in the area of fusion. No other instances of cervical vertebral fusion were identified in reviewing radiographs of 1225 other dogs with cervical intervertebral disc extrusion evaluated at our hospital. Clinical signs resolved in all dogs after surgical removal of extruded intervertebral disc material. Information from these four dogs suggest vertebral fusion may predispose adjacent discs to herniation.     

Magnetic resonance imaging findings in histologically confirmed Pug dog encephalitis

The purpose of the study was to describe magnetic resonance (MR) imaging features of histologically confirmed necrotizing encephalitis in four Pugs and to compare those findings with MR imaging characteristics of necrotizing encephalitis in other breeds. All dogs had the following common findings: lesions restricted to the forebrain, both cerebral hemispheres diffusely but asymmetrically affected, lesions affected gray and white matter resulting in loss of distinction between both, most severe lesions in occipital and parietal lobes, lesions were irregularly T2-hyperintense and T1-isointense to slightly T1-hypointense, and no cavitation. There were various degrees of contrast enhancement of brain and leptomeninges. Asymmetry of lateral ventricles and midline shift was seen in one dog each. Two dogs had brain herniation, which may have contributed to the progression of neurologic signs. Hyperintensity on T2-weighted and fluid attenuated inversion recovery images in the hippocampus and piriform lobe was consistent with excitotoxic edema, whereas similar imaging features in other forebrain areas corresponded to areas of inflammation or liquefaction on histopathology. In comparison with necrotizing encephalitis in other canine breeds, Pug dog encephalitis has some unique MR imaging features. Therefore, these characteristics cannot be applied to other breeds, nor should imaging features of necrotizing encephalitis of other canine breeds be used for interpretation of MR images in Pug dogs.     

Clinical and magnetic resonance imaging features of canine compressive cervical myelopathy with suspected hydrated nucleus pulposus extrusion

Objective : To describe clinical and magnetic resonance imaging features in dogs with compressive cervical myelopathy due to acute suspected hydrated nucleus pulposus extrusion. Design : Retrospective case series. Animals : Ten dogs with compressive cervical myelopathy caused by acute suspected hydrated nucleus pulposus extrusion. Materials and Methods : Medical records and magnetic resonance images of dogs evaluated from 2005 through 2010 were reviewed. The inclusion criteria were clinical signs compatible with cervical myelopathy, magnetic resonance imaging of the spine performed within 48 hours after onset, magnetic resonance imaging findings consistent with compressive suspected hydrated nucleus pulposus extrusion, complete medical records and follow‐up information. Results : Six dogs were presented with acute onset tetraplegia and four dogs with acute onset of nonambulatory tetraparesis. Compromised respiratory function was present in three dogs. Compressive suspected hydrated nucleus pulposus extrusion was found on magnetic resonance imaging at the C4‐C5 (n=6), C3‐C4 (n=3) and C5‐C6 (n=1) intervertebral disc spaces. Seven dogs underwent surgical treatment and three dogs were treated conservatively. All dogs except one regained ambulatory status within two weeks after the onset, and had a favourable outcome. Clinical Significance : Compressive myelopathy caused by acute suspected hydrated nucleus pulposus extrusion has not been reported previously and, even though neurological deficits can be severe on presentation, outcome is favourable.     

SPINAL MAST CELL TUMORS IN DOGS: IMAGING FEATURES AND CLINICAL OUTCOME OF FOUR CASES

Published information regarding canine vertebral column mast cell tumors (MCTs) is limited. The objectives of this study were to report clinical and advanced imaging findings for a group of dogs with confirmed spinal MCT. Inclusion criteria for this retrospective case series were dogs with spinal magnetic resonance imaging (MRI) or computed tomography (CT) scans and a histological diagnosis of spinal MCT. Clinical, imaging, treatment, and outcome data were recorded. Four dogs met inclusion criteria. One dog had primary spinal MCT and three dogs had metastatic spinal MCT. All four dogs presented for paraspinal hyperesthesia and subacute progressive or acute myelopathy. All CT and MRI lesions were extradural. Two cases exhibited distinct masses in the epidural space. In one case, an epidural tumor invaded from the paravertebral musculature. One case exhibited polyostotic lesions indistinguishable from multiple myeloma by MRI. One dog with a primary epidural low‐grade MCT remains clinically normal 4 years postoperatively, following adjunctive lomustine. An epidural high‐grade MCT, metastatic from a cutaneous tumor, recurred within 2 months of surgery despite adjunctive vinblastine. Two high‐grade cases with concurrent visceral involvement were euthanized immediately after imaging. In dogs, MCT should be considered as a differential diagnosis for a progressive painful myelopathy and CT or MRI evidence of an extradural spinal lesion (epidural, paravertebral, or polyostotic). While more often associated with cutaneous or disseminated disease, MCT may also occur as a primary tumor of the epidural space in dogs.