Materials and Methods

The medical records and imaging database of the Veterinary Teaching Hospital at Washington State University were retrospectively cross-referenced for both presumptive and confirmed cases of diskospondylitis in dogs from 1997 through 2010. There were no age, sex, breed, or geographic restrictions. Inclusion criteria for the study included a complete clinical history and thorough neurologic examination, MRI of the vertebral column, and bacteriologic cultures of urine, blood, CSF, or intervertebral disk. Corresponding survey radiographs of the spine were evaluated if available. Those patients having undergone prior vertebral column surgery at the neuroanatomic site of interest were excluded from the study.

Patients were assigned a neurologic score from 1 to 5 similar to a modified Frankel score used in a previous study¹¹; however, no distinction was made between superficial and deep pain sensation. Neurologic deficits were scored on a progressive 5-point scale (1 = spinal pain without neurologic deficits; 2 = deficits evident, but dog could walk unassisted; 3 = dog was nonambulatory but maintained voluntary motor function; 4 = voluntary motor function was absent, but dog retained nociception caudal to the lesion; and 5 = complete paralysis with the loss of nociception caudal to the lesion).

All patients had multiplanar MRI with a 1.0-T magnet.^a Images for each case were evaluated by all authors. T2-weighted, T1-weighted, T1-weighted postcontrast (gadopentate dimeglumine^b), and STIR images in standard planes were reviewed and included when available. Specific imaging characteristics were documented on a prefabricated MRI scoring sheet. These included the number and location of affected intervertebral disk spaces and the signal intensity of the affected vertebral endplate, vertebral body, intervertebral disk space, paravertebral soft tissues, and spinal cord over the affected disk. Signal intensity was described as hyperintense, hypointense, isointense, or mixed, compared with the surrounding normal structures. Additionally, the presence or absence of intervertebral disk distortion, vertebral endplate irregularity, spinal cord or cauda equina compression, pathological fracture, subluxation, T2-weighted intramedullary spinal cord intensity, and contrast enhancement were documented. If contrast enhancement was present, it was classified as homogenous or heterogeneous. Involvement of paravertebral soft tissues, degree of paravertebral soft tissue contrast enhancement, and degree of spinal cord hyperintensity were described as either focal (< 1 vertebral body) or diffuse (> 1 vertebral body). Associated nervous system compression was classified as mild (< 25% of the vertebral canal diameter), moderate (25% to 50%), or marked (> 50%). Corresponding radiographic characteristics, including the presence of endplate lysis and sclerosis and the presence or absence of bony reaction, were evaluated and included when available. All MRI and radiographic scoring was made via reviewer consensus.

Statistical analysis—The signalment, clinical features, and MRI characteristics were analyzed with descriptive statistics. Spearman rank correlation coefficient was used to determine whether there was a relationship between the neurologic score and the presence and degree of intramedullary hyperintensity, the number of compressive diskospondylitis lesions, and the severity of spinal cord or cauda equina compression. This correlation coefficient was also used to determine whether there was a relationship between the duration of clinical signs and the intensity of the affected endplates and intervertebral disk spaces on both the T2 and STIR sequences and the presence of contrast enhancement of the endplates and intervertebral disk spaces. Values of P < 0.05 were considered significant.

Results

Twenty-three dogs meeting the selection criteria were included in the study. Large-breed dogs (> 25 kg [55 lb]) were overrepresented and accounted for 21 of the cases. All but 1 were purebred dogs and included the following breeds: Labrador Retriever (n = 5), Boxer (3), Doberman Pinscher (3), German Shepherd Dog (3), and 1 each of the following: Akita, Saint Bernard, Great Pyrenees, Great Dane, Mastiff, Weimaraner, Pembroke Welsh Corgi, Maltese, and mixed. The majority of dogs (15/23) were > 5 years of age. The mean age of all dogs was 6.8 years (median, 7 years; range, 5 months to 13 years), with 12 males (5 neutered and 7 sexually intact) and 11 females (8 spayed and 3 sexually intact).

Three dogs had signs of spinal pain as the only physical or neurologic exam abnormality at the time of initial evaluation (neurologic score 1). The majority of dogs (12/23) were ambulatory paraparetic (neurologic score 2). Five dogs had such severe paresis that they were no longer ambulatory (neurologic score 3). Three dogs were plegic with intact nociception (neurologic score 4). No dogs were found to be plegic with a loss of nociception caudal to the lesion (neurologic score 5). The duration of clinical signs prior to initial evaluation was variable; however, the majority of dogs (13/23) had clinical signs noted by owners for ≤ 1 month. The majority of dogs (15/23) had a single lesion identified, 3 dogs had 2 lesions, 2 dogs had 4 lesions, 1 dog had 6 lesions, 1 dog had 10 lesions, and finally 1 dog had 11 lesions. Of the 56 affected intervertebral disk spaces, 39 were located in the thoracic spine. Of the remaining sites, 9 were at the lumbosacral intervertebral disk space, 4 were in the lumbar spine, 3 were in the cervical spine, and 1 was at the thoracolumbar junction.

An infectious organism was identified in 18 of 23 cases. Results of bacteriologic culture of blood were positive in 14 of 17 cases. Results of bacteriologic culture of urine were positive in 9 of 20 cases. Results of bacteriologic culture of both urine and blood were positive in 4 of 14 cases. An infectious organism was not isolated from any of the 3 fine-needle aspirates of affected disk spaces or from the 3 sites from which tissue was surgically obtained. Bacterial organisms were identified in 17 of 18 cases, and multiple bacteria were identified in 7 of these cases. A mixed bacterial and fungal infection occurred in 1 patient. In 1 dog, a single fungal organism was identified. Coagulase-positive staphylococcal species were isolated most commonly, with Staphylococcus intermedius identified in 8 dogs and Staphylococcus aureus in 2. Coagulase-negative staphylococcal species were also common and were identified in 6 dogs. Streptococcal species (Streptococcus canis, Streptococcus mutans, and others) were isolated in 5 cases, Escherichia coli in 3 cases, and Corynebacterium sp, Micrococcus sp, Eubacterium aerofaciens, Enterococcus sp, and Actinomyces sp in 1 case each. Two fungal organisms were identified and included Phialemonium obovatum and Paecilomyces sp in 1 case each.

Cerebrospinal fluid was obtained from 14 of the 23 dogs and was abnormal in 10 dogs. Abnormalities were nonspecific and included an elevated CSF microprotein concentration in 9 dogs (range, 51.6 to 578.5 mg/dL; reference limit, < 25 mg/dL). Neutrophilic pleocytosis (23 cells/L; reference limit, < 5 cells/μL) was found in 1 dog; mixed cell pleocytosis (12 cells/μL) and mixed lymphocytic and neutrophilic pleocytosis (20 cells/μL) were also found in 1 dog each. All 3 dogs with a pleocytosis had a concurrently elevated CSF microprotein concentration. Concurrent disease conditions identified at the time of initial evaluation included endocarditis (n = 1), pyelonephritis (3), protein-losing nephropathy (1), prostatitis (2), prostatic abscess (1), polyarthritis (1), lower urinary tract infection (9), and hypothyroidism (1).

Of the 56 diskospondylitis sites identified in the study, MRI characteristics of only 27 sites were reported because multiple adjacent similarly affected sites in a single dog were reviewed collectively. The most common MRI characteristics of the 27 sites were summarized (Table 1; Figures 1–3). In addition, vertebral endplate irregularity and erosion was observed at 25 of 27 sites. Intervertebral disk distortion was observed at 19 of 27 sites. Intervertebral disk distortion could not be evaluated at the remaining 8 sites due to severe disk space collapse. Four sites had evidence of subluxation, and a pathological fracture was identified at 1 site. No significant correlation was found between the intensity of the vertebral endplates on T2-weighted (P = 0.878) or STIR (P = 0.591) sequences and the duration of clinical signs prior to the performance of advanced imaging. Also, no significant correlation was found between the intensity of the intervertebral disk space on T2-weighted (P = 0.502) or STIR (P = 0.258) sequences and the duration of clinical signs prior to the performance of advanced imaging. Similarly, no significant correlation was found between the duration of clinical signs and presence of contrast enhancement of the intervertebral disk spaces (P = 0.328) or vertebral endplates (P = 0.052).

View Full Size

Sagittal T2-weighted MRI image of the caudal thoracic vertebrae (A) and transverse T2-weighted (B) and STIR (C) MRI images at the level of the T8–9 vertebral endplate or disk space of an 11-month-old male Doberman Pinscher. A—The T8–9 intervertebral disk space is collapsed, and the adjacent vertebral endplates are severely eroded, are of mixed signal intensity, and cannot be clearly distinguished from material within the T8–9 disk space. There is proliferative material of mixed signal intensity dorsal to the T8–9 intervertebral disk space causing moderate, focal compression of the spinal cord (arrow). B and C—The vertebral endplate is irregularly margined and of mixed signal intensity, with patchy areas of hyperintensity (arrows), compared with normal vertebral endplates.

Citation: Journal of the American Veterinary Medical Association 242, 3; 10.2460/javma.242.3.359

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Dorsal STIR MRI image of the thoracolumbar vertebrae (A) and transverse T1-weighted (B) and T1-weighted postcontrast (C) images at the level of the T12–13 intervertebral disk space of a 9.5-year-old spayed female Akita Mix. A—The caudal endplate of T12 and cranial endplate of T13 are irregular and hyperintense (arrow) in comparison with normal vertebral endplates. B and C–Notice the heterogenous contrast enhancement of the affected vertebral endplate (arrow).

Citation: Journal of the American Veterinary Medical Association 242, 3; 10.2460/javma.242.3.359

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Sagittal T2-weighted (A) and STIR (B) MRI images of the thoracolumbar vertebrae of an 8-year-old castrated male Weimaraner. There are minimal signal changes in the paravertebral soft tissues both dorsal and ventral to the affected disk space on T2-weighted images, whereas these tissues appear markedly abnormal with mixed hyperintensities on STIR sequences in the same imaging plane (arrows).

Citation: Journal of the American Veterinary Medical Association 242, 3; 10.2460/javma.242.3.359

• Download Figure
• Download figure as PowerPoint slide

Six dogs had no sites of spinal cord, cauda equina, or other nerve root compression associated with diskospondylitis lesions. Of the remaining 17 dogs, there were 22 sites of spinal cord or cauda equina compression identified. Thirteen dogs had 1 site of compression, 3 had 2 sites of compression, and 1 had 3 sites of compression. Of the 22 sites of compression, 10 were mild, 6 were moderate, and 6 were severe. A significant (P = 0.011) direct correlation (r = 0.6373) was found between the number of compressive lesions secondary to diskospondylitis and the neurologic score of the patient at the time of initial evaluation. Patients with a higher neurologic score (those with more severe deficits) had a higher number of compressive diskospondylitis lesions. Similarly, a significant (P = 0.011) direct correlation (r = 0.6010) was also found between the overall total percentage of spinal cord or cauda equina compression secondary to diskospondylitis and the neurologic score of the patient. Patients with greater percentage canal compression had higher neurologic scores.

Ten dogs had MRI evidence of concurrent spinal cord compression or disease unrelated to the diskospondylitis lesions: mild to moderate multifocal intervertebral disk disease, mild focal intervertebral disk disease, thoracic spinal cord atrophy, mild to moderate lumbosacral disease with foraminal stenosis, traumatic disk rupture in association with a vertebral body fracture, and syringohydromyelia.

Ten of the 23 diskospondylitis sites had changes to the spinal cord parenchyma visible as a focal (6/10) or diffuse (4/10) intramedullary hyperintensity. At 6 of 10 sites in which an intramedullary hyperintensity was present, there was an associated extradural compressive lesion. Two sites had mild compression, 3 had moderate compression, and 1 had severe compression. Although changes in spinal cord intensity were noted at 10 of 23 sites, no significant correlation was found between the presence of an intramedullary hyperintensity and the severity of patient neurologic score (P = 0.272) or the focal or diffuse nature of the hyperintensity and the patient neurologic score (P = 0.882).

Corresponding survey radiographs of the spine were available in 16 of 23 cases and were abnormal at every affected disk space concurrently identified on MRI. Endplate lysis was the most common finding, followed by endplate sclerosis and bony reaction.

Discussion

The results of the present study indicated that diskospondylitis in dogs has a characteristic MRI appearance, and in some patients, MRI may aid in the identification of severe spinal cord compression, which could warrant surgical intervention. The primary source of infection in small animal patients with diskospondylitis is infrequently determined, although urogenital infections, abscesses, open wounds, and respiratory tract and oral cavity infections are frequently implicated.¹² Of these, the urinary tract is generally accepted to be the most likely source of infection in most cases.¹² In the present case series of 23 dogs, 14 dogs had a concurrently identified disease process that may have acted as a source of infection. Only 2 dogs had clinicopathologic evidence of a lower urinary tract infection at the time of initial evaluation; however, results of bacteriologic cultures were positive in 9 of 20 samples submitted.

In this case series, bacterial infections were more common than fungal infections, and multiple bacterial infections occurred frequently. This is similar to studies^13–15 in which coagulase-positive staphylococcal species (S intermedius and S aureus) are isolated most commonly. Other bacterial organisms cultured from this case series included coagulase-negative staphylococcal species, E coli, Corynebacterium sp, and Actinomyces sp, which are also similar to reports^16,17 in dogs. Among fungal infections, Aspergillus and Paecilomyces spp are most commonly cultured,¹⁷ and Paecilomyces sp was the causative agent in 1 fungal infection in this case series.

Organism identification in diskospondylitis can prove to be a challenging task. Combining bacteriologic cultures of urine and blood is thought to provide the best chance of identifying the causative organism, with overall success rates ranging widely (between 30% and 78%).¹⁸ Attempts at bacterial isolation were successful in this case series, with bacteriologic culture of blood proving to be most valuable. More invasive techniques aimed at organism identification have been attempted and include fluoroscopically guided percutaneous needle aspiration of the affected intervertebral disk and surgical biopsy, which have reported success rates of 75% and 100%, respectively.¹⁹ In this study, needle aspiration of an affected disk space yielded negative results of bacteriologic cultures at all 3 sites in which it was performed as did surgically obtained biopsies from 3 sites. Cerebrospinal fluid analysis proved to be a sensitive diagnostic test; findings were abnormal in 10 of 14 cases. However, the abnormalities were quite varied and nonspecific. Results of bacteriologic culture of CSF were negative in all cases, making the diagnostic value of this test somewhat debatable. Reports of specific CSF findings in diskospondylitis in the veterinary literature are lacking, and microbial culture from CSF appears difficult.²⁰

Clinical signs in affected dogs are often nonspecific and highly variable. The most common signs in the present series included lethargy, anorexia, weight loss, fever, and signs of spinal pain. The duration of clinical signs prior to initial evaluation are also reported to be extremely variable, although most patients in this case series were affected for < 1 month. Affected dogs may be neurologically normal, or their systemic signs of illness may be accompanied by various degrees of neurologic impairment, spanning from signs of pain only to paralysis with absent nociception. Several pathological processes can account for neurologic deficits in this disease and include extrusion of intervertebral disk material secondary to collapse of an affected disk space, osseous or soft tissue proliferation within the vertebral canal in response to chronic inflammation, vertebral subluxation or pathological fracture secondary to marked bone lysis, and secondary meningitis or myelitis.^21,22 The incidence of neurologic deficits was much higher in this case series, compared with a previous report,²⁰ with 20 of 23 patients having various degrees of paresis or paralysis in addition to spinal hyperesthesia. The high incidence of neurologic impairment is likely a reflection of the fact that this retrospective case series included only those cases referred to a veterinary teaching hospital for neurologic evaluation.

Veterinary reports^10,23 have documented male dogs to be twice as likely as female dogs to be affected. No sex predilection was noted in this case series, with males and females being equally affected. There was also no difference noted between sexually intact and neutered patients. The mean age of affected dogs reported in the literature ranges from 4.1 to 9 years, and it is generally accepted that the risk increases with age and is greatest for dogs > 10 years of age.^10,23 Similarly, the mean age of dogs in this study was 6.8 years (median, 7 years) and most dogs were > 5 years of age. Only 4 dogs were > 10 years of age. In prior reports,^20,23 large-breed dogs, including dogs of the Great Dane, Boxer, Rottweiler, German Shepherd Dog, and Doberman Pinscher breeds, appear to be at a higher risk than are small-breed dogs and mixed-breed dogs. Similarly, large-breed dogs were overrepresented in this study, with Labrador Retrievers being the single most commonly affected breed.

Any area of the vertebral column can be affected by diskospondylitis. However, lesions in the lumbar and thoracic spine are much more commonly reported than are lesions in the cervical spine.²⁰ In this case series, most lesions were located in the thoracic spine and multiple lesions were a frequent finding. This finding is in accordance with a study²⁰ in which multiple lesions, typically involving the thoracic and lumbar spine in combination, were found. Historically, the lumbosacral space has been the most commonly reported site involved in cases of singular diskospondylitis lesions.^22,24,25 However, solitary thoracic lesions were more common in this case series, affecting 7 of 14 sites, compared with isolated lumbosacral lesions in only 5 of 15 sites.

Conventionally, the diagnosis of diskospondylitis has relied on the presence of characteristic radiographic findings. Lateral and ventrodorsal projections of the spine may reveal a loss of definition of the vertebral end plate margins, narrowing or collapse of the intervertebral disk space, and various degrees and combinations of endplate lysis, sclerosis, and bony proliferation.¹² It has been well established that radiographic evidence of diskospondylitis may lag behind the onset of clinical signs by as much as 2 to 6 weeks.^3,23,24 Thus, acutely affected patients may have equivocal or even normal radiographic findings, as in a study² performed in 2001. As such, MRI has been proposed as a more sensitive imaging modality for detecting early cases of diskospondylitis in dogs with normal or equivocal radiographic findings.^{1,8–10,24,25}

The use of MRI in the diagnosis of diskospondylitis in dogs has been infrequently reported in the veterinary literature.^1,7–9 This is in contrast to human medicine, where MRI is the modality of choice for inflammatory and infectious disease of the spine, including diskospondylitis, for which MRI characteristics have been well described.^4,6 Overall, although some of the MRI characteristics of diskospondylitis lesions in this case series (Table 1) were similar to reports^5,7–9 in both the veterinary and human literature, several notable differences were observed. The vertebral endplates in this case series were more likely to be of mixed signal intensity on T2-weighted images, containing multifocal areas of hyperintensity, whereas the vertebral bodies were most commonly hypointense to isointense. The intervertebral disks and disk spaces were most commonly diffusely hyperintense on T2-weighted and STIR sequences (Figures 1 and 2). Changes in paravertebral soft tissues were most commonly noted on T2-weighted and STIR sequences in which tissues most commonly appeared diffusely hyperintense, similar to results reported in other studies.^8,9 At 6 sites in which the paravertebral soft tissues were isointense or normal on T2-weighted images, changes in signal intensity were noted on STIR sequences (Figure 4).

Many of the vertebral endplates had contrast enhancement, which was most often heterogenous (Figure 3). Heterogenous contrast enhancement of the intervertebral disk or intervertebral disk space was also noted at many affected sites, whereas most vertebral bodies did not have enhancement. Paravertebral soft tissue enhancement occurred infrequently. Contrast enhancement was noted more commonly in those cases affected for ≥ 8 weeks. This finding was not significant; therefore, differentiating between acute and chronic lesions on the basis of contrast enhancement and other signal characteristics was not possible.

Intramedullary spinal cord intensity was only apparent on T2-weighted sequences and was slightly more likely to be focal than diffuse in nature. No significant correlation could be found between the presence of an intramedullary hyperintensity and the severity of associated extradural compression or the neurologic status of the patient.

Static spinal cord compression was very common in this case series. A correlation was found between the number of compressive lesions and the neurologic score of the patient and the overall degree of spinal cord or cauda equina compression secondary to diskospondylitis and the neurologic score of the patient. These findings suggest that the incidence of severe spinal cord compression may have previously been underestimated via other contrast radiographic procedures. Furthermore, with the increased sensitivity of MRI in identifying spinal cord compression and the documented direct correlation between spinal cord or cauda equina compression and neurologic score of the patient, more aggressive surgical intervention in these cases may need to be considered because decompression may positively influence patient outcome. Therefore, MRI should be considered for those patients with diskospondylitis with severe neurologic deficits or those that are refractory to conventional medical management.