Magnetic Resonance Imaging and Neurological Findings in Dogs with Disc-Associated Cervical Spondylomyelopathy (63 cases)

Background: Canine cervical spondylomyelopathy can be separated into osseous and disc-associated (DA-CSM) forms. Our aim was to describe the magnetic resonance imaging (using a high-eld scanner) and neurological ndings in dogs with DA-CSM and investigate a relationship between these ndings. Results: Sixty-three dogs were included: 60/63 (95%) were large breeds, with Doberman Pinschers and males over-represented (70%). Mean and median age at the time of diagnosis was 7.25 and 7.2 years (range 0.41 – 12 years). Chronic signs were noted in 52/63 (83%) dogs, with proprioceptive ataxia the most common. Main site of spinal cord compression was commonly C6-C7 or C5-C6. Thirty-six (57%) dogs had various sites of spinal cord compression. Most dogs younger than 6 years of age had a single affected. Foraminal stenosis was present in 51/63 dogs (81%). T2-weighted hyperintensity was present in 40/63 dogs (63%). Eighty-eight percent of the articular processes showed degenerative changes, which correlated strongly with intervertebral disc degeneration. Ligamentum avum hypertrophy was seen in 38% of dogs. No correlation was observed between neurologic signs and number of affected sites. A moderate positive correlation was observed between severity of spinal cord compression and neurologic grade (r 0.48; p<0.001). Conclusions: DA-CSM was predominantly observed in older, male Dobermans, with chronic neurologic signs, with compressive lesions located in the caudal cervical vertebral region. Although less common, DA-CSM was also seen in dogs 3 years of age or younger (8%). Single compressive lesions were more common in dogs younger than 6 years of age. Most dogs had concomitant changes (e.g.: ligamentum avum hypertrophy and foraminal stenosis) that may affect neurologic signs. Most dogs with ligamentum avum hypertrophy were 6 years or older. A correlation was observed between severity of spinal cord compression and neurologic grade; dogs with more severe location of compressive lesions, main compressive lesion (site with the greatest reduction in cross-sectional area and T2-weighted hyperintensity when present), severity of the compression (grade 1 = mild: less than 25% of the diameter of the spinal cord; grade 2 = moderate: 25–50%, grade 3 = severe: greater than 50%), presence of intervertebral disc degeneration, presence of spinal cord signal changes on T1 and T2-weighted images, presence of ligamentum avum hypertrophy, intervertebral foraminal stenosis, articular degenerative and of of intervertebral disc presence of spondylosis, presence of spinal cord signal changes, presence and severity of intervertebral foraminal stenosis, presence of signs of articular process degenerative changes, and presence of ligamentum avum at the same site of main compression. presence and severity of intervertebral foraminal stenosis, presence of signs of articular process degenerative changes.


Results
Sixty-three (27%) out of 232 dogs diagnosed with CSM during the study period met the inclusion criteria.
Overall, 24/63 (38%) dogs had concomitant ligamentum avum hypertrophy causing spinal cord compression (Fig. 2). This was located at the same site as the intervertebral disc protrusion in 13/24 (54%) dogs, while in 7/24 (29%) dogs they were located at a different site and 4/24 (17%) dogs had ligamentum hypertrophy both at the same site of intervertebral disc protrusion and at a different site.
When comparing dogs with single vs. multiple affected sites, dogs with ligamentum avum hypertrophy more commonly had multiple sites of spinal cord compression (p < 0.001). No other statistical signi cance was observed. Of note, whether dogs had single or multiple sites affected had no relationship detected with neurologic grade or signal changes.
When investigating correlations among neurologic and imaging parameters, a moderate positive correlation was observed between severity of spinal cord compression and neurologic grade (r 0.48; p < 0.001), between age and number of protruded discs (r 0.54; p < 0.001), number of intervertebral discs with total degeneration (r 0.47; p < 0.001), and spondylosis (r 0.48; p < 0.001). A moderate correlation was also observed between the number of protruded discs causing spinal cord compression and number of intervertebral discs with signs of total degeneration (r 0.45; p < 0.001) and presence of ligamentum avum hypertrophy (r 0.46; p < 0.001).
Regarding comparison between neurologic grades, a signi cant difference was observed (p < 0.001). Dogs with spinal cord compression with a severity score of 3 (18 dogs) had a neurologic grade of 3 or higher (9/18 were grade 3, 6/18 were grade 4, 3/18 were grade 5). Dogs with spinal cord compression with a severity score of 2 (27 dogs) mostly had a neurologic grade between 2 and 4 (9/27 were grade 4, 7/27 were grade 2, and 7/27 were grade 3). As for dogs with spinal cord compression with a severity of 1 (18 dogs), the majority had neurologic grades of 1 and 2 (7/18 were grade 2 and 6/18 were grade 1). Of note, there was one dog with spinal cord compression severity score of 1 and one dog with severity score of 2 that presented with neurologic grade of 5. The former had an acute presentation and the latter had acute worsening of chronic signs.
A signi cant difference was seen when comparing grouped neurologic scores for severity of spinal cord compression (p = 0.0049). The majority of dogs with spinal cord compression severity score of 1 (15/18) and 2 (17/27) had neurologic grade 1 to 3 (Group A).
As for correlation between degenerative changes of the intervertebral disc and articular processes, a very strong positive correlation was observed between the number of intervertebral discs with total degeneration and the number of articular processes with irregular surface and signs of subchondral sclerosis (r 0.90; p < 0.05). A very strong correlation was also found between the number of intervertebral discs with total degeneration and the total number of articular process surfaces with any signs of degenerative changes (r 0.84; p < 0.05) and between the total number of intervertebral discs with signs of degenerative changes and the number of articular processes with irregular surface and signs of subchondral sclerosis (r 0.81; p = 0.05).
Foraminal stenosis was present in some degree in 51/63 dogs (81%) in at least one location. A weak correlation was noted between tetraparesis and the presence of foraminal stenosis (r 0.26; p = 0.04). Spondylosis was noted in 33/63 dogs (52%), and a moderate positive correlation was observed between age and spondylosis (r 0.48; p < 0.001).
Regarding degenerative changes to the articular processes, 56/63 (89%) dogs had some degree of subchondral sclerosis with reduced or absent amount of synovial uid noted. A total of 319 locations were evaluated. Of these, 279/319 (88%) had some degree of subchondral sclerosis and 280/319 (88%) had evidence of reduced or absent synovial uid on MRI.
Percentages have all been rounded to the nearest whole.

Discussion
In this study, we report the rst high-eld MRI case series of dogs with disc-associated CSM. Most of the dogs in this series were older (68%), primarily Doberman Pinschers (70%), male (70%), and had a chronic history of progressive signs (83%). These ndings are in alignment with observations from other reports [10][11][12][13][14]. We did observe DA-CSM also affecting 8% of young dogs, which was infrequently reported [12,13]. As expected in DA-CSM, compressive lesions were centered in the caudal cervical vertebral region (C6-C7, C5-6) in 90% of the dogs. Proprioceptive ataxia was the most common clinical sign (78%), followed by paresis (58%) and cervical pain (49%). Interestingly, 55% of dogs with cervical hyperesthesia noted on physical exam had this noticed by the owners, while approximately half of these presented with cervical hyperesthesia alone (neurologic grade 1).
We hypothesized that dogs with multiple sites of spinal cord compression would be more severely affected, but this was not con rmed. We found that the majority of dogs (57%) had more than one site of spinal cord compression, with 37% of these having two sites of compression. Dogs with ligamentum avum hypertrophy more often had multiple sites of spinal cord compression. No other statistical signi cance was observed. Of note, whether dogs had single or multiple sites affected had no relationship detected with neurologic grade or signal changes. In dogs with osseous-associated CSM, a correlation between number of affected sites and neurological signs has also not been observed [7]. Perhaps the presence of less severe sites of spinal cord compression would not su ciently worsen the patient's neurological presentation.
As for our secondary hypothesis, that severity of compression of the spinal cord would be associated with a worse neurologic grade, a moderate positive correlation was observed between severity of spinal cord compression and neurologic grade. Previous studies with osseous-associated CSM [7] and DA-CSM [15] have not found a correlation between severity of compression and neurologic score. The latter study was done using a low-eld MRI using 21 dogs [15].
Studies looking at intervertebral disc herniations in dogs have also tried investigating a correlation between degree of spinal cord compression and neurological grade [16][17][18][19]. Despite a noted trend of increasing spinal cord compression seen on MRI and more severe neurologic grade in dogs with acute thoracolumbar disc herniations [18], a correlation between these two variables has not been con rmed [16]. A correlation was, however, observed between presenting neurologic grade of dogs with disc extrusion in the cervical region [19]. This discrepancy may in part be explained due to the anatomical differences between the cervical and thoracolumbar regions of the vertebral column [19]. Also, factors such as initial concussive force of the extrusion and length of the compression likely also play important roles in acute intervertebral disc extrusions [18,19]. Although in DA-CSM, concussive force and length of the compression do not commonly play a role, other equally complex forces may help explain the discrepancies found when dealing with DA-CSM [15]. The presence of dynamic lesions, for instance, may lead to divergences between spinal cord compression in neutral position and maximum spinal cord compression, since dynamic MRI has shown that extension and exion of the cervical vertebral column can affect the degree of spinal cord compression [20]. Total number of dogs investigated, variations in the neurologic grading system used, and different observer and measuring methods in each study may also help explain differences between studies.
Except for dogs with severe spinal cord compression (grade 3) where all dogs were classi ed as at least having moderate signs (grade 3), dogs with mild to moderate spinal cord compression had less clearly demarked distribution. Factors that may have contributed to these exceptions include evolution of the lesion (acute or chronic) and a possible dynamic component.
Most of the dogs in this study (63%) had spinal cord signal changes in the form of T2W hyperintensity. This percentage is similar to the overall average (56.5%) published for dogs with either form of CSM [3,8,21,22] and that reported speci cally in dogs with DA-CSM [12]. T1W hypointensity was not as prevalent, and was only observed in 20% of dogs with signal changes. Correlations observed for T2W hyperintensity with neurologic grade and with severity of spinal cord compression, as well as between T1W hypointensity and duration of clinical signs seem to be in line with previous suppositions, where T2W hyperintensity would be associated with clinical relevance of the lesion [15] and T1W hypointensity with chronicity [23]. Spinal cord signal changes, as represented by an association of T2W hyperintensity and T1W hypointensity, have been associated with gray matter changes such as motor neuron loss and necrosis and have been considered indicative of a worse prognosis [24][25][26]. Further studies are still needed to fully understand how the presence of spinal cord signal changes may be related to neurological status and prognosis in dogs.
A total of 38% dogs had concomitant ligamentum avum hypertrophy causing some degree of spinal cord compression, with 95% of those dogs having multiple sites of compression. Overall, 64% of dogs with multiple affected sites had ligamentum avum hypertrophy. Ligamentum avum hypertrophy causing spinal cord compression has been previously observed in dogs with disc-associated CSM [10,12]. A moderate correlation was observed between the presence of ligamentum avum hypertrophy and the number of intervertebral disc protrusions, but a reason for why these would occur in some locations where intervertebral disc degeneration or protrusion is present, but not others, is unknown. Perhaps this is a re ection of biomechanical changes in a speci c site as well as the effect of age and increasing number of degenerated intervertebral discs. The fact that practically half (49%) of dogs 6 years of age or older had ligamentum avum hypertrophy does seem to suggest that age may play a role in this nding.
Almost all dogs (98%) had intervertebral disc degeneration, with a higher prevalence of partial intervertebral disc degeneration. Also, most (80%) dogs with a single affected site were younger than 6 years of age while the majority (53%) of dogs with multiple sites affected were 6 years or older. This was not unexpected since most dogs in the study population were middle-aged or older dogs [10,12,13].
Statistically, there was a moderate correlation between age and number of protruded discs, as well as degenerated discs.
Foraminal stenosis was also prevalent, with some degree of stenosis observed in 81% of the dogs in this study. A weak correlation was noted between tetraparesis and foraminal stenosis, which may suggest an association between these two, however, there was no correlation between tetraparesis and maximum severity of foraminal stenosis.
The high percentage (89%) of sites with degenerative changes to the articular process joints was an unexpected nding in dogs with purely disc-associated CSM and no proliferative changes to the articular processes. A strong correlation was observed between the number of intervertebral discs with total degeneration and number of articular processes with irregular surface and signs of subchondral sclerosis. Interestingly, a previous study observed a weak correlation between intervertebral disc degeneration and degenerative changes to the articular synovial uid, but not with surface changes; however, that study was done using a comparatively smaller population of 13 Great Danes, where the majority (12/13) had spinal cord compression due to articular process changes. Another difference, likely secondary to breed and type of CSM, was the median age of dogs in that study was 3.9 years [22]. While it is di cult to make a proper comparison between that study and ours, it is interesting to note that both found a correlation between changes to articular process joints and intervertebral disc degeneration. It is unclear how several of these degenerative changes interact and develop, and which would be the primary determining factor. The authors believe a con uence of anatomical, degenerative, and biomechanical changes are interwoven in the pathogenesis of the investigated changes in DA-CSM.
The main limitation of this study is its retrospective design, which is associated with all limitations common to retrospective studies. Another limitation would be the small number of dog breeds other than Doberman Pinscher, but this most likely represents the distribution of dogs with DA-CSM since the dogs herein presented were selected from a large cohort at the authors' institution.

Conclusions
Overall, disc-associated CSM was mostly observed in older, male Doberman Pinschers, with a chronic history of neurologic signs (proprioceptive ataxia was the most common nding, followed by paresis). Though not as common, it should be noted that 8% were younger dogs (< 3 years). Cervical hyperesthesia was noted in approximately half the dogs. Compressive lesions were located in the caudal cervical vertebral region (C6-C7, C5-6) in the vast majority of dogs. A large number of dogs had concomitant changes such as ligamentum avum hypertrophy and intervertebral foraminal stenosis that may affect neurologic signs. The majority of dogs with ligamentum avum hypertrophy were 6 years or older.
A correlation was observed between severity of spinal cord compression and neurologic grade, where dogs with more severe spinal cord compression were more likely to have a higher neurologic grade.
Dogs with more sites of spinal cord compression were not shown to have more severe neurologic involvement. Unexpectedly, a very high percentage of dogs with purely DA-CSM had degenerative changes in the articular processes. A possible biomechanical or genetic relationship between degenerative changes in articular processes, ligamentum avum, and intervertebral discs warrants further investigation.

Methods
Records were searched from January 2005 to August 2019 for dogs with a con rmed diagnosis of cervical spondylomyelopathy based on clinical and magnetic resonance imaging (MRI) ndings in a single academic center. Criteria for being included in the present study were a diagnosis of CSM caused by intervertebral disc protrusion (with or without ligamentum avum hypertrophy) and availability of clinical data and MRI for review. Exclusion criteria was the presence of concomitant spinal cord or nerve root compression caused by osseous proliferation of the articular facets (with or without ligament hypertrophy), by laminar thickening (with or without ligament hypertrophy). Dogs were also excluded if clinical signs resulted solely from ligamentum avum hypertrophy or intervertebral foraminal stenosis.
Data obtained from medical records included breed, age, gender, duration of clinical signs (acute when < 1 week, chronic when ≥ 1 week before diagnosis), presence of ataxia and/or paresis, presence of cervical hyperesthesia, and neurologic grade (1 to 5). The neurologic grading system was adapted from a previously published system [9]: grade 1, cervical hyperesthesia only; grade 2, mild pelvic limb ataxia or paresis ± thoracic limb involvement; grade 3, moderate pelvic limb ataxia or paresis ± thoracic limb involvement; grade 4, marked pelvic limb ataxia or paresis with thoracic limb involvement; grade 5, tetraparesis with inability to stand or walk without assistance.
Magnetic resonance images were acquired using a high-eld, 3.0 T MRI scanner (Achieva 3.0 T, Philips Healthcare) and a surface coil. Dogs were positioned in dorsal recumbency with the cervical region extended in neutral position. Images were acquired using a turbo spin-echo technique. Overall, these were the minimum protocol used: transverse T1-weighted and T2-weighted images were obtained. Repetition time (TR) and echo time (TE) were as follows: T1-weighted, TR = 650 ms, TE = 8 ms; T2-weighted, TR = 4000 ms, TE = 120 ms. The number of acquisitions was 2. The ip angle was set at 90°. The eld of view was 20 cm with matrix dimensions of 200 × 192 mm. Slice thickness was set at 3 mm with no interslice interval. The entire cervical vertebral region was included (C1-T2) in the sagittal images. Transverse images were available for 319 (84%) out of 378 intervertebral disc spaces between C2 and T1.
Magnetic resonance images were reviewed by a board-certi ed neurologist with extensive expertise and experience in evaluating MRI studies of dogs with CSM. The following data was recorded: location of compressive lesions, main compressive lesion (site with the greatest reduction in cross-sectional area and T2-weighted hyperintensity when present), severity of the compression (grade 1 = mild: less than 25% of the diameter of the spinal cord; grade 2 = moderate: 25-50%, grade 3 = severe: greater than 50%), presence of intervertebral disc degeneration, presence of spinal cord signal changes on T1 and T2weighted images, presence of ligamentum avum hypertrophy, intervertebral foraminal stenosis, signs of articular process degenerative changes, and presence of spondylosis. These changes were classi ed according to previously published studies [3,9,22,27,28]. Articular process joints were evaluated as a pair per location (C2-C3 through C7-T1). The appearance of the intervertebral discs was classi ed as normal (P rmann grade I), partial degeneration (grade II or III), or total degeneration (grade IV or V) [29].
We investigated an association between age, gender, severity of neurologic signs (neurologic grade), duration of clinical signs, history (chronic or acute), main site of spinal cord compression, number of intervertebral discs with signs of degeneration, number of intervertebral disc protrusions, presence of spondylosis, presence of spinal cord signal changes, presence and severity of intervertebral foraminal stenosis, presence of signs of articular process degenerative changes, and presence of ligamentum avum at the same site of main compression.
We also compared dogs with single vs. multiple sites of spinal cord compression for: age, gender, severity of neurologic signs (neurologic grade), duration of clinical signs, history (chronic or acute), number of intervertebral discs with signs of degeneration, number of intervertebral disc protrusions, presence of ligamentum avum hypertrophy, presence of spondylosis, presence of spinal cord signal changes, presence and severity of intervertebral foraminal stenosis, presence of signs of articular process degenerative changes.
For statistical analyses, dogs were also grouped according to neurologic grade as follows: grouped neurologic grade A included dogs graded from 1 to 3 and grouped neurologic grade B included dogs graded 4 and 5. Groups A and B were then compared for the same characteristics as the comparison between non-grouped neurologic grades.
Data were analyzed using SAS statistical software (version 9.4, SAS Institute Inc., Cary, NC). Descriptive statistics were calculated using Proc FREQ, and mean, median and range were reported. Association between listed variables were tested using chi-squared test of independence or Fisher's exact test as appropriate and Spearman's rank correlation coe cient (r) was used to calculate the relationship between variables. Correlation strength were classi ed as values < 0.19 considered very weak correlation, 0.   Sagittal T2-weighted (A) and T1-weighted (B) and transverse T1W (C) and T2W (D) magnetic resonance images from an 8-year-old Doberman Pinscher diagnosed with disc-associated cervical spondylomyelopathy. Spinal cord compression is observed due to intervertebral disc protrusion at C6-C7 (asterisk) and intervertebral disc protrusion and ligamentum avum hypertrophy at C4-C5 (arrow). Note T2W hyperintensity of the spinal cord parenchyma at C6-C7.