J Vet Intern Med 2007;21:12901298

Magnetic Resonance Imaging Findings and Clinical Associations in 52 Dogs with Suspected Ischemic Myelopathy
Luisa De Risio, Vicki Adams, Ruth Dennis, Fraser McConnell, and Simon Platt
Background: The magnetic resonance imaging (MRI) features of ischemic myelopathy have been described in the human literature and in a small number of cases in the veterinary literature. Hypothesis: The aims of this study were to identify associations among MRI findings, timing of imaging, and presenting neurologic deficits in a large series of dogs with a presumptive diagnosis of ischemic myelopathy. Animals and Methods: The medical records and MR images of dogs with a presumptive diagnosis of ischemic myelopathy (20002006) were reviewed retrospectively. Inclusion criteria were acute onset of nonprogressive and nonpainful myelopathy, 1.5-tesla MRI of the spine performed within 7 days of onset, and complete medical records and follow-up information. Presumptive diagnosis was based on history, as well as clinical, MRI, and cerebrospinal fluid (CSF) findings. The extent of the lesion on MRI was assessed as the following: (1) the ratio between the length of the hyperintense area on sagittal T2-weighted images and the length of C6 or L2 vertebral body, and (2) the maximal cross-sectional area of the hyperintense area on transverse T2-weighted images as a percentage of cross-sectional area of the spinal cord. Results: Fifty-two dogs met the inclusion criteria. MRI findings were abnormal in 41 dogs and normal in 11 dogs. The presence of MRI abnormalities was not significantly associated with the timing of imaging (P 5 .3) but was associated with ambulatory status on presentation (P 5 .04). Severity of signs on presentation was associated with extent of the lesion on MRI (P 5 .02). Conclusion and Clinical Importance: The severity of signs on presentation is associated with the presence and the extent of the lesion on MRI. Key words: Canine; Fibrocartilaginous embolization; Magnetic resonance imaging; Spinal infarct.

schemic myelopathy has been described in several species including humans,15 dogs,618 cats,13,1924 horses,25 pigs,26 sheep,27 turkeys,28 and tayra.29 In these reports, ischemic myelopathy was most commonly caused by fibrocartilaginous emboli histochemically identical to the nucleus pulposus of the intervertebral disk. Histopathologic findings included the presence of fibrocartilage in the lumen of spinal cord arteries or veins and associated ischemic necrosis of the spinal cord parenchyma.1,2,622,2429 Several theories have been proposed to explain how the fibrocartilaginous material enters the vascular system.30 These include direct penetration of nucleus pulposus material into the vessels of the spinal cord or into the vertebral vessels, presence of embryonic remnant vessels within the nucleus pulposus, neovascularization of the degenerated intervertebral disk, and mechanical herniation of nucleus pulposus into the vertebral bone marrow with sub-

Centre for Small Animal Studies, Animal Health Trust, Newmarket, Suffolk, UK (De Riso, Dennis, McConnell, Platt); and the Centre for Preventive Medicine, Animal Health Trust, Newmarket, Suffolk, UK (Adams). Dr McConnell is currently affiliated with the Diagnostic Imaging Service, Faculty of Veterinary Science, University of Liverpool, Liverpool, UK. Dr Platt is currently affiliated with the Department of Small Animal Medicine, College of Veterinary Medicine, University of Georgia, Athens, GA. Previously presented in part at the XIX Symposium of the European Society of Veterinary Neurology, Barcelona, Spain, September 2930, 2006. Reprint requests: Luisa De Risio, Neurology/Neurosurgery Unit, Centre for Small Animal Studies, Animal Health Trust, Lanwades Park, Kentford, Newmarket, Suffolk, CB8 7UU, UK; e-mail: luisa.derisio@aht.org.uk. Submitted November 15, 2006; Revised February 1, 2007, April 23, 2007; Accepted May 18 2007. Copyright E 2007 by the American College of Veterinary Internal Medicine 0891-6640/07/2106-0019/$3.00/0

sequent retrograde entrance into the vertebral venous sinuses.30 In addition, it has been suggested that fibrocartilage may arise from metaplasia of the vascular endothelium that later ruptures and eventually embolizes within the spinal cord vessels.30 Ischemic myelopathy also may result from material other than fibrocartilage that obstructs the intrinsic spinal vascularization, such as thrombi or bacterial, parasitic, or neoplastic emboli.30 Dogs with ischemic myelopathy typically present with an acute onset of nonpainful, nonprogressive (after the 1st 24 hours), and often asymmetric clinical signs. Definitive diagnosis can only be made by histopathology of the affected spinal cord segment. Antemortem diagnosis of ischemic myelopathy is based on history, clinical findings, and exclusion of other causes utilizing survey radiographs, cerebrospinal fluid (CSF) analysis, myelography, computed tomography, and magnetic resonance imaging (MRI).18,30 MRI is particularly helpful in supporting the antemortem diagnosis of ischemic myelopathy because, in addition to excluding other causes of myelopathy, findings may reveal signalintensity changes compatible with spinal cord infarction.23,24,3140 The MRI features of ischemic myelopathy have been described in the human literature and in a small number of cases in the veterinary literature.23,24,3140 These MRI findings include a focal, relatively sharply demarcated intramedullary lesion, involving predominantly the gray matter that appears hyperintense to normal spinal cord gray matter on T2weighted fast spin-echo (FSE) images and iso- or hypointense to normal spinal cord gray matter on T1weighted FSE images. Various degrees of contrast enhancement may be present on T1-weighted FSE images, depending on the time interval between onset of clinical signs and MRI.30,3440 Timing of MRI also may influence the presence of MRI changes. Indeed in

MRI and Ischemic Myelopathy

1291

humans, it has been reported that conventional MRI may not allow detection of the ischemic spinal cord lesion in the 1st 48 hours after the onset of neurologic signs.3,5,3638 Ischemic myelopathy recently has been confirmed histopathologically in dogs with normal findings on MRI that were imaged in the acute stages of the disease.a Diffusionweighted (DW) MRI may increase the sensitivity and specificity for diagnosis of spinal cord infarction in the early stages of the disorder.41,42 The aims of this study were to describe the clinical and diagnostic findings in dogs with a presumptive diagnosis of ischemic myelopathy and to identify associations among MRI findings (eg, presence, site, lateralization, and extent of the lesion), timing of imaging, and presenting neurologic deficits (eg, localization, lateralization, and severity of signs).

Materials and Methods

The medical records and MR images of dogs with a presumptive diagnosis of ischemic myelopathy that presented at the Animal Health Trust between January 1, 2000, and June 30, 2006, were reviewed. Inclusion criteria were acute (,24 hours) onset of nonprogressive and nonpainful myelopathy, 1.5-tesla MRI of the spine performed within 7 days of onset, complete medical records, and follow-up information. Presumptive diagnosis was based on history and clinical, CSF, and MRI findings consistent with those previously reported for nontraumatic ischemic myelopathy.23,24,3040,a Dogs with no detectable changes on MRI, and history, clinical, and CSF findings consistent with ischemic myelopathy also were included. The following information was retrieved from the medical records: age; sex; breed; body weight; physical activity before onset of signs; onset and course of clinical signs; medications received before presentation; general physical and neurologic examination findings; the results of CBC, serum biochemistry profile, coagulation profile, thyroid panel, blood pressure measurement, urinalysis, serology for Toxoplasma gondii and Neospora caninum, thoracic radiographs, abdominal ultrasound examination, CSF analysis, CSF polymerase chain reaction (PCR) for canine distemper virus, T gondii, and N caninum; and the time interval between onset of signs and MRI. Cases were categorized into 6 groups according to the time from onset of clinical signs to MRI: ,12 hours, 1324 hours, 2548 hours, 4972 hours, 73120 hours, and 121168 hours after clinical onset. On the basis of neurologic

findings, dogs were assigned a neurologic score from 0 (normal) to 5 (most severe degree of neurologic dysfunction) (Table 1). MRI was performed with a 1.5-tesla Signa EchoSpeed MRI scanner.b T2-weighted FSE images were acquired in the dorsal, sagittal, and transverse planes in all dogs. T1-weighted FSE, contrast-enhancedc T1-weighted FSE, and T2* gradient-echo (GRE) images were obtained in 1 or more planes in the majority of dogs, at the discretion of the radiologist. MRI was performed from C1 to T2 vertebrae in dogs with a cervical or cervicothoracic neurolocalization and from T1 to S3 vertebrae in dogs with a thoracolumbar or lumbosacral neurolocalization. All MR images were reviewed to a consensus by 2 board-certified radiologists, blinded to the clinical findings. Images were evaluated on a work station or computer with proprietary software.d Dogs with MRI findings not consistent with those reported for nontraumatic ischemic myelopathy in veterinary and human literature were excluded. Specific criteria were established in order to exclude dogs with MRI findings consistent with acute noncompressive nucleus pulposus extrusion and spinal cord concussion. These criteria included the presence of focal hyperintensity in the spinal cord overlying an intervertebral disk with reduction in volume and signal intensity of the nucleus pulposus on T2-weighted MR images, reduced intervertebral disk space with evidence of ruptured dorsal annulus on T2* GRE images, and the presence of epidural fat disruption and extraneous material within the vertebral canal dorsal to the abnormal disk with minimal or absent spinal cord compression. In all dogs in which both radiologists agreed on the presumptive radiologic diagnosis of ischemic myelopathy, they assessed the following MRI variables: presence, site, lateralization, and extent of the intramedullary hyperintense lesion on sagittal and transverse T2-weighted images. The extent of the lesion on MRI was defined as the following: (1) the ratio between the length of the lesion (hyperintense area on sagittal T2-weighted images) and the length of either the C6 (cervical lesions) or the L2 (thoracolumbar lesions) vertebral body (Figs 1A,B; 3A,B), and (2) the maximal cross-sectional area of the lesion (hyperintense area on transverse T2-weighted images) as a percentage of the cross-sectional area of the spinal cord (Figs 2A,B; 4A,B). In addition, the presence of spinal cord swelling at the level of the lesion and the degree of involvement of gray matter, white matter, or both were assessed in T2-weighted FSE images. Spinal cord swelling was diagnosed when the subarachnoid space was attenuated. Presence and signal characteristics of the lesion in T1-weighted FSE and T2* GRE images were reviewed. The presence of contrast enhancement was recorded in dogs that underwent contrast studies.

Table 1.

Neurologic scoring by area of neurolocalization.

Neurolocalization

Score 0 1 2 3 4

C1-C5 or C6-T2 normal minimal gait abnormalities* ambulatory hemi or tetraparesis non ambulatory hemi or tetraparesis 6 monoplegia non ambulatory hemiplegia 6 tetraparesis tetraplegia

T3-L3 normal minimal gait abnormalities* ambulatory mono or paraparesis non ambulatory paraparesis or monoplegia paraplegia 6 urinary and/or faecal incontinence 6 tail plegia paraplegia 6 urinary and/or faecal incontinence 6 tail plegia loss of nociception

L4-S3 normal minimal gait abnormalities* ambulatory mono or paraparesis non ambulatory paraparesis or monoplegia paraplegia 6 urinary and/or faecal incontinence, normal/decreased perineal reflex, tail plegia 6 loss of nociception (only of the tail) paraplegia loss of nociception in one or both hind limbs, 6 urinary and/or faecal incontinence, absent perineal reflex, tail plegia and loss of nociception

*mild ataxia 6 proprioceptive deficits 6 mild paresis.

1292

De Risio et al

Fig 1. Unmarked (A) and marked (B) sagittal T2-weighted fast spin-echo images of the C4C7 vertebral segments. The length of the hyperintense area and the length of C6 vertebral body are outlined as indicated (B), and the computer softwared calculated these lengths in millimeters. In this case, the ratio between the length of the hyperintense area and the length of C6 vertebral body is 41/12 5 3.4. The site of CSF collection and the results of CSF analysis (eg, red blood cell count, total and differential white blood cell [WBC] count, protein content) were recorded. CSF protein content was considered normal with ,0.25 g/L for cisternal samples and ,0.40 g/L for lumbar samples. In case of pleocytosis (.5 WBCs per microliter), the increased cell count was graded as mild (550 cells), moderate (50200 cells), or marked (.200 cells).43

Fig 2. Unmarked (A) and marked (B) transverse T2-weighted fast spin-echo images at the level of maximal transverse extent of the lesion shown in Fig 1. The cross-sectional area of the hyperintensity and the cross-sectional area of the spinal cord are outlined manually as indicated (B). The computer softwared calculated the areas of the outlined regions (hyperintensity, 42 mm2; spinal cord, 54 mm2). In this case, the maximal crosssectional area of the lesion is 77.8% of the cross-sectional area of the spinal cord. used to examine associations among the presence of MRI abnormalities, time interval between onset of signs, MRI and CSF analysis (normal or abnormal), and severity of clinical signs on presentation. Dogs were grouped according to whether they were ambulatory (neurologic score, 02) or nonambulatory (neurologic score, 35) on presentation. Results are presented as relative risk (RR). Cross-tabulations and Kappa statistics as a test of agreement were used to examine associations between clinical neurolocalization and site of lesion on MRI and clinical and MRI asymmetry of the lesion. Spearman rank correlation coefficients

Statistical Analysis
Descriptive statistics for continuous variables are presented as median (minimum, maximum) values with frequencies reported for categoric variables. Cross-tabulations and Fisher exact tests were

MRI and Ischemic Myelopathy

1293

Fig 3. Unmarked (A) and marked (B) sagittal T2-weighted fast spin-echo images from the caudal aspect of T13 to the cranial aspect of L4 vertebral segments. The length of the hyperintense area and the length of L2 vertebral body are outlined as indicated (B). In this case, the ratio between the length of the hyperintense area and the length of L2 vertebral body is 22/25 5 0.88. were used to identify associations among the severity of clinical signs on presentation, the ratio of length of the lesion and the length of C6 or L2 vertebral body, and the maximal cross-sectional area of the lesion as a percentage of the cross-sectional area of the spinal cord. Wilcoxon rank sum tests were used to examine the association of CSF analysis (normal or abnormal) with the extent of the lesion on MRI (ratio of the length of the lesion and the length of C6 or L2 vertebral body and maximal cross-sectional area of the lesion as a percentage of the cross-sectional area of the spinal cord). The level of significance was set at P , .05 in all instances. Results are presented with 95% confidence intervals (CI) where applicable.

Fig 4. Unmarked (A) and marked (B) transverse T2-weighted fast spin-echo images at the level of the maximal transverse extent of the lesion shown in Fig 3. In this case, the maximal crosssectional area of the lesion is 60% of the cross-sectional area of the spinal cord.

Results
The medical records and MR images of 78 dogs with a presumptive diagnosis of ischemic myelopathy at the Animal Health Trust between January 1, 2000, and June

30, 2006, were reviewed. Fifty-two dogs met the inclusion criteria. Of the 26 dogs that were excluded, 20 had MRI findings consistent with acute noncompressive nucleus pulposus extrusion and spinal cord concussion, 2 had incomplete medical records, 2 had incomplete MRI studies, and 2 had concurrent spinal disorders (syringohydromyelia, intervertebral disk protrusions). Ten of these 52 dogs had been reported in a previous study.40 Breeds included the following: Labrador Retriever (14), German Shepherd dog (5), Miniature Schnauzer (4), Boxer (4), Staffordshire Bull Terrier (4), Golden Retriever (3), Irish Wolfhound (2), Yorkshire Terrier (2), English Bull terrier (2), and 1 each

1294

De Risio et al

of Basset Hound, Bearded Collie, border collie, Bullmastiff, Chesapeake Bay Retriever, collie cross, Giant Schnauzer, Greyhound, Jack Russell terrier, lurcher, Newfoundland, and rough collie. The median age at diagnosis was 6 years (range, 3 months to 11 years 11 months). Fifteen dogs were female (8 of which were spayed), and 37 were male (15 of which were neutered). The median body weight was 30 kg (range, 2.570 kg). All 52 dogs had acute (,24 hours) onset of neurologic signs. Within the 1st 24 hours after onset, the signs were reported by the owner to be static in 34 (65%) dogs and progressive in 18 dogs (35%). In the 28 dogs that presented in the 1st 24 hours after onset, the owners reported improvement in 14 dogs and no changes since onset in 14 dogs. In 15 (29%) dogs, the owners reported that the dog was performing some form of physical activity (eg, walking, running, playing) at the time of onset. In 8 (15%) dogs, onset of signs did not seem to be associated with any type of physical activity. In the remaining 29 (56%) dogs, there was no information in the medical record on physical activity at the time of onset. Treatment before presentation included corticosteroids (prednisolone, dexamethasone, betamethasone, or methylprednisolone sodium succinate) in 26 dogs, nonsteroid anti-inflammatory drugs (NSAIDs) such as carprofen or meloxicam in 12 dogs, and corticosteroids and NSAIDs in 6 dogs. Ten dogs also received a single administration of antibiotics. Eight patients had no treatment before presentation. None of the patients were continued on corticosteroids or NSAIDs once admitted to the Animal Health Trust. Oral antibiotic treatment (clavulanated amoxicillin) was continued for 2 weeks only in 1 dog with an unrelated tail wound. The median neurologic score at presentation was 3 (range, 25). Neurologic score and neurolocalization on presentation are summarized in Table 2. Of the 14 dogs with a T3-L3 neurolocalization, 4 had T3-L3 signs (cutoff of the cutaneous trunci reflex in the thoracolumbar region, normal-to-increased patellar reflex, and normal tail tone and perineal reflex) and concurrent L6-S1 signs (decreased-to-absent withdrawal in 1 or both hind limbs). These 4 dogs were severely paraparetic or paraplegic with intact nociception. MRI findings revealed signal changes consistent with ischemic myelopathy (focal intramedullary lesion, hyperintense on T2-weighted and hypo- or isointense on Table 2. Neurologic score of 52 dogs by neurolocalization.
Neurolocalization Neurologic Score 2 3 4 5 Total C6-T2 0 6 8 1 15 T3-L3 6 5 3 0 14a L4-S3 5 7 9 2 23 Total 11 18 20 3 52

Table 3. Time interval between the onset of signs and MRI.

Number of Dogs with Time Interval (h) ,12 1224 2548 4972 73120 121168 Total Normal MRI 3 3 2 3 0 0 11 Abnormal MRI 4 14 13 4 2 4 41 Total 7 17 15 7 2 4 52

MRI, magnetic resonance imaging.

T1-weighted images) in 41 dogs and did not reveal any abnormalities in the remaining 11 dogs (Table 3). No statistically significant association could be identified between the presence of MRI abnormalities and the time interval between onset of signs and MRI (Fishers exact test, P 5 .3). By grouping dogs according to whether they were ambulatory or nonambulatory on presentation, an association between the severity of clinical signs on presentation and the presence of MRI abnormalities was identified (Table 4). Nonambulatory dogs with a neurologic score $3 were 1.6 times more likely to have abnormal MRI findings compared with ambulatory dogs with a score of 2 (RR 5 1.6; 95% CI, 0.90 2.72; Fishers exact test, P 5 .04). In the 41 dogs with abnormal MRI findings, a hyperintense lesion on T2-weighted images was identified in the cervical or cervicothoracic spinal cord in 15 dogs and in the thoracolumbar or lumbosacral spinal cord in 26 dogs. In these 41 dogs, the site of the lesion identified by MRI agreed with clinical neurolocalization in 32 dogs (78%; 95% CI, 6289). Of the remaining 9 dogs, MRI identified the lesion in either the C1-C5 (2 dogs) or the T3-L3 (7 dogs) spinal cord segments, whereas clinical examination findings revealed decreased withdrawal in the front limbs (2 dogs) or in the hind limbs (7 dogs), respectively. In 3 of the 4 dogs with T3-L3 neurolocalization and decreased withdrawal reflex in 1 or both hind limbs, the site of the MRI lesion was consistent with the site of the cutaneous trunci reflex cut-off (T12, T13, and L1, respectively), whereas the 4th dog had normal findings on MRI. In all 9 dogs with discrepancy between clinical and MRI localization of the lesion, the radiologist carefully re-reviewed the MR Table 4. The relationship between presence of MRI abnormalities and neurologic status (ambulatory status) on presentation.
Neurologic Score on Presentation $3 5 nonambulatory 2 5 ambulatory Total MR Images Abnormal 35 6 41 Normal 6 5 11 Total 41 11 52

a Four of these 14 dogs also had a decreased withdrawal in 1 or both hind limbs.

MRI, magnetic resonance imaging.

MRI and Ischemic Myelopathy

1295

images in order to identify a possible subtle 2nd lesion within the intumescences, but such a lesion was not identified in any of these dogs. Clinical neurolocalization had revealed a moderate level of agreement with the site of the lesion identified by MRI (k 5 0.52; 95% CI, 0.220.81; n 5 41; P 5.001). Neurologic signs were symmetric in 7 dogs and asymmetric in 45 dogs with lateralization to the right in 18 dogs and to the left in 27. Of the 41 dogs with abnormal findings on MRI, the lesion was symmetric in 10 dogs and asymmetric in 31 dogs; and, of these, 18 lateralized to the left and 13, to the right. Clinical and MRI asymmetry of the lesion was consistent in all but 3 dogs and had very good statistical agreement (k 5 0.80; 95% CI, 0.481.00; n 5 41; P , .0001). In these 3 dogs, clinical lateralization was very subtle (eg, minimal voluntary motor activity in 1 limb), and the radiologists could not identify a specific side in which the lesion was more prominent. In the 15 dogs with a cervical or cervicothoracic lesion, the median ratio of the length of the hyperintense lesion on sagittal T2-weighted images and the length of the C6 vertebral body was 1.8 (range, 0.57.8). In the 26 dogs with a thoracolumbar or lumbosacral lesion, the median ratio between the length of the hyperintense lesion on sagittal T2-weighted images and the length of the L2 vertebral body was 2.2 (range, 0.66.7). The ratio of the length of the hyperintense lesion on sagittal T2weighted images and the length of C6 or L2 vertebral body was associated with the severity of clinical signs on presentation (Spearman r 5 0.37, P 5 .02). The maximal cross-sectional area of the hyperintense lesion on transverse T2-weighted images as a percentage of the cross-sectional area of the spinal cord ranged from 20% to 100% with a median of 75% and was associated with the severity of clinical signs on presentation (Spearman r 5 0.43, P 5.02). T2-weighted studies revealed spinal cord swelling in 36 of 41 dogs with abnormal MRI findings. In all 41 dogs with abnormal findings on MRI, the lesion involved the gray matter of the spinal cord, and in 36 of these, there also was involvement of the adjacent white matter. Of the 41 dogs with a hyperintense lesion on T2-weighted images, findings of T1-weighted FSE studies revealed isointensity of the lesion in 39 dogs and hypointensity in 2 dogs that underwent MRI 24 and 60 hours after onset, respectively. GRE studies were performed in 27 dogs; the lesion was isointense to normal spinal cord gray matter in 17 (63%) and hyperintense in 10 (37%). In 4 of these 10 dogs, findings of GRE studies also revealed patchy areas of decreased signal, suggestive of hemorrhage. Contrast-enhanced T1-weighted MRI studies were performed in 40 dogs. Five of these had various degrees of contrast enhancement of the lesion. Time intervals between onset of signs and MRI in these 5 dogs were 24 hours in 2 dogs; and 48 hours, 4 days, and 6 days in each of the other 3 dogs. CBC and serum biochemistry profiles were performed in all dogs and did not disclose any specific abnormalities. Additional tests included coagulation profiles

(3 dogs), thyroid panels (4 dogs), blood pressure measurement (5 dogs), urinalysis (10 dogs), and chest radiographs and abdominal ultrasound examination (3 dogs), of which all findings were considered within normal limits. Serology for T gondii and N caninum was performed in 4 dogs, and all were seronegative. CSF analysis was done in 32 dogs (61.5%). CSF was collected from the cerebellomedullary cistern in 5 dogs, from the lumbar subarachnoid space in 26 dogs, and from both sites in 1 dog. CSF was normal in 17 dogs, abnormal in 14, and contaminated by blood in 1 dog. Of the 14 dogs with abnormal CSF, all 14 had increased total protein content (range, 0.502.39 g/L; median, 0.83 g/L), and 6 had pleocytosis (784 WBC per microliter; median, 12 WBC per microliter). Of these 6 dogs, 3 had mild neutrophilic pleocytosis, 1 had moderate neutrophilic pleocytosis, and 2 had mild mixed pleocytosis. In 1 dog, blood contamination prevented diagnostic interpretation. CSF PCR for canine distemper virus, T gondii, and N caninum was performed in 4 of the 6 dogs with pleocytosis, and all were negative. The other 2 dogs with CSF pleocytosis were seronegative for T gondii and N caninum. CSF analysis was normal in all dogs with normal findings on MRI. All dogs with abnormal CSF also had abnormal MRI findings. The presence of CSF abnormalities was associated with the presence of MRI abnormalities. Dogs with abnormal CSF were more likely to have abnormal MRI findings (RR 5 1.42; 95% CI, 1.041.93; P 5 .048). The presence of CSF abnormalities was not associated with the extent of the lesion on MRI (P . .2). Only 2 of the dogs in the current study have undergone postmortem examination, and in both dogs, the MRI findings were abnormal, and histopathology disclosed spinal cord ischemic infarction secondary to fibrocartilaginous embolization.

Discussion
The breed, age, history, onset, and progression of clinical signs in the current series of dogs are consistent with previous veterinary reports on ischemic myelopathy.618,30 The male-to-female ratio was 2.5 : 1 in our case series. A similar sex distribution has been reported previously in 1 study of Miniature Schnauzers with ischemic myelopathy.17 However, a predisposition for males has not been observed in other reports on dogs with ischemic myelopathy.14,18,30 Interestingly, in our study, the youngest dog was a 3-month-old Irish Wolfhound. Ischemic myelopathy due to fibrocartilaginous embolization has been histopathologically confirmed in dogs of this breed as young as 8 weeks old.44 Of the 52 dogs included in our study, 2 were Yorkshire Terriers and 4 were Miniature Schnauzers. Ischemic myelopathy after fibrocartilaginous embolization has been reported most commonly in large-breed dogs; however, this disorder also has been described in small-breed dogs and particularly in Miniature Schnauzers.6,12,17,39,45,46 The antemortem diagnosis of ischemic myelopathy is one of exclusion.30 The role of MRI in supporting the clinical suspicion of ischemic myelopathy has been

1296

De Risio et al

reported in the human literature and in a small number of cases in the veterinary literature.23,24,3140,a In our study, we have described the MRI findings of 52 dogs with a presumptive diagnosis of ischemic myelopathy, 2 of which later had a histologically confirmed diagnosis. Eleven of these dogs, imaged between 12 and 72 hours (median, 1224 hours) after onset of signs, had normal MRI findings. This suggests that conventional MRI may not allow detection of the ischemic lesion in the early stages of the disease. Indeed, in the human literature, it is widely accepted that in the acute phase of ischemic myelopathy, findings of conventional MRI may be normal.3,5,3638 In addition, a recent report has described histopathologically confirmed ischemic myelopathy in dogs with normal MRI findings.a Detection of abnormal signal changes during the 1st 2448 hours is dependent on the size of the infarction, the degree of ischemic insult, and the availability of high-contrastresolution MRI with minimal artifact from patient motion.37 DW MRI may increase the sensitivity and specificity for diagnosis of spinal cord infarction in the early stages of the disorder.41,42,47,48 However, DW MRI of the spine presents a technical challenge because of the relatively small size of the spinal cord, low spatial resolution, and the magnetic susceptibility to artifacts caused by vascular and CSF pulsation.37,45,47 Serial MRI is widely used in human medicine to detect an ischemic lesion that was not apparent in the acute stage of the disease.3338 Repeated MRI studies also may have a role in veterinary medicine; however, the need for repeated anesthesia is a limiting factor. In the present study, no statistically significant association could be identified between the presence of an abnormal MRI findings and the time interval between onset of signs and MRI. This is most likely due to the small number of dogs in each time-related group and associated lack of statistical power. The severity of clinical signs and particularly the ambulatory status on presentation were associated with the presence of MRI abnormalities. This finding can be explained given that detection of MRI changes is dependent on the size of the infarction and the degree of ischemic insult37 and assuming that severity of clinical signs is related to the size of the ischemic lesion. To the best of our knowledge, no studies on ischemic myelopathy previously have looked at the association between the severity of neurologic signs on presentation and the extent of the lesion on MRI. In our study, we also found a small but statistically significant correlation of extent of the lesion on both sagittal and transverse T2-weighted MR images with the severity of clinical signs on presentation. Another interesting finding of our study was the discrepancy between clinical and MRI localization of the lesion in 9 dogs. In all these 9 dogs, MRI identified the lesion in either the C1-C5 (2 dogs) or the T3-L3 (7 dogs) spinal cord segments, but the clinical neurolocalization was either to the cervical (2 dogs) or to the lumbar (7 dogs) intumescences on the basis of a decreased withdrawal reflex in the front or hind limbs, respectively. Interestingly, 3 of the 7 dogs with decreased withdrawal

reflex in 1 or both hind limbs and a MRI lesion within the T3-L3 region had an interruption of the cutaneous trunci reflex consistent with the site of the MRI lesion, normalto-increased patellar reflexes, and normal perineal reflexes. The presence of a lesion extending from either the T12, T13, or L1 spinal cord segments to the L6-S1 region, sparing the L4-5 lower motor neurons and not apparent on MRI, is possible, but unlikely. Another possible explanation for these clinical findings is the presence of multiple ischemic events in different regions of the spinal cord. Concurrent brain stem and spinal cord fibrocartilaginous emboli have been documented histopathologically in a sheep, in a dog, and in humans.27,46,49 In addition, multiple fibrocartilaginous emboli determining ischemic lesions of different severity, extending from the thoracic to the sacral spinal cord segments, have been described.6,9 A careful review of the MR images of our cases did not reveal any signs of a possible concurrent lesion within the intumescences. However, normal findings on imaging do not necessarily exclude an ischemic event, especially a small one.37,a Fourteen of the 32 dogs that underwent CSF analysis in our study had increased protein concentrations, and 6 of these 14 dogs also had neutrophilic or mixed pleocytosis (median, 12 WBC per microliter). Similar CSF findings have been reported previously in dogs with presumptive or histologically confirmed ischemic myelopathy.16,18,30 No previous study, however, has reported an association between the presence of CSF abnormalities and the presence of MRI changes consistent with ischemic myelopathy. In dogs with nonspecific CSF changes and suspected ischemic myelopathy based on history and clinical findings, negative PCR on CSF and MRI changes consistent with ischemic myelopathy can help support the antemortem diagnosis of this disease.

Footnotes
a

Stein VM, Wagner F, Bull C, et al. Findings of magnetic resonance imaging in suspected canine fibro-cartilaginous embolization. European Society of Veterinary Neurology XIX Symposium, Barcelona, Spain, September 2930, 2006:157 (abstract) b 1.5-tesla Signa EchoSpeed MRI, GE Healthcare, Milwaukee, WI c Gadolinium: Omniscan (gadodiamide), Nycomed, Oslo, Norway or MultiHance (gadobenate dimeglumine); Bracco, Milan, Italy d e-Film, Merge eMed, Milwaukee, WI

Acknowledgment
The work was done at the Centre for Small Animal Studies, Animal Health Trust, Newmarket, UK.

References
1. Feigin I, Popoff N, Adachi M. Fibrocartilaginous venous emboli to the spinal cord with necrotic myelopathy. J Neuropathol Exp Neurol 1965;24:6374.

MRI and Ischemic Myelopathy 2. Bocknek WL, Bach JR, Alba AS, Cravioto HM. Fibrocartilaginous emboli to the spinal cord: A case report. Arch Phys Med Rehabil 1990;71:754757. 3. McLean JM, Palagallo GL, Henderson JP, et al. Myelopathy associated with fibrocartilaginous emboli (FE): Review and two suspected cases. Surg Neurol 1995;44:228235. 4. Tosi L, Rigoli G, Beltramello A. Fibrocartilaginous embolism of the spinal cord: A clinical and pathogenetic reconsideration. J Neurol Neurosurg Psychiatry 1996;60:5560. 5. Han JJ, Massagli TL, Jaffe KM. Fibrocartilaginous embolism: An uncommon cause of spinal cord infarctionA case report and review of the literature. Arch Phys Med Rehabil 2004;85: 153157. 6. Griffiths IR. Spinal cord infarction due to emboli arising from the intervertebral disc in the dog. J Comp Pathol 1973;83:225232. 7. Zaki FA, Prata RG, Willian JK. Necrotizing myelopathy in five great Danes. J Am Vet Med Assoc 1974;165:10801084. 8. Zaki FA, Prata RG. Necrotizing myelopathy secondary to embolization of herniated intervertebral disk material in the dog. J Am Vet Med Assoc 1976;169:222228. 9. DeLahunta A, Alexander JW. Ischemic myelopathy secondary to presumed fibrocartilaginous embolism in nine dogs. J Am Anim Hosp Assoc 1976;12:3748. 10. Hayes MA, Creighton R, Boysen BG, Holfield N. Acute necrotizing myelopathy from nucleus pulposus embolism in dogs with intervertebral disk degeneration. J Am Vet Med Assoc 1978;173:289295. 11. Gill CW. Fibrocartilaginous embolic myelopathy in a dog. Can Vet J 1979;20:273278. 12. Kornegay JN. Ischemic myelopathy due to fibrocartilaginous embolism. Comp Cont Educ 1980;2:402405. 13. Bischel P, Vandevelde M, Lang J. Linfarctus de la moelle epiniere a la suite dembolies fibrocartilagineuses chez le chien et le chat. Schweiz Arch Tierheilk 1984;126:387397. 14. Gilmore DR, deLahunta A. Necrotizing myelopathy secondary to presumed or confirmed fibrocartilaginous embolism in 24 dogs. J Am Anim Hosp Assoc 1986;23:373376. 15. Dyce J, Houlton JEF. Fibrocartilaginous embolism in the dog. J Small Anim Pract 1993;34:332336. 16. Cauzinille L, Kornegay JN. Fibrocartilaginous embolism of the spinal cord in dogs: Review of 36 histologically confirmed cases and retrospective study of 26 suspected cases. J Vet Intern Med 1996;10:241245. 17. Hawthorne JC, Wallace LJ, Fenner WR, et al. Fibrocartilaginous embolic myelopathy in miniature schnauzers. J Am Anim Hosp Assoc 2001;37:374383. 18. Gandini G, Cizinauskas S, Lang J, et al. Fibrocartilaginous embolism in 75 dogs: Clinical findings and factors influencing the recovery rate. J Small Anim Pract 2003;44:7680. 19. Zaki FA, Prata RG, Werner LL. Necrotizing myelopathy in a cat. J Am Vet Med Assoc 1976;169228229. 20. Turner PV, Percy DH, Allyson K. Fibrocartilaginous embolic myelopathy in a cat. Can Vet J 1995;36:712713. 21. Scott HW, OLeary MT. Fibrocartilaginous embolism in a cat. J Small Anim Pract 1996;37:228231. 22. Abramson CJ, Platt SR, Stedman NL. Tetraparesis in a cat with fibrocartilaginous emboli. J Am Anim Hosp Assoc 2002;38:153156. 23. MacKay AD, Rusbridge C, Sparkes AH, et al. MRI characteristics of suspected acute spinal cord infarction in two cats, and a review of the literature. J Feline Med Surg 2005;7:101107. 24. Mikszewski JS, Van Winkle TJ, Troxel MT. Fibrocartilaginous embolic myelopathy in five cats. J Am Anim Hosp Assoc 2006;42:226233.

1297

25. Taylor HW, Vandevelde M, Firth EC. Ishemic myelopathy by fibrocartilaginous emboli in a horse. Vet Pathol 1977;14: 479481. 26. Tessaro SV, Doige CE, Rhodes CS. Posterior paralysis due to fibrocartilaginous embolism in two weaner pigs. Can J Comp Med 1983;47:124126. 27. Jeffery M, Wells GA. Multifocal ischemic encephalomyelopathy associated with fibrocartilaginous emboli in the lamb. Neuropathol Appl Neurobiol 1986;12:231244. 28. Stedman NL, Brown TP, Rowland GN. Intravascular cartilaginous emboli in the spinal cord of turkeys. Avian Dis 1998;42:423428. 29. Renner MS, Bryant W, Kennedy GA. Fibrocartilaginous emboli in a tayra (Eira barbara): A case report. J Zoo Wildl Med 1998;29:470473. 30. Cauzinille L. Fibrocartilaginous embolism in dogs. Vet Clin North Am Small Anim Pract 2000;30:155167. 31. Brown E, Virapongse C, Gregorios JB. MR imaging of cervical cord infarction. J Comput Assist Tomogr 1989;13:920922. 32. Nagashima C, Nagashima R, Morota N, Kobayashi S. Magnetic resonance imaging of human spinal cord infarction. Surg Neurol 1991;35:368373. 33. Hirono H, Yamadori A, Komiyama M, et al. MRI of spontaneous spinal cord infarction: Serial changes in gadoliniumDTPA enhancement. Neuroradiology 1992;34:9597. 34. Fortuna A, Ferrante L, Acqui M, Trillo G. Spinal cord ischemia diagnosed by MRI. J Neuroradiol 1995;22:115122. 35. Simonson TM, Yuh WT. Stroke and cerebral ischemia. In: Edelman RR, Zlatkin MB, Hesselink JR, eds. MRI Clinical Magnetic Resonance Imaging, Vol. 1. 2nd ed. Philadelphia, PA: WB Saunders; 1996:767786. 36. Weidauer S, Nichtweiss M, Lanfermann H, Zanella FE. Spinal cord infarction: MR imaging and clinical features in cases. Neuroradiology 2002;44:851857. 37. Luo CB, Chang FC, Teng MM, et al. Magnetic resonance imaging as a guide in the diagnosis and follow-up of spinal cord infarction. J Chin Med Assoc 2003;66:8995. 38. Masson C, Pruvo JF, Cordonneir C, et al. Spinal cord infarction: Clinical and magnetic resonance imaging findings and short term outcome. J Neurol Neurosurg Psychiatry 2004;75: 14311435. 39. Grunenfelder FI, Weishaupt D, Green R, et al. Magnetic resonance imaging findings in spinal cord infarction in three small breed dogs. Vet Radiol Ultrasound 2005;46:9196. 40. Abramson CJ, Garosi L, Platt SR, et al. Magnetic resonance imaging appearance of suspected ischemic myelopathy in dogs. Vet Radiol Ultrasound 2005;46:225229. 41. Zhang J, Huan Y, Qian Y, et al. Multishot diffusionweighted imaging features in spinal cord infarction. J Spinal Disord Tech 2005;18:277282. 42. Fujikava A, Tsuchiya K, Koppera P, et al. Case report: Spinal cord infarction demonstrated on diffusion-weighted MR imaging with a single-shot fast spin-echo sequence. J Comput Assist Tomogr 2003;27:415419. 43. Tipold A. Cerebrospinal fluid. In: Braund KG, ed. Clinical Neurology in Small Animals: Localization, Diagnosis and Treatment. Ithaca, NY: International Veterinary Information Service; 2003. Available at: www.ivis.org. Accessed September 1, 2006. 44. Junker K, van den Ingh T, Bossard MM, et al. Fibrocartilaginous embolism of the spinal cord (FCE) in juvenile Irish Wolfhounds. Vet Q 2000;22:154156. 45. Ueno H, Shimizu J, Uzuka Y, et al. Fibrocartilaginous embolism in a chondrodystrophoid breed dog. Aust Vet J 2005;83:142144. 46. Axlund TW, Isaacs AM, Holland M, et al. Fibrocartilaginous embolic encephalomyelopathy of the brainstem and

1298

De Risio et al 48. Shinoyama M, Takahashi T, Shimizu H, et al. Spinal cord infarction demonstrated by diffusion-weighted magnetic resonance imaging. J Clin Neurosci 2005;12:466468. 49. Kepes JJ, Reynard JD. Infarction of spinal cord and medulla oblongata caused by fibrocartilaginous emboli: A case report. Virchows Arch 1973;361:185193.

midcervical spinal cord in a dog. J Vet Intern Med 2004;18: 765767. 47. Kuker W, Weller M, Klose U, et al. Diffusion-weighted MRI of spinal cord infarction-high resolution imaging and time course of diffusion abnormality. J Neurol 2004;251: 818824.