• Open Access

Intramedullary Spinal Cord Neoplasia in 53 Dogs (1990–2010): Distribution, Clinicopathologic Characteristics, and Clinical Behavior


  • T.E. Pancotto,

    Corresponding author
    1. Department of Small Animal Clinical Sciences, Virginia-Maryland Regional College of Veterinary Medicine, Blacksburg, VA
    • Corresponding author: Theresa E. Pancotto, DVM, MS, DACVIM, Department of Small Animal Clinical Sciences, Virginia-Maryland Regional College of Veterinary Medicine, 225 Duck Pond Drive, Mail code 0442, Blacksburg, VA 24061; e-mail: tepdvm@vt.edu.

    Search for more papers by this author
  • J.H. Rossmeisl Jr.,

    1. Department of Small Animal Clinical Sciences, Virginia-Maryland Regional College of Veterinary Medicine, Blacksburg, VA
    Search for more papers by this author
  • K. Zimmerman,

    1. Department of Biomedical Sciences and Pathobiology , Virginia-Maryland Regional College of Veterinary Medicine, Blacksburg, VA
    Search for more papers by this author
  • J.L. Robertson,

    1. Department of Biomedical Sciences and Pathobiology , Virginia-Maryland Regional College of Veterinary Medicine, Blacksburg, VA
    Search for more papers by this author
  • S.R. Werre

    1. Department of Biomedical Sciences and Pathobiology , Virginia-Maryland Regional College of Veterinary Medicine, Blacksburg, VA
    Search for more papers by this author



Intramedullary neoplasms of the canine spinal cord are infrequently reported.


To describe distribution, clinicopathologic characteristics, radiographic findings, and clinical features of canine intramedullary spinal tumors.


Retrospective series of histologically confirmed canine intramedullary spinal tumors. Contingency tables were generated for categorical variables (breed, sex, treatment, pain, chief complaint, localization, histology, imaging, and site). Associations were assessed by Fisher's exact, Wilcoxon rank sum test, t-test, and one-way ANOVA.


Intramedullary spinal cord tumors comprised 16% (53/331) of all tumors of the spinal cord. Primary tumors were diagnosed in 66% (35/53) of cases, with neuroepithelial-origin tumors comprising 51% (18/35) of all primary neoplasms. Intraparenchymal metastases of transitional cell carcinoma and hemangiosarcoma accounted for 66% (6/18 each) of all secondary tumors. Primary tumors were more likely to affect younger dogs. Dogs with intramedullary metastases were most commonly presented for primary myelopathic signs (8/18, 44%). The majority of all tumors (52.8%) occurred in the T3-L3 spinal cord segments. All dogs with cervical neurolocalization had primary tumors. Dogs with metastatic lesions had a shorter duration of clinical signs before presentation, but there was no difference in survival time between dogs with primary as compared with secondary tumors.


Intramedullary spinal cord tumors are uncommon. Primary intramedullary spinal cord tumors are more common than secondary intramedullary spinal cord tumors and tend to occur in the cervical spinal cord of younger dogs. Intramedullary metastases occur in older dogs, are rarely asymptomatic, and neurologic dysfunction is a common clinical presentation. Dogs with primary tumors may have a protracted clinical course compared with those with intramedullary metastases.







Intramedullary spinal cord tumors are described infrequently, and no recent studies provide data on the distribution, clinicopathologic characteristics, and clinical behavior of primary as compared with secondary intramedullary spinal cord tumors. Spinal cord tumors can be classified based on circumferential location relative to the dura and histopathology. Intramedullary (IM) tumors are least common, with a reported distribution of approximately 15%.[1-5] Extradural (ED) tumors are reported with an overall distribution of approximately 50% and intradural-extramedullary (ID-EM) tumors making up the remaining approximately 35%.[2-5] This circumferential distribution is similar in humans.[6, 7]

Intramedullary spinal cord tumors present a diagnostic and therapeutic challenge. Spinal cord tumors can cross from 1 compartment to another, making antemortem diagnosis difficult even with advanced imaging.[8] Intramedullary spinal cord tumors may not be readily amenable to surgery, although successful surgical removal has been described.[9-11] Intramedullary spinal cord tumors have also been reported as being nonpainful and rapidly progressive.[4, 5, 12-15] Historically, large breed dogs have been reported to be more susceptible to developing primary spinal neoplasms.[3, 16] No sex[4, 17-25] or age predisposition has been demonstrated,[4, 17, 26, 27] but there may be a tendency for younger animals to develop tumors of neural origin.[4] In children, intramedullary tumors (especially astrocytomas) may be more common than in adults.[6] A possible predisposition for intramedullary tumors to occur at the cervico-thoracic junction in dogs was identified previously in a single study.[4]

The goals of this study were to describe the distribution of histologically confirmed canine spinal tumors, and to report the age and breed of affected dogs for intramedullary tumors, as well as tissue of origin of the mass, vertebral level of the mass, radiographic findings, and clinical behavior. We hypothesized that primary neuroepithelial tumors would have a predilection to occur in the cervical spinal cord segments of young dogs and that metastatic tumors would occur most often in the thoracolumbar spinal cord segments. We also hypothesized that intramedullary spinal cord neoplasms would be associated with an acute and rapidly progressive clinical course and that clinical signs associated with spinal cord dysfunction would be common in dogs with metastatic tumors.

Materials and Methods

Medical records at the Virginia-Maryland Regional College of Veterinary Medicine from January 1990 to December 2010 were searched for dogs with histologically confirmed spinal cord tumors. For intramedullary spinal cord tumors, the following information was recorded: breed, body weight, age, sex, primary complaint, the presence or absence of pain on spinal palpation, neurolocalization, duration of signs before evaluation, treatment, survival time, spinal cord segment in which the tumor was located, primary versus metastatic origin, histologic tumor type, diagnostic imaging results, and location of additional tumor sites. Veterinary pathologists blinded to the original diagnoses reviewed H&E-stained slides from each case and assigned a morphologic diagnosis. In 7 cases, slides were unavailable for review; these cases were included based on descriptions and diagnoses provided in necropsy reports. In an additional 3 cases, the original pathology reports were unavailable for review or incomplete, but these cases were included because slides were available for review by study pathologists.

Intramedullary spinal cord tumors are tumors that occur within the parenchyma of the spinal cord, the medulla spinalis, and are referred to as intramedullary tumors in this study. Primary intramedullary spinal cord tumors are of neural origin, arise within the medulla spinalis, and are referred to as primary tumors. Secondary intramedullary spinal cord tumors are metastatic tumors that occur within the medulla spinalis and are referred to as secondary tumors.

Statistical Analysis

Normal probability plots showed that age was normally distributed, whereas body weight, duration of signs, and survival were skewed. Subsequently, age was summarized as mean (±standard deviation). Body weight, duration of signs, and survival days were summarized as medians (range). Contingency tables were generated for the categorical variables including breed, sex, treatment, the presence of hyperpathia during spinal palpation, primary complaint, localization, histologic type, diagnostic imaging modality, and tumor site.

Association between age and tumor origin (primary versus secondary) was tested by a 2-sample t-test, whereas the association between age and histologic type was tested by one-way analysis of variance. Associations between body weight or duration of clinical signs and tumor origin were tested by the Wilcoxon rank sum test. Associations between categorical variables (primary complaint and tumor origin, primary complaint and histologic type, location and tumor origin, location and histologic type, localization and tumor origin, breed and tumor origin, spinal pain and tumor origin) were assessed by Fisher's exact test. Statistical significance was set to alpha = .05. All analyses were performed by a commercial statistical software package.1


In total, 331 histologically confirmed tumors were identified involving the spinal cord. Extradural spinal cord tumors accounted for 61% (203/331), intradural-extramedullary spinal cord tumors 23% (76/331), and intramedullary spinal cord tumors 53/331 (16%). Intramedullary spinal cord neoplasms were diagnosed by necropsy in 48/53 cases and by surgical biopsy in 5/53 cases. Thirty-five tumors (66%) were primary tumors and 18 (34%) were secondary tumors. Ependymomas (n = 13) were the most common tumor and accounted for 25% of all intramedullary tumors and 37% of all intramedullary primary tumors. Other primary tumors included astrocytoma (n = 7), nephroblastoma (n = 4), chordoma (n = 4), oligodendroglioma (n = 3), hemangioma (n = 2), ganglioglioma (n = 1), and teratoma (n = 1). Metastatic transitional cell carcinoma (TCC) and hemangiosarcoma were most common among secondary tumors (6/18 each, 33%). Other secondary tumors included pheochromocytoma (n = 2), mammary carcinoma (n = 1), pancreatic carcinoma (n = 1), prostatic carcinoma (n = 1), and sarcoma of unknown etiology (n = 1).

There were 27 spayed females and 26 castrated males. Mixed breed dogs were most common 18/53 (34%). Mixed breeds weighed between 6 and 42 kg with a median of 12.5 kg. Other breeds included Labrador Retriever (4), Beagle (2), Boston Terrier (2), Chihuahua (2), Dalmatian (2), German Shepherd (2), Husky (2), Pug (2), and one of each of the following: Bassett Hound, Bloodhound, Border Collie, American Bulldog, Golden Retriever, Great Dane, Jack Russell Terrier, Lhasa Apso, Maltese, Mastiff, Miniature Poodle, Miniature Schnauzer, Pomeranian, Standard Poodle, Rat Terrier, Shih Tzu, and Yorkshire Terrier. There were no significant differences among the breed (P = 1.0), sex (P = .8431), or body weight (P = .1488) distributions of dogs with primary or secondary tumors (Table 1).

Table 1. Clinical characteristics of intramedullary spinal cord tumors
Tumor TypeaHistologic Tumor TypeNumber of CasesMean Age (years)Mean Body Weight (kg)Neurolocalization
  1. a

    1, Primary; 2, Secondary.

  2. b

    BW not recorded for 1 case.

  3. c

    BW not recorded for 3 cases.


















2Transitional cell carcinoma61132c









2Mammary carcinoma11012T3-L3(1)
2Pancreatic carcinoma1132T3-L3(1)
2Prostatic carcinoma1129L4-S3(1)
2Sarcoma of unknown etiology1928L4-S3(1)

The mean age of all dogs with intramedullary neoplasia was 7.6 ± 3.6 years. Dogs with primary tumors were significantly (P < .0001) younger (mean age, 5.9 ± 2.8 years) than dogs with secondary tumors (mean age, 10.8 ± 1.6 years). The duration of clinical signs before presentation was recorded for 49 dogs and ranged from 1 to 236 days (median, 18 days). For primary tumors, the duration of clinical signs ranged from 3 to 236 days (median, 28 days). For secondary tumors, the duration of clinical signs ranged from 1 to 55 days (median, 11 days), which was significantly shorter than for dogs with primary tumors (P = .0012).

Forty-one (77%) dogs were referred for myelopathic signs, 7 (13%) for systemic signs, and 5 (9%) for signs referable to another organ (Table 2). Of the dogs with primary tumors, 95% (33/35) were presented for myelopathic signs. Of the dogs with secondary tumors, 44% (8/18) had myelopathic signs. The remaining dogs with secondary tumors 22% (5/18) had nonspecific clinical signs including anorexia, collapse, and fever, and 22% (5/18) had a primary complaint of primary organ dysfunction without overt myelopathic signs including hemoabdomen, stranguria, and hematuria.

Table 2. Associations between categorical clinical characteristics and tumor type
VariableCategoriesTotal NPrimary TumorsSecondary TumorsP Value
% (95% CI)N% (95% CI)N
  1. a

    CI not defined for cells with zero counts.

  2. b

    Statistically significant.

SexFS2766.7 (46.0–83.5)1833.3 (16.5–54.0)9.8431
MN2665.4 (44.3–82.8)1734.6 (17.2–55.7)9
Spinal painAbsent1764.7 (38.3–85.8)1135.3 (14.2–61.7)61.0000
Present3265.6 (46.8–81.4)2134.4 (18.6–53.2)11
Unknown475.0 (15.9–31.7)325 (0.6–80.6)1
Chief complaintSpinal cord dysfunction4180.5 (65.1–91.2)3319.5 (8.8–34.9)8<.0001b
Systemic signs728.6 (3.7–71.0)271.4 (29.0–96.3)5
Organ-specific50a0100 (47.8–100.0)5
Neuroanatomic localizationUnknown10a0100 (2.5–100)1.0028b
C1-C512100 (73.5–100.0)120a0
C6-T22100 (15.8–100.0)20a0
T3-L32860.7 (40.6–78.5)1739.3 (21.5–59.4)11
L4-S2933.3 (7.5–70.1)366.7 (29.9–92.5)6
Multifocal1100 (2.5–100.0)10a0

A T3-L3 neurolocalization was most common, accounting for 52.8% (28/53) of all tumors (Fig 1). Dogs with a cervical neurolocalization were significantly (P = .0028) more likely to have a primary tumor than a secondary tumor. For primary tumors, 12/35 dogs had a neurolocalization of C1-5, 3/35 were C6-T2, 17/35 were T3-L3, 3/35 were L4-S3, and 1 dog had a multifocal neurolocalization with masses identified at vertebral level C7-T1 and T3-4 on postmortem examination. For secondary tumors, 1/18 dogs did not have neurolocalization recorded, but a metastatic tumor was identified in the C4-5 spinal segments on postmortem examination, 11/18 were T3-L3, and 6/18 were L4-S3.

Figure 1.

Longitudinal histopathologic distribution of primary and secondary intramedullary tumors.

The presence or absence of hyperpathia during spinal palpation was documented in 36/53 dogs; 24/36 had primary tumors with 21/24 exhibiting signs of hyperpathia and 3/24 without hyperpathic signs. Twelve dogs had secondary tumors, with 11/12 exhibiting signs of hyperpathia and 1/11 without signs. There was no association between the presence of hyperpathia during spinal palpation and tumor origin (P = 1.0).

Of the 53 cases, 24/35 with primary tumors and 14/18 with secondary tumors underwent at least 1 type of diagnostic imaging of the vertebral column (Table 3). Among the 24 dogs with primary tumors, vertebral column radiography was performed in 25% (6/24) and did not identify clinically relevant abnormalities in any case. Myelography was performed in 37.5% (9/24) of dogs with primary tumors, and considered abnormal in 78% (7/9) cases. Myelographic abnormalities were nonspecific and included segmental (n = 4) or focal (n = 3) intramedullary cord expansion. CT-myelography was performed in 2/24 dogs with primary tumors and identified a focal region of intramedullary spinal cord expansion in 1 dog. The 2nd dog had multifocal intervertebral disc herniation associated with spinal cord compression, but no evidence of an intramedullary mass lesion (Fig 2D, E, and G). Intramedullary masses were identified in 100% (7/7) of dogs in which MRI was performed including the dog that had a CT-myelogram (Fig 2 and 3).

Table 3. Frequency of vertebral column diagnostic imaging by tumor type
Tumor TypeHistopathologic Diagnosis (Total study n)Diagnostic Imaging Abnormal Studies/Total Performed
PrimaryEpendymoma (n = 13)0/22/30/01/12/2
Astrocytoma (n = 7)0/22/30/01/12/2
Oligodendroglioma (n = 3)0/10/00/00/02/2
Nephroblastoma (n = 4)0/01/10/00/01/1
Chordoma (n = 4)0/11/10/00/00/0
Ganglioglioma (n = 1)0/01/10/00/00/0
MetastaticTransitional cell carcinoma (n = 6)1/20/00/01/12/2
Hemangiosarcoma (n = 6)2/20/00/01/12/2
Pheochromocytoma (n = 2)2/20/00/00/00/0
Prostatic carcinoma (n = 1)1/10/01/10/00/0
Sarcoma unknown etiology (n = 1)1/10/00/00/00/0
Figure 2.

Imaging and pathology of cervical astrocytoma. T2W axial (A) and right parasagittal (B) images illustrating an eccentric, lobular hyperintense intramedullary mass at the C4 level and multifocal cervical intervertebral disc protrusions. (C) Subgross section of the spinal cord at C4, demonstrating effacement of the spinal cord parenchyma by the mass (H&E). The mass was not detected on myelography (D and E) and CT myelography (G) performed 1 month before MRI. (F) The mass is composed of a monomorphic population of gemistocytic astrocytes (H&E).

Figure 3.

Canine multisegmental ependymoma. An elongated fusiform, intramedullary mass extends from T2 to T7, with peritumoral cyst formation at T7. (A) Sagittal T1W postcontrast image. The mass is predominantly T1W hypointense relative to the spinal cord (white arrows), and the cystic portion demonstrates mild peripheral heterogeneous contrast enhancement. The mass occupies a central location within the spinal cord and appears hyperintense in the dorsal planar (B), sagittal (C), and axial (D) T2W images. (E) The mass consists of sheets of ovoid neoplastic cells with uniform and markedly chromatic nuclei, indistinct cytoplasm, and perivascular pseudorosette (black arrows) formation, and intratumoral hemorrhage (H&E).

MRI findings for (Table 3) astrocytomas and oligodendrogliomas were similar and characterized by ovoid to elliptical mass lesions that were well marginated, located eccentrically in the spinal cord, and associated with variable degrees of spinal cord expansion (Fig 2). Astrocytomas and oligodendrogliomas were T1 iso- to hypointense, T2 and STIR hyperintense, and moderately contrast enhancing. Ependymomas (n = 2) appeared as focal to multisegmental, fusiform, centrally located lesions that were heterogeneously T1 iso- to hypointense, T2 hyperintense, and markedly contrast enhancing (Fig 3).

Vertebral column radiography was performed in 8/18 dogs with secondary tumors. Abnormalities were observed in 7/8 studies, including pulmonary metastases (n = 2), retroperitoneal and or abdominal masses (n = 2), splenic mass (n = 1), pleural effusion (n = 1), and geographic vertebral body lysis of L5 (n = 1). Four dogs with secondary tumors underwent MRI imaging, 2 with TCC and 2 with hemangiosarcoma. Both cases of TCC were characterized by focal IM masses that were T1 isointense, T2 hyperintense, and mildly but uniformly contrast enhancing and associated with mild-to-moderate expansion of the spinal cord. In addition, the medial iliac lymph nodes in both TCC cases were enlarged and T2 and STIR hyperintense (Fig 4). Secondary hemangiosarcoma tumors had a variable appearance on T1 images, including both hypo- and isointensity, and were typically hyperintense on T2 images (Fig 5), although 1 lesion contained intermixed areas of T2 hypointensity thought to represent a flow void, hemorrhage, or both. Secondary hemangiosarcoma tumors demonstrated marked contrast enhancement.

Figure 4.

Transitional cell carcinoma with intraspinal metastases. Sagittal pre- (A) and postcontrast (B) T1W MRI images of the lumbar spine demonstrating contrast-enhancing ovoid mass at the L3 level (arrowheads). (C) Sagittal STIR images demonstrating enlarged and heterogeously hyperintense medial iliac lymph nodes (arrows), which contained TCC metastases at necropsy. (D) Subgross (E) and higher magnification (E; inset D) images of the L3 spinal cord illustrating eccentric and ovoid intramedullary mass containing islands of neoplastic transitional epithelial cells (H&E).

Figure 5.

Multifocal intramedullary hemangiosarcoma metastases. Multifocal ovoid, contrast enhancing intramedullary lesions are present at T7 and T13 (arrows) on the sagittal (A and E) and axial (B and F) postcontrast T1W images. The lesion is isointense on the sagittal precontrast T1W (C) and hyperintense on the axial T2W (D) image. (G) Subgross specimen of T13-L1 spinal cord obtained at necropsy 5 weeks after MRI demonstrating hemorrhagic metastatic intramedullary lesion (*) in the right lateral funiculus with associated liquefactive necrosis of the surrounding gray and white matter.

Twenty-seven of 53 dogs received some form of treatment including palliative treatment, surgery, radiation, chemotherapy, surgery and radiation, or surgery and chemotherapy. Treatments were too variable to analyze. Survival times were known for 22 dogs and ranged from 1 to 202 days. Regardless of treatment, median survival time (MST) was 20 days (range 1–202). For primary tumors, MST was 23 days (range 1–202) and for secondary tumors, MST was 16.5 days (range 2–113). This difference was not significant (P = .4038).

In 3/35 dogs with primary tumors, coexistent malignancies were identified at necropsy, including lymphoma (n = 1), cardiac hemangiosarcoma (n = 1), and pulmonary adenocarcinoma (n = 1). Necropsy identified intracranial metastases in 2/13 dogs with secondary tumors (hemangiosarcoma, n = 1; TCC, n = 1). Five dogs with secondary tumors had no description of the brain in the necropsy report.


These results demonstrate that intramedullary spinal cord tumors are uncommon. Primary tumors occur more often in young dogs and in the cervical spinal cord segments. Secondary tumors were found throughout the thoracolumbar spinal cord segments. The duration of clinical signs before diagnosis is shorter for dogs with metastatic cancer than those with primary tumors. Dogs with primary tumors are frequently presented for clinical signs associated with spinal cord involvement as are a substantial number of dogs with secondary tumors. Both primary tumors and secondary tumors are associated with a poor prognosis.

Fifty-six intramedullary spinal cord tumors have been reported in the veterinary literature.[1, 2, 4, 9-13, 15, 22, 25, 26, 28-47] Primary tumors were documented in 41 cases [1, 2, 4, 9-12, 15, 22, 28-36, 38, 39, 41, 42, 47] and secondary tumors in 15 cases.[2, 13, 25, 26, 37, 43-47] The mean age, recorded for 49/55 cases, was 4.7 years. If separated by primary versus secondary masses, the mean age was 4.0 and 6.3 years, respectively. This is comparable to our results with younger dogs (mean age, 5.9 years) being significantly more likely to develop primary tumors compared to dogs with secondary tumors (mean age, 10.8 years).

In humans, secondary tumors account for 8.5% of all CNS metastases with lung carcinomas (54%) and breast carcinomas (11%) accounting for the majority.[7] Like our population, humans with secondary tumors usually are elderly.[7] Nearly half of all secondary tumors in human patients are associated with brain metastases,[48] whereas brain metastases were found in 2/13 dogs examined in this study. In people with secondary tumors, 35% present with neurologic dysfunction as the first sign of malignancy.[49] Similarly, 44% of dogs with secondary tumors had myelopathic signs. Approximately 1% of people with secondary tumors are asymptomatic.[49] One case in this study was not neurologically evaluated and was presented with hemoabdomen as the chief complaint. All other cases had abnormal neurologic examination findings. Thus, asymptomatic spinal cord metastases appear to be rare in dogs (1/53, 2%).

Only 5% of secondary tumors in humans are diagnosed antemortem, and affected patients present more acutely and progress more rapidly than patients with primary tumors.[7] Clinical progression in previous veterinary reports ranged from 1 to 504 days (mean, 55.7 days; median, 21 days).[4, 9-13, 15, 22, 25, 26, 30-32, 34-43, 45, 46] Our data also indicated a variable clinical course. Thus, a protracted clinical course does not exclude an intramedullary tumor, although consideration should be given to primary tumors because secondary tumors had a more acute clinical course, particularly when myelopathy was the chief complaint.

The spinal cord level of the tumor was available for 37 reported cases in the literature; primary and secondary tumors were found in each region of the spinal cord. Thirteen masses occurred in the cervical spinal cord segments: 6 with C1-5 location,[13, 25, 29, 33, 40, 45, 47] 4 in the C6-T2 [13, 15, 22, 35] segments, and 2 were not specified beyond cervical myelopathy.[12, 34] Twenty-five masses occurred in the thoracolumbar spinal cord segments: 20 with a T3-L3 location [9-11, 22, 25, 30-32, 38, 39, 41-46] and 5 with a L4-S3 location.[2, 28, 36, 37, 47] One dog had multiple masses in the spinal cord.[13] Consistent with these findings, T3-L3 spinal cord segments were the most common areas of neurolocalization in this study. However, in our population, 100% of dogs with cervical localization on antemortem neurologic evaluation had primary tumors and only 1 dog with a subclinical metastatic cervical lesion occurred in the entire cohort.

Signs of hyperpathia on spinal palpation were described in the literature in 16 dogs with intramedullary tumors,[9, 11, 12, 15, 22, 25, 29, 34-37, 39, 40, 45] such signs were absent in 9 cases,[9, 13, 28, 32, 33, 41, 46] and they were not recorded in 32 cases. Our data indicate that the majority (32/36) of dogs with intramedullary tumors exhibited signs consistent with hyperpathia and there was no association with tumor origin. In people with primary tumors, pain is a common and chief symptom (72%), followed by motor, sensory, and urinary dysfunction.[50] Discomfort associated with the vertebral column is a later development in people with secondary tumors, in whom motor dysfunction is most prevalent (93%) with reports of pain in 24–33%.[7] One possible explanation of the development of pain in association with intramedullary lesions is altered neurotransmitter modulation attributable to remodeling or destruction of the dorsal horn.[51] In people, this is referred to as syringomyelic syndrome,[50] where there is dissociation of sensory perception and proprioception with or without concurrent motor neuron dysfunction. A mass effect could also contribute to stretching of the meninges, dorsal nerve roots, or both, leading to discomfort.

MRI identified intramedullary tumors in all cases in which it was performed, including a dog that had a prior negative CT-myelogram. The MRI characteristics of the neoplasms identified in this study were similar to those reported for similar tumors in humans, suggesting that MRI findings may provide insight into the specific type of canine intramedullary tumors. For example, ependymomas were centrally located, fusiform in shape, markedly contrast enhancing, and contained peritumoral cysts.[50]

Retrospective studies have inherent limitations. Although we had a substantial cohort of cases, the total number still is relatively low, limiting the power of the study. We cannot ensure that complete examination of the spinal cord was done in every necropsy case, and therefore the occurrence of asymptomatic neoplasia may be underestimated. Therapeutic regimens were also highly variable, from immediate euthanasia to surgery, radiation, and chemotherapy, rendering it difficult to infer any potential effects of treatment on outcome. Information about severity of neurologic dysfunction was not available and no tumors were originally graded using World Health Organization classification. In humans, both the severity of neurologic disease and histologic grade of malignancy are important when establishing a prognosis.[7, 50, 52] There are inherent difficulties in accurately quantifying pain and establishing its source in veterinary medicine. We recognize that multiple concurrent and potentially painful conditions may have coexisted with intramedullary tumors in the cohort of dogs described here, complicating the ability to attribute pain to a particular lesion. However, given the precedents in the veterinary literature citing the absence of pain [22, 53] associated with intramedullary lesions, we felt it valuable to include pain in the spectrum of possible clinical signs. Finally, advanced imaging was done in relatively few cases limiting interpretation of the imaging data.

In conclusion, primary tumors comprise the majority of intramedullary spinal cord neoplasms, they are more common in the cervical spinal cord of young dogs, and affected dogs often have a protracted clinical course. Secondary tumors occur in older dogs with signs of malignancy frequently found on survey radiographs (7/8). These patients are presented frequently for signs of spinal cord dysfunction rather than primary organ dysfunction. Intramedullary neoplasms are associated frequently with hyperpathia. Prognosis regardless of tumor origin is considered poor.


Funded by Maria Garst Memorial Cancer Research Grant.

Conflict of Interest: Authors disclose no conflict of interest.


  1. 1

    SAS Version 9.2, Cary, NC