Materials and Methods

Case selection—Medical records were obtained from the database of the Laboratory of Pathology and Toxicology of the School of Veterinary Medicine at the University of Pennsylvania (January 1986 to July 2005) and the medical record database at the University of Tennessee Veterinary Teaching Hospital (January 1980 to July 2005). The dates encompassed all of the information in its entirety for these institutions. Histopathologic diagnoses were reviewed by 2 board-certified veterinary pathologists (SHS and TJVW), who conducted a histologic examination on preserved tissues. Cats with histologically confirmed primary or metastatic tumors of the spinal cord or tumors that caused spinal cord disease by local extension from adjacent tissues were included in the study.

Medical records review—Location of a tumor within the spinal cord was recorded and assigned to 1 or more of 3 anatomic divisions (cervical, thoracic, or lumbosacral segment of the spinal cord, as determined on the basis of exit of the nerve roots). The lumbosacral segment comprised the lumbar, sacral, and coccygeal segments of the spinal cord. For lymphosarcoma, extraneural sites of the tumor and location within the vertebral canal (extradural, intradural, or both) were also recorded. Data collected from the medical records also included signalment; clinical history; and results of neurologic examination, diagnostic imaging, and clinical pathologic evaluation.

Statistical analysis—Statistical analyses were performed by use of commercially available software.^a Normally distributed continuous variables were described as mean and SD, and an independent sample t test was used to compare continuous variables among groups. The assumption of homogeneity was tested for each variable, and when variance among groups was not met, data were logarithmically (log₁₀) transformed. Categoric variables were described as percentages, and comparisons among categoric variables were performed by use of the χ² test or Fisher exact test, when appropriate. For all comparisons, a value of P < 0.05 was considered significant.

Spinal cord tumors were classified into 2 groups (lymphosarcoma and other; binary data) and were modeled relative to the effect of the continuous variable age (0.5 to 16 years), duration of clinical signs prior to diagnosis (log₁₀ duration), and number of CNS locations (1, 2, 3, or 4 for the 3 spinal cord regions and 1 brain location). Logistic probability of a cat having lymphosarcoma was modeled by use of logistic regression analysis, and independent variables were included in the model when they were significant, as determined on the basis of the Wald χ² statistic. The following model was used:

View Expanded

View Expanded

Results

Eighty-five cats met the criteria for inclusion in the study (68 from Pennsylvania and 17 from the University of Tennessee). This included 56 cats reported in another study⁵ and 29 newly identified cats. Initially, 89 cats were identified, then 2 cats (1 from Pennsylvania and 1 from the University of Tennessee) were excluded because the tumor (ie, lymphosarcoma) was limited to the nerve roots and the spinal cord was not affected. Two other cats from the University of Tennessee were excluded because tissues were not available for histologic examination.

The study population comprised 69 domestic shorthair cats; 9 domestic longhair cats; 2 Siamese; and 1 each of a Persian crossbred cat, Devon Rex, Angora, and Abyssinian; information on breed was not available for 1 cat. Female (n = 39) and male (43) cats were equally represented in the study; information on sex was not available for 3 cats. Thirty-eight cats were castrated males, 5 were sexually intact males, 31 were spayed females, and 8 were sexually intact females. Mean ± SD age at initial examination for all cats was 8.1 ± 4.4 years (median, 9 years; range, 0.5 to 16 years). Mean body weight was 4.61 ± 1.80 kg (10.14 ± 3.97 lb), with a median of 4.41 kg (9.70 lb) and a range of 0.80 to 11.02 kg (1.76 to 24.25 lb).

Tissue samples were collected for 52 cats during a complete postmortem examination, whereas tissue sections were collected for 33 cats during a surgical procedure or postmortem examination. Spinal cord tumors were classified into 11 groups on the basis of cell of origin (Table 1). Among the 8 glial tumors, there were 3 astrocytomas (including 1 gemistocytic and 1 granular cell astrocytoma), 2 oligodendrogliomas, 1 ependymoma, 1 anaplastic giant-cell tumor, and 1 poorly differentiated glial tumor. The 2 metastatic tumors were a thyroid gland C-cell tumor, which metastasized to the lungs, spleen, and vertebrae, and a melanoma of the iris, which metastasized to the liver, kidneys, lymph nodes, bones, and vertebral canal. Tumors affected the lumbosacral segment of the spinal cord in 39 cats, thoracic segment of the spinal cord in 34 cats, and cervical segment of the spinal cord in 28 cats. Thoracic and lumbosacral locations were significantly (P = 0.01; χ² test) more common in cats with lymphosarcoma, compared with locations for cats with other types of tumors. In the group of cats with spinal cord tumors other than lymphosarcoma, 6 of 8 glial tumors affected the cervical segment of the spinal cord, and 4 of 5 fibrosarcomas affected the thoracic segment of the spinal cord. In 15 cats, the tumor affected multiple regions of the spinal cord; significantly (P < 0.001) more of these cats had lymphosarcoma (n = 13), compared with only 2 cats in which other types of tumors affected multiple segments of the spinal cord. The tumor also affected the brain in 10 cats (6 cats with lymphosarcoma, 2 cats with an astrocytoma, and 2 cats with a histiocytic tumor). The brain was affected in a significantly (P = 0.01) higher number of cats with lymphosarcoma, compared with the number of cats with other types of tumors.

Table 1—

Number and percentage of various types of tumors and age of onset of clinical signs in 85 cats with tumors that affected the spinal cord.

View Table

Information on the location of lymphosarcoma within the vertebral canal and other extraneural organs was also recorded. Location of lymphosarcoma was described as intradural when neoplastic cells were found in the leptomeninges or spinal cord parenchyma. Histologic examination revealed extradural and intradural components in 29 of 33 (87.9%) cats with lymphosarcoma; 10 of 29 (34.5%) cats with lymphosarcoma had an exclusive extradural location, and 3 (10.3%) cats with lymphosarcoma had an exclusive intradural location. The brain was examined for 14 cats, and neoplastic cells were found in the brain of 6 (42.9%) cats with lymphosarcoma in the spinal cord. Twenty-six cats with lymphosarcoma had a complete postmortem examination, and lymphosarcoma was found in extraneural locations in 22 (84.6%) of these cats. The most common extraneural location of lymphosarcoma was the bone marrow (n = 10 cats), followed by the kidneys (9), liver (8), skeletal muscle (7), spleen (6), lymph nodes (6), vertebrae (4), and heart (4). Lymphosarcoma was found in 1 extraneural location in 5 cats, 2 locations in 3 cats, 3 locations in 7 cats, and 4 locations in 4 cats; in the remaining 3 cats, lymphosarcoma was found in 5, 6, and 9 extraneural locations, respectively.

Cats with lymphosarcoma were significantly (P = 0.001; Student t test and Satterthwaite unequal variances) younger (mean ± SD, 5.9 ± 4.8 years; median, 4.25 years) than cats with other spinal cord tumors (mean ± SD, 9.5 ± 3.5 years; median, 9.9 years). Univariate analysis revealed that cats with tumors other than lymphosarcoma had a normal distribution for age, with 80% of those cats between the ages of 5.5 and 14 years. In contrast, cats with lymphosarcoma had a bimodal distribution for age, with 50% of them < 4.2 years old and 25% between 11 and 16 years of age.

The ratio of neutered to sexually intact cats was lower in cats with lymphosarcoma (2.9:1) than in cats with other spinal cord tumors (9.2:1). However, there was no difference for sex and sexually intact or neutered status between cats with lymphosarcoma or cats with other tumor types, and the higher number of sexually intact cats in the lymphosarcoma group correlated with the younger age of these cats.

Mean body weight was 4.11 ± 1.80 kg (9.04 ± 3.97 lb) for cats with lymphosarcoma (median, 3.90 kg [8.59 lb]) and 5.11 ± 1.70 kg (11.24 ± 3.75 lb) for cats with other types of tumors (median, 4.61 kg [10.14 lb]). Body weight did not differ significantly between the 2 groups.

Information on clinical history was available for 78 of 85 (91.8%) cats. The most common initial clinical sign was weakness or paralysis affecting the hind limbs. Cats with lymphosarcoma had a higher prevalence of nonspecific signs such as anorexia, lethargy, and weight loss (Figure 1). A complete neurologic examination was performed by a board-certified veterinary neurologist or veterinarian in a neurology residency training program for 61 of 85 (71.8%) cats. The most common neurologic abnormality was signs of pain associated with the spinal cord or vertebral column, which was detected in 37 of 61 (60.7%) cats (Figure 2). Signs of pain associated with the spinal cord or vertebral column were slightly more common in cats with tumors other than lymphosarcoma (21/33 [63.6%]); however, there was not a significant difference between these groups.

View Full Size

Initial clinical signs in 78 cats with tumors (30 cats with lymphosarcoma [white bars] and 48 cats with other types of tumors [black bars]) that affected the spinal cord. Numbers in parentheses are the number of cats with the indicated clinical sign; numbers do not sum to 78 because most cats had > 1 clinical sign.

*Detected on general physical examination; represents a range of weakness to lack of voluntary movements. †Represents signs of pain associated with the spinal cord or vertebral column. ‡Coughing, sneezing, and nasal discharge.

Citation: Journal of the American Veterinary Medical Association 232, 2; 10.2460/javma.232.2.237

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Neurologic examination findings in 61 cats with tumors (28 cats with lymphosarcoma [white bars] and 33 cats with other types of tumors [black bars]) that affected the spinal cord. *Represents signs of pain associated with the spinal cord or vertebral column. †Represents asymmetric neurologic signs, such as hemiparesis or paraparesis that was worse in 1 limb. ‡Represents signs detected during neurologic examination conducted by a board-certified veterinary neurologist or veterinarian in a neurology residency training program. LMN = Lower motor neuron. UMN = Upper motor neuron. See Figure 1 for remainder of key.

Citation: Journal of the American Veterinary Medical Association 232, 2; 10.2460/javma.232.2.237

• Download Figure
• Download figure as PowerPoint slide

Mean duration of clinical signs prior to diagnosis for all cats was 49.7 ± 75.9 days (median, 27.5 days; range, 1 to 455 days). Cats with lymphosarcoma had a significantly (P = 0.011; Student t test of log₁₀ duration) shorter duration of clinical signs (mean, 24.3 ± 26.9 days; median, 15 days) than did cats with other types of spinal cord tumors (mean, 69 ± 93.5 days; median, 30 days). Most (29/31 [93.5%]) cats with lymphosarcoma had a duration of clinical signs of < 60 days.

The logistic model for the population of cats based on age, duration of clinical signs prior to diagnosis, and number of CNS locations affected was significant (Wald χ² = 17.6; 3 df; P < 0.001), and effectiveness of the model for predicting the binary response (lymphosarcoma or other tumor type) was 90% (index of rank correlation c = 0.90). The model used was logit(P) = (3.52 × NUM) − (0.3178 × age) − (1.6828 × log₁₀ duration) + 0.1119.

Results of a CBC and serum biochemical analysis were available for 43 cats. In 2 cats with lymphosarcoma, atypical lymphocytes were detected during examination of blood films. In 1 cat with multiple myeloma, total calcium concentration was increased (16.4 mg/dL; reference range, 9.1 to 11.2 mg/dL) and serum electrophoresis revealed polyclonal gammopathy. Bone marrow aspirates were obtained from 7 cats with lymphosarcoma. In 1 cat, bone marrow examination led to the diagnosis of lymphosarcoma, whereas in another cat, bone marrow examination led to suspicion of a plasma cell tumor. In 2 cats, an abnormal ratio for myeloid to erythroid precursors was recorded. In the remaining 3 cats, results of cytologic examination of bone marrow were considered within expected limits.

In 3 cats with lymphosarcoma, fine-needle aspiration of mediastinal, peritoneal, and axillary masses was performed, and all of them were diagnostic. Fine-needle aspiration was performed in 3 cats with osteosarcoma. In those cats, examination of an FNA of a lymph node in 1 cat yielded negative results, and results of examination of FNAs of vertebral lesions detected during radiography in 2 cats were suggestive of a neoplastic process. Examination of an FNA of a mesenteric lymph node revealed neoplastic plasma cells in 1 cat with multiple myeloma.

Biochemical analysis and cytologic evaluation of CSF were performed in 11 cats with lymphosarcoma, which revealed lymphoblasts in 1 cat. In 4 cats with lymphosarcoma, CSF was contaminated by blood, whereas CSF analysis revealed albuminocytologic dissociation in 2 cats and an increase in protein content and neutrophilic pleocytosis in 2 cats. In 2 other cats with CSF pleocytosis, the differential cell count was not available. Analysis of CSF samples was also performed in 7 cats with tumors other than lymphosarcoma. Neutrophilic pleocytosis was detected in 2 cats with spinal cord astrocytomas and 1 cat with osteosarcoma, and albuminocytologic dissociation was found in CSF obtained from the lumbar region of a cat with chondrosarcoma. In 3 cats (1 with a meningioma, 1 with a peripheral nerve sheath tumor, and 1 with a plasma cell tumor), results of CSF biochemical analysis and cytologic examinations were considered within expected limits.

Information on serologic tests was available for 42 of 85 (49.4%) cats (23 with lymphosarcoma and 19 with other types of tumors). Cats were tested for FeLV (n = 41), FIV (19), Toxoplasma gondii (8), Cryptococcus neoformans (4), and feline coronavirus (4). Of the cats with spinal cord lymphosarcoma, 13 of 23 (56.5%) had positive results for FeLV, 1 of 7 (14.3%) had positive results for FIV, and 1 of 3 (33.3%) had positive results for feline coronavirus. One cat with a spinal cord ependymoma was seropositive for T gondii (IgG titer, 1:1,024). Results of the other serologic tests were negative, including negative results for the FeLV test in 18 cats with tumors other than lymphosarcoma.

Reports on diagnostic imaging were available for 53 of 85 (62.4%) cats (20 with lymphosarcoma and 33 with other types of tumors). Survey radiography in 19 cats revealed a mass within the thoracic cavity of 3 cats (2 cats with lymphosarcoma and 1 cat with a bronchial carcinoma and a spinal cord fibrosarcoma) and a lytic vertebral lesion in 2 cats with osteosarcoma. A mass was also detected during examination of survey abdominal radiographs in 1 cat with lymphosarcoma and 1 cat with a plasma cell tumor. Examination of survey radiographs of the vertebral column performed in 8 cats with lymphosarcoma revealed 2 soft tissue masses (1 associated with a vertebra and 1 associated with a kidney) and a lytic vertebral lesion. Lytic lesions were also detected in 14 of 18 survey radiographs of vertebral columns of cats with tumors other than lymphosarcoma. In cats with osteosarcoma, a lytic lesion was detected in 8 of 9 radiographs of the vertebral column (including 2 lesions detected on examination of thoracic radiographs). Four of these lytic lesions were associated with a pathologic fracture that was diagnosed on examination of survey radiographs in 3 cats and confirmed by CT in 1 cat.

Abdominal ultrasonography was performed in 3 cats with lymphosarcoma. One cat had an enlarged lymph node, and nodules were associated with the spleen and pancreas in the other 2 cats. During postmortem examination, lymphosarcoma was detected in the spleen and lymph nodes of 2 of these cats and pancreatitis was diagnosed in 1 of these cats.

Advanced imaging of the vertebral column was performed for 33 cats. Myelography was performed in 21 cats (11 cats with lymphosarcoma and 10 cats with other types of tumors). Myelography revealed an extradural mass (5 cats with lymphosarcoma and 5 cats with other types of tumors), attenuation of the contrast columns (4 cats with lymphosarcoma and 2 cats with other types of tumors), deviation of the contrast column suggestive of an intramedullary mass (1 cat with lymphosarcoma that infiltrated the leptomeninges and 2 cats with glial tumors), and interruption of the contrast columns in 1 cat with lymphosarcoma and another cat with a malignant peripheral nerve sheath tumor that invaded the spinal cord.

Computed tomography was performed in 3 cats with lymphosarcoma and 7 cats with other types of tumors. A CT scan conducted after myelography allowed identification of an extradural mass in 1 cat with lymphosarcoma. In 2 other cats with lymphosarcoma, CT revealed an enlargement of the lumbar region of the spinal cord without enhancement in 1 cat and focal contrast enhancement extending over the length of 2 lumbar vertebrae in the other cat. Results of CT were considered within expected limits in 2 cats with a glioma. In a cat with a granular cell astrocytoma, CT revealed a diffuse enlargement of the cervical segment of the spinal cord without contrast enhancement. A CT scan after myelography revealed a mildly enhanced intradural or intramedullary mass in a cat with a malignant peripheral nerve sheath tumor and an extradural mass in a cat with a chondrosarcoma. Computed tomography confirmed a pathologic vertebral fracture in a cat with osteosarcoma and a lytic lesion in a cat with a metastatic thyroid gland carcinoma.

Magnetic resonance imaging was performed in 7 cats. In 1 cat with a plasma cell tumor in the CNS, results of MRI were within expected limits, and the only information available in 2 cats with osteosarcoma in a vertebra was that MRI allowed the identification of a mass. One cat with a meningioma had a strong contrast-enhancing mass, 1 cat had an oligodendroglioma that appeared as a ring-enhancing intramedullary mass, and 2 other cats with tumors of neuroectodermal origin appeared as intradural or intramedullary contrast-enhanced masses.

A definitive antemortem diagnosis was achieved for tumors in 16 of 85 (18.8%) cats (11 cats with lymphosarcoma and 5 cats with other types of tumors). The diagnosis was reached via examination of surgical biopsy specimens for 7 cats (4 cats with lymphosarcoma, 1 cat with a meningioma, and 2 cats with osteosarcoma of the vertebra) and by examination of FNAs for 5 cats (3 cats with lymphosarcoma, 1 cat with a plasma cell tumor, and 1 cat with a metastatic thyroid gland C-cell tumor). Finally, a definitive diagnosis of lymphosarcoma was achieved by cytologic examination of bone marrow (n = 1), a blood film (2), and CSF (1).

Specific treatment was instituted in 4 cats, and information on survival time was available for 3 cats (2 cats with lymphosarcoma and 1 cat with a meningioma). One cat was euthanized 38 days after the diagnosis of lymphosarcoma because of recurrence of clinical signs; this was after administration of a single dose of radiation and a mixture of drugs that included prednisone, cyclophosphamide, and vincristine. Another cat with lymphosarcoma survived 60 days after decompressive surgery and treatment with clindamycin, interferon, prednisone, asparaginase, and cytarabine. The cat with a spinal cord meningioma was lost to followup monitoring 7 months after decompressive surgery; at 7 months after surgery, the owner reported that the cat was doing well without any subsequent treatments.

Discussion

The retrospective study reported here was conducted to determine the prevalence of and characterize tumors that affected the spinal cord of 85 cats. In other studies,^2,3,5-7 investigators have reported that lymphosarcoma is the most common type of spinal cord tumor of cats. Other tumors, such as osteosarcoma, meningioma, and various types of gliomas, reportedly only sporadically affect the spinal cord of cats^1,4,8-21; however, they accounted for 28 (32.9%) of the tumors in our study.

Information on signalment and clinical signs was reviewed to assist in reaching an antemortem diagnosis and differentiating lymphosarcoma from other tumors that affect the spinal cord. A histopathologic diagnosis was required for cats to be included in the study, and most of the cats had a postmortem examination with detailed information on the extent of the tumors. A limitation of this study was that only a limited amount of information was available regarding response to treatment (a definitive antemortem diagnosis was only achieved in 16 cats, and specific treatment was only instituted in 4 of these cats). Other limitations were that many tumors were represented by < 10 cats and that conclusions with regard to signalment and clinical signs could not be derived; however, information on signalment and clinical signs was helpful in differentiating lymphosarcoma from other spinal cord tumors. Diagnosing lymphosarcoma in the spinal cord in vivo is important because lack of an antemortem diagnosis and institution of specific treatment is probably one of the factors that has limited the information on long-term follow-up results of cats with lymphosarcoma and other spinal cord tumors in the veterinary literature.^2,3,22

In the study reported here, cats with lymphosarcoma were younger at initial examination, had a shorter duration of clinical signs, and had lesions in more regions of the CNS, compared with results for cats with other types of tumors. Based on these findings, a logistic model of spinal cord tumors in this population of cats was created, and this model correctly predicted 25 of 31 (80.6%) cats with lymphosarcoma as being in the lymphosarcoma group and 37 of 40 (92.5%) cats with other types of tumors as being in the nonlymphosarcoma group. On the basis of this logistic regression model, cats with lesions in 3 or 4 CNS locations and with a 2-week history of clinical signs had a 98% chance of having lymphosarcoma. Cats with lesions in 2 locations that were < 10 years old and had a 10-day duration of clinical signs had a 90% chance of having lymphosarcoma, but similar cats (ie, lesions in 2 locations and < 10 years old) that had a duration of clinical signs of 100 days had only a 65% chance of having lymphosarcoma. Cats with only 1 lesion that were > 6 years old and had a 14-day duration of clinical signs had less than a 50% chance of having lymphosarcoma. For this last group, when cats have a focal spinal cord lesion during initial examination, additional clinical information would appear to be useful in diagnosing spinal cord lymphosarcoma. The number of CNS locations in our logistic model was derived only from information obtained during necropsy because information from advanced imaging and clinical localization techniques was available for only a few cats in this retrospective study. Ideally, the clinical usefulness of our model should be tested in a prospective study that would include the use of advanced diagnostic tools, such as MRI.

The most common initial clinical sign for cats with spinal cord tumors was paraparesis or paraplegia. A higher percentage of cats with lymphosarcoma had weakness and other clinical concerns that were nonspecific, such as anorexia, lethargy, weight loss, and signs of respiratory tract infection. This could be explained by the fact that 13 of 23 (56.5%) cats with lymphosarcoma had positive results for FeLV and that this virus may induce immunosuppression. Also, in 22 of 26 (84.6%) cats with spinal cord lymphosarcoma, this tumor type was found in other organs, such as the kidneys, liver, and heart, which could have accounted for some of the nonspecific signs.

During neurologic examination, the most common clinical sign in all cats with spinal cord tumors was signs of pain associated with the spinal cord or vertebral column, which was detected in 37 of 61 (60.7%) cats, despite the fact that signs of pain associated with the spinal cord were among the initial clinical signs in only 12 of 78 (15.4%) cats. Signs of pain associated with the spinal cord may not have been recognized by owners and referring veterinarians, such as for 4 cats with initial clinical signs of inappropriate vocalization and hiding behavior, which could be signs suggestive of discomfort. Signs of pain associated with the spinal cord were slightly more common in cats with tumors other than lymphosarcoma, whereas asymmetric clinical signs (eg, hemiparesis or paraparesis worse in 1 limb) and weakness were more common in cats with lymphosarcoma. This could be explained with the more common localization of lymphosarcoma to the thoracic and lumbosacral regions of the spinal cord. Compressive lesions within the cervical portion of the vertebral canal (eg, intervertebral disc extrusion) often cause cervical hyperesthesia and minimal or a total lack of neurologic deficits because of the relative larger diameter of the cervical portion of the vertebral canal, compared with the diameter for the thoracic and lumbar regions.

Findings of clinical pathologic evaluation assisted investigators in reaching an antemortem diagnosis of lymphosarcoma in 7 of 33 (21.2%) cats. Lymphoblasts were detected during cytologic examination of FNAs obtained from masses (n = 3), peripheral blood films (2), CSF (1), or bone marrow (1). Cytologic examination of masses and bone marrow also led to the antemortem diagnosis of a plasma cell tumor (n = 2 cats) and an osteosarcoma (1).

Regarding radiographic examinations, evaluation of survey radiographs of the vertebral column was helpful in reaching an antemortem diagnosis of spinal cord tumors, especially for tumors in the vertebrae, such as osteosarcoma, where lytic lesions were detected in 8 of 9 cats. A lytic lesion was detected only in the vertebra of 1 of 7 cats with lymphosarcoma and in 14 of 18 cats with other spinal cord tumors that had survey radiographs of the vertebral column. Also, at the postmortem examination, lymphosarcoma was found within the vertebrae of 4 of 26 (15.4%) cats in the study; therefore, lymphosarcoma should be lower on the list of differential diagnoses for cats with lytic vertebral lesions detected during radiographic examination.

All myelograms in the study reported here revealed abnormalities, but masses with intradural or intramedullary location could often not be differentiated solely on the basis of myelography. A clear extradural lesion was reported in 5 of 11 cats with lymphosarcoma and 4 of 9 cats with other spinal cord tumors; therefore, the finding of an extradural lesion during myelography cannot be used to differentiate lymphosarcoma from other spinal cord tumors.

Computed tomography confirmed a lytic lesion in 1 cat and revealed a pathologic fracture in another cat. A CT scan performed after myelography provided more information on location of the tumor in 2 cats. It is likely that the availability of CT and, especially, MRI will increase the ability of clinicians to localize and diagnose spinal cord tumors in cats.¹⁴

Abdominal ultrasonography should also be recommended when lymphosarcoma in the CNS is suspected because of the high percentage of cats with extraneural lymphosarcoma. Information on abdominal ultrasonographic findings was available for only 3 cats with lymphosarcoma. A mass was detected in each of those cats, and in 1 cat, cytologic examination of an FNA led to an antemortem diagnosis of lymphosarcoma.

Postmortem examination revealed that lymphosarcoma commonly involved multiple spinal cord segments, whereas fibrosarcoma and glial tumors were most commonly localized to the thoracic and cervical spinal cord segments, respectively. A possible explanation for the location of fibrosarcomas is that some of them may have been vaccine-associated neoplasms. In the past, vaccines were administered SC in tissues of the interscapular space, and fibrosarcomas in the cats of the study reported here were diagnosed between 1983 and 1994. There is no explanation for the preferential location of glial tumors to the cervical segment of the spinal cord, and it may have been coincidental as a result of the limited number of gliomas in the study (n = 8 cats). There are few reports of spinal cord gliomas in cats. In 3 case reports,^19-21 the tumors were localized to the lumbar (n = 2 cats) and cervical (1) spinal segments.

In the study reported here, lymphosarcoma was often found in both the extradural and intradural space, and this finding differs from that of other studies^2,3,23 of spinal cord lymphosarcoma in cats on which a postmortem examination was performed. In those reports, spinal cord lymphosarcoma had an exclusive extradural location in 22 of 23 (95.7%),²³ 11 of 13 (84.6%),² and 20 of 23 (86.9%) cats,³ respectively. In another study,²⁴ investigators found that spinal cord lymphosarcoma had an exclusive extradural location in 5 of 13 (38.5%) cats, which is comparable to the 10 of 29 (34.5%) cats in the study reported here. Similar to results of other studies,^2,3,23,24 lymphosarcoma was found to commonly affect extraneural locations in the cats of our study. The most common extraneural locations for lymphosarcoma were bone marrow and the kidneys, as previously reported,²⁵ followed by the liver, skeletal muscle, spleen, lymph nodes, vertebrae, and heart. In another study,²³ lymphosarcoma was detected in tissues other than nervous tissues in 23 of 23 cats with lymphosarcoma of the CNS. Among the extraneural locations in that study were the kidneys, lymph nodes, bone marrow, and vertebrae. In 2 other studies, the most common extraneural site was the kidneys (10/23 cats with extraneural lymphosarcoma; all 10 cats)³ and the kidneys (11/13 cats with extraneural lymphosarcoma; 8 cats), followed by the bone marrow (7), liver (6), lymph nodes (5), and spleen (3).² Finally, lymphosarcoma was found in extraneural sites (12/18 cats with extraneural lymphosarcoma) in the kidneys, liver, lymph nodes, and bone marrow.²⁴ Similar locations for extraneural lymphosarcoma were also reported in our study, and similar to the aforementioned studies, bone marrow and the kidneys were the most common locations affected in 10 of 22 (45.5%) and 9 of 22 (40.9%) cats with extraneural lymphosarcoma, respectively. This information on lymphosarcoma location may help clinicians to make decisions regarding clinical staging and treatment of spinal cord lymphosarcoma. Spinal cord lymphosarcoma was most likely secondary to multicentric involvement in 22 of 26 (84.6%) cats in our study, and on the basis of findings of clinical pathologic evaluation and radiographic examination, it appears that performing examination of a bone marrow aspirate, cytologic examination of any mass detected on physical or radiographic examination, or abdominal ultrasonography will provide the best likelihood of obtaining an antemortem diagnosis. The treatment of cats with extradural lymphosarcoma via surgical excision, local radiation treatments, or a combination of both may not be rewarding considering that 19 of 29 (65.5%) cats in our study had intradural lymphosarcoma and that 13 of 26 (50%) had multiple spinal cord regions affected.

In the study reported here, cats with spinal cord lymphosarcoma had a lower age at the time of initial examination (mean ± SD, 5.9 ± 4.8 years; median, 4.25 years), a shorter duration of clinical signs (mean ± SD, 24.3 ± 26.9 days; median, 15 days), and lesions in more locations of the CNS, when compared with results for cats with other types of spinal cord tumors. This information was used to create a logistic model for this population of cats, and results of the model could be used to assist interpretation when a neoplastic process that affects the spinal cord of a cat is suspected. This model and additional clinical information derived from the physical and neurologic examinations, diagnostic imaging, and clinical pathologic evaluation reported here appeared to be helpful in differentiating lymphosarcoma from other types of tumors that affect the spinal cord of cats.