Introduction

Solitary vertebral masses are uncommon in dogs. The most common malignant primary bone tumors of the spine are usually monostotic (e.g., osteosarcoma, chondrosarcoma, hemangiosarcoma, and fibrosarcoma), whereas masses that have metastasized to the spine are usually polyostotic¹ and more commonly originate from carcinomas and sarcomas. Benign solitary vertebral masses are less frequent and comprise benign bone tumors (e.g., osteochondroma, chondroma)² and nonneoplastic bone proliferation (inflammatory, reactive tissue). Infectious processes less frequently affect a single vertebra, although they may occasionally manifest as a solitary lesion.

Reports of solitary vertebral masses describing radiographic and computed tomography (CT) features abound in the veterinary literature.^3–5 Vertebral masses are often detected when animals present with neurological evidence of spinal cord compression. Myelography, with its inherent potential complications, may be necessary alongside these techniques to characterize the degree of spinal cord compression. MRI offers better soft-tissue contrast resolution than CT and radiography, enabling an earlier detection of changes such as infiltration of bone or soft tissue, and a more detailed evaluation of the spinal cord without the need for myelography. MRI has become more readily available in veterinary medicine; however, reports describing the MRI features of solitary vertebral mass lesions in the literature are sparse, often limited to a single case or small case series. At the time of writing this manuscript, solitary vertebral lesions described with MRI in the veterinary literature include two osteosarcomas,⁶ two infiltrative lipomas,^7,8 an osteochondroma,⁹ a hemangiosarcoma,¹⁰ two lymphomas,¹¹ a plasmacytoma,¹² a myxosarcoma,¹³ a chondroma,¹⁴ a squamous cell carcinoma,¹⁵ and a paraganglioma of the cauda equina.¹⁶

No study gathering MRI findings of larger numbers of cases with solitary vertebral masses is currently available in the veterinary literature. The primary aim of this study was to describe the MRI features of solitary vertebral masses in dogs, for which a histopathologic or cytologic diagnosis was reached. A secondary purpose was to identify any potential MRI features enabling differentiation between malignant and benign lesions.

Materials and Methods

MRI

MRI examinations were performed using a low-field MRI unit^a with animals under general anesthesia maintained on isoflurane gas and oxygen, following differing premedication and induction protocols depending on the anesthetist’s preference.

For each dog, the spinal region scanned comprised at least the region corresponding to the neurolocalization (i.e., C1–C5). However, depending on the size of the animal relative to the coil used for the examination, a larger proportion of the spine was usually included on the scan; therefore, the number of vertebrae included on the scan was recorded for each dog.

MR sequences varied between individuals. All studies included a T2-weighted (T2W) sequence in sagittal plane and at least one of the following: T2W, short tau inversion recovery (STIR), T1-weighted (T1W), T1W gradient echo, in various planes. Sequence parameters are detailed in Supplementary Table I. In some patients, T1W sequences were repeated following IV administration of 0.2 mL/kg body weight of gadoterate meglumine or 0.1 mL/kg body weight of gadobutrol^b,c. Postcontrast T1W images were obtained in various planes, with more than one plane acquired in 6 of the 13 dogs and all three planes available in 1 dog.

The MR images were reviewed by a board-certified radiologist (A.C.) and radiology resident (E.M.H.) by consensus, blinded to the clinical findings, diagnosis, and outcome, using an image analysis software^d.

The features assessed included the following: (1) Vertebra affected and specific anatomic areas involved (body, laminae, pedicles, spinous process, transverse processes). (2) Type of vertebral involvement (lateralized, symmetric, circumferential) and delineation (expansile or confined; smooth or irregular). A mass was defined as “lateralized” if located predominantly on one side of the vertebra, “symmetric” if extending in a similar fashion on both sides of the vertebra, and “circumferential” if involving the vertebral body, both laminae and pedicles. A mass was described as “expansile” if it extended beyond the limits of the cortex of the vertebra and “confined” if the margins of the vertebra were not crossed. The expansile or confined nature of the vertebral mass was further characterized by the percentage of cross-sectional vertebral involvement, calculated as the ratio “cross-sectional area of the mass/cross-sectional area of the vertebra” on T2W images (method detailed in Figure 1). (3) MR signal intensity of the mass was described as hypointense, isointense, or hyperintense compared with the nearest adjacent vertebra with no degenerative change (i.e., no evidence of Modic type II changes). The signal of the mass was recorded as homogeneous or heterogeneous. Evidence of enhancement following contrast medium administration was recorded as none, mild, moderate, strong, or rim in nature. (4) Local invasiveness into surrounding structures was recorded as invasion into the vertebral canal or into the adjacent paraspinal muscles by the mass or by extension of an abnormal signal into these structures. (5) Cortical destruction, detected as interruption of the signal void vertebral margins. (6) Sclerosis of surrounding bone, visualized as T1W and T2W hypointensity within the vertebral medullary cavity. (7) Spinal cord compression, recorded as the percentage of cross-sectional reduction, calculated using the formula [1 − (cross-sectional area of the cord at the level of the maximal compression ÷ cross-sectional area of noncompressed spinal cord)] × 100. Cross-sectional areas were measured by drawing a region of interest on the outer border of the spinal cord on the transverse image. The region of measurement of the noncompressed spinal cord was determined subjectively at a level where no spinal cord compression was visible on the transverse image from the same acquisition. (8) Signal hyperintensity within the spinal cord on T2W images, relative to normal spinal cord, recorded as present or absent. The length of the affected cord measured in vertebral body lengths was also recorded. (9) Meningeal enhancement was recorded as present or absent. (10) Evidence of pathologic fracture, detected as a visible fracture line within the vertebral mass, with discontinuity of the cortex and/or change in contour of the vertebra or a shortening of the vertebral body compared with adjacent vertebrae.

View Figure

• Download as Powerpoint
• Download Figure

Statistical analysis for comparisons between malignant and benign groups and between confirmed osteosarcomas and other malignant tumors were performed. For each variable, differences in the frequency of observations in patients with a malignant diagnosis and in patients with a benign diagnosis were tested for significance (α = 0.05) using a Fisher exact test. Differences in percentage of vertebral involvement and spinal cord compression in the different groups were tested for significance (α = 0.05) using a Student t test.

Results

Signalment, Clinical Presentation, and Diagnosis

A total of 150 dogs with a solitary vertebral mass were found in the database, including 24 with a cytologic or histopathologic examination performed; 20 dogs finally met the inclusion criteria, after 4 cases were excluded because of inconclusive diagnostic results.

Fifteen animals were diagnosed with a malignant tumor using histopathology (n = 14) or cytology (n = 1), and five were diagnosed with a benign mass with histopathology. Details of signalment and diagnosis are given in Table 1.

TABLE 1 Signalment and Diagnosis of the 20 Cases

View Fullscreen

The malignant group comprised 15 dogs with a mean age of 9.2 yr (range 5.3–11.5 yr) and a mean body weight of 21.4 kg (range 8–37 kg). Clinical signs were chronic (>7 days) in 9/15 animals and acute in 6/15. Evolution was progressive in 11/15. Twelve animals were ambulatory, and 3/15 were nonambulatory. On admission, clinical signs included spinal pain (13/15), ataxia (6/15), paraparesis (6/15), tetraparesis (2/15), paraplegia (2/15), lethargy (1/15), and absent nociception (1/15). Neurolocalization corresponded to a C1–T2 myelopathy in two cases, C6–T2 in three cases, T3–L3 in eight cases, and L4–S3 in three cases. One dog had clinical signs localized to both C1–T2 and T3–L3 segments.

The five dogs of the benign group had a mean age of 5.4 yr (range 0.5–11.8 yr) and a mean body weight of 25.9 kg (range 10–37.9 kg). All animals of this group had chronic clinical signs (>7 days), which were progressive in three. Neurologic signs on admission were ataxia (3/5), tetraparesis (3/5), neck pain (3/5), and paraparesis (1/5). One dog had no neurological deficits (pain only). Two dogs had a lesion neurolocalized to the C1–C5 segment of the spine, one to C6–T2, one to both C1–C5 and C6–T2 segments, and one to T3–L3.

MRI Findings

Three dogs had the entire spine scanned, and 15.7 vertebrae per animal were imaged on average. The malignant neoplasias were found in all spinal segments; however, the benign lesions were found at C1 (3/5), T2 (1/5), and T12 (1/5) only. MRI features for each case are detailed in Supplementary Table II.

Table 2 shows MRI findings for each group and for confirmed osteosarcoma, which was the most common tumor of the malignant group (6/15 cases).

TABLE 2 MRI Findings in the Malignant Group, Benign Group, and Confirmed Osteosarcomas

View Fullscreen

TABLE 2 (Continued)

View Fullscreen

In three dogs, the percentage of cross-sectional involvement could not be measured because no transverse image was acquired (case 7), because of subluxation of fragments of the vertebral body secondary to a pathologic fracture (case 1), and because of the symmetrical nature of the mass and the unique shape of C1 (case 18).

Spinal cord compression was greater in the malignant group (range 0–56.3%, median 28.1%, interquartile range 4–47.2%) compared with the benign group (range 0–29%, median 10.6%, interquartile range 1.9–25.2%); however, the difference was not statistically significant.

Approximately half of the cases in each group showed an intramedullary hyperintense signal at the level of the vertebral mass on T2W images, ranging from focal to three vertebral bodies in length for the malignant group; however, this was always focal in the benign group.

Five vertebral fractures were suspected in the malignant group (cases 1, 5, 7, 9, 19), whereas none was observed in the benign group. Three cases were suspected to have a pathological fracture because of evident shortening of the vertebral body without a visible fracture line.

Statistical Analysis

Statistically significant (P <.05) parameters are given in Table 3. For the comparison of osteosarcomas versus other malignant neoplasias, the nonconfirmed osteosarcoma case (case 9) was not included in either of the two groups to avoid bias in the results.

TABLE 3 MRI Features Reaching Statistical Significance (P < 0.05) with the Fisher Exact Test When Comparing Malignant and Benign Groups and Osteosarcomas with Other Nonosteosarcoma Malignant Lesions

View Fullscreen

Involvement of the vertebral body; hyperintensity on T2W, STIR, T1W, and T1W gradient echo sequences; and evidence of cortical destruction were significantly associated with malignancy (P <.05). A hypointense signal on T1W gradient echo sequence was significantly associated with benign masses (P < .05). Although not statistically significant, T2W hypointensity or isointensity, STIR hypointensity or isointensity, and T1W hypointensity were also more frequently associated with benign masses (P = [.05–.0526]). The presence of bone sclerosis was significantly associated with osteosarcomas compared with other malignant masses (P < .05).

Discussion

In this study, MRI features of a series of solitary malignant and benign vertebral masses were described, and potential MRI features were identified to help differentiate between malignant and benign lesions. To the authors’ knowledge, this is the first case series of small animals with a solitary vertebral mass described with MRI.

Two subgroups of malignant neoplasia in this study can be compared to individual case reports in the veterinary literature. Osteosarcoma was the most commonly encountered histologic tumor type. Tendency to destroy the vertebral cortex, expand, and invade adjacent tissues, as well as altered MRI signal intensities, reflected intrinsic characteristics of osteosarcomas, which tend to be both lytic and proliferative (Figures 2A–D). These findings are consistent with the three previously reported cases of vertebral osteosarcomas described with MRI^6,17; however, all malignant masses other than the hemangiosarcoma (case 4) had similar features. Case 4 was the only case with a definitive diagnosis of hemangiosarcoma in our study and was a confined mass at T13, with hyperintense and heterogeneous signal on T2W and STIR sequences, isointense and homogeneous signal on T1W images, and rim enhancement, with absence of central enhancement thought to represent central necrosis. This differs from case 5, a sarcoma at T11 that was suspected to be a hemangiosarcoma, which had strong internal contrast enhancement and an expansile nature with cortical destruction—MRI features that were close to a previously described case of vertebral poorly differentiated hemangiosarcoma.¹⁰

View Figure

• Download as Powerpoint
• Download Figure

Two rare subtypes of vertebral neoplasia that are not commonly identified in the veterinary literature were identified in this study: malignant peripheral nerve sheath tumor (MPNST) and extradural leiomyosarcoma. Case 8 was diagnosed with an MPNST with secondary invasion of the adjacent vertebra, and this diagnosis was also considered the most likely in another dog (case 17) in whom a poorly differentiated spindle cell tumor was detected histopathologically. To the authors’ knowledge, MPNST invading vertebral bone has not been previously reported in the veterinary literature. In humans, intraosseous MPNSTs are rare and most often result from secondary bone invasion by an extraosseous tumor rather than being primarily intraosseous.¹⁸ Signs of vertebral involvement by an MPNST include bone (especially pedicle) erosion, vertebral body scalloping, widening of the neural foramen, and expansile vertebral body lesion, typically contrast enhancing.^19,20 In both dogs from our study, the mass was expansile, associated with cortical destruction, invading surrounding muscles and the vertebral canal, and T2W and STIR hyperintense to adjacent vertebral bodies, with involvement of the pedicles. The vertebral body was affected and strongly contrast enhancing in the dog who was administered a contrast medium (case 16).

To the best of our knowledge, no report of leiomyosarcoma with vertebral involvement has been described previously in the veterinary literature. Case 6 presented with a symmetric mass involving both pedicles, laminae, and spinous process. The mass was expansile, destroying the vertebral cortex, invading the vertebral canal, and hyperintense to adjacent vertebral bodies on T1W, T2W, and STIR images. An extradural leiomyosarcoma has been shown to originate in the paraspinal muscles of the neck, causing extradural compression of the spinal cord and secondarily eroding and infiltrating the fourth cervical vertebra in a woman.²¹ Another human patient presented with a primary leiomyosarcoma of C7 with secondary pathologic compression fracture of the vertebra.²² Vertebral metastasis from primary leiomyosarcomas have also been described in humans,²³ although given the lack of further lesions, a metastatic etiology is not suspected in our case.

Only five dogs of the study presented with a benign solitary vertebral mass, reflecting that benign masses are less frequently encountered. Only one juvenile dog (case 18) had a confirmed benign tumor (osteochondroma) with a clinical presentation and MRI findings similar to a recent case report of a solitary vertebral osteochondroma of C2 in a 10 wk old French bulldog.⁹ The case diagnosed with fibro-osteochondromatous proliferation (case 15) was also thought to represent an osteochondroma, although it was challenging to categorize it with certitude histologically. Recurrence of the mass with more aggressive MRI features 19 mo after debulking surgery might indicate a malignant transformation of the osteochondroma, as previously described²⁴; however, this was not confirmed histopathologically. The other three cases of the benign group represented reactive bone proliferation rather than true benign tumors. Case 14 demonstrated common characteristic features from the benign group, illustrated in Figures 2E–H.

In our study, most benign masses were encountered in the cervical spine, including three located in the first cervical vertebra, whereas spinal osteochondromas are usually most often found in the thoracic region in small animals.⁹ The discrepancy may be because of the fact that only one of the benign cases was confirmed as a benign neoplastic process (although a second case was a suspected osteochondroma as well), and the low number of cases included in the study is not necessarily representative of the distribution in the whole population. Similarly, the fact that dogs in the benign group were relatively old is likely because of the fact that the majority of the benign lesions in our study were reactive processes rather than benign neoplasia. On the contrary, malignant tumors were more evenly distributed along the spine, which is consistent with previous reports.¹

Vertebral fractures may be identified on MRI by detecting discontinuity of the vertebral cortex as well as changes in contour and signal intensity of the vertebra.²⁵ Fracture lines might be easier to visualize on T1W images because of the better depiction of cortical anatomy. In our study, five dogs had MRI features compatible with a vertebral fracture (Figure 2A). Fractures were only found in dogs with malignant tumors (two osteosarcomas, two sarcomas, one plasmacytoma) in the present study. It is possible that the frequency of vertebral fractures in this population could have been underestimated.

The main limitation of the study is the limited number of cases included, which reflects the relatively low incidence of solitary vertebral masses causing clinical signs. Because of the low number of cases and imbalance between case numbers in the malignant and benign groups, care should be taken in the interpretation of the statistical analysis. The absence of some reportedly common histologic types of solitary vertebral tumors (e.g., chondrosarcomas) is also likely due to the overall relatively low number of cases; therefore, the cases reported here cannot be representative of all solitary spinal masses. The high proportion of masses compressing the spinal cord is probably a recruitment bias, especially for benign masses, because associated neurologic signs led to referral of the patient.

The retrospective nature of the study accounts for the incompleteness of some medical records and variability in the MR protocols used. Details of numbers of dogs for whom each sequence was acquired are detailed in Supplementary Table I. T1W postcontrast sequences were found particularly useful for assessing the extent of the vertebral masses but were only available in 13 of the 20 cases. One case had no transverse image acquired, limiting the ability to characterize the degree of spinal cord compression and invasion of the vertebral canal. The slice thickness also varied among protocols, leading to more volume averaging and potentially impacting on measurements and signal intensity characteristics with thicker slices. However, the slice thickness was relatively homogeneous and comprised between 3 and 4 mm in all sagittal and transverse plane spin echo sequences, except for two dogs weighing more than 30 kg, for which 4.5 mm- and 5 mmthick slices were acquired for T2W transverse images. Given that some patients had only a portion of the spine imaged, it is possible that some of these neoplasias included were not solitary spinal lesions. However, to mitigate this, patients were confirmed as having complete imaging of any spinal segment to which there was clinical localization and in which lesions diagnosed on histopathology or cytology were consistent with a solitary mass. The single plasmacytoma included (case 1) had the entire axial skeleton included on the imaging, because sporadic reports of solitary vertebral plasmacytomas have been published in the veterinary literature,^12,26 confirming this was not multiple myeloma.

Conclusion

We described MRI features of 20 dogs who presented with a solitary vertebral mass, for which a histopathologic or cytologic diagnosis was reached. MRI enabled a thorough evaluation of changes in soft tissues and bony structures. A solitary vertebral mass was significantly more likely to be malignant when the mass involved the vertebral body, when the mass was T2W hyperintense and/or STIR hyperintense and/or T1W hyperintense and/or hyperintense on T1W gradient echo sequence, and if there was evidence of cortical destruction. Conversely, a solitary vertebral mass was more likely to be benign if the mass was hypointense on T1W gradient echo sequence. The only significant variable between confirmed osteosarcomas and other malignant tumors was the presence of bone sclerosis, which was found more frequently with osteosarcoma.

We found statistically significant features that may help differentiate between malignant and benign masses, but further studies with larger cohorts of animals are needed to further explore these findings and evaluate a greater number of tumor types.