Imaging findings of a haemangiosarcoma in a cervical vertebra of a horse



An 18-year-old Warmblood gelding was presented with a recent, abrupt onset of ataxia. Transcranial magnetic stimulation (TMS) was performed and revealed a delayed response in both thoracic and pelvic limbs. Radiographic examination of the cervical region demonstrated osteolytic lesions in the vertebral body and arch of the third cervical vertebra (C3). A cervical myelogram revealed spinal cord swelling in this region on the lateral projection. Computed tomographic (CT) myelography was performed, which showed osteolytic lesions and circumferential thinning of the contrast column due to extra-dural compression at the cranial cervical region. These findings were compatible with an aggressive bone lesion compressing the spinal cord. Primary or secondary neoplasia and osteomyelitis were considered in the differential diagnosis. Because of the severity of the findings and poor prognosis, the horse underwent euthanasia. Post mortem histopathological diagnosis was a haemangiosarcoma involving the cranial cervical vertebrae and epidural space. This report demonstrates the additional value of CT myelography on the extent and exact location of cervical vertebral compressive lesions in the horse. This information can be useful for prognosis, biopsy and/or planned surgery of these lesions.


Haemangiosarcoma is a malignant tumour of the vascular endothelium and rarely occurs in the horse. This tumour type is more commonly seen in small animals and is mainly detected at the heart base, spleen, liver and lungs. In horses, they are mainly reported in the musculoskeletal and respiratory systems (Van Pelt et al. 1972; Von Reinachter 1978; Valentine et al. 1986; Johns et al. 2005).

Clinical signs are determined by tumour location. As haemangiosarcomas rapidly metastasise it is often difficult to determine the primary site of the tumour and early diagnosis and aggressive therapy are mandatory (Waugh et al. 1977; Frye et al. 1983; Hargis and McElwain 1984; Johnson et al. 1988; Sweeney and Gilletee 1989).

This case report describes the clinical, imaging and pathological findings in a gelding with a haemangiosarcoma at the level of the cranial cervical region, characterised by a recent onset of ataxia. To the authors' knowledge this is the first report describing CT myelography of a haemangiosarcoma in the cervical region in the horse.

Clinical history and clinical findings

An 18-year-old Warmblood gelding was presented with a recent, abrupt onset of ataxia. The horse was stabled and had not previously shown clinical signs. The patient showed a normal body condition, normal rectal temperature, heart and respiratory rates. Mental status was unremarkable and the horse did not show involvement of the cranial nerves. Marked deficits of proprioception were noticed at the walk and at trot. During circling, circumduction of both outer pelvic limbs was visible. Head elevation caused the horse to scuff the hooves of the thoracic limbs.

Transcranial magnetic stimulation (TMS) was performed to evaluate the integrity of the descending motor pathway. This is a noninvasive, nonpainful test. For stimulation of the brain cortex, a coil1 with a diameter of 70 mm generating a magnetic field of approximately 4 Tesla, was used. From the brain cortex, signals are transmitted along the spinal cord. Magnetic motor evoked potentials (MMEP) are measured in both the fore- and hindlimbs (extensor carpi radialis and tibialis cranialis muscles). Morphology of the complexes and latency time are registered. In case of disruption of this pathway, the evoked potentials can have an abnormal morphology and/or the onset latency can be prolonged. In this case, the evoked potentials had an abnormal morphology and latency time was clearly prolonged in both thoracic and pelvic limbs, with the delay being more pronounced on the left side compared to the right (latency times: Left front 26.2 ms, right front 23.4 ms [reference front legs: 20–21 ms], left hind 51.4 ms and right hind 43.9 ms [reference hind legs: 36–38 ms]). Abnormal complexes and/or latency times in all 4 limbs suggested a lesion in the cervical spinal cord.

Lateral radiographic projections of the cervical region were obtained showing heterogeneous radiolucencies at the caudoventral aspect of the body and ventral aspect of the vertebral arch of C3 (Fig 1). No other radiographic abnormalities were detected on the precontrast radiographs. A cervical myelogram under general anaesthesia (triple drip: combination of α2-agonist, myorelaxans and dissociative anaestheticum) was performed. A puncture of the cisterna magna was performed and the same volume cerebrospinal fluid as the injected volume of contrast medium was removed. Iohexol (Omnipaque 2402, 20 ml/100 kg bwt) was injected. The head of the horse was lifted for approximately 5 min. The cervical myelogram revealed thinning of the dorsal and ventral contrast medium columns at the level of C3 compatible with spinal cord swelling on a lateral projection (Fig 2). CT myelography was performed immediately after the cervical myelogram using a single-slice helical CT scanner3 (image settings: 120 kV; 140 mA; matrix size: 512 × 512; contiguous slices of 5 mm with detail algorithm). At CT, the vertebral body of C3, including both transverse processes and part of the vertebral arch, showed a mottled, heterogeneous density. The floor of the vertebral canal at the level of C3 was interrupted. The contrast column showed a circumferential decrease in thickness at the mid and caudal aspects of C3 due to extra-dural compression (Fig 3). With CT, discrete, small hypodense foci were also detected at the caudal aspect of the vertebral body of C2.

Figure 1.

Left-right lateral projection of the second (C2), third (C3) and fourth (C4) cervical vertebrae. The body and ventral part of the vertebral arch of C3 show several rounded, ill-defined, radiolucent areas (arrows). No abnormalities are detected at C2 and C4.

Figure 2.

Left-right lateral projection of the second (C2), third (C3) and fourth (C4) cervical vertebrae after injection of contrast medium in the cisterna magna . At the level of C3 the ventral and dorsal contrast columns are thinned and bulging outwards resulting in an image of spinal cord swelling (arrow).

Figure 3.

Transverse CT images (bone window) at the level of the caudal aspect of the second vertebra (C2) (a), mid (b) and caudal (c) aspects of the third cervical vertebra (C3). Right is at the left, dorsal at the top. a) Discrete hypodense foci are present in the vertebral body of C2 (arrows). The contrast column is normal at this level. b) The vertebral body shows several hypodense foci, the transverse processes and ventral aspect of the vertebral arch are also involved in the process. The contrast column is circumferentially thinned and shows an ovoid shape due to lateral compression. c) The ventral floor of the vertebral canal is interrupted (arrowhead). The left aspect of the vertebral body shows an irregular defect (asterisk). The contrast column is circumferentially thinned and the body of C3 shows multiple hypodense areas.

The imaging findings suggest an aggressive bone lesion involving mainly C3 with a space-occupying lesion in the vertebral canal compressing the spinal cord. Primary or secondary neoplasia and osteomyelitis were considered in the differential diagnosis.

Because of the severity of the findings and poor prognosis the horse was subjected to euthanasia.

Post mortem findings and diagnosis

At post mortem examination the vertebral bodies of C2–C4 showed osteolysis and several haemorrhagic areas. The dorsal cortex of the vertebral body of C3 showed an indentation of 1.5 cm left to the midline. In this defect a haemorrhagic mass (measuring 1 × 2.5 × 1 cm) was present, filling the epidural space and compressing the spinal cord. Pronounced intramuscular haemorrhage was observed ventral to C3 and C4. The haemorrhagic mass and surrounding bone samples were fixed in formalin (10%) and submitted for histological examination.


The formalin-fixed tissue was processed, embedded in paraffin wax, sectioned at 4 µm, stained with haematoxylin and eosin (H&E) and then examined according to standard techniques.


Paraffin embedded tissues were sectioned at 4 µm, mounted on coated slides4 and allowed to dry for 1 h at 60°C and then overnight at 37°C. Thereafter the sections were deparafinised and incubated with polyclonal rabbit anti-human von Willebrand factor5 (1/200) a marker for endothelial cells. A standard immunohistochemical avidin biotin complex method with diaminobenzidine as chromogen was used (Envision)5. Samples of the spleen and large intestine, used as positive controls, were run concurrently for the antibody tested.


The lesions in the vertebral bodies of C2, C3, C4 and of the mass in the epidural space were consistent with a haemangiosarcoma (Fig 4). It was unclear whether the origin of the mass was in the vertebral body with infiltration in the epidural space or vice versa.

Figure 4.

Section of the third cervical vertebra (a) and mass in the epidural space (b) stained with haematoxylin-eosin (bar 500 µm). a) showing proliferation of large blood-filled vascular spaces lined by single layers of endothelial cells (1) and necrotic bone tissue (2). (normal bone tissue: 3). b) the mass is surrounded by the dura mater (1), the mass itself consists of a large amount of adipose tissue (2) and blood-filled vascular spaces (3).


Tumours of vascular origin are very rare in the horse. A prevalence of 0.058% is reported in the literature (Kennedy and Brown 1993). Haemangiosarcoma seems to be more frequent than haemangioma and middle-aged to older horses are predominantly affected, although it may appear at any age (6 months to 27 years have been previously reported in the literature) (Johns et al. 2005).

Plasmacellular myeloma, fibrosarcoma, lymphosarcoma and melanoma have been described at the level of the vertebral canal and can cause vertebral compression, while haemangiosarcoma is a rare cause of neurological signs in horses. Three other case reports described a haemangiosarcoma as the cause of spinal cord compression. In these cases, osteolysis of the vertebral bodies was present and a cervical myelogram showed an extradural space-occupying lesion. Also in these cases the musculature adjacent to the cervical vertebrae was involved (Kennedy and Brown 1993; Newton-Clarke et al. 1994; MacGillivray et al. 2003). In the previously described cases, cervical myelography could show the extra-dural extension of the mass, located dorsally or ventrally, resulting in a clear dorsal or ventral deviation of the contrast columns. In our case, however, the mass was located laterally. On myelography lateral compressive lesions cause widening of the spinal cord and narrowing of the contrast medium column on the lateral projection. These lesions are difficult to confirm with a ventrodorsal projection in the horse because of numerous superimpositions and increased scattered radiation, both of which decrease radiographic detail (Tucker and Gavin 2002). As also demonstrated in this case, CT myelography may be useful in distinguishing lateral compressive lesions of the spinal cord from an intramedullary lesion. Lateral compressive lesions appear as a circumferential loss of contrast agent (Moore et al. 1992). In our patient, only CT detected the lesions present at the level of C2, which were defined by post mortem examination. This finding demonstrates the high sensitivity of CT for bone lesions. However, the lesions described at the level of C4 were not noticed. Reducing the slice thickness could increase the sensitivity of the examination. Computed tomographic myelography yields the most sensitive evaluation and qualitative information regarding the source, severity and location of spinal cord compression (Moore et al. 1992). Nevertheless, CT is an expensive technique, not widely available that requires general anaesthesia. However, standing CT of the head and neck of the sedated horse is possible in a few facilities (Nelson 2008). Gantry size limits the investigation of the cervical region to C4–C5; however, using a large bore C6 can be included in the examination. In this case the patient was already anaesthetised for the myelogram and C3 fitted perfectly in the gantry.

A previous report of CT myelography of the cervical region of the horse dealt with cervical stenotic myelopathy. Myelography and CT myelography were highly sensitive to detect spinal cord compression, but CT myelography was more accurate in the demonstration of the source, severity and location of this compression (Moore et al. 1992).

The general osteolytic appearance of C3, with compression of the spinal cord could have been caused by an infectious or neoplastic process. The absence of cervical pain and any other sign of an infectious process in our patient suggested a neoplastic process to be more likely. Previous studies reported that ante mortem diagnosis of a haemangiosarcoma is difficult. In a study by Southwood et al. (2000) only 4 of the 35 patients with a disseminated haemangiosarcoma were diagnosed ante mortem. Ante mortem diagnosis is possible only when a mass is readily recognisable and accessible for biopsy (Johns et al. 2005). CT-guided biopsy would have been an option for diagnosis in this case.

Treatment of haemangiosarcoma is generally unrewarding in all species. Reported treatments involve surgical excision, radiation and chemotherapy. In most cases treatment is considered more palliative than curative and nonsteroidal anti-inflammatory drugs can be used (Southwood et al. 2000).

Osseous haemangiosarcoma should be included in the differential diagnosis of compressive lesions in the vertebral canal. Computed tomographic myelography can give additional information on the extent and exact location of these lesion and this information can be useful for prognosis, biopsy and/or surgery.

Authors' declaration of interests

No conflicts of interest have been declared.

Manufacturers' addresses

1 Magstim 200 Whitland, UK.

2 General Electric Co. Healthcare, Milwaukee, Wisconsin, USA.

3 GE Prospeed, General Electric Co., Medical Systems, Milwaukee, Wisconsin, USA.

4 SuperFrost Plus, Menzel-Gläser, Braunschweig, Germany.

5 Dakocytomation, DK-2600 Glostrup, Denmark.