Suspected primary hematomyelia in 3 dogs

Abstract

Primary hematomyelia refers to hemorrhage occurring within the spinal cord without an identifiable etiology. Clinical signs, magnetic resonance imaging characteristics, and histopathological findings are described. Diagnosis was made through histological analysis and rule-outs for underlying factors. Following removal of the hematoma, neurologic deficits improved, although some residual deficits persisted.

Spinal cord dysfunction can result from a multitude of diseases and clinicians often must make a rapid diagnosis in order to prevent additional neurologic injury and implement appropriate therapy. The cornerstone of diagnosis is the neurologic examination, which localizes the disease within the nervous system. However, many diseases present with similar neurologic deficits and time course implying that auxiliary diagnostics are required to determine the nature of the disease. In particular, it is well-recognized that animals that present with transverse myelopathies usually require advanced spinal imaging to clarify diagnosis and suggest treatment plans (1).

Intramedullary lesions in the spinal cord of dogs are relatively uncommon (2,3). Many historical reports have identified primary or secondary neoplastic lesions, such as ependymoma, astrocytoma, and nephroblastoma (4). In this report we describe 3 dogs each of which was identified as having an intramedullary spinal lesion subsequently diagnosed as a hematoma. This condition, although rare, is a possible differential diagnosis for any space-occupying lesion within the spinal cord parenchyma.

Materials and methods

Following surgical diagnosis of hematomyelia in Case 3 herein, a search was made through medical records at Iowa State University (ISU) and Washington State University (WSU) veterinary hospitals between the years 1997 and 2012 for similar cases. Cases were included if hematomyelia without an underlying cause (such as disc herniation, traumatic injury) was identified after surgical exploration, and further suggested by histologic analysis, without the diagnosis of primary coagulation disorders. In total 3 cases were identified: 2 from WSU and 1 from ISU (index case).

The first 2 dogs were imaged with a 1.0 Tesla magnetic resonance imaging system (Gyroscan T10-NT; Philips Medical Systems, Shelton, Connecticut, USA) and the third dog was imaged with a 1.5 Tesla MR system (Signa Excite^®; GE Healthcare, Waukesha, Wisconsin, USA). All dogs were imaged with phased array spine coils.

Image series evaluated for the first 2 dogs included T1-weighted (T1W), T2-weighted (T2W), and T1-weighted SE images immediately following intravenous administration of 0.2 mL/kg gadopentetate dimeglumine (Magnevist; Bayer Healthcare Pharmaceuticals, Wayne, New Jersey, USA). In addition, a single-shot turbo spin echo was acquired in Case 1. In Case 2, a T2-weighted fluid attenuated inversion recovery sequence (T2W FLAIR) was obtained. All images were acquired in standard axial and sagittal planes with the patient in dorsal recumbency.

Image series evaluated for Case 3 in all 3 standard planes included T1W spin echo, T2W fast spin echo, T2W single shot fast spin echo, T2-weighted fluid attenuated inversion recovery sequence (T2W FLAIR), T2* gradient echo (T2*W), and T1W immediately following gadolinium contrast administration as described above. Fat saturation was applied to the T2W sequence in the dorsal plane. The patient was positioned in dorsal recumbency.

Standards of care consistent with institutional guidelines were maintained in all cases. Owners’ consent was obtained for all diagnostic and surgical procedures.

Case descriptions

Case 1

Dog 1 was an 8-month-old female Labrador retriever presented to WSU Veterinary Teaching Hospital (VTH) for progressive tetraparesis, worsening gradually over 7 d. On examination the dog appeared quiet, was reluctant to move freely, had increased muscle tone in all 4 limbs, and was unable to rise from a recumbent position. The dog had a kyphotic posture with her neck positioned in flexion. Neurological findings of delayed postural reactions in all limbs with normal spinal and cranial nerve reflex function were suggestive of a lesion within the 1st to 5th cervical spinal cord segments. Additionally, there was pain on cervical palpation. A complete blood (cell) count (CBC) and serum chemistry analysis were unremarkable.

Magnetic resonance imaging (MRI) of the cervical and cranial thoracic spine revealed a distinctly marginated tubular-shaped intramedullary mass lesion extending from the mid-body of the first thoracic vertebra to the mid-body of the second thoracic vertebra. On transverse views the lesion involved approximately 80% of the transverse diameter of the cord and was asymmetrical with a predominance on the left side. On T2W images the central portion of the mass was isointense and was surrounded by a non-uniform hyperintense ring (Figure 1). The mass was uniformly hyperintense on T1W images and did not enhance following intravenous gadolinium, consistent with a diagnosis of subacute hematoma. The surprising location of the lesion given the neurologic examination was attributed to the loss of triceps function that can be associated with lesions in the cranial thoracic spinal cord without affecting the flexor reflexes.

Figure 1.

Case 1, T2-weighted sagittal (A) and axial (B) magnetic resonance images illustrating a well-marginated tubular shaped intradural mass lesion extending from the mid-body of the first to the second thoracic vertebra. The mass has a heterogenous signal intensity. The white line in (A) represents the level of image (B).

Cerebrospinal fluid collected from the lumbar region revealed a mild elevation in protein (0.4 g/L; reference: < 0.25 g/L) with low cellularity (no white blood cells observed with 270 erythrocytes/μL on fluid analysis; 19 mature lymphocytes, 4 mononuclear cells, and 2 hypersegmented neutrophils observed on cytospin). Despite blood contamination there was no evidence of erythrophagocytosis; no microorganisms or atypical cells were identified.

Surgical exploration via dorsal laminectomy commencing cranially at the 7th cervical vertebra revealed an expanded and blue-green discoloration of the dural tube. Durotomy and mid-sagittal myelotomy exposed a dark blue-green mass of gelatinous consistency within the spinal cord parenchyma. This abnormal tissue was removed through a combination of manual disruption, suction aspiration, and fragment removal using forceps. Once the bulk of the lesion was removed, a biopsy was collected from the adjacent lesion-spinal cord interface. Abnormal tissues were submitted for histologic evaluation and cultured for infectious agents. Histological examination identified an organized hematoma. No abnormalities were present in the lesion-spinal cord interface. Blood clotting times (one-stage prothrombin time and activated partial thromboplastin time) and von Willebrand factor antigen were within reference ranges. Aerobic and anaerobic bacterial and fungal cultures from the mass and the surrounding cavity did not grow any organisms.

The dog recovered the ability to walk without support (ambulatory tetraparesis) during the following 20 d with no signs of post-operative complications. At 4 mo after surgery the dog was walking but had clinically obvious ataxia and paraparesis.

Case 2

Dog 2 was a 6-year-old male boxer presented to WSU VTH for progressive paraparesis. Signs had begun in the right pelvic limb and had progressed to involve both pelvic limbs gradually over the ensuing 3 wk. The referring veterinarian had initiated medical therapy using dexamethasone sodium phosphate and methocarbamol (Robaxin tablets; Actient Pharmaceuticals, Lake Forest, Illinois, USA) without a change in the clinical sign progression. For 48 h immediately prior to presentation, there was hematuria, vomiting, regurgitation, and decreased appetite.

Conscious proprioceptive deficits in both pelvic limbs with intact spinal reflexes were present on neurologic examination, consistent with a lesion within the 3rd thoracic to the 3rd lumbar spinal cord segments. Cranial nerve and thoracic limb examinations were normal. A cardiac arrhythmia was noted during physical examination. Right arrhythmogenic cardiomyopathy was subsequently identified through electrocardiogram and echocardiogram.

A distinctly marginated ovoid intramedullary mass extending from the caudal endplate of the tenth thoracic vertebra to the cranial endplate of the eleventh thoracic vertebra was identified through MRI of the thoracolumbar spinal cord (Figure 2). The mass was predominantly right-sided, occupying approximately 95% of the cord at its greatest diameter. The mass had rounded and tapering longitudinal margins, which were hyperintense on T2W sequences. On T2W images the mass appeared uniformly hyperintense compared to the spinal cord. The mass was mildly hyperintense on T1W images without attenuation on T1W FLAIR sequence images. There was no contrast enhancement of the lesion following intravenous gadolinium administration.

Figure 2.

Case 2, T2-weighted sagittal (A), axial (B), and axial fluid attenuated inversion recovery sequence (C) magnetic resonance images of a distinctly marginated intramedullary mass that encompasses approximately 95% of the spinal cord diameter immediately dorsal to the intervertebral disk space between the 10th and 11th thoracic vertebrae. Note the T2-weighted high signal intensity that persists within the mass on the fluid attenuated inversion recovery sequence images. The white line in (A) represents the level of image (B) and (C).

There was increased protein concentration (0.61 g/L; reference: < 0.50 g/L) and creatine kinase (0.35 μkat/L; reference interval: 0.17 to 0.30 μkat/L) in CSF collected from the lumbar region with a normal cell count and no abnormal cells or infectious organisms. The lesion was surgically explored through a hemilaminectomy centered over the lesion. Durotomy and myelotomy via hemilaminectomy between the 9th and 12th thoracic vertebrae revealed an apparent hematoma that was removed using a combination of micromanipulators and suction aspiration. Histologic analysis of this abnormal tissue revealed erythrocytes with scant fibrin and few hemosiderin containing macrophages, consistent with a blood clot. Anaerobic, aerobic, and fungal cultures from the abnormal tissue were negative. Coagulation times (one-stage prothrombin time and activated partial thromboplastin time) were within reference ranges.

After surgery the dog showed gradual recovery during a hospitalization period of 22 d, at which time the dog could walk short distances but required hindquarter (sling) support to squat to defecate. The dog was subsequently lost to follow-up.

Case 3

Dog 3 was an 18-month-old Siberian husky mixed-breed referred to ISU Lloyd Veterinary Medical Center with an approximately 48 h history of progressive pelvic limb neurologic dysfunction. Initial signs included lethargy and ataxia which progressed to non-ambulatory paraparesis, incontinence, and loss of voluntary tail movement. Spinal and thoracic radiographs, CBC, serum biochemistry panel, and SNAP 4DX test (Anaplasma, Ehrlichia, B. burgdorferi, Heartworm; IDEXX Laboratories, Westbrook, Maine, USA) were all unremarkable.

Neurologic examination detected bilaterally depressed flexor reflexes in the pelvic limbs plus reduced anal tone and perineal reflex consistent with a lesion within the 6th lumbar to the 3rd sacral spinal cord segments. The bladder was easily expressed by manual palpation. Thoracic limbs, panniculus reflex, cranial nerve examination, and mentation were all normal.

A distinctly marginated non-contrast enhancing oblong intramedullary mass was observed on MRI of the lumbar spinal cord. The lesion extended from the cranial portion of the fourth lumbar vertebra to the mid-body of the 5th lumbar vertebra (Figure 3). On T2W scans the lesion was isointense to the spinal cord, surrounded by a narrow hypointense region with a hyperintense region cranial to the lesion. T1W images revealed an isointense lesion, and suspected spinal cord swelling with loss of epidural fat and CSF in the area surrounding the lesion.

Figure 3.

Case 3, T2-weighted sagittal (A), T2-weighted dorsal with fat saturation (B) and axial T2* gradient echo (C) images of an intramedullary mass over the bodies of the 4th and 5th lumbar vertebrae, demonstrating the laminated appearance of the lesion. The white line in (A) represents the level of image (C).

At the cranial edge of the mass a hypointense linear feature extended cranially within the dorsal spinal cord to the level of the intervertebral disc between lumbar vertebrae 3 and 4. This linear feature was hypointense to the mass and spinal cord. On the T2*W sequence the hypointense line remained uniformly hypointense. This region was continuous with a thin hypointense region immediately surrounding the mass along its entire length.

The lesion was surgically exposed via a dorsal laminectomy centered over the 5th lumbar vertebra. Mid-sagittal myelotomy revealed a dark red friable lesion with evidence of surrounding chronic hemorrhage, which was excised using micromanipulators and suction aspiration. Histological analysis of the excised friable lesion revealed a combination of erythrocytes, scant leukocytes, and strands of fibrin, consistent with clotted blood. There was no evidence of neoplastic cells or other organized structures.

The dog recovered increasing pelvic limb function and was able to ambulate with minimal support by 8 d after surgery. The dog continued to improve and at 9 mo after surgery was ambulating well with mild to moderate hind limb paresis. However, the dog continued to have episodes of urinary and fecal incontinence.

Discussion

Hematomyelia (intramedullary spinal hemorrhage) is a rarely reported cause of myelopathy in dogs. In humans, in which it is also uncommon (5), the condition is subdivided based on traumatic and non-traumatic inciting events. Non-traumatic hematomyelia is a rare clinical problem which can be further subdivided into primary (idiopathic) and secondary hematomyelia. Causes of secondary hematomyelia include vascular malformations, infarcts, neoplasia, aneurysm, syrinx, or coagulation disorders (6). Clinical signs in humans vary with location and severity but usually begin with severe spinal pain.

Most cases of hematomyelia reported in animals, while not specifically classified as such, have features characteristic of secondary hematomyelia. A variety of underlying structural or coagulation disorders have been identified, for instance abnormal growth of blood vessels such as hamartomas, which are intramedullary spinal lesions consisting of an overgrowth of normal vascular tissue and fibrillar stroma that can cause a mass effect and lead to secondary hematomyelia (7). Arteriovenous malformations can have a similar presentation to primary hematomyelia on both gross examination and advanced imaging (8). Arteriovenous malformations differ from primary hematomyelia histologically in that they are composed of tangled masses of blood vessels as opposed to a simple blood clot. Arteriovenous malformations and other vascular malformations can have a similar appearance on MRI (9) and differentiation requires histopathology to identify the specific nature of the lesion (10).

Other vascular related abnormalities leading to myelopathies have also been described in dogs. These include aneurysm of the venous sinus leading to spinal cord compression (11), arteriovenous malformations leading to hypoxic insult (12), hemangiomas leading to a mass effect within the spinal cord (12), and cavernous angioma (13). Intraspinal hemorrhage has also been observed in dogs suffering from congenital (14) and acquired (15) coagulopathies. Other causes of secondary hematomyelia include iatrogenic (16), and Angiostrongylus vasorum infection (17).

There are only 2 previously reported canine cases with clinical features consistent with primary hematomyelia (Table 1) (18,19). In the first case the authors concluded that it was likely a vascular malformation though no abnormal vascular tissue was identified, and, unlike the 3 cases reported here, a relapse of clinical signs occurred at 6 mo after surgery (18). In a more recent case the dog had a progressive history of approximately 6 mo, differing greatly in time scale from the 3 cases reported here (19).

Table 1.

Summary of imaging findings in all reported cases of hematomyelia in dogs

Case	Signalment	Presenting signs	Imaging technique and findings	Vertebral location	Post-operative outcome
Martin et al, 1986 (18)	3-year-old (FS) collie-mix	3-week progressive cervical pain + tetraparesis	Myelogram Intramedullary lesion	C3–C4	Improved 6-weeks mild weakness 6-months relapse episode responded well to corticosteroids
Thibaud et al, 2008 (19)	4-year-old (M) Jack Russell terrier	6-month progressive cervical pain + tetraparesis	MRI T1W — Hyperintense T2W — Hypointense + hyperintense periphery Non-contrast enhancing	C4	Improved 18-months mild residual deficits
Dog 1 Barker et al, present study	8-month-old (F) Labrador retriever	1-week progressive tetraparesis	MRI T1W — Uniform hyperintense T2W — Centrally isointense Non-contrast enhancing	T1–2	Improved 4-months ambulatory tetraparesis
Dog 2 Barker et al, present study	6-year-old (M) boxer	3-week progressive paraparesis	MRI T1W — Mildly hyperintense T2W — Uniform hyperintense Non-contrast enhancing	T10–11	Improved 22-days ambulatory with hind limb support
Dog 3 Barker et al, present study	18-month-old (FS) Siberian husky	2-day progressive paraparesis with urinary incontinence	MRI T1W — Uniform isointense T2W — Centrally isointense Non-contrast enhancing	L4–5	Improved 9-months ambulatory with mild paresis Persistent urinary + fecal incontinence

F — female; M — male; FS — female spayed; MRI — magnetic resonance imaging; T1W — T1-weighted MRI sequence scans; T2W — T2-weighted MRI sequence scans.

Considerable variability was observed in MRI characteristics of the lesions in the 3 cases in the current report (Table 2). The only consistent findings for all 3 cases were the absence of contrast enhancement and mild T2W hyperintensity associated with the poles of the lesion. There are no MRI findings, even T2* sequences, that would definitively distinguish primary hemorrhage from other intramedullary lesions with a hemorrhagic component.

Table 2.

Summary of the expected and observed magnetic resonance imaging (MRI) findings in the cases in this report

Dog	Time from onset of clinical signs to imaging (days)	T1 expected	T1 actual	T2 expected	T2 actual
1	7	Hyperintense	Hyperintense	Hypointense	Isointense plus hyperintense ring
2	21	Hyperintense	Mildly hyperintense	Hyperintense plus hypointense rim	Hyperintense
3	2	Iso- to hyperintense	Isointense	Hypointense	Isointense plus hypointense border

All intensities are in comparison to spinal cord.

Variability in the MRI appearance of lesions in the central nervous system is well-recognized, although in general intramedullary tumors are T1W hypointense and T2W hyperintense to the spinal cord with evidence of contrast enhancement (4,20). The MRI appearance of central nervous system hemorrhage is known to vary with time because of changes in iron redox state within hemoglobin and red blood cells (21). Given the suspected duration of signs, dog 1 would have been expected to have had a lesion that appeared hyperintense on T1W and hypointense on T2W sequences, dog 2 hyperintense on T1W and hyperintense with a hypointense rim on T2W sequences, and dog 3 iso- to hyper-intense on T1W and hypointense on T2W sequences.

As summarized in Table 2, the appearance of the lesions differed from that expected. Although the reasons are difficult to explain, it is well-recognized that the MRI appearance of hemorrhage might vary with the field strength of the magnet (22). A further likely cause is the potential for continual slow bleeding in hematomyelia cases, which would explain the progressive nature of the clinical signs, particularly in dogs 1 and 2. This could also be an explanation for the greater variability in the periphery of the lesion, a potential site for newly accumulating hemorrhagic fluid. Edema, non-hemorrhagic fluid accumulation, venous hypertension or ischemic damage are also probable contributing factors.

The T2*W sequence has been used to identify hemorrhage, which appears hypointense or as a signal void (23). This appearance has been described in cases of secondary hematomyelia in the dog (16). Case 3 was the only one in which a T2*W scan was performed and demonstrated the hypointense region expected of hemorrhagic material (Figure 3). This appears to be the only report in the veterinary literature that describes the appearance of primary hematomyelia on a T2*W scan. In future, it is possible that MR angiography may be able to differentiate vascular malformations or angiomas from primary hematomyelia.

Additional diagnostic testing did not reveal evidence of concomitant intramedullary lesions. Cases 1 and 2, in which CSF analysis was performed, revealed elevated protein, which is non-specific. Other reported cases of spinal cord hemorrhage showed similar results (15,19). Blood clotting times and Von Willebrand factor were normal in the cases in which they were applied, although these tests alone do not provide a complete investigation of clotting status. In the cases described in this report, no evidence of a hemostatic disorder was detected, and no evidence of abnormal hemorrhage or petechiation was observed before, during, or after surgery.

In these 3 cases histological analysis of the removed mass was consistent with organized hemorrhage. As associated hemorrhage has been reported with cavernous angiomas (13), hamartomas (7,24), and intramedullary cavernous malformations (25), it is conceivable that this was the underlying cause in these cases. However, most such lesions consist predominantly of abnormal vascular tissue, together with a small amount of associated hemorrhage. The absence of any organized structure or vessel formation on histological examination, coupled with the lack of contrast enhancement on MRI would make the presence of a vascular abnormality very unlikely.

Surgical intervention is associated with improvement in neurologic function in this series and other reports (18,19). In humans, surgical intervention has also been successful (26), although recovery is related to severity of the condition. The failure to improve fully and persistence of urinary and fecal incontinence have been reported in other cases of severe injury to the spinal cord, despite recovery of ambulation (27).

Medical therapy was not pursued in any of the cases presented here because of the progressive clinical signs, the need to biopsy the lesions, and/or to potentially remove suspected malignant masses within the cord. There are limited data in the human literature for medical therapy for primary hematomyelia, which is considered a surgical disease (5). The case discussed by Thibaud et al (19) initially responded to corticosteroid before requiring surgery, which is consistent with a report in a human (28). The dog described by Martin et al (18) also responded to corticosteroid therapy after a relapse episode, although the neurologic signs were significantly less severe at that time.

The 3 cases described here represent a rare condition leading to myelopathy in dogs. They show considerable variability in onset, progression, and MRI appearance and also vary from previously described cases of suspected primary hematomyelia. Primary hematomyelia should be considered a differential for intramedullary lesions in dogs. Though complete recovery was not achieved, the clinical improvement with surgical intervention highlights the value of performing exploratory and decompressive surgery for cases with a similar presentation. CVJ