Presumed cervical subperiosteal vertebral hemorrhage in nine greyhounds

Abstract

Background

Spontaneous subperiosteal vertebral hemorrhages (SSVHs) are rarely documented in veterinary literature, with only a few cases reported in dogs.

Hypothesis/Objectives

Describe the clinical presentation, imaging findings, and outcome of presumptive cervical SSVHs in dogs and perform a prospective cadaveric study describing the morphology of lesions.

Animals

Nine greyhounds presented with acute onset of cervical myelopathy.

Methods

Multicenter descriptive study. The databases were searched for dogs that underwent magnetic resonance imaging (MRI), computed tomography (CT), or both of cervical vertebral column with SSVH as the main or differential diagnosis.

Results

Nine middle-aged greyhounds met the inclusion criteria. On MRI, all lesions were at third cervical vertebra or fourth cervical vertebra or both vertebral bodies. Most were bilateral, symmetrical, and ventrolaterally located, causing spinal cord compression. Lesions showed homogeneous T2W hyperintense and T1W isointense to hyperintense signal relative to gray matter; they were non-contrast-enhancing and exhibited a peripheral rim of susceptibility artifact on T2^*W images. On CT images, extradural lesions were homogeneously hyperattenuating (60-80 Hounsfield Units). All dogs had an acute onset of clinical signs and were treated medically. Clinical improvement was observed in all cases, and all nonambulatory dogs became ambulatory within 9 days. Clinical signs did not recur during the follow-up period (median, 254 days; range, 5-1217 days).

Conclusions and clinical importance

Spontaneous subperiosteal vertebral hemorrhage should be included in the differential diagnosis of acute onset cervical myelopathy in greyhounds. Imaging characteristics can aid in differentiating subperiosteal hemorrhage from other extramedullary or extradural lesions. The prognosis is excellent despite the severity of neurologic impairment at presentation.

Introduction

Subperiosteal hemorrhage is an accumulation of extravasated blood between the periosteum and the bone.¹ The periosteum is a layer of vascularized connective tissue covering nonarticular surfaces of all bones and merges with the attachments of tendons and ligaments.² Subperiosteal hemorrhages are rarely reported in veterinary literature. They have been reported to affect the calvarium in 5 dogs,^3,4 and the thoracolumbar vertebral column in 3 dogs.^1,5 In humans, they occur most commonly in the calvarium,⁴ with the orbit being frequently affected,⁶ but they also have been reported anecdotally in other bones (eg, ilium,⁷ tibia,⁸ and vertebra ^9,10).

The main causes of subperiosteal hemorrhage in humans include trauma and neoplasia, whereas spontaneous subperiosteal hemorrhage has been linked to coagulopathy, hypertension, or increased intraocular pressure.^6–8,11,12 In adults, the orbital area is commonly affected, likely because of the specific features of the periorbital anatomy. The periorbital periosteum is not tightly fixed to the underlying bone, and the diploic veins that traverse the space between periosteum and orbital bone are susceptible to rupture.¹¹

Spontaneous subperiosteal vertebral hemorrhage has been described in 3 dogs. A myelogram was performed in 2 dogs, with a bilateral symmetric extradural compressive lesion present at the level of L3.⁵ The histopathologic diagnosis was made at necropsy. Magnetic resonance imaging was performed in the remaining dog, in which surgery confirmed spontaneous subperiosteal vertebral hemorrhage (SSVH) at the level of L1.¹ So far, SSVH has not been reported in the cervical vertebral column, and the computed tomography (CT) characteristics of SSVH have not been described.

The purpose of our study was to document presumed SSVHs in the cervical vertebral column and describe their clinical and imaging features.

Materials and methods

Ours was a retrospective and prospective study. The retrospective component of study was approved by the Research Ethics Committee of the School of Biodiversity, One Health and Veterinary Medicine, College of Medical, Veterinary and Life Sciences, University of Glasgow (EA41/23). Medical records of 7 referral institutions from November 2008 to December 2023 were reviewed, and databases were searched for dogs that underwent magnetic resonance imaging (MRI) or CT or both of cervical vertebral column with SSVH as the main differential diagnosis on the imaging report. The prospective component was a cadaveric study. The animal used in cadaveric study was donated by the owner following an approved deceased animal donation program at the Universitat Autònoma de Barcelona (UAB) for anatomical education and research purposes.

Medical records were reviewed by a European College of Veterinary Neurology (ECVN) resident under supervision (I.O.), and relevant data were retrieved. Long-term follow-up was obtained by contacting owners by telephone or email and using validated questionnaire responses supplemental material questionnaire.

The MRI findings were reviewed by a board-certified radiologist (C.M.) and a board-certified neurologist (J.B.). Inclusion criteria included presence of (1) an extradural lesion, (2) in a broad base with adjacent vertebral body (with no evidence of epidural fat signal between lesion and vertebra), and (3) with extension limited to length of vertebral body (not crossing intervertebral foramina). The presence of other additional or connected lesions within the epidural space was not an exclusion criterion. The MRI studies included at least T2-weighted (T2W) images in sagittal and transverse planes, T1-weighted (T1W) images at least in the transverse plane, and T2-weighted gradient recalled echo (T2^*W) images in the transverse plane. Homogeneity and attenuation were recorded on CT images. Post-contrast imaging was not required. The exact MRI characteristics of the lesions are presented in Appendix S1.

To understand shape and boundaries of lesions, anatomical study of cervical cross-sections was performed by V.A. on the cadaver of a mixed breed female dog (18 kg). The cadaver for dissection was preserved using 10% formaldehyde solution (200 mL/kg) injected within 24 h after euthanasia via common carotid artery and subsequently stored at 4-6°C for several weeks. The cervical vertebral column, including epaxial musculature, was isolated by gross dissection and frozen at −20°C for transverse sectioning. The 1-cm-thick section of the third cervical vertebra (C3) was thawed and analyzed under a magnifying glass to identify the periosteal layer. The epidural and subperiosteal spaces were explored to identify anatomical boundaries that would shape lesions. The subperiosteal space was filled with paper pulp to mimic a subperiosteal space-occupying lesion.

Results

Nine dogs from 4 referral institutions in the United Kingdom (UK) and Spain met the inclusion criteria, all of which were greyhounds. The study period spanned 15 years (November 2008 to December 2023).

Clinical findings

Of the 9 greyhounds in the study, 6 were females (4 spayed and 2 intact) and 3 were males (all neutered). The median age was 4 years and 8 months (range, 2-6 years). Weight was available in 7 dogs, with a median of 30.3 kg (range, 27.4-40.1 kg). All clinical findings are summarized in Table S1.

Two dogs had episodes of cyanosis and gasping before presentation, requiring intubation by the referring veterinarians. Another dog had multiple jaw-champing episodes after admission (Table S1).

Physical examination identified signs of hemorrhagic diathesis (eg, petechiae, ecchymoses) in 2 dogs. These were located on the skin of the abdominal wall in one dog and affected the sclera and prepuce in the other dog.

Dogs were presented with peracute (<24 h, 6/9) or acute (1-3 days, 3/9) onset of cervical myelopathy. None of the dogs sustained a major traumatic event before presentation. No dog had shown improvement in the clinical signs before presentation to the referral center (median, 24 h; range, 4-48 h). One dog was ambulatory tetraparetic, 4 were nonambulatory tetraparetic, and 4 were tetraplegic. Proprioception was impaired in all 4 limbs in 8 nonambulatory dogs and hopping was mildly delayed in all limbs in one ambulatory dog. The neuroanatomical localization was to the C1-C5 (fifth cervical vertebra) spinal cord segments in 5 dogs (normal thoracic limb withdrawal reflexes); the remaining 4 dogs localized to the C6-T2 spinal cord segments (absent or decreased thoracic limb reflexes). In a single dog, which was nonambulatory tetraparetic, and anisocoria with miosis was noted in the left eye. Cervical discomfort was noted on neurologic examination in 3 cases. All dogs had intact nociception in their limbs.

Clinicopathological findings

Hematology performed in all dogs showed no remarkable changes; platelet counts were normal in all dogs. Serum biochemistry showed mild to moderate increases of ALT activity in 2 dogs; in 2 dogs, mild nonspecific electrolyte imbalances were noted, and mildly increased CK activity was observed in 2 dogs. One dog had mildly increased serum creatinine concentration.

At least one test assessing coagulopathy was performed in most (7/9) of the cases. Two dogs had prolonged buccal mucosa bleeding time (BMBT), which was tested in 3 dogs. Coagulation times (prothrombin time [PT] and activated partial thromboplastin time [aPTT]) were performed in 7 dogs (including the dogs with prolonged BMBT) and were all normal. An Angiostrongylus vasorum SNAP test was done in 7 dogs and was negative in all of them. Von Willebrand factor (vWF) was measured in 2 dogs and was decreased in one dog, that also had focal scleral hemorrhage and ecchymosis of the lateral prepuce.

Systemic blood pressure was measured in 2 dogs and was normal in both. The blood pressure measurement technique used (eg, oscillometric, Doppler) was not recorded. One dog had thoracic radiographs and abdominal ultrasonography performed, both of which were unremarkable. Cerebrospinal fluid analysis (CSF) was performed in 1 dog, and was reported to be normal. A summary of the clinicopathological findings is provided in Table S2.

Magnetic resonance imaging findings

All dogs underwent MRI (1.5 T, Magnetom Siemens, Munich, Germany; Philips Intera/Achieva, Amsterdam, Netherlands; Vantage Elan Canon, Otawara, Tochigi, Japan), and a CT (16-slice Somatom, Siemens, Erlangen, Germany) scan additionally was performed in 1 dog. All dogs had extradural lesions, according to the inclusion criteria. Seventeen compressive lesions were observed in 9 dogs. The lesions extended along the vertebral body of C3 in 7 dogs and fourth cervical vertebra (C4) in 4 dogs. Two dogs had lesions affecting both C3 and C4. All lesions were well-defined with regular margins, and none of the lesions crossed the midline ventrally. Fifteen lesions were lenticular (15/17), and 2 were crescent-shaped (2/17). All lesions but one reached the level of the dorsal aspect of the pedicle, with 8 of 17 reaching the dorsal lamina. In 6 dogs, the lesions were bilateral and symmetrical. Three dogs had unilateral lesions, with one of these having lesions in 2 separate locations. One dog had both unilateral and bilateral lesions. The crescent-shaped lesions were unilateral. All bilateral lesions were ventrolateral, and 2 unilateral lesions were latero-lateral (Table S3).

The MRI lesions were homogeneously T2W hyperintense in all dogs but one (8/9). The T1W signal was hyperintense in 4 dogs (4/9) and iso- to hyperintense in 5 (5/9) dogs. T1W postcontrast (0.1 mmol/kg gadobenate meglumine [Dotarem, Guerbert Princeton, NJ, USA]) images were available in 7 dogs, none of which showed contrast enhancement (Figures 1 and 2). In all dogs but one (8/9), a thin hypointense peripheral rim was observed at the interface between the lesion and the epidural fat in T1W and T2W spin echo sequences. A complete (8/9) or incomplete (1/9) rim of susceptibility artifact on T2^*W sequence was present in all dogs. A focal signal void was observed within the caudal aspect of the lesion in a dog with a unilateral SSVH, corresponding to the only case with a heterogeneous lesion on T2W.

Figure 1.

(Case 7) Transverse MRI images of bilateral symmetrical ventrolateral extradural lesions in the cervical vertebral column, at the level of C3. The lesions were T2W (A) and T2^*W (B) hyperintense, T1W (C) isointense to hyperintense to the spinal cord gray matter, with no significant contrast enhancement on T1W fat saturation images (D). Note the lenticular shape, well-defined, and regular margins of the lesions, the susceptibility artifact with an incomplete peripheral rim (B) and the resulting moderate spinal cord compression. Abbreviations: C3 = third cervical vertebra; MRI = magnetic resonance imaging; T1W = T1 weighted; T2W = T2 weighted; T2^*W = T2^* weighted/gradient echo.

Figure 2.

(Case 2) Transverse (A-D) and dorsal CISS (E) MRI images of a right-sided unilateral extradural lesion at the level of C3, consistent with subperiosteal hemorrhage. The lesion was T2W (A), T2^*W (B), and T1W (C) hyperintense to the spinal cord gray matter, with no significant contrast enhancement on T1W post-contrast images (D). Note the crescent shape, well-defined, and regular margins of the lesion, the susceptibility artifact with a complete peripheral rim (B) and the resulting moderate spinal cord compression. Abbreviations: C3 = third cervical vertebra; CISS = constructive interference in steady state; MRI = magnetic resonance imaging; T1W = T1 weighted; T2W = T2 weighted; T2^*W = T2^* weighted/gradient echo.

The spinal cord was compressed in all dogs. As assessed by the cross-sectional area, the extradural lesion resulted in mild spinal cord compression in 3 dogs and moderate compression in 7 dogs, with 1 dog having a mildly compressive lesion at C3 and a moderately compressive lesion at C4. An ill-defined focal T2W intramedullary hyperintensity was present in 5 cases (5/9) and was associated with mildly compressive lesions in 3 dogs and moderately compressive lesions in 2 dogs. All lesions caused at least partial obliteration of the internal vertebral venous plexuses. No clinically relevant dilatation of the venous plexuses was identified cranial or caudal to the lesions in nearly all cases (8/9; Table S3). In this single case, the internal vertebral venous plexuses and the paravertebral vessels on the opposite site of the spinal compression were mildly distended cranial and caudal to the lesion. Finally, other MRI findings included mild caudal cervical intervertebral disc protrusions in 2 dogs (2/9) and mild central canal dilatation cranial and caudal to the lesion in one dog (1/9). Table S4 summarizes the MRI findings.

Computed tomography findings

The CT findings were concordant with the MRI findings. The lesion in this dog was homogeneously hyperattenuating (60-80 Hounsfield Units) and non-contrast (2 mL/kg iohexol, [Omnipaque Injection, GE Healthcare Ltd, Giles, UK]) enhancing, extending along the length of the C3 vertebral body (Figure 3).

Figure 3.

(Case 5) Pre-contrast CT images of a bilateral symmetrical homogeneously hyperattenuating ventrolateral lesion, extending along the length of C3 vertebral body, consistent with subperiosteal hemorrhage. Sagittal (A), transverse (B), and dorsal (C) plane reconstructions of the cervical vertebral column in a modified soft tissue window (WL: 80; WW: 234; 1 mm slice thickness). Note the reference lines (A) for the transverse and dorsal planes. Abbreviations: C3 = third cervical vertebra; CT = computed tomography.

Treatment and outcome

All dogs were treated medically (with supportive care and exercise restriction). Tranexamic acid (Teva, Castleford, UK, Manx Healthcare Ltd, Warwick, UK) was used short-term in 2 dogs. One dog had an 18-day course of 2 antithrombotic drugs. Anti-inflammatory drugs were used in 3 dogs (1/9 meloxicam [Metacam, Boehringer Ingelheim, Bracknell, UK], 2/9 paracetamol [Paracetamol, Crescent Pharma, Basingstoke, UK]) and other forms of analgesia were used in 3 dogs (gabapentin). Medical treatments and outcomes are summarized in Table S5. For the 8 dogs (8/9) that were presented tetraplegic or nonambulatory tetraparetic, the time to ambulation varied from 24 h to 9 days, with a median of 2 days.

Follow-up for 4 weeks or more was available for 8 dogs (8/9). The median follow-up for all dogs was 254 days (range, 5-1217). One dog was euthanized 775 days after presentation because of a suspected osteosarcoma in the left thoracic limb. Complete recovery was achieved in 4 dogs (4/8). In the remaining 4 dogs, residual generalized proprioceptive ataxia was present. Owners did not perceive that the ataxia affected their dogs’ quality of life. Follow-up for over 180 days (6 months) was available in 5 dogs (median, 590 days; range, 254-1217). None of the dogs with follow-up for longer than 6 months experienced recurrence of the clinical signs (eg, becoming acutely nonambulatory tetraparetic).

One dog (case 4), that did not receive antithrombotic treatment, underwent a repeat MRI one month after diagnosis, which showed complete resolution of the previous lesion. This dog was reported to have remained mildly ataxic.

Cadaveric study

Analysis of the 1-cm-thick section of the C3 under a magnifying glass showed that the periosteal layer can easily detach from the vertebral arch and vertebral body. At the level of the dorsal longitudinal ligament, located on the dorsal surfaces of the vertebral bodies, between the ventral internal vertebral venous plexuses, the periosteal layer was strongly adhered to the ligament. Apart from the anatomical structures already described, no additional structures were identified that could explain the shape of the lesions. After white paper pulp was placed in the subperiosteal space, the shape of the lesions closely resembled those observed on MRI images (Figure 4).

Figure 4.

Images of a cross-section at the level of the C3 cervical vertebra with the spinal cord (A and B) and without the spinal cord (C and D). In images E and F, white paper pulp has been introduced into the subperiosteal space unilaterally (E) and bilaterally (F), trying to imitate the lesions closely resembling those observed on MRI images. 1, ventral internal vertebral venous plexuses; 2, epidural ligaments; 3, dorsal longitudinal ligament; 4, periosteal layer; 5, dural layer; and 6, white paper pulp. Abbreviations: C3 = third cervical vertebra; MRI = magnetic resonance imaging.

Discussion

Our study summarizes the clinical presentation, imaging findings, treatment, and outcome of 9 greyhounds with presumed cervical SSVH, a rarely described condition in both the veterinary and human medical literature.^1,5,9,13,14 In our study, all dogs were middle-aged (median age, 4 years 8 months) greyhounds. Previously reported dogs with SSVH included a 13-month-old Doberman pinscher, a 7-year-old Labrador retriever,⁵ and a 4-year-old greyhound.¹ Therefore, greyhounds seem predisposed.

Underlying coagulopathy, systemic hypertension, or vascular malformations are known to account for increased risk of hemorrhage.^6–8,14 One greyhound in our study had low vWF (vWF, 7%; reference interval, 70%-180%), but 10% of healthy greyhounds have vWF concentrations < 30%.^15,16 Furthermore, greyhounds have a known tendency for increased bleeding risk^15,17 because of hyperfibrinolysis,¹⁵ which may be a predisposing cause for the development of SSVH. This possibility could not be confirmed in the dogs in our study because thromboelastographic testing¹⁷ was not performed. Two dogs had prolonged BMBT with normal PT and aPTT, indicating a primary hemostasis defect,¹⁸ one of which had low vWF. This possibility was not evaluated in the second dog. Hypertension also can be a risk factor for developing spontaneous bleeding,¹⁹ but unfortunately, blood pressure was only measured in 2 dogs in our study. In humans, spontaneous subperiosteal hematomas without traumatic or iatrogenic cause are reported.^7,12 Potential risk factors for subperiosteal hematomas in people include underlying coagulopathy, hypertension, vascular malformations,^7,12 and neurofibromatosis.⁸ To our knowledge, SSVH has only been described once in humans.^9,20

The vertebral canal and spinal cord represent anatomic constraints for the expansion of pathological lesions within the epidural space. The subperiosteal nature of the lesions we describe would limit the hemorrhage to the length of the vertebral arch, although extension into the epidural space or the concomitant presence of an epidural hematoma may occur.¹ The periosteum is a layer of connective tissue covering the nonarticulated surfaces of all bones.² There is a current debate in human and veterinary medical studies that the lining of the vertebral canal may not be composed of the typical periosteum, as it does not contain any osteoblasts, osteoclasts, or osteoprogenitor cells. Therefore, several other terms are used for this fibrous lining (eg, peridural membrane, fibrous lining, and cervical fibrous sheath).^9,10,20–22

Anatomical structures (eg, periosteum, ligaments, and venous system) and their differences along the vertebral column (cervical, thoracic, and lumbar region) may explain the pattern, shape, and distribution of the lesions. In agreement with the previous case reports,^1,5 we did not observe any ventral to dorsal compression of the spinal cord. It is likely that the dorsal longitudinal ligament prevents the ventral spread of the subperiosteal bleed, as it tightly connects to the dorsal surface of the vertebral bodies.^2,20,22 Additionally, the meningo-vertebral ligament,⁹ which connects the ventral surface of the inner dura to the median bony ridge of the dorsal aspect of the vertebral bodies near the dorsal longitudinal ligament, forms a ventral anatomic boundary within the epidural space.²² This ligament is widest in the cervical vertebral column²² and has been proposed to be responsible for the bilobed shape of the neoplasms and hydrated nucleus pulposus extrusion (HNPE) in the ventral vertebral canal.²² This factor would explain why the lesions in these cases do not cross the midline.

The T1W hyperintense signal of the lesions on MRI, with the respective hyperattenuating appearance on CT, is consistent with hemorrhage or thrombosis.²² Cervical vertebral venous sinus thrombosis has been described previously in 3 dogs with similar MRI^23,24 and CT²⁵ characteristics to our cases, with lenticular bilaterally symmetric ventrolateral extradural compression of the spinal cord. On MRI, in contrast to our cases, the lesions exhibited T2W signal heterogeneity, intralesional susceptibility artifacts on T2^*W images, and some degree of inhomogeneous contrast enhancement. In humans, T2^*W imaging is the most reliable sequence for detecting venous sinus thrombosis, typically identifying dilated sinuses with central heterogeneous susceptibility artifacts. Conversely, in our cases, susceptibility artifacts were predominantly confined to the periphery.^26,27

In agreement with a previous case report,¹ the internal vertebral venous plexuses were considered to be at least partially obliterated at the level of the hemorrhage. When assessing the dorsal extension of the lesion within the vertebral canal, most lesions extended up to the dorsal aspect of the pedicle or even reached the level of the dorsal lamina with a mild or moderate degree of spinal cord compression. It is unlikely that the vertebral venous sinus could expand to this extent and produce the observed degree of spinal cord compression. Furthermore, it seems unlikely for a distended sinus to have a crescent shape. Interestingly, the thrombotic cases previously reported were linked to generalized dilatation of the cervical vertebral sinuses or a multifocal pattern affecting nonadjacent vertebrae (C2 [second cervical vertebra] and possibly C5). In contrast, the cases described here seem to be localized, with a predilection for vertebral bodies (C3 and C4), and in case of multiple lesions, involved adjacent vertebrae.

The previously described cases of SSVH^1,5 occurred in the thoracolumbar vertebrae, with bilateral and dorsal (but not ventral) compression of the spinal cord caused by the hemorrhage and detachment of the periosteum. Surprisingly, the cases we described lacked dorsal spinal cord compression or detachment of the periosteum, and 2 vertebrae were simultaneously affected in some dogs. The reasons for these findings remain unknown and made us question whether the lesion was truly a SSVH.

Clinically, dogs with cervical vertebral venous sinus thrombosis and the dogs described here behaved differently. Cases of cervical vertebral venous sinus thrombosis presented with pain as the main clinical sign, which often had been ongoing for more than 1 week. Furthermore, a predisposing factor for the hypercoagulable state (eg, chronic prednisolone treatment) was identified in all 3 previously described cases of thrombosis. Our cases had acute onset of marked neurologic dysfunction, lacked disorders or treatments that predispose to a hypercoagulable state, and all dogs rapidly improved (of which only one received antithrombotic treatment). The acute onset, rapid clinical improvement, and resolution of the lesion on repeat MRI without antithrombotic treatment in one dog support the hemorrhagic nature of the disease.

Determination of the location of the hemorrhage to the extradural or subperiosteal space on imaging studies can be challenging.¹ Spinal epidural hematomas more frequently have been described than SSVH in the human and veterinary medical literature.^13,14 Differentiating between epidural and subperiosteal hematomas in imaging studies remains challenging, particularly in cases of unilateral compressive lesions, which were observed in 4 dogs (4/9). In our study, none of the cases was associated with intervertebral disc herniations, which is the most common pathology associated with epidural hematoma.^28–30 Spontaneous epidural hematomas have been described in dogs,³¹ but in contrast to our cases, they usually had a more diffuse appearance and were not confined to the length of a vertebral body but rather spanned multiple intervertebral spaces.³² Similarly, a defect in the periosteum can result in leakage of blood to the epidural space. Epidural hemorrhage also extends along the nerve roots into the intervertebral foramina, and often occurs in a dorsolateral location.³⁰ Some features are common to both epidural and subperiosteal hemorrhage, such as direct continuity with adjacent osseous structures and displacement of the epidural fat toward the spinal cord (capping).³⁰

The MRI appearance of spinal hematomas varies according to the stage of hemoglobin degradation and whether hemorrhage is clotted or non-clotted.¹⁷ Consistent with previous reports of thoracolumbar spinal epidural hematomas, most dogs in our study (8/9) exhibited homogeneous signal intensity across all MRI sequences, with only one case showing mild T2W heterogeneity. The observed pattern (T2W hyperintensity, T1W iso- to hyperintensity relative to gray matter, and lack of central but presence of peripheral susceptibility rim on T2^*W images) is characteristic of hyperacute (<24 h) or late subacute (7-14 days) hemorrhagic stages.³¹ The clinical onset of the cases aligns with the hyperacute phase of hemorrhage. During this phase, diamagnetic oxyhemoglobin predominates, producing minimal susceptibility effects and appearing as intermediate-to-high signal intensity on T2^*W sequences. Peripheral conversion to deoxyhemoglobin within hours introduces paramagnetic effects, resulting in the characteristic low-signal peripheral rim.³³

In most (8/9) cases of subperiosteal vertebral hemorrhage described here and in the case reported previously,¹ a peripheral inner hypointense rim also is observed in spin echo sequences (T2W and T1W) between the lesion and the epidural fat. This lining could be consistent with a previous observation,³⁴ where the periosteum was visualized as a T1W hypointense line within the epidural space. Alternatively, as noted above, this rim appeared as a peripheral susceptibility artifact on T2^*W images in all cases, suggesting that its appearance also may be attributed to the hemorrhagic nature of the lesion.³⁵

In dogs with moderate to marked neurologic deficits (eg, nonambulatory paretic or plegic), surgical treatment traditionally has been recommended in presence of clinically relevant spinal cord lesions, especially for intervertebral disc disease.³⁶ However, for certain types of compressive extradural lesions such as HNPE, the outcomes of medical and surgical treatment are comparable.³⁷ The overall outcome for cases in our study was excellent with supportive medical treatment and median recovery time to ambulation was 2 days (range,1-9 days). This finding is similar to previously reported times to ambulation in dogs with cervical HNPE treated medically (median, 1-3 days; range, 1-30 days^37–39). Contrary to compressive cervical myelopathy caused by extrusion of a degenerated intervertebral disc, the rapid favorable response to the medical management of the described cases can be explained by a time-limited, mild to moderate spinal cord compression caused by the hematoma.⁴⁰ The peracute to acute onset of moderate to severe neurologic deficits in our cases in the absence of severe spinal cord compressions suggest a substantial role of the spinal cord contusion by the rapidly developing hematoma,^41,42 whereas the swift recovery likely reflects rapid reabsorption of the lesion, similar to what is seen in cases of HNPE^39,42 and as confirmed by the dog that underwent repeat MRI. This possibility likely explains the very favorable outcome, without surgical intervention in the described cases.

Limitations of our study arise from its retrospective and multicenter nature. There were differences in ancillary tests performed (eg, coagulation status, CSF analysis, coagulation assessment), imaging protocols used, and therapeutic approaches applied. Furthermore, no dogs were treated surgically, and therefore the diagnosis of subperiosteal vertebral hematoma remained presumptive, with other less likely differential diagnoses being venous sinus thrombosis or epidural hematoma. The overall rapid neurologic improvement and positive outcome achieved in these cases prevented a surgical or histopathologic diagnosis being achieved. The rapid onset, imaging features, rapid clinical improvement, and imaging resolution of the lesions, all support the likely hemorrhagic nature of the disease. The small sample size did not allow for statistical analyses. Additionally, the cadaveric evaluation was performed on a mixed-breed dog and not a greyhound, but the anatomy of this particular area likely does not differ markedly among dog breeds.

Conclusion

Cervical SSVHs is a rare cause of spinal cord compression, to which greyhounds seem predisposed. An acute onset of cervical myelopathy in the breed or possibly in other sighthounds should raise suspicion of this disease. We describe the imaging findings with lesions affecting C3 and C4, with most lesions being bilaterally symmetrical and ventrolateral, but unilateral latero-lateral lesions may occur. On MRI, lesions were mostly homogeneous, T2W hyperintense, T1W isointense to hyperintense, with no clinically relevant contrast enhancement and with a peripheral ring of susceptibility artifact on T2^*W images. Despite most dogs being nonambulatory tetraparetic or tetraplegic, the outcome was excellent with medical treatment. Given the good outcome associated with medical management, histopathologic confirmation is unlikely to be obtained. Nevertheless, we describe a very consistent population of dogs, clinical signs, imaging findings, and outcome.

Abbreviations

1.5 T

1.5 tesla

ALT

alanine aminotransferase

aPTT

activated partial thromboplastin time

BMBT

buccal mucosa bleeding time

C1

first cervical vertebra (atlas)

C3

third cervical vertebra

C4

fourth cervical vertebra

C5

fifth cervical vertebra

CK

creatine kinase

CSF

cerebrospinal fluid

CT

computed tomography

HNPE

hydrated nucleus pulposus extrusion

L3

third lumbar vertebra

MRI

magnetic resonance imaging

PT

prothrombin time

SSVH

spontaneous subperiosteal vertebral hemorrhage

T1W

T1 weighted

T2W

T2 weighted

T2^*W

T2^* weighted/gradient echo

vWF

von Willebrand factor

Author contributions

Ivona Orgonikova (Data curation, Writing—original draft, Writing—review & editing), Claudia Mallol (Conceptualization, Data curation, Formal analysis, Methodology, Supervision), Rodrigo Gutierrez-Quintana (Data curation, Writing—review & editing), Vicente Aige (Formal analysis, Investigation, Methodology, Writing—review & editing), Pablo Amengual-Batle (Data curation, Writing—review & editing), Emili Alcoverro (Data curation, Writing—review & editing), Kiterie Faller (Conceptualization, Writing—review & editing), Tomás Elvira (Data curation, Writing—review & editing), and Josep Brocal (Conceptualization, Methodology, Resources, Supervision)

Conflicts of interest

Ivona Orgonikova is an employee of North Downs Specialist Referrals, which is part of Linnaeus Veterinary Limited and Mars Veterinary Health, a provider of veterinary services.

Funding

Linnaeus Veterinary Limited, which is part of Mars Veterinary Health, supported the costs of the Open Access Publication Charges.

Off-label antimicrobial declaration

The authors declare no off-label use of antimicrobials.

Institutional animal care and use committee or other approval declaration

Research Ethics Committee of School of Biodiversity, One Health and Veterinary Medicine, College of Medical, Veterinary and Life Sciences, University of Glasgow (EA41/23).

Human ethics approval declaration

The authors declare human ethics approval was not needed.