Spinal subdural haemorrhage secondary to strenuous exercise and warfarin, complicated by acute ischaemic stroke

Abstract

Spinal subdural haemorrhage or haematoma (SSDH) is a rare condition that is often overlooked and missed on initial presentation due to its non-specific features that may mimic other more common pathologies. It is associated with high morbidity and mortality rates, with few evidence-based management principles, particularly during the subacute stages of recovery. In this report, we detail a case of SSDH associated with exercise and anticoagulation therapy, which was complicated by acute ischaemic stroke. SSDH should be suspected in cases of acute back pain without a clear alternative cause, particularly in coagulopathic individuals. Following treatment, early recommencement of anticoagulation therapy may be justified in certain cases where indicated, after careful consideration of the affected individual’s risk profile.

Background

Spinal subdural haemorrhage or haematoma (SSDH) is a rare, poorly understood and highly comorbid condition that has only been described in case reports and case series since it was first reported in the literature in 1948.¹ Typical causes of SSDH include trauma, iatrogenic procedures and vasculopathies, while rarer causes include intracranial extension, infection and neoplasms (see table 1). However, approximately 40% of reported spontaneous cases have failed to identify a specific cause.² Between 1948 and 2013, a total of 150 cases of acute, non-traumatic, spontaneous SSDH were reported in the literature.³ Of these, the overall mortality rate exceeded 10% and nearly half of cases were associated with poor neurological recovery. Anticoagulation therapy and coagulopathies due to conditions such as leukaemia, haemophilia and thrombocytopenia represent significant risk factors and are associated with over 50% of SSDH cases.

Table 1.

Reported aetiologies of spinal subdural haematoma

Cause	Examples
Trauma	Fall,33 motor vehicle collision,34 gunshot wound35
Iatrogenic	Spinal surgery,36–38 diagnostic lumbar puncture,39 spinal anaesthesia40
Vasculopathies	Arteriovenous malformation,41 arteriovenous fistula,42 eosinophilic granulomatosis with polyangiitis,43 granulomatosis with polyangiitis,44 primary angiitis of the central nervous system45
Neoplasms	Meningioma46
Infection	Dengue fever47
Intracranial	Cranial surgery,48 arachnoid cyst rupture,49 aneurysm rupture50 51
Idiopathic	Bending over,11 exercise

While there now appear to be established consensus opinions for the diagnosis and initial treatment of SSDH, very little guidance is available to optimise the subacute stages of management, particularly in those with pre-existing comorbidities that necessitate ongoing anticoagulation therapy. In this report, we demonstrate a case of a large SSDH that developed during exercise in a therapeutically anticoagulated individual, whose recovery was complicated by acute ischaemic strokes.

Case presentation

A man in his early 50s presented with sudden-onset, sharp, central, upper thoracic back pain radiating to his central chest and posterior neck while weightlifting. This was associated with headache, nausea and vomiting. His medical background included hypertension and a metallic aortic valve replacement nearly 20 years prior for aortic stenosis, which has been managed with daily warfarin therapy. He was otherwise physically fit and independent, with no recent surgery, trauma or medication regimen changes.

While in hospital, he developed photophobia with his headache and his pain began to radiate down his right lower limb. Neurological deficits did not occur until approximately 17 hours after the onset of his back pain; suprapubic pain due to urinary retention of over 1.5 L was noted, followed by right lower limb numbness, paraesthesia and weakness, which quickly progressed to paralysis. Physical examination revealed reduced pinprick sensation below the right T4 and left T12 dermatomes, bilateral lower limb paraesthesia including in the L5-S1 distribution, a positive right-sided Babinski’s sign and loss of anal sphincter tone. His motor function deteriorated over the next few hours and his left lower limb also weakened to 1/5 on the muscle strength scale. However, his bilateral lower limb reflexes were otherwise intact, and he had no clonus, change in muscle tone or saddle anaesthesia. He demonstrated no cranial nerve or upper limb neurological deficits.

Differential diagnosis

Initial investigation results established no clear cause for his original presenting symptoms. Blood tests showed normal serial troponin levels, a therapeutic international normalised ratio (INR) of 3.4 and a platelet count within normal limits. Acute myocardial infarction was excluded based on his troponin and ECG results. A CT aortogram revealed no evidence of an acute aortic pathology, mediastinal haematoma or pericardial effusion. A CT brain scan with angiogram demonstrated no acute intracranial pathology, such as parenchymal haemorrhage or large-vessel occlusion. However, a subsequent MRI scan of his spine following the onset of his neurological deficits revealed a large, hyperacute subdural haematoma spanning a craniocaudal distance of 115 mm from the T2 to T6/T7 vertebral levels, causing compression and displacement of the spinal cord to the left (see figures 1 and 2). Further evidence of subdural haemorrhage was seen extending from the C6/C7 vertebral level to the sacrum.

Figure 1.

Preoperative T2-weighted MRI spine sagittal views. (A) Lumped, loculated, heterogenous material at the T2–T6/T7 vertebral levels in keeping with hyperacute subdural haematoma. (B, C) Hypointense material seen extending to the sacrum with dependent layering below the T8 vertebral level.

Figure 2.

Preoperative T2-weighted MRI spine axial views. (A) Haematoma causing mass effect with compression and displacement of the spinal cord to the left at the T2 vertebral level with sparing of the dorsal epidural fat. (B) Classic appearance of the inverted Mercedes-Benz sign at the T4 vertebral level. (C) Inverted Mercedes-Benz sign highlighted in red.

Treatment

A T1–T7 laminectomy, durotomy and evacuation of his intradural haematoma was performed just over 8 hours after the onset of his neurological symptoms following reversal of warfarin with 1500 IU of Prothrombinex-VF (20 IU/kg) to decrease his INR to 1.6. His spinal cord was heavily congested, cyanotic and bulging, but non-pulsatile (see figure 3). On intradural exploration, the haematoma was visualised and evacuated, with evidence of extension into the subarachnoid space. Dilated vessels were noted, but otherwise no obvious vascular lesion was identified. A postoperative spinal digital subtraction angiogram (DSA) also did not demonstrate any vascular abnormality. On postoperative day (POD) 1, his left lower limb muscle strength had recovered to 4/5, while his right lower limb remained paralysed, with only occasional flickers of contraction.

Figure 3.

Intraoperative image demonstrating a significant haematoma in the subdural space.

He was managed postoperatively with a 48-hour course of oral dexamethasone totalling 32 mg. Mean arterial pressure (MAP) was maintained at >85 mm Hg for the first 72 hours. Anticoagulation therapy with heparin infusion targeting an activated partial thromboplastin time (aPTT) of 50–60 s was commenced on POD 3, though therapeutic levels were not attained until POD 6. His usual warfarin therapy was recommenced on POD 11. Unfortunately, on POD 12, he developed acute left-sided homonymous hemianopia, with CT scans showing an acute cerebral infarct due to a right posterior cerebral artery P2 occlusion (see figure 4). This was managed conservatively by ceasing warfarin to minimise the complications of possible haemorrhagic transformation, commencing high-dose statin therapy and increasing the heparin aPTT target to 60–80 s. Warfarin was eventually recommenced on POD 27 without complications, with a target INR of 2.5–3.0, which is slightly below his usual target of 2.5–3.5.

Figure 4.

Trace diffusion-weighted MRI head axial view 7 days after onset of left homonymous hemianopia showing multiple areas of diffusion restriction, particularly in the right occipital lobe, consistent with acute ischaemic infarction.

Outcome and follow-up

He was transferred to the rehabilitation unit 7 weeks after operation and subsequently discharged home 3 weeks later. He was able to mobilise independently without aid for 30 metres, but his left-sided homonymous hemianopia has persisted. At 6 weeks of follow-up, progress MRI scan showed no residual haematoma, obvious lesional pathology or significant remnant intrinsic cord signal change (see figure 5). His lower limb strength was 4/5 and 4+/5 in right ankle dorsiflexion and right hip flexion, respectively, and he has regained independence with personal activities of daily living.

Figure 5.

Follow-up T2-weighted MRI spine sagittal views showing no residual haematoma or persistent spinal cord signal changes at the (A) cervical or upper thoracic levels, and at the (B) lower thoracic and lumbar levels.

Discussion

Spinal haemorrhage is traditionally classified based on the anatomical location of the haemorrhage, where it may exist within the epidural, subdural or subarachnoid spaces, within the spinal cord itself (intramedullary, also known as haematomyelia), or extend across multiple compartments. Unlike the brain, the meningeal layers of the spine are less pronounced, which makes it difficult to precisely identify the location of spinal haemorrhage on radiological studies. In 2001, Hausmann et al⁴ coined the term ‘spinal intradural-extramedullary haemorrhage (SIEH)’ to categorise both subdural and subarachnoid haemorrhages into a single entity due to the difficulty in distinguishing between these two conditions radiologically, and because of their high prevalence of coexistence at just over 50% of cases,³ which is supported by cerebrospinal fluid (CSF) studies showing half of supposedly pure subdural haematomas demonstrating xanthochromia on pathology.⁵ However, this term has yet to gain widespread usage.

The incidence of SSDH has not been accurately deduced due to its rarity, but it is believed that they are the rarest form of spinal haemorrhage, making up only 4% of total cases.⁶ Spinal epidural haemorrhage is the most common, with an estimated incidence of 0.1 in 100 000 individuals per year.⁷ Case series have established that SSDH may have a slight female predominance of approximately 1.25:1 and a bimodal age distribution, with prevalence of cases highest between the first and second decades of life, likely due to haematological conditions such as leukaemia,³ and between the fourth and sixth decades of life, likely due to age-related comorbidities such as vasculopathies and the increased use of anticoagulation therapy.^{2 3} SSDH most commonly occurs in the thoracic spine, though this is influenced by its cause.^{2 3 8} For example, haemorrhage due to iatrogenic procedures such as spinal anaesthesia tends to occur in the thoracolumbar spine.

The exact pathophysiology behind atraumatic, spontaneous SSDH remains unknown, but various theories have been hypothesised. It is believed that elevated intra-abdominal or intrathoracic pressures result in a subsequent increase in the intravascular pressure of the valveless, radiculomedullary veins that cross the subarachnoid and subdural spaces.^{9 10} If this pressure is not neutralised by a concomitant rise in CSF pressure, then the delicate network of small extra-arachnoid vessels located along the inner surface of the dura may rupture, resulting in subdural haemorrhage and haematoma formation, causing compression on the spinal cord. Other authors argue that unlike in the brain, the subdural space of the spine contains only small, low-pressure vessels that are unlikely to be the source of any significant haemorrhage.^{4 11} It has also been proposed that given the high prevalence of coexisting subdural and subarachnoid haemorrhages, the source of subdural haemorrhage may originate from the more vascular subarachnoid space before extending through the arachnoid membrane and into the subdural space.^{11 12} However, this cannot explain the several cases that have been reported where surgical explorations have revealed large, isolated subdural haematomas with intact arachnoid mater.¹⁰ As such, the reverse mechanism has also been hypothesised, whereby a large dural border haematoma may invade the subarachnoid space through a break in the subarachnoid mater.^{10 11} Overall, in the absence of a contributing vasculopathy, a culprit blood vessel has not been universally identified. We were also unable to identify a causative vascular lesion in our case, with anticoagulation therapy and perhaps intense exercise precipitating sudden changes in intravascular pressure serving as the only risk factors. There is agreement that the presence of a coagulopathy in isolation is insufficient to cause SSDH, and that a second precipitating factor must be present, such as an iatrogenic procedure or physical insult.^{6 8}

Prompt diagnosis of SSDH is crucial for the early initiation of treatment to preserve neurological function and prevent permanent disability. The initial presentation for atraumatic, spontaneous SSDH can be highly variable and non-specific. Cases typically involve acute, sudden-onset back pain. Associated headache or signs of meningism such as neck stiffness or photophobia are rare but may indicate the presence of multicompartmental haemorrhage with extension into the subarachnoid space,¹³ which was present in our case. Left untreated, radicular symptoms eventually develop, including lower limb pain and paraesthesia, which can rapidly progress to myelopathy, causing paralysis and autonomic disturbances such as bladder and bowel dysfunction.^{2–4 10} The duration from onset of initial symptoms to the development of neurological deficits can range from a few hours up to as long as 3 weeks.^{3 11} However, chronic cases of SSDH are less prevalent, accounting for approximately only one-fifth of cases.⁹ In most acute cases, neurological deficits will likely be evident 10–26 hours after the initial onset of back pain.¹⁰

MRI is currently the gold standard for the radiological diagnosis and characterisation of SSDH.^{4 10} The classic appearance is that of an intradural, extramedullary heterogenous mass best seen at the levels of the denticulate ligaments on axial views, which may demonstrate a characteristic tristar-shaped appearance within the thecal sac known as the ‘inverted Mercedes-Benz’ sign. This is the result of the bilateral denticulate ligaments and the midline dorsal septum becoming highlighted by a large, circumferential haemorrhage encased within the intradural space as it begins to displace and compress the exiting neural elements that descend the spinal cord.^13–15 SSDH may be distinguished on imaging from other forms of spinal haemorrhage based on both the shape and extent of the haematoma, and its relationship with surrounding anatomical structures (see table 2). For example, on sagittal imaging, SSDH has a clumped, loculated and concave shape, which differentiates it from the smooth biconvex contour of epidural haematomas.^{10 13 16} Epidural haematomas usually extend over only two to three vertebrae, while SSDH may extend up to 20 vertebral levels, with nearly half of cases extending over five levels.^{3 11} Epidural haematomas tend to efface the epidural fat, displace the dura mater inward and make direct contact with the surrounding osseous structures, all of which are features that are not seen in pure SSDHs.¹¹

Table 2.

Typical CT and MRI features of spinal haemorrhage^{16 52 53}

Haemorrhage characteristics	Type of spinal haemorrhage
	Epidural	Subdural	Subarachnoid	Intramedullary
Shape	Biconvex mass lesion	Crescenteric, clumped, loculated, streaking	Longitudinally diffuse (common) or localised (rare)	Variable
Margins	Sharp, well demarcated, contacts bone, displaces dura inward	Irregular, tapered superior and inferior margins	Layered (common) or mass-like (rare)	Often bound within central grey matter
Extent	Usually 2–3 vertebrae, may extend into neural foramina	Often >5 vertebrae	May layer within distal lumbosacral thecal sac	Usually focal, localised at areas of traumatic impact
Other	Effaced epidural fat	Inverted Mercedes-Benz sign	Associated fluid levels, arachnoiditis	Accompanying acutely expanding cord oedema

Other imaging modalities that are commonly employed for SSDH include CT and DSA, both of which were used in our case. Plain CT is a useful initial investigation due to its accessibility and its ability to rapidly identify and compartmentalise acute spinal haemorrhage. For example, CT can quickly help to differentiate the haemorrhage from the adjacent, lower density epidural fat. Although CT is less effective in assessing the spinal cord and identifying neural compromise, it is a valuable modality in complementing and aiding diagnosis in cases where MRI becomes less reliable, such as in multicompartmental spinal haemorrhage and when the blood appears heterogenous.^{16 17} A spinal DSA may also be performed if spinal vascular malformations are suspected as the underlying cause, such as in cases without history of trauma, iatrogenic procedures or haematological disorders.⁴ In our case, retrospective review of the initial CT aortogram performed approximately 5 hours prior to the onset of neurological deficits already showed subtle evidence of abnormality within the thoracic spine (see figure 6). Unfortunately, this finding was missed at the time given his cardiovascular risk factors and lack of neurological symptoms, with the clinical priority focused on ruling out an aortic pathology. There are multiple case reports of either misdiagnosis or delayed diagnosis of spinal haemorrhage,^{18 19} which usually arise due to its subtle presenting symptoms that may mimic other more common and life-threatening conditions—for example, myocardial infarction²⁰—and a general lack of awareness of the condition due to its rarity. Ultimately, a high clinical index of suspicion for spinal haemorrhage should be raised in individuals with acute and persistent back pain even in the absence of neurological symptoms, particularly in those who are anticoagulated or coagulopathic, and after more common causes have been excluded.

Figure 6.

Preoperative CT aortogram. (A) Sagittal view showing subtle cord hyperdensity at the T2–T5 vertebral levels. (B) Axial view showing subtle partial inverted Mercedes-Benz sign at the T4 vertebral level.

Treatment of SSDH consists of either surgical decompression, percutaneous drainage or conservative medical management.^{8 10 11} The decision is made based on the severity of neurological deficits, the anatomical location of the haemorrhage and the individual’s pre-existing comorbidities.⁸ Surgical decompression involving laminectomy, durotomy and evacuation of the haematoma is indicated in those with severe or acutely deteriorating neurological deficits. In acute, non-traumatic SSDH the mean time from diagnosis to surgical intervention was 73.3 hours, with 63% of patients undergoing surgery within 48 hours of symptom onset and 25% undergoing surgery over 48 hours after diagnosis.² Percutaneous drainage may be considered in dorsally located haematomas within the lumbar spine,^{11 21} but is limited by its inability to evacuate larger blood clots and is also associated with risks such as further iatrogenic injury to the blood vessels.¹⁰ Successful cases of conservative medical management resulting in spontaneous recovery have also been reported in those who present with mild or improving neurological symptoms.^{22 23} This typically involves the use of a short course of steroids³ and serial radiological and neurological assessments.¹³ In all cases, the underlying cause of the SSDH should be identified and promptly managed.

There is currently no high-quality evidence for optimal blood pressure parameters in acute spinal haemorrhage. In acute spinal cord injury, current guidelines based on level III evidence recommend a MAP of greater than 85 mm Hg for 5–7 days.^{24 25} This has been shown to correlate with improved neurological recovery by theoretically ensuring adequate spinal cord perfusion pressure and preventing secondary injury. In the context of associated spinal haemorrhage, a 2020 retrospective review concluded that elevated MAPs over 85 mm Hg for 7 days were not associated with a significantly increased risk of haemorrhagic expansion in those with cervical spine haemorrhagic contusions.²⁶

There are also very limited guidelines for recommencement of anticoagulation therapy in spinal haemorrhage. Postoperatively, three decisions need to be made depending on the risk of ongoing postoperative bleeding and whether the individual was previously on anticoagulation therapy for a pre-existing pathology: (1) the anticoagulation agent to be used, (2) the timing of commencement and (3) the dosage or therapeutic target. Our decision to use unfractionated heparin is due to its short half life and relative ease of reversibility should postoperative haemorrhage occur. There is minimal high-level evidence in the current literature to guide the timing of commencement and the therapeutic target, though a 2010 systematic review investigating brain and spinal haemorrhage in those at increased risk of thromboembolism concluded that commencing anticoagulation therapy 72 hours after presentation serves as an ideal compromise to minimise both the risk of rehaemorrhage and thromboembolic events.²⁷ The study showed that spinal haemorrhage has a relatively low risk of rehaemorrhage but does pose a significant risk for thromboembolic complications. Of the 31 cases of spinal haemorrhage included in the review, there were no cases of rehaemorrhage, but a 32% rate of thromboembolic complications. In comparison, the 294 cases of brain haemorrhage had an overall rehaemorrhage rate of 13% and a thromboembolic complication rate of only 5%. Therefore, those with pre-existing pathologies that necessitate anticoagulation should be considered for recommencement as soon as possible. For example, earlier commencement should be considered in individuals with atrial fibrillation and a high CHA₂DS₂-VASc score. Similarly, earlier commencement at a higher therapeutic target should be considered for those with mechanical heart valves, such as in our case.

Most published case reports of SSDH with associated anticoagulation therapy have not commented on the timing of recommencement or any dosage adjustments that were made. In one case, warfarin therapy used for previous ischaemic stroke was recommenced at a lower INR target on POD 7, and then increased to the usual INR target on POD 30 without complications.³ In our case, the impact of our anticoagulation regimen on the development of subsequent ischaemic stroke is difficult to interpret given the acute infarction occurred shortly after recommencement of warfarin therapy with heparin bridging. We hypothesise that our initial aPTT target of 50–60 s may have been too low. Earlier recommencement of an individual’s usual anticoagulation regimen may perhaps be considered in future similar cases given SSDH is associated with a very high risk for thromboembolic complications due to the hypercoagulable state and the physically debilitating neurological motor impairments that it produces. We recommend regular deep vein thrombosis screening and discussion with other specialists such as cardiologists and haematologists to help assess the overall risk for thromboembolic complications on a case-by-case basis.

As is the case with acute spinal cord injury, the current literature does not provide a consensus on the role of steroids in SSDH with spinal cord compromise.^{28 29} Steroids are believed to reduce inflammation and oedema, thereby minimising secondary cord compression and injury. Due to the rarity of SSDH, there are not enough comparative data to support or oppose the use of steroids in this setting. However, a review of the case studies shows that steroids are frequently used, particularly in conservative management of SSDH, with regimens such as oral prednisolone,³⁰ pulse therapy methylprednisolone³ and dexamethasone³¹ reported. The use of high-dose steroids postoperatively has also been reported.³² The possible theoretical benefits of steroid use must be weighed against its risks, particularly in the context of the individual’s comorbidities and prognosis.

Factors that have been found to be associated with worse prognosis include severe neurological deficits on presentation, pre-existing coagulopathies and the presence of a contributing iatrogenic procedure.³ Conflicting conclusions have been drawn regarding whether subarachnoid extension confers a worse prognosis.^{3 8 10} Postoperative MRI may be used to assess for evidence of spinal cord injury and to monitor for haemorrhage reoccurrence to provide a further insight into prognosis.¹⁰ Despite the historically high rates of mortality and morbidity, more recent cases indicate that these figures may be improving. Acute, non-traumatic cases from the last two decades have shown mortality and morbidity rates of 1.3% and 28% respectively.³ When limited to both non-traumatic and non-iatrogenic cases, even more favourable outcomes are seen, with 59% of cases resulting in full recovery or regaining independence.² Cases managed conservatively have higher rates of favourable outcomes compared with those managed surgically, likely because they tend to be of a lower acuity on initial presentation.² Despite early surgical intervention in our case, neurological recovery was slow likely due to the severity of the haemorrhage.

Learning points.

• Spinal subdural haemorrhage or haematoma (SSDH) is a challenging condition to diagnose and manage, especially when it occurs spontaneously and in the presence of other competing comorbidities.

• A high clinical index of suspicion for spinal haemorrhage should be raised in presentations of acute back pain in coagulopathic individuals once other more common diagnoses have been excluded, even in the absence of neurological deficits.

• While MRI remains the gold standard imaging modality for the diagnosis of SSDH, plain CT is more readily accessible and can also be used to rapidly diagnose cases.

• There exist very few evidence-based principles to optimise the subacute stages of care due to the rarity of SSDH, but recent reports have demonstrated improving prognosis.

• Where indicated, early recommencement of anticoagulation therapy may be justified given the high risk for thromboembolic complications, but this should be evaluated on a case-by-case basis.

Ethics statements

Patient consent for publication

Consent obtained directly from patient(s)