An expedited evaluation in patients with acute spinal cord compression is paramount because reversal of tetraparesis or paraparesis is time locked.1 Beyond a certain interval the symptoms may remain complete, with no prospect of future ambulation or bladder control.
Patients with spinal cord compression from malignant disease often have some degree of ambulation at first evaluation. In the Memorial Sloan-Kettering series, 50% of the patients presented ambulatory, 35% paraparetic, and 15% paraplegic at the time of diagnosis.2 In addition, it has been estimated that 30% of patients with epidural spinal cord compression from metastatic cancer become paraplegic within 1 week.3 This observation clearly indicates a dynamic process that possibly can be halted or partly reversed. Unfortunately, unacceptable delay in diagnosis, referral, and investigation occurs in patients with spinal cord compression.4
Acute management of spinal cord compression and the priorities of evaluation are discussed in this chapter, but the usual considerations in patients presenting with ambulation difficulties have already been discussed in Chapter 2. Injuries severe enough to damage the spinal cord are commonly associated with head, abdominal, or chest trauma; and management of traumatic spinal cord injury is discussed in Chapter 19.
Neurologic Assessment of Acute Spinal Cord Compression
Neurologic examination should localize the lesion in patients with acute paraplegia or tetraplegia. Sensory abnormalities localize in the vertical plane (cervical, lumbar, sacral) and, when combined with other long tract signs, point to localization in the horizontal plane (extradural, intradural, or intramedullary).5Clinical clues helpful in localization are found in Appendix 12.1.
The major cord syndromes are summarized in Table 12.1
All sensory modalities should be tested (pinprick, position, and vibration sense; light touch with a wisp of cotton; pressure touch; and temperature tested with a cold or hot piece of metal [e.g., warmed under running hot water]). Abnormal pinprick is usually interpreted as touch without identification of a sharp sting and is most valuable in localizing segments. When a tuning fork is unavailable to test vibration, at least position sense should be tested. Normally, movement of a few degrees in the position of the toe joints should be easily appreciated. In addition, tactile discrimination should be tested, and normally a 2- to 3-cm difference between two points should be appreciated. Normal function suggests intact posterior column tracts but also nerve root function.
Saddle anesthesia (S3–S5) is an indication of a conus medullaris lesion, which can be accurately delineated but may be missed with superficial examination in a supine patient. The sensory loss is often dissociated, with sparing of touch but loss of pinprick. Absence of dissociation suggests involvement of the cauda equina, not just the conus.
Sacral sparing of the sensory symptoms is an important sign because it implies a centrally located intramedullary lesion. (The representation of the sacral fibers is very peripheral in the cord; thus, pinprick and temperature sensation may be spared in acute central cord lesions.)
Table 12.1. Major Acute Spinal Cord Syndromes
Figure 12.1 Abnormal sensory patterns in acute spinal cord disease. (Modified from Byrne TN, Waxman SG: Spinal Cord Compression: Diagnosis and Principles of Management. Philadelphia: FA Davis, 1990, p. 39. By permission of the publisher.)
Dissociating sensory lesions may be further localized in the horizontal plane. Brown-Séquard syndrome is strongly indicative of extramedullary compression, but it may occur in patients with cancer and radiation myelopathy. Its clinical hallmark is loss of pain and temperature sensation opposite the lesion, with loss of position and vibration and more prominent leg weakness at the level of the lesion. The patient often may be puzzled by numbness in one leg and weakness in the other. Brown-Séquard syndrome is rarely uniform in presentation, but marked unilateral leg weakness with Babinski's sign and lack of position recognition of the toe should point to acute extra-medullary compression. The classic patterns of sensory loss in myelopathies are depicted in Figure 12.1.
Muscle strength should be graded with the British Medical Research Council scale (Appendix 12.2) in index muscles. These are proximally iliopsoas, gluteus, quadriceps, and hamstring muscles and distally tibialis anterior and posterior, peronei, gastrocnemius, soleus, and extensor and flexor muscles of toes. Further progression of weakness can be easily assessed by this validated grading system.
Tendon, abdominal, and anal reflexes are usually unelicitable in patients with acute paraplegia from a spinal cord lesion. Abdominal reflexes involve the T7–T12 segment; however, absence of these reflexes is not particularly helpful in localization, and they are absent in most obese patients. The cremaster reflex involves the L1–L2 arc, and the anal reflexes involve the S2–S4 arc; both reflexes have localizing value in determining the segment of involvement in the spinal cord.
Immediate assessment of the bladder is warranted. Sensation of bladder distention may be lost, resulting in overflow incontinence. Detrusor areflexia can be expected with perianal anesthesia, absence of the bulbocavernosus reflex (an unpleasant but important reflex triggered when a squeeze of the glans penis is followed by contraction of the bulbocavernous muscle assessed by palpation), and poor anal tone or loss of voluntary control of the anal sphincter. Distention of the bladder should be prevented by immediate catheterization.
Pain is common in acute spinal cord compression. However, significant destructive and compressive spinal lesions may be virtually painless. Pain that is worse with lying down may signal an epidural spinal tumor and can be explained by additional traction from lengthening of the spine in the supine position.6Excruciating pain closely associated in time with the development of acute paraplegia or tetraplegia should suggest intra-medullary, subarachnoid, or acute epidural hemorrhage, particularly in patients receiving anticoagulation. Equally important to recognize is a spinal epidural abscess, in which acute paraparesis or tetraparesis can evolve in hours. Acute chest pain followed by paraplegia may be due to aortic dissection. Pain in the lower back area may be referred from a dissecting abdominal aneurysm; it may begin in the lower lumber spine and be followed by acute paraplegia from spinal cord infarction. In young patients, acute low back pain preceding acute paraplegia may indicate fibrocartilaginous emboli to the spinal cord from thoracic disk herniation.7 Pain referred to the abdomen is often experienced by patients with acute spinal cord lesions, who may feel they are strapped into a corset.
Pain should be classified as local, referred, radicular, or funicular. Local spinal percussion pain (deep, boring) in the thoracolumbar spine should be evaluated by having the patient turn to the side and carefully tapping on the spinous processes with a reflex hammer. Acute radicular pain (sharp, stabbing) should be further confirmed by straight leg testing and a forceful cough or Valsalva maneuver. Funicular pain (burning, stabbing, electrical) is a less clearly characterized pain sensation of burning, jolting, and jabbing without clear localization, often occurring with sudden movements of the spine. The pain may signal intramedullary disease (e.g., tumor or demyelination).
Two neurosurgical emergencies need special mention not only because recognition may be difficult but also because presentation mimics common disorders seen in the emergency department. First, epidural spinal abscess is caused in 50% of the patients by Staphylococcus aureus infection. Drug use and chronic alcoholism predispose to diskitis and osteomyelitis, which may extend to the epidural space. Recognition is difficult because most confused and delirious patients have signs suggesting sepsis or acute bacterial meningitis. Local back tenderness may not be prominent, but paraparesis and loss of voluntary muscles and sphincters may rapidly become defining features in patients admitted to the emergency department. Blood cultures have a much higher yield in identifying the organism than cerebrospinal fluid (CSF) and can be isolated from blood in at least 30% of cases.8 CSF examination in the emergency department—done to document or exclude bacterial meningitis—may also be potentially dangerous because shifts in CSF pressure that displace the spinal cord may cause sudden worsening of paraparesis. Second, epidural spinal hematoma may present with acute chest pain or pain between the shoulder blades. The pain has been described as a dagger thrust (le coup de poignard) and is rapidly followed by tingling, the development of a sensory demarcation, and often Brown-Séquard syndrome. This type of pain in combination with use of warfarin or tissue plasminogen activator, epidural block, or recent multilevel spine surgery should immediately point to this diagnosis.9,10,11,12,13 Tetraparesis or paraparesis follows. Presentation with arm weakness and neck pain only has been reported.13 Spontaneous spinal subarachnoid hematoma, although rare, may lead to paralysis when located dorsally in the spinal cord. A ventral type of spinal subarachnoid hematoma12 has a much more benign presentation and resolves spontaneously. Diagnoses to consider in paraplegic patients and acute chest or lumbar pain are listed in Table 12.2.
Neuroimaging in Acute Spinal Cord Compression
A plain radiograph of the spine is useful because it quickly identifies bone destruction from metastatic disease and the consequences for stability of the spine. Plain radiographs can appear misleadingly normal in approximately 25% of patients with documented metastatic spinal cord compression; furthermore, plain radiographic abnormalities may not correspond to the location of the tumor, often showing cord compression at a much higher or lower thoracic level.
Table 12.2. Acute Chest or Lumbar Pain with Paraplegia
Bone scan with technetium 99m diphosphonate is occasionally used for screening and as a supplementary test,14 but magnetic resonance imaging (MRI) of the spine, with specific attention to the level determined by clinical localization, should be considered the standard in acute spinal cord compression.15,16 MRI of the spine can classify abnormalities as intramedullary or extra-medullary, in which the lesions are often intradural. Often more than one lesion is involved, supporting a policy of MRI of the whole spine in these patients.15
For reference purposes, a normal MRI of the cervical, thoracic, and lumbar regions of the spine, with T1- and T2-weighted images, is shown in Figure 12.2. An adequate MRI study of the spine should with T1- and T2-weighted images with thin (4–5 mm) sections.
Several important features can be identified on MRI of the spine. On T1-weighted images, bone marrow in the vertebral bodies produces a high intensity but a low signal of the cortical bone. T1-weighted images may underestimate the width of the spinal canal because CSF characteristics are of low signal as well. The nerve roots may emerge on axial slices against the high-intensity signal of epidural fat and low intensity of CSF. Disks also have a low T1 signal. The spinal cord signal is intermediate but higher than that of surrounding CSF.
On T2-weighted images, the CSF is bright (also called “the myelographic effect”). The intravertebral disks are brighter. The nerve roots are much better appreciated on T2 images because of the distinctive bright signal of the CSF.
Motion artifacts may produce hyperintense or hypointense bands (phantom images or harmonics) suggesting a cavity in the cord or neoplasm.
Gadolinium does not penetrate the central nervous system; therefore, if the blood–brain barrier is intact, the spine should not become enhanced. T1-weighted images enhance the basivertebral veins, epidural venous plexus, and spinal ganglion. Necrosis in the spine appears as a high-intensity signal in T1-weighted images after gadolinium injection. Because tumor has a high signal enhancement, gadolinium is useful in further evaluation of intramedullary, intradural, and extramedullary lesions.
Figure 12.2 A–D: Normal T2 and T2 characteristics of sagittal and axial magnetic resonance images of the cervical, thoracic, and lumbar regions of the spine.
Nerve root enhancement with gadolinium ordinarily does not appear unless disease is present but occasionally is observed when a dose of very high contrast is used (0.3 mmol/kg of body weight). Enhancement of the spinal nerve roots is an important finding, and several patterns have been described.17 Diffuse enhancement of the cauda can occur in leptomeningeal metastasis, most often associated with systemic malignant disease, such as breast, lung, or skin cancer. However, diffuse enhancement can be seen in inflammatory polyradiculopathy, such as cytomegalovirus radiculopathy in acquired immunodeficiency syndrome (AIDS). Tuberculosis should be considered in persons from endemic areas and, more recently, in patients with AIDS. Epidural compression may be caused by granuloma formation, which is apparent as thickening of the nerve roots. Virtually any leptomeningeal infection can cause enhancement, including Mycobacterium tuberculosis infection and cysticercosis.18 Sarcoidosis should be considered when enhancement is linear at the nerve roots.19
In spinal cord compression from cancer, vertebral compression fractures may not coexist with an epidural mass. Malignant lesions on MRI most often have a low-intensity signal on T1-weighted images and a high-intensity signal on T2-weighted images (as noted earlier, normal adult marrow has a high signal intensity on T1 and an intermediate intensity on T2images). Contrast enhancement increases the sensitivity of detecting malignant lesions in further defining epidural mass effect, which may not be evident on unenhanced images.20,21
Figure 12.3 Composite magnetic resonance images of the most common causes of spinal cord compression: epidural metastasis (A), epidural abscess (B), epidural hematoma (C,D), and granulomatous disease and thickening of nerve roots (E).
The diagnosis of epidural hematoma has been greatly facilitated by MRI. A high T2-weighted signal often identifies a hematoma that may be scattered throughout the spinal canal, with various degrees of compression at different levels. However, a hyperacute hematoma (within an hour or so) may be isointense on T1-weighted images.22
MRI is the preferred test in epidural abscess, and sometimes after gadolinium enhancement, compartmentalization becomes evident.
Acute spinal cord syndromes may be caused by infarction or an arteriovenous malformation.23,24 Arteriovenous malformation may be located in the dura and cause significant backlog of venous flow and a dramatic swelling of the cord.24,25,26
Figure 12.4 Common magnetic resonance images in patients with acute spinal cord syndromes but without spinal cord compression. A: Spinal cord infarction. B,C: Spinal cord swelling from dural arteriovenous malformation, with resolution after surgical extirpation.
The characteristic MRI findings in these disorders are shown in Figures 12.3 and 12.4. Further visualization of the different compartments in the spine is demonstrated in Figure 12.527 for additional orientation.
Discovery of a mass compressing the spinal cord will lead to further radiologic studies, including chest radiograph or computed tomographic (CT) mammogram, abdominal echocardiogram or CT, and any other tests, particularly positron emission tomographic (PET) scanning, focused on the disclosure of a primary tumor.
The pathophysiologic mechanism of cord compression is poorly understood, but recent insights may provide an avenue of treatment (Box 12.1).
The approach to acute spinal cord compression is determined by its cause, but immediate surgical management is an established route in patients with an epidural abscess localized at a few levels, epidural hematoma, or extradural metastasis with rapidly evolving neurologic deterioration. Its benefit lies in preservation of at least partial mobility and, equally important, complete bladder function. Outcome also depends on the ability to prevent complications and treat nonneurologic problems (lungs, skin, bladder) early.
Figure 12.5 Localization of metastatic lesions in compartments inside the spinal canal: intramedullary process (A); leptomeningeal process (B); process in vertebral body extending into the epidural space (C); para-vertebral process (D); and epidural process (E). (Modified from Byrne TN: Spinal cord compression from epidural metastases. N Engl J Med 1992;327:614. By permission of the Massachusetts Medical Society.)
Box 12.1. Pathophysiology of Metastatic Cord Compression
Spinal cord compression from metastatic disease may be caused by vascular congestion28 due to venous occlusion of the paravertebral venous plexus in the epidural space. Vasogenic edema is an early feature, caused by a breakdown of the blood-spinal cord barrier. The following sequence of events after compression has been documented.29 After 3 hours of cord compression, selective demyelination occurs without axonal damage. It evolves over 24 hours and is associated with production of prostaglandin. In experimental settings, sustained spinal cord compression for 3 hours resulted in persistently absent somatosensory evoked potentials and much less chance for recovery.30 Experimental blocking of serotonin receptors not only inhibits prostaglandin production but also delays the onset of paraplegia. Prolonged compression results in irreversible cord ischemia. If the epidural mass suddenly enlarges from hemorrhage or an extensively infiltrated vertebral body suddenly collapses, acute spinal cord compression may progress very rapidly. Further spinal cord damage may occur if the tumor mass encases radicular arteries.
Box 12.2. Radiotherapy in Spinal Canal Tumors
The radiation field is determined by the extent of involvement and includes two vertebral levels above and Wow the lesion. With this extended field, early local recurrence is less likely, A common radiation dose is 30 Gy in 10–20 fractions administered in 2–4 weeks. The effect is greatest in radiosensitive tumors, such as lymphoma, seminoma, myeloma, Ewing's sarcoma, and neuroblastoma, and less in breast and prostate cancers.
If paraplegia existed for approximately 1 week, recovery of ambulation can be expected 3–6 months later. Recovery is more rapid in patients with gradual onset of paraplegia over weeks. Reirradiation in relapsing patients (“infield” recurrence) frequently preserves ambulation. However, reirradiation in nonambulatory patients may result in the ability to ambulate in only a few.
Box 12.3. Corticosteroids
Dexamethasone used in patients with metastatic epidural spinal cord compression decreases water content of the spinal cord, reduces epidural swelling, and reduces tumor mass in lymphoma (it may even “disappear”). However, its most dramatic clinical effect is on pain reduction, often within hours of intravenous injection. Dexamethasone has a 4 hour half-life, and repeated doses are needed. High doses (intravenous bolus of 100 mg followed by 96 mg orally for 3 days) or high doses with gradual reduction (96 mg intravenously tapered in 14 days) may not be more effective than 10 mg intravenously followed by 16 mg orally for 7 clays and tapered in 2 weeks. There is good evidence to support the use of high-dose dexamethasone therapy in conjunction with radiotherapy.38 Pain relief is more complete with higher doses. Corticosteroids significantly decrease gastric: pH and may rapidly lead to pseudo-obstructive ileus from constipation. These side effects may be reduced by stool softeners, antacids, and H2 blockers. Serious early side effects are psychosis, hyperglycemia, gastric ulcer bleeding, gastrointestinal perforation, and masking of clinical signs of infections.
Surgery should be the preferred approach when the primary tumor is unknown and histologic diagnosis is needed.31,32,33 If vertebral collapse coincides with spinal cord compression, the chances for ambulation are lower and the potential for further deterioration after surgery is real. Spinal stabilization techniques may overcome this situation. These techniques include vertebral body resection, rod stabilization, and anterior (abdominal or thoracic) decompression. Epidural metastatic lesions often can be treated effectively only by surgery with anterior–posterior resection with instrumentation.34 Marginal life expectancy and the degree of metastasis often preclude major surgery.
Approximately 70% of patients with malignant spinal extradural compression remain or become mobile after surgical treatment.35 Radiotherapy is preferred in patients with known radiosensitive tumors (Box 12.2).5 A short fractionated course, often with corticosteroids, is appropriate in most patients.36 Single-fraction therapy should be considered when the aim of treatment is palliation of pain only. Reirradiation in patients with recurrent spinal cord compression from cancer also preserves ambulation. In one study, ambulation was achieved in two-thirds of patients but median survival was 4 months.37 Primary chemotherapy has been advocated for lymphoma, myeloma, and germ cell tumors but in most patients is combined with radiotherapy.
Dexamethasone (Box 12.3) is given to all patients with metastatic cord compression (100 mg intravenous push followed by 16 mg orally daily in divided doses) until definitive management has been determined.
Patients with an epidural hematoma need fresh-frozen plasma and vitamin K for immediate reversal of anticoagulation to international normalized ratio (INR) levels within the normal range. Multiple doses and plasma infusions may be needed to reach an INR level of <1.5, which is satisfactory for surgical exploration. Patients with a high risk for cardioembolization (e.g., metallic heart valve) may tolerate short-term discontinuation of anticoagulation, but experience is limited. Reinstitution of anticoagulation is usually considered 1 week after surgery.39 Spontaneous complete resolution of spinal epidural hematoma has been reported in approximately 7 of more than 250 cases reported in the literature, but identifying these patients and accurately predicting a benign course remain open to doubt.40,41,42
Three major factors predict favorable postoperative recovery in spontaneous epidural hematoma: incomplete cord syndrome, decompression within 36 hours in patients with complete cord syndrome, and decompression within 48 hours in patients with incomplete cord syndrome.43 Rapid onset of paraplegia is not predictive of outcome and should not discourage surgical intervention.
1. Helweg-Larsen S: Clinical outcome in metastatic spinal cord compression. A prospective study of 153 patients. Acta Neurol Scand 94:269, 1996.
2. Posner JB: Neurologic Complications of Cancer. Philadelphia: FA Davis, 1995.
3. Stark RJ, Henson RA, Evans SJ: Spinal metastases. A retrospective survey from a general hospital. Brain 105:189, 1982.
4. Husband DJ: Malignant spinal cord compression: prospective study of delays in referral and treatment. BMJ 317:18, 1998.
5. Byrne TN, Waxman SG: Spinal Cord Compression: Diagnosis and Principles of Management. Philadelphia: FA Davis, 1990.
6. Nicholas JJ, Christy WC: Spinal pain made worse by recumbency: a clue to spinal cord tumors. Arch Phys Med Rehabil 67:598, 1986.
7. Pal B, Johnson A: Paraplegia due to thoracic disc herniation. Postgrad Med J 73:423, 1997.
8. Tang HJ, Lin HJ, Liu YC, et al.: Spinal epidural abscess: experience with 46 patients and evaluation of prognostic factors. J Infect 45:76, 2002.
9. Sawin PD, Traynelis VC, Follett KA: Spinal epidural hematoma following coronary thrombolysis with tissue plasminogen activator. Report of two cases. J Neurosurg 83:350, 1995.
10. Stoll A, Sanchez M: Epidural hematoma after epidural block: implications for its use in pain management. Surg Neurol 57:235, 2002.
11. Kou J, Fischgrund J, Biddinger A, et al.: Risk factors for spinal epidural hematoma after spinal surgery. Spine 27: 1670, 2002.
12. Komiyama M, Yasui T, Sumimoto T, et al.: Spontaneous spinal subarachnoid hematoma of unknown pathogenesis: case reports. Neurosurgery 41:691, 1997.
13. Sakamoto N, Yanaka K, Matsumaru Y, et al.: Cervical epidural hematoma causing hemiparesis. Arch Neurol 60:783, 2003.
14. Tryciecky EW, Gottschalk A, Ludema K: Oncologic imaging: interactions of nuclear medicine with CT and MRI using the bone scan as a model. Semin Nucl Med 27:142, 1997.
15. Cook AM, Lau TN, Tomlinson MJ, et al.: Magnetic resonance imaging of the whole spine in suspected malignant spinal cord compression: impact on management. Clin Oncol (R Coll Radiol) 10:39, 1998.
16. Husband DJ, Grank KA, Romaniuk CS: MRI in the diagnosis and treatment of suspected malignant spinal cord compression. Br J Radiol 74:15, 2001.
17. Georgy BA, Snow RD, Hesselink JR: MR imaging of spinal nerve roots: techniques, enhancement pat terns, and imaging findings. AJR Am J Roentgenol 166:173, 1996.
18. Zee CS, Segall HD, Boswell W, et al.: MR imaging of neurocysticercosis. J Comput Assist Tomogr 12:927, 1988.
19. Nesbit GM, Miller GM, Baker HL Jr, et al.: Spinal cord sarcoidosis: a new finding at MR imaging with Gd-DTPA enhancement. Radiology 173:839. 1989.
20. Lim V, Sobel DF, Zyroff J: Spinal cord pial metastases: MR imaging with gadopentetate dimeglumine. AJNR Am J Neuroradiol 11:975, 1990.
21. Sellwood RB: The radiological approach to metastatic cancer of the brain and spine. Br J Radiol 45:647, 1972.
22. Henderson RD, Pittock SJ, Piepgras DG, et al.: Acute spontaneous spinal epidural hematoma. Arch Neurol 58: 1145, 2001.
23. Anderson NE, Willoughby EW: Infarction of the conus medullaris. Ann Neurol 21:470, 1987.
24. Criscuolo GR, Oldfield EH, Doppman JL: Reversible acute and subacute myelopathy in patients with dural arteriovenous fistulas. Foix-Alajouanine syndrome reconsidered. J Neurosurg 70:354, 1989.
25. Oldfield EH, Doppman JL: Spinal arteriovenous malformation. J Neurosurg 70:354, 1989.
26. Symon L, Kuyama H, Kendall B: Dural arteriovenous malformations of the spine. Clinical features and surgical results in 55 cases. J Neurosurg 60:238, 1984.
27. Byrne TN: Spinal cord compression from epidural metastases. N Engl J Med 327:614, 1992.
28. Siegal T: Spinal cord compression: from laboratory to clinic. Eur J Cancer 31A:1748, 1995.
29. Carlson GD, Minato Y, Okada A, et al.: Early time-dependent decompression for spinal cord injury: vascular mechanisms of recovery. J Neurotrauma 14:951, 1997.
30. Carlson GD, Gorden CD, Oliff HS, et al.: Sustained spinal cord compression. Part I. Time-dependent effect on long-term pathophysiology. J Bone Joint Surg Am 85:86, 2003.
31. Wiley RG: Neurological Complications of Cancer. New York: Marcel Dekker, 1995.
32. Schiff D, Batchelor T, Wen PY: Neurologic emergencies in cancer patients. Neurol Clin 16:449, 1998.
33. Gokaslan ZL: Spine surgery for cancer. Curr Opin Oncol 6:449, 1998.
34. Sundaresan N, Sachdev VP, Holland JF, et al.: Surgical treatment of spinal cord compression from epidural metastasis. J Clin Oncol 13:2330, 1995.
35. Huddart RA, Rajan B, Law M, et al.: Spinal cord com pression in prostate cancer: treatment outcome and prognostic factors. Rodiother Oncol 44:229, 1997.
36. Maranzano E, Latini P: Effectiveness of radiation therapy without surgery in metastatic spinal cord compression: final results from a prospective trial. Int J Radiat Oncol Biol Phys 32:959, 1995.
37. Schiff D, Shaw EG, Cascino TL: Outcome after spinal reirradiation for malignant epidural spinal cord compression. Ann Neurol 37:583, 1995.
38. Loblaw DA, Laperriere NJ: Emergency treatment of malignant extradural spinal cord compression: an evidence-based guideline. J Clin Oncol 16:1613, 1998.
39. Phoung LK, Wijdicks EFM, Sanan A: Spinal epidural hematoma and high thromboembolic risk: between Scylla and Charybdis. Mayo Clin Proc 74:147, 1999.
40. Silber SH: Complete nonsurgical resolution of a spontaneous spinal epidural hematoma. Am J Emerg Med 14:391, 1996.
41. Cirek B, Guven MB, Akalan N: Spontaneous spinal epidural hematoma. Arch Phys Med Rehabil 80:125, 1999.
42. Kumar R, Gerber C: Resolution of extensive spinal epidural haematoma with conservative treatment. J Neurol Neurosurg Psychiatry 65: 949, 1998.
43. Groen RJ, van Alphen HA: Operative treatment of spontaneous spinal epidural hematomas: a study of the factors determining postoperative outcome. Neurosurgery 39: 494, 1996.