Clinical manifestations

Presentation and course

Back pain is the initial complaint in up to 97% of patients with epidural spinal cord compression (68; 118); hyperalgesia is a complaint in up to 65.7% (118). It is unusual for patients with metastatic epidural spinal cord compression to present without pain, but patients with cord compression from lung or renal metastases and lymphoma do so more frequently (06; 41; 115). Pain may precede neurologic symptoms by days to as long as 3 or more years (06). The median duration of pain before the development of neurologic signs varies from 7 to 23 weeks (38; 115). Pain is often described as localized, radicular, or both. Local pain is considered secondary to stretching of the pain-sensitive cortical bone and periosteum but can also be seen acutely in pathologic fracture (109). Local pain is usually constant, relentlessly progressive, and exacerbated by coughing, sneezing, bending, or exercise (38; 68). Pain made worse by or only present with movement can be a warning that there is underlying instability in the thoracic spine or the thoracolumbar junction because of pathologic movement of unstable kyphosis (87; 65). Although the worsening of pain on recumbency is common with metastatic epidural spinal cord compression, the absence of this characteristic symptom should not falsely reassure the practitioner to exclude the possibility of spinal cord compression (38). In fact, in one prospective study of 319 patients with spinal cord compression, only 19% reported exacerbation of pain when lying flat (68). Radicular pain often develops after localized pain and is described as shooting or burning. It can be present in the thoracic, cervical, or lumbosacral areas. The pain is frequently bilateral, more commonly in the lumbosacral regions, and can indicate localization (49; 114). However, pain can also be misleading as to localization; in one study, just over 50% of patients experienced pain at a level different than the location of their spinal cord compression (68). Referred pain, distant from the lesion and nondermatomal, occasionally occurs, such as an L1 vertebral mass causing sacroiliac joint pain (114). These symptoms can be misleading and to a workup focused on the wrong anatomic area (89).

Weakness is present in roughly 80% of patients with metastatic epidural spinal cord compression at presentation. At the time of diagnosis, between 33% to 52% of patients are ambulatory, 35% are paraparetic, and 15% are paraplegic (05). Once weakness is present, progression is often rapid, and 30% of patients with weakness become paraplegic within 1 week (109). Thus, rapid diagnosis and treatment is crucial in any patient with suspected metastatic spinal cord compression. The rate of progression of weakness depends on the tumor growth rate (06; 49). Weakness is usually bilateral and symmetric (38). Those with lesions above the conus will have weakness in a pyramidal pattern with lower extremity flexors and upper extremity extensors predominantly affected and often exhibit hyperreflexia and spasticity. Those with lesions at the level of the cauda equina or with nerve root impingement will have weakness associated with hyporeflexia and decreased tone.

Bladder and bowel symptoms are a later finding but are frequently present at the time of diagnosis. About 20% of patients have mild bladder dysfunction (urgency and frequency), and as many as 48% have more severe dysfunction like retention or incontinence (05). Symptoms of hesitancy and retention tend to be more common than urgency and incontinence. Autonomic disturbance is a bad prognostic sign as it implies bilateral cord or root damage and is usually associated with moderate to severe weakness. Objective sensory disturbance is found in 68% of patients at the time of diagnosis (68). The severity of sensory loss almost always mirrors the severity of motor weakness (06; 14). The sensory loss usually begins distally in the lower extremities and ascends, with patients complaining of "pins and needles" or that "my feet are asleep" (89). A sensory level is identified in up to 52% of patients but, similar to radicular pain, can be poorly correlative to the level of the compressive lesion. In one study, the sensory level varied by up to 10 dermatomes above or below the lesion (68). Lhermitte sign, the sensation of an electrical shock spreading down the spinal column and into the arms and legs with flexion of the neck, is also experienced in up to 15% of patients (87). In some patients, ataxia has also been reported as the primary manifestation of epidural spinal cord compression without other deficits (44), potentially leading to diagnostic failure and causing treatment delay.

Prognosis and complications

Determining the expected prognosis at the time of diagnosis is important, as prognosis can guide treatment decisions in the individual patient. Prognosis is influenced by neurologic function at the time of treatment, speed of neurologic progression, time between development of neurologic deficits and treatment, time from diagnosis of primary malignancy to development of epidural spinal cord metastases, and aggressiveness of the tumor histology.

In general, overall survival in patients with metastatic epidural spinal cord compression is about 6 to 8.9 months (109; 118; 07). In a prospective study of 153 patients, pretreatment ambulatory patients had a median survival of 7 months, and pretreatment nonambulatory patients had a median survival of 1.5 months (49). The general median survival range for pretreatment ambulatory patients is 7 to 12 months, and for pretreatment nonambulatory patients it is 1 to 4 months.

The degree of neurologic dysfunction at presentation is the most significant prognostic variable for recovery of neurologic function (76; 49). The most important prognostic factor for regaining ambulation after treatment is the prior neurologic status. Sixty percent to 99% (with an average of 81%) of patients who were ambulatory at presentation remain so after treatment (36). In contrast, only 6% to 67% (average 32%) of nonambulant patients regain the ability to walk after treatment (36). Between 30% and 45% of patients who are nonambulatory with antigravity proximal leg function will regain ambulation, whereas only 5% of patients who have no antigravity proximal function will walk again (48; 128). A more recent study reported slightly better outcomes, revealing that 55% (n=17) of patients who were nonambulatory before surgery became ambulatory after decompression (119).

Overall, a worse American Spinal Injury Association (ASIA) score at presentation and postoperatively are associated with worse health-related quality of life and decreased overall survival (09). Generally, ambulatory patients have less pain than nonambulatory patients (128). The likelihood of postoperative ambulation also increases if treatment begins less than 12 hours after loss of ambulation and if patients have sparing of bowel and bladder function and sacral sensation (17). However, rapid onset of symptoms and quick progression are bad prognostic variables (13). Ambulatory patients with slow symptom development have improved survival rates and a likelihood of posttreatment ambulation (58; 01). In one study, patients whose motor deficits developed over more than 14 days were more likely to regain or maintain ambulation at the end of treatment (58% ambulatory before radiation, 86% ambulatory after radiation); in contrast, those who developed symptoms in fewer than 7 days were less likely to be walking at the end of treatment (100% ambulatory before treatment, 35% after radiation) (93). One theory behind this finding is that rapid development of neurologic dysfunction may represent compression and injury due to arterial infarction, whereas slow development of neurologic deficits may represent venous congestion. In addition, pathologic compression fractures were associated independently with a decreased ambulatory status. In children with metastatic epidural spinal cord compression, prognosis for recovery from complete motor and sensory loss was significantly better than in adults, with 50% of children becoming ambulatory after surgical decompression and medical therapy (62).

Another important factor in survival is tumor type. Breast and prostate carcinoma, myeloma, and lymphoma are associated with longer survival times and higher rates of posttreatment ambulation than other cancer types (78; 58; 98). In a retrospective review of 148 patients, those with breast cancer and hematological malignancy showed the highest median overall survival of 42.8 months, whereas lung cancer patients showed a worse prognosis with median overall survival of 4.2 months (118). Amelot and colleagues also reported similar median overall survival of 5.9 months in patients with lung spinal metastasis; however, the group reported better survival in patients carrying EGFR-positive status (03). Interestingly, immunotherapy was not an independent prognostic factor of survival. Myeloma, lymphoma, and breast carcinoma have almost an 80% initial response rate to epidural spinal cord compression treatment, and those with myeloma, lymphoma, breast, or prostate cancer have a median survival rate of 10 months (38; 41; 77). Only 25% of patients with lung or renal carcinoma and melanoma respond to any treatment modality and have a median survival rate of about 3 months (38; 77). One study found that the posttreatment ambulation rate in patients with breast cancer was 69%, and another that only 15% of patients with lung cancer were ambulatory after treatment (05; 78).

In addition to tumor type, the location and size of the lesion naturally influence prognosis. Upper thoracic and cervical thoracic lesions and a circumferential angle of spinal cord compression greater than 180 degrees have been associated with a higher risk of poor postoperative neurologic outcomes (67). Axial T2-weighted images at the site of the most severe compression have been used to classify the severity of the epidural spinal cord compression (ESCC score), which includes four grades from 0 to 3 (11):

Bilsky scale presents a radiographic classification for the severity of cord compression and treatment guidance; however, MRI findings of spinal cord compression are not always correlated with the severity of patients’ symptoms in clinical practice and may not stand as an independent factor for prognosis and treatment guidance (121).

Multiple scoring systems have been proposed for predicting pretreatment prognosis, including the SORG Classic, SORG nomogram, Tomita score, Tokuhashi score, Bauer score, Rades score, Linden score, and Katagiri score. They were all developed using a combination of the known predictive factors that contribute to outcome. Some predict outcome after a certain definitive treatment (ie, radiation therapy). Unifying characteristics of these scoring systems include primary site of cancer and visceral involvement and prediction of survival time, usually on the matter of fewer than 6 months, 6 to 12 months, and greater than 12 months. A review evaluating all of these scoring systems found that the SORG nomogram was most accurate at predicting 30- and 90-day mortality, whereas the original Tokuhashi score was best at predicting 365-day survival (02; 57). A separate study revealed that a “good prognosis” per the Rades score predicted the best outcomes in a cohort of patients that underwent stereotactic body radiotherapy for postoperative reirradiation of metastatic epidural spinal cord compression (53). Table 1 shows Rades and colleagues’ survival rate based on prognostic factors for patients with metastatic epidural spinal cord compression treated with radiation therapy.

Additionally, the Spinal Instability Neoplastic Score (SINS) shows good reliability for predicting instability and evaluating for surgical candidacy in patients with spinal metastasis (30; 31; 16; 84). The SINS score (Table 2) includes one clinical factor of pain and five radiographic parameters (lesion location, bone lesion quality, spinal alignment, collapse of vertebral body, and involvement of posterolateral spinal elements).

Table 1. Significant Prognostic Factors and Corresponding Scores in Patients With Metastatic Epidural Spinal Cord Compression Treated With Radiation Therapy

Table 2. Spinal Instability Neoplastic Score (SINS) system

Clinical vignette

A 55-year-old male was 2 years out from diagnosis of prostatic carcinoma by radical prostatectomy and had developed midback pain 8 months previously that had gotten worse over the preceding 5 months. The midback pain was worse when the patient was lying supine and was not made worse by activity. Over the course of 2 weeks, the patient had a second type of pain radiating around his upper abdomen that felt like he had "a rope tightened around his chest." Later, he developed difficulty arising from chairs and then had to use a walker for assistance. He denied having arm weakness, sensory complaints, and bowel or bladder difficulty. On neurologic examination, there was tenderness to palpation in the area of the T6 to T8 vertebral bodies. Mental status and cranial nerves were normal. The patient walked with a waddling gait and was unable to arise from a chair without using his hands. On motor exam, upper extremity strength was 5/5 bilaterally. In the lower extremities, the tone was normal, with strength 4/5 at the iliopsoas and 5/5 distally. Reflexes were 2+ and symmetrical in the upper and lower extremities, with Babinski signs present bilaterally. Sensory examination was normal except for minimally decreased vibratory sense in the lower extremities. Straight-leg raising, reverse straight-leg raising, and Patrick sign were negative. Rectal tone was normal. The preliminary diagnosis was epidural spinal cord compression, and the patient was treated emergently with dexamethasone 100 mg intravenously and then started on dexamethasone 24 mg by mouth four times daily. The patient underwent an emergency MRI scan showing two areas of vertebral involvement: one with epidural spinal cord compression at T8 and the second with asymptomatic metastases at L3. Four hours later, the patient underwent anterior vertebral body resection at T8, with the placement of titanium rods from T6 to T10. The patient tolerated the procedure well except for a postoperative urinary tract infection. Steroids were tapered over 2 weeks, and strength returned to normal at postoperative day 7. A few weeks after surgery, the patient received radiation therapy at a dose of 30 Gy, from T6 to L5, over 2 weeks.

Biological basis

Etiology and pathogenesis

Primary locations of involvement of epidural spinal metastasis. The epidural space is a true space between the dura enclosing the spinal cord and the bony canal, which contains fat, connective tissue, and a rich paravertebral venous plexus that drains the vertebrae and intervertebral spaces. Epidural metastases may originate from the vertebral columns, with the vertebral body as the most common site (85% to 89%) but also the paravertebral tissue (10% to 15%) and, rarely, the epidural space itself. The location of epidural metastatic involvement is important for the surgical treatment of epidural spinal cord compression and is associated with the primary site of neoplasm. Direct extension of epidural metastasis from a vertebral body often arises from carcinoma of the breast, lung, prostate, or plasma cell. Epidural lesions extending through the intervertebral foramina often arise from an adjacent prevertebral lymph node in lymphoma, sarcoma, or neuroblastoma (18).

Mechanism of metastatic disease spreading to epidural spinal lesion. Three proposed mechanisms are present for metastatic epidural spinal cord lesions. The most common mechanism (80%) is hematogenous arterial spread to bone marrow. The vertebral body is the most common location for bony metastasis of the vertebral columns due to the presence of red marrow throughout its lifetime and the high vascularity compared to other bony parts of the vertebral column. The lesion often starts in the vertebral body and may grow posteriorly, affecting the posterior elements, including the pedicles and lamina. Metastatic lesions may also arise directly into the epidural space through the neural foramina from the paravertebral tissues (61), such as lymph nodes in lymphoma. Another probable mechanism of metastatic epidural spinal cord lesion is retrograde venous spread from the primary site via the Batson paravertebral plexus. The valveless epidural Batson venous plexus is a network of bidirectional flow between pelvic veins and the thoracic internal vertebral plexus. It is postulated that an increase in abdominal pressure (such as during Valsalva maneuver) can cause retrograde flow from the abdomen and pelvis into the epidural venous plexus, seeding tumor cells into the epidural space. Once in the epidural space, the tumor usually grows along the path of least resistance. Two main compression patterns have been identified, including predominant soft tissue epidural disease and bony collapse with direct neural compression. The epidural lesion can encircle or encroach on the thecal sac; compress the spinal cord or vascular structure, leading to cord compression or infarction of the spinal cord; or occlude the venous plexus of the epidural space, leading to vasogenic edema of the cord. Additionally, destruction of the vertebral body can result in deformity from direct compression of the bony fragment on the spinal cord or instability of the spinal canal from the collapse of involved vertebrae.

Mechanism underlying spinal cord injury from epidural metastasis. Spinal cord injury due to epidural spinal metastasis occurs due to factors including vasogenic edema, hemorrhage, cord infarction, inflammatory factors, and even demyelination. During the early stage of epidural spinal cord compression, axonal swelling and edema of the spinal cord are observed with preserved spinal cord blood flow. As spinal cord compression progresses, increased swelling and additional mechanical compression compromise spinal cord blood flow. As the blood flow decreases to a critical level in the later stage, spinal cord infarction and hemorrhage can be observed, leading to irreversible injury. Administration of steroids or nonsteroidal anti-inflammatory agents decreases the vasogenic edema and prostaglandin E2 and produces objective improvement in weakness, suggesting that vasogenic edema and the release of inflammatory factors play a significant role in neurologic dysfunction (123; 110). Animal models have been used to demonstrate morphologic features of cord damage and subsequent recovery. Following 3 hours of cord compression in cats, selective demyelination without axonal disruption evolves over the subsequent 21 hours and continues for 1 week. Most demyelinated fibers show evidence of remyelination within 1 month (39). If compression is produced slowly over 48 hours and maintained for 7 days, it is still possible to recover from paralysis, suggesting that demyelination is also an important factor in development of neurologic symptoms with epidural metastatic spinal cord compression.

Epidemiology

The overall incidence of metastatic epidural spinal cord compression can only be estimated, as many patients are never diagnosed either due to the development of symptoms after the decision to forgo further testing or therapy has been made or due to the presence of asymptomatic lesions. In a population-based study, the likelihood of developing symptomatic epidural spinal cord compression in the 5 years preceding death was 2.5% and ranged from 7.9% in patients with multiple myeloma to 0.2% in patients with pancreatic cancer (71). A retrospective study of hospitalized patients cited an annual 3.4% incidence of metastatic epidural spinal cord compression (74). Another retrospective study focused on regional treatment centers cited a slightly higher incidence of 4.4% to 6% (05).

In up to 23% of cases, metastatic epidural spinal cord compression is the initial manifestation of malignancy (106; 68). Multiple myeloma accounts for 11.1% of this patient population (65). Among patients with metastatic epidural spinal cord compression, the most common underlying tumors are lung, prostate, multiple myeloma, non-Hodgkin lymphoma, and breast (74). These five types of malignancy, in addition to sarcomas and renal cancers, account for over 70% of cases of metastatic epidural spinal cord compression. In the pediatric population, sarcoma, neuroblastoma, and lymphoma have been reported as the most frequent causes of spinal cord compression (101). The thoracic spine is the most common location of metastases, accounting for approximately 60% of cases, with 30% in the lumbosacral spine and 10% in the cervical spine (21). Multiple sites of metastatic epidural spinal cord compression occur in 17% to 30% of patients (124). This is particularly common in breast and prostate cancer and less common in lung cancer (115). The average age at diagnosis of metastatic epidural spinal cord compression has varied from 53 to 65 years. Sex differences reflect the underlying primary neoplasm (eg, breast and prostate).

Prevention

To improve clinical outcomes, it is important to identify patients before the appearance of neurologic symptoms or signs. The most important prognostic factor for functional outcome is neurologic function before treatment. As only about 50% of patients are ambulatory at diagnosis, earlier detection and diagnosis are essential. Failure to recognize metastatic epidural spinal cord compression and delay in treatment can result in irreversible neurologic dysfunction, including paraplegia, sensory loss, and bowel and bladder dysfunction. Patients with known malignancy should be advised to inform their physician if they develop new back pain, which is the initial complaint in up to 97% of patients who go on to develop metastatic epidural spinal cord compression. High-risk patients are those with known malignancy and recent-onset back pain and patients who are not known to have malignancy but who have new radicular or localized thoracic back pain worse on recumbency or associated with spinal tenderness.

Among oncological patients, metastatic epidural spinal cord compression is the initial manifestation in 20% of cases in those with unknown cancer history (105). Physicians must focus on history and physical examinations, noting symptoms such as fever, weight loss, malaise, and chills in the setting of new back pain.

Unfortunately, diagnosis of metastatic epidural spinal cord compression is often delayed, even in patients with known malignancy and new back pain. In a published retrospective analysis of 301 patients who eventually presented to a specialized cancer center, for all patients (those with and without previously diagnosed malignancy), the median delay in treatment from onset of back pain was a little over 2 months (51). The median delay in treatment from onset of symptoms of spinal cord compression was 14 days; when analyzed, 3 days were due to patient delays in seeking medical care, 3 days were due to general practitioner delays, and 4 days were due to hospital delays (51). Rates of treatment delivery within 24 hours of presentation to a medical professional (considered the target) were only 30% for those who presented to a general practitioner, 21% for those who presented to a district general hospital, and 87% who presented to a specialized treatment center (51). Of note, a retrospective review of 467 patients with metastatic epidural spinal cord compression showed no correlation between the severity of paralysis and the ESCC score (determined via MRI) (121), highlighting the importance of the initial evaluation and physical exam for determining prognosis.

Radiation therapy for bony metastases without metastatic epidural spinal cord compression is believed to prevent the development of epidural metastases. In prostate cancer, irradiation of the lumbar spine coincidental to the irradiation of paraaortic nodes and the normally radiated pelvic area prevented or delayed the development of the lumbar spine metastases, which might significantly reduce cauda equina compression (56). Strontium-89 (Metastron) and the nitrogen-containing bisphosphonate olpadronate have also been found to decrease the incidence of metastatic epidural spinal cord compression in patients with hormone-refractory prostate cancer metastatic to the skeleton but to a lesser extent than radiotherapy (112).

Differential diagnosis

Metastatic epidural spinal cord compression is a challenging diagnosis to make promptly, given its initial nonspecific symptom of back pain. The differential diagnosis for metastatic epidural spinal cord compression is broad and includes both benign and more concerning causes. When a patient presents with new back pain, multiple differential diagnoses should be considered, including trauma, degenerative changes, inflammatory processes, vascular malformation, infection, and oncologic pathologies.

A history of previous radiation therapy, trauma, or infection is important. Patients with a history of prior radiation therapy are susceptible to developing radiation myelopathy characterized by ascending numbness and weakness in a hemicord pattern. Symptoms usually present 9 to 15 months after radiation, and the risk of development increases with greater total radiation dose and fraction size. History of minor trauma increases the suspicion of a musculoskeletal cause of pain, such as muscle strain and disc disease. Vascular malformations can also present similarly to epidural spinal cord compression. Dural arteriovenous fistulas, the most common vascular malformation in the spinal cord, can produce myelopathy as a result of direct increase in venous pressure or following hemorrhage. Patients often present with progressive gait imbalance, ascending lower extremity weakness, numbness, paresthesias, low back pain, and eventually bowel and bladder dysfunction (54). Dural arteriovenous fistulas may be identified by their "snakelike" appearance on myelography or by demonstrating flow voids or characteristic signal changes of hemorrhage on magnetic resonance imaging (73). Nontraumatic epidural hematomas are rare, but patients taking anticoagulants have an increased risk of this complication (103). Chemotherapy increases the risk of infections, and patients with a history of intravenous drug abuse, hematogenous infection, and vertebral osteomyelitis should raise concern for an underlying epidural abscess. The differentiation between epidural abscess and metastasis is often difficult (14). Epidural abscess is more frequently posteriorly situated and will often cover multiple vertebral body segments compared to metastatic epidural cord lesions. If there is vertebral collapse due to an infective cause, the disc space is frequently destroyed, whereas metastatic vertebral disease usually spares the disc space. An epidural abscess may be associated with increased systemic white blood cell count, fever, or cerebrospinal fluid pleocytosis, but because the dura is an effective barrier, the cerebrospinal fluid may be normal. Blood cultures often yield the correct organism, but an image-guided biopsy is sometimes required to differentiate between abscess and malignancy.

Other malignancy-related etiologies should also be considered in patients with new back pain and known malignancy, including vertebral metastases without epidural spinal cord compression, malignant plexopathy, leptomeningeal metastases/carcinomatous meningitis, and intramedullary metastases. Malignant plexopathy usually presents with radicular pain followed by weakness and sensory disturbance (hyperesthesia or hypoesthesia) in the correlating distribution of the plexus. Breast and lung cancers are most commonly associated with brachial plexopathy, whereas gynecological malignancies and lymphomas are associated with lumbosacral plexopathies. Diagnosis can be confirmed with an MRI or CT of the affected area. Leptomeningeal metastases or carcinomatous meningitis usually present with headaches, altered mental status, multiple cranial nerve palsies, cauda equina syndrome, and even seizures. Some patients also have radicular pain, and though MRI may or may not always reveal enhancement, CSF studies are diagnostic (elevated opening pressure, low glucose, elevated protein, pleocytosis). Intramedullary metastases are much less common than epidural metastases and are most common in lung cancer. They usually present with sensory disturbances followed by motor symptoms, as lesions are most commonly located in the posterior cord (109). Some patients present with Brown-Sequard syndrome, which can clinically distinguish intramedullary from epidural cord lesions. Brown-Sequard syndrome is characterized by weakness and impairment to light touch and vibration unilateral to the lesion, with contralateral impairment of pinprick and temperature (109).

Diagnostic workup

The most predictive prognostic factor in outcome is neurologic function at time of initiation of treatment; thus, timely diagnosis is essential. The goal is to quickly establish radiographic evidence of thecal sac compression and begin treatment within 24 hours of presentation. Optimally, diagnosis can be made and treatment started prior to permanent spinal cord damage. MRI is typically considered the diagnostic imaging modality of choice. CT myelogram is an acceptable alternative when MRI is not readily available. Plain x-rays are inexpensive, easily obtained, and were historically the imaging modality of choice before the advent of more sensitive and specific MRI and CT.

Plain x-rays are quite good at evaluating bony metastases but not at evaluating the spinal cord and surrounding soft tissues. Metastatic epidural spinal cord compression most commonly occurs at the site of vertebral involvement on plain x-ray, especially where there is evidence of vertebral collapse. Plain films can help identify abnormalities such as compression fractures with significant height loss or kyphosis caused by large lytic lesions. If there is mechanical neck or back pain, flexion-extension plain films determine any dynamic instability (32). Most common findings on x-rays include pedicle erosion, paravertebral soft shadow, vertebral collapse, and pathologic fracture or dislocation (87). In the past, if there was back pain or a localizing sign and spinal x-ray was abnormal, the probability of epidural disease was 0.9, but if the x-ray was normal, the probability was only 0.1 (100; 88). More recently, x-rays were found to have a 10% to 17% false-negative rate (05). This rate of missed metastatic epidural spinal cord compression is unacceptable; thus, more advanced imaging is required for any patient with possible epidural spinal cord compression. In one study, Kienstra and colleagues prospectively looked at 170 cases with cancer and back pain to identify a subpopulation of patients in whom MRI can be omitted because of a low risk of spinal epidural metastases (59). From a detailed neurologic history and examination, with the hypothesis that missing any case of spinal epidural metastases is unacceptable, no neurologic criteria were able to diagnose all cases. Therefore, the authors concluded that all patients should have an MRI scan unless the probability is extremely low.

MRI is noninvasive, effectively demonstrates metastatic epidural spinal cord compression and gives a clear image of the spinal cord to better diagnose intramedullary disease.

Complete MRI, including sagittal and axial T1- and T2-weighted images and contrast-enhanced images with gadolinium, are necessary for maximum sensitivity (60).

All patients with suspected metastatic epidural spinal cord compression should have their entire spine imaged. Asymptomatic second areas of metastatic epidural spinal cord compression are identified with MRI and are important for treatment planning. In a retrospective study of 337 cases, failure to image either the thoracic or lumbosacral spine when symptomatic disease was elsewhere would have missed 21% of secondary metastatic lesions, whereas failure to image the cervical spine with symptomatic disease elsewhere would have missed only 1% of cases (107). Anatomical definition of vertebral and extraspinal disease on MRI is necessary when planning a surgical procedure. Control of pain before myelography or MRI may help to prevent movement artifact.

Myelography or CT myelography is an invasive procedure involving lumbar puncture and radio-opaque dye injection, typically followed by CT scan. Neither myelography nor CT/myelography visualizes the spinal cord parenchyma, but they do detect intradural and extradural lesions. Bony lesions are also well-defined on CT scan. Clinical deterioration may occur directly following myelography; however, it is difficult to separate this from natural history of disease, and in one small study, no patients who underwent CT myelogram experienced this potential complication (43). If there is a complete block following lumbar injection, a second puncture for a cervical myelogram or MRI is necessary to visualize the upper limit of the block and to exclude second lesions.

Magnetic resonance imaging has largely replaced CT/myelography as the procedure of choice. Although there are little data directly comparing the two, many studies support that MRI is at least equivalent if not superior to CT myelography (43; 69; 52). Intradural extramedullary metastases are equally well diagnosed by gadolinium-enhanced MRI or myelography, with roughly similar sensitivities and specificities (43). Relative advantages to MRI over CT myelography include that it is noninvasive and provides better detail of the surrounding soft tissues, spinal cord parenchyma, leptomeningeal deposits, and the cauda equina. Because it is noninvasive, MRI has fewer risks than CT/myelography in patients with intracranial mass lesions or bleeding tendencies. The main advantage of CT myelography is that spinal fluid can be obtained for testing, which may help to exclude or confirm alternative diagnoses (eg, carcinomatous meningitis). Where there has been previous surgery or scoliosis, MRI sagittal images can be difficult to interpret due to artifact and surrounding scar tissue. Claustrophobic patients may not tolerate MRI, and those with an incompatible ferromagnetic implant cannot be scanned. Due to its invasiveness and inability to visualize the spinal cord parenchyma, CT myelography is the second imaging choice after MRI. However, it is advisable to get the readily available test because the patient may deteriorate while waiting for proper investigation.

Other imaging modalities such as bone scans, plain CT, and PET scans are sometimes obtained, but these are all inferior to MRI and CT myelography and should not be used as a screening tool in the evaluation of potential metastatic epidural spinal cord compression. CT imaging is excellent for bony anatomy. Radionucleotide bone scans use a radioactive substance to identify diseased bones that show up as “hot spots.” It is superior to x-rays in identifying bony metastatic lesions and has the benefit of visualizing the entire spine in one scan. However, lytic tumors or lesions characteristic of multiple myeloma are not well visualized on bone scans. Furthermore, bone scans provide no information about the thecal sac. PET scans use a radiotracer to identify areas of glucose uptake. Thus, PET scans are better at identifying bony lesions as they can visualize both osteolytic and osteoblastic lesions and are better at identifying lesions in the early stages of growth. Similar to bone scans, PET scans do not visualize the thecal sac. Lastly, plain CT scans are good for visualization of the bony spine but do not show the spinal cord or thecal sac.

Simple investigations such as chest radiograph, prostate-specific antigen, mammography, abdominal ultrasound, or abdominal and chest computer-assisted tomography scan may immediately demonstrate the primary malignancy in patients without a known primary malignancy. Additionally, if the cause of epidural spinal cord compression is uncertain, computer-assisted tomography-guided biopsy of a paraspinal or epidural mass or percutaneous needle biopsy of a collapsed vertebral body is often helpful in establishing a diagnosis. Percutaneous needle biopsy of the vertebral lesion under fluoroscopy is a good way to obtain tissue if the vertebral lesion is the initial presentation of malignancy. It was positive in 18 of 19 patients with no immediate complications (106).

Management

General treatment of the patient with epidural metastatic spinal cord compression includes immediate administration of steroids, followed by definitive therapy comprising one or more of the following: surgery, laser interstitial thermal therapy, cryotherapy, external beam radiation, stereotactic body radiotherapy or radiosurgery, and chemotherapy. The treatment plan will depend on the characteristics of the individual patient. Other general management principles include adequate pain control, prevention of deep vein thromboses in the nonambulatory patient, and treatment of urinary retention or incontinence and constipation. Pain control can be achieved with a combination of steroids, neuropathic analgesics for those with radicular pain (such as gabapentin), and opiates. Patients with metastatic spinal cord compression are at an increased risk of deep venous thrombosis and pulmonary embolus, especially in those who are nonambulatory. Prophylactic subcutaneous heparin, antiembolic stockings, or compression pumps will help reduce morbidity and mortality. In the presence of urinary retention or constipation, intermittent or permanent catheterization should be considered, and laxatives or suppositories should be initiated early in the course of admission, especially for those requiring opiates for pain control. Standard care should be taken to prevent pressure sores when nursing patients with paraparesis or paraplegia.

Emergency measures. Patients with cancer, back pain, and an abnormal progressing neurologic examination demonstrating myelopathy or radiculopathy should receive glucocorticoids and undergo emergency MRI or computer-assisted tomography or myelography, whichever is more readily available.

Stable patients with uncertain, mild, or stable neurologic findings can be scanned urgently over the next 24 hours, and if epidural metastatic spinal cord compression is confirmed, should also be started on steroids. Dexamethasone is thought to downregulate the production of vascular endothelial growth factor and prostaglandin E2, which causes a decrease in spinal cord edema and delays the onset of neurologic decline (65). A dexamethasone dose-response effect has been demonstrated in an animal model of metastatic epidural spinal cord compression, producing decreased spinal cord water content, reduced epidural swelling, and transient clinical improvement (24). High-dose glucocorticoids are usually part of the standard therapy, but there is little evidence to support high-dose versus lower-dose steroids, and in fact, higher doses have been associated with significant side effects. High-dose intravenous dexamethasone side effects include vaginal burning, elevation in blood pressure, glucose intolerance, increased risk of infections, gastrointestinal bleeding and perforation, and electrolyte disturbance; caution should be used in those suspected of having an infection or gastrointestinal symptoms.

Three randomized clinical trials have attempted to address the utility of steroids. One study addressed outcomes comparing high-dose steroids with placebo, and the other two studies were focused on outcomes comparing high versus moderate-dose steroids. In a study conducted by Sorensen and associates, 57 patients were randomized to dexamethasone 96 mg once followed by 24 mg four times daily for 3 days, then tapered over 10 days versus placebo with normal saline; all patients also received standard radiation therapy (113). Ambulation posttreatment was intact in 81% of patients who received steroids versus 63% of those who did not. At 6 months after treatment, 59% of patients in the steroid group were still walking, whereas only 33% in the placebo group were. There was no difference in median survival between the two groups. A prospective, randomized, double-blind trial of 37 patients compared a single high dexamethasone dose (100 mg intravenously) with a conventional initial dose (10 mg intravenously), both subsequently followed by 4 mg orally every 6 hours (16 mg total per day). This study did not demonstrate a significant difference in ambulation, bladder function, or pain control (125). Side effects were not commented on in this study. Another study of 20 patients compared dexamethasone 96 mg daily versus 16 mg daily for 2 days followed by a rapid taper over the following 15 days (40). All patients also underwent standard radiation therapy. Though small, this study showed significant adverse events in five of nine patients in the high-dose group and four of 11 patients in the moderate-dose group and no differences in pain control or ambulation. Lastly, in a nonrandomized series of 28 patients who received high-dose steroids (96 mg dexamethasone followed by a 14-day taper), four serious adverse events occurred (gastrointestinal bleeding or perforation) (46). Due to the incidence of side effects, the next 38 patients in the series received 16 mg dexamethasone followed by a 14-day taper, in whom only three side effect events were documented, none of which were serious. Outcomes between the two groups were similar. A Cochrane meta-analysis of the second, third, and fourth trials listed was conducted and revealed that there is no difference in outcome between high-dose and moderate-dose steroids, but there was a higher incidence of serious adverse events in the higher dose groups (34).

In general, patients with rapidly progressive symptoms, paraparesis, or paraplegia should be considered for high-dose dexamethasone followed by a taper. A moderate-dose regimen may also be considered based on the data above that suggest no significant difference in outcome between the two regimens but a higher incidence of side effects with the higher doses. Some patients with small lesions without spinal cord compression may even forgo the use of steroids completely. In one study, 20 patients with metastatic epidural spinal cord compression without myelopathy on clinical examination and no evidence of massive spine invasion on imaging were treated without steroids during 10 fractions of 300 Gy/fraction without clinical deterioration (78). Regardless of the choice of steroid dose, definitive treatment such as chemotherapy, radiation therapy, or surgery (or a combination of one or more) should be planned after diagnosis. Choice of definitive treatment depends on radiosensitivity of the tumor, presence of spinal instability, and degree of spinal cord compression.

One of the most-used decision paradigms for selecting appropriate management for spinal metastatic disease is the NOMS framework, which takes into account neurologic, oncologic, mechanical, and systemic considerations to optimize treatment (32).

The neurologic assessment of the decision framework considers the degree of epidural spinal cord compression and radiculopathy or myelopathy. Based on imaging, the Spine Oncology Study Group designed a 6-point grading scale for the degree of epidural spinal cord compression. Axial T2 imaging at the site of the most severe compression is used to assign a grade of 0, 1a, 1b, 2, or 3. Grade 0 signifies metastatic lesions confined to bone; Grade 1 is for tumors extending into the epidural space without compression of the spinal cord; Grade 2 refers to spinal cord compression where cerebrospinal fluid is still visible; and Grade 3 is for cases with complete cerebrospinal fluid obstruction due to the degree of tumor compression (32).

The oncologic assessment of the NOMS framework evaluates the responsiveness of tumors, which is dependent on histology, to available treatment options. Specifically, this looks at how responsive the tumor is to radiation therapy as this is the least invasive and most effective approach to local tumor control (32).

The mechanical assessment of the NOMS framework examines spinal stability. Although there is no consensus on the definition of what qualifies for spinal instability, the Spine Oncology Study Group (SOSG), an international cohort of 30 expert spine oncologists, developed the Spine Instability Neoplastic Score (SINS) after conducting a systematic literature review to identify the clinical and radiographic characteristics that influence spinal instability (29). This classification system uses six components, including the location of the vertebral metastasis, mechanical pain, blastic or lytic lesions, subluxation or deformity, vertebral collapse, and posterior element compromise, to determine a composite score predictive of spinal instability. Patients with scores exceeding a threshold are considered at risk of spinal instability and should be considered for surgery. Prospective clinical validation of the Spine Instability Neoplastic Score is still pending, but one independent interobserver and intraobserver evaluation of 30 patients using the score reported interobserver reliability of 0.797 (15). A more recent retrospective review of 285 patients with metastatic spinal disease revealed that Grade 3 epidural spinal cord compression was significantly associated with an unstable Spine Instability Neoplastic Score (p< 0.001) (22).

Lastly, the systemic assessment portion of the NOMS framework determines the ability of individual patients to tolerate various treatment modalities based on the burden of metastatic disease (32). The most well-known and commonly used scoring system to assist physicians in predicting survival is the Tokuhashi score (32).

Surgery. The role of surgery in epidural metastatic spinal cord tumors is being redefined. In 1978, a retrospective, nonrandomized series of 235 patients with metastatic epidural spinal cord compression concluded that radiation therapy alone is as effective as decompressive laminectomy and radiation therapy (38). This study was instrumental at the time in changing initial therapy for metastatic epidural spinal cord compression from surgery followed by radiation therapy to radiation therapy alone. More recently, a randomized prospective comparison of lesion-directed surgery with stabilization followed by radiation therapy versus radiation therapy alone (86) confirmed a significant advantage for lesion-directed surgery and continued the renewed interest in a directed surgical approach based on the site and level of metastatic epidural spinal cord compression. Surgery is a major undertaking in patients with metastatic disease who have limited life expectancy. Nevertheless, surgery has been advocated to help the rapidly deteriorating patient to decompress the spinal cord and nerve roots, correct spinal instability, relieve pain, maintain continence, and promote early mobilization. Due to Patchell and colleagues’ 2005 study, surgery followed by radiation is often proposed as the definitive treatment plan for metastatic epidural spinal cord compression, depending on prognosis, especially in those with spinal instability. Surgical intervention (in addition to adjunct radiation therapy or chemotherapy) provides sustained improvement in neurologic and functional outcomes for patients who have at least a 3-month life expectancy (27).

Choice of surgical intervention depends on the location of the tumor in regards to both spinal level (cervical, thoracic, lumbar), location within the spine (anterior or posterior column), and prognosis. Many old trials compared decompressive laminectomy with radiation therapy, irrespective of anterior column involvement. Decompressive laminectomy alone for metastatic epidural spinal cord compression in the presence of vertebral body involvement is now contraindicated, as it can worsen or create spinal instability, with only a 3% probability of recovery of ambulation in one study (28). The only indications for decompressive laminectomy are tissue diagnosis and removal of posteriorly situated epidural deposits in the absence of vertebral body disease. Posterior stabilization via fusion may be required following decompressive laminectomy to prevent spinal instability. Unilateral posterolateral pedicular decompression with or without stabilization can also be used as a palliative measure or for those with pain secondary to pathologic fracture. In one study, 32 patients with severe metastatic disease and both anterior and posterior column cervical involvement underwent posterior decompression and fusion as a palliative measure. Of these 32 patients, 94% had improved spinal stability, pain, and neurologic function (85).

With advancement in surgical techniques, more aggressive tumor resections and reconstructions are available to address spinal cord compression, pain, and instability in appropriate candidates. Corpectomy, in which all or part of the vertebral body is removed, or anterior decompression followed by vertebral body reconstruction, are techniques employed to address anterior column disease. Both approaches can be technically challenging, as the anterior column is difficult to reach and, in the cervical spine, requires an anterior approach through the neck musculature, in the thoracic spine an anterolateral approach through the ribs, and in the lumbar region an anterolateral approach through the retroperitoneal plane. Such techniques often have longer recovery times than a posterior decompressive laminectomy and carry an additional risk of respiratory or abdominal complications. Patients undergoing corpectomy were found to have a greater 30-day postoperative complication rate, secondary to postoperative anemia and pulmonary complications compared to patients who underwent laminectomy (P< 0.0001) (04). However, paraplegic patients traditionally have not significantly improved with posterior decompressive laminectomy, but anterior vertebral body resection shows promise in paraplegic patients (111; 116; 117).

In one case series, clinical outcomes following anterior decompression in patients with a single anteriorly situated epidural metastasis are remarkably good. Fifty-two percent of these patients who are in good general health and have either failed radiation or have a radioresistant tumor will have improved ambulation and a median survival of 16 months (111). Operative mortality is 7%, and surgical morbidity is 11%. Spinal instability develops in 5% of patients, and recompression at the initial site will eventually occur in 22% of patients. As expected, there is a significant increase in complications and a reduction in survival for nonambulatory patients after a corpectomy (102).

In other surgical series of vertebral body resection, pain improved in 60% to 97% and neurologic function in 55% to 97% (116; 28; 83). If back pain is due to vertebral involvement with spinal instability and the patient has otherwise limited metastatic disease, spinal stabilization should be strongly considered.

One study has compared the benefit of surgical intervention followed by radiation therapy versus radiation therapy alone (86). In this randomized multi-institutional study of patients with metastatic epidural spinal cord compression, patients who were surgical candidates with only one level of cord compression were randomized to lesion-directed surgery with stabilization and radiation therapy or radiation therapy alone. Patients treated with surgery and radiation therapy retained the ability to walk for 122 days versus 13 days with radiation therapy alone (p=.006). Significantly more patients in the surgery group (42/50, 84%) than in the radiotherapy group (29/51, 57%) were able to walk after treatment (odds ratio 6.2, p=0.0001). The need for corticosteroids and opiates was significantly reduced in the surgery group. The 30-day mortality rate was 6% in the surgery group versus 14% in the radiation therapy alone group. The survival time was significantly greater in the surgical group, 126 days versus 100 days (p=.033), and the maintenance of continence was significantly longer, 156 days versus 14 days (p=.016). However, inclusion criteria for patients were very restrictive, leading to slow enrollment; patients had to have a single epidural mass at one location with one symptom or sign referable to the mass and not be paraplegic for greater than 48 hours, and have a 3-month or more life expectancy to be randomized. Cauda equina compression was not considered in this study (86). Only 25% of patients with radiographic epidural spinal cord compression at Johns Hopkins were felt to meet these criteria (10).

Not all data support surgical intervention followed by radiation therapy over radiation therapy alone. One retrospective cohort study in which patients were matched on multiple prognostic factors did not find any significant difference in outcome between those who had undergone surgery followed by radiation versus those who had undergone radiation therapy alone. The two groups had a similar ability to ambulate after treatment, regaining of ambulation, survival time, 1-year local control rates, and motor function (94). Thus, if patients meet the clinical and radiographic criteria in the Patchell study, they should be very strongly considered for directed anterior surgical resection with stabilization followed by radiation therapy. However, if patients do not meet the strict criteria of the Patchell study, surgery followed by radiation therapy should be at least considered, taking into account factors such as spinal instability, biology of the tumor (radioresistant vs. radiosensitive), and prognosis. Ultimately, more randomized control trials comparing the two treatment strategies are needed to determine if surgery plus radiation therapy results in better or worse outcome than radiation therapy alone.

Alternatively, minimally invasive surgical appropriates have also been studied. One less invasive therapeutic approach is percutaneous vertebroplasty and kyphoplasty. This involves the injection of polymethylmethacrylate (PMMA) directly into the vertebral body or after expansion of the vertebral body with a balloon device. One retrospective study showed a significant decrease in blood loss (P< 0.002) for patients who underwent kyphoplasty with short posterior instrumentation (n=50, mean blood loss 650 mL) compared to traditional long instrumentation (n=70, mean blood loss 2100 mL) (19). Another study compared a “mini-open approach” to the “traditional open approach” and found that the mini-open group was superior to the open group regarding estimated blood loss, blood transfusion, hospital stay, complications, and pain score. KPS, Frankel scores, and 24-month survival were similar in both groups (129).

A separate single-institution study in Japan showed that the mean Barthel Index for activities of daily living improved from 53.5 preoperatively to 71.5 postoperatively for patients treated with minimally invasive spine stabilization using percutaneous pedicle screws (122). The same group compared posterior decompression and stabilization, stabilization without decompression, and radiotherapy alone in the treatment of epidural spinal cord compression 1b or severe cases with ASIA E spinal metastasis and found no statistically significant difference in paralysis after intervention but still advocated for the use of minimally invasive spine stabilization, as it can lead to shorter time to adjuvant cancer therapy and better postoperative ADL outcomes (80).

Laser interstitial thermal therapy and cryoablation. Laser interstitial thermal therapy is a minimally invasive technique that has been proposed as an alternative to traditional surgery for the treatment of metastatic epidural spinal cord compression. One study enrolled 19 patients with metastatic epidural spinal cord compression and no motor deficits to receive MRI-guided spinal laser interstitial thermal therapy. At 2-month follow-up, 16 of the 19 patients revealed significant decompression of the neural component, whereas three patients had progressive disease (one requiring surgery and the other two requiring repeated LITT) (120). A more recent matched group study at a single institution compared patients with metastatic epidural spinal cord compression treated with open surgery (n=40) and laser interstitial thermal therapy (n=40) (23). Progression-free survival and overall survival were similar between groups but did not reach statistical significance. However, the laser interstitial thermal therapy cohort had a smaller postoperative decrease in the extent of epidural spinal cord compression but a lower estimated blood loss (117 vs. 1331 ml, p< 0.001), shorter hospital length of stay (3.4 vs. 9 days, p< 0.001), lower overall complication rate (5% vs. 35%, p=0.003), fewer days until radiotherapy or spinal stereotactic radiosurgery (7.8 vs. 35.9, p< 0.001), and systemic treatment (24.7 vs. 59 days, p=0.015) (23). Similarly, a separate group has published a small case series (n=7) where MRI-guided percutaneous cryoablation of vertebral metastases resulted in successful neural decompression (42). One hundred percent of patients had complete tumor ablation, and 80% had complete relief of preprocedural pain. Future studies will be needed to directly compare laser interstitial thermal therapy and imaging-guided cryoablation with traditional surgery and radiation therapy/stereotactic radiosurgery.

These modalities provide less rapid decompression of the tumor than traditional surgical techniques. Additionally, with laser interstitial thermal therapy, there is theoretical concern regarding heat injury to the spinal cord that needs to be considered when planning treatment.

Radiotherapy. Before the advent of more advanced surgical techniques, radiation therapy was often the first-line treatment for patients with metastatic epidural spinal cord compression. Now, radiation treatment is considered first-line in those who are not surgical candidates and for those following surgical decompression. Various radiation treatments are available, including external beam radiation therapy (EBRT), radiosurgery or stereotactic radiotherapy, and intensity-modulated radiotherapy.

With EBRT, radiation treatment targets the involved vertebral body as well as one level above and below, any paravertebral tumor, and the spine. A level above and below the area of metastatic involvement is commonly included in the radiation field to prevent marginal recurrence (55). However, more recent data suggest recurrence is most likely to occur at more distant sites, and utility of irradiating at a level above and below the site of involvement is being questioned (48; 63).

Response to EBRT is highly dependent on radiosensitivity of the tumor, with lymphoma, myeloma, small-cell lung cancer, breast, and prostate being radiosensitive; and kidney, colon, non-small cell lung cancer, and melanoma being more radioresistant (76; 92). Patients with radiosensitive tumors have better and longer response rates to radiation than those with radioresistant tumors. In one study, 75% of patients with radiosensitive tumors who were nonambulatory but could raise their legs off the bed became ambulant after radiotherapy, but only 34% of comparable patients with radioresistant tumors became ambulant after radiotherapy (38). Those with radiosensitive tumors have a longer duration of improvement, and in one study, those with radiosensitive tumors maintained motor response for 11 months, and those with radioresistant tumors only 3 months (77). Both radiosensitive and radioresistant tumors can have significant improvement in pain after EBRT. In one study, roughly 70% of all patients treated with EBRT achieved partial or complete regression of pain (77).

The sensitivity of the spinal cord to radiation limits the prescribed amount of therapy, and the spinal cord dose should always be calculated as well as the dose to the involved vertebral body. The incidence of permanent radiation injury to the spinal cord directly correlates with the total dose and fraction size (81). Other factors such as the presence of hypertension, advanced age, and immunocompromised state may lower the spinal cord’s tolerance to radiation therapy and increase the risk of radiation myelitis or myelopathy (109). Traditionally, stereotactic radiosurgery is only given to targets more than 3 mm from the spinal cord. However, a small retrospective review (n=20) revealed the use of stereotactic radiosurgery delivered to spinal metastatic lesions less than 3 mm from the spinal cord without a single case of resultant myelitis (82). This may reflect the duration of follow-up and the limited survival in this patient population. A separate phase 1 trial revealed no cases of radiation myelitis in a cohort that received a relaxed cord dmax of 16 Gy (37). Although the risk of radiation myelitis from stereotactic radiosurgery is known to be low, particularly if recommended dose metrics are followed (20), further studies will be needed to evaluate the safety of stereotactic radiosurgery at different doses and with more liberal margins of stereotactic radiosurgery delivery near the spinal cord.

The optimal dose and fractionation regimen for metastatic epidural spinal cord compression remain unknown. Each plan constructed represents a compromise between delivery of the highest dose achievable to improve tumor control, a desire to achieve palliation as expediently as possible, number of treatments (fractions), and the intrinsic radiosensitivity of the spinal cord. In fact, there may be no generally optimal plan, but consideration should be given to prognosis when deciding between a short-course versus long-course treatment (95). Dosing schedules can range from single large fractions to smaller fractions divided into a protracted treatment course. Shorter course therapy with a larger fraction size is more convenient to the patient as it requires less overall treatment time and fewer visits to the treatment center. One prospective, nonrandomized study compared short-course radiation therapy (8 Gy in 1 fraction or 20 Gy in 5 fractions) in 114 patients versus long-course radiation therapy (3 Gy in 10 fractions, 2.5 Gy in 15 fractions, or 2 Gy in 20 fractions) in 117 patients (96). The patients were similar in all common prognostic factors. This study found a significant difference in progression-free survival rates in the more protracted treatment group (at 12 months 72% in the long-course group and 55% in the short-course group). There also was a significant difference in local control, with 77% achieving local control in the long-course group and 61% in the short-course group. There was no difference in overall survival or functional outcome. In patients with poor prognoses, outcomes after 1 × 8 Gy and 5 × 4 Gy were not significantly different. In patients with favorable prognoses, need for in-field reRT was greater after 1 × 8 Gy (90). Secondary analysis also revealed that 4 Gy × 5 appeared noninferior to 3 Gy × 10 regarding pain and distress relief (97). A similar analysis found that in patients not undergoing surgical decompression, there was no statically significant difference in post-treatment quality of life when patients received a 10 Gy single fraction compared to receiving 20 Gy x 5 fractions (66). Though studies suggest that a more protracted course is desirable in terms of progression-free survival and local control, secondary analysis suggests that pretreatment prognosis should be considered when selecting a treatment course. This concept supports previous findings from a randomized control trial of 300 patients with a life expectancy of less than 6 months (75). These patients were randomized to short-course therapy (8 Gy in 2 fractions) or to long-course therapy (5 Gy in 3 fractions then 3 Gy in 5 fractions). There was no difference in treatment response, duration of response, survival, or toxicity between the two groups. Thus, in general for patients with a short life expectancy, short-course of EBRT should be considered as it offers similar palliation and response, with the benefit of providing convenience to the patient in terms of overall treatment time and number of clinic visits. For those with a longer life expectancy and better pretreatment predicted response to EBRT (ie, favorable histology, slow progression of motor deficits, oligometastatic disease, and absence of visceral metastases), a more protracted course should be considered to offer longer progression-free survival and better local control (92; 98; 96).

Stereotactic body radiosurgery (SBRS) or stereotactic body radiotherapy (SBRT) are techniques that deliver high-dose per fraction radiation to a well-defined target (126; 104; 47). With SBRS and SBRT, radiation exposure of the spinal cord and surrounding normal tissue is minimized, allowing higher treatment doses. Although EBRT provides effective treatment for radiosensitive tumors, there is emerging evidence that SBRS and SBRT can provide good local control and pain relief even in those with radioresistant histologies (35). A 2016 consensus statement formulated by a panel of 15 radiation oncologists and five neurosurgeons suggested that postoperative SBRT is indicated for radio-resistant primary lesions, disease confined to one to two levels, or prior overlapping radiotherapy (99). In one series, 85 lesions were treated in 62 patients with stereotactic radiosurgery with a single fraction of SBRT (median of 16 Gy) (104). There was a 65% reduction in epidural tumor volume 2 months after treatment and an increase of 55% to 76% in thecal sac patency (104). Neurologic function improved in 81% of patients. High doses of SBRT must be used to achieve such results, meaning that SBRT can only be used in epidural spinal cord metastases in the absence of high-grade cord compression. In another study of 500 patients treated with SBRS, 65 of those treated with SBRS had no prior radiation or surgery (35). In these 65 patients, a majority achieved pain control at 1-month follow-up, and radiographic control in 90% of patients (35). No patients had evidence of neurologic symptoms due to spinal cord toxicity at follow-up ranging from 3 to 53 months (35). SBRT or SBRS may also be considered in patients as adjuvant definitive treatment after surgical decompression. In a retrospective study of 187 patients with surgical decompression followed by SBRT, local tumor control was maintained in 82% of patients (64). Those who underwent high-dose single SBRT had a slightly higher recurrence rate, with 1-year progression rate of less than 10%, whereas those who received a high-dose hypofractionated course had a 1-year progression rate of less than 5% (64). SBRS or SBRT can be considered definitive treatment either alone or following surgical decompression. The authors from this study advocate for “hybrid therapy” (64), where separation surgery followed by concomitant stereotactic radiosurgery is a superior approach to traditional laminectomy followed by stereotactic radiosurgery, as the former provides safer margins for stereotactic radiosurgery delivery (08).

In recurrent epidural spinal cord compression, SBRT or SBRS, EBRT, surgery, or chemotherapy may be considered. A second course of palliative EBRT has been used to treat recurrent same-segment spinal cord compression, with 74% of patients ambulatory at the onset of reirradiation and 78% on completion of radiation therapy (108). Sixty-nine percent were ambulatory for a median duration of 4.7 months following completion of reirradiation, and the median survival was 4.2 months. Reirradiation carries minimal risk of radiation myelopathy for patients with progressive recurrent epidural disease. As mentioned above, in Gerszten and colleagues’ 2007 study of 500 patients treated with SBRS, 344 had recurrent disease previously treated with EBRT. Of this subset of 344 patients, 86% achieved long-term radiographic control and pain control, and none demonstrated evidence of spinal cord toxicity (35). If patients develop a neurologic deficit during radiation therapy and the deficit is unresponsive to steroid dose increase, a directed surgical approach can be considered. In the 2005 Patchell study, these patients did less well than those who underwent initial surgical intervention. A “good prognosis” per the Rades score predicted the best outcomes in a cohort of patients that underwent SBRT for postoperative reirradiation of metastatic epidural spinal cord compression (53).

Several small investigations regarding the use of computed tomography-guided iodine-125 seed implantation therapy as an alternative palliative measure for metastatic epidural spinal cord compression found the intervention to be safe (n=28) with no neurologic sequelae (72) and efficacious at relieving pain, as the Numerical Rating Scale (which assesses the degree of pain) scores before implantation and at postoperative 3 and 6 months were 7.81±0.74, 2.04±1.10, and 1.81±0.79, respectively, (P< 0.05). Further, KPS scores before treatment and at 3 months and 6 months postoperatively were 66.30±6.88, 85.93±9.31, and 87.91±8.56, respectively (P< 0.05) (50). Median survival time in this cohort was 10 months, with an 82% response rate (70). Randomized controlled trials are needed to directly compare the clinical efficacy of computed tomography-guided radioactive iodine-125 seed implantation with other already accepted methods (discussed above) in patients with metastatic epidural spinal cord compression.

Chemotherapy. Chemotherapy may have a place in treating patients with metastatic epidural spinal cord compression. The use of chemotherapy relies on the chemosensitivity of the tumor. Chemotherapy can be considered for patients who have chemosensitive tumors but is often used as an adjuvant therapy combined with radiation or surgery. However, some patients with extremely chemosensitive tumors can have an excellent response with chemotherapy alone and may not require other definitive treatments. In one study, 20 of 23 patients with metastatic epidural spinal cord compression secondary to breast cancer who were treated primarily with chemotherapy experienced a resolution of signs and symptoms (45). Chemotherapy alone has had the most success in patients who are treated early and in those with tumors such as germ cell and hematological malignancies (127; 33). A plasmacytoma was found to be responsive to a combination of bortezomib, cyclophosphamide, and dexamethasone (79). Chemotherapy may also be considered for patients who have already received radiation or surgery and are not candidates for either treatment. Complete resolution of paraparesis following intravenous chemotherapy has been reported in patients with breast cancer who have failed radiation (12).

Immunotherapy. Per trends in medical oncology, immunotherapy use is increasing, particularly immune checkpoint blockade, to treat several cancers. One study revealed that the use of radiation therapy to treat vertebral metastatic disease is well tolerated in patients on a PD-1 inhibitor. Numbness and weakness were more likely to improve in patients who received radiation therapy after starting treatment with a PD-1 inhibitor when compared to patients who received radiation therapy before starting treatment with a PD-1 inhibitor (71% v 17%, P=0.04) (26). Of note, most patients in this study did not have epidural spinal cord compression (n=12/37) but instead had spine metastases without cord compression. Future studies will be needed to further assess immunotherapy used for cancers that are causing metastatic epidural spinal cord compression.

Rehabilitative care. Most patients with metastatic cord compression regain some function compared to patients with ischemic or traumatic spinal cord injury (65). During acute hospitalization, patients should be mobilized as soon as possible to avoid complications of immobility. A physical medicine and rehabilitation specialist should be involved in coordinating patient care. Management of bowel and bladder dysfunction should also be managed appropriately. Patients experiencing urinary retention should have scheduled straight catheterization, whereas those with sphincter incontinence should receive regular-timed voiding and bladder training (65). Bowel regimens for retention should include an osmotic agent (polyethylene glycol) combined with motility agents (senna). For fecal incontinence, bulking agents such as psyllium should be implemented (65). Additionally, patients with compromised sensation require pressure-relieving bedding and regular weight shifts to protect areas prone to breakdown, such as calcanei and ischial tuberosities (65).