Up to 95% of adult and 80% of pediatric patients with EC present with pain (10,32). The difference in pain prevalence between adults and children may reflect greater difficulty in the pain assessment of and the underreporting of pain in children. Pain may precede other symptoms and signs of EC by 1 year (11). This interval may vary by tumor type; it is generally shorter for lung cancer than breast cancer (33). Overall, patients experience pain for an average of 4 to 5 months before presentation (3).

Pain may be local at the site of pathology or referred in a nonradicular or a radicular (dermatomal) distribution or have combined features. Radicular or root pain is reported in 90% of lumbosacral EC, 79% of cervical compression, and 55% of thoracic cord compression (33). Radicular pain may be bilateral in thoracic lesions and is often described as a tight band around the chest or abdomen. It is important to note that radicular pain may be experienced in only one part of a dermatome (as a partial suspended sensory level). When a nerve root lesion produces chest or abdominal pain, the complaint may be mistakenly identified as referred pain of visceral origin. Radicular lesions are usually associated with segmental findings on examination. Nonradicular referred pain may be associated with vague paresthesias and tenderness at the painful site. Pain may be continuous at rest and markedly aggravated by body movements (incident pain). Although local pain from a vertebral lesion is worsened with loading due to upright posture, pain due to EC is often greatly increased by lying supine. A lesion confined to the vertebral body may also produce nonradicular referred pain.
Disease at C7 may refer pain to the interscapular region, and pain due to disease at L1 may be referred to the iliac crests, hips, or sacroiliac region. Sacral disease often causes midline pain radiating to the buttocks, which is made worse with sitting. Radicular pain, in particular, may be paroxysmal, spontaneous, or provoked by movement or sensory stimulation. Valsalva’s maneuver may produce or aggravate both local and radicular pain. Pain on neck flexion or straight leg raising implies dural traction. Lhermitte’s sign (electric shock-like pain in the spine) is a symptom of spinal cord dysfunction. Compression of the cervical spinal cord rarely produces funicular pain, which is pain referred to the lower extremities, thorax, or abdomen as a band of paresthesias. “Pseudoclaudication” of legs may be an isolated lumbar root symptom (2).

The neurologic findings associated with EC also vary considerably. There can be extensive epidural tumor with no neurologic findings on examination. Upper motor neuron weakness may occur with lesions of the spinal cord (above the L1 vertebral body). This finding is present in 75% of patients with EC at diagnosis (10). Sensory changes occur in approximately half of patients at presentation, including paresthesias and sensory loss, which can be segmental or below the level of injury. Sensory complaint without pain is exceedingly rare. Bladder and bowel dysfunction are evident in more than half of patients on presentation with EC; constipation usually precedes urinary retention or incontinence (2).

The physical examination begins with the observation of posture, spinal curvature, symmetry of paraspinal muscles, extremities, and skin. The practitioner may appreciate tenderness of the spinous processes on palpation or percussion, although this may not correlate with the level of spinal disease. Gibbus deformity and vertebral misalignments are frequently palpable; actual crepitus of the spine is unusual. Tenderness or spasm of the paraspinal muscles may also be noted. Urinary retention may be demonstrated by bladder percussion. Laxity of the anal sphincter may be apparent on digital rectal examination. Specific areas of sacral or coccygeal tenderness may be identified by external palpation, rectal, or pelvic examination.

Spinal maneuvers to elicit pain should be performed carefully. Thoracic and abdominal radicular pain may be provoked on lateral flexion and rotation of the trunk. Increased pain on neck flexion and straight leg raise sign may be “pseudomeningeal” signs of dural traction due to epidural tumor. If neck rigidity is present, the examiner should use extreme caution with range-of-motion maneuvers. Muscle spasm may be triggered by bony instability of the cervical spine, and forced movements may dislodge bony fragments, causing acute spinal cord or brainstem injury.

The neurologic examination reveals positive findings in most patients with EC. The examination should include assessment of mental status, cranial nerves, motor function, reflexes, sensation, coordination, and gait. Proximal lower extremity weakness may be initially evident only as difficulty in rising from a chair. Although weakness due to upper motor neuron dysfunction is usually associated with increased tone and hyperreflexia, acute “spinal shock” can cause a flaccid areflexic paralysis. In the subacute phase of recovery from spinal shock, “mass reflexes” appear consisting of flexor spasms, hyperhidrosis, and piloerection due to autonomic dysfunction. Lower motor neuron weakness may be accompanied by flaccidity, atrophy, muscle fasciculations, and hyporeflexia. A cervical lesion can produce segmental hyporeflexia in the arm or arms and increased reflexes below. Lesions above the pyramidal decussation of the corticospinal tracts in the lower brainstem may be associated with the loss of contralateral abdominal reflexes; lesions below the decussation produce loss of ipsilateral abdominal reflexes. Segmental motor dysfunction due to thoracic nerve root disease may produce asymmetric abdominal muscle contraction and loss of abdominal reflexes. Beevor’s sign (upward movement of the umbilicus on attempted flexion of the trunk) indicates a lesion at or near the T10 thoracic level. Lesions of the roots of the upper lumbar plexus produce hip flexion weakness and a dropped knee jerk reflex; lesions of the roots to the lower lumbar plexus may produce foot drop and diminished ankle jerk reflex. Loss of bulbocavernosus and anal reflexes may accompany spinal cord conus and cauda equina lesions (2).

Although the sensory examination may help in determining the level of epidural disease, EC results in a broad variation of sensory dysfunction, with incomplete lesions being the rule. The level of reduced sensation may be determined to be up to five segmental levels below, or one to two segments above, the level of cord compression. A sensory level on the trunk sparing the sacral dermatomes may occur in up to 20% of patients with thoracic or high lumbar compression (2). Suspended partial sensory levels or unilateral bands of sensory loss may be seen with spinal cord lesions up to the brainstem. Facial numbness may be due to upper cervical lesions. Lesions of the upper thoracic nerve roots may result in Horner syndrome, with autonomic dysfunction of the face and upper extremity. Compression of the conus of the spinal cord may produce sensory loss in the saddle area (buttocks and perineum) without lower extremity symptoms or signs.

Gait ataxia is an uncommon isolated sign of spinal cord compression. Other unusual features are signs of raised intracranial pressure; facial paresis, lower extremity fasciculations, or sciatica with cervical tumor; nystagmus with thoracic tumor; spinal myoclonus; an inverted knee jerk reflex; and “painful legs and moving toes” (11).

The selection of specific imaging tests is guided by the clinical presentation. Several imaging methods are available to confirm EC. Because the correct interpretation of symptomatic and asymptomatic lesions on diagnostic imaging studies requires a thorough knowledge of the patient’s clinical
presentation, it is strongly recommended that clinicoradiographic correlation be made by the examining physician. In each individual case, the “neurologic urgency” for further diagnostic tests must be modified according to the potential for treatment, the patient’s condition, and overall prognosis (Fig. 39.1).

Plain radiographs confirm tumor and assess structural stability of the spinal elements. In the patient with cancer at risk for spinal metastases with neck, shoulder, or upper extremity pain, flexion and extension views of the cervical spine should not be forced. Although plain radiographs are >90% sensitive and 86% specific for demonstrating abnormalities in the patient with symptomatic spinal metastases, autopsy series suggest that up to 25% of spinal lesions are invisible on radiography (2). False negatives occur because of mild degree of pathology, because of poor visualization
(e.g., the first thoracic vertebra), or because the abnormality is missed on interpretation. The false-positive rate for interpreting collapsed vertebrae as malignant may be as high as 20% (34).

It is estimated that a 30% to 50% change in bone mass is needed before plain films become abnormal (32). On anterior/posterior view, spinal radiographs may show pedicle erosion (the “winking owl” sign), increased interpeduncular distance, paraspinal widening, or paraspinal soft tissue shadow. On lateral view, vertebral collapse (wedging of the body), scalloped bodies, disk space destruction, a narrow spinal canal, hypertrophied facets, and disk calcification may be seen. Oblique views are needed to discriminate spondylotic osteophytic encroachment from tumor causing foraminal abnormality (5). Vertebral collapse and pedicle erosion >50% are especially predictive of EC. On plain radiography, multiple vertebral involvement is noted in up to 86% of patients with spinal tumor (5) and in >30% of patients with EC.

Computed tomography (CT) scan may be useful to better delineate pathology using restricted fields of view (2). CT is superior to other imaging techniques for demonstrating cortical bone architecture (4). Before the availability of MRI, CT scan in combination with myelography was considered the gold standard for demonstrating the level and extent of epidural disease. CT myelography may be considered if the index of suspicion for epidural disease is high and other imaging studies are normal or if MRI cannot be interpreted or performed. Lumbar puncture should precede cervical puncture in most cases. Injection of air to supplement contrast medium may better image a CSF block. If the upper and lower extent of the block cover a long spinal segment, myelography may be repeated after treatment to determine if multiple discrete lesions are present and to better define radiotherapy portals. If repeated imaging is anticipated, oil-based contrast medium may be used to allow for follow-up radiographic imaging without repeated punctures. Another advantage of myelography over other diagnostic imaging tests is the collection of CSF for analysis. However, there is a risk of worsening neurologic function after dural puncture in the patient with partial CSF block, due to the “coning” of the spinal cord as the pressure below the block is relieved. This risk may be as high as 15% (2,11). It is therefore recommended that under these conditions, corticosteroids be administered before dural puncture.

Radionuclide bone scintigrams reveal a 5% to 10% change in bone tissue (32). Bone scintigrams are more sensitive than radiographs except in multiple myeloma (11). However, they are not as specific as radiographs in identifying the level of EC. False positives may be due to nonmalignant skeletal conditions and false negatives due to lytic lesions, for example, myeloma or solid tumors such as lung and melanoma, and prior radiation therapy. If the entire skeleton is involved by tumor, no contrast in the radionuclide uptake may be appreciated. New technology of immunoscintigraphy may prove to be more sensitive (2).