According to current understanding of the pathophysiology of metastatic disease, in order for metastasis to occur a complex multistep process must occur involving genetic instability, acquisition of malignant phenotype, growth of the tumor cell, extravasation from the local environment, dissemination in the circulation, adhesion in the new environment, angiogenesis, and new focal colonization. One early explanation for the relatively high frequency of prostate carcinoma metastatic to the thoracolumbar spine involves retrograde communication from the prostatic venous plexus through a valveless vertebral venous plexus which parallels the caval system, the so-called Batson’s plexus [12]. Since the original description, this venous plexus has been invoked in explaining the pathophysiology of spinal metastasis for many different tumor types. However, if metastatic spread were predominantly along this pathway, one would expect metastatic tumor deposits to be along the course of the end venules in each vertebra, namely in the posterior and middle vertebral body. In fact, the metastatic distributions are considerably more varied, and no correlation has been found between the anatomy of vertebral venous drainage and vertebral metastatic deposits. Thus, other more complex distribution mechanisms likely exist which depend on both arterial and venous pathways, the primary tumor location, and molecular based adhesion and colonization properties [13, 14]. Other mechanisms for metastatic tumor invasion of the spine include direct local extension from organs such as the lung or kidney, or invasion via cerebrospinal fluid seeding.

The Weinstein-Boriani-Biagini system (Fig. 29.1) [15] was designed as an anatomic classification for primary tumors of the spine but is a useful descriptive scheme for any spinal lesion including metastatic foci. It allows for precise axial and radial localization of the lesion within the extra-osseus, intra-osseous, intra-canal/extradural, and intradural spaces. Although up to 95 % of metastatic spinal lesions occur extradurally [9], it is important to note that intradural or intramedullary metastases do occur [16, 17] and their treatment should involve a surgical team specifically trained in the care of intra-dural neoplasms.

Each segment of the mobile spine has unique anatomic properties and implications for the treatment of metastatic disease. For example, one should recognize that metastatic epidural compression at the spinal cord level in the zone of the cervical or thoracic spine has more serious implications for neurologic injury than nerve-root-level compression below the conus medullaris (typically caudal to L1 or L2). Different zones of the mobile spine also have different structural properties. In the cervical spine, the intrinsic mobility of each spinal motion segment may predispose to early instability-related myelopathic progression, whereas in the thoracic spine the stabilizing effect offered by the costosternal architecture may temporize instability-related pain and cord injury. In contrast, while the lower lumbar and lumbosacral spine may tolerate more spinal canal encroachment due to the root-level anatomy, bony destruction and instability may be particularly poorly tolerated from a pain standpoint given the high mechanical loads placed upon this region of the spine.

Another anatomic consideration is the surrounding vascular anatomy. The vertebral artery or great vessel architecture may dictate the laterality and orientation of the surgical approach. Also, one may consider preoperative angiographic mapping when planning possible sacrifice of key segmental or radiculomedullary vessels such as the Adamkiewicz artery, since their disruption could lead to spinal cord ischemia [18, 19].

Average life expectancy from the time of diagnosis with bony spinal metastases is short (often less than 12 months), with variation based on histology, extent of disease, neurologic status, performance status, and other factors [20–22]. Although many texts and expert opinions cite a 3-month predicted survival as a rough cutoff for surgical candidacy, no hard rules exist to define when and in whom a more aggressive approach should be pursued. In addition, there are many different surgical approaches now available, each with their own unique morbidity and mortality risks. Multiple authors have attempted to risk-stratify patients by predicted survival in order to identify a priori those who are not likely to survive for long periods of time and thus may not be appropriate for open surgical intervention. Several of the more popular scoring systems are listed in Table 29.2, along with their stratified survival data. A recent analysis of these scoring systems retrofit to a large, single center dataset identified the Bauer and modified Bauer prognostic scale as the best instruments for distinguishing between good, moderate, and poor survival prognoses [22]. While these scoring systems help predict survival for groups of patients, it should be stressed that they are merely tools to assist the care team in estimating survival and planning treatment. It is notoriously difficult to predict any individual patient’s survival, and given the constant evolution of medical, surgical, and radiotherapies, each treatment plan should be highly individualized by the multidisciplinary care team.

Most patients with metastatic spine disease first experience back pain [23], but by the time they present for care, up to 85 % may have a true neurologic complaint [24]. Factors which distinguish neoplastic pain from more common degenerative complaints include acuity of onset, progressive quality, night pain, accompanying constitutional symptoms, weight loss, or other suggestive disease-specific symptoms such as a breast mass, goiter, or hematuria. Carefully distinguishing rest or night pain from activity-related mechanical pain is especially important because the latter may indicate mechanical instability. A detailed neurologic exam should be performed by a qualified practitioner versed in care of the spine and should include examination of sharp, dull, and light touch sensation, motor grading, and reflex evaluation. Ambulatory status and a sensorimotor level should be established if applicable. Radiculopathy should be distinguished from myelopathy or frank spinal cord injury. Signs of myelopathy such as gait dysfunction, difficulty with fine motor coordination, hyperreflexia, and bowel or bladder dysfunction may be subtle and should be carefully investigated.

Since up to 20 % of initial presentations of metastatic disease will present with a spine-related complaint [25], physicians caring for the spine should be familiar with the general work-up of metastatic disease. In the presence of multiple skeletal metastases of unknown origin, history and physical exam along with computed tomography scans of the chest, abdomen, and pelvis identify nearly 80 % of primary tumors [26]. Further advanced imaging, specialized laboratory tests, and biopsy modalities may ensue if the diagnosis continues to be in doubt.

Great care must be taken when one encounters a solitary aggressive-appearing neoplastic lesion of the spine. Although metastatic disease is possible and is indeed amongst the more likely entities, one must also consider that the lesion could be a primary malignant tumor of bone. Intralesional excision or unplanned biopsy in this setting can have disastrous consequences and make future treatment difficult. Table 29.3 lists common entities in the differential diagnosis for a solitary neoplastic lesion of the bony spine.

Initial imaging should consist of upright standing full-length scoliosis-style films of the spine in the coronal and sagittal planes. This allows for visualization of the symptomatic area in question, other abnormal areas which may be asymptomatic, and for an assessment of coronal and sagittal alignment. The latter become important when assessing tumor-related instability and in planning reconstructive surgery. If standing films are not obtainable due to patient intolerance, full length sitting films may substitute. It should be noted that plain films typically detect bony destruction only in the late, progressive state, for example when more than 50 % of the vertebral bone has been replaced [27]. For context, the classically described “winking owl” sign, or loss of the pedicular density on AP radiographs, usually can’t be seen unless nearly the entire pedicle cortex had been replaced by tumor.

Technetium-99 scintigraphy and in some cases fluorodeoxyglucose positron emission tomography (FDG-PET) scanning may contribute to systemic disease staging, but are not usually first-line modalities in the work-up of metastatic disease to the spine. They are sensitive but not specific for metastatic disease. The most useful advanced imaging modalities in the spine are computed tomography (CT) and magnetic resonance (MR) imaging. The former is useful in demonstrating detailed bony architecture and as such is able to identify subtle pathologic fracture or corticocancellous destruction earlier than plain radiographs, bone scanning, or MR. In addition, CT may demonstrate intralesional calcifications in the case of chondrosarcoma or chordoma, phleboliths and normal intervening marrow architecture in the case of hemangioma, and other tumor-specific findings. For metastatic disease in general, it is best used to understand the extent of bony destruction. The main disadvantage is the relatively large amount of ionizing radiation exposure.

Conversely, MR is an excellent soft-tissue imaging modality which emits no ionizing radiation and should be used liberally to delineate the extent of metastatic disease and its relation to the neural elements. We recommend MR imaging of the entire spinal column when confronted with a spinal neoplasm, as this is a sensitive and specific modality for identifying synchronous lesions [28], and helps with treatment planning. Generally speaking, it is advantageous to use intravenous gadolinium contrast to aid visualization of tumor vascularization and help differentiate solid vascularized tumor from cystic necrosis, scar, or fluid-filled cavities. MR can also be helpful in differentiating de novo or recurrent metastatic bone disease from osteoporotic compression fractures or radionecrosis [29–32].

If the diagnosis is in doubt after thorough noninvasive work-up or if the lesion is solitary, biopsy should be strongly considered. We and others [33] feel that the best outcomes occur when the biopsy is performed by experienced interventionalists or proceduralists at the final treating institution, as part of a multidisciplinary treatment plan that takes into account the expected histology and the likely surgical approach for resection.

While fine needle aspirate may demonstrate neoplastic cells, it does not offer preservation of tissue architecture and thus CT-guided core-needle biopsy has become the standard modality. This is typically performed via a posterior transpedicular approach and is useful for most vertebral body tumor locations except for those posterior central locations where access is difficult. Cultures should be sent from most tumor biopsies to rule out infection, and depending on the expected histology, advanced histopathologic techniques can be employed such as with flow cytometry in lymphoma. Needle biopsy of the spine through hollow viscera or along tracts of tissue that cannot be readily excised during definitive surgery is discouraged given the theoretical concern for needle tract contamination and tumor cell implantation.

Open biopsy may be required in the case of equivocal needle biopsy results or progressive neurologic deficit requiring urgent concomitant decompression of the neural elements, but this procedure should be undertaken with great care given the potential for tumor contamination and compromised future treatment options.

Metastatic epidural cord compression is a clinical scenario that deserves separate commentary because of the propensity for permanent neurologic injury and disability. Compression of the neural elements can generally occur via three mechanisms: direct tumor compression, bony retropulsion or deformity due to pathologic fracture, or more rarely, impingement from osteoblastic bone response. The Spine Oncology Study Group [34] has laid out an anatomic classification system of metastatic epidural spinal cord compression (MESCC) based on axial T2 MRI that can be broken down into six categories: grade 0 (bone only), grade 1a (tumor in epidural space but no thecal sack compression), grade 1b (thecal sack compression but no touching cord), 1c (cord abutment without deformation), 2 (cord deformation with some surrounding cerebrospinal fluid (CSF) visible), and 3 (no remaining visible CSF). It is impossible to direct treatment based on this or any other one classification system alone because of the interaction of other factors such as the neurologic exam and mechanical instability, but the authors suggest tumors which fall into the latter two grades be considered “high grade” and require open surgical decompression prior to radiotherapy given the proximity of tumor to neural elements. Conversely, they suggest other grades be considered for radiotherapy prior to surgery in the absence of mechanical instability.

Since the life expectancy of patients with metastatic disease to the spine is short, the goals of treatment are typically palliation of pain, maintenance of ambulatory function via preservation of spinal stability, and optimization of quality of life. In certain settings, goals may also include optimization of neurologic recovery or local tumor control. Three patient specific factors should be considered when selecting the most appropriate treatment : the neurologic status, the mechanical stability of the spine, and the systemic status of disease. Aspects of this framework have been widely described, but a treatment algorithm based on the NOMS (neurologic, oncologic, mechanical, and systemic) framework was recently crystallized by Laufer et al. [35], and forms the basis for the adapted treatment framework we lay out in Fig. 29.2. Rather than presenting a rigid algorithm, we present positive prognostic factors in each of the three domains which help inform the invasiveness of intervention along the displayed hierarchy of possible treatments. As has been stated, treatment should be highly individualized, and many factors contribute to selecting the most appropriate course.

The neurologic status of the patient forms the basis of the acuity of treatment, and the presenting neurologic status is one of the best predictors of posttreatment neurologic status [36]. Although the optimal timing of treatment for vertebral metastases has yet to be defined, it intuitively follows that treatment is most advantageous prior to progressive neurologic involvement, when options become more limited. In a prospective audit by the Scottish Cord Compression Study Group of 319 patients who developed malignant cord compression, the median time interval between back pain in cancer patients and development of neurologic symptoms was 66 days [37]. Thus, the patient’s neurologic condition should be carefully established. For those patients with an evolving or progressive neurologic deficit, urgent surgery within 48 h of the onset of symptoms [38] seems to result in the best neurologic outcomes. An urgent operative intervention in this clinical setting is also supported by various studies in the traumatic spinal cord injury literature [39]. In the case of an evolving, cord-level neurologic deficit it is our practice to intervene with surgical decompression immediately or as soon as the patient can be medically optimized for surgery.

The mechanical stability of the spine is also a major consideration in planning treatment, and it becomes the principle issue when no neurologic deficit exists. Clinical signs of impending or frank mechanical instability include abrupt new axial or radicular pain, especially that which worsens with activity. Severe cases may present with changes in clinical alignment such as with horizontal gaze in the cervical spine or ability to stand upright in the thoracolumbar spine. Multiple studies have attempted to identify risk factors for thoracolumbar instability using finite element, biomechanical, cadaveric, and clinical designs. These are outlined in Weber’s recent systematic review [40]. Consistent factors which seem to predict higher risk for compression or burst fracture in the thoracolumbar spine include larger tumor volume, location of tumor in the pedicles and other posterior elements, loss of integrity of the ribcage or costovertebral junction, poor baseline bone mineral density, baseline sagittal imbalance, and high spinal loads such as with obesity and strenuous activity. Combining consensus expert opinion with published data, the Spine Oncology Study Group recently created a novel scoring system for predicting spinal instability (Table 29.4, adapted from [41]). While no prospective validation of this schema has been performed, it appears to be a reliable and retrospectively validated scoring system [42, 43], with scores from 0 to 6 predicting stability, scores from 13 to 18 predicting instability, and scores from 7 to 12 having indeterminate stability.