Stereotactic Body Radiotherapy for Spinal Metastases

Rationale for Spine Stereotactic Body Radiotherapy

The term radiosurgery has been reserved for ablative doses of radiation directed to intracranial tumors with stereotactic localization and delivered typically in a single fraction. Thanks to the development of high-precision linear accelerators equipped with image guidance and robotics, radiosurgery can now be delivered to extracranial tumors in one to five fractions, a process termed stereotactic body radiotherapy (SBRT).

For patients with spine and paraspinal metastases, the traditional treatment has been to deliver palliative conventional external beam radiation (cEBRT) to reduce pain and analgesic requirements while improving or maintaining local tumor control, mobility, neurologic function, and mechanical stability. Palliative cEBRT treatment courses were designed to be short, typically lasting no more than 2 weeks, so as to reduce interruptions in the overall oncologic care. cEBRT regimens for spine metastases include schedules such as 8 Gy in one fraction, 20 Gy in five fractions, and 30 Gy in 10 fractions. Although this approach is effective in reducing pain, complete pain response and local control rates have been reported to be modest at best, particularly for complex spinal metastases.

Spinal SBRT was developed in order to improve on historic local control rates, especially for radioresistant disease (such as renal, melanoma, sarcoma, or colon histologies), complex bulky tumors with paraspinal and/or epidural extension (mass type), or lesions previously irradiated. Typical total doses include 16 to 24 Gy in a single fraction, 24 Gy in two fractions, and 30 to 40 Gy in three to five fractions. This technique requires a high level of technical rigor, which was realized not only with specialized delivery platforms such as the CyberKnife (Accuray Inc.) but also with mainstream linear accelerators owing to the advent of image guidance, robotic technology, subcentimeter multileaf collimators, and sophisticated treatment planning systems. At present, there have been significant publications for approximately a decade detailing outcomes specific to spine SBRT, guidelines reported by organizations such as the AOSpine, and the International Spine Radiosurgery Consortium ; there are also randomized trials in the process of completion.

Initially, the fear associated with spine SBRT involved the risk of overdosing the spinal cord, as the target and the cord are in close proximity and the steepest dose gradient is intentionally at that anatomic junction. A second challenge involves treatment planning, as the spinal cord’s tolerance is lower than the prescribed dose; therefore the trade-off with spine SBRT lies in maximizing coverage of the target while respecting this organ at risk. Spinal cord tolerance guidelines in both the de novo and reirradiation settings ^, have resulted in standardization of the technique and approach with accepted parameters for tumor coverage. As a result, spinal cord myelopathy is considered a rare event and local control rates of ∼80% to 90% are frequently reported. The third major challenge lies in the target and healthy bone being the same target volume. This has led to a better understanding of radiation-induced fractures and improved reporting of a dose-complication relationship.

We have now obtained mature outcomes with spine SBRT, further understanding of the radiobiology of SBRT, and an in-depth understanding of the adverse event profile. These are the focus of this chapter.

Treatment Planning Overview

The critical goal of spine SBRT is to allow for dose escalation within the target volume while keeping the dose to critical organs, namely the spinal cord, within safe constraints. Given the steep dose gradient that is purposefully created between the spinal cord and tumor and that positional deviations of 1 to 2 mm can dramatically increase the dose received by the spinal cord, near-rigid immobilization is of utmost importance.

When linear accelerator (LINAC)–based isocentric treatments are being delivered, near-rigid body immobilization has been shown to reduce intrafraction motion compared with less rigid devices and is thus recommended. If target lesions are located in the cervical or upper thoracic spine, a five-point head and shoulder mask should be employed for immobilization. Exceptions to the need for strict near-rigid immobilization include technologies that incorporate real-time imaging with feedback for positional corrections such as those inherent to the CyberKnife, although these are still recommended.

Radiotherapy plans are fundamentally created based on a treatment planning simulation CT scan. For the purposes of spine SBRT, the CT scan acquired should be thin slice (1 to 2.5 mm; we recommend 1 mm) and coregistered with a simulation MRI performed on the same day or in close proximity as the simulation CT scan. The tumor and organs at risk (OARs) both require MRI for delineation. The axial MRI scan should be thin slice, volumetric, obtained without gadolinium, and include axial T1- and T2-weighted sequences. Scanning limits should encompass at least one healthy vertebral body above and below the target spinal segment or segments. A planning CT myelogram may be indicated in situations where the fusion is unreliable; for example, if the patient has severe scoliosis or if implanted hardware results in significant imaging artifact. ^, The published recommendations from the SPIne response assessment in Neuro-Oncology (SPINO) group provide full details on these imaging requirements.

The contouring of targets and relevant OARs is based primarily on MRI; however, CT images can be useful when bony structures are being delineated and for differentiating between lytic and blastic osseous metastases. Gross soft tissue and osseous disease is encompassed by the gross tumor volume (GTV). The clinical target volume (CTV) encompasses the GTV and is expanded to include anatomy that is at risk of harboring subclinical disease ( Fig. 90.1 ). Finally, a planning target volume (PTV) is generated as an isotropic expansion on top of the CTV to account for setup error and treatment delivery uncertainties; a 2-mm margin is typically used at our institution. Exclusion of the PTV from the spinal cord is practiced by some; this simply changes the coverage that is accepted for planning purposes. Ultimately, coverage is maximized while also respecting the strict constraints to the OARs.

For the CTV, adhering to the International Spine Radiosurgery Consortium guidelines will inform which segments of the involved vertebral segment need to be included; this represents a standard of care. In cases where there is paraspinal or epidural soft tissue extension, we recommend a CTV margin of 5 mm beyond the GTV both in the axial and craniocaudal planes while respecting known anatomic boundaries to tumor spread so as to reduce the rate of marginal failure. OAR delineation depends on the spinal levels involved and may include but is not limited to the spinal cord and/or thecal sac, larynx, pharynx, trachea, lungs, esophagus, stomach, bowel, liver, kidneys, and sacral nerve roots. As an additional precaution, we recommend that a 1.5-mm expansion be generated on the spinal cord contour, known as a planning at risk volume (PRV), and we apply evidence-based cord constraints ^, ( Table 90.1 ) to that structure as a strict dose limit. At the level of cauda equina, the thecal sac is contoured. No PRV expansion is required and dose limits are applied to the thecal sac.

It is mandatory that treatments be delivered using image guidance. Our protocol involves acquiring a kilovoltage cone-beam CT to identify and correct initial patient setup errors with a correction threshold of 1 mm and 1 degree. A 6-degrees-of-freedom (3 degrees of translation, 3 degrees of rotation) robotic couch enables precise corrections to the patient setup. Treatment dose is discussed in the sections to follow.

Patient/Tumor Assessment Overview

The safe and effective delivery of spine SBRT is dependent on careful patient selection in a multidisciplinary setting with collaboration among radiation oncologists, spine surgeons, and medical oncologists. The Neurologic Oncologic Mechanical and Systemic (NOMS) framework outlines the essential considerations in this decision. Table 90.1 summarizes the ideal criteria and contraindications for spine SBRT.

During the initial assessment, key features to elicit on history include the patient’s performance status, comorbidities, disease burden, and a thorough characterization of lesion-specific pain if present. Patients with an Eastern Cooperative Oncology Group (ECOG) performance score ≥3 are generally unsuitable for SBRT spine. Exceptions would include patients who are considered EGOG 3 solely based on neurologic deficits from their spinal disease. Other guidelines have recommended considering cEBRT or best supportive care if the Karnofsky Performance Score (KPS) is below 40 or if the patient’s life expectancy is less than 2 months. The combination of these factors into a recursive partitioning analysis (RPA) can make for a better selection of patients ; however, given the enormous resources required for spine SBRT, more refined predictive algorithms are needed to determine optimal patient selection.

For spine metastases, differentiating between biologic and mechanical back pain is critical for patient selection. Biologic back pain is generally worse in the morning and evening, can wake the patient up at night, occurs irrespective of position, and improves with steroid administration. This pain generally responds well to radiotherapy. Mechanical pain is worse with movement or loading of the spine (e.g., the pain is worse during bumpy car rides or when the patient is lifting an object from the ground); it improves with rest but not with steroids. Such pain suggests structural spinal instability, which may not respond well to radiotherapy alone.

Characterizing a patient’s level of neurologic deficit is also important. The American Spinal Injury Association (ASIA) impairment scale is a validated standardized tool that can be used to describe deficits due to tumor as well. The scale ranges from grade E (normal) to grade A (complete loss of sensory and motor function) and has been shown to be strongly predictive of functional outcomes. Patients with grade A deficits have a 92% negative predictive probability for independent ambulation, whereas those with grade D deficits have a 97% positive predictive probability of regaining independent ambulation.

Radiographic evaluation should be completed using a whole-spine MRI with particular focus on assessing mechanical stability and the presence and degree of spinal cord compression. Furthermore, to understand the disease burden in the spine, a full spine MRI is required as a baseline assessment.

According to the Spine Oncology Study Group (SOSG), spinal instability is the “the loss of spinal integrity as a result of a neoplastic process that is associated with movement-related pain, symptomatic or progressive deformity, and/or neural compromise under physiologic loads.” The Spinal Instability Neoplastic Score (SINS) was developed by the SOSG to standardize and simplify the evaluation of spinal stability. The six components of the SINS score consist of the following: spine location, mechanical pain, bone lesion quality, spinal alignment, vertebral body collapse, and posterolateral involvement of spinal elements ( Table 90.2 ). Scores between 0 to 6, 7 to 12, and 13 to 18 indicate stability, indeterminate, and instability, respectively. Patients with a score of 7 to 18 should be referred for consideration of surgical stabilization or percutaneous cement augmentation. ^, The score has been externally validated and has a specificity and sensitivity of 80% and 95%, respectively.

The six-point Epidural Spinal Cord Compression (ESCC) scale developed by Bilsky et al. provides a consistent cross-disciplinary metric for describing the severity of spinal cord compression. Grade 0 ESCC indicates bone-only disease. Grade 1 ESCC is further divided into three subgroups: 1a, which denotes impingement of the thecal sac but no deformation; 1b, which is deformation of the thecal sac alone; and 1c, or deformation of the thecal sac with spinal cord abutment (but no compression). Grade 2 ESCC is defined as spinal cord compression with partial obliteration of the cerebrospinal fluid (CSF) space around the cord at the site of compression. Grade 3 ESCC is spinal cord compression with complete obliteration of the CSF space. Patients with grades 2 and 3 ESCC are considered to have high-grade epidural disease and should undergo surgical decompression before SBRT, which is discussed in further detail later.

Stereotactic Body Radiotherapy for De Novo Spine Metastases

The dose regimen selected for spine SBRT varies considerably in the published literature; it typically ranges from one to five fractions. A recent guideline from the ISRS reported by Hussain et al. specific to de novo spine SBRT summarizes the literature and outcomes; it is noted that the bulk of the literature consists of retrospective reviews and a few prospective observational single-arm studies. Ultimately the crude local control rate at 1 year was 78.6% (range, 73% to 88%). However, since that review, a randomized phase II trial from Germany has been reported where 55 patients were randomized to receive 24 Gy in a single fraction of SBRT versus 30 Gy in 10 fractions of cEBRT. A trend was observed with respect to complete response (CR) rates for pain at 3 months favoring SBRT over cEBRT (43.5% vs. 17.4%, P = .057, respectively). At 6 months a significant difference was observed in the SBRT arm, with a CR rate of 52.6% versus 10% in the cEBRT arm ( P = .003). The rate of vertebral compression fracture (VCF) was 28% in the SBRT arm and 5% in the cEBRT arm at 6 months. This trial was not powered a priori and remains hypotheses generating; it is encouraging nonetheless. Larger phase III randomized trials are currently in progress and will ultimately inform the community of the advantages of spine SBRT in comparison with cEBRT.

At our institution, practice transitioned from single-fraction spine SBRT to delivering 24 Gy in two fractions. This was in response to the significant VCF rates seen with single-fraction SBRT, which have been as high as 40%. We recently published our experience in treating 279 de novo spinal metastases with 24 Gy in two fractions and demonstrated 1- and 2-year cumulative VCF rates of 8.5% and 13.8%, respectively. Local failure rates were favorable: 9.7% at 1 year and 17.6% at 2 years. Overall survival rates were 73.1% and 60.7% at 1 and 2 years, respectively. A multi-institutional phase III randomized trial (SC.24) comparing SBRT (24 Gy in two fractions) with cEBRT (20Gy in five fractions) is under way, and its results are highly anticipated ( ClinicalTrials.gov identifier: NCT02512965). It is important to note that at present the optimal dose fractionation has yet to be compared in a randomized trial specific to spinal metastases.

Postoperative Stereotactic Body Radiotherapy

The predominant indications for surgery in the metastatic spine consist of high-grade symptomatic ESCC (ESCC grade 2 or 3) and/or the presence of mechanical instability.

With respect to ESCC, Patchell et al. compared decompressive surgery followed by cEBRT with cEBRT alone for symptomatic patients with ESCC confined to one spinal region (several contiguous vertebral levels were eligible as well). They reported improved overall survival, enhanced durability of ambulation, and decreased reliance on corticosteroids and opioid analgesics with the combination of surgery and radiotherapy. Furthermore, a prospective multi-institutional study showed improved quality-of-life outcomes following surgery for patients with ESCC.

With respect to instability, patients with impending or frank instability are typically stabilized either with an open procedure for placement of surgical instrumentation or a minimally invasive procedure that stabilizes the vertebral body through cement injection and/or percutaneous hardware placement. As discussed previously, the SINS classification system is designed to identify patients with mechanical instability, and it continues to be adopted in the oncologic community and in clinical trials for appropriate patient selection. There are no randomized trials specific to unstable patients comparing radiotherapy to surgery. However, there are nonrandomized comparative data that do support better pain and quality-of-life outcomes in patients initially stabilized in the presence of mechanical pain and VCF versus the use of radiotherapy alone (data under submission, personal communication, A. Sahgal).

The question that has arisen as a result of SBRT relates to postoperative radiotherapy management. Traditionally these patients would have been treated with 30 Gy in 10 fractions of cEBRT. When patients have undergone a significant operation (e.g., surgical stabilization and tumor debulking), there is compelling motivation to consider postoperative SBRT, which theoretically maximizes tumor control, particularly if there is residual tumor. Several early adopters began programs for postoperative SBRT and now have limited data to understand the outcomes. A study from Al-Omair et al. reported retrospective postoperative SBRT outcomes from the University of Toronto. High rates of local control (of 81%) were observed at 1 year, which is consistent with the data in general, including a recent prospective study from M.D. Anderson Cancer Center. Of interest in the Al-Omair analyses was the impact of postoperative epidural disease. When grade 2 or 3 epidural disease was downgraded to a 0 or 1 by aggressive surgical debulking, outcomes were better than if disease remained at Bilsky 2. These data validate increasing attention to the management of epidural disease, and groups such as those at Memorial Sloan Kettering have been advocating for separation surgery to maximize spine SBRT outcomes. From that group, the largest postoperative SBRT series was reported by Laufer et al. following separation surgery; it reports a cumulative 1-year local control rate of 83.6% and a median time to local progression of 4.8 months (range 0.2 to 38.3 months). Ultimately the decision for surgery is major, and we have to limit morbidity and delays in the overall oncologic care that can arise from open invasive surgery. Therefore we have developed a minimally invasive technique for separation surgery known as minimal access spine surgery (MASS), where decompression occurs percutaneously through a tube-based retractor system and stabilization with cement. Our initial experience has been reported.

In response to the paucity of evidence, an international consensus study was recently published; it surveyed 15 radiation oncologists and 5 neurosurgeons with expertise in postoperative spine SBRT. There was agreement that indications for postoperative spine SBRT included (1) radioresistant primary tumor histology, (2) disease confined to one or two vertebral levels, and (3) cases with overlapping or adjacent volumes with previous cEBRT. Contraindications to postoperative spine SBRT included (1) residual disease after surgery causing ≥ ESCC grade 2 spinal cord compression, (2) disease involving more than three vertebral levels, and (3) complete spinal cord injury (ASIA grade A). In these cases, if treatment is indicated, postoperative cEBRT should be favored over SBRT.

Special considerations are required in planning postoperative spine SBRT in the presence of epidural disease. As described by Chan et al., patterns of epidural disease on both pre- and postoperative MRI images must be examined carefully, as they influence the patterns of epidural recurrence and should guide CTV delineation. If epidural disease is present only in the anterior sector on both pre- and postoperative MRIs, a horseshoe-shaped CTV (sparing the posterior sector) is appropriate. When there is epidural disease in both the anterior and posterior sectors, a donut-shaped distribution (no epidural sparing) is recommended ( Fig. 90.2 ).