- Email: [email protected]
Marginal Recurrence Requiring Salvage Radiotherapy After Stereotactic Body Radiotherapy for Spinal Metastases
International Journal of
Radiation Oncology biology
physics
www.redjournal.org
Clinical Investigation: Metastases
Marginal Recurrence Requiring Salvage Radiotherapy After Stereotactic Body Radiotherapy for Spinal Metastases Shlomo A. Koyfman, M.D.,* Toufik Djemil, Ph.D.,* Michael J. Burdick, M.D.,y Neil Woody, B.Sc.,* Ehsan H. Balagamwala, B.A.,* Chandana A. Reddy, M.S.,* Lilyana Angelov, M.D.,z John H. Suh, M.D.,* and Samuel T. Chao, M.D.* *Department of Radiation Oncology, Cleveland Clinic, Cleveland, Ohio; yDepartment of Radiation Oncology, Tulane Medical Center, New Orleans, Louisiana; and zDepartment of Neurosurgery, Cleveland Clinic, Cleveland, Ohio Received Mar 4, 2011, and in revised form May 20, 2011. Accepted for publication May 26, 2011
Summary The adoption of small or no treatment margins characteristic of spine SBRT for bony metastases raises concern for a risk of marginal recurrence. In this study of over 200 treated patients MR occurred in 26 (12.5%) treated lesions, at a median time of 7.7 months after SBRT. Patients with paraspinal disease at the time of SBRT and those treated with < 16 Gy had higher rates of MR.
Introduction: We sought to quantify and identify risk factors associated with margin recurrence (MR) requiring salvage radiotherapy after stereotactic body radiation therapy (SBRT) for spinal metastases. Methods: We retrospectively reviewed patients with spinal metastases who were treated with single-fraction SBRT between 2006 and 2009. Gross tumor was contoured, along with either the entire associated vertebral body(ies) or the posterior elements, and included in the planning target volume. No additional margins were used. MR was defined as recurrent tumor within one vertebral level above or below the treated lesion that required salvage radiotherapy. Only patients who presented for 3-month post-SBRT follow-up were included in the analysis. Fine and Gray competing risk regression models were generated to identify variables associated with higher risks of MR. MR was plotted using cumulative incidence analysis. Results: SBRT was delivered to 208 lesions in 149 patients. Median follow-up was 8.6 months, and median survival was 12.8 months. The median prescribed dose was 14 Gy (10 e16 Gy). MR occurred in 26 (12.5%) treated lesions, at a median time of 7.7 months after SBRT. Patients with paraspinal disease at the time of SBRT (20.8% vs. 7.6% of patients; p Z 0.02), and those treated with <16 Gy (16.3% vs. 6.3% of patients, p Z 0.14) had higher rates of MR. Both variables were associated with significantly higher risk of MR on multivariate analysis. Conclusion: SBRT for spinal metastases results in a low overall rate of MR. The presence of paraspinal disease at the time of SBRT and a dose of <16 Gy were associated with higher risks of MR. Ó 2012 Elsevier Inc. Keywords: Marginal recurrence, Spinal metastases, Stereotactic body radiotherapy
Reprint requests to: Samuel T. Chao, M.D., Department of Radiation Oncology, Desk T28, Taussig Cancer Institute, Cleveland Clinic, Cleveland, OH 44195; Tel: (216) 445-7876; Fax; (216) 445-1068; E-mail: [email protected] Int J Radiation Oncol Biol Phys, Vol. 83, No. 1, pp. 297e302, 2012 0360-3016/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.ijrobp.2011.05.067
Presented in part at the 52nd Annual Meeting of the American Society of Radiation Oncology, San Diego, CA, Oct. 31-Nov. 4, 2010. Conflict of interest: none.
298
Koyfman et al.
Introduction Stereotactic body radiation therapy (SBRT), which uses highly conformal techniques to deliver large doses of radiation in one to five fractions, is being increasingly used in the treatment of spinal metastases (1, 2). Spine SBRT has transformed from an invasive procedure in its infancy to a noninvasive procedure due to a variety of image guidance systems, including infrared cameras with external fiducials, stereoscopic imaging, cone beam-computed tomography (CT), tomotherapy, and CT-on-rails (3e9). Several investigators have reported excellent local (i.e., in-field) control with this technique, with some reports of rates higher than 90% (1). However, departure from conventionally larger radiation fields and adoption of small or no treatment margins characteristic of spine SBRT have generated concern for the possibility of increased risk of marginal recurrence (MR) that may not be reflected in local control outcomes. This study sought to specifically quantify the risk of MR after treatment with SBRT and identify variables that are associated with an increase in that risk.
Methods and Materials This institutional review board-approved study included patients treated with single-fraction SBRT who had a histological diagnosis of cancer and radiographic evidence of spinal metastases. Histological confirmation of metastatic disease was not required. Patients with primary spinal tumors (e.g., sarcoma) were excluded. Indications for SBRT included symptoms, such as pain or minor neurological deficits, or epidural extension of tumor. Patients presenting with rapidly progressive neurological compromise were not treated with SBRT. The presence of paraspinal disease, defined as any disease in the soft tissues outside of the vertebral body or bony posterior elements, was not a contraindication to treatment. SBRT was used as both primary treatment and as salvage treatment after conventional radiotherapy. Only patients with 3 months of post-SBRT follow-up were included in this study to ensure that radiographic and/or symptomatic progression was confirmed to be related to tumor progression rather than treatment-related edema. All patients underwent CT simulation in the supine position. Patients with lesions at T4 and above were immobilized using a 5-point thermoplastic head mask, while those with lesions below T4 were immobilized in a Bodyfix mold (Medical Intelligence; Elekta, Stockholm, Sweden) with a vacuum bag for additional immobilization. The simulation CT and high-definition magnetic resonance images (MRI) were obtained in 1.5-mm slices. Axial T1-weighted, T2-weighted, and short T1-weighted inversion recovery (STIR) MRI sequences were then fused to the treatment planning CT to aid with target and spinal cord delineation. A clinical target volume was contoured that included the entire vertebral body, for lesions involving the vertebral body, or all of the posterior elements, for lesions involving the lamina, pedicles, or spinous process. No additional margins were added to the clinical target volume, which was identical to the planning target volume (PTV). Based upon the suggestion by Ryu et al. (10), two slices (3e4 mm) of the spinal cord was contoured above and below the PTV (10). Treatment planning was initially performed using BrainScan (Brainlab, Munich, Germany) and, more recently, iPlan (Brainlab). Image guidance was accomplished using the Exactrac system before treatment delivery (7). No more than 10%
International Journal of Radiation Oncology Biology Physics of the contoured spinal cord or thecal sac organs at risk was allowed to receive more than 10 Gy or 12 Gy, respectively (10). The conformality index (prescription isodose volume/target volume) and heterogeneity index (maximum dose/prescribed dose) were calculated for each patient. As data for the safety and efficacy of spine SBRT emerged, the prescription dose for primary SBRT was gradually escalated to 16 Gy, our current standard. All other dosimetric parameters and planning techniques have remained consistent over time. Patients were seen 1 to 2 months after treatment and then followed every 3 months with a clinical examination and spinal MRI. Margin recurrence was defined as recurrent tumor within one vertebral level above or below the treated lesion, with some component outside the original PTV, which required salvage radiotherapy (Fig. 1). To confirm MR, each treatment plan was retrospectively reviewed to verify that at least part of the re-treated area was not included in the original PTV. Local progression of disease that was entirely within the PTV and marginal failures that did not result in reirradiation were not scored as MR. Salvage radiotherapy was delivered with further SBRT or conventional radiotherapy, as clinically indicated. Indications for salvage radiotherapy included radiographic progression along with worsening symptoms, epidural disease, or both. When marginal recurrences were salvaged with a second SBRT treatment to the area adjacent to the original PTV, the second SBRT treatment was not assessed for whether or not it resulted in MR and was not included in our analysis. Univariate analysis and multivariate analysis were performed using competing risk regression with death as the competing event to identify whether the following variables were associated with MR: SBRT dose, treatment volume, maximum tumor diameter, renal cell histology, and presence of paraspinal disease. Rates for MR were calculated using cumulative incidence analysis.
Results A total of 208 lesions treated in 149 patients were included in this study. Patient and disease characteristics are described in Table 1, and treatment characteristics are detailed in Table 2. The median age of the cohort was 58.5 years old (range, 8e90), median Karnofsky performance statuswas 80 (range, 40e100), median follow-up was 8.6 months (range, 2e48), and median survival was 12.8 months. At the time of SBRT, paraspinal disease was present 37% of the time. The median prescribed dose was 14 Gy (range, 10e16 Gy). Seventy-nine lesions (38%) were treated with 16 Gy. SBRT was used as primary therapy in 71% of cases, while 29% of patients were salvage treatment cases after conventional radiotherapy. Margin recurrence occurred in 26 (12.5%) treated lesions at a median time-to-salvage treatment of 7.7 months. The 6-, 12-, and 18-month cumulative incidence rates for MR were 3.9%, 9.5%, and 14.5%, respectively (Fig. 2). These rates were significantly higher when paraspinal disease was present at the time of SBRT (Fig. 3; p Z 0.02). This finding was consistent at 6 (6.7% vs. 2.3%), 12 (17% vs. 5.1%), and 18 month follow-up (22.1% vs. 7.6%). The pattern of MR for lesions that involved the paraspinal musculature varied. While half of these lesions progressed in the adjacent paraspinal soft tissues, the other half progressed in the adjacent vertebral body or epidural space without evidence of soft tissue progression. The risk of MR was also nonsignificantly higher when the prescription dose was less than 16 Gy than when
Volume 83 Number 1 2012
Salvage RT after SBRT for spinal metastases
299
Fig. 1. (aed) Representative axial (a) and sagittal (b) CT planning images of a 36-year-old female patient with newly diagnosed lung cancer that had metastasized to L4, with epidural and paraspinal extension, treated with SBRT (16 Gy) are shown. The patient later developed metastasis with associated partial vertebral body collapse at L3, which was also treated with SBRT (14 Gy). Axial (c) and sagittal (d) images of this treatment are shown. The PTV is delineated in red, the 16 Gy isodose line is in blue, the 14 Gy isodose line is in green, and the 10 Gy isodose line is in purple. The cauda equina is contoured in gold. the prescribed dose was 16 Gy (16.3% vs. 6.3%, respectively, p Z 0.14) (Fig. 4). Univariate analysis revealed that the presence of paraspinal disease, larger maximum tumor diameter, and a dose less than 16 Gy were all associated with increased risk of MR (Table 3). On multivariate analysis, only the presence of paraspinal disease (hazard ratio [HR], 2.82; p Z 0.02) and an SBRT dose less than 16 Gy retained significance (HR, 2.69; p Z 0.05). Renal cell histology, salvage SBRT after conventional Table 1
Patient and disease characteristics
Variable No. of patients/lesions treated Median age (years) Median Karnofsky performance status Median follow-up (months) % of patients with histology Renal cell cancer Non-small-cell lung cancer Breast cancer Other % of patients with paraspinal disease at the time of SBRT
No. of patients or percentage of patients 149/208 58.5 80 8.6 25 17 10 48 37
radiotherapy, and disease volume were not associated with an increased risk of MR.
Discussion SBRT has emerged as a safe and efficacious treatment for spinal metastases, and its use is increasing. Despite the use of highly conformal techniques without the addition of planning margins, our results demonstrate that overall, the risk of MR requiring salvage radiotherapy is reasonably low. We defined MR as disease progression within one vertebral level above or below the treated area that required salvage radiotherapy. We chose one vertebral level above or below the treated area to try and capture the margin that would have been included in a conventional radiotherapy field. By limiting MR to those patients who required salvage radiotherapy for either symptomatic progression or new epidural disease, we sought to select a clinically meaningful endpoint that captured the physical, psychological, and financial costs to the patient that are incurred by the need for retreatment and that allowed us to refine our ability to more accurately counsel patients about the risks and benefits of spine SBRT. That endpoint may also serve as a benchmark for future studies aimed at comparing the risks of MR after conventional radiotherapy versus the risk of MR after SBRT. While we did not perform such a comparative study, numerous Phase III randomized trials that have compared various dose fractionation
300
International Journal of Radiation Oncology Biology Physics
Koyfman et al.
Table 2
Treatment characteristics Variable
No. or percentage
Median dose (Gy) Median V100 Median conformality index Median heterogeneity index % of patients receiving SBRT as the initial treatment
14 (range, 10e16) 94.8 1.37 (range, 1.0e4.0) 1.13 (range, 1.01e2.55) 71
Abbreviation: V100 Z volume receiving at least 100% of the prescription dose.
Fig. 3. Cumulative incidence of MR following spine SBRT by the presence of paraspinal disease is shown. regimens have reported retreatment rates. In Radiation Therapy Oncology Group (RTOG) protocol 97e14, at 3 years, only 9% of patients treated with 30 Gy in 10 fractions required reirradiation compared to 18% of patients who required retreatment after being treated with a single fraction of 8 Gy (p < 0.001) (11). While some studies have suggested that this difference is more reflective of physician bias than true treatment failures and that these data represent local failures more than margin failures, a crude comparison to our cohort demonstrates that SBRT compares favorably to conventional single-fraction radiotherapy, despite the significant proportion of radioresistant histologies in our cohort. The introduction of highly conformal radiotherapy is often accompanied by the apprehension that reducing treatment margins will lead to an increase in margin failures. Regarding spinal SBRT, these early concerns were seemingly allayed by several publications that described MR as a rare phenomenon (Table 4). While some investigators have reported no margin failures, others have found that while MR does happen, it appears to be unusual (5, 12). In an early report reviewing their patterns of failure after treating 61 lesions in 49 patients with spine SBRT, Ryu et al. (13) noted that 3 patients (4%) had recurrence immediately outside the PTV. Interestingly, all 3 patients harbored associated paraspinal masses that had progressed at the time of margin failure. Gerszten et al. (14) published outcomes for 48 patients with renal cell cancer treated with spine SBRT and reported that therapy failed for 6 patients who had recurrences at the margins of the SBRT fields. As a result of that phenomenon, the authors subsequently “increased the PTV to include more of the adjacent normalappearing vertebral body” (14). Chang et al. (4) identified the possibility that the two primary mechanisms of failure in their report of 63 patients treated with spine SBRT were either at the osseous margin posterior to the site of vertebral body treatment or in the epidural space, which is often underdosed in order to meet
spinal cord planning constraints. While that study does not explicitly discuss marginal recurrence, the authors do describe several cases in which in- and out-of-field failure were associated with large pre- or paravertebral soft tissue masses. It is also important to note that those studies define margin failure radiographically; unlike our study, those reports did not require retreatment to be considered MR. Our results demonstrate that while MR is unusual, patients with paraspinal disease appear to be at significantly increased risk for MR. This is a novel finding. While the presence of paraspinal disease may simply be a proxy for increased tumor bulk, which is generally associated with greater radioresistance, a different physiologic explanation for this observation appears plausible. While intervertebral discs are thought to be an effective anatomic barrier to tumor spread, once a tumor has broken through the bony cortex and accessed the paraspinal soft tissues, there are no longer anatomic boundaries to tumor spread. For this reason, radiotherapy for tumors that involve the soft tissues, such as head and neck cancers and sarcomas, routinely incorporates additional clinical target volume margins to account for potential microscopic spread that is not appreciated radiographically. Additionally, while SBRT often provides significant “spillover” of reasonably high doses laterally into the paraspinal tissues, the use of coplanar beams ensures that very little dose is deposited outside the PTV in the superior and inferior directions, where microscopic paraspinal disease may be present. While less statistically conclusive, it also appears that a dose of <16 Gy may be associated with an increased risk MR. While it can be argued that this finding may be more reflective of an SBRT learning curve, with our most recent cohort of patients having been treated with higher doses, the fact that our methods of target
Fig. 2. SBRT.
Fig. 4. Cumulative incidence of MR after SBRT by treatment dose is shown.
Cumulative incidence is shown for MR following spine
Volume 83 Number 1 2012
Salvage RT after SBRT for spinal metastases
Table 3 Competing risk regression analysis for factors associated with the risk of MR Variable Univariate analysis Paraspinal disease Renal cell histology Maximum tumor diameter Treatment volume Dose (<16 Gy vs. 16 Gy) SBRT dose SBRT after conventional radiotherapy Multivariate analysis Paraspinal disease Maximum tumor diameter Dose (<16 Gy vs. 16 Gy)
Hazard ratio (95% CI)
p value 0.01 0.39 0.04 0.34 0.06 0.09 0.50
2.90 1.44 1.19 1.00 2.54 0.88 1.31
(1.33e6.34) (0.63e3.32) (1.00e1.40) (1.00e1.01) (0.96e6.73) (0.76e1.02) (0.60e2.89)
0.02 0.50 0.05
2.82 (1.16e6.84) 1.07 (0.88e1.31) 2.69 (1.00e7.22)
Abbreviation: CI Z confidence interval.
delineation and treatment planning have been extremely consistent throughout the study period lends credence to the validity of this observation. Other recent studies have also suggested a dose response phenomenon in these patients (15). While there are various dose and fractionation schedules currently employed in spinal SBRT, we have adopted 16 Gy in a single fraction as our standard, similar to that of the open RTOG 0631 study (16). Whether there is any associated benefit to further dose escalation remains unresolved. There are several limitations to this study. As a retrospective review of a single-institution experience, this study is subject to selection and physician biases. However, because it includes a large cohort of patients with a variety of different tumor types, who were
Table 4
301
treated in the primary and reirradiation settings, this study is wide ranging and captures the heterogeneity of patients and diseases that are treated with spine SBRT, thereby providing a global assessment of MR. Also, with the exception of the prescription dose, all other technique-related variables remained consistent throughout the study period. Another potential shortcoming of this study is that it may underestimate the true incidence of radiographic MR, as salvage radiotherapy was required to meet our MR definition. However, this definition allowed us to identify those recurrences that most, if not all, investigators would interpret as clinically meaningful, rather than include radiographic-only failures that might never have had an impact on a patient’s quality of life or survival. Finally, our finding that paraspinal disease is associated with higher MR rates may not apply to other SBRT techniques that incorporate marginal expansions. The strengths of this study lie in its large size, consistent treatment approach, and systematic method of characterizing a clinically meaningful yet often underreported endpoint. Moreover, we describe a novel finding: the presence of paraspinal disease at the time of SBRT is associated with a significant increase in the risk of MR and the resultant physical, psychological, and financial costs to the patient that are associated with disease progression and retreatment. Based on these results, we believe that expanded treatment margins may be prudent in these patients. For patients with paraspinal disease not being treated with conventional radiotherapy, we currently employ a nonuniform expansion of 3 mm around the soft tissue component of the target volume, without any associated expansion adjacent to the spinal cord. By doing so, we are increasing our dosimetric coverage of potential microscopic spread of soft tissue disease without compromising our ability to meet our spinal cord constraints. Finally, our data suggest that SBRT doses of at least 16 Gy are preferred and appear to be associated with less MR.
Target volume delineation used for spine SBRT and patterns of failure
Study, center (ref[s].) Current Study, Cleveland Clinic Chang et al., MDACC (4) Gerszten et al. Pittsburgh, PA (5) Gerszten, et al. Pittsburgh, PA (14) Nelson et al., Duke Univ. (12)
No. of treatments
Target volume definition
208
All visible tumor and associated subsitesz All visible tumor and associated subsitesz All visible tumor All visible tumor All visible tumor and associated subsitesz All visible tumor and associated subsitesz All visible tumor and associated subsitesz
74 500 48 21
Ryu et al., Henry Ford (13)
61
Yamada et al., MSKCC (17)
103
Additional planning margin (mm)*
Median prescribed dose (Gy)
Median no.of fractions
Incidence (%) of margin failurey
0
14
1
12.6
2
27
3
“Several”
0 0 6
20 20 21
1 1 3
NR 10 0
0
16
1
4
2e3
24
1
NR
Abbreviations: NR Z not reported; MSKCC Z Memorial Sloan Kettering Cancer Center; MDACC Z M.D. Anderson Cancer Center. * This expansion was not done in the area immediately adjacent to the spinal cord. y The denominator for this value is the total number of treatments included in the cited study, not the total number of patients. This value also represents the crude rate, rather than the actuarial rate. z Associated subsites refers to the inclusion of additional areas at risk for microscopic extension of disease. Similar to the current target delineation procedure on RTOG 0631, the entire vertebral body would be included for anterior lesions, the entire posterior elements for lesions involving the pedicles, or spinous processes, and at times, some combination would be used based on the location of gross disease, with care not to encircle the spinal cord when possible.
302
Koyfman et al.
Conclusions Highly conformal SBRT for spinal metastases results in a low overall rate of MR that compares favorably to that resulting from treatment with conventional radiotherapy. The presence of paraspinal disease at the time of SBRT and an SBRT dose of less than 16 Gy were associated with significantly higher risks of MR. In addition to the routine use of 16 Gy, the use of nonuniform treatment margin expansions should be considered in patients with paraspinal disease.
References 1. Sahgal A, Larson DA, Chang EL. Stereotactic body radiosurgery for spinal metastases: A critical review. Int J Radiat Oncol Biol Phys 2008;71:652e665. 2. Potters L, Kavanagh B, Galvin JM, et al. American Society for Therapeutic Radiology and Oncology (ASTRO) and American College of Radiology (ACR) practice guideline for the performance of stereotactic body radiation therapy. Int J Radiat Oncol Biol Phys 2010;76:326e332. 3. Bilsky MH, Yamada Y, Yenice KM, et al. Intensity-modulated stereotactic radiotherapy of paraspinal tumors: A preliminary report. Neurosurgery 2004;54:823e831. 4. Chang EL, Shiu AS, Mendel E, et al. Phase I/II study of stereotactic body radiotherapy for spinal metastasis and its pattern of failure. J Neurosurg Spine 2007;7:151e160. 5. Gerszten PC, Burton SA, Ozhasoglu C, Welch WC. Radiosurgery for spinal metastases: Clinical experience in 500 cases from a single institution. Spine 2007;32:193e199. 6. Hamilton AJ, Lulu BA. A prototype device for linear acceleratorbased extracranial radiosurgery. Acta Neurochir Suppl 1995;63: 40e43.
International Journal of Radiation Oncology Biology Physics 7. Ryu S, Fang Yin F, Rock J, et al. Image-guided and intensitymodulated radiosurgery for patients with spinal metastasis. Cancer 2003;97:2013e2018. 8. Sheehan JP, Shaffrey CI, Schlesinger D, Williams BJ, Arlet V, Larner J. Radiosurgery in the treatment of spinal metastases: Tumor control, survival, and quality of life after helical tomotherapy. Neurosurgery 2009;65:1052e1062. 9. Wang JZ, Rice R, Mundt A, Sandhu A, Murphy K. Image-guided stereotactic spine radiosurgery on a conventional linear accelerator. Med Dosim 2010;35:53e62. 10. Ryu S, Jin JY, Jin R, et al. Partial volume tolerance of the spinal cord and complications of single-dose radiosurgery. Cancer 2007;109: 628e636. 11. Hartsell WF, Scott CB, Bruner DW, et al. Randomized trial of shortversus long-course radiotherapy for palliation of painful bone metastases. J Natl Cancer Inst 2005;97:798e804. 12. Nelson JW, Yoo DS, Sampson JH, et al. Stereotactic body radiotherapy for lesions of the spine and paraspinal regions. Int J Radiat Oncol Biol Phys 2009;73:1369e1375. 13. Ryu S, Rock J, Rosenblum M, Kim JH. Patterns of failure after single-dose radiosurgery for spinal metastasis. J Neurosurg 2004; 101(Suppl 3):402e405. 14. Gerszten PC, Burton SA, Ozhasoglu C, et al. Stereotactic radiosurgery for spinal metastases from renal cell carcinoma. J Neurosurg Spine 2005;3:288e295. 15. Damast S, Wright J, Bilsky M, et al. Impact of dose on local failure rates after image-guided reirradiation of recurrent paraspinal metastases. Int J Radiat Oncol Biol Phys 2010. 16. RTOG 0631 protocol. http://www.rtog.org/members/protocols/0631/ 0631.pdf. Accessed December 12, 2010. 17. Yamada Y, Bilsky MH, Lovelock DM, et al. High-dose, single-fraction image-guided intensity-modulated radiotherapy for metastatic spinal lesions. Int J Radiat Oncol Biol Phys 2008;71:484e490.
Radiation Oncology biology
physics
www.redjournal.org
Clinical Investigation: Metastases
Marginal Recurrence Requiring Salvage Radiotherapy After Stereotactic Body Radiotherapy for Spinal Metastases Shlomo A. Koyfman, M.D.,* Toufik Djemil, Ph.D.,* Michael J. Burdick, M.D.,y Neil Woody, B.Sc.,* Ehsan H. Balagamwala, B.A.,* Chandana A. Reddy, M.S.,* Lilyana Angelov, M.D.,z John H. Suh, M.D.,* and Samuel T. Chao, M.D.* *Department of Radiation Oncology, Cleveland Clinic, Cleveland, Ohio; yDepartment of Radiation Oncology, Tulane Medical Center, New Orleans, Louisiana; and zDepartment of Neurosurgery, Cleveland Clinic, Cleveland, Ohio Received Mar 4, 2011, and in revised form May 20, 2011. Accepted for publication May 26, 2011
Summary The adoption of small or no treatment margins characteristic of spine SBRT for bony metastases raises concern for a risk of marginal recurrence. In this study of over 200 treated patients MR occurred in 26 (12.5%) treated lesions, at a median time of 7.7 months after SBRT. Patients with paraspinal disease at the time of SBRT and those treated with < 16 Gy had higher rates of MR.
Introduction: We sought to quantify and identify risk factors associated with margin recurrence (MR) requiring salvage radiotherapy after stereotactic body radiation therapy (SBRT) for spinal metastases. Methods: We retrospectively reviewed patients with spinal metastases who were treated with single-fraction SBRT between 2006 and 2009. Gross tumor was contoured, along with either the entire associated vertebral body(ies) or the posterior elements, and included in the planning target volume. No additional margins were used. MR was defined as recurrent tumor within one vertebral level above or below the treated lesion that required salvage radiotherapy. Only patients who presented for 3-month post-SBRT follow-up were included in the analysis. Fine and Gray competing risk regression models were generated to identify variables associated with higher risks of MR. MR was plotted using cumulative incidence analysis. Results: SBRT was delivered to 208 lesions in 149 patients. Median follow-up was 8.6 months, and median survival was 12.8 months. The median prescribed dose was 14 Gy (10 e16 Gy). MR occurred in 26 (12.5%) treated lesions, at a median time of 7.7 months after SBRT. Patients with paraspinal disease at the time of SBRT (20.8% vs. 7.6% of patients; p Z 0.02), and those treated with <16 Gy (16.3% vs. 6.3% of patients, p Z 0.14) had higher rates of MR. Both variables were associated with significantly higher risk of MR on multivariate analysis. Conclusion: SBRT for spinal metastases results in a low overall rate of MR. The presence of paraspinal disease at the time of SBRT and a dose of <16 Gy were associated with higher risks of MR. Ó 2012 Elsevier Inc. Keywords: Marginal recurrence, Spinal metastases, Stereotactic body radiotherapy
Reprint requests to: Samuel T. Chao, M.D., Department of Radiation Oncology, Desk T28, Taussig Cancer Institute, Cleveland Clinic, Cleveland, OH 44195; Tel: (216) 445-7876; Fax; (216) 445-1068; E-mail: [email protected] Int J Radiation Oncol Biol Phys, Vol. 83, No. 1, pp. 297e302, 2012 0360-3016/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.ijrobp.2011.05.067
Presented in part at the 52nd Annual Meeting of the American Society of Radiation Oncology, San Diego, CA, Oct. 31-Nov. 4, 2010. Conflict of interest: none.
298
Koyfman et al.
Introduction Stereotactic body radiation therapy (SBRT), which uses highly conformal techniques to deliver large doses of radiation in one to five fractions, is being increasingly used in the treatment of spinal metastases (1, 2). Spine SBRT has transformed from an invasive procedure in its infancy to a noninvasive procedure due to a variety of image guidance systems, including infrared cameras with external fiducials, stereoscopic imaging, cone beam-computed tomography (CT), tomotherapy, and CT-on-rails (3e9). Several investigators have reported excellent local (i.e., in-field) control with this technique, with some reports of rates higher than 90% (1). However, departure from conventionally larger radiation fields and adoption of small or no treatment margins characteristic of spine SBRT have generated concern for the possibility of increased risk of marginal recurrence (MR) that may not be reflected in local control outcomes. This study sought to specifically quantify the risk of MR after treatment with SBRT and identify variables that are associated with an increase in that risk.
Methods and Materials This institutional review board-approved study included patients treated with single-fraction SBRT who had a histological diagnosis of cancer and radiographic evidence of spinal metastases. Histological confirmation of metastatic disease was not required. Patients with primary spinal tumors (e.g., sarcoma) were excluded. Indications for SBRT included symptoms, such as pain or minor neurological deficits, or epidural extension of tumor. Patients presenting with rapidly progressive neurological compromise were not treated with SBRT. The presence of paraspinal disease, defined as any disease in the soft tissues outside of the vertebral body or bony posterior elements, was not a contraindication to treatment. SBRT was used as both primary treatment and as salvage treatment after conventional radiotherapy. Only patients with 3 months of post-SBRT follow-up were included in this study to ensure that radiographic and/or symptomatic progression was confirmed to be related to tumor progression rather than treatment-related edema. All patients underwent CT simulation in the supine position. Patients with lesions at T4 and above were immobilized using a 5-point thermoplastic head mask, while those with lesions below T4 were immobilized in a Bodyfix mold (Medical Intelligence; Elekta, Stockholm, Sweden) with a vacuum bag for additional immobilization. The simulation CT and high-definition magnetic resonance images (MRI) were obtained in 1.5-mm slices. Axial T1-weighted, T2-weighted, and short T1-weighted inversion recovery (STIR) MRI sequences were then fused to the treatment planning CT to aid with target and spinal cord delineation. A clinical target volume was contoured that included the entire vertebral body, for lesions involving the vertebral body, or all of the posterior elements, for lesions involving the lamina, pedicles, or spinous process. No additional margins were added to the clinical target volume, which was identical to the planning target volume (PTV). Based upon the suggestion by Ryu et al. (10), two slices (3e4 mm) of the spinal cord was contoured above and below the PTV (10). Treatment planning was initially performed using BrainScan (Brainlab, Munich, Germany) and, more recently, iPlan (Brainlab). Image guidance was accomplished using the Exactrac system before treatment delivery (7). No more than 10%
International Journal of Radiation Oncology Biology Physics of the contoured spinal cord or thecal sac organs at risk was allowed to receive more than 10 Gy or 12 Gy, respectively (10). The conformality index (prescription isodose volume/target volume) and heterogeneity index (maximum dose/prescribed dose) were calculated for each patient. As data for the safety and efficacy of spine SBRT emerged, the prescription dose for primary SBRT was gradually escalated to 16 Gy, our current standard. All other dosimetric parameters and planning techniques have remained consistent over time. Patients were seen 1 to 2 months after treatment and then followed every 3 months with a clinical examination and spinal MRI. Margin recurrence was defined as recurrent tumor within one vertebral level above or below the treated lesion, with some component outside the original PTV, which required salvage radiotherapy (Fig. 1). To confirm MR, each treatment plan was retrospectively reviewed to verify that at least part of the re-treated area was not included in the original PTV. Local progression of disease that was entirely within the PTV and marginal failures that did not result in reirradiation were not scored as MR. Salvage radiotherapy was delivered with further SBRT or conventional radiotherapy, as clinically indicated. Indications for salvage radiotherapy included radiographic progression along with worsening symptoms, epidural disease, or both. When marginal recurrences were salvaged with a second SBRT treatment to the area adjacent to the original PTV, the second SBRT treatment was not assessed for whether or not it resulted in MR and was not included in our analysis. Univariate analysis and multivariate analysis were performed using competing risk regression with death as the competing event to identify whether the following variables were associated with MR: SBRT dose, treatment volume, maximum tumor diameter, renal cell histology, and presence of paraspinal disease. Rates for MR were calculated using cumulative incidence analysis.
Results A total of 208 lesions treated in 149 patients were included in this study. Patient and disease characteristics are described in Table 1, and treatment characteristics are detailed in Table 2. The median age of the cohort was 58.5 years old (range, 8e90), median Karnofsky performance statuswas 80 (range, 40e100), median follow-up was 8.6 months (range, 2e48), and median survival was 12.8 months. At the time of SBRT, paraspinal disease was present 37% of the time. The median prescribed dose was 14 Gy (range, 10e16 Gy). Seventy-nine lesions (38%) were treated with 16 Gy. SBRT was used as primary therapy in 71% of cases, while 29% of patients were salvage treatment cases after conventional radiotherapy. Margin recurrence occurred in 26 (12.5%) treated lesions at a median time-to-salvage treatment of 7.7 months. The 6-, 12-, and 18-month cumulative incidence rates for MR were 3.9%, 9.5%, and 14.5%, respectively (Fig. 2). These rates were significantly higher when paraspinal disease was present at the time of SBRT (Fig. 3; p Z 0.02). This finding was consistent at 6 (6.7% vs. 2.3%), 12 (17% vs. 5.1%), and 18 month follow-up (22.1% vs. 7.6%). The pattern of MR for lesions that involved the paraspinal musculature varied. While half of these lesions progressed in the adjacent paraspinal soft tissues, the other half progressed in the adjacent vertebral body or epidural space without evidence of soft tissue progression. The risk of MR was also nonsignificantly higher when the prescription dose was less than 16 Gy than when
Volume 83 Number 1 2012
Salvage RT after SBRT for spinal metastases
299
Fig. 1. (aed) Representative axial (a) and sagittal (b) CT planning images of a 36-year-old female patient with newly diagnosed lung cancer that had metastasized to L4, with epidural and paraspinal extension, treated with SBRT (16 Gy) are shown. The patient later developed metastasis with associated partial vertebral body collapse at L3, which was also treated with SBRT (14 Gy). Axial (c) and sagittal (d) images of this treatment are shown. The PTV is delineated in red, the 16 Gy isodose line is in blue, the 14 Gy isodose line is in green, and the 10 Gy isodose line is in purple. The cauda equina is contoured in gold. the prescribed dose was 16 Gy (16.3% vs. 6.3%, respectively, p Z 0.14) (Fig. 4). Univariate analysis revealed that the presence of paraspinal disease, larger maximum tumor diameter, and a dose less than 16 Gy were all associated with increased risk of MR (Table 3). On multivariate analysis, only the presence of paraspinal disease (hazard ratio [HR], 2.82; p Z 0.02) and an SBRT dose less than 16 Gy retained significance (HR, 2.69; p Z 0.05). Renal cell histology, salvage SBRT after conventional Table 1
Patient and disease characteristics
Variable No. of patients/lesions treated Median age (years) Median Karnofsky performance status Median follow-up (months) % of patients with histology Renal cell cancer Non-small-cell lung cancer Breast cancer Other % of patients with paraspinal disease at the time of SBRT
No. of patients or percentage of patients 149/208 58.5 80 8.6 25 17 10 48 37
radiotherapy, and disease volume were not associated with an increased risk of MR.
Discussion SBRT has emerged as a safe and efficacious treatment for spinal metastases, and its use is increasing. Despite the use of highly conformal techniques without the addition of planning margins, our results demonstrate that overall, the risk of MR requiring salvage radiotherapy is reasonably low. We defined MR as disease progression within one vertebral level above or below the treated area that required salvage radiotherapy. We chose one vertebral level above or below the treated area to try and capture the margin that would have been included in a conventional radiotherapy field. By limiting MR to those patients who required salvage radiotherapy for either symptomatic progression or new epidural disease, we sought to select a clinically meaningful endpoint that captured the physical, psychological, and financial costs to the patient that are incurred by the need for retreatment and that allowed us to refine our ability to more accurately counsel patients about the risks and benefits of spine SBRT. That endpoint may also serve as a benchmark for future studies aimed at comparing the risks of MR after conventional radiotherapy versus the risk of MR after SBRT. While we did not perform such a comparative study, numerous Phase III randomized trials that have compared various dose fractionation
300
International Journal of Radiation Oncology Biology Physics
Koyfman et al.
Table 2
Treatment characteristics Variable
No. or percentage
Median dose (Gy) Median V100 Median conformality index Median heterogeneity index % of patients receiving SBRT as the initial treatment
14 (range, 10e16) 94.8 1.37 (range, 1.0e4.0) 1.13 (range, 1.01e2.55) 71
Abbreviation: V100 Z volume receiving at least 100% of the prescription dose.
Fig. 3. Cumulative incidence of MR following spine SBRT by the presence of paraspinal disease is shown. regimens have reported retreatment rates. In Radiation Therapy Oncology Group (RTOG) protocol 97e14, at 3 years, only 9% of patients treated with 30 Gy in 10 fractions required reirradiation compared to 18% of patients who required retreatment after being treated with a single fraction of 8 Gy (p < 0.001) (11). While some studies have suggested that this difference is more reflective of physician bias than true treatment failures and that these data represent local failures more than margin failures, a crude comparison to our cohort demonstrates that SBRT compares favorably to conventional single-fraction radiotherapy, despite the significant proportion of radioresistant histologies in our cohort. The introduction of highly conformal radiotherapy is often accompanied by the apprehension that reducing treatment margins will lead to an increase in margin failures. Regarding spinal SBRT, these early concerns were seemingly allayed by several publications that described MR as a rare phenomenon (Table 4). While some investigators have reported no margin failures, others have found that while MR does happen, it appears to be unusual (5, 12). In an early report reviewing their patterns of failure after treating 61 lesions in 49 patients with spine SBRT, Ryu et al. (13) noted that 3 patients (4%) had recurrence immediately outside the PTV. Interestingly, all 3 patients harbored associated paraspinal masses that had progressed at the time of margin failure. Gerszten et al. (14) published outcomes for 48 patients with renal cell cancer treated with spine SBRT and reported that therapy failed for 6 patients who had recurrences at the margins of the SBRT fields. As a result of that phenomenon, the authors subsequently “increased the PTV to include more of the adjacent normalappearing vertebral body” (14). Chang et al. (4) identified the possibility that the two primary mechanisms of failure in their report of 63 patients treated with spine SBRT were either at the osseous margin posterior to the site of vertebral body treatment or in the epidural space, which is often underdosed in order to meet
spinal cord planning constraints. While that study does not explicitly discuss marginal recurrence, the authors do describe several cases in which in- and out-of-field failure were associated with large pre- or paravertebral soft tissue masses. It is also important to note that those studies define margin failure radiographically; unlike our study, those reports did not require retreatment to be considered MR. Our results demonstrate that while MR is unusual, patients with paraspinal disease appear to be at significantly increased risk for MR. This is a novel finding. While the presence of paraspinal disease may simply be a proxy for increased tumor bulk, which is generally associated with greater radioresistance, a different physiologic explanation for this observation appears plausible. While intervertebral discs are thought to be an effective anatomic barrier to tumor spread, once a tumor has broken through the bony cortex and accessed the paraspinal soft tissues, there are no longer anatomic boundaries to tumor spread. For this reason, radiotherapy for tumors that involve the soft tissues, such as head and neck cancers and sarcomas, routinely incorporates additional clinical target volume margins to account for potential microscopic spread that is not appreciated radiographically. Additionally, while SBRT often provides significant “spillover” of reasonably high doses laterally into the paraspinal tissues, the use of coplanar beams ensures that very little dose is deposited outside the PTV in the superior and inferior directions, where microscopic paraspinal disease may be present. While less statistically conclusive, it also appears that a dose of <16 Gy may be associated with an increased risk MR. While it can be argued that this finding may be more reflective of an SBRT learning curve, with our most recent cohort of patients having been treated with higher doses, the fact that our methods of target
Fig. 2. SBRT.
Fig. 4. Cumulative incidence of MR after SBRT by treatment dose is shown.
Cumulative incidence is shown for MR following spine
Volume 83 Number 1 2012
Salvage RT after SBRT for spinal metastases
Table 3 Competing risk regression analysis for factors associated with the risk of MR Variable Univariate analysis Paraspinal disease Renal cell histology Maximum tumor diameter Treatment volume Dose (<16 Gy vs. 16 Gy) SBRT dose SBRT after conventional radiotherapy Multivariate analysis Paraspinal disease Maximum tumor diameter Dose (<16 Gy vs. 16 Gy)
Hazard ratio (95% CI)
p value 0.01 0.39 0.04 0.34 0.06 0.09 0.50
2.90 1.44 1.19 1.00 2.54 0.88 1.31
(1.33e6.34) (0.63e3.32) (1.00e1.40) (1.00e1.01) (0.96e6.73) (0.76e1.02) (0.60e2.89)
0.02 0.50 0.05
2.82 (1.16e6.84) 1.07 (0.88e1.31) 2.69 (1.00e7.22)
Abbreviation: CI Z confidence interval.
delineation and treatment planning have been extremely consistent throughout the study period lends credence to the validity of this observation. Other recent studies have also suggested a dose response phenomenon in these patients (15). While there are various dose and fractionation schedules currently employed in spinal SBRT, we have adopted 16 Gy in a single fraction as our standard, similar to that of the open RTOG 0631 study (16). Whether there is any associated benefit to further dose escalation remains unresolved. There are several limitations to this study. As a retrospective review of a single-institution experience, this study is subject to selection and physician biases. However, because it includes a large cohort of patients with a variety of different tumor types, who were
Table 4
301
treated in the primary and reirradiation settings, this study is wide ranging and captures the heterogeneity of patients and diseases that are treated with spine SBRT, thereby providing a global assessment of MR. Also, with the exception of the prescription dose, all other technique-related variables remained consistent throughout the study period. Another potential shortcoming of this study is that it may underestimate the true incidence of radiographic MR, as salvage radiotherapy was required to meet our MR definition. However, this definition allowed us to identify those recurrences that most, if not all, investigators would interpret as clinically meaningful, rather than include radiographic-only failures that might never have had an impact on a patient’s quality of life or survival. Finally, our finding that paraspinal disease is associated with higher MR rates may not apply to other SBRT techniques that incorporate marginal expansions. The strengths of this study lie in its large size, consistent treatment approach, and systematic method of characterizing a clinically meaningful yet often underreported endpoint. Moreover, we describe a novel finding: the presence of paraspinal disease at the time of SBRT is associated with a significant increase in the risk of MR and the resultant physical, psychological, and financial costs to the patient that are associated with disease progression and retreatment. Based on these results, we believe that expanded treatment margins may be prudent in these patients. For patients with paraspinal disease not being treated with conventional radiotherapy, we currently employ a nonuniform expansion of 3 mm around the soft tissue component of the target volume, without any associated expansion adjacent to the spinal cord. By doing so, we are increasing our dosimetric coverage of potential microscopic spread of soft tissue disease without compromising our ability to meet our spinal cord constraints. Finally, our data suggest that SBRT doses of at least 16 Gy are preferred and appear to be associated with less MR.
Target volume delineation used for spine SBRT and patterns of failure
Study, center (ref[s].) Current Study, Cleveland Clinic Chang et al., MDACC (4) Gerszten et al. Pittsburgh, PA (5) Gerszten, et al. Pittsburgh, PA (14) Nelson et al., Duke Univ. (12)
No. of treatments
Target volume definition
208
All visible tumor and associated subsitesz All visible tumor and associated subsitesz All visible tumor All visible tumor All visible tumor and associated subsitesz All visible tumor and associated subsitesz All visible tumor and associated subsitesz
74 500 48 21
Ryu et al., Henry Ford (13)
61
Yamada et al., MSKCC (17)
103
Additional planning margin (mm)*
Median prescribed dose (Gy)
Median no.of fractions
Incidence (%) of margin failurey
0
14
1
12.6
2
27
3
“Several”
0 0 6
20 20 21
1 1 3
NR 10 0
0
16
1
4
2e3
24
1
NR
Abbreviations: NR Z not reported; MSKCC Z Memorial Sloan Kettering Cancer Center; MDACC Z M.D. Anderson Cancer Center. * This expansion was not done in the area immediately adjacent to the spinal cord. y The denominator for this value is the total number of treatments included in the cited study, not the total number of patients. This value also represents the crude rate, rather than the actuarial rate. z Associated subsites refers to the inclusion of additional areas at risk for microscopic extension of disease. Similar to the current target delineation procedure on RTOG 0631, the entire vertebral body would be included for anterior lesions, the entire posterior elements for lesions involving the pedicles, or spinous processes, and at times, some combination would be used based on the location of gross disease, with care not to encircle the spinal cord when possible.
302
Koyfman et al.
Conclusions Highly conformal SBRT for spinal metastases results in a low overall rate of MR that compares favorably to that resulting from treatment with conventional radiotherapy. The presence of paraspinal disease at the time of SBRT and an SBRT dose of less than 16 Gy were associated with significantly higher risks of MR. In addition to the routine use of 16 Gy, the use of nonuniform treatment margin expansions should be considered in patients with paraspinal disease.
References 1. Sahgal A, Larson DA, Chang EL. Stereotactic body radiosurgery for spinal metastases: A critical review. Int J Radiat Oncol Biol Phys 2008;71:652e665. 2. Potters L, Kavanagh B, Galvin JM, et al. American Society for Therapeutic Radiology and Oncology (ASTRO) and American College of Radiology (ACR) practice guideline for the performance of stereotactic body radiation therapy. Int J Radiat Oncol Biol Phys 2010;76:326e332. 3. Bilsky MH, Yamada Y, Yenice KM, et al. Intensity-modulated stereotactic radiotherapy of paraspinal tumors: A preliminary report. Neurosurgery 2004;54:823e831. 4. Chang EL, Shiu AS, Mendel E, et al. Phase I/II study of stereotactic body radiotherapy for spinal metastasis and its pattern of failure. J Neurosurg Spine 2007;7:151e160. 5. Gerszten PC, Burton SA, Ozhasoglu C, Welch WC. Radiosurgery for spinal metastases: Clinical experience in 500 cases from a single institution. Spine 2007;32:193e199. 6. Hamilton AJ, Lulu BA. A prototype device for linear acceleratorbased extracranial radiosurgery. Acta Neurochir Suppl 1995;63: 40e43.
International Journal of Radiation Oncology Biology Physics 7. Ryu S, Fang Yin F, Rock J, et al. Image-guided and intensitymodulated radiosurgery for patients with spinal metastasis. Cancer 2003;97:2013e2018. 8. Sheehan JP, Shaffrey CI, Schlesinger D, Williams BJ, Arlet V, Larner J. Radiosurgery in the treatment of spinal metastases: Tumor control, survival, and quality of life after helical tomotherapy. Neurosurgery 2009;65:1052e1062. 9. Wang JZ, Rice R, Mundt A, Sandhu A, Murphy K. Image-guided stereotactic spine radiosurgery on a conventional linear accelerator. Med Dosim 2010;35:53e62. 10. Ryu S, Jin JY, Jin R, et al. Partial volume tolerance of the spinal cord and complications of single-dose radiosurgery. Cancer 2007;109: 628e636. 11. Hartsell WF, Scott CB, Bruner DW, et al. Randomized trial of shortversus long-course radiotherapy for palliation of painful bone metastases. J Natl Cancer Inst 2005;97:798e804. 12. Nelson JW, Yoo DS, Sampson JH, et al. Stereotactic body radiotherapy for lesions of the spine and paraspinal regions. Int J Radiat Oncol Biol Phys 2009;73:1369e1375. 13. Ryu S, Rock J, Rosenblum M, Kim JH. Patterns of failure after single-dose radiosurgery for spinal metastasis. J Neurosurg 2004; 101(Suppl 3):402e405. 14. Gerszten PC, Burton SA, Ozhasoglu C, et al. Stereotactic radiosurgery for spinal metastases from renal cell carcinoma. J Neurosurg Spine 2005;3:288e295. 15. Damast S, Wright J, Bilsky M, et al. Impact of dose on local failure rates after image-guided reirradiation of recurrent paraspinal metastases. Int J Radiat Oncol Biol Phys 2010. 16. RTOG 0631 protocol. http://www.rtog.org/members/protocols/0631/ 0631.pdf. Accessed December 12, 2010. 17. Yamada Y, Bilsky MH, Lovelock DM, et al. High-dose, single-fraction image-guided intensity-modulated radiotherapy for metastatic spinal lesions. Int J Radiat Oncol Biol Phys 2008;71:484e490.