Brain metastasis is the most common malignant

CLINICAL ARTICLE J Neurosurg 126:735 743, 2017 Impact of the radiosurgery prescription dose on the local control of small (2 cm or smaller) brain metastases Alireza M. Mohammadi, MD, 1,2 Jason L. Schroeder, MD, 5 Lilyana Angelov, MD, 1,2 Samuel T. Chao, MD, 3 Erin S. Murphy, MD, 3 Jennifer S. Yu, MD, PhD, 3 Gennady Neyman, PhD, 3 Xuefei Jia, MS, 4 John H. Suh, MD, 3 Gene H. Barnett, MD, 1,2 and Michael A. Vogelbaum, MD, PhD 1,2 1 The Rose Ella Burkhardt Brain Tumor and Neuro-Oncology Center, Departments of 2 Neurosurgery and 3 Radiation Oncology, and 4 Quantitative Health Science, Neurological Institute and Taussig Cancer Institute, Cleveland Clinic, Cleveland; and 5 Department of Neurosurgery, University of Toledo Medical Center, Toledo, Ohio OBJECTIVE The impact of the stereotactic radiosurgery (SRS) prescription dose (PD) on local progression and radiation necrosis for small ( 2 cm) brain metastases was evaluated. METHODS An institutional review board approved retrospective review was performed on 896 patients with brain metastases 2 cm (3034 tumors) who were treated with 1229 SRS procedures between 2000 and 2012. Local progression and/or radiation necrosis were the primary end points. Each tumor was followed from the date of radiosurgery until one of the end points was reached or the last MRI follow-up. Various criteria were used to differentiate tumor progression and radiation necrosis, including the evaluation of serial MRIs, cerebral blood volume on perfusion MR, FDG-PET scans, and, in some cases, surgical pathology. The median radiographic follow-up per lesion was 6.2 months. RESULTS The median patient age was 56 years, and 56% of the patients were female. The most common primary pathology was non small cell lung cancer (44%), followed by breast cancer (19%), renal cell carcinoma (14%), melanoma (11%), and small cell lung cancer (5%). The median tumor volume and median largest diameter were 0.16 cm 3 and 0.8 cm, respectively. In total, 1018 lesions (34%) were larger than 1 cm in maximum diameter. The PD for 2410 tumors (80%) was 24 Gy, for 408 tumors (13%) it was 19 to 23 Gy, and for 216 tumors (7%) it was 15 to 18 Gy. In total, 87 patients (10%) had local progression of 104 tumors (3%), and 148 patients (17%) had at least radiographic evidence of radiation necrosis involving 199 tumors (7%; 4% were symptomatic). Univariate and multivariate analyses were performed for local progression and radiation necrosis. For local progression, tumors less than 1 cm (subhazard ratio [SHR] 2.32; p < 0.001), PD of 24 Gy (SHR 1.84; p = 0.01), and additional whole-brain radiation therapy (SHR 2.53; p = 0.001) were independently associated with better outcome. For the development of radiographic radiation necrosis, independent prognostic factors included size greater than 1 cm (SHR 2.13; p < 0.001), location in the corpus callosum (SHR 5.72; p < 0.001), and uncommon pathologies (SHR 1.65; p = 0.05). Size (SHR 4.78; p < 0.001) and location (SHR 7.62; p < 0.001) but not uncommon pathologies were independent prognostic factors for the subgroup with symptomatic radiation necrosis. CONCLUSIONS A PD of 24 Gy results in significantly better local control of metastases measuring < 2 cm than lower doses. In addition, tumor size is an independent prognostic factor for both local progression and radiation necrosis. Some tumor pathologies and locations may also contribute to an increased risk of radiation necrosis. https://thejns.org/doi/abs/10.3171/2016.3.jns153014 KEY WORDS brain metastases; Gamma Knife; radiation necrosis; local progression; predictive factors; tumor volume; oncology; stereotactic radiosurgery Brain metastasis is the most common malignant brain tumor and occurs in up to one-third of patients with systemic cancer. 17,30,33,38 Multiple sources of high-quality evidence have shown the effectiveness of stereotactic radiosurgery (SRS) for the treatment of brain metastases. 3,4,10,40 As a result, SRS alone or in combination with other treatment modalities, such as surgery or whole-brain radiation therapy (WBRT), has become the mainstay of treatment for patients with brain metastases, particularly for the treatment of smaller tumors. 5,7,14,20,25 ABBREVIATIONS KPS = Karnofsky Performance Scale; MTD = maximum tolerated dose; PD = prescription dose; RTOG = Radiation Therapy Oncology Group; SHR = subhazard ratio; SRS = stereotactic radiosurgery; WBRT = whole-brain radiation therapy. SUBMITTED December 24, 2015. ACCEPTED March 8, 2016. INCLUDE WHEN CITING Published online May 27, 2016; DOI: 10.3171/2016.3.JNS153014. AANS, 2017 J Neurosurg Volume 126 March 2017 735

A. M. Mohammadi et al. To determine the relationship between SRS dose and toxicity, Radiation Therapy Oncology Group (RTOG 90-05) was designed as a dose-escalation study and included separate arms based on the size of the target tumor. 36 The maximum tolerated dose (MTD) for lesions with a maximum linear diameter between 3 and 4 cm was found to be 15 Gy to the tumor margin, while for tumors between 2 and 3 cm it was 18 Gy. However, for tumors smaller than 2 cm in maximum diameter, MTD was not reached even at a dose of 24 Gy to the tumor margin. The RTOG 90-05 investigators were reluctant to increase the treatment dose beyond 24 Gy, and hence this dose became the de facto MTD. Despite the fact that the actual MTD was not reached for tumors smaller than 2 cm in maximum diameter, not all SRS centers use 24 Gy as their preferred dose for these tumors. Various dosing schedules are being used in practice, especially for lesions smaller than 2 cm, in part because of the lack of high-quality evidence that examines the treatment efficacy versus risk of radiation necrosis associated with a marginal tumor dose of 24 Gy. 23,31,47 Furthermore, it should be noted that the RTOG 90-05 study included patients who had received prior fractionated radiotherapy; many patients today are being treated with SRS as the primary treatment and hence have not had prior cranial exposure to therapeutic radiation. 3 To further understand the relationship of the risks to benefits associated with a prescription dose (PD) of 24 Gy, we analyzed our large, institutional review board approved clinical database that includes patients with brain metastases that were treated with SRS. Our goal was to identify prognostic factors, including the impact of radiosurgery PD, on the rates of local tumor control, and radiation necrosis when SRS is used to treat brain metastases smaller than or equal to 2 cm. This is the largest series ever reported, with longitudinal follow-up of more than 3000 brain metastases, on the response to treatment of small brain metastases. Methods Between 2000 and 2012, more than 1450 patients were treated with Gamma Knife radiosurgery at the Cleveland Clinic for more than 4700 brain metastases. The inclusion criteria for our study consisted of adult patients with brain metastases treated with Gamma Knife radiosurgery during the study period (4700 tumors). Exclusion criteria were 1) maximum dimension (based on the largest diameter on T1-weighted postcontrast MRI or CT if no MRI was performed as part of the Gamma Knife planning scans) greater than 2 cm (900 tumors), 2) treatment with PD less than 15 Gy (30 tumors), and 3) lack of posttreatment MR images available for review (696 tumors). Therefore, the final study cohort included 896 patients who underwent 1229 radiosurgery treatments for 3034 brain metastases. Of note, among the 900 lesions that were larger than 2 cm and were excluded from the study, approximately 320 patients exclusively had lesions measuring larger than 2 cm. However, more than 550 lesions larger than 2 cm were found in approximately 330 patients with multiple brain metastases who had some lesions 2 cm or smaller and some larger than 2 cm. For the purposes of this study, only 2 cm or smaller lesions were considered for the analysis and included. Most patients (680 cases; 73%) underwent 1 radiosurgery procedure, while 216 patients (24%) had 2 (16%) or more (8%) procedures (maximum 6 SRS sessions). This study was approved by the institutional review board of Cleveland Clinic. The primary end points were local progression and onset of at least radiographic evidence of radiation necrosis after SRS. The subgroup of patients with symptomatic radiation necrosis was evaluated independently as well. We followed every individual lesion on every postoperative MRI study and comparisons were made based on the unior bidimensional measurements per the neuroradiology reports and comparisons with the Gamma Knife or prior scan measurements. In cases in which these factors were not reported clearly or there was conflicting information, additional measurement was performed by one of the authors (A.M.M.). We used a variety of techniques to differentiate between local progression and radiation necrosis, which can often appear similar on conventional contrastenhanced MRI. These techniques included any of the following: repeat imaging at intervals ranging from 1 to 3 months; MR perfusion studies to evaluate relative cerebral blood volume in the area of the lesion; FDG avidity on PET imaging; and, in select cases, surgical pathology. Ultimately, the behavior of the tumor on serial radiographic evaluation was the most reliable method for distinguishing radiographic radiation necrosis from tumor progression. 11 In complicated situations for example, when various imaging modalities produced conflicting results we used the consensus of our multidisciplinary brain tumor board to render a diagnosis. Despite utilizing this multimodality approach, we were not able to reliably differentiate between local progression and radiation necrosis in 72 lesions (e.g., in some cases, the patient died after the first posttreatment MRI study due to progressive systemic disease without having later follow-up scans or a perfusion study), and the lesions were excluded from further analysis. All lesions that met the inclusion and exclusion criteria were followed after SRS. Tumor response (enlargement, stable, shrinkage, resolution) was assessed for all lesions on the first post-srs MRI study (4 8 weeks after SRS), and thereafter all lesions were just followed for possible enlargement (i.e., one of the end points) or until the last available radiographic follow-up. Overall, the median radiographic follow-up of the lesions was 6.2 months. We evaluated various patient and tumor characteristics as prognostic factors for either local control or the development of at least radiographic radiation necrosis, including primary pathology, prior treatment with either or both SRS and WBRT to the same lesion, location of the tumor, and the patient s clinical status at the time of treatment. Additionally, treatment data, including PD, prescription isodose line, tumor volume, maximum diameter, and the conformality ratio, were evaluated. SRS was performed using Gamma Knife models B, C, 4C, and Perfexion (Elek ta AB) during the study period based on our institutional protocol. 25 Records associated with the treatment of 896 patients were reviewed. Table 1 summarizes the patient and treatment characteristics. Demographically, 502 patients (56%) 736 J Neurosurg Volume 126 March 2017

Radiosurgery dose and local control of small brain metastases TABLE 1. Patient and first SRS characteristics Factor n (%) Sex Male 394 (44) Female 502 (56) Primary site Non small cell lung cancer 393 (44) Breast 166 (19) Renal cell 123 (14) Melanoma 98 (11) Small cell lung cancer 46 (5) Uncommon pathologies 70 (8) Other systemic metastases No 208 (23) Yes (single metastases) 348 (39) Yes (multiple metastases) 340 (38) Controlled systemic disease at SRS 326 (36) KPS score 90 100 466 (52) 70 80 381 (43) 50 60 49 (5) Recursive partitioning analysis class 1 139 (16) 2 708 (79) 3 49 (5) Graded prognostic assessment class 0 1 203 (23) 1.5 2.5 571 (64) 3 89 (10) 3.5 4 33 (4) Neurological function status Asymptomatic 336 (37) Mild symptoms 437 (49) Moderate/severe symptoms 123 (14) Chemotherapy in the month prior to SRS 475 (53) Additional WBRT before or after index SRS 338 (38) No. of targets 1 431 (48) 2 4 364 (40) 5 123 (14) No. of SRS treatments 1 680 (76) 2 145 (16) 3 42 (5) 4 17 (2) 5 12 (1) were female. The median age at primary cancer diagnosis was 56 years (range 7 88 years), and the median age at neurological involvement was 59 years (range 16 89 years). The median interval from primary diagnosis to neurological involvement with brain metastasis was 12 months (range 0.3 4 years), and the median interval from the diagnosis of brain metastasis to the initial radiosurgery was 2.1 months (range 0 10.1 years). The distribution of the primary cancer histologies was non small cell lung cancer in 393 patients (44%), breast cancer in 166 patients (19%), renal cell carcinoma in 123 patients (14%), melanoma in 98 patients (11%), and small cell lung cancer in 46 patients (5%). The remaining patients (70 patients; 8%) had other less common types of cancer that were grouped as uncommon pathologies, including colon cancer (21 patients; 34 lesions), esophageal cancer (7 patients; 17 lesions), gynecological cancer (15 patients; 27 lesions), urothelial cancer (7 patients; 15 lesions), sarcoma (4 patients; 9 lesions), cancer of the salivary glands (3 patients; 10 lesions), unknown primary (7 patients; 17 lesions), and other (6 patients; 18 lesions). Statistical Considerations The primary end points of the study were the time to local progression and time to radiographic or symptomatic radiation necrosis, which were measured from the date of SRS. Overall survival, which was also measured from the date of SRS, was calculated. There is a fairly high risk of death from systemic disease without central nervous system local progression or radiation necrosis in this population, and therefore the competing risks methods as described by Fine and Gray were used for both the univariate and multivariate analyses. 15 Stepwise variable selection with p = 0.05 as the criterion for both entry and retention in a model was used to identify independent prognostic factors. All analyses took into account possible correlations resulting from some patients having more than 1 lesion and/or more than 1 SRS procedure. The results from these analyses were summarized by the subhazard ratio (SHR) for the outcome of interest (local control or radiation necrosis). The categorical data were summarized as frequency counts and percentages; the Kaplan-Meier method was used to summarize overall survival; and other measured factors were summarized as medians and ranges. For convenience, measured factors such as maximum tumor diameter and volume were dichotomized using a recursive partitioning algorithm. All data analyses were performed using SAS (version 9.2, SAS Institute, Inc.) and Stata (version 12.1, StataCorp LP). Results Descriptive Data The brain was the only site of metastases in 208 patients (23%). The other 688 patients (77%) harbored extracranial metastases at the time of SRS. These were distributed as metastases to a single extracranial organ in 348 patients (39%) and to multiple organs in 340 patients (38%). Five hundred seventy patients (64%) had systemic disease progression at the time of SRS. Most patients (475 cases; 53%) had received chemotherapy within the month prior to their initial radiosurgery procedure for the treatment of systemic disease. At the initial SRS treatment, patients tended to show good Karnofsky Performance Scale (KPS) scores. A KPS score of 90 to 100 was documented for 466 patients J Neurosurg Volume 126 March 2017 737

A. M. Mohammadi et al. (52%) and 265 patients (30%) had a KPS score of 80. Additionally, 708 patients (79%) were classified as recursive partitioning analysis Class II and 571 patients (64%) in graded prognostic assessment Group 1.5 to 2.5. Overall, there were 1229 radiosurgery procedures for the treatment of 3034 lesions. Most of the 896 patients (641 patients; 81%) had 1 to 2 tumors treated; however, 11% of the cohort (111 of 896) had treatment for 5 or more tumors (range 1 16). Tumors were located primarily within the supratentorial compartment (2373; 78%) with 17 lesions occurring within the corpus callosum (0.5%). Most of the infratentorial lesions were located in the cerebellum (589 lesions; 19%), with an additional 74 lesions located within the brainstem (2%). The median tumor volume was 0.16 cm 3 (range 0.0003 7.2 cm 3 ), and the median of the maximum linear diameter was 0.8 cm (range 0.02 2 cm). A maximum diameter larger than 1 cm was observed in 1019 lesions (34%). The PD at the tumor margin ranged from 15 to 24 Gy with the following distribution: 216 lesions (7%) received 15 to 18 Gy, 408 lesions (13%) received 19 to 23 Gy, and 2410 lesions (80%) received 24 Gy. The median treatment isodose line was 56%. For purpose of definition within this study, boost radiation was considered to have been used when a particular lesion received both WBRT and SRS with no local progression having occurred between the 2 treatments (regardless of any potential distant intracranial failure). Using this definition, 1041 lesions (34%) were treated with boost radiation. Outcome The estimated median overall survival was 14.9 months after SRS, and 748 patients (83%) died during follow-up. In 338 patients (45%), death was caused by systemic progression with radiographically stable intracranial disease. Intracranial disease progression caused death in 134 patients (18%) with stable systemic disease and another 166 patients (22%) with concurrent systemic progression. In 110 patients (15%), the cause of death was indeterminate. The median radiographic follow-up of the entire cohort was 6.2 months. New intracranial lesions were observed in 445 patients (45%) after a median of 10.2 months. In more than half of these cases (246 patients; 55%), intracranial progression was accompanied by concurrent systemic cancer progression. Overall, 87 patients (10%) experienced local progression of a total of 104 lesions (3%) at some point in the course of their disease. The diagnosis of local progression was made based on continuous lesion enlargement on multiple, serial follow-up MRI scans (60 lesions; 59%), positive MR perfusion studies (i.e., elevated cerebral blood volume) (11 lesions; 10%), or positive FDG-PET scans (5 lesions; 5%), in addition to continuous enlargement on follow-up scans or the pathological evaluation of surgical specimens in 28 lesions (27%). In 148 patients, 17% had radiographic radiation necrosis observed at some point during the course of their disease in a total of 199 lesions (7%). One hundred six lesions (53% of lesions with radiographic radiation necrosis and 4% of all lesions) in 83 patients with radiation necrosis (56% of radiographic radiation necrosis patients and 9% of all patients) were symptomatic, and the other 93 lesions (47% of lesions with radiographic radiation necrosis and 3% of all lesions) in 65 patients (44% of radiographic radiation necrosis patients and 8% of all patients) were asymptomatic. The primary criterion for the diagnosis of radiographic radiation necrosis was the spontaneous stabilization or shrinkage of the evolving lesion on follow-up MRI (155 lesions; 77% of cases of radiographic radiation necrosis); 66 of these lesions (33% of cases of radiographic radiation necrosis) had relevant negative FDG-PET or MR perfusion results as well. Seventeen lesions (9%) were pathologically shown to be cases of radiation necrosis. The diagnosis of radiation necrosis was more complicated for 29 lesions (14%), as the information provided by multiple imaging modalities was contradictory. The consensus of our multidisciplinary tumor board was used to make the diagnosis of radiographic radiation necrosis for 16 lesions (8%) that had progressive shrinkage on follow-up MRI scans but positive FDG-PET or MR perfusion results. Similarly, tumor board consensus was used for the diagnosis of radiation necrosis in 13 lesions (6%), which showed continuous enlargement over multiple MRI scans but simultaneously had multiple negative FDG-PET and/ or MR perfusion scans. Statistical Analysis Univariate analysis was performed and revealed that the tumor diameter (Fig. 1) is a prognostic factor for all 3 end points, which consisted of local progression (> 1 vs 1 cm; SHR 2.32; p < 0.001), radiographic radiation necrosis (SHR 2.13; p < 0.001), and also the subgroup with symptomatic radiation necrosis (SHR 4.87; p < 0.001). Tumor volume and the conformality index also had similar results for all 3 end points. Several different locations of the brain, including the brainstem, were evaluated and compared with other locations according to our end points, with no statistically significant difference except for 1 location (the corpus callosum). Tumors located in the corpus callosum, compared with other locations of the brain, were shown to have more radiographic (SHR 4.90; p < 0.001) and symptomatic (SHR 7.70; p < 0.001) radiation necrosis with no impact on tumor progression. In addition, a PD (Fig. 1) of 24 Gy (SHR 2.03; p = 0.004) and additional WBRT before or after SRS (SHR 2.46; p = 0.001) were prognostic factors for tumor progression with no impact on radiation necrosis (Table 2). Multivariate analysis revealed that maximum tumor diameter (> 1 vs 1 cm; SHR 2.32; p < 0.001), PD ( 24 vs 24 Gy; SHR 1.84; p = 0.01), and boost radiation treatment (no vs yes; SHR 2.53; p = 0.001) were independent prognostic factors for local progression (Table 3). Independent predictors for radiographic radiation necrosis were (again) maximum tumor diameter (SHR 2.13; p < 0.001) as well as lesion location within the corpus callosum (yes vs no; SHR 5.72; p < 0.001) and uncommon pathologies (yes vs no; SHR 1.65; p = 0.05) (Table 3). For symptomatic radiation necrosis, the maximum tumor diameter (SHR 4.78; p < 0.001) as well as lesion location within the corpus callosum (SHR 7.62; p < 0.001) were independent prognostic factors (Table 3). Discussion Recommendations for the PD for treating brain metas- 738 J Neurosurg Volume 126 March 2017

Radiosurgery dose and local control of small brain metastases FIG. 1. Cumulative hazards of tumor diameter and PD (24 Gy) on different outcomes after SRS: local progression (A and D), radiographic radiation necrosis (B and E), and symptomatic radiation necrosis (C and F). RT = radiation therapy. tases with SRS have taken into account target tumor size. RTOG 90-05 was a dose-escalation study for SRS, which stratified patients into 3 arms based on the size of each target tumor at the time of treatment. For lesions larger than 2 cm, the MTDs for SRS were clearly defined as the doselimiting toxicities that were encountered when higher doses were used. On the other hand, the maximum tolerated SRS dose was not reached for tumors smaller than 2 cm in diameter. The investigators nonetheless stopped the study for this cohort at a PD of 24 Gy. 36 Subsequently, Vogelbaum et al. retrospectively evaluated more than 200 patients who were treated with SRS according to the RTOG 90-05 criteria and showed that the group of lesions treated with 24 Gy had better local control compared with those treated with 18 and 15 Gy. 45 That study did not directly address the risk of radiation necrosis in those patients. Additionally, the tumors treated with 18 and 15 Gy were larger lesions, and the relationship between treatment dose and lesion size makes it unclear if radiation dose alone accounts for better control of smaller lesions. 8 Despite the fact that an SRS treatment dose of 24 Gy has been shown to produce better local control of brain metastases and the lack of evidence demonstrating an increased risk radiation necrosis associated with this PD, many centers have been reluctant to use 24 Gy for lesions smaller than 2 cm and as a result doses between 18 and 24 Gy are being used for such lesions. 9,22,26,37,42 We now report on the largest series to date of patients with brain metastases treated with SRS. We performed a lesion-by-lesion analysis for tumors 2 cm or smaller to evaluate the complex relationships between treatment dose, tumor size, risk of local tumor progression, and risk of developing radiation necrosis. Our results showed strong and independent relationships between tumor size and both the risk of tumor progression and the risk of developing radiation necrosis (radiographic or symptomatic) at a cutoff point of 1 cm (p < 0.001). Tumor location within the corpus callosum (p < 0.001) and SRS treatment for uncommon pathologies (p = 0.05) were additional independent prognostic factors for the development of radiation necrosis, while planned boost SRS before or after WBRT was prognostic of better local control (p = 0.001). In evaluating the impact of radiation dose, a PD of 24 Gy compared with lower PDs was found to be an independent prognostic factor for improved local control (p = 0.01), but not for radiation necrosis. Chang et al. reviewed 135 patients with 153 lesions (2 cm or less in maximum diameter) who were treated with PD of at least 20 Gy and observed that tumors smaller than 1 cm had a significantly better control rate than tumors measuring 1 to 2 cm (p = 0.001). 8 This relationship between tumor size and the local control rate has been supported by other studies as well. 6,23,32 The relationship between PD and local control also has been evaluated in a number of studies. In addition to Vogelbaum et al. s study, which showed superior results from treatment at 24 Gy compared with lower doses, multiple studies provide evidence for better local control after higher radiosurgery doses. 8,31,32,45,46 Wiggenraad et al., in a systemic review of the literature published between 1990 and 2009, and Rodrigues et al. in a recursive partitioning analysis, concluded that SRS PDs of 21 Gy or higher are associated with better local control. 31,46 Finally, there is Level 1 evidence showing that boost Gamma Knife following WBRT leads to better local control compared with Gamma Knife or WBRT alone. 3,4 J Neurosurg Volume 126 March 2017 739

A. M. Mohammadi et al. TABLE 2. Statistically significant results from the univariate analysis Factor Local Progression Radiographic Radiation Necrosis Symptomatic Radiation Necrosis SHR (95% CI)* p Value SHR (95% CI)* p Value SHR (95% CI)* p Value Maximum tumor diameter >1 cm vs <1 cm 2.32 (1.51 3.56) <0.001 2.13 (1.54 2.94) <0.001 4.78 (2.94 7.76) <0.001 Tumor volume >0.1 ml vs <0.1 ml 2.76 (1.47 5.17) 0.002 2.08 (1.34 3.24) 0.001 3.97 (2.20 7.19) <0.001 Isodose line <55% vs >55% 1.37 (0.89 2.11) 0.15 1.64 (1.16 2.31) 0.005 2.0 (3.22 1.23) 0.005 Conformality index <2.0 vs >2.0 1.86 (1.11 3.12) 0.02 1.81 (1.24 2.65) 0.002 2.0 (3.22 1.23) <0.001 PD <24 Gy vs 24 Gy 2.03 (1.26 3.26) 0.004 1.19 (0.77 1.83) 0.43 1.04 (0.56 1.92) 0.90 Additional radiation No vs yes 2.46 (1.45 4.17) 0.001 1.34 (0.96 1.87) 0.09 0.98 (0.62 1.57) 0.95 Tumor location Corpus callosum vs other 3.19 (0.79 12.96) 0.10 4.90 (2.15 11.16) <0.001 7.55 (2.66 21.4) <0.001 Pathology Uncommon vs other 0.46 (0.15 1.44) 0.18 1.82 (1.10 3.01) 0.02 2.14 (1.08 4.24) 0.03 * SHR is the hazard ratio for local progression (or radiation necrosis) when death from any cause is considered a competing risk. Values > 1.0 indicate that the first level of a factor is associated with an increased risk of progression (radiation necrosis) relative to the second level. Values < 1.0 indicate that the first level is associated with a lower risk. Several previous studies have shown that radiation necrosis occurs more frequently after the treatment of larger brain metastases. 1,6,16,27 However, an association between the risk of radiation necrosis and the treatment of tumors in specific locations or uncommon pathologies has not been identified previously in large series. While our observations that lesion location in the corpus callosum and a specific group of pathologies are an interesting addition TABLE 3. Results of the multivariate analysis Factor SHR (95% CI)* p Value Local progression Maximum tumor diameter (>1 cm vs 2.32 (1.51 3.57) <0.001 1 cm) PD (<24 Gy vs 24 Gy) 1.84 (1.15 2.95) 0.01 Additional radiation (no vs yes) 2.53 (1.49 4.29) 0.001 Radiographic radiation necrosis Maximum tumor diameter (>1 cm vs 2.13 (1.55 2.93) <0.001 1 cm) Tumor location (corpus callosum vs 5.72 (2.28 14.3) <0.001 other) Pathology (uncommon vs other) 1.65 (1.01 2.70) 0.05 Symptomatic radiation necrosis Maximum tumor diameter (>1 cm vs 4.78 (2.95 7.75) <0.001 1 cm) Tumor location (corpus callosum vs other) 7.62 (2.38 24.4) <0.001 * Values > 1.0 indicate that the first level of a factor is associated with an increased risk of progression to radiation necrosis. to the literature, these specific results must be interpreted cautiously because they were not the common features of brain metastases in our series. Indeed, there were only 17 tumors (of a total of 3034 tumors) located in the corpus callosum. Our observation that 5 of these tumors developed radiation necrosis (30%) is remarkable because this is a much higher percentage than for any other location, and it was significant on the multivariate analysis (p < 0.001). Our observation of an association between uncommon pathology and the risk of radiation necrosis similarly needs to be viewed cautiously as it does not prove that all tumors with uncommon pathologies have a higher chance of developing radiation necrosis after SRS treatment. For example, there were no cases of radiation necrosis after treating 34 metastases from colon cancer or 9 sarcomas. However, radiation necrosis did occur more frequently following the treatment of 3 esophageal metastases (18%), 7 gynecological metastases (26%), and 3 urothelial metastases (20%). Of note, despite the very large size of the series, we did not observe an increased risk of radiation necrosis with increased radiation dose according to either the univariate or multivariate analysis. Overall, only 7% of the more than 3000 lesions reviewed in our study exhibited radiographic radiation necrosis, which is a reasonably small percentage of lesions given the doses that were used and the very high rate of local control observed. A review of the literature reveals that only 1 study has shown an increased risk of radiation toxicity with a PD of 25 Gy. 28 To the best of our knowledge, therefore, for tumors of 2 cm or smaller in maximal diameter, there is no evidence that shows an increased risk of developing radiation necrosis with a PD of 24 Gy or lower, and our study provides compelling evi- 740 J Neurosurg Volume 126 March 2017

Radiosurgery dose and local control of small brain metastases dence that a PD of 24 Gy provides the optimal balance of local control and risk of radiation injury. 33,36,42 There are important limitations to this study. First, this is a retrospective study and hence there is the possibility for treatment bias. However, as a group, we have largely followed the RTOG 90-05 dosing scheme, which has helped to bring uniformity to our SRS practice. Having said that, more than 20% of the lesions were treated with radiation doses less than 24 Gy. Many of the decisions were made for the individual patients per the discretion of the radiation oncologist and surgeon. Circumstances that may have led to reduced PD in some patients included a large number of tumors, multiple prior sessions of radiosurgery, the close proximity of 2 tumors to each other, the prior use of whole-brain radiotherapy, and proximity to the optic apparatus. The only consistently used dose-reduction regimen is in patients with tumors located in the brainstem (of note, no differences in outcome were observed in tumor progression or radiation necrosis in the brainstem lesions compared with other locations). Because this was a retrospective analysis, we do not have records of the thought processes that went into the dose reduction for each case. Probably, the most important limitation of this study relates to the difficulty of differentiating between tumor progression and radiation necrosis, which affects all of the subsequent analysis. Unfortunately, there is no single radiological diagnostic imaging modality for differentiating between tumor progression and radiation necrosis after radiosurgery. 11,29,41 A variety of MRI sequences (including perfusion, diffusion-weighted imaging, and spectroscopy) as well as metabolic imaging modalities (FDG-PET and SPECT) have been suggested to provide modest accuracy. 2,12,13,18,19,21,24,34,35,41,43,44 Despite significant improvements in imaging modalities over time, this diagnostic dilemma often still remains when trying to differentiate between tumor progression and radiation necrosis. 18,19,29,39,43,46 Surgical pathology is the most reliable method for distinguishing between progression and radiation necrosis; however, there are the attendant surgical risks and its use is influenced by selection bias. 43 In our series, 375 lesions showed enlargement on any posttreatment MRI. For 104 of these lesions (28%), there was adequate radiographic and clinical evidence to confirm tumor progression. An additional 175 tumors (45%) demonstrated clinical and radiographic behavior typical of radiation necrosis, or this diagnosis was confirmed by biopsy. The remaining lesions showed conflicting results on imaging studies or had inadequate follow-up. Of these, we were able to classify 29 more lesions (8%) as progression or radiation necrosis based on the consensus of our multidisciplinary tumor board. However, for 72 lesions (19% of enlarging lesions), a definitive classification as either progression or radiation necrosis could not be made, and as a result we had to exclude these lesions from further analysis. However, even if we had classified all of these 72 cases as radiation necrosis, our overall rate of 7% would have increased to only 9%. Particularly in light of the fact that our local control rate was 97%, and most of these cases of radiation necrosis were only detected on radiography and did not produce clinical symptoms, this rate of radiographic radiation necrosis would support a clinically acceptable risk-benefit ratio. Conclusions Based on our study which, to the best of our knowledge, is the largest reported series on small brain metastases with lesion follow-up for brain metastases with a maximum diameter of 2 cm or less the use of a PD of 24 Gy for SRS produces exemplary local control with a less than 10% risk of radiographic radiation necrosis per lesion. The risk of radiation necrosis is less dependent on PD than on lesion size, pathology, and the location of the tumor. Based on these results and our longstanding experience with use of the RTOG 90 05 dosing regimen, we recommend administering a PD of 24 Gy to the tumor margin for all lesions smaller than 2 cm in maximum diameter, unless contraindicated by proximity to the hypothalamus or optic apparatus or location in the brainstem. References 1. Alexander E III, Moriarty TM, Davis RB, Wen PY, Fine HA, Black PM, et al: Stereotactic radiosurgery for the definitive, noninvasive treatment of brain metastases. J Natl Cancer Inst 87:34 40, 1995 2. Alexiou GA, Tsiouris S, Kyritsis AP, Polyzoidis KS, Fotopoulos AD: Brain SPECT by 99m Tc-tetrofosmin for the differentiation of tumor recurrence from radiation injury. J Nucl Med 49:1733 1734, 2008 (Letter) 3. Andrews DW, Scott CB, Sperduto PW, Flanders AE, Gaspar LE, Schell MC, et al: Whole brain radiation therapy with or without stereotactic radiosurgery boost for patients with one to three brain metastases: phase III results of the RTOG 9508 randomised trial. Lancet 363:1665 1672, 2004 4. Aoyama H, Shirato H, Tago M, Nakagawa K, Toyoda T, Hatano K, et al: Stereotactic radiosurgery plus whole-brain radiation therapy vs stereotactic radiosurgery alone for treatment of brain metastases: a randomized controlled trial. JAMA 295:2483 2491, 2006 5. Asher AL, Burri SH, Wiggins WF, Kelly RP, Boltes MO, Mehrlich M, et al: A new treatment paradigm: neoadjuvant radiosurgery before surgical resection of brain metastases with analysis of local tumor recurrence. Int J Radiat Oncol Biol Phys 88:899 906, 2014 6. Blonigen BJ, Steinmetz RD, Levin L, Lamba MA, Warnick RE, Breneman JC: Irradiated volume as a predictor of brain radionecrosis after linear accelerator stereotactic radiosurgery. Int J Radiat Oncol Biol Phys 77:996 1001, 2010 7. Brennan C, Yang TJ, Hilden P, Zhang Z, Chan K, Yamada Y, et al: A phase 2 trial of stereotactic radiosurgery boost after surgical resection for brain metastases. Int J Radiat Oncol Biol Phys 88:130 136, 2014 8. Chang EL, Hassenbusch SJ III, Shiu AS, Lang FF, Allen PK, Sawaya R, et al: The role of tumor size in the radiosurgical management of patients with ambiguous brain metastases. Neurosurgery 53:272 281, 2003 9. Chang EL, Selek U, Hassenbusch SJ III, Maor MH, Allen PK, Mahajan A, et al: Outcome variation among radioresistant brain metastases treated with stereotactic radiosurgery. Neurosurgery 56:936 945, 2005 10. Chang EL, Wefel JS, Hess KR, Allen PK, Lang FF, Kornguth DG, et al: Neurocognition in patients with brain metastases treated with radiosurgery or radiosurgery plus whole-brain irradiation: a randomised controlled trial. Lancet Oncol 10:1037 1044, 2009 11. Chao ST, Ahluwalia MS, Barnett GH, Stevens GH, Murphy ES, Stockham AL, et al: Challenges with the diagnosis and treatment of cerebral radiation necrosis. Int J Radiat Oncol Biol Phys 87:449 457, 2013 J Neurosurg Volume 126 March 2017 741

A. M. Mohammadi et al. 12. Chernov MF, Hayashi M, Izawa M, Usukura M, Yoshida S, Ono Y, et al: Multivoxel proton MRS for differentiation of radiation-induced necrosis and tumor recurrence after gamma knife radiosurgery for brain metastases. Brain Tumor Pathol 23:19 27, 2006 13. Dequesada IM, Quisling RG, Yachnis A, Friedman WA: Can standard magnetic resonance imaging reliably distinguish recurrent tumor from radiation necrosis after radiosurgery for brain metastases? A radiographic-pathological study. Neurosurgery 63:898 904, 2008 14. Ellis TL, Neal MT, Chan MD: The role of surgery, radiosurgery and whole brain radiation therapy in the management of patients with metastatic brain tumors. Int J Surg Oncol 2012:952345, 2012 15. Fine JP, Gray RJ: A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 94:496 509, 1999 16. Flickinger JC, Lunsford LD, Kondziolka D, Maitz AH, Epstein AH, Simons SR, et al: Radiosurgery and brain tolerance: an analysis of neurodiagnostic imaging changes after gamma knife radiosurgery for arteriovenous malformations. Int J Radiat Oncol Biol Phys 23:19 26, 1992 17. Khuntia D, Brown P, Li J, Mehta MP: Whole-brain radiotherapy in the management of brain metastasis. J Clin Oncol 24:1295 1304, 2006 18. Kickingereder P, Dorn F, Blau T, Schmidt M, Kocher M, Galldiks N, et al: Differentiation of local tumor recurrence from radiation-induced changes after stereotactic radiosurgery for treatment of brain metastasis: case report and review of the literature. Radiat Oncol 8:52, 2013 19. Kim YZ, Kim DY, Yoo H, Yang HS, Shin SH, Hong EK, et al: Radiation-induced necrosis deteriorating neurological symptoms and mimicking progression of brain metastasis after stereotactic-guided radiotherapy. Cancer Res Treat 39:16 21, 2007 20. Kondziolka D, Flickinger JC, Lunsford LD: Radiosurgery for brain metastases. Prog Neurol Surg 25:115 122, 2012 21. Larsen VA, Simonsen HJ, Law I, Larsson HB, Hansen AE: Evaluation of dynamic contrast-enhanced T1-weighted perfusion MRI in the differentiation of tumor recurrence from radiation necrosis. Neuroradiology 55:361 369, 2013 22. Lutterbach J, Cyron D, Henne K, Ostertag CB: Radiosurgery followed by planned observation in patients with one to three brain metastases. Neurosurgery 52:1066 1074, 2003 23. Minniti G, Clarke E, Lanzetta G, Osti MF, Trasimeni G, Bozzao A, et al: Stereotactic radiosurgery for brain metastases: analysis of outcome and risk of brain radionecrosis. Radiat Oncol 6:48, 2011 24. Mitsuya K, Nakasu Y, Horiguchi S, Harada H, Nishimura T, Bando E, et al: Perfusion weighted magnetic resonance imaging to distinguish the recurrence of metastatic brain tumors from radiation necrosis after stereotactic radiosurgery. J Neurooncol 99:81 88, 2010 25. Mohammadi AM, Recinos PF, Barnett GH, Weil RJ, Vogelbaum MA, Chao ST, et al: Role of Gamma Knife surgery in patients with 5 or more brain metastases. J Neurosurg 117 (Suppl):5 12, 2012 26. Molenaar R, Wiggenraad R, Verbeek-de Kanter A, Walchenbach R, Vecht C: Relationship between volume, dose and local control in stereotactic radiosurgery of brain metastasis. Br J Neurosurg 23:170 178, 2009 27. Nakamura JL, Verhey LJ, Smith V, Petti PL, Lamborn KR, Larson DA, et al: Dose conformity of Gamma Knife radiosurgery and risk factors for complications. Int J Radiat Oncol Biol Phys 51:1313 1319, 2001 28. Nedzi LA, Kooy H, Alexander E III, Gelman RS, Loeffler JS: Variables associated with the development of complications from radiosurgery of intracranial tumors. Int J Radiat Oncol Biol Phys 21:591 599, 1991 29. Parvez K, Parvez A, Zadeh G: The diagnosis and treatment of pseudoprogression, radiation necrosis and brain tumor recurrence. Int J Mol Sci 15:11832 11846, 2014 30. Patchell RA: The management of brain metastases. Cancer Treat Rev 29:533 540, 2003 31. Rodrigues G, Zindler J, Warner A, Lagerwaard F: Recursive partitioning analysis for the prediction of stereotactic radiosurgery brain metastases lesion control. Oncologist 18:330 335, 2013 32. Schomas DA, Roeske JC, MacDonald RL, Sweeney PJ, Mehta N, Mundt AJ: Predictors of tumor control in patients treated with linac-based stereotactic radiosurgery for metastatic disease to the brain. Am J Clin Oncol 28:180 187, 2005 33. Schüttrumpf LH, Niyazi M, Nachbichler SB, Manapov F, Jansen N, Siefert A, et al: Prognostic factors for survival and radiation necrosis after stereotactic radiosurgery alone or in combination with whole brain radiation therapy for 1 3 cerebral metastases. Radiat Oncol 9:105, 2014 34. Serizawa T, Saeki N, Higuchi Y, Ono J, Matsuda S, Sato M, et al: Diagnostic value of thallium-201 chloride singlephoton emission computerized tomography in differentiating tumor recurrence from radiation injury after Gamma Knife surgery for metastatic brain tumors. J Neurosurg 102 Suppl:266 271, 2005 35. Shah R, Vattoth S, Jacob R, Manzil FF, O Malley JP, Borghei P, et al: Radiation necrosis in the brain: imaging features and differentiation from tumor recurrence. Radiographics 32:1343 1359, 2012 36. Shaw E, Scott C, Souhami L, Dinapoli R, Kline R, Loeffler J, et al: Single dose radiosurgical treatment of recurrent previously irradiated primary brain tumors and brain metastases: final report of RTOG protocol 90-05. Int J Radiat Oncol Biol Phys 47:291 298, 2000 37. Shehata MK, Young B, Reid B, Patchell RA, St Clair W, Sims J, et al: Stereotactic radiosurgery of 468 brain metastases < or = 2 cm: implications for SRS dose and whole brain radiation therapy. Int J Radiat Oncol Biol Phys 59:87 93, 2004 38. Soffietti R, Rudā R, Mutani R: Management of brain metastases. J Neurol 249:1357 1369, 2002 39. Soussain C, Ricard D, Fike JR, Mazeron JJ, Psimaras D, Delattre JY: CNS complications of radiotherapy and chemotherapy. Lancet 374:1639 1651, 2009 40. Stafinski T, Jhangri GS, Yan E, Menon D: Effectiveness of stereotactic radiosurgery alone or in combination with whole brain radiotherapy compared to conventional surgery and/ or whole brain radiotherapy for the treatment of one or more brain metastases: a systematic review and meta-analysis. Cancer Treat Rev 32:203 213, 2006 41. Stockham AL, Tievsky AL, Koyfman SA, Reddy CA, Suh JH, Vogelbaum MA, et al: Conventional MRI does not reliably distinguish radiation necrosis from tumor recurrence after stereotactic radiosurgery. J Neurooncol 109:149 158, 2012 42. Valéry CA, Cornu P, Noël G, Duyme M, Boisserie G, Sakka LJ, et al: Predictive factors of radiation necrosis after radiosurgery for cerebral metastases. Stereotact Funct Neurosurg 81:115 119, 2003 43. Verma N, Cowperthwaite MC, Burnett MG, Markey MK: Differentiating tumor recurrence from treatment necrosis: a review of neuro-oncologic imaging strategies. Neuro Oncol 15:515 534, 2013 44. Vidiri A, Guerrisi A, Pinzi V, Fabi A, Mirri MA, Pompili A, et al: Perfusion computed tomography (PCT) adopting different perfusion metrics: recurrence of brain metastasis or radiation necrosis? Eur J Radiol 81:1246 1252, 2012 45. Vogelbaum MA, Angelov L, Lee SY, Li L, Barnett GH, Suh JH: Local control of brain metastases by stereotactic 742 J Neurosurg Volume 126 March 2017

Radiosurgery dose and local control of small brain metastases radiosurgery in relation to dose to the tumor margin. J Neurosurg 104:907 912, 2006 46. Wiggenraad R, Verbeek-de Kanter A, Kal HB, Taphoorn M, Vissers T, Struikmans H: Dose-effect relation in stereotactic radiotherapy for brain metastases. A systematic review. Radiother Oncol 98:292 297, 2011 47. Yu JB, Schulder M, Knisely J: Radiosurgical dose selection for brain metastasis. Prog Neurol Surg 25:139 147, 2012 Disclosures The authors report the following. Dr. Chao has received honoraria from Varian. Dr. Neyman is a consultant for Elekta AB. Dr. Suh receives travel and lodging reimbursements from Elekta. Author Contributions Conception and design: Vogelbaum, Mohammadi, Schroeder, Angelov, Chao, Murphy, Yu, Neyman, Suh, Barnett. Acquisition of data: Mohammadi, Schroeder. Analysis and interpretation of data: Vogelbaum, Mohammadi, Jia. Drafting the article: Vogelbaum, Mohammadi, Schroeder. Critically revising the article: all authors. Reviewed submitted version of manuscript: all authors. Approved the final version of the manuscript on behalf of all authors: Vogelbaum. Statistical analysis: Jia. Administrative/technical/material support: Vogelbaum, Mohammadi. Study supervision: Vogelbaum, Angelov, Chao, Murphy, Yu, Suh, Barnett. Supplemental Information Previous Presentations This manuscript was presented as a platform presentation at the 82nd Annual Scientific Meeting of the American Association for Neurological Surgeons in San Francisco, 2014. Correspondence Michael A. Vogelbaum, Cleveland Clinic Foundation, Neurological Institute, 9500 Euclid Ave., ND40, Cleveland, OH 44195. email: vogelbm@ccf.org. J Neurosurg Volume 126 March 2017 743