We have previously reported good clinical results

J Neurosurg 113:48 52, 2010 Gamma Knife surgery as sole treatment for multiple brain metastases: 2-center retrospective review of 1508 cases meeting the inclusion criteria of the JLGK0901 multi-institutional prospective study Clinical article To r u Se r i z a w a, M.D., Ph.D., 1 Ma s a a k i Ya m a m o t o, M.D., Ph.D., 2 Ya s u n o r i Sa t o, Ph.D., 3,4 Yo s h i n o r i Hi g u c h i, M.D., Ph.D., 5 Os a m u Na g a n o, M.D., Ph.D., 6 Ta k u ya Ka w a b e, M.D., 2 Sh i n j i Ma t s u d a, M.D., Ph.D., 6 Ju n i c h i On o, M.D., Ph.D., 5 Na o k a t s u Sa e k i, M.D., Ph.D., 3,4 Man a b u Ha t a n o, M.D., Ph.D., 1 a n d Ta t s u o Hi r a i, M.D. Ph.D. 1 1 Tokyo Gamma Unit Center, Tsukiji Neurological Clinic, Tokyo; 2 Mito Gamma House, Katsuta Hospital, Hitachinaka; 3 Clinical Research Center, Chiba University Hospital; 4 Department of Neurological Surgery, 5 Graduate School of Medicine, Chiba University, Chiba; and 6 Gamma Knife House, Chiba Cardiovascular Center, Ichihara, Japan Object. The authors retrospectively reviewed the results of Gamma Knife surgery (GKS) used as the sole treatment for brain metastases in patients who met the eligibility criteria for the ongoing JLGK0901 multi-institutional prospective trial. They also discuss the anticipated results of the JLGK0901 study. Methods. Data from 1508 consecutive cases were analyzed. All of the patients were treated at the Gamma Knife House of Chiba Cardiovascular Center or the Mito Gamma House of Katsuta Hospital between 1998 and 2007 and met the following JLGK0901 inclusion criteria: 1) newly diagnosed brain metastases, 2) 1 10 brain lesions, 3) less than 10 cm 3 volume of the largest tumor, 4) no more than 15 cm 3 total tumor volume, 5) no findings of CSF dissemination, and 6) no impairment of activities of daily living (Karnofsky Performance Scale score < 70) due to extracranial disease. At the initial treatment, all visible lesions were irradiated with GKS without upfront whole-brain radiation therapy. Thereafter, gadolinium-enhanced MR imaging was performed every 2 3 months, and new distant lesions were appropriately retreated with GKS. Patients were divided into groups according to numbers of tumors: Group A, single lesions (565 cases); Group B, 2 4 tumors (577 cases); and Group C, 5 10 tumors (366 cases). The differences in overall survival (OS) were compared between groups. Results. The median age of the patients was 66 years (range 19 96 years). There were 963 men and 545 women. The primary tumors were in the lung in 1114 patients, gastrointestinal tract in 179, breast in 105, urinary tract in 66, and other sites in 44. The overall mean survival time was 0.78 years (0.99 years for Group A, 0.68 years for Group B, and 0.62 years for Group C). The differences between Groups A and B (p < 0.0001) and between Groups B and C (p = 0.0312) were statistically significant. Multivariate analysis revealed significant prognostic factors for OS to be sex (poor prognostic factor: male, p < 0.0001), recursive partitioning analysis class (Class I vs Class II and Class II vs III, both p < 0.0001), primary site (lung vs breast, p = 0.0047), and number of tumors (Group A vs Group B, p < 0.0001). However, no statistically difference was detected between Groups B and C (p = 0.1027, hazard ratio 1.124, 95% CI 0.999 1.265). Conclusions. The results of this retrospective analysis revealed an upper CI of 1.265 for the hazard ratio, which was lower than the 1.3 initially set by the JLGK0901 study. The JLGK0901 study is anticipated to show noninferiority of GKS as sole treatment for patients with 5 10 brain metastases compared with those with 2 4 in terms of OS. (DOI: 10.3171/2010.8.GKS10838) Ke y Wo r d s metastatic brain tumor stereotactic radiosurgery Gamma Knife surgery whole-brain radiation therapy Abbreviations used in this paper: GK = Gamma Knife; GKS = GK surgery; HR = hazard ratio; JLGK = Japan Leksell Gamma Knife; JROSG = Japanese Radiation Oncology Study Group; KPS = Karnofsky Performance Scale; RPA = recursive partitioning analysis; RTOG = Radiation Therapy Oncology Group; SRS = stereotactic radiosurgery; UMIN = University Hospital Medical Information Network; WBRT = whole-brain radiation therapy. We have previously reported good clinical results of SRS alone using GKS for patients with multiple (even 5 or more) brain metastases. 3 11 According to the Japanese Radiation Oncology Study Group (JROSG) Study 99-1, reported by Aoyama et al. 1 in 2006, the efficacy of SRS alone for 1 4 brain metastases has been confirmed. However, we GK sur- 48 J Neurosurg / Volume 113 / December 2010

GKS alone for 1 10 brain metastases in JLGK0901-eligible cases geons have some doubts about this upper number limit of only 4 lesions. Therefore, the Japan Leksell Gamma Knife (JLGK) Society has planned a prospective multiinstitutional study (UMIN ID 0000001812, http://www. umin.ac.jp/) for selected patients to establish evidence of the efficacy of GKS as the sole treatment for brain metastases in patients with 5 10 brain lesions, as previously reported in a single GK center. 6 In this paper, we discuss the anticipated results based on a retrospective review of more than 1500 patients in 2 GK centers, focusing on statistical analysis of overall survival, which is the primary end point of the JLGK0901 study. Methods Of 3716 patients with metastatic brain tumors treated with GKS between 1998 and 2008 at the Chiba Cardiovascular Center and Katsuta Hospital by 2 of the authors (T.S. and M.Y.) as the chief surgeons, 1508 patients satisfied the following JLGK0901 inclusion criteria: 1) newly diagnosed brain metastases, 2) 1 10 brain lesions, 3) less than 10 cm 3 volume of the largest tumor, 4) less than 15 cm 3 total tumor volume, 5) no MR imaging findings of CSF dissemination, and 6) no impaired activities of daily living (< 70 of KPS score) due to extracranial disease. Data pertaining to these 1508 cases were retrospectively analyzed for the present study. At initial treatment, all lesions were irradiated with GKS without upfront WBRT. Gadolinium-enhanced MR imaging was performed every 2 3 months until disease progression. The dates of death were documented by the patients primary physicians. The standard peripheral doses were 22 Gy if the tumor volume was < 4.0 cm 3 and 20 Gy if it was 4.0 but < 10.0 cm 3. We changed the dose depending on tumor pathology, physical status, tumor location, extracranial disease status, and so on. The dose was also reduced if the tumor was located adjacent to SRS-risk organs (such as the brainstem, optic apparatus, cochlear, and facial nerves). The interval from the date of GKS treatment until the date of death (OS) was calculated by the Kaplan-Meier method. We divided patients into 3 groups according to number of tumors: Group A, single tumor (565 cases); Group B, 2 4 tumors (577 cases); and Group C, 5 10 tumors (366 cases). Differences between groups were assessed using the log-rank test. Covariates that emerged as significant on univariate analyses were included in the multivariate model and then verified by stepwise methods in the final model. We calculated the HR of the following covariates: GK center (Chiba, Mito), patient age ( 65 years, < 65 years), patient sex (male, female), primary tumor site (lung, breast, and others), extracranial disease status (controlled, active), pretreatment KPS score (< 70, 70), number of brain lesions (Group A, 1; Group B, 2 4; and Group C, 5 10), initial neurological symptoms (yes, no), total tumor volume ( 4 cm 3, > 4 cm 3 ), and RPA classification (Classes I, II, and III). Using multivariate analyses, we calculated the adjusted HR of Group C to Group B. All computations were planned, and the tests were 2-sided. A p value < 0.05 was considered to be statistically significant. All statistical analyses were performed using SAS 9.2 (SAS Institute Inc.). J Neurosurg / Volume 113 / December 2010 Results Patient characteristics are summarized in Table 1. The overall mean survival time was 0.78 years (0.99 years in Group A, 0.68 years in Group B, and 0.62 years in Group C), as shown in Fig. 1. The mean survival times for each subgroup of covariates are presented in Table 2. The association between prognostic factors and OS was evaluated by univariate analysis in Table 3. Multivariate analysis revealed 4 statistically significant poor prognostic factors: multiple brain metastases, male sex, lung cancer origin, and low RPA class. The strongest significant prognostic factor was the RTOG-RPA classification, as shown in Fig. 2. Table 4 lists the final results of multivariate analysis, which revealed no statistically significant differences between Groups B and C (p = 0.1027, HR 1.124, 95% CI 0.999 1.265). Among 1299 patients who died, 452 (34.8%) required salvage GKS for new distant lesions. For numerous new distant lesions and/or CSF dissemination, salvage WBRT was performed in 36 cases (2.8%). Discussion Background of the JLGK 0901 Study The JROSG 99-1 study reported by Aoyama et al. 1 found no significant differences in OS, neurological death, or functional preservation between SRS alone and SRS with upfront WBRT, although new lesions and tumor progression were observed significantly more frequently in patients treated with SRS alone. These results confirmed the efficacy of SRS alone for patients with 1 4 brain metastases. However, no evidence supporting SRS alone for multiple brain tumors (that is, more than 3 4) has yet been obtained. Indeed, SRS using the GK has been widely applied to multiple brain metastases, as reported especially by Japanese groups. 3 11 In previous papers, the first author of the present paper (T.S.) and colleagues 4,5,9 have insisted that factors limiting the application of GKS as sole treatment include not only the number but also the size of lesions, as well as the presence of CSF dissemination and total tumor volume. As the first step to bridging the gap between broad clinical application and limited evidence of efficacy for multiple brain tumors, we, the JLGK Society, undertook a prospective multiinstitutional study of GKS without upfront WBRT to establish evidence that such a treatment strategy is feasible for 5 10 brain metastases. Validity of Selected Cases in the JLGK0901 Study Based on the considerable experience and vast literature on SRS alone using the GK in treating multiple brain metastases in Japan, the following intracranial conditions are not considered to be good indications for GKS alone: 3 11 1) volume of the largest tumor of 10 cm 3 or more, 2) total tumor volume greater than 15 cm 3, 3) presence of CSF dissemination, 4) numerous (> 10) brain metastases, and 5) impaired activities of daily living (KPS score < 70) due to extracranial disease. Thus, the JLGK0901 study committee set the above 49

T. Serizawa et al. TABLE 1: Characteristics of 1508 cases* Characteristic Value GK center (no. of pts) Chiba 579 Mito 929 age (yrs) median 66 mean 64.7 range 19 96 sex (no. of pts) M 963 F 545 extracranial disease (no. of pts) controlled 256 active 1252 pretreatment KPS score median 100 range 60 100 primary tumor site (no. of pts) lung 1114 GI tract 179 breast 105 kidney 66 other 44 no. of brain lesions median 2 mean 3.2 range 1 10 total tumor volume (cm 3 ) median 2.7 mean 3.9 range 0.1 15.0 RTOG-RPA classification Class I 113 Class II 1365 Class III 30 * GI = gastrointestinal; pts = patients. as exclusion criteria. Patients with sarcoma, lymphoma, and unknown primary cancers were also excluded. The protocol stipulates follow-up, including enhanced MR imaging and neurological examinations, every 3 months at a minimum. Statistical Analysis of the JLGK0901 Study This prospective study is designed to prove the noninferiority of OS in patients with several (5 10) metastases (Group C) compared with OS in those with 2 4 lesions (Group B). With 80% statistical power to achieve a significant level of 0.025 for a 1-sided test with 2 years of recruitment and an additional 1-year follow-up, a planned total sample size of 1200 has been determined for this prospective trial. Noninferiority will be confirmed if the Fig. 1. Overall survival curves according to tumor number group. Mean survival time (MST) was 0.78 years (0.99 years in Group A, 0.68 years in Group B, and 0.62 years in Group C). There were significant differences between all 3 pairs of groups (p < 0.0001 for A vs B, p = 0.0312 for B vs C, and p < 0.0001 for A vs C). TABLE 2: Mean survival times by subgroup* Covariate Group MST (yrs) GK center Chiba Mito 0.821 0.745 age (yrs) 65 <65 0.751 0.841 sex M F 0.638 1.027 primary tumor site lung GI tract breast urinary tract other 0.810 0.537 0.942 1.005 0.780 extracranial disease controlled active 1.799 0.638 pretreatment KPS score 70 <70 0.797 0.249 no. of brain lesions Group A (1) Group B (2 4) Group C (5 10) 0.986 0.682 0.621 initial neurol Sx yes no 0.591 0.903 total tumor vol 4 cm 3 >4 cm 3 0.871 0.616 RTOG-RPA classification Class I Class II Class III 2.218 0.720 0.249 * MST = Mean Survival Time; neurol = neurological. 50 J Neurosurg / Volume 113 / December 2010

GKS alone for 1 10 brain metastases in JLGK0901-eligible cases TABLE 3: Results of univariate analysis Covariate Comparison p Value HR 95% CI GK center Chiba vs Mito 0.2024 1.078 0.961 1.209 age (yrs) 65 vs <65 0.0043 1.173 1.051 1.309 sex M vs F <0.0001 1.455 1.297 1.632 primary organ GI vs lung <0.0001 1.486 1.256 1.761 extracranial disease active vs controlled <0.0001 2.336 2.000 2.725 pretreatment KPS score <70 vs 70 <0.0001 2.955 2.040 4.279 no. of brain lesions 1 vs 2 4 <0.0001 1.318 1.161 1.495 5 10 vs 2 4 0.0316 1.166 1.013 1.342 initial neurol Sx yes vs no <0.0001 1.466 1.312 1.639 total tumor vol (cm 3 ) >4 vs 4 <0.0001 1.374 1.159 1.577 RTOG-RPA classification Class II vs I <0.0001 2.639 2.096 3.322 Class III vs II <0.0001 2.717 1.876 3.937 upper limit of the CI for the HR does not exceed 1.3, which is in accordance with the noninferiority margin. This trial has been registered by the Japanese ethics committee (UMIN). Expected Results of the JLGK 0901 Study Results of our final multivariate analysis model showed the HR for OS in Group C versus Group B in our sample of 1508 previously treated patients to be less than 1.3, allowing us to conclude that the 5 10 brain metastases group will not have results inferior to those of the 2 4 lesions group with a noninferiority margin of 7.5%. 2 Secondary end points of the JLGK 0901 study include neurological death, functional independence, emergence of new lesions, and higher function. These results will bring more useful information about GKS as sole treatment for up to 10 brain metastases. Fig. 2. Overall survival curves according RTOG-RPA classification. Mean survival time was 0.78 years (2.22 years in Class I, 0.72 years in Class II, and 0.25 years in Class III). There were significant differences between all pairs of groups (p < 0.0001). TABLE 4: Final results of multivariate analysis Covariate Comparison p Value HR 95% CI no. of tumors 1 vs 2 4 <0.0001 0.701 0.630 0.780 2 4 vs 5 10 0.1027 1.124 0.999 1.265 sex M vs F <0.0001 1.605 1.448 1.780 primary site breast vs lung 0.0047 1.474 1.176 1.848 other vs lung 0.0541 1.144 1.020 1.284 RTOG-RPA Class II vs I <0.0001 2.624 2.186 3.222 classification Class III vs II <0.0001 7.254 5.039 10.44 J Neurosurg / Volume 113 / December 2010 Conclusions The OS for patients with 5 10 brain metastases was almost the same as that of patients with 2 4 brain lesions in our retrospective study. We anticipate that these results will be confirmed by the JLGK 0901 study, which is a well-designed prospective multi-institutional controlled trial. If this study proves noninferiority between 2 4 and 5 10 brain metastases, Level 2 evidence of the efficacy of GKS alone for 5 10 brain metastases will be established. Disclosure The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper. Author contributions to the study and manuscript preparation include the following. Conception and design: Serizawa, Yamamoto, Sato, Higuchi. Acquisition of data: Serizawa, Yamamoto, Higuchi, Nagano, Kawabe, Matsuda, Ono, Saeki, Hatano, Hirai. Analysis and interpretation of data: Sato, Higuchi. Drafting the article: Serizawa. Critically revising the article: Serizawa. Reviewed final version of the manuscript and approved it for submission: all authors. Statistical analysis: Serizawa, Sato. Study supervision: Serizawa. 51

T. Serizawa et al. References 1. 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 2. Schoenfeld DA, Richter JR: Nomograms for calculating the number of patients needed for a clinical trial with survival as an endpoint. Biometrics 38:163 170, 1982 3. Serizawa T: Metastatic brain tumors: lung cancer, in Yamamoto M (ed): Japanese Experience with Gamma Knife Radiosurgery. Basel: Karger, 2008, pp 142 153 4. Serizawa T, Higuchi Y, Ono J, Matsuda S, Iuchi T, Nagano O, et al: Gamma Knife surgery for metastatic brain tumors from lung cancer without prophylactic whole brain radiation therapy, in Kondziolka D (ed): Radiosurgery, Vol 6. Basel: Karger, 2006, pp 186 198 5. Serizawa T, Higuchi Y, Ono J, Matsuda S, Nagano O, Iwadate Y, et al: Gamma Knife surgery for metastatic brain tumors without prophylactic whole-brain radiotherapy: results in 1000 consecutive cases. J Neurosurg 105 (Suppl):86 90, 2006 6. Serizawa T, Hirai T, Nagano O, Higuchi Y, Matsuda S, Ono J, et al: Gamma knife surgery for 1 10 brain metastases without prophylactic whole-brain radiation therapy: analysis of cases meeting the Japanese prospective multi-institute study (JLGK0901) inclusion criteria. J Neurooncol 98:163 167, 2010 7. Serizawa T, Iuchi T, Ono J, Saeki N, Osato K, Odaki M, et al: Gamma knife treatment for multiple metastatic brain tumors compared with whole-brain radiation therapy. J Neurosurg 93 (3 Suppl):32 36, 2000 8. Serizawa T, Ono J, Iichi T, Matsuda S, Sato M, Odaki M, et al: Gamma knife radiosurgery for metastatic brain tumors from lung cancer: a comparison between small cell and non small cell carcinoma. J Neurosurg 97 (5 Suppl):484 488, 2002 9. Serizawa T, Saeki N, Higuchi Y, Ono J, Iuchi T, Nagano O, et al: Gamma knife surgery for brain metastases: indications for and limitations of a local treatment protocol. Acta Neurochir (Wien) 147:721 726, 2005 10. Serizawa T, Yamamoto M, Nagano O, Higuchi Y, Matsuda S, Ono J, et al: Gamma Knife surgery for metastatic brain tumors. A 2-institute study in Japan. J Neurosurg 109 (Suppl): 118 121, 2008 11. Yamamoto M, Ide M, Nishio S, Urakawa Y: Gamma Knife radiosurgery for numerous brain metastases: is this a safe treatment? Int J Radiat Oncol Biol Phys 53:1279 1283, 2002 Manuscript submitted May 29, 2010. Accepted August 24, 2010. Address correspondence to: Toru Serizawa, M.D., Ph.D., Tokyo Gamma Unit Center, Tsukiji Neurological Clinic, 1-9-9 Tsukiji, Chuo-ku, Tokyo 1040045, Japan. email: gamma-knife.serizawa@ nifty.com. 52 J Neurosurg / Volume 113 / December 2010