Gamma knife stereotactic radiosurgery for renal cell carcinoma and melanoma brain metastasiscomparison

Transcription

1 See discussions, stats, and author profiles for this publication at: Gamma knife stereotactic radiosurgery for renal cell carcinoma and melanoma brain metastasiscomparison of dose response ARTICLE JANUARY 2013 CITATION 1 READS 92 9 AUTHORS, INCLUDING: hong-yiou Lin Beaumont Health System 5 PUBLICATIONS 30 CITATIONS Yoichi Watanabe University of Minnesota Twin Cities 87 PUBLICATIONS 425 CITATIONS SEE PROFILE SEE PROFILE Charles R. Watts Mayo Clinic Health System, La Crosse, United 25 PUBLICATIONS 376 CITATIONS SEE PROFILE Available from: Charles R. Watts Retrieved on: 09 April 2016

2 Jour. of Radiosurgery and SBRT, Vol. 2, pp Reprints available directly from the publisher Photocopying permitted by license only 2013 Old City Publishing, Inc. Published by license under the OCP Science imprint, a member of the Old City Publishing Group. CLINICAL INVESTIGATION Gamma knife stereotactic radiosurgery for renal cell carcinoma and melanoma brain metastases comparison of dose response Hong-Yiou Lin PhD 1, Yoichi Watanabe PhD 1, L. Chinsoo Cho MD MS 1, Jianling Yuan MD PhD 1, Hunt MD 2, Paul W. Sperduto MD MPP 3, Aviva Abosch MD PhD 2, Charles R. Watts MD PhD 2 and Chung K. Lee MD 1 1 Department of Radiation Oncology, University of Minnesota Medical School, Minneapolis, Minnesota, USA 2 Department of Neurosurgery, University of Minnesota Medical School, Minneapolis, Minnesota, USA 3 University of Minnesota Medical Center-Fairview Gamma Knife Center, Minneapolis, Minnesota, USA Correspondence to: Dr. Yoichi Watanabe, Department of Radiation Oncology, 420 Delaware St. SE MMC-494, University of Minnesota, Minneapolis, MN, 55455, USA, Phone: Fax: watan016@umn.edu (Received: December 29, 2012; Accepted: March 12, 2013) Background: Metastatic melanoma appears to have inferior local control (LC) than renal cell carcinoma (RCC) after stereotactic radiosurgery (SRS) to the brain. Objective: To retrospectively examine RCC vs. melanoma LC dose response. Methods: Follow-up data were available for 88 patients (RCC=38; melanoma=50) with 235 tumors (RCC=92; melanoma=143) treated with Gamma Knife SRS between Dec to Aug LC was compared among RCC vs. melanoma and then at each margin dose ( 18Gy, 20Gy, 22Gy, and 24Gy). Patient survival and toxicity were analyzed. Median follow-up was 9.8 months (RCC) and 5.4 months (melanoma). Results: Patient characteristics were similar between RCC vs. melanoma with respect to gender, age, KPS, GPA, lesions per patient, and tumor volume. For all margin doses, LC at 6 months was 98.6% (RCC) vs. 79.2% (melanoma). When broken down by margin dose, at 18 Gy (P<0.0001) and 20 Gy (P=0.02), RCC had better LC compared to melanoma. At 22 Gy, LC were similar between the two histologies (P=0.19). At 24 Gy, melanoma had better LC than RCC (P=0.02). Tumor volumes were similar between RCC vs. melanoma at each margin dose (P>0.05). Small melanoma tumors (<4ml) exhibited LC dose dependence. Median survival was 16.1 months (RCC) and 9.6 months (melanoma). Toxicity was not significantly different between the two histologies and margin doses. Conclusions: RCC has significantly better LC than melanoma after SRS. Higher doses could be used for melanoma tumors <4ml to improve melanoma LC. Keywords: Radiosurgery; melanoma; renal cell carcinoma; local control. Financial support and presentations: Dr. Yoichi Watanabe received a grant from the University of Minnesota Medical Center-Fairview to support this manuscript. Portions of this manuscript were presented as posters at the 14th Leksell Gamma Knife Society meeting, Quebec, Canada, May 2008; 9th Congress of the International Stereotactic Radiosurgery Society, Seoul, South Korea, June 7-11, 2009; 15th Leksell Gamma Knife Society meeting, Athens, Greece, May 17-20, 2010; and these authors contributed equally to this project. Journal of Radiosurgery and SBRT Vol

3 Yoichi Watanabe et al. 94th annual meeting of the American Radium Society, Las Vegas, Nevada, April 28-May 2, ABBREVIATIONS GKSRS, Gamma Knife stereotactic radiosurgery; GPA, graded prognostic assessment; KPS, Karnofsky performance scale; LC, local control; LGP, Leksell Gamma Plan; RCC, renal cell carcinoma; SEM, standard error of the mean; WBRT, whole brain radiation therapy. INTRODUCTION Renal cell carcinoma (RCC) and melanoma frequently metastasize to the brain [1-3]. These brain metastases are managed by surgery, glucocorticosteroids, chemotherapy, whole brain radiation therapy (WBRT), or stereotactic radiotherapy/radiosurgery (SRT/SRS) including Gamma Knife stereotactic radiosurgery (GKSRS) [4-6]. Historically, brain metastases carried a median survival of 2-3 months [7]. However, the use of SRS has improved local control (LC) of treated brain metastases [8]. SRS enables focused targeting of tumors with a high radiation dose while minimizing radiation to surrounding normal brain tissue. SRS is usually a minimally invasive procedure that is completed in one session. Investigators have reported worse local control for melanoma compared to RCC after SRS treatment LC for RCC is around 91%, while melanoma LC is only about 70.5% [9-25]. Thus, melanoma LC has much room for improvement while RCC LC after SRS is quite good. The reported differences in LC between RCC and melanoma after SRS treatment raise several questions. First, is RCC local control superior to melanoma at all margin doses? If not, then are there doses that would allow melanoma to have similarly good LC to RCC. Second, can superior LC in RCC be explained by smaller tumor size? If not, then maybe dose significantly affects local control. Third, does RCC or melanoma demonstrate significant dose dependence in terms of LC? If yes, then maybe prescribing a higher dose would result in improved local control. Fourth, are maximum margin doses allowed per RTOG recommendations being prescribed [26, 27]? If not, then there might be room to safely increase margin dose to improve local control while practicing within the standard of care. Answering the above four questions would allow physicians to better prescribe the margin dose for melanoma brain metastases in SRS. Ultimately, melanoma LC might be improved to the excellent level seen for RCC. As a result, we retrospectively analyzed the data of patients treated at the Gamma Knife Center of the University of Minnesota Medical Center-Fairview. MATERIALS AND METHODS Patient selection This study received approval from the University of Minnesota Institutional Review Board (Study # 0710M19564). Patients with RCC or melanoma brain metastases treated with GKSRS, from Dec to August 2012, for whom follow-up imagings were available, were included in this study. A total of 156 melanoma and RCC patients were treated from Dec to August 2012, but only 88 of them had follow-up imaging done at UMMC-Fairview and were included in this study. Patient survival data were obtained from hospital records or the Social Security Death Index, and duration of survival after radiosurgery was recorded. Stereotactic radiosurgery Leksell Gamma Knife Model 4C (Elekta AB, Stockholm, Sweden) was used for stereotactic radiosurgery of RCC or melanoma lesions. After local anesthesia and additional mild sedation as needed, an MRI-compatible frame was affixed to the skull. Patients then underwent MR imaging with gadolinium contrast for treatment planning. The GKSRS prescription dose was based on tumor size, and the prescription isodose level ranged from 45% to 90% [26, 27]. Taking RTOG protocol into consideration, tumors measuring less than 2 cm in diameter received a maximum of 24 Gy, 2 to 3 cm received at most 18 Gy, and those measuring 3 to 4 cm received a maximum of 15 Gy [27]. Some patients received steroids on the day of treatment to decrease the risk of early brain swelling based on physicians clinical judgment. Tumor volume quantification Tumor volume was approximated using the equation for an ellipsoid. The craniocaudal (a), anteriorposterior (b), and mediolateral (c) diameters of the tumors were measured on MR images, and the tumor volumes were computed using the ellipsoid equation: 194 Journal of Radiosurgery and SBRT Vol

4 Melanoma and RCC stereotactic radiosurgery dose response 4π abc. To confirm that tumor volumes could be 3 8 adequately approximated as ellipsoids, a subset of tumors was measured using both the Leksell Gamma Plan software and the ellipsoid equation, and the correlation coefficient between these two methods was 0.89 indicating good agreement between the two approaches. Tumor volume at the last follow-up visit was compared with volume at the time of GKSRS treatment. Necrosis or bleeding inside the contrast enhanced tumor was included in volume quantifications, but lesions clearly indicating hemorrhage or necrosis outside the tumor were excluded. Response to radiation was categorized using a modified RECIST criterion: complete response (no evidence of residual tumor), partial response ( 30% decrease in volume), stable disease ( 20% increase in volume), and progressive disease (>20% increase in volume) [28, 29]. LC was defined as complete response, partial response, and stable disease. For comparison with other published studies, crude local control was defined as LC at the end of the study. Evaluation of toxicity Adverse events were quantified using the National Cancer Institute s Common Terminology Criteria for Adverse Events (CTCAE) guidelines [30]. Side effects were considered as a result of radiation treatment if the signs and symptoms were new after treatment, dramatically worsened since treatment, or could not be attributed to new metastases. Radiation necrosis, hemorrhage, and brain edema were confirmed through MRI or CT imaging. Statistical analysis All statistical tests except multivariate Cox proportional hazards regression were performed with Graphpad Prism 5 (Graphpad Software Inc., La Jolla, CA). SPSS (IBM SPSS, Armonk, New York) was used for Cox proportional hazards regression. Sex and treatment modality (GKSRS with or without WBRT) were analyzed using Fisher s exact chi-square tests. Age, Karnofsky performance score (KPS), Graded Prognostic Assessment (GPA) [31], WBRT doses, GKSRS doses, average number of tumors per patient, tumor volumes, and complications were compared between RCC vs. melanoma using student s t-tests. Multivariate Cox regressions were used to identify factors significantly associated with LC or survival in both RCC and melanoma. LC and survival were evaluated using Kaplan- Meier curves and Mantel-Cox log-rank tests. Toxicity was compared using two-way ANOVA. Error bars represent standard error of the mean (SEM), and P<0.05 was deemed statistically significant. RESULTS Patient characteristics Follow-up imagings for 88 out of 156 patients were readily available in our hospital s computer system and only those 88 patients were included in this study. Patient characteristics are presented in Table 1. There were 38 RCC patients with 92 RCC tumors, and 50 melanoma patients with 143 melanoma tumors. RCC and melanoma patients had similar profiles in terms of gender (P=0.118), age (P=0.269), KPS (P=0.33), GPA (P=0.35), average number of tumors per patient (P=0.462), and tumor volume (P=0.888). Some patients were treated with GKSRS more than once due to recurrence or additional brain metastases and retreated patients were only counted once in tallying patient numbers in Table 1. Some patients received prior WBRT before SRS, but the proportion of patients receiving prior WBRT was not significantly different between the two histologies (P=0.111). For patients that received prior WBRT, the WBRT doses were similar between RCC and melanoma patients (P=0.347). WBRT doses for RCC were 35.3 Gy (mean) and 37.5 Gy (median) whereas doses for melanoma were 37 Gy (mean) and 37.5 Gy (median), respectively. Median duration of follow-up was 9.8 months for RCC (range months) and 5.4 months for melanoma (range months) patients. RCC 25 th and 75 th percentile follow-up imagings were 6.2 and 20.1 months, respectively. For melanoma, the 25 th and 75 th percentile follow-ups were 3.3 and 11.3 months, respectively. Median survivals were 16.1 months for RCC and 9.6 months for melanoma patients. Local control of RCC vs. melanoma Melanoma and RCC tumors had comparable mean (P=0.888) and median volumes as well as number of tumors per patient (P=0.462; Table 1). The two histologies were also treated with similar SRS margin doses (P=0.82), modality (P=0.111; SRS only vs. SRS+WBRT), and WBRT doses (P=0.347; Table 1). Despite these similarities, local control was significantly better for RCC compared to melanoma (P<0.0001; Figure 1). At 6 months, 98.6% of RCC vs. 79.2% of melanoma lesions exhibited LC. Crude LC Journal of Radiosurgery and SBRT Vol

5 Yoichi Watanabe et al. Table 1. Patient Characteristics Renal Melanoma Total number of patients Total number of tumors Tumor volume (cc) Mean ± SEM 2.54 ± ± 0.55 (P = 0.888) Median Gender (P = 0.118) Male Female Age Median (P = 0.269) Range KPS Median (P = 0.33) Range GPA Median (P = 0.35) 3 3 Range Tumors per patient Mean (P = 0.462) Median 2 2 Range Treatment Modality (P = 0.111) Gamma Knife only Gamma Knife + WBRT GKSRS margin dose Mean (P = 0.82) 19.3 ± ± 0.23 Median 20 (range 10 24) 20 (range 12 24) WBRT dose Mean (P = 0.347) 35.3 Gy 37 Gy Median 37.5 Gy 37.5 Gy Range Gy Gy Number of lesions treated at each SRS margin dose 18 Gy Gy Gy Gy Follow-up (months) Median 9.8 (range ) 5.4 (range ) 25th Percentile th Percentile Median Survival (months) Journal of Radiosurgery and SBRT Vol

6 Melanoma and RCC stereotactic radiosurgery dose response Figure 1. Local control for RCC and melanoma. RCC exhibited better LC. The percentage of RCC or melanoma tumors demonstrating local control was plotted versus time. RCC lesions had better local control than melanoma metastases. P<0.0001, Mantel-Cox log-rank test. (LC at the end of the study) was also superior for RCC (83.7%) compared to melanoma (75.5%). Local control at each margin dose for RCC vs. melanoma To better understand how treatment dose affected local control, we compared RCC vs. melanoma at each margin dose. The tumor volumes were similar between the two histologies at each margin dose compared (P=0.825 for 18 Gy; P=0.192 for 20 Gy; P=0.824 for 22 Gy; and P=0.686 for 24 Gy). At margin doses of 18 Gy or less, RCC LC was significantly better than melanoma (P<0.0001; Figure 2A). For 20 Gy margin dose, RCC LC remained superior to melanoma (P=0.02; Figure 2B). However, at 22 Gy, RCC and melanoma had statistically similar LC (P=0.19; Figure 2C). For the highest margin dose, 24 Gy, melanoma LC became significantly better than RCC (P=0.02; Figure 2D). Thus, with similar tumor volumes, RCC had better LC than melanoma when margin doses were 20 Gy. However, RCC no longer had superior LC than melanoma with margin doses 22 Gy. Local control for small (<4 ml) vs. large ( 4 ml) tumors Our multivariate analysis indicated that volume significantly affected local control (P=0.044; Table 2). The size of tumors was also a factor in deciding the margin dose prescribed. Thus, tumor volume must be taken into consideration to properly study the effect of dose on local control. Since we followed RTOG recommendations that tumors less than 2 cm in diameter could receive a maximum margin dose of 24 Gy, we decided to use 2 cm as a cut off to separate large and small tumors. 2 cm in diameter translated into a sphere or ellipsoid with volume of 4.2 ml π * * = For simplification purposes, 4 ml instead of 4.2 ml was chosen. Melanoma tumors smaller than 4 ml exhibited significant dose dependence with tumors treated with 24 Gy showing the best local control followed by 22 Gy & 20 Gy, and then 18 Gy having the worst LC (P=0.014; Figure 3A). Melanoma tumors 4 ml did not show significant dose dependence in terms of local control (P=0.55; Figure 3B). RCC tumors smaller than 4 ml did not have significant dose dependence (P=0.281; Figure 3C). There were not enough RCC tumors larger than 4 ml for statistical analysis. In summary, only melanoma tumors <4 ml exhibited significant dose dependence with higher doses resulting in better local control. Patient Survival There were no significant differences in survival between RCC and melanoma patients (P=0.06; Figure 4A). Since KPS was found to significantly affect survival in the Cox multivariate analysis (P=0.017; Table 2), we used KPS 90 as a cutoff to study patient survival. Patients with KPS 90 had significantly better survival than ones with KPS<90 (P<0.0001; Figure 4B). In terms of how local control affected patient survival, our Cox multivariate analysis indicated that local control did not significantly affect patient survival (P=0.264; Table 2). Journal of Radiosurgery and SBRT Vol

7 Yoichi Watanabe et al. Figure 2. RCC vs. melanoma LC at each margin dose. A, 18 Gy or less. RCC and melanoma tumors treated with margin dose 18 Gy had similar volumes (P=0.825), but RCC local control was significantly better than melanoma s (P<0.0001). B, 20 Gy. RCC and melanoma tumors treated with margin dose 20 Gy again had similar volumes (P=0.192), but RCC local control was superior to melanoma s (P=0.02). C, 22 Gy. RCC and melanoma metastases treated with margin dose 22 Gy were similar in volume (P=0.824), and the two histologies had similar local control (P=0.19). D, 24 Gy. RCC and melanoma tumors were similar in size (P=0.686), but melanoma tumors now had better LC than RCC (P=0.02), which was in contrast to lower doses. Volumes were compared using two tailed t-tests, and LC was analyzed with Mantel-Cox log-rank tests. Cox multivariate analysis for determinants of local control and survival To investigate the variables that significantly affected LC or patient survival, multivariate Cox regressions were performed. Histology (P<0.001), treatment modality (SRS only vs. SRS+WBRT; P=0.015), margin dose (P=0.023), number of metastases (P=0.033), and tumor volume (P=0.044) were found to significantly affect local control (Table 2). Only melanoma tumors were included in the margin dose multivariate Cox regression since Figure 3C showed no correlation between dose and LC for RCC tumors. Meanwhile, patient survival was significantly affected by only KPS scores and treatment modality (SRS only vs. SRS+WBRT) (Table 2). Since the Cox multivariate analysis indicated that number of metastases treated with SRS significantly affected LC, we separated patients that had 3 vs. 4 SRS treated brain metastases. Local control was significantly better when there were 3 metastases (P=0.004; Supplemental Figure 2). 198 Journal of Radiosurgery and SBRT Vol

8 Melanoma and RCC stereotactic radiosurgery dose response Figure 3. Dose dependence for small (<4 ml) vs. large ( 4 ml) tumors in each histology. A, small melanoma tumors. Melanoma tumors <4 ml demonstrated significant dose dependence in terms of local control (P=0.014). 18 Gy had the lowest while 24 Gy had the highest LC with 20 Gy and 22 Gy being in between. B, large melanoma tumors. Melanoma tumors 4 ml did not demonstrate significant LC dose dependence (P=0.55). Due to a smaller quantity of large melanoma tumors, only two groups were compared ( 18 Gy vs. 20 Gy) instead of at each margin dose. C, small RCC tumors. RCC tumors <4 ml did not show significant dose dependence (P=0.281). In fact, tumors treated with the highest margin dose (24 Gy) had the worst LC with other doses displaying similar LC. Local control was analyzed with Mantel-Cox log-rank tests. Journal of Radiosurgery and SBRT Vol

9 Yoichi Watanabe et al. Figure 4. Patient survival for RCC and melanoma. A, RCC vs. melanoma. Patient survival was similar between RCC and melanoma (P = 0.06). B, effects of KPS on survival. Patients with KPS 90 demonstrated significantly better survival than those with KPS<90 (P<0.0001). Survival was compared using the log-rank test. KPS, Karnofsky performance score. Adverse reactions Toxicities experienced by patients ranged from CTCAE grade 1 to 3, and no patients experienced grade 4 or 5 toxicities after SRS. First, we examined whether there was a correlation between margin dose and toxicity, and the data indicated no significant differences in toxicity between each margin dose (P=0.58; Figure 5A). Next, the differences in toxicity observed between RCC and melanoma patients were examined, and no significant differences were found (P=0.83; Figure 5B). Hemorrhage at the treatment site or radionecrosis were seen in 10.5% of the RCC and 8% of the melanoma patients. Effect of modality (SRS only vs. SRS+WBRT) on local control and survival Since our multivariate analysis indicated prior WBRT significantly affected both SRS local control and patient survival, the effect of WBRT on LC and survival was studied. When melanoma and RCC tumors were analyzed together, tumors that received 200 Journal of Radiosurgery and SBRT Vol

10 Melanoma and RCC stereotactic radiosurgery dose response Table 2. Multivariate Cox Regression P Value Variables Parameter LC Survival Gender male vs. female Age continuous Histology renal vs. melanoma <0.001*** Modality SRS only vs. SRS +WBRT 0.015* 0.014* Margin Dose # continuous 0.023* Max Dose continuous # of Metastases continuous 0.033* KPS continuous * Tumor Volume continuous 0.044* Local Control yes vs. no Not tested GPA 0 4 Not tested *, P<0.05; **, P<0.01; ***, P<0.001 #, only melanoma tumors were included in the analysis for margin dose. SRS+WBRT actually had worse LC than ones that received SRS only (P=0.0004; Supplemental Figure 1A). Similarly for survival, patients that received both WBRT and SRS had worse survival outcomes compared to ones that only received SRS (P=0.003; Supplemental Figure 1B). DISCUSSION Our retrospective analysis showed that RCC had better local control than melanoma when the two histologies were not compared by margin dose (P<0.0001; Figure 1). When RCC vs. melanoma were compared at each margin dose, it was found that at doses 22 Gy, melanoma LC was similar if not better than RCC s. The two histologies have similar tumor volumes at each margin dose. Small melanoma tumors (<4 ml 2 cm diameter) demonstrated significant dose dependence in terms of LC. Since RTOG recommendations allow for up to 24 Gy in tumors 2 cm, and melanoma LC needs to be improved, higher margin doses could be used for small melanoma tumors to increase LC while staying within RTOG recommendations [27]. Local control in RCC vs. melanoma Our finding that RCC has better LC than melanoma is in agreement with results from other institutions that have made similar retrospective comparisons [14, 24, 25]. However, previous publications did not compare RCC vs. melanoma at each margin dose as it was done in this study. Comparisons at each margin dose would better contrast LC of the two histologies at each dose used. Tumor sizes were similar between the two histologies at each dose, which meant that size could not account for differences in LC. Melanoma LC was inferior to RCC only with margin doses 20 Gy. Margin doses 22 Gy resulted in melanoma LC similar to or better than RCC. Thus, melanoma was more radioresistant than RCC, and higher doses could be used for melanoma to improve melanoma LC after SRS. Dose dependence of local control Small melanoma tumors (<4 ml 2 cm diameter) demonstrated significant dose dependence in terms of LC. Large melanoma tumors ( 4 ml) did not show significant dose dependence, and this might be due to a smaller sample size. Small RCC tumors (<4 ml) also did not show dose dependence, which could be due to either inadequate sample size or 18 Gy was sufficient for small RCC tumors and treatment with higher doses did not further improve local control. Our finding that melanoma LC demonstrated dosedependence was in agreement with other studies that examined brain metastases from lung, colon, and breast primaries [23, 32]. Journal of Radiosurgery and SBRT Vol

11 Yoichi Watanabe et al. Figure 5. Toxicity from SRS. Using CTCAE criteria, toxicity was categorized as grade 1 or grade 2-3. A, toxicity at each dose. The frequency of grade 1 and grade 2-3 toxicities were not significantly different between the four margin doses (P=0.58). B, toxicity between RCC vs. melanoma patients. The severity and frequency of toxicity were not significantly different between the two histologies (P=0.38). Toxicity was analyzed using two-way ANOVA. Four points from this study support a need of higher margin doses to treat small melanoma tumors than doses for RCC. First, melanoma LC is only 79.2% at 6 months while RCC LC was 98.6%. Second, small melanoma tumors (<4 ml) demonstrated significant LC dose dependence. The majority of melanoma tumors were small (<4 ml) in this study the median melanoma tumor volume was 0.55 ml, which was similar to other published studies [14, 20, 21, 23, 25]. Third, melanoma LC was not inferior to RCC with doses 22 Gy. Fourth, RTOG recommendations allow for up to 24 Gy for tumor diameters 2 cm, which is approximately 4 ml in volume [27]. 202 Journal of Radiosurgery and SBRT Vol

12 Melanoma and RCC stereotactic radiosurgery dose response Supplemental Figure 1. Effect of prior WBRT on SRS local control and patient survival. A, SRS local control. Patients that received prior WBRT had worse local control of the SRS treated tumors (P=0.0004). B, patient survival. Patients with prior WBRT had worse survival after SRS treatment (P=0.003). LC and survival were analyzed with log-rank tests. WBRT, whole brain radiation therapy. Supplemental Figure 2. Effect of number of tumors on local control. Patients with 1-3 tumors treated with SRS experienced better local control than those that had 4-8 metastases. P=0.004, log-rank test. SRS, stereotactic radiosurgery; WBRT, whole brain radiation therapy. Journal of Radiosurgery and SBRT Vol

13 Yoichi Watanabe et al. Toxicity Neither margin dose nor histology seemed to significantly correlate with toxicity. This might be due to the retrospective nature of this study because margin doses were partly chosen to minimize toxicity while still maximizing local control. The frequency of hemorrhage and radionecrosis were similar to a previously published study [14]. Survival Using the log-rank test, no significant differences in survival were observed between RCC vs. melanoma, but the P value was close to being significant (P=0.06). The median survival was longer for RCC (16.1 months) than melanoma (9.6 months). Our median RCC survival was similar to what has been reported by Brown et al. [15]. Our melanoma survival was slightly longer by about 3 months compared to previously published reports (Table 4). Our patient survival data was obtained using the social security data base, which should be accurate and comprehensive. Chang et al. reported that 66% of the melanoma patients succumbed due to neurological causes from either local or distant failures [14]. Our results suggested that higher doses for melanoma metastases would improve local control, but SRS would not affect distant control. Additionally, improvements in LC might improve patients quality of life by decreasing symptoms due to tumor enlargement. GPA is a tool to estimate survival in patients with brain metastases from RCC or melanoma as well as several other histologies [31]. However, our multivariate regression did not show GPA to significantly correlate with patient survival. This discrepancy could most likely be explained by differences in sample size 88 patients in this study vs reported by Sperduto et al. [31]. Lastly, our finding that KPS 90 was associated with improved survival was consistent with previous reports [8, 16]. Tumor volume measurements Tumor volume was approximated as an ellipsoid. This approach took into consideration all three axes of the tumor instead of only the maximal diameter. Correlation coefficient between the calculated ellipsoidal volume and the volume measured by using the Leksell Gamma Plan, which is based on outlining twodimensional tumor contours across the entire volume of a tumor was 0.89 indicating very good correlation. When the data was available, disease progression was tracked with sequential follow-up imaging, and tumor size from the last follow-up MRI was used in this study. Comparison with previous studies Table 3 summarized published RCC stereotactic radiosurgery studies. For comparison purposes, the median of the published studies was tabulated at the bottom of Table 3. In terms of RCC, this report had slightly more patients and a longer follow-up duration than the median of the previously published studies. The median margin dose used in this study (20 Gy) was also slightly higher than other reports (18 Gy), but the rate of local control appeared to be similar to other studies. The median survival (16.1 months) after radiosurgery was longer than the median of other studies (7.6 months). However, Brown et al. reported similar survival data at 14.2 months [15]. Table 4 summarized published melanoma SRS studies. For comparison purposes, the median values of the published studies were summarized at the bottom of Table 4. For melanoma, our patient number and follow-up duration were similar to the median values of previously published studies. The median margin dose of this study was 20Gy, which was slightly higher than the median of the previously published studies at 18Gy. The crude LC was 75.5% in this study, which was slightly better than the median of the previously published studies (70.5%). Schomas et al. used 25Gy median margin dose, and 6 month LC was 85% whereas Lwu et al used 21Gy margin dose, and 1 year LC was 75%--both better than the 70.5% LC with 18 Gy (median of previous studies). The survival after SRS was 9.6 months in this study, and was a little longer than the 6.5 months for the median of published melanoma SRS reports. Effect of prior WBRT on local control and survival Patients that received WBRT in this study had the WBRT done prior to SRS. The group that received prior WBRT usually had multiple tumors too numerous to treat with SRS. Thus, patients treated with SRS+WBRT had more tumors than patients treated with SRS only. Having more tumors might have resulted in poorer LC as was observed with the SRS+WBRT group in this study (Supplemental Figure 1A). This decreased LC with having more tumors is further supported by Supplemental Figure 2 where LC was lower when 4-8 tumors were treated with SRS vs. 1-3 tumors (P=0.004). Patients with 1-3 SRS treated tumors received a mean dose of 19.6 Gy while patients with 4-8 tumors were 204 Journal of Radiosurgery and SBRT Vol

14 Melanoma and RCC stereotactic radiosurgery dose response Table 3. Comparisons with Results of Other Published Studies Renal Cell Carcinoma Author & Year Margin dose; Gy (mean or median) # of patients Median FU (mo) Crude LC (%) Median OS (mo) Schoggl et al., (mean) Goyal et al., (median) Payne et al., 2000 # 18 (median) Wowra et al., (mean) 21 (median) 75 at least 6 months Sheehan et al., (median) 69 not stated 96 6 Chang et al., (median) Schomas et al., (median) (1 yr) 6 Brown et al., (mean) Kano et al., (median) 158 not stated Lo et al., (median) Lwu et al., (median) (1 yr) not stated 21.5 (mean) Median of previous studies Current study 20 (median) 19.3 (mean) #, Only 12 out of 21 patients had follow-up imaging performed (crude) 98.6 (6 mo) 16.1 Table 4. Comparisons with Results of Other Published Studies Melanoma Author & Year Gieger et al., 1997 Margin dose; Gy (mean or median) 18 (median) 18 (mean) # of patients Median FU (mo) Crude LC (%) Median OS (mo) Mori et al., (mean) Seung et al., (median) 55 not stated 89 (6 mo) 77 (1 yr) Chang et al., (median) Schomas et al., (median) (1 yr) 6 Gaudy-Marqueste et al., (median) 25.6 (mean) (6 mo) 69 (1 yr) Powell et al., (median) Liew et al., (median) Lo et al., (median) Lwu et al., (median) 21.5 (mean) (1 yr) not stated Median of previous studies Current study 20 (median) 19.4 (mean) (crude) 79.2 (6 mo) 9.6 Journal of Radiosurgery and SBRT Vol

15 Yoichi Watanabe et al. treated with 18.9 Gy, but the difference did not achieve statistical significance (P=0.13; t-test). Another reason why LC was lower in the SRS+WBRT group was that time spent receiving WBRT was not counted only the duration after SRS treatment was measured. In other words, the SRS+WBRT patients have had brain metastases longer than SRS only patients at the time of SRS treatment. Similarly, SRS+WBRT patients had worse survival than SRS only because of the higher tumor burden (P=0.003; Supplemental Figure 1B). Another reason why survival was lower in the SRS+WBRT group was that survival in this report measured duration lived after SRS treatment. The length of time SRS+WBRT patients spent receiving WBRT (prior to SRS) was not included in survival measurements. Weaknesses of this study This was a retrospective analysis of about 6 years of data from the University of Minnesota. Approximately half of our patients treated did not have follow-up imaging at UMMC-Fairview and thus could not be included in this study. This report could be strengthened by a larger sample size or data from other institutions. Ideally, a prospective randomized trial is better suited to answer questions about whether melanoma brain metastases demonstrate significant LC dose dependence to SRS treatment or the best dose to use for small melanoma tumors (<4 ml). Additional weaknesses included shorter duration of follow-up for melanoma than RCC, variations in length of followup between patients, inability to distinguish radiation necrosis from tumor recurrence on MRI, difficulty assessing volume for tumors of complex shape, and lack of detailed information on quality-of-life following treatment. CONCLUSION Melanoma brain metastases are known to be more radioresistant than RCC. This study showed that melanoma s increased radioresistance could be overcome by higher margin doses, and small (<4 ml 2 cm diameter) melanoma tumors, but not RCC, exhibited dose dependence. Furthermore, RTOG recommendations allow for up to 24 Gy in small melanoma tumors ( 2 cm diameter <4 ml volume). Therefore, small melanoma tumors (<4 ml) could be treated with higher doses, i.e. 22 Gy, to improve local control. Further research at different centers is needed to confirm these findings. ACKNOWLEDGEMENTS We are grateful to Drs. Kathryn Dusenbery, John Mullan, Terry Hood, Gregg Dyste, and Stephen Haines, for their treatment of patients whose data were analyzed in this study. We would especially like to thank Pam Caflisch, Gamma Knife nurse clinician, for assisting in patient care operation of the Gamma Knife Center. Assistance from former students, David D. Nelsen, Pamela Sooriyan, and Jami Johnson, is also greatly appreciated. DISCLOSURE The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper. REFERENCES 1. Lassman AB, DeAngelis LM. Brain metastases. Neurol Clin 2003;21(1):1 23, vii. 2. Marshall ME, Pearson T, Simpson W, Butler K, McRoberts W. Low incidence of asymptomatic brain metastases in patients with renal cell carcinoma. Urology 1990;36(4): Gay PC, Litchy WJ, Cascino TL. Brain metastasis in hypernephroma. J Neurooncol 1987;5(1): Elaimy AL, Mackay AR, Lamoreaux WT, Fairbanks RK, Demakas JJ, Cooke BS, Peressini BJ, Holbrook JT, Lee CM. Multimodality treatment of brain metastases: an institutional survival analysis of 275 patients. World J Surg Oncol 2011;9: Clarke JW, Register S, McGregor JM, Grecula JC, Mayr NA, Wang JZ, Li K, Gupta N, Kendra KL, Olencki TE, Cavaliere R, Sarkar A, Lo SS. Stereotactic radiosurgery with or without whole brain radiotherapy for patients with a single radioresistant brain metastasis. Am J Clin Oncol 2010;33(1): Bindal AK, Bindal RK, Hess KR, Shiu A, Hassenbusch SJ, Shi WM, Sawaya R. Surgery versus radiosurgery in the treatment of brain metastasis. J Neurosurg 1996;84(5): Markesbery WR, Brooks WH, Gupta GD, Young AB. Treatment for patients with cerebral metastases. Arch Neurol 1978;35(11): Mori Y, Kondziolka D, Flickinger JC, Logan T, Lunsford LD. Stereotactic radiosurgery for brain metastasis from renal cell carcinoma. Cancer 1998;83(2): Schoggl A, Kitz K, Ertl A, Dieckmann K, Saringer W, Koos WT. Gamma-knife radiosurgery for brain metastases of renal cell carcinoma: results in 23 patients. Acta Neurochir (Wien) 1998;140(6): Goyal LK, Suh JH, Reddy CA, Barnett GH. The role of whole brain radiotherapy and stereotactic radiosurgery on brain 206 Journal of Radiosurgery and SBRT Vol

16 Melanoma and RCC stereotactic radiosurgery dose response metastases from renal cell carcinoma. Int J Radiat Oncol Biol Phys 2000;47(4): Payne BR, Prasad D, Szeifert G, Steiner M, Steiner L. Gamma surgery for intracranial metastases from renal cell carcinoma. J Neurosurg 2000;92(5): Wowra B, Siebels M, Muacevic A, Kreth FW, Mack A, Hofstetter A. Repeated gamma knife surgery for multiple brain metastases from renal cell carcinoma. J Neurosurg 2002;97(4): Sheehan JP, Sun MH, Kondziolka D, Flickinger J, Lunsford LD. Radiosurgery in patients with renal cell carcinoma metastasis to the brain: long-term outcomes and prognostic factors influencing survival and local tumor control. J Neurosurg 2003;98(2): Chang EL, Selek U, Hassenbusch SJ, 3rd, Maor MH, Allen PK, Mahajan A, Sawaya R, Woo SY. Outcome variation among radioresistant brain metastases treated with stereotactic radiosurgery. Neurosurgery 2005;56(5):936 45; discussion Brown PD, Brown CA, Pollock BE, Gorman DA, Foote RL. Stereotactic radiosurgery for patients with radioresistant brain metastases. Neurosurgery 2008;62 Suppl 2: Kano H, Iyer A, Kondziolka D, Niranjan A, Flickinger JC, Lunsford LD. Outcome Predictors of Gamma Knife Radiosurgery for Renal Cell Carcinoma Metastases. Neurosurgery 2011;69(6): Gieger M, Wu JK, Ling MN, Wazer D, Tsai JS, Engler MJ. Response of intracranial melanoma metastases to stereotactic radiosurgery. Radiat Oncol Investig 1997;5(2): Mori Y, Kondziolka D, Flickinger JC, Kirkwood JM, Agarwala S, Lunsford LD. Stereotactic radiosurgery for cerebral metastatic melanoma: factors affecting local disease control and survival. Int J Radiat Oncol Biol Phys 1998;42(3): Seung SK, Sneed PK, McDermott MW, Shu HK, Leong SP, Chang S, Petti PL, Smith V, Verhey LJ, Wara WM, Phillips TL, Larson DA. Gamma knife radiosurgery for malignant melanoma brain metastases. Cancer J Sci Am 1998;4(2): Gaudy-Marqueste C, Regis JM, Muracciole X, Laurans R, Richard MA, Bonerandi JJ, Grob JJ. Gamma-Knife radiosurgery in the management of melanoma patients with brain metastases: a series of 106 patients without whole-brain radiotherapy. Int J Radiat Oncol Biol Phys 2006;65(3): Liew DN, Kano H, Kondziolka D, Mathieu D, Niranjan A, Flickinger JC, Kirkwood JM, Tarhini A, Moschos S, Lunsford LD. Outcome predictors of Gamma Knife surgery for melanoma brain metastases. Clinical article. J Neurosurg 2011;114(3): Powell JW, Chung CT, Shah HR, Canute GW, Hodge CJ, Bassano DA, Liu L, Mitchell L, Hahn SS. Gamma Knife surgery in the management of radioresistant brain metastases in high-risk patients with melanoma, renal cell carcinoma, and sarcoma. J Neurosurg 2008;109 Suppl: 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 2005;28(2): Lo SS, Clarke JW, Grecula JC, McGregor JM, Mayr NA, Cavaliere R, Kendra KL, Gupta N, Wang JZ, Sarkar A, Olencki TE. Stereotactic radiosurgery alone for patients with 1-4 radioresistant brain metastases. Med Oncol 2011;28 Suppl 1:S Lwu S, Goetz P, Monsalves E, Aryaee M, Ebinu J, Laperriere N, Menard C, Chung C, Millar BA, Kulkarni AV, Bernstein M, Zadeh G. Stereotactic radiosurgery for the treatment of melanoma and renal cell carcinoma brain metastases. Oncol Rep Andrews DW, Scott CB, Sperduto PW, Flanders AE, Gaspar LE, Schell MC, Werner-Wasik M, Demas W, Ryu J, Bahary JP, Souhami L, Rotman M, Mehta MP, Curran WJ, Jr. 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 2004;363(9422): Shaw E, Scott C, Souhami L, Dinapoli R, Kline R, Loeffler J, Farnan N. Single dose radiosurgical treatment of recurrent previously irradiated primary brain tumors and brain metastases: final report of RTOG protocol Int J Radiat Oncol Biol Phys 2000;47(2): Eisenhauer EA, Therasse P, Bogaerts J, Schwartz LH, Sargent D, Ford R, Dancey J, Arbuck S, Gwyther S, Mooney M, Rubinstein L, Shankar L, Dodd L, Kaplan R, Lacombe D, Verweij J. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). Eur J Cancer 2009;45(2): Therasse P, Arbuck SG, Eisenhauer EA, Wanders J, Kaplan RS, Rubinstein L, Verweij J, Van Glabbeke M, van Oosterom AT, Christian MC, Gwyther SG. New guidelines to evaluate the response to treatment in solid tumors. European Organization for Research and Treatment of Cancer, National Cancer Institute of the United States, National Cancer Institute of Canada. J Natl Cancer Inst 2000;92(3): Cirillo M, Venturini M, Ciccarelli L, Coati F, Bortolami O, Verlato G. Clinician versus nurse symptom reporting using the National Cancer Institute-Common Terminology Criteria for Adverse Events during chemotherapy: results of a comparison based on patient s self-reported questionnaire. Ann Oncol 2009;20(12): Sperduto PW, Kased N, Roberge D, Xu Z, Shanley R, Luo X, Sneed PK, Chao ST, Weil RJ, Suh J, Bhatt A, Jensen AW, Brown PD, Shih HA, Kirkpatrick J, Gaspar LE, Fiveash JB, Chiang V, Knisely JPS, Sperduto CM, Lin N, Mehta M. Summary Report on the Graded Prognostic Assessment: An Accurate and Facile Diagnosis-Specific Tool to Estimate Survival for Patients With Brain Metastases. Journal of Clinical Oncology 2012;30(4): 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 2009;23(2): Journal of Radiosurgery and SBRT Vol