J Neurosurg 111:825 831, 2009 The results of resection after stereotactic radiosurgery for brain metastases Clinical article Hi d e y u k i Ka n o, M.D., Ph.D., 1,3 Do u g l a s Ko n d z i o l k a, M.D., F.R.C.S.C., 1,3 Os c a r Zo r r o, M.D., 1,3 Jav i e r Lo b a t o -Po l o, M.D., 1,3 Jo h n C. Fl i c k i n g e r, M.D., 2,3 a n d L. Da d e Lu n s f o r d, M.D. 1,3 Departments of 1 Neurological Surgery, 2 Radiation Oncology, and the 3 Center for Image-Guided Neurosurgery, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania Object. Radiosurgery for brain metastasis fails in some patients, who require further surgical care. In this paper the authors goal was to evaluate prognostic factors that correlate with the survival of patients who require a resection of a brain metastasis after stereotactic radiosurgery (SRS). Methods. During the last 14 years when surgical navigation systems were routinely available, the authors identified 58 patients who required resection for various brain metastases after SRS. The median patient age was 54 years. Prior adjuvant treatment included whole-brain radiation therapy alone (17 patients), chemotherapy alone (9 patients), both radiotherapy and chemotherapy (10 patients), and prior resection before SRS (8 patients). The median target volumes at the time of SRS and resection were 7.7 cm 3 (range 0.5 24.9 cm 3 ) and 15.5 cm 3 (range 1.3 81.2 cm 3 ), respectively. Results. At a median follow-up of 7.6 months, 8 patients (14%) were living and 50 patients (86%) had died. The survival after surgical removal was 65, 30, and 16% at 6, 12, and 24 months, respectively (median survival after resection 7.7 months). The local tumor control rate after resection was 71, 62, and 43% at 6, 12, and 24 months, respectively. A univariate analysis revealed that patient preoperative recursive partitioning analysis classification, Karnofsky Performance Scale status, systemic disease status, and the interval between SRS and resection were factors associated with patient survival. The mortality and morbidity rates of resection were 1.7 and 6.9%, respectively. Conclusions. In patients with symptomatic mass effect after radiosurgery, resection may be warranted. Patients who had delayed local progression after SRS (> 3 months) had the best outcomes after resection. (DOI: 10.3171/2009.4.JNS09246) Ke y Wo r d s stereotactic radiosurgery gamma knife cancer brain metastases surgery Abbreviations used in this paper: KPS = Karnofsky Performance Scale; NSCLC = non small cell lung cancer; RPA = recursive partitioning analysis; SCLC = small cell lung cancer; SRS = stereotactic radiosurgery; WBRT = whole-brain radiation therapy. Br a i n metastases occur in 20 40% of all patients with systemic cancer. 2,6 Approximately 30 40% of patients with systemic cancer harbor a solitary brain tumor. 11,12,17 Patients who have 1 or 2 brain metastases appear to have better outcomes than those with 3 or more brain metastases. 12,17 Treatment options include corticosteroids, WBRT, SRS, and/or resection, alone or in combination. Radiosurgery is frequently used to manage metastases as initial treatment or as adjuvant therapy to resection and/or WBRT. 16 Stereotactic radiosurgery alone is a successful treatment strategy in 70 90% of patients. 1,3,7,19 If SRS fails, it can be repeated if the tumor is small. In others it may not be appropriate because of toxicity or diagnostic concerns. 3 Additional surgical removal for recurrent brain metastases has been used successfully after an initial surgical removal. 4,20 Delayed resection may be necessary in patients whose tumors progress despite SRS. Our purpose was to define the benefits and risks of delayed resection in this patient cohort. Methods Patient Population During a 14-year interval, 2884 patients with metastatic brain tumors underwent SRS at our center. We retrospectively identified 58 patients who underwent subsequent resection of a brain metastasis that progressed after SRS. The series included 31 men and 27 women with a median age of 54 years (range 24 80 years). Twenty-five patients (43%) had other brain metastases (range 2 6) 825
H. Kano et al. that were treated at the time of SRS. Eight patients had undergone resection of their brain metastases before radiosurgery (Table 1). Five patients underwent SRS twice for the same brain metastasis. Prior adjuvant management included WBRT alone (in 17 patients), chemotherapy alone (in 9), and both WBRT and chemotherapy (in 10). The primary histological types were NSCLC (in 21 patients [36%]), SCLC (in 3 [5%]), melanoma (in 15 [26%]), breast cancer (in 9 [16%]), gastrointestinal cancer (in 5 [9%]), renal cell carcinoma (in 4 [7%]), and sarcoma (in 1 patient [2%]). Tumors were located in the frontal lobe (in 27 patients [47%]), parietal lobe (in 7 [12%]), temporal lobe (in 7 [12%]), occipital lobe (in 8 [14%]), basal ganglia (in 2 [5%]), and cerebellum (in 7 [12%]). Twenty-six patients (45%) had extracranial metastases at the time of SRS. All patients were assigned a KPS score, which was obtained at the time of SRS and resection. Patients were also categorized according to the RPA of the Radiation Therapy Oncology Group. 8 The median interval between the primary site diagnosis and the diagnosis of the brain metastases was 22.0 months (range 0.5 196 months). TABLE 1: Characteristics in 58 patients with brain metastases Characteristic No. of Patients (%) sex male 31 (53) female 27 (47) median age in yrs (range) 54 (24 80) prior WBRT alone 17 (29) prior chemotherapy alone 9 (16) prior WBRT & chemotherapy 10 (17) no prior WBRT or chemotherapy 22 (38) resection pre-srs total 6 (10) partial 2 (3) tumor location frontal 27 (47) parietal 7 (12) temporal 7 (12) occipital 8 (14) basal ganglia 2 (5) cerebellum 7 (12) primary cancer lung (NSCLC) 21 (36) lung (SCLC) 3 (5) melanoma 15 (26) breast 9 (16) GI tract 5 (9) kidney 4 (7) sarcoma 1 (2) Radiosurgical Technique Our radiosurgical technique has been described in detail in previous reports. 3 In brief, patients underwent application of an imaging-compatible stereotactic head frame after administration of local anesthesia supplemented by intravenous sedation. High-resolution MR imaging then was performed. Images of the tumor were obtained using contrast-enhanced volume acquisition sequences, supplemented by T2-weighted images. The SRS target was defined as the contrast-enhanced tumor volume. The median KPS score and RPA class were 90 (range 60 100) and 2 (range 1 2), respectively, at the time of SRS. The median tumor volume at the time of SRS was 7.7 cm 3 (range 0.5 24.9 cm 3 ). The median tumor diameter and maximum edema measurements at the time of SRS were 2.3 cm (range 0.4 5.2 cm) and 3.5 cm (range 0.4 9.1 cm), respectively (Table 2). A median of 5 isocenters (range 1 12 isocenters) was used for dose planning. The median prescription dose delivered to the tumor margin was 16.5 Gy (range 12 20 Gy). The maximum dose varied from 24 to 50 Gy (median 33.0 Gy). Stereotactic radiosurgery was performed using a model U, B, C, or 4-C Leksell Gamma Knife (Elekta, Inc.). Surgical Indications Patients were evaluated clinically by using MR imaging at intervals of 1 3 months after SRS. The decision for craniotomy and resection after SRS was based on evidence of clinical deterioration and associated imaging progression. Neuroimaging indications included an enlarging lesion, hemorrhage, and symptomatic mass effect unresponsive to medical management that included corticosteroids. Patients were reevaluated clinically and by imaging at intervals of 1 3 months after resection. The follow-up MR images were compared with the preoperative images and tumor dimensions were measured in axial, sagittal, and coronal planes. The edema size was measured by the largest dimension at the time of SRS or later resection. The median follow-up intervals after the initial cancer diagnosis, from initial SRS, and from resection (in months) were 45.4 (range 8.6 226), 13.3 (range 0.7 115) and 7.6 (range 0.03 105), respectively. Statistical Analysis For statistical analysis we constructed Kaplan-Meier plots for survival and local tumor control rate using the dates of diagnosis, SRS, and resection. The local tumor control rate and overall survival time were calculated from the day of resection using the Kaplan-Meier method. We conducted our statistical analyses using the TABLE 2: Tumor characteristics at the time of SRS and resection Mean/Median (range) Characteristic at SRS at Resection tumor vol (cm 3 ) 8.9/7.7 (0.5 24.9) 18.1/15.5 (1.3 81.2) max tumor diameter 23.4/22.9 (4.0 52.0) 31.5/30.0 (12 120) (mm) edema vol (ml) 39.8/35.0 (0.2 193) 80.1/59.0 (4.4 376) max edema diameter (mm) 37.7/35.0 (4.0 91.0) 57.3/50.7 (16.0 130) 826
Resection after radiosurgery Mann-Whitney U-test and the Kruskal-Wallis test to assess factors that might influence the length between SRS and resection (Table 3). Univariate analysis was performed on the Kaplan-Meier curves using the log rank statistic with p < 0.05 set as significant. We performed multivariate analysis using the Cox proportional hazards model with p < 0.05 set as significant. Standard statistical processing software (version 15.0, SPSS, Inc.) was used. This retrospective study was approved by the University of Pittsburgh Institutional Review Board. We performed a univariate analysis by using the logrank test to assess factors that might influence survival. The following variables were assessed: age ( 65 vs < 65 years), sex (male vs female), KPS score before surgery ( 70 or < 70), status of primary cancer (stable vs active), RPA at the time of SRS (Class 1 vs 2), RPA at the time of surgery (Class 1 vs 2 vs 3), prior SRS for the same tumor (yes vs no), prior surgery for the same tumor (yes vs no), prior WBRT (yes vs no), prior chemotherapy (yes vs no), prior WBRT and/or chemotherapy (yes vs no), tumor resection rate (partial vs total), primary cancer site (melanoma vs others, lung vs others, and breast vs others), tumor location (frontal vs others), distant tumor recurrence after SRS (yes vs no), tumor volume at the time of SRS ( 8 vs < 8 cm 3 : median tumor volume 7.7 cm 3 ), margin dose of SRS ( 18 vs < 18 Gy: margin dose of 18 Gy was standard dose for brain metastasis), edema diameter ( 3.5 vs < 3.5 cm: median edema diameter 3.5 cm), multiple brain metastases at the time of surgery (yes vs no), the TABLE 3: Relationship between the interval from radiosurgery and surgery, and variables Characteristic Variable p Value* age at SRS >65 yrs vs <65 yrs 0.774 sex male vs female 0.646 status of primary cancer stable vs active 0.047 RPA at the time of SRS 1 vs 2 0.077 prior op yes vs no 0.944 prior SRS yes vs no 0.893 prior WBRT yes vs no 0.161 prior chemotherapy yes vs no 0.180 prior WBRT &/or chemotherapy yes vs no 0.032 primary cancer melanoma vs others 0.010 lung vs others 0.137 breast vs others 0.0004 tumor location frontal vs others 0.773 multiple brain metastases pre-srs yes vs no 0.525 MRI finding pre-srs solid vs necrosis, 0.825 mixed regular vs irregular 0.089 margin tumor vol at SRS (cm 3 ) >8 vs <8 0.003 margin dose of SRS (Gy) >18 vs <18 0.743 edema size at SRS (cm) >3.5 vs <3.5 0.827 * Based on the Mann-Whitney U-test. Significant. interval between SRS and tumor recurrence after SRS ( 3 vs < 3 months), MR imaging findings before surgery (solid vs mixed), and pathological findings (all tumor vs mixed) (Table 4). Results At a median follow-up of 7.6 months, 8 patients (14%) were living and 50 patients (86%) had died. The overall survival after the initial primary diagnosis was 98, 59, and 43% at 1, 3, and 5 years, respectively. The median survival after the initial diagnosis was 43.5 months. The survival after SRS was 93, 62, and 28% at 6, 12, and 24 months, respectively, with a median survival after SRS of 14.5 months (Fig. 1). The survival after subsequent resection was 65, 30, and 16% at 6, 12, and 24 months, respectively. The median survival after resection was 7.7 months (Fig. 1). Cause of Death Thirty-one patients (62%) died of systemic disease progression. Nineteen patients (38%) died of brain metastasis progression, including 15 who died of subsequent metastasis progression after both SRS and resection. Four patients died of progression of other brain metastases. Patient Characteristics Affecting Time to Resection The median interval between SRS and resection was 7.2 months (range 0.3 27.7 months). In the Mann-Whitney U-test, active primary cancer site (p = 0.047) and larger TABLE 4: Univariate analysis of survival after resection Characteristic Variable p Value* age at time of op >65 yrs vs <65 yrs 0.982 sex male vs female 0.423 preop KPS score >70 vs <70 0.016 status of primary cancer stable vs active 0.023 RPA at time of op 1 vs 2 vs 3 <0.0001 prior op yes vs no 0.139 prior SRS for same tumor yes vs no 0.760 prior WBRT yes vs no 0.651 prior chemotherapy yes vs no 0.935 prior WBRT &/or chemotherapy yes vs no 0.837 tumor resection rate partial vs total 0.563 primary cancer melanoma vs others 0.883 NSCLC vs others 0.416 breast vs others 0.205 tumor location frontal vs others 0.603 distant tumor recurrence yes vs no 0.588 multiple brain metastases yes vs no 0.994 preresection interval btwn SRS & resection >3 mos vs <3 mos 0.007* preresection MRI finding solid vs mixed 0.093 pathological finding all tumor vs mixed 0.787 * Log-rank test. Significant 827
H. Kano et al. Fig. 2. Box-and-whisker plots of the interval between SRS and resection for all primary cancer sites. Horizontal bars inside boxes indicate median values. Error bars indicate farthest points that are not outliers. In the Mann-Whitney U-test, for melanoma there was a significantly shorter interval between SRS and resection, compared with NSCLC and breast cancer. For breast cancer there was a significantly longer interval between SRS and resection, compared with NSCLC and melanoma. GI = gastrointestinal; RCC = renal cell carcinoma. Fig. 1. Upper: Kaplan-Meier estimate of survival curve after the initial SRS. Lower: Kaplan-Meier estimate of survival curve after resection. tumor volume (p = 0.003) at the time of SRS were significantly associated with a shorter interval between SRS and later resection (Table 3). However, breast cancer was significantly associated with a longer interval between SRS and resection (p = 0.0004). The median times from SRS to resection were 6.7 versus 7.0 months for active primary cancer site versus controlled primary cancer site and 4.0 versus 8.3 months for a tumor volume of 8 versus < 8 cm 3 at the time of SRS. Primary Cancer Site Affecting Time to Resection, not Patient Survival In the Kruskal-Wallis test, we analyzed the relationship between each primary cancer origin (all cancer types) and interval from SRS to resection. Primary cancer origin was significantly associated with survival (p = 0.001). On the other hand, the SRS target volume, margin dose, and pre-srs edema size were not significantly associated with time between SRS and resection. The 3 major primary cancers in this series were NSCLC (in 21 patients), melanoma (in 15), and breast cancer (in 9). The median interval between SRS and resection was 8.7 months (range 1.6 24.5 months) for NSCLC, 3.1 months (range 0.3 11.0 months) for melanoma, and 17.6 months (range 7.0 27.7 months) for breast cancer. In the interval between SRS and resection, there were significant differences for each primary cancer site (p = 0.002 for NSCLC vs melanoma, p = 0.010 for NSCLC vs breast cancer, p = 0.0001 for melanoma vs breast cancer; Mann- Whitney U-test) (Fig. 2). In the Kruskal-Wallis test, we analyzed the relationship between each primary cancer origin (all cancer types) and survival. Primary cancer origin was not associated with survival (p = 0.138). Indications for Resection The reasons for resection after failed SRS included clinical deterioration caused by progressive neurological signs or symptoms (in 50 patients), imaging evidence of tumor progression (in 3), or intractable seizures (in 5). Clinical deterioration and/or tumor progression was detected at a median of 4.6 months after SRS (range 0.3 24.6 months). Twenty-three patients (40%) had multiple brain metastases before resection. Results of Resection The median KPS score and RPA Class were 80 (range 50 90) and 2 (range 1 3), respectively, at the time of resection. The median tumor volume at the time of resection was 15.5 cm 3 (range 1.3 81.2 cm 3 ). The median tumor diameter and maximum edema diameter at the time of resection were 3.0 cm (range 1.2 12 cm) and 5.1 828
Resection after radiosurgery Fig. 3. Kaplan-Meier estimate of local tumor progression curve in all patients after resection. cm (range 1.6 13 cm), respectively (Table 2). One month after resection, the median KPS score improved to 90. Twelve patients underwent a partial tumor resection and 46 patients underwent a total tumor resection. The local tumor progression rate after resection was 71, 62, and 43% at 6, 12, and 24 months, respectively (Fig. 3). Histological examinations of the 58 resected tumors confirmed residual tumor in 32 (55%) and mixed tumor and radiation effect in 26 (45%). Despite resection, the tumor recurred locally in 18 patients. Nineteen patients developed a new distant brain metastasis at a median of 1.9 months (range 0.03 12.6 months) after resection. New distant brain metastases after SRS were not associated with patient survival (p = 0.588). The mortality rate after resection was 1.7% (1 patient). This patient with melanoma died of a remote intracranial hemorrhage 2 days after craniotomy. The morbidity rate associated with resection was 6.9% (4 patients). Two patients developed a new hemiparesis, 1 had dysphasia and hemiparesis, and 1 had transient ataxia. Management After Resection Among the 18 patients with local tumor progression after resection, 3 patients did not receive further treatment in the face of rapid progression of systemic disease. One patient underwent additional resection 3 months later, 2 patients underwent WBRT, 4 underwent additional SRS, and 2 patients received chemotherapy. Five patients underwent radiation therapy and boost repeat SRS, 2 patients underwent another resection and boost RT, and 1 patient underwent another surgical removal and boost SRS 2 months after resection. Predictors of Improved Overall Survival At the time of resection, 22 patients (38%) were in RPA Class 1, 30 (52%) were in RPA Class 2, and 6 (10%) were in RPA Class 3. The median survival of patients in RPA Class 1 was 9.4 months (95% CI 6.8 12.0 months) Fig. 4. Kaplan-Meier estimate of survival curve after resection by the RPA classification. Note the significant impact of RPA class on the overall survival after resection. after resection, 7.6 months for RPA Class 2 (95% CI 5.6 9.6 months), and 3.1 months for RPA Class 3 (95% CI 0.5 5.7 months). Patients in RPA Class 1 had a better prognosis than those in RPA Class 2 (p = 0.007) or RPA Class 3 (p < 0.0001) (Fig. 4) (Table 5). Fifteen patients (26%) underwent resection for the same tumor < 3 months after SRS, and 43 patients (74%) had resection > 3 months after radiosurgery. The median interval between SRS and resection in the entire series was 7.7 months (range 0.3 27.7 months). The median survival times of those who underwent resection 3 months versus < 3 months were 8.8 months (95% CI 7.0 10.6 months) and 5.8 months (95% CI 4.8 6.8 months), respectively (p = 0.007) (Fig. 5). Patients undergoing a resection 3 months after SRS had 6-, 12- and 24-month overall survivals of 74, 39, and 20%, respectively. No patient having a resection within 3 months of radiosurgery lived > 1 year. We also found that stable primary extracranial cancer status before resection (p = 0.023) and KPS score 70 (p = 0.016) were significantly associated with better survival (Table 4). We could not identify a relationship between tumor histology and survival after resection. TABLE 5: The scoring system used to determine the patient s RPA class* RPA Class No. of Patients (%) Postop MST (mos) Death RR p Value p Value 1 22 (38) 9.4 1.0 NA NA 2 30 (52) 7.6 2.4 0.007 0.009 3 6 (10) 3.1 3.1 <0.0001 0.0002 * MST = median survival time; NA = not available; RR = relative risk. Log-rank test. Cox proportional hazard. Referent group to which other groups are compared. 829
H. Kano et al. Interval Between the Initial Treatment and Resection Bindal et al. 4 reported that reoperation for recurrent brain metastases after the initial resection prolonged survival and improved quality of life. Their median time from first craniotomy to diagnosis of recurrence (time to recurrence) was 6.7 months (range 1.2 28.8 months). The time to recurrence after initial resection was significantly associated with overall survival in a multivariate analysis (p = 0.008). The median survival for patients eligible for reoperation was 11.5 months. The patients who underwent a second reoperation survived a median of 8.6 additional months versus 2.8 months for those who did not undergo a second operation (p < 0.0001). They identified the following 5 prognostic factors: the presence of active systemic disease, KPS score 70, time to recurrence < 4 months, age 40 years, and primary tumor type of breast or melanoma. Vecil et al. 23 reported that the median time from SRS of an index lesion to craniotomy for its removal was 5.2 months (range 0.3 34.4 months) in those patients who required it. The time from SRS to resection was not associated with overall survival in that study (p = 0.76). In our series, the median interval between SRS and resection was 7.1 months (range 0.3 27.7 months). A shorter interval between SRS and resection was significantly associated with melanoma (p = 0.011), larger tumor volume at the time of SRS (p = 0.0001), and larger edema size at the time of SRS (p = 0.007) (Table 3). Fig. 5. Kaplan-Meier estimate of overall survival curve after resection by the subgroup of the interval between SRS and tumor recurrence after SRS ( 3 vs < 3 months). The interval between SRS and tumor recurrence after SRS of < 3 months was significantly associated with poor survival (p = 0.001). Discussion The most common treatment approaches for patients with brain metastases are resection, WBRT, and SRS, either alone or in combination. Although WBRT is considered by some as a standard of care, the role of SRS is continually increasing. Compared with SRS alone, the addition of WBRT to SRS did not improve survival for patients with < 4 brain metastases, but intracranial relapse occurred more frequently in those who did not receive WBRT. 1 Stereotactic radiosurgery advocates believe that salvage SRS is a better option than repeat WBRT. 7,10,13 Initial resection also has an important role in carefully selected patients with brain metastasis who then receive postoperative WBRT. 18,22 The presence or absence of active extracranial cancer strongly influences the results of clinical studies of all brain metastases. 8,9,18 Risk of SRS There are some risks associated with SRS. 15 Petrovich et al. 19 reported that 4.7% of patients required resection for symptomatic radiation effects. Brown et al. 5 reported a 5% rate of radiation necrosis, a 2% rate of leukoencephalopathy, and a 12% rate of peritumoral edema. Additional SRS We previously reported that repeat radiosurgery can be performed for recurrent tumors with minimal CNS toxicity and with reasonable tumor control. 3 Tumor control was significantly better (p = 0.01) for benign tumors (6 of 6 [100% actuarial rate at 4 years]) than for malignant tumors (7 of 20 [35% actuarial rate at 3 years]). The retreatment volume for radiosurgery correlated significantly with the probability of neurological decline (any cause) (p = 0.02). 3 We continue to advocate repeat radiosurgery for tumors that remain small or are in a critical location if initial SRS fails. Such failure is identified when progress is noted serially on imaging studies typically spaced over several months. Resection After SRS Vecil et al. 23 reported on 61 patients who underwent resection after radiosurgery for brain metastases. They noted that resection favorably affected local control and overall survival. In this study, the single most important factor influencing duration of survival was the patient s RPA classification. The median survival time of RPA Class 2 patients was 11.1 months, and for 7 RPA Class 3 patients it was 2.4 months. However, most patients were in RPA Class 2 (72%), and the other grades were underrepresented. In a retrospective series of 32 patients who underwent resection after SRS, Truong et al. 21 reported that patients who underwent SRS and required resection had a better prognosis than patients who did not undergo subsequent resection (p < 0.0001). It is likely that selection bias was present, as patients in better condition or those who had lived longer underwent resection. In our series the median survival time for RPA Class 1 patients was 9.1 months, for RPA Class 2 patients it was 7.7 months, and for Class 3 patients it was 3.1 months, although our study was retrospective in design and patients were selected for treatment. Vecil et al. 23 reported that the relative risk of death for RPA Class 1 versus Classes 2 and 3 was 5.0 (p = 0.03) and 17.4 (p < 0.001). They found that the RPA Class 3 patients had poor outcomes after resection. In our series the relative risk of death for RPA Class 1 versus Classes 2 and 3 was 2.1 (p = 0.024) and 4.2 830
Resection after radiosurgery (p = 0.001). Bindal et al. 4 found that the interval between the initial resection and local recurrence of < 4 months was an important prognostic factor (p = 0.008). They did not use the RPA classification. We found an additional factor that was an important predictor of survival: an interval between SRS and resection of < 3 months. Survival Benefit for Patients Undergoing Repeat Resection After SRS In 1200 patients with brain metastases who underwent WBRT and/or chemotherapy Gaspar et al. 8 reported that the median survival of RPA Classes 1, 2, and 3 was 7.1, 4.2, and 2.3 months, respectively. Our series had slightly higher median survivals in each RPA class than Gaspar s series. Our median survival of 7.7 months for patients who underwent reoperation after SRS is similar to a median survival of 6 10 months in the randomized control trials that compared resection plus WBRT with WBRT only, suggesting that prognosis of resection after SRS is similar to that after an initial resection plus WBRT. 14,18,22 Conclusions Carefully selected patients can attain a survival benefit after resection of a brain metastasis that has progressed despite prior radiosurgery. The RPA class, systemic disease status, and a short interval from radiosurgery to resection are the most important predictors of outcome. Disclosure The work described in this report was funded by a grant (to H.K.) from the Osaka Medical Research Foundation for Incurable Diseases. Drs Lunsford, and Kondziolka are consultants with AB Elekta. Dr. Lunsford is a stockholder in AB Elekta. 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. Arnold SM, Patchell RA: Diagnosis and management of brain metastases. Hematol Oncol Clin North Am 15:1085 1107, vii, 2001 3. Bhatnagar A, Heron DE, Kondziolka D, Lunsford LD, Flickinger JC: Analysis of repeat stereotactic radiosurgery for progressive primary and metastatic CNS tumors. Int J Radiat Oncol Biol Phys 53:527 532, 2002 4. Bindal RK, Sawaya R, Leavens ME, Hess KR, Taylor SH: Reoperation for recurrent metastatic brain tumors. J Neurosurg 83:600 604, 1995 5. Brown PD, Brown CA, Pollock BE, Gorman DA, Foote RL: Stereotactic radiosurgery for patients with radioresistant brain metastases. Neurosurgery 51:656 657, 2002 6. Cairncross JG, Kim JH, Posner JB: Radiation therapy for brain metastases. Ann Neurol 7:529 541, 1980 7. Flickinger JC, Loeffler JS, Larson DA: Stereotactic radiosurgery for intracranial malignancies. Oncology (Williston Park) 8:81 86, 94, 97 98, 1994 8. Gaspar L, Scott C, Rotman M, Asbell S, Phillips T, Wasserman T, et al: Recursive partitioning analysis (RPA) of prognostic factors in three Radiation Therapy Oncology Group (RTOG) brain metastases trials. Int J Radiat Oncol Biol Phys 37:745 751, 1997 9. Gaspar LE, Scott C, Murray K, Curran W: Validation of the RTOG recursive partitioning analysis (RPA) classification for brain metastases. Int J Radiat Oncol Biol Phys 47:1001 1006, 2000 10. Larson DA, Bova F, Eisert D, Kline R, Loeffler J, Lutz W, et al: Current radiosurgery practice: results of an ASTRO survey. Task Force on Stereotactic Radiosurgery, American Society for Therapeutic Radiology and Oncology. Int J Radiat Oncol Biol Phys 28:523 526, 1994 11. Lohr F, Pirzkall A, Hof H, Fleckenstein K, Debus J: Adjuvant treatment of brain metastases. Semin Surg Oncol 20:50 56, 2001 12. Lutterbach J, Bartelt S, Ostertag C: Long-term survival in patients with brain metastases. J Cancer Res Clin Oncol 128: 417 425, 2002 13. Mathieu D, Kondziolka D, Flickinger JC, Fortin D, Kenny B, Michaud K, et al: Tumor bed radiosurgery after resection of cerebral metastases. Neurosurgery 62:817 824, 2008 14. Mintz AH, Kestle J, Rathbone MP, Gaspar L, Hugenholtz H, Fisher B, et al: A randomized trial to assess the efficacy of surgery in addition to radiotherapy in patients with a single cerebral metastasis. Cancer 78:1470 1476, 1996 15. Modha A, Shepard SR, Gutin PH: Surgery of brain metastases is there still a place for it? J Neurooncol 75:21 29, 2005 16. Noel G, Proudhom MA, Valery CA, Cornu P, Boisserie G, Hasboun D, et al: Radiosurgery for re-irradiation of brain metastasis: results in 54 patients. Radiother Oncol 60:61 67, 2001 17. Patchell RA, Tibbs PA, Regine WF, Dempsey RJ, Mohiuddin M, Kryscio RJ, et al: Postoperative radiotherapy in the treatment of single metastases to the brain: a randomized trial. JAMA 280:1485 1489, 1998 18. Patchell RA, Tibbs PA, Walsh JW, Dempsey RJ, Maruyama Y, Kryscio RJ, et al: A randomized trial of surgery in the treatment of single metastases to the brain. N Engl J Med 322: 494 500, 1990 19. Petrovich Z, Yu C, Giannotta SL, O Day S, Apuzzo ML: Survival and pattern of failure in brain metastasis treated with stereotactic gamma knife radiosurgery. J Neurosurg 97:499 506, 2002 20. Sundaresan N, Sachdev VP, DiGiacinto GV, Hughes JE: Reoperation for brain metastases. J Clin Oncol 6:1625 1629, 1988 21. Truong MT, St Clair EG, Donahue BR, Rush SC, Miller DC, Formenti SC, et al: Results of surgical resection for progression of brain metastases previously treated by gamma knife radiosurgery. Neurosurgery 59:86 97, 2006 22. Vecht CJ, Haaxma-Reiche H, Noordijk EM, Padberg GW, Voormolen JH, Hoekstra FH, et al: Treatment of single brain metastasis: radiotherapy alone or combined with neurosurgery? Ann Neurol 33:583 590, 1993 23. Vecil GG, Suki D, Maldaun MV, Lang FF, Sawaya R: Resection of brain metastases previously treated with stereotactic radiosurgery. J Neurosurg 102:209 215, 2005 Manuscript submitted February 16, 2009. Accepted April 16, 2009. Please include this information when citing this paper: published online May 8, 2009; DOI: 10.3171/2009.4.JNS09246. Address correspondence to: Douglas Kondziolka, M.D., De - partment of Neurological Surgery, University of Pittsburgh Suite B-400, UPMC Presbyterian, 200 Lothrop Street, Pittsburgh, Pennsylvania 15213. email: kondziolkads@upmc.edu. 831