Stereotactic radiosurgery, with or without whole-brain radiation therapy, has become a valued management choice for patients with brain metastases, although their median survival remains limited. In patients who receive successful extracranial cancer care, patients who have controlled intracranial disease are living longer. The authors evaluated all brain metastasis in patients who lived for ≥ 4 years after radiosurgery to determine clinical and treatment patterns potentially responsible for their outcome.
Six hundred seventy-seven patients with brain metastases underwent 781 radiosurgery procedures between 1988 and 2000. Data from the entire series were reviewed; and, if patients had ≥ 4 years of survival, then they were evaluated for information on brain and extracranial treatment, symptoms, imaging responses, need for further care, and management morbidity. These long-term survivors were compared with a cohort who lived for < 3 months after radiosurgery (n = 100 patients).
Forty-four patients (6.5%) survived for > 4 years after radiosurgery (mean, 69 mos with 16 patients still alive). The mean age at radiosurgery was 53 years (maximum age, 72 yrs), and the median Karnofsky performance score (KPS) was 90. The lung (n = 15 patients), breast (n = 9 patients), kidney (n = 7 patients), and skin (melanoma; n = 6 patients) were the most frequent primary sites. Two or more organ sites outside the brain were involved in 18 patients (41%), the primary tumor plus lymph nodes were involved in 10 patients (23%), only the primary tumor was involved in 9 patients (20%), and only brain disease was involved in 7 patients (16%), indicating that extended survival was possible even in patients with multiorgan disease. Serial imaging of 133 tumors showed that 99 tumors were smaller (74%), 22 tumors were unchanged (17%), and 12 tumors were larger (9%). Four patients had a permanent neurologic deficit after brain tumor management, and six patients underwent a resection after radiosurgery. Compared with the patients who had limited survival (< 3 mos), long-term survivors had a higher initial KPS (P = 0.01), fewer brain metastases (P = 0.04), and less extracranial disease (P < 0.00005).
Approximately 20–40% of patients with systemic malignancies develop brain metastases.1–3 There appears to be a trend toward an increasing incidence of brain disease, and patients with systemic malignancies can live longer because of earlier diagnosis and/or better treatment. Many brain tumors are now identified in asymptomatic patients as part of screening neuroimaging studies. The two most commonly used treatments, whole-brain radiation therapy (WBRT)1, 4–7 and stereotactic radiosurgery (SR), extend survival from 3 months to 5 months and from 7 months to 13 months, respectively, depending on tumor type.
Surgical resection can be a valuable approach for patients with larger symptomatic tumors. Although the value of radiosurgery in obtaining local tumor control and extending survival compared with WBRT alone has been substantiated,8–25 the status of long-term survivors and the reasons for their better outcome remain unclear. Numerous studies have confirmed that the extent of extracranial disease correlates directly with survival. What remains unclear is whether successful or unsuccessful care for systemic malignancy, together with aggressive brain tumor care, can lead to prolonged survival. We are faced with the challenge of improving survival for patients with brain metastases and managing late complications or recurrences that previously were unseen as our overall management of cancer improves.
MATERIALS AND METHODS
Six hundred seventy-seven patients with brain metastases underwent 781 gamma knife radiosurgery procedures (Elekta Instruments, Atlanta, GA) between 1988 and 2000 at the University of Pittsburgh. We reviewed data from this entire series and evaluated patients who had ≥ 4 years of survival for information on brain and extracranial treatment, tumor subtypes, radiosurgery dosimetric data, symptoms, imaging responses, need for further care, and management morbidity. Specific inclusion criteria for radiosurgery included 1) a greatest tumor dimension ≤ 3.5 cm; 2) no major, sustained neurologic deficit because of mass effect; and 3) medically refractory seizures. Patients with larger tumors that caused significant neurologic symptoms that did not improve with corticosteroids underwent a craniotomy and resection if the tumor was in an accessible location.
Patients were followed from the time of treatment until death. A listing of all 677 patients with median survival is provided in Table 1. Clinical and imaging characteristics of the 44 patients who survived for > 4 years are shown in Table 2. The mean patient age of these 44 patients was 53 years at the time of radiosurgery. Nineteen patients were female (43%), and 25 patients were male (57%). The median Karnofsky performance score (KPS) was 90 (range, 60–100). Long-term survivors were compared with a cohort of patients from the same clinical experience who had the shortest survival (< 3 mos; n = 100 patients).
Table 1. Tumor Type and Survival in 677 Patients with Brain Metastases after Radiosurgery
No. of patients (%)
Median survival (mos)
Unknown primary tumor
Table 2. Characteristics of 44 Long-Term Survivors with Brain Metastases
No. of patients (%)
WBRT: whole-brain radiotherapy.
Tumor type (n = 44 patients)
Tumor location (n = 133 patients)
Other irradiation at first radiosurgery
Only brain metastasis
Primary tumor only
Primary tumor plus lymphadenopathy
Organs involved, ≥ 2
Prior management included WBRT and/or tumor resection in some patients. Twenty-five patients (56%) had completed prior WBRT (mean dose, 30 Gray [Gy]), a mean of 5 months before radiosurgery. Radiosurgery was indicated because of either local or new tumor progression. Thirteen patients received WBRT and boost radiosurgery during the same 1-month period as part of a planned strategy that was determined by the referring medical or radiation oncologist. Seventeen tumors had been resected previously with radiosurgery used for local recurrence; for 9 of those tumors WBRT also was given, but subsequent progression occurred.
All patients underwent radiosurgery under local anesthesia and mild intravenous sedation. Stereotactic magnetic resonance imaging (MRI) was used for target planning unless there was a contraindication to MRI; then, computed tomography was used. A single or multiple isocenter plan was used to construct a conformal irradiation volume to the tumor margin. The neurologic surgeon, radiation oncologist, and medical physicist performed dose selection and planning. The maximum and marginal doses varied from 20 Gy to 40 Gy (mean, 32.4 Gy) and from 11 Gy to 22 Gy (mean, 17.7 Gy), respectively. The mean tumor volume was 2.8 mL (range, 0.15–15.3 mL) and was measured by using computer software. All patients were discharged within 24 hours after radiosurgery.
After radiosurgery, clinical follow-up data were obtained during office evaluations of the treated patients. Patients who lived more than a 2-hour drive from our center were followed by their referring oncologist. Follow-up imaging studies were scheduled at 2 months, 5 months, 8 months, and 12 months for the first year; every 4 months during the second year; and every 6 months thereafter. The survival rate, local control rate, intracranial control rate, and complication rates after radiosurgery were calculated. Local tumor control was defined as control of the original brain metastasis treated with radiosurgery. Factors that affected survival and tumor control were assessed with the Cox regression model. Factors that affected survival included age, gender, KPS, recursive partitioning analysis (RPA) classification, active extracranial disease, tumor volume, marginal dose, tumor location, number of metastases, primary tumor status, and primary disease sites. Factors that affected tumor control included age, KPS, marginal dose, tumor volume, primary tumor status, and primary sites. A final multivariate analysis was calculated using a stepwise forward logistic regression.
The median survival (to death or to the last office visit) for the entire cohort of 677 patients was 12 months (mean, 14.6 mos). Of the 44 patients who lived for > 4 years after radiosurgery, the median survival was 68 months (mean, 68.6 mos; range, 48–156 mos). Sixteen patients remained alive at the time of last follow-up (maximum, 156 mos).
Of the 28 patients who died later, only 1 patient died from an intracranial event, which was related to a brainstem hemorrhage at 53 months, after his third radiosurgery. The bleed occurred 5 months after radiosurgery for a pontine melanoma. All other patients died of progressive systemic disease.
Forty-four patients who had long-term survival underwent radiosurgery for 133 brain metastases. This included 96 tumors that were irradiated without prior resection and 17 tumors that were resected (either a residual or recurrent tumor). At last follow-up, 99 tumors had regressed (74.4%), 22 tumors remained unchanged in volume (16.5%), and 12 tumors had enlarged (9%) (Figs. 1–3). Resection after radiosurgery was performed on 6 tumors (4.5%) because of progressive enlargement on serial imaging studies associated with worsened neurologic symptoms. These resections were performed 5 months, 13 months, 19 months, 21 months, 22 months, and 23 months after radiosurgery. WBRT was administered later for 4 tumors (3%), and additional radiosurgery was performed on 2 tumors. Thus, 12 tumors required further treatment of some kind (9%).
Over time, many patients underwent additional radiosurgery procedures, because new brain metastases were identified (Fig. 4). Only 23 patients (52%), who underwent 1 radiosurgery each, became long-term survivors. Eight patients required 2 radiosurgeries, another 8 patients required 3 radiosurgeries, 1 patient required 4 radiosurgeries, 1 patient required 5 radiosurgeries, 2 patients required 6 radiosurgeries, and 1 patient (who lived for ≥ 10 yrs) required 7 radiosurgeries. Each procedure was performed for a tumor in a different brain location.
Temporary or permanent complications were evaluated in these long-term survivors. Transient headache related to intracranial edema was noted in nine patients, with nausea (three patients), imbalance (two patients), and arm or leg weakness (two patients) also identified. Imaging studies showed that 8 patients (18%) had asymptomatic edema (in a total of 133 imaged tumors). Permanent neurologic deficits were noted in 4 patients (9%). These included left leg weakness after a tumor resection in one patient, mild hemiparesis after a tumor resection in one patient, one patient with a partial gaze paresis, and one patient who was comatose as a result of the pontine melanoma hemorrhage described earlier.
Comparison with Patients who Had the Shortest Survival
During this clinical experience, 100 patients died < 3 months after radiosurgery. Our objective was to compare outcomes in the best patients with those who had the poorest survival. Compared with the group that survived for > 4 years, a multivariate analysis showed no differences in age; gender; percentage of lung carcinoma, melanoma, or renal cell carcinoma; radiosurgery margin dose; use of prior WBRT; volume of the largest tumor; or total tumor volume (Table 3). The patients who lived for > 4 years had a higher preradiosurgery KPS (100 vs. ≤ 90; P = 0.017; odds ratio per 10-point KPS change, 1.67; 95% confidence interval [95% CI], 1.60–1.74), fewer metastases (P = 0.039; odds ratio per tumor, 0.604; 95% CI, 0.373–0.977), and less extracranial disease burden, which was measured as the number of involved organ systems (P < 0.00005; odds ratio per organ involved, 0.052; 95% CI, 0.075–0.181). Patients with 2 brain tumors fared almost as well as patients who had 1 tumor, but increasing numbers of tumors were associated with a lesser chance of surviving past 4 years. Table 4 compares the short-term and long-term survivors in categories and notes the significant differences.
Table 3. Univariate and Multivariate Logistic Regression Analyses of Factors that Potentially Predicted Long-Term Survival (> 4 Yrs) in 44 Patients versus Short-Term Survival (< 3 mos) in 100 Patients after Radiosurgery for Brain Metastases
Table 4. Comparison of Significant Differences Between 100 Short-Term Survivors (< 3 Mos) and 44 Long-Term Survivors (> 4 Yrs) after Radiosurgery for Brain Metastasis
No. of patients (=)
Extent of systemic disease
Karnofsky performance status
No. of brain metastases treated at first radiosurgery
Selected patients with brain metastasis live much longer than the expected median survival of 7–13 months noted after comprehensive multimodal therapy. This survival range, which is strongly dependant on tumor histology, has been limited mainly by progression of extracranial disease. RPA showed that survival was related to numerous clinical factors including neurologic status and extent of cancer.26, 27 However, today, a minority of patients with metastatic brain carcinoma die because of intracranial progression or complications. Previously, we observed a median survival of 16 months after radiosurgery for patients with metastasis from nonsmall cell lung carcinoma if they had no active, systemic malignancy and 7 months if they did.23 In addition, important factors for improved survival included female gender, preoperative KPS, adenocarcinoma histologic subtype, and longer time from lung carcinoma diagnosis.23 In an earlier study, we found a median survival of 26 months in patients with nonsmall cell lung carcinoma if they had no active, systemic disease; no neurologic deficit; and a small, nonnecrotic tumor.14 The overall imaging-defined tumor control rate was 84%.
Surgical resection followed by WBRT also has been associated with a survival benefit compared with WBRT alone. Two randomized trials demonstrated such a benefit for patients with a single brain metastasis.28, 29 However, a third study that included more patients with active extracranial disease failed to show a benefit from surgery.30 A later trial showed that resection alone led to poorer local tumor control if WBRT was not delivered.31 Those studies emphasized not only the importance of brain tumor control but also the strong negative effect of extracranial disease. In a large retrospective review from the Mayo Clinic, Smalley et al. recommended that surgical resection should not be considered routinely in patients with active, systemic malignant disease.32 However, many surgeons perform such resections in this setting in an attempt to reverse neurologic disability and improve quality of life. Thus, how should physicians use information on disease extent to guide brain tumor management? Perhaps it is not so much the extent of disease but, rather, the availability of specific treatments that should be considered.
In the current study, we found that a larger extracranial carcinoma burden was seen more commonly in patients who had the poorest survival. Conversely, the majority of patients who survived for > 4 years had active disease that involved ≥ 2 organ systems at the time they underwent radiosurgery. In addition, patients with adverse prognostic factors, such as active, systemic disease and multiple brain metastases, who do not become long-term survivors are likely to experience improved quality of life compared with patients who are treated with WBRT alone. Therefore, a nihilistic approach to patients with active malignancies and brain disease should not be the norm. Rather, we must work to manage effectively both the brain and body sites of disease.
To our knowledge, little has been written regarding extended survival in patients with brain metastasis. Most of the information available provides some 2-year survival data, with longer term actuarial calculations in some series. Lutterbach et al. noted a 2-year survival rate of 13% after radiosurgery, with median survivals of 13.4 months, 9.3 months, and 1.5 months for patients in RPA Class 1, Class 2, and Class 3, respectively.33 In a separate series, Chang et al. noted a 2-year survival rate of 19% and an overall median survival of 11.4 months.34 Petrovich et al. noted 1-year, 2-year, and 3-year actuarial survival rates of 33%, 16%, and 10%, respectively, after gamma knife radiosurgery, with the best survival noted in patients with breast carcioma.20 In a report from Radiation Therapy Oncology Group Study 90-05, Shaw et al. noted a 2-year survival rate of 15%, but only 2 patients (1%) were alive after 5 years.35 In a series from the University of Florida, Ulm et al. noted a 2-year actuarial survival rate of 19% and a median survival of 9 months.36 Sneed et al. noted a 2-year survival rate of 25% in their series from San Francisco, but the 3-year survival after radiosurgery alone was zero.37 Pirzkall et al. noted no difference in 2-year survival rates after radiosurgery alone or after radiosurgery plus WBRT (P = 0.75), but they observed that no active extracranial disease strongly favored better survival (P < 0.001).21 Thus, across various reported series, the rate of 2-year survival after radiosurgery ranged from 10% to 20%. In the current series, the 4-year survival rate was 6.5% indicating that approximately 1 in 3 patients who live for 2 years can be expected to double that result. Such patients who live to 4 years almost always have controlled extracranial disease at the 2-year mark after radiosurgery, whether or not they had controlled disease at the time of radiosurgery.
Tumor histology is an important factor in predicting survival. Improved survival for different tumor types depends on several issues, including the extent of disease at brain tumor presentation (a frequent problem for patients with primary gastrointestinal carcinoma) and the variety or efficacy of treatments available for extracranial carcinoma (limited for patients with metastatic melanoma). The various median survivals associated with different tumor histologies are listed in Table 1. However, tumor histology should not be considered a barrier to obtaining long-term survival, because all 4 of the most frequent tumor types were present among our long-term cohort (Table 2). In a similar fashion, deep brain location does not limit the opportunity for prolonged survival. The distribution of tumors seen in patients with 4-year survival was similar to the typical distribution in any large brain metastasis population. These patients had characteristics that were similar in many ways to patients who had the most limited survival (< 3 mos). However, because the factors of better neurologic condition at initial management, fewer metastases, and less systemic disease are so important, but not absolute, we must continue to offer effective, comprehensive cancer care.
The results of the current study showed that, in patients with extended survival, there is a high likelihood that additional brain tumor care will be necessary during the course of their lifetime.38 Twelve of 133 tumors required further management, which included resection, radiosurgery, or WBRT. However, 48% of patients underwent additional radiosurgeries for new brain tumors, including 5 patients who underwent ≥ 4 procedures over time. It is noteworthy that the patient who underwent 7 radiosurgeries and survived for 10 years was a participant in our randomized trial of WBRT plus radiosurgery versus WBRT alone.15 Varlotto et al. studied patients from our center who had at least 1-year survival.25 The rate of local tumor control in their study was 90% at 1 year and 63% at 5 years, with 2.8% and 11.4% rates of postradiosurgery clinical sequelae, respectively.
In patients who have tumor progression or findings of new tumor, salvage therapy should be considered immediately, including WBRT, repeat radiosurgery, or resection. Thus, periodic imaging examinations (every 3–4 mos during the first 2 yrs and every 6 mos thereafter) are recommended after radiosurgery. In the current study, most of the new brain metastases were controlled with additional radiosurgery. The decision to provide additional brain tumor treatment was based mainly on the neurologic condition of the patient, together with a plan to continue treatment of any extracranial malignancy.
To our knowledge to date, WBRT has been the most commonly used treatment for patients with brain metastases.39 In this series, which reflected patient care up to the year 2000, most patients received WBRT just before radiosurgery. However, in the hope of avoiding short-term toxicity and longer term cognitive problems in patients with limited brain disease, now, we often withhold initial WBRT.1, 6, 17, 40–42 Because patients with fewer brain metastases and limited extracranial disease live longer, they are at risk for the development of new brain metastases with extended follow-up. For patients who have recurrent tumors after radiosurgery alone, later WBRT or repeat radiosurgery remain valuable options. Although WBRT may control micrometastases that are not seen on MRI (as shown by the higher remote tumor control rate), later WBRT or radiosurgery usually is effective as salvage therapy. We believe that initial avoidance of WBRT will lead to improved patient quality of life. A recent survey that evaluated patient and family opinions after radiosurgery alone or radiosurgery preceded by WBRT found that patients who received WBRT noted more cognitive and constitutional side effects.42
Although the expected survival of patients with brain metastases may be limited, selected patients with malignant disease who receive effective intracranial and extracranial care can have prolonged, good-quality survival. The extent of extracranial disease at the time of radiosurgery is predictive of outcome but does not necessarily mean that patients cannot live for years if treatment is effective.
The authors thank the many medical oncologists, radiation oncologists, surgeons, primary physicians, nurses, and administrative staff whose efforts helped to achieve extended survivals in these patients.