Brain metastases (BM) are the most common intracranial tumors in adults. To the authors' knowledge, established prognostic factors for survival after the diagnosis of BM in breast cancer patients do not take into account HER–2 status, which may have increasing relevance in the trastuzumab therapy era.
The authors identified 83 patients with breast cancer and new parenchymal BM diagnosed between January 1, 2001 and December 31, 2005 who were treated at Massachusetts General Hospital. Survival was estimated using the Kaplan-Meier method and curves were compared using the log-rank test. A Cox proportional hazards model was used to determine independent predictors of survival.
The median overall survival from the time of BM was 8.3 months. On univariate analysis, HER-2-positive patients were found to have prolonged survival after BM compared with HER-2-negative patients (17.1 months vs 5.2 months). Patients with triple negative disease had a median survival of 4.0 months, compared with 11.2 months for all other patients. Additional predictors of improved survival on univariate analysis included ≤3 BM, controlled or absent systemic disease, and controlled local disease. On multivariate analysis, only HER-2 status, number of BM, and local disease status remained independent predictors of survival.
Breast cancer is the most frequent neoplasm in women and the second leading cause of cancer-related deaths in women in the U.S. The past 20 years have seen a 24% decrease in breast cancer mortality rates that has been attributed to earlier detection by mammography screening as well as improvements in adjuvant systemic therapy.1 Despite advances in survival for patients with early–stage disease, brain metastases (BM) remain an important cause of morbidity and mortality in patients with advanced disease. Approximately 15% to 20% of patients with metastatic breast cancer will be affected by BM during the course of their disease.
Historically, median survival after the diagnosis of BM ranges from 3 months to 6 months, regardless of the type of primary cancer, with cause of death divided equally between neurologic and systemic disease progression. Known prognostic factors of survival after BM include age, functional status, number of BM, systemic disease status, and the interval between the primary cancer diagnosis and the development of BM. For patients with newly diagnosed breast cancer, hormone receptor (HR) and HER-2 status have established prognostic value.2, 3 However, to our knowledge, their utility after the diagnosis of BM is not well characterized.
There are several reasons to believe that survival may be improving for patients with BM. For the subset of patients with good functional status and a single BM, surgical resection or stereotactic radiosurgery (SRS) combined with whole brain radiotherapy (WBRT) may improve both neurologic progression–free and overall survival by 2 months to 6 months over WBRT alone.4–6 In addition, treatment options for systemic disease control in patients with breast cancer have increased, particularly for women with HER-2-positive tumors, leading to improved survival after a diagnosis of metastatic disease.7–9
Whether these advances have translated into improved survival for patients with BM from breast cancer is unknown. We sought to study a recent cohort of patients with breast cancer and newly diagnosed BM to determine the prognostic impact of HER-2 status and the duration of survival.
MATERIALS AND METHODS
We identified all patients with a primary diagnosis of breast cancer who developed new BM between January 1, 2001 and December 31, 2005 using 2 Massachusetts General Hospital (MGH) institutional databases. This study was reviewed and approved by the Partners Human Research Committee. All patients had histologically confirmed breast cancer and parenchymal BM, defined as the presence of ≥1 intra-axial–enhancing lesions on gadolinium–enhanced magnetic resonance imaging (MRI) of the brain or contrast–enhanced computed tomography (CT) scans believed by the radiologist and treating physicians to represent metastatic disease. A total of 154 patients were identified on the initial search. Patients were excluded if they were referred to MGH exclusively for surgery or SRS and did not maintain care at MGH afterward (n = 40 patients), if they had isolated dural (n = 16 patients) or leptomeningeal (n = 6 patients) metastases with no evidence of parenchymal BM, or if there was uncertainty regarding the primary tumor responsible for the BM (n = 9 patients). The remaining 83 patients formed the basis of the current study. Data regarding 29 of these patients has been previously published in preliminary form.10
The paper and electronic inpatient and outpatient medical records of all patients were reviewed, and data were collected regarding demographic and clinical characteristics including patient age, TNM classification of the primary tumor,11 grade, estrogen receptor (ER) and progesterone receptor (PR) status, HER-2 receptor amplification, treatment of the primary tumor, date of first distant metastasis, date of first BM, symptoms at the time of BM diagnosis, treatment and extracranial disease activity at the time of BM, organ involvement at the time of BM, distribution and imaging characteristics of BM, treatment, time to disease recurrence, treatment at disease recurrence, survival, and cause of death. Follow-up took place through April 10, 2007.
HR status was considered positive if either the ER or PR immunohistochemical (IHC) stains were positive. Patients were considered HER-2 positive if they were 3+ on IHC testing or 2+ on IHC with gene amplification identified by fluorescence in situ hybridization testing. Disease activity at the time of BM diagnosis was considered active if chest, abdominal, and pelvic CT scans; positron emission tomography, and/or bone scans within 4 weeks of the BM diagnosis demonstrated new sites of extracranial metastatic disease or disease progression at previously known sites of disease. The duration of response was measured from the day of the initial diagnostic CT or MRI scan until the day of radiologic disease progression.
Death was attributed to central nervous system (CNS) disease progression if the patient had radiologic disease progression or worsening neurologic symptoms at the time of last follow-up and had radiologic documentation of stable extracranial disease within 3 months of death. Death was attributed to systemic disease progression if the patient had active and symptomatic extracranial disease at the time of last follow-up and had no clinical or radiologic evidence of CNS disease progression within 3 months of death. In all other cases, death was attributed to both CNS and systemic disease progression unless death occurred >3 months after last follow-up, in which case the cause of death was considered indeterminate. Deaths unrelated to breast cancer were also noted.
Frequencies and descriptive statistics of demographicand clinical variables were obtained. Categorical variables were compared using the Fisher exact test and chi-square test, and continuous variables were compared using the Student t test or Wilcoxon rank sum test, as appropriate. Survival from the time of BM to death or last follow-up was estimated using the Kaplan-Meier method, and survival curves were compared using the log-rank test. Multivariate predictors of survival were determined using the Cox proportional hazards model. Hazards ratios and their 95% confidence intervals (95% CIs) were computed. The proportional hazards assumption was verified by testing the significance of time-varying covariates for each variable in the final model. A 2-sided P < .05 was considered to be statistically significant. All analyses were conducted using SAS statistical software (version 9.2; SAS Institute Inc, Cary, NC).
A total of 83 patients with breast cancer and new BM were identified. Baseline patient characteristics according to HER-2 status are listed in Table 1. The mean age at the time of diagnosis of breast cancer for the entire cohort was 49.2 ± 13 years and did not vary significantly by HER-2 or HR status. The majority of patients had underlying invasive ductal carcinoma (n = 75 patients) with high-grade features (n = 51 patients) at the time of initial diagnosis. Approximately 67% of patients had T1 or T2 disease, 54% (n = 45 patients) were lymph node positive, and 7% (n = 6 patients) had metastatic disease at the time of the primary presentation. HR and HER-2 status were known in 100% and 96% of patients, respectively. Approximately 36% of patients were HER-2 positive and 25% were triple negative, which together accounted for greater than half of all patients followed (Fig. 1). The majority of patients received neoadjuvant or adjuvant chemotherapy (n = 59 patients) and radiotherapy (n = 53 patients). Among patients with HR-positive disease, approximately half (n = 25 patients) received adjuvant hormonal therapy.
Table 1. Baseline Patient Characteristics
HER-2 positive (N = 30)
HER-2 negative (N = 50)
Total (N = 83)
No. of patients (%)
No. of patients (%)
No. of patients (%)
SD indicates standard deviation; HR: hormone receptor.
The mean age at the time of diagnosis of BM was 54.3 ± 13 years and did not vary significantly by HER-2 or HR status (Table 2). The median time from the diagnosis of breast cancer to first BM was 47 months (range, 0–269 months). Patients with triple negative disease had a significantly shorter time to first BM compared with all other patients (25 months vs 60 months; P = .004). The brain was the first site of metastasis in 24% of patients and was accompanied by other sites of metastatic disease in the majority of cases. Active systemic disease was very common at the time of BM, with 73% of patients having either new or progressive disease developing outside of the brain within 1 month of the diagnosis of BM. Seventeen patients (20%) had active local disease in the breast or chest wall at the time of BM. The most common sites of extracranial metastasis were the lung (65%), bone (63%), lymph node (48%), and liver (40%). The median number of extracranial sites of disease at the time of BM was 3 (range, 0–6 extracranial sites). Only 3 patients had absent extracranial disease at the time of first diagnosis of BM.
Table 2. Patient Characteristics at the Time of Brain Metastasis
HER-2 positive (N = 30)
HER-2 negative (N = 50)
Total (N = 83)
No. of patients (%)
No. of patients (%)
No. of patients (%)
SD indicates standard deviation.
Mean age at diagnosis of brain metastasis (± SD), y
54.0 ± 12.4
53.6 ± 13.0
54.3 ± 12.8
Systemic disease status
Controlled on treatment
Local disease status
Sites of extracranial metastasis
No. of brain metastases
Location of brain metastases
Approximately 75% of patients were receiving active treatment for breast cancer at the time of BM diagnosis. Among HER-2-positive patients with previously metastatic disease at the time of BM, 73% (n = 16 patients) were receiving trastuzumab. Of these, 50% had controlled systemic disease at the time of the diagnosis of BM. The median number of prior chemotherapy regimens for all patients was 2 (range, 0–8 chemotherapy regimens).
BM were single in 23 patients (28%) and multiple in the remaining 60 patients. The number of BM did not vary significantly by HER-2, HR, or triple negative status. Approximately 76% of patients were symptomatic at the time of diagnosis. The proportion of patients diagnosed with asymptomatic BM ranged from 20% of HER-2-positive patients to 38% of patients with triple negative disease. Cerebellar metastases were common; 13% of patients (n = 11) had isolated cerebellar BM and an additional 45% (n = 37 patients) had both supratentorial and cerebellar BM at the time of diagnosis. Overall, 58% of patients had at least 1 cerebellar BM at the time of the initial presentation. Cerebellar BM were slightly more common in HER-2-positive patients (50%) compared with HER-2-negative patients (44%), but the difference was not statistically significant.
The majority of patients (83%) were treated with WBRT either as initial therapy (n = 63 patients) or at the time of disease recurrence (n = 6 patients). Nineteen patients (23%) underwent a craniotomy, and all but 1 received postoperative WBRT. Approximately one–quarter of patients (n = 19) were treated with SRS to ≥1 BM as initial therapy, either alone or in combination with WBRT. An additional 11 patients were treated with SRS at the time of disease recurrence. The proportion of patients treated with SRS did not vary significantly by HER-2 or HR status. The majority of patients received ≥1 cycles of systemic chemotherapy after the diagnosis of BM (n = 56 patients). Among HER-2-positive patients, 80% (n = 24) received trastuzumab after the diagnosis of BM.
Disease recurrence or progression in the brain before death was diagnosed by CT or MRI in 36% of patients (n = 30). An additional 17 patients had progressive neurologic symptoms at death but did not have a confirmatory CT or MRI scan. Using the Kaplan-Meier method, assuming disease recurrence at the time of death in these additional patients, the median time to disease recurrence was 8.5 months (95% CI, 5.0–13.6 months). The time to disease recurrence was significantly longer in patients with HER-2-positive tumors compared with those with HER-2-negative tumors (13.5 months vs 4.4 months; P = .009).
Seventy-seven patients (93%) had died by the end of follow-up. The median overall survival from the time of BM diagnosis was 8.3 months (95% CI, 4.8–11.5 months) and the 1-year survival rate was 36% (Fig. 2). On univariate analysis, HER-2-positive patients hadprolonged survival after the diagnosis of BM compared with HER-2-negative patients (17.1 months vs 5.2 months; P = .001). There was no significant difference noted with regard to overall survival based on HR status (11.0 months for HR-positive vs 5.0 months for HR-negative patients; P = .41). The 21 patients with triple negative disease had a median survival of 4.0 months, compared with 11.2 months for all other patients (P = .04) (Fig. 3). Additional predictors of improved survival on univariate analysis included ≤3 BM (11.5 months vs 5.7 months; P = .03), controlled or absent systemic disease (15.9 months vs 5.9 months; P = .04), and controlled local disease (10.5 months vs 3.4 months; P = .02). On multivariate analysis, only HER-2 status, the number of BM, and local disease status remained as independent predictors of survival, after adjusting for age, HR status, local disease control, and status of systemic disease (Table 3). Additional adjustments for craniotomy or SRS did not appear to change the magnitude or significance of the hazards ratios appreciably, and therefore these factors were not included in the final model.
Table 3. Predictors of Survival After Diagnosis of Brain Metastasis
Reference groups: HER-2-negative, hormone receptor negative, not triple negative, ≤3 BM, absent or controlled systemic disease, and controlled local disease.
No. of BM
Systemic disease status
Local disease status
Cause of Death
The cause of death could be determined in 62 patients (75%). Only 9 patients (15%) had isolated CNS disease progression with stable systemic disease at the time of death. Twenty-seven patients (44%) had evidence of both systemic and CNS disease progression, and 24 patients (38%) died of systemic disease progression with stable CNS disease. The remaining 2 patients died of causes unrelated to breast cancer (infection). The proportion of patients who died of CNS versus systemic disease progression did not appear to vary significantly by HER-2 or HR status.
We performed a single–institution, retrospective cohort study of all breast cancer patients with newly diagnosed BM within the last 5 years and found a median overall survival of 8.3 months from the time of diagnosis of BM, which compares favorably with the median overall survival of 4 to 6.5 months reported in historical prospective and retrospective analyses of breast cancer patients treated through the early 1990s.12–16 The majority of patients in the current study cohort received multimodality treatment, including cranial irradiation and systemic chemotherapy and/or targeted drugs after the diagnosis of BM. These data support the premise that survival after a diagnosis of BM in breast cancer patients is improving with modern treatment modalities.
In the current study, prolonged survival after a diagnosis of BM appears to be driven by the HER-2-positive patient subset. Patients with HER-2-positive disease were found to survive significantly longer after a diagnosis of BM compared with those with HER-2 negative disease, independent of previously recognized prognostic factors (hazards ratio of 0.37; 95% CI, 0.19–0.71). Although lead-time bias from asymptomatic screening in approximately 25% of the patients may have contributed to prolonged survival in the overall cohort, it should not preferentially affect the HER-2-positive patient subset, 80% of whom were diagnosed on the basis of neurologic symptoms.
Several recent studies have examined survival after the diagnosis of BM in HER-2-positive patients (Table 4), although to our knowledge the majority have not included a HER-2-negative comparison group.10, 17–20 Bendell et al studied 122 breast cancer patients who were treated with trastuzumab and found that 34% (n = 42 patients) developed CNS disease with a median survival of 13 months from the date of BM diagnosis.17 Similarly, Stemmler et al reported a median survival of 13 months in 42 patients with HER-2-positive disease who were being treated with trastuzumab at the time of BM.19 In an effort to determine whether patients receiving trastuzumab are more likely to develop BM compared with those not receiving the drug, Lai et al studied 343 breast cancer patients and found no difference with regard to the rate of BM between the 2 groups (48.1% and 46.6%, respectively).21 The median survival after the diagnosis of BM was 24.9 months for patients who received trastuzumab (47% of whom were HER-2 positive) and 26.3 months for those who did not (14% of whom were HER-2-positive) (P = .7).
Table 4. Survival After Diagnosis of BM in Patients With HER-2-Positive Breast Cancer
The reasons behind the survival advantage for patients with HER-2-positive disease are not immediately clear. The majority (80%) of patients with HER-2-positive disease either began or continued to receive trastuzumab in combination with various chemotherapeutic agents after a diagnosis of BM. Trastuzumab has proven antitumor effects in vitro and is reported to produce a significant survival advantage both alone and in combination with chemotherapy in women with HER-2-positive metastatic breast cancer.22 Therefore, the observation of prolonged survival after BM may reflect improved disease control related to trastuzumab therapy. It is interesting to note that patients with HER-2-positive tumors were no more likely to die of isolated CNS disease progression than those with HER-2-negative tumors. This finding suggests that we are not simply observing the creation of a “sanctuary site,” wherein trastuzumab allows for progressive CNS disease while keeping systemic disease under good control.
HER-2-positive patients also had a longer time to CNS disease recurrence compared with HER-2-negative patients in the current study cohort. Although it is possible that HER-2-positive BM are biologically more responsive to cranial radiotherapy, this appears unlikely and to our knowledge has not been described previously. Alternatively, better control of systemic disease with trastuzumab combined with traditional chemotherapy may prolong time to reseeding of the CNS by hematogenous tumor burden. It is also possible that trastuzumab contributes to local CNS disease control by penetrating the CNS at sites of local blood-brain barrier (BBB) disruption. In animal models, BM measuring >0.5 cm are associated with reduced interactions between pericytes and glia in peritumoral capillaries, longer junctions between endothelial cells, and disruption of the BBB.23
In addition to HER-2-positive tumors, which comprised 36% of the current study cohort, triple negative tumors were also very common (25%). This suggests overrepresentation of triple negative histology in the BM population compared with newly diagnosed breast cancer patients, in whom the prevalence of triple negative disease is estimated to be only 10% to 15%.24 A recent pathologic study by Gaedcke et al found that 29 of 85 surgical BM samples (34%) from 1 institution were HER-2 positive and 22 of 85 (22%) had a basal phenotype (HR/HER-2 negative and basal cytokeratin [CK] positive).25 Similar to HER-2-positive tumors, triple negative tumors have a propensity for earlier visceral metastasis and confer a poorer survival compared with other breast cancer subtypes.2, 26, 27 Survival for the triple negative disease patient subgroup in the current study was only 4 months, most likely reflecting an advanced and treatment-refractory disease state at the time of BM diagnosis.
The number of BM and local disease control were also found to be independent predictors of survival in the current analysis. Patients with >3 metastases had a significantly poorer survival compared with those with <3 BM. It is noteworthy that survival was not found to be significantly different in patients with 1 vs. 2 to 3 BM, a finding that has been reported previously in patients undergoing surgical resection for a dominant lesion.28, 29 This may support the notion that patients with otherwise favorable prognostic factors and a limited number of BM (≤3) benefit from similarly aggressive efforts toward local disease control in the brain.
We report a surprisingly high prevalence of cerebellar metastases in the current study cohort, with 57% of patients having at least 1 cerebellar BM at the time of diagnosis. This is much higher than the 15% reported by Delattre et al in a study of 288 patients with BM from a variety of primary cancer types, although this study depended entirely on CT imaging, which is less sensitive than MRI for detecting disease in the posterior fossa.30 The reasons for a cerebellar preponderance in patients with breast cancer are not known. The craniospinal venous system anastomoses with the sacral, pelvic, and prostatic venous plexus, providing a potential pathway for the rostral metastatic spread of pelvic malignancies to the posterior fossa.30, 31 However, a predilection for the cerebellum has not been previously reported in breast cancer to our knowledge. As knowledge of the mechanisms of invasion, proliferation, metastatic cell signaling, and angiogenesis in BM increases, we may find that the cerebellum represents a particularly hospitable microenvironment for BM from breast cancer, and this observation deserves attention in future studies.
There are several potential limitations in the current study. First, although patients referred only for surgery or SRS were excluded, the residual selection bias of a tertiary care center may remain and may have inflated the median survival for the overall cohort. In addition, we could not draw conclusions regarding the superiority of various treatment modalities because of a limited sample size and concern that residual confounding by indication would remain despite adjustment for known prognostic factors. This study was underpowered to determine whether the delivery of trastuzumab after a diagnosis of BM impacted survival in HER-2-positive patients because only 6 patients did not receive trastuzumab in this setting, although we suspect it may have played a role. Finally, we were not able to adjust for functional status in the current analysis because Karnofsky performance status data were not collected prospectively and could not be reliably determined from a retrospective review of the clinical record. However, it appears unlikely that functional status differed so greatly between HER-2-positive and HER-2-negative subgroups so as to explain away the highly significant difference observed with regard to overall survival.
The results of the current study demonstrate that patients with HER-2-positive disease have a significantly prolonged survival after a diagnosis of BM compared with those with HER-2-negative and/or triple negative disease. The median overall survival of the entire cohort was modestly improved compared with historical controls from the 1980s and 1990s. This suggests that progress is being made with improved systemic agents, specifically trastuzumab in HER-2-positive patients, along with modern surgical and radiotherapy techniques. However, despite this progress greater than half of the patients in the current study died of CNS disease progression, either in isolation or combined with systemic disease progression. Better treatments are needed for patients with BM. At the same time, an improved understanding of the delayed effects of treatment is necessary to properly counsel patients regarding the risk-benefit ratio of various treatment modalities.