Introduction

Breast cancer is the second most common cause of brain metastases, with 10–20% of patients experiencing metastasis to the brain [1, 2]. Recently, however, a trend toward increased central nervous system (CNS) recurrence has been noted: increases of up to 25–34% [3–6] compared with historical rates of 10–20%, owing to the induction of more sensitive and accurate diagnostic methods, the development of improved adjuvant and palliative therapy regimens, and improved survival [7–9].

The median time from diagnosis of breast cancer to CNS metastasis has been reported to be 33 months [2, 10]. Brain metastasis is usually associated with aggressive tumor behavior, negative hormone-receptor status, a relatively young age (premenopausal), and the presence of lung and liver metastases. However, no conclusion can yet be drawn as to whether there is a specific biology that drives a tumor to preferentially metastasize to the brain, or whether the likelihood of brain metastases is simply related to a general metastatic potential.

The median survival of untreated patients may be as short as 1–2 months [2, 11, 12]. After whole-brain radiation therapy (WBRT), an increase in survival of 3–6 months has been reported [13–15]. WBRT is a well-established treatment modality that results in transitory amelioration of neurological deficits in 60–85% of patients. Despite the use of WBRT, however, the prognosis for patients with brain metastases remains poor, with median survival of only 3–6 months, and this figure has not changed over the past 25 years [2, 10, 16]. Selected patients with a small tumor burden, preferably a solitary brain metastasis that can be treated with surgery as well as radiotherapy, may exhibit extended survival [17].

Prognostic factors for the occurrence and clinical outcome of brain metastases remain to be identified. The present retrospective analysis was performed to contribute to our understanding of the natural history of brain metastases in breast cancer, to assess the prognostic significance of patient-related-factors and to evaluate the efficacy of various treatment modalities.

Patients and methods

Of the 7,872 patients histologically diagnosed with breast cancer between January 1990 and July 2006 at the Asan Medical Center, 198 patients with solitary or multiple brain metastases were included in this retrospective study. CNS lesions were diagnosed by computed tomography (CT) or magnetic resonance imaging (MRI) of the brain. Patients with leptomeningeal metastasis or dural metastases without parenchymal brain metastatic lesions were not included in this study. We reviewed the medical records and pathologic data of these 198 patients to summarize the clinicopathological characteristics and effects of different treatments. Performance status at time of diagnosis of brain metastases was estimated retrospectively using the Eastern Cooperative Oncology Group (ECOG) scale.

Categorical variables were compared using the χ2 test or Fisher’s exact test, where appropriate. The brain-metastasis-free interval was defined as the time from histological diagnosis of the primary breast lesion to the time of objective evidence of the brain metastasis. The time of brain metastasis was defined as that time when brain CT or MRI scan was taken. Overall survival time after brain metastasis was measured from the time of detection of the brain metastasis to the time of death or last follow-up. Univariate analysis was performed to identify prognostic factors for survival by modeling Kaplan-Meier survival curves, and the difference in survival was compared using the log-rank test. Multivariate analysis was performed using the Cox proportional hazards regression model. SPSS program (version 12.0) was used for all analysis, and a P-value less than 0.05 was considered statistically significant.

Results

Patient characteristics

Patient characteristics of the 198 breast cancer patients with brain metastases are presented in Table 1. The median age at diagnosis of brain metastasis was 45 years (range 26–78 years), and 151 (75.7%) had ECOG performance status 1 or 2. The brain was the first site of metastatic involvement with or without extracranial metastases in 62 patients (31%), the second in 74 patients (37%), and at least the third in 61 patients (31%). Brain metastases were solitary in 55 patients and multiple in 143 patients. The most common additional metastatic sites at time of diagnosis of brain metastases were the lungs (37.4%) and bones (35.4%).

Table 1 Patient characteristics
Full size table

Analysis of primary breast lesions showed that 61.6% were estrogen receptor (ER) 0–3 (scale of 7) and 20.7% were ER 4–7; 64.1% were progesterone receptor (PR) 0–3 (scale of 7) and 17.7% were PR 4–7; and 36.9% were human epidermal growth factor receptor-2 (HER-2/neu) expression 0–2+ and 34.3% were HER-2/neu expression 3+. Information about primary lesion size was available for 152 (76.8%) patients; of these, 120 (60.6%) had tumors of ≤5 cm. About 42% of these patients had fewer than three positive lymph nodes. A total of 29 patients (15%) received no adjuvant chemotherapy, 47 (24%) were treated with a CMF (cyclophosphamide, methotrexate, 5-fluorouracil) regimen, 48 (24%) with a CAF (cyclophosphamide, doxorubicin, 5-fluorouracil) regimen, 27 (14%) with an AC-based (doxorubicin/cyclophosphamide) regimen, and 29 (15%) with other regimens (Table 1).

Treatment modalities

Whole brain radiation, usually as primary treatment (75.3%), but also as WBRT after gamma-knife or surgery (4%), was administered to 157 (79.3%) patients. A total of 7 (3.6%) patients underwent resection of solitary brain metastases, 22 (11%) patients underwent gamma-knife surgery, three patients underwent intrathecal chemotherapy (1.5%), and 9 (4.6%) received no treatment.

Radiation treatment consisted of a total dose of 3,000 cGy administered in 10 daily fractions of 300 cGy over a period of 2 weeks. Almost all patients received dexamethasone after diagnosis, which was continued throughout subsequent treatment and then gradually tapered after completion of treatment depending on performance status.

Interval from primary breast tumor to brain metastasis and prognostic factor of survival

The cumulative incidence rates of brain metastases at 1, 3, and 5 years were 15.7%, 58.9%, and 81.7%, respectively. The median interval from primary diagnosis to brain metastases was 32.3 months. Overall median survival of the 198 patients was 5.6 months (95% CI 4.7–6.5 months), with only 23.1% of patients surviving for more than 1 year (Fig. 1). The patients with a long metastasis-free interval (>5 years) showed small and clinically no significant difference (9.8 months versus 5.3 months, P = 0.06, Fig. 2) in survival from diagnosis of brain metastases compared with those with a shorter metastasis-free interval (≤5 years).

Fig. 1
figure 1

Overall survival (OS) after brain metastasis

Full size image
Fig. 2
figure 2

There was a difference in survival from diagnosis of brain metastasis (P = 0.06) between patients with a long metastasis-free interval (>5 years, dashed line) and patients with a shorter metastasis-free interval (≤5 years, solid line)

Full size image

The time from primary diagnosis of breast cancer until diagnosis of brain metastasis and survival time were analyzed according to the patient and tumor characteristics (Table 2). Some of the primary tumor characteristics did affect the interval duration from primary tumor to brain metastasis; however, they had no effect on survival once brain metastasis occurred. Age less than 45 years was associated with a significantly shorter interval from primary diagnosis to brain metastasis compared with age more than 45 years. However, there was no difference in median survival between the two age groups: 5.6 and 5.5 months, respectively. An ER status of 0–3 and a PR status of 0–3 were associated with significantly shorter intervals from primary diagnosis to brain metastases compared with ER status 4–7 and PR status 4–7, respectively; however, there were no significant differences in median survival between the groups. HER-2-positive patients did not develop brain metastases earlier than HER-2-negative patients. Tumor size and lymph-node status of the primary lesion, both of which are known risk factors for distant dissemination, did not affect the time to brain metastasis or the survival from brain metastases. A high histology grade was associated with earlier brain metastases and a slightly but not significantly shorter survival. Lymphovascular invasion of the primary lesion did not affect the interval to brain metastasis. A total of 148 patients received adjuvant chemotherapy depending on the stage of the primary breast lesion. The intracranial metastases-free interval and median survival were not significantly different in the patients who received adjuvant chemotherapy compared with those who did not.

Table 2 Median metastasis-free interval and survival data
Full size table

Patients with a good performance (ECOG performance status ≤2) had a median survival time of 6.5 months (range 0.1–44.6 months) after diagnosis of brain metastases, whereas patients with a poor performance (ECOG performance status ≥3) had a median survival of 3.2 months (range 0.1–20.2 months; P < 0.001).

Patients without extracranial metastases had a median survival of 9.0 months, whereas patients with extracranial metastases had a median survival of 5.3 months (P = 0.02). Patients with a solitary brain lesion had median survival of 9.2 months, which was significantly longer than the 5.3 month median survival time of patients with multiple brain lesions (P = 0.01). The median overall survival time was 5.4 months for patients who received WBRT-based therapy (WBRT or WBRT plus surgery/gamma-knife surgery), 14.9 months for patients who received surgery/gamma-knife surgery only, and 2.1 months for those who received no treatment (P < 0.001, Table 2, Fig. 3). The most common sites of extracranial disease were the lung, bone and liver. Univariate analyses revealed that patients with bone metastases at the time of brain metastases had shorter survival than patients with metastases at other sites (P = 0.02).

Fig. 3
figure 3

Overall survival for different treatment modalities: surgery/gamma-knife treatment (black dashed line, n = 29); whole brain radiation therapy (WBRT) (gray dashed line, n = 157); no treatment (black solid line, n = 9)

Full size image

Multivariate analysis showed that ECOG performance status (RR = 0.704, 95% CI 0.482–1.028, P = 0.069), number of brain metastases (RR = 0.682, 95% CI 0.459–1.014, P = 0.058), treatment modalities (RR = 1.686, 95% CI 1.022–2.781, P = 0.041), and palliative systemic chemotherapy after brain metastases (RR = 1.871, 95% CI 1.353–2.586, P < 0.001, Table 3) were the only significant predictors for survival.

Table 3 Multivariate analysis of factors prognostic for survival
Full size table

Brain metastases as initial site of recurrence

A total of 62 patients (31.3%) had brain metastasis as their first site of recurrence, and 21 of these patients had brain metastasis without extracranial disease. Of these 21 patients, 7 had solitary brain metastasis and 14 had multiple brain metastases. Of the seven patients with solitary brain metastasis, two had surgical resection, four had gamma-knife surgery without additional WBRT, and one patient had resection followed by WBRT. Of the 14 patients with multiple brain metastases, 12 received WBRT and 2 received gamma-knife treatment. Median survival was 19.3 months for the patients with solitary metastasis and 7.0 months for those with multiple brain metastases (Table 4, P = 0.008).

Table 4 Treatment modalities and secondary relapse sites in patients with brain metastasis (n = 21) as initial recurrence without extracranial disease
Full size table

At the time of this analysis, 4 of the 7 patients with solitary brain metastasis were alive, (survival of 12.0, 13.2, 16.6 and 40.0 months). Three patients died; one of whom had a leptomeningeal relapse at 11.1 months and two who had brain relapses at 14 and 16 months, respectively.

Systemic treatment after brain metastases

About half of the patients received systemic chemotherapy with or without hormonal treatment after the diagnosis of brain metastases. The median survival was 7.8 months (range 0.1–28.7 months) in patients who received palliative chemotherapy versus 3.6 months (range 0.1–44.6 months) for those who did not (Fig. 4, P = 0.001).

Fig. 4
figure 4

Systemic chemotherapy after brain metastasis: palliative chemotherapy (black dashed line, n = 89), no treatment (black solid line, n = 104)

Full size image

Discussion

In the present study, we retrospectively reviewed the clinical features and treatment of breast cancer patients with brain metastases. This is a single institutional study of Korean patients analyzing the clinical features, survival and prognostic factors of brain metastasis. The median age of the patients at the time of first diagnosis of brain metastases was 45 years. This was not unexpected because the median age of breast cancer at our institution for the period of 1989–2004 was 46 years, and it is well known that the peak incidence of breast cancer in Korea occurs at an earlier age than in western countries [18, 19]. The median interval from primary diagnosis to brain metastasis was 32.3 months, which is in agreement with previous reports [2, 10]. DiStefano et al. reported a median interval between primary diagnosis and CNS metastasis of 33 months.

Several small-scale studies have reported that young age [20, 21] and negative hormone receptor status [22–24] are the main risk factors for brain metastasis. Recently, Tham et al. reported a comparative analysis of risk factors in CNS metastasis patients (n = 383) compared with patients who had no CNS metastasis (n = 2302). They found that ER-negativity, young age, p53 mutation and EGFR, but not Her-2, overexpression were associated with increased risk. They also found that lymph-node status and adjuvant chemotherapy had little effect on the risk of brain metastasis [25]. Interestingly, of the above factors, only HER-2 status had an effect on survival after brain metastasis.

We did not perform a comparative analysis to identify the risk factors for brain metastasis compared with no brain metastasis. However, we did investigate characteristics that could have affected the tumor-free interval from primary breast cancer to brain metastasis, and we found that certain features such as young age, ER-negativity, PR-negativity, and high histology grade were associated with shorter intervals. These findings, indicating that certain factors are associated with increased risk of early brain metastasis, may have similar significance to the results of Tham et al. [25]. Also in agreement with the results of Tham et al., the tumor characteristics associated with a shorter interval to the development of brain metastasis did not affect survival.

In the present study, the overall median survival of the 198 patients was 5.6 months (95% CI 4.7–6.5 months), with only 23.1% of patients surviving for more than 1 year. Tham et al. reported that survival after CNS metastases was poor, with a median survival of 5.5 months, and only 25% of patients were alive at 1 year. Although their study included a large number of patients, they did not investigate whether performance status, the number of brain metastases or the treatment modality were prognostic indicators of survival. Lentzsch et al. used multivariate analysis to show that the performance status, number of brain metastases, dose of radiation and primary tumor size were prognostic factors for survival [26]. In the present study, patients with solitary metastasis also had a significantly better outcome than those with multiple metastases (9.2 months versus 5.3 months, respectively).

Certain treatment modalities seem to be associated with better outcome. The median overall survival time was 5.4 months for patients who received WBRT, 14.9 months for those who received surgery/gamma-knife treatment, and 2.1 months for those who received no treatment. However, this finding should be interpreted with care because the choices of treatment were highly correlated with, and dependent on, the disease status.

About half of the patients received systemic treatment such as palliative chemotherapy after the diagnosis of brain metastases. Patients who received systemic treatment after local treatment for brain lesion had a longer survival than the patients who did not: 7.8 months versus 3.6 months, respectively (P = 0.001). However, this finding warrants further prospective study.

In summary, we found that ECOG performance status, number of brain metastases, treatment modalities, and palliative systemic chemotherapy after brain metastasis were significantly associated with survival. The features of the primary breast lesion did not affect survival once brain metastasis had occurred; however, patients with solitary brain lesions should be managed aggressively.