Brain Metastases from Breast Cancer: Focus on Risk Factors, Treatment, and Clinical Outcome

Introduction

Among the worldwide female population, breast cancer accounts for 29% of the total new cancer cases diagnosed and 15% of the cancer deaths. To date, breast cancer is the most frequently diagnosed tumor and the second leading cause of cancer death among women ( ).

The American Cancer Society assessed that 160,000 women in the United States are living with metastatic breast cancer with an estimated median survival rate within the range of 3 years. Over the last four decades, the survival of breast cancer patients has increased mainly due to early detection and improvement in adjuvant treatment. Initial reports suggested an improvement in metastatic breast cancer survival between 1974 and 2000 ( ). However, starting from the late 1990s, a significant shift toward more unfavorable risk factors was observed in metastatic breast cancer, and patients developing metastatic disease seem to have a more aggressive disease ( ).

The vast majority of brain metastases are detected due to symptoms suggesting central nervous system (CNS) involvement, such as headache, vomiting, dizziness, and ataxia ( ). According to the literature, the incidence of brain metastases in breast cancer patients ranges from 10% to 16% ( ); however, the total incidence of brain metastases may be underestimated. The CEREBEL trial, a trial specifically testing whether lapatinib plus capecitabine was superior to trastuzumab plus capecitabine in prevention of CNS lesions in metastatic breast cancer patients after trastuzumab failure, carefully screened asymptomatic patients for brain metastases. Asymptomatic brain metastases were detected in almost 20% of patients ( ).

A review of the literature showed that the risk of developing brain metastases increased with increasing stage at the initial diagnosis. Only 2.5% of patients diagnosed with early disease developed CNS disease, whereas 7.6% of patients diagnosed with regional disease and 13.4% of patients presenting with stage IV disease were found to have brain involvement ( ).

More than 80% of brain metastases were detected later during the course of metastatic disease. The diagnosis of CNS involvement as first manifestation of breast cancer or synchronously with first diagnosis of systemic disease could be considered a relatively rare event. Overall, one-third of patients were diagnosed with polymetastatic brain disease (four or more metastatic lesions), one-third with oligometastatic brain disease (between two and three lesions), and one-third with one single brain lesion ( ).

Risk Factors for Brain Metastases Development

A large study with a median follow up of 15 years showed that the incidence of brain metastases varied in different breast cancer subtypes. Patients with epidermal growth factor receptor 2 (HER2)-positive disease had the highest incidence of CNS involvement, followed by triple negative (hormone receptor (HR)-negative and HER2-negative) both basal-like and nonbasal-like, luminal B–like (HR-positive, HER2-negative, high proliferation rate), and luminal A–like (HR-positive, HER2-negative, low proliferation rate) subtypes ( ).

In the Graesslin nomogram ( ), age, histological grade, HR status, HER2 status, number of noncentral nervous system metastatic sites, and interval between date of primary diagnosis and date of first metastases were associated with the risk of developing brain lesions. Several data supported the same results ( ) and young women with HR-negative, high proliferation rate, and/or HER2-positive disease were confirmed to be the subgroup with the higher risk of developing brain metastases.

A study that evaluated the correlation between quantitative HER2 protein expression ratio by fluorescence in situ hybridization and risk for brain metastases in 142 consecutive patients who received trastuzumab (a humanized monoclonal antibody directed at the HER2 ectodomain)-based therapy for HER2-positive metastatic breast cancer showed that time to brain metastases strongly correlate with total HER2 protein expression ( ). HER2 positivity seems to be the strongest risk factor for brain metastases ( ). Indeed, among 413 patients who died during follow-up on the HERA (HERceptin adjuvant) trial, 47% were diagnosed with CNS relapse in the trastuzumab arm and 57% in the control arm (Pestalozzi et al., 2013). Moreover, among patients with HER2-positive breast cancer, the risk of brain metastases is further elevated in the setting of HR negativity. In a recent retrospective analysis of 473 metastatic breast cancer patients ( ), estrogen receptor (ER) and progesterone receptor (PR) negativity were associated with an increased risk of development of CNS lesions (odds ratio (OR) 4.06, 95% CI 2.38–6.91; P < 0.001 and OR 2.21, 95% CI 1.31–3.71; P = 0.003, respectively).

The loss of HR positivity seems to be a common event in brain metastases. Duchnowska and colleagues evaluated the conversion rate of ER and PR receptor from primary tumor to corresponding brain metastases in a consecutive series of 120 breast cancer patients ( ) and found the conversion rate of ER and PR in brain metastases was 29%, mostly from positive to negative. Apparently, patients who received endocrine therapy had the highest conversion rate from hormone receptor positive to negative disease. However, it is important to note that the conversion did not affect survival. The authors also investigated HER2 status; conversion of HER2 occurred in 14% of patients and it was more balanced either way. No impact of the use of chemotherapy or trastuzumab was observed. Patients with metastatic triple-negative breast cancer were also at high risk of brain metastases, with an estimated incidence from 25% to 46% ( ).

One of the major differences between HER2-positive and triple-negative breast cancer patients seems to be the time to brain relapse. A hospital-based tumor registry of 2441 women with breast cancer treated between 1998 and 2006 showed that the median time to development of brain metastases was shortest in the triple-negative group (22 months), followed by the HER2-positive group (30 months), followed by the luminal group (63.5 months) ( ). In contrast, brain involvement in patients with triple-negative disease frequently occurs simultaneously with extracranial disease progression ( ). There is also evidence to suggest that once brain metastases are discovered, patients with triple-negative tumor have a shorter time to death.

Mechanisms of Brain Dissemination

The number of different genes expressed among subtypes of breast cancer confirms that the underline biology of the subtypes is different. This heterogeneity also plays an important role in the metastatic potential and pattern of relapse of the different subtypes. For example, brain metastases occur most frequently in nonluminal subtypes, liver relapse is associated with HER2-positive tumors, and lung relapse occurs often within the basal-like subtypes.

The mechanisms underlying brain dissemination remain poorly understood. Through comparative genome-wide expression analysis, a 17-gene signature was identified as associated to brain dissemination in two independent breast cancer datasets ( ). In particular, ST6GALNAC5, a sialyltransferase, was found to facilitate transmigration across the blood–brain-barrier and to be specifically involved in the brain metastatic potential, whereas the cyclooxygenase (COX) 2 and the epidermal growth factor receptor (EGFR) ligand HBEGF also conferred lung metastatic potential. Other intracellular pathways associated with metastases to the brain include CXCR4/CXCL12, VEGF, PI3K, and Notch. An extensive analysis detected 37 proteins differentially expressed between primary breast tumors with brain metastases and primary breast tumors without brain metastases. In particular, the combination of GRP94, FN14, and inhibin was observed to have a higher probability to predict brain metastases compared with HER2 alone ( ).

Once tumor cells have migrated across the blood–brain-barrier, an essential step of metastases formation is at the vascular branch points. The brain microenvironment represents a unique system of stromal elements, such as pericytes, astrocytes, and glial cells, which may interfere with both metastatic colonization and therapeutic response. In particular, in vitro culture experiments suggest that a brain inflammatory response with extensive reactive gliosis may play a role in promoting brain metastases colonization and growth ( ).

Prognosis of Breast Cancer Patients With Brain Metastases

In clinical practice, the patient’s overall general condition is generally defined by the performance status, using either the Karnofsky performance status (KPS) score or the Eastern Cooperative Oncology Group (ECOG) system. The Radiation Therapy Oncology Group (RTOG) Recursive Partitioning Analysis (RPA) was developed to assist in assigning patients with newly diagnosed brain metastases to the appropriate prognostic group to guide treatment decision making and clinical trial design. The scale assigns patients with brain metastases to RPA class 1, 2, or 3 based on KPS, age, and burden of disease. Patients with KPS score 70–100, age less than 65, controlled primary, and metastases to the brain only are classified as RPA class 1; patients with KPA less than 70 are classified as RPA class 3; and all others are classified as RPA class 2. According to the literature, patients in class 1 have a median survival of 7.1 months, class 2, 4.2 months, and class 3, 2.3 months ( ).

More modern prognostic scales have been developed including the graded prognostic assessment (GPA). This scale assigns a score of 0, 0.5, or 1 for age, KPS, number of CNS lesions, and presence or absence of extracranial metastases. The sum score is then used to determine the median survival in months, ranging from 2.6 months for a score of 0–1 to 11 months for a score of 3.5–4.0. Comparison of the RPA, GPA, and other prognostic indices found the GPA and RPA to have the most statistically significant difference between categories. In addition, the GPA was the least subjective, most quantitative, and easiest to use. The GPA has been further refined and a diagnosis-specific GPA for breast cancer assigns scores based on KPS, age greater than or less than 60, and breast cancer subtype ( ) ( Table 21.1 ). The sum of the points for each prognostic factor is the GPA for an individual patient. Patients with a GPA score of 0–1.0 have a median survival time (mST) of 3.4 months, patients with a GPA score of 1.5–2.0 have an mST of 7.7 months, patients with a GPA score of 2.5–3.0 have an mST of 15.1 months, and patients with a GPA score of 3.5–4.0 have an mST of 25.3 months.