Introduction

Breast cancer is the second leading cause of brain metastases (BM) after lung cancer, and the leading cause of BM in women [1]. Up to 30% of the patients with progressive breast cancer will develop BM [1, 2], most often within an interval of 1–3 years after initial breast cancer diagnosis [1, 3]. In contrast, less than 0.5% of the patients are diagnosed with BM at time of their primary breast cancer diagnosis [4].

The primary breast cancer molecular subtype [as determined by the expression of estrogen receptor (ER), progesterone receptor (PR) and human epidermal growth factor receptor 2 (HER2)] is a significant clinical factor that may predict systemic treatment efficacy and patient’s prognosis [5]. The current recommendations in the metastatic setting are to consider evaluation of receptor phenotype in a metastatic lesion to determine the possible discordance with the primary tumor, as such changes may have clinically relevant significance for treatment decision making [6]. However, there is limited data about the discordance rates between the breast cancer primary and the BM, [7,8,9] and the potential clinical implications of such findings.

A pooled analysis from 48 studies reported a discordance between the primary tumor and systemic metastases rate of 20% for ER [95% confidence interval (CI) 16–35%], 33% for PR (95% CI 29–38%), and 8% for HER2 (95% CI 6–10%). For all ER-, PR-, and HER2-enriched patients, the proportion of tumors shifting from positive expression to negative was higher than the rate of gaining expression of a certain receptor. However, the analysis did not indicate the site of the recurrent/metastatic tumor [10].

In the current study, we performed a multi-institutional data analysis to evaluate the discordance rates for the expression of ER, PR, and HER2 between the primary breast tumor and subsequent BM.

Methods

This was a retrospective study, pooling data collected by 11 institutions in 8 different countries. The centers were initially approached in June 2016, and data collection was completed in November 2016.

The data were collected from medical records of patients with breast cancer who were diagnosed with BM. Only patients for whom the pathology specimens and/or report from both the primary lesion and the BM were available (including evaluation of receptor status) were included in this study. The information was recorded in a predefined-standardized format, without patient identifiers. Each institution was responsible to follow their institutions’ ethical guidelines and gain ethics committee approval for the study.

Data collected included: patient-related factors (e.g., demographics); primary disease-related factors (e.g., histology type, receptor status, proliferation index, and stage at diagnosis); treatment-related factors (e.g., type of systemic therapy, given at diagnosis as well as for metastases); BM-related factors (e.g., date of occurrence, receptor status, proliferation index when available); and evaluation of receptor status from systemic metastases (i.e., extra-cranial) within the 3-month range from evaluation of BM (before or after).

Hormone receptor [HR, estrogen and/or progesterone] status (positive or negative), HER2 overexpression, and proliferation index rate (Ki67) were determined according to institutional guidelines [immunohistochemically (IHC) and/or fluorescence in situ hybridization (FISH)]. The The Mann–Whitney test was used to analyze the differences in the timing from the diagnosis of the primary tumor to develop BM. Bonferroni correction was applied, thus p < 0.025 was considered statistically significant. R version 3.2.3 was used for statistical analyses.

Results

A total of 167 breast cancer patients that developed BM was included. Patients’ demographics and disease-related information are listed in Table 1.

Table 1 Patients’ demographics and disease-related information
Full size table

Out of 167 breast cancer patients with BM, 129 patients (77%) had complete receptor information (ER, PR, and HER2) of both primary tumor and BM. Of these, 25 patients (19%) had a change in receptor status (Table 2): 17/129 (58.6%) patients with HR-positive primary had ER/PR change, 11/129 (8.5%) with HER2-positive primary had HER2 change. The proportion of tumors shifting from positive expression to negative was higher than the rate of gaining expression of a certain receptor (13% vs. 8%). Table 3 summarizes loss or gain of receptor for BM compared to the primary tumor in the whole group (n = 167).

Table 2 Out of 129 patients that had complete receptor information from the primary tumor and brain metastasis; 25 (21%) patients had discordance between the primary and the brain metastasis
Full size table
Table 3 Loss or gain per receptor for brain metastases compared to the primary tumor
Full size table

The median time to BM according to the primary tumor subtype was 5.6 years (range 0–13) for HR-positive/HER2-negative; 4.2 years (range 0–17) for HR-positive/HER2-positive; 3 years (range 0–47) for HR-negative/HER2-positive; and 2.3 years (range 0–17) for triple-negative primary breast tumors. These differences were not statistically significant for triple negative (TG) versus all others (p = 0.073), but were found significant for TG or HR negative/HER2 positive versus HR positive/HER2 negative or HR positive/HER2 positive (p = 0.0014).

Median time from diagnosis of primary breast cancer to BM in patients without receptor conversion was 3 years (range 0–47) compared to 2.45 years (range 0.6–17) in patients with conversion (p = 0.31).

Of the total 167 patients, 51 had an available Ki67 index for the primary tumor, of which 44 patients (86%) showed a Ki67 index of > 20% and 7 patients (14%) of < 20%. A total of 36 patients had a Ki67 index for both the primary tumor and the BM. Of these, 29 patients (80.5%) had a Ki67 index > 20% in both primary and BM tumors. Discordance in Ki67 was found in 7 patients: in 5 patients (14%) the primary tumor Ki67 index was < 20% while the BM Ki67 index was > 20%, and in 2 patients (5.5%) the primary tumor Ki67 index was > 20% while the BM Ki67 index was < 20%.

Change in receptor expression in relation to Ki67 index was seen in 5 patients [out of the 29 patients (17%) with both primary and BM with Ki67 > 20%]: 4 had loss of HR and 1 had gain of HR and in 2 patients [out of 5 patients (40%) with higher Ki67 index in BM]: 1 had loss of HR and 1 had gain of HR. None of the patients in which the Ki67 index in the BM was lower than the primary tumor had changes in receptor expression.

Out of 167 patients, five patients (3%) have had an assessment of receptor status from systemic metastases within a 3-month period from the evaluation of BM. Data of full receptor status of the primary tumor, systemic metastases (extra-cranial, within a 3-month period from BM), and BM was available for four patients; one had all 3 receptors in agreement (positive expression of all three receptors in primary, systemic metastases, and BM); one patient had a TN phenotype of the primary and gained HER2 overexpression in both systemic and brain metastasis.

Out of the 25 patients with changes in BM receptor expression, data about systemic therapy changes were available for 15 patients (60%). Of these, for 12 patients (80%), the systemic therapy was modified according to the BM receptor status.

Discussion

Breast cancer is one of the most common cancers and comprises a large fraction of our patients. It is widely accepted that receptor discrepancies between the primary breast tumor and its systemic metastases (extra-cranial) exist, and that it might have significance for treatment decisions [6]. Yet, the molecular phenotype of BM has received little attention in the scientific literature, even though craniotomy/biopsy (therefore accessibility to evaluate the receptor status of the BM) is often part of the diagnosis and treatment of these patients.

Our study represents the largest cohort evaluating discrepancies in biological marker expression of primary breast cancer and subsequent BM. A major finding of our study is that one out of five (20%) patients with breast cancer BM had a receptor discrepancy between the primary tumor and the subsequent BM, with loss of hormone receptors (ER and/or PR) expression, and gain of HER2 overexpression as the most commonly observed changes.

Several studies have reported similar findings with up to a 30% of discordance rate between the primary breast tumor and subsequent BM [2, 8, 9, 11, 12]. Gaining expression of HER2 in the BM was estimated to be in up to 16–18% of the cases [8, 9].

The observation that the common changes are loss of hormone receptor expression and gain of HER2 overexpression in BM compared to the primary tumor was also supported by a recent novel report published by Priedigkeit and colleagues using gene expression testing [7]. In that report, HER2 alteration was the most frequent observed change, showing an at least twofold increase in mRNA expression in BM compared to the primary breast tumor [7]. The authors indicated that there was a robust and significant enrichment of HER2 alterations, specifically in breast cancer BM (24%) and that these changes were not found to be significantly overexpressed in other metastatic sites (13%) [7]. Moreover, the Estrogen Receptor Gene 1 (ESR1) was the most recurrently down-regulated gene, with twofold and fourfold expression decrease in 9 out of 20 samples evaluated [7]. These findings might indicate a clonal selection favoring TN- and HER2-positive cancer cells that preferentially seed the CNS and develop BM [13,14,15].

Similar to other studies, the predominant histological primary tumor in our cohort was invasive ductal carcinoma, and 86% of the primary tumors had high proliferation index (Ki67 > 20%), suggesting an aggressive behavior of highly proliferating primaries [9, 15, 16]. However, our small sample size did not allow us to further analyze whether this finding is important for receptor conversion.

A limitation of our study is that a central review of the specimens was not performed; therefore, some of the reported discordances might be a result of inter- and intra-observer variability in assays between the two sets of samples tested (primary versus metastases). Yet, the rate of such variabilities has been estimated in the literature to be only around 6% [17] and our results were supported by the findings using gene expression testing of BM versus primary breast cancer [7].

It is well-accepted that breast cancer patients are a heterogenous group and that the molecular subtype is an important prognostic and predictive factor. The primary subtype has also significance for the timing of developing BM from primary diagnosis. With shortest time to develop BM seen in the TG and for HR-negative/HER2-positive patients. Moreover, a recent population study showed that the incidence of BM at time of breast cancer diagnosis was highest in these two groups and suggested to consider brain-imaging as screening for this population [4].

Additionally, the primary subtype is a known prognostic for survival in breast cancer patients with BM, and integrated in the graded prognostic assessment [18]. Therefore, we should strive to better understand if discrepancies in receptor expression contribute to these differences in patients’ survival. In some cases, these discrepancies may represent an undiagnosed extra-cranial metastatic phenotype that necessitate treatment modification, while in others, it might explain the unresponsiveness of intracranial disease to systemic therapy. There are numerous preclinical and clinical studies evaluating the activity of systemic therapy for controlling breast cancer BM, with early reports that systemic therapy may contribute to intracranial disease control [19,20,21], therefore, exploring these changes and better understanding resistance pathways may provide clinical benefit for these patients.

At this time, we advise receptor status assessment of BM in all breast cancer patients with available histology as it may have significant implications for therapy. For patients with receptor conversion (e.g., HER2 gain), we advise to consider adapting systemic therapy accordingly. Further studies are needed to understand the clinical implications of these changes and potential treatments.