Breast cancer is the solid tumor most commonly associated with leptomeningeal metastases (LM) [1]. Although the absolute incidence is low, patients with LM often present with debilitating symptoms, including pain, headache, cranial nerve palsies, limb weakness, and alterations in mental status, and therefore the clinical implications are significant [2]. Unfortunately, despite multimodality therapy, the prognosis is generally poor. At present, standard therapies for LM do not take into account tumor subtype, though targeted, subtype-specific therapies are increasingly being employed in the treatment of patients with advanced breast cancer without active CNS involvement.

In this context, as reported in this issue of Breast Cancer Research and Treatment, Park and colleagues set out to describe concordance in HER2 status between primary tumors and subsequent LM. Notably, of 49 patients diagnosed with LM on the basis of positive cytology at their institution, 32 patients had inadequate cellularity (less than 20 malignant cells per slide) for testing by fluorescence in situ hybridization (FISH), and one patient was excluded because of lack of information on the primary tumor, leaving only 16 cases (33%) for analysis. Additionally, although immunohistochemistry (IHC) for HER2 was available for primary tumors in all cases, FISH results from the primary tumor were only available for comparison in two cases (who had 2+ IHC). Nevertheless, and not unexpectedly, the authors observed a high level of concordance for HER2 in LM, compared to matched, primary tumors. Specifically, of four patients with evidence of HER2 amplification in malignant cells from the CSF, all four were IHC 3+ in the primary tumor. Eleven of 12 HER2-negative cases were also concordant. Closer examination of the data indicates that the single “discordant” case occurred in the setting of IHC 2+ and FISH ratios that were quite similar between primary and LM (HER2 to CEP17 ratio = 2.1 and 1.74, respectively). Given the potential for tumor heterogeneity and sampling issues related to the limited amount of CSF sample, it is not clear that the case truly represents a functional shift in tumor subtype.

This study adds to the literature indicating high concordance of HER2 expression between primary tumors and their corresponding metastases. In a series of 29 patients with breast cancer who had matched specimens from primary tumor and brain metastasis (BM), the overall concordance rate for HER2 was 97% [3]. A similar study of 43 patients treated at Memorial Sloan Kettering Cancer Center revealed no significant differences in ER, PR, HER2, EGFR, or CK5/6 expression between paired primary and BM specimens [4]. Looking more generally at other sites of metastasis, the data indicate conservation of HER2 status in most cases [5]. MacFarlane et al. [6] examined 160 paired specimens and found only nine cases (5.6%) in which the HER2 status changed. Liedtke et al. [7] examined 211 paired specimens and found that the correlation coefficient for HER2 FISH was 0.895. Even when multiple metastases are tested in a single patient, HER2 status is generally conserved across lesions [8]. Collectively, the data substantiate the concept that HER2 is a central player in tumor maintenance, including in patients with CNS relapse.

What is the clinical relevance of the findings? First, how common are LM in women with HER2-positive breast cancer? The same group has previously characterized the clinicopathological characteristics of 805 patients with metastatic breast cancer treated at the National Cancer Center Hospital (Goyang, Korea), of whom 126 went onto develop CNS recurrences [9]. Although 29% of patients with HER2-positive breast cancer relapsed in the CNS, the incidence of LM was only 3%. Similar results were observed in a series of 122 patients treated with trastuzumab for metastatic breast cancer at Dana-Farber Cancer Institute, where only two patients (5%) presented with LM in the absence of parenchymal lesions [10]. The largest experience comes from the RegistHER observational study, in which the reported incidence of LM among 1,009 patients with HER2-positive metastatic breast cancer was 2.3% [11]. By contrast, LM involvement appears to be somewhat more frequent in patients with triple-negative or invasive lobular tumors [12–16]. In summary, the available data indicate that LM involvement is a relatively rare manifestation of HER2-positive breast cancer.

From a therapeutic standpoint, data from this small study generally support the practice of basing treatment upon characteristics of the primary tumor. In their discussion, the authors take this one step further, and postulate that their findings justify consideration of intrathecal (IT) trastuzumab, and cite responses in patients with HER2-positive LM that have been reported elsewhere in the literature [17–19]. With respect to the use of IT trastuzumab, several points should be noted. First, the present study was limited to a description of HER2 concordance and did not specifically evaluate treatments for LM. No prospective trials have investigated subtype-specific treatments for LM from breast cancer. Unfortunately, given the small patient population, large, or even moderately-sized clinical trials targeted to this population are unlikely to ever be mounted. Next, close reading of the case reports indicates that many (though not all) patients with reported responses to IT trastuzumab also received multiple concurrent therapies, including IT methotrexate, IT thiotepa, and systemic chemotherapy, making it difficult to isolate an effect of trastuzumab [17–19]. Most importantly, trastuzumab is not formulated for IT administration and per the manufacturer, “should not be given via intra-CSF administration” [20]. The formulation contains the excipient trehalose, and the provided diluent contains 1.1% benzyl alcohol as a preservative, both of which can make IT administration potentially unsafe. In terms of other approaches, capecitabine has reported activity against LM, both alone, and in combination with intravenous trastuzumab [21, 22]. Whether the combination of lapatinib and capecitabine would offer additional value is unknown, though the combination has some activity against parenchymal brain metastases [23, 24].

From a translational perspective, there is a significant opportunity to evaluate tumor cells collected from CSF for markers beyond ER, PR, and HER2. In their paper, Park and co-workers demonstrate that FISH is feasible by destaining of previously stained CSF cytology slides, which raises several intriguing possibilities. For example, MET amplification can confer resistance to EGFR targeted therapies via activation of ErbB3 signaling, and is an adverse prognostic factor in breast cancer [25, 26]. Moreover, MET is rapidly upregulated in preclinical models of acquired trastuzumab resistance [27]. Given that the majority of patients with HER2 positive LM will have previously received trastuzumab, retrospective interrogation of the CSF malignant cells for MET amplification may offer an opportunity to evaluate whether this mechanism is operative in vivo. Additionally, in a prospective fashion, (particularly in patients with Ommaya reservoirs), malignant cells could be captured using flow cytometric methods in order to explore serial changes in a broad variety of biomarkers associated with treatment and upon the development of resistance.

In sum, results of this small study indicate a high level of concordance in HER2 status between primary tumors and LM. The study was not designed to evaluate whether HER2-directed therapies in patients with HER2-positive LM offer additional value over standard therapies alone, and should not necessarily change clinical practice. However, the study does point out the feasibility of molecular characterization of malignant cells from CSF and opens up opportunities for further investigations.