Introduction
The incidence of brain metastases in patients with metastatic breast cancer (MBC) appears to be increasing over time, due in part to improved control of systemic disease, prolonged survival, and enhanced detection [
1]. Brain metastases lead to substantial morbidity—in patients with MBC, including cerebral edema, headaches, seizures, motor impairment, speech difficulty, and mental disturbances [
2]. Systemic therapy has limited efficacy in treating brain metastases, possibly due to poor penetration of the blood–brain barrier (BBB), expression of drug efflux pumps in the BBB, enriched abundance of ErbB ligands, or acquired resistance following treatment with multiple prior regimens [
3]. In addition to systemic therapy, standard treatments for brain metastases include whole-brain radiation, stereotactic radiosurgery, and, for eligible patients with solitary lesions, surgical resection [
4]. Despite these interventions, median overall survival (OS) is poor, ranging from 3–30 months, depending on breast cancer subtype and treatment [
5].
Approximately 20% of all breast cancers overexpress human epidermal growth factor receptor 2 (HER2) [
6,
7]. In RegistHER, a prospective, observational study of 1012 patients with HER2-positive MBC, central nervous system (CNS) metastases were documented in 37.3% of patients over a median follow-up of 29 months [
8]. Other analyses found brain metastases to be present in up to 55% of patients with HER2-positive MBC [
5,
9]. Amplification and/or overexpression of HER2 may play a role in the occurrence or progression of brain metastases. In a mouse model of MBC, HER2 overexpression promoted outgrowth of breast tumor-derived brain metastases, with the number of large brain metastases increasing 3-fold in mice inoculated with high HER2-expressing versus low HER2-expressing human breast cancer cells [
10]. In a study of more than 600 patients with MBC, HER2-positivity was a significant and independent risk factor for subsequent development of brain metastases [
11]. Moreover, in a study comparing HER2 mRNA levels in unlinked archival brain metastases and primary breast tumors, HER2 mRNA was found, on average, to be 5-fold more abundant in brain metastases than in primary tumors [
10].
HER2-targeted agents, such as trastuzumab [
8,
12], lapatinib [
13,
14], and trastuzumab emtansine (T-DM1) [
15,
16], have been shown to improve outcomes in patients with HER2-positive MBC and CNS metastases, including leptomeningeal or brain parenchymal lesions [
7,
8]. In registHER, patients administered trastuzumab exhibited a median OS of 17.5 months from the date of CNS disease diagnosis compared with 3.8 months for patients not receiving trastuzumab [
8]. Moreover, multivariate analysis showed trastuzumab to be a significant independent predictor of survival [
8]. In another retrospective study of women with HER2-positive MBC and CNS metastases, median OS was 11.6 months among those who received trastuzumab at the time of, or prior to, CNS lesion diagnosis compared with 6.1 months among women who did not receive trastuzumab (
p = 0.03) [
12]. It is unclear, however, whether the improvements in OS stem from control of systemic, extra-cranial disease, or from direct effects of trastuzumab on brain lesions.
Although efficacy of systemic therapy for treating brain metastases may be limited by the inability of HER2-targeted therapies to access the brain, animal studies show that the BBB is likely compromised by brain lesions [
17]. Moreover, in patients with HER2-positive breast cancer, accumulation of trastuzumab was 17.5-fold higher in brain metastases than in normal brain tissue [
18]. As it is unclear what role access plays in treating brain metastases in patients with HER2-positive MBC, we investigated the extent of trastuzumab delivery, as well as efficacy of trastuzumab alone or in combination with a PI3K (phosphatidylinositol 3-kinase) inhibitor, and T-DM1 in experimental models of HER2-positive brain lesions.
Discussion
In retrospective clinical studies, trastuzumab was demonstrated to prolong OS in patients with brain metastases from HER2-positive MBC [
4,
8,
12]. It remains unclear if increased OS is due to effective control of systemic, extra-cranial disease or a more direct effect on brain metastases [
4,
12]. A key outstanding question has been to what degree access to the brain impacts the efficacy of trastuzumab in these lesions.
Although the Fo5 and Fo2-1282 models involve direct intracranial injection of tumor cells, rather than seeding of brain lesions from systemic circulation, these models were selected for evaluation of HER2-targeted agents in established brain metastases. Steeg et al. [
3] demonstrated the involvement of HER2 signaling in outgrowth of breast cancer-derived experimental brain metastases, but not in initiation of these lesions [
10]; outgrowth therefore appears to be the key process to capture in a model used to evaluate the effects of HER2 inhibition. This outgrowth process is well represented in the Fo5 and Fo2-1282 models.
The Fo5 and Fo2-1282 models used in our experiments also differ from those utilized by Steeg et al. in a significant way: while the model tumors based on 231-BR cells are grown in athymic nude mice [
26,
27], our models use immune-competent mice and therefore lend themselves particularly well to evaluating the efficacy of trastuzumab, an antibody whose mechanism of action includes the recruitment of immune effector cells and activation of antibody-dependent cell-mediated cytotoxicity or ADCC [
28]. Overall, the efficacy results reported here are similar to those obtained by Steeg et al. [
3] with another HER2-targeted agent, lapatinib, in an animal model of breast cancer-derived brain metastasis. As with trastuzumab, lapatinib inhibited metastatic outgrowth of brain lesions at high doses, but inhibition was incomplete [
26].
In the Fo2-1282 mouse model of HER2-positive breast cancer-derived brain metastases, 3-fold higher systemic doses of muMAb4D5/trastuzumab were required to achieve efficacy in brain tumor grafts compared with those in the mammary fat pad. The reduced efficacy of muMAb4D5 in treating brain grafts did not appear to result from lack of access, as PET imaging showed 89Zr-trastuzumab to localize equivalently in brain and mammary grafts. Furthermore, 89Zr-trastuzumab localization in the HER2-positive tumor graft was significantly greater compared to normal brain tissue and muMAb 4D5 was demonstrated to accumulate in Fo2-1282 brain grafts at known therapeutic concentrations.
There are several hypotheses put forward as to the reduced efficacy of trastuzumab in Fo2-1282 brain lesions compared with mammary tumors: the inability of immune effector cells to access the brain lesion, thereby impairing ADCC; the presence of ErbB ligands in the brain microenvironment circumventing HER2 inhibition by trastuzumab; or activation of compensatory signaling pathways. Although we did not directly investigate the access of effector cells to brain grafts, previous reports show the activity of immune cell-dependent therapies in preclinical models of glioma [
29], as well as in patients with melanoma brain metastases [
30]. These observations are consistent with our hypothesis that lack of access to effector cells does not explain reduced trastuzumab efficacy in brain grafts.
Our results further suggest that the reduced efficacy of muMAb 4D5 in Fo2-1282 brain grafts compared with mammary tumors may arise from incomplete pathway suppression or hyper-activation of downstream signal transduction pathways, as combined treatment with muMAb 4D5 and the brain-penetrant PI3K/mTOR inhibitor GNE-317 was more effective than either drug alone. One possible explanation for the diminished muMAb 4D5 response is the presence of brain-specific ligands that mediate resistance to HER2 inhibition. Multiple redundant HER family ligands mediate insensitivity to trastuzumab or other HER2-targeted agents [
31], and it is likely that signaling driven by these ligands converges on important downstream cell-survival pathways such as PI3K. Recent reports suggest that resistance to anti-cancer tyrosine kinase inhibitors is frequently triggered by the presence of additional receptor tyrosine kinase ligands [
31], and in vitro experiments suggest that ligand-driven activation of alternative receptors in the HER family may present a recurrent mechanism of resistance in breast cancer cells [
32].
An alternate approach to circumventing insensitivity to anti-HER2 therapies is to target a potent cytotoxic agent to HER2-positive tumors by utilizing a HER2-directed ADC. T-DM1 was demonstrated to have superior anti-tumor activity compared with trastuzumab in HER2-positive preclinical models [
25]. Improved survival after T-DM1 treatment was demonstrated in mice with experimental lesions from the Fo5 model, a model that does not respond to trastuzumab [
25]. In the phase III EMILIA trial of patients with HER2-positive MBC previously treated with trastuzumab and a taxane, T-DM1 showed improved PFS and OS compared with lapatinib plus capecitabine [
33]. Importantly, in a subset of EMILIA study participants with asymptomatic CNS metastases at baseline, T-DM1 was associated with significantly improved survival compared with lapatinib and capecitabine [
34].
In conclusion, our data provide a rationale for the clinical evaluation of higher-dose trastuzumab, T-DM1, or combination therapy with two or more targeted agents for the treatment of brain metastases in patients with HER2-positive MBC. To this end, a trial designed to assess the efficacy of high-dose trastuzumab, combined with pertuzumab, in patients with HER2-positive MBC and CNS progression post-radiotherapy (NCT02536339) is currently enrolling patients.
Acknowledgements
Support for third-party medical writing assistance for this manuscript was provided by Genentech, Inc. The authors gratefully acknowledge the expertise of Kat Parsons-Reponte for propagation of the Fo2-1282 model and Dr. Franklin Peale for histological analysis.