HER2‐positive breast cancer brain metastasis: A new and exciting landscape

Abstract Background Brain metastases (BrM) incidence is 25% to 50% in women with advanced human epidermal growth factor receptor 2 (HER2)‐positive breast cancer. Radiation and surgery are currently the main local treatment approaches for central nervous system (CNS) metastases. Systemic anti‐HER2 therapy following a diagnosis of BrM improves outcomes. Previous preclinical data has helped elucidate HER2 brain trophism, the blood‐brain/blood‐tumor barrier(s), and the brain tumor microenvironment, all of which can lead to development of novel therapeutic options. Recent findings Several anti‐HER2 agents are currently available and reviewed here, some of which have recently shown promising effects in BrM patients, specifically. New strategies driven by and focusing on brain metastasis‐specific genomics, immunotherapy, and preventive strategies have shown promising results and are under development. Conclusions The field of HER2+ breast cancer, particularly for BrM, continues to evolve as new therapeutic strategies show promising results in recent clinical trials. Increasing inclusion of patients with BrM in clinical studies, and a focus on assessing their outcomes both intracranially and extracranially, is changing the landscape for patients with HER2+ CNS metastases by demonstrating the ability of newer agents to improve outcomes.

patients but 30% in autopsy reports. 4,14 The progressive improvement of systemic treatment of breast cancer led to increased survival as reports emerged of a higher incidence of BrM (28%-48%) in stage IV patients treated with trastuzumab. 7 Aiming to characterize CNS progression in patients with breast cancer in the clinical era of trastuzumab, a multicenter cohort of 1012 patients newly diagnosed with metastatic HER2+ breast cancer was followed in a prospective observational study from 2003 to 2006. 15 Overall, 37% (377/1012) of patients developed CNS metastases, 7.5% (75/1012) at the initial diagnosis of metastatic disease, and 10.5% (106/1012) as the sole, initial site of progression. Trastuzumab was the main anti-HER2 therapy available at the time, and only 5.5% of the patients had received it prior to study entry; however, 93% of patients received it during the follow-up period, and before the first diagnosis of CNS metastases. The pivotal clinical trials that evaluated the adjuvant use of trastuzumab reported an overall low incidence of CNS as the first site of metastatic disease, with mixed results regarding a possible protective effect of trastuzumab. 16,17 An analysis of CNS relapses as the first event or at any time in the HERA trial data, which had a median follow-up of 4 years, confirmed that the frequency of CNS relapses as the first recurrence event was similar between the group given 1 year of trastuzumab (2%, 37/1703 patients) and the observation group (2%, 32/1698 patients). 18 Nevertheless, a subgroup analysis of 413 patients with available data regarding sites of progression after initial recurrence showed an increased incidence of CNS relapse in patients that did not receive trastuzumab compared to those who received trastuzumab (57%, 129/227 patients vs 47%, 88/186 patients; P = .06, respectively), again possibly related to improved systemic disease control, resulting in lower rates of CNS seeding. Overall, the magnitude of benefit derived from anti-HER2 systemic therapies in controlling systemic disease with improvements in OS cannot be understated. Moreover, receipt of anti-HER2 therapy has also been associated with improved outcomes following a diagnosis of BrM. In addition, retrospective analyzes have consistently shown better outcomes for patients with HER2+ BrM, especially when also hormone receptor positive, in comparison to those with the triple negative subtype, likely due to the availability of effective, targeted therapies. [20][21][22] 2 | PRECLINICAL STUDIES

| Mouse models of HER2+ BCBrM
Several methods of modeling BCBrM in mice exist, each with its own benefits and caveats regarding the specific source of cancer cells (human vs mouse), tumor generation method (eg, direct intracranial implantation vs systemic inoculation), and analytical approach (eg, bioluminescence vs histology). As seen in Table 1, careful selection of the specific model must match the specific question(s) being tested, and resulting data must be appropriately interpreted based on the methods used. Some examples of important questions to consider include: • Is an intact immune system required for this therapeutic intervention? If yes, then a syngeneic model with mouse cancer cells is needed.
• Is this gene/pathway of interest involved in the early metastatic process (eg, intravasation or colonization)? If yes, then direct intracranial implantation is not appropriate.
• Is the ability to monitor/detect individual BrM cells or micrometastases necessary? If yes, then bioluminescence is not appropriate.
In the field of HER2+ breast cancer, the story of HER2 actually started in the brain and has been extensively characterized in animal models. The role of HER2 in cancer was first identified in a rat brain tumor model, and development of the HER2-targeting MAb trastuzumab in animal models has provided a road map for future antibody-based therapeutics (reviewed in Reference 23). Studies of HER2+ BCBrM have utilized the full arsenal of mouse models, ranging from direct implantation of human-derived BC cells into the brains of nude mice to intravenous and intracardiac injection of cells to spontaneous metastasis models. More recently, the field has also been leveraging patient-derived xenografts (PDXs) in immunocompromised mice. Early PDX models derived from tumorspheres of HER2+ primary core needle biopsies demonstrated a capacity for spontaneous metastasis to the brain as well as other organs. 24 PDX models derived from HER2+ BCBrM have also been generated. 25 However, the increasing T A B L E 1 Comparison of some common methods for studying breast cancer brain metastases (BrM) in animal models importance of immunotherapies in the clinic has highlighted the need for more immunocompetent models. Syngeneic models of HER2+ BC that spontaneously metastasize to the brain, such as one recently characterized, are poised to become more standard in the field. 26

| Biology of HER2+ BCBrM
Extensive work in HER2+ animal models has provided potential explanations for why this subtype of breast cancer has a predilection for CNS recurrence and subsequent BrM. HER2, as an oncogene itself, may drive brain trophism, as HER2 induces a more mesenchymal state in breast cancer cells, increasing invasiveness and metastatic potential. 27 Induced expression of HER2 in experimental models increases the size of BrM from intracardiac injections, and may alter the spatiotemporal growth of BrM within different brain regions toward favoring more posterior areas. 28,29 HER2's ability to increase BrM may be due in part to proposed interactions between HER2 and other receptors. Interactions between and signaling from HER2 and its family members, notably the epidermal growth factor receptor (EGFR) and HER3, have been implicated as driving factors in BCBrM (reviewed in Reference 30). BCBrM often overexpress HER2 and HER3, and can express mutated EGFR, even relative to primary tumors and other metastases. [31][32][33][34][35] The brain microenvironment contains several HER family ligands, including neuregulins, which can cause dimerization and activation of these receptors in brain metastatic cells. 31,36 The HER2:HER3 association appears to be particularly important in BCBrM, as it may drive BrM through the release of matrix metalloproteases that can disrupt the blood-brain barrier (BBB). 36 Furthermore, the interaction between HER2 and HER3 is enhanced by Src activation and may be a mechanism of resistance to HER2-targeting agents. 37 Preclinical models have shown that HER2 can also heterodimerize with the neutrotropin receptor TrkB and be activated by the neutrotrophic factor BDNF, suggesting that paracrine signaling increases survival of HER2+ BCBrM. 38 Interestingly, estrogen present in premenopausal women may further drive this BDNF/TrkB signaling, as well as the migratory and invasive capacity of breast cancer cells through paracrine signaling from ERα-expressing astrocytes in the brain, further driving the growth of BrM. 39,40 Preclinical studies have also shown that HER2, along with EGFR, in BCBrM can alter proliferation by modulating DNA topoisomerase I through nucleolar localization of heparanase (HPSE). 41 Additional features of HER2+ BC may contribute to its predilection for BrM. Truncated glioma-associated oncogene homolog 1 (TGLI1) is highly expressed in HER2+ BC and has been shown to increase the incidence of BrM. 42 TGLI1 may also contribute to radioresistance by increasing stemness and creating a "metastasis-friendly" microenvironment through activation of astrocytes. 42 Fatty acid binding protein 7 (FABP-7) is a lipid binding protein found specifically in the brain. However, FABP-7 is also expressed in BC cells, particularly  43 FABP7 is thought to induce a more glycolytic, metastatic, and pro-angiogenic state in BC cells, thereby enhancing the survival of HER2+ BC in the foreign brain microenvironment. 43 Thus, beyond just HER2 expression and its interaction with other receptors, other aspects of HER2+ BC biology likely contribute to the brain metastatic potential of this subtype.

| Blood-brain and brain-tumor barriers
The BBB is composed of tight junctions between various brain cells to prevent substances, including most cancer treatments, from crossing into the brain from circulation. However, this barrier is often altered in BrM, including changes in many of the cells that make up the barrier, leading to the concept of a substantially different blood-tumor barrier (BTB). 44 Seminal papers in different BCBrM mouse models, including HER2-expressing models, demonstrated that the BTB is compromised in the vast majority of experimental BCBrM, enabling increased but heterogenous, and still often subtoxic, levels of drug uptake by BrM relative to normal brain tissue. [45][46][47] Indeed, trastuzumab reaches preclinical brain tumors, but is not as effective at controlling intracranial disease in the clinical setting. 48 Surprisingly, the distribution of trastuzumab does not appear to be dictated by vascular architecture, or lack thereof, in BrM. 47 Small molecule HER2 inhibitors do not peform substantially better. Lapatinib does achieve higher levels in intracranial tumors of HER2+ BCBrM mouse models relative to normal brain, but the elevated levels in brain tumors are short-lived (<12 hours), heterogenous due to differential permeability of the BTB, and below the concentrations reached in extracranial metastases. 49 This limited exposure is due, in part, to active removal of lapatinib by P-glycoprotein (Pgp) and breast cancer resistance protein (Bcrp) efflux pumps inherent to the BBB. 50 Historically, efficacy of systemic treatments with most HER2-targeting agents against BrM has been somewhat limited and inconsistent, though newer generation HER2-targeting agents may circumvent these issues. 51 Several approaches have been tested in HER2+ BCBrM mouse models to improve access of drugs through the BBB and BTB. Studies have demonstrated the feasibility of using MRI-guided focused ultrasound in HER2+ BCBrM mouse models to disrupt the BBB and thereby increase trastuzumab delivery to BrM. 52 Focused ultrasound has also been used with microbubbles to increase the accessibility of both chemotherapy and an antibody-drug conjugate (ADC) in a HER2 + BCBrM mouse model. 53

| Human tissue-based studies
Analyzes of patient specimens have further suggested an important role for HER2 in the biology of BrM. HER2 was identified as one of four "brain metastasis selected markers" for CTCs in patients with metastatic BC, where CTCs expressing these markers had increased propensity to spread to the brain. 71

| CLINICAL MANAGEMENT OF HER2-POSITIVE BrM
The care of patients with HER2+ BCBrM is complex and requires a multidisciplinary team to determine optimal therapy (ie, local or systemic), timing of therapy, and the best management of sequelae of therapy (ie, radiation therapy necrosis) (see algorithm, Figure 1). In addition, as patient symptom burden can be high, incorporation of palliative care early is also recommended as part of the multidisciplinary team, in addition to radiation oncology, medical oncology, and neurosurgery. 73

| Local therapy
Local therapy modalities, including neurosurgical resection, stereotactic radiosurgery (SRS), and/or whole-brain radiotherapy (WBRT) remain the cornerstone of therapy for BrM. 74 Neurosurgical resection offers survival benefit when associated with adjuvant radiation therapy, more so in patients with good performance status, controlled systemic disease, and a solitary brain lesion. [75][76][77] While SRS is preferred F I G U R E 1 Suggested algorithm for multidisciplinary management of care for patients with HER2+ breast cancer brain metastases. BCBrM: breast cancer brain metastases; MBC: metastatic breast cancer; THP: Taxotere (Docetaxel) + Herceptin (Trastuzumab) + Perjeta (Pertuzumab); T-DM1: ado-trastuzumab emtansine (Kadcyla) in cases with a limited number of BrM, the upper limit of number of lesions remains controversial. 78 Finally, for patients with multiple, diffuse BrM, WBRT is the recommended treatment modality, but has fallen out of favor in recent years due to observed negative impacts on longer-term neurocognition.
It is important to highlight that most prospective trials evaluating local treatment of BrM included mainly NSCLC patients, with only a small proportion (~10%-15%) incorporating breast cancer patients. [79][80][81] Overall, these trials demonstrated that adding WBRT to initial surgery or SRS decreased intracranial disease recurrence without affecting OS (median 7 to 10 months, P = .42 to P = .93). Conversely, WBRT has been reported to worsen quality of life and neurocognitive function, particularly in patients with prolonged survival. [82][83][84] In those cases, neurocognitive decline is progressive and untreatable. Preventive strategies using memantine and hippocampal avoidance have shown improvements in neurocognitive decline. 85

| Lapatinib/capecitabine
Lapatinib is a small molecule tyrosine-kinase inhibitor (TKI) of EGFR and HER2, and is able to cross the BTB. 45

| Tucatinib/trastuzumab/capecitabine
Tucatinib is a TKI that inhibits HER2 in a reversible way. It has shown promising activity in combination with capecitabine and trastuzumab in a phase I trial, which included notable response in BrM. 100 Building on that, the HER2CLIMB phase III trial was developed and results were recently reported. 101 The trial randomized 612 patients with

| Pertuzumab
Pertuzumab is a MAb that binds to the extracellular domain II of HER2 and inhibits the dimerization of HER2 with other HER family receptors, especially HER3. In this way, it acts in synergy with trastuzumab. It was demonstrated to prolong OS when offered in combination to trastuzumab and docetaxel as first line treatment for metastatic HER2+ BC in the Cleopatra phase III, randomized, controlled trial. 103 The incidence of BrM as the first site of disease progression was evaluated in an exploratory analysis, and found to be similar in the pertuzumab arm and the placebo arm (13.7% and 12.6%). 104 Pertuzumab also showed some benefit when added to chemotherapy and trastuzumab in the adjuvant setting in patients with highrisk HER2+ BC. 105 In a recent update, with a median 74.1 months follow-up, the incidence of CNS metastases as invasive disease first recurrence was not different with or without use of adjuvant pertuzumab, 2% in both arms. 106

| TDM-1 (trasutuzumab emtansine)
T-DM1 is an ADC containing emtansine (DM1), a microtubuleinhibitory agent, linked to trastuzumab. It was the first ADC agent to be approved for treatment of HER2+ BC. Several case reports have described the activity of T-DM1 in CNS metastases. 107  Considering the issues of drug penetration in the CNS and the biologic cascade promoting BrM, preclinical data using a mouse xenograft model of BCBrM demonstrates that temozolomide administered in a preventive fashion can prevent the development of BrM. In these models, temozolomide did not result in reduction in established BrM. 118 Based on this observation, a phase I/II clinical trial for secondary prevention of BrM in HER2+ BC has been developed, enrolling patients after an initial local therapy to receive T-DM1 with or without temozolomide, with the goal to prevent and decrease incidence of new BrM. 119 This represents a new study design of "secondary prevention" in BrM clinical trials which could also be utilized in the development of clinical protocols for other subtypes of BCBrM.
As immunotherapy has emerged as one of the most promising ways to approach cancer therapy over the last decade, HER2+ and  (Table 2). One interesting approach is the use of chimeric antigen receptor-engineered T cells (CART) to target HER2+ BCBrM, demonstrated to be effective with intraventricular delivery of HER2-CAR constructs in xenograft models 68 and currently being investigated in clinical trials (Table 2).

| CONCLUSIONS
BrM are a frequent clinical challenge for patients with advanced HER2+ breast cancer. The continuous development of newer, brain permeable, anti-HER2 therapeutic options has steadily improved the impact of systemic therapy for patients with metastatic HER2+ breast cancer. In parallel, there has been an increased awareness of BrM as a clinically unmet need for this subtype of breast cancer. The complexity of CNS biology and the unique local microenvironment coupled with the historically limited availability of clinical trials for patients with CNS involvement has contributed to a poorer prognosis for this population in the past. This picture is slowly and steadily changing, in large part due to a paradigm shift resulting in the inclusion of patients with BrM in large, randomized, phase 3 clinical trials such as HER2Climb. Enrollment of patients with BrM in future clinical trials evaluating promising, brain permeable, HER2-targeted therapies should be at the forefront to maintain this forward momentum, with the goal of continuing to improve our patients' survival and quality of life in a meaningful way.

ACKNOWLEDGMENTS
The authors thank Translating Duke Health for supporting the Duke Center for Brain and Spine Metastases. We also thank the patients in these studies and their families.

ETHICAL STATEMENT
Not applicable. The other authors have no conflicts requiring disclosure.

DATA AVAILABILITY STATEMENT
Data sharing is not applicable to this article as no new data were created or analyzed in this study.