Brain Metastasis from HER2-Positive Breast Cancer: An Evolving Landscape

In this issue of Clinical Cancer Research, Kabraji and colleagues present the results of a hybrid study with preclinical and clinical evaluation of the activity of trastuzumab deruxtecan (T-DXd) on brain metastases (BM) from advanced HER2-positive breast cancer (1).

HER2 is upregulated in BM compared with primary tumors (2). BM occur in up to 50% of patients with HER2-positive metastatic breast cancer: this incidence raised over time due to better control of systemic disease with HER2-directed therapies, such as the mAbs trastuzumab and pertuzumab, that led to a prolongation of survival.

Research on targeted treatments for BM has initially focused on small-molecule tyrosine kinase inhibitors (TKI), based on the hypothesis that the low molecular weight would have allowed a better penetration of the blood–brain barrier in comparison with large molecules, such as mAbs. However, clinical results were modest (3). Single-agent therapy with the reversible HER2/EGFR TKI lapatinib yielded a central nervous system (CNS) response rate of 6% in patients with BM progressing after local therapy, while the combination with capecitabine yielded a CNS response rate of 20%–38% in patients progressing after prior radiotherapy and 66% in naïve patients. The irreversible inhibitor of the HER2 family neratinib yielded as monotherapy a response of 8% in patients progressing after radiotherapy, while in combination with capecitabine CNS response rate increased to 33% in pretreated patients and 49% in naïve patients.

The modest clinical results of these small molecules may be partially explained by a limited penetration of the blood–tumor barrier (BTB). In mouse models, BTB is heterogeneously compromised leading to a suboptimal distribution of drugs within the tumor (4). Lapatinib achieves higher levels in intracranial metastases than in normal brain, but the elevated levels are short lived (<1–2 hours), heterogeneous and lower than in extracranial metastases. Moreover, a removal of lapatinib by the efflux pumps (P-glycoprotein, and breast cancer resistance protein), that are upregulated in the BTB (5). However, in addition to drug delivery limitations, drug resistance in HER2-positive breast cancer BM (BCBM) is related to the potential molecular divergence with the primary tumor (6), and cross-talk with cells of microenvironment (astrocytes, neurons, microglia; refs. 7, 8).

Tucatinib, a novel HER2-specific TKI, showed promising activity in an early clinical trial in combination with trastuzumab and capecitabine, and the phase III trial HER2CLIMB compared trastuzumab and capecitabine with tucatinib or placebo (9). A total of 291 patients with either active or stable BM were included. The CNS response rate in the evaluable subset (55 patients in the tucatinib group and 20 in the placebo group) was 47% versus 20% with a median duration of response of 6.8 versus 3.0 months, respectively. The median CNS progression-free survival (PFS) was also improved by tucatinib, with 9.9 versus 4.2 months in all patients with BM, and 9.5 versus 4.1 months in patients with active BM.

Studies with radiolabeled mAbs and mouse models strongly support the penetration of HER2 antibody and HER2 antibody–drug conjugates (ADC) across the BTB (4). T-DM1 is an ADC containing emtansine (DM1), a microtubule inhibitory agent linked to trastuzumab (T), with an intracranial response rate of 30%–49%. A novel ADC is T-DXd, composed of an anti-HER2 antibody, a cleavable tetrapeptide-based linker, and a cytotoxic topoisomerase I inhibitor, whose activity is based on a higher therapeutic index, bystander killing due to dispersion of payload and antitumor immune activity.

Thus far, five studies have been published on the clinical activity of T-DXd in BM from advanced HER2-positive breast cancer (Table 1; refs. 1, 10–13). Intracranial response rate in active BM accounted for 46%–73%, with complete response representing 4%–13%, partial response 54%–73%, and stable disease 13%–33%. Median intracranial PFS ranged from 13 to 18 months, and reached 75% at 12 months in the study of Kabraji and colleagues (1). With regard to stable BM after local treatment, the DESTINY-Breast03 study (11) reported an intracranial response rate of 64%, with an intracranial PFS of 15 months. Overall, these results are extremely promising; however, the small heterogeneous sample size and short follow-up of the studies call for confirmative trials. More definitive data will come from DESTINY-Breast12, a multicenter phase IIIb/IV trial, investigating T-DXd in up to 500 patients, either with or without BM at baseline. Another ongoing phase Ib/2 is investigating the combination of T-DXd with durvalumab or paclitaxel or both. Of particular interest appears the cohort 7, which is exploring the combination of T-DXd with tucatinib, that is, the combination of the two most active drugs.

How can we best evaluate and predict the biological and clinical efficacy of novel drugs in earlier studies before embarking in randomized studies? One can use patient-derived xenograft (PDX) models from human BCBM, that were demonstrated to recapitulate the genomic characteristics of matched patient tissue (4). This approach has been employed by Kabraji and colleagues (1), who demonstrated in a HER2 BCBM PDX model that T-DXd was able to reduce tumor growth and prolong survival, and this was confirmed in a T-DM1–resistant model. Interestingly, the biological activity of T-DXd consisted in stimulating the apoptosis while not reducing the proliferative capacity of tumor cells. Another approach is represented by phase 0 and window of opportunity trials (14), that consist in the administration of a clinical dose of an investigational drug for a short period (commonly 1 week) before surgery, followed by pharmacokinetic and pharmacodynamic analyses on the resected tissue.

An issue of growing importance is the potential efficacy of HER2-directed therapies in patients with metastatic or advanced breast cancer, who have been defined as HER2 negative (15): approximately 50% of these patients have actually low levels of HER2 expression (i.e., HER2-low). In the phase III DESTINY-Breast04 trial T-DXd improved PFS (median, 9.9 months) compared with physician's choice of chemotherapy for patients with HER2-low, and no differences in PFS were found between patients with IHC1+ and IHC2+/ISH− tumors (10). A phase II study with biomarker analysis (DAISY trial) is investigating T-DXd regardless of HER2 status. In a preliminary analysis, responses occur in both HER2-low (33%) and HER2-nul (30.6%) cohorts in comparison with 69% in HER2-over cohort (16). Thus far, there are no data available on HER2-low expression in BCBM. This is the reason why it is of interest the finding of Kabraji and colleagues (1) that T-DXd was able to reduce tumor growth and prolong survival also in a HER2-low BCBM PDX model, and moreover they detected a partial response, even if of brief duration, in a single patient.

A critical issue in designing clinical trials in BM is the evaluation of response and progression. The RANO (Response Assessment in Neuro-Oncology) group proposed standard criteria, based on MRI, neurologic status and need for steroids to improve the comparability between different trials (17). New trials on targeted agents should focus on cohorts of patients with BM, and include neurocognitive and QoL evaluation. Volumetric assessments on MRI should be preferred to standard measurements, and liquid biopsy with the analysis of cell-free DNA needs to be included in clinical trials on BCBM to measure in cerebrospinal fluid and blood alterations of molecular biomarkers, such as HER2, EGFR, PIK3CA, CDKN2A (18). Finally, some novel techniques to bypass the BTB are promising: one is the MRI-guided focused ultrasound to enhance delivery of mAbs (trastuzumab, pertuzumab, and T-DM1) to BM (19), but more robust data are needed.

No disclosures were reported.