Multi-institutional report of trastuzumab deruxtecan and stereotactic radiosurgery for HER2 positive and HER2-low breast cancer brain metastases
Introduction
Approximately 15% of patients with hormone receptor positive (HR+), human epidermal growth factor receptor 2 (HER2) negative metastatic breast cancer (MBC) and 30% of patients with HER2+ or triple negative breast cancer (TNBC) will develop brain metastases (BM)1. The development of brain metastases is associated with poor prognosis and a detriment to quality of life2.
Local therapies have long been the standard of care management of BM. Stereotactic radiosurgery (SRS) is employed for limited intracranial disease and provides excellent local control, approaching or exceeding 90%3,4,5; however, patients often develop new metastases outside of the irradiated field. SRS is also complicated, primarily, by radiation necrosis (RN), though with modern dose regimens and constraints, its incidence is <10%6.
Advancements in systemic therapies have provided improved survival, resulting in an increased cumulative risk of BM7. Additionally, newer systemic agents, particularly HER2-targeted therapies, have intracranial activity. The development of HER2-targeted tyrosine kinase inhibitors (TKIs) such as neratinib, lapatinib, and tucatinib, and HER2-targeted monoclonal antibodies and antibody drug conjugates (ADCs) have been associated with improved survival and, in some cases, improved intracranial control8,9,10. However, an increased frequency of RN with the combination of SRS and ADCs has been reported as a concern11,12.
Trastuzumab deruxtecan (T-DXd) is an ADC consisting of a HER2-targeted monoclonal antibody with a topoisomerase I inhibitor, bound by a tetrapeptide linker selectively cleaved by lysosomes upregulated in tumor cells, with a high drug to antibody ratio (8:1). The topoisomerase inhibitor has high membrane permeability, theorized to enhance the antitumor activity via the bystander effect on neighboring tumor cells regardless of HER2 expression. Indeed, T-DXd has demonstrated favorable disease control and survival outcomes in the single arm DESTINY-Breast0113, and comparatively to physician’s choice of therapy (capecitabine plus trastuzumab or capecitabine plus lapatinib) and trastuzumab emtansine (T-DM1) in the DESTINY-Breast02 and DESTINY-Breast03 trials, respectively14,15,16. A recent development is also the efficacy of such agents in HER2-low patients, defined as immunohistochemistry (IHC) 1+ or 2+ without amplification on in-situ hybridization (ISH), as reported in DESTINY-Breast0417,18.
Notably, these trials reported favorable disease control in patients with asymptomatic, or stable and treated BM in a pooled analysis of DESTINY-Breast01, −02, and −0319. DESTINY-Breast01 reported a subgroup analysis of patients with BM, noting 78% received prior local therapy and required a 60-day washout period prior to T-DXd initiation20. Since these trials excluded active or progressing BM, further investigation was conducted by the TUXEDO-1 and DEBBRAH trials, with favorable response rates, PFS, and quality of life21,22. However, details regarding prior or subsequent RT are not reported, and a combination strategy has yet to be investigated prospectively regarding both safety and efficacy. Additionally, reports regarding management of HER2-low BM are lacking, as DEBBRAH has yet to report outcomes in this subgroup and DESTINY-Breast04 reported a 25% intracranial ORR in stable, treated BM only18.
In this report, we detail multi-institutional oncologic and safety outcomes of combination T-DXd and SRS for the management of both HER2+ and HER2-low BM.
Methods
We identified patients who received T-DXd at three institutions, located in Tampa, FL, Miami, FL and Columbus, OH following Moffitt Cancer Center, Miami Cancer Institute, and Ohio State University IRB approvals, from institution-specific databases. Informed consent was waived given the retrospective nature of the study. We included patients who received T-DXd for MBC and SRS for the management of new and progressive BM, beginning in August 2020. All patients received the FDA approved starting dose, or they were included even if they were dose reduced per treating physician discretion. Patients were excluded if SRS was more than 3 months prior to the initiation or after the discontinuation of T-DXd, if they had diagnosed leptomeningeal carcinomatosis prior to T-DXd, or if they received whole brain radiation therapy alone. HER2-low was defined as per DESTINY-Breast04, as IHC 1+ or 2+ without amplification on ISH. Patients were followed until November 2023.
Stereotactic (SRS) radiosurgery technique
Patients were immobilized using thermoplastic masks or stereotactic frames, depending on the treatment technique. For linear accelerator-based treatments, a planning CT scan was obtained and co-registered with T1-weighted contrast enhanced MRI, obtained at most 2 weeks prior to the first day of treatment; for those treated with Gamma Knife-based SRS, MR-based planning was performed with the planning scan performed up to 48 h prior to treatment23. Gross tumor volume (GTV) was delineated as the contrast-enhancing disease. A uniform 1-2 mm expansion was used to create the planning target volume (PTV) for linear accelerator-based SRS but no PTV expansion was used for Gamma Knife-based SRS24. Treatment plans were reviewed by a radiation oncologist and a neurosurgeon. Treatment was delivered using 3D conformal arcs or intensity-modulated radiation therapy, using linear accelerators with 6MV X-Rays, or using a Leksell Gamma Knife Icon (Elekta, Stockholm, Sweden). Daily image guidance was performed using cone-beam CT. Intrafraction motion was tracked using ExacTrac (Brainlab AG, Munich, Germany), an optical surface monitoring system (OSMS), or a high-definition motion management (HDMM) system with an infrared stereoscopic camera (Elekta AB, Stockholm, Sweden). SRS was considered concurrent with T-DXd if delivered during T-DXd treatment and up to 1 month after the date of last infusion.
Follow-up
Patients were followed with physical examination by a radiation oncologist, medical oncologist, neuro-oncologist, or neurosurgeon with contrast-enhanced brain MRIs every 2–4 months after treatment. Local failure was defined as contrast enhancement within a previously irradiated volume, with a ≥20% increase that remained consistent or demonstrated continued progression on subsequent imaging as defined by RANO-BM criteria25. Distant failure was defined as new metastases or leptomeningeal enhancement outside of previously irradiated volumes. Extracranial failure was defined as new or enlarging lesions on FDG PET/CT, CT thorax/abdomen/pelvis, or technetium 99-m bone scan. Symptomatic radiation necrosis was diagnosed by MR perfusion, MR spectroscopy, and/or pathology and following multidisciplinary discussion including available clinical, radiologic, and pathologic information, along with the onset of new or progressive neurologic symptoms.
Statistical analysis
The cumulative incidence of radiation necrosis was calculated using death as a competing event. Overall survival (OS) and progression-free survival (PFS) were estimated using the Kaplan–Meier method from the time of T-DXd initiation. Local and distant intracranial control (DIC) were measured from the date of SRS or the first date of fractionated treatments. Cox Proportional Hazards Model was used to identify significant prognostic variables. Variables that were significant (p < 0.05) on univariable analysis (UVA) were included in the multivariable analysis (MVA). Statistical analysis was performed using JMP 17 (SAS Institute Inc, Cary, NC).
Results
Patient and treatment characteristics
Patient characteristics are summarized in Table 1. A total of 215 lesions were treated over 48 SRS courses in 34 patients. Radiation treatment details are listed in Table 2. Median follow up from T-DXd initiation was 13.9 months (range 0.9–31.8) and 11.8 months (range 0.1–30.2) from each SRS course. Nineteen (56%) patients had received prior intracranial RT. One hundred seventy-two (80%) lesions were treated with single fraction SRS and 43 (20%) lesions were treated with fractionated SRS (fSRS). One hundred seventeen (54%) lesions received concurrent SRS and T-DXd. Median GTV size was 0.05 cc (range 0.002–33.04 cc) and median PTV size was 0.30 cc (0.04–65.01cc). Median dose for single fraction SRS was 22 Gy (range 15–24 Gy), and 24 Gy (range 15–35 Gy) in a median 3 fractions (range 3–5) for fSRS. Six (3%) lesions were treated post-operatively.
Toxicity
Symptomatic radiation necrosis occurred in 3 (1%) lesions in 3 patients. The cumulative incidence of radiation necrosis at 24 months per lesion was 2.1% (Fig. 1A) and per patient 11% (Fig. 1B). Two lesions received concurrent T-DXd and SRS, and one received SRS prior to T-DXd. Lesion size was not significantly associated with incidence of symptomatic radiation necrosis (Fisher’s Exact Test p = 0.12 for lesion size ≤0.05 cc vs >0.05 cc; p = 0.93 for lesion size ≤14 cc vs >14 cc; logistic regression p = 0.59). One lesion was treated with linear accelerator-based SRS to 20 Gy, with a GTV size of 0.12 cc and PTV size of 1.12 cc. This patient was treated with Vitamin E and pentoxifylline, and initiated dexamethasone and Boswellia extract. Another patient received prior WBRT of 37.5 Gy in 15 fractions at initial BM diagnosis and received fSRS, 24 Gy in 3 fractions to a GTV of 6.7 cc and PTV 15.1 cc, approximately 1 year later. This patient also received T-DM1 for 7 months prior to T-DXd but did not receive any intracranial RT while on T-DM1. RN was diagnosed 2 months after SRS, which was rapidly progressive and treated with dexamethasone for 1.9 months with initial improvement followed by deterioration. The patient was noted to have both systemic and intracranial progression and elected for hospice care prior to the planned initiation of bevacizumab for radiation necrosis. RN may also have been a contributing factor in her death. The third lesion received Gamma Knife-based SRS to 20 Gy, with a GTV size of 0.27 cc. This patient required a prolonged dexamethasone taper for 2.5 months due to intracranial symptomatology. No other unexpected toxicities were noted.
Cumulative incidence of symptomatic radiation necrosis (A) per lesion and (B) per patient.
Local and distant control
Eight lesions (4%) developed local recurrence. LC rates at 12- and 24-months were 97% (95% CI 93–99%) and 88% (95% CI 75–94%), respectively, Fig. 2. No differences were noted in local control by HER2 status, surgical resection, SRS/FSRT, lesion size, or prior systemic therapies, Supplementary Table 1. One patient had local recurrence of two separate lesions, which were treated during the same session and were noted to have recurred on the same follow up MRI.
Kaplan–Meier curve for local control.
Median DIC was 7.3 months (95% CI 3.7–9.1), with a 12-month rate of 27% (95% CI 14–46%) and 24-month rate of 6% (95% CI 1–31%). HER2-low disease was associated with distant intracranial failure (HR 3.1, 95% CI 1.4–6.8, p = 0.006), with 12-month DIC rates of 43% for HER2+ disease and 9% for HER2-Low, Fig. 3A. Concurrent SRS and T-DXd was associated with improved DIC compared to SRS after T-DXd (HR 0.15, 95% CI 0.03–0.76, p = 0.02). On MVA, both remained associated with distant intracranial failure (adjusted HR for HER2-low vs HER2+ 2.5, 95% CI 1.1–5.6, p = 0.03; adjusted HR for SRS concurrent vs after 0.14, 95% CI 0.03–0.72, p = 0.02), Table 3.
Kaplan–Meier curves for (A) distant intracranial control (B) systemic progression free survival by HER2 status.
Four (12%) patients developed leptomeningeal disease, at a median of 9.6 months (range 1.4–19.6 months) after T-DXd initiation. One of the four patients underwent prior craniotomy and surgical resection of a BM, that then received post operative fSRS of 24 Gy in 3 fractions to a 30.2 cc post-operative cavity GTV and 49.6 cc PTV.
Systemic PFS
Eighteen patients (53%) progressed extracranially. Median systemic PFS was 7 months (95% CI 3.3–11.8 months), with 12- and 24-month systemic PFS rates of 32% (95% CI 17–51%) and 27% (95% CI 14–47%), respectively. HER2-low disease was significantly associated with systemic progression or death, with 12-month systemic PFS rates of 48% for HER2+ and 6% for HER2-low (HR 4.1, 95% CI 1.6–10.7, p = 0.004), Fig. 3B, Table 4.
Overall survival
At the time of analysis, 12 (35%) patients had died. Median overall survival was not reached (NR) (95% CI 10.7 – NR), with 12- and 24-month OS rates of 64% (95% CI 45–80%) and 53% (95% CI 33–72%), respectively. OS rates trended towards significant difference for HER2+ and HER2-low disease, with 12-month OS rates of 43% and 9%, respectively (HR for death 2.4, 95% CI 0.7-8.0, p = 0.14). Death was not significantly associated with age, KPS at T-DXd initiation, hormone receptor status, receipt of prior tucatinib or T-DM1, or surgical resection, Supplementary Table 2.
Discussion
In this report of combination SRS and T-DXd, we note several findings. First, T-DXd with SRS did not appear to increase the frequency of symptomatic RN. Second, the combination of T-DXd and SRS resulted in favorable outcomes for breast cancer BM, with excellent local control that is comparable to SRS alone26. Third, HER2-low patients had significantly worse outcomes with T-DXd compared to HER2+ patients, with worse systemic PFS and DIC with a trend towards worse OS.
Intracranial activity of T-DXd has been demonstrated in the setting of BM on subgroup analyses as well as the TUXEDO-1 and DEBBRAH trials18,20,21,22,27. It is important that outcomes with established local therapies be investigated to assess for potential toxicity as well as synergy to optimize outcomes for these advanced breast cancer patients. Reports of SRS with concurrent ADCs including T-DM1, T-DXd, and sacituzumab-govitecan note a potential increased risk of symptomatic radiation necrosis11,12. However, in these series most patients received T-DM1, 53% in the study from Lebow et al. and 92% in the study from Koide et al. Higher rates of RN have been reported with T-DM1 and SRS, with a proposed mechanism of T-DM1 targeting reactive astrocytes and increased radiation-induced cytotoxicity and astrocytic swelling via upregulation of Aquaporin-4 (Aqp4)28,29,30. Our prior series of T-DM1 and SRS reported a single case of symptomatic radiation necrosis at a median follow up of 13.2 months31.
Given the differences in payload, pharmacokinetics, and the definition of concurrent treatment, it is important to assess each ADC independently with SRS to delineate risks and potential benefit more clearly. T-DM1 has a half-life of approximately 4 days, whereas T-DXd has a half-life of 6 days, and sacituzumab-govitecan of only 16 h32,33,34. Therefore, a biological definition of concurrent treatment differs depending on the pharmacokinetics of each agent and the data from one cannot be extrapolated to another. Additionally, this definition has been inconsistent regardless of agent. With a cumulative incidence of symptomatic radiation necrosis at 24 months of 11% per patient, our rates of radiation necrosis with TDX-d are lower than those reported by Koide and Lebow et al.11,12. However, it is important to note the potentially devastating impact of symptomatic necrosis, as radiation necrosis may have been a contributing factor in one patient’s death in this series. This patient did have concurrent intracranial and systemic progression and received prior WBRT to 37.5 Gy. As new ADCs such as TDX-d and sacituzumab govitecan show potential efficacy in the management of BM, it is important that potential synergy, efficacy, and risks be assessed with each individually18,20,21,22,27,35.
Preliminary efficacy of TDX-d alone in the setting of BM has been reported. Results from the subgroup analysis of TDX-d in the DESTINY-Breast03 Trial revealed a median PFS of 15 months compared to 3 months for those treated with T-DM1 with BM. We noted significant differences in systemic PFS between HER2+ and HER2-low disease. In the DESTINY-Breast04 Trial a total of 24 patients with BM were treated with HER2-low disease with a median duration of treatment 6.6 months. Stable, treated BM made up 5% of the patient population this trial. Although cross trial comparisons are difficult, comparisons of the BM populations in the DESTINY-BREAST-03 and 04 Trials appear to be in line with differences noted in our study based on HER2 status. These differences in systemic PFS were also noted in intracranial responses post SRS with significant differences noted in DIC post SRS. This also led to a trend in OS for improved survival in HER2+ disease.
We report excellent local control in our series with no differences based on HER2 positivity. A quarter of patients in our study were symptomatic with a 24-month local control rate of 88% across all 215 lesions treated. TUXEDO-1 investigated active BM, either new and untreated, or progressive after prior local therapy, and reported a RANO-BM intracranial response rate (ORR-IC) of 73% for HER2+ disease only, subdivided as 100% for new lesions and 66.7% for prior treated lesions. Prior treated lesions were 60% of the study group. The DEBBRAH trial reported ORR-IC of 50% for new, untreated BM and 44% for lesions that progressed after prior local therapy. Our data reveals excellent local control with the receipt of SRS and the need for further prospective data on the potential synergy with combination therapy. Timing of SRS with TDX-d should be further explored. Given the potential intracranial response to TDX-d, TDX-d could be assessed upfront to reduce the size and volume of brain metastases treated with SRS. Limitations of the current study include its retrospective design with a heterogeneous group of patients with respect to timing of TDX-d and SRS and treatment characteristics.
Combination SRS and T-DXd was well tolerated in our series and provides excellent local control for both HER2+ and HER2-low breast cancer BM. HER2-low breast cancer BM have worse outcomes with combination SRS and T-DXd compared to HER2+ disease. Prospective investigation is warranted to establish a clear treatment paradigm for this highly vulnerable patient population.
Responses