EGFR as a potent CAR T target in triple negative breast cancer brain metastases

There is currently no curative treatment for patients diagnosed with triple-negative breast cancer brain metastases (TNBC-BM). CAR T cells hold potential for curative treatment given they retain the cytolytic activity of a T cell combined with the specificity of an antibody. In this proposal we evaluated the potential of EGFR re-directed CAR T cells as a therapeutic treatment against TNBC cells in vitro and in vivo. We leveraged a TNBC-BM tissue microarray and a large panel of TNBC cell lines and identified elevated epidermal growth factor receptor (EGFR) expression. Next, we designed a second-generation anti-EGFR CAR T construct incorporating a clinically relevant mAb806 tumor specific single-chain variable fragment (scFv) and intracellular 4-1BB costimulatory domain and CD3ζ\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\zeta$$\end{document} using a lentivirus system and evaluated in vitro and in vivo anti-tumor activity. We demonstrate EGFR is enriched in TNBC-BM patient tissue after neurosurgical resection, with six of 13 brain metastases demonstrating both membranous and cytoplasmic EGFR. Eleven of 13 TNBC cell lines have EGFR surface expression ≥ 85% by flow cytometry. EGFR806 CAR T treated mice effectively eradicated TNBC-BM and enhanced mouse survival (log rank p < 0.004). Our results demonstrates anti-tumor activity of EGFR806 CAR T cells against TNBC cells in vitro and in vivo. Given EGFR806 CAR T cells are currently undergoing clinical trials in primary brain tumor patients without obvious toxicity, our results are immediately actionable against the TNBC-BM patient population.


Introduction
Breast cancer is the most frequently diagnosed cancer among American females and is the second leading cause of cancerrelated deaths with a predicted 43,250 deaths this year [1]. Triple-negative breast cancer (TNBC) is a highly aggressive subtype of breast cancer which is defined by the lack of estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) expression. It accounts for 15-20% of all breast cancer cases and is known for its high recurrence rate and high ability to metastasize resulting in poor clinical prognosis [2]. The most common systemic therapy for metastatic TNBC (mTNBC) is chemotherapy [3], but median overall survival (OS) is limited to 12 to 18 months [4]. Only 20-40% of mTNBC are PD-L1 positive and candidates for checkpoint inhibitor immunotherapy [5,6]. Worse yet, the incidence of brain metastases (TNBC-BM) is as high as 46% among patients with advanced TNBC [7]. Median survival from the diagnosis of TNBC-BM is only 4.9 months [7,8]. Despite many clinical trials, no current standard therapy exists that provides long-term control or cure for TNBC-BM patients [7,9,10].
Chimeric Antigen Receptor(CAR) T cell therapy combines the cytolytic potency of a T cell with the antigen specificity of an antibody single-chain (scFv) domain. Although CAR T cells have had remarkable clinical success in liquid tumors [11,12], significant hurdles still exist in their application against brain tumors [13], a major obstacle being identifying appropriate tumor antigen that has low to no expression in normal tissues. EGFR is highly expressed in 13-76% [14][15][16][17][18][19] of TNBC. Monoclonal antibody [20][21][22] and smallmolecule targeted therapy against EGFR [23,24] have failed in mTNBC, due to the lack of cytolytic-effect, the activation of downstream compensatory oncogenic signaling pathways, and the co-expression of other onco-receptors [25][26][27]. CAR T cells targeting breast-to-brain metastases hold promise due to their cytolytic effect against tumor cells. Preclinical models of HER2 CAR T cells have shown anti-tumor activity in HER2 + xenograft mouse models of breast-to-brain metastases [28]. As a result, HER2 + breast-to-brain metastatic patients are undergoing clinical study for efficacy of intracranial delivered HER2 + CAR T cells (NCT03696030). However, there are currently no clinical trials investigating CAR T cells targeting TNBC-BM patients. EGFR may be an attractive target in TNBC-BM tissue. It is known that TNBCs expressing EGFR are prone to metastasize to the brain compared to those lacking EGFR [29]. In a study of 33 breast cancer patients, EGFR expression was found in 39% patients who had brain metastasis [30], although the percentage of patients with TNBC was not stated. Herein, we explore the percentage expression of EGFR in primary TNBC-BM tissue, as well as evaluate the preclinical efficacy of EGFR as a CAR T target in TNBC-BM mouse models.
EGFR CAR T cells have been explored in preclinical TNBC models [31,32], however, the scFv is based on the anti-EGFR cetuximab antibody that targets both oncogenic EGFR and skin-keratinocyte EGFR [33,34], thus raising concern for on-target off-tumor toxicity. Indeed, studies in cetuximab based EGFR CAR T demonstrate toxicity in human skin orthotopic xenografts [35]. To address this problem, we employed a CAR T construct employing the EGFR mAb806 antibody (EGFR806), which is tumorrestricted to EGFR as a result of oncogene amplification [36][37][38]. Furthermore, mAb806 kills TNBC cells in vitro when conjugated with toxin [39]. The mAb806 antibody has been tested in clinical trials in EGFR + tumors with mild toxicity [40]. Intracranial delivered EGFR806 CAR T cells are currently under clinical study in pediatric brain tumors (NCT03638167), with no dose limiting toxicity in preliminary results [41]. Further, EGFR expression in normal adult human brain is low to negligible, reducing the risk of ontarget off-tumor toxicity [42][43][44].
Here, we have developed second generation EGFR806 CAR T cells for the treatment of TNBC-BM. To our knowledge, this is the first CAR T cell targeting metastatic TNBC using this mAb806 derived scFv. We evaluated therapeutic efficacy of intracranial tumoral (ICT) delivery of CAR T cells at the site of tumor in orthotopic human tumor xenograft models. Our data provide support for potential clinical application of EGFR806 CAR T cells targeting TNBC-BM in human clinical trials.

EGFR806 CAR T construct design
The EGFR806 targeted scFv sequence was derived from antibody Mab806 and cloned into second generation CAR T construct ( Fig. 2A). A second generation construct was employed, as this format has demonstrated safety in multiple brain tumor related clinical trials. The extracellular spacer domain comprised of the double mutated IgG4EQ [45], followed by CD4 transmembrane domain, 41BB co-stimulatory domain and CD3ζ cytolytic domain. The CAR construct was cloned in a pLenti6.3/V-5-DEST backbone under the control of CMV promoter (GeneArt;Thermo Fisher Scientific, Waltham, MA). Full length amino acid sequence is described in supplemental Table 2.

Lentivirus production
HEK293T cells (Takara) were plated in T-300 flasks (TP90301; Midwest Scientific, St. Louis, MO), a day prior to transfection of the CAR and enhanced Green Fluorescent Protein-Firefly-Luciferase (eGFP-Ffluc, 119816; AddGene, Watertown, MA) plasmid constructs separately. Transfection was done at 70% confluence using CalPhos Mammalian Transfection Kit (631312; Takara Bio) with LV-MAX Lentiviral Packaging Mix (A43237; GibcoTM). Media was renewed the next day morning with addition of 0.5 mM Sodium butyrate ((B5887; Sigma-Aldrich, St. Louis, MO) and supernatant collected in 50 mL tubes 72 h later. The tubes were centrifuged at 800 g for 10 min to remove cellular debris and filtered through a 0.45 µm filter. The supernatants were combined and centrifuged at 6080 g at 4℃ for 24 h. The lentivirus pellet was resuspended in 4% Lactose (L5-500; Fisher Chemical; Thermo Fisher Scientific, Waltham, MA, USA) in PBS (10,010-031; Gibco) solution, aliquoted and stored in -80℃ freezer for further use. All chemical solutions were filtered in 0.2 µm.

Human CAR T cell production
Peripheral blood samples from healthy human male and female patients < 45 years of age without prior cancer diagnosis were collected in vacutainer tubes ± EDTA, followed by centrifugation at 4 °C set at 1300 g for 10 min, followed by collection of buffy coat. Untouched human CD4 T cells were isolated from buffy coat using Dynabeads® (11346D; Invitrogen; Thermo Fisher Scientific, Waltham, MA) according to the manufacturer's instructions. Briefly, RBC lysis buffer was added to the buffy coat to lyse red blood cells and then magnetic beads were added to the PBMCs. After washing and spinning down tagged cells, Dynabeads® were added to the cells to negatively select the CD4 population. Finally, the supernatant was centrifuged to collect untouched CD4 T cells. The T cells were then cultured in gas-permeable tissue 24-well culture plates (80192 M; Wilsonwolf, St Paul, MN) at a concentration of 2 × 10 6 cells/well in 1 mL of LymphoONE™ T-Cell Expansion Xeno-Free media (WK552S; Takara) supplemented with 10% FBS, 50 U/mL IL2 (200-02; Peprotech, Cranbury, NJ), 0.5 ng/mL IL15 (200-15; peprotech) and one-time 20µL T Cell Trans-Act™ (130-128-758; Miltenyi Biotec, Gaithersburg, MD). EGFR806 CAR T lentivirus particles are added on D1 at multiplicity of infection (MOI) of 6. Media was doubled on D2 (2 × cytokines were added on this day only) & D3, and 3 mL added on D4 for a total of 7 mL. On D5 and D6 half of the media was replaced with fresh media (1 × cytokines were added in each of these days). Cells were collected on Day 7, followed by CAR expression by flow cytometry, then frozen down for later use in cryopreservation media (07930; CryoStor® StemCell Technologies, Cambridge, MA). CAR expression was identified by flow cytometry using antihuman IgG, Fcγ fragment specific (109-546-008 Jackson Immuno Research, West Grove, Pennsylvania).

Generation of stable TNBC cell lines expressing firefly luciferase
MDA-MB-231 cells were transduced with lentivirus to express eGFP-Ffluc under Puromycin (VWR) selection. Cells were then flow sorted (FACs Aria II, BD Biosciences, San Jose, CA) based on GFP expression (90%) and cultured as described above while maintaining puromycin selection. For Perforin and Granzyme CAR T activation assays, 50,000 MDA-MB-231 cells were plated per well of a 48 well plate and allowed to attach overnight. 100,000 CD4 CAR T cells and untransduced Mock T cells were added to the cells the following morning and allowed to incubate for 3 h. Brefeldin A was added for to the co-culture for another 5 h after which the T cells were collected in a V-Bottom 96 well plate and cells were washed thrice before adding extracellular antibodies and allowed to incubate at 4C for 30 min. After 2 washes, cells were fixed with Fixation Buffer (BioLegend #420801) in the dark for 20 min at room temperature. Cells were centrifuged at 350 g for 5 min and washed with Cell Staining Buffer. Samples were left with Cell Staining Buffer on them and left at 4C overnight. The fixed cells were suspended in Intracellular staining Perm Wash Buffer (BioLegend #421002) and centrifuged at 350 g for 5 min twice. The cells were than incubated with antibodies against Perforin (Biolegend #353312) and GranzymeB (Biolegend #372208) in Perm Wash Buffer for 20 min in the dark at room temperature. Cells were washed twice with Intracellular staining Perm Wash Buffer and centrifuged at 350 g for 5 min. Cells were finally resuspended in FSS and Flow was run.

Immunoblot
TNBC cell lines and CRISPR KO Lines were interrogated for EGFR expression by Immunoblot. In brief, cell lines were grown to confluence and lysed using RIPA buffer (89901; Thermo Fisher Scientific), with Protease and Phosphatase inhibitor (1861281; Thermo Fisher Scientific) and EDTA (1861274; Thermo Fisher Scientific). These lysates had protein concentrations quantified using a BCA Protein Assay Kit (23227; Thermo Fisher Scientific) on a Nanodrop (ND-2000; Thermo Fisher Scientific) according to manufacture instructions. Lysates were then prepared and run using Separation

In vitro killing assay
Human TNBC cells were co-cultured with EGFR806 CAR T cells and mock T cells at specified effector cell: TNBC cell ratios. After the specified duration of co-culture, the supernatant is collected in tubes and attached tumor cells detached using Cellstripper™ (25-056-CI; Corning) for 15-30 min incubation while shaking 37C. All cells are transferred to tubes and were stained for CD3, and a Live/Dead viability dye as described above for evaluating killing by the CAR T cells. CAR T cells were identified by anti-IgG to target hinge region of CAR (109-096-008; JacksonImmuno Research Labs). Flow cytometry was acquired on Cytek Aurora and data analyzed using FlowJo software.

Mouse tumor studies
Female NSG (NOD.Cg-PrkdcscidIl2rgtm1Wjl/SzJ) mice, 8-11 weeks of age were purchased from the Jackson Laboratory (Jackson Laboratory; Farmington, CT). All mice experiments were approved by the Institutional Animal Care and Use Committee (IACUC).
The intracranial implantation of tumor cells has been previously described [46]. Briefly, mice were anesthetized by exposure to isoflurane. A burr hole was drilled on the skull at 2 mm lateral of the center line, 0.5 mm anterior to the bregma and 1 × 10 5 MDA-MB-231 cells expressing luciferase suspended in 3ul of PBS was injected orthotopically in 1 ul increments at 2.5 mm, 2 mm, and 1.5 mm deep from the dura. Engraftment was verified by bioluminescent imaging (IVIS Spectrum imager; PerkinElmer, Waltham, MA) using intraperitoneal injection of 200ul (100 mM) D-Luciferin one day before CAR T cell injection. The mice were randomized into groups based on the BLI signal and 1 × 10 6 EGFR806 CAR T cells or mock treatment injected intracranially at the same burr hole at 0.5ul per incremental depth (2.75 mm, 2.5 mm, 2.25 mm, 1.75 mm, 1.5 mm, & 1.25 mm) site next day. Tumor growth was monitored by IVIS imager and flux signals analyzed using Living Image®software (v4.5.5; PerkinElmer). Mice with BLI negative tumors underwent repeat BLI imaging to confirm absence of BLI signal. Additional dates for luminescent imaging were chosen to capture the exponential growth of tumor in non-treated mice based on previous work with the tumor kinetics. After the control mice were all euthanized, imaging was halted and mice were monitored for survival. At D100 the mice were imaged once more. At this time the luminescence seen in two CAR treated mice had disappeared leading to them to be imaged again in the subsequent timepoints after the machine had been serviced to ensure that it was still functioning. At desired time points or at moribund status, mice were euthanized, and brain tissues were processed for H&E histology as described below.

Immunohistochemistry staining
A human tissue microarray was constructed from resected brain metastases of 13 patients in duplicate and two positive controls (placenta, and kidney) and one negative control (tonsil) for EGFR expression. Pathologist confirmed tissue prior to TMA construction, and pathologist confirmed IHC scoring. Anti-EGFR antibody (Emab-134 ab264540 Abcam; Cambridge, UK) was used for IHC staining according to manufacturer instruction. EGFR immunoreactivity was scored by a clinical pathologist and quantified based on the percentage of tumor cells exhibiting weak (1 +), moderate (2 +), or strong (3 +) intensity staining of cell surface and cytoplasmic staining. The H score is obtained by the formula: (3 × percentage of strongly staining cells) + (2 × percentage of moderately staining cells) + percentage of weakly staining cells, giving a range of 0 to 300. Each patient was assigned a separate H-score for both membrane and cytoplasmic staining.
For murine H&E, mice were euthanized at indicated time points and were perfused with ice cold PBS followed by 4% PFA. Whole brains were dissected and incubated in 4% PFA for 3 days, followed by 70% ethanol for 3 days before being embedded in paraffin. Transverse Sects. (10 μm thick) were cut and stained with hematoxylin and eosin.

Statistical analysis
All statistical analysis was performed using Prism software (GraphPad v9). Data are represented as Mean ± SD as stated 1 3 in the figure legends. Biological significance was determined by student's T-test or using one-way ANOVA with multiple comparisons. For mouse survival studies, differences between groups were assessed by log-rank (Mantle-Cox) test. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

EGFR is highly expressed in TNBC-BM patient tissues and cell lines
We generated a tissue microarray of resected metastatic brain tumor tissue for 13 TNBC-BM patients and H-score was assigned individually for both surface and cytoplasmic EGFR expression (Fig. 1A, supplemental Fig s1 and Table 1). This result demonstrates high membranous and cytoplasmic expression of EGFR in 6 of 13 patients. Strikingly, 3 patients have near maximum membranous H-score of 300. Next, we evaluated surface expression of EGFR in a panel of TNBC breast cancer lines (Fig. 1B). Eleven of 13 tested cell lines had EGFR mean fluorescence intensity (MFI) greater than 85%, and only one cell line (MDA-MB-453) was negative for EGFR. In order to confirm these findings, we also interrogated the 13 cell lines by immunoblot. 12 of 13 cell lines demonstrated high levels of EGFR expression relative to EGFR knockout control line (Fig. 1C)

EGFR806 CAR T cells demonstrate effective in vitro tumor killing
We first designed a second-generation CART cell using the EGFR806 scFv to target the oncogenic EGFR expression on TNBC-BM patient tissue samples (Fig. 2). Mab806 is restricted to oncogenic amplification of EGFR [36][37][38] and has been utilized for treating glioblastoma patients in clinical trials [46]. The Vh and Vl domain is followed by double mutated IgG spacer (IgG4EQ) to reduce Fc receptor recognition [45]. The endodomain consist of CD4 transmembrane domain, 4-1BB costimulatory domain and CD3ζ cytolytic domain (supplemental table S2). CAR expression was identified by flow cytometry using anti-human IgG, Fcγ fragment specific (109-546-008 Jackson Immuno Research, West Grove, Pennsylvania) (Fig. 2B,C). We transduced human CD4 cells with the EGFR806 lentivirus. We chose CD4 cells given their superior long-term persistence and recursive killing potential compared to CD8 T cells alone or combination CD4-CD8 T cells [48,49].
We examined the in vitro killing of the EGFR806 CAR T cells against MDA-MB-231 over time course 24, 48, and 96 h at Effector to Tumor cell ratio of 1:1 (Fig. 3A).  (Fig. 3B). Next, we evaluated specificity of EGFR806 CAR T cells cytolysis of against MDA-MB-231 scarmble control vs EGFR CRISPR-Cas9 KO MDA-MB-231cells (Fig. 4A, B). Cytolytic ability of EGFR806 CAR T cells was lost upon EGFR KO, confirming the specificity of the EGFR806 CAR. Next, we evaluated EGFR806 CAR killing using other cell lines in addition to MDA-MB-231 (MDA-MB-468 and HCC70 TNBC) cells by co-culturing tumor cells and CAR T cells (Fig. 4C). DAPI staining was our gating strategy to determine the percentage of dead cells by flow cytometry. Statistically significantly higher frequency of live cells was observed with EGFR806 CAR T cells with respect to Mock CD4 T cells as control for all the lines.

TNBC-BM tumor regression after ICT delivery of EGFR CAR T cells in vivo
To evaluate the tumor targeted killing of EGFR806 CAR T cells, NSG mice were intracranially injected with human TNBC cell line MDA-MB-231 cells transduced with firefly luciferase. Mice underwent bioluminescent imaging (BLI) one day prior to CAR T injection (Day 5 post tumor implantation) to quantify tumor burden, as seen in schema (Fig. 5A). Human CD4 CAR T cells were thawed from a lentivirus production that was frozen D7 after transduction, and flow cytometry was done to confirm viability and CAR expression. EGFR806 CAR T cells treated mice were eradicated of tumor without any BLI evidence of tumor recurrence Mice with BLI negative tumors underwent repeat BLI imaging to confirm absence of BLI signal. (Fig. 5B,  C), statistically significant survival (Fig. 5D)  Data are represented as mean ± SD and were analyzed by student's t-test. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001 disease burden and hemorrhage in control mice ( Fig. 5E and  supplementary Fig. s2).

Discussion
TNBC brain metastases is incurable with standard therapy. Autologous T cell therapy holds promise for a potential curative treatment. However, tumor associated antigens (TAAs) are often heterogenous in solid tumors, limiting single TAA targeted CAR T cells due to antigen loss [50,51]. Identifying highly enriched TAAs is critical for targeting the maximal tumor population. Previous studies have demonstrated EGFR expression in breast cancer brain metastases in approximately 40% of patients, although the TNBC population was not stated [29,52]. Our data further supports EGFR as a viable target in TNBC brain metastases, as our TMA is specific for TNBC-BM and demonstrates high membrane and cytoplasmic EGFR expression in six of 13 patients. Because cytoplasmic TAA expression is generally hidden from CAR T cells, future strategies may include tyrosine kinase inhibitors to enhance TAA surface presentation and CAR killing [53].
To our knowledge, our study is the first to demonstrate EGFR mAb806 CAR T cell mediated killing of TNBC cells in vitro and in vivo. This is a clinically meaningful observation, as EGFR mAb806 targets oncogenic over-expressed EGFR with limited toxicity to endogenous EGFR, including brain astrocytes [46,54]. Indeed, EGFR806 CAR T cells are currently undergoing ICT delivery in pediatric brain tumor patients (Brainchild-02 NCT03638167) with no obvious toxicity in early reports [41]. Thus, there is significant potential to rapidly translate this finding into early-phase clinical evaluation for TNBC-BM patients.
Our EGFR806 CAR had maximum killing at 96 h compared to earlier time point at 1:1 ratio. This is consistent with prior study evaluating mAb806 based EGFR CAR killing of glioblastoma cells at earlier time points at higher E:T ratio, and similar anti-tumor activity in vivo murine brain tumor models [46]. Future studies will be required to further optimize our CAR T construct, such as optimization of hinge region (IgG4EQ, CD8a, CD28, IgG hinge only), or additional co-stimulatory domain. Data are represented as mean ± SD and were analyzed by student's t-test. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001 Further, our data support local-regional intracranial delivery of CAR T cells. There are two major routes of intracranial delivery: intracranial ventricular (ICV) and intracranial tumor (ICT). In the clinical setting, an ommaya reservoir is placed subcutaneously and attached with cathether placement in the ventricle (ICV) or tumor resection cavity (ICT). This way, CAR T cells can be injected into ommaya subcutaneously, and allow for repeated access of CSF for molecular correlative studies. Other benefits include CAR T bypass of blood-brain-barrier as well as mitigating risk of extra-cranial CAR T toxicity. Priceman et al. are using ICV delivery of HER2 CAR T cells for treatment of breast-to-brain metastasis [28] in an ongoing clinical trial (NCT03696030). Similarly, with respect to patients with TNBC brain metastases, the major delivery route will likely include ICV, as CAR T cells injected within the CSF will be able to circulate through CSF and targets multiple foci of disease.  Interestingly, there may be synergy in co-targeting EGFR and HER2 re-directed CAR T cells in breast to brain metastases based upon co-expression from patient brain metastases [55]. An "OR" gated CAR, such as a dual or single chain bispecific CAR T construct design may mitigate tumor associated antigen escape. However, this does increase risk of on-target off-tumor toxicity. Advanced logic gated "AND" gated CAR T design, such as Syn-Notch may help mitigate toxicity [56]. Further, other potential tumor associated antigen targets for TNBC CAR T cells have been reviewed elsewhere [57].
Our EGFR806 CAR T cells were produced from CD4 human T cells and demonstrate effective long term tumor eradication and mouse survival. It is believed that efficacy of adoptive cell therapy can be often attributed to CD8 T cells [58] and infusion of CD8 derived CD19 CAR T cells alone is sufficient for long-term B-cell eradication [58,59]. However, recent clinical data demonstrate long-term (10 year+) persistence of CD19 CAR T cells is mostly CD4 based [60]. In support of this observation, CD4 (vs CD8) based IL13R∝2 redirected IgG4(EQ)-41BBζ CAR T cells in preclinical GBM models demonstrate improved long term persistence and recursive killing [49]. Further, in other in vitro and in vivo studies of CD4 CAR T cells demonstrate similar effectiveness in directly killing tumor cells [48,61,62], with less activation induced cell death compared to CD8 T cells [48,49,63].
Given the ongoing clinical safety reports in these patients, our data further supports clinical evaluation of intracranial delivered CD4 T derived EGFR806 CAR T to TNBC-BM patients. Future studies will help further interrogate TNBC brain tissue for other TAAs that may complement heterogenous EGFR expression, as well as TNBC-BM specific tumor micro-environment related immune-suppressive pathways that may be targeted in conjunction with EGFR.
Our study has limitations. Our TNBC cell lines tested demonstrate 100% surface EGFR expression, however the patient TNBC-BM tissue demonstrates more heterogenous membrane EGFR expression. Future studies of combination with EGFR tyrosine kinase inhibitors (TKI) may enhance surface EGFR expression and sensitize to EGFR806 CAR T cells. This is a promising strategy, as brain-penetrant EGFR TKIs are undergoing clinical study [64]. Future studies utilizing heterogenous tumor EGFR expression will help identify mechanisms of tumor resistance to EGFR806 CAR T. Furthermore, our TMA did not assess the EGFR expression in the primary tumor. It is not yet clear the association between primary breast tumor and TNBC-BM for EGFR expression with regards to EGFR enrichment once metastatic to brain.
Author contributions All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by DA, SS, SS, JJ, and CW, BS, and JZ. The first draft of the manuscript was written by DA and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.
Funding This work was funded by the University of Kansas Cancer Center.

Data availability
The original contributions presented in the study are included in the article/supplementary material. Further inquiries can be directed to the corresponding author. All data will be made available to interested parties upon reasonable request.

Declarations
Competing interests The authors have no relevant financial or nonfinancial interests to disclose.
Ethical approval This study was performed in line with the principles of the Declaration of Helsinki and its later amendments or comparable ethical standards. Ethical approval of this study was granted by the University of Kansas Medical Center Institutional Review Board (IRB) and IACUC on September 21, 2021.
Informed consent Informed consent was obtained from all University of Kansas Medical Center Biorepository Core Facility for all patient tissue presented in this study.

Consent to participate
Informed consent was obtained from all individual participants included in the study.

Consent to publish
The authors affirm that human research participants provided informed consent for de-identified immunohistochemistry in Fig. 1A.