Abstract PD2-06: Implications of Heterogeneity in Breast Tumor Cell MHC-I Expression on Immunity and Therapeutic Resistance

Background Immune checkpoint inhibitors (ICIs) targeting the PD-1/L1 axis are approved in early-stage treatment for triple-negative breast cancer (TNBC), but only a fraction of patients benefit. Tumor expressed antigens bound to major histocompatibility complex-I (MHC-I) are required for CD8-mediated tumoricidal activity, and thus, response to anti-PD-1/L1 targeted ICI. However, many breast tumors downregulate, or heterogeneously express, MHC-I, making them less susceptible to ICIs. Tumor cells downregulating MHC-I may be effectively targeted by natural killer (NK) cells due to ‘missing self’ signals. However, this heterogeneity in MHC-I expression is poorly modeled in most preclinical studies, limiting our understanding of how to overcome ICI resistance in the context of heterogeneous MHC-I expression, as is often observed clinically. Objective We aimed to 1) quantitatively delineate how intratumoral heterogeneity in MHC class I expression affects immune responses and immunotherapy outcomes in mouse models and 2) determine whether targeting inhibitory signals on NK cells can overcome ICI resistance in MHC-I heterogenous TNBC. Methods We performed quantitative immunofluorescence for MHC-I, CD8, CD56, and pan-cytokeratin on breast cancer tumors from diverse subtypes (n=314) to obtain single-cell-resolution MHC-I expression and spatial information of tumor and immune cells. Fluorescence intensity and spatial analysis were processed to output individual tumor/stromal cell MHC-I expression and the composition of the local tumor microenvironment. To determine the functional effect of MHC-I heterogeneity in vivo, we generated a CRISPR-guided B2m knockout (B2m-null) in a murine orthotopic model (EMT6). We then combined MHC-I-proficient and MHC-I-deficient isogenic lines at various ratios to model how populations of MHC-I loss affected the immune microenvironment. We also assessed a second, intrinsically heterogenous (MHC-I expression) TNBC model E0771. To evaluate changes in the microenvironment, we used flow cytometry and an immune NanoString panel to evaluate gene expression patterns in tumor cells and infiltrating immune cells. Results TNBC patients had the highest MHC-I expression level across tumor cells, but also the highest variability and probability of demonstrating bimodal MHC-I expression (consisting of high and low/absent cells within a single tumor). ER+ tumors were unimodally low. Using spatial analysis, we identified that heterogenous MHC-I tumors had significantly higher levels of infiltrating NK cells(Paired T-Test: p=0.03). In murine models, even 10% or less of MHC-I-null (B2m-null) EMT6 cells in the tumor injection resulted in acquisition of ICI resistance. Interestingly, heterogeneity in expression of MHC-I resulted in a substantial infiltration by NKG2A+ NK cells compared to MHC-I high or -low tumors (Student T-Test: p=.002). Activation of these infiltrating NK cells via anti-NKG2A and anti-PD-L1 combination treatment restored complete responses in heterogeneous EMT6 tumors, and significantly extended survival in both E0771 (Mantel-Cox: p< 0.0001) and EMT6 models (Mantel-Cox: < 0.0001). Additionally, anti-NKG2A and anti-PDL1 combination treatment improved complete response in the heterogenous MHC-I EMT6 model to 30% and in the parental EMT-6 tumors to 70%. Conclusion Combined therapy with anti-NKG2A (targeting NK cells) and anti-PD-L1 (targeting CD8+ T cells) can restore immunotherapy responses and overcome resistance due to lack of MHC-I expression in tumor cell subpopulations.

Citation Format: Brandie C. Taylor, Xiaopeng Sun, Justin Balko, Paula Gonzalez-Ericsson, Melinda Sanders. Implications of Heterogeneity in Breast Tumor Cell MHC-I Expression on Immunity and Therapeutic Resistance [abstract]. In: Proceedings of the 2022 San Antonio Breast Cancer Symposium; 2022 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2023;83(5 Suppl):Abstract nr PD2-06.

245 Host myeloid response to tumor and immunotherapy is associated with heterogeneity in outcomes to anti-PDL1

Background

Immune checkpoint inhibitors (ICI) improve patient survival in some cancer types but yield limited success in breast cancer. Phase-III clinical trials in triple-negative breast cancer (TNBC) patients, who harbor extensive tumor-infiltrating lymphocytes, demonstrate increased progression-free survival (IMpassion130) and pathologic complete response (KEYNOTE-522). Consequently, combinations of ICI and chemotherapy have been FDA-approved for metastatic TNBC patients. However, the therapeutic benefit of ICI alone and the most efficacious chemotherapy combinations are poorly characterized. We sought to model ICI response in vivo to elucidate the mechanisms of immunotherapy efficacy in breast cancer and ascertain the therapeutic benefits of different chemotherapeutic combinations with ICI.

Methods

Using an immunocompetent EMT6 orthotopic mammary tumor model, we investigated the efficacy of single-agent immunotherapy and in combination with standard-of-care chemotherapy (paclitaxel [PAC] or doxorubicin [DOX]). We used single-cell RNA sequencing and bulk RNA and T-cell receptor (TCR) sequencingto assess the cellular landscape of the primary tumor in response to combinatorial therapeutic strategies and identify systemic genetic alterations and T-cell expansion, respectively.

Results

Single-agent anti-PD-L1 robustly suppressed primary tumor growth (p =0.0046) and extended survival (p<0.0001) beyond the isotype control. Chemotherapy demonstrated moderate therapeutic efficacy without potentiating the benefit of single-agent anti-PD-L1. Interestingly, despite using a genetically identical murine tumor model/host, anti-PD-L1 induced heterogeneous responses, from complete response to intrinsic resistance. Longitudinal analysis of peripheral blood from heterogeneously responding mice uncovered myeloid cell recruitment signatures corresponding to transient responses ultimately converting to resistance. We identified specific clonal T cell expansion present only in responders. Single-cell transcriptomic profiling of the tumor microenvironment revealed increased T cells and natural killer cells and reduced regulatory T cells in the combination groups versus chemotherapy alone, although this did not translate into improved benefit. Gene-set enrichment analysis on infiltrating T cells identified a robust signature of cytotoxic T cell activation characterized by a significant enrichment in inflammatory pathways in both single-agent anti-PD-L1 and in combination with chemotherapy.

Conclusions

We identify a heterogeneously ICI-responsive in vivo model that emulates TNBC patient response to combinatorial ICI approaches. We describe single-agent ICI efficacy in upregulating cytotoxic immune cell infiltration and expansion within the primary tumor that diminishes tumor growth and enhances survival. Moreover, this study describes differential responses in a genetically similar host, which reflects heterogeneous patient response to ICI. Further characterization may identify systemic biomarkers and tumor antigen-specific T cell clones to accurately predict immunotherapy response in patients and uncover mechanisms for sensitizing refractory tumors to ICI

318 Enforced tumor specific MHC-I heterogeneity in triple negative breast cancer drives immunotherapy resistance

Background

Despite the broad success of immune checkpoint inhibition (ICI e.g. anti-Programmed Death Ligand-1 [PD-L1]) in cancer treatment, tumor-intrinsic factors leading to intrinsic and acquired resistance are poorly understood. Tumor specific MHC-I expression is indispensable for anti-PD-1/L1 response as complete loss of MHC-I via B2M deletion results in inability of CD8+ T cells to recognize tumor-associated antigens. However, MHC-I is heterogeneously downregulated or lost in many tumor types. Tumor cell destruction can also occur through non-synaptic mechanisms, in a so-called ‘field effect’. Therefore, we modeled heterogeneous loss of MHC-I expression in breast cancer and experimentally evaluated how heterogeneous MHC-I loss affects response to anti-PD-L1 therapy.

Methods

We performed quantitative immunofluorescence for MHC-I and Pan-CK on breast cancer tumors (n=410). To determine the functional effect of MHC-I heterogeneity on anti-PD-L1 response, we used an immunocompetent EMT6 orthotopic mammary tumor model which ubiquitously expresses MHC-I at baseline. Using CRISPR/Cas9, we engineered EMT6 cells with B2m loci excision resulting in complete knockout of MHC-I on the cell surface. We then orthotopically implanted B2m-comtetent and B2m-KO cells at varying inoculum ratios (100:0, 90:10, 50:50, 10:90, 0:100) into syngeneic Balb/C mice and assessed immune responsiveness and efficacy of checkpoint inhibition. Additionally, to look at how loss of MHC-I affects the tumor microenvironment we will use the PanCancer Immune NanoString panel (n=770 genes) to evaluate gene expression patterns in tumor cells and infiltrating immune cells.

Results

In patient samples, we identified high diversity in MHC-I expression across all clinical subtypes, with triple negative breast cancer (TNBC) having the highest MHC-I expression. Chemotherapy-treated tumors had higher MHC-I levels than untreated tumors. In mice when 10% of cells were B2m-KO, we observed a 50% reduction in complete eradication of EMT6 tumors with aPD-L1 treatment and reduced disease-stabilization and no complete responses when a 50% mixture of MHC-I deficient cells. An increasing percentage of B2m KO leads to worse outcomes overall and a decrease in infiltrating T cells.

Conclusions

Our work suggests that there is an ICI-responsive phenotype that is driven by heterogeneity in MHC-I expression levels. As little as 10% of tumor specific MHC-I loss can lead to therapeutic resistance and a decrease in complete responders. This represents that early TNBC may be less responsive to single-agent PD-L1 due to specific percentages of MHC-I loss. MHC-I expression can influence therapy outcomes and potentially lead to novel observations of how to overcome lack of, or limited, MHC-I expression.

812 Erythema nodosum-like toxicity in an immunotherapy treated patient is accompanied by oligoclonal memory activated CD4 T cells

Background

Immune checkpoint inhibitors (ICIs) are increasingly used to treat advanced malignancy but can be associated with immune related adverse events (irAE). Here we present a case report of a rare dermatologic toxicity occurring in a melanoma patient with isolated brain metastasis. After surgical resection, the patient was treated with ipilimumab (anti-CTLA-4) and nivolumab (anti-PD-1) combination therapy followed by single agent nivolumab with ongoing, excellent response. During nivolumab, the patient developed an erythema nodosum (EN)-like irAE. The condition resolved after potassium iodine treatment and nivolumab therapy was resumed. To understand the pathogenesis of this irAE, we examined samples from this patient's blood, brain metastasis and tissue biopsy of the EN toxicity.

Methods

RNA and T cell receptor (TCR) sequencing on the patient's brain metastasis and site of irAE were performed. We also performed RNA sequencing on 3 non-ICI EN patients. RNA in situ hybridization (RNAish) for CD4, CD8 and granzyme B, and the most abundant TCR identified was conducted on the patient's site of toxicity. Single cell RNA/TCR sequencing was carried out on the patient's peripheral blood mononuclear cells (PBMC) at baseline, 3 weeks after ipilimumab and nivolumab combination therapy, during EN toxicity and after resolution.

Results

RNAish showed that the most abundant TCR (20% of total TCR sequencing reads at the site of toxicity) colocalized with CD4 at the site of toxicity. According to CIBERSORT deconvolution, the site of toxicity had high memory activated CD4 T cells and low M2 macrophage infiltration, which is different from the brain metastasis and non-ICI-induced EN cases. Compared to non-ICI EN, the EN skin biopsy was also enriched for interferon response and inflammation related genes. In the peripheral blood, cytotoxic CD8 T cells clonally expanded during EN toxicity, accompanied by a decrease in naïve/memory CD4 T cells. The TCR repertoire in the site of toxicity did not overlap with that in the tumor or PBMC.

Conclusions

We found oligoclonal memory activated CD4 T cells are enriched at the site of toxicity, suggesting their association with EN toxicity. The unique TCR repertoire, gene expression profile and immune cell composition at the site of toxicity could indicate that the EN toxicity is distinct from the anti-tumor immunity and analogous non-ICI autoimmunity. Future work will focus on determining the antigen for this irAE and determining its relevancy to other skin toxicities and EN autoimmune conditions.

Ethics Approval

IRB 100178 and 161485

Consent

Approval under IRB 100178

Beyond Programmed Death-Ligand 1: B7-H6 Emerges as a Potential Immunotherapy Target in SCLC

INTRODUCTION

The programmed death-ligand 1 (PD-L1) immune checkpoint inhibitors, atezolizumab and durvalumab, have received regulatory approval for the first-line treatment of patients with extensive-stage SCLC. Nevertheless, when used in combination with platinum-based chemotherapy, these PD-L1 inhibitors only improve overall survival by 2 to 3 months. This may be due to the observation that less than 20% of SCLC tumors express PD-L1 at greater than 1%. Evaluating the composition and abundance of checkpoint molecules in SCLC may identify molecules beyond PD-L1 that are amenable to therapeutic targeting.

METHODS

We analyzed RNA-sequencing data from SCLC cell lines (n = 108) and primary tumor specimens (n = 81) for expression of 39 functionally validated inhibitory checkpoint ligands. Furthermore, we generated tissue microarrays containing SCLC cell lines and patient with SCLC specimens to confirm expression of these molecules by immunohistochemistry. We annotated patient outcomes data, including treatment response and overall survival.

RESULTS

The checkpoint protein B7-H6 (NCR3LG1) exhibited increased protein expression relative to PD-L1 in cell lines and tumors (p < 0.05). Higher B7-H6 protein expression correlated with longer progression-free survival (p = 0.0368) and increased total immune infiltrates (CD45+) in patients. Furthermore, increased B7-H6 gene expression in SCLC tumors correlated with a decreased activated natural killer cell gene signature, suggesting a complex interplay between B7-H6 expression and immune signature in SCLC.

CONCLUSIONS

We investigated 39 inhibitory checkpoint molecules in SCLC and found that B7-H6 is highly expressed and associated with progression-free survival. In addition, 26 of 39 immune checkpoint proteins in SCLC tumors were more abundantly expressed than PD-L1, indicating an urgent need to investigate additional checkpoint targets for therapy in addition to PD-L1.

Abstract GS1-06: FGFR1 associates with gene promoters and regulates gene transcription: Implications for endocrine resistance in ER+/FGFR1-amplified breast cancer

Background: FGFR1 amplification occurs in ~ 15% of ER+ breast cancers. In these tumors, nuclear FGFR1 has been shown to interact with DNA, but its role in transcriptional regulation is unclear. Thus, we investigated the genomic role of FGFR1 in ER+/FGFR1-amplified breast cancer. Results: FGFR1 ChIP-Seq detected 4,412 DNA binding sites in CAMA1 ER+/FGFR1-amplified breast cancer cells cultured in estrogen-free conditions. Of these binding sites, 67% were enriched at promoter regions. ChIP-qPCR confirmed FGFR1 binding to several promoter regions in a second ER+/FGFR1-amplified cell line, MDA-MB-134, and a patient derived xenograft, HCI-011. To determine the nuclear FGFR1 interactome, we performed FLAG immunoprecipitation of mixed nuclear and chromatin fractions of CAMA1 cells transduced with a 3XFLAG-FGFR1 plasmid, followed by mass spectrometry (MS) of FLAG antibody pulldowns. MS revealed RNA Polymerase II subunits among the top nuclear FGFR1 interacting proteins. FGFR1 mainly bound Pol II phosphorylated on Ser5 (Pol II S5P), a marker of transcription initiation, in CAMA1, MDA-MB-134 and HCI-011 cell extracts. Pol II S5P ChIP-Seq revealed that 2,867/4,412 (65%) FGFR1 peaks were shared with Pol II S5P. ChIP-Seq also showed that 95% of FGFR1 peaks overlapped with both H3K4me3 and H3K27ac, markers of active transcription. Consistent with these results, RNA-Seq of CAMA1 cells showed that expression of FGFR1-bound genes was markedly higher than non FGFR1-bound genes (p&lt;0.0001), suggesting that FGFR1 binds to actively transcribed genes. In addition to Pol II, MS detected FOXA1 among FGFR1 interacting proteins. ChIP-Seq analysis revealed FOXA1 enriched at FGFR1-bound loci. siRNA-mediated FOXA1 knockdown reduced FGFR1 distribution to several genomic loci in CAMA1 cells, as measured by FGFR1 ChIP-Seq, suggesting that FOXA1 mediates FGFR1 recruitment to chromatin. We next transduced MCF-7 cells with an FGFR1(SP-)(NLS) plasmid, where the NLS sequence forces nuclear import of the resulting protein. To determine the role of FGFR1 on transcriptional regulation, we used Binding and Expression Target Analysis (BETA), integrating FGFR1 ChIP-Seq and RNA-Seq results from MCF7FGFR1(SP-)(NLS) vs MCF7EV cells. This analysis predicted a direct role for genomic-bound FGFR1 in activating gene expression (p=8.01e-6). MCF7FGFR1(SP-)(NLS) cells were markedly less sensitive to fulvestrant compared to control cells. Gene Set Enrichment Analysis (GSEA) of the 1,009 genes upregulated in MCF7FGFR1(SP-)(NLS) cells and bound by FGFR1 at a genomic level revealed a strong enrichment of estrogen response early (q=2.2e-44) and late (q=6.4e-33) genes, suggesting that nuclear FGFR1 induces an ERα-associated transcriptional profile that may contribute to endocrine resistance. Finally, an expression signature associated with nuclear FGFR1 correlated with endocrine resistance in three cohorts of patients with ER+ breast cancer treated with aromatase inhibitors. We next studied the effect of growth factor stimulation on FGFR1 transcriptional function. Stimulation with FGF2 enhanced nuclear FGFR1 import in CAMA1 cells, as well as FGFR1-Pol II S5P association. Notably, these effects were not abrogated by treatment with the FGFR1 inhibitor erdafitinib. ChIP-Seq revealed that erdafitinib did not impair the FGFR1 genomic distribution. These results do not support a causal link between the FGFR1 activated TK and the receptor’s activity in the nucleus. Conclusions: We have demonstrated a role for nuclear FGFR1 in transcriptional regulation in breast cancer. FGFR1-induced gene expression contributes to endocrine resistance and is not affected by FGFR TKIs. These findings provide a rationale for developing treatment strategies to inhibit nuclear FGFR1 in ER+/FGFR1-amplified breast cancer.

Citation Format: Alberto Servetto, Rahul Kollipara, Luigi Formisano, Chang-Ching Lin, Kyung-min Lee, Dhivya R Sudhan, Ariella B Hanker, Sumanta Chatterjee, Angel Guerrero-Zotano, Paula Gonzalez-Ericsson, Saurabh Mendiratta, Hiroaki Akamatsu, Nicholas James, Ralf Kittler, Carlos L Arteaga. FGFR1 associates with gene promoters and regulates gene transcription: Implications for endocrine resistance in ER+/FGFR1-amplified breast cancer [abstract]. In: Proceedings of the 2020 San Antonio Breast Cancer Virtual Symposium; 2020 Dec 8-11; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2021;81(4 Suppl):Abstract nr GS1-06.

Pitfalls in assessing stromal tumor infiltrating lymphocytes (sTILs) in breast cancer

Stromal tumor-infiltrating lymphocytes (sTILs) are important prognostic and predictive biomarkers in triple-negative (TNBC) and HER2-positive breast cancer. Incorporating sTILs into clinical practice necessitates reproducible assessment. Previously developed standardized scoring guidelines have been widely embraced by the clinical and research communities. We evaluated sources of variability in sTIL assessment by pathologists in three previous sTIL ring studies. We identify common challenges and evaluate impact of discrepancies on outcome estimates in early TNBC using a newly-developed prognostic tool. Discordant sTIL assessment is driven by heterogeneity in lymphocyte distribution. Additional factors include: technical slide-related issues; scoring outside the tumor boundary; tumors with minimal assessable stroma; including lymphocytes associated with other structures; and including other inflammatory cells. Small variations in sTIL assessment modestly alter risk estimation in early TNBC but have the potential to affect treatment selection if cutpoints are employed. Scoring and averaging multiple areas, as well as use of reference images, improve consistency of sTIL evaluation. Moreover, to assist in avoiding the pitfalls identified in this analysis, we developed an educational resource available at www.tilsinbreastcancer.org/pitfalls.

Report on computational assessment of Tumor Infiltrating Lymphocytes from the International Immuno-Oncology Biomarker Working Group

Assessment of tumor-infiltrating lymphocytes (TILs) is increasingly recognized as an integral part of the prognostic workflow in triple-negative (TNBC) and HER2-positive breast cancer, as well as many other solid tumors. This recognition has come about thanks to standardized visual reporting guidelines, which helped to reduce inter-reader variability. Now, there are ripe opportunities to employ computational methods that extract spatio-morphologic predictive features, enabling computer-aided diagnostics. We detail the benefits of computational TILs assessment, the readiness of TILs scoring for computational assessment, and outline considerations for overcoming key barriers to clinical translation in this arena. Specifically, we discuss: 1. ensuring computational workflows closely capture visual guidelines and standards; 2. challenges and thoughts standards for assessment of algorithms including training, preanalytical, analytical, and clinical validation; 3. perspectives on how to realize the potential of machine learning models and to overcome the perceptual and practical limits of visual scoring.

Abstract P2-06-09: Identification of oncogenic gene fusions in triple-negative breast cancer metastatic to the brain

BACKGROUND: As many as 24% of breast cancer (BC) patients develop metastasis to the brain and 27% of these occur in patients with triple negative (TN) BC. The brain (#2) and lung (#1) are the most common sites of TNBC metastasis. Half of women with metastatic (m) TNBC will die within one year. Novel therapies for women with mTNBC, especially to the brain, are desperately needed. TNBCs often exhibit genomic instability, resulting in gene fusions and other chromosomal rearrangements. Such alterations are long recognized as oncogenic drivers and effective drug targets in hematologic and other solid malignancies. We aimed to identify these alterations in mTNBC with Segmental Transcript Analysis (STA), a novel population-based algorithm ranking genes based on likelihood of harboring a gene rearrangement by identifying abrupt changes in transcript expression.

RESULTS: We performed RNAseq on 25 mTNBC, 4 matched primary BC, 2 normal cerebellum and 2 normal cerebral cortex samples. TNBC subtype analysis showed an enrichment of basal-like (BL) 2 and mesenchymal (M) subtypes among mTNBC to the brain compared to primary (p) BC in the TCGA. A principal component analysis demonstrated mTNBC and their matched primaries clustered together, indicating the pairs were more closely related to each other genetically than to other mTNBCs. When the gene expression patterns of pBC were compared with their brain metastases following adjustment for contamination with normal brain tissue, an upregulation of mesenchymal genes (p63/p73 axis) and primitive/embryonic genes (NOTCH and PAX family) were noted in the brain mTNBC. We confirmed a modest to marked upregulation of p63 expression by immunohistochemistry across all TNBC subtypes, most notably in BL1 and BL2. STA found three gene rearrangements: EHF-WT1, NF1-VWDE and JAG1-Chr20. Two of these are potential driver gene rearrangements. The first, in a tumor of M subtype, results from fusion of the 5'UTR of EHF to WT1. The resulting fusion transcript is predicted to encode a truncated WT1 protein; loss of the N-terminal regulatory region of WT1 would likely cause constitutive activation. We also identified a truncating rearrangement in the Notch ligand JAG1 in a tumor of LAR subtype. JAG1 is a membrane bound ligand for the Notch receptor and the breakpoint occurs just distal to the DSL domain essential for Notch interaction. This rearrangement occurs within an intergenic chromosomal region, introduces an abrupt stop codon and generates a truncated protein lacking a transmembrane domain. An intact signal peptide should localize the protein to the endoplasmic reticulum; however, loss of the transmembrane domain would likely lead to secretion rather than membrane insertion. This alteration is predicted to constitutively stimulate Notch activity in a paracrine fashion. If the rearrangement functions in this manner, neutralizing antibodies to the N-terminal portion of JAG1 should have therapeutic efficacy.

DISCUSION: EHF-WT1 and JAG1-Chr20 are potential driver rearrangements novel to brain mTNBC. The JAG1 rearrangement predicts constitutive production of Notch transcription factors and is potentially targetable. Continued interrogation of mTNBC for discovery of additional oncogenic fusions driving BC biology is warranted.

Citation Format: Gonzalez-Ericsson P, Lehmann B, Mobley B, Chen Y-Y, Pietenpol J, Sanders ME. Identification of oncogenic gene fusions in triple-negative breast cancer metastatic to the brain [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P2-06-09.