Triple negative breast cancer (TNBC): non-genetic tumor heterogeneity and immune microenvironment: emerging treatment options

Abstract

Triple-negative breast cancer (TNBC) is an aggressive subtype characterized by extensive intra-tumoral heterogeneity, and frequently develops resistance to therapies. Tumor heterogeneity and lack of biomarkers are thought to be some of the most difficult challenges driving therapeutic resistance and relapse. This review will summarize current therapy for TNBC, studies in treatment resistance and relapse, including data from recent single cell sequencing. We will discuss changes in both the transcriptome and epigenome of TNBC, and we will review mechanisms regulating the immune microenvironment. Lastly, we will provide new perspective in patient stratification, and treatment options targeting transcriptome dysregulation and the immune microenvironment of TNBC patients.

Introduction and Current therapy for TNBC

TNBC lack the expression of both estrogen receptor (ER) and progesterone receptor (PR), and do not have amplification of the epidermal growth factor receptor 2 (ERBB2) gene or overexpression of the ERBB2 protein. For decades, chemotherapy has been the main first line therapeutic option for TNBC patients. In the early-stage setting, multiagent chemotherapy is most commonly given prior to surgery (neoadjuvant therapy) and the tumors show a high response rate with 40–50% of the tumors having no residual invasive disease in the breast or in the lymph nodes – so called a pathological complete response (pCR) (1). While the patients whose tumors achieve a pCR have low recurrence rates (e.g. < 15% at 10 years), those with residual disease have high recurrence rates (overall about 50% at 10 years) and approaching 80% for those with a large burden of tumor after neoadjuvant treatment of paclitaxel followed by fluorouracil, doxorubicin and cyclophosphamide (2). In early-stage patients, the addition of adjuvant Capecitabine therapy for those patients who do not achieve a pCR significantly decreases recurrences and improves overall survival (OS) (3–5). In the metastatic setting, the disease is incurable and again chemotherapy is the main therapeutic option with median OS of 2–3 years (1).

There have been recent advances that include targeted agents that have improved the outcomes for both early and metastatic TNBC. In the adjuvant setting for early-stage TNBC patients who carry germline mutations in BRCA1 or BRCA2, the addition of the PARP inhibitor Olaparib to patients who do not achieve a pCR (or to those who do not get neoadjuvant therapy who are high risk) significantly decreases recurrences and shows an improved OS (6). In metastatic disease, patients with germline mutations in BRCA1 and BRCA2 also benefit from treatment with PARP inhibitors in first- or second-line treatment, showing higher response rates than chemotherapy and longer progression free survival (PFS) compared to standard chemotherapy (7,8). While the use of PARP inhibitors for metastatic disease in these patients has clear benefit, the OS has not been significantly improved with a median of ~19 months, but there may be subgroups of patients such as those who have not received chemotherapy in which OS is improved (9). Another new targeted agent recently approved for patients with metastatic TNBC is the antibody drug conjugate Sacituzumab govitecan, an antibody targeting the human trophoblast cell surface antigen 2 (Trop-2) conjugated to the topoisomerase inhibitor SN-38. When given in the third line setting, this drug improved PFS and OS compared to standard chemotherapy (10). While OS is improved, the median survival with Sacituzumab govitecan was only 12.1 months.

The addition of immune checkpoint blockade (ICB) (Figure 1) to 1^st and 2^nd line therapies for various malignancies has resulted in clinical gain for many cancer patients, including those with solid tumors such as TNBC (11). TNBC displays a high level of programmed cell death-ligand 1 (PD-L1), which led researchers to design multiple TNBC clinical trials with PD-L1 inhibitors (12) and subsequent FDA approval of programmed cell death protein-1 (PD-1) ICB for treating TNBC. The clinical studies showed significant improvement in patients with early disease (13). In these cases, the anti-PD-1 pembrolizumab was added to the neoadjuvant chemotherapy and then continued as adjuvant therapy after surgery (14). The addition of pembrolizumab increases the pCR rate to >60% and significantly improves PFS (14,15). The data on OS with the addition of pembrolizumab is immature but OS appears to be improved. The recurrence benefit appears to be mostly in the patients who do not achieve a pCR with the neoadjuvant therapy (16). The benefit to pembrolizumab in the early-stage setting is seen both in tumors that do or do not express PD-L1 (14,15). In the advanced setting, ICB has shown improvements in PFS as first line treatment of PD-L1+ TNBC but not in PD-L1 negative TNBC compared to chemotherapy (17,18). The data to date have not shown a statistically significant increase in OS with median OS in the 15–25-month range. The use of ICB in later lines for metastatic TNBC has limited activity (19).

Figure 1. Immune checkpoint blockade (ICB).

The blockade of immune check point inhibitors takes the advantage of specific ligand-receptor interaction such as PDL1-PD1. The immune checkpoint inhibitory receptors and ligands are indicated in T cells and cancer cells. PD1: programmed cell death protein 1; PD-L1: programmed death-ligand 1; LAG3: lymphocyte activating 3; TCR: T-cell receptor; MHC: major histocompatibility complex; CD: cluster of differentiation; CTLA4: cytotoxic T-lymphocyte associated protein 4.

Overall, despite the advances listed above, patients with TNBC, especially those with metastatic disease, have short OS and continues to have the worst outcomes of all major subtypes of breast cancer. New strategies for therapeutic intervention including the testing of novel drug combinations for better performance are essential.

Treatment resistance and relapse, evidence of non-genetic mechanisms of transcriptome changes in intratumor heterogeneity

The evolutionary principles imparted by genetic intra-tumor heterogeneity are clear: oncogene mutations, amplifications, loss of function mutations in tumor suppressors, as well as large-scale chromosomal alterations are all well-established mechanisms which have previously been shown to drive cancer evolution and produce subpopulations of cancer cells for metastasis dissemination and outgrowth. It has long been thought that these genetic alterations are likely responsible for the expansion of resistant variants due to selective pressure (20–28).

However, recent progress demonstrate that distinct genomic alterations alone cannot fully explain the emergence or expansion of therapy-resistant cancer cells. Rather, epigenetic and transcription reprogramming are now understood to be key factors in driving tumor heterogeneity and evolution (29–34). Advances of single-cell technologies (scRNAseq and ATAC) have allowed the profiling of tumors at unprecedented depth and enable the dissection of non-genetic mechanisms that contribute to intra-tumor heterogeneity and malignant clonal fitness. In patient-derived-xenograft models of breast cancer, rare metastatic cells during seeding were identified by the global transcriptomic changes using single-cell RNA sequencing (35). The authors found both primary tumors and micrometastases display transcriptional heterogeneity but micrometastases harbor a distinct transcriptome program conserved across patient-derived-xenograft models that is highly predictive of poor survival of patients, which was the upregulation of mitochondrial oxidative phosphorylation (35).

Transcriptome changes have also been well recognized in populations of treatment resistant cancer cells; TNBC frequently develops resistance and relapse after chemotherapy, targeted therapy, or immunotherapy. One study examined samples before and after treatment from breast cancer patient who received neoadjuvant chemotherapy with and without Bevacizumab, an anti-VEGF antibody, and concluded that chemotherapy-induced transcriptional reprogramming of tumor cells are important mechanisms for subclonal adaptations to treatment (29). A study of 20 TNBC patients during neoadjuvant chemotherapy (NAC) show that resistant genotypes were pre-existing and adaptively selected by NAC, and interestingly, the same genotype cells further developed distinct subpopulations at the transcriptional level (31). In TNBC patient derived xenograft (PDX) models treated with neoadjuvant chemotherapies, the residual tumors showed transcriptomes, proteomes and histological traits which were distinctive from untreated tumors while still maintaining the genetic subclonal architectures. In addition, subclonal analysis of serial biopsies from TNBC patients before and after neoadjuvant chemotherapies found residual tumor-specific transcriptomic profiles enriched in oxidative phosphorylation pathway without major alterations in genetic subclones (36). These studies emphasize the vital roles of non-genetic heterogeneity in therapeutic resistance and metastatic progression of TNBC.

Epigenetic mechanisms in TNBC transcriptome regulation

Epigenetic dysregulation that leads to transcriptome changes has been recognized as a key player in tumor initiation and progression. Cancer associated epigenetic changes include well documented DNA hypermethylation and hypomethylation, dysregulation of covalent histone modifications, as well as disrupted chromatin architecture (37–43). Epigenetic dysfunction is also a key mechanism regulating how the tumor interacts with its microenvironment (44), and the immune landscape of cancer (45). The role of epigenetic changes as a primary non-genetic mechanism driving tumor heterogeneity has been increasingly recognized, and we now appreciate that epigenetic changes give rise to stable and heritable therapy-resistant subpopulations of tumor cells.

DNA methylome is the best understood epigenetic change in human cancers including promoter CpG island DNA hypermethylation of tumor suppressor genes and global DNA hypomethylation (37,38,41,46). Genome-wide DNA methylation profiling of 1538 breast cancers found subtype-specific DNA methylation signatures and DNA methylation-mediated epigenomic instability correlated with tumor grade, stage and poor prognosis of patients (47). DNA methylation profiling of TNBC tumors further identifies DNA methylation signatures for lymph node metastases and biomarkers which predict patient response to neoadjuvant chemotherapy (48,49). In the study with TNBC PDX models, the chemo-resistant and post-therapy residual tumors showed alterations in DNA methylation and expression of anthracycline resistant genes, such as SCL25A30 and NNMT (50). Our preclinical studies have also demonstrated that the de novo methyltransferase DNA methyltransferase 3B (DNMT3B) was elevated in metastatic TNBC cells in response to an inflammatory microenvironment (51), suggesting a critical molecular mechanism that links the metastatic microenvironment and epigenetic alterations of metastatic progression.

Dysregulation of covalent histone modifications is another major epigenetic mechanism that have been shown to be critical in transcriptomic reprogramming (40,43). High-throughput technologies have identified heterogeneity of histone modifications in TNBC that acquired resistance to chemotherapy. TNBC cells with chronic paclitaxel exposure gave rise to chemo-resistant subpopulations, which was regulated through increased H3K27me3 at the transposable elements thus blocked the transcription of dsRNA LINE1 and HERV that are important for the viral mimicry (52). Interestingly, an enhancer of zeste homolog 2 (EZH2) inhibitor, which decreases the trimethylation of histone H3K27, sensitizes the resistant cells to paclitaxel (52). This data is supported by another study which showed that EZH2 inhibition increased efficacy of paclitaxel chemotherapy in murine TNBC model (53). This study further identified transcriptional repression of major histocompatibility complex class I (MHC-I) by H3K27me3 in therapy resistant subpopulation (53). In addition, a preclinical study demonstrated high EZH2 expression in tumor initiating cells upregulates the transcription factors SOX2, OCT4, Nanog (54). EZH2 inhibition decreased metastasis in TNBC PDX models (54). These studies suggest H3K27me3, a relatively stable histone modification, could be a histone mark for tumor heterogeneity and chemoresistance. However, a contradicting study showed that loss of H3K27me3, through decreased recruitment of polycomb repressive complexes, promoted gene expression of IGF2BP3 and HOXD which are associated with capecitabine-resistance (55). It is not clear what underlies this discrepancy, but further studies are warranted.

Chromatin remodeling is one of the epigenetic mechanisms that significantly regulate transcriptome changes (42), and transcription factors plays a key role in establishing chromatin accessibility (56). Transcription factor-mediated changes in chromatin remodeling has been shown in multiple studies of TNBC, e. g. accessibility of SOX transcription factors is increased in stem/progenitor cells, but not in differentiated cells. This suggests that chromatin remodeling is critical in stem/progenitor development, dedifferentiation, and invasive characteristics of these cancer cells (57). Chromatin structure heterogeneity was also found in breast cancer subtypes, with TNBC cells displaying lysyl oxidase-like 2 (LOXL2) induced H3K4 oxidation and chromatin compaction resulting in inactivation of DNA damage response genes (58). In addition, super-enhancer heterogeneity has been demonstrated between breast cancer subtypes and FOXC1 has been identified as a key regulator of TNBC-specific super-enhancers (59). Furthermore, chromatin remodeling mediated by MUC-1 regulated JUN/AP-1 and ARID1A/BAF was uncovered at the enhancers of stemness genes in TNBC cells (60). These data suggest wide-spread chromatin deregulation driving various aspects of TNBC biology. Together, epigenetic regulators, including DNA methyltransferases and histone modifiers, chromatin modulators can regulate therapeutic responses through various mechanisms.

The crosstalk between tumor and the immune microenvironment: implication for TNBC immunotherapy

The immune microenvironment of tumors, described as chronic inflamed wounds, co-evolves with the tumor clonal selection and expansion ultimately resulting in metastatic colonization. The host anti-tumor inflammatory and immune responses have been increasingly appreciated, which lead to the use of immunotherapy as an effective cancer treatment strategy----most commonly with ICB. However, it is limited by low response rates and resistance/relapse to ICB with agents such as anti-cytotoxic T-lymphocyte associated protein 4 (CTLA-4) and anti-PD-1/PD-L1 (61,62). Strong evidence from immunogenomics analyses of more than 10,000 tumors across cancer types reveals a correlation of genomic aberrations with immune subtypes (45).

The immune phenotyping of TNBC have revealed unprecedent insight in the tumor ecosystem and to the success of therapeutic treatment. One study showed that large variations of neutrophils and macrophages that define ‘immune subtypes’ of TNBC, including neutrophil-enriched (NES) and macrophage-enriched subtypes (MES) using multiple murine models and clinical datasets (63). These TNBC immune subtypes exhibit variable responses to ICB. In particular, MES contains predominantly macrophages and exhibits variable responses to ICB. NES exhibits systemic and local accumulation of immunosuppressive neutrophils and is resistant to ICB. Interestingly, a MES-to-NES conversion mediated acquired ICB resistance of initially sensitive MES models (63). Tumor intrinsic resistance to tamoxifen and radiotherapy through the expression of interferon-stimulated genes was found via interaction with tumor-infiltrating lymphocyte (64). A single-cell atlas using mass cytometry analysis of 144 human breast tumors and 50 non-tumor tissue samples revealed high frequencies of PD-L1⁺ tumor-associated macrophages and exhausted T cells in high-grade ER⁺ and ER⁻ tumors (65). Very recently, a single-cell and spatially resolved transcriptomics analysis deconvoluted large breast cancer cohorts to stratify them into nine ecotypes. In particular, the molecular signatures from the PD-L1/PD-L2⁺ macrophage populations, mesenchymal cells as well as stromal-immune niches identified the ecotypes with unique cellular compositions and clinical outcomes (66). The E3 ecotype characterizes the basal subtype and showed poor 5-year survival compared with other ecotypes (66).

Apparently, the link between distinct genetic alterations in tumors and the composition of the immune landscape is critical in cancer biology. Emerging work from several groups suggest that silencing or mutation of tumor suppressors (TSs) is a critical contributor. TSs are well known as autonomous transcriptional and signaling regulators that negatively modulate cell cycle and survival and thus inhibit neoplastic transformation and tumor growth (67). As such, TSs counteract the growth promoting activity of oncogenes. Aside from mutation of TS genes as predisposition to cancers such as p53, PTEN, APC, BRCA1/2 (68)Wooster, et al., 1995; de la Chapelle, 2004; Hollander, Blumenthal, & Dennis, 2011; (69–72), the expression of TSs could be decreased or silenced through epigenetic mechanisms e.g. promoter hypermethylation (39,73–75). It is proposed that loss of TSs such as TGFβR2, PTEN, P53 and APC induces tumor autonomous inflammatory reprogramming and produce factors that promote an inflammatory and immune suppressive tumor microenvironment (76–78). Ongoing research using Cas9-sgRNA loss of function screening further support that TS silencing in tumor cells alters host inflammatory and immune responses (ongoing research in Yang lab). The immune microenvironment ranged from ‘cold’, noninflamed tumors to massively infiltrated landscapes depending on the genetic makeup in the tumor. These results apparently have important therapeutic implications. Although not from breast cancer studies but in prostate cancers, the immune-cell composition is driven by the loss of the critical tumor suppressor gene Pten, either alone or in combination with the loss of Trp53, Zbtb7a or Pml (79). Thus, relationship analyses between tumor intrinsic properties and immune cells revealed characteristics of ecosystems related to immunosuppression and poor prognosis.

Stratification of breast cancer patients per transcriptome changes and the immune landscape, potential treatment options

A significant development for TNBC treatment in recent years is the ICB approval by the FDA, with many clinical studies showing promises and at the same time challenges due to resistance and relapse (Table 1). Somatic mutations may predict resistance mechanisms or suggest sensitivity to ICB monotherapy as tumor mutation burdens (TMB) positively correlate with the ICB efficacy. However, TMB alone is not a perfect response biomarker (80). Rather non-TMB factors, such as replication stress response (81) or tumor-specific HMC-II expression (82) predicted clinical benefits of ICB in TNBC. Could we stratify patients based on biomarkers of epigenome, transcriptomes, and immune profiling, to overcome the low response rate and therapeutic resistance and relapse?

Table 1.

Clinical trials of ICBs in TNBC

TNBC Clinical trials		Drug	Efficacy in all patients	Efficacy in PD-L1+ patients
GeparNUEVO	Early TNBC	Nab-paclitaxel +/− Durvalumab	pCR: 53.4% vs 44.2%	pCR: 58% vs 50.7%
Keynote522*	Early TNBC	Carboplatin/paclitaxel + AC x4 +/− pembrolizumab	pCR: 64.8% vs 51.2% (P < 0.001) EFS at 18 month: 91.3% vs 85.3%	pCR: 68.9%. Vs 54.9%
NeoTRIPaPDLl	Early TNBC	Carboplatin/paclitaxel +/− atezolizumab	pCR: 43.5% vs 40.8%	pCR: 51.9% vs 48%
IMpassion031	Early TNBC	Paclitaxel + AC × 4 +/− atezolizumab	pCR: 57.6% vs 41.4% p = 0.0044	pCR: 68.8% vs 49.3% p = 0.021
I-SPY2	Early TNBC	Paclitaxel +/− pembrolizumab	pCR: 60% vs 22%
Kenote173	Early TNBC	pembrolizumab + Chemotherapy	pCR: 60%
NeoTRIPaPDL1	Early TNBC	Nab-paclitaxel/carboplatin +/− atezolizumab	pCR: 43.5% vs 40.8%	pCR: 51.9% vs 48.0%
IMpassion130	1st-line metastatic TNBC	Paclitaxel +/− atezolizumab	PFS: 7.2 vs 5.5 months	PFS: 7.5 vs 5.0 months
Keynote355**	1st-line metastatic TNBC	Chemotherapy +/− pembrolizumab	PFS: 7.5 vs 5.6 months	PFS CPS > 10: 9.7 vs 5.6 months
IMpassion131	1st-line metastatic TNBC	Paclitaxel +/− atezolizumab	PFS: 5.7 vs 5.7 months	PFS: 5.7 vs 6.0 months
COLET	1st-line metastatic TNBC	Cobimetinib(MEK inhibitor) + atezolizumab + paclitaxel	ORR: 31.7%	ORR: 44%
NCT03800836	1st-line metastatic TNBC	Ipatasertib/atezolizumab + paclitaxel	ORR: 73.0%	ORR: 82%
ENHANCE1	1st to 3rd-line metastatic TNBC	Eribulin + pembrolizumab	ORR: 23.4%	ORR: 34.5%
Kenote163	1st to 4th-line metastatic TNBC	Niraparib + pembrolizumab	ORR: 29.0%	ORR: 32%
Kenote899	2nd to 6th-line metastatic TNBC	GX-I7(long acting IL-7) + pembrolizumab	ORR: 13.3%
ENCORE602	2nd and later-line metastatic TNBC	atezolizumab +/− entinostat	PFS: 1.68 vs 1.51 months

pCR: pathological complete response; EFS: event free survival; PFS: progression-free survival; OS: overall survival; ORR: objective response rate.

^*Clinical study leading to the FDA approval of pembrolizumab in early TNBC patients.

^**Clinical study leading to the FDA approval of pembrolizumab in advanced TNBC patients.

Epigenetic dysregulation is recently recognized as one of the key hallmarks of cancer (83). Epi-drugs to target epigenetic modifiers (Figure 2) are actively being considered as potential ways to modulate therapeutic response to a wide range of therapeutic approaches including chemotherapy and immune based therapy. For example, DNA methylation inhibitors (e.g., Azycytidine and Decitabine), broad-spectrum histone deacetylase (HDAC) inhibitors (e.g., Entinostat, Belinostat and Panobinostat) and bromodomain and extra-terminal protein (BET) inhibitors (e.g., I-BET762 and ZEN003694) have been developed and investigated for treatment efficacy in clinical studies (84). Hematological malignancies have shown significant clinical benefits with multiple FDA-approved drugs, such as HDAC inhibitor for T cell lymphoma and DNMT inhibitors for acute myeloid leukemia and chronic myelomonocytic leukemia (85,86). While preclinical studies showed encouraging results of epigenetic drugs (epi-drugs) in solid tumors, including TNBC, their efficacy in patients has been disappointing (87). For example, the trial of atezolizumab (PD-L1 inhibitor) in combination with Entinostat (HDAC inhibitor) did not achieve clinical benefit in advanced TNBC patients (NCT02708680). A clinical trial to re-sensitize endocrine therapy-resistant patients by combination of DNMT inhibitor and Entinostat also did not achieve clinical benefits, although in the preclinical studies the re-activation of ER signaling in TNBC cells by the epigenetic drugs were observed (88,89). However, EZH2 was found to be a leading candidate of epigenetic modulators associated with the TNBC, and high expression of EZH2 predicts poor overall survival in TNBC patients. EZH2 inhibition differentiates EZH2(high) basal cells to a luminal-like phenotype by derepressing GATA3 and renders them sensitive to endocrine therapy (54). Specific targeting of EZH2 catalytic domain SET by small molecule suppressed distant metastasis in TNBC (54), while genetic disruption of EZH2 enhance tumorigenesis (90), emphasizing the fine-tuned inhibition of EZH2 to achieve proper clinical benefits. Several other studies also demonstrated that alterations of epigenome and epigenetic modifiers are good biomarkers for patient stratification for chemoresistance (52,91–93) (Table 2, Epigenetic markers). In addition, epigenetic drugs can directly affect differentiation and activation of immune cells, such as LSD1 inhibitor induced M1-like macrophage polarization (94), to alter tumor immune microenvironments, which were extensively discussed in review article by Yang et al. (95). These studies suggest that the therapeutic targeting of epigenetic enzymes could be used to decrease intra-tumoral cellular heterogeneity and treatment resistance. However, non-specificity and low efficacy in targeting epigenetic modifiers remains a significant concern.

Figure 2. Epigenetic drugs (Epi-drugs) and their targets.

Inhibitors of DNA methyltransferases (DNMT1, 3A and 3B) targeting DNA methylation; inhibitors of histone writers (histone acetyltransferases or HATs), histone erasers (histone deacetylases or HDACs) and histone readers (bromodomain proteins or BRDs) targeting histone modification; Inhibitors of PRC2 subunits (EZH2 and EED) targeting enzymatic activity and H3K27me3 binding pocket for chromatin remodeling. BET: bromodomain and extra-terminal protein; GCN5: general control non-repressed 5 protein; PCAF: P300/CBP-associated factor; MYST: HAT family of MOZ, Ybf2, Sas2 and Tip60; SIRT: Sirtuin; BRDT: bromodomain testis associated; PRC2: polycomb repressive complex 2; EZH2: enhancer of zeste homolog 2; EED: embryonic ectoderm development.

Table 2.

Patients stratification based on emerging epigenetic/transcriptional biomarkers and propose treatment options

Patient stratification based on epigenetic/transcriptional markers		Predicted responses to therapy			Proposed treatment option	Reference
		Chemotherapy	Immunotherapy	Targeted therapy
Epigenetic markers	Increased H3K27me3	Chemotherapy resistance			Chemo + EZH2 inhibitor	{Deblois, 2020 #2784}
	High EZH2	Platinum-based chemo response ↑				{Puppe, 2019 #2801}
	Low KDM4B	Anthracycline resistance				{Seoane, 2019 #2800}
	High DNA methylation of FERD3L, TRIP10	Chemotherapy resistance				{Pineda, 2019 #2808}
	High LSD1		ICB resistance		LSD1 inhibitor + ICB	{Qin, 2019 #2796}
	High DNA methylation of MHC-I		ICB resistance		DNMT inhibitor + ICB	{Luo, 2018 #2795}
	High DNMT1–3				decitabine	{Yu, 2018 #2809}
Transcriptional markers	High S100A	Chemotherapy resistance			Chemo + KDM6A inhibitor	{Lu, 2020 #2797}
	High MYCN	Chemotherapy resistance			BET + MEK inhibitor	{Schafer, 2020 #2798}
	Low IRF9	Chemotherapy resistance				{Brockwell, 2019 #2803}
	High expression of 4 gene signature (HLF, CXCL13, SULT1E1, GBP1)	Chemotherapy response ↑				{Criscitiello, 2018 #2804}
	High SLFN1/RB1 loss	TOP1 inhibitor response ↑				{Coussy, 2020 #2799}
	High HAGE	Anthracycline response ↑				{Abdel-Fatah, 2016 #2802}
	High TYRO3		ICB resistance		TYRO3 inhibitor + ICB	{Jiang, 2021 #2765}
	High MAL2		PD-L1 ICB resistance		MAL2 inhibitor + PD-L1 ICB	{Fang, 2021 #2806}
	High FGFR1–4				FGFR inhibitor	{Sanchez- Guixe, 2022 #2805}
Immune microenvironment signature	Increased CXCL1/5		ICB resistance		ICB+CXCR2 inhibitor	{Horn, 2020 #2779}
	NES and MES immune subtype		ICB response ↑		CCR2 inhibitor for MES subtype	{Kim, 2019 #2792}
	Increased adaptive immune and INFγ-related signature		ICB response ↑			{Jiang, 2019 #2771}
	Increased TGFbeta and CD4 T cells				CD4-TGFbeta trap	{Li, 2020 #2702}
	High macrophage signature			PARP inhibitor resistance	CSF1R-Ab + PARP inhibitor	{Mehta, 2021 #2756}

KDM4B: lysine demethylase 4B; LSD1: lysine-specific demethylase 1; FERD3L: fer3 like BHLH transcription factor; TRIP10: thyroid hormone receptor interact 10; MYCN: N-myc; IRF9: interferon regulatory factor 9; HLF: hepatic leukemia factor; SULT1E1: sulfotransferase family 1E member 1; GBP1: guanylate binding protein 1; SLFN1: schlafen 1, a negative regulator of cell cycle; HAGE: helicase antigen gene; TYRO3: tyro3 protein tyrosine kinase; MAL2: myelin and lymphocyte protein 2; NES: neutrophil enriched signature; MES: macrophage enriched signature; CSF1R: colony stimulating factor 1 receptor.

This difficulty could be overcome by careful patient selection, by identification of specific transcriptome changes associated with tumor resistance and heterogeneity. Transcriptional profiling of clinical and preclinical TNBC tumors has identified biomarkers associated with chemotherapy or immunotherapy resistance (96–99) (Table 2, Transcriptional markers). A preclinical study of TNBC E0771 mouse model identified transcriptional biomarkers that differentiate responders from nonresponders, and were associated with patient response to ICB (100). A recent study shows tumors with high TYRO3 expression exhibited anti-PD-1/PD-L1 resistance in mouse model and in patients, and inhibition of TYRO3 promoted tumor ferroptosis and sensitized resistant tumors to anti-PD-1 therapy (101). It is well known that TGF-β suppresses T helper 2 (TH2)-cell and myeloid cell mediated cancer immunity (102–104). For tumors that display increased TGF-β signaling, cancer immunotherapy via bispecific receptor decoy, ibalizumab, a CD4 antibody fused to a TGF-β-neutralizing TGFβR2 extracellular domain, was found to produce an effective anti-tumor immunity (105). Simultaneous targeting of TGF-β/PD-L1 by bintrafusp alpha (a bifunctional fusion protein of anti PD-L1 antibody and TGF-β RII extracellular domain) in combination with radiotherapy reprogrammed the tumor immune microenvironment and eradicated immune therapy resistant tumors in preclinical study (106). In addition, transcriptomic profiling of TNBC tumors uncovered an immunomodulatory subtype by elevated adaptive immune signature and INFγ-related pathways (107).

Inhibitors modulating immune microenvironment have provided additional options (Figure 3). In fact, patient stratification based on integrated genotypic-immunophenotypic analyses has been proposed as necessary for successful clinical trials and tailored precision immunological therapies (79). Defining ‘immune subtypes’ of TNBC, including NES and MES, helped to alleviate the immunotherapy resistance by C-C motif chemokine receptor 2 (CCR2) inhibition (63). In addition, inhibition of colony stimulating factor 1 receptor (CSF1R), a critical mediator tumor associated macrophage (TAM) recruitment and function has shown significant efficacy (108). CSF1R inhibitor significantly inhibited immune-suppressive macrophages and sensitized PARP inhibitor resistant TNBC tumors (109). The combined treatment of CSF1R inhibitor with low - dose cyclophosphamide led to durable regression of TNBC (110). TNBC are enriched in tumor associated neutrophils (TAN) through TGF-β regulation of C-X-C motif ligand (CXCL) chemokine ligands to the CXC chemokine receptor 2 (CXCR2) receptor (111). Targeting CXCR1/2 and TGF-β enhanced PD-L1 ICB through remodeling of the tumor microenvironment thus drove antitumor immunity (112). Furthermore, deconvolution of transcriptome data further revealed an immune microenvironment signature that predicts the patient response to ICBs (63,107,109) (Table 2, Immune microenvironment signature). No doubt that the multiplex profiling of immune cell subtypes likely allows for a more effective personalizing ICB-focused therapies in TNBC (113).

Figure 3. Inhibitors modulating the immune microenvironment.

Inhibitors targeting the interaction of cytokine/chemokine and their receptors in inflammatory monocytes, macrophages, neutrophils, as well as tumor cells. CSF1: colony stimulating factor 1; CSF1R: colony stimulating factor 1 receptor; CXCL1/5: C-X-C motif ligand 1 and 5; CXCR2: CXC chemokine receptor 2; CCL2: C-C motif chemokine ligand 2; CCR2: C-C motif chemokine receptor 2.

Scientific advances in recent years have substantially improved clinical outcome of breast cancer patients. However, metastatic TNBC is still detrimental disease with high rate of treatment resistance and relapse. TNBC displays extensive intra-tumoral heterogeneity and plasticity. In addition, the “cold” immune microenvironment in many TNBC, and the immune heterogeneity from different individuals, and from different metastatic organ sites are some of the most daunting challenges. Patient stratification based on transcriptome and epigenome changes, as well as characterization of immune microenvironment, together with chemotherapy, targeted therapies and/or immunotherapy will allow optimizing tailored precision immunotherapy for TNBC patients.

Abbreviation

BET

Bromodomain and extra-terminal protein

CCR2

C-C motif chemokine receptor 2

CSF1R

Colony stimulating factor 1 receptor

CTLA-4

Cytotoxic T-lymphocyte associated protein 4

CXCL

C-X-C motif ligand

CXCR

CXC chemokine receptor

DNMT3B

DNA methyltransferase 3B

ER

Estrogen receptor

ERBB2

Epidermal growth factor receptor 2

EZH2

Enhancer of zeste homolog 2

HDAC

Histone deacetylase

ICB

Immune checkpoint blockade

MES

Macrophage-enriched subtype

MHC-I

Major histocompatibility complex class I

NAC

Neoadjuvant chemotherapy

NES

Neutrophil-enriched subtype

OS

Overall survival

pCR

Pathological complete response

PFS

Progression free survival

PD-1

Programmed cell death protein 1

PD-L1

Programmed cell death-ligand 1

PDX

Patient derived xenograft

PR

Progesterone receptor

TAM

Tumor associated macrophage

TAN

Tumor associated neutrophil

TH2

T helper 2

TMB

Tumor mutation burden

TNBC

Triple negative breast cancer

TS

Tumor suppressor