Breast Cancer Immunity: It is TIME for the Next Chapter

Abstract

Our ability to interrogate the tumor immune microenvironment (TIME) at an ever-increasing granularity has uncovered critical determinants of disease progression. Not only do we now have a better understanding of the immune response in breast cancer, but it is becoming possible to leverage key mechanisms to effectively combat this disease. Almost every component of the immune system plays a role in enabling or inhibiting breast tumor growth. Building on early seminal work showing the involvement of T cells and macrophages in controlling breast cancer progression and metastasis, single-cell genomics and spatial proteomics approaches have recently expanded our view of the TIME. In this article, we provide a detailed description of the immune response against breast cancer and examine its heterogeneity in disease subtypes. We discuss preclinical models that enable dissecting the mechanisms responsible for tumor clearance or immune evasion and draw parallels and distinctions between human disease and murine counterparts. Last, as the cancer immunology field is moving toward the analysis of the TIME at the cellular and spatial levels, we highlight key studies that revealed previously unappreciated complexity in breast cancer using these technologies. Taken together, this article summarizes what is known in breast cancer immunology through the lens of translational research and identifies future directions to improve clinical outcomes.

The immune landscape in cancer is among the key determinants of disease prognosis and clinical outcome. Powered by cutting edge technologies, a growing collection of studies have revealed the complexity of the breast tumor immune microenvironment (TIME) and illuminated regulatory mechanisms that can potentially be leveraged for clinical benefit. In this article, we will discuss the state-of-the-art in breast cancer immunology; the approaches that are used for characterizing the tumor immune landscape; how the TIME differs between breast cancer subtypes; and identify emerging opportunities for modulating the immune microenvironment with the goal of eliminating tumors.

IMMUNE CELLS AND MECHANISMS IN BREAST CANCER

The breast cancer TIME is highly heterogeneous, varies between and within clinical subtypes, and contains immune cells from both lymphoid and myeloid origins (Yuan et al. 2021). T cells, B cells, and natural killer (NK) cells constitute major lymphoid lineages, while macrophages, monocytes, dendritic cells (DCs), and neutrophils comprise the predominant myeloid populations found in the breast cancer TIME. These cell types each exhibit heterogeneous functional states that can directly eliminate tumors or actively modulate the immune response to establish a protumorigenic or an antitumorigenic niche. The spatial composition and complexity of the TIME is increasingly recognized to distinguish breast cancers and their outcomes. The complexity of the TIME in breast cancer raises both new opportunities in disease management and challenges in developing appropriate research models. Here, we will summarize key findings from clinical and preclinical studies on the involvement of various immune cell types in regulating the outcome in breast cancer.

Cytotoxic and Helper T Cells

The role of T lymphocytes has been a predominant focus in breast cancer research. T cells are key effectors of adaptive immunity. CD8⁺ cytotoxic T lymphocytes (CTLs) directly induce tumor cell death, while CD4⁺ helper T cells orchestrate the functions of other immune cells within the TIME. CTL-mediated killing of a cancer cell involves antigen-specific recognition followed by the secretion of cytotoxic molecules including granzymes and perforin, as well as release of proinflammatory effector cytokines such as interferon γ (IFN-γ) (Barry and Bleackley 2002). Although T cells are the predominant tumor-infiltrating lymphocytes (TILs) in breast cancer (Ruffell et al. 2012), both numbers and composition of this infiltrate vary within and between tumors. Studies of large retrospective cohorts of breast cancer patients have shown that increased CD8⁺ T cells are generally associated with good outcome in breast cancer patients (Ali et al. 2014a). However, studies have shown that the infiltration of CD8⁺ T cells throughout the tumor bed or their retention in the stroma might differentially impact cancer outcomes, highlighting the importance of understanding the breast cancer immune contexture (Ali et al. 2014a; Khoury et al. 2018). Analyses of patient samples at the single-cell level, albeit with smaller cohorts, have started to demonstrate this immune heterogeneity in breast cancer. For example, examination of eight breast tumors suggested a higher heterogeneity in tumor-infiltrating immune cells from lymphoid and myeloid lineages when compared to matched normal breast tissue (Azizi et al. 2017). Furthermore, this work implies that T-cell clonality and phenotype are uniquely shaped by the tissue context, and that immune biomarkers in blood may not represent an accurate snapshot of the TIME. Notably, in breast tumors with a high-TIL content, a subset of CD8⁺ T cells exhibited a tissue-resident memory (TRM) phenotype characterized by CD103 positivity in addition to high expression of immune checkpoint and effector molecules (Savas et al. 2018). Further, the TRM gene-expression signature derived from CD8/CD103 dual positive cells more strongly predicted survival for breast cancer patients than did CD8 expression alone, suggesting that TRM cells may play a key role in cancer immunosurveillance and control. Interestingly, the development of the TRM phenotype in this cohort did not rely on pretreatment with immune checkpoint blockade (ICB) or neo-adjuvant chemotherapy. While the TRM phenotype was not exclusive to patients experiencing long-term survival in breast cancer, analysis of data from melanoma patients treated with ICB indicated that the TRM signature is enriched at baseline in responders. Furthermore, TRM-specific genes were significantly up-regulated during treatment with anti-PD1 ICB, suggesting that TRM may be key effectors of ICB response (Savas et al. 2018).

The importance of the composition of CD8⁺ T-cell infiltrate in combating tumor progression was also shown in a study examining the dynamics of the immune landscape during progression from precancerous breast lesions, ductal carcinoma in situ (DCIS), to invasive ductal carcinoma (IDC) (Gil Del Alcazar et al. 2017). This work revealed enhanced diversity and activation of CD8⁺ CTLs at microinvasive DCIS transition stages, whereas, in more advanced IDCs from the same patients, these cells exhibited exhaustion with a lower activation state and a less diverse T-cell receptor (TCR) repertoire. Thus, both the quantity and quality of tumor-infiltrating CTLs play an important role in controlling tumor progression. Critical roles of CD8⁺ CTLs in antitumor immunity have also been functionally confirmed in preclinical breast cancer models, where depletion of these cells leads to tumor progression (Gómez-Aleza et al. 2020; Lai et al. 2021; Li et al. 2021). In parallel, CD8⁺ CTLs have been shown to play important roles at distant metastatic sites. This was initially suggested by studies showing that breast cancer cells disseminate early during tumor development, but immunosurveillance at distant metastatic loci mediated mainly by CD8⁺ T cells prevents metastatic outgrowth (Eyles et al. 2010). A recent study identified CD39⁺PD1⁺CD8⁺ T cells in primary tumors as well as dormant metastatic nodules in mice (Tallón de Lara et al. 2021). These cells kept tumors dormant through secretion of TNF-α and IFN-γ and blocked metastatic outgrowth when adoptively transferred prior to intravenous tumor cell injection in an experimental metastasis model. These studies suggest that CD8⁺ T cells are key players in the immune system for controlling tumor progression and metastasis.

Compared to CTLs, the involvement of CD4⁺ helper T cells in breast cancer immunosurveillance appears to be more complex. A study modeling breast cancer using two distinct mouse strains (BALB/c and C57BL/6) revealed that CD4⁺ T cells have tumor-promoting effects and have negative prognostic associations with disease outcome in patients (Huang et al. 2015). Supporting this view, in an “immune-humanized” mouse model of triple-negative breast cancer (TNBC), CD4⁺ T cells were shown to infiltrate tumors and differentiate into immunosuppressive regulatory T cells (Tregs), which promote tumor growth through their immune-suppressive function (Su et al. 2017). Here, blocking the recruitment of naive CD4⁺ T cells into tumors reduced the ability of Tregs to accumulate and suppressed tumor growth. Increased Treg infiltration was also shown during progression from DCIS to IDC concomitant with CTL exhaustion (Gil Del Alcazar et al. 2017). On the other hand, in an HER2⁺ mouse model of breast cancer, treatment with an antibody-drug conjugate (ADC) targeting HER2 (trastuzumab-emtansine) elicited a strong antitumor immune response and led to an increase in Tregs in animals responding to treatment (Müller et al. 2015). However, in this case, the increase in Tregs was concomitant with a higher level of TILs with an antitumor T helper 1 (Th1) phenotype, which uniquely rendered tumors susceptible to anti-CTLA-4/PD-1 ICB immunotherapy. ICB therapy takes advantage of blocking inhibitory signals for T cells, such as those conveyed through the CTLA-4 or PD-1 receptors (Buchbinder and Desai 2016).

The Th1/Th2 paradigm for helper T-cell activation was defined as early as 1986 based on the cellular secretory phenotype (Mosmann et al. 1986), and it has been used extensively for studying the immune response, although some T-cell phenotypes may not exactly fit into these polarized classes (Muraille and Leo 1998). Th1 cells primarily secrete proinflammatory cytokines IL-12, TNF, and IFN-γ, which are important for antitumor immunity, while Th2 cells are characterized by the production of IL-4, IL-5, and IL-10, which are typically anti-inflammatory and counterproductive for antitumor immunity (Kim and Cantor 2014). Tregs can produce some of the same anti-inflammatory cytokines to modulate the immune response. Interestingly, Tregs were shown to be important for achieving a therapeutic window in which antitumor immunity can occur in the absence of autoimmunity, in the context of ICB (Müller et al. 2015). The importance of Th1 CD4⁺ T cells in breast cancer was further supported by identification of a stem-like, Th1-skewed CD4⁺ T-cell population that was essential for eliminating tumors that had metastasized to lungs in the PyMT mouse model of breast cancer (Fig. 1; Lai et al. 2021). On the other hand, another study found no significant association between metastasis-free survival and bulk CD4 or Foxp3 (key transcription factor for Tregs) expression within human TNBC (Schmidt et al. 2018), suggesting that bulk tumor mRNA analysis approaches may not be optimal for untangling complex relationships within the TIME.

Figure 1.

Mammary-selective overexpression of polyomavirus middle-T antigen (MMTV-PyMT) is a commonly used preclinical research model for breast cancer. These multicolor immunofluorescence images show metastatic lesions in mouse lungs 2 weeks after injecting MMTV-PyMT cells intravenously as part of an experimental metastasis assay. (A,B) Metastatic tumors in wild-type (FVB) mouse hosts are poorly infiltrated by immune cells (A) or exhibit an immune-excluded phenotype (B), similar to what is often observed in human metastatic breast cancer. (C,D) Genetic deletion of a negative immune regulator called “short-form Ron kinase” (Ron SF^−/−) results in swarming of metastatic tumors in the lungs by CD4⁺ and CD8⁺ T cells. Notably, these cells expressed TCF1, a transcription factor associated with increased effector and stem-like features (Lai et al. 2021). The robust infiltration of tumors with TCF1-expressing CD4⁺ and CD8⁺ T cells led to a lower tumor burden and long-term survival in mice. These findings suggest that preclinical research models are invaluable for identifying and validating new immunotherapy targets in breast cancer.

NK Cells

NK cells are innate lymphoid cells whose responses are controlled by a set of activating and inhibitory immune receptors. NK cells can be activated through a “missing-self” mechanism where the lack of MHC-I on the target cell surface results in failure to engage inhibitory NK receptors, leading to release of cytotoxic molecules. Thus, NK cells are critical for elimination of cancer cells that evade CTL recognition by down-regulating MHC-I expression (Guillerey et al. 2016). Multiple studies have shown that NK cells play an important antitumorigenic role in breast cancer, and NK cell dysfunction is associated with advanced disease. For instance, locally advanced breast cancer samples contain fewer numbers of NK cells compared to healthy breast tissue, and neoadjuvant chemotherapy was shown to increase NK cell numbers particularly in patients with pathological responses to treatment (Verma et al. 2015). Similarly, enhanced peripheral NK cell activity correlates with the disappearance of axillary node metastases in breast cancer after neoadjuvant chemotherapy (Kim et al. 2019). Furthermore, a small clinical trial involving high dose chemotherapy with subsequent autologous hematopoietic stem cell transplantation indicated that residual disease can be eliminated upon adoptive immunotherapy with IL-2 and NK cells in metastatic breast cancer (deMagalhaes-Silverman et al. 2000). Unsurprisingly, tumors can develop various mechanisms to evade NK-mediated cytotoxicity, including altered expression of NK cell receptors and ligands and secretion of immunosuppressive cytokines such as TGF-β (Mamessier et al. 2011; Slattery et al. 2021). Interestingly, NK cells in the peripheral blood of advanced breast cancer patients were shown to be enriched in immature and noncytotoxic subsets (Mamessier et al. 2013), suggesting that the tumor-derived factors can shape NK phenotypes. A recent study indicated that such factors not only can inhibit NK cell cytotoxicity, but they can also educate NK cells to a prometastatic state (Chan et al. 2020). Interestingly, protumorigenic effects of tumor-educated NK cells were blocked by antibodies specific to TIGIT, a negative immune checkpoint receptor that is currently being targeted in clinical trials (Chan et al. 2020). In addition to TIGIT, PD-1 can also be expressed on NK cells in tumor-draining lymph nodes (Frazao et al. 2019), opening interesting avenues for targeting the NK cell axis using ICB therapy in cancer.

Macrophages/Monocytes

Macrophages are highly diverse innate immune cells of the mononuclear phagocyte system and play critical roles in mediating the balance between cancer progression and antitumor immunity. Macrophage accumulation in breast tumors is often associated with poor survival in patients; however, these analyses rely heavily on general markers of bulk macrophages such as CD68. In reality, macrophages are much more diverse and can adopt both pro- and antitumorigenic identities. Historically, macrophages have been described to exist along a phenotypic continuum that ascribes a polarized activation status as either “M1” or “M2” (Mills et al. 2000). M1 macrophages are classically activated, often expressing proinflammatory markers including inducible nitric oxide synthase (iNOS) or CD80/CD86, and in cancer these are considered to be antitumorigenic. In contrast, M2-macrophages are alternatively activated, and express anti-inflammatory markers, CD163, Arginase-1 (Arg1) or CD206, and exhibit protumorigenic behavior. Many studies that describe tumor-associated macrophages (TAMs) assume an M2-like activation status; however, it is now appreciated that macrophages can adopt features of both M1 and M2 cells simultaneously, particularly when making comparisons between in vitro and in vivo models (Orecchioni et al. 2019). Therefore, this polarization paradigm has been dropped in favor of a broader appreciation for macrophage diversity and heterogeneity within tumors (Murray et al. 2014; Ginhoux et al. 2016).

TAMs are known to be phenotypically and ontogenically diverse (Lavin et al. 2015). Tissue-resident macrophages arise from the embryonic yolk sac during development and their replenishing ability is tissue-dependent (Ginhoux and Guilliams 2016). On the other hand, monocyte-derived macrophages are recruited from the bone marrow in response to inflammatory cues and can evolve from two developmentally distinct precursors (granulocyte-monocyte progenitors [GMPs] or monocyte-DC progenitors [MDPs]) (Yáñez et al. 2015, 2017). Within the fatty breast tissue, macrophages also display spatial and functional diversity in their activation status in accordance with the metabolic status of adipose tissue. Most macrophages within the breast adipose tissue, particularly in women with a high body mass index (BMI), are associated with adipocyte hypertrophy, where they form inflammatory crown-like structures (CLSs) that are associated with elevated breast cancer risk, distant recurrence, and cancer mortality (Iyengar et al. 2016; Koru-Sengul et al. 2016; Carter et al. 2018). Adding further complexity to macrophage diversity, CLS macrophages are lipid-laden and exhibit unique features, including high proliferation and expression of CD11c, CD9, or TREM2 (Patsouris et al. 2008; Amano et al. 2014; Hill et al. 2018; Jaitin et al. 2019). Moreover, the mammary gland undergoes various remodeling events during adult life, for example during lactation, which relies heavily on macrophage phagocytosis during postlactational involution (O'Brien et al. 2012). Therefore, macrophages within breast tumors are highly diverse despite many studies using broadly defining markers to study their functional contribution to breast cancer.

In preclinical models of breast cancer, macrophages represent the predominant myeloid component of the TIME, and macrophage depletion is sufficient to blunt tumor progression and reinvigorate antitumor immunity by the adaptive immune system (DeNardo et al. 2011; Strachan et al. 2013; Franklin et al. 2014; Ruffell et al. 2014). In the widely studied MMTV-PyMT model, at least two distinct macrophage populations have been described, including CD11b^hi MHC-II^hi tissue-resident macrophages with low proliferative capacity, and CD11b^lo MHC-II^hi macrophages that are more proliferative and are directly associated with the tumor (Dawson et al. 2020). Both macrophage subsets appear to be peripherally derived and replenished by inflammatory Ly6C^hi monocytes, which express CCR2 and CX3CR1 (Geissmann et al. 2003; Franklin et al. 2014; Dawson et al. 2020); however, it is unclear whether Ly6C^hi GMP- versus MDP-derived monocytes differentially contribute to these macrophage pools (although their high expression of MHC-II might suggest so) (Yáñez et al. 2017). Macrophage recruitment to breast tumors appears to be in part dictated by spatial features of the tumor microenvironment, such as hypoxia (Movahedi et al. 2010). Indeed, mechanistic studies have explored the profound effect of nutrient gradients on macrophage localization and activation states (Carmona-Fontaine et al. 2017; Gilmore et al. 2021), as well as the consequences of macrophage metabolism on nutrient partitioning between other cells within the tumor microenvironment, such as activated T cells (Vitale et al. 2019; Reinfeld et al. 2021). The robust heterogeneity of macrophages in cancer suggests there may be untapped therapeutic potential in reprogramming or targeting specific subsets of these cells in cancer.

Neutrophils

Similar to macrophages, there is a growing appreciation for the heterogeneity of neutrophils both in steady-state and inflammatory contexts (Ng et al. 2019; Hedrick and Malanchi 2022). Neutrophils are highly cytotoxic polymorphonuclear granulocytes that act as first responders during inflammation and comprise up to 70% of all peripheral leukocytes in humans. Although studies have historically viewed neutrophils as short-lived phenotypically homogenous cells, emerging research is demonstrating significant functional diversity based on physiologic context (e.g., tissue type [Casanova-Acebes et al. 2018; Ballesteros et al. 2020; Xie et al. 2020], time of day [Casanova-Acebes et al. 2013, 2018; Adrover et al. 2019, 2020], sex [Blazkova et al. 2017; Gupta et al. 2020; Lu et al. 2021], age [Wenisch et al. 2000; Hazeldine et al. 2014; Tomay et al. 2018; Lu et al. 2021], and microbiome [Clarke et al. 2010; Balmer et al. 2014; Deshmukh et al. 2014; Khosravi et al. 2014; Zhang et al. 2015]) as well as pathologic context (e.g., cancer-related surgical stress or infection [Cools-Lartigue et al. 2013; Tohme et al. 2016; Najmeh et al. 2017], smoking [Hosseinzadeh et al. 2016; Albrengues et al. 2018; Wculek et al. 2020b; Tyagi et al. 2021], and obesity [Xia et al. 2011; Nagareddy et al. 2014; McDowell et al. 2021]). Neutrophils develop in the bone marrow from GMPs and are replenished diurnally through granulopoiesis (Evrard et al. 2018). It is estimated that humans produce up to 1 billion new neutrophils per kg per day at steady state and up to 10 billion during inflammatory settings (Ley et al. 2018). Normally, neutrophils are released from the bone marrow as fully mature effector cells in response to granulocyte macrophage (GM)/granulocyte-colony stimulating factor (G-CSF) and CXCR2 chemokines such as IL-8; however, tumors chronically produce high levels of these factors (Kowanetz et al. 2010; Casbon et al. 2015; Coffelt et al. 2015; Wculek and Malanchi 2015; Hsu et al. 2019), causing neutrophils to be prematurely released as committed progenitors (Evrard et al. 2018; Zhu et al. 2018). Within tumors, multiple neutrophil identities have been described (Zilionis et al. 2019), although in breast cancer, similar in-depth characterization of neutrophil heterogeneity has yet to be fully defined.

Many studies on neutrophils and breast cancer have focused on their contribution to metastatic disease. In the MMTV-PyMT and KEP (K14cre;Cdh1^F/F;Trp53^F/F) preclinical models, neutrophil accumulation has been shown to precede the arrival of tumor cells within the premetastatic niche, where they promote colonization and outgrowth, such that neutrophil depletion with anti-Ly6G antibodies is sufficient to blunt metastatic progression (Coffelt et al. 2015; Wculek and Malanchi 2015). In the D2.0R breast tumor dormancy model, neutrophils contribute to awakening of dormant metastases within the lung by releasing neutrophil extracellular DNA traps (NETs) in response to inflammatory triggers within the host (e.g., tobacco, acute infection) (Park et al. 2016; Albrengues et al. 2018). Similarly, in the 4T1 model, it has been shown that neutrophils can cluster with circulating tumor cells within the periphery to promote cell-cycle progression during metastasis (Szczerba et al. 2019), while simultaneously blocking NK-mediated antitumor immune surveillance and facilitating extravasation (Spiegel et al. 2016). Emerging work now suggests that distinct genetic drivers of breast cancer may predispose to neutrophilic inflammation and facilitate metastasis. Notably, a comparison of 16 genetic mouse models of breast cancer demonstrated that loss of p53 within cancer cells and subsequent secretion of WNT ligands is a trigger for macrophage-dependent neutrophil infiltration and prometastatic functions (Wellenstein et al. 2019). Similar to macrophages, neutrophil metabolism has also been shown to play an important role in influencing antitumor immunity; although neutrophils are largely glycolytic, they adapt to the glucose-restricted microenvironment by engaging alternative metabolic pathways to sustain a protumoral phenotype in 4T1 breast cancer models (Rice et al. 2018; Hsu et al. 2019; Ombrato et al. 2019; Li et al. 2020). These studies shed light on the multifaceted effects of neutrophils at various stages of the metastatic cascade, and functional interdependency of cells within the TIME. However, because of their strong circadian rhythmicity, it is likely that developing neutrophil-targeted immunotherapies will require chronopharmacological considerations, thus making these cells challenging to target clinically.

Dendritic Cells

Derived from lymphoid and myeloid origins, DCs are key players responsible for bridging the innate and adaptive immunity. DCs are present in peripheral tissues and lymphoid organs, and they specialize in presenting extracellular and intracellular antigens to prime T-cell responses (Collin et al. 2013; Lee et al. 2017). Three major DC subsets were defined in humans based on distinct expression patterns of transcription factors IFN regulatory factor 4 (IRF4) and IRF8: plasmacytoid DCs (pDCs, IRF4⁺IRF8⁺), conventional DC1 (cDC1, IRF4^–IRF8⁺), and conventional DC2 (cDC2, IRF4⁺IRF8^–) (Guilliams et al. 2016; Collin and Bigley 2018). In addition to these, specialized DCs can participate in the immune response including Langerhans cells in squamous epithelial tissues and monocyte-derived inflammatory DCs (Mo-DCs) (Collin and Bigley 2018). As professional antigen-presenting cells, DCs can phagocytose extracellular antigens and present them on MHC-II molecules for priming CD4⁺ helper T-cell responses (Cabeza-Cabrerizo et al. 2021). However, one of the key features of DCs in the context of tumor immunity is antigen cross-presentation in which antigens from dying tumor cells can be internalized and loaded onto MHC-I molecules to induce CD8⁺ killer T-cell responses (Albert et al. 1998; Joffre et al. 2012). Antigen presentation capabilities of DCs are regulated by costimulatory ligands expressed on their surfaces, and a recent study in melanoma showed that antitumor functions of DCs may be circadian in nature through the regulation of CD80 expression (Wang et al. 2023). While most DC subsets are evolutionarily conserved across multiple mammals, the functional capacities of these cells and their surface markers may differ between species (Vu Manh et al. 2015; Guilliams et al. 2016). Studies in mice and humans have shown that DCs can exhibit a range of activation phenotypes and regulate immunological tolerance as well as inflammatory responses (Mellman and Steinman 2001; Anderson et al. 2017). Particularly, immature DCs presenting self-antigens can induce immunologic tolerance, whereas non-self-antigen-loaded mature DCs can assume a migratory phenotype to initiate a cascade of immunostimulatory reactions in lymphoid organs. The critical roles of DCs in regulating immunity have generated interest in modulating their functions for an effective antitumor immune response (Wculek et al. 2020a). Various therapeutics targeting DCs and/or their activation phenotypes have been clinically approved including tumor antigen-pulsed DC vaccine in prostate cancer (Kantoff et al. 2010), GM-CSF-expressing oncolytic virus in melanoma (Bommareddy et al. 2017), and TLR7/8 ligands in nonmelanoma skin cancers (Drobits et al. 2012). There are no DC-targeting therapies approved for breast cancer yet, but studies indicate that DCs can be dysregulated in this disease, suggesting possible intervention points.

DCs isolated from the peripheral blood of lymph nodes of breast cancer patients were shown to be dysfunctional and unable to induce T-cell activation in mixed lymphocyte reaction experiments (Satthaporn et al. 2004; Gervais et al. 2005). In this study, DCs from a wide range of breast cancer subtypes expressed lower levels of MHC-II and costimulatory molecules suggesting that they exist in an immature state. Other groups examined DCs within the TIME and revealed that DCs within the tumor bed have an immature phenotype, whereas mature DCs are located at the peritumoral sites (Bell et al. 1999; Coventry et al. 2002). A more recent study focusing on intratumoral DC subsets in luminal breast cancer and TNBC showed subset-specific gene-expression patterns when compared to DCs from the uninvolved tissues (Michea et al. 2018). Interestingly, the vast majority of differentially up-regulated genes in cancer-infiltrating DCs were found to be subset-specific, suggesting that there is not a strong tissue-specific gene-expression pattern shared among DC populations within breast cancer. Importantly, DCs in TNBC were enriched for immunological signatures including the IFN pathway, whereas DCs in luminal cancer were enriched for wound healing and extracellular matrix reorganization pathways (Michea et al. 2018). In addition to transcriptomic signatures, the composition of DCs within the breast TIME can have differential impact on clinical outcomes. For instance, DCs expressing a CD1a invariant MHC-I molecule were shown to be associated with reduced metastasis and longer survival (Coventry and Morton 2003; Giorello et al. 2021; Szpor et al. 2021). On the other hand, infiltration of breast tumors with CD123⁺ pDCs correlated with poor relapse-free and overall survival (Treilleux et al. 2004). In a later study, the negative effects of tumor-associated pDCs were attributed to Treg activation, and intratumoral administration of TLR7 ligand was shown to reverse pDC-mediated tumor promotion (le Mercier et al. 2013). These findings suggest that cancer subtype, spatial features within the tumor, and the context of the TIME can have a significant impact on the DC phenotypes in breast cancer. Indeed, the preconditioning of the DCs and/or the tumor microenvironment can be an effective approach to strengthen antitumor immunity. For instance, tumor-antigen-pulsed cDC1-type polarized DCs were shown to secrete high levels of IL-12 and expand tumor-specific CD4⁺ and CD8⁺ T cells in HER2⁺ breast cancer in clinical trials (Czerniecki et al. 2007). Similarly, coadministration of antigen-pulsed DCs and cytokine-induced killer (CIK) cells (CD3⁺CD56⁺ T cells generated in vitro from peripheral blood lymphocytes using anti-CD3 monoclonal antibody, IL-2, IL-1α, and IFN-γ) enhanced antitumor immune responses and prolonged survival in multiple studies (Wang et al. 2014; Shevchenko et al. 2020). Taken together, although the mechanisms by which DCs participate in the regulation of antitumor immunity remain to be fully elucidated, there is an active interest for modulating DCs as a cancer immunotherapy.

HIGH-THROUGHPUT APPROACHES FOR CHARACTERIZING THE BREAST CANCER TIME

The last few decades have witnessed an explosive growth in the amount of cancer genomic data. Historically, bulk tissue gene-expression microarrays, transcriptome analyses via bulk RNA sequencing, and mutation profiling through whole-exome sequencing have uncovered several complexities in breast cancer. The METABRIC microarray study, involving nearly 2000 primary tumor samples, was among the first comprehensive characterization efforts in breast cancer (Curtis et al. 2012). This work integrated copy number analysis with gene-expression profiling and identified cis- and trans-acting genomic aberrations. This high dimensional data defined novel patient subgroups with varying prognoses and intratumoral immune involvement. Shortly after the landmark METABRIC study, a comprehensive analysis of 825 breast cancer samples was released by the TCGA network, which involved DNA copy number arrays, reverse phase protein arrays, DNA methylation and exome sequencing, and mRNA and miRNA-expression profiling (Koboldt et al. 2012). Both studies reported overlapping and complementary findings on two independent data sets and defined the molecular characteristics of human breast cancer. Many subsequent studies have used these data sets and found tumor-specific immune signatures for predicting survival and immunotherapy response in cancer (Charoentong et al. 2017; Ock et al. 2017; Auslander et al. 2018; Cristescu et al. 2018).

Although the data gathered from these bulk tissue analyses have been indispensable for new discoveries and are still heavily used, findings can be difficult to attribute to particular cell types or individual phenotypic states. With the advent of single-cell analysis technologies, it is now clear that the TIME contains highly diverse immune lineages with considerable functional heterogeneity. For example, as discussed above, several landmark studies using single-cell transcriptomics in breast cancer have led to new discoveries that have deepened our understanding of TAM and T-cell biology, as well as the heterogeneity of malignant cells (Azizi et al. 2017; Lai et al. 2021; Pal et al. 2021). Others have further expanded our knowledge about the breast cancer TIME by interrogating the single-cell spatial proteome through high-dimensional mass cytometry approaches (Schulz et al. 2018; Ali et al. 2020; Jackson et al. 2020). These studies have demonstrated that cellular neighborhoods in breast cancer are predictive of patient outcomes, and that spatial examination of the TIME can better segregate patient subsets compared to current clinical classifications (Ali et al. 2020; Jackson et al. 2020). Taken together, integration of multiple technologies has enabled an unprecedented understanding of the breast cancer TIME. The increasing availability of these high-dimensional data sets will also continue to fuel the development of visual analysis interfaces and dashboards to further facilitate data sharing and dissemination (Fig. 2). Importantly, iterative analyses of large amounts of data from these studies using sophisticated computational methods, such as machine learning and deep learning, will surely open new avenues for research in the years ahead.

Figure 2.

The rapid accumulation of single-cell transcriptomics data has led to the development of user-friendly web interfaces to facilitate data exploration and dissemination. This screenshot was obtained from the Broad Institute's Single Cell Portal (singlecell.broadinstitute.org), which hosts 436 single-cell RNA sequencing data sets at the time of this writing, and at least 10 of these data sets were derived from breast cancer studies. The Single Cell Portal enables users to interact with complex multidimensional data in a point-and-click manner to visualize single-cell clusters and expression of genes of interest. In this example, more than 100,000 cells from 26 different breast tumor samples are shown based on Wu et al. (2021). Such visual analysis solutions are key for bridging the gap between disciplines and increasing data usability.

IMMUNE LANDSCAPES OF BREAST CANCER SUBTYPES

Breast cancer has been clinically characterized as estrogen and progesterone receptor-positive (ER⁺, PR⁺, or hormone receptor-positive, HR⁺), human epidermal growth factor receptor 2-positive (HER2⁺), or the TNBC subset, which lacks expression of ER, PR, and HER2. Examination of gene-expression patterns in resected tumors demonstrated the heterogeneity beyond these classifications and indicated that breast cancer can be divided into at least five prognostically variant subtypes, including luminal-A, luminal-B, basal, HER2⁺, and normal breast-like (Perou et al. 2000; Sørlie et al. 2003). ER⁺ tumors are characterized by a relatively high expression of genes associated with the luminal subtype, while TNBC tumors are mostly basal in nature (Perou et al. 2000). Analyses of breast tumors from patients and preclinical research models have revealed that the tumor subtype affects the composition of immune infiltration. Although the prevalence of TILs is higher in HER2⁺ and TNBC subtypes (Ali et al. 2014a), TIL infiltration and abundance is a prognostic biomarker for all breast cancer subtypes and an independent predictor of response to neoadjuvant chemotherapy (Loi et al. 2013). Guidelines are now established by the international TILs working group on scoring TILs in breast cancer (El Bairi et al. 2021), and in 2019 the WHO officially listed TILs as one of the biomarkers for the clinicopathological analysis of breast cancer with efforts to include TIL scores into practice to further define disease subsets.

HR⁺ Breast Cancer

Historically, HR⁺ breast tumors have been considered immunologically cold as there are few T cells associated with these tumors (Loi et al. 2014). However, other immune cells can be associated with the TIME in HR⁺ breast cancer. In an analysis of over 11,000 HR⁺ breast tumors, the immune parameter that correlated most significantly with poor clinical outcome was the presence of TAMs (Ali et al. 2016; Bense et al. 2016). Protumorigenic M2-like TAMs have been shown to induce endocrine therapy resistance in HR⁺ breast cancer cells in vitro and in vivo through NF-κB- and IL-6-dependent signaling pathways (Castellaro et al. 2019). In contrast, a higher fraction of M1-like TAMs in HR⁺ breast cancer (which can mediate phagocytosis and cross-presentation of antigen to T cells) correlated with a higher pathologic complete response (pCR) rate as well as prolonged disease-free survival and overall survival (Bense et al. 2016). Macrophage depletion has shown promise in solid tumors including breast cancer (Wesolowski et al. 2019), and identifying strategies to harness the antitumor potential of macrophages may offer new potential opportunities for the treatment of HR⁺ breast cancer. Due to the low abundance of CD8⁺ T cells in HR⁺ tumors, this subtype is unlikely to benefit from current ICB therapies that aim to reinvigorate a previously existing but ineffective T-cell response. However, since dying cells can elicit an enhanced immune response through release of tumor antigens, coupling therapies for HR⁺ tumors, such as CDK4/6 inhibitors, together with ICB agents may be a useful therapeutic approach (Zhang et al. 2019).

HER2⁺ Breast Cancer

HER2⁺ breast cancer is a heterogeneous disease with HR⁺/HER2⁺ and HR^–/HER2⁺ subsets. Among different therapeutic approaches, immunotherapies represent a relevant option for HER2⁺ breast cancer patients, both in the adjuvant and metastatic setting. The concept that TILs are a good prognostic indicator for HER2⁺ breast cancer is generally accepted (Denkert et al. 2018), although the median TIL levels are lower in HR⁺/HER2⁺ tumors than in HR^–/HER2⁺ tumors (Solinas et al. 2017). Conversely, elevated immunoregulatory cells, such as Tregs and M2-like macrophages, have been associated with poor prognosis (Bates et al. 2006; Mahmoud et al. 2012). HER2-targeting monoclonal antibodies such as trastuzumab and pertuzumab have dramatically improved the clinical outcome in HER2⁺ breast cancer by, in part, activating the immune response through antibody-dependent cellular cytotoxicity (Demonty et al. 2007). This implies that HER2⁺ tumors that progress on these now standard-of-care therapies, may already have protumorigenic immune microenvironments. Murine models of HER2⁺ breast cancer have been useful for understanding the mechanisms of trastuzumab resistance and characterizing the immune cell dynamics during ineffective antitumor immunity. For example, recent work involving mammary-specific expression of Neu, the rat homolog of HER2 (MMTV-Neu model), tested the efficacy of combining trastuzumab and CDK4/6 inhibitors and showed an enrichment of an immunosuppressive myeloid cell population in tumors that are resistant to the therapy (Wang et al. 2019a). In a search to find efficacious drug combinations, another study involving a similar model reported that PD1-blocking antibodies augment anti-HER2 therapy responses (Stagg et al. 2011). Many clinical trials are now underway to test immune therapies in HER2-positive tumors and are covered in companion literature (see Emens and Loi 2023).

TNBC Subtype

TNBC tumors are associated with the highest TILs of all breast cancer subtypes, likely reflecting the higher mutational complexity of TNBC (Karaayvaz et al. 2018; Bao et al. 2021). Although great patient-to-patient heterogeneity exists within TNBC, leukocytes can comprise more than 60% of all cells, and 40%–60% of the tumor immune infiltrate is composed of T cells (Gil Del Alcazar et al. 2017). To define their immunological state, solid tumors have been categorized into three groups: “immune-desert” (cold) tumors largely devoid of lymphocytes, “immune-excluded” where lymphocytes are present in the peritumoral stroma only, and “immune-infiltrated/inflamed” (hot) tumors (Chen and Mellman 2017). Among these, immune-cold breast cancers are characterized by the worst survivability, immune-excluded tumors exhibit somewhat poor prognosis, and the immune-hot tumors are associated with better outcomes (Ali et al. 2014a,b). These observations suggest that where the immune cells are found in the tumor matters, and they provide an important insight as to why immune signatures alone from bulk gene-expression data from tumor biopsies may be confounding (Tofigh et al. 2014; Gruosso et al. 2019). Nevertheless, bulk gene-expression data are more readily available in the literature at this time and efforts to decipher the complexity of TIME in these data sets continue. For instance, a landmark study investigating bulk gene-expression data from The Cancer Genome Atlas (TCGA) across different tumor types defined six unique immune signatures (C1–6) within human cancers (Thorsson et al. 2018). Among these, the basal molecular subtype of breast cancer, which corresponds to the majority of TNBC, was shown to be enriched in C1 (wound healing, 60.5%) and C2 (IFN-γ-dominant, 34%) states (Thorsson et al. 2018), indicating the heterogeneity of the immune infiltration in TNBC.

A spatial examination of the TIME has provided a further understanding of immunological heterogeneity within the tumor architecture of TNBC (Fig. 3), which can have a significant impact on outcomes. Spatial enrichment analyses show “immune-infiltrated/inflamed” TNBC tumors in which Gzmb⁺CD8⁺ cytotoxic T cells are distributed throughout the tumor bed and exhibit a type 1 IFN signature, M1-like macrophages, and elevated expression of immune inhibitory molecules PD-L1 and indoleamine 2,3-dioxygenase (IDO) have a better prognosis than TNBC tumors in which immune cells are separated from the cancer cells (Gruosso et al. 2019). These “immune-excluded” tumors displayed hallmarks of exhausted CD8⁺ T cells, elevated levels of protumorigenic M2-like macrophages, as well as neutrophils and IL-17⁺ T cells, which is consistent with the promotion of metastatic spread in the PyMT murine model of breast cancer by IL-17-expressing γδT cells (Coffelt et al. 2015). However, using imaging mass cytometry and machine learning, another study categorized TNBC tumors into “mixed” and “compartmentalized” subsets in which immune cells are found within the tumor or spatially separated, respectively, and reported that a subset of “compartmentalized” tumors has improved survival (Keren et al. 2018). Notably, the expression patterns of immune checkpoint molecules differed between these spatial configurations. In particular, PD-1/PD-L1 interactions occurred between CD4⁺ T cells and macrophages in compartmentalized tumors, while they were more frequent between CD8⁺ CTLs and tumor cells in mixed tumors (Keren et al. 2018). The heterogeneous spatial composition of the TIME and its correlation with differential outcomes were recently shown by the ARTEMIS clinical trial (NCT02276443) designed to assess response to initial neoadjuvant therapy using diagnostic imaging followed by personalized targeted therapy in TNBC. The authors reported a shorter physical distance between malignant cells and CD3⁺ T cells in patients experiencing pCR compared to patients with residual cancer burden (Yam et al. 2021). Interestingly, patients responding to neoadjuvant chemotherapy tended to have more clonal intratumoral T-cell populations and higher PD-L1 positivity prior to treatment in this study. These data suggest that the cellular and spatial context is an important determinant within the TIME. It is important to note that the choice of the analytical method can have a strong impact on the predictive molecular signatures. At the other end of the spectrum of intratumoral inflammation, “immune-cold” TNBC is characterized by the absence of CD8⁺ T cells, increased fibrosis, TGF-β-regulated gene-expression signature, M2-polarized macrophages, as well as elevated expression of the immunosuppressive PD-L1 family member, B7-H4, but not PD-L1 (Altan et al. 2018; Gruosso et al. 2019). In contrast, B7-H4 was absent in inflamed PD-L1-positive tumors, suggesting this could be a novel immunotherapy target in TNBC (Gruosso et al. 2019; Li et al. 2019b).

Figure 3.

Highly multiplexed imaging of the breast tumor microenvironment reveals the complexity of the immune infiltration. Representative images from imaging mass cytometry (IMC) on human breast tumor tissues from four patients (one patient per column). Each column depicts a patient tumor (1 mm² area) from the indicated histological immune subtype (as defined in Gruosso et al. 2019). Each row represents different combinations of markers within each image (right color legends).

Large-scale immune classification from transcriptomic data from a cohort of 299 patients (METABRIC study) supported the notion that TNBC can be subdivided into these three TIME categories based on immune-specific gene-expression profiles to predict survival and sensitivity to PD-1 blockade (Bareche et al. 2020; Hu et al. 2021). Interestingly, multiple studies reported the presence of tertiary lymphoid structures, organized lymph node–like structures at the peritumoral regions, and their positive association with survival in TNBC (Martinet et al. 2011; Gu-Trantien et al. 2013). Recent studies used imaging mass cytometry to further dissect the features of intratumoral lymphoid and vascular structures and contributed to the complex picture of the TIME in breast cancer (Danenberg et al. 2022).

Despite success in other cancer types, ICB is not yet optimally effective in TNBC, and various strategies have been proposed to boost antitumor immune mechanisms (Vonderheide et al. 2017). Additionally, efforts are underway to more accurately stratify patients with inflamed tumors to obtain better immunotherapy outcomes. Importantly, a recent clinical trial indicated that preconditioning with cisplatin or doxorubicin can lead to up-regulation of inflammatory genes and improve anti-PD-1 responses in TNBC, suggesting that the TIME can be rewired to elicit an effective antitumor state (Voorwerk et al. 2019). Furthermore, several ADCs are currently approved or being evaluated in the clinic for the treatment of breast cancer (Nagayama et al. 2020). In these studies, different types of cytotoxic payloads are conjugated to antibodies against molecular targets, including HER2, TROP-2, LIV-1, and EGFR, that are variably expressed in TNBC. It is possible that increasing tumor cell death using ADCs may modulate the TIME in a way that makes immunotherapy more effective. Taken together, it is becoming increasingly clear that the highly heterogeneous TIME in TNBC provides both opportunities and challenges for reaching the full potential of immunotherapy.

EMERGING OPPORTUNITIES TO MODULATE THE TIME IN BREAST CANCER

Immune Checkpoint Inhibitors

As we better understand the immunoregulatory mechanisms in breast cancer, we are becoming able to devise new therapeutic approaches to improve patient outcomes. The clinical success of ICB therapies such as anti-PD-1 and anti-CTLA-4 in a few cancer types has placed immunotherapy in the center of attention for cancer therapy in general (Callahan et al. 2016). PD-1 and CTLA-4 immune checkpoint receptors are up-regulated in activated T cells and their engagement with the ligands PD-L1/PD-L2 and CD80/CD86, respectively, results in the inhibition of T-cell functions (Wei et al. 2018). Antibody-mediated blockade of these receptors increases T-cell infiltration into tumors and the production of immune effector molecules (Curran et al. 2010). Although ICB can lead to complete tumor clearance in some cases, the majority of patients do not benefit from therapy (Wolchok et al. 2017). In the context of breast cancer, ICB is approved only for TNBC but still has minimal efficacy in that subtype (Villacampa et al. 2022). A phase 1b clinical trial testing the PD-1 inhibitor pembrolizumab in TNBC reported an 18.5% overall response rate, even though patients were enrolled based on PD-L1 expression in the stroma or tumor (Nanda et al. 2016). Among the key challenges of ICB therapy in breast cancer are nonexistent or modest TIL levels in the “immune-cold” tumors, low tumor antigenicity, and the presence of compensatory immunoregulatory mechanisms in the TIME (Vonderheide et al. 2017). There are many current studies now investigating the effects of combining ICB with other anticancer agents and targeting other immune-modulatory processes in breast cancer, some of which are discussed below. For further information on clinical breast cancer immunotherapy, we refer the reader to Emens and Loi (2023).

Myeloid-Targeted Therapies

In light of the profound contributions of myeloid cells to the progression and metastasis of solid tumors, there is an emerging interest in developing immunotherapies that target specific subsets of these cells. Notably, CSF-1R inhibitors have been used extensively both in preclinical models and in patients to deplete or reprogram macrophages in cancer. Blockade of CSF-1R signaling has been proven to have multiple benefits, including blunting tumor growth and metastasis (Strachan et al. 2013; Klemm et al. 2021) and enhancing chemotherapeutic response (DeNardo et al. 2011; Ruffell et al. 2014; Olson et al. 2017). Building on these findings, studies have now begun to explore the therapeutic potential of targeting specific macrophage ontogenies in cancer. For example, targeting TAMs in breast cancer models is effective to restore antitumor adaptive immune responses, whereas targeting resident mammary macrophages is not (Franklin et al. 2014). Similarly, anti-CD49d has been identified as a putative approach to target monocyte-derived macrophages in mouse models of breast cancer metastasis to brain, while leaving the nonreplenishing tissue-resident microglia population intact (Bowman et al. 2016). As an alternative approach to depleting specific macrophage populations, studies have also developed methods to reprogram macrophages to adopt an antitumorigenic identity. For example, class IIa HDAC inhibitors can reprogram monocytes and macrophages to limit tumor progression in preclinical breast cancer models, while leaving lymphocyte functions intact (Lobera et al. 2013; Guerriero et al. 2017). Class IIa HDAC inhibition elicits a robust antitumor effect by enhancing macrophage phagocytosis and synergizes with chemotherapy and ICB. These discoveries and others highlight the potential value in reprogramming the immunosuppressive myeloid niche to leverage existing standard-of-care treatments or immunotherapies.

Similarly, there is growing interest in targeting neutrophils in the context of cancer, although there are complex considerations. First, neutropenia is often a consequence of systemic chemotherapy, and it is treated clinically with recombinant G-CSF to boost white blood cell counts. Although necessary to avoid serious risks from infection, it is unclear whether G-CSF supplementation has counterproductive effects in cancer patients by stimulating neutrophil production and activation. Fortunately, retrospective analysis of breast cancer patients with brain metastasis has shown that G-CSF treatment does not exacerbate metastatic disease (Fujii et al. 2021); however, this has yet to be expanded to additional metastatic contexts, particularly lung or liver where the majority of preclinical work has been done. Second, although preclinical models have focused on neutralizing antibodies against the canonical neutrophil marker Ly6G, this causes rapid and robust neutropenia, which is not translatable to patients. Therefore, therapies that target protumorigenic neutrophil functions are receiving more attention, such as NET inhibitors like recombinant DNase1. Although these have not yet been evaluated in cancer patients, emerging clinical trials are now using NET inhibitors for the first time in patients with severe COVID-19 (NCT04409925, NCT04359654, NCT04445285, NCT04541979, NCT04432987, NCT04402944). Finally, given the robust fluctuations in neutrophil frequency and effector function in response to the circadian clock (Casanova-Acebes et al. 2018; Adrover et al. 2019), it is likely that chronopharmacology-based approaches will need to be considered for the development of neutrophil-targeted therapies in the clinical setting.

Emerging Targets and Strategies to Reinvigorate Antitumor Immunity

Resistance to chemotherapy and metastasis are recognized as main factors of mortality in breast cancer. Studies have reported that up to 20% of treatment-naive patients show response to single-agent anti-PD-L1 therapy (Adams et al. 2019), which is similar to first-line chemotherapy response rates in TNBC (Li et al. 2019a). This raises an interesting point about potential immune-mediated mechanisms of tumor clearance in chemotherapy. TNBC patients that respond to PD-L1 ICB are predicted to be those that display an inflamed TIME, which reflects 30%–40% of TNBC patients (Bareche et al. 2020; Hu et al. 2021), and it remains to be determined whether specific chemotherapeutic agents can potentiate immunotherapy.

The TIME is a complex network of internal and external components that are sculpted during tumor immune evolution (Finak et al. 2008). Recent multiplex and single-cell analyses of immune microenvironments demonstrate that T cells within the TIME can be found in transitionary activation states as well as terminal differentiation and exhaustion associated with signatures of hypoxia (Azizi et al. 2017). Moreover, many immune cells display covariant expression patterns for the costimulatory and coinhibitory molecules, which suggests targeting one alone may not be a sufficient immunotherapy approach. For example, Foxp3⁺ Tregs that express coinhibitory CTLA-4 along with other coinhibitory molecules, TIGIT, and the costimulatory receptor GITR have been demonstrated to selectively inhibit proinflammatory Th1 and Th17 but not Th2 responses (Joller et al. 2014). Similarly, macrophage populations are demonstrated along a continuum, with enrichment in M2-like macrophages for the scavenger receptor MARCO and coinhibitory receptor B7-H3 in humans and mouse models (Kos et al. 2022). The B7 family of immune checkpoint molecules, of which PD-L1 is a member (B7-H1), has been recognized as crucial modulators of adaptive immunity. However, members of this family, such as the coinhibitory molecules B7-H3 and B7-H4, can be expressed both in immune and tumor cells and often correlate with immune-cold TNBC subtypes (Altan et al. 2018; Gruosso et al. 2019; Bareche et al. 2020). Thus, these molecules may both promote tumor-intrinsic mechanisms of progression in addition to stimulating an immunosuppressive tumor microenvironment (Podojil and Miller 2017).

In addition to complexity within the immune cell subsets, studies have shown that human breast cancer cells can aberrantly express genes classically thought to be immune cell–specific, such as components of class-II antigen presentation machinery (MHC-II), and this can correlate with favorable prognosis and long-term disease-free survival (Andres et al. 2016). Mechanistic investigations in mice showed that MHC-II expression in breast cancer cells promote antigen presentation to CD4⁺ T cells and elicitation of Th1 response leading to tumor rejection (Mortara et al. 2006). Hence, accurately stratifying patients for ICB and improving the therapeutic outcomes require a fuller understanding of the complex TIME in breast cancer patients as well as tumor cell capacity to display immunogenic antigens.

It is increasingly appreciated that targeted therapies can also modulate the TIME and thus can be considered as unorthodox immunotherapeutics (Petroni et al. 2020). For instance, CDK4/6 inhibitors intended for cell-cycle arrest can lead to immunostimulation through multiple mechanisms. One of these mechanisms may involve induction of cancer cell senescence, which is characterized by secretion of chemokines and expression of ligands responsible for activating NK and T cells (Textor et al. 2011; Iannello et al. 2013). CDK4/6 inhibitors can also lead to up-regulation of antigen presentation machinery and improve ICB responses in a senescence-independent manner (Goel et al. 2017; Schaer et al. 2018). In addition to effects on tumor cells, cell-cycle inhibitors can directly regulate immune cells within the TIME. Studies in murine breast and lung cancer models have shown that CDK4/6 inhibition can block Treg functions and enhance antitumor T-cell responses (Deng et al. 2018; Schaer et al. 2018). These findings indicate that targeted therapeutics have multiple mechanisms of action and can potentially be combined with other immunotherapeutics for improved outcomes in cancer.

Considerations for Clinical Translation of Emerging TIME Therapies

As emerging immunotherapies continue their development in the preclinical setting, an important hurdle that we will face is the practicality of translating treatments to the clinic where patients are immunologically diverse. This is akin to comparative studies conducted between wild mice and laboratory mice, which demonstrated a striking enhancement in immune activity in wild mice driven by a high degree of environmental pathogen exposure (Abolins et al. 2017). These seminal discoveries called into question the relevance of laboratory mice to study immunological diseases, where experimental conditions are more sterile and all physiologic variation is tightly controlled (e.g., diet, exercise, temperature, genetics, etc). Indeed, recent clinical discoveries have shed light on how immunological diversity among humans impacts cancer immunotherapy. For example, the gut microbiome, which is heavily influenced by diet and other lifestyle factors, is emerging as a major regulator of ICB efficacy both in mouse (Gopalakrishnan et al. 2018; Matson et al. 2018; Routy et al. 2018) and human (McQuade et al. 2018; Baruch et al. 2021) disease settings. Similarly, preclinical and epidemiological studies have shown that obesity is associated with an unexpected improvement in response to ICB (McQuade et al. 2018; Wang et al. 2019b; Kichenadasse et al. 2020), owing to its chronic inflammatory effects that are otherwise usually harmful. Similar observations have been made in aging studies, where it has been discovered that older melanoma patients respond to ICB better than younger patients, which may be due to a higher CD8⁺/Treg cell ratio (Kugel et al. 2018; Fane and Weeraratna 2020). Additionally, various aspects of lifestyle, such as stress or circadian disturbances, also affect immunological responses and cancer outcomes. For example, stress hormones can lead to activation of neutrophils, which can awaken dormant tumor cells and potentially act as a trigger for recurrence (Perego et al. 2020). Taken together, these findings raise the question of whether epigenetic reprogramming of immune cells via trained immunity—uniquely shaped by a lifetime of pathogen exposures, vaccinations, chronic diseases, diet, and other lifestyle factors within each person—explains why differential responses to cancer immunotherapies have been so challenging to predict at the individual level (Christ et al. 2018; Kaufmann et al. 2018; Mitroulis et al. 2018; Chavakis et al. 2019; Kalafati et al. 2020; Netea et al. 2020). They also highlight the importance of understanding disparities in cancer outcomes through an intersectional lens, by taking into consideration age, gender identity, racialization, weight, socioeconomic status, and other factors that may influence treatment access and efficacy (Murthy et al. 2004; Brady and Weeraratna 2020; Carpten et al. 2021).

FUTURE DIRECTIONS

Developing technologies have provided us with an unprecedented view of the breast cancer TIME. Not surprisingly, as we examine more patient samples at an ever-increasing resolution, we are discovering previously unappreciated complexities of breast cancer and how the immune system participates in disease control and progression. Although bulk transcriptomic and genomic analyses have equipped us with essential insights about disease mechanisms, the tumor immunology arena is undoubtedly shifting toward interrogating the TIME at single-cell resolution in situ (Giesen et al. 2014; Ali et al. 2020; Jackson et al. 2020; Andersson et al. 2021; Wu et al. 2021; Danenberg et al. 2022), and in three dimensions (Kuett et al. 2022). Technology advancements are now building toward four dimensions (Crainiciuc et al. 2022). Such approaches will reveal critical interactions between tumor and immune cells, which confer differential disease outcomes. Notably, the analysis of high-dimensional data obtained in these experiments requires the development of sophisticated algorithms that necessitate the integration of artificial intelligence and machine learning into biomedical research (Butler et al. 2018; Chen et al. 2018; Bao et al. 2021). On this front, the availability of a wide collection of large data sets of patient samples with clinical outcome information will be important for revealing the strongest predictors as well as validating molecular targets to effectively combat breast cancer. In addition to examining the TIME, more sophisticated assessment of corresponding systemic factors such as serum biomarkers can further help distinguish patient subsets with different risk profiles. Defining the key features of the local and systemic immune dysregulation in cancer can also help us classify the disease into differentially actionable groups, prioritize treatment, and aid clinical trial design.

Complementing clinical research initiatives, it remains important to study breast cancer in preclinical animal models where mechanistic experimentation is possible. Many of the key immune cells and mechanisms are conserved between mouse and human, as suggested by comparable antitumor immune responses during CTLA-4 blockade between the two species (Leach et al. 1996; Wolchok et al. 2017). This mechanistic conservation was further supported by a study comparing human and mouse immune responses during sepsis, although this study also identified some species-specific effects in inflammation (Godec et al. 2016). Thus, differences between mouse and human immune systems must be considered when interpreting findings from preclinical studies. For example, there may be value in incorporating more immunological diverse preclinical models in cancer research, such as outbred or wild mice, rather than relying on inbred pathogen-free models (Abolins et al. 2017). A study involving outbred mice that express human HER2 recently revealed that MHC-IB-dependent NK cell responses can counteract Treg functions (Wei et al. 2020). Alternatively, to develop clinically relevant research models, other groups have created immune-humanized mice in which tumors are established simultaneously with the human immune system in immunocompromised mice. Humanized mice can be created through multiple methods including injecting mature immune cells, or transferring hematopoietic stem cells alone or in combination with supportive cytokines and lymphoid organ fragments (Covassin et al. 2013; Katano et al. 2015; Jespersen et al. 2017; Scherer et al. 2021). While they can be overly reductionist, immune-humanized mouse models are valuable for studying specific cellular interactions and answering precisely defined research questions. Since immune-humanized mouse models offer a relatively tractable system for experimental manipulations, they can be useful for testing new therapies. Taken together, validating the findings from preclinical models will continue to be important for the next generation of discoveries in breast cancer immunology.

Finally, we would like to provide a few humble opinions on how to further increase the impact of breast cancer research. First, as almost all approved therapies have been developed as a result of investigating the basic mechanisms that underpin biology, we are confident that our ever-increasing collective knowledge will continue to be a driver of new advancements in the clinic. Notably, the most effective research initiatives are the ones combining sophisticated preclinical models with clinical research infrastructures to establish a translational research ecosystem. Thus, it is essential to foster connections between the bench and the bedside by establishing environments conducive to collaborations. Second, expanding biobanks with patient samples and data collected throughout the spectrum of clinical care will be essential for uncovering yet unknown mechanisms of immune dysfunction in cancer. However, as new approaches and technologies emerge, we may have to rethink how the patient tissues are acquired and stored for future analyses. Traditionally, pathological assessments and bulk transcriptomic/genomic analyses were performed on formalin-fixed, paraffin-embedded (FFPE) tumor samples. While paraffin embedding of fixed tissues can have advantages for histology and long-term storage, it may not allow interrogating the TIME in its native spatial configuration or at single-cell resolution. It is becoming increasingly clear that both the quality and quantity of the immune infiltrate can have a significant impact on cancer outcomes, and, thus, preserving the features of collected tissues will be important for future investigations. Third, a multiparameter fingerprinting of the host milieu may inform risk assessment and help devise personalized clinical trials. Often, patients are enrolled in clinical trials based on the expression of a specific immunotherapy target in the stroma or the tumor bed. While it is sensible to test an immunotherapy agent on patients with a clear expression of the specific molecular target such as PD-L1, many of these “good candidates” still fail to respond to the therapy. With future research, we may be able to shift from this suboptimal categorization approach to qualifying patients for treatment based on specific molecular and cellular fingerprints to achieve the best therapeutic outcomes. A remaining challenge in this context is the heterogeneity of cancer cells and the infiltrating immune cells, which makes it difficult to find parallels even among a large number of patients. Eventually, this can pose an obstacle for conducting clinical trials at high statistical power, perhaps suggesting new statistical methods might also be needed beyond what has been traditionally used.

Last but not least, clinical research would not be possible without the consent and participation of patients and their families. In this context, a critical responsibility befalls both basic scientists and physicians to educate the public about the importance of basic and translational research and clinical trials. To improve patient participation and compliance, in addition to increasing scientific literacy across all walks of life, we must also recognize preexisting biases and disparities in healthcare access. This means looking at patient populations from a holistic perspective and considering other social determinants of health, including socioeconomic status, racial background, social networks, and environmental factors to improve outcomes. Furthermore, to make the new findings applicable to a wide range of patients, research designs must be diversified to account for the effects of age, race, and gender. Analyzing samples from a diverse group of patients can help us develop more relevant biomarkers and discover new therapeutic targets, which would, in turn, improve overall outcomes by creating positive public feedback for future research initiatives. Taken together, we believe that advanced technologies in basic science and robust clinical research infrastructures will continue to reveal new facets in breast cancer immunology and lead to the development of next-generation therapeutics to improve outcomes in breast cancer.