Relevance of tumor-infiltrating lymphocytes in breast cancer

Chemotherapy and radiotherapy are frontline management options for breast cancer and the underlying immunogenic component of these agents make them attractive candidates for neoadjuvant therapy. Preclinical studies of chemotherapy and radiotherapy have revealed the unexpected ability of the immune system to contribute to the success of treatment. There is an abundance of experimental and, more recently, clinical evidence [37‐41] showing that chemotherapies are more efficient if they successfully re-activate immune surveillance through the elimination of immunosuppressive cells or through the promotion of danger signals released by the death of tumor cells, hence triggering a long-term immune response against residual tumor cells [42, 43]. Several findings have delineated a sequential series of events driving immunogenic tumor cell death that results in the activation of innate and adaptive immune responses [44]. During immunogenic cell death, dying tumor cells release danger signals such as adenosine triphosphate (ATP) and high-mobility group protein B1 (HMGB1), which act to prime an antitumor immune response [45, 46]. In landmark studies by Zitvogel and colleagues, anthracycline therapy was less effective in mice deficient in P2X7 receptors, NLRP3, and caspase-1, while oxaliplatin was dependent on HMGB1-mediated activation of toll-like receptors (TLRs). The activation of antigen-presenting cells such as DCs by ATP (P2X7) or HMGB1 (TLR4) leads to enhanced production of IL-1β, and consequently activation of both innate and adaptive immune responses [23]. Interestingly, the observation of enhanced antitumor immunity following treatment extends to various chemotherapies, including gemcitabine [47‐49], cyclophosphamide [50, 51], paclitaxel [52], and doxorubicin [53‐55]. Induction of tumor cell death by chemotherapy (e.g., gemcitabine) also enhances tumor antigen cross-presentation [56]. Furthermore, certain chemotherapies have direct effects on immune cells themselves. For example, low dose cyclophosphamide selectively reduces the number of immunosuppressive Treg cells while sparing immune effector cells [51]. These findings are summarized in Table 1. Many preclinical studies have described immune-enhancing activity in response to chemotherapies [57]. However, it is evident that not all chemotherapies induce immunogenic cell death. As such, it is apparent that combinations with specific immune-inducing or immune-targeted inhibitors are necessary to promote tumor regression and immunogenic cell death in these cases. Thus, despite initial theories suggesting that breast cancer is not an immunogenic disease, recent studies have confirmed and consolidated understanding of the underlying immunological component in breast cancer, thus revealing promising targets and novel therapies for treatment of this disease.

The importance of the immune system in breast cancer is increasingly being recognized, owing to the observation by several groups that the presence of TILs is a prognostic indicator for higher rates of pathological complete responses (pCRs) to neoadjuvant chemotherapy (NAC) [37, 58‐60]. Notably, Denkert and colleagues [41] were able to first show in a large-scale analysis of 1058 patients’ biopsies that TIL⁺ tumors achieved a pCR rate of 40–42 % following NAC, whereas TIL⁻ tumors achieved a pCR of only 3–7 %. This distinction was based upon comparing lymphocyte predominant breast cancer (LPBC; defined as ≥60 % TILs in either the intratumoral or stromal compartment) to the rest of the population. Interestingly, this analysis revealed that TILs within either the stromal or intratumoral compartment was of value as a prognostic indicator. Subsequently, this has also been shown in another dataset of 580 triple negative breast cancer (TNBC) and HER2-positive breast cancers (59.9 % in LPBC vs 33.8 % in non-LPBC) [41] as well as in a meta-analysis of 996 patients where an immune gene module indicative of a Th1 response was prognostic of pCR in all cancer subtypes [61]. The protective role of a Th1 response was also reported by an independent group who found that the infiltration of T-bet⁺ lymphocytes correlated with a favorable prognosis [62]. TILs have been reported to predict pCR in a prospective study of HER2-negative breast cancers in patients treated with NAC, confirming the prognostic value of TILs in the response to chemotherapy [63].

From a clinical perspective, the observation suggests that higher levels of TILs are associated with higher responses to chemotherapy, seemingly independent of oestrogen receptor (ER), progesterone receptor (PR), and HER2. There is in fact no biologically appropriate TILs cut-point as all studies have shown that the TIL marker is prognostic on a continuous scale: each 1 % or decile increment is associated with a further increase in the rate of pCR. The biological reasons for this observation of chemosensitivity remain poorly understood. Perhaps pre-existing immune antitumor responses are better placed to be able to clear tumor cells after chemotherapy has depleted local immunosuppression or Treg cells. An increase in TILs in the breast cancer post-NAC is also associated with improved outcomes [64]. Notably, however, although TILs correlate with pCR to NAC in all breast cancer subtypes, the correlation between TILs at diagnosis and disease-free or overall survival was only significant in TNBC and HER2-positive breast cancers, although the reasons for this are not fully understood [39, 64‐66]. The studies indicating a positive relationship between TILs and responses to NAC are summarized in Table 2. TILs are important in the efficacy of trastuzumab [65, 67, 68] and, moreover, trastuzumab treatment results in the activation or recruitment of multiple immune cell lineages and increases the susceptibility of tumor cells to antibody dependent cytotoxicity (ADCC) [67, 69] (reviewed by [70]). A study looking at TILs at baseline and correlation with both pCR and disease-free endpoints highlights that high levels of TILs at diagnosis bodes for an improved outcome regardless of attainment of pCR [68]. Those with low levels of TILs and residual tumor at surgery post-NAC and anti-HER2 therapy had the worst outcomes of all, suggesting that it is this group that needs more effective antitumor and immune-enhancing strategies. The fact that TILs are highly associated with responses to chemotherapy and trastuzumab suggests that the presence of TILs, both pre-NAC and post-NAC, should be taken into account for patient treatment decisions—a possible schema for this is shown in Fig. 2. The potential synergistic immune effects between chemotherapy and immunotherapies also suggest that combination approaches in the neoadjuvant setting, prior to surgery, may also be beneficial.

Recently, efforts have been made to sub-divide the immune infiltrate into lineage subsets to determine the prognostic value of each immune cell type. The presence of CD8⁺ cells in the tumor infiltrate prior to the onset of NAC predicted pCR in several studies [41, 71‐75]. The presence of Tregs prior to NAC has also been shown to be a prognostic indicator of pCR [71‐73, 76]. Although this seems somewhat counterintuitive given the suppressive role of Tregs, it is probably because FOXP3⁺ infiltrate is also significantly associated with CD8⁺ infiltrate. More informative is the observation that the presence of Tregs following NAC has a significant negative correlation with pCR and disease-free or overall survival [77‐79]. Indeed, the ratio of CD8 to Tregs following NAC has been shown to be a strong predictor of clinical responses [78]. The increased CD8 to Treg ratio likely facilitates the acquisition of CD8⁺ T-cell effector functions such as Granzyme B expression, which is elevated post-NAC [78, 80]. In this study they found that the proportion of CD8⁺ lymphocytes remained stable pre-NAC and post-NAC whereas Tregs were significantly reduced in the post-NAC samples. Limited work has been done on the prognostic significance of other immune cell subtypes although the presence of CD4⁺ [72] and CD20⁺ [74] lymphocytes pre-NAC is also associated with pCR. Another study found that the presence of T follicular helper (T_FH) cells, which function to attract and promote the formation of memory B cells, was associated with improved responses to chemotherapy in breast cancer. These T_FH cells localized to peritumoral tertiary lymphoid structures, indicating there may be localized orchestration of antitumor immune responses in certain breast cancers [81]. The link between TIL infiltrate and patient outcome has led to heightened interest in utilizing immune-modulating strategies for patients with breast cancer. Other immune subsets of interest include γδ-T cells and Th17 cells. γδ-T cells show positive correlations with CD4⁺ FOXP3⁺ T cells, and are associated with poor outcomes and advanced tumor stage, as well as low disease-free and overall survival in patients with breast cancer [82]. However, further work is needed given the apparent antitumor role of γδ-T cells in some settings [83, 84]. In a small study of 30 patients, Th17 cells negatively correlated with disease stage [85] but further work is needed to evaluate their effect on immune responses to NAC [86]. Accordingly, further work is needed to determine whether these subsets are potential targets for immunotherapy.

The strong prognostic significance of TILs in breast cancer opens up important questions for patient management. In particular, it is likely that patients with high TILs post-NAC can benefit from strategies designed to enhance the immune response against the tumor. As discussed above, NAC has the potential to increase the CD8⁺ to Treg ratio, which correlates with the likelihood of pCR [87, 78]. Interestingly, certain chemotherapies, including cyclophosphamide [42, 51, 88], used to treat patients in the cohort studied by Ladoire et al., have been shown to specifically reduce the number of Tregs in preclinical models. Thus, it is worth considering the immunological consequences of any NAC regime. For example, concurrent depletion of Tregs through the use of immunotherapies such as anti-CTLA-4 mAb [89] may enhance the efficacy of NAC in inducing antitumor immune responses. There may also be clinical benefit in depleting other immunosuppressive populations such as MDSCs or tumor-associated macrophages [16]. A histopathological examination of breast cancer tissue pre-NAC and post-NAC by DeNardo et al. revealed that the presence of CD68⁺ macrophages inversely correlated with CD8⁺ TIL following NAC. Furthermore, depletion of this subset enhanced chemotherapy in preclinical models of breast cancer [90]. A recent study showed that a monocyte/DC metagene analysis held similar predictive value to NAC as the T cell/NK cell module [91]. Therefore, further work evaluating this in humans is warranted although the heterogeneity of this population makes characterization more complex than the T-cell infiltrate.

Although NAC is capable of inducing CD8⁺ infiltrate within the tumor microenvironment, the efficacy of the antitumor immune responses is likely limited by the presence of immunosuppressive networks. For example, a recent retrospective analysis of 94 post-NAC biopsies demonstrated that PD-L1 expression correlates with TIL infiltrate and pCR [40], which is in agreement with studies from other groups evaluating PD-L1 mRNA expression [41, 92]. This is likely because PD-L1 is expressed following IFNγ stimulation and so is a surrogate marker of an immune response. It should be noted, however, that other groups have reported that PD-1 and PD-L1 expression are correlated with poor prognosis in patients with breast cancer [93‐95]. Nevertheless, these studies consistently observed PD-L1 and PD-1 expression in a subset of breast tumors; because PD-L1 expression has previously been shown to be a marker for the efficacy of anti-PD-1 in other tumor types [96], it is possible that PD-L1 expression may be a predictive marker for patients likely to have good responses to a combination of NAC and anti-PD-1. It was also recently reported that expression of B7-H4, another member of the B7 family, is a marker of good prognosis in patients with breast cancer [97]. Further work is need to evaluate the potential to modulate B7-H4 because it has been reported to inhibit T-cell function [98, 99] and yet B7-H4 expression was reported to limit tumor growth in this study [97]. Immune checkpoint blockade with anti-PD-1 [100] and anti-CTLA-4 [101] has already been shown to enhance the efficacy of chemotherapy and radiotherapy [102] in preclinical models and phase I clinical trials [103], and so there is clear rationale to further evaluate these combinations in patients. Further work will investigate the potential of combining NAC with other immune modulators in breast cancer given that that these immunosuppressive pathways appear upregulated in TIL⁺ breast cancers [41].

The success of checkpoint inhibitors is highly dependent on the extent of a pre-existing immune response against the tumor, and the ability of immune-targeted therapies to re-stimulate/activate immune subsets. Because conventional therapies such as chemotherapy and radiotherapy result in the release of tumor-associated antigens (increased antigenicity) from the apoptotic tumor cells, the increased immunogenicity provides a promising target for immunotherapies [103]. These tumor-associated antigens, released from tumor cells after chemotherapy, are taken up by antigen-presenting cells or DCs and used to stimulate downstream effector T cells capable of recognizing and lysing tumor cells [103]. Accordingly, it is evident that subsequent immunotherapy is likely to be a beneficial endeavor, provided that the initial conventional therapy used is able to effectively “prime” tumor cells to express and present foreign tumor antigens. A study conducted by Twyman-Saint Victor et al. demonstrated that radiation is able to enhance the diversity of the T-cell receptor repertoire of intratumoral T cells. This process was found to underpin the synergy of radiotherapy and anti-PD-1/anti-CLTA-4 in patients with melanoma [103]. Notably, high-dose radiotherapy and high-dose chemotherapy can ablate immune function. Therefore, careful consideration must be given to the dosage and the timing of these therapies when combined with immunotherapy.

While immunotherapy is an attractive adjuvant therapy for patients presenting with high TIL pre-NAC or post-NAC, an alternative approach is required for the treatment of patients with poorly immunogenic cancer types and low-level TILs at diagnosis. Liu et al. recently used preclinical models of melanoma to investigate the immunogenic function of the MAPK pathway inhibitors trametinib and dabrafenib in combination with immunomodulatory antibodies, including PD-L1/PD-1 and CTLA-4. They reported the potential for synergy between these therapies [104]. The combination of trametinib with anti-PD-1 increased tumor-infiltrating CD8⁺ T cells in CT26 mouse colon carcinoma tumors, as well as downregulating immunosuppressive factors, upregulating HLA molecules, and increasing immune responses in the tumors yes [104]. Dabrafenib had no suppressive action on the function of CD4⁺ and CD8⁺ T cells, whereas trametinib had mild partial or transient inhibitory effects on T-cell proliferation fine [104]. This phenomenon is supported by findings of Callahan et al. demonstrating that RAF inhibitors in BRAF wild-type tumors caused hyperactivation of ERK signaling, thereby enhancing T-cell activation and signaling [105]. While targeted blockade of checkpoint inhibitors has proven to be effective, recent evidence suggests that agonists may also be crucial in TIL stimulation. CD8⁺ T-cell responses require T-cell receptor activation in addition to co-stimulation provided through ligation of tumor necrosis factor receptor family members 4-1BB (CD137) and OX40 (CD134). Studies using the OX40 agonist in combination with either anti-4-1BB, anti-PD-L1, anti-CTLA-4, and immunization in sarcoma, melanoma, hepatoma, and breast cancer models have shown significant survival benefit by boosting the T-cell response [106‐109]. One hurdle to the application of immunotherapy in breast cancer is the apparent low immunogenicity of breast cancers, perhaps owing to the low mutation rate of breast cancer when compared to cancers such as melanoma, which show a high response rate to immunotherapy [110‐112]. One approach is to use targeted inhibitors designed to induce immunogenicity in tumors. For example, Kim et al. proposed the notion of epigenetic modulation, suggesting that MHC I-related genes are downregulated through epigenetic silencing of tumor cells [16]. They investigated this by combining anti-PD-1 and anti-CTLA-4 antibodies with 5-azacytidine (a DNA methyltransferase inhibitor) and entinostat (a class I histone deacetylase inhibitor). This study revealed complete eradication of tumors in murine 4 T1 (breast) and CT26 (colorectal) models [16]. Similarly, studies investigating DC vaccination in melanoma have shown that combination therapy with PI3K inhibitors has significant pro-inflammatory effects via TLR ligands that support antitumor immunity [113]. The results of these studies suggest that combining targeted inhibitors with immunotherapy may be a promising option for clinical treatment of breast cancer for patients with high-level TILs. However, alternative therapeutic options are still required for patients with low-level TILs. Because patients with high TILs show increased response rates to NAC, one approach is to use strategies that enhance T-cell trafficking to the tumor site. Denkert et al. reported a correlation between CXCL9/CXCL13 expression and the number of TILs in patients with breast cancer [59]. Therefore, treatments that enhance the expression of these chemokines may enhance the efficacy of NAC. Alternatively, adoptive cell therapy (ACT) utilizing TILs or chimeric antigen receptor T cells has emerged as a promising regimen for the treatment of certain cancers, including melanoma [114‐118], with recent reports of long-term remission in some patients [119]. In a study conducted by Koya et al., combined treatment of vemurafenib plus ACT of lymphocytes genetically modified with a T-cell receptor recognizing ovalbumin resulted in superior antitumor responses [120]. While absolute numbers of T cells infiltrating the tumor did not increase with vemurafenib treatment, they found that the combination with ACT increased the functionality of antigen-specific T lymphocytes [120]. The findings of these studies suggest that BRAF inhibitors may be effective in combination with ACT; however, evaluation of ACT in combination with other inhibitors is also necessary. The utility of this approach in breast cancer is yet to be determined. One possibility is the targeting of the HER2 antigen [121, 122]. Ishikawa et al. [123] investigated the effects of ACT in addition to anti-CTLA-4, and were able to show that the combination enhanced antitumor activity. Taken together, there is a large body of evidence demonstrating the potential utility of TIL-associated immunotherapies for breast cancer in the future.