It has been well established that the immune system plays a crucial role in host defense against tumor progression. Unlike solid tumor types such as melanoma, breast cancer is not usually considered an immunogenic; however, recent evidence suggests that tumor immunogenicity is important in the biology of breast cancer and its response to treatment. Tumor-infiltrating lymphocytes (TILs) have provided insight into the immunogenicity of breast tumors and have been shown to be useful as a prognostic biomarker in early-stage triple-negative and HER2-positive breast cancer.

Cancer immunoediting is believed to be a central component in the evolutionary processes that shape an established tumor, and therefore also plays a critical role in the determination of the resulting immunogenic phenotype (1, 2). Immunoediting comprises three processes, namely, elimination, equilibrium, and escape. Elimination encompasses the concept of immunosurveillance, where the host immune system recognizes tumor antigens and destroys the developing tumor. It is characterized by the presence of CD8+ T lymphocytes, CD4+ type 1 helper T lymphocytes, and natural killer cells. When this elimination process is not successful, the tumor enters the equilibrium phase, whereby the tumor cell clones with superior survival abilities are selected in a Darwinian fashion. Escape occurs when the immunologically sculpted tumor proliferates in an uncontrolled manner and becomes clinically detectable. During escape, FOXP3+ T regulatory lymphocytes, CD4+ type 2 T helper lymphocytes, and myeloid-derived suppressor cells can be seen in the tumor microenvironment (3).

Tumor-infiltrating lymphocytes are immune cells that have been observed in many solid tumors, including breast cancer. They are a population of cells comprising a mixture of cytotoxic T cells and helper T cells, as well as B cells, macrophages, natural killer cells, and dendritic cells. Whilst not discussed extensively in this manuscript, gene expression and flow cytometry profiles demonstrate that TIL quantity, as assessed semiquantitatively on a hematoxylin and eosin (H&E) stained slide, represents a surrogate for a pre-existing favourable host anti-tumor activated T cell response (4, 5).

Tumor-infiltrating lymphocytes are more commonly observed at higher levels in triple-negative and HER2-positive tumors compared with estrogen receptor positive, HER2-negative breast cancer (6–8). This indicates that TNBC and HER2-positive subtypes are more immunogenic, and for reasons yet to be fully elucidated are able to generate more robust T cell anticancer responses. Lately the clinical importance of the presence of TILs has been reported, with associations with improved prognosis and higher response rates to neoadjuvant therapy in the early-stage setting (9, 10).

Biomarkers in cancer must undergo thorough evaluation in terms of analytical validity, clinical validity, and clinical utility before they can be adopted as part of clinical practice (11). Analytical validity refers to the robustness of the test, in terms of accuracy, reliability, and interobserver reproducibility (12). The clinical validity relates to the extent to which the biomarker is associated with prognosis or response to treatment and refers to the ability of a tumor biomarker to divide one population into groups that differ either biologically or clinically. Importantly, clinical validity alone is insufficient to recommend that the biomarker be used to guide clinical treatment decisions. Clinical utility is the usefulness of the biomarker in terms of directing clinical decisions and patient management.

As the interest in TILs as an immunological biomarker grows, so too does the need for standardizing TIL evaluation. The International Immuno-Oncology Biomarker Working Group has developed a practice guideline and tutorial on how to score TILs in breast cancer (13). Currently, most clinical research groups world wide interested in TIL evaluation in breast cancer are using this guideline in their studies, which will promote comparability of studies. This will avoid a plethora of different methods assessing the same biomarker, as has been the experience with Ki-67, although for Ki-67 international efforts at standardization are also underway (14).

Tumor-infiltrating lymphocytes in breast cancer lesions can be seen in both the intratumoral and stromal areas. Intratumoral TILs have cell-to-cell contact with carcinoma cells, while stromal TILs are located in the fibrous stroma adjacent to the tumor cells. Evaluation of intratumoral TILs is generally more difficult and less reproducible than evaluation of stromal TILs but the separation of these two variables may have biological connotations. The current recommendation is therefore to use the percentage of stromal TILs as the principal parameter, where the numerator is the area occupied by TILs and the denominator is the total stromal area, not the number of stromal cells. All mononuclear cells (lymphocytes and plasma cells) are scored, but polymorphonuclear white blood cells are excluded (13).

The guidelines of the International Immuno-Oncology Working Group clearly state that TILs should be assessed as a continuous variable rather than as a categorical variable (13). Having TILs documented as a continuous variable enables it to be implemented in algorithms encompassing other prognostic clinical–pathological variables. Loi and colleagues have demonstrated that TILs incorporated as such in an algorithm containing nodal status, tumor size, and age enables making predictions of the 5-year predicted distant disease-free survival (15) (Figure 1). Where a binary variable is needed, the term lymphocyte predominant breast cancer (LPBC) can be used to describe cancers which comprise more lymphocytes than tumor cells. Previous reports of LPBC have variably utilized a threshold between 50 and 60% stromal lymphocytes; however data since the original publications suggest that this may not be the best cutpoint as this level of immune infiltration represents less than 10% of primary breast cancers. This is considerably lower than that of melanoma and non-small cell lung cancer in which this level of lymphocytic infiltration is present in around 25% of tumors.

Tumor-infiltrating lymphocytes can be easily and are most commonly assessed using light microscopy of H&E stained slides. More complex techniques have been used, such as flow cytometry (16) and multicolor immunohistochemical staining with multispectral imaging (17), as well as gene expression analysis measured using quantitative polymerase chain reactions, gene expression microarrays, and RNA sequencing (18, 19). These techniques can provide more information regarding the different lymphocyte subpopulations; however, gene expression data are limited with regard to identifying specific T cell subsets. All of these techniques remain in the research setting due to cost, lack of robust analytical validity, and requirements for certain types of tissue processing.

The International Immuno-Oncology Working Group conducted two ring studies to evaluate interobserver variability of TIL assessment (20). The aim was to calculate the interclass correlation coefficient (ICC) for assessment of TILs by different pathologists. In the first ring study, 34 pathologists assessed 60 slides for stromal TILs. While the interobserver agreement was relatively good, the prespecified endpoint of an ICC above 0.7 was not reached (ICC: 0.70; 95% CI: 0.62–0.78). A software-guided image evaluation system was developed for the second ring study, where 28 pathologists evaluated 60 slides. The ICC improved to 0.89 (95% CI: 0.85–0.92). This suggests that TIL evaluation can be reproducible among pathologists and that computer-assisted systems may be valuable in this regard. A study by Swisher et al. (21) also found that TIL evaluation can be undertaken with acceptable interobserver agreement. Four pathologists assessed 75 TNBC specimens. The kappa statistic for interobserver agreement was 0.57, which was interpreted as moderate agreement.

Nevertheless, some hurdles still remain with regard to standardizing TIL assessment. For example, TIL infiltration patterns within a single tumor can be quite heterogeneous (Figure 2). It is well known that the eye of the pathologist may not be able to quantify reliably the spatial variation in infiltration patterns. In this respect, machine learning algorithms may potentially be of added value (22). Regardless, TIL evaluation by H&E remains the most pragmatic for many community laboratories in the absence of data supporting other methods.

As the majority of experience in TIL assessment in breast cancer has been in primary lesions, less is known about immune influences in the advanced metastatic setting. Important differences in the immune infiltrates between primary tumors and metastases have recently been reported. Studies of paired primary and metastatic lesions have noted significantly lower levels of TILs in metastases (23–25) (Figure 3). This is in keeping with lower TIL levels described in metastases in lung cancer (26) and renal cell carcinoma (27). The TILs working group guideline (13) methodology for TIL assessment has been used to evaluate TILs in the advanced setting in a retrospective analysis of the CLEOPATRA study (28). This study also found lower TIL levels in metastatic lesions than in their paired primaries, although there were not enough paired samples to adequately power this comparison. The site of metastases may also have an important influence on the immune infiltrate, with higher TIL levels having been observed in lung metastases than in bone, liver, or skin lesions. The International Immuno-Oncology Working Group will soon publish guidelines on scoring TILs in the metastatic setting and in the residual disease setting. Overall, these observations lend evidence to the hypothesis that breast cancer must adapt and learn how to evade the immune system as well as cause significant immunosuppression in the host in order to progress to the advanced disease setting. This has implications for the development of immunotherapies in advanced breast cancer.

Tumor-infiltrating lymphocytes have potential clinical utility in breast cancer in a number of different domains. Since TILs are read on simple H&E stained slides, it is not surprising that there is a tremendous interest to investigate their importance for the clinical setting.

We see the clinical utility of TILs in the following ways:

Schmid et al. (48) recently reported the results of a phase 1a study of atezolizumab monotherapy in 115 metastatic TNBC patients. Higher overall response rates and longer overall survival were observed in patients with higher TIL levels. In the group with TILs ≤ 10%, median OS was 6.6 months, compared with 12.6 months for those with TILs > 10%. Of note, TILs here were evaluated differently and not as in the current guidelines. TILs were defined as immune infiltrate as a percentage of the tumor area. While further refinement of immune markers is likely, TILs evaluated on H&E slides could represent an easy first step for many laboratories to determine which patients may benefit from checkpoint blockade as a single agent. Data in this area will undoubtedly increase in the next few years.

Tumor-infiltrating lymphocytes as a robust prognostic factor may potentially lead to the identification of those patients with excellent prognosis and thus may guide adjuvant treatment de-escalation efforts. For example, in HER2+ patients with high-TIL levels, it may be appropriate to consider using only a taxane and trastuzumab rather than maximal therapy with combination anthracycline, taxane, trastuzumab, and pertuzumab (10). In the pooled analysis of 991 primary TNBC patients, those patients with node negative disease and TILs 20% or more (which is around the average level) had excellent 5-year distant metastases-free survival (92% CI: 87–97%) (Figure 1). TILs may have potential utility in the design of clinical trials, by appropriately grouping patients into those who may benefit from standard treatments, and those who require novel therapeutic strategies (49).

The neoadjuvant model represents an attractive model for drug development (50). What remains to be clearly elucidated is how the change in pCR induced by new agents will correlate with improvements in disease-free survival in the adjuvant setting (51, 52). Given that TILS are prognostic in the residual disease setting, we hypothesize that incorporation of TILS into the pCR/residual disease endpoint may improve prediction of those who do well, and help us understand how a novel agent performs. Incorporation into existing models such as residual cancer burden (53) may improve these surrogate endpoints as not all patients who do not achieve pCR relapse and identification of these patients that most benefit from the new agent will aid adjuvant trial design and statistical calculations.

A lack of the presence of antitumor immunity bodes for a poor prognosis and poor response to therapy. Therefore, a key area of interest is how to modulate the tumor microenvironment in order to convert a “low-TIL” tumor to a “high-TIL” tumor. Therapeutics could stimulate antigenicity and allow the augmentation of the antitumor immune response with checkpoint inhibitors. Potential approaches include multimodal combinations with chemotherapy, targeted therapies, radiotherapy, personalized vaccines, and adoptive T cell therapies (49).

The combination of conventional chemotherapeutic agents and immunotherapies is currently being investigated. A phase III trial of nab-paclitaxel combined with the anti-PD-L1 inhibitor atezolizumab in advanced TNBC is underway (ClinicalTrials.gov.NCT02425891), as is a phase III study of pembrolizumab plus chemotherapy (ClinicalTrials.gov NCT NCT02819518).

Ras–MAPK pathway activation has been described as a mechanism of promoting immune evasion in TNBC (43). Genetic alterations in Ras–MAPK signaling have been significantly associated with lower TIL levels. MEK inhibition can upregulate major histocompatibility complex and PD-L1 expression in TNBC cells both in vivo and in vitro. Thus, this combination is a very attractive hypothesis to investigate in patients. The combination of the MEK inhibitor cobimetinib, taxane chemotherapy, and atezolizumab is currently being investigated in advanced TNBC patients (ClinicalTrials.gov NCT02322814).

The efficacy of trastuzumab in HER2-positive breast cancer has traditionally been thought to be due to reductions in signaling downstream of the HER2 receptor. However, it is now understood that antibody-dependent cell-mediated cytotoxicity as well as T cell-mediated cytotoxicity play a central role in its action (54–59). Studies of immunotherapies combined with anti-HER2 treatments are currently recruiting and the results are eagerly awaited (ClinicalTrials.gov NCT02318901, NCT02605915, NCT02924883).

Tumor-infiltrating lymphocytes, as evaluated on H&E slides, has been shown in numerous studies now to be a reliable and reproducible marker of pre-existing antitumor immunity in breast cancer. It is clear that higher levels of TILs are associated with improved prognosis in early and advanced stage TNBC and HER2-positive breast cancer, as well as a higher probability of achieving pCR in the neoadjuvant setting. Analysis of TILs in residual disease specimens after neoadjuvant therapy has also been shown to have prognostic value. The evaluation of TILs as a biomarker in breast cancer is likely to be extended from the research domain to the clinical setting in the future.

As the body of evidence of the clinical validity of TILs grows, so too does the appreciation of the importance of the immune system in the biology and outcome of breast cancers. There remain considerable challenges in optimizing host antitumor immunity, and combination of checkpoint inhibitors with traditional and novel therapies is an area of active investigation.

Preparation/drafting of manuscript: LW, PS, SJL, BV, RS, and SL; figures: LW, RS, and SL.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.