In this study, we evaluated multiple immune cell fractions in both breast cancer tissues and matched blood samples, then observed that there was a partial correlation in the composition of immune cells at the tumor site and in the peripheral blood. In the comprehensive analysis of the association between each immune cell lineage fraction with hTIL and hPD-L1, we found for the first time that clinically diagnosed breast cancers, especially in cases with higher hTIL or positive hPD-L1 were already in a state in which innate immune responses were poorly involved or had escaped, and acquired immune responses were active. We also found that Non-B-cell antigen-presenting cell fractions were involved primarily in the PD-L1 pathway in breast cancer microenvironments.
To the best of our knowledge, most of the analyses of the immune cell compositions of breast cancer tissues using a multicolor FCM had 10 colors or less [32, 37], and that there were only two studies with more than 11 colors [38, 39]. Although the reactivities of the labeled antibodies were not always the same, and a direct comparison was not possible, there was a similar distribution of the leukocyte infiltrations in the tumor tissue in our study and a previous study with a distribution median of 218 CD45+ TIL/mg of tumor tissue (interquartile range: 85–445 CD45+ TIL/mg) . Although there are very few reports of systematic examinations of leukocyte compositions in breast cancer tissue, some studies have reported a ratio of total T being 86% (mean)  or 75% (median)  of the leukocytes (CD45+ cells) in breast cancer tissues, suggesting that T cells accounted for the majority of TILs . In our study, the proportion of total T cells in the leukocytes in the tumors was 57.3% (mean), which was slightly lower than the proportion in previous reports probably due to the difference in the antibody used and the gating strategy. But consistent with the fact that T cells account for the majority of leukocytes in the tumor tissues. In three previous studies, the proportions of CD19+ B cells in CD45+ TIL were found to be 8% (mean), 4.58% (median), and approximately 10% (mean), respectively [32, 37, 38], which was similar to the results of our study. With regard to other lineages, the findings of a previous study showed that CD14+/CD40+/CD163+ M2 macrophages were 0.06% (median), CD11b+/CD15+/HLA-DR- MDSCs were 1.19% (median), and CD56+ NK was 2.33% (median) . However, the number of studies were small, and the definitions of each lineage did not match those in our study, therefore valid comparisons could not be made. No comparable reports were found for the remaining lineages.
This is the first study to show an association between the immune cell composition of blood and that of breast cancer tissues. The immune cell composition of blood showed a partial correlation with the tumor tissues (Fig. 2) and the percentages of the immune cell fractions showed some differences between the tumor tissues and blood (Fig. 3). While this suggested that the composition of tumor-infiltrating immune cells may be estimated using blood samples, there were significant differences between them. Each also has a different function or clinical significance.
Although there was a significant difference between the subtypes, hTIL was shown to correlate with some clinicopathological factors including subtypes [3–5], prognoses, and responses to chemotherapy [4, 6, 7] in breast cancer. In our study, higher hTIL scores were associated with high-grade tumors, ER-negativity, higher Ki67 positive ratios, and hPD-L1 positivity (Table S6). Many studies have also reported that ER-positive breast cancer is the least immune-infiltrated subtype, which was consistent with our results [2, 5]. However, there are some controversies with regard to other clinicopathological factors, and results differ from study to study [2, 4, 5, 41]. To date, no studies have systematically assessed the relationship between hTIL and the immune cell fraction using FCM. In this study, we demonstrated that hTIL was associated with not only the degree of leukocyte infiltration in the tumor tissues but also the composition (Fig. 4). We found that while there were positive correlations between the hTIL scores and percentages of total T, CD4+ T, and CD8+ T (Fig. 4b–d), there were negative correlations between the hTIL scores and NK and NKT (Fig. 4k, m). Therefore, we speculate that a higher hTIL not only reflects the amount of immune cell infiltration, but also reflects the state in which acquired immunity is activated, relative to innate immunity in clinically diagnosed breast cancers. Furthermore, they may suggest that there is still room for the development of cancer immunotherapy to promote acquired immune responses in breast cancers other than triple negative breast cancer.
As mentioned previously, PD-L1 plays a significant role in the immune tolerance mechanisms that suppress T-cell activation [8, 9], and its expression is suggested to reflect ongoing (or active) immune responses or in addition to immunosuppression via the PD-1/PD-L1 pathway . The hPD-L1 was shown to correlate with some clinicopathological factors, including subtypes . It is also a clinically approved predictive marker for atezolizumab in triple-negative advanced breast cancer . In the present study, hPD-L1 positivity was associated with ER-negative diseases and higher hTIL scores but no significant association with the other factors, probably due to the small cohort size (Table S7). Although PD-L1 expression in multiple types of immune cells or tumor cells has been reported [8, 9], there is no consensus as to which immune cell fraction is responsible for the substantial function of the PD-L1 pathway in breast cancer. The findings of only one report that evaluated the PD-L1 expression in CD4 + T and CD8 + T, B showed that the overall proportion of the PD-L1 positive TILs was very low and could only be detected in a small number of tumors . In the present study, we found that a substantial proportion of PD-L1 positive immune cells were non-B-cell antigen-presenting cell fractions such as Mo/Mφ, CD16+ Mo, MDSC, DC, and mDC fractions (Fig. 5a), and that the PD-L1 positive ratios were significantly higher in tumor tissues than in blood (Fig. 5b–l), suggesting that these fractions were involved primarily in the PD-L1 pathway in breast cancer tissue. In addition, we found that hPD-L1 positive tumors exhibited increased leukocyte infiltration in tumor tissues (Fig. 6a), and that the hPD-L1 reflected PD-L1 expression in Mo/Mφ, CD16+ Mo, DC, and mDCs (Fig. 7a–k). These results suggested that hPD-L1 expression indicates the activation status of the immune tolerance mechanism that occurs in non-B-cell antigen-presenting cells in response to an increased immune cell infiltration, mainly effector cells which secrets IFN-gamma to induce PD-L1 expression on various cells, into the breast cancer microenvironment.
The FCM analysis findings will be useful in the exploration of new immune-related factors in breast cancer. Briefly, by evaluating the expression of candidate proteins related to tumor immunity by IHC and analyzing them together with these data, function of candidate proteins may be verified. Currently, we are focusing on some candidate proteins as immuno-regulatory factors in breast cancer and further analyzing FCM data of this study to validate their immunological functions.
This study had several limitations. A relatively small number of patients were enrolled in this study. In addition, in a pilot study, it was found empirically that the number of cells required for FCM was not sufficient in cases of ER-positive breast cancer, especially in cases with lower Ki67s. In addition, cases of small tumor sizes and post-NAC with pathological CRs were excluded due to technical problems in the collection of the tumor tissues. Therefore, it should be noted that there was an inevitable bias in the enrollment of the cases; it differed from the general breast cancer cohort in terms of larger invasive tumor sizes, more ER-negative cases, and higher Ki67 cases (Table S2). Although, as mentioned above, the significance of the TIL is suggested to be different between subtypes, a subgroup analysis could not be performed because of the small sample size. It is recommended that in future studies, more samples be collected and more detailed analyses be performed.