The disease modulating properties of tumor-infiltrating lymphocytes (TILs) and other inflammatory cells (TIICs) continue to be of great research interest in breast cancer, both on a mechanistic level and as prognostic biomarkers [25, 26]. Although triple-negative breast cancer (TNBC) is more frequently addressed in this context, we have already been able to demonstrate a surprisingly strong prognostic relevance of various TIICs in the cohort of patients with early luminal breast cancer investigated here [12, 14-16]. In the present study we extended this analysis with the addition of CD66b+ tumor associated neutrophil granulocytes (TANs).
TANs, similar to TAMs, can exhibit different polarization states and thus exert both pro- and anti-tumor effects in the tumor microenvironment [10]. In multiple solid cancers, a high density of TANs has been described as a negative prognostic factor [11]. Data on their prognostic relevance in breast cancer is mixed and further clarification is needed: Boissiere-Michot et al. reported no prognostic relevance of CD66b+ TANs in a mixed subtype as well as a triple-negative cohort [27, 28]. Geng et al. showed that in tumors treated with neoadjuvant chemotherapy, an increase in CD66b+ cells was prognostically unfavorable, but there were no conclusive results regarding pretherapeutic TAN density [29]. Wang et al. showed a negative prognostic effect, especially in TNBC, of parenchymal CD66b+ TANs, only [30]. Two other studies also found a correlation between CD66b+ TAN density and worsened prognosis in cohorts with mixed subtypes [30, 31].
A consistent finding of the above-mentioned studies was that the various molecular subtypes of breast cancer displayed different degrees of TAN infiltration, and that infiltration was most pronounced in TNBC. This highlights that an evaluation of prognostic relevance as well as the establishment of cut-off values for prognostic groups should be performed separately according to subtypes. To our knowledge, this is the first study to evaluate the prognostic relevance of CD66b+ TANs exclusively in the early luminal subtype, which is the most frequently treated subtype and stage.
The results of our work suggest that increased CD66b+ cell density both in the stromal and intraepithelial compartment of central tumor tissue from early luminal breast cancers is a negative predictive factor associated with tumor progression. This is consistent with the above-mentioned results of studies in other subtypes of breast cancer [27, 28, 30, 31]. In addition, we also made intriguing novel observations in tissue beyond the primary tumor: Although there was limited availability of normal tissue samples in this study, a negative prognostic relevance of CD66b+ cells was also observed in these samples. And even in surgically removed lymph nodes, which were free from metastasis, we could identify a small group of patients with strongly increased CD66b+ cell density and significantly elevated long-term risk for recurrence or distant metastasis. Correlation of cell densities with clinical parameters also revealed that neutrophil infiltration in lymph nodes correlated with more aggressive clinical parameters. Thus, in the luminal subtype of breast carcinoma, increased infiltration with CD66b+ neutrophils in the primary tumor itself, as well as in adjacent normal tissue and lymph nodes, was associated with worse disease outcome. This suggests that the ability of breast cancers to recruit neutrophil granulocytes and to influence their behavior may extend far beyond the classical tumor microenvironment.
The mechanisms by which TANs contribute to cancer progression have been a subject of ongoing oncoimmunological interest in recent years. They include immunosuppression as well as promotion of tumor growth, angiogenesis and distant metastasis, which is mediated via a CD90-TIMP-1 loop and the G-CSF-RLN2-MMP-9 axis in breast cancer [9, 30, 31]. Moreover, estrogen alters the activity and gene expression of neutrophils to contribute to tumor formation and growth [32, 33]. Furthermore, TANs might also play a role in resistance to radiotherapy, which is a key element of breast-conserving treatment strategies in breast cancer [34].
We were particularly interested in the aspect of immunosuppression and possible interactions with other inflammatory cells, which we already quantified in previous studies. We found the most consistent intratumoral correlation between M2-like TAM density and CD66b+ TAN density. This is in contrast to the results of Boissiere-Michot et al. who could not report any association between the density of TAMs and TANs in their work [27]. However, in their study, the pan-macrophage marker CD68 was applied, whereas here CD163 was used to specifically identify immunosuppressive M2-like TAMs. Thus, there may be a dependency or mutual recruitment between M2-like TAMs and TANs, a process which has also been described in literature [35]. In hepatocellular cancer, Zhou et al. reported the recruitment of tumor-associated macrophages by TANs as a mechanism of tumor progression [36]. Tumor-induced G-CSF release, which as described by Sheng et al. promotes metastasis in TANs via the G-CSF-RLN2-MMP-9 axis, also affects the polarization of TAMs to the immunosuppressive M2-like phenotype [31].
One of the main observations of our previous work on TAMs was that in early luminal breast cancer, a combined analysis of cytotoxic M1-like and immunosuppressive M2-like TAMs, i.e. the overall polarization status of the macrophage population, had the most pronounced association with prognosis [15]. A particularly unfavorable constellation of macrophage polarization was the combination of high M2-like density and low M1-like density (M2-shifted) or, in other words, a low M1-like/M2-like ratio. This was contrasted by patients with tumors with a higher M1/M2 ratio and almost ideal response to therapy. Interestingly, dividing the cohort into two groups based on M1/M2 ratio in our present study revealed that within the group with a low M1/M2 ratio, CD66b+ TAN density was no longer prognostically relevant. In the remaining tumors with higher M1/M2 ratios, CD66b+ TANs were still significantly associated with reduced disease-free survival. This may reflect alternative primary mechanisms of immunosuppression: While some tumors succeed in recruiting and repolarizing the macrophage population to the suppressive M2-like phenotype (low M1/M2 ratio, Type A), others primarily utilize the progression-promoting properties of TANs (Type B) (Figure 5). These tumors express a prognostically favorable M1/M2 ratio but the anti-tumor effects of the M1-like TAM phenotype is counteracted by suppressive TANs.
In experimental studies in both cervical carcinoma and pancreatic adenocarcinoma, results suggest that there is a complementary immunosuppressive dynamic between the two myeloid cell types [37, 38]. Macrophage depletion via CCR2 blockade led to compensatory influx of neutrophils into the tumor, and only blockade of both cell types led to initiation of a tumor-directed immune response. A similar mechanism may also be relevant in early luminal breast cancer. Due to the immunohistochemical methodology of the present study, this hypothesis is of course speculative at this point and requires experimental confirmation.
Strengths of this study were the precise detection of cell densities separately in the stromal and intraepithelial compartments, as well as the uniform clinical characteristics of the patient cohort with exclusively luminal breast carcinoma, which were treated within the framework of a clinical trial. The relatively small cohort of patients with a low number of recurrences and metastases entailing a degree of statistical uncertainty was the major weakness of our analysis.