Inflammatory mediators in breast cancer: Coordinated expression of TNFα & IL-1β with CCL2 & CCL5 and effects on epithelial-to-mesenchymal transition

This study included 38 healthy individuals who were diagnosed with benign breast disorders that included fibrocystic changes, hyperplastic changes and benign tumors (this group is termed herein "Benign"). The study also included 88 breast cancer patients, who were divided to three groups: (1) 30 patients with DCIS. Since DCIS is a less malignant form of disease that rarely develops to IDC [62‐64], the DCIS group included patients in whom disease has not progressed to IDC in the course of 5-10 years of follow up; (2) 23 patients with IDC, who remained disease free in the course of 5-10 years of follow up (IDC-no-relapse); (3) 35 patients with IDC who relapsed with metastases or local tumor/s, or who died of breast cancer, in 5-10 years of follow up (IDC-with-relapse).

The characteristics of the study patients who were diagnosed with DCIS, IDC-no-relapse and IDC-with-relapse are presented in Table 1. Tumor status (T1, etc.) was determined by the guidelines of the Cancer Staging Manual of the American Joint Committee on Cancer. The patients were treated and followed in Tel Aviv Sourasky Medical Center and in Meir Medical Center. The study was approved by the Helsinki Committees of both Centers as a retrospective study based on information obtained from files of unidentified patients (included in the category of studies which do not require informed consent from study patients).

In this study, we have used archived paraffin blocks of patients, all obtained at the time of diagnosis. Serial sections (5 μm thick) were prepared from the blocks and processed for immunohistochemistry (IHC) by the Pathology Departments of Tel Aviv Sourasky Medical Center and Meir Medical Center.

The biopsy sections were deparaffinized, dehydrated in xylene and graded alcohols, rinsed in PBS and stained by primary antibodies. All antibodies used in IHC were commercial monoclonal antibodies, and their ability to recognize their respective antigens was confirmed in ELISA assays (data not shown). Staining by primary monoclonal antibodies for human CCL5 (PeproTech, Rocky Hill, NJ; Cat# 500-M75) and to human CCL2 (R&D Systems, Minneapolis, MN; Cat# MAB679) was performed by two protocols that showed similar sensitivity and specificity: (1) Sections were incubated with hyaluronidase at 37°C for 1 hr, then treated with 3% H₂O₂ for 10 min at room temperature (RT). After rinsing in PBS, non-specific binding was blocked by incubating the sections with normal goat serum at 37°C for 30 min. Then, the sections were stained over night with the primary antibodies at 4°C, except for sections of DCIS patients in which the staining was performed for 4 hr (the time was reduced in this case due to some background staining that was observed only in the DCIS biopsies stained by the different antibodies). (2) Sections were immersed in a staining dish containing citrate buffer at 95-100°C for 22 min, then cooled for 20 min. After washing with PBS, the sections were blocked with blocking solution (0.05% sodium azide and 0.05% Tween 20 in PBS). Next, the sections were incubated with the different primary antibodies for 1 hr at RT, blocked by 3% H₂O₂ for 10 min at RT, and washed thoroughly in PBS.

Staining for TNFα and IL-1β was performed according to the second procedure detailed above, using monoclonal antibodies to human TNFα (PeproTech; Cat# 500-M26) and to human IL-1β (Exalpha, Watertown, MA; Cat# L140M).

For all four factors (CCL2, CCL5, TNFα and IL-1β), the staining by secondary antibodies was performed as follows: sections were washed thoroughly in PBS, stained with biotinylated anti broad spectrum secondary antibody, or alternatively, by anti mouse secondary antibody (both having similar sensitivity and specificity) for 10 min at RT. After additional washing in PBS, the sections were stained by HRP-streptavidin for 10 min at RT. Then, the sections underwent additional washings in PBS and were incubated with Diaminobenzidine for 10 min at RT, washed and counterstained by incubation with hematoxylin for 5 sec at RT. After washing in H₂O, the sections were dehydrated in graded alcohols and mounted to cover slides.

In all cases, negative controls included substitution of the primary antibodies by PBS, or by non-relevant isotype controls. The specificity of the antibodies against TNFα and IL-1β was verified using mouse IgG1 as an isotype control for the antibodies (that shared the same isotype). Also, since the antibodies against CCL5 share the same isotype with the antibodies against CCL2, but there were biopsies that stained with one but not with the other, they served as internal controls for each other's specificity (In the 88 breast cancer patients: 9% of the biopsies stained for CCL2 but not for CCL5, and 25% of the biopsies stained for CCL5 but not for CCL2; In all other cases, the expression of CCL2 was coordinated with that of CCL5).

All slides were submitted to light microscopy, and the staining pattern in the cells was evaluated in a blind manner by a pathologist having expertise in breast cancer, and by a pathology-experienced researcher. The tumors were evaluated for % chemokine/cytokine-positive cells in the tumors (0-100%) and intensity of expression of the chemokines/cytokines (1 = low intensity, 2 = medium intensity, 3 = high intensity). TNFα and IL-1β were also evaluated for score of expression, determined as "% cytokine-positive cells in the tumors × the intensity of cytokine expression in each of the tumors". In line with the evaluations of many other markers in IHC, a threshold for positive staining was set. Based on our past experience, non-specific staining (exhibited as rare/sporadic/random stainings) could account for up to 5% of the epithelial cells in the biopsy, therefore the cut-off for positivity was set on 5%.

The data on expression of Her2-neu, estrogen receptor α (ER) and progesterone receptor (PR) that are included in Table 1 were provided by Tel Aviv Sourasky Medical Center and by Meir Medical Center, where the patients were treated and followed.

In this study, a comparison was made between the incidences of expression the different factors in normal breast epithelial duct cells of the Benign group and the tumor cells in DCIS, IDC-no-relapse and IDC-with-relapse groups (Fig.[2]). This analysis was performed by Chi-square or two-tailed Fisher's exact test as applicable. Hochberg's GT2 method for multiple comparisons was applied for pair-wise comparisons between groups of patients.

The differences between the incidences of expression of the different factors in normal breast epithelial cells that were adjacent to tumor cells in the cancer biopsies, as compared to the tumor cells in the DCIS, IDC-no-relapse and IDC-with-relapse groups (Table 2), were determined by McNemar test (except for TNFα expression in "IDC-no-relapse" group, that because of limitations of the McNemar test was analyzed by Wilcoxon).

Chi-square or Fisher's exact test were used in order to determine the significance of associations between factors that belonged to "Group 1" (TNFα & IL-1β) and factors that belonged to "Group 2" (CCL2 & CCL5) (Fig.[3]).

Group means were compared using a one-way Analysis of Variance (Fig. [7]and Table 3). Whenever a significant effect of group was observed, pair-wise comparisons between groups were performed using the Ryan-Einot-Gabriel-Welsch Multiple Range Test. Levene's Test for Homogeneity of Variances was applied to compare dispersion between groups.

Logistic regression model was applied to assess the potential use of Her2-neu, ER and PR with cytokines as risk factors for disease relapse.

All statistical analyses were performed using the SAS for Windows version 9.1.3.

The T47D and MCF-7 human breast carcinoma cell lines were grown in DMEM medium as described previously [20]. The growth medium was replaced by serum-deficient medium, and the cells were grown to confluency in the absence or in the presence of TNFα or IL-1β for 72 hr.

E-cadherin expression was determined in live cells by mouse antibodies against human E-cadherin (Santa Cruz Biotechnology, Santa Cruz, CA), followed by FITC-conjugated antibodies against mouse IgG (Jackson ImmunoResearch Laboratories, West Grove, Pennsylvania). Baseline staining was obtained by adding the appropriate buffer to the cells instead of primary antibody, and by staining with non-relevant isotype-treated antibodies. Determination of vimentin expression was performed in methanol-treated cells by mouse antibodies against human vimentin (Santa Cruz Biotechnology), followed by FITC-conjugated antibodies against mouse IgG (Jackson ImmunoResearch Laboratories). Baseline staining was obtained by non-relevant isotype-matched antibodies. Stainings were determined with a Becton Dickinson FACSort (Mountain View, CA) using the CellQuest software.

The expression of β-catenin was determined by specific antibodies, along with nuclei staining by DAPI. The cells were fixed by 8% paraformaldehyde for 15 min. Following treatment by 0.2% triton for 10 min, blocking was performed with PBS containing 2% BSA for 30 min. Primary mouse antibodies to human β-catenin were then used (Santa Cruz Biotechnologies), followed by FITC-conjugated mouse antibodies against mouse IgG. Baseline staining was obtained by adding the appropriate buffer to the cells instead of primary antibody. Similarly fixed and permeabilized cells were incubated with phalloidin-Alexa 488 (Molecular Probes, PoortGebouw, The Netherlands) for detection of actin organization. Stained cells were detected by LSM 510 Meta confocal microscope (Carl Zeiss AG, Oberkochen, Germany).

Biopsy sections were obtained from the total of 126 individuals (as indicated in "Methods") at the time of diagnosis, and were stained by specific antibodies that recognized human CCL2, CCL5, TNFα or IL-1β. Figure 1 displays representative examples of the staining patterns of the four proteins in biopsy sections of healthy individuals (the "Benign" group), and of DCIS, IDC-no-relapse and IDC-with-relapse patients. While the expression of CCL2, CCL5, TNFα or IL-1β was only minimally detected in infiltrating leukocytes of all groups of patients, the four factors were clearly observed in breast tumor cells, with mainly a cytoplasmic staining pattern. Based on the above, we have performed detailed analysis of the expression patterns of the four factors in breast epithelial cells, normal and malignant, in biopsy sections of healthy individuals and of breast cancer patients, as indicated below.

Figure 2 provides information on the incidences of expression of CCL2, CCL5, TNFα and IL-1β in normal breast epithelial cells of Benign patients, and in the tumor cells in patients diagnosed with DCIS, IDC-no-relapse and IDC-with-relapse. All four factors were minimally detected in normal breast epithelial duct cells in the healthy individuals of the Benign group (expression incidences ranging between 5.3% and 10.5%). A significant elevation was denoted in the expression of CCL2, CCL5, TNFα and IL-1β in the tumor cells in all groups of cancer patients (DCIS, IDC-no-relapse and IDC-with-relapse), when compared to their expression in the normal breast cells in biopsies of the Benign patients, with significance value of p < 0.001 in all cases.

In parallel, we determined the expression of CCL2, CCL5, TNFα and IL-1β in normal breast epithelial duct cells that were in proximity to the tumor cells in biopsies of patients with DCIS, IDC-no-relapse and IDC-with-relapse (Table 2). In these three groups of cancer patients, we found low incidence of expression of the inflammatory chemokines and inflammatory cytokines in the normal cells adjacent to the tumor cells, while they were expressed at significantly higher incidence in the tumor cells (incidences and p values are presented in Table 2). The incidence of CCL2, CCL5, TNFα and IL-1β expression in the normal epithelium was not significantly different between DCIS and IDC-no-relapse, and also did not differ significantly between DCIS and IDC-with-relapse patients.

Taken together, these findings indicate that the expression of the inflammatory chemokines and inflammatory cytokines was acquired by breast epithelial cells upon their malignant transformation.

Next, we compared the three groups of breast cancer patients (DCIS, IDC-no-relapse, IDC-with-relapse) with respect to the incidence of expression of CCL2, CCL5, TNFα and IL-1β in the tumor cells. The results of Figure 2 clearly show that the expression of the inflammatory chemokines CCL2 and CCL5 remained high and constant in the tumor cells of DCIS, IDC-no-relapse and IDC-with-relapse patients. No significant differences were denoted between DCIS, IDC-no-relapse and IDC-with-relapse patients, with respect to the incidence of CCL2 and CCL5 expression in the tumor cells.

A high incidence of expression in the tumor cells was also denoted for TNFα and IL-1β in the three groups of cancer patients: DCIS, IDC-no-relapse and IDC-with-relapse. The results of Figure 2 indicate that the incidence of TNFα expression was 50% in DCIS patients and was significantly increased to 85.7% in IDC-with-relapse patients (p = 0.005); Similarly, the incidence of IL-1β was significantly raised from 60% in DCIS to 88.6% in IDC-with-relapse patients (p = 0.017). No other significant differences were found between the different groups of breast cancer patients with respect to TNFα and IL-1β expression in the tumor cells (namely: IDC-no-relapse vs. DCIS; IDC-no-relapse vs. IDC-with-relapse).

Together, these findings indicate that there was high incidence of CCL2, CCL5, TNFα and IL-1β expression in the tumor cells in the three groups of cancer patients, and that the expression of TNFα and IL-1β was further elevated in the IDC-with-relapse group. The rare expression of all four factors in normal breast epithelial cells, and the high incidence of their expression in tumor cells in biopsies of all groups of cancer patients, indicate that the expression of the inflammatory chemokines CCL2 & CCL5 is coordinated with that of the inflammatory cytokines TNFα & IL-1β along stages of tumor development and progression in breast cancer.

Based on the above results, we asked whether there are associations between the inflammatory chemokines CCL2 & CCL5 and the inflammatory cytokines TNFα & IL-1β in patients representing different stages along the progression process of breast cancer. To this end, we divided the four factors to two groups, based on their published functional interactions in breast tumor cells: Group 1 - The inflammatory cytokines TNFα & IL-1β, which promote the release of CCL2 & CCL5 by breast tumor cells; Group 2 - The inflammatory chemokines CCL2 & CCL5, whose release by malignant breast cells is increased by TNFα and IL-1β (Please refer to Additional Files 1 and 2, as well as to references [65‐70]. These analyses were performed in the tumor cells in the three groups of cancer patients: DCIS, IDC-no-relapse and IDC-with-relapse.

When we analyzed the DCIS group, we found that factors belonging to Group 1 (TNFα & IL-1β) "tended" to be expressed in the same biopsies in which the factors of group 2 were expressed (CCL2 & CCL5) (Figure 3a). For example, the study of CCL2 showed that CCL2 was co-expressed with TNFα and IL-1β in 33.3% of the DCIS biopsies. In contrast, in only 10% of the DCIS patients CCL2 was detected without co-expression of TNFα and IL-1β. This finding indicates that there was a preferential expression of CCL2, as a representative of Group 2, with factors that belonged to Group 1, namely TNFα & IL-1β.

Thereafter, analyses that were performed for patients diagnosed with IDC-no-relapse have shown that the expression of CCL2 & CCL5 was coordinated with that of TNFα & IL-1β (Figure 3b). This was substantiated by significant associations that were detected between factors of Group 1 (TNFα & IL-1β) and Group 2 (CCL2 & CCL5), with p = 0.036.

A somewhat different pattern of expression was obtained for IDC-with-relapse patients (Figure 3c). As in the DCIS and IDC-no-relapse groups, the incidence of CCL2 and CCL5 in the IDC-with-relapse group remained at the level of 50-65%; however, the incidence of TNFα and IL-1β expression in the IDC-with-relapse group was increased to 85-90% (Figure 2). Therefore, in many of the patients of this group, the expression of TNFα & IL-1β was coordinated with that of CCL2 & CCL5, but in others the expression of TNFα & IL-1β was not accompanied by expression of CCL2 & CCL5. This has led to insignificant associations (p = 0.627) between Group 1 (the inflammatory cytokines TNFα & IL-1β) and Group 2 (the inflammatory chemokines CCL2 & CCL5) in the IDC-with-relapse group.

To conclude, in patients that belonged to the DCIS and IDC-no-relapse groups, the expression of the inflammatory chemokines was coordinated and significantly associated with that of the inflammatory cytokines. In contrast, in the IDC-with-relapse group there was sub-population of patients in which the expression of CCL2 & CCL5 did not accompany the expression of TNFα & IL-1β. In this group of IDC-with-relapse patients, the expression of TNFα & IL-1β was further elevated and was highly prevalent, supporting an important role for TNFα and IL-1β in disease progression and recurrence. Furthermore, these results suggest that in this specific sub-group of IDC-with-relapse patients, the two cytokines are coordinated with other tumor-supporting factors that replace CCL2 & CCL5 in promoting disease relapse.

The findings on the elevated expression of TNFα and IL-1β in the IDC-with-relapse group suggest that these two cytokines support disease progression. In addition to the tumor-promoting activities of TNFα and IL-1β that have already been described, it is possible that these two cytokines also have functional interactions with CCL2 and CCL5, by that promoting disease course. In line with this possibility, our findings have shown that TNFα and IL-1β up-regulated the release of CCL2 and CCL5 by breast tumor cells (Additional Files 1 and 2; also suggested by findings of references [65‐70]). Such a cross-talk may lead to up-regulated expression of CCL2 and CCL5 in tumors expressing TNFα and/or IL-1β, thus possibly leading to increased tumor-supporting activities of the two chemokines in breast tumors (see Discussion, below).

In addition to their ability to promote the release of CCL2 and CCL5 by the tumor cells, it is possible that TNFα and IL-1β act directly to elevate processes that are required for local recurrence or metastasis formation, which identify the IDC-with-relapse group of patients. Recent studies indicate that acquisition of self-renewal properties that are required for formation of recurrent local tumors, as well as metastasis formation, are promoted by Epithelial-to-Mesenchymal Transition (EMT) processes [71‐77]. Based on the above, we determined the possibility that TNFα and IL-1β promote EMT processes in the tumor cells, thus pushing forward disease recurrence and progression.

To investigate this possibility, we tested the ability of TNFα and IL-1β to induce in the tumor cells properties that are typical of EMT [71‐77]. In these experiments, we used the T47D and MCF-7 cells, which represent a non-advanced stage of breast malignancy [78‐82], and thus can serve as an appropriate platform for induction of EMT processes that are associated with more progressed disease.

First, we stimulated the cells by TNFα and IL-1β (please see "Note" in legend to Figure 4 on cytokine concentrations used in these analyses), and determined the ability of the cytokines to induce reduction in the membranous expression of E-cadherin, an adhesion molecule which is important in establishing cell-to-cell contacts. The results of Figures 4a and 4b demonstrate that TNFα potently induced a very typical property of EMT, namely reduction in E-cadherin expression at the cell membrane of the tumor cells. This activity of TNFα was induced in both the T47D and the MCF-7 cells in a dose-dependent manner. In parallel, stimulation by IL-1β has also led to reduced expression of E-cadherin at the plasma membrane, but only in the T47D cells (Figure 4c).

Based on these results and on additional preliminary findings (data not shown), we suspected that TNFα was a more potent inducer of EMT than IL-1β, and that its effects were more prominent on MCF-7 cells than on T47D cells. Therefore in depth analyses were continued with TNFα, analyzed on MCF-7 cells only. Further analyses indicated that TNFα has led to substantial decrease in the expression of β-catenin at the cell membrane of the tumor cells (Figure 4d), an event which is typical of EMT. Also, the stimulation by the cytokine has promoted the expression of vimentin, a mesenchymal feature whose expression is elevated in cells undergoing EMT (Figure 5a). In parallel, the tumor cells have acquired cellular protrusions, accompanied by extensive re-organization of the actin cytoskeleton (Figures 5b and 5c, respectively).

In addition to the above, TNFα has induced in the tumor cells elevated adhesion to substrate, a step necessary for establishment of productive metastases at remote organs (Figure 6a). Importantly, following TNFα stimulation, the cells have acquired the most critical and metastasis-relevant property of EMT, namely increased migratory and invasive properties (Figures 6b and 6c), by that substantiating the ability of the cytokine to induce EMT in breast tumor cells.

Taken together, the above results indicate that TNFα, and possibly also IL-1β - although to a lower extent - induced EMT properties in the tumor cells. Of note, similar activities of the two cytokines were obtained in an unrelated ongoing study in our laboratory, performed on a different set of breast cells (data not shown; manuscript in preparation), further substantiating the EMT-promoting activities of TNFα and IL-1β. Also, preliminary analyses that we have performed indicated that CCL2 and CCL5 were not capable of inducing EMT-related processes in the breast tumor cells (data not shown).

Since the potent EMT-promoting activities of TNFα may push forward processes of increased metastasis formation and disease recurrence, we further addressed the characteristics of TNFα activity. In view of reports showing that the EMT phenotype may be transient and needs continuous exposure to EMT-inducing factors [72], we asked if induction of EMT by TNFα required constant stimulation by the cytokine. To determine this issue, we have taken cells in which EMT was induced by TNFα (as indicated by reduced expression of E-cadherin at the cell membrane), and extended their growth for additional 72 hr with or without TNFα. The results of Figure 7 clearly indicate that the EMT phenotype was reversible if the cells were deprived of TNFα, while in cells that were grown continuously with the cytokine the EMT phenotype was preserved.

The above findings suggest that breast tumor cells benefit from high and constant presence of TNFα, because the persistence of the cytokine may induce EMT processes that lead to progressed and recurred disease. In such a case, we would expect that TNFα would be more persistent in tumors of IDC patients who relapsed with metastases or local tumors, compared to IDC-no-relapse or to DCIS patients.

To find out if this was indeed the case, we determined in each of the three groups of breast cancer patients (DCIS, IDC-no-relapse, IDC-with-relapse) the parameters of % TNFα-positive cells in the tumors, and the score of TNFα expression. The results presented in Figure 8a and in Table 3 demonstrate that there was a significant elevation in % TNFα-positive cells/patient in the IDC-with-relapse group when it was compared to DCIS and to the IDC-no-relapse groups, where p = 0.0008 for % TNFα-positive cells, and p = 0.0331 for TNFα score (as indicated in the legend to Table 3).

Of note, similar analyses that were performed for IL-1β have also shown highly significant persistence of this cytokine in tumors of IDC-with-relapse patients, with p < 0.0001 for % IL-1β-positive cells, and p = 0.0004 for IL-1β score (Figure 8b and Table 3). These results suggest that persistence of IL-1β in the tumors is also important for driving forward processes of progression and recurrence. Based on our data, these activities of IL-1β are not necessarily related to EMT, because this cytokine was not as effective in this function as TNFα. Rather, IL-1β may acquire activities that are related for example to elevated angiogenesis, as previously suggested [28, 31, 83]. In addition, it is possible that IL-1β acts jointly with other pro-malignancy factors to promote disease progression and recurrence. Such additional factors could be related to the response of the tumor cells to growth factors that are found at the tumor microenvironment, or to hormonal stimulation.

Supporting this possibility are analyses that we have performed, using the data obtained on the expression of Her2-neu, ER and PR in the patients. In this analysis, we have compared between the IDC-no-relapse and the IDC-with-relapse groups. Statistical analysis that was performed for IL-1β, taking into account the status of Her2-neu, ER and PR, all combined (currently each of these three factors is used in the clinic as a predictor of disease progression and recurrence in breast cancer), indicated that IL-1β was a significant risk factor for disease relapse: p = 0.0402, with 95% confidence limits of 1.098 and 59.93 and odds ratio 1 value of 8.112 (also, of the different clinical markers, PR was statistically different between the IDC-no-relapse and IDC-with-relapse groups, with p = 0.0391). Similar analyses performed for TNFα, CCL2 and CCL5 did not reveal significant positive correlation with disease progression.

The findings on IL-1β strongly support the possibility that when the tumor-derived and/or microenvironment-derived characteristics have higher cancer-promoting properties, IL-1β can further promote progression-related processes in breast cancer.

To date, the relationships between inflammatory chemokines and inflammatory cytokines in breast cancer were not elucidated, and the associations between them during the process of breast cancer development and progression were not analyzed. At this setting, the findings of our study are important because they indicate that the expression of the inflammatory chemokines CCL2 & CCL5 and of the inflammatory cytokines TNFα & IL-1β is coordinated in crucial stages along the process of disease progression. Moreover, the expression of all four inflammatory factors is minimal in normal breast epithelial cells, and is simultaneously acquired by the cells once malignant transformation has taken place, namely from the DCIS stage and on. These results suggest that events dictated by genetic/epigenetic alterations in the tumor cells, or by the microenvironment, lead to a synchronized up-regulation in the expression of several inflammatory mediators together, by transformed breast epithelial cells.

The coordinated expression of CCL2 & CCL5 and of TNFα & IL-1β along stages of breast cancer development and progression is important because the activities of the four factors are not fully overlapping (Refs [3, 5‐24] for CCL2 & CCL5, [25‐59] for TNFα and IL-1β). Therefore, it is possible that each of these factors contributes its own share to disease course, alongside with the others. Together, the coordinated presence of CCL2 & CCL5 and of TNFα & IL-1β may support malignancy, possibly also due to spatio-temporal cross-interactions between them. Indeed, our findings (Additional Files 1 and 2) and published studies [65‐70] indicate that TNFα and IL-1β up-regulate the release of CCL2 and CCL5 by breast tumor cells. Furthermore, recent findings obtained in our studies suggest that TNFα and IL-1β exert tumor-promoting activities that are connected to the ability of CCL2 and CCL5 to function as cancer-supporting factors. Consequently, it is possible that the pro-tumorigenic activities of the inflammatory cytokines and of the inflammatory chemokines depend on each other, and/or are complementary to one another. This possibility emphasizes the need to determine whether combined inhibition of all factors together will lead to improved limitation of tumor growth.

Moreover, since our findings indicate that the coordinated array of cytokines and chemokines takes action in about 50-70% of the patients (depending on the group), it is possible that these factors can act along with other inflammatory and tumor-supporting factors to promote tumor growth and metastasis, as further discussed below.

In the group of IDC-with-relapse patients, the incidence of CCL2 & CCL5 expression was similarly high to that of DCIS and IDC-no-relapse group (50-65%), while that of TNFα & IL-1β has reached about 85-90%. Therefore, while the expression of CCL2 & CCL5 was coordinated with TNFα & IL-1β in many of the patients belonging to this group, there was a sub-population of patients in which the expression of TNFα & IL-1β was not accompanied by CCL2 & CCL5. While we need to understand the reasons for this lack of coordination in this specific sub-group of patients (please see discussion below), this observation indicates that TNFα & IL-1β are required, and play important roles in disease progression, and suggests that they cooperate with other factors at the tumor microenvironment that substitute for the lack of CCL2 & CCL5.

Of interest in this respect are the reasons for the lack of expression of CCL2 & CCL5 in the specific sub-population of IDC-with-relapse patients who do express TNFα & IL-1β. This lack of coordination between CCL2 & CCL5 and TNFα & IL-1β in this sub-population is intriguing mainly because TNFα and IL-1β were found to stimulate the release of CCL2 and CCL5 by breast tumor cells. The explanation could be given, at least partly, by consideration of the extent to which receptors for TNFα and IL-1β are expressed by the tumor cells, and of which type. Although the expression of TNFα & IL-1β receptors by breast tumor cells was already documented [29, 31, 37, 43, 44], it is possible that in this specific sub-group of the IDC-with-recurrence patients, the tumor cells do not express the required receptors for TNFα & IL-1β, or express non-signaling receptors. Under such conditions, TNFα & IL-1β would not act on the tumor cells to promote the release of CCL2 & CCL5, leading to lack of associations between the two cytokines and CCL2 & CCL5.

Accordingly, in our future investigations we will perform detailed analysis of the expression of receptors for the two cytokines (e.g. TNFRI, TNFRII, IL-1RI) in the course of disease progression in breast cancer. Such an analysis may enable us to better identify the basis for the patterns of expression of the inflammatory cytokines and chemokines in different stages of disease.

As indicated above, the incidence of TNFα & IL-1β was further increased in IDC patients in whom disease has relapsed, suggesting that these two cytokines push forward processes of tumor progression and recurrence. Our findings support this possibility because they indicate that TNFα, and to a lower extent also IL-1β, induce EMT properties in the tumor cells.

In addition, our findings demonstrate that in order to undergo EMT, the tumor cells had to be constantly stimulated by TNFα. These results suggest that in situ, EMT processes would be facilitated by high persistence of TNFα at the tumor site. Under such conditions, we would expect high prevalence of TNFα in tumors of patients belonging to the IDC-with-relapse group, compared to IDC-no-relapse or to DCIS patients. Indeed, here we have obtained an important indication to the roles of TNFα in disease recurrence and progression, because there was significantly higher prevalence of TNFα in tumors of the IDC-with-relapse group than in the other two groups of cancer patients. In addition, significantly elevated persistence of IL-1β was also observed in the IDC-with-relapse patients. Although IL-1β was not a strong stimulant of EMT, it is possible that this cytokine supports tumor recurrence or metastasis by other means, such as increased angiogenesis, as indicated by several published studies [28, 31, 83], or by having joint activity with other factors. Supporting this possibility is the finding that in a specific setting of Her2-neu, ER and PR expression in the tumors, IL-1β was identified as a risk factor for disease recurrence, suggesting that it can act jointly with other pro-malignancy factors to promote disease progression in breast cancer.