全文

Introduction

With roughly 287,000 new cases per year endometrial cancer (EC) is the sixth most frequent cancer in women worldwide. The burden of EC is more severe in developed nations making it the most common malignancy of the female genital tract in high-income countries with approximately 142,000 new cases annually. Even more alarming is the fact that EC incidence rates continued to increase over the last several years. The rise is, however, confined to type I EC and thus most likely attributable to the growing prevalence of overweight. EC has been divided into two groups with a different biological behavior and clinical course. Estrogen-driven obesity-related type I endometrioid adenocarcinoma accounts for the majority of cases (75–80%); the less common type II cancer is of serous papillary or clear-cell histology and not linked to estrogen exposure. While type II carcinomas carry a poor prognosis, type I malignancies have an excellent outcome due to their mostly early presentation and an efficient treatment consisting of total hysterectomy and bilateral salpingo-oophorectomy with systematic pelvic/paraaortic lymphadenectomy in selected cases. Adjuvant radio- or brachy- as well as chemotherapy has been shown to be of benefit in high-risk cases. Mortality is primarily related to advanced or recurrent disease for which effective therapeutic options are still lacking. Although radio-, brachy- or chemotherapy may achieve a transient treatment response the median survival in this group is less than one year [1]. Thus, new and effective therapeutic strategies are needed.

 

It is well known that tumor microenvironments harbor an extensive immune infiltrate that is able to promote tumor growth. In particular macrophages appear to constitute attractive targets for treatment, as they are one of the most abundant stromal immune cell populations. A macrophage-affecting therapeutic approach is further reasonable based on the assumption that host-tumor immune interactions might be important for the reversion of the malignant phenotype by reestablishing normal control mechanisms [2]. Macrophages originate from monocytes and undergo specific differentiation in response to the local context. Recently, using transcriptomic technology, at least nine distinct macrophage activation states have been identified, a fact that expands the dichotomous framework of classical activated type 1 (M1) and alternatively activated type 2 macrophages (M2) [3]. Tumor-associated macrophages (TAMs), attracted and sustained by the neoplasm, share pro- and antiinflammatory properties and are therefore described as a separate subset. They are able to support the malignancy by enhancing immune suppression, hem-/lymphangiogenesis, invasion and metastasis [4]. Accordingly, increased infiltration of TAMs has been reported to correlate with unfavorable prognosis in a plethora of tumors [5] including EC [6] and [7]. To identify TAMs in EC samples we used the monocyte/macrophage-restricted expression of the type I transmembrane glycoprotein CD163, which belongs to the scavenger receptor cysteine-rich superfamily [8]. CD163 expression is upregulated along the macrophage differentiation pathway and has been shown to be specifically induced in TAMs by immunosuppressive signals [9]. We simultaneously evaluated an immunosuppressive effector cell population, referred to as regulatory T cells (Treg cells) and phenotypically identified by Foxp3 expression. The accumulation of these cells has also been shown to be associated with poor outcome not only in EC [10] but also in other tumor entities [11] and [12] Further, to accurately assess lymph vessel density (LVD) and lymphovascular space invasion (LVI) the antibody D2-40 against the glycoprotein podoplanin, which is predominantly expressed on lymphatic endothelium, was used.

 

We first evaluated the role of immunosuppression in EC biology by correlating TAMs and Treg cells with features of aggressive tumor behavior, namely advanced FIGO stages, high tumor grade, LVD, LVI and lymph node metastasis. Further, the accumulation of suppressive immune infiltrates was tested as a prognosticator of recurrence-free and overall survival. The goal of our study was to build the basis for new immunotherapeutic regimens that may help to improve disease outcome.

 

Material and methods

Patients and specimen

The study population consisted of a retrospective sample of 163 patients with EC (all histologic subtypes of endometrioid adenocarcinoma that are grades 1–3) diagnosed at the University of Bonn between 1995 and 2008. Tumors with a serous papillary or clear-cell histology and those confined to the endometrium (formerly FIGO IA) were excluded. Patients were treated according to the S2k guideline of the German Cancer Society and the German Society for Gynecology and Obstetrics [13]. Surgical management consisted of total hysterectomy and bilateral salpingo-oophorectomy with or without systematic pelvic/paraaortic lymphadenectomy. Adjuvant chemotherapy was performed in 7 of 18 (39%) advanced cases (FIGO stages III/IV) using taxane/platinum- or anthracycline/ifosfamide-based regimens; 56 patients (34.4%) received vaginal brachytherapy and 33 women (20.2%) were treated with pelvic or whole-abdomen radiotherapy. Clinical parameters were collected from medical records; follow-up data were updated until February 2014. The histological cell type was determined according to the World Health Organization (WHO) criteria. The 2010 revision of the International Federation of Gynecology and Obstetrics (FIGO) system was used to assign tumor grade and stage. Detailed patient baseline characteristics are listed in Table 1. The study was approved by the Institutional Review Board of the University of Bonn.

 

Tissue microarray (TMA) design

For immunological profiling TMAs were constructed from archival formalin-fixed paraffin-embedded tissue specimen. Tumor areas were identified in sections stained with hematoxylin and eosin (HE). One representative 1.2 mm core biopsy was randomly obtained from the selected zones of tumor nests of the corresponding tissue specimen and arrayed in paraffin blocks.

 

Immunohistochemistry

Immunohistochemical staining of D2-40, CD163 and Foxp3 was performed on serial 4 μm sections using an automated staining system (DAKO TechMate 500; DAKO, Glostrup, Denmark). Labeling was visualized with the streptavidin-biotin-peroxidase/DAB technique (DAKO). Supplementary Table S1 reports the antibodies used for immunohistochemistry.

 

Assessment of immunohistochemical staining

Immunostained cells were analyzed with the Axio Observer D1 microscope (Zeiss, Jena, Germany) and the AxioVision 4.7 software (Zeiss). To avoid interobserver variability specimens were either analyzed by the same person (Foxp3: T.H.A., LVI: N.B.S., LVD: K.K.) or quantified with a semi-automated computerized method (CD163; AxioVision 4.7 software). Tissue sections were evaluated masked to clinical details. Whole slides were screened for the presence of LVI using the strict criterion of cancer cell clusters within D2-40+ lymphovascular spaces [14] and [15]. To determine LVD the area of the highest neovascularization at the tumor invasive front (hot spot) was selected in each specimen and D2-40+ vessels were quantified within an area of 0.6 mm2[16]. TMAs were used to assess the immunoreactivity of CD163 and Foxp3 by analyzing the whole core area. TAMs were defined as cells with CD163+ membranous/cytoplasmic staining; the soluble form of CD163 was excluded from the analysis. Since the irregular shape of macrophages makes it difficult to distinguish individual TAMs within a cell cluster, CD163 immunoreactivity was recorded digitally in accord with previous studies [17]. Focusing the microscope on the typical amount of CD163 coloring set the cut-off. Staining intensity was defined as the area above threshold for CD163 immunolabeling in percent of the whole image indicating the total amount of CD163+ macrophages. Treg cells were identified by anti-Foxp3 antibodies; the number of immunoreactive cells with a nuclear staining was counted visually [18]. Immunocyte infiltrates both within the tumor (intraepithelial) and the adjacent stromal compartment (periepithelial) were recorded as one.

 

Statistical analysis

Statistical analysis was carried out using SPSS version 21 (IBM Corp, Armonk, NY, USA). Comparisons between continuous data were performed using the Mann–Whitney U and the Kruskal–Wallis tests. Pearson's correlation coefficient r was used for association analysis. To identify the immunogenic factor that correlated best with lymphogenic metastasis we used binary logistic regression analysis. The median value of immunostaining was used as a cut-off point to assign tumors into high and low CD163 and Foxp3 reactivity groups. Further, the median number of lymph vessels served as a cut-off value to separate cases into high and low LVD groups. Age > 60 years [19] and obesity defined as a body mass index (BMI) ≥ 30 kg/m2 [20] have been shown to result in a poorer prognosis and were also used as thresholds. Cumulative survival analysis (recurrence-free survival, RFS; overall survival, OS) was performed using the Kaplan–Meier method; curves were compared with the log-rank test. Multivariate survival analysis was performed using the Cox's proportional hazard regression model. The median follow-up time was calculated using the reverse Kaplan–Meier estimator [21]. Results with a p-value < 0.05 were considered to be significant.

 

Results

Endometrial adenocarcinoma is characterized by an immunosuppressive microenvironment

Cancer-associated inflammation comprises the infiltration of immunocytes that have the potential to either promote or inhibit tumor initiation and progression. The aim of our study was to identify and characterize an immunosuppressive milieu in EC that might support cancer growth. We observed the presence of antiinflammatory cells of both myeloid and T cell origin. In detail, CD163+ TAMs and Foxp3+ Treg cells were detected in all EC specimens examined (Table 1). Tumors could be divided into tissue specimen with a high (Fig. 1A) or a low amount of CD163+ TAMs (Fig. 1B) and a high (Fig. 1C) or a low (Fig. 1D) density of Foxp3+ Treg cells, respectively. Analyzed immune cell populations were found both within the tumor epithelium and the underlying stroma. However, the infiltration of tolerogenic cells showed an overall trend toward higher values in the stroma compared with the epithelium. Further, the recruitment of TAMs in EC was found to correlate with the accumulation of Treg cells suggesting an amplification loop, which may sustain a tolerogenic microenvironment in EC (Fig. 1E).

 

 

A high amount of TAMs but not of Treg cells is associated with poor prognostic factors

In order to gain an improved understanding of the biological behavior of immunosuppressive cells in EC and their clinical relevance, we correlated the amount of TAMs and Treg cells with classical histopathological factors that are related to patient outcome. We found the count of TAMs to increase with progressing tumor stage (p < 0.02; Fig. 2A) and grade (p < 0.01; Fig. 2B). Of note, the highest levels of TAMs were observed in patients with aggressive tumor biology, namely FIGO stage III/IV tumors (p < 0.03; Fig. 2A) and grade 3 carcinomas (p < 0.01; Fig. 2B). In contrast, the amount of Treg cells was neither associated with tumor stage ( Fig. 2C) nor with grade ( Fig. 2D). Based on the unopposed estrogen theory, which hypothesizes that exposure to estrogen may augment mitotic activity of endometrial cells ultimately leading to cancer, we tested the association of different socio-biological risk factors related to hyperestrogenism with the presence of immunosuppressive cells [15]. We were, however, unable to relate TAMs with a higher postmenopausal age ( Supplementary Fig. S1A) or obesity ( Supplementary Fig. S1B). Likewise, Treg cells were equally distributed across ages ( Supplementary Fig. S1C) and weight categories ( Supplementary Fig. S1D).

 

 

 

TAMs support tumor cell spreading via lymphatic routes

Based on our results that TAMs but not Treg cells prognosticated the presence of aggressive EC biology we postulated that TAM-induced lymphangiogenesis might be the cause for the difference. TAMs unlike Treg cells are well known to promote the formation of new lymph channels by paracrine and cell autonomous mechanisms consequently facilitating lymphatic metastasis [22]. Indeed, using the highly specific lymphatic endothelium marker D2-40 (Fig. 1F) an increased LVD in the invasive front of EC was detected in cases with a high number of TAMs (p < 0.003; Fig. 3A). The invasion of cancer cells into lymph vessels was observed in 27% of the specimen ( Table 1) and correlated with an elevated amount of TAMs (p < 0.001; Fig. 3B). Consequently, lymph node metastasis preferentially occurred in tumors with a dense infiltration of TAMs (p < 0.03; Fig. 3C). With regard to Treg cells, a high amount was associated with the presence of LVD (p < 0.002; Fig. 3D) and LVI (p < 0.02; Fig. 3E) without being related to lymph node metastasis ( Fig. 3F). Accordingly, using binary logistic regression analysis only the presence of TAMs appeared to be an important indicator of LVD, LVI and lymph node metastasis ( Supplementary Table S2).

 

 

A high TAM count is indicative of short recurrence-free and overall survival

In the next step, we asked whether immunosuppressive cells would be of use as prognosticators of poor patient outcome. We found the RFS to be significantly prolonged in the group with low TAM density (Fig. 4A). The amount of TAMs was also able to distinguish between OS rates relating the high content to an unfavorable outcome (Fig. 4B). However, the analysis of Treg cell infiltration showed no significant differences in RFS (Fig. 4C) and OS (Fig. 4D) among patients with low or high Treg cell count. Due to an uneven stage distribution in our cohort with a low amount of FIGO stage III/IV disease, further prognostic analysis was performed for tumors confined to the uterus. Even in the subgroup of FIGO stage I/II tumors, the extent of TAM infiltration was found to be a prognostic factor for RFS (Supplementary Fig. S2A). However, OS in low-risk disease was similar between patients with high and low TAM count (Supplementary Fig. S2B). The amount of Treg cells did not affect recurrence and survival curves (Supplementary Fig. S2C, D).

 

 

Consistent with our results above, univariate analysis identified the following risk factors of RFS and OS: enhanced CD163 immunoreactivity, high FIGO stage and grade, abundant presence of LVI and lymph node metastasis (Table 2). High LVD was associated only with the risk of tumor recurrence. In the multivariate model for RFS CD163 immunostaining remained significant. Of note, a high amount of TAMs increased the risk of recurrence by 8.31 times. The multivariate analysis of OS, however, identified only the FIGO stage as an independent prognostic factor.

 

 

Discussion

Our study demonstrates that EC is characterized by a marked infiltration of immunocytes indicative of a tumor-specific immune response. Thus, in accord with previous results EC is considered to be an immunogenic tumor. Malignant cells of the endometrium have been shown to induce cytotoxic CD8+ T cell responses [23]. Moreover, EC frequently expresses the tumor antigen KU-CT-1, which is able to elicit antigen-specific CD4+ T cells and antibodies [24]. However, signals that defeat antitumor immunity are often more influential in the cancer microenvironment. Accordingly, our study identified a substantial amount of tolerogenic cells of both myeloid and T cell origin in all specimens analyzed. These data are consistent with those of other studies reporting the presence of antiinflammatory cell populations in EC, including TAMs [6], [7], [25], [26], [27], [28] and [29] and Treg cells [10], [30] and [31]. Moreover, we observed a positive correlation between quantities of TAMs and Treg cells in our cohort indicating an ongoing immunoregulation in the tumor microenvironment. Several findings support the hypothesis of crosstalk between TAMs and Treg cells. In detail, TAMs may produce CCL20, which promotes Treg cell recruitment to the tumor [32]. Likewise, Treg cells are able to secrete IL-4, IL-13 and IL-10 that polarize macrophages toward the M2 phenotype [33].

 

Although these data suggest, TAMs and Treg cells act as synergistic amplifiers of immunosuppression that foster tumor growth and progression, only TAMs accumulated in advanced FIGO stages and high tumor grade. Our observations are in accord with published results that reported the association of TAMs with poor prognostic factors in EC [6], [25], [26] and [27]. Only one study differed from these results and was unable to establish a positive relationship between TAMs and histopathologic features presumably due to a small and heterogeneous cohort of type I and II carcinomas [29]. Taken together, our findings suggest that TAMs play a crucial role in EC by promoting aggressive tumor behavior. This hypothesis is further strengthened by the observation that TAMs increased linearly with disease progression from endometrial hyperplasia to EC and were thus more frequently identified in EC than in benign endometrial tissue [6] and [29].

 

Despite the lack of functional studies that address the mechanism by which TAMs promote EC development, previous histopathologic reports demonstrated a positive association between the number of TAMs and Ki-67-assessed EC cell proliferation [25] and [27]. Provascular properties capable of promoting tumor cell survival have been attributed to TAMs in EC since microvessel density and levels of the proangiogenic molecules IL-8 and VEGF correlated with the macrophage count [6], [25], [28] and [34]. Thus, in addition to their known immunoregulatory function that dampens the tumor-specific immune response TAMs also play a role in the vascular network by influencing both hem- and lymphangiogenesis. In contrast to its sister process only little data are available on TAM-induced lymphatic remodeling in EC. Our study revealed a positive relationship between TAMs and the whole process of lymphatic spread including lymphatic sprouting, LVI and lymph node metastasis. It is well established that a high LVD is a prerequisite for cancer cell dissemination. Therefore, lymph vessels were counted in the tumor invasive front, considered most crucial for malignant cell spread [35]. The range of lymph vessels determined in our study was in accord with previous reports [36] and [37]. Consistent with our results, increased LVD has been correlated with a dense infiltration of TAMs in a number of cancer entities [38]. However, until now no data are available on TAM-associated LVD in EC. Following the metastatic cascade theory the invasion of tumor cells into lymph vessels is the next critical step, which we found to be associated with a high macrophage count. These data are similar to published results although previous results were based entirely on the evaluation of HE sections [6]. Thus, the strength of our investigation lies in the use of a D2-40-based assessment of lymph vessels, which is known to improve the detection rate of LVI [14]. Consistent with our findings of TAM-promoted LVD and LVI TAMs were also closely associated with the metastatic spreading to the regional lymph nodes, which is in accord with earlier studies [6], [25] and [26].

 

Further support for our hypothesis that EC is able to recruit growth-promoting and prolymphangiogenic TAMs to become more aggressive comes from the observation that patients with a high amount of TAMs have a poor prognosis. This macrophage-specific influence on survival has also been found by other authors [6] and [7]. Our multivariate analysis identified a high density of TAMs to be an independent prognostic factor of RFS but not of OS. Although our study had a long-term follow-up, it may have been too short to detect independent prognostic factors of OS based on the fact that our cohort included a high amount of FIGO stage I/II cases. Moreover, treatment with curative intent in relapsed cases might have influenced survival.

 

The clinical importance of Treg cells in EC remains unclarified. The lack of prognostic significance is supported by a recent report that was unable to link Treg cells to histological parameters [31]. Other studies, however, showed that high Treg cell numbers are associated with unfavorable clinicopathological factors [10] and [30]. One of the reasons for these conflicting results might be the use of antibodies that target different Foxp3 epitopes finally resulting in various staining patterns [18]. Moreover, T cell receptor activation in conventional T cells may lead to a transient Foxp3 expression without inducing immunosuppression indicating that not all Foxp3+ cells are Treg cells [39].

 

The strength of our investigation lies in the use of a macrophage-specific marker for TAMs. Using CD163, the current study is the largest series published to date examining immunosuppression in EC and the first study reporting a statistical significant prognostic relevance of TAMs. A potential limitation of our study is the retrospective design with a relatively small number of patients with advanced disease or lymph node metastasis. This low rate can be attributed to the well known fact that at diagnosis approximately 75% of women have a tumor confined to the uterus [40]. However, the amount of LVI, a process that takes place prior to the spread of tumor cells to the lymph nodes, was adequately high and correlated significantly with the presence of TAMs. As of the retrospective nature of our study further prospective research is needed to confirm the prognostic role of TAMs in EC.

 

Taken together, the balance of inflammatory mechanisms in EC, particularly in advanced and high-grade stages, appears to be polarized to an immunosuppressive milieu that is characterized by a pronounced accumulation of CD163+ cells. These TAMs promote immune escape, lymphatic sprouting, LVI and lymphatic metastasis and thereby affect recurrence and survival. Thus, to improve the outcome in advanced and relapsed tumors new therapeutic agents that target TAMs might be of high benefit. Future functional studies of TAMs in EC would allow understanding the tumor-promoting co-evolution of cancer and myeloid cells ultimately resulting in immunotherapeutic approaches that are directed toward the lymphangiogenic signaling in TAMs and most likely used as adjuvants along with standard cytotoxic regimens.