Presence of Dendritic Cell Subsets in Sentinel Nodes of Breast Cancer Patients Is Related to Nodal Burden

BACKGROUND: Sentinel lymph nodes (SLNs) are both the first site where breast cancer (BC) metastases form and where anti-tumoral immunity develops. Despite being the most potent antigen-presenting cells, dendritic cells (DCs) located in a nodal tissue can both promote or suppress immune response against cancer in SLNs. METHODS: In SLNs excisions obtained from 123 invasive BC patients, we performed immunohistochemistry (IHC) for CD1a, CD1c, DC-LAMP, and DC-SIGN to identify different DCs populations. Then we investigated the numbers of DCs subsets in tumor-free, micrometastatic, and macrometastatic SLNs with the use of a light microscope. RESULTS: We observed that CD1c+ and DC-SIGN+ DCs were more numerous in SLNs with a larger tumor size. More abundant intratumoral DC-LAMP+ population was related to a higher number of metastatic lymph nodes. Conversely, more abundant CD1a+ DCs were associated with a decreasing nodal burden in SLNs and a lower number of involved lymph nodes. Moreover, densities of the investigated DC populations differed with respect to tumor grade, HER2 overexpression, hormone receptor status, and histologic type of BC. CONCLUSIONS: According to their subtype, DCs are associated with either lower or higher nodal burden in SLNs from invasive BC patients. These relationships appear to be dependent not only on the maturation state of DCs but also on the histological and biological characteristics of the tumor.


Introduction
Metastases in lymph nodes are one of the most important adverse prognostic indicators in invasive breast cancer (BC) patients [1]. Sentinel lymph nodes (SLNs) are the first lymphatic organs on the lymph flow from a primary tumor; therefore, they are regarded as sites where regional breast cancer metastastes occur first [1][2][3]. Recently, sentinel lymph node biopsy (SLNB) has displaced axillary lymph node dissection (ALND) in the management of primary operable and clinically node-negative BC, as it provides comparable staging information but spares many side effects associated with ALND without deterioration in patient survival rates [1,4]. Evaluation of nodal burden in SLNs is based on the size of the largest metastatic focus, which is assigned to one of three categories: isolated tumor cells (tumor diameter < 0.2 mm or < 100 cancer cells), micrometastasis (0.2-2 mm), or macrometastasis (>2 mm) [1,3]. The diameter of secondary tumor correlates negatively with patient survival rates [3].
Despite being the most likely location of BC spread, SLNs are the sites where the anti-cancer immune response develops. This process is initiated by dendritic cells (DCs), the most potent antigen-presenting cells. DCs loaded with tumor-derived molecules migrate from the primary tumor site to SLNs, where they are able to present antigens and subsequently activate T-cells [2,3,5]. The notion of SLNs as the site of activation of antitumoral immune response in BC was postulated by Poindexter et al, who observed higher levels of interleukin (IL) 10 and IL-12 in SLNs than in other uninvolved lymph nodes [5]. Higher numbers of DCs in tumor-free and metastatic cervical and endometrial cancer SLNs as compared to their non-sentinel counterparts backed up the hypothesis of SLNs as more immunologically active organs than other LNs [6]. However, anti-tumor immune reaction in SLNs is also modulated by factors produced by tumor cells and their microenvironment; this includes impairment of antigen presentation by DCs and effect on their maturation status [2,3,5]. When immature, the main function of DCs is antigen processing and maintenance of immune tolerance. After a contact with antigen and additional stimulation, DCs maturate, migrate from the peripheral tissue to lymphoid organs, and acquire the ability to induce immune response [7]. It was shown that maturation of DCs may be in part dependent on VEGF expression by breast cancer cells [8]. However, a study carried out in melanoma cases showed that even mature DCs retain their antigenprocessing functions [9].
To date, several subtypes of DCs have been distinguished. Although there is no universal marker for DC identification, functional and maturation status of DCs is determined by the set of expression of certain proteins [7] that allow approximate classification of DC subtypes. Mature DCs in cooperation with T helper and cytotoxic cells were shown to contribute to cancer cell killing. The proposed role of DCs in this process was supply of cytokines and other "contact signals" that launch anti-tumoral immunity [10]. On the other hand, prevalence of DC subtypes associated with humoral immune response and immunological tolerance was observed in LNs from non-small cell lung cancer patients and suggested as one of the mechanisms of tumor-escape. Moreover, reduction of DCs in patient blood (as compared with healthy donors) pointed to the systemic impact of cancerderived molecules on DC number [11]. Therefore, immune suppression is not confined to the primary tumor site, but may extend to the circulation and secondary lymphoid organs also [12]. It was suggested that immune suppression in SLNs may precede the establishment of nodal metastases [2] and that immunosuppression follows anti-tumoral response (regarded as maturation of DCs and activation of specific T-cells) in BC [13][14][15]. Thus, it is not surprising that DCs are considered as one of the targets of immunotherapeutic approaches for BC patients [2,7,13,15].
Recently we have shown that the content of respective DCs subtypes in BC tumors differs according to BC molecular subtype and clinicopathological features [16]. In this study, we investigated the densities of several DC populations in SLNs of invasive breast cancer patients with respect to the size of nodal metastasis and other prognostic parameters in this malignancy.

Results
Clinicopathological characterization of the study group is summarized in Table 1.

Relationships between Densities of DC Populations and Metastatic Burden of SLNs
When micro-and macrometastatic SLNs were compared, we found significantly higher densities of CD1c+ and DC-SIGN+ DCs at tumor margin of the latter (U Mann-Whitney test: p < 0.003 and p < 0.035, respectively; Figure 1B,C, Table 2). Investigation of positive SLNs revealed that higher intratumoral DC-LAMP+ cell densities were associated with occurrence of metastasis in more than one LN (0.75 ± 1.38 for >1 positive SLNs vs. 0.28 ± 1.04 for 1 positive SLN, p < 0.015; Figure 1E). With regard to type of metastasis in SLN, we observed that CD1a+ DCs tended to decrease their numbers in distant area from tumor-free SLNs, SLNs with micrometastases to SLNs with macrometastasis (p = 0.055; Figure 1A, Table 2). Moreover, there was a tendency towards more abundant CD1a+ DCs in distant area of tumor-free than in positive SLNs (63.25 ± 31.16 vs. 50.85 ± 29.46, p = 0.054, Figure 1D). The intratumoral area of macrometastases showed the most abundant infiltration of CD1a+, followed by DC-LAMP+, CD1c+, and DC-SIGN+ DCs ( Figure 2, Table 2).   box is mean ± standard error (SE) and whiskers are mean ± standard deviation (SD). (B-E) U Mann-Whitney test: the central point is arithmetical mean, box is mean ± 2xSE and whiskers are mean ± 0.95xSD. p-value < 0.05 was considered significant. micro-and macrometastatic SLNs, p = 0.002; (C) CD1c+ DC densities at tumor margin in micro-and macrometastatic SLNs, p = 0.031; (D) CD1a+ DC density in lymphoid tissue ("distant area") with reference to presence or absence of secondary tumor in SLN, p = 0.054; (E) intratumoral DC-LAMP+ DC density with reference to presence or absence of metastases in lymph nodes other than investigated SLN, p = 0.013. (A) Kruskal-Wallis ANOVA test: the central point is arithmetical mean, box is mean ± standard error (SE) and whiskers are mean ± standard deviation (SD). (B-E) U Mann-Whitney test: the central point is arithmetical mean, box is mean ± 2xSE and whiskers are mean ± 0.95xSD. p-value < 0.05 was considered significant.

Discussion
There is a great body of evidence that DCs are a heterogenous group of APCs with several subsets that differ in their function and role in malignant disease. The selected markers comprise different types of DCs regarding their maturational as well as functional status and can be used to approximately identify DC populations (Table 3).   CD1 encompasses a family of molecules that are responsible for the presentation of self and foreign lipid antigens on APCs. The respective members of CD1 family present a distinct repertoire of lipid antigens [26,43]. CD1a is localized mainly in early endosomes and in recycling compartments of DCs [43]. Expression of CD1a has been for a long time attributed to immature populations of DCs; however, a body of evidence showed that CD1a molecules can be found on DCs in general [2]. High expression of CD1a is a hallmark of Langerhans cells (LCs) [19]. In melanoma SLNs, CD1a+ LCs were proven to express molecules associated with mature phenotype [20]. In a study carried out by van de Ven et al., CD1a+ DCs turned out to be poor activators of T cells despite their mature phenotype. Moreover, this population in SLNs from melanoma patients was identified as skin-derived and migratory cells that eventually contribute to immune tolerance [18]. Accumulation of immature CD1a+ DCs in metastatic oral squamous cell carcinoma (OSCC) LNs was associated with both increase in regulatory T cells and a drop of cytotoxic cell numbers, supporting the hypothesis of the immunosuppressive microenvironment that in turn contributes to cancer spread [17].
It was shown that the cells of invasive BC attract immature CD1a+ DCs and their precursors, subsequently impair maturation of these cells, and eventually decrease their ability to activate T lymphocytes [21]. In contrast, higher densities of CD1a+ DCs both in the stroma of primary invasive tumors and in LNs were associated with absence of nodal metastases and good prognosis [22,23]. Similarly, in our recent study we showed that a higher density of intratumoral CD1a DCs was associated with longer progression-free survival in BC patients [16]. These would support the hypothesis that higher levels of DCs evoke functional anti-cancer responses [22]. Similarly, a decrease in CD1a+ cells in SLNs was associated with tumor involvement [24]. These findings are in accordance with our results, which pointed out an association between higher densities of CD1a+ DCs in lymphoid tissue and negative nodal status or lower metastatic burden in SLNs. Since such a difference was not observed by Kohrt et al. in other axillary LNs [24], it was postulated that alterations in immune profile are dependent on tumor burden in the sentinels exclusively. Furthermore, the authors suggested that changes in the immune profile occur before metastasis established and observed that decreased CD1a+ cells in non-sentinel LNs were associated with disease recurrence [24]. On the other hand, some authors did not observe any relationship between CD1a+ DCs density and metastatic status of SLNs [5,12,25].
Regarding other prognostic indicators in breast cancer, we noted that higher numbers of CD1a+ DCs in lymphoid tissue were associated with a lower tumor histological grade and HER2 negativity. This is partially in contrast with Poindexter et al.'s study, who found higher numbers of CD1a+ DCs in tumor-containing than in tumor-free SLNs from G3 cancers [5]. Correlation between CD1a level in LNs and expression of hormone receptors was also observed by some authors [22]. Although we did not make such observations, we noted that breast cancer metastases of lobular histology are more abundantly infiltrated by CD1a+ DCs than NOS cancers.
Little is known about the role of CD1c+ DCs in malignancies, particularly in BC tissues or SLNs. The CD1c molecule is expressed on Langerhans cells and some populations of B cell surfaces [26]. Inflammatory CD1c+ DCs are present in several tissues, including solid tumor infiltrate and lymph nodes [27]. In cancer patients CD1c+ cells were shown to represent DCs that are able to stimulate cytotoxic CD8+ and CD4+ helper T cells [27,28]. On the other hand, human blood CD1c+ DCs also were identified as producers of immunosuppressive and regulatory factors such as IL-10 and IDO in response to stimulation by E. coli bacteria [29]. CD1c+ DCs can be generated at the tumor site; their elevated numbers were noted in lung adenocarcinoma tumors as compared with normal lung tissue [30]. On the other hand, hepatocellular cancer (HCC) patient blood CD1c+ cells were identified as myeloid, IL-12 producing DCs, less abundantly represented in circulation of HCC patients than in healthy controls [31]. Similarly, lower numbers of epidermal CD1c+ DCs were found in invasive and in situ melanoma tissues in comparison with dysplastic nevi [32]. In non-small cell lung cancer, higher numbers of myeloid CD1c+ cells in tumor tissue were related to worse survival. The authors offered an explanation of tumor tissue acting like a trap, which attracts DCs but prevents their migration to LNs [11]. In our study, higher numbers of CD1c+ DCs located in close proximity to cancer islets were associated with greater diameter of metastasis as well as with HR-positive and HER2 overexpression. Thus, we propose that this DC subpopulation supports nodal spread of breast cancer and supposedly plays a more important role in progression of ER, PR, and/or HER2-positive cancers.
Lysosome-associated membrane protein (DC-LAMP) is expressed on mature DCs [2,20]. In melanoma SLNs, lower counts of mature DC-LAMP+ DCs were associated with unfavor-able prognostic factors such as ulceration. As their higher numbers were related to longer survival, maturation of DCs became regarded as guarantees for long-term antimetastatic protection [35]. As opposed to this, in primary melanoma lesions, the presence of DC-LAMP+ DCs in the peritumoral area was accompanied by the infiltrate of resting naïve T-cells that pointed to a lack of T-cell stimulation [20]. High DC-LAMP+ DCs numbers of mature morphology were observed in melanoma metastatic LNs and their density in SLNs negatively correlated with additional melanoma metastases in non-SLNs. Therefore, close proximity of DC-LAMP+ DCs to T cells and tumor cells was regarded as proof of active immune reaction [34]. In tumor-free LNs from OSCC patients, numbers of DC-LAMP+ cells were higher when cancers were non-metastatic [33].
In breast cancer, more dense DC-LAMP+ cell infiltration in SLNs was associated with a lack of nodal metastases [25]. The observed numbers of these DC populations were higher in tumor-free and metastatic SLNs than in other LNs, suggesting that SLNs are sites of more potent immune activity before and early in metastasis development [25,44]. The lower proportion of mature DC-LAMP+/immature CD1a+ cells in positive breast carcinoma SLNs as compared with negative ones suggested that the maturation of DCs is arrested [25]. Moreover, lower DC-LAMP+/cytotoxic cells and higher T regs/DC-LAMP+ DCs ratios in metastatic than in negative SLNs implied impaired antigen presentation, resulting in lower cytotoxicity against tumors as well as accumulation of T regs [25]. The study using another marker of DC maturity, CD83, showed similar expression of mature DCs in tumornegative SLNs and non-SLNs, but higher expression of co-stimulatory molecules in the latter, suggesting that immune response is suppressed before development of BC metastasis, but enhanced after the tumor is established in SLN [14]. We noted that higher infiltration of DC-LAMP+ cells in tumor islets and at tumor front is associated with more advanced nodal disease, favoring a hypothesis of insufficient or tumor-promoting immune response in BC.
Another molecule associated with DCs-DC-SIGN-is a C-type lectin receptor expressed on the surface of both immature and mature cells located in dermis and mucosal tissues [36]. Some authors associated its expression with the monocyte-derived [3] or interstitial [19] population of DCs. In primary melanoma tumors, expression of DC-SIGN and DC-LAMP was mutually exclusive [20]. This observation is in strong contrast to our results, which showed a positive correlation between these two populations in SLNs. Such an inconsistency can be explained by the notion that in peripheral tissue and in lymph nodes DC-SIGN+ represent immature and mature phenotypes of DCs, respectively [7]. However, the low expression of DC-SIGN in melanoma patient LNs suggested the loss of its expression during the maturation and migration of DCs [18]. Furthermore, we noted the positive correlation between numbers of immature CD1a+ and mature DC-LAMP+ DCs, implying distinct roles of DCs populations in lymphoid tissue or the primary tumor site. DC-SIGN is responsible for immune regulation of DCs as well as their adhesion, migration, maturation, and T cell activation. The molecule is one of the factors that contribute to immune escape of tumors [36]. DC-SIGN is also expressed on tumor-associated and monocyte-derived macrophages. Its expression can be induced by both cancer cells and fibroblasts. As a result of interaction between DC-SIGN+ macrophages and cancer cells, immunosuppressive cytokines are released from the former, which eventually promote tumor progression [38,39]. Likewise, differentiation of immunosuppressive DC-SIGN+ DCs in prostate cancer tissue was orchestrated primarily by factors derived from adjacent stroma [40]. After stimulation with LPS, DC-SIGN+ DCs were shown to dramatically increase the secretion of pro-tolerant cytokine IL-10 [37].
Information on the role of DC-SIGN+ DCs in invasive mammary tumors is modest. In primary breast tumors, immature myeloid DC-SIGN+ DCs were associated with worse survival [41] and early recurrence [42]. Since in OSCC intratumorally located immature dermal DC-SIGN+ cells were associated with worse survival, it was concluded that this DC subset fails to combat tumor effectively [33]. In primary cutaneous melanoma, immature dermal DC-SIGN DCs were increased in comparison with normal skin or naevi. In SLNs, these cells were also abundant and located primarily around high endothelial venules, although the accumulation of mature DCs was also observed [20]. In invasive BC SLNs, we observed relationship between higher DC-SIGN+ DCs infiltration of secondary tumor border and greater metastasis size, suggesting protumorigenic properties of the cells.
To sum up, we conclude that CD1a+ DCs show protective activity against cancer progression while CD1c+, DC-LAMP+ and DC-SIGN+ subsets favor tumor spread. Based on these, we might suppose that CD1a DCs are related to the better prognosis while CD1c+, DC-LAMP+, and DC-SIGN+ subsets are related to worse prognosis. Moreover, the balance between these populations can be at least partially dependent on biologic features of tumors.

Selection of Cases
The study group consisted of primary invasive breast cancer patients who underwent tumor excision with following SLNB and/or ALND at the University Hospital in Cracow. The patients who received neoadjuvant chemotherapy were excluded from the study. The study group included patients treated at our institution between 2010 and 2020; the criteria for qualification for neoadjuvant therapy significantly changed over this period, as more patients than nowadays are treated with systemic therapy post-surgery. Our group included also 2 patients with pT4 tumors excised palliatively to avoid unstoppable bleeding.
The archival hematoxylin-eosin-stained slides were re-evaluated and one representative, well-preserved specimen was chosen for immunohistochemistry. Nottingham Histologic Grade system was used for grading, and the 8th edition of AJCC system was used for staging [45].

Detection of SLNs and Identification of Nodal Metastases
Lymphatic mapping was performed using the double tracer technique: the conventional blue dye method (Patent Blue V) combined with the isotope technique [46,47].
The commercially available Nanocoll ® kit (human albumin 500 µm/vial; GE Healthcare, Chicago, IL, USA) was used for technetium marking. The administered 99 mTc Nanocoll solution was injected subcutaneously in the perialeolar region according to protocol, 3 to 18 h prior to the surgery.
Blue dye was injected periareolarly on a table, 10 to 15 min before surgery, followed by a massage of the injection site. During surgery, a hand-held scintillation counter (Gamma Finder ® II) was applied to identify SLNs. Any nodes with 10% or more of the ex vivo count of the hottest node and/or any nodes with at least one blue afferent lymphatic vessels derived from the breast were removed and designated as SLNs. A successful SLNB was defined as a procedure in which at least one true sentinel node was identified to be either blue and/or radioactive.
The tumor burden of SLNs was determined histologically based on the size of the largest metastatic focus, which was assigned to one of the three categories: isolated tumor cells (tumor diameter < 0.2 mm or < 100 cancer cells), micrometastasis (0.2-2 mm), or macrometastasis (>2 mm) and the number of involved lymph nodes. When several lymph nodes were identified as sentinels, the most representative one was selected on the basis of its size and tumor content.
For specimens with HER2 status 2+ in immunohistochemistry, fluorescence in situ hybridization (FISH) was conducted. FISH was performed using a PathVysion HER-2 DNA Probe Kit II (Abbott Molecular, Des Plaines, IL, USA) according to the manufacturer's protocol. The red Locus Specific Identifier (LSI) HER-2/neu and green Centromere Enumeration Probe (CEP 17) signals were counted on a fluorescence microscope equipped with specific filter sets and HER-2/neu to CEP17 ratio > 2.0 was considered as HER2/neu amplification [48].
Positive ER and PR expression thresholds were set when ≥1% of neoplastic cells showed positive immunostaining. The threshold for discriminating between low and high Ki67 expression was set at ≥20% of positive cells. Scoring of the HER2 staining was performed by a standard method [48].

Evaluation of DCs Densities in SLNs
First, the slides were scanned at low power (magnification of 100×) on a Nikon Labophot-2 optical microscope (Tokyo, Japan) in the search of the areas of highest infiltration of positively-stained cells. Then, DC populations were counted in 3 to 5 high-power fields (magnification of 400×, HPF) in "hot-spots" and their numbers were averaged. The densities of DCs were assessed in 3 compartments of SLN: within the islets of metastatic tumor ("intratumoral"), at the border of metastasis ("tumor margin"; no farther than 1 HPF from tumor edge), and in the "distant area", located >1 HPF from the edge of metastasis. In tumor-free SLNs, the DC densities were evaluated in lymphoid tissue (termed "distant area").

Statistical Analysis
Distributions were tested for normality with the Chi-square test. To assess the differences between groups, ANOVA Kruskal-Wallis and U Mann-Whitney tests were performed. The correlations between groups were evaluated by using the Spearman rank test. All analyses were performed using Statistica 13 (StatSoft Inc., Tulsa, OK, USA). In brackets,