In this retrospective study we determined the density of several immune cell types: OX40+ activated T lymphocytes, FOXP3+ regulatory T cells, DC-LAMP+ mature dendritic cells and CD123+ plasmacytoid DCs in sentinel lymph nodes of patients with malignant melanoma. Density values of each cell type in SLNs vs. lymph nodes obtained from block dissection, as well as those in tumor-free vs. metastatic SLNs were compared. Furthermore, results were evaluated with regard to associations with patient and tumor parameters and the outcome of the disease.
In our cohort of 100 sentinel nodes, FOXP3+ lymphocytes proved the most abundant cell type, greatly outnumbering OX40+ activated T cells. This finding could be considered a sign of immunosuppression. However, OX40 is a very specialized marker expressed primarily by recently activated CD4+ T lymphocytes showing preferential accumulation at tumor sites in cancer patients and is thought to represent tumor antigen specific T cells [26, 27]. Since it is not a general marker of all activated T cells, the expression of OX40 is expected to be restricted. Moreover, most immune cell types studied were present in higher amount in the sentinel nodes compared to 37 non-sentinel nodes from a subset of the patients, not only FOXP3+ Tregs but also OX40+ activated T cells and DC-LAMP+ mature DCs.
Our results were mainly based on interindividual comparisons of labeled cells in SLNs vs. NSLNs. We did not perform statistical analysis comparing SLN-NSLN pairs of the same patients because of the low number of cases with available NSLN samples. Moreover, since it was a retrospective study, we did not have access to non-sentinel nodes of sentinel-negative patients. On the other hand, in our study, similarly to others comparing sentinel and non-sentinel nodes from the same patients, NSLNs derived from the same lymph node basin as the SLN. Therefore, they were likely also affected by the primary tumor albeit to a lesser degree than the SLN. Nevertheless, our results indicate a state of functional activity in the sentinel lymph nodes, or at least no unambiguous signs of immune suppression compared to NSLNs.
The question whether sentinel nodes are immunologically competent or suppressed is a matter of debate, with no consensus reached yet. According to Cochran’s group, SLNs draining melanoma and breast cancer show evidence of immunosuppression, with lower number of paracortical dendritic cells (DCs) and lower expression of costimulatory molecules compared to NSLNs [2, 3]. Other studies examining various DC markers, however, found either higher amount of DCs in SLNs or no difference from NSLNs or control lymph nodes in melanoma, breast, gastric or oral cancers [4–6, 9–12] (Table 1). One reason for this discrepancy could be the relatively small sample size in many of the studies comparing SLNs and NSLNs, hampering the drawing of reliable conclusions. Also, while some of these investigations evaluated metastatic and tumor-free SLNs separately, others did not make such distinction which makes the results difficult to compare. Nevertheless, most comparative studies mentioned above, including the few involving higher case numbers, seem to indicate that sentinel lymph nodes are not suppressed (Table 1). It should be noted, however, that signs of immune activation and those of dysfunction can be detected at the same time in some of the studies, including ours. Moreover, the balance of activation and suppression most probably changes with time, parallel to tumor progression.
Analysis of our results according to tumor positivity of sentinel nodes resulted in no significant difference in the case of three of the four markers studied, including FOXP3, in contrast to some [14, 15], but not all  other studies on melanoma SLNs (Table 1). The reason for this discrepancy is not known, but may partly be explained by methodological differences. On the other hand, in our present study positive nodes contained higher amount of OX40+ activated T lymphocytes compared to negative ones, which may be related to in situ activation of T cells recognizing tumor antigens. In accordance with this finding, a previous report on axillary lymph nodes of breast cancer patients revealed preferential accumulation of OX40+ cells only in positive nodes .
We did not find significant difference between metastatic and tumor-free SLNs in the density of DC-LAMP+ mature dendritic cells or CD123+ plasmacytoid DCs. Results concerning dendritic cell subtypes in SLNs are contradictory in the case of melanoma and other tumor types as well (Table 1). A recent paper described decreased number of CD11c+ dendritic cells but elevated amount of CD86+ mature DCs in positive compared to negative sentinel lymph nodes in melanoma . In breast cancer, mature DCs detected by the CD83 or DC-LAMP marker were present in lower amount in tumor-positive SLNs compared to tumor-negative ones while the density of DCs stained for CD1a was similar [9, 12]. On the other hand, there was no difference in the number of either S100+ or DC-LAMP+ DCs in metastatic vs. tumor-free SLNs in gastric carcinoma [11, 13]. The prevalence of plasmacytoid DCs (pDCs) was found more prominent in tumor-containing SLNs in one study on melanoma  while no difference was detected between metastasis-positive and -negative SLNs in another  or in breast cancer .
In our patient cohort, density of DC-LAMP+ mature DCs, CD123+ plasmacytoid DCs and OX40+ activated T cells in sentinel nodes did not show significant association with disease progression or survival. No previous studies have been reported on a potential prognostic effect of the amount of pDCs or activated T cells in sentinel nodes in any tumor types. In a study on tumor-positive SLNs in melanoma, the accumulation of DC-LAMP+ DCs was shown to have a statistically significant but weak impact on survival, and did not prove independent prognostic factor . On the other hand, infiltration of primary melanomas by DC-LAMP+ mature dendritic cells and OX40+ activated T cells proved strong independent predictors of survival according to our previous studies [21, 22]. The above findings on the more prominent prognostic effect of these cell types at the site of the primary melanoma may indicate the importance of immune cells residing in the primary tumor in the development of antitumor immune response.
On the other hand, the amount of FOXP3+ cells infiltrating primary melanomas did not show association with the outcome of the disease according to our previous study . In contrast to these results, our present findings indicate that quantification of FOXP3+ cells in SLNs may provide information with prognostic impact. High mean density of these cells was associated with significantly shorter progression-free and overall survival. A novel, intriguing finding of our study is that high density of FOXP3+ cells in sentinel nodes was associated with disease progression and shorter survival only in cases with positive SLN status. Moreover, univariate and multivariate Cox regression analyses evaluating mean cell density values as continuous variables showed associations with survival only in the case of positive SLN status and not in patients with negative SLNs or in the whole patient group. A recent study by Speeckaert et al., not evaluating cases with different SLN status separately, also could not prove the independent prognostic effect of FOXP3+ cells in sentinel nodes of melanoma patients . Although the sentinel-negative but progressing cases in our study represented a minority of the samples (7 patients), which in theory might have influenced the result of such comparisons, differences in PFS and OS were much larger in the SLN-positive group compared to those in the whole cohort. These results indicate that Tregs may exert their suppressor function more efficiently in metastatic nodes, at least at a degree that would be reflected in influencing the outcome of the disease. Reasons for this difference could include the relative abundance of Tregs in positive nodes as suggested by some studies [8, 13–15], or an elevated activity of tumor-specific immune reactions in the presence of tumor antigens.
Priming of antitumor T-cell responses is generally believed to take place primarily in draining lymph nodes via cross presentation of tumor antigens by dendritic cells capturing antigens at the primary tumor site and migrating to lymph nodes. However, in experimental models direct priming of T lymphocytes by tumor cells in the lymph nodes has been described [28, 29]. In melanoma patients, precursor frequency of peripheral blood CD8+ T cells recognizing melanocyte differentiation antigen epitopes was found increased in stage III and IV vs. stage I and II, indicating the importance of the development of lymph node metastases in triggering T-cell mediated antitumor immunity . On the other hand, it has also been shown that T-cell activation by dendritic cells may occur extranodally, at tumor sites . In accordance with this finding, a study analyzing T-cell clonotypes revealed that the majority of T cells in tumor-negative SLNs are not clonally expanded , in contrast to primary melanomas and metastatic lesions showing the presence of multiple clonotypic TCR transcripts [32, 33]. In the light of these data, the results of our present and previous studies could be interpreted as suggesting the prognostic importance of immune markers in tumoral compartments of melanoma patients, both in the primary tumor and in metastatic sentinel lymph nodes.