The other mechanism involved is represented by soluble immunosuppressive factors in the aqueous humor
[38]. They are: transforming growth factor (TGF)-β, α-melanocyte-stimulating hormone (α-MSH), calcitonin gene-related peptide (CGRP), vasoactive intestinal protein (VIP), and indoleamine 2,3 dioxygenase (IDO). TGF-β inhibits the activation of macrophages, T lymphocytes, and natural killer (NK) cells and enhances the tolerance of antigen-presenting cells (APCs)
[34]. TGF-β is also required for CTLA-4 up-regulation on CD8
+ T cells. CTLA-4 stimulation leads to T cell inactivation and generation of T regulatory (Treg) cells
[39]. α-MSH reduces neutrophil activities and stimulates Tregs. α-MSH and CGRP downregulate the production of pro-inflammatory factors. VIP inhibits NK-cells mediated cytolysis. VIP and IDO inhibit T cell activation
[34].
When antigens enter the eye, a unique mechanism of immune privilege is involved: it is termed “anterior chamber-associated immune deviation” (ACAID), an immunomodulatory phenomenon involving the eye, but also the thymus, spleen, and sympathetic nervous system
[41]. In mouse models an antigen injected into the anterior chamber of the eye is captured by ocular APCs. Then, these APCs migrate to the spleen and thymus, inducing immunomodulatory cells, such as Tregs, regulatory B cells, and NK T cells, which cause immune deviation. The sympathetic nervous system promotes the maintenance of a functional ACAID
[42].
3. Immune Infiltrate in Uveal Melanoma
The crosstalk between tumor and microenvironment influences the inflammatory response: cancer cells interact with both the innate and the adaptive immune system and use immune cells for tumor survival and protection from immunological attacks. The main immune cells in uveal melanoma are the M2-type macrophages, which foster tumor growth through angiogenesis and immunosuppression
[43][44].
Monocytes/macrophages of M2 lineage exceed T cells in tumor-infiltrating immune cells. They secrete anti-inflammatory cytokines (IL-10 and TGF-β), which inhibit dendritic cell activation, as well as T- and NK-cell functions
[35][43] (
Figure 1). Treg and UM cells trigger inflammation and M2-polarization through the production of factors, such as CCL22, CCL2, VEGF, M-CSF, TGF-β, IL-6, IL-10, CCL17, CCL22, PGE2, and endothelial monocyte-activating polypeptide (EMAP)-II
[35][43][45]. Moreover, M2-macrophages can secrete soluble factors, which enhance the invasive capabilities of neoplastic cells, such as the melanoma inhibitory activity (MIA), a growth-regulatory protein, which inhibits the cellular adhesion to the extracellular matrix
[46].
Figure 1. Tumor infiltrating immune cells in primary tumor (left panel) and in liver metastasis (right panel). In primary uveal melanoma (UM), CD8+ T cells and fewer CD4+ T cells are present, but M2 polarized macrophages are predominant. M2 macrophages stimulate Tregs through IL-10, CCL2, and CCL17. M2 macrophages suppress CD8+, NK, and DC secreting TGF-beta and IL-10. In liver metastasis, CD8+ cells surround but do not infiltrate the tumor mass, while CD4+ cells are present in perivascular aggregates. Furthermore, Tregs are stimulated by TGF-beta and IL-10 produced by MDSC.
The analysis of uveal melanoma suspensions has revealed that the majority of tumor infiltrating lymphocytes (TILs) are CD8
+ T cells with fewer CD4
+ T cells
[47]. Lagouros reported a limited number of Tregs in primary UM
[48]. Nevertheless, the count of Tregs within primary tumor correlates with the development of systemic metastases. In addition, Tregs and cyclooxygenase-2 (COX-2) expression in primary tumors are associated with a poor prognosis
[49]. Tregs are recruited by IL-10, chemokine (C–C motif) ligand (CCL) 17, and CCL22 produced by M2 macrophages
[35].
Tumor-associated macrophages (TAMs) are a negative prognostic factor for UM
[50]. Indeed, TAMs are associated with highly malignant tumors characterized by negative prognostic features, such as epithelioid cells, high microvascular density, intense pigmentation, and larger size
[51]. TAMs seem to promote tumor growth through angiogenesis and metastatic spread
[50][52].
In primary UM, the expression of HLA class I and II exerts a prognostic role
[53]. It has been reported that, in primary tumors, down-regulation of HLA class I is a mechanism for evading CD8
+ cell cytotoxicity. Therefore, tumors should be more sensitive to NK cells
[54][55][56][57]. Nevertheless, only 50% of primary UM expresses MHC class I related chain (MIC) A and B, which are the ligands for NK cell receptor (NKG2D). In UM metastases, MIC A and B are absent. Consequently, the activity of NK cells in metastatic UM is limited. Vetter reported a single case of MIC expression induced after chemotherapy, suggesting a possible role for immunotherapy following cytotoxic therapy
[54]. It has also been proved that UM cells are capable of producing the macrophage migration inhibitory factor (MIF), a cytokine that inhibits cytolytic activity of NK cells, contributing to tumor growth and metastatic spread
[58][59][60]. The expression of FasL is a further mechanism in UM explaining the escape from NK cells
[61]. Additionally, TILs and TAMs of UM produce IL-2 and IL-15, which bind receptors on tumor cells, promoting UM cell growth and reducing sensitivity to NK cell activity
[62].
Moreover, the loss of the maturation marker CD-40 on APCs has been observed in primary UM. This lack does not allow correct lymphocyte T-mediated anti-tumor activity because of an inadequate functioning of APCs to induce T cell activation
[63].
Overall, “inflammatory phenotype” has been proposed to define UM with infiltrating macrophages and lymphocytes in addition to a high expression of HLA class I and II molecules. It identifies tumors with a worse prognosis
[43][44].
It is doubtless that the immune cells in the UM inflammatory phenotype do not stimulate an antitumor response. They contribute to angiogenesis, immunosuppression, tumor growth, and metastatic spread
[64]. Moreover, comparing the circulating immune cells between primary and metastatic UM, a weaker immune surveillance has been spotted in case of metastases. Lower circulating CD3
−CD56
dim NK cells, CD8
+, and NK T cells, as well as an increase in Tregs and MSDCs have been detected during the metastatic phase. Furthermore, metastases are associated with increasing plasma levels of several miRNAs involved in immune regulation
[65].
In the metastatic sites, CD4
+ cells are present in perivascular aggregates, while CD8
+ cells are scarce and mainly surround the liver metastases
[66][67][68][69]. Tumor cells and resident hepatic cells recruit in metastatic sites two different types of myeloid-derived suppressor cells (MDSCs): monocytic MDSCs and polymorphonuclear MDSCs. Monocytic MDSCs, more frequent than polymorphonuclear MDSCs in metastases, promote neoplastic growth after their differentiation towards TAMs. MDSCs are able to induce Tregs releasing IL-10 and TGF-β
[35][66] (
Figure 1).
Nevertheless, Rothermel found that a subset of TILs associated with a lack of melanin pigmentation is more effective in antitumor response, similarly to TILs in cutaneous melanoma
[70].
3.1. The Role of PD-1/PD-L1 Interaction in UM
UM cell lines are able to suppress T-cell activation by the expression of PD-L1 (B7-H1)
[71][72]. However, in vivo a limited constitutive expression of PD-L1 on tumor cells and PD-1 on TILs has been proven in metastatic uveal melanoma compared with cutaneous melanoma
[73]. As a matter of fact, in the metastatic sites, only 5% of uveal melanoma shows the expression of PD-L1, whereas PD-1 is expressed in about 51% of TILs
[74]. Regarding primary UM, 40% of PD-L1 expression on tumor cells is reported in patients with metastatic disease
[75].
PD-L1 expression indicates an active interaction between the tumor and the adaptive immune cells. PD-L1 expression is induced by IFN-γ produced by activated CD8
+ cells. Consequently, PD-L1 on tumor cells depends on the presence of activated TILs PD-1
+ [74]. The lower PD-L1 expression in UM is not due to a loss of function. Indeed, it is known that uveal melanoma cells do not lose the ability to upregulate PD-L1 in response to IFN-γ. In a preclinical model, PD-L1 is able to inhibit T cell proliferation through a lower secretion of IL2
[76].
Among PD-1/PD-L1 patterns, in UM samples the two more frequent patterns are PD-1
−/PD-L1
− and PD-1
+/PD-L1
−, representing immunological tolerance, with, respectively, absence or functional suppression of TILs in the tumor microenvironment. Differently, in cutaneous melanoma the dominant subgroup is the PD-1
+/PD-L1
+, resulting in immune-competent and active TILs. This subgroup seems to be absent in UM. TILs with antitumor function are less frequent than regulatory TILs in UM, resulting in immune tolerance both in the primary site and metastases.
[74]. PD-1/PD-L1 axis is probably not one of the most relevant mechanisms to avoid immune response in uveal melanoma. This hypothesis can explain the poor response to anti-PD-1 therapy.
3.2. IDO and Immune Escape
The enzyme indoleamine 2,3 dioxygenase (IDO) controls tryptophan degradation, influencing both the innate and the adaptive immune system for lymphocytes proliferation, activation, and survival. IDO is able to suppress T and NK cells, generate Tregs, and promote tumor angiogenesis.
In uveal melanoma IDO is not constitutively expressed either in primary or in metastatic cells. However, IFN-γ upregulates IDO mRNA and protein in UM cells, inducing the production of enzymatically and biologically active IDO
[77][78]. We might conclude that the induction of IDO by IFN-γ is a defensive mechanism of uveal melanoma cells in response to the presence of T lymphocytes and NK cells. Upregulation of the IDO gene and protein expression can contribute to an immune-privileged microenvironment promoting immune escape. In advanced cutaneous melanoma, epacadostat, an IDO1 inhibitor, in addition to anti-PD-1 treatment, does not improve the clinical outcomes of anti-PD-1 therapy alone
[79]. The effectiveness of IDO inhibitors in UM could be limited by the lack of a constitutive expression. Nevertheless, we cannot exclude a role of IDO inhibitors in selected patients or in special conditions inducing IDO expression (i.e., previous treatments).