Different Subsets of Tumor-Infiltrating Lymphocytes Correlate with Macrophage Influx and Monosomy 3 in Uveal Melanoma

Abstract

Purpose.: In contrast to many other malignancies, in uveal melanoma (UM) the presence of an immune infiltrate is associated with a bad prognosis. An analysis of the different functional phenotypes of tumor-infiltrating leukocytes (TIL) and a comparison with the genetic background of the tumors may help to explain this apparent anomaly.

Methods.: We performed a comprehensive immunohistochemical study by evaluating the density of CD8⁺ and CD4⁺ T lymphocytes, forkhead box p3 (Foxp3⁺) regulatory T cells (Tregs), and CD68⁺ and CD68⁺CD163⁺ macrophages in 43 cases of UM in relation to tumor characteristics. Expression of the chemokines CCL2, CCL17, and CCL22 in cultured human UM cells and peripheral blood monocytes was analyzed by quantitative PCR (qPCR).

Results.: The presence of TILs was highly variable between tumors and was dominated by CD8⁺ T cells with fewer CD4⁺ T cells and Tregs. When tumors were infiltrated by immune cells, the infiltrate generally comprised all different subsets of lymphocytes (P < 0.001) and M2 macrophages (P < 0.001). Different T-cell ratios did not influence clinical outcome. In addition, the presence of TIL correlated with the loss of one chromosome 3 (P < 0.04). UM cells express CCL2 and CCL22, two chemokines known to mediate trafficking of immune cells to the tumor.

Conclusions.: All studied subtypes of tumor-infiltrating immune cells were collectively increased and showed an association with monosomy of chromosome 3 suggesting that tumor intrinsic factors control the leukocyte influx, possibly through local chemokine secretion.

Introduction

Uveal melanoma (UM) is the most common form of cancer in the eye of adults, and is a highly malignant disease. Regarding prognosis, Callender's ¹ description revealed that an epithelioid cell type is associated with an unfavorable outcome. Over time, different molecular techniques to determine prognosis of uveal melanocytic neoplasms have been developed. ² A specific cytogenetic profile (i.e., the presence of monosomy of chromosome 3 and other cytogenetic markers) is related with a poor prognosis. ³ Furthermore, a specific gene expression profile, based on a molecular characterization of primary UM, can also be used to determine prognosis. ⁴ All these (intrinsic) properties help to differentiate between UM with a more or less favorable clinical course of disease. 

Evaluation of the prognostic significance of tumor-infiltrating leukocytes (TIL) in other human cancers revealed that increased numbers of TIL are often associated with a better prognosis. ⁵ In UM, the presence of substantial numbers of leukocytes does not represent an effective immunological antitumor response. Rather, a pronounced infiltration of UM by, for example, lymphocytes, is associated with a poor prognosis. ^6,7 This may be related to the immunoregulatory influence of the intraocular microenvironment. ⁸ For instance, similar to the situation in experimental anterior chamber associated immune deviation (ACAID), human ocular tumor cells may escape from the eye and induce regulatory T cells (Tregs) that subsequently suppress the local and systemic immune response against the tumor cells. Tregs are a subset of CD4⁺ T cells and are required for the maintenance of self-tolerance, and they are broadly characterized by their expression of the nuclear transcription factor forkhead box p3 (Foxp3). Increased numbers of these immunosuppressive lymphocytes have been detected in a variety of malignant tumors. ⁹ In UM, Tregs have been described in 12%–24% of cases. ^10,11 In these studies, the presence of monosomy 3 was not analyzed, or extrapolation of trends was limited due to the small number of tumors found to contain Tregs. However, in 50 COX-2-positive tumors, the presence of Foxp3⁺ cells was an independent predictive factor for worse overall survival. ¹¹  

Another nonmutually exclusive option might be the regulation of local immune responses by macrophages. In UM, an unfavorable prognosis is also associated with an increased density of CD68⁺ macrophages. ¹² Tumor-associated macrophages (TAMs) are involved in the regulation of angiogenesis, which can promote tumor growth. ¹³ On the other hand, macrophages that have less immune-stimulating molecules can induce Tregs, thereby suppressing tumor immunity. Vice versa, Tregs may support the differentiation of monocytes to tumor-promoting M2 macrophages. ¹⁴ In humans, M2 macrophages are characterized by a higher expression of CD163 and the majority of macrophages in human UM carry this characteristic. ¹⁵ In a murine ocular tumor model, intratumoral accumulation of M2 macrophages was shown to foster tumor growth in elderly mice. ¹⁶  

In order to gain a more profound insight in the interplay between the different immune cells in UM and their influence on the clinical outcome of disease, we determined the presence of different infiltrating immune cells, by counting the number of subtypes of intratumoral T cells and macrophages in a cohort of 43 primary tumors, which were characterized for chromosome 3 status. Our data indicate that all different types of immune cells studied were collectively increased in the presence of monosomy 3, leading to the conclusion that, specifically tumors with poor prognosis, support the influx of immune cells, including those with an immunosuppressive function such as Foxp3⁺ Tregs and M2 macrophages. 

Methods

Study Population

Tissue specimens were obtained from 43 patients who had undergone a primary enucleation for UM between the years of 1999 and 2004 at the Leiden University Medical Center (LUMC), Leiden, The Netherlands. Patient data and survival were updated until December 2010 from the patients' charts and from the database of the Integral Cancer Center West. Survival was termed the interval between enucleation and death from UM or the interval between enucleation and the last observation for surviving patients. The research protocol followed the current revision of the tenets of the Declaration of Helsinki. The present study was performed with the same group of tumors as described previously. ¹⁵  

Immunohistology

Enucleated eyes were fixed in 4% buffered neutralized formalin for 48 hours. After embedding in paraffin, 4-μm serial sections were made and mounted on a slide. Hematoxylin-eosin–stained 4-μm sections were reviewed by one ocular pathologist for pathological diagnosis and evaluated for histologic parameters, which included largest basal diameter (in millimeters), prominence (in millimeters), cell type according to the modified Callender classification, ¹⁷ ciliary body involvement, and intrascleral tumor growth. These parameters were used for classification in the TNM category/stage. ¹⁸  

For HLA class I staining, we used the mouse monoclonal antibodies HCA2 and HC10 (anti–HLA-A and anti–HLA-B/C, respectively), while for class II/ HLA-DR we used HLA-DR (Tal.1B5) antibody, as described previously. ¹⁹ The number of HLA-positive cells was estimated at 100× magnification and expressed as a percentage of the total number of tumor cells. 

Fluorescent Immunostaining of Tumor-Infiltrating Leukocytes

Phenotypic characterization of lymphocytes was performed using triple fluorescent immunostaining. A previously developed technique for simultaneous immunofluorescence (IF) staining of different epitopes was applied to 4-μm formalin-fixed, paraffin-embedded tissue sections. ²⁰ In brief, deparaffinized and citrate antigen retrieval-treated sections were stained by a mixture of the antibodies ab828 (rabbit polyclonal, anti-CD3; Abcam, Cambridge, MA), 4B11 (mouse monoclonal IgG2b, anti-CD8; Novocastra, Valkenswaard, The Netherlands), and the anti-FoxP3 antibody (IgG1, clone 236A/E7; Abcam) for the detection of Tregs. As secondary antibodies to visualize the lymphocytes, we used a combination of fluorescent antibody conjugates (goat anti-rabbit IgG-Alexa Fluor 546, goat anti-mouse IgG2b-Alexa Fluor 647, and goat anti-mouse IgG1-Alexa Fluor 488; Molecular Probes, Invitrogen, Breda, the Netherlands). Counting the number of CD3⁺CD8⁻ cells by this technique is a proven good marker for the presence of CD4⁺ T cells. ²¹ We did not stain for NK cells as they are rare in UM. ²² Images were captured with a confocal laser scanning microscope (LSM510; Carl Zeiss Meditec, Jena, Germany) in a multitrack setting. All images were stack size 368.5 × 368.5 μm. A microscope objective (PH2 Plan-NEOFluar 25x/0.80 Imm Korr; Zeiss) was used. Ten images were scanned per slide, and each scan represented one square optical field (area, 0.137 mm²): positive cells were counted in these randomly selected, high-power (250×) fields by two of the authors (IHGB and THKV). The mean was calculated and the SDs were analyzed; these showed no outliers. Counts of intratumoral infiltrating lymphocytes were represented as the number of cells per millimeters squared. 

Macrophages in these tumors were identified and measured using double IF staining, with mAbs directed against CD68 and for the M2 type, CD163, as performed and described previously. ¹⁵ The amount of staining was objectively determined in pixels per millimeters squared by the image analysis software program, Stacks (Department of Molecular Cell Biology, LUMC, Leiden, The Netherlands). 

Chromosome 3 Status

Analysis of chromosome 3 status (standard cytogenetic analysis and fluorescence in situ hybridization on isolated nuclei) of this patient material was described previously. ¹⁹  

Quantitative PCR of CCL2, CCL17, and CCL22

Tregs (and other immune cells) are known to be recruited into the tumor site through certain chemokines released by tumor and surrounding cells. Therefore, we analyzed expression of chemokines CCL2, CCL17, and CCL22 in UM cell cultures and monocytes, as extracting macrophages from fresh UM tissue has proven to be a technical challenge. Fresh tissue from four tumors, obtained immediately after enucleation, was placed in Amniochrome Pro Medium (Lonza Group Ltd., Basel, Switzerland) to develop a primary cell culture. Peripheral blood mononuclear cells (PBMC) were isolated from peripheral blood of three control subjects using ficoll-amidotrizoaat. CD14⁺ cells were magnetically labeled with CD14 MicroBeads and separated according to the manufacturer's protocol (Miltenyi Biotec, Auburn, CA). RNA was extracted with an RNeasy Mini Kit (Qiagen, Valencia, CA). Primers were designed with Beacon Designer (Biosoft, Palo Alto, CA). Quantitative PCR (qPCR) was performed in duplicate according to our standard laboratory protocol, as described previously. ²³ Glyceraldehyde 3-phosphate dehydrogenase (GAPDH), β-actin, ribosomal protein (RPL) 13, and ribosomal protein S (RPS)11 were initially included, and RPL13 and RPS11, as determined with the geNorm software (qBase, Bio-Rad Laboratories, Inc., The Netherlands), were selected as suitable reference genes. The calculated values were the normalized values of each sample. 

Statistical Analysis

All statistical analyses were performed with a statistical software program (PASW Statistics 17.0; SPSS Inc., Chicago, IL, US). Wilcoxon rank sum test (Mann-Whitney U test) for nonparametric analysis was used for determining associations of the number of TILs with clinical variables. In statistical correlation analysis, the Spearman's rank test was used. Cumulative survival rate was calculated by the Kaplan-Meier method and analyzed by the logrank test. Univariate Cox proportional models were used to determine the hazard ratio (HR). 

Results

Patients

At the time of enucleation, the mean age of the 43 patients within our study group was 60 years (range 27–88 years). At the end of the follow-up period, 17 patients had died from UM metastases, 23 were alive, and two died of causes unrelated to the primary disease but showed no evidence of metastases; in one case, the cause of death was unknown. The mean follow-up at the time of analysis was 92 months (range 62–133 months). 

Subtypes of Tumor-Infiltrating Lymphocytes

For a comprehensive analysis of TILs, we performed triple IF staining, using anti-CD3 (red), anti-CD8 (blue), and anti-Foxp3 (green) monoclonal antibodies. We identified and measured the number of CD8⁺ cytotoxic T cells (CD3⁺CD8⁺T cells), CD4⁺ helper T cells (CD3⁺CD8⁻Foxp3⁻T cells), and Treg (CD3⁺CD8⁻Foxp3⁺) cells on sections from 43 primary UMs. In Figure 1, an example of the analysis of CD3⁺ (red), CD3⁺CD8⁺ (purple) T-cell, and CD3⁺Foxp3⁺ (red with green nucleus) infiltration by confocal microscopy is shown. CD8⁺ Foxp3⁺ T cells were rarely present, and were therefore not enumerated. In general, all tumors contained all the different subtypes studied, but the number of infiltrating T cells varied enormously between tumors (range, 1–1834 per mm²) (Table 1). CD3⁺CD8⁺ cells were identified in all samples with a mean score of 163 per mm² (range 1–1566 positive cells). CD3⁺CD8⁻ (CD4⁺) cells were identified in 39 samples (91%) with a mean score of 42 per mm² (range 1–268 positive cells). Naïve CD4 T cells may differentiate into one of several lineages of T helper (Th) cells (including Th1, Th2, Th17), and into Tregs, as defined by their pattern of cytokine production and function. These cells can influence their environment. By combining three antibodies in one staining experiment, we were able to ensure the Tregs were not missed. Separated on the basis of their Foxp3 expression level, the CD4⁺ cells consisted of Foxp3⁺ cells, identified in 26 samples (61%), with a mean score of 20 per mm² (range, 1–158), and Foxp3⁻ cells (defined by exclusion), which were identified in 39 samples (91%), with a mean score of 22 per mm² (range, 1–151). Infiltration by Tregs was paralleled by other types of immune cells in the tumor; Spearman rank analysis revealed significant correlations between the number of Tregs and the number of CD4⁺Th cells (Spearman correlation coefficient (r) = 0.82, P < 0.001), as well as with the number of CD8⁺T cells (r = 0.89, P < 0.001). The number of Tregs was also associated with the number of macrophages detected by CD68 and CD163. Interestingly, the number of Tregs showed a stronger correlation with the number of CD68⁺CD163⁺ M2 macrophages (r = 0.81, P < 0.001) (Fig. 2) than with the total amount of CD68⁺ macrophages (r = 0.76, P < 0.001). All correlation coefficients are shown in Table 2. 

Figure 1. 

 

Analysis of T cell subsets in human UM. (A) Immunostaining with anti-Foxp3: intratumoral infiltration of a Foxp3⁺ cell (red nuclear staining) (original magnification, ×200). (B) Immunofluorescence staining using three antibodies directed against CD3 (red), CD8 (blue), or Foxp3 (green). Merged image: CD3⁺CD8⁺ Foxp3⁻ T cells are purple, CD3⁺CD8⁻Foxp3⁺ regulatory T cells show a nuclear green Foxp3 staining accompanied with surface red CD3 staining, and CD3⁺CD8⁻Foxp3⁻ helper T cells are red. (original magnification, ×250). (C) Distribution of CD3, CD8, and Foxp3 IF staining in 43 UMs. Infiltrating T cells were counted in 10 representative fields per tissue section and the number of cells per millimeters squared was calculated.

Figure 2. 

 

A comparison between the number of infiltrating regulatory T cells and M2 macrophages in 43 observations (logarithmic graph). A significant r value of 0.81 was seen, and the variability of M2 macrophages can be explained in approximately 65% (r ²) by the variability of the regulatory T cells.

Table 1. 

 

Baseline Characteristics of Patients and Histological Data of Primarily Enucleated Eyes

Categorial Variables	Baseline Data		Associations (P)
	N = 43	%	CD3+ Total	CD8+	CD4+ Total	CD4 Th	Foxp3+
Sex
Male	23	53	0.65	0.73	0.61	0.56	0.44
Female	20	47
Prognostic groups
Stage I	3	7	0.86	0.84	0.64	0.20	0.89
Stage IIA	9	21
Stage IIB	13	30
Stage IIIA	15	35
Stage IIIB	3	7
Cell type
Spindle	10	23	0.03	0.03	0.04	0.09	0.02
Mixed + epithelioid	33	77
Ciliary body involvement
Not present	25	58	0.19	0.32	0.11	0.16	0.046
Present	18	42

Numerical Variables		Baseline Data		Correlations (P)
		Mean	±SD	CD3+ Total	CD8+	CD4+ Total	CD4 Th	Foxp3+
Age at enucleation (years)		60	(±15)	0.60	0.86	0.39	0.71	0.26
Diameter (in mm)		13	(±3)	0.12	0.08	0.16	0.16	0.07
Prominence (in mm)		8	(±2)	0.57	0.46	0.72	0.97	0.44
CD68+ (pixels × 103 /mm2)		125	(±76)
CD68+CD163+ (pixels × 103 /mm2)		97	(±56)
CD3+ total (cells/mm2)		205	(±358)
CD8+ T cells		163	(±301)
CD3+CD8−(CD4+) T cells		42	(±64)
CD3+CD8−Foxp3− Th cells		22	(±34)
CD3+CD8−Foxp3+ Tregs		20	(±35)

 The numbers positively for CD3⁺ lymphocytes, consisting of CD8⁺ T cells, CD4⁺ Th cells, and CD4⁺Foxp3⁺ Tregs compared with other parameters. Only the P values are shown. P values for the comparison with categorical data were obtained by Wilcoxon rank sum test (Mann-Whitney U test), and for the numerical data by Spearman's correlation (two-tailed). P ≤ 0.05 are shown in italics.

Table 2. 

 

Correlation between Different Infiltrating Immune Cells (T Cells and Macrophages) as Well as with HLA Expression

		CD8	CD4	CD4 Th	CD4Foxp3	CD68	CD68CD163	HC10	HCA2	HLA-DR
CD3 total	r	.984	.959	.890	.917	.692	.769	.775	.440	.493
	P	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	0.003	0.001
CD8	r		.914	.839	.894	.663	.742	.758	.400	.467
	P		<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	0.008	0.002
CD4 total	r			.956	.929	.697	.765	.755	.451	.495
	P			<0.001	<0.001	<0.001	<0.001	<0.001	0.002	0.001
CD4 Th	r				.824	.607	.689	.665	.365	.405
	P				<0.001	<0.001	<0.001	<0.001	0.016	0.007
CD4Foxp3	r					.760	.807	.701	.435	.519
	P					<0.001	<0.001	<0.001	0.004	<0.001
CD68	r						.958	.639	.363	.517
	P						<0.001	<0.001	0.017	<0.001
CD68CD163	r							.652	.342	.528
	P							<0.001	0.025	<0.001
HC10	r								.746	.456
	P								<0.001	0.002
HCA2	r									.403
	P									0.007

r  = two-tailed Spearman correlation coefficient, with 43 observations.

Associations with Variables

There were significantly higher amounts of TILs in tumors containing epithelioid cells compared with spindle cells. The mean numbers of different TILs and associations with clinical parameters are described in Table 1. 

In the current group, on average, 35% of the tumor cells stained positive for HLA-B/C (HC10; range, 0%–100%), 42% for HLA-A (HCA2; range, 0%–100%) and 21% for HLA-DR (range, 5%–100%). A significant positive correlation was observed between the expression of HLA class I and II on the tumor cells and the numbers of infiltrating cells of all TIL subsets (r = 0.37–0.78; P < 0.02, Table 2). 

Loss of a copy of chromosome 3 was present in 26 of 43 tumors. This analysis revealed that tumors with monosomy 3 contained higher numbers of intratumoral CD3⁺ lymphocytes (P = 0.006), CD8⁺ T cells (P = 0.01), CD4⁺ T cells (P = 0.006), and its subtypes CD4⁺ Foxp3⁻ Th cells (P = 0.04) and CD4⁺Foxp3⁺ Tregs (P = 0.004) (Wilcoxon rank sum test), while TILs were sparingly detected in the tumors without loss of one chromosome 3 (Fig. 3). The mean numbers of different TIL in the different prognostic groups are described in Table 3. 

Figure 3. 

 

Tumors characterized by detrimental monosomy 3 contain significantly (all P < 0.05) higher numbers of different subtypes of intratumoral T cells (D3 = disomy for chromosome 3, and M3 = monosomy 3).

Table 3. 

 

UM Groups Based on Presence or Absence of Monosomy 3 Were Compared Regarding Different Types of TILs and Ratios between These Immune Cells

	Chromosome 3 Status
	Normal (n = 17)		Monosomy (n = 26)
	Mean	SD	Mean	SD	P
Immunofluorescence staining
CD3+ total (cells/mm2)	46	100	309	425	0.006
CD8+ T cells	26	53	252	360	0.01
CD3+CD8−(CD4+) T cells	20	47	56	70	0.006
CD3+CD8−Foxp3+ Tregs	4	11	30	42	0.004
CD3+CD8−Foxp3- Th cells	15	36	27	33	0.04
CD8+/regulatory T-cell ratio	6.2	3.4	9.2	7.8	0.22
CD4+/regulatory T-cell ratio	3.4	2.3	3.2	2.3	0.85
CD4+Foxp3− Th/Treg-cell ratio	3.0	2.1	2.4	2.5	0.11
CD8+/CD4+ T-cell ratio	2.7	2.4	3.6	2.7	0.33
CD8+/CD4+Foxp3− T-cell ratio	2.9	2.4	10.2	12.5	0.10
CD68+ macrophages (pixels × 103)	83	82	153	58	0.001
CD68+CD163+ M2 (pixels × 103)	68	61	116	44	0.002
Immunohistochemical staining
HLA-DR (%)	10	7	28	26	0.002
HC10 (%)	16	22	48	33	0.001
HCA2 (%)	22	26	54	25	0.001

 For the numerical parameters we show mean ± SD. P values for numerical data were obtained by Wilcoxon rank sum test (Mann-Whitney U). P ≤ 0.05 are shown in italics.

Survival Analysis

The Kaplan-Meier method and the logrank test were used to analyze the correlation between TIL subtypes, HLA class I and II expression, and patient survival. For logrank testing and Cox regression analysis, appropriate parameters were divided into categorical variables: TIL and macrophage groups were based on the median (50th percentile) of positive immune cells per millimeters squared; 25% was used as a cutoff point to dichotomize the HLA expression variable. A significantly worse patient survival was associated with a high total number of macrophages as M2 macrophages (P = 0.01), but not with increased numbers of Tregs. A trend towards significance for worse survival was seen with a high number of total CD3⁺ T cells (P = 0.07) and CD3⁺CD8⁺ T cells (P = 0.07). 

Previous studies in other malignancies show the relevance of the ratios between different immune cells for survival, ^21,24–26 indicating that especially the proportion between the different subtypes of immune cells within the tumor of each individual patient is important. When we evaluated the ratios between different TILs, no significant difference in survival was seen (data not shown). 

No significant association between HLA expression and survival was found. 

A significant association was seen between decreased survival and monosomy 3 (P < 0.001, logrank testing). P values and HRs are shown in Table 4. 

Table 4. 

 

Associations between Different Inflammatory Markers, Tumor Intrinsic Properties, and Survival in 43 Cases of UM

	Kaplan-Meier (LogRank Test)		Cox Univariate
	χ2	P	P	HR	95% Confidence Interval
Immunologic determinants
Immunofluorescence staining
CD3+ total	3.3	0.07	0.08	2.4	0.9–6.6
CD8+ cells	3.3	0.07	0.08	2.4	0.9–6.6
CD4+ (CD3+CD8−)	1.2	0.28	0.29	1.7	0.6–4.5
CD4+Foxp3+ Tregs	0.7	0.41	0.41	1.5	0.6–3.9
CD4+Foxp3− Th	3.1	0.08	0.09	2.4	0.9–6.4
CD68+ macrophages	6.0	0.01	0.03	5.2	1.2–22.9
CD68+CD163+ M2 macrophages	6.4	0.01	0.03	5.5	1.2–24.0
Tumorcel properties
Immunohistochemical staining
HCA2	1.9	0.17	0.19	2.1	0.7–6.6
HC10	1.5	0.22	0.23	1.9	0.7–5.0
HLA-DR	0.1	0.73	0.73	1.2	0.4–3.8
Monosomy of chromosome 3	13.5	<0.001	0.007	16.5	2.2–125.5

Expression of CCL22, CCL17, and CCL2

We wondered what determined the influx of specific TIL and, therefore, investigated the production of different chemokines, such as macrophage-derived chemokine (MDC/CCL22) and thymus- and activation-regulated chemokine (TARC/CCL17), both of which play a role in Treg migration, ²⁷ and monocyte chemoattractant protein-1 (MCP-1/CCL2), ²⁸ a chemokine that attracts and activates mononuclear cells and can be produced by the M2 macrophages themselves. We measured the expression of these chemokines in primary tumor cultures and monocytes by real-time PCR. Both CCL22 and CCL2 were expressed in all freshly cultured UM cells as well as in monocytes. Expression of CCL22 was higher in monocytes than in any of the UM cultures, while CCL2 showed the opposite result (Fig. 4). CCL17 showed extremely low expression in all samples tested (data not shown). 

Figure 4. 

 

Quantitative PCR shows CCL22 and CCL2 expression in UM cells and CD14⁺ monocytes. Especially primary UM cultures (n = 4) expressed CCL2 mRNA, while CCL22 was more strongly expressed in monocytes (n = 3).

Discussion

Previous studies have observed associations between the presence of a leukocytic infiltrate and a poor prognosis in UM, but such studies did not look at the complex diversity and balance of the infiltrating leukocytes. Recent reports suggested that thymic-derived CD4⁺CD25⁺ Treg (Foxp3⁺ lymphocytes) participate in the control of tumor immunity, and the presence of these cells in UM has been reported, but the relationship between the Treg population and other immune cell types or with important prognostic parameters, such as monosomy 3, has not yet been clarified in this malignancy. In order to understand whether the presence of TIL may be associated with nonimmunological tumor characteristics, we grouped the UMs according to their chromosome 3 status. Our study shows a clear association between monosomy 3, a highly reliable indicator of poor prognosis in UM, ²⁹ and the presence of a complex immune infiltrate, including CD8⁺ and CD4⁺ lymphocytes, Foxp3⁺ Tregs, and the pro-angiogenic and immunosuppressive M2 macrophages. ¹⁵  

UM cells, both primary and metastatic, are poor stimulators of proliferative responses by allogeneic lymphocytes, ³⁰ while by using autologous skin melanoma cell lines, generation of tumor-specific T cells from autologous mixed lymphocyte tumor cell cultures was possible. ³¹ This suggests that UM cells hamper the expansion of tumor-specific T cells either via direct interactions or by secretion of soluble factors. In a study by Ksander et al., ³² TILs were recovered from a series of human choroidal melanomas and expanded in culture media containing IL-2. In these experiments, tumor-specific T cells were detected among TILs from some tumors, but proliferation of TILs was limited. Furthermore, to function optimally, CD8⁺ T cells usually require CD4⁺ Th1 cells. ³³ Although in our study the number of CD4⁺ TILs correlated significantly with the presence of CD8⁺ T cells, we counted a relatively lower number of CD4⁺ than CD8⁺ TILs, similar to other tumors. ^21,25 The inability to proliferate or to exert other effector functions may be limited due to presence of immunosuppressive cells (e.g., Tregs and M2 macrophages). 

The increase of infiltrating Tregs in UM seems due to an increase in the total number of T lymphocytes. Using our immunofluorescent technique, we see more Treg-positive samples than in previously published studies using immunohistochemical stainings. ^10,11 When looking at our numbers of Foxp3⁺ cells, we frequently found only a few cells: in eight tumors there were less than five Foxp3⁺ cells per millimeter squared. It is possible that using our technique, the cells could be more easily detected. In contrast, in another Foxp3 study (Khatib et al. 2011; ARVO Meeting Abstract 1452), the authors were able to detect Foxp3⁺ Tregs in approximately 77% of posterior UM samples. As Tregs comprise less than 10% of the T-cell population within the UMs, one may think that Foxp3⁺ infiltrating cells would not play a dominant role. In addition, we did not find a correlation between Foxp3 status and TNM classification, suggesting that Tregs do not influence the progression of cancer. However, in an experimental corneal transplantation study it was demonstrated that the functional status of Tregs is more related to allograft outcome than their numbers. ³⁴ The essential role of Foxp3 in regulating the suppressor function of Tregs has been well documented, ³⁵ and all our counted intratumoral Tregs express Foxp3, indicating that part of the tumor-infiltrating CD4⁺ T cells may well be able to suppress local immunity and, thus, have immunosuppressive properties. However, to adequately evaluate antitumor immune functions, multiple factors should be analyzed simultaneously. ³⁶ The purpose of this study, therefore, was to determine if there is an association between Tregs, CD4⁺ and CD8⁺ T cells, and macrophages in UM. HLA peptide complexes at the cell surface of tumor cells allow for recognition by and interactions with T cells. This is the first report that has clarified the correlation of all these different leukocytic subsets and HLA expression in UM, and indicates that this infiltration by subtypes of lymphocytes and macrophages increased simultaneously in the inflamed tumors. 

Excessive inflammatory responses may be dampened by adaptive immune responses. ³⁷ Suppression of melanoma-specific CD8⁺ T-cell responses by ocular TAMs would provide another explanation of the association between a poor prognosis and the presence of a macrophage infiltrate. ¹² Two different types of macrophages can be identified, of which the M1 phenotype is pro-inflammatory and the M2 phenotype is associated with tissue repair and the production of anti-inflammatory cytokines. ³⁸ M2 macrophages are known to be able to directly induce Tregs, resulting in suppression of tumor-specific cytotoxic T-cells. ³⁹ We have previously shown that the majority of macrophages in UM belong to the M2 type, ¹⁵ and now show a strong correlation between the presence of Tregs and M2 macrophages. This correlation has also been observed in intrahepatic cholangiocarcinoma ⁴⁰ and gastrointestinal stromal tumors. ⁴¹  

The selective accumulation of Tregs and macrophages in the tumor microenvironment suggests that this process is tumor driven, and it may be that TIL influx is controlled by local chemokine secretion. CCL22 secreted by ovarian cancer cells and macrophages in the tumor microenvironment is known to induce selective migration of Treg. ²⁷ In a mouse skin melanoma transplantation model, Treg migration towards melanomas was related to secretion of CCL2, and not CCL22. ⁴² CCR4, the chemokine receptor that recognizes CCL22 and its alternative ligand CCL2, is broadly expressed on other immune cells that, therefore, can be attracted to the tumor. As UMs, as well as monocytes, produce these chemokines, the tumor as well as its TAM may be the source of these chemokines and may induce immune cell trafficking to tumors. The possibility that chemokine-mediated immune cell migration to tumor tissue may be blocked by targeting of either the chemokines or their specific receptors, provides opportunities to prevent TIL accumulation in the tumor microenvironment and suppression of tumor development. ⁴³  

In thyroid tumors, where many of the genetic tumor-initiating events have been identified, the oncogenes that drive tumorigenesis were proven to be able to induce an inflammatory program. ⁴⁴ We did not find an association between mutations in GNAQ and GNA11 and infiltrate (data not reported), which was to be expected, as the presence of these mutations is not associated with monosomy 3, ⁴⁵ and the molecular pathways by which monosomy 3 induces an inflammatory program remains, therefore, as yet, largely unknown. However, some oncogenes may directly activate inflammatory pathways in tumor cells, ⁴⁴ and have been studied in UM as, for example, the MYC oncogene (located on 8q24.1) ⁴⁶ : only tumors with monosomy 3 showed amplification of c-myc and this suggests a unique pathway of genetic progression, possibly involving attraction of inflammatory cells. RAS and BRAF mutations have been described as oncogene-induced promoters of an inflammatory microenvironment, but mutations in these genes are usually absent in UM. ⁴⁷ Our data suggest that loss of chromosome 3 may be associated with the initiation of inflammation. Moreover, infiltrating immune cells, such as macrophages, can increase the genomic instability of malignant cells as well as promote tissue remodeling and angiogenesis, through secretion of effector molecules, ⁴⁸ suggesting a self-reinforcing mechanism. 

In our study of the tumor microenvironment, UMs with poor prognosis are characterized by a brisk inflammatory infiltrate containing CD8⁺ and CD4⁺ T cells, Foxp3⁺ regulatory T cells, as well as macrophages, in an environment with an increased HLA class I and II expression. The phenotypic analysis of the TIL cell population and macrophages show that this infiltration is collectively increased, and that the balance of different immune cells is of no relevance. In addition, the presence of an immune infiltrate is associated with monosomy of chromosome 3. Therefore, intrinsic malignant properties of the tumor may lead to the production of leukocyte-attracting chemokines. However, the causal relationship between these processes remains elusive. 

Acknowledgments

We thank Frans A. Prins for his help with the confocal microscope and Geert Haasnoot, BSc, for his help with the statistical analysis. 

Footnotes

 Supported by grants from the Board of Directors of the Leiden University Medical Center (IHGB), Stichting Nederlands Oogheelkundig Onderzoek, Rotterdamse Stichting Blindenbelangen, and the Landelijke Stichting voor Blinden en Slechtzienden.

Footnotes

 Disclosure: I.H.G. Bronkhorst, None; T.H.K. Vu, None; E.S. Jordanova, None; G.P.M. Luyten, None; S.H. van der Burg, None; M.J. Jager, None