- Email: [email protected]
Ovarian cancer tumor infiltrating T-regulatory (Treg) cells are associated with a metastatic phenotype
Gynecologic Oncology 116 (2010) 556–562
Contents lists available at ScienceDirect
Gynecologic Oncology j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / y g y n o
Ovarian cancer tumor infiltrating T-regulatory (Treg) cells are associated with a metastatic phenotype☆ Jason C. Barnett a, Sarah M. Bean b, Regina S. Whitaker a, Eiji Kondoh a,c, Tsukasa Baba a,c, Shingo Fujii c, Jeffrey R. Marks d, Holly K. Dressman e, Susan K. Murphy a,b,e, Andrew Berchuck a,e,⁎ a
Division of Gynecologic Oncology/Department of Obstetrics and Gynecology, Duke University Medical Center, USA Department of Pathology, Duke University Medical Center, USA Department of Gynecology and Obstetrics, Kyoto University, Kyoto, Japan d Department of Surgery, Duke University Medical Center, Duke University, Durham, NC, USA e Duke Institute for Genome Sciences and Policy, Duke University, Durham, NC, USA b c
a r t i c l e
i n f o
Article history: Received 4 July 2009 Available online 14 December 2009 Keywords: Ovarian cancer Tumor infiltrating lymphocytes
a b s t r a c t Objective. The objective of this study was to examine the clinicopathologic correlates of T-regulatory (Treg) cell infiltration in serous ovarian cancers and to define gene signatures associated with high Tregs. Methods. Tumor infiltrating Treg and cytotoxic T-cells (CTLs) were quantitated in 232 primary serous ovarian cancers by immunostaining for FOXP3 and CD8. Expression microarray analysis was performed in a subset of 48 advanced cancers with the highest and lowest numbers of infiltrating Tregs and a genomic signature was developed using binary regression. ANOVA analysis was performed to assess the most differentially expressed genes and these genes were further assessed using Ingenuity Pathway Analysis (IPA) software. Results. High Treg infiltration in ovarian cancers was associated with high grade (p b 0.0001), advanced stage (p = 0.004) and suboptimal debulking (p b 0.04), but not with survival. In contrast, high tumor infiltrating CD8 CTL infiltration was associated with favorable survival (median survival 48.7 vs. 34.6 months, p = 0.01). A microarray-based genomic signature for high tumor-infiltrating Treg cells had a 77% predictive accuracy using leave-one-out cross-validation. ANOVA of microarray data revealed the antigen presentation pathway as the most differentially expressed canonical pathway (p b 0.00001) between cancers with high and low Treg cells. Conclusions. These data suggest that there may be an association between increased Treg cell infiltration in ovarian cancers and advanced stage. Increased Treg infiltration is characterized by a genomic signature enriched with several immunologic pathway genes. Therapeutic strategies that reduce tumor infiltrating Treg cells are under investigation and may prove useful in ovarian cancers with high numbers of these cells. © 2009 Elsevier Inc. All rights reserved.
Introduction The immune response to solid malignant tumors likely plays an influential role in cancer development and progression. Interactions between the tumor and the host immune response may lead to an environment that allows cancers to evade tumor-specific immune responses, resulting in tumor growth and metastasis [1]. Mechanisms of immune evasion include changes in the type, quality and response of tumor antigen presentation cells, alterations of immune costimulatory and co-inhibitory molecules within the tumor microenvironment, and altered ratios of different subsets of tumor infiltrating
☆ Presented in abstract form at the Society of Gynecologic Oncologists (SGO) 40th 2009 Annual Meeting on Women's Cancer. ⁎ Corresponding author. Division of Gynecologic Oncology, Duke University Medical Center, Box 3079 Durham, NC 27710, USA. Fax: +1 919 684 8719. E-mail address: [email protected] (A. Berchuck). 0090-8258/$ – see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.ygyno.2009.11.020
lymphocytes [1]. Epithelial ovarian cancer is a malignancy that has a propensity to present at an advanced stage, often with metastasis throughout the peritoneal cavity and to the retroperitoneal lymph nodes. This advanced presentation is associated with a poor prognosis with only about a 20% 5-year overall survival for stage III and IV disease [2]. The abundant lymphocytic infiltrate seen in many ovarian cancers suggests that the immune response plays a significant role in the biology of this disease, and several studies indicate that the predominant subtype of infiltrating lymphocyte may affect prognosis [3–9]. The importance of this paradigm was first highlighted by Zhang et al. in their 2003 report demonstrating that high numbers of CD3+ tumor infiltrating cytotoxic T lymphocytes were associated with improved survival in advanced ovarian cancers [3]. T-regulatory cells (CD4+CD25+ T-cells), also called Treg cells, are a specific subpopulation of lymphocytes that have immunosuppressive properties [1] Treg cells make up 5–10% of total CD4+ T-cells and normally act to help suppress inappropriate immune responses that
J.C. Barnett et al. / Gynecologic Oncology 116 (2010) 556–562
could lead to autoimmune disorders [7]. FOXP3 (forkhead box transcription factor 3) is a marker that distinguishes Treg cells and its expression can be evaluated immunohistochemically [7–8]. Infiltration of ovarian cancers by Treg cells has been shown to suppress tumor-specific T cell immune response and contribute to tumor growth [4]. Evidence suggests that trafficking of Treg cells into ovarian cancers may be mediated by the chemokine CCL22 produced by the tumor cells and macrophages in the microenvironment [4]. In addition, high numbers of Treg cells in ovarian cancers have been associated with poor survival in two prior studies [4,7]. The extent of metastatic disease at initial presentation is generally the most important prognostic factor related to cancer survival. Lack of immunological control of early cancer growth may be one factor contributing to the tendency for ovarian cancers to present at an advanced stage. To examine the hypothesis we evaluated the relationship between numbers of Tregs and stage as well as other clinical features and outcome in ovarian cancer. Our group has previously used expression microarray data to develop signatures that predict outcome in ovarian cancer [10]. In the present study, microarray-based genomic signatures were developed that correlate with high Treg infiltration in ovarian cancers. Alterations in molecular pathways that accompany high Treg infiltration were also defined using these microarray data. Methods Ovarian cancer specimens Primary epithelial ovarian cancer specimens were obtained from 232 women who were treated at Duke University Medical Center between January 1995 and December 2003. Specimens were acquired through the IRB approved tissue acquisition protocol and were frozen at −70 °C in the Duke Gynecology Oncology frozen tissue bank. All specimens were obtained with patient consent at the time of initial primary cytoreductive surgery. Frozen tumor specimens were formalin-fixed, paraffin embedded, and histologically evaluated by a gynecologic pathologist to confirm the pathologic diagnosis. Gene expression microarray data was previously generated as described for a subset of 74 specimens from patients with advanced ovarian cancer [11]. Immunohistochemistry Immunohistochemistry for FOXP3 was performed on 232 primary tumor specimens. Formalin-fixed, paraffin-embedded tissues were serially sectioned in 4–5 μm thick sections and placed in 0.01 M citrate buffer, pH 6.0, for a 15 min heat-induced antigen retrieval in a Decloaking Chamber (Biocare Medical, Concord, CA). Mouse monoclonal antibodies to FOXP3 (clone 236A/E7, AbD Serotec, Raleigh, NC) and to CD8 (clone 144B, abcam, Cambridge, MA) were used to identify Tregs and CTLs, respectively. Each were used separately at a dilution of 1:200 and incubated overnight at 4 °C followed by the two step Universal 4plusHPR horseradish peroxidase kit (Biocare Medical). Slides were then counterstained with hematoxylin. Tumor infiltrating T-cells staining positive for FOXP3 and CD8 were each manually counted in the three most densely stained 200× high power fields (hpf). For each case, the location of positive staining tumor infiltrating T-cells (intraepithelial vs. stromal) was noted and counted separately. The mean number of tumor infiltrating T-cells per hpf was evaluated as a continuous variable. Microarray analysis Gene microarray expression data was previously generated as described in Berchuck et al. [11] using Affymetrix U133A GeneChip arrays (22,283 probe sets can be found at: http://www.affymetrix.
557
com/support/technical/byproduct.affx?product=hgu133, Santa Clara, CA). Briefly, 10 μg of total RNA was used in the development of the Affymetrix cRNA target and was hybridized to the Affymetrix U133A GeneChip array following the protocols defined in the Affymetrix User Manual. Statistical analysis Clinicopathologic features were analyzed using the Mann– Whitney test for continuous nonparametric variables and Fischer's exact test was used to compare categorical variables. Two-tailed pvalues were used and statistical significance was set at p = 0.05. Survival analysis was performed using the Log-rank (Mantel-Cox) Test. Analyses were done with Graphpad Prism 5 (Graphpad Software Inc., La Jolla, CA). The Affymetrix CEL files from 74 tumor specimens (CEL files for tumor samples and clinical data can be accessed online1) were used to generated RMA normalized data followed by scan date batch correction using methods in the Partek Genomic Suite software (Partek Inc., St. Louis, MI). Normalized gene expression data from the 17 tumors with the highest number of infiltrating intraepithelial Treg cells (20 or greater/hpf, N80th percentile) were compared with the 31 tumors with the lowest number of infiltrating Treg cells (zero/hpf) using analysis of variance with correction for multiple comparisons and a false discovery rate q-value cutoff of 0.05. Expression of each gene was reported as a fold change comparison between high and low Treg infiltration. Affymetrix probe identifications were imported into Ingenuity Pathway Analysis (IPA) software (Ingenuity Systems, Redwood, CA) with their respective fold changes; the software was used to analyze significant pathway, network, and disease associations. IPA is commercially available software that uses literaturebased associations to evaluate microarray data. A gene expression signature for infiltrating Treg cells was developed using a binary regression model that has been described previously [12]. The expression signature represents a group of genes as a single expression profile and is here derived as the top principal component, or metagene. Prediction analysis of the expression data was performed using MATLAB software as previously described for the analysis of breast cancer samples [12]. When predicting infiltrating Treg cells, gene selection and identification is based on training the data and finding those genes most highly correlated to infiltrating Treg cells. Each signature summarizes its constituent genes as a single expression profile and is here derived as the first principal component of that set of genes (the factor corresponding to the largest singular value), as determined by a singular value decomposition. Given a training set of expression vectors (of values across metagenes) representing two biological states (high vs. low infiltrating Treg cells), a binary probit regression model is estimated using Bayesian methods. Bayesian fitting of binary probit regression models to the training data then permits an assessment of the relevance of the metagene signatures in within-sample classification, and estimation and uncertainty assessments for the binary regression weights mapping metagenes to probabilities of infiltrating Treg cells [11]. Results The clinicopathologic characteristics of the 232 patients are reported in Table 1. Median age of the patient population was 58 years. Patients had predominantly advanced stage (stage III/IV, 80%) and high grade (grade 2/3, 75%) disease. The most common histologic subtype was serous (69%) and 47% of the patients with stage III/IV cancer were optimally cytoreduced (b1 cm maximal diameter residual tumors) at primary surgery. All patients have been followed
1
http://data.genome.duke.edu/Treg.
558
J.C. Barnett et al. / Gynecologic Oncology 116 (2010) 556–562
Table 1 Clinicopathologic features of 232 patients with epithelial ovarian cancer. Number (%) Age, years Median Range Behavior Borderline Grade 1 Grade 2 Grade 3 FIGO stage I II III IV Unknown Histology Serous Endometrioid Mucinous Clear cell Other Debulking status Optimal Suboptimal Unknown
58 19–88 39 20 90 83
(17%) (9%) (39%) (36%)
24 (12%) 13 (7%) 127 (66%) 27 (14%) 2 (1%) 161 (69%) 30 (13%) 23 (10%) 12 (5%) 6 (3%) 72 (47%) 73 (47%) 9 (6%)
for more than 5 years and 139 patients (60%) have died. Median survival for the entire group was 47 months. Tumor infiltrating T-cells (both Treg and CTLs) were quantified both in the stromal and epithelial regions of the tumor specimen. Fig. 1 shows representative pictures of immunostaining and Table 2 details the relationship between Treg and CTL numbers and tumor behavior and stage. There was no FOXP3 staining of tumor cells. Higher numbers of infiltrating Treg cells were seen with advanced stage disease irrespective of whether the Treg cells were counted only among the epithelial component or in the stroma and epithelium together. The relationship with stage was also seen when grade 1 and borderline tumors, which are typically more indolent, were eliminated from the analysis. Increased numbers of either Treg cells or CTLs were also associated with higher grade regardless of their location (tumor vs. stroma) (Table 2). Mean number of intraepithelial and stromal Treg cells were higher in suboptimally debulked patients in comparison to optimally debulked patients (56.1 vs. 42.2/hpf, p = 0.04) while there was no association between infiltrating CTLs and debulking status.
There was no difference in overall survival when comparing advanced stage cases with high (≥ 20 Treg/hpf) and low Treg infiltration (median survival 32 versus 36 months for high versus low numbers of Treg cells, p = 0.67). Likewise, comparison of cases with high (≥2) versus low ratio of CTL/Treg also was not associated with outcome, although there was a trend towards improved survival with higher ratios (median survival 54 vs. 46 months for high versus low CTL/Treg, p = 0.15). Using the median value in advanced cases as a cutoff (≥2.1 CTL/hpf), higher intraepithelial CTL tumor infiltration was associated with improved survival with a median survival of 49 months compared to only 35 months for patients with lower numbers of CTLs (p = 0.01) (Fig. 2). Microarray gene expression data was previously generated for 74 advanced stage tumors [11]. Using the 48 cancers with the highest and lowest Treg infiltration, a genomic signature was generated predictive of high Treg infiltrating tumors. A heat map showing the 200 most differentially expressed genes as determined by a binary regression model between ovarian cancers with high versus low Treg cell tumor infiltration is shown in Fig. 3. This signature had a 77% overall accuracy to predict tumors with high versus low Treg infiltration using leave-one-out cross validation, correctly predicting 76% (13/17) of tumors with high Treg infiltration and 77% (24/31) with low Treg infiltration (p b 0.001). ANOVA with correction for multiple comparisons among the same tumors revealed 59 differentially expressed genes. Of these 59 most differentially expressed genes, 56 overlapped and were incorporated within the 200 genes used in the binary regression predictor set (Table A, supplementary data). These genes were imported into Ingenuity Pathway Analysis software. The canonical pathway most highly associated with differential expression of these genes was the antigen presentation pathway (4 of 39 genes, p b 0.00001) (Fig. 4). The four genes differentially expressed in this pathway were all upregulated in cancers with high Treg infiltration (HLA-DPA1, HLA-DPB1, TAP1, and TAPBP). Besides the antigen presentation pathway, the biological processes most associated with high Treg infiltration were immunological disease and cell mediated immune response, with 21 upregulated genes and one downregulated gene in the 35 gene network illustrated in Fig. 4 (list of genes detailed in Table B, supplementary data). Discussion Advanced ovarian cancer has a poor prognosis due to its propensity to present at an advanced stage after the tumor has
Fig. 1. Immunostaining for T-cells in ovarian cancers. Representative cancers with high infiltrating Treg cells and CTLS showing both intraepithelial (blue arrow) and stromal (red arrow) infiltration.
J.C. Barnett et al. / Gynecologic Oncology 116 (2010) 556–562
559
Table 2 Relationship between mean Tcells per high powered field and clinicopathologic features. Behavior
Intraepithelial Tregs Intraepithelial Tregs (grade 2/3 only) Intraepithelial + stromal Tregs Intraepithelial + stromal Tregs (grade 2/3 only) Intraepithelial CTLs Intraepithelial + stromal CTLs Ratio intraepith. CTLs/Tregs Ratio intraepith. + strom. CTLs/Tregs
Stage
Borderline/grade 1
Grade 2/3
p-value
I/II
III/IV
p-value
1.4 ± 3.8
9.6 ± 15.1
b0.0001
18.6 ± 49.7
49.3 ± 56.5
b0.001
10 ± 24.6 25.5 ± 50.2 17.7 ± 44.1 2.3 ± 3.1
17.8 ± 35.7 52.9 ± 76.5 2.7 ± 4.7 1.4 ± 3.4
0.05 0.004 0.28 0.26
4.1 ± 9.5 3.4 ± 4.5 30.8 ± 55.8 23.6 ± 28 18.7 ± 36.7 34.2 ± 58.2 16.4 ± 43.9 1 ± 1.3
10.0 ± 15.4 10.0 ± 15.5 52.4 ± 59 52.2 ± 58.6 17.8 ± 35.9 58.2 ± 79.2 2.6 ± 4.8 1.5 ± 3.7
0.004 b0.06 0.005 0.02 0.10 0.03 0.13 0.26
metastasized extensively in the abdomen. In an immunopathogenesis model for tumorigenesis, a cancer must overcome immunosurveillance and escape detection to grow and spread [1]. In a relatively large study of 232 ovarian carcinomas, we found that as stage and grade increased, the number of tumor infiltrating Treg cells also increased. As Treg cells normally function to suppress the immune system and prevent autoimmune disease, increased Treg tumor infiltrate in cancers may reflect an acquired ability to disarm immune responses that would otherwise inhibit progression. This model is further supported by the finding that suboptimally debulked cancers had higher numbers of tumor infiltrating T-regulatory cells, which suggests that Treg infiltration may facilitate the growth of bulky metastases that are difficult to resect. Zhang et al. first documented the prognostic significance of tumor infiltrating lymphocytes in ovarian cancer. High levels of CD3+ cytotoxic T-cells were associated with a dramatic improvement in both progression free and overall survival [3]. High Treg expression in ovarian cancers has also been reported to be a poor prognostic factor. Wolf et al. found that in 99 ovarian carcinomas, average survival was 77 months in tumors with low FOXP3 RNA expression versus only 28 months in those with high expression [7]. Curiel et al. used multicolor confocal microscopy to measure Treg cells in 70 ovarian cancers and found that those with the highest infiltration had a 25 fold higher death hazard in comparison with tumors with the lowest numbers [4]. In contrast, studies that have used immunohistochemistry to evaluate Treg infiltration have reported conflicting results. Leffers et al. showed improved survival with the presence of tumor infiltrating Treg cells [6] while two other studies showed no association with survival [5,13]. One of these latter studies by Sato et al. included detailed information regarding the number of tumor infiltrating Treg cells, and the results were similar to what was observed in our study (median Treg cells = 0.7/ hpf) [5]. At least two studies have examined the ratio of different subpopulations of tumor-infiltrating lymphocytes in ovarian carcinomas. In both, increased ratios of CD8+ T-cells/Treg cells were associated with improved survival [5–6]. While our results are not statistically significant (p = 0.15), there was an 8 month improvement in median overall survival for patients having tumors with higher ratios of CTL/Treg cells. Our results also corroborate reports that higher numbers of CD8+ infiltrating lymphocytes are associated with favorable survival [5–6]. Studies have also evaluated the relationship between the specific location of the tumor infiltrating lymphocytes and prognosis. A recent report in colon cancer showed that while higher Treg infiltration was associated with an improvement in survival, increased numbers of Treg cells in the adjacent normal colon epithelium were associated with diminished survival [14]. Sato et al. examined both intraepithelial and stromal infiltration in ovarian cancer and showed that only intraepithelial T-cell infiltration has prognostic significance [5]. Curiel et al. found that a majority of intratumoral T-regulatory cells were in close proximity to CD8+ T-cells, suggesting that the immunosuppressive properties of Treg cells may be mediated through direct contact [4]. In light of these reports, we focused on intraepithelial infiltration in the development of a genomic signature associated with increased T-regulatory cells—even though our study showed that
higher numbers of Treg cells were associated with advanced stage and high grade, regardless of their location. Besides establishing the prognostic significance of tumor infiltrating lymphocytes, another important question is: what determines the predominant lymphocytic subpopulation in ovarian cancers? The study by Zhang et al. suggests that higher levels of interferon gamma (IFNγ) and interleukin-2 (IL-2) and the chemokines induced by these cytokines are necessary for higher densities of tumor lymphocyte infiltration [3]. In a murine model of fibrosarcomas, Treg cells suppressed CD8+ T-cells through IFNγ production. They also
Fig. 2. Relationship between tumor infiltrating lymphocytes and survival. (a) T regulatory cells. (b) Cytotoxic T cells. (c) CTL/Treg ratio.
560
J.C. Barnett et al. / Gynecologic Oncology 116 (2010) 556–562
Fig. 3. Genes most differentially expressed and leave-one-out cross validation predictions between ovarian cancers with low and high Treg infiltration. (a) Columns correspond to cancers, with those from cancers with low infiltrating Tregs depicted to the left (n = 31) of the vertical black line and those from cancers with high infiltrating Tregs (n = 17) to the right. The 200 differentially expressed genes determined by the binary regression model are plotted in rows. High gene expression is represented in red and low expression in blue, while intermediate levels of expression are represented in shades of dark yellow through green to light blue. (b) A horizontal line is plotted at 0.5 indicating the classification threshold used for determining leave-one-out cross validation predictive accuracy (cases with probability of high Treg infiltration greater than 0.5 were declared high Treg infiltrating tumors; those less than 0.5 as low Treg infiltrating tumors). The horizontal axis lists the 48 samples with low (empty circles) and high (filled circles) Treg infiltrating tumors used in the regression. The corresponding values on the vertical axis are estimated classification probabilities with corresponding 90% probability intervals marked as lines to indicate uncertainty about these estimated values. The analysis and prediction for each sample are based on the subset of 200 most discriminatory genes. (c) A single-variable scatter plot of a significance test of the predicted probabilities of high infiltrating Treg cells (P b 0.001, Mann–Whitney U-test).
demonstrated reversal of this suppression by blocking the effects of interleukin 10 (IL-10) and transforming growth factor Beta (TGF-β), cytokines implicated in Treg recruitment [15]. TGF-β was also found to convert CD4+CD25− T-cells into CD4+CD25+ (Treg cells) when added to ovarian cancer cell lines [16], and Wei et al. showed that administration of IL-2 to patients with ovarian cancer affects the concentration and function of Treg cells in peripheral blood [17]. Curiel et al. demonstrated that ovarian cancer cells in mice produce chemokine ligand 22 (CCL22), while Treg cells express the receptor to CCL22–chemokine receptor 4 (CCR4). They showed that migration of Treg cells into ovarian tumors could be blocked using antibodies to CCL22, suggesting that this chemokine is a chemoattractant necessary for Treg recruitment [4]. Although these cytokines and chemokines may have a role in recruitment of tumor infiltrating lymphocytes, there was no difference in mRNA expression of these genes between
samples with high vs. low Treg infiltration in our study. However, several immunologic genes were differentially expressed, including increased expression of the chemokines CXCL10 (IP-10) and CCL5 (RANTES), both of which are involved in T-cell recruitment [18,19]. Further studies of the relationship between these chemokines, Tregulatory cells, and the ratios of different tumor infiltrating lymphocytes (CD8, CD3) could add insight into the immunologic microenvironment associated with immune evasion. Although immune cells comprise a relatively small fraction of most non-hematologic human cancers, genomic signatures have been described that involve differential expression of genes involved in immune response. In this regard, Tothill et al. defined six distinct ovarian cancer subtypes based on signatures from gene expression microarrays. One of these was an “immunological subtype” that was characterized by differential expression of several immunologic related genes and these
Fig. 4. Schematics depicting antigen presentation pathway and immune related gene networks. (a) Diagram of antigen presentation pathway showing upregulated genes shaded red in cancers with high Treg infiltration (©2000–2008 Ingenuity Systems, Inc.). (b) Diagram depicting a 35 gene network involving immunological disease, antigen presentation, and cell mediated immune response with 21 upregulated genes (red) and one downregulated gene (green) in cancers with high Treg infiltration. Differentially expressed genes are listed with respective fold changes in supplementary data table B (©2000–2008 Ingenuity Systems, Inc.).
J.C. Barnett et al. / Gynecologic Oncology 116 (2010) 556–562
561
562
J.C. Barnett et al. / Gynecologic Oncology 116 (2010) 556–562
cancers demonstrated higher numbers of tumor infiltrating CD3+ lymphocytes and more favorable survival [20]. In the present study, we demonstrated the existence of a genomic signature for high Treg infiltration that had a 77% predictive accuracy using leave-one-out cross validation. We also showed that the most differentially expressed genes between cancers with the highest and lowest infiltrating Treg cells are involved in pathways and biologic processes related to immune function. The antigen presentation pathway was the canonical pathway most highly associated with high Treg infiltration. This pathway is used by the cell to prepare and then present antigenic oligopeptide fragments on the surface of the cell to nearby T-cells allowing for its activation [21]. A limitation of the present study is the inclusion of only CD8 and FOXP3 markers in the evaluation of tumor infiltrating lymphocytes. While CD3 and CD4 are markers that add further insight into the immunologic tumor microenvironment, and have been included in other studies [3,5], the focus of the present study was to evaluate Tregulatory cells, their relationship to CTLs, and the genomic expression profile of tumors with increased numbers of Tregs. Another limitation is that CD8 is also a marker of certain subsets of dendritic antigen presentation cells [22]. Further staining could delineate the specific type and level of cytotoxicy of these CD8+ tumor infiltrating cells. Additionally, previous studies have also suggested a role for CD8+ Tregulatory cells; evaluation of this lymphocytic subset could also provide further insight into tumor immunity [23]. A better understanding of the mechanisms that underlie host immune tolerance to cancers is of paramount importance in developing immunologic treatment strategies. Currently, the FDA has approved denileukin diftitox, a ligand toxin fusion of full length IL2 and diphtheria toxin, for treatment of CD25+CD4+ cutaneous Tcell leukemia and lymphoma. This drug works through depletion of CD25+ Treg cells and is being tested in treatment of other cancers, including ovarian cancers [1]. FOXP3 has also been studied as a therapeutic target with vaccination against FOXP3 being associated with enhanced tumor immunity [24]. Overcoming immune tolerance is an immunotherapeutic strategy with potential promise in the treatment of ovarian cancer. A better understanding of the mechanisms underlying attenuated tumor immunity may facilitate therapeutic approaches that complement existing cytotoxic therapies. Conflict of interest statement The authors declare there are no conflicts of interest.
Acknowledgments This work was supported by the University of Alabama Ovarian Cancer Specialized Program of Research Excellence (SPORE), the Gail Parkins Ovarian Cancer Research Fund (AB), and the American Cancer Society. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.ygyno.2009.11.020.
References [1] Zou WP. Immunosuppressive networks in the tumor environment and their therapeutic relevance. Nat Rev, Cancer 2005;5:263–74. [2] Heintz AP, Odicino F, Maisonneuve P, et al. Carcinoma of the ovary. FIGO 6th Annual Report on the Results of Treatment in Gynecological Cancer. Int J Gynaecol Obstet 2006;95:S161–92. [3] Zhang L, Conejo-Garcia JR, Katsaros D, et al. Intratumoral T cells, recurrence, and survival in epithelial ovarian cancer. N Engl J Med 2003;348:203–13. [4] Curiel TJ, Coukos G, Zou L, et al. Specific recruitment of regulatory T cells in ovarian carcinoma fosters immune privilege and predicts reduced survival. Nat Med 2004;10:942–9. [5] Sato E, Olson SH, Ahn J, et al. Intraepithelial CD8+ tumor-infiltrating lymphocytes and a high CD8+/regulatory T cell ratio are associated with favorable prognosis in ovarian cancer. Proc Natl Acad Sci U S A 2005;102:18538–43. [6] Leffers N, Gooden Mm, de Jong RA. Prognostic significance of tumor-infiltrating Tlymphocytes in primary and metastatic lesions of advanced stage ovarian cancer. Cancer Immunol Immunother 2009;58:449–59. [7] Wolf D, Wolf AM, Rumpold H, et al. The expression of the regulatory T cell-specific forkhead box transcription factor FoxP3 is associated with poor prognosis in ovarian cancer. Clin Cancer Res 2005;11:8326–31. [8] Bignone PA, Banham AH. FOXP3+ regulatory T cells as biomarkers in human malignancies. Expert Opin Biol Ther 2008;8:1897–920. [9] Nelson BH. The impact of T-cell immunity on ovarian cancer outcomes. Immunol Rev 2008;222:101–16. [10] Berchuck A, Iversen ES, Luo J, et al. Microarray analysis of early stage serous ovarian cancers demonstrates profiles predictive of favorable outcome. Clin Cancer Res 2009;15:2448–55. [11] Berchuck A, Iversen ES, Lancaster JM, et al. Patterns of gene expression that characterize long-term survival in advanced stage serous ovarian cancers. Clin Cancer Res 2005;15:3686–96. [12] West M, Blanchette C, Dressman H, et al. Predicting the clinical status of human breast cancer by using gene expression profiles. Proc Natl Acad Sci U S A 2001;98: 11462–7. [13] Shah CA, Allison KH, Garcia RL, Gray H, Goff BA, Swisher EM. Intratumoral T cells, tumor-associated macrophages, and regulatory T-cells: Association with p53 mutations, circulating tumor DNA and survival in women with ovarian cancer. Gynecol Oncol 2008;109:215–9. [14] Salama P, Phillips M, Grieu F, et al. Tumor-infiltrating FOXP3 T regulatory cells show strong prognostic significance in colorectal cancer. J Clin Oncol 2009;27: 186–92. [15] Ping Y, Lee Y, Liu W, et al. Intratumor depletion of CD4+ cells unmasks tumor immunogenicity leading to the rejection of late-stage tumors. J Exp Med 2005;2015:779–91. [16] Li X, Ye F, Chen H, Lu W, Wan X, Xie X. Human ovarian carcinoma cells generate CD4+CD25+ regulatory T cells from peripheral CD4+CD25− T cells through secreting TGF-B. Cancer Lett 2007;253:144–53. [17] Wei S, Kryczek I, Edwards RP, et al. Interleukin-2 administration alters the CD4 +FOXP3+ T-cell pool and tumor trafficking in patients with ovarian carcinoma. Cancer Res 2007;67:7487–94. [18] Nesbeth Y, Scarlett U, Cubillos-Ruiz J, et al. CCL5-Mediated endogenous antitumor immunity elicited by adoptively transferred lymphocytes and dendritic cell depletion. Cancer Res 2009;69:6331–8. [19] Dufour JH, Dziejman M, Liu MT, et al. IFN-gamma-inducible protein 10 (IP-10; CXCL10)-deficient mice reveal a role for IP-10 in effector T cell generation and trafficking. J Immunol 2002;168(7):3195–204. [20] Tothill RW, Tinker AV, George J, et al. Novel molecular subtypes of serous and endometrioid ovarian cancer linked to clinical outcome. Clin Cancer Res 2008;14: 5198–208. [21] Weinberg RA. Crowd control: tumor immunology and immunotherapy. The Biology of Cancer. Garland Science. Taylor and Francis Group, LLC; 2007. [22] Kronin V, Wu L, Gong S, et al. DEC-205 as a marker of dendritic cells with regulatory effects on CD8 T cell responses. Int Immunol 2000;12(5):731–5. [23] Xystrakis E, Dejean AS, Bernard I, et al. Identification of a novel natural regulatory CD8 T-cell subset and analysis of its mechanism of regulation. Blood 2004;104 (10):3294–301. [24] Nair S, Boczkowski D, Fassnacht M, Pisestky D, Gilboa E. Vaccination against the Forkhead family transcription factor Foxp3 enhances tumor immunity. Cancer Res 2007;67:371–80.
Contents lists available at ScienceDirect
Gynecologic Oncology j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / y g y n o
Ovarian cancer tumor infiltrating T-regulatory (Treg) cells are associated with a metastatic phenotype☆ Jason C. Barnett a, Sarah M. Bean b, Regina S. Whitaker a, Eiji Kondoh a,c, Tsukasa Baba a,c, Shingo Fujii c, Jeffrey R. Marks d, Holly K. Dressman e, Susan K. Murphy a,b,e, Andrew Berchuck a,e,⁎ a
Division of Gynecologic Oncology/Department of Obstetrics and Gynecology, Duke University Medical Center, USA Department of Pathology, Duke University Medical Center, USA Department of Gynecology and Obstetrics, Kyoto University, Kyoto, Japan d Department of Surgery, Duke University Medical Center, Duke University, Durham, NC, USA e Duke Institute for Genome Sciences and Policy, Duke University, Durham, NC, USA b c
a r t i c l e
i n f o
Article history: Received 4 July 2009 Available online 14 December 2009 Keywords: Ovarian cancer Tumor infiltrating lymphocytes
a b s t r a c t Objective. The objective of this study was to examine the clinicopathologic correlates of T-regulatory (Treg) cell infiltration in serous ovarian cancers and to define gene signatures associated with high Tregs. Methods. Tumor infiltrating Treg and cytotoxic T-cells (CTLs) were quantitated in 232 primary serous ovarian cancers by immunostaining for FOXP3 and CD8. Expression microarray analysis was performed in a subset of 48 advanced cancers with the highest and lowest numbers of infiltrating Tregs and a genomic signature was developed using binary regression. ANOVA analysis was performed to assess the most differentially expressed genes and these genes were further assessed using Ingenuity Pathway Analysis (IPA) software. Results. High Treg infiltration in ovarian cancers was associated with high grade (p b 0.0001), advanced stage (p = 0.004) and suboptimal debulking (p b 0.04), but not with survival. In contrast, high tumor infiltrating CD8 CTL infiltration was associated with favorable survival (median survival 48.7 vs. 34.6 months, p = 0.01). A microarray-based genomic signature for high tumor-infiltrating Treg cells had a 77% predictive accuracy using leave-one-out cross-validation. ANOVA of microarray data revealed the antigen presentation pathway as the most differentially expressed canonical pathway (p b 0.00001) between cancers with high and low Treg cells. Conclusions. These data suggest that there may be an association between increased Treg cell infiltration in ovarian cancers and advanced stage. Increased Treg infiltration is characterized by a genomic signature enriched with several immunologic pathway genes. Therapeutic strategies that reduce tumor infiltrating Treg cells are under investigation and may prove useful in ovarian cancers with high numbers of these cells. © 2009 Elsevier Inc. All rights reserved.
Introduction The immune response to solid malignant tumors likely plays an influential role in cancer development and progression. Interactions between the tumor and the host immune response may lead to an environment that allows cancers to evade tumor-specific immune responses, resulting in tumor growth and metastasis [1]. Mechanisms of immune evasion include changes in the type, quality and response of tumor antigen presentation cells, alterations of immune costimulatory and co-inhibitory molecules within the tumor microenvironment, and altered ratios of different subsets of tumor infiltrating
☆ Presented in abstract form at the Society of Gynecologic Oncologists (SGO) 40th 2009 Annual Meeting on Women's Cancer. ⁎ Corresponding author. Division of Gynecologic Oncology, Duke University Medical Center, Box 3079 Durham, NC 27710, USA. Fax: +1 919 684 8719. E-mail address: [email protected] (A. Berchuck). 0090-8258/$ – see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.ygyno.2009.11.020
lymphocytes [1]. Epithelial ovarian cancer is a malignancy that has a propensity to present at an advanced stage, often with metastasis throughout the peritoneal cavity and to the retroperitoneal lymph nodes. This advanced presentation is associated with a poor prognosis with only about a 20% 5-year overall survival for stage III and IV disease [2]. The abundant lymphocytic infiltrate seen in many ovarian cancers suggests that the immune response plays a significant role in the biology of this disease, and several studies indicate that the predominant subtype of infiltrating lymphocyte may affect prognosis [3–9]. The importance of this paradigm was first highlighted by Zhang et al. in their 2003 report demonstrating that high numbers of CD3+ tumor infiltrating cytotoxic T lymphocytes were associated with improved survival in advanced ovarian cancers [3]. T-regulatory cells (CD4+CD25+ T-cells), also called Treg cells, are a specific subpopulation of lymphocytes that have immunosuppressive properties [1] Treg cells make up 5–10% of total CD4+ T-cells and normally act to help suppress inappropriate immune responses that
J.C. Barnett et al. / Gynecologic Oncology 116 (2010) 556–562
could lead to autoimmune disorders [7]. FOXP3 (forkhead box transcription factor 3) is a marker that distinguishes Treg cells and its expression can be evaluated immunohistochemically [7–8]. Infiltration of ovarian cancers by Treg cells has been shown to suppress tumor-specific T cell immune response and contribute to tumor growth [4]. Evidence suggests that trafficking of Treg cells into ovarian cancers may be mediated by the chemokine CCL22 produced by the tumor cells and macrophages in the microenvironment [4]. In addition, high numbers of Treg cells in ovarian cancers have been associated with poor survival in two prior studies [4,7]. The extent of metastatic disease at initial presentation is generally the most important prognostic factor related to cancer survival. Lack of immunological control of early cancer growth may be one factor contributing to the tendency for ovarian cancers to present at an advanced stage. To examine the hypothesis we evaluated the relationship between numbers of Tregs and stage as well as other clinical features and outcome in ovarian cancer. Our group has previously used expression microarray data to develop signatures that predict outcome in ovarian cancer [10]. In the present study, microarray-based genomic signatures were developed that correlate with high Treg infiltration in ovarian cancers. Alterations in molecular pathways that accompany high Treg infiltration were also defined using these microarray data. Methods Ovarian cancer specimens Primary epithelial ovarian cancer specimens were obtained from 232 women who were treated at Duke University Medical Center between January 1995 and December 2003. Specimens were acquired through the IRB approved tissue acquisition protocol and were frozen at −70 °C in the Duke Gynecology Oncology frozen tissue bank. All specimens were obtained with patient consent at the time of initial primary cytoreductive surgery. Frozen tumor specimens were formalin-fixed, paraffin embedded, and histologically evaluated by a gynecologic pathologist to confirm the pathologic diagnosis. Gene expression microarray data was previously generated as described for a subset of 74 specimens from patients with advanced ovarian cancer [11]. Immunohistochemistry Immunohistochemistry for FOXP3 was performed on 232 primary tumor specimens. Formalin-fixed, paraffin-embedded tissues were serially sectioned in 4–5 μm thick sections and placed in 0.01 M citrate buffer, pH 6.0, for a 15 min heat-induced antigen retrieval in a Decloaking Chamber (Biocare Medical, Concord, CA). Mouse monoclonal antibodies to FOXP3 (clone 236A/E7, AbD Serotec, Raleigh, NC) and to CD8 (clone 144B, abcam, Cambridge, MA) were used to identify Tregs and CTLs, respectively. Each were used separately at a dilution of 1:200 and incubated overnight at 4 °C followed by the two step Universal 4plusHPR horseradish peroxidase kit (Biocare Medical). Slides were then counterstained with hematoxylin. Tumor infiltrating T-cells staining positive for FOXP3 and CD8 were each manually counted in the three most densely stained 200× high power fields (hpf). For each case, the location of positive staining tumor infiltrating T-cells (intraepithelial vs. stromal) was noted and counted separately. The mean number of tumor infiltrating T-cells per hpf was evaluated as a continuous variable. Microarray analysis Gene microarray expression data was previously generated as described in Berchuck et al. [11] using Affymetrix U133A GeneChip arrays (22,283 probe sets can be found at: http://www.affymetrix.
557
com/support/technical/byproduct.affx?product=hgu133, Santa Clara, CA). Briefly, 10 μg of total RNA was used in the development of the Affymetrix cRNA target and was hybridized to the Affymetrix U133A GeneChip array following the protocols defined in the Affymetrix User Manual. Statistical analysis Clinicopathologic features were analyzed using the Mann– Whitney test for continuous nonparametric variables and Fischer's exact test was used to compare categorical variables. Two-tailed pvalues were used and statistical significance was set at p = 0.05. Survival analysis was performed using the Log-rank (Mantel-Cox) Test. Analyses were done with Graphpad Prism 5 (Graphpad Software Inc., La Jolla, CA). The Affymetrix CEL files from 74 tumor specimens (CEL files for tumor samples and clinical data can be accessed online1) were used to generated RMA normalized data followed by scan date batch correction using methods in the Partek Genomic Suite software (Partek Inc., St. Louis, MI). Normalized gene expression data from the 17 tumors with the highest number of infiltrating intraepithelial Treg cells (20 or greater/hpf, N80th percentile) were compared with the 31 tumors with the lowest number of infiltrating Treg cells (zero/hpf) using analysis of variance with correction for multiple comparisons and a false discovery rate q-value cutoff of 0.05. Expression of each gene was reported as a fold change comparison between high and low Treg infiltration. Affymetrix probe identifications were imported into Ingenuity Pathway Analysis (IPA) software (Ingenuity Systems, Redwood, CA) with their respective fold changes; the software was used to analyze significant pathway, network, and disease associations. IPA is commercially available software that uses literaturebased associations to evaluate microarray data. A gene expression signature for infiltrating Treg cells was developed using a binary regression model that has been described previously [12]. The expression signature represents a group of genes as a single expression profile and is here derived as the top principal component, or metagene. Prediction analysis of the expression data was performed using MATLAB software as previously described for the analysis of breast cancer samples [12]. When predicting infiltrating Treg cells, gene selection and identification is based on training the data and finding those genes most highly correlated to infiltrating Treg cells. Each signature summarizes its constituent genes as a single expression profile and is here derived as the first principal component of that set of genes (the factor corresponding to the largest singular value), as determined by a singular value decomposition. Given a training set of expression vectors (of values across metagenes) representing two biological states (high vs. low infiltrating Treg cells), a binary probit regression model is estimated using Bayesian methods. Bayesian fitting of binary probit regression models to the training data then permits an assessment of the relevance of the metagene signatures in within-sample classification, and estimation and uncertainty assessments for the binary regression weights mapping metagenes to probabilities of infiltrating Treg cells [11]. Results The clinicopathologic characteristics of the 232 patients are reported in Table 1. Median age of the patient population was 58 years. Patients had predominantly advanced stage (stage III/IV, 80%) and high grade (grade 2/3, 75%) disease. The most common histologic subtype was serous (69%) and 47% of the patients with stage III/IV cancer were optimally cytoreduced (b1 cm maximal diameter residual tumors) at primary surgery. All patients have been followed
1
http://data.genome.duke.edu/Treg.
558
J.C. Barnett et al. / Gynecologic Oncology 116 (2010) 556–562
Table 1 Clinicopathologic features of 232 patients with epithelial ovarian cancer. Number (%) Age, years Median Range Behavior Borderline Grade 1 Grade 2 Grade 3 FIGO stage I II III IV Unknown Histology Serous Endometrioid Mucinous Clear cell Other Debulking status Optimal Suboptimal Unknown
58 19–88 39 20 90 83
(17%) (9%) (39%) (36%)
24 (12%) 13 (7%) 127 (66%) 27 (14%) 2 (1%) 161 (69%) 30 (13%) 23 (10%) 12 (5%) 6 (3%) 72 (47%) 73 (47%) 9 (6%)
for more than 5 years and 139 patients (60%) have died. Median survival for the entire group was 47 months. Tumor infiltrating T-cells (both Treg and CTLs) were quantified both in the stromal and epithelial regions of the tumor specimen. Fig. 1 shows representative pictures of immunostaining and Table 2 details the relationship between Treg and CTL numbers and tumor behavior and stage. There was no FOXP3 staining of tumor cells. Higher numbers of infiltrating Treg cells were seen with advanced stage disease irrespective of whether the Treg cells were counted only among the epithelial component or in the stroma and epithelium together. The relationship with stage was also seen when grade 1 and borderline tumors, which are typically more indolent, were eliminated from the analysis. Increased numbers of either Treg cells or CTLs were also associated with higher grade regardless of their location (tumor vs. stroma) (Table 2). Mean number of intraepithelial and stromal Treg cells were higher in suboptimally debulked patients in comparison to optimally debulked patients (56.1 vs. 42.2/hpf, p = 0.04) while there was no association between infiltrating CTLs and debulking status.
There was no difference in overall survival when comparing advanced stage cases with high (≥ 20 Treg/hpf) and low Treg infiltration (median survival 32 versus 36 months for high versus low numbers of Treg cells, p = 0.67). Likewise, comparison of cases with high (≥2) versus low ratio of CTL/Treg also was not associated with outcome, although there was a trend towards improved survival with higher ratios (median survival 54 vs. 46 months for high versus low CTL/Treg, p = 0.15). Using the median value in advanced cases as a cutoff (≥2.1 CTL/hpf), higher intraepithelial CTL tumor infiltration was associated with improved survival with a median survival of 49 months compared to only 35 months for patients with lower numbers of CTLs (p = 0.01) (Fig. 2). Microarray gene expression data was previously generated for 74 advanced stage tumors [11]. Using the 48 cancers with the highest and lowest Treg infiltration, a genomic signature was generated predictive of high Treg infiltrating tumors. A heat map showing the 200 most differentially expressed genes as determined by a binary regression model between ovarian cancers with high versus low Treg cell tumor infiltration is shown in Fig. 3. This signature had a 77% overall accuracy to predict tumors with high versus low Treg infiltration using leave-one-out cross validation, correctly predicting 76% (13/17) of tumors with high Treg infiltration and 77% (24/31) with low Treg infiltration (p b 0.001). ANOVA with correction for multiple comparisons among the same tumors revealed 59 differentially expressed genes. Of these 59 most differentially expressed genes, 56 overlapped and were incorporated within the 200 genes used in the binary regression predictor set (Table A, supplementary data). These genes were imported into Ingenuity Pathway Analysis software. The canonical pathway most highly associated with differential expression of these genes was the antigen presentation pathway (4 of 39 genes, p b 0.00001) (Fig. 4). The four genes differentially expressed in this pathway were all upregulated in cancers with high Treg infiltration (HLA-DPA1, HLA-DPB1, TAP1, and TAPBP). Besides the antigen presentation pathway, the biological processes most associated with high Treg infiltration were immunological disease and cell mediated immune response, with 21 upregulated genes and one downregulated gene in the 35 gene network illustrated in Fig. 4 (list of genes detailed in Table B, supplementary data). Discussion Advanced ovarian cancer has a poor prognosis due to its propensity to present at an advanced stage after the tumor has
Fig. 1. Immunostaining for T-cells in ovarian cancers. Representative cancers with high infiltrating Treg cells and CTLS showing both intraepithelial (blue arrow) and stromal (red arrow) infiltration.
J.C. Barnett et al. / Gynecologic Oncology 116 (2010) 556–562
559
Table 2 Relationship between mean Tcells per high powered field and clinicopathologic features. Behavior
Intraepithelial Tregs Intraepithelial Tregs (grade 2/3 only) Intraepithelial + stromal Tregs Intraepithelial + stromal Tregs (grade 2/3 only) Intraepithelial CTLs Intraepithelial + stromal CTLs Ratio intraepith. CTLs/Tregs Ratio intraepith. + strom. CTLs/Tregs
Stage
Borderline/grade 1
Grade 2/3
p-value
I/II
III/IV
p-value
1.4 ± 3.8
9.6 ± 15.1
b0.0001
18.6 ± 49.7
49.3 ± 56.5
b0.001
10 ± 24.6 25.5 ± 50.2 17.7 ± 44.1 2.3 ± 3.1
17.8 ± 35.7 52.9 ± 76.5 2.7 ± 4.7 1.4 ± 3.4
0.05 0.004 0.28 0.26
4.1 ± 9.5 3.4 ± 4.5 30.8 ± 55.8 23.6 ± 28 18.7 ± 36.7 34.2 ± 58.2 16.4 ± 43.9 1 ± 1.3
10.0 ± 15.4 10.0 ± 15.5 52.4 ± 59 52.2 ± 58.6 17.8 ± 35.9 58.2 ± 79.2 2.6 ± 4.8 1.5 ± 3.7
0.004 b0.06 0.005 0.02 0.10 0.03 0.13 0.26
metastasized extensively in the abdomen. In an immunopathogenesis model for tumorigenesis, a cancer must overcome immunosurveillance and escape detection to grow and spread [1]. In a relatively large study of 232 ovarian carcinomas, we found that as stage and grade increased, the number of tumor infiltrating Treg cells also increased. As Treg cells normally function to suppress the immune system and prevent autoimmune disease, increased Treg tumor infiltrate in cancers may reflect an acquired ability to disarm immune responses that would otherwise inhibit progression. This model is further supported by the finding that suboptimally debulked cancers had higher numbers of tumor infiltrating T-regulatory cells, which suggests that Treg infiltration may facilitate the growth of bulky metastases that are difficult to resect. Zhang et al. first documented the prognostic significance of tumor infiltrating lymphocytes in ovarian cancer. High levels of CD3+ cytotoxic T-cells were associated with a dramatic improvement in both progression free and overall survival [3]. High Treg expression in ovarian cancers has also been reported to be a poor prognostic factor. Wolf et al. found that in 99 ovarian carcinomas, average survival was 77 months in tumors with low FOXP3 RNA expression versus only 28 months in those with high expression [7]. Curiel et al. used multicolor confocal microscopy to measure Treg cells in 70 ovarian cancers and found that those with the highest infiltration had a 25 fold higher death hazard in comparison with tumors with the lowest numbers [4]. In contrast, studies that have used immunohistochemistry to evaluate Treg infiltration have reported conflicting results. Leffers et al. showed improved survival with the presence of tumor infiltrating Treg cells [6] while two other studies showed no association with survival [5,13]. One of these latter studies by Sato et al. included detailed information regarding the number of tumor infiltrating Treg cells, and the results were similar to what was observed in our study (median Treg cells = 0.7/ hpf) [5]. At least two studies have examined the ratio of different subpopulations of tumor-infiltrating lymphocytes in ovarian carcinomas. In both, increased ratios of CD8+ T-cells/Treg cells were associated with improved survival [5–6]. While our results are not statistically significant (p = 0.15), there was an 8 month improvement in median overall survival for patients having tumors with higher ratios of CTL/Treg cells. Our results also corroborate reports that higher numbers of CD8+ infiltrating lymphocytes are associated with favorable survival [5–6]. Studies have also evaluated the relationship between the specific location of the tumor infiltrating lymphocytes and prognosis. A recent report in colon cancer showed that while higher Treg infiltration was associated with an improvement in survival, increased numbers of Treg cells in the adjacent normal colon epithelium were associated with diminished survival [14]. Sato et al. examined both intraepithelial and stromal infiltration in ovarian cancer and showed that only intraepithelial T-cell infiltration has prognostic significance [5]. Curiel et al. found that a majority of intratumoral T-regulatory cells were in close proximity to CD8+ T-cells, suggesting that the immunosuppressive properties of Treg cells may be mediated through direct contact [4]. In light of these reports, we focused on intraepithelial infiltration in the development of a genomic signature associated with increased T-regulatory cells—even though our study showed that
higher numbers of Treg cells were associated with advanced stage and high grade, regardless of their location. Besides establishing the prognostic significance of tumor infiltrating lymphocytes, another important question is: what determines the predominant lymphocytic subpopulation in ovarian cancers? The study by Zhang et al. suggests that higher levels of interferon gamma (IFNγ) and interleukin-2 (IL-2) and the chemokines induced by these cytokines are necessary for higher densities of tumor lymphocyte infiltration [3]. In a murine model of fibrosarcomas, Treg cells suppressed CD8+ T-cells through IFNγ production. They also
Fig. 2. Relationship between tumor infiltrating lymphocytes and survival. (a) T regulatory cells. (b) Cytotoxic T cells. (c) CTL/Treg ratio.
560
J.C. Barnett et al. / Gynecologic Oncology 116 (2010) 556–562
Fig. 3. Genes most differentially expressed and leave-one-out cross validation predictions between ovarian cancers with low and high Treg infiltration. (a) Columns correspond to cancers, with those from cancers with low infiltrating Tregs depicted to the left (n = 31) of the vertical black line and those from cancers with high infiltrating Tregs (n = 17) to the right. The 200 differentially expressed genes determined by the binary regression model are plotted in rows. High gene expression is represented in red and low expression in blue, while intermediate levels of expression are represented in shades of dark yellow through green to light blue. (b) A horizontal line is plotted at 0.5 indicating the classification threshold used for determining leave-one-out cross validation predictive accuracy (cases with probability of high Treg infiltration greater than 0.5 were declared high Treg infiltrating tumors; those less than 0.5 as low Treg infiltrating tumors). The horizontal axis lists the 48 samples with low (empty circles) and high (filled circles) Treg infiltrating tumors used in the regression. The corresponding values on the vertical axis are estimated classification probabilities with corresponding 90% probability intervals marked as lines to indicate uncertainty about these estimated values. The analysis and prediction for each sample are based on the subset of 200 most discriminatory genes. (c) A single-variable scatter plot of a significance test of the predicted probabilities of high infiltrating Treg cells (P b 0.001, Mann–Whitney U-test).
demonstrated reversal of this suppression by blocking the effects of interleukin 10 (IL-10) and transforming growth factor Beta (TGF-β), cytokines implicated in Treg recruitment [15]. TGF-β was also found to convert CD4+CD25− T-cells into CD4+CD25+ (Treg cells) when added to ovarian cancer cell lines [16], and Wei et al. showed that administration of IL-2 to patients with ovarian cancer affects the concentration and function of Treg cells in peripheral blood [17]. Curiel et al. demonstrated that ovarian cancer cells in mice produce chemokine ligand 22 (CCL22), while Treg cells express the receptor to CCL22–chemokine receptor 4 (CCR4). They showed that migration of Treg cells into ovarian tumors could be blocked using antibodies to CCL22, suggesting that this chemokine is a chemoattractant necessary for Treg recruitment [4]. Although these cytokines and chemokines may have a role in recruitment of tumor infiltrating lymphocytes, there was no difference in mRNA expression of these genes between
samples with high vs. low Treg infiltration in our study. However, several immunologic genes were differentially expressed, including increased expression of the chemokines CXCL10 (IP-10) and CCL5 (RANTES), both of which are involved in T-cell recruitment [18,19]. Further studies of the relationship between these chemokines, Tregulatory cells, and the ratios of different tumor infiltrating lymphocytes (CD8, CD3) could add insight into the immunologic microenvironment associated with immune evasion. Although immune cells comprise a relatively small fraction of most non-hematologic human cancers, genomic signatures have been described that involve differential expression of genes involved in immune response. In this regard, Tothill et al. defined six distinct ovarian cancer subtypes based on signatures from gene expression microarrays. One of these was an “immunological subtype” that was characterized by differential expression of several immunologic related genes and these
Fig. 4. Schematics depicting antigen presentation pathway and immune related gene networks. (a) Diagram of antigen presentation pathway showing upregulated genes shaded red in cancers with high Treg infiltration (©2000–2008 Ingenuity Systems, Inc.). (b) Diagram depicting a 35 gene network involving immunological disease, antigen presentation, and cell mediated immune response with 21 upregulated genes (red) and one downregulated gene (green) in cancers with high Treg infiltration. Differentially expressed genes are listed with respective fold changes in supplementary data table B (©2000–2008 Ingenuity Systems, Inc.).
J.C. Barnett et al. / Gynecologic Oncology 116 (2010) 556–562
561
562
J.C. Barnett et al. / Gynecologic Oncology 116 (2010) 556–562
cancers demonstrated higher numbers of tumor infiltrating CD3+ lymphocytes and more favorable survival [20]. In the present study, we demonstrated the existence of a genomic signature for high Treg infiltration that had a 77% predictive accuracy using leave-one-out cross validation. We also showed that the most differentially expressed genes between cancers with the highest and lowest infiltrating Treg cells are involved in pathways and biologic processes related to immune function. The antigen presentation pathway was the canonical pathway most highly associated with high Treg infiltration. This pathway is used by the cell to prepare and then present antigenic oligopeptide fragments on the surface of the cell to nearby T-cells allowing for its activation [21]. A limitation of the present study is the inclusion of only CD8 and FOXP3 markers in the evaluation of tumor infiltrating lymphocytes. While CD3 and CD4 are markers that add further insight into the immunologic tumor microenvironment, and have been included in other studies [3,5], the focus of the present study was to evaluate Tregulatory cells, their relationship to CTLs, and the genomic expression profile of tumors with increased numbers of Tregs. Another limitation is that CD8 is also a marker of certain subsets of dendritic antigen presentation cells [22]. Further staining could delineate the specific type and level of cytotoxicy of these CD8+ tumor infiltrating cells. Additionally, previous studies have also suggested a role for CD8+ Tregulatory cells; evaluation of this lymphocytic subset could also provide further insight into tumor immunity [23]. A better understanding of the mechanisms that underlie host immune tolerance to cancers is of paramount importance in developing immunologic treatment strategies. Currently, the FDA has approved denileukin diftitox, a ligand toxin fusion of full length IL2 and diphtheria toxin, for treatment of CD25+CD4+ cutaneous Tcell leukemia and lymphoma. This drug works through depletion of CD25+ Treg cells and is being tested in treatment of other cancers, including ovarian cancers [1]. FOXP3 has also been studied as a therapeutic target with vaccination against FOXP3 being associated with enhanced tumor immunity [24]. Overcoming immune tolerance is an immunotherapeutic strategy with potential promise in the treatment of ovarian cancer. A better understanding of the mechanisms underlying attenuated tumor immunity may facilitate therapeutic approaches that complement existing cytotoxic therapies. Conflict of interest statement The authors declare there are no conflicts of interest.
Acknowledgments This work was supported by the University of Alabama Ovarian Cancer Specialized Program of Research Excellence (SPORE), the Gail Parkins Ovarian Cancer Research Fund (AB), and the American Cancer Society. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.ygyno.2009.11.020.
References [1] Zou WP. Immunosuppressive networks in the tumor environment and their therapeutic relevance. Nat Rev, Cancer 2005;5:263–74. [2] Heintz AP, Odicino F, Maisonneuve P, et al. Carcinoma of the ovary. FIGO 6th Annual Report on the Results of Treatment in Gynecological Cancer. Int J Gynaecol Obstet 2006;95:S161–92. [3] Zhang L, Conejo-Garcia JR, Katsaros D, et al. Intratumoral T cells, recurrence, and survival in epithelial ovarian cancer. N Engl J Med 2003;348:203–13. [4] Curiel TJ, Coukos G, Zou L, et al. Specific recruitment of regulatory T cells in ovarian carcinoma fosters immune privilege and predicts reduced survival. Nat Med 2004;10:942–9. [5] Sato E, Olson SH, Ahn J, et al. Intraepithelial CD8+ tumor-infiltrating lymphocytes and a high CD8+/regulatory T cell ratio are associated with favorable prognosis in ovarian cancer. Proc Natl Acad Sci U S A 2005;102:18538–43. [6] Leffers N, Gooden Mm, de Jong RA. Prognostic significance of tumor-infiltrating Tlymphocytes in primary and metastatic lesions of advanced stage ovarian cancer. Cancer Immunol Immunother 2009;58:449–59. [7] Wolf D, Wolf AM, Rumpold H, et al. The expression of the regulatory T cell-specific forkhead box transcription factor FoxP3 is associated with poor prognosis in ovarian cancer. Clin Cancer Res 2005;11:8326–31. [8] Bignone PA, Banham AH. FOXP3+ regulatory T cells as biomarkers in human malignancies. Expert Opin Biol Ther 2008;8:1897–920. [9] Nelson BH. The impact of T-cell immunity on ovarian cancer outcomes. Immunol Rev 2008;222:101–16. [10] Berchuck A, Iversen ES, Luo J, et al. Microarray analysis of early stage serous ovarian cancers demonstrates profiles predictive of favorable outcome. Clin Cancer Res 2009;15:2448–55. [11] Berchuck A, Iversen ES, Lancaster JM, et al. Patterns of gene expression that characterize long-term survival in advanced stage serous ovarian cancers. Clin Cancer Res 2005;15:3686–96. [12] West M, Blanchette C, Dressman H, et al. Predicting the clinical status of human breast cancer by using gene expression profiles. Proc Natl Acad Sci U S A 2001;98: 11462–7. [13] Shah CA, Allison KH, Garcia RL, Gray H, Goff BA, Swisher EM. Intratumoral T cells, tumor-associated macrophages, and regulatory T-cells: Association with p53 mutations, circulating tumor DNA and survival in women with ovarian cancer. Gynecol Oncol 2008;109:215–9. [14] Salama P, Phillips M, Grieu F, et al. Tumor-infiltrating FOXP3 T regulatory cells show strong prognostic significance in colorectal cancer. J Clin Oncol 2009;27: 186–92. [15] Ping Y, Lee Y, Liu W, et al. Intratumor depletion of CD4+ cells unmasks tumor immunogenicity leading to the rejection of late-stage tumors. J Exp Med 2005;2015:779–91. [16] Li X, Ye F, Chen H, Lu W, Wan X, Xie X. Human ovarian carcinoma cells generate CD4+CD25+ regulatory T cells from peripheral CD4+CD25− T cells through secreting TGF-B. Cancer Lett 2007;253:144–53. [17] Wei S, Kryczek I, Edwards RP, et al. Interleukin-2 administration alters the CD4 +FOXP3+ T-cell pool and tumor trafficking in patients with ovarian carcinoma. Cancer Res 2007;67:7487–94. [18] Nesbeth Y, Scarlett U, Cubillos-Ruiz J, et al. CCL5-Mediated endogenous antitumor immunity elicited by adoptively transferred lymphocytes and dendritic cell depletion. Cancer Res 2009;69:6331–8. [19] Dufour JH, Dziejman M, Liu MT, et al. IFN-gamma-inducible protein 10 (IP-10; CXCL10)-deficient mice reveal a role for IP-10 in effector T cell generation and trafficking. J Immunol 2002;168(7):3195–204. [20] Tothill RW, Tinker AV, George J, et al. Novel molecular subtypes of serous and endometrioid ovarian cancer linked to clinical outcome. Clin Cancer Res 2008;14: 5198–208. [21] Weinberg RA. Crowd control: tumor immunology and immunotherapy. The Biology of Cancer. Garland Science. Taylor and Francis Group, LLC; 2007. [22] Kronin V, Wu L, Gong S, et al. DEC-205 as a marker of dendritic cells with regulatory effects on CD8 T cell responses. Int Immunol 2000;12(5):731–5. [23] Xystrakis E, Dejean AS, Bernard I, et al. Identification of a novel natural regulatory CD8 T-cell subset and analysis of its mechanism of regulation. Blood 2004;104 (10):3294–301. [24] Nair S, Boczkowski D, Fassnacht M, Pisestky D, Gilboa E. Vaccination against the Forkhead family transcription factor Foxp3 enhances tumor immunity. Cancer Res 2007;67:371–80.