Tumor-Infiltrating Foxp3⁻CD4⁺CD25⁺ T Cells Predict Poor Survival in Renal Cell Carcinoma

In the United States, an estimated 38,890 new cases of renal cell carcinoma (RCC) are predicted to occur in 2006, with 12,840 deaths (1). Despite the progress in radiographic testing, almost one third of patients present with advanced disease, either locally advanced or metastatic. Surgery remains the mainstay treatment of localized disease with excellent long-term survival. Unfortunately, median survival with metastatic disease is only 13 months with a 5-year survival of <10% (2). Consequently, it is imperative to identify alternative medical and surgical treatment modalities to effectively manage advanced RCC.

RCC is considered to be an immunogenic cancer, with pathologic specimens frequently harboring large numbers of tumor-infiltrating lymphocytes (TIL; refs. 3, 4). Reports of spontaneous regression and limited responses to cytokine therapies support this view. Biological modifiers, including interleukin-2 (IL-2) and IFN-α, have, at best, induced modest rates of response against metastatic RCC, ranging from 10% to 20% (2, 5). Evolving therapies with stem cell transplantation (6, 7) and tumor vaccines (8) have also exhibited promise in preliminary investigations. Although these novel immune-based approaches show some clinical activity, it seems that their effectiveness is limited by one or more mechanisms, such as tumor-induced immune suppression in the tumor microenvironment.

One mechanism by which cancers evade immune destruction is by recruiting regulatory cells into the tumor microenvironment, including regulatory T cells (Tregs). In the last few years, there has been a resurgence of interest among immunologists about T-cell–mediated immune suppression. T cells were first shown to suppress immune responses in the 1970s (9, 10). Although some work continued through the 1980s (11, 12), the failure to identify antigen-specific factors and unique identifying molecules that caused immune suppression led to stagnation of this field. Interest was rekindled in the mid-1990s when Sakaguchi et al. (13) showed that a small population of CD4⁺ T cells that coexpressed the IL-2 receptor α-chain (CD25) could control autoreactive CD4⁺ T cells in vivo. In vitro murine studies have since confirmed that certain CD4⁺CD25⁺ T cells can function as potent inhibitors of T-cell–mediated immunity (14, 15), and subsequent studies have revealed a similar CD4⁺CD25⁺ T-cell population in humans (16, 17).

Our current understanding now suggests that Tregs constitute a family of specialized T cells that can be divided into two broad subsets: natural and adaptive. The natural Treg population develops as a separate T-cell lineage in the thymus (18, 19). These cells are antigen specific and populate in the periphery to guard against autoimmune reactions. The second subset of “adaptive” Treg cells seems to develop from mature, peripheral CD4⁺ T cells in response to tissue-specific or foreign antigens and mediates regulation through the release of soluble cytokines (IL-10; refs. 18, 19). However, one problem with the study of Tregs is the lack of specific cell markers that distinguish them from other lymphocytes. Recently, the ability to characterize Tregs improved with the identification of the transcription factor Foxp3, which has been shown to be a critical component of natural Treg generation and function (20). Murine and human natural CD4⁺CD25⁺ Tregs express high levels of Foxp3 compared with other lymphocytes, permitting its use as a relatively specific marker. Although research into the effect of Tregs on human cancer has been limited, several key studies have emerged in recent years showing their importance. Cancer patients seem to have increased numbers of both peripherally circulating and tumor-associated CD4⁺CD25⁺ Tregs (21, 22). Murine studies have shown that antibody- or immunotoxin-mediated depletion of CD25⁺ cells can enhance tumor immunity and rejection (23, 24). Curiel et al. (25) were the first to show a direct link between Tregs and human tumor antigen-specific T-cell responses and positively correlate with poorer prognosis. In the current study, we evaluated intratumoral CD4⁺CD25⁺Foxp3⁺ Tregs as well as CD4⁺CD25⁺Foxp3⁻ T cells in RCC. Our findings reveal that it is the Foxp3⁻ subset of CD4⁺CD25⁺ T cells and not the Foxp3⁺ subset that correlates with worse pathologic features of RCC and cancer-specific survival.

Patient selection. On approval from the Institutional Review Board, we identified 196 patients treated with radical nephrectomy or nephron-sparing surgery for unilateral, sporadic clear cell RCC between 2000 and 2002 who had fresh-frozen tumor tissue available for study (26). Of these, 170 (86.7%) patients had an adequate tumor specimen available for staining. There was no difference in patient outcome between those with and without tissue available for study (P = 0.372, log-rank test). The cancer-specific survival rates (SE, number still at risk) at 2 years for patients with and without tissue available for study were 83.7% (2.9%, 134) and 88.5% (6.3%, 20), respectively.

Pathologic features. The pathologic features examined included histologic subtype, tumor size, the 2002 tumor-node-metastasis (TNM) stage groupings, nuclear grade, coagulative tumor necrosis, the presence of lymphocytic infiltration, and the presence of tumor B7-H1 expression. The microscopic slides from all specimens were reviewed by a urologic pathologist (J.C.C.) without knowledge of patient outcome. Histologic subtype was classified according to the Union Internationale Contre le Cancer, American Joint Committee on Cancer, and Heidelberg guidelines (27, 28). Nuclear grade was assigned using standardized criteria as described previously (29). Histologic tumor necrosis was defined as the presence of any microscopic coagulative tumor necrosis (30). B7-H1 is a costimulatory glycoprotein whose aberrant expression by tumor cells has been shown to increase the risk of RCC-specific death (26). B7-H1 immunohistochemical staining and quantification were done as described previously (26).

Immunohistochemistry. Frozen sections of RCC were brought to room temperature and fixed in −20°C acetone for 10 min and air dried at room temperature for 30 min. Sections were rehydrated in PBS and blocked for endogenous avidin/biotin using an Avidin/Biotin blocking kit (Vector Laboratories, Burlingame, CA). Sections were immunostained with mouse anti-human CD25 IgG1 at 1:10 dilution (DAKO, Carpinteria, CA) and visualized with Fluorescein Avidin DCS at a 1:300 dilution (Vector Laboratories) followed by staining with mouse anti-human CD4 IgG1 at a 1:100 dilution (BD PharMingen, San Diego, CA) and visualized with Texas Red Avidin DCS at a 1:400 dilution (Vector Laboratories) using the multiple immunofluorescent staining method of the MOM kit (Vector Laboratories). For Foxp3 immunostaining, sections were blocked for avidin/biotin and incubated for 5 min in 0.1% goat serum in PBS and then incubated with rabbit anti-human Foxp3 IgG at a 1:75 dilution (Abcam, Cambridge, MA) for 30 min. Sections were incubated with biotinylated antirabbit IgG at a 1:200 dilution (Vector Laboratories) for 30 min, washed with PBS, and incubated for 5 min with AMCA Avidin D diluted at 1:800 (Vector Laboratories). Coverslips were mounted onto the sections using Prolong Gold mounting medium (Invitrogen, Carlsbad, CA).

Confocal microscopy and quantification of Tregs. Using a LSM 510 confocal microscope (Zeiss, Jena, Germany) at a 400× objective, photomicrographs were taken of 170 tissue specimens. A total of 12 photomicrographs of random nonoverlapping high powered fields were taken for every section. Images were viewed using LSM Image Browser software at a screen resolution of 1,280 × 1,024 pixels. T cells were counted by randomly choosing 10 of the 12 photomicrographs that were then analyzed for the presence or absence of CD4⁺, CD25⁺, and Foxp3⁺ stained cells within identically photographed fields. Cells were considered Foxp3⁺ based on strong intranuclear expression (Fig. 1A). Cells were considered positive for CD4 or CD25 based on the presence of intense circular membranous staining (Fig. 1B). Numbers of CD4⁺CD25⁺Foxp3⁻ and CD4⁺CD25⁺Foxp3⁺ T cells (in merged photomicrographs) were divided by total numbers of CD4⁺ T cells to obtain percentages of CD4⁺CD25⁺Foxp3⁻ and CD4⁺CD25⁺Foxp3⁺ T cells within RCC tumor specimens.

Statistical methods. Comparisons of the presence and percentage of CD4⁺CD25⁺Foxp3⁻ and CD4⁺CD25⁺Foxp3⁺ T cells with pathologic features of interest were evaluated using χ² and Fisher's exact tests. Cancer-specific survival was estimated using the Kaplan-Meier method. The duration of follow-up was calculated from the date of nephrectomy to the date of death or last follow-up. Cause of death was determined from the death certificate and physician correspondence. The associations of the presence of CD4⁺CD25⁺Foxp3⁻ and CD4⁺CD25⁺Foxp3⁺ T cells with death from RCC were evaluated using Cox proportional hazards regression models univariately and after adjusting for lymphocytic infiltration, positive tumor B7-H1 expression, and the Mayo Clinic Stage, Size, Grade, and Necrosis Score, a prognostic, multivariate, composite score specifically developed for patients with clear cell RCC (31). Statistical analyses were done using the SAS software package (SAS Institute, Cary, NC). All tests were two sided and P values <0.05 were considered statistically significant.

Tumor and blood samples. RCC tissue specimens were obtained intraoperatively from patients who either underwent radical nephrectomy or nephron-sparing surgery for unilateral, sporadic RCC, pathologically determined to be the clear cell histologic subtype on frozen section.

TILs were isolated from a 1- to 10-cm specimen of tumor tissue obtained intraoperatively using sterile technique on kidney removal. Tissue was enzymatically dissociated in digestion medium, consisting of HBSS with 30 units/mL hyaluronidase V, 30 units/mL DNase I type IV (both from Sigma, St. Louis, MO), 400 units/mL collagenase type IV, and antibiotics (100 units/mL penicillin, 100 μg/mL streptomycin, and 100 μg/mL gentamicin, all from Life Technologies, Carlsbad, CA). Tissue was then placed in a 60-mm tissue culture disc with a sterile stirring bar with 5 mL of the digestion medium and incubated at 37°C for 30 min. Every 30 min, medium was removed and cells were washed twice with PBS + 5% fetal bovine serum (Life Technologies, Langley, OK). This step was repeated six times. After the last cycle, cell pellets were pooled and centrifuged through a sucrose density gradient (Lymphoprep, Accurate Chemical and Scientific Corp., Westbury, NY). The buffy coat layer, representing the TIL population, was collected and washed twice with PBS + 5% fetal bovine serum. The TIL population was counted and cryopreserved at −80°C for later staining.

Normal donor buffy coat cells were obtained from the Mayo blood bank and peripheral blood mononuclear cells were obtained by sucrose gradient density centrifugation as described above.

Flow cytometry staining. Antibodies were purchased from BD Biosciences (San Diego, CA) unless otherwise stated. Before staining, the frozen TILs were thawed at 37°C for 2 min, washed, and counted. The TILs were adjusted to 10⁷ cells/mL in RPMI 1640 + 10% fetal bovine serum containing 1.0 μg/mL anti-CD3 antibody with 0.7 μg/mL BD GolgiStop and 1.0 μg/mL BD GolgiPlug (BD Biosciences) for 5 h. Cells were then washed with and resuspended in fluorescence-activated cell sorting buffer [PBS with 3% of fetal bovine serum with 1 mmol/L EDTA (Life Technologies, Carlsbad, CA)]. Fc receptors were blocked with 2% human serum for 20 min at 4°C. Cells were stained with the appropriate surface antibodies for 30 min at 4°C in the dark and washed twice. TILs were then subjected to intracellular staining for Foxp3, IL-10, or transforming growth factor-β intracellular staining using a commercial kit (eBioscience, San Diego, CA) following the manufacturer's directions.

Surface expression of immune molecules was quantified by flow cytometry using fluorescence-activated cell sorter (Becton Dickinson FACSCalibur, Franklin Lakes, NJ) using CellQuest software. Cell areas of interest were gated, and ∼5 × 10⁴ events were scored and included for analyses, although typical collection events ranged from 10 × 10⁴ to 20 × 10⁴ events. Matched isotype controls were used to determine threshold for positive expression. CD4⁺CD25⁺ lymphocytes were gated, and percentage of positive for intracellular IL-10 and transforming growth factor-β expression was determined. The appropriate nonparametric test was used to assess group differences (Kruskal-Wallis) and to compare two populations (Wilcoxon rank-sum).

Because the study was limited by the availability of tumor specimens, no a priori attempts at power calculations or sample size requirements were made. Statistical analyses were done using the SAS software package. All tests were two sided and P values <0.05 were considered statistically significant.

Pathologic features and patient follow-up. A summary of pathologic features for the 170 patients studied is shown in Table 1. At last follow-up, 46 patients had died, including 37 who died from RCC at a median of 1.4 years following nephrectomy (range, 0-4.4). Among the 124 remaining patients, the median duration of follow-up was 3.7 years (range, 0-5.7). Two patients underwent treatment with IL-2 for metastatic disease. Consistent with previous findings using paraffin-embedded clear cell RCC specimens (32), the presence of lymphocytic infiltration in this cohort was significantly associated with death from RCC [risk ratio (RR), 2.14; 95% confidence interval (95% CI), 1.04-4.43; P = 0.039; Fig. 2A].

CD4⁺CD25⁺Foxp3⁺ T cells are found in a minority of clear cell RCC tumors and are not associated with poorer survival. Forty-three (25.3%) tumors had CD4⁺CD25⁺Foxp3⁺ T cells. The mean percentage of CD4⁺CD25⁺Foxp3⁺ cells for these 43 cases was 4.1% (median, 2.5; range, 0.4-25.0). The associations of CD4⁺CD25⁺Foxp3⁺ cells with pathologic features of the tumor are summarized in Table 2. The presence of CD4⁺CD25⁺Foxp3⁺ cells was not significantly associated with death from RCC univariately (RR, 1.28; 95% CI, 0.63-2.59; P = 0.493), after adjusting either for the presence of lymphocytic infiltration (RR, 1.19; 95% CI, 0.59-2.42; P = 0.627) or for positive tumor B7-H1 expression (RR, 1.01; 95% CI, 0.50-2.07; P = 0.972) or after adjusting for the Mayo Clinic Stage, Size, Grade, and Necrosis Score (RR, 1.20; 95% CI, 0.59-2.44; P = 0.612).

CD4⁺CD25⁺Foxp3⁻ T cells are associated with cancer-specific survival and pathologic features predictive of outcome from clear cell RCC. One hundred forty-three (84.1%) tumors had CD4⁺CD25⁺Foxp3⁻ T cells. The mean percentage of CD4⁺CD25⁺Foxp3⁻ cells (of all CD4⁺ T cells counted) for these 143 cases was 12.1% (median, 8.0; range, 0.4-60.0). There were 98 (57.7%) cases with 5% or more CD4⁺CD25⁺Foxp3⁻ cells and 64 (37.7%) cases with 10% or more CD4⁺CD25⁺Foxp3⁻ cells. The associations of CD4⁺CD25⁺Foxp3⁻ cells with pathologic features of the tumor are summarized in Table 3. CD4⁺CD25⁺Foxp3⁻ cells were significantly associated with tumor size (P = 0.02), TNM pathologic stage (P = 0.012), and coagulative tumor necrosis (P = 0.031).

The percentage of CD4⁺CD25⁺Foxp3⁻ cells (evaluated as a continuously scaled variable) was significantly associated with death from RCC (RR, 1.03; 95% CI, 1.01-1.05; P = 0.007). Each 1% increase in the percentage of CD4⁺CD25⁺Foxp3⁻ cells increased the risk of death from RCC by 3%. However, for illustrative purposes and ease of interpretation, the percentage of CD4⁺CD25⁺Foxp3⁻ cells was also analyzed using a cutpoint of 10% (i.e., halfway between the mean and median). The indicator variable for ≥10% CD4⁺CD25⁺Foxp3⁻ cells was significantly associated with death from RCC univariately (RR, 2.60; 95% CI, 1.35-4.98; P = 0.004; Fig. 2B). Cancer-specific survival (SE, number at risk) at 1, 2, and 3 years following nephrectomy for patients having tumors with <10% CD4⁺CD25⁺Foxp3⁻ cells were 94.3% (2.2%, 99), 88.6% (3.1%, 92), and 86.5% (3.4%, 72), respectively, compared with 88.5% (4.1%, 54), 75.0% (5.6%, 42), and 65.4% (6.3%, 29), respectively, for patients having tumors with ≥10% CD4⁺CD25⁺Foxp3⁻ cells (P = 0.004). The association with death persisted after adjusting for the presence of lymphocytic infiltration (RR, 2.53; 95% CI, 1.32-4.87; P = 0.005), after adjusting for positive tumor B7-H1 expression (RR, 2.53; 95% CI, 1.32-4.85; P = 0.005), and after adjusting simultaneously for the presence of lymphocytic infiltration and positive B7-H1 expression (RR, 2.45; 95% CI, 1.27-4.71; P = 0.007). However, the indicator variable for ≥10% CD4⁺CD25⁺Foxp3⁻ cells was not significantly associated with death from RCC after adjusting for the Stage, Size, Grade, and Necrosis Score (RR, 1.29; 95% CI, 0.65-2.56; P = 0.466).

IL-10 is up-regulated in CD4⁺CD25⁺Foxp3⁻ intratumoral T cells (TILs) but not in normal donors. To determine whether CD4⁺CD25⁺Foxp3⁻ could have inhibitory function, we did intracellular flow cytometry. Expression of IL-10 and Foxp3 in ccRCC TILs and healthy donors was assessed and the results are depicted in Fig. 3A and B. A higher percentage of Foxp3⁻ CD4⁺CD25⁺ T cells expressed IL-10 (median, 43.8%; range, 10.6-70.2) compared with the Foxp3⁺ subpopulation (median, 7.2%; range, 0.6-23.7; P = 0.026). Similarly, Foxp3⁻ CD4⁺CD25⁺ T cells exhibited greater expression of transforming growth factor-β (median, 20.7%; range, 7.1-49.0) compared with Foxp3⁺ T cells (median, 8.8; range, 0.3-26.6), although this did not reach statistical significance (P = 0.064).

We have reported previously that increased lymphocytic infiltration is independently associated with death from RCC (32). That study, coupled with a report that Tregs were associated with poor outcome in ovarian cancer (25), provided the impetus to identify specific lymphocyte populations associated with RCC survival. To our knowledge, this study is the first report examining the relationship between TIL Foxp3⁺ and Foxp3⁻ expression among CD4⁺CD25⁺ T cells and their effect on cancer-specific survival in patients with RCC.

Given that CD4⁺CD25⁺Foxp3⁺ Tregs are elevated in the periphery of RCC patients compared with normal subjects and may be associated with blocking peripheral immune responses to RCC antigens (21), we expected this population of cells to be increased in TILs among patients with more aggressive forms of RCC. Contrary to our expectations, however, CD4⁺CD25⁺Foxp3⁺ cells were identified only in a minority of patients (25%) and not found to be significantly associated with RCC pathology or patient outcome. In human malignancies, it remains controversial as to whether infiltrating Foxp3⁺ Tregs are associated with disease biology. Alvaro et al. (33) found that increased Foxp3 expression within reactive lymphocytes in Hodgkin's lymphoma patients was associated with improved survival. In contrast, Wolf et al. (34) have reported recently that intratumoral Foxp3 expression in ovarian cancer was associated with an increased risk of death. In mice, Foxp3 has an important role in immunity and is required for thymic production of CD4⁺CD25⁺ Tregs. Retroviral transduction of murine CD4⁺CD25⁻ T cells with Foxp3 induces a immunosuppressive phenotype, showing a critical role for Foxp3 in the generation and function of peripheral Tregs (35). In humans, however, the biology of Foxp3 is different (36). In contrast to the mouse, ectopic expression of Foxp3 renders human CD4⁺ T cells anergic but incapable of suppressing activation of CD4⁺CD25⁻ T cells (37). Thus, either Foxp3 cooperates with other factors to impart the CD4⁺ Treg phenotype or it is not involved. In any case, in RCC, it seems that, if in fact Tregs are among the intratumoral immune effectors, the presence of Foxp3 is not a marker of poorer outcome.

A clinically relevant finding in the current study was that patients with increased tumor infiltration by CD4⁺CD25⁺Foxp3⁻ T cells were 2.5 times more likely to die from RCC compared with patients with little or no infiltration of these cells. The increased presence of these cells within RCC tumors was also associated with advanced TNM stage, tumor size, and presence of coagulative tumor necrosis, all of which are known independent predictors of RCC death (31, 38). We also found that increased lymphocytic infiltration was associated with a poorer outcome (Fig. 2A), consistent with our previous findings in a different cohort (32). Importantly, even when we accounted for lymphocytic infiltration in our multivariate model, CD4⁺CD25⁺Foxp3⁻ T cells continued to be associated with cancer-specific death (RR, 2.53; P = 0.005). To our knowledge, the CD4⁺CD25⁺Foxp3⁻ T-cell phenotype has not been associated with an adverse outcome in any other human malignancy.

Why this specific subgroup of T cells is associated with worse RCC outcome remains unclear. It is conceivable that CD4⁺CD25⁺Foxp3⁻ T cells may represent a previously uncharacterized group of immunoinhibitory T cells that block antitumor immune effectors. In recent years, several studies have postulated the existence of Foxp3⁻ inducible Tregs (39–41). These inducible Tregs apparently acquire regulatory capabilities independent of Foxp3. For example, a recent study by Marshall et al. (42) strongly suggests that Hodgkin's lymphomas induce a subset of Tregs, called Tr1, which are capable of producing IL-10, similar to the cells we identified by flow cytometry. Other studies have shown that inducible Tregs can express CD25 in the absence of Foxp3 expression (36, 43).

A major challenge in interpreting CD25 expression, however, is that CD25 is the high-affinity IL-2 receptor α-subunit and is transiently up-regulated on effector T cells following activation. Hence, another possibility is that CD4⁺CD25⁺Foxp3⁻ cells are activated T cells, which increase in number in response to more immunogenic and aggressive tumors. Recent studies in breast cancer suggest that less immunogenic tumors are less aggressive, as shown in a study by Madjd et al. (44) who showed that tumors with reduced HLA class I expression behave less aggressively. This concept is consistent with our findings of superior survival in RCC patients harboring low levels of TILs (32). Last, tumors with more activated T cells could be more aggressive due to T-cell–derived cytokines or other factors that promote tumor growth. Cytokines that are known to increase the growth of renal cancers include tumor necrosis factor-α and IL-6 (45–47).

Immune modulation, both locally in the tumor microenvironment and systemically, provides a new window of opportunity in RCC therapeutics. Anti-CD25 monoclonal antibodies have been developed for clinical use in humans for the treatment of acute cellular rejection in allograft transplants (48, 49). Elimination of CD25⁺ T cells with anti-CD25 antibodies may unmask effective antitumor immunity and lead to an augmented antitumor response. Because CD25 is a component of the high-affinity IL-2 receptor, the use of IL-2 has also been proposed as a potential agent against Tregs. Denileukin difitox (Ontak) is an IL-2 diphtheria toxin fusion protein that has been shown to directly kill Tregs (36, 50). Dannull et al. (50) showed that denileukin difitox–mediated Treg suppression augmented vaccine responses against antigens likely due to down-regulation of Tregs. In another study, Cesana et al. (21) showed recently that suppression of CD4⁺CD25⁺ T-cell populations in RCC patients was associated with an objective response following high-dose IL-2 treatment. Collectively, these findings suggest that modulating specific T-cell subsets (e.g., Tregs) may be an effective approach for treating advanced RCC.

In conclusion, CD4⁺CD25⁺Foxp3⁺ Tregs were found in a minority of RCC specimens (25%) compared with CD4⁺CD25⁺Foxp3⁻ T cells (84%). CD4⁺CD25⁺Foxp3⁻ T cells were associated with higher TNM staging, larger tumor size, presence of coagulative tumor necrosis, and poorer cancer-specific survival. Further characterization of these subsets of T cells may help elucidate the immunologic milieu within RCC tumors, potentially leading to more targeted and effective immunotherapeutic strategies.