The immune system is potentially able to develop immune responses against self antigens and tumor-associated antigens. Regulatory T cells (Tregs) play a crucial role in regulating the balance between the attack and tolerance of such antigens. Regulatory CD4+ CD25+ FOXP3+ T cells are a subset of T cells mediating immunological tolerance by suppressing the activation of autoreactive T cells. Tregs not only inhibit the development of autoimmune-mediated diseases such as inflammatory bowel disease or intestinal graft-versus-host disease [8,9] but are also suspected of impeding immunosurveillance against autologous tumor cells [10]. Tregs have been found to be increased in the peripheral blood of patients suffering from various malignant diseases. Wolf et al. reported an increase of Tregs in peripheral blood of patients with epithelial malignancies [11]. Sasada et al. [12] described an increased percentage of Tregs among lymphocytes in peripheral blood of patients with gastrointestinal cancers. An increase of Tregs in peripheral blood has also been reported for hepatocellular cancer and non-small cell lung cancer [13,14]. Kono et al. [15] found higher Tregs in patients with advanced gastric and esophageal cancer than in those with limited disease. Ichihara et al. [16] reported an increase of Tregs not only in the peripheral blood of patients with gastric and esophageal cancer but also in the tumor itself compared to healthy mucosa. Moreover, they demonstrated a higher Treg infiltration in advanced gastric/esophageal cancer [16]. All of this suggests a role of Tregs in tumor development, which may be related to immunosuppression in humans.

Several mouse studies have shown that depletion of regulatory T cells improves anti-tumor immunity [17-19]. For example, Ko et al. [20] found FOXP3+CD25+CD4+ Tregs predominantly in growing tumors in mice that had not received an agonistic monoclonal antibody against glucocorticoid-induced tumor necrosis factor receptor family-related protein (GITR), thus suggesting that tumor-infiltrating natural Tregs may hamper effective tumor immunity. A similar conclusion can be drawn from the work of Turk et al. [21], which shows that melanomas were immunologically rejected after Treg inhibition, and an improvement of anti-tumor T cell response was obtained when Tregs had been depleted before immunization of mice. In a murine colon cancer model, Ghiringelli et al. [22] demonstrated that Tregs suppress anti-tumor immunity and that this effect can be circumvented when Tregs are depleted by cyclophosphamide. This stimulated intensified research on the interaction between Tregs and adoptive CD8+ T cell transfer. Antony et al. [23] described a reduced effect of adoptively transferred tumor-antigen-directed CD8+ T cells when given together with Tregs in mice. Interestingly the in vitro induction of tumor-antigen-specific (anti-NY-ESO-1) antibodies required Treg depletion from seronegative cancer patients and healthy donors [24]. This is in accordance with data by Rosenberg & Dudley [25], which suggest that Treg depletion may also be a prerequisite for a clinically effective adoptive T cell transfer in humans.

As mentioned above, Tregs are suspected of reducing T cell activity, but their role in the development of tumor-antigen-specific T cell responses has not yet been investigated in detail. Furthermore, it is not known whether the intratumoral presence of Tregs has an impact on the antigen-specific T cell response, clinical course, and survival of CRC patients. With regard to tumors of other organs, Albers et al. [26] described an increased number of regulatory and antigen-specific T cells in head and neck cancer. Yu et al. [27] observed Treg accumulation at the tumor site. They were able to show in a murine fibrosarcoma model that Treg inhibition resulted in a changed cytokine milieu in the tumor and an eradication of tumors. Viguier et al. (28) reported that the number of FOXP3 expressing regulatory T cells is increased in human metastatic melanoma lymph nodes. These cells inhibited in vitro proliferation and cytokine production of CD8 +T cells [28]. It was shown for ovarian cancer that tumoral Tregs suppress tumor-specific T cell immunity and contribute to reduced survival [29]. Another study on ovarian cancer patients found that not only intraepithelial CD8+ tumor-infiltrating lymphocytes but also a high CD8+/CD25+FOXP3+ regulatory T cell ratio (i.e., low relative FOXP3 infiltration) is associated with a favorable prognosis [30]. Moreover, Wolf et al. [31] reported that a high expression level of FOXP3 in ovarian cancer is associated with a poorer prognosis. Briefly, the latter three studies suggest that high Treg infiltration is associated with a poorer prognosis in ovarian cancer. Overall, targeting Tregs may have an important impact on immunotherapeutic anti-cancer strategies and the clinical outcome of cancer patients [19].

Tregs infiltrated all 40 CRC samples with higher Treg frequency in the stroma than in the tumor epithelium. In agreement with prior studies in various malignant diseases [16,26-28], we found significantly higher numbers of Tregs in CRC than in normal colon tissue. Although counts differ from patient to patient and, in particular, are relatively low in the epithelium, we found an approximately 20-fold increase of infiltrating Tregs in CRC compared to healthy colon.

Increased regulatory T cells in peripheral blood have been associated with advanced gastrointestinal tumor [15]. Interestingly, we found an approximately 2-fold increase of Tregs at the tumor site in CRC patients with limited compared to metastatic disease (p < 0.05). This difference is mainly based on a higher stromal Treg infiltration in limited CRC. The reason for this observation is unknown. However, it underlines the importance of in situ analyses. We speculate that migration of Tregs away from the primary tumor site during metastatic spread might be important in this setting, which would explain the Treg increase in peripheral blood as described by others. Furthermore, our data show that local Treg infiltration in malignant disease may be a dynamic process, with increased numbers as an initial response in limited disease followed by a relative decrease after metastatic spread.

In CRC, the average Treg infiltration rate was slightly higher in patients without systemic TAA-specific T cell response. Although these differences were not significant, this may be a first indication of a T cell response-suppressing role of tumor-infiltrating Tregs in CRC. In this context, the above described increased Treg infiltration in limited disease is particularly interesting because systemic T cell responses have been detected significantly less often in patients with limited disease (3, 4). This might suggest an indirect interaction between systemic T cell response and Treg infiltration. Of note, the herein presented T cell responses against the TAA EpCAM, her-2/neu, and CEA represent only a small proportion of potential T cell targets in CRC.

Evaluating a possible interaction between Treg infiltration and survival required a stage-corrected analysis due to the stage-dependency of survival and Treg infiltration. No association was found between total Treg infiltration and survival neither with nor without stage correction. In a descriptive post-hoc analysis a weakly better survival was found for patients with higher epithelial Treg infiltration after stage correction. This subset analysis must not be over-interpreted and must be seen with great caution because of multiple testing and the low number of epithelial Treg in CRC. Nonetheless, assuming that high Treg infiltration reduces anti-tumor immunity and thus survival, our results may seem surprising at first glance; and, e.g., in ovarian cancer, high Treg infiltration seems to be an adverse prognostic factor [29-31]. However, contrasting results are reported for various neoplasms. Erdman et al. [42] have shown that adoptive transfer of Tregs into mice significantly inhibited the development of microbially induced colon carcinoma. In a study on human Hodgkin lymphoma, it was shown that low FOXP3+ cell numbers correlated with negative clinical outcome [43]. Additionally, a recent study demonstrated that high numbers of tumor infiltrating FOXP3-positive regulatory T cells are associated with improved survival in follicular lymphoma [44]. It is possible that in some malignant disease – so far preferentially lymphoproliferative or bacteria-induced diseases – infiltrating Tregs may have a protective effect by reducing/delaying the development of an aggressive and cytotoxic, potentially proliferogenic cytokine milieu, which is the basis for the inflammation-driven progress of malignant diseases [45,46]. In accordance with our study, Pages et al. [47] measured FOXP3 at the RNA level in CRC and found no difference in survival comparing high and low expression.

Since CD8+ T cell tumor infiltration has been described as a positive prognostic factor in CRC [32-37], we performed an analysis of CD8 T cell infiltration in relation to disease stage and survival. First, based on our in situ approach, it was interesting to see that the increased CD8 infiltration in CRC compared to healthy colon was based on increased stromal CD8 tumor infiltration. Furthermore, we found a reversion of CD8 T cell distribution from higher epithelial CD8 infiltration in healthy colon to higher stromal CD8 infiltration in CRC suggesting a CD8 T cell shift to the stroma during tumor development. In accordance with the above mentioned previous studies, we found a trend for better survival in CRC patients with higher stromal CD8 infiltration. This may be due to an increase of CD8 infiltration in limited disease compared to metastatic disease as we found in our study. Thus, for further analyses on the influence of CD8 infiltration of CRC, it has to be taken into account that CD8 infiltration may be higher in the limited stages of the disease. Since ratios between Tregs and CD8 T cells in relation to survival were used in a prior study in ovarian cancer [30], we analyzed the association of the Treg/CD8 ratio with survival. No correlation was found. This difference between findings in ovarian cancer and CRC may be due to specific biological characteristics of each tumor type during progression with predominantly local spread in ovarian cancer and a higher tendency for systemic metastasis (e.g. to the liver) in CRC. While no correlation was found between stromal Tregs and CD8 infiltration in our study, it was interesting to see that Tregs and CD8 infiltration weakly correlated in the epithelial tissue.

Regulatory T cells are seen as a hindrance for clinically successful tumor-directed T cell responses particularly in discussions on active specific immunization and adoptive T cell transfer in cancer patients [25]. Our data do not directly support this assumption. There are only indirect links between T cell response and Treg infiltration: the Treg infiltration is higher in patients without systemic TAA-specific T cell response at a non-significant level, and Tregs are increased in patients with limited disease, in whom T cell responses are less frequent. We only begin to understand the relationships between tumor-infiltrating Treg, stage of disease, and systemic T cell response against TAA; and they have to be further investigated.