The prognostic effects of tumor infiltrating regulatory T cells and myeloid derived suppressor cells assessed by multicolor flow cytometry in gastric cancer patients

The prognostic effects of tumor infiltrating lymphocytes (TILs), especially regulatory T cells (Tregs) and myeloid derived suppressing cells (MDSCs) are inconclusive in gastric cancers. We investigated the frequencies of TILs including CD8+ T cells, CD45+CD4+CD25± FOXP3+ Tregs, CD45+CD11b+ CD14+ HLA−DR− MDSCs in 28 gastric cancer tissues by using multicolor flow cytometry. In gastric cancer tissue, the percentage of Tregs among the CD4+ T cell subset was substantially increased compared to that of Tregs among peripheral blood CD4+ T cells from the controls. High frequency of CD8+ T cells among CD3+ T cells correlated with increased overall survival (OS) (p = 0.005). High frequency of Tregs among CD4+ T cells correlated with increased OS (p < 0.001), and disease-free survival (DFS) (p = 0.039) and was an independent prognostic factor in OS (Hazard ratio: 0.047; 95% confidence interval, 0.006-0.372; p = 0.004). High frequency of MDSCs among total examined cells correlated with decreased OS (p = 0.027) and was an independent prognostic factor in OS (Hazard ratio 8.601; 95% confidence interval, 1.240-59.678; p = 0.029). We have demonstrated that high levels of Tregs among tumor-infiltrating CD4+ T cells were favorable, but an increased proportion of MDSCs was an adverse independent prognostic factor in gastric cancer. Our results may provide important insights for future immunotherapy in gastric cancer.


INTRODUCTION
Gastric cancer (GC) is the fourth most commonly diagnosed cancer and the second most common cause of cancer-related deaths worldwide [1,2]. Despite efforts to introduce new treatment modalities such as surgery combined with chemotherapy or chemo-radiotherapy, the control of GC at an advanced stage remains difficult [3]. Considering the high mortality rate of GC in advanced stage or at recurrence, new therapeutic strategies are urgently needed.
Mechanisms underlying the dynamic interplay between immune cells and tumor progression have been studied during past decades. The accumulated data indicate that the outcome of an immune response toward a tumor is largely determined by the type of immune response elicited [4]. Earlier studies have shown survival benefit of tumor infiltrating lymphocytes (TILs) [5][6][7][8], and this suggests that TILs are effective on suppressing tumor progression. Thereafter, subgrouping of CD3+ T cells into cytotoxic T cells and regulatory T cells (Tregs) have made further understanding of relationship between TILs and cancer progression. Infiltration of CD8+ cytotoxic T cells into tumors has been reported as a favorable prognostic factor in many cancers including GCs [9][10][11][12][13][14][15][16][17]. Tregs regulate immune reactivity including tumor-specific immune responses [18]. Tregs were initially characterized by the CD4+CD25+ phenotype, but the most specific cell marker of Tregs identified to date is the nuclear transcription factor known as Forkhead box protein P3 (FOXP3) [19][20][21][22][23]. Many studies have shown that a high density of Tregs, which usually evaluated by quantification of FOXP3+ T cells by using immunohistochemistry (IHC), was associated with a poor outcome, and these results were explained on the basis of the suppressive effects of Tregs on antitumor cytotoxic T cells [24][25][26]. However, many controversial results have been reported [24,27]. Moreover, the controversial prognostic significance of Tregs may differ depending on the specific tumor types. According to a recent meta-analyses, the prognostic impacts of Treg infiltration is not consistent [27]. For example, Treg is a favorable prognostic factor in colorectal carcinoma while it is an adverse prognostic factor in hepatocellular carcinoma [24]. In GC, the prognostic significance of Tregs has been controversial. Many studies reported Tregs as adverse prognostic factor in GC [25,[27][28][29], while several opposite results suggesting Tregs as a favorable prognostic factor have been reported so far [18,30,31].
Myeloid-derived suppressor cells (MDSCs) are a heterogeneous population of immature myeloid cells and are involved in inhibiting both innate and adaptive immune responses [32,33]. In cancer biology, MDSCs have been shown to potently inhibit antitumor immunity in vivo through diverse mechanisms [32,33]. MDSCs have been generally defined as a CD11b+ CD33+ CD14+ HLA− DR− myeloid cell population in human cancer patients [33]. In GC, two previous studies have shown that the numbers of MDSCs are increased in the blood of cancer patients compared with healthy individuals, and this increase was associated with adverse clinical outcomes [34,35]. However, the clinical significance of MDSCs in GC tissue remains completely unknown because specific phenotype of MDSCs cannot be evaluated by IHC in cancer tissue.
In this study, we examined the frequencies of tumor infiltrating immune cell subsets by multi-color flow cytometry, which enabled us to define more specific roles of immune cells as well as more objective quantification in GC, and investigated the clinical significance of immune cells, especially Treg and MDSC.

Distribution of immune cell subsets in gastric cancer tissue
The frequencies of the various immune cell subsets in the GC tissue are summarized in Table 1. The frequencies of immune cell subsets in PBMCs of 8 healthy individuals measured using same immunophenotyping panels are displayed. In GC tissue samples, CD45+ hematopoietic cells occupied 8.5% (median value, range 0.8-29.4%) of the total counted cells, and lymphocytes and myeloid cells occupied 68.4% and 31.6% of the CD45+ cells, respectively.

High frequency of CD8+ T cells among CD3+ T cells correlates with increased overall survival
The CD8 high /Total patient subgroup was associated with longer DFS and OS compared with the CD8 low / Total patient subgroup, but the tendency did not reach statistical significance ( Figure 1A, 1D). The proportion of the CD8 high /CD3 group was tended to increase according to the advancing N stage (p = 0.023) ( Table 3). The CD8 high /CD3 group showed longer OS (p = 0.005) and a tendency towards longer DFS than the CD8 low /CD3 group (p = 0.056 ) (Figure 2A, 2D).

High frequency of Tregs among CD4+ T cells correlates with increased overall survival and is an independent prognostic factor in overall survival
The Treg high /Total group showed significantly longer OS (p = 0.048) and a tendency of longer DFS compared to the Treg low /Total ( Figure 1B, 1E). The Treg high / CD4 group showed longer OS (p < 0.001)) and DFS (p = 0.039) than the Treg low /CD4 group ( Figure 2B, 2E) and remained a significant predictor of OS in the multivariate analysis (HR: 0.065; 95% CI, 0.009-0.491; p = 0.008) ( Table 4).

Low frequency of MDSCs among total examined cells correlates with increased overall survival and is an independent prognostic factor in overall survival
The MDSC low /Total group was more frequent in old age patients and the proportion of MDSC high /Total group decreased as the TNM stage advanced ( Table 2). The MDSC low /Total group showed a significantly longer OS (p = 0.027) and a tendency of longer DFS compared to the MDSC high /Total groups (p = 0.151) ( Figure 1C, 1F). MDSC/Total was an independent prognostic factor (HR 8.601; 95% CI, 1.240-59.678; p = 0.029) in OS (Table 4). MDSC high /CD45+ group showed better DFS rate (p = 0.038) and a tendency of better OS than MDSC low / CD45+ group (p = 0.142) ( Figure 2C, 2F).

DISCUSSION
In this study, we explored the frequencies of various immune cell subsets in GC tissues by multicolor flow cytometry and evaluated the prognostic value of intratumoral CD8+ cytotoxic T cell, Treg and MDSC frequencies. We found that GC tissues contained significant proportions of immune cells including Tregs and MDSCs and the percentages of Tregs among CD4+ T cells and of MDSCs among myeloid cells    were substantially higher when compared with those in peripheral blood from healthy individuals. Moreover, the high proportion of Tregs among CD4+ T cells was an independent prognostic factor for longer OS, whereas the increased proportion of MDSCs among total examined cells was associated with shorter OS. Infiltration of CD8+ cytotoxic T cells into tumors has been reported as a favorable prognostic factor in many cancers including GCs [10][11][12][13][14][15][16][17]. Here, we report that higher frequencies of CD8+ T cell among tumorinfiltrating CD3+ T cells are associated with longer OS. The prognostic significance of Tregs infiltration remains controversial and seems different depending on the specific tumor type. Many studies have shown that a high density of Tregs in IHC results is associated with a poor outcome, and these results were explained on the basis of the suppressive effects of Tregs on antitumor cytotoxic T cells [24][25][26]. However, a few recent studies have highlighted the protective effects of Tregs, especially in gastric or colorectal cancers occurring in an inflammatory background caused by microorganisms [9,14,[28][29][30]. In this context, Tregs were shown to inhibit the immune reaction of cytotoxic T cells in response to bacteria rather than cancer cells, and, thereby, to suppress cancer progression by controlling inflammation. In GC, the results of several studies on TILs focused on Tregs are controversial [9,28,[31][32][33]. This discrepancy could be explained by differences in technical issue; specific FOXP3 antibodies, scoring or statistical process. Moreover, most of the previous studies defined Tregs as FoxP3+ cells by IHC. Unlike previous studies using a single marker to detect a specific type of immune cell by IHC followed by manual quantification of positive cells, we used flow cytometry to analyze the cell populations at the single-cell level. This approach made possible to identify cell lineages or subsets by using multiple monoclonal antibodies at the same time and utilizing objective quantification with an automated system. So, overview of the immune cell population from the cancer tissue was possible and we determined the percentage of immune cells among the total number of examined cells as well as the proportion of each immune cells among their belonging subsets of immune cells.
In this study, we could define tumor infiltrating Tregs in detail as CD45+CD4+CD25± and FOXP3+ cells

cells (A, D), regulatory T cells (Tregs) (B, E) and Myeloid derived suppressor cells (MDSCs) (C, F) among total examined cells.
and found that a higher frequency of Tregs among total examined cells was associated with a longer OS, and this result is in line with several prior reports [9,33]. Furthermore, we found that the proportion of Tregs among CD4+ T cells was an independent favorable prognostic factor, which could not be evaluated by traditional IHC method. In addition to traditional concept of Tregs suppressing antitumor cytotoxic T cells, several other subsets of Treg have been identified, and direct tumor-killing Tregs activity has been reported [34,35]. The functional relevance of Treg-mediated tumor killing remains unknown. Further functional studies of Tregs may elucidate their role in the antitumor response and may help to explain the observed associations with the disease prognosis.
MDSCs are generally defined based on the combined expression of several markers and cannot simply be identified by IHC in cancer tissue. Therefore, MDSCs have been rarely studied in cancer tissue. Recently, Sun et al. reported that CD33−HLA−DR− MDSC were increased in the peripheral blood and tumor tissue of colorectal cancer patients compared to the peripheral blood of healthy controls [36]. They also found that an increased frequency of MDSCs was correlated with nodal/distant metastasis and tumor stage. In GC, two previous studies showed increased MDSCs in PBMCs from cancer patients and that MDSC frequency was associated with adverse clinical outcomes [37,38]. In the present study, we found that high MDSC percentages among total cells were an independent adverse prognostic factor in patients with advanced GC. This result may be explained by the role of MDSCs as inhibitors of antitumor immunity [39][40][41][42].
Although this prospective study provides valuable insights on de novo presence and relative quantity of multiple immune cell subsets in GC tissues, current study has limitations as follow: First, it was performed in relatively small number of patients with short follow up period. Second, the cutoffs, which were determined with the highest statistical significance for patient survival in presents study, need to be validated in an independent cohort. Third, the location of the immune cells within the tumor remains unknown whether the immune cells are in the intraepithelial, intratumoral or peritumoral stroma. To improve our understanding of the complex role of immune cells in GC, further studies combining both flow cytometry and IHC on a larger number of cases are necessary.
In conclusion, we have demonstrated that high levels of Tregs among tumor-infiltrating CD4+ T cells CD45+ leukocytes (C, F). www.impactjournals.com/oncotarget were favorable, but an increased proportion of MDSCs was an adverse independent prognostic factor in GC.

Figure 2: Disease-free and overall survival curves of gastric cancer patients according to the proportions of CD8+ T cells among CD3+ T cells (A, D), regulatory T cells (Tregs) among CD4+ T cells (B, E), and Myeloid derived suppressor cells (MDSCs) among
Our results may provide important insights for further classification of GC based on immunological feature and also for future immunotherapy in GC.

Patients
Tissue samples from 28 patients with GC undergoing curative resection of the primary tumor at the Department of Surgery at the Samsung Medical Center between June 2010 and September 2011 were included in this study (18 males, 10 females; ages ranging from 37 to 88 years). The research protocol was approved by the Institutional Review Board. Written informed consent was received from all patients. Tumor stage and grading were classified according to the 7th edition of the TNM Classification of the International Union Against Cancer [43]. Patients with other malignant diseases in their medical history and patients treated with neoadjuvant chemotherapy were excluded from this study. Six patients had postoperative concurrent chemo-radiation therapy and 13 patients performed postoperative chemotherapy. The median follow-up period was 29 months (range 10-42 months).
Tumor tissues were taken from the center of the cancer immediately after surgical removal and histologic confirmation. For preparation of single cell suspensions, tumor samples were mechanically dissected and washed with PBS. The cell suspension was passed through a sterile 40 µm nylon filter (BD Falcon, Heidelberg, Germany), and single cells were pelleted and suspended in PBS. Data from peripheral blood mononuclear cells from 8 healthy individuals were used as a control.

Multicolor flow cytometry
Multicolor immunofluorescence and multiparameter flow cytometric analyses were used to characterize the single-cell suspensions. To distinguish  The cells were incubated with the surface marker antibodies for 30 min at 4°C and washed twice with PBS. To conduct intracellular staining, 1 ml of fixation and permeabilization mixture (eBioscience) was added, and the cells were incubated on ice for an hour. After incubation, the cells were washed twice with PBS. After washing, the cells were mixed with the antibodies and incubated at room temperature for 15 minutes. After the final wash, the cells were resuspended in PBS buffer and analyzed with a FACS Canto II Flow Cytometer (BD Biosciences; FACS DIVA software ver. 6.1.3).

Statistics
Pearson's chi-square test, Fisher's exact test, and the Cochran-Armitage test were used as appropriate for statistical analysis. Disease-free survival (DFS) was defined as the time from surgery to the first relapse of cancer, occurrence of a second primary tumor, or death from any cause. Overall survival (OS) was measured from the date of surgery to the date of death. The OS and DFS were calculated using the Kaplan-Meier method, and survival curves were compared by the log-rank method.
The Cox proportional hazard model was used to evaluate the association between clinicopathologic factors and survival for multivariate analysis. In the multivariate analysis, variables showing p values of less than 0.3 in the univariate analysis were included as covariates. All statistical analyses were performed using SPSS software (SPSS Inc., Chicago, IL, USA) or R software (version 3.03), and p values less than 0.05 were considered to be statistically significant.