Journal of Cancer Research and Clinical Oncology

, Volume 139, Issue 11, pp 1845–1852

The appearance of Tregs in cancer nest is a promising independent risk factor in colon cancer

Authors

  • Wei Xu
    • Department of SurgeryThe Affiliated Hospital of Xuzhou Medical College
  • Hao Liu
    • Department of SurgeryThe Affiliated Hospital of Xuzhou Medical College
  • Jun Song
    • Department of SurgeryThe Affiliated Hospital of Xuzhou Medical College
  • Hai-Xiao Fu
    • Department of SurgeryThe Affiliated Hospital of Xuzhou Medical College
  • Lei Qiu
    • Department of SurgeryThe Affiliated Hospital of Xuzhou Medical College
  • Bao-Fu Zhang
    • Jiangsu Key Laboratory of Biological Cancer TherapyXuzhou Medical College
  • Hui-Zhong Li
    • Jiangsu Key Laboratory of Biological Cancer TherapyXuzhou Medical College
    • Jiangsu Key Laboratory of Biological Cancer TherapyXuzhou Medical College
    • Jiangsu Key Laboratory of Biological Cancer TherapyXuzhou Medical College
Original Paper

DOI: 10.1007/s00432-013-1500-7

Cite this article as:
Xu, W., Liu, H., Song, J. et al. J Cancer Res Clin Oncol (2013) 139: 1845. doi:10.1007/s00432-013-1500-7

Abstract

Purpose

To investigate the prognostic value of tumor-infiltrating regulatory T cells (Tregs) in the distribution of cancer nest, cancer stroma and normal mucosa and FOXP3-positive cancer cells in colon cancer patients after resection.

Methods

Paraffin blocks of operation resection of primary adenocarcinoma of colon were obtained from ninety patients. The distribution of tumor-infiltrating Tregs was detected by tissue microarray and immunohistochemistry staining technique to evaluate the prognostic effects by Kaplan–Meier and Cox regression analysis using median values as cutoff.

Results

The intratumoral Tregs counts were significantly higher than that in corresponding normal mucosa tissues (P < 0.001); the Tregs counts in cancer nest were significantly lower than that in corresponding cancer stroma tissues (P < 0.001); the increased intratumoral Tregs counts were associated with favorable prognosis (P < 0.05); the presence of Tregs in cancer nest was associated with unfavorable prognosis and was an independent prognostic factor for overall survival (P < 0.05). The appearance of FOXP3-positive cancer cells was associated with worse prognosis (P < 0.05). In addition, the frequency of the presence of FOXP3-positive cancer cells was higher in patients with lymphatic invasion (P < 0.001) and lower in patients with early TNM stage (P < 0.01).

Conclusions

The higher tumor-infiltrating Tregs counts are closely associated with the improved prognostic effects of colon carcinoma. Tregs play different roles in cancer nest and cancer stroma. And the appearance of Tregs in cancer nest is a promising independent risk factor for overall survival in colon carcinoma. FOXP3-positive cancer cells may also be a risk factor for overall survival in colon carcinoma.

Keywords

Colon carcinomaTregsFOXP3PrognosticTissue microarray

Introduction

Colorectal cancer is one of the most common malignant tumors in the world. Despite improved progress that has been made in basic and clinical research, the overall survival of colorectal cancer did not improve significantly (Mitry et al. 2008). Previous studies, which focused their key point on the tumors characteristics, tried to change gene and phenotype of cancer cells to inhibit the progress of the tumors. Recently, tumor immune microenvironment was reported to play an important role in the development of the tumor (Galon et al. 2007). In particular the regulatory T cells (Tregs), these cells were first reported by (Sakaguchi et al. 1995) in both mice and humans as a key contributor to the maintenance of immunological tolerance and the prevention of autoimmunity. In tumor tissues, Tregs could migrate to tumor locality and inhibit anti-tumor immune response by influencing the effect of CD4+ and CD8+ T cells, NK cells and myeloid cells, through different mechanisms (Sakaguchi et al. 2010). And a series of studies have revealed that the high tumor-infiltrating Tregs counts often associate with a poor prognosis in breast (Bates et al. 2006), ovarian (Curiel et al. 2004), lung (Petersen et al. 2006), hepatocellular (Gao et al. 2007), renal cell (Li et al. 2009), pancreatic (Hiraoka et al. 2006), gastric (Perrone et al. 2008) and cervical (Jordanova et al. 2008) cancers. In these studies, FOXP3, which belongs to the family of forkhead box transcription factors, has been a specific surface marker of Tregs (Hori et al. 2003).

Surprisingly, seven clinical studies have showed that (Salama et al. 2009; Yoon et al. 2012), in patients bearing colorectal carcinoma, a high tumor-infiltrating Tregs counts are associated with an improved prognosis. And they tried to explain this paradox by a hypothesis that Tregs can indirectly inhibit cancer growth by restraining T cell-mediated inflammatory anti-microbial response that can promote tumor growth in colorectal cancer, although Tregs are considered as a major cause of protecting cancer cells from anti-tumor immunity because of the ability to block the T cell-mediated anti-tumor immune response in colorectal cancer (Clarke et al. 2006).

Recent studies found that FOXP3 was not only expressed in Tregs, but also in several types of malignant tumor cells, including pancreatic (Hinz et al. 2007), breast (Zuo et al. 2007), lung (Tao et al. 2012) and prostate (Wang et al. 2009) cancers. And FOXP3-positive cancer cells may have ability to escape immune surveillance in pancreatic cancer, which likes the function of Tregs (Hinz et al. 2007). But in breast cancer, this type of cells is reported as favorable prognostic factor (Ladoire et al. 2011a, b). In colorectal cancer, high FOXP3 expression of the cancer cells was associated with poor prognosis compared to patients with low FOXP3 expression (Kim et al. 2013).

However, how to evaluate the ability of inhibiting anti-tumor immunological response that Tregs have in the microenvironment of colorectal cancer is less clear. In this study, the prognostic value of Tregs in the distribution of cancer nest, cancer stroma and normal mucosa and FOXP3-positive cancer cells was investigated. We found that Tregs play different roles in cancer nest and cancer stroma, that is the first report by now, and the appearance of Tregs in cancer nest is a promising independent risk factor in colon carcinoma.

Materials and methods

Patients and samples

The tissue microarray that includes 90 patients of primary colon adenocarcinoma was obtained from Shanghai Outdo Biotech (Shanghai, China). And each patient enrolled has two cores, one is the cancer tissue, the other is the normal mucosa tissue (>5 cm from the tumor margin). All of the patients, who underwent operative resection for colon cancer between July 2006 and May 2007, did not accept any neoadjuvant therapy or postoperative radiotherapy and chemotherapy. This study contains 9 cases of TNM stage I, 50 cases of TNM stage II, 29 cases of TNM stage III and 2 cases of TNM stage IV. Histologic type was determined according to the WHO classification, and the tumor staging was determined according to the TNM classification system of American Joint Committee on Cancer and International Union against Cancer (the seventh edition). Follow-up was completed in August 2012. The median follow-up was 65 months (63–73 months). This study was approved by the research ethics committee and agreed by each patient with written consent.

Immunohistochemistry of TMA

TMA slides were deparaffinized and rehydrated using xylene and graded alcohol washes. After heat-mediated antigen retrieval via the microwave method and a series of blocking with 3 % H2O2 and normal goat serum, the tissue microarray was incubated with primary mouse monoclonal to FOXP3 with 1:100 dilution (Abcam, Cambridge, MA) overnight at 4 °C. And then, biotinylated secondary antibody followed by horseradish peroxidase (HRP)-conjugated streptavidin and DAB-chromogen was applied according to the manufacturer’s instructions and counterstained with hematoxylin. Negative control was made using PBS to take the place of the primary antibodies in another serial section. Two independent investigators reviewed the tissue microarray independently for a random selection of 5 high-power fields (HPF, 400×) each point. The nuclear staining was considered as the Tregs in cancer nest, cancer stroma and normal mucosa, and the cytoplasmic staining which was more than 20 % of examined fields in cancer nest was considered as the FOXP3-positive cancer cell (Tao et al. 2012). The average count of Tregs of 5 HPF was recorded for further analysis.

Statistical analysis

Statistic analysis was performed with SPSS 16.0 software (SPSS, Chicago, IL). Means and standard deviations (\( \bar{x} \pm S \)) were used for the descriptive statistics of enumeration data. For the comparison of two groups of enumeration data, two independent samples t test, two paired samples t test, χ2 test or Fisher’s exact test were adopted appropriately. Kaplan–Meier method and log-rank test were used for survival analysis. Univariate and multivariate analyses were based on the Cox regression. Two-tailed P < 0.05 was considered to be statistically significant.

Results

Analysis on the distribution of Tregs and the expression status of FOXP3

The counts of the Tregs which were detected by positive nuclear staining of FOXP3 (Fig. 1a, b) ranged from 0 to 78.0 (mean 16.1, median 11.0) per HPF in tumor tissue, from 0 to 18.0 (mean 2.7, median 1.0) per HPF in cancer nest. Of the 90 patients analyzed, low expression levels and high expression levels of intratumoral Tregs were 76.7 % (69/90) and 23.3 % (21/90). Cancer cells expression FOXP3-negative and expression FOXP3-positive were 53.3 % (48/90) and 46.7 % (42/90). The intratumoral Tregs counts were significantly higher than that in corresponding normal mucosa tissues (P < 0.001); the Tregs counts in cancer nest were significantly lower than that in corresponding cancer stroma tissues (P < 0.001) (Table 1). These suggested that Tregs may play a key role in the development of colon cancer. In addition, FOXP3-positive cancer cells, which were detected by cytoplasmic staining in the cancer nest (Fig. 1c, d), were found in 15 of 90 patients.
https://static-content.springer.com/image/art%3A10.1007%2Fs00432-013-1500-7/MediaObjects/432_2013_1500_Fig1_HTML.jpg
Fig. 1

Representative images of FOXP3 immunohistochemical staining in human colon cancer. a Tregs in cancer tissue were detected with positive nuclear staining of FOXP3 (redandblack arrow), and tumor-infiltrating Tregs were detected, respectively, in cancer nest (red arrow) and in cancer stroma (black arrow); b Tregs in normal mucosa tissue were also detected with positive nuclear staining of FOXP3 (red arrow); c FOXP3 positive staining in the cytoplasm of colon cancer cells; d FOXP3 negative staining in the cytoplasm of colon cancer cells. Magnification ×400

Table 1

The distribution of Tregs of 90 colon cancer patients in intratumoral, normal mucosa, cancer stroma and cancer nest

Location

N

Minimum

Maximum

Mean

Median

Treg count (per HPF)

t

P

Intratumoral (total)

90

0

78.0

16.1

11.0

16.08 ± 16.70

3.95

<0.001

Normal mucosa

0

112.0

7.0

3.0

6.97 ± 14.67

Cancer stroma

90

0

78.0

13.4

7.0

13.39 ± 16.17

6.019

<0.001

Cancer nest

0

18.0

2.7

1.0

2.67 ± 4.54

Association of the distribution of Tregs and the expression status of FOXP3 with clinicopathologic features

According to the cutoff value of the count of Tregs in cancer nest, cancer stroma and normal mucosa, 90 patients were divided into two groups separately. The count of Tregs in cancer tissue did not correlate with age, gender, tumor size or the number of tumors. Patients with higher pathological differentiation, no lymphatic invasion and earlier tumor-node-metastasis (TNM) stage showed higher tumor-infiltrating Tregs (Table 2). In addition, the presence of FOXP3-positive cancer cells did not correlate with age, gender, tumor size, the number of tumor, pathological differentiation or the count of tumor-infiltrating Tregs. The frequency of the presence of FOXP3-positive cancer cells was higher in patients with lymphatic invasion or lower in patients with early TNM stage (Table 3).
Table 2

Association of Tregs count with clinicopathologic features of 90 colon cancer patients

Group

N

The count of Tregs

t

P

Age (years)

 <70

43

15.44 ± 18.16

−0.344

0.732

 ≥70

47

16.66 ± 15.41

Gender

 Male

47

16.34 ± 18.62

0.155

0.877

 Female

43

15.79 ± 14.52

Pathological grading

 I + I − II + III

56

18.98 ± 19.40

2.525

0.013

 II − III + III

34

11.29 ± 9.32

Tumor size (cm3)

 <35

45

16.56 ± 17.45

0.27

0.788

 ≥35

45

15.60 ± 16.09

The number of tumor

 1

82

15.78 ± 16.76

−0.539

0.591

 2

8

19.13 ± 16.82

Lymphatic invasion

 No

53

19.38 ± 20.02

2.606

0.011

 Yes

37

11.35 ± 8.44

TNM stage

 I + II

58

18.98 ± 19.38

2.801

0.006

 III + IV

32

10.81 ± 8.05

Table 3

Association of FOXP3 expression in cancer cells with clinicopathologic features of 90 colon cancer patients

Group

Expression of FOXP3 in cancer cells

χ2

P

FOXP3 (+)

FOXP3 (−)

Age (years)

 <70

6 (14.0 %)

37 (86.0 %)

0.436

0.509

 ≥70

9 (19.1 %)

38 (80.9 %)

Gender

 Male

5 (10.6 %)

42 (89.4 %)

2.574

0.109

 Female

10 (23.3 %)

33 (76.7 %)

Pathological grading

 I + I − II + II

6 (10.7 %)

50 (89.3 %)

3.782

0.052

 II − III + III

9 (26.5 %)

25 (73.5 %)

Tumor size (cm3)

 <35

8 (17.8 %)

37 (82.2 %)

0.08

0.777

 ≥35

7 (15.6 %)

38 (84.4 %)

The number of tumor

 1

13 (15.9 %)

69 (84.1 %)

0.439

0.508

 2

2 (25.0 %)

6 (75.0 %)

Lymphatic invasion

 Yes

12 (32.4 %)

25 (67.6 %)

11.244

<0.001

 No

3 (5.7 %)

50 (94.3 %)

TNM stage

 I + II

5 (8.6 %)

53 (91.4 %)

7.603

0.006

 III + IV

10 (31.2 %)

22 (68.8 %)

Tumor-infiltrating Tregs

 High

2 (9.5 %)

19 (90.5 %)

1.006

0.316

 Low

13 (18.8 %)

56 (81.2 %)

Tregs play different roles in cancer nest and cancer stroma by Kaplan–Meier survival analyses of 90 colon cancer patients

A total of 90 patients had completed follow-up and clinical information. To evaluate the prognosis, Kaplan–Meier survival analyses were used. The results revealed that the increased intratumoral Tregs counts were associated with favorable prognosis (P < 0.05) (Fig. 2a); the presence of Tregs in cancer nest was associated with unfavorable prognosis (P < 0.05) (Fig. 2b); the increased Tregs counts in cancer stroma were seemingly associated with favorable prognosis but no statistical significance (P = 0.103) (Fig. 2c); the reduced Tregs counts in normal mucosa were seemingly associated with favorable prognosis but also no statistical significance (P = 0.206) (Fig. 2d); the expression of FOXP3 in cancer cells was associated with a poor prognosis (P < 0.05) (Fig. 2e).
https://static-content.springer.com/image/art%3A10.1007%2Fs00432-013-1500-7/MediaObjects/432_2013_1500_Fig2_HTML.gif
Fig. 2

Kaplan–Meier survival analyses of 90 colon cancer patients. a High intratumoral Tregs counts correlate with an improved overall survival (P = 0.020, log-rank test). b The presence of Tregs in cancer nest was associated with unfavorable prognosis (P = 0.011, log-rank test). c High Tregs counts in cancer stroma were seemingly associated with favorable prognosis but no statistical significance (P = 0.103, log-rank test). d Low Tregs counts in normal mucosa were seemingly associated with favorable prognosis but also no statistical significance (P = 0.206, log-rank test). e The expression of FOXP3 in cancer cells was associated with a poor prognosis (P = 0.035, log-rank test). Cum cumulative

The presence of Tregs in cancer nest is an independent risk factor for overall survival

We also used univariate Cox proportional hazards regression model to estimate the crude hazard ratios (HRs) of Tregs or each clinicopathological variable on patient survival. The log-rank test and univariate Cox regression analyses revealed that pathological grading, tumor size, the number of tumor, lymphatic invasion, TNM staging, tumor-infiltrating Tregs in cancer tissue, the presence of Tregs in cancer nest and the expression of FOXP3 in cancer cells were significantly associated with overall survival in all 90 patients (Table 4).
Table 4

Univariate Cox proportional regression analysis on 5-year overall survival of 90 colon cancer patients

Group

Overall survival

Hazard ratio

95 % CI

P

Age (<70 years vs. ≥70 years)

1.594

0.855–2.972

0.143

Gender (male vs. female)

0.917

0.499–1.684

0.78

Pathological grading (I + I − II + II vs. II − III + III)

1.936

1.056–3.551

0.033

Tumor size (<35 cm3 vs. ≥35 cm3)

2.145

1.148–4.009

0.017

The number of tumor (1 vs. 2)

2.854

1.261–6.459

0.012

Lymphatic invasion (no vs. yes)

3.849

2.034–7.282

<0.001

TNM stage (I + II vs. III + IV)

3.049

1.651–5.630

<0.001

Tumor-infiltrating Tregs (low vs. high)

0.348

0.137–0.886

0.027

Tregs in cancer nest (yes vs. no)

2.193

1.174–4.096

0.014

Tregs in cancer stroma (low vs. high)

0.603

0.326–1.117

0.108

Tregs in normal mucosa (low vs. high)

1.533

0.784–2.995

0.212

Expression of FOXP3 in cancer cells (no vs. yes)

2.102

1.033–4.280

0.04

To further validate the prognostic value of Tregs and FOXP3 expression in colon cancer cells, multivariate analysis was performed, where pathological grading, tumor size, the number of tumor, lymphatic invasion, TNM stage, tumor-infiltrating Tregs, Tregs in cancer nest and expression of FOXP3 in cancer cells were included. And the results showed that pathological grading (HR 2.238; 95 % confidential interval (CI) 1.075–4.659; P = 0.031), tumor size (HR 3.146; 95 % CI 1.528–6.479; P = 0.002), the number of tumor (HR 2.914; 95 % CI 1.192–7.123; P = 0.019) and the presence of Tregs in cancer nest (HR 2.179; 95 % CI 1.091–4.349; P = 0.027) were significant factors (Table 5). The Cox regression model indicated that the presence of Tregs in cancer nest is an independent risk factor for overall survival.
Table 5

Multivariate Cox regression analysis on 5-year overall survival of 90 colon cancer patients

Group

Overall survival

Hazard ratio

95 % CI

P

Pathological grading (I + I − II + II vs. II − III + III)

2.238

1.075–4.659

0.031

Tumor size (<35 cm3 vs. ≥35 cm3)

3.146

1.528–6.479

0.002

The number of tumor (1 vs. 2)

2.914

1.192–7.123

0.019

Lymphatic invasion (no vs. yes)

2.600

0.816–8.288

0.106

TNM stage (I + II vs. III + IV)

1.284

0.380–4.340

0.687

Tumor-infiltrating Tregs (low vs. high)

0.581

0.205–1.641

0.305

Tregs in cancer nest (yes vs. no)

2.179

1.091–4.349

0.027

Expression of FOXP3 in cancer cells (no vs. yes)

1.284

0.562–2.935

0.553

Discussion

Tregs have the ability to inhibit the immune response against tumors mediated by T cells in colorectal cancer (Clarke et al. 2006), which can protect the growth behavior of tumor. This finding seems to be consistent with other outcome studies in various human carcinomas (Curiel et al. 2004; Gao et al. 2007; Hiraoka et al. 2006; Jordanova et al. 2008; Li et al. 2009; Perrone et al. 2008; Petersen et al. 2006), in which the high accumulation of Tregs often results in a poor prognosis. But conclusions contrary to the previous reports in other cancers were established by recent studies in which high count of tumor-infiltration Tregs was associated with improved survival in colorectal cancer (Frey et al. 2010; Pages et al. 2005; Salama et al. 2009). And 216 cases of colon carcinomas from participants in 5-FU-based adjuvant trials were analyzed to find that among CD8+low tumors, FOXP3+high cases had significantly improved overall survival compared to FOXP3+low cases, but Tregs counts in cancer epithelial did not significantly influence overall survival (Yoon et al. 2012).

In our study, through the precise location of Tregs in the microenvironment of colon cancers which were not treated with neoadjuvant radiochemotherapy or postoperative radiochemotherapy, we found that the higher tumor-infiltrating Tregs counts are closely associated with the improved prognostic effects. The reason of this discrepancy may be attributed to the existence of microbiological flora in colonic lumen which can infiltrate the tumor through the increased permeability of the physical structure of the colonic mucosal barrier (Ladoire et al. 2011a, b). The tumor-infiltrating microbe can cause the inflammatory anti-microbial response, which involves Th17 cells, mediated by T cells. This can stimulate angiogenesis through VEGF production and induce inflammatory reaction triggered by cytokines like IL-17, IL-1, IL-6 and TNF-α, with subsequent proangiogenic (Numasaki et al. 2003) and tumor-enhancing effects (Bromberg and Wang 2009), thus contributing to the progress of tumor. This anti-microbial response can be attenuated by Tregs (Ladoire et al. 2011a, b), which may give Tregs the anti-tumorigenic effects indirectly in colon cancer and explain the relationship, unlike other carcinomas, between the higher tumor-infiltrating Tregs counts and improved prognosis.

But more precise location of Tregs in the microenvironment of colon cancer, whereas the cancer nest and cancer stroma were investigated separately, showed that the appearance of Tregs in cancer nest was correlated with a poor outcome based on both univariate and multivariate analyses. We surmised that Tregs in cancer nest may inhibit anti-tumor immunity and help the cancer cells to escape from immune surveillance maybe through the direct contact with the cancer cells. We firstly report that, unlike other carcinomas, Tregs in colon cancer, whereas the tumor-infiltrating Tregs have specificity not only for tumor-specific antigens but also for the commensal microflora, play different roles in cancer nest and cancer stroma, mainly for inhibition of inflammatory anti-microbial response in cancer stroma and mainly for suppression of anti-tumor response in cancer nest.

In addition, we also found that FOXP3 could also be expressed in colon cancer cells detected by cytoplasmic staining. Correlated with clinicopathological parameters and overall survival of patients, the frequency of the presence of FOXP3-positive cancer cells was higher in patients with lymphatic invasion or lower in patients with early TNM stage. The expression of FOXP3 in cancer cells was associated with a poor prognosis, which is similar to previous studies (Kim et al. 2013).

In conclusion, our results have shown that the prognosis could be assessed on the basis of tumor-infiltrating Tregs and tumor cells FOXP3 expression status that may be predictors for overall survival in colon carcinoma. What is more, the appearance of Tregs in cancer nest is a promising independent risk factor for overall survival in colon carcinoma. Our findings may be beneficial to design a new immunotherapy targeting Tregs, especially in different locations, and FOXP3-positive cancer cells in colon cancer.

Conflict of interest

We declare that we have no conflict of interest.

Copyright information

© Springer-Verlag Berlin Heidelberg 2013