Fax: (011) 822-3410-0929
Clinical impact of tumor-infiltrating lymphocytes for survival in stage II colon cancer
Version of Record online: 3 NOV 2010
Copyright © 2010 American Cancer Society
Volume 116, Issue 22, pages 5188–5199, 15 November 2010
How to Cite
Lee, W.-S., Park, S., Lee, W. Y., Yun, S. H. and Chun, H.-K. (2010), Clinical impact of tumor-infiltrating lymphocytes for survival in stage II colon cancer. Cancer, 116: 5188–5199. doi: 10.1002/cncr.25293
- Issue online: 3 NOV 2010
- Version of Record online: 3 NOV 2010
- Manuscript Accepted: 9 FEB 2010
- Manuscript Revised: 24 JAN 2010
- Manuscript Received: 30 NOV 2009
- colon neoplasm;
- forkhead box P3;
- tumor-infiltrating lymphocyte;
- cluster of differentiation 45 isoform RO
The most reliable prognostic factor in colon cancer is the TNM classification. The objective of this study was to assess and compare the prognostic role of tumor-infiltrating lymphocytes (TILs) in stage II colon cancer.
Immunohistochemistry was used to assess the density of TILs that were positive for cluster of differentiation 3 (CD3) (T-cell coreceptor), CD45 isoform RO (CD45RO) (protein tyrosine phosphatase), nuclear transcription factor forkhead box P3 (FOXP3), and CD25 (a type I transmembrane protein) according to tumor site (intraepithelial and stromal) in samples from 87 patients who had stage II colon cancer. These variables were evaluated for their association with histopathologic features along with overall survival (OS) and disease-free survival (DFS).
Intraepithelial CD3-posititve (CD3+), CD45RO+, CD25+, and FOXP3+ TILs were associated significantly with better DFS (P = .049, P = .009, P = .013, and P = .001, respectively). The estimated 5-year OS rates for patients who had high-density CD45RO+ and FOXP3+ expression was 100% for both compared with 79.2% and 78.8% for patients who had low-density CD45RO+ and FOXP3+ expression (P = .017 and P = .040, respectively). A significant prognostic factor for both OS and DFS was high-density stromal CD45RO+ lymphocytic infiltration (OS: P = .031; relative risk [RR], 0.134; 95% confidence interval [CI], 0.015-1.164; DFS: P = .013; RR, 0.198; 95% CI, 0.055-0.710); whereas intraepithelial FOXP3+ expression was an independent prognostic factor for DFS (P = .032; RR, 0.108; 95% CI, 0.014-0.821).
FOXP3+ and CD45RO+ TILs demonstrated independent prognostic significance for survival in the current investigation. These results may help to improve the prognostication of early stage colon cancer. Cancer 2010. © 2010 American Cancer Society.
Colon cancer accounts for 10% to 15% of all cancers. It is the second leading cause of cancer deaths in Western countries1 and the fourth most common cancer in Korea.2 Currently, the most reliable prognostic factor in colon cancer is the TNM stage classification. Although the prognosis for individual patients depends on the extent of disease, the prognosis for patients with the same disease stage varies widely, especially in those with stage II and III tumors.3-6 Given a significantly reduced recurrence rate from adjuvant chemotherapy, postoperative chemotherapy has become the standard treatment for stage III colon cancer.6, 7 However, the role of adjuvant chemotherapy remains controversial for patients with stage II colon cancer and currently is not the standard of care for this group. The overall survival rate (OS) for patients with stage II colon cancer after surgery alone ranges from 70% to 80%,8 and the risk of recurrence ranges from 25% to 30%. Recently, the American Society of Clinical Oncology (ASCO) convened an expert panel to develop guidelines regarding adjuvant therapy for patients with stage II disease.9 The panel identified 37 randomized controlled trials and 11 meta-analyses of adjuvant therapy for colon cancer and selected 12 trials, each of which included a surgery-alone control arm and at least 1 arm that received a 5-fluouroracil (5-FU)-based chemotherapy regimen. On the basis of that analysis, the ASCO panel concluded that routine adjuvant chemotherapy is not recommended, because the absolute 5-year survival rate was not improved by >5% in those patients.9 Although adjuvant therapy is recommended for patients who have high-risk stage II disease, the survival benefit has not been established.10
Recently, it was reported that the biologic behavior of colon cancer is associated with a unique immune response by the colon cancer microenvironment and that patient outcomes are governed predominantly by the immune response at the primary tumor site.11 This suggests that a precise evaluation of local immune response may be useful for predicting prognosis12 and even may have prognostic value superior to and independent of the TNM classification.11 Naito et al13 were the first to demonstrate that infiltrating cytotoxic T cells that were positive for the transmembrane glycoprotein cluster of differentiation 8 (CD8) were a prognostic factor in colon cancer. T cells that were positive for the protein tyrosine phosphatase CD45 isoform RO (CD45RO) also reportedly had prognostic significance.14, 15 Oberg et al14 reported that a high density of CD45RO-positive T cells in lymph node metastases from colon cancer was associated with an improved prognosis, and Pages et al15 demonstrated that a high density of CD45RO-positive cells within a tumor was associated with decreased invasiveness, lower disease stage, and improved survival.
The most specific T-regulatory cell marker that has been identified to date is the nuclear transcription factor forkhead box P3 (FOXP3). This finding clearly provides evidence that the host immune response plays an important role in determining the outcome of patients with colon cancer.16-18 A high degree of tumor-infiltrating FOXP3-positive T cells has been associated with a poor outcome in various solid tumors, such as pancreatic cancer,19 ovarian cancer,20 and hepatocellular carcinoma.21
The objective of the current study was to assess and compare the prognostic role of tumor-infiltrating lymphocytes (TILs) in stage II colon cancers. We evaluated the type and density of TILs that were able to predict disease-free survival (DFS) and OS in patients who had long-term follow-up information.
MATERIALS AND METHODS
Patients and Specimens
Between January 1996 and January 2006, 4974 patients underwent curative resection for colon carcinoma at the Samsung Medical Center, Sungkyunkwan University. Of these patients, 1154 were diagnosed with pathologic stage II colon cancer according to the American Joint Committee on Cancer staging system.22 Ten archival specimens per year were selected randomly. The criteria for inclusion were the availability of formalin-fixed, paraffin-embedded tissue and a complete set of clinical and pathologic information, including age, sex, tumor size, lymphatic invasion, vascular invasion, and perineural invasion. Among the 100 randomly selected archival specimens, 87 archival specimens fulfilled the inclusion criteria and were available for the current analysis.
Surveillance for disease recurrence after surgery included physical examinations, serum carcinoembryonic antigen (CEA), chest radiography, and spiral abdominal computed tomography scans obtained every 6 months for the first 3 years and annually thereafter. Other examinations, such as colonoscopy and abdominal ultrasound, were selected and performed every 6 to 12 months, depending on the patient's status. The median follow-up duration was 125 months (range, 42.5-168.5 months).
Formalin-fixed, paraffin-embedded sections were dewaxed in xylene, rehydrated through a graded series of alcohol, and placed in an endogenous peroxide block for 15 minutes. Antigen retrieval was performed using a Micromed T/T Mega Histoprocessing Labstation (Milestone Inc., Atlanta, Ga) for 30 minutes in target retrieval solution (citrate buffer, pH 6.0; Dako, Glostrup, Denmark) and ethylene diamine tetracetic acid buffer, pH 8.0. Then, the sections were allowed to cool to room temperature and were incubated in 3% hydrogen peroxide for 20 minutes. Endogenous peroxidase activity was blocked by incubating the sections in 0.3% hydrogen peroxide in methanol for 15 minutes. Next, the slides were washed 3 times in wash buffer and incubated in 5% normal goat serum (Vector Laboratories Inc., Burlingame, Calif) for 60 minutes to block the nonspecific antibody binding sites. The rabbit monoclonal antibody used was antihuman CD3 (T-cell coreceptor; 1:50 dilution; Thermo Fisher Scientific, Waltham, Mass). The mouse monoclonal antibodies used were CD45RO (1:500 dilution; Thermo Fisher Scientific), rat antihuman FOXP3 monoclonal antibody clone PCH1011 (eBioscience, San Diego, Calif), and CD25 (a type I transmembrane protein; 1:10 dilution; Novocastra, Newcastle Upon Tyne, United Kingdom). Then, the sections were treated with biotinylated secondary antibody (Envision Detection Kit; DakoCytomation, Glostrup, Denmark) for 60 minutes, and washed 3 times in wash buffer. The diaminobenzidine substrate kit (Vector Laboratories) was used, and the sections were counterstained with hematoxylin (Dako).23
Quantitative Analysis of TILs
The numbers of TILs were counted using a computerized image analysis system (IMT i-Solution Inc., Vancouver, British Columbia, Canada) that included a DP70 Digital camera (Olympus, Tokyo, Japan) installed on an Olympus light microscope (Olympus BX51) and attached to a personal computer. Five independent microscopic fields (at ×400 magnification), representing the densest lymphocytic infiltrates, were selected for each patient sample to ensure representativeness and homogeneity. Each independent microscopic field contained an approximately 80% cancer ratio. The results were expressed as the mean number of cells (±standard error) in 1 computerized ×400 microscopic field (0.0935 mm2/field). The mean density was used for statistical analysis. The evaluation of TILs was performed by 1 pathologist who had no knowledge of the clinicopathologic data.
Comparisons between groups were performed using the Pearson chi-square test. For all immunohistochemical markers, the cutoff for defining a subgroup was the mean value. Survival curves were estimated using the Kaplan-Meier product-limit method, and the significance of differences between survival curves were determined using the log-rank test.
For multivariate analyses, the Cox regression model was used to identify independent prognostic factors for OS and recurrence. Differences with P values <.05 were considered statistically significant. OS was calculated from the date of colon resection and the date of either death or last follow-up, and DFS was measured from the date of colon resection to the date of either tumor recurrence or last follow-up.
TILs were detected within cancer epithelium (intratumoral) and in stroma (peritumoral). At low-power magnification, the distribution of immune cell infiltration was not relatively homogeneous within cancer epithelium or in stroma. Lymphocytes infiltrated colon cancer tissue in a diffuse manner or in lymphoid aggregates, and heavy lymphocytic infiltrates were observed in stromal areas. Five independent spots that were selected from each tumor had a good level of homogeneity for stained cell density. Representative images and statistics on immunohistochemical variables are provided in Table 1 and Figure 1. By using mean numbers, all specimens were classified into low-density and high-density groups for each marker (CD3-positive, CD45RO-positive, FOXP3-positive, and CD25-positive) according to tumor site (intraepithelial or stromal). We also grouped specimens using combined variables, that is, high-density or low-density CD3/CD45RO (CD3H/CD45ROH or CD3L/CD45ROL, respectively), CD25H/CD45ROH or CD25L/CD45ROL, and FOXP3H/CD3H/CD45ROH or FOXP3L/CD3L/CD45ROL (Table 2).
|Variable||No. of Patients|
Correlation of Immunohistochemical Variables With Clinicopathologic Features
In total, there were 87 patients with colon cancer. Their median age was 53 years (range, 23-76 years), and 43.6% were men. TIL density was not correlated with CEA levels, tumor size, or the presence of vascular invasion. FOXP3H was related significantly to less lymphatic invasion (P = .024). FOXP3H also had tendency toward less vascular invasion (P = .057) (Table 3).
|Variable||CD45RO+ TILs||CD3+ TILs||CD25+ TILs||FOXP3+ TILs|
|Tumor size, cm||.043||.175||.923||.367|
The OS and DFS rates were 89.1% and 82.3% at 5 years, respectively, for all patients. At univariate analysis, vascular invasion had prognostic significance for OS and DFS (P = .009 and P = .019, respectively) (Table 4). Intraepithelial CD3-positive, CD45RO-positive, and CD25-positive TILs were not associated significantly with OS. However, intraepithelial CD3-positive, CD45RO-positive, CD25-positive, and FOXP3-positive TILs were associated significantly with better DFS (P = .049, P = .009, P = .013, and P = .001, respectively) (Fig. 2) compared with patients who were positive for low-density TILs (log-rank test). Stromal CD45RO-positive, CD25-positive, and FOXP3-positive TILs significantly predicted OS compared with the presence of low-density TILs (P = .010, P = .017, and P = .036, respectively; log-rank test) (Fig. 3). The estimated 5-year OS rates for patients who had high-density CD45RO-positive and FOXP3-positive TILs was 100% compared with 79.2% and 78.8% for patients who had low-density CD45RO-positive and FOXP3-positive TILs (P = .017 and P = .040, respectively) (Table 5, Fig. 4). CD3-positive, CD45RO-positive, and FOXP3-positive lymphocytes infiltrating the stromal area significantly predicted DFS.
|Variable||No. of Patients (%)||P|
|All patients||87 (100)|
|Median age [range], y||53.0 [23-76]|
|Median [range], ng/mL||3.70 [0.5-205.0]|
|Tumor size, cm|
|Mean size [range], cm||6.58 [2.5-12.0]||.457||.062|
|Variable||Estimated 5-Year OS, %||P||Estimated 5-Year DFS, %||P|
|CD3+ high/lowa||100 vs 81.0||.039||100 vs 69.2||.003|
|CD45RO+ high/low||100 vs 76.1||.017||95.8 vs 65.3||.002|
|CD25+ high/low||100 vs 76.7||.037||100 vs 72.1||.014|
|FOXP3+ high/low||100 vs 78.8||.040||100 vs 69.6||.005|
|CD3/CD45RO||100 vs 68.1||.027||100 vs 51.9||.002|
|CD25/CD45ROb||100 vs 61.5||.058||100 vs 49.0||.021|
|FOXP3/CD3/CD45ROc||100 vs 64.3||.035||100 vs 45.7||.005|
The combined effects of low versus high numbers of stromal and intraepithelial CD45RO-positive, FOXP3-positive, and CD3-positive TILs also were evaluated. A significant advantage for OS was identified for the CD3H-positive, CD45ROH-positive, FOXP3H-positive, and CD25H-positive groups (high-density groups) and for high-density groups with combined TIL variables (Table 5, Fig. 4). The density of various TILs appeared to be useful for defining subgroups of patients with an unfavorable prognosis.
A forward conditional Cox regression model was used to delineate significant prognostic factors for survival. A prognostic factor for OS and DFS was high-density stromal CD45RO-positive lymphocytic infiltration (OS: P = .031; relative risk [RR], 0.134; 95% confidence interval [CI], 0.015-1.164; DFS: P = .013; RR, 0.198; 95% CI, 0.055-0.710) (Table 6). Intraepithelial FOXP3-positive lymphocytic infiltration was independent prognostic factor for DFS (P = .032; RR, 0.108; 95% CI, 0.014-0.821).
|Stromal TIL density(low vs high)|
|Intraepithelial TIL density (low vs high)||.465||0.443||0.500-0.392|
|Stromal TIL density (low vs high)|
|Intraepithelial TIL density (low vs high)|
A multivariate model was developed to test independent prognostic factors and the density of TILs (Table 7). In the first model (Model A), only pathologic prognostic variables were included, and vascular invasion was identified as the strongest prognostic factor. The second model included this feature (vascular invasion) with the density of each T-cell marker and revealed that FOXP3H-positive vascular invasion (P = .027; RR, 0.143; 95% CI, 0.065-0.847) and CD45ROH-positive vascular invasion (P = .014; RR, 0.235; 95% CI, 0.018-1.101) were the strongest prognostic factors.
At the time they undergo surgical resection, approximately 30% to 40% of patients with colorectal cancer are diagnosed with stage II disease.24 There is growing evidence that the prognosis for certain patients who have stage II colon cancer with unfavorable prognostic factors can be improved by adjuvant therapy.6, 25 Accordingly, there clearly is a need to delineate predictive factors to guide the identification of patients with stage II disease who are likely to experience a recurrence. The CD45RO-positive markers of cytotoxic immune response were associated previously with an improved prognosis in patients with colon cancer.12, 13, 26, 27 In agreement with these earlier reports, our current results confirmed the importance of CD45RO-positive cell density as a prognostic factor. The novelty of this study is that it demonstrated the prognostic importance of CD3-positive, CD45RO-positive, CD25-positive, and FOXP3-positive lymphocytes in a population of patients with stage II colon cancer.
Univariate survival analysis confirmed the poor prognostic factors for survival associated with conventional pathologic markers of adverse outcome (Table 4). FOXP3H was associated with less lymphatic invasion (P = .024) and had a tendency toward less vascular invasion (P = .057) (Table 3). Vascular invasion is a relatively putative pathologic marker of colon cancer invasiveness. This result may sustain the hypothesis that T-regulatory cells may modify colon cancer cells in ways that potentiate their invasiveness.
The better prognosis associated with high densities of stromal and intraepithelial FOXP3-positive TILs, stromal CD45RO-positive TILs, and stromal CD25-positive TILs is in agreement with earlier studies.14, 28 OS for patients who had high-density TILs was longer than for patients who had low-density TILs (CD3-positive, P = .039; CD45RO-positive, P = .017; CD25-positive, P = .037; FOXP3-positive, P = .040) (Table 5). This suggest that the diversity of clinical outcomes among patients with stage II colon cancer may be caused by differences in TIL density, underlining the importance of TILs as predictors of clinical outcome. It is believed that T-cell mediated adaptive immunity plays a major role in antitumor immunity and preventing tumor recurrence.11 Correale et al29 reported results from a phase 2 trial indicating that progressive increases in lymphocyte and eosinophil counts, amplification in central memory, a marked depletion of immunosuppressive regulatory T cells, and activation of colon cancer-specific cytotoxic T cells are tightly correlated with antitumor activity. Thus, enhancing antitumor immunity by augmenting the infiltration of stromal and intraepithelial TILs locally into tumors may be a reasonable treatment strategy for colon cancer and for suppressing metastasis. The precise mechanisms by which regulatory T cells suppress immune cell functions remain unclear. Currently, the better outcomes observed among patients who had colon cancer with high-density FOXP3-positive TILs cannot be explained. Zuo et al reported recently that the expression of FOXP3-positive TILs in breast cancer is an X-linked suppressor gene and a novel transcriptional repressor of the breast cancer oncogene SKP2 (S-phase kinase-associated protein 2) that suppresses breast tumor cell growth.30, 31 The expression of FOXP3 in tumor cells raises the possibility not only that TILs (such as regulatory T cells) influence antitumor immunity32, 33 but also that the tumor cells themselves modulate T-cell function through FOXP3, indicating its value as a potential prognostic factor. The apparent discrepancy between these results remains to be investigated further with emphasis on functional studies of their role in the antitumor microenvironment.
The survival benefit from adjuvant therapy for patients with stage II colon cancer is controversial. On the basis of a randomized trial, the 5-year OS was 80% for the control group and 82% for the adjuvant therapy group of patients with stage II colon cancer. Moreover, the data on DFS had similar trends, suggesting a possible but small benefit.34 The decision to offer adjuvant chemotherapy to patients with stage II colon cancer after they undergo curative surgical resection is difficult and involves weighing the risks of toxicity and complications with the potential improvement in curability. Hence, the discovery and validation of such novel prognostic factor for OS and DFS are of utmost importance for the management of stage II colon cancers. Multivariate models (Table 7) suggest that CD45RO-positive and FOXP3-positive TILs had stronger independent prognostic significance than conventional predictive factors, such as vascular invasion and perineural invasion. Multivariate analysis with dichotomized high-density groups (ie, stromal high-density plus intraepithelial high-density vs low-density groups for each marker and combination of markers) was not possible because the number of events was too low for statistical analysis, and the coefficients did not converge. Although the results need to be validated in additional large-scale trials, the current study suggests that T-cell markers may facilitate the identification of high-risk patients with early stage colon cancer who may benefit most from adjuvant chemotherapy.
Several study groups have conducted similar studies using tissue microarrays.35, 36 In those studies, 2 representative areas (0.6 mm to 2 mm) of tumor (the center and the invasive margin or tumor border) were selected to investigate the presence of TILs. Although tissue microarrays allow the simultaneous assessment of large cohorts, the relatively small area investigated may represent only a small cell population of the total tumor volume.37 To address this limitation, in the current study, we independently investigated 5 microscopic fields that had the densest lymphocytic infiltrates to ensure representativeness and homogeneity. However, cautious interpretations are needed for TILs, because the presence of specific T-cell subsets also may be influenced by other components in the tumor microenviroment.35 Given the finding that 5-FU–based adjuvant chemotherapy is associated with morbidity and decreased quality of life in a significant proportion of patients, intraepithelial and stromal TILs may facilitate selecting patient subgroups that may not benefit from adjuvant chemotherapy. A prospective analysis of intraepithelial and stromal expression profiles of groups of immune-related genes using quantitative real-time reverse transcriptase polymerase chain reaction is warranted to further investigate local immune response.
In conclusion, for the first time, the current results have elucidated the prognostic implications of the density of various types of TILs in a uniform population with stage II colon cancer. We conclude that FOXP3-positive and CD45RO-positive TILs have independent prognostic significance. These results may provide a novel independent predictor for prognosis and may help to predict clinical outcomes and refine the selection of patient subgroups that have an unfavorable prognosis. Furthermore, the stimulation of T-cell populations may be an effective strategy in patients with colon cancer to reduce recurrence and prolong survival.
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.
- 2Cancer incidence in Korea. Cancer Res Treat. 2002; 34: 405-408., , , et al.
- 22Greene FL, Page DL, Fleming ID, et al, eds. AJCC Cancer Staging Manual. 6th ed. New York: Springer-Verlag; 2002.
- 29Immunity feedback and clinical outcome in colon cancer patients undergoing chemoimmunotherapy with gemcitabine + FOLFOX followed by subcutaneous granulocyte macrophage colony-stimulating factor and aldesleukin (GOLFIG-1 Trial). Clin Cancer Res. 2008; 14: 4192-4199., , , et al.
- 34[No authors listed] Efficacy of adjuvant fluorouracil and folinic acid in B2 colon cancer. International Multicentre Pooled Analysis of B2 Colon Cancer Trials (IMPACT B2) Investigators. J Clin Oncol. 1999; 17: 1356-1363.