Increased Intratumor Vα24-Positive Natural Killer T Cells: A Prognostic Factor for Primary Colorectal Carcinomas

Natural killer T (NKT) cells have been recently recognized as a novel subset of lymphoid cells (1). NKT cells, which share the characteristics of both T and natural killer cells, constitute a unique class of the T lymphocyte lineage. They have an extremely restricted T-cell receptor (TCR) repertoire consisting of an invariant Vα24-Jα18 chain (formerly Vα24-JαQ) paired with a Vβ11 chain in human peripheral blood (2, 3). These cells recognize glycolipid antigens, such as α-galactosylceramide presented by MHC class I–like molecules CD1d on antigen-presenting cells (4–8). It has been assumed that NKT cells play a role in immunoregulation. When activated, NKT cells immediately produce a large amount of proinflammatory T helper 1 cytokines (e.g., IFN-γ and tumor necrosis factor-α) and antiinflammatory T helper 2 cytokines [e.g., interleukin (IL)-4, IL-10, and IL-13; refs. 4, 7, 9–12]. NKT cells have been implicated in the control of infection (13–15), inflammatory bowel diseases (16, 17), and abortion (18), as well as in transplantation tolerance (19, 20), the suppression of autoimmune diseases (21–24), and the regulation of allergic disorders (25, 26). Moreover, NKT cells have been postulated to play protective as well as tolerogenic roles in tumor immunity (12, 27–34).

NKT cells are mainly localized in the thymus, bone marrow, liver, and spleen, but they are rare in the lymph nodes (4, 34). In the gastrointestinal tract, Bannai et al. (35) found a considerable number of NKT cells intraepithelially in the colon of mice. More recently, O'Keeffe et al. (36) reported that CD161⁺ T cells are the major natural killer receptor–positive cell population in the intestine and that the natural killer receptor–positive T cells are not much more frequent in colorectal carcinomas than in adjacent tumor-free tissue. However, few reports have been available on human NKT cells in normal and neoplastic colon.

In the present study, to evaluate the prognostic potential of intratumor NKT-cell infiltration in colorectal carcinomas, we investigated intratumor TCR-Vα24-positive cells in 103 primary colorectal carcinoma samples.

Study population. The clinical records of patients who had been diagnosed with primary colorectal carcinoma and had undergone resection with curative intent at the Department of Surgery and Surgical Basic Science, Kyoto University, between 1996 and 1997, were carefully examined. After excluding those who had died within 30 days of the surgery and those who had died from causes unrelated to colorectal carcinoma, a total of 103 patients (65 males and 38 females) underwent full survival analysis over a median follow-up time of 1,914 days. The mean age of the patients at diagnosis was 64.9 ± 12.2 years. In the tumor-node-metastasis (TNM) classification system (37), 37% were pTNM stage I, 21% were stage II, 29% were stage III, and 13% were stage IV. None had received preoperative chemotherapy or radiotherapy. Patients with stage III or IV colorectal carcinomas were given postoperative chemotherapy consisting of 5-fluorouracil. Written informed consent was obtained from all the patients.

Immunohistochemistry. Immunohistochemistry was carried out using the catalyzed signal amplification system (DAKO, Carpinteria, CA) according to the manufacturer's instructions with minor modifications. In brief, 4-μm sections of formalin-fixed and paraffin-embedded tissues were deparaffinized and rehydrated, and an antigen retrieval procedure was done using Target-Retrieval Solution (DAKO). Sections were incubated overnight at 4°C with a 1:100 diluted primary antibody. For primary reagents, we used monoclonal antibodies C15 (TCR-Vα24, Immunotech SA, Marseilles, France), C21 (TCR-Vβ11, Immunotech), MC-1 (CD56, DAKO), NK1 (CD57, DAKO), DX12 (CD161, BD Biosciences, San Diego, CA), CH11 (CD69, Lab Vision, Fremont, CA), MD-1 (IFN-γ, Biosource, Camarillo, CA), anti-granzyme B (WAKO, Osaka, Japan), and the polyclonal antibody to FasL (Histofine, Nichirei, Tokyo, Japan). The endogenous peroxide was quenched by incubation with 3.0% hydrogen peroxide/methanol after the sections had been incubated with the second antibody. Horseradish peroxidase–conjugated extravidin (Sigma, St. Louis, MO) was used instead of streptavidin. Nuclei were counterstained with Mayer's hematoxylin (WAKO). For immunofluorescence staining, phycoerythrin-conjugated streptavidin and fluorescein-streptavidin (Vector Laboratories, Burlingame, CA) were used in place of streptavidin. Twenty-four sections of adjacent normal colorectal mucosa were also examined. The negative control consisted of sections that were treated with the same protocol but without the primary antibody. Specimens were examined under Olympus BX50 and Olympus AX80TR microscopes (Olympus, Tokyo, Japan).

Histologic evaluation of tumor and natural killer T cells. Histologic grading of the colorectal carcinomas was done according to the criteria of the WHO (38). Lymphatic invasion and venous invasion were evaluated following the rules proposed by the Japanese Research Society for Cancer of the Colon and Rectum (39).

Human Vα24⁺ NKT cells express invariant T-cell receptor composed of Vα24-Jα18 (5, 6). It could be assumed that the majority of Vα24-positive cells represent invariant Vα24⁺ NKT cells. Vα24⁺ NKT cells were counted in five randomly selected ×200 fields of the tumor mass and in adjacent non–tumor mucosa or submucosa by two investigators who were blinded to the clinical data. NKT cells infiltrating in the tumor were defined as Vα24⁺ cells found simultaneously with tumor tissues in the same field to avoid counting the incidental bystanders as much as possible. The degree of Vα24⁺ NKT-cell infiltration was scored as either low if less than seven Vα24⁺ NKT cells were found in five intratumor fields, or high if seven or more Vα24⁺ NKT cells were found. This number was selected to give the highest significance in the statistical analysis of prognosis.

Statistical analysis. Statistical analyses were done with StatView-J 5.0 software (SAS Institute, Cary, NC). The relationship between clinicopathologic factors and NKT-cell infiltration was evaluated by the χ² test. The unpaired Mann-Whitney U test was used to compare the number of NKT cells in normal mucosa and in colorectal tumors. The Kaplan-Meier method was used to obtain survival curves. The log-rank test was used to compare survival based on NKT-cell infiltration. Multivariate analysis was done by the Cox proportional hazard regression model. P < 0.05 was considered statistically significant.

Increased Vα24⁺ natural killer T cells in colorectal carcinomas. Combined immunostaining with fluorescence-labeled antibodies revealed that an increased number of NKT cells bearing a T-cell receptor repertoire of Vα24 and Vβ11 infiltrated colorectal carcinomas (Fig. 1A). Vα24⁺ NKT cells expressed combinations of Vβ11, CD56, CD57, or CD161 (Fig. 1B, C, D, and E). In normal colorectal tissues, a few Vα24⁺ NKT cells were found mainly in the interstitial region of the submucosa (data not shown) and sometimes in the colonic lymphoid follicle and/or intraepithelial region (data not shown). Vα24⁺ NKT cells were large lymphoid cells with multilobulated nuclei (Fig. 1F, inset).

The number of Vα24⁺ NKT cells in colorectal carcinomas (15.2 ± 16.3 cells/5 HPF; n = 103) was significantly higher than that in normal mucosa (2.6 ± 3.7, n = 24; P = 0.0003). Vα24⁺ NKT cells were found also in the tumor-infiltrating vessels (data not shown).

Activation of Vα24⁺ natural killer T cells in colorectal carcinomas. Vα24⁺ NKT cells infiltrating colorectal carcinomas seemed activated because they were positively stained for the early activation markers, CD69 and FasL (Fig. 2A and B). They also expressed active effector molecules IFN-γ and granzyme B (Fig. 2C and D). The fraction of CD69-positive NKT cells in colorectal carcinomas was 81.9%, whereas in the normal mucosa, it was 12.5% (P < 0.001). The fraction of FasL-positive cells was 76.6% in the former and 21.4% in the latter (P < 0.001).

Vα24⁺ natural killer T-cell infiltration and colorectal carcinoma clinicopathologic factors. Among colorectal carcinoma patients, the degree of Vα24⁺ NKT-cell infiltration varied from case to case. Figure 1F shows representative colorectal carcinomas with high NKT-cell infiltration. The colorectal carcinoma patients were subdivided into two groups, designated high (65 patients) and low (38 patients), according to the degree of NKT-cell infiltration. We compared the clinicopathologic variables in these two groups (Table 1). There was no significant correlation between NKT-cell infiltration and age, tumor location, histologic differentiation, lymphatic invasion, venous invasion, tumor depth, distant metastasis, or TNM stage (Table 1). However, the patients with high NKT-cell infiltration had a significantly lower incidence of regional lymph node metastasis (21 of 65) than those with low NKT-cell infiltration (20 of 38; P = 0.042). Furthermore, many more female than male patients showed high NKT-cell infiltration (P = 0.037; Table 1).

Vα24⁺ natural killer T-cell infiltration as a colorectal carcinoma prognostic factor. As seen in Fig. 3A, overall survival was found to be significantly higher in patients with high NKT-cell infiltration (n = 65) than in those with low NKT-cell infiltration (n = 38; P = 0.018). The univariate analysis revealed that lymph node metastasis (P = 0.0011), NKT-cell infiltration (P = 0.024), and distant metastasis (P < 0.0001) were primary factors influencing favorable overall survival. Multivariate analysis showed that NKT-cell infiltration (P = 0.033) and distant metastasis (P < 0.0001) were independent prognostic factors for survival (Table 2).

Subsequently, we compared the disease-free survival rate in the two groups after exclusion of the 13 patients with stage IV disease. As shown in Fig. 3B, the patients with high NKT-cell infiltration (n = 58) had a significantly longer disease-free survival rate than those with low NKT-cell infiltration (n = 32; P = 0.0006). The univariate analysis revealed that venous invasion (P = 0.036), lymph node metastasis (P = 0.0081), and NKT-cell infiltration (P = 0.0031) were significant prognostic factors. Multivariate analysis showed that only the grade of NKT-cell infiltration (P = 0.0064) was an independent prognostic factor for disease-free survival (Table 3).

The role of NKT cells in tumor immunity has not been studied extensively. NKT cells are currently classified into three types (40): type I, classical NKT cells; type II, non-classical NKT cells; and NKT-like cells or CD1d-independent NK1.1⁺ T cells. Classical NKT cells expressed TCR-Vβ11 and/or CD161 (6, 40). Human classical NKT cells used an invariant TCR-Vα24, whereas mouse NKT cells used TCR-Vα14 (1, 28). In contrast, the T-cell receptor V regions used by type II and NKT-like cells are highly diverse. The observation that intratumor Vα24⁺ T cells invariably coexpressed TCR-Vβ11 and other natural killer markers indicates that they are invariant Vα24⁺ NKT cells.

In this study, we found that the density of Vα24⁺ NKT cells in colorectal carcinomas increased significantly. Thus far, other groups have reported a remarkable decrease of invariant NKT cells in peripheral blood in patients with advanced cancers (41, 42). We observed that Vα24⁺ NKT cells actually existed in the luminal space of tumor-infiltrating vessels. These observations suggest that Vα24⁺ NKT cells could migrate from the peripheral blood to the tumors.

The present study indicates that the accumulation of intratumor NKT cells provide a better prognosis for colorectal carcinoma patients. However, the effector mechanism of antitumor activity has remained speculative. We found that the higher fraction of Vα24⁺ NKT cells in colorectal carcinomas expressed activation marker CD69 than those in normal mucosa, and probably execute local cytotoxicity against tumor cells by IFN-γ and apoptosis-inducing molecules including FasL, perforin, and granzyme B (29). In this way, NKT cells may play a role in the primary defense against colorectal carcinomas by induction of tumor cell death. Alternatively, the increase of NKT cells in the tumors may represent a surrogate marker of antitumor activities.

Coca et al. (43) have previously reported that colorectal carcinomas with higher CD57⁺ natural killer cell infiltration have better prognosis. Because the CD57⁺ cell population includes NKT cells, the present observation is partly compatible with their conclusion.

Another interesting feature is that Vα24⁺ NKT cells play a role in the inhibition of lymph node metastasis of colorectal carcinomas. Although Vα24⁺ NKT cells are normally rare in lymph nodes (4, 34), a large number of these cells appear in the metastasis-free swollen lymph nodes of colorectal carcinoma patients. The precise mechanism for Vα24⁺ NKT cells to inhibit lymph node metastasis has remained obscure. However, it is possible that activated Vα24⁺ NKT cells migrate to regional lymph nodes, where they injure the metastasizing colorectal carcinoma cells and consequently prevent lymph node metastasis.

Statistically, patients with higher intratumor NKT-cell infiltration showed significantly longer overall and disease-free survival rates. This is consistent with the fact that these patients had much less lymph node metastasis. Our findings support intratumor infiltration of NKT cells as a promising variable to predict the prognosis of patients with colorectal carcinoma.