Biological Significance of Isolated Tumor Cells and Micrometastasis in Lymph Nodes Evaluated Using a Green Fluorescent Protein–Tagged Human Gastric Cancer Cell Line

Detection of microscopic residual disease by immunostaining and PCR has been a subject of thorough research for more than a decade. Recently, the International Union against Cancer (UICC) and the American Joint Committee on Cancer defined minute metastases identified by these techniques as either isolated tumor cells or micrometastasis. Based on its criteria, isolated tumor cell was defined either as single tumor cells scattered in the stroma or a small cell cluster <0.2 mm in diameter. Micrometastasis was identified as a metastatic focus ranging from 0.2 to 2 mm in diameter (1). Isolated tumor cells or micrometastasis in the lymph nodes reportedly can be detected at the incidence of 10% to 50% from the specimens of node-negative gastric cancer patients (2). Several immunohistochemical studies have reported that isolated tumor cells or micrometastasis in the lymph nodes did not affect the survival of gastric cancer patients (3–6). Other studies, including immunohistochemical and molecular biological method, however, have reported lower survival of gastric cancer patients with than without micrometastasis (7–11). Thus, the effects on patient outcome of isolated tumor cells or micrometastasis in the regional lymph nodes remain controversial. To clarify the clinical significance of micrometastases, a basic understanding of the early process of micrometastasis development and the positive and negative regulatory mechanism underlying growth of micrometastasis in vivo using an animal model is crucial.

Lymph node metastasis models have been less available than hematogenous metastasis models. To date, however, the number of reported lymph node metastasis models are gradually increasing, including lung, breast, prostate, colon, and oral cancer cell lines (12–17), and there are a few lymph node metastasis models of gastric cancer in nude mice as well (18, 19). Nevertheless, such conventional models pose some difficulties for micrometastasis research because laborious histologic or molecular biological examination is necessary for detection of micrometastasis after sacrifice of mice. The green fluorescent protein (GFP) gene, cloned from the genome of the jellyfish Aequoreavictoria, is known to yield a bright, stable green fluorescence in live tissue (20). Chishima et al. (21) first used GFP for detecting tumor cells at a single cell level in vivo and Yang et al. (22) reported whole-body imaging of tumors and metastases in intact mice with GFP technology. Since then, a variety of GFP-expressing metastasis models, which enable fluorescence imaging of metastatic growth noninvasively and real-time analysis of metastatic processes in living mice, have been reported (23, 24). As for lymph node metastasis, however, only three models tagged with GFP have been reported to our knowledge (25–27). We previously developed a human gastric cancer cell line tagged with GFP gene (GCIY-EGFP) as a peritoneal micrometastasis model (28, 29) and also found it to possess a high metastatic potential for regional lymph nodes as evidenced by inguinal lymph node metastases after s.c. inoculation of tumor cells into nude mice.

In the current study, using this model, we investigated the early development of lymph node metastasis from isolated tumor cells to micrometastasis and sought to evaluate the influence of resection of the primary tumor along with immunologic treatment and chemotherapy on micrometastasis formation. We here discuss the biological and clinical significance of isolated tumor cells and micrometastasis in the lymph nodes in gastric cancer patients.

Seven- to eight-week-old male athymic nude mice of KSN strain were obtained from Shizuoka Laboratory Animal Center (Hamamatsu, Japan) and maintained under specific pathogen-free conditions. All animal experiments were done under the experiment protocol approved by the Ethics Review Committee for Animal Experimentation of the Aichi Cancer Center. S-1 (Tegafur/5-chloro-2,4-dihydroxypyrimidine/potassium oxonate = 1:0.4:1), a dihydropyrimidine dehydrogenase inhibitory fluoropyrimidine, was purchased from Taiho Pharmaceutical, Inc. (Tokyo, Japan), and was dissolved in 0.5% carboxymethyl cellulose in saline.

The GCIY cell line, a poorly differentiated human gastric carcinoma cell line, was obtained from the RIKEN Cell Bank. This line was transfected with the pEGFP-C1 plasmid (Clontech Laboratories, Palo Alto, CA) using the FuGENE6 transfection reagent (Roche Diagnostics, Basel, Switzerland). A subline with bright GFP fluorescence (GCIY-EGFP) was used in this study as described previously (28). The cell line was maintained in DMEM containing 10% fetal bovine serum (Life Technologies, Grand Island, NY) with 100 units/mL penicillin and 100 μg/mL streptomycin sulfate (Sigma-Aldrich, St. Louis, MO) and cultured in a humidified 5% CO₂ incubator at 37°C.

Exponentially growing GCIY-EGFP cells were harvested with trypsin/EDTA, washed with HBSS, and resuspended in HBSS. A tumor cell suspension (5 × 10⁶/0.2 mL) was s.c. injected into the lower abdominal flank of nude mouse. The growth of primary s.c. tumor and the metastasis of inguinal lymph nodes in living mice were monitored weekly externally and noninvasively with a small, convenient fluorescent microscope system (GFP checker), which consists of an LG-PS halogen source that produces blue light (excitation: 450-490 nm) and a dissecting microscope with a cut filter (emission: 530 nm; SZ40-GFP, Olympus, Tokyo, Japan; ref. 30). Using this detection system, the time course of the development of lymph node metastases was first examined. Mice were inoculated with tumor cells, and five mice each were sacrificed at 1, 2, 3, and 5 weeks postinoculation. The inguinal lymph nodes were examined internally through skin incision in situ and then surgically removed, and metastases in the isolated nodes were examined by the GFP checker at ×20 to ×40 magnification. The metastasized lymph nodes were cut along an appropriate line under the guide with blue light illumination, fixed in formalin, embedded in paraffin, and micrometastasis in the lymph nodes was confirmed histopathologically and immunohistochemically. The maximum diameter of the metastasis foci in the lymph nodes was measured under a dissecting microscope on the GFP image captured on a Windows PC with Nikon Capture 3 software. If multiple metastatic foci were present within a lymph node, the largest one in diameter was measured.

To elucidate the biological significance of isolated tumor cells and micrometastasis, the following three consecutive experiments were conducted.

Experiment 1: Effects of primary tumor resection. Mice were s.c. inoculated with 5 × 10⁶ GCIY-EGFP cells. Primary s.c. tumors were then surgically resected 1, 2 (isolated tumor cells stage), 3, and 5 (micrometastasis stage) weeks postinoculation. Lymph node metastasis was monitored externally by the blue light illumination. Mice were sacrificed 4 weeks after surgery and the inguinal lymph nodes were removed. The presence or absence of metastasis in the lymph nodes and the size of the largest metastatic foci in each group were measured with a GFP checker.

Experiment 2: Effects of anti-asialo GM1 antibody treatment. Mice were s.c. inoculated with 5 ×10⁶ GCIY-EGFP cells, and the rabbit anti-asialo GM1 antibody (400 μg/0.2 mL/mouse; Wako Pure Chemicals, Osaka, Japan) or control normal rabbit IgG (Upstate, Lake Placid, NY) was injected i.p. twice a week for 3 weeks starting from 1 day before inoculation of tumor cells (total seven times). Mice were sacrificed 3 weeks postinoculation and the volume of primary tumor and the incidence and the size of inguinal node metastasis were examined by the GFP checker. The volume of primary tumor was calculated by the following formula, 1/2LW², where L is the tumor maximum diameter and W is the diameter at right angle to the axis. In the subsequent study, the primary tumor was resected 2 weeks after s.c. inoculation of 5 ×10⁶ GCIY-EGFP cells. Mice were injected with either the anti-asialo GM1 antibody or control IgG in the same schedule as described above. Mice were sacrificed 3 weeks after resection, and the incidence and the size of inguinal node metastases were assessed by the GFP checker.

Experiment 3: Effects of postoperative chemotherapy. Primary s.c. tumors were resected 3, 5 (micrometastasis stage), and 7 (macroscopic metastasis stage) weeks after inoculation. The mice were then orally administered S-1 (20 mg/kg/d) or vehicle (0.5% carboxymethyl cellulose) through a gastric tube five times a week for 4 weeks, starting from 1 day after resection. Mice were sacrificed 4 weeks after resection, and metastases of the inguinal lymph nodes were assessed by the GFP checker.

Keratin immunohistochemical staining of micrometastasis in the lymph nodes was done using mouse monoclonal antibody (mAb) against a broad spectrum of human cytokeratin (AE1/AE3, DAKO, Copenhagen, Denmark) with optimal dilution as described previously (31).

To confirm the depletion of natural killer (NK) cells with anti-asialo GM1 antibody, fluorescence-activated cell sorting analysis was done with peripheral blood leukocytes of nude mice with or without anti-asialo GM1 treatment (one i.p. injection, 400 μg/0.2 mL/mouse). Peripheral blood leukocytes were harvested by the density gradient method with HISTOPAQUE (Sigma-Aldrich) and washed twice with fluorescence-activated cell sorting buffer (5 mmol/L EDTA and 5 mg/mL bovine serum albumin containing PBS), and reacted on ice with FITC-conjugated anti-mouse Pan-NK cells mAb (BD Biosciences, San Diego, CA) for 30 minutes. After washing twice with fluorescence-activated cell sorting buffer, cells were then incubated with red phycoerythrin-conjugated anti-mouse CD19 mAb (BD Biosciences) on ice for an additional 30 minutes. The labeled cells were washed and the intensity of fluorescence was evaluated by a flow cytometer, FACSCalibur (BD Biosciences).

The statistical significance of differences in data for metastases between groups was determined by Student's or Welch's two-tailed t tests. Differences in the incidence of lymph node metastasis between groups were analyzed with χ² test or Fisher's exact test.

Development of regional lymph node metastasis after s.c. injection. Inguinal lymph node metastases could be externally visualized with blue light illumination in intact mice noninvasively (Fig. 1A and B) and the metastasis could be confirmed by internal visualization after skin incision in live mice (Fig. 1C). Detection limit was 0.2 to 0.4 mm in diameter by whole-body imaging. With this GFP imaging system, inguinal lymph node metastases could be externally detected at the incidence of 15%, 65%, 75%, and 100% in tumor-bearing mice at 1, 2, 3, and 5 weeks after s.c. inoculation of GCIY-EGFP cells, respectively (Fig. 2I, black columns). Dissecting fluorescence microscopic observation of the removed lymph nodes enabled more sensitive detection of metastases in the lymph node even in the early stage (1 and 2 weeks) after injection of tumor cells (Fig. 2A-D). Consequently, the detection rate of metastasis was increased up to 100% at all stages (Fig. 2I, gray columns). The maximum diameter of the metastatic foci in the nodes was linearly increased with time (i.e., 60, 180, 420, and 650 μm at 1, 2, 3, and 5 weeks postinjection, respectively; Fig. 2J), indicating that minute metastases within 2 weeks postinjection were <0.2 mm in diameter and, therefore, classified as isolated tumor cells, whereas metastases during 3 to 5 weeks postinjection ranged from 0.4 to 1.0 mm in diameter and were classified as micrometastasis based on the UICC criteria. The presence of such isolated tumor cells and micrometastasis in the lymph nodes was confirmed by the immunohistochemical staining for keratin (Fig. 2E-H). In summary, isolated tumor cells generated in the inguinal lymph nodes of all the mice by 1 week postinoculation grew continuously to develop into micrometastasis by 3 weeks postinjection in this model.

Effects of primary tumor resection on growth of isolated tumor cells and micrometastasis in lymph nodes (experiment 1). Experimental protocols using this micrometastasis model were schematically represented in Fig. 3A. When the primary tumor was surgically resected 1 or 2 weeks postinoculation of tumor cells, the incidence of nodal metastasis 4 weeks after resection was dramatically decreased (0%) compared with the incidence of nodal metastases (100%) in sham-operated control mice (Fig. 4A-B, E-F, and I). However, when the resection was done 3 or 5 weeks after inoculation, inhibition of metastasis was limited and the incidence of nodal metastasis remained 60% and 100%, respectively (Fig. 4C-D, G-H, and I). In addition, in mice whose primary tumor was resected 5 weeks postinoculation, the size of metastatic foci in the nodes was significantly increased after primary tumor resection (P < 0.01; Fig. 4J), indicating progressive growth of micrometastasis at later stages irrespective of primary tumor removal. These results suggest preferential regression of isolated tumor cells, but not micrometastasis, in the lymph nodes by primary tumor removal.

Effects of anti-asialo GM1 antibody treatment on spontaneous regression of isolated tumor cells in lymph nodes (experiment 2). To clarify the mechanism by which isolated tumor cells in the lymph nodes was regressed by the removal of primary tumor, we examined the effect on isolated tumor cells regression of anti-asialo GM1 antibody, which is capable of depleting NK cells effectively in vivo (Fig. 3B). Flow cytometric analysis confirmed selective and significant depletion of Pan-NK positive fraction (NK cells) in the peripheral blood of nude mice by the treatment (one i.p. injection) with anti-asialo GM1 antibody compared with the blood from mice injected with control IgG. CD19-positive fraction (B cells) and double-negative fraction (macrophage) were unchanged by the antibody treatment (Fig. 5A). We then tested the influence of the anti-asialo GM1 treatment on the primary tumor growth and the development of inguinal lymph node metastasis. The results showed that the primary tumor volume and the maximum diameter of metastatic foci in the lymph nodes were significantly increased with anti-asialo GM1 treatment (P < 0.05 and P < 0.01, respectively) compared with control treatment (Fig. 5B). Furthermore, complete recovery from abrogation of inguinal lymph node metastasis by the primary tumor removal was achieved by treatment with anti-asialo GM1 as evident from the presence of micrometastasis in the inguinal lymph nodes at 100% incidence (Fig. 5C and D). These results indicate that the spontaneous regression of isolated tumor cells occurring after resection of the primary tumor in the immunodeficient nude mice is mainly due to the antitumor activity of the NK cells.

Effects of postoperative chemotherapy on growth of micrometastases in lymph nodes (experiment 3). As described above, the antitumor effect of NK cells proved to be mild and actually limited to isolated tumor cells, but not to micrometastasis. Therefore, we then tested whether chemotherapy following resection of the primary tumor can eliminate NK cell–resistant micrometastasis in the lymph nodes. In this experiment, the primary tumors were surgically resected 3, 5, and 7 weeks after inoculation of tumor cells, followed by oral S-1 administration from 1 day postoperation for 4 weeks (Fig. 6A). The respective incidence of inguinal lymph node metastasis in mice whose primary tumor was resected 3, 5, and 7 weeks postinjection was 70%, 90%, and 100% without S-1 treatment, against 40%, 80%, and 100%, respectively, with S-1 treatment (Fig. 6B), indicating that the smaller the micrometastasis in size, the higher sensitivity to S-1. The maximal diameters of the metastatic foci in the inguinal lymph nodes in S-1-treated mice whose primary tumor was resected 3, 5, and 7 weeks postinjection were 0.51 mm (not significant), 0.54 mm (P < 0.05), and 2.0 mm (not significant), respectively, against 0.68, 2.0, and 2.3 mm, respectively, without S-1 treatment (Fig. 6C). These results indicate higher sensitivity of early-stage (5 weeks) micrometastasis to S-1 than that of late-stage (7 weeks) micrometastasis.

Isolated tumor cells as defined by the UICC (2002) is characterized by qualitative features, such as the lack of contact with lymph sinus, extrasinusoidal stromal reaction, and lack of tumor cell proliferation. Therefore, isolated tumor cells is not always regarded as true lymph node metastasis and is currently recorded as pN0(i+). However, several investigators who have reported the prognostic significance of isolated tumor cells and micrometastasis to the regional lymph nodes are not convinced of the appropriateness of this stage classification (7–9). Horstmann et al. (7) explained the reasons for this dispute in their report that morphologic criteria of isolated tumor cells are somewhat subjective and may not be reproducible, including false positives. Another reason for the discrepancy is poor understanding of the etiology and development of lymph node micrometastasis because of the difficulty in establishing an adequate in vivo model for isolated tumor cells and micrometastasis. In the present study, we established an attractive model of nodal micrometastasis, featuring a highly metastatic GFP-transfected gastric cancer cell line. In this model, not only micrometastasis but also isolated tumor cells in the inguinal lymph nodes could be detected in all mice 1 week after inoculation of tumor cells. However, no tumor cells were detected in lymph nodes immediately (1 day) after tumor cell inoculation, indicating that this is not an experimental but a spontaneous micrometastasis model that reflects lymphatic metastasis in human cancer tissues and, therefore, is useful for analysis of early-phase development of metastasis through lymphatics.

The most remarkable finding in the current study was the spontaneous regression of isolated tumor cells in the inguinal lymph node by resection of the primary lesion. This regression was no longer observed when sufficient time (>3 weeks) had passed after inoculation because isolated tumor cells develops into micrometastasis by that time. This finding supports the UICC choice of treating isolated tumor cells as pN0, and may have a role in solving the enigma described by Fukagawa et al. (4), namely that the presence of micrometastasis in the lymph node of the resected specimens had no prognostic significance among pT2 stage gastric carcinoma patients. Lymph nodes in the Fukagawa series had been extensively removed en bloc by thorough lymphadenectomy. Therefore, the excellent prognosis among patients with micrometastasis may simply reflect the benefit of systemic lymphadenectomy. However, the current data indicate that isolated tumor cells cannot survive in the isolated lymph nodes if continuous recruitment of cancer cells and nutrient supply from the primary tumor through lymphatic channels are interrupted by the resection of primary tumor, suggesting that isolated tumor cells depend for their growth on the primary tumor and are not equivalent to an established metastasis. Therefore, a good prognosis in patients with micrometastasis in the lymph node may be due to the innocent nature of isolated tumor cells as well as extensive lymphadenectomy.

Another finding of interest in this study is that anti-asialo GM1 antibody treatment promotes the growth of isolated tumor cells so as to develop into micrometastasis in the lymph nodes after removal of primary tumor. This suggests that the factor most responsible for the regression of isolated tumor cells after primary tumor resection is the immunologic activity conferred by NK cells against metastatic tumor cells in the lymph nodes. NK cells are known to play a major role in innate immunity and to figure importantly in immune surveillance for tumor cells in man and animals (32). NK cells can target tumor cells with reduced expression of class I MHC and confer cytotoxic activity against tumor cells independent of tumor-related antigens (33). Several previous studies indicated that the depletion of NK cells with anti-asialo GM1 antibody led to growth promotion of a primary tumor with a different origin in nude mice (34–37). Consistent with this finding, we found enhancement of primary tumor growth by treatment with anti-asialo GM1 in the present study. We further showed that NK cell–mediated immunity plays an important role in the inhibition of lymph node metastasis, especially in the early stage of metastasis development, including isolated tumor cells. Conversely, minimal residual disease must grow to a certain extent (micrometastasis level according to the UICC) to overcome the host immune response and proliferate on its own, suggesting that micrometasasis in the early stage may be a good therapeutic target to prevent recurrence in the form of lymph node metastasis by immunotherapy that potentiates NK cell activity, such as interleukin 2 or interleukin 12 (38, 39). NK cell may be enhanced via negative feedback mechanism by the allograft of human tumor cells lacking MHC class 1 on their surface in immunodeficient nude mice (40). Therefore, such an allograft system consisting of human tumor cells transplanted to nude mouse may be somewhat artificial in terms of the deficiency of ADCC and CTL activity, suggesting the limitation of the present result to a clinical setting. Further study using an isomatched (mouse/mouse) system is needed to overcome this problem and to understand the precise relationship between micrometastasis in the lymph nodes and immune response.

Finally, the effect of postoperative adjuvant chemotherapy was assessed in the current model of node metastasis. When the primary tumor in mice was resected 3 weeks after tumor cell inoculation without postoperative chemotherapy, the incidence of residual disease in the inguinal lymph nodes was 70%. This was reduced to 40% by postoperative adjuvant chemotherapy with oral S-1 starting from 1 day postoperation. This moderate decrease in the incidence of residual disease by chemotherapy was no longer observed when the primary lesion was allowed to grow for >5 weeks. However, the size of the metastatic foci in the lymph nodes was significantly (P < 0.05) diminished by chemotherapy if the treatment was initiated within 5 weeks after inoculation. These results suggest the importance and effectiveness of early initiation of treatments after metastasis as has been proved in our previous studies (28, 30, 41).

To conclude, a novel and useful animal model of lymph node micrometastasis using a GFP-tagged human gastric cancer cell line was established. Using this experimental model, we showed for the first time that isolated tumor cells in the lymph nodes regressed spontaneously after resection of the primary tumor, possibly through antitumor activity of the NK cells. Nodal metastasis that developed into micrometastasis was no longer NK cell sensitive and proliferated after resection of the primary tumor. However, it was relatively vulnerable to surgical intervention and subsequent postoperative adjuvant chemotherapy.

We thank K. Asai for expert technical assistance.