Role of Vascular Endothelial Growth Factor C Expression in the Development of Lymph Node Metastasis in Gastric Cancer

Gastric cancer is still a leading cause of death worldwide (1). Despite surgeons’ efforts to remove lymph node metastasis by aggressive resection, many patients have died of lymph node metastasis (2, 3, 4). Actually, >80% of patients with advanced gastric cancer have lymph node metastasis, and the remote lymph nodes, such as the para-aortic nodes, are involved in 20% of gastric cancers. Accordingly, control of lymph node metastasis is the most important strategy for the treatment of gastric cancer.

However, it is very difficult to predict or to diagnose the precise lymph node metastasis preoperatively or intraoperatively (5). The accuracy of diagnosis of lymph node metastasis by computed tomography and ultrasonography is reported to be low, with a rate of 10–20% (6). New methods to increase the accuracy rates of detection of lymph node metastasis using molecular biological techniques have been reported. In gastric cancer, we reported previously that the expressions of c-erbB-2 (7), epidermal growth factor receptor (8), c-met (9), autocrine motility factor receptor (10), and urokinase-type plasminogen activator and its receptor (10) are closely associated with lymph node metastasis. These proteins are now believed to have a great role in the intravasation of cancer cells into lymphatic capillary through destruction of stromal elements and lymphatic vessel or high motile activity. However, the expressions of these genes are not associated with the proliferation of lymphatic vessels. Lymph node metastasis is strongly related with the lymphatic invasion in the primary tumor, and the proliferation and dilation of lymphatic vessels are frequently found in the stroma of gastric cancer with nodal involvement.

More recently, Alitalo and colleagues (11) reported that VEGF-C² is a potential lymphoangiogenic factor and that this factor can selectively induce the hyperplasia of the lymphatic vasculature.

Here, we report the close relationship between VEGF-C expression and lymph node metastasis in gastric cancer. To the best of our knowledge, there is no published article on the relationship between VEGF-C tissue status and lymph node metastasis in cancer. This study suggests that the detection of VEGF-C protein in the primary gastric cancer may represent a potential risk of lymph node metastasis and poor prognosis.

The nine human cancer cell lines used in this study were from the following sources: prostatic adenocarcinoma PC-3 and gastric carcinomas MKN-28, MKN-45, KATO III, MKN-7, and AZ521 were from the Japanese Cancer Research Resources Bank; gastric carcinomas KKLS and NKPS were kindly provided by Dr. M. Mai (Cancer Research Institute, Kanazawa University, Kanazawa, Japan); and gastric cancer cell line TMK-1 was a kind gift from E. Tahara (Department of Pathology, Hiroshima University, Hiroshima, Japan). All cell lines except for AZ521 were maintained in RPMI 1640 supplemented with 10% heated-inactivated fetal bovine serum (Life Technologies, Inc.) at 37°C and 5% CO₂; AZ521 was incubated with MEM with 10% fetal bovine serum.

One hundred seventeen patients with primary gastric cancer who were diagnosed and treated in the Department of Surgery II, Kanazawa University Hospital, between 1990 and 1997 were entered into the study. All of the patients underwent gastrectomy with lymph node dissection. Among them, 25 patients had early gastric cancer (T₁ tumor), in which the tumor invasion is confined to the mucosa or submucosa.

All of the resected primary tumors and regional lymph nodes were histologically examined by H&E staining according to the general rules of the Japanese Classification of Gastric Carcinoma (12).

Immediately after resection of primary tumor, samples of ∼5 mm in diameter were taken from 32 primary tumors and each adjacent normal mucosa, and these samples were stored at −80°C. In addition, small pieces of tissue ∼5–8 mm in diameter were sampled from the primary tumor and fixed in acetone at −20°C overnight, dehydrated in acetone at 4°C for 15 min and acetone at room temperature for 15 min, and then cleared in methyl benzoate for 30 min and in xylene in a vacuum evaporating emitter (AMeX method; Ref. 13).

Anti-VEGF-C (C-20) and anti-Flt4 antibodies (15, 21) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-VEGF-C antibody (C-20) is an affinity-purified goat polyclonal antibody that is raised against a peptide corresponding to amino acids 136–155 mapping at the COOH terminus of the VEGF-C precursor of human origin. Anti-Flt4 antibody is an affinity-purified rabbit polyclonal antibody raised against a peptide corresponding to amino acids 1279–1298 mapping at the COOH terminus of the precursor form of Flt4 of human origin.

Immunohistochemistry was performed as follows. Paraffin sections (3–4 μm thick) were deparaffinized and then placed in a solution of absolute methanol and 2% hydrogen peroxide for 30 min. They were subsequently washed in distilled water, rinsed for PBS, and treated with 10-fold diluted blocking serum (rabbit) for 20 min for blocking of nonspecific reaction. The slides were then incubated overnight at 4°C in a humidified chamber with anti-VEGF-C antibody diluted 1:100 in PBS. After the overnight treatment, the slides were incubated with biotinylated goat antirabbit IgG for 20 min (Vectastain ABC kits; Vector Laboratories, Burlingame, CA) and then with premixed ABC (Vector Laboratories) reagent for 20 min. Immunostaining was performed by incubating the slides in diaminobenzidine (DAKO) solution containing 0.06 mm diaminobenzidine and 2 mm hydrogen peroxide in 0.05% PBS (pH 7.6) for 5 min. After chromogen development, the slides were washed, counterstained with methyl green, dehydrated with alcohol and xylene, and mounted in a routine fashion. Negative control were performed in all cases by omitting the first antibody. Only cases in which at least 20% of tumor cells were immunoreactive were scored as positive.

RT-PCR analysis was carried out according to the modifications by Conboy et al. (14). Briefly, total RNAs were extracted from primary gastric cancer and the normal mucosa, using Isogen (Nippon Gene, Tokyo, Japan; Ref. 16). The prepared RNA (1 μg) was mixed with the oligo(dT) (50 pmol), incubated for 15 min at 68°C, and then quickly chilled in an ice bath for 5 min. The RNA samples were reverse-transcribed at 42°C for 60 min into first-strand cDNA in reverse transcription solution [50 mm Tris-HCl (pH 8.3), 40 mm KCl, 8 mm MgCl₂, 0.5 mm each dNTP, 225 μg/ml BSA, 5 mm DTT, 20 units of RNasin (Promega, Madison, WI), and 4 units of avian myeloblastosis virus reverse transcriptase (Life Sciences, St. Petersburg, FL)] with a total volume of 20 μl. The cDNA samples were incubated at 95°C for 5 min to inactivate the reverse transcriptase and then chilled. The samples were amplified by the addition of 80 μl of PCR mixture [50 mm Tris-HCl (pH 8.3), 40 mm KCl, 8 mm MgCl₂, 0.5 mm each dNTP, 50 pmol of each sense and antisense primer, and 2.5 units of Taq polymerase (Takara, Tokyo)]. The amplification was performed for 1.5 min at 94°C, 2 min at 48°C, and 2 min at 72°C for 3 cycles, followed by 25 cycles of 40 s at 94°C, 1.5 min at 48°C, and 1.3 min at 72°C. The products were electrophoresed on 2% agarose gels and then transferred to a nylon membrane filter. The transferred products were hybridized overnight to a ³²P-end-labeled probe specific for the target cDNA fragment (Southern blotting). The autoradiogram was exposed for 4–5 h with two intensifying screens at −80°C.

Specific primers for the VEGF-C gene, targeting an 867-bp fragment, were: VEC-S3, 5′-AGTTTTGCCAATCACACTTCCTG-3′; and VEC-A3, 5′-GTCATTGGCAGAAAACCAGT-CTT-3′. The probe oligonucleotide was VEC-A2 (5′-CTTTCTGTACATTCACAGGCACA-3′). Primers for the flt4 gene were: Flt-4-1, 5′-AGCCATTCATCAACAAGCCT-3′; and Flt-4-2, 5′-GGCAACAGCTGGATGTCATA-3′ (PCR product, 298 bp). The probe was Flt-4-probe (5′-TTCCTTTCCAACCCCTTCCTGGTGCACATC-3′; Ref. 17). As internal standards, the β-actin gene and the transferrin receptor gene were used. Primer pairs specific to the β-actin gene (18), targeting a 865-bp fragment, were used: β-act-5, 5′-TTGAAGGTAGTTTCGTGGAT-3′; and β-act-11, 5′-GAAAATCTGGCACCACACCTT-3′. β-act-7 (5′-ACTGACTACCTCATGAAGAT-3′) was used as the probe. Primers for transferrin receptor (TFR), targeting a 415-bp fragment, were: TFR2-S, 5′-ACAGACTCTACATGTAGGAT-3′; and TFR4-A, 5′-AAACCTTGAAGTTGCTGGTA-3′. The probe was TFR4-P (5′-TATCCCTCTAGCCATTCAGT-3′; Ref. 19).

Twelve μl of protein sample (total protein 20 μg) were mixed with 6 μl of sample buffer [50 mm Tris-HCl (pH 6.5), 10% glycerol, 2% SDS, and 0.1% bromphenol blue] prior to separation by 10% SDS-polyacrylamide gel. After completion of electrophoresis, samples were transferred to polyvinylidene difluoride membrane filters (Immobilon; Millipore, Bedford, MA).

Immunoreactivity were performed using anti-VEGF antibody (C-20) and anti-VEGFR-3 (Flt4) antibody as primary antibodies and a peroxidase-conjugated secondary antibody. The transferred samples were conjugated with anti-VEGF-C antibody (C-20; 1:1000), anti-Flt4 antibody (1:1000), and anti-β-actin monoclonal antibody (Sigma Chemical Co., St. Louis, MO) for 2 h at room temperature. As a negative control, primary VEGF-C antibody, which was preabsorbed with synthetic VEGF-C peptide (Santa Cruz Biotechnology) overnight at a ratio of 1:1, was reacted with the membrane filter. VEGF-C peptide was used as a positive control. After washing with TBS-Tween 20 with 0.05% skim milk to block a non-specific reaction, membranes were incubated with peroxidase-conjugated Affipure F(ab′)₂ fragment rabbit antigoat IgG F(ab′) fragment (Jackson ImmunoResearch Laboratory, Inc.; 1:5000), and antirabbit IgG horseradish peroxidase linked F(ab′) fragment (Amersham Life Science; 1:5000) for VEGF-C, Flt4, and β-actin for 40 min at room temperature, respectively. The antigen and antibody complexes were detected using ECL Western blotting detection reagent (Amersham Life Science).

All statistical calculations were carried out using SPSS statistical software. Data are presented as means ± SD. The χ² test was used to analyze data. The outcomes of the different groups of patients were compared by the generalized Wilcoxon test. A multivariate model using Cox stepwise regression analysis was used to evaluate the statistical strength of independent association between selected covariates and patient survival. Values with a P of ≤0.05 were considered to be statistically significant.

Among eight gastric cancer cell lines, NKPS, TMK-1, AZ521, and MKN-45 expressed VEGF-C mRNA, but VEGF-C expression was not detected in KATO III, MKN-7, KKLS, and MKN-28 (Fig. 1). However, flt4 mRNA was expressed from KKLS only. PC-3 overexpressed not only VEGF-C but also flt4 mRNA (Fig. 1).

A series of 32 gastric cancers and adjacent normal mucosa was examined for the expression of VEGF-C and flt4 mRNA by RT-PCR (Fig. 2). VEGF-C was expressed in 15 (47%) and 4 (13%) of 32 gastric cancers and the normal gastric mucosa, respectively (Table 1). In contrast, flt4 mRNA was expressed in 17 (53%) and 20 (63%) of 32 gastric cancers and normal gastric mucosa, respectively.

There was a significant correlation between VEGF-C mRNA expression and flt4 mRNA expression in primary tumors (Table 1).

In gastric cancer cell lines, C-20 recognized main band of M_r 58,000 precursor of VEGF-C and proteolytic forms of M_r 31,000, 29,000, and 21,000 molecules (Fig. 3). In the clinical specimens, a M_r 31,000 molecule was predominant (Fig. 4), and the intensity of VEGF-C expression in cancer tissue was usually stronger than that of the normal counterpart. By preabsorption of VEGF-C antibody with synthetic VEGF-C peptide, M_r 55,000 and 31,000 bands, corresponding to VEGF-C peptides and VEGF-C, respectively, disappeared.

In gastric cancer cell lines, however, Flt4-specific rabbit IgG did not recognize a M_r 125,000 band that corresponded to Flt4 protein (Fig. 5). In contrast, a M_r 125,000 Flt4 band was detected in the clinical specimen.

In normal gastric epithelium, VEGF-C was weakly expressed in the cytoplasm of parietal cells, but almost all epithelial cells, stromal cells, and smooth muscle did not show VEGF-C immunoreactivity (Fig. 6). In gastric cancers, VEGF-C was observed as diffuse cytoplasmic staining (Figs. 7 and 8). Among 117 examined tumors, 30 (26%) tumors showed high expression of VEGF-C protein.

Among 27 primary tumors, cancer cells were positive for Flt4 immunoreactivity only in one tumor, which was poorly differentiated adenocarcinoma. The other 26 tumors did not show any Flt4 immunoreactivity in cancer cells. However, cancer cell stroma always showed Flt4 immunoreactivity, and Flt4 was strongly expressed in the cytoplasm of lymphatic endothelial cells (Fig. 9).

Table 2 shows a correlation between clinicopathological parameters and VEGF-C mRNA expression in the primary tumors. VEGF-C mRNA expression in primary tumor showed a statistically significant relationship with lymph node metastasis, lymphatic invasion, and venous invasion. However, there was no relationship between flt4 mRNA expression in primary tumors and lymph node metastasis, lymphatic invasion, or venous invasion (Table 3).

As shown in Table 4, a strong correlation was found between VEGF-C protein expression and lymph node status, lymphatic invasion, venous invasion, and infiltrating patterns. Tumors with high expression of VEGF-C protein had significantly higher incidence of the involvement of the lymph node metastasis (P < 0.05). In addition, the incidence of positive lymphatic invasion in tumors with high expression of VEGF-C was significantly higher than that in tumors with low VEGF-C expression.

Table 5 shows the correlation between VEGF-C tissue status and the grade of lymph node metastasis. Distant lymph node stations were more frequently involved in VEGF-C-positive tumors than in VEGF-C-negative tumors.

Fig. 10 shows the survival curves of patients according to the VEGF-C tissue status. Patients with high expression of VEGF-C (n = 30) had a significantly poorer prognosis than did those in low VEGF-C expression group (n = 87). When the VEGF-C tissue status and six clinicopathological parameters, which have been considered as the good prognostic indicators, were entered into the regression model, depth of wall invasion (T), lymph node metastasis (N), and VEGF-C tissue status emerged as independent prognostic parameters (Table 6). Among seven prognosticators, VEGF-C tissue status was ranked third as an independent risk factor for death.

VEGF family is the only growth factor that is specific for vascular endothelial cells, and it consists of VEGF-A, VEGF-B, and VEGF-C. Among them, VEGF-C is ranked first as a lymphoangiogenic factor (20), which induces proliferation of lymphatic endothelial cells and lymphatic vessels via activating VEGFR-3 (Flt4) or VEGFR-2 (KDR) on the cell membrane of endothelial cells (21). However, the activity in proliferating endothelial cells via VEGF-C and VEGFR-2 loop is considered to be decreased in adult tissue. Accordingly, the main receptor of VEGF-C in adult tissue is considered Flt4. Oh et al. (22) reported that VEGF-C can induce proliferation of lymphatic endothelial cells and development of new lymphatic sinuses on the avian chorioallantoic membrane. It has been considered that the mechanism of the neogenesis of the lymphatic channel is activation of VEGFR-3 by VEGF-C, which is abundant in the proliferating endothelial cells of vascular sprouts and branching vessels of embryonic tissues (23).

Neogenesis and dilation of the lymphatic vessels are often found in the stroma of gastric cancer tissue, and intravasation of cancer cells in the lymphatic vessels is known as lymphatic invasion. Thus far, the mechanism of the dilation of lymphatic vessels in gastric cancer tissue has been speculated to be the obstruction of lymphatic channel by the cancer cell emboli. However, the mechanisms of the lymphoangiogenesis in cancer tissue have remained unclear.

It is generally accepted that the cancer cell invasion into the lymphatic vessels is established by the destruction of stroma around the lymphatic vessels via activation of matrix-digesting enzymes, which are produced by cancer cells or fibroblasts (24, 25), and by the enhanced motility activity of cancer cells via motility factor and their receptor, such as AMF/AMFR (26) or c-met/HGF (9, 27).

This study clearly demonstrated the close relationship between VEGF-C expression and lymphatic invasion and lymph node metastasis. Tumors with high expression of VEGF-C had more remote lymph node involvement than did those with low VEGF-C expression. Considering that VEGF-C is selectively expressed in cancer cells but not in normal gastric mucosa or stromal elements, VEGF-C produced from gastric cancer may induce lymphovascular neogenesis and dilation of lymphatic vessel in the stroma of primary tumor (23). Immunohistological study has shown that the VEGF-C receptor flt4 was expressed on the lymphatic endothelial cells in the stroma of primary tumors and normal mucosa. Furthermore, there was a close correlation between flt4 and VEGF-C mRNA expression in the primary tumors, and immunohistochemical examination showed that cancer cells in the lymphatic vessels frequently expressed VEGF-C. In contrast, flt4 mRNA was not detected in eight examined gastric cancer cell lines, and this finding was in accordance with immunohistological evidence showing that almost all gastric cancer cells did not express Flt4. Furthermore, Flt4 immunoreactivity was only found on the lymphatic endothelial cells in the stroma of primary tumors and stomach wall. These results strong suggest the possibility that VEGF-C produced by cancer cells stimulate the lymphoangiogenesis and dilation of lymphatic vessels by activating Flt4, which is expressed on the lymphatic endothelial cells. Joukov et al. (28) reported that mature VEGF-C (M_r 21,000) also increases the permeability of lymphatic vessels as well as the migration and proliferation of endothelial cells. Our Western blot analysis in the clinical specimens showed that the mature VEGF-C was found in the cancer cells. Accordingly, VEGF-C may enhance the intravasation of cancer cells via loosening the mutual attachment of lymphatic endothelial cells (28). These results strongly suggest that cancer cells producing VEGF-C may have a higher potential for the intravasation into the lymphatic vessel, resulting in the lymph node involvement.

This study also demonstrated that the patients with VEGF-C-positive tumor had a poorer prognosis than did those with VEGF-C-negative tumor. Furthermore, VEGF-C tissue status was confirmed to be a significantly independent factor for poor prognosis. It has been speculated that the reason is the prevalent progression by the higher lymphatic spread and more infiltrating spread in VEGF-C-positive tumors than in the VEGF-C-negative tumors. Further studies of the proliferative and motility activity of VEGF-C-producing tumors should be performed.

In summary, this study demonstrates the close relationship between VEGF-C tissue status and lymphatic spread and postoperative prognosis in gastric cancer. VEGF-C-producing cancer cells may induce the proliferation and dilation of the lymphatic vessel in the stroma around them. In these circumstances, cancer cells can easily invade the lymphatic vessel, even if these cells have low production of matrix-digesting enzymes and scatter factors. The preoperative determination of VEGF-C tissue status by RT-PCR or immunohistochemistry using endoscopically biopsied materials may be useful in deciding the extent of surgical lymph node resection and postoperative chemotherapy.