Gastric cancer is the most common malignant tumor of the digestive system in China and its effective therapy depends on early diagnosis. The biomarkers that are currently available for the early diagnosis of gastric cancer are limited, and the specificity and sensitivity of the biomarker assays are not high. Therefore, the investigation of the molecular mechanisms underlying the invasion and metastasis of gastric cancer and the search for associated biomarkers are key areas in basic and clinical research (Hippo et al., 2002). It is well known that cancer is typically an angiogenesis-dependent pathological process, involving prominent angiogenesis during the initiation and development of tumors. The established blood supply subsequently promotes the growth, invasion, and metastasis of cancer (Xia et al., 2006). Hepatocyte growth factor (HGF) is closely associated with mitotic division and cell migration, as well as angiopoiesis. In addition, HGF can enhance the movement and invasive ability of cancer cells (Han et al., 1996). Through binding with its receptor c-Met, HGF can activate vascular endothelial cells (VECs), resulting in the proliferation and migration of VECs, which participate in neoangiogenesis in cancer tissues (Lasagna et al., 2006). On the other hand, basic fibroblast growth factor (bFGF) participates in tumor neovascularization. bFGF can stimulate proliferation and differentiation of VECs, promoting angiogenesis and enhancing vascular permeability, thereby also promoting the metastasis of cancer cells (Gan et al., 2006). Recent studies have shown that both HGF and bFGF are overexpressed in various cancers, such as lung cancer (Kuhn et al., 2006), hepatoma (Qin and Tang, 2002), breast cancer (Hirtenlehner et al., 2002), prostatic cancer (Gravdal et al., 2006), and colon cancer (Kammula et al., 2007), demonstrating a close association with angiogenesis, as well as with the progress of cancer. However, to date, there are very few reports investigating the role of both cytokines in the clinical prognosis of patients with gastric cancer. In this study, the expression of HGF and bFGF in gastric cancer tissues from Chinese patients was assayed to investigate the relationship between them and among other pathological parameters, such as microvascular density (MVD) and invasion-metastasis. These studies aimed to identify and evaluate new biomarkers for the prognosis of gastric cancer.
We have investigated the correlations among hepatocyte growth factor (HGF) mRNA expression, basic fibroblast growth factor (bFGF) mRNA expression, tumor microvessel density (MVD), and clinical pathological features of gastric cancer in Chinese patients. In situ hybridization was used to detect the expression of HGF and bFGF mRNAs, and immunohistochemistry was used to detect CD34 in 105 gastric cancer tissues and in 20 normal control tissues. The rate of HGF mRNA expression in normal gastric tissues (25%) was significantly lower than that (57.1%) in tumor tissues (P < 0.01). The rates of HGF mRNA and bFGF mRNA expression and MVD in T3–T4 stage tissues were higher than those in T1–T2 stage tissues (P < 0.01); the HGF mRNA expression rate was directly correlated with the bFGF mRNA expression rate (P < 0.05), and they were also directly correlated with MVD (P < 0.01). The mean survival time and the 5-year survival rate of patients who were positive for expression of HGF mRNA and bFGF mRNA and who had a MVD ≥ 39.5/0.72 mm2 were significantly shorter than those who did not express HGF mRNA and bFGF mRNA and who had a MVD <39.5/0.72 mm2. Both HGF and bFGF may participate in angiogenesis in gastric cancer and may be involved in tumor invasion and metastasis. HGF and bFGF mRNA expression can be used as useful parameters to evaluate the prognosis of gastric cancer. Anat Rec, 2009. © 2009 Wiley-Liss, Inc.
MATERIALS AND METHODS
Specimens and Reagents
Specimens were obtained from 105 patients with gastric cancer who underwent gastrointestinal surgery in the People's Hospital of Zhejiang Province, China, during October 1986–November 1998. Full follow-up data over 5 years, until November 2003, was recorded for all patients. The median age was 57.6 (38–78), and the ratio of male to female was 2:1 (70/35). According to the WHO classification criteria of gastric cancer (2002), 71 cases were tubular adenocarcinomas, 17 were papillary adenocarcinoma, nine were mucus adenocarcinomas, and eight were signet-ring cell carcinoma. Sixty-three cases presented a high and moderate degree of differentiated carcinoma and 42 presented low and undifferentiated carcinoma. Forty-eight cases showed expansive type growth and seven showed infiltrative type growth. According to the staging criteria for gastric cancer, 20 cases were in T1 stage, 24 in T2, 29 in T3, and 32 in T4. Seventy-six cases were with vessel invasion, 29 were without vessel invasion, 70 were with lymph gland metastasis, 35 were without lymph gland metastasis, 42 were with distant metastasis (24 peritoneal metastasis and 18 liver metastasis), and 63 were without distant metastasis. Twenty control specimens were taken from nontumorous gastric mucosa without hyperplasia or atypical hyperplasia and 5 cm away from the edge of a tumor. All specimens were approved for inclusion in the experiment according to the guidelines of the Ethics Committee of the People's Hospital of Zhejiang province.
Reagents: (1) HGF and bFGF digoxigenin-labeled oligonucleotide probes and an in situ hybridization kit were purchased from Wuhan Boside, China. The detected sequences for HGF (NCBI accession number: NM_000601.4) mRNA were 5′-AGACA CCACA CCGGC ACAAA TTCTT GCCTG-3′, 5′-ATTTG GAATG GAATT CCATG TCAGC GTTGG-3′, and 5′-GTAGC ATATT ATGCA AAATG GATAC ACAAA-3′. The detected sequences for bFGF mRNA (NCBI accession number: NM_002006.4) were 5′-GGCTT CTTCC TGCGC ATCCA CCCCG ACGGC-3′, 5′-AGCAG AAGAG AGAGG AGTTG TGTCT ATCAA-3′, and 5′-TGGAA TCTAA TAACT ACAAT ACTTA CCGGT-3′. (2) Immunohistochemical reagents: mouse anti-human CD34 antibody and streptavidin-peroxidase-biotin (SP) reagent were purchased from Zymed and used at a working concentration of 1:50.
In Situ Hybridization Assay
Paraffin sections were deparaffinized with dimethyl benzene, followed by hydration through rinsing in a graded series of alcohol. Sections were then incubated with 3% H2O2 at room temperature for 10 min, digested with 3% pepsin diluted in citric acid at 37°C for 20 min, and then washed for 5 min three times with 0.5 mol/L PBS. For prehybridization, each section was incubated with 20 μL of prehybridization solution at 40°C for 2 hr, then 20 μL hybridization solution was added and the sections were incubated at 42°C overnight (16 hr) (the probe concentration was 2 μg/mL). Sections were subsequently washed for 5 min three times with 2× SSC (standard saline citrate) at 37°C and for 5 min three times in 0.2× SSC at 37°C. Then, the sections were blocked with blocking solution at room temperature for 30 min. Biotinylated mouse anti-human digoxigenin-labeled IgG (Wuhan Boside, China) was added onto the sections and incubated at 37°C for 60 min, followed by washing for 2 min three times with 0.5 mol/L PBS. Streptavidin-biotin-complex was then added and incubated at 37°C for 20 min, followed by washing for 5 min three times with 0.5 mol/L PBS. Finally, the sections were incubated with biotinylated peroxidase (Zymed) at 37°C for 20 min, followed by washing for 5 min four times with 0.5 mol/L PBS. Sections were then stained with DAB, contrast stained with hematoxylin, dehydrated, and mounted. The negative controls included hybridization solution alone, without probe, and specimen pretreatment with RNase.
Immunohistochemical studies were performed following the manufacturer's instructions (SP, Zymed). Five microns sections of tissue were deparaffinized and dehydrated. Endogenous peroxide activity was blocked with 3% hydrogen peroxide for 20 min. After rinsing three times with 0.01 mol/L PBS, pH = 7.4, sections were incubated with 10% normal goat serum at room temperature for 10 min to block nonspecific reactions and then incubated for 2 hr with anti-CD-34 antibody. After rinsing in PBS for 5 min, sections were incubated with biotinylated secondary antibody for 15 min at room temperature, then incubated with ABC complex for 20 min at room temperature, then rinsed in PBS for 5 min, stained with DAB (3,3-diaminobenzidine), and finally washed in PBS. PBS replaced anti-CD-34 antibody as a negative control. The positive control was the section provided with the kit.
Cytoplasm that was positive for the expression of HGF mRNA was brown-yellow. All slides were observed under an Olympus Light Microscope and representative photographs were taken by Nikon 4500 camera. Five fields of vision were observed under high-power magnification, and 200 cells were counted each field. Based on the ratio of positive cells to total cells, levels of expression were classified as follows: positive cells <10% (−), 11%–50% (+), 51%–75% (++), and >75% (+++). The same criteria were used to classify bFGF expression.
MVD Assessment in Tumor Tissues
Initially, under low-power magnification, sections were scanned to find fields with cleanly stained endotheliocytes and cancer cells, and the microvessels were concentrated and backgrounds were clear. In tumor fields, a brown-yellow–stained cell or a cluster of endotheliocytes was counted as a blood vessel. Vessels were enumerated by counting the number of discrete stained structures within each field. In a 200 mm2 field, five random fields were chosen in which the MVD was assessed based on the CD34 staining, expressed as ± s. Values for high MVD (MVD ≥ 39.5/0.72 mm2) and low MVD (MVD <39.5/0.72 mm2) were determined according to the mean MVD (39.5/0.72 mm2) counted from 105 specimens of gastric cancer.
SPSS 13.0 statistical software was used for statistical analysis. The t-test was used for quantitative data analysis, and the χ2 test and Fisher's exact test were used for qualitative data analysis. Kaplan-Meier analysis was used for survival data. P < 0.05 was considered as statistically significant.
Expression of HGF mRNA in Normal and Cancerous Gastric Tissues
Expression of HGF mRNA in normal and cancerous gastric tissues was studied (Table 1). The expression of HGF mRNA in normal or noncancerous tissue is generally weak and focal, and largely limited in the cytoplasm of epithelial cells of deep gastric pits with some nonspecific signals in the lamina propria while the surface epithelial cells are negative (Fig. 1). The signals were only detected in 5 of 20 cases (25%) of noncancerous gastric mucosa. In contrast, the signals of HGF mRNA were detected in 60 of 105 cases with much stronger brown-yellow coarse cytoplasmic staining (Fig. 2). The rate of HGF mRNA expression is 57.1%, which is significantly higher than that in the normal or noncancerous gastric tissue (χ2 = 6.954, P < 0.01).
|Clinicopathologic parameters||N||HGF mRNA expression||χ2||P||bFGF mRNA expression||χ2||P||t||P|
|−ve||+ve||−ve||+ve||MVD N/0.72 mm2|
|Expansive||48||24||24||27||21||34.70 ± 15.23|
|Infiltrative||57||21||36||14||43||43.42 ± 12.64|
|G1+G2||63||28||35||26||37||55.72 ± 21.56|
|G3+G4||42||17||25||15||27||60.70 ± 20.73|
|T1∼T2||44||29||15||31||13||30.84 ± 13.66|
|T3∼T4||61||16||45||10||51||45.64 ± 11.69|
|No||29||19||10||25||4||25.69 ± 10.11|
|Yes||76||26||50||16||60||44.68 ± 12.33|
|No||35||22||13||25||10||27.07 ± 11.33|
|Yes||70||23||47||16||54||45.62 ± 11.69|
|No||63||40||23||37||26||31.30 ± 12.97|
|Yes||42||5||37||4||38||52.19 ± 6.42|
Expression of bFGF mRNA in Normal and Cancerous Gastric Tissues
Expression of bFGF mRNA in normal and cancerous gastric tissue showed very similar pattern of staining as seen in that of HGF (Table 1). Expression of bFGF mRNA (Fig. 3) was seen in 61 of 105 cases (64%), whereas only 2 of 20 normal or noncancerous tissues had weakly positive signals (10%, Fig. 4). The difference of the expression is also statistically significant (χ2 = 17.501, P < 0.01).
HGF and bFGF mRNA Expression in the Advanced Gastric Cancers
HGF and bFGF mRNA expression and their relationship with the tumor stage, presence or absence of lymphovascular or distant organ invasion were further analyzed. Strong signals of both HGF and bFGF mRNA expressions were detected in the cancer cells invading into the muscular layer, gastric serosa, and greater omentum. The rates of both HGF and bFGF mRNA expressions were higher in stage T3–T4 tissues compared with T1–T2 tissues (χ2 = 16.434, P < 0.01 and χ2 = 31.389, P < 0.001, respectively). The expression rates of HGF and bFGF mRNA were also higher in the tissues with vascular invasions (χ2 = 8.401, P < 0.01 and χ2 = 37.439, P < 0.01, respectively) or lymphatic invasions (χ2 = 8.575, P < 0.01 and χ2 = 23.129, P < 0.01, respectively) than with the expression in the tissues without such invasions. The higher rates of HGF and bFGF mRNA were also observed in the tissues with liver and peritoneal metastasis (χ2 = 27.384, P < 0.01 and χ2 = 25.636, P < 0.001, respectively). In addition, bFGF mRNA expression was also higher in the tissues with invasive growth pattern compared with the tissues with expansive growth (χ2 = 10.994, P < 0.01). However, the association was not observed when it comes to the cancer growth pattern and the degree of differentiation of gastric cancer for both HGF and bFGF (P > 0.05). Cox multivariate analysis showed that HGF and bFGF mRNA expression can be used as independent prognostic markers of gastric cancer (P < 0.05) (Table 2).
|Clinicopathological parameters||Hazard ratio||95% confidence interval||P-value|
|Lymph node metastasis||3.023||1.716–5.325||0.000|
MVD in the Advanced Gastric Cancers
The relationship between MVD and other pathological parameters of gastric cancer progress was also investigated. In both control and cancer tissues, VECs were stained immunohistochemically with anti-CD34. Inside the tumor tissues or around the cancer nests, there were abundant capillary channels that were stained positive with anti-CD34 and often showed comma-shaped, tubular, or trabecular-like patterns (Figs. 5, 6). The MVD values in tissues with invasive growth were higher compared with the tissues with expansive growth (χ2 = 10.105, P < 0.01). Similarly, the values in stage T3–T4 tissues were higher than those in stage T1–T2 tissues (χ2 = 5.961, P < 0.01) and higher in the tissues with vascular invasion compared with the tissues without (χ2=7.394, P < 0.01). The values were also higher in the tissues with lymphatic metastasis compared with the tissues without (χ2 = 3.819, P < 0.05) and higher in tissues with liver and peritoneal metastasis compared with the tissues without (χ2 = 10.578, P < 0.01). However, there was no statistically significant difference in MVD when well or moderately differentiated cancers were compared with those undifferentiated cancers (χ2 = 0.965, P > 0.05).
The Association of HGF and bFGF mRNA Expression With MVD
The MVD value in HGF mRNA-expressing gastric cancer tissue (44.30 ± 13.31/0.72 mm2) was higher compared with that in tissues that did not express HGF (32.95 ± 13.54/0.72 mm2; P < 0.01) (Table 3). In addition, the MVD value in bFGF mRNA-expressing cancer tissue (46.09 ± 11.52/0.72 mm2) was also higher compared with that in tissues that did not express bFGF (29.41 ± 12.47/0.72 mm2) (P < 0.05). A direct correlation was found when a comparison was made by Pearson's correlation analysis among expression of HGF mRNA and bFGF mRNA and MVD (HGF mRNA vs. bFGF mRNA, P < 0.05; HGF mRNA vs. MVD, P < 0.01; bFGF mRNA vs. MVD, P < 0.01).
|Group||N||MVD (N/0.72 mm2)||t||P|
|–ve||45||32.95 ± 13.54|
|+ve||60||44.30 ± 13.31|
|–ve||41||29.41 ± 12.47|
|+ve||64||46.09 ± 11.52|
Prognostic Roles of mRNA Expressions of HGF and bFGF as Well as MVD
The utilities of HGF and bFGF mRNA expression as well as MVD values for assessing the clinical prognosis of the gastric cancer were further analyzed. The mean survival times (months) of the patients with negative and positive expression of HGF mRNA were 113.33 ± 7.03 and 33.06 ± 3.57 (P < 0.01), respectively, and the corresponding 5-year survival rate was 83.57% (38/45) and 16.21% (10/64) (P < 0.01), respectively (Fig. 7A and Table 4). These data suggested that positive expression of HGF mRNA in the cancerous tissue may be a poor prognostic factor.
|Group||N||Mean survival time (months)||5-year survival rate (%)||P|
|–||45||113.33 ± 7.03||83.57|
|+||60||43.43 ± 6.04||19.89|
|–||41||117.55 ± 6.73||90.00|
|+||64||34.02 ± 3.60||36.90|
|<39.5||67||124.20 ± 5.32||95.70 (44/46)|
|≥39.5||38||35.17 ± 4.85||23.70 (14/59)|
Similar findings were also seen when relationship between bFGF mRNA expression and prognosis was analyzed. The mean survival times (months) of the patients with negative and positive expression of bFGF mRNA were 118.04 ± 6.52 and 33.06 ± 3.57, respectively, and the corresponding 5-year survival rates were 87.06% (38/41) and 16.21% (10/64) (P < 0.01), respectively (Table 4 and Fig. 7B). These data also suggested that positive expression of bFGF mRNA in the cancerous tissues may be another independent poor prognostic factor.
Similar findings were also demonstrated when relationship between MVD values and survival rate was analyzed. The mean survival times (months) of the patients with MVD values < 39.5/0.72 mm2 and ≥39.5/0.72 mm2 were 124.20 ± 5.32 and 35.17 ± 4.85, respectively, and the corresponding 5-year survival rates were 95.70% (44/46) and 23.70% (14/59) (P < 0.01), respectively (Fig. 7C). The data clearly suggested that an MVD value ≥39.5/0.72 mm2 can also independently predict a poor prognosis in patients with gastric cancer.
Gastric cancer is one of the most frequently occurring cancers. In 2002, the worldwide incidence of gastric cancer was 22/100,000 in males and 10.4/100,000 in females. The mortality rate of gastric cancer in males is 16.3/100,000 and in females is 7.9/100,000. There is a high incidence of gastric cancer in China. In 2005, the incidence of gastric cancer in China was 37.1/100,000 in males and 17.4/100,000 in females. The annual incidence of new gastric cancer cases in China is 400,000, and the annual patient death rate was up to 300,000. Gastric cancer is the third most common cancer in China (Yang, 2006). Currently, the only effective therapy for gastric cancer is surgical operation. Even so, the 5-year survival rate of gastric cancer is still lower than 40%, owing to relapse and metastasis. It is expected that through understanding the molecular mechanisms underlying the invasion and metastasis of gastric cancer, improved clinical therapies can be derived to raise patient survival rates. To date, only a few prognostic parameters are available for patients with gastric cancer. In this study, we investigated HGF mRNA and bFGF mRNA expression, in combination with pathological and follow-up data to determine whether HGF and bFGF expression can be used as prognostic parameters for patients with gastric cancer.
HGF is mainly generated by mesenchymal cells, as a polypeptide growth factor. It has multiple functions, including the stimulation of mitotic division in epithelia and endotheliocytes (Rosen et al., 1994; Rasola et al., 2007) and promoting morphogenesis of kidney tubules and regeneration of blood vessel endothelium (Boros and Miller, 1995). Recent studies have shown that HGF is also an endotheliocyte-specific growth factor, which can promote recovery of injured endotheliocytes, promote cell migration, and inhibit cell apoptosis (Takahashi et al., 2005). Nakamura and Niwa (2005) found that the stimulating effect of HGF on endotheliocyte proliferation was stronger than that of bFGF or VEGF. HGF, secreted from VECs in an autocrine manner, activates a phosphotyrosine of the c-Met receptor, which subsequently binds to specific signal proteins, leading to cell division, proliferation, migration, and forming a lumen-like structure in VECs (Lee et al., 2008). Meanwhile, HGF can indirectly enhance angiogenesis by inducing the expression of VEGF and bFGF (Wojta et al., 1999), and an in vivo study showed that HGF can induce neoangiogenesis in mouse cornea and hypodermic tissues (Yanai et al., 2006).
Our results show that most normal gastric mucosa did not express HGF mRNA. In contrast, the rate of HGF mRNA expression in gastric cancer tissues was up to 57.1%, and it increased with the depth of cancer invasion. The expression level of HGF mRNA was raised significantly in gastric cancer patients with metastasis of lymph gland, liver, and peritoneum. Moreover, patients with high levels of HGF mRNA expression had increased MVD, and the 5-year survival rate of patients expressing HGF mRNA was significantly lower compared with that of patients that did not express HGF mRNA. These results indicate that HGF may have significant effects on enhancing the invasive potential of cancer cells and of promoting angiogenesis in tumors; HGF has previously been shown to be closely associated with the growth and metastasis of cancer (Hu et al., 2007; Brychtova et al., 2008). Previous studies have demonstrated that HGF mRNA was overexpressed in cancers of the lung (Korobko et al., 2007), breast (Sheen-Chen et al., 2005), liver (Lasagna et al., 2006), pancreas (Kitajima et al., 2008), colon (Li and Shan, 2005), and thyroid (Vesely et al., 2004). Zhang et al. (2007) studied 60 cases of gastric cancer tissues and 20 cases of nontumorous tissues using immunohistochemical methods and demonstrated that the expression level of HGF increased gradually from nontumorous gastric mucous membrane to the primary focus of gastric cancer and was then further increased in lymphatic metastases. This showed that the expression of HGF was associated with the incidence of gastric cancer and lymph node metastasis. We suggest that HGF mRNA expression can be used clinically as a prognostic parameter for gastric cancer patients.
bFGF is an important angiogenesis factor that plays a role in promoting cell division and proliferation and enhancing angiogenesis through binding to the fibroblast growth factor receptors. It has a close association with cancer, particularly solid tumor cancers (Bremnes et al., 2006). Cancer cells can produce bFGF by themselves and they can also induce VECs to produce bFGF, therefore, bFGF can be produced via autocrine and paracrine secretion to promote the division of endotheliocytes, resulting in enhanced angiogenesis. From our observations of bFGF expression in gastric cancer tissues, very low or no expression of bFGF was seen in normal gastric mucous membranes, but the bFGF expression rate was up to 64% in gastric cancer tissues. Furthermore, higher expression of bFGF was associated with invasive growth, stage T3–T4, vessel invasion, lymph gland metastasis, and liver and peritoneal metastasis. In addition, higher expression of bFGF was associated with higher MVD, indicating that expression of bFGF was closely associated with angiogenesis. In advanced cancers, overexpression of bFGF favors angiogenesis, speeding up invasion and metastasis. Our study showed that patients with high bFGF expression had a low 5-year survival rate, indicating that bFGF can also be used as a prognostic indicator for patients with gastric cancer.
Our study showed that both HGF mRNA and bFGF mRNA were expressed in gastric cancer tissues, and that their expression was significantly correlated with cancer progression. In addition, both HGF mRNA and bFGF mRNA levels correlated with MDV values. Therefore, we postulate that both HGF mRNA and bFGF mRNA may work synergistically to promote angiogenesis in gastric cancer. This would lead to the proliferation and metastasis of cancer cells, that is, gastric cancer tissues can produce plentiful HGF to stimulate the proliferation of endotheliocytes, while cancer tissues can induce the expression of bFGF. Both HGF and bFGF then cooperate to stimulate the angiogenesis of capillaries (Tomita et al., 2003). In conclusion, on the basis of the results of this study, we suggest that the expression levels of HGF mRNA and bFGF mRNA can be used as prognostic indicators for gastric cancer in clinical settings.