The aim of the present study is to clarify the critical roles of vasohibin in cervical carcinomas. We investigated the expression ratios of vasohibin and vascular endothelial growth factor (VEGF) receptor-2 on endothelium and microvessel density, lymphatic vessel density (LVD) by immunohistochemistry. Sixty-one squamous cell carcinoma (SCC), 18 mucinous adenocarcinoma (Adenocarcinoma), 38 carcinoma in situ (CIS), and 35 normal cervical epithelium were collected. We investigated the expression of vasohibin and compared it with the expression of VEGF receptor-2 (VEGFR-2, KDR/flk-1), and CD34 in the stromal endothelium. Expression of VEGF was counted using the histological score (H score). D2-40 was used as a marker for lymphatic endothelial cells to investigate LVD. The microvessel density of the normal cervical epithelium was significantly lower than that of CIS, SCC, and Adenocarcinoma (P < 0.05). The expression ratio of vasohibin in the normal cervical epithelium was significantly lower than that of SCC and Adenocarcinoma (P < 0.05). The expression ratio of VEGFR-2 of the normal cervical epithelium was significantly lower than that of SCC and Adenocarcinoma (P < 0.05). The LVD of the normal cervical epithelium was significantly lower than that of CIS, SCC, and Adenocarcinoma (P < 0.05). For normal cervical epithelium, CIS, and SCC, there was a moderate correlation between the expression percentage of vasohibin and the expression percentage of VEGFR-2 (P < 0.05, r2 = 0.3018). This is the first study to elucidate the correlation between the expression of vasohibin in the stromal endothelial cells and the expression of VEGFR-2 in human cervical carcinomas. (Cancer Sci 2011; 102: 446–451)
Cervical carcinoma is a common gynecologic malignancy in women worldwide. In 2008 in the USA, an estimated 11 070 new cases of invasive cervical cancer were diagnosed and 3870 cancer-related deaths were expected to occur; this represents approximately 1% of cancer deaths in women.1 The morbidity of cervical cancer, especially in women younger than 40 years old, is rapidly increasing and is the worst in Japan.2 It is well recognized that angiogenesis (the process of forming new vessels) is required for tumor growth and enables the hematogenous spread of tumor cells throughout the cervical cancer. Several studies document the association between the MVD and/or the extent of endothelial proliferation, the pre-cancer and tumor stage, and the invasive disease of the cervical cancer.(3–6) The expression of various angiogenesis stimulators, such as VEGFs, angiopoietins, and thymidine phosphorylase, has been reported in cervical cancer and dysplasia.(7,8) Several published reports similarly report stromal LVD in cervical neoplasia.(9–11) Angiogenesis is determined by the local balance between angiogenic stimulators and inhibitors. However, the significance of endogenous angiogenesis inhibitors in cervical cancer is poorly documented.
We isolated a novel angiogenesis inhibitor, vasohibin, which is specifically expressed in ECs. Its basal expression in quiescent ECs is low; however, its production is induced by angiogenic stimuli such as VEGF-A and fibroblast growth factor-2, and it inhibits angiogenesis in an autocrine manner.(12,13) We therefore propose that vasohibin acts as a negative feedback regulator to inhibit angiogenesis. VEGF-A is the most important factor for angiogenesis and most VEGF-A-mediated signals for angiogenesis are transduced through VEGF receptor-2 (VEGFR-2).(14) We observed that the VEGF-A-mediated induction of vasohibin was preferentially mediated through the VEGFR-2 signaling pathway.(15) In human cancer tissues, we have already reported a significant correlation between the expression ratio of vasohibin and VEGFR-2 in the vascular endothelial cells of endometrial cancer and breast cancer.(16–18)
In the present study, we aimed to elucidate the significance of vasohibin in human cervical cancer and in the normal cervical squamous epithelium. We investigated the correlations between a tumor’s biology (including vasohibin) and the clinical prognostic factors for cervical cancer (such as MVD, LVD, and the expression rate of vasohibin and VEGFR-2 in the endothelial cells). The histological type, lymph node metastasis, vessel involvement, and staging were the most important prognostic factors for cervical cancer. Our analysis revealed a significantly positive correlation between vasohibin and VEGFR-2 in cervical cancer. This is the first study to profile the expression of vasohibin as a negative feedback regulator of angiogenesis in human cervical cancer.
Materials and Methods
Tissue specimens and clinical data. There were 61 SCC, 18 mucinous adenocarcinoma (Adenocarcinoma), 38 CIS, and 35 normal cervical epithelium. Of the 61 SCC: 11 were stage IA (lymph node metastasis 0/11, vessel involvement 0/11, parametrium invasion 0/11); 29 were stage IB (lymph node metastasis 6/29, vessel involvement 11/29, parametrium invasion 0/29); 12 were stage IIA (lymph node metastasis 3/12, vessel involvement 7/12, parametrium invasion 0/12); and nine were stage IIB (lymph node metastasis 4/9, vessel involvement 6/9, parametrium invasion 8/9). Of the 18 mucinous adenocarcinomas: three were stage IA (lymph node metastasis 0/3, vessel involvement 1/3, parametrium invasion 0/3); 11 were stage IB (lymph node metastasis 1/11, vessel involvement 0/11, parametrium invasion 0/11); and four were stage IIB (lymph node metastasis 3/4, vessel involvement 4/4, parametrium invasion 3/4). We investigated the expression of vasohibin, and compared it with the expression of VEGFR-2 (KDR/flk-1)and CD34 (as a marker for vascular endothelial cells) in the stromal vascular endothelium. We used VEGF to indicate the presence of carcinoma cells and normal squamous epithelium. To investigate MVD, CD34 was used as a marker for vascular endothelial cells. To investigate LVD, D2-40 was used as a marker for lymphatic endothelial cells. Immunohistochemical assessment was classified as negative or positive, based on the staining intensity.
Tissue specimens were retrieved from the surgical pathology files of Tohoku University Hospital (Sendai, Japan). The Ethics Committee at Tohoku University School of Medicine (Sendai, Japan) approved the protocol for this study.
The average age was 43.1 years in patients with SCC, 47.1 years old in patients with adenocarcinomas; and 41.0 years old in patients with CIS. Each patient provided written, informed consent before her surgery. None of the patients who were examined had received irradiation, hormonal therapy, or chemotherapy prior to surgery. The clinicopathological findings of the patients (including age, histology, stage, grade, and preoperative therapy) were retrieved by an extensive review of the charts. The standard primary treatment for cervical carcinoma at the Tohoku University Hospital was surgery, which consisted of abdominal radical hysterectomy and pelvic lymphadenectomy. The lesions were classified according to the Histological Typing of Female Genital Tract Tumors by the World Health Organization and staged in accordance with the International Federation of Gynecology and Obstetrics system.(19,20) Patients with subtypes other than CIS, SCC, or mucinous adenocarcinoma, or patients who had second primary carcinoma, were excluded from this series.
All specimens were routinely processed (i.e., 10% formalin and fixed for 24–48 h), embedded in paraffin, and thin sectioned (3 μm).
Immunohistochemical staining and the scoring of immunoreactivity. We carried out immunohistochemical staining for vasohibin, VEGFR-2, CD34 (as a marker for vascular endothelial cells), and D2-40 (as a lymphatic vessel marker). The presence of VEGF-A was assessed in normal squamous epithelium and in carcinoma cells. Paraffin-embedded tissue sections from human endometrial cancers were deparaffinized, rehydrated, and incubated with 3% H2O2 for 10 min to block endogenous peroxidase activity. The sections were incubated for 30 min at room temperature in a blocking solution of 10% goat serum (Nichirei Biosciences, Tokyo, Japan). They were then stained for 12 h at 4°C with primary antibodies, followed by staining for 30 min at room temperature with secondary antibodies. The primary antibodies were all mouse mAbs and were used as follows: 2 μg/mL anti-human vasohibin mAb, anti-VEGFR-2 (Santa Cruz Biotechnology, Santa Cruz, CA, USA) diluted at 1:100; anti-CD34 (Dako, Copenhagen, Denmark) diluted at 1:200; anti-D2-40 (Dako) diluted at 1:100; and anti-VEGF-A (Lab Vision, Fremont, CA, USA) diluted at 1:100. We previously described a mouse mAb against a synthetic peptide corresponding to the 286–299 amino acid sequence of vasohibin.(15) The positive control slide for CD34 antigen was prepared from paraffin-fixed breast cancer tissue that contained a high microvessel density. Nuclei were counterstained with hematoxylin.
Three investigators (K.A., T.E., and K.Y) independently evaluated the immunohistochemical staining of the entire group of tissue sections. They were blinded to the clinical course of the patients. The average number counted by the three investigators was used for subsequent analysis.
We carefully examined the areas where cancer cells had come in contact with or had invaded the stroma. After first scanning the immunostained section at low magnification, microvessels having CD34-positive signals were counted. The areas with the greatest number of distinctly highlighted microvessels were selected. Any cell clusters with CD34-positive signals were regarded as a single countable microvessel, regardless of whether the lumen was visible. Unstained lumina were considered artifacts, even if they contained blood or tumor cells. Microvessel density was assessed by light microscopy in areas of invasive tumor containing the highest number of capillaries and small venules per area (i.e., neovascular hot spots), in accordance with an original method.(21) Lymphatic vessel density was investigated by D2-40-positive signals, using the same procedure described as above.
Immunostaining for vasohibin and VEGFR-2 was evaluated in the serial thin sections. Positive immunoreactive signals for vasohibin and VEGFR-2 in CD34-positive microvessels were counted; from this the positive ratios of vasohibin and VEGFR-2 in microvessels were calculated. Using an established antibody, immunohistochemistry detected the protein expression of a selected angiogenic factor (i.e., VEGF-A). For this marker, the intensity of cytoplasmic staining and the proportion of positive tumor cells were recorded. A staining index (with values 0–9) was calculated as the product of the staining intensity (0–3) and the area of positive staining (1, <10%; 2, 10–50%; 3, >50%).(22)
Statistical analysis. Statistical analyses such as the Student’s t-test and the Pearson’s correlation coefficient test were carried out using StatView (version 4.5, Abacus Concepts Inc, Piscataway, NJ, USA). A P-value of <0.05 was considered significant.
Figure 1 shows that the cells stained positively for vasohibin, VEGFR-2, CD34, and D2-40 on the normal cervix, CIS, SCC, and Adenocarcinoma. There were no significant differences between clinical factor (lymph node metastasis, vessel involvement, parametrium invasion) and MVD, LVD, the vasohibin expression ratio in the microvessels, the VEGFR-2 expression ratio in the microvessels, and the histology score of VEGF-A expression.
Microvessel density. CD34-positive MVD (number of vessels per mm2) were: 34.7 ± 2.5 in normal cervical epithelium; 46.1 ± 2.5 in CIS; 45.7 ± 3.5 in SCC; and 73.85 ± 8.4 in Adenocarcinoma (Fig. 2A). The MVD of Adenocarcinoma was significantly higher than that of the normal cervical epithelium, CIS and SCC. The MVD of normal cervical epithelium was significantly lower than that of CIS and SCC.
Lymphatic vessel density. D2-40-positive LVD (number of vessels per mm2) was: 3.7 ± 0.8 in the normal cervical epithelium; 7.3 ± 0.8 in CIS; 17.6 ± 1.8 in SCC; and 34.6 ± 6.5 in Adenocarcinoma (Fig. 2B). The LVD of Adenocarcinoma was significantly higher than that of the normal cervical epithelium, CIS, and SCC. The LVD of SCC was significantly higher than that of the normal cervical epithelium and CIS. The LVD of CIS was significantly higher than that of the normal cervical epithelium.
Vasohibin expression ratio in microvessels. The positive ratio of vasohibin in microvessels was 53.6 ± 4.2% in normal cervical epithelium; 59.1 ± 4.6% in CIS; 71.5 ± 3.2% in SCC; and 79.3 ± 4.0% in Adenocarcinoma (Fig. 2C). The vasohibin-immunopositivity of the Adenocarcinoma microvessels was significantly higher than that of the normal cervical epithelium, CIS, and SCC. The vasohibin-immunopositivity of the SCC microvessels was significantly higher than that of the normal cervical epithelium and CIS.
VEGFR-2 expression in microvessels. The positive ratio of VEGFR-2 in microvessels was 42.6 ± 3.8% in the normal cervical epithelium; 42.1 ± 4.3% in CIS; 52.6 ± 2.6% in SCC; and 58.2 ± 4.7% in adenocarcinoma (Fig. 2D). The VEGFR-2 immunopositivity in Adenocarcinoma microvessels was significantly higher than that of the normal cervical epithelium, CIS, and SCC. The VEGFR-2 immunopositivity of SCC microvessels was significantly higher than that of the normal cervical epithelium and CIS.
Histology score of VEGF-A expression. VEGF-A expression was present in the cytoplasm of epithelial cells. The VEGF-A positive ratio of cytoplasmic staining intensity was: 37.9 ± 7.3 in the normal squamous epithelium; 58.6 ± 9.1 in CIS; 88.5 ± 8.7 in SCC; and 52.8 ± 16.3 in Adenocarcinoma (Fig. 2E). The VEGF-A positive ratio of cytoplasmic staining intensity in SCC was significantly higher than that of the normal squamous epithelium, CIS, and Adenocarcinoma. The VEGF-A positive ratio of cytoplasmic staining intensity in CIS was significantly higher than that of the normal squamous epithelium.
Correlation between vasohibin and VEGFR-2 positive ratios in microvessels. We emphatically analyzed through normal cervix–CIS–SCC as the multi-step carcinogenesis model. A moderate positive correlation existed between vasohibin and VEGFR-2 positive ratios in the microvessels of the normal cervical epithelium, CIS, and SCC (P < 0.05, r2 = 0.301) (Fig. 3).
Using immunohistochemistry, we examined the vascular density of cervical cancer and compared it with that of the normal cervix. Our results with MVD and LVD in CIS and SCC were almost identical with some previous reports.(3–5,23) Therefore, this proved the validity of our investigation. The major common perception of cervical cancer is that a CIS integrates the human papillomavirus and progresses to SCC in several years. Some reports indicate that the MVDs of CIS and SCC are greater than the MVD of the normal cervix. Therefore, we thought that it was important in our investigation to consider the vessel number in the normal cervix and to compare it with that of CIS and SCC. In accordance with our expectation, the MVD and LVD in the normal cervix were significantly lower than in CIS and SCC. These results suggested that hypervascularity is an important phenomenon of the progression of cervical squamous neoplasia. One published report showed that the VEGF intensity of SCC of the cervix increases in correlation with the clinical stage;(24) however, in the study there was no significance between the clinical features and the MVD, LVD, vasohibin expression ratio, VEGFR-2 expression ratio, and histological score (H score) of VEGF-A. In our study, we only found significant histological differences in the normal cervix, CIS, SCC, and Adenocarcinoma. Of course, it is important to investigate more cases of cervical neoplasia, and we suggest that peritumoral stromal angiogenesis is strongly ruled by the histological type of tumor. The VEGF-A positive ratio of cytoplasmic staining intensity in the normal cervical epithelium was significantly lower than that in CIS and in SCC. This result clearly showed that the cytoplasmic expression level of VEGF-A in squamous epithelium cells and squamous neoplasia was increased in accordance with the progression.
It is generally accepted that, at the same clinical stage, Adenocarcinoma has a poorer prognosis than SCC.(25–29) However, angiogenesis in Adenocarcinoma has been analyzed relatively little because of the minor frequency of Adenocarcinoma. In this study, we compared normal cervix, CIS, SCC, and Adenocarcinoma. We clearly showed that MVD and LVD were significantly higher in Adenocarcinoma than in SCC. This finding might contribute to understanding the poorer prognosis of adenocarcinoma of the cervix. To our regret, in this study we did not establish the significance between LVD and lymph node metastasis, because of insufficient number of cases. To clarify the issue, the same investigation should be repeated in a higher number of cases. However, the H score of VEGF-A of Adenocarcinoma was lower than that of SCC. We believe that it was difficult to detect VEGF-A in the cytoplasm of Adenocarcinoma because the cytoplasm was rich in mucus. In an in vitro study, there are many methods of measuring proteins in the cytoplasm, so we must examine measuring proteins in paraffin-embedded tissues except immunohistochemistry. Similar to MVD, LVD in this study was significantly increased in SCC and Adenocarcinoma. Some studies have reported the presence of peritumoral lymphatic vessels in cervical cancers and show that a high LVD is strongly associated with aggressive cervical cancer features such as lymph node metastasis, vessel involvement, parametrium invasion, and poor prognosis.(30,31) Although we expected that the LVD would be correlated with lymph node metastasis, there was no significant correlation between the two.
We examined the expression of vasohibin, as described in our previous report.(16) Vasohibin is induced in ECs by stimulation with VEGF.(12,15) Indeed, the expression of vasohibin in tumor vessels was shown to be correlated with that of VEGF in human breast cancer and human non-small cell lung carcinoma.(17,18) However, because of the difficulty of immunostaining VEGF-A in adenocarcinoma, we did not show such correlation in this study. VEGFR-2 receptor transduces most of the angiogenesis-related signals in ECs. The VEGF/VEGFR-2 signaling pathway is also important for the induction of vasohibin in Ecs.(15) We previously revealed that the VEGF-A-mediated induction of vasohibin was preferentially mediated through the VEGFR-2 signaling pathway.(15) In our results, the expression ratio of vasohibin in vascular endothelial cells increased as the tissue progressed from a normal cervix to SCC to Adenocarcinoma. In addition, the expression ratio of VEGFR-2 in vascular endothelial cells increased as the tissue progressed from a normal cervix to SCC to Adenocarcinoma. This correlation of vasohibin and VEGFR-2 is also evident in endometrial adenocarcinoma(16) and human breast cancer.(17) Thus, the pathological studies from human cancer specimens, including our present one, support the theory on the negative feedback role of vasobihin in cancers. Lu et al.(32) recently reported that the increased expression of Zeste homolog 2 (EZH2) predicts poor prognosis of ovarian cancers. This poor prognosis was explained by EZH2-mediated silencing of vasohibin expression in tumor vessels. As the level of vasohibin expression is considerably variable (Fig. 3), it would be interesting to see whether lower expression of vasohibin in tumor vessels predicts poor prognosis.
It has been established that there is a significant relationship between cervical papillomavirus infection and cervical cancer. It is also clear that CIS progresses to SCC in several years. In our study, a moderate positive correlation existed between the positive ratios of vasohibin and VEGFR-2 expression in the normal cervix, CIS, and SCC. As the expression of vasohibin increased in correlation with the expression of VEGFR-2, vasohibin could be an important biomarker of angiogenesis in the normal cervix and in cervical cancer. However, further investigations are required to clarify the precise role of vasohibin in regulating anti-angiogenic activity in the normal cervix and in cervical disorders.
We previously reported the function of vasohibin as a novel angiogenesis inhibitor.(12–15,18,33) The profiling of vasohibin expression in human cancer tissues is being accumulated.(18,33,34) In the future, vasohibin could potentially have medical application as a novel angiogenesis inhibitor.
The authors thank Miss Keiko Abe and Miss Emi Endo for preparing material. This work was supported by a Grant-in-Aid from the Kurokawa Cancer Research Foundation and a Grant-in-Aid for Scientific Research on Priority Areas from the Ministry of Education, Science and Culture, Japan and a Grant-in-Aid from the Ministry of Health, Labor and Welfare, Japan, and the 21st Century COE Program Special Research Grant (Tohoku University) from the Ministry of Education Science, Sports and Culture, Japan.
The authors have no conflict of interest to declare.