Research ArticleResearch Article

Tissue Microarray Study of Vasculogenic Mimicry in Bi-directional Differentiated Malignant Tumors

Xishan Hao, Baocun Sun, Shiwu Zhang and Xiulan Zhao
Chinese Journal of Clinical Oncology February 2004, 1 (1) 4-9;

Abstract

OBJECTIVE To determine if vasculogenic mimicry (VM) exists in bi-directional differentiated malignant tumors.

METHODS The blood supply models for bi-directional differentiated tumors were studied with immunohistochemical and PAS double–staining techniques. New sections were made from 158 paraffin–embedded bi–directional malignant-tumor samples, including melanoma (high malignancy n= 30, low malignancy n=30); synoviosarcoma(SS) (high malignancy n=26, low malignancy n=13); acinar rhabdomyosarcoma (AR) (high malignancy n=16, low malignancy n=13); malignant mesothelioma (MM) (n=26), and epithelioid sarcoma (ES)(n=4). Tissue microarrays were made. The representative points in the paraffin sections were labeled and two tissue microarrays were made, one included 60 cases of melanoma, and the other included the other tumors. Immunohistochemical staining of the platelet–endothelial cell adhesive molecule(CD31 antigen) and periodic acid Schiff(PAS) staining were conducted. The areas were calculated of vessel-like channels consisting of CD31 antigen–positive tumor cells and of PAS positive materials. The VM was studied using the data obtained.

RESULTS Some of these bi–directional tumor cells secreted PAS-positive materials and CD31 positive materials. The walls of the VM consisted of PAS-positive materials lined with CD31 negative tumor cells with red blood cells inside the channel, whereas the walls of the epithelium–dependent vessels were comprised of CD31 positive materials. The positive areas of CD31 were significantly less than that of PAS (P<0.01). The number of cases with VM in highly malignant tumors was greater than that found in the lowly malignant tumors.

CONCLUSIONS Bi–directional differentiated malignant tumor cells have the ability to auto–transform and might interact with the extracellular matrix to form a vessel channel system which mimics blood vessels for transporting blood. That process is called VM. Results in this study show that bi-directional differentiated tumors develop a nutritional supply by VM to satisfy the demand for the rapid growth, and thus acquire higher invasive ability and malignancy.

KEYWORDS:

keywords

Currently a topic of great world-wide interest is that of blood supplies to tumors as related to angiogensis [1]. Angiogenesis, the process by which a tumor’s blood supply is increased by forming vessels which enter a tumor tissue, has been a putative model, but some researchers believe that angiogensis may not be the only mechanism by which various tumors acquire a blood supply. A number of melanoma with high invasive and metastatic ability can form vessel–like channels called VM[2–4] The characteristics of VM are: no vascular endothelial lining on the wall of the channel, and tumor cells mimic angiogensis to produce a line of tumor cells which form non–en dothelial channels through which blood flows. There is enwrapped around the channel a new thick layer of basal membrane of PAS–positive materials. Current studies on the molecular mechanism concerning VM are still in an initial stage, however considerable evidence has shown that this manner of tumor cells obtaining a blood supply is unique and related to the biological behavior of tumor invasion and metastasis. By analyzing and comparing the morphological and biological character of melanoma, it was presumed that VM may exist in some bi–directional differentiated malignant tumors. The double–staining of CD31 and PAS used in this study showed that there were many negative CD31 lined cells with a PAS positive vessel–like pattern in highly malignant and bi–directional differentiated tumors. Further study showed that the VM, which is related to tumor malignancy, also existed in bi-directional differentiated malignant tumors. In addition we discovered that tumor cells also expressed the CD31 antigen and PAS positive materials, which may be the fundamental substances by which tumor cells participate in constituting VM channels through matrix remodeling.

MATERIALS AND METHODS

Materials

Altogether 158 cases of bi–directional differentiated malignant tumors, including melanoma 60 cases, SS 39 cases, AR 29 cases, MM 26 cases, and ES 4 cases, were chosen from tissue samples embedded in paraffin which had been deposited in the Tianjin Tumor Hospital from 1980–2000. For each patient who died of a bi–directional differentiated malignant tumor entire follow-up data were available.

Methods

The 158 paraffin samples of bi–directional differentiated malignant tumors were sectioned for review and diagnosis. In order to avoid orientation errors and to facilitate the study, we labeled the representative points in the paraffin sections relating to the H&E sections, after which two new tissue microarrays were made from the 158 paraffin samples. One new chip was made from 60 melanomas, the lattice number was 248, including 8 control lattices. The other SS (n=39), AR (n=29), MM(n=26), and ES(n=4) were used to make another microarray, in which the lattice number was 400, including 8 control lattices. Each sample was made of 4 sampling points to avoid point leaking and section sliding in the process of making points and immunohistochemical staining. CD31 and PAS double–staining was conducted on the tissue microarrays. The silica slides were purchased from Zhongshan Company, cover glasses were soaked in 50ml of 75% ethanol containing a drop of concentrated hydrochloric acid.

Immunohistochemical and histochemical double–staining

Mouse monoclonal anti-CD31 (clone JC/70A) antibodies were used in this study at dilutions of : 1:400.

Four–micrometer sections from formalin–fixed, paraffin–embedded tissue were mounted on poly–L–lysine-coated slides. The slides were air-dried and the tissue deparaffinized. Endogenous peroxidase activity was blocked with 3% hydrogen peroxide in 50% methanol for 10 min at room temperature. The sections were rehydrated and washed with PBS and then pretreated with boiling in citrate buffer (0.01 M citric acid, pH 6.0) for 20 min in a microwave oven. After nonspecific binding sites were blocked by exposure to 2% normal goat serum in phosphate–buffered saline (PBS) (20 min at 37°C), the sections were incubated overnight at 4°C with mouse monoclonal anti–CD31. After this, the sections were rinsed with PBS, incubated with biotinylated goat anti-mouse IgG for 20 min at 37°C, incubated with 3,3’–diaminobenzidine chromogen for 10 min at room temperature, and washed with distilled water. Sections incubated with CD31 were deoxidized by exposure to 1% sodium periodate for 10 min. The sections were then rinsed with distilled water for 5 min and incubated with periodic acid–Schiff (PAS) for 15 min. Finally, all of the sections were counterstained with hematoxylin.

Counting methods

After double–staining counting was conducted using an OLMPUS CH-2 microscope with 16xl6D gridding magnified 400x. There were a total of 256 total cross points on the intersecting (Pt) grid. A cross point was counted just when the CD31 and PAS positive areas covered the cross point, whereas a cross point would not be counted when the CD31 and PAS positive areas were found outside of the cross points. CD31 and PAS counting must be conducted in the same viewing area. In the same tumor, counting of the Pc of CD31 and PAS must be conducted equally. Each side of the micro square(d) is the same and the test area is At=Pc*d2.

Statistical methods

A SPSS for Windows 10.0 statistical software package was used to conduct ranks sum test for dependent samples. A P<0.05 was considered significant.

RESULTS

PAS positive patterns in bi–directional differentiated tumors

Employing CD31 and PAS double staining of bi–directional differentiated malignant tumors, such as melanoma, SS, MM, ES and AR, PAS positive vessel–like patterns similar to those existing in VM of human uveal melanoma were found in this study [2]. Tumor cells which had a PAS pattern lined the inner vessel walls and were negative for CD31. Arrangement of the PAS–positive substances of the vessel–like pattern was linear, parallel, crossed, and rainbow–like (not completely closed), closed and network–like (Fig. 1). Furthermore, a more significant phenomenon could be seen: on one side of the papillary MM, the surfaces of the papillary tumor cells were covered with a layer of PAS positive materials, and a great number of random red blood cells could be seen between the two papillae. It appeared that the papillary tumor cells constructed the structure of the vessel wall, and the space between two papillae constituted a vascular lacuna. On the surface of the papillae no covered PAS positive materials and endothelium cells were seen (Fig. 2).

Fig. 1.
Fig. 1.

Various PAS patterns showing red blood cells; double–staining, ×200.

Fig. 2.
Fig. 2.

Papillary MM. Red blood cells located between papillae and on the surface of papillae covered by PAS-positive material; double-staining. ×200.

The tumor cellular expression of CD31 and PAS–positive materials

In the lattices of the bi-directional differentiated tumor tissue microarrays, besides the VM structure described by Maniotis et al [2], the cells of a bi-directional differentiated tumor can express the CD31 antigen and PAS positive materials. Microscopic examinations revealed that CD31 positive materials were located in the tumor cell cytoplasm. Whether it was a highly malignant or lowly malignant tumor, whether the tumor consisted of epithelioid cells or shuttle–like cells, the tumor cells all expressed CD31 positive materials (Fig. 3). PAS positive substances were distributed diffusively in the cytoplasm in the form of granules, a patch, even more as a circle of PAS–positive materials around a single tumor cell or several tumor cells. More typically seen was that CD31 and PAS materials were secreted at the same time by some giant tumor cells of an AR. Furthermore, we could see that orbicular, irregular or shuttle–like tumor cells adhered in a closed PAS–positive pattern in which red blood cells existed. In some microarray lattices, CD31–positive cells were scarcely visible, but only closed or network–like PAS positive patterns existed in the tumor tissue. Necrosis was not detected and infiltrative inflammatory cells were also not found around the VM (Fig. 4).

Fig. 3.
Fig. 3.

Tumor cells secreted CD31 positive materials; double–staining, ×200.

Fig. 4.
Fig. 4.

PAS positive materials around a single cell or several tumor cells without obvious necrosis; double-staining, ×200.

Other evidence of VM in melanomas

In the lattices of highly malignant melanomas, many shuttle cells whose shape was similar to vascular endothelium could be seen in a PAS–positive pattern, whereas CD31 staining was negative. These cells were identified to be melanoma cells by HMB45 staining.

The pattern area of CD31 and PAS-positive materials formed by bi-directional differentiated malignant tumors.

Through the lattice counting method, the area of patterns made by CD31 and PAS positive materials in bi–directional differentiated malignant tumors were calculated and statistical analysis conducted. The statistical results of CD31 and PAS positive areas are shown in Table 1. The difference between CD31 and PAS positive areas of bi-directional differentiated malignant tumors was statistically significant (P<0.01). The area of PAS positive materials was obviously more than that of CD31 positive materials (Table 1).

View this table:
Table 1.

Comparison of immunohistochemical staining with CD31 and PAS positive materials in bi–directional differentiated tumors

Based on the immunohistochemical staining results, atypia, the extent of necroses, the number of mitosis per visual field and prognosiss the 158 cases of bi-directional differentiated malignant tumors were divided into two groups: One was highly malignant; the other one was lowly malignant. Results of the statistical study showed that among 39 cases of highly malignant SS and AR, 16 cases had VM; while in 29 cases of lowly malignant SS and AR, VM could be seen in only 5 cases. The difference between the two has statistical significance.

DISCUSSION

In 1999, on the basis of a microcirculatory study of human uveal melanoma, Maniotis and Folberg et al.[2–5] found a completely different tumor angiogenesis pathway from the classical one, a new tumor blood supply system which was independent of endothelium and could transfer blood. This system was formed by melanoma cells mimicking vessel wall structure through self–transformation and the interaction with the extracellular matrix. In this way, the tumor can reconstruct a microcirculation and thereby obtain a supply of blood from the connecting host blood vessels [2,6,7]. The existence for VM was demonstrated by three–dimentional (3D) cell culture and confocal laser scanning microscopy. Its characteristic was: these vascular channels are composed of a basement membrane that stains positive with the PAS reagent in the absence of endothelial cells and fibroblasts. Immunohistochemical study showed that its inner wall was lined with tumor cells, but not endothelium. Red blood cells were discovered in these PAS-positive channels.

Our study found that VM existed not only in uveal melanoma of human eyes, but also in melanoma from other parts of the body. Melanoma has the ability of bi–directional differentiation. The tumor cells in these tumors possess features of both endothelial cells, (which line the vessels), and mesenchymal cells, which secrete the extracellular matrix proteins, such as PAS-positive material and CD31 antigen. From this view point, we propose that tumors having the ability of bi–directional differentiation might have VM. The object of this study was to determine if VM exists in the bi–directional differentiated malignant tumors, such as SS, MM, ES and AR, besides melanomas. Biochemically these tumor cells can produce epithelial cytoskeletal proteins, such as cytokeratin, and mesenchymal cytoskeletal proteins, such as vimentin. VM may thus reflect a defect in the cell differentiation process, so that tumors with VM channels are poorly differentiated or are dedifferentiated. This view is supported by the finding that the molecular profile of aggressive cutaneous and uveal melanoma cells in the presence of VM is similar to that of pluripotent, embryonic-like stem cells. It is also possible that SS, MS, and AR are dedifferentiated. This is suggested by the fact that sarcomas are derived from mesenchymal tissues, which can secret matrix proteins such as CD31 and PAS–positive material. The difference from the methods we used compared to foreign studies was that we conducted CD31 and PAS double–staining in the same sections. By doing this, one could readily determine whether it was a real blood vessel or VM. By this technique it is possible to show more directly that VM exists in bi-directional differentiated malignant tumor tissue.

Former studies have shown that the platelet–endothelial cell adhesion molecule (PECAM–1), that is the CD31 antigen, is a type of linking molecule in the endothelium which participates in many physiological processes in the body, and mediates interaction of many ligands. The CD31 antigen was generally distributed on endothelium-dependent vessels, and its immunohistochemical identification can be regarded as a marker vessel. Our study found that bi–directional differentiated tumor cells could express CD31 positive materials (64.1% of SS, 57.7% of MM, 75.0% of ES and 40.7% of AR) and PAS positive materials (64.1% of SS, 84.6% MM, of 75.0% of ES and 81.5% of AR). CD31 and PAS–positive materials are important elements which constitute vessel structure. CD31 and PAS positive materials secreted by tumor cells with VM may be stimulated by the need for more oxygen and nutrients to meet the requirements of rapid growth that takes place in highly malignant tumors. It may be difficult for tumor cells to satisfy their requirements if they only rely on epithelium–dependent vessels to transfer oxygen and nutrients. Based on these circumstances, tumor cells form a channel system which can transfer blood by self–transformation and generating extracellular matrix(CD31 and PAS positive materials) and thereby reconstruct a tumor microcirculation to gain an additional blood supply from connecting host blood vessels. CD31 is regarded as a kind of intercellular connective molecule. The reason why tumor cells could secrete CD31 could be explained theoretically: through self–transformation, mimicking endothelium and generating extracellular matrix, tumor cells could connect to each other and line PAS positive materials on the inner channel wall. Through consecutive sections, in the tissue microarrays examined in our study, some of the tumor tissue did not express any CD31 staining positive points, and the only thing that could be seen was a great amount of PAS positive materials constructed in VM. Its inner wall was lined with tumor cells and necrosis was not found.

CK8 and CK18 immunohistochemocal staining in the tissue microarrays served to define the malignant degree of melanoma. The results were very similar to that of Maniotis et al[2]. With regard to the other bi-directional differentiated tumors, the degree of malignancy could be divided on the basis of the tumor cells, atypia, the number of mitosis (more than 3 mitosis per visual field can be regarded that this tumor was highly malignant) and whether or not necrosis was present. The results demonstrated that the existence of VM in highly malignant tumors was very common; while in lowly malignant tumors, it was seldom seen. In the highly malignant bi–directional differentiated tumors, the distributed range of PAS positive vessel channels was much wider, the density was much higher, and necrotic foci were not found. The tumor tissues can acquire enough blood supply by VM in contact directly with normal blood circulation to nurture extensive metastases. Clinical follow–up data indicated that most of the patients with bi–directional differentiated malignant tumors with VM died of tumor cells metastases, especially metastases from vessels. It would be more conducive for the growth of tumor cells if they develop vascular metastases in various places that link VM with a normal vascular channel. The structure of the VM constructed directly by tumor cells without endothelium maybe related to a high level of vascular metastasis[8]. The follow-up data also indicate that the 3–year survival rate of SS and AR with VM was only 23.81% (among 21 cases, the survival time of 5 cases exceeded 3-years after operation). Therefore it can be seen that the rate of formation of VM is connected to the behavior of highly malignant tumor cells and prognosis. Beside bi–directional differentiated malignant tumors, Kobayashi et al [9]. proved that VM also existed in other highly malignant tumor tissue, such as liver cancer and prostate cancer. It was not clear whether the therapy aimed at endothelium was effective in reducing the growth of the tumors with VM.

The results of studies on the biochemical and molecular mechanisms concerning VM offer the possibility of opening new routes for tumor therapy [10]. By preventing tumor cells from secreting adhesion molecules, suppressing the synthesis of tumor–invasion-related proteases and inhibiting the synthesis and secretion of the extracellular matrix, development of this newly formed blood supply of tumor cells may be blocked. If these goals can be achieved, this group of highly invasive malignant tumors, which spread by vascular channel metastasis, may be cured[11–20].

Footnotes

  • This work was supported by National Nature Science Foundation (30070378).

  • Received February 8, 2004.
  • Accepted April 6, 2004.

REFERENCES

  1. ↵
    1. Foss AJE,
    2. Alexander RA,
    3. Jefferies LW, et al.
    Microvessel count predicts surival in uveal melanoma. Cancer Res, 1996;56:2900–2903.
    OpenUrlAbstract/FREE Full Text
  2. ↵
    1. Maniotis AJ,
    2. Folberg R,
    3. Hess A, et al.
    Vascular channel formation by human melanoma cells in vivo and vitro: VM. Am J Pathol, 1999;155: 739–752.
    OpenUrlCrossRefPubMed
    1. Folberg R,
    2. Pe’er J,
    3. Gruman LM, et al.
    The morphologic characteristics of tumor blood vessels as a marker of tumor progression in primary human uveal melanomas, a matched case-control study. Hum Pathol, 1992;23:1298–1305.
    OpenUrlCrossRefPubMed
  4. ↵
    1. Folberg R,
    2. Rummelt V,
    3. Parys–Van Ginderdeuren R, et al.
    The prognostic value of tumor blood vessel morphology on primary uveal melanoma. Opthalmology, 1993;100:1389–1398.
    OpenUrlCrossRefPubMed
  5. ↵
    1. Folberg R,
    2. Mehaffey M,
    3. Gardner LM, et al.
    The microcirculation of choroidal and ciliary body melanomas. Eye, 1997;11:227–238.
    OpenUrlPubMed
  6. ↵
    1. Folberg R,
    2. Augsburger JJ,
    3. Gamel JW, et al.
    Fine–needle aspirates of uveal melanomas and prognosis. Am J Opthalmology, 1985;100:645–657.
    OpenUrl
  7. ↵
    1. Rummelt V,
    2. Mehaffey MG,
    3. Ampbell RG, et al.
    Microcirculation architecture of metastases from prime ciliary body and choroidal melanomas. Am J Ophthalmal, 1998;126: 303–305.
    OpenUrlCrossRefPubMed
  8. ↵
    1. Foss AJE,
    2. Alexander RA,
    3. Hungerford JL, et al.
    Re–assessment of the PAS patterns on uveal melanoma. Br J Ophthalmol, 1997;81:240–246.
    OpenUrlAbstract/FREE Full Text
  9. ↵
    1. Kobayashi H,
    2. Shirakawa K,
    3. Kawamoto S, et al.
    Rapid accumulation and internalization of radio labeled herceptin in an inflammatory breast cancer xenograft with VM predicted by the contrast–enhanced dynamic MRI with the macromolecular contrast agent G6–(JB4M–Gd) (256). Cancer Res, 2002;62:860–866.
    OpenUrlAbstract/FREE Full Text
  10. ↵
    1. Seftor EA,
    2. Meltzer PS,
    3. Kirschmann DA, et al.
    Molecular determinants of human uveal melanoma invasion and metastasis. Clin Exp Metastasis, 2002;19: 233–246.
    OpenUrlCrossRefPubMed
  11. ↵
    1. Hess AR,
    2. Seftor EA,
    3. Gardner LM, et al.
    Molecular regulation of tumor cell VM by tyrosine phosphorylation: role of epithelial cell kinase (Eck/EphA2). Cancer Res, 2001;61: 3250–3255.
    OpenUrlAbstract/FREE Full Text
    1. Holash J,
    2. Maisonpierre PC,
    3. Compton D, et al.
    Vessel cooption, regression and growth on tumors mediated by angiopoientins and VEGF. Science, 1999;284: 1994–1998.
    OpenUrlAbstract/FREE Full Text
    1. Sanz L,
    2. Feijoo M,
    3. Blanco B,
    4. Serrano A,
    5. Alvarez–Vallina L.
    Generation of non–permissive basement membranes by anti-laminin antibody fragments produced by matrix-embedded gene–modified cells. Cancer Immunol Immunother, 2003;52:643–647.
    OpenUrlCrossRefPubMed
    1. Hendrix MJ,
    2. Seftor EA,
    3. Hess AR,
    4. Seftor RE.
    Vasculogenic mimicry and tumour–cell plasticity: lessons from melanoma. Nat Rev Cancer, 2003;3:411–421.
    OpenUrlCrossRefPubMed
    1. Sanz L,
    2. Feijoo M,
    3. Blanco B,
    4. Serrano A,
    5. Alvarez–Vallina L.
    Generation of non–permissive basement membranes by anti-laminin antibody fragments produced by matrix–embedded gene–modified cells. Cancer Immunol Immunother, 2003;52:643–647.
    OpenUrlCrossRefPubMed
    1. Hendrix MJ,
    2. Seftor EA,
    3. Kirschmann DA,
    4. Quaranta V,
    5. Seftor RE.
    Remodeling of the microenvironment by aggressive melanoma tumor cells. Ann N Y Acad Sci, 2003; 995:151–161.
    OpenUrlCrossRefPubMed
    1. Hendrix MJ,
    2. Seftor EA,
    3. Hess AR,
    4. Seftor RE.
    Molecular plasticity of human melanoma cells. Oncogene, 2003;22: 3070–3075.
    OpenUrlCrossRefPubMed
    1. Shirakawa K,
    2. Kobayashi H,
    3. Sobajima J,
    4. Hashimoto D,
    5. Shimizu A,
    6. Wakasugi H.
    Inflammatory breast cancer: vasculogenic mimicry and its hemodynamics of an inflammatory breast cancer xenograft model. Breast Cancer Res, 2003;5:136–139.
    OpenUrlCrossRefPubMed
    1. Dupuy E,
    2. Hainaud P,
    3. Villemain A,
    4. Bodevin–Phedre E,
    5. Brouland JP,
    6. Briand P,
    7. Tobelem G.
    Tumoral angiogenesis and tissue factor expression during hepatocellular carcinoma progression in a transgenic mouse model. J Hepatol, 2003;38:793–802.
    OpenUrlCrossRefPubMed
  20. ↵
    1. Lloyd RV,
    2. Vidal S,
    3. Horvath E,
    4. Kovacs K,
    5. Scheithauer B.
    Angiogenesis in normal and neoplastic pituitary tissues. Microsc Res Tech, 2003;60:244–250.
    OpenUrlCrossRefPubMed
Back to top

In this issue

Print
Download PDF
Email Article
Citation Tools

Jump to section

Keywords