Flow cytometric quantification of tumour endothelial cells; an objective alternative for microvessel density assessment.

Assessment of microvessel density by immunohistochemical staining is subject to a considerable inter-observer variation, and this has led to variability in correlation between microvessel density and clinical outcome in different studies. In order to improve the method of microvessel density measurement in tumour biopsies, we have developed a rapid, objective and quantitative method using flow cytometry on frozen tissues. Frozen tissue sections of archival tumour material were enzymatically digested. The single-cell suspension was stained for CD31 and CD34 for flow cytometry. The number of endothelial cells was quantified using light scatter- and fluorescence-characteristics. Tumour endothelial cells were detectable in a single cell suspension, and the percentage of endothelial cells detected in 32 colon carcinomas correlated highly (r=0.84, P<0.001) with the immunohistochemical assessment of microvessel density. Flow cytometric endothelial cells quantification was found to be more sensitive especially at lower levels of immunohistochemical microvessel density measurement. The current method was found to be applicable for various tumour types and has the major advantage that it provides a retrospective and quantitative approach to the angiogenic potential of tumours.

Quantitative assessment of angiogenesis and tumor vessel architecture by computer-assisted digital image analysis: effects of VEGF-toxin conjugate on tumor microvessel density.

Tumor growth is angiogenesis dependent. As a consequence, strategies aimed at disrupting this mechanism are heavily investigated. Several angiogenesis assays are used to directly compare the efficacy of anti-angiogenic compounds. However, objective assessment of new vascular growth has been difficult to achieve. The aim of this study was to test and develop a computer-assisted image analysis method that would give an unbiased quantification of the microvessel density. Human tumors were grown in athymic mice and tumor biopsies were taken after a weeklong treatment with VEGF-toxin conjugate. Frozen tumor sections were prepared and stained with PE-conjugated anti-CD-31 antibodies and vessels were imaged with a fluorescence microscope. Vessel density was analyzed by quantifying PE-positive pixels per recorded field. In addition, images were further processed to investigate morphological differences by an automated binarization and skeletonization protocol. This procedure allowed the computer-assisted estimation of important angiogenic parameters such as total vessel number, length, and branch points. Based on these indices, differences in the angiogenic response between control tumors and those treated with VEGF-toxin conjugate were readily detected (P < 0.007 for all parameters). More importantly, computer-generated measurements correlated well with manual microvessel counts and showed significantly less variation. Our results suggest that computer-assisted image analysis represents a rapid, objective, and alternative method for the quantitative assessment of tumor angiogenesis and vessel architecture.