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.
Copyright 2000 Academic Press.