A computational approach to compare microvessel distributions in tumors following antiangiogenic treatments

Abstract

Experimental approaches currently used to quantify the activity of antiangiogenic treatments in cancer therapy do not generally address the importance of spatial distribution of microvessels in target tissues. We report a new computerized method to assess tumor vascularization by quantifying the distribution of functional microvessels as revealed by in vivo staining with sulfosuccinimidyl-6-(biotinamido) hexanoate. Our approach was based on pixel dilation of digital images of blood vessels and addressed the space-filling property of the vessel layouts. This was practically achieved computing the number of dilation cycles (Halo index) needed to permeate a pre-defined amount of each image. Our approach was validated on human tumor xenografts in nonobese diabetic/severe combined immunodeficient mice treated with the antiangiogenic drug sorafenib. For each experimental model, area normalization allowed the unbiased comparison of several hundreds of images showing different amounts of vascular tissue. In two different tumor types, comparison of Halo values showed statistically significant differences between control and sorafenib-treated samples. Conversely, this effect was not observed in samples from an additional xenograft known to resist the antiangiogenic treatment. By separating the analysis of vessel area from the quantification of vessel distributions, our approach can potentially contribute to a better evaluation of the antiangiogenic or vascular-disrupting activity of new drugs or treatments.