Development of a novel analytical pipeline to characterize morphological and molecular features of blood vessels within marginal regions of glioblastoma
Source: PubMed Central Open Access, NCBI / U.S. National Library of Medicine
Abstract Background Glioblastomas (GBM) are highly invasive tumors with marginal regions comprising unresectable functional brain infiltrated by tumor cells. Effective drug delivery to these regions is crucial, but lack of understanding of the structural and functional characteristics of their vasculature is impeding drug development. We aimed to develop a bespoke analytical pipeline that could be used to characterize molecular and morphological features of the blood-brain barrier within marginal regions of GBM by analyzing image descriptors extracted from multiplex colorimetric imaging of human samples. s1 Methods Multiplex immunohistochemical consecutive staining of key vascular antigens was performed on human samples of GBM and adjacent brain to determine the morphology and composition of blood vessels. A neural network was utilized to segment the vessels, and multiple image descriptors extracted to characterize and classify them with a Linear Discriminant model. s2 Results Multiplexed immunohistochemistry was optimized for vessel related-antigens CD31, laminin, claudin-5, smooth muscle actin, platelet-derived growth factor beta, and glial fibrillary acidic protein. Multiple parameters analyzed from the segmented blood vessels were modeled into four distinct categories, linked to region-specific molecular and morphological factors. Margin regions exhibited the lowest vessel density and a heterogenous mix of vessels with some unique to the region and others similar to tumor
Abstract
Abstract Background Glioblastomas (GBM) are highly invasive tumors with marginal regions comprising unresectable functional brain infiltrated by tumor cells. Effective drug delivery to these regions is crucial, but lack of understanding of the structural and functional characteristics of their vasculature is impeding drug development. We aimed to develop a bespoke analytical pipeline that could be used to characterize molecular and morphological features of the blood-brain barrier within marginal regions of GBM by analyzing image descriptors extracted from multiplex colorimetric imaging of human samples. s1 Methods Multiplex immunohistochemical consecutive staining of key vascular antigens was performed on human samples of GBM and adjacent brain to determine the morphology and composition of blood vessels. A neural network was utilized to segment the vessels, and multiple image descriptors extracted to characterize and classify them with a Linear Discriminant model. s2 Results Multiplexed immunohistochemistry was optimized for vessel related-antigens CD31, laminin, claudin-5, smooth muscle actin, platelet-derived growth factor beta, and glial fibrillary acidic protein. Multiple parameters analyzed from the segmented blood vessels were modeled into four distinct categories, linked to region-specific molecular and morphological factors. Margin regions exhibited the lowest vessel density and a heterogenous mix of vessels with some unique to the region and others similar to tumor core or normal brain vessels. s3 Conclusions We established a multiplex immunohistochemical staining protocol and pipeline to identify blood vessels and analyze their composition. The pipeline and the preliminary quantitative data it generated will facilitate more comprehensive characterization of margin-specific blood vessels, with implications for drug development. s4