A spatial statistical framework for the parametric study of fiber networks: application to fibronectin deposition by normal and activated fibroblasts

Abstract

Changes in the spatial landscape of the extracellular matrix (ECM) in health and disease significantly impact the surrounding tissues. Quantifying the spatial variations in the fibrillar architecture of major ECM proteins could enable a profound understanding of the link between tissue structure and function. We propose a method to capture relevant ECM features using graph networks for fiber representation in normal and tumor-like states of 4 alternatively spliced isoforms of fibronectin (FN) associated with embryonic development and disease. Then, we construct graph-derived statistical parametric maps, to study the differences across variants in normal and tumor-like architectures. This novel statistical analysis approach, inspired from the analysis of functional magnetic resonance imaging (fMRI) images, provides an appropriate framework for measuring and detecting local variations of meaningful matrix parameters. We show that parametric maps representing fiber length and pore orientation isotropy can be studied within the proposed framework to differentiate among various tissue states. Such tools can potentially lead to a better understanding of dynamic matrix networks within the tumor microenvironment and contribute to the development of better imaging modalities for monitoring their remodeling and normalization following therapeutic intervention. Author SummaryDue to the complex architectural diversity of biological networks, there is an increasing need to complement statistical analyses with a qualitative and local description. The extracellular matrix (ECM) is one such network for which fiber arrangement has a major impact on tissue structure and function. Thus, a flexible numerical representation of fibrillar networks is needed for accurate analysis and meaningful statistical comparison of ECM in healthy and diseased tissue. First, we propose a versatile computational pipeline to study fiber-specific features of the ECM with graph networks. Then, we introduce a novel framework for the statistical analysis of graph-derived parametric maps, inspired from the statistical analysis of fMRI parametric maps. This analysis is useful for the quantitative/qualitative comparison of ECM fiber networks observed in normal and tumor-like, or fibrotic states. These methods are applied to study networks of fibronectin (FN), a provisional ECM component that dictates the organization of matrix structure. From 2D confocal images we analyzed architectural variations among 4 alternatively spliced isoforms of FN, termed oncofetal FN, that are prevalent in diseased tissue. We show how our approach can be used for the computation and statistical comparison of heterogeneous parametric maps representing FN variant-specific topological/geometrical features. These methods may be further developed and implemented into tumor tissue ECM profiling to decipher the specific roles of ECM landscapes and their remodeling in disease.