ISPAT-3D: Spatially Varying Conditional Volumetric Network Estimation for 3D Tumor Imaging

Area of Science:

• Computational biology
• Bioinformatics
• Cancer research

Background:

• The tumor microenvironment's spatial organization critically influences immune responses and cancer progression.
• Current methods for analyzing cell interactions in multiplexed tissue images are limited to 2D and overlook spatial autocorrelation.
• Understanding 3D spatial relationships is essential for accurate cancer diagnostics and therapeutics.

Purpose of the Study:

• To introduce ISPat-3D, a novel hierarchical Bayesian framework for analyzing 3D multiplexed cancer imaging data.
• To recover spatially varying, zone-specific cell-type interaction networks within the tumor microenvironment.
• To overcome the limitations of 2D analyses by incorporating spatial autocorrelation and volumetric information.

Main Methods:

• ISPat-3D partitions tissue volumes into tumor intensity zones.
• It employs anisotropic Gaussian processes for cell type modeling with distinct spatial scales.
• The framework utilizes multi-study factor analysis and extracts partial correlation networks from precision matrices.

Main Results:

• Simulations confirm ISPat-3D's ability to accurately recover shared and zone-specific interaction structures with high statistical power.
• Application to colorectal and breast cancer datasets revealed zone-specific immune cell modules and fibroblast-endothelial interactions.
• The analysis identified critical 3D spatial interactions, including shifts in T cell regulation and angiogenic remodeling, not detectable in 2D.

Conclusions:

• ISPat-3D provides a powerful framework for dissecting complex 3D spatial interactions within the tumor microenvironment.
• The method reveals critical zone-specific cellular communication patterns influencing cancer progression and immune evasion.
• This approach enhances our understanding of tumor heterogeneity and offers new avenues for targeted cancer therapies.