Multiparameter persistent homology landscapes identify immune cell spatial patterns in tumors

Area of Science:

• Computational biology
• Topological data analysis
• Cancer research

Background:

• Analyzing complex, heterogeneous, and noisy biological data, such as tissue samples, presents significant challenges.
• Topological data analysis (TDA) has emerged as a powerful mathematical framework for extracting information from data shapes.
• Multiparameter persistent homology (MPH) is an extension of TDA theory designed to handle heterogeneous data.

Purpose of the Study:

• To apply MPH landscapes, a statistical tool derived from MPH theory, to analyze spatial patterns in complex biological systems.
• To evaluate the performance of MPH landscapes against traditional spatial statistics and one-parameter persistent homology.
• To investigate the spatial distribution of immune cells within the tumor microenvironment.

Main Methods:

• Computation of MPH landscapes using noisy data from agent-based model simulations of immune cells.
• Application of MPH landscapes to digital histology images of head and neck cancer.
• Quantification of intratumoral immune cell spatial patterns, including voids and distributions.
• Integration of TDA with multi-modal data, such as immune cell locations and oxygenation levels.

Main Results:

• MPH landscapes demonstrated superior performance compared to existing spatial statistics and one-parameter persistent homology for analyzing simulated immune cell infiltration.
• Analysis of head and neck cancer histology revealed distinct spatial patterns for regulatory T cells and macrophages, with regulatory T cells exhibiting more prominent voids.
• MPH landscapes successfully integrated and interrogated diverse data types and scales within the tumor microenvironment.

Conclusions:

• MPH landscapes provide a robust and effective method for quantifying and characterizing features within the tumor microenvironment.
• This approach enhances our understanding of immune cell spatial organization in cancer.
• MPH landscapes hold significant potential for analyzing complex biological data across various scales and modalities.