Abstract 6211: AI-guided spatial multi-omics integration reveals immune and stromal heterogeneity underlying population disparities in prostate cancer

Key Points

• Investigate biological mechanisms behind prostate cancer disparities between African American and European American men using AI-guided multi-omics integration.
• Developed an AI-based framework for spatial characterization of the tumor microenvironment.
• Employed CODA for histological segmentation of prostate tumor sections from African American and European American patient samples.
• Integrated spatial transcriptomics with segmentation maps for biological annotation of the tumor microenvironment.
• Conducted functional analysis of distinct tumor microenvironment niches.
• AA tumors displayed higher immune infiltration and inflammatory activity compared to EA tumors.
• Integration with spatial proteomics data revealed immune clusters and discordance between transcript expression and protein signaling.
• Achieved over 90% classification accuracy for tumor compartment analysis.