Spatial Genetic Mapping of Normal Human Prostate Tissue Identifies Epithelial Compartments Enriched for Prostate...

Previous genome-wide association studies have identified more than 400 prostate cancer (PCa) susceptibility loci. However, the tissue and spatial epithelial contexts through which inherited PCa risk may operate in the normal prostate remain incompletely understood. We analyzed normal human prostate spatial transcriptomic data using a gsMap-based framework to localize PCa association signals across spatially resolved prostate compartments. Spatial spots were clustered using SpaGCN and then manually curated and annotated based on matched histological images and canonical prostate marker gene expression. Because individual spatial spots may contain multiple cells, annotations were interpreted as compartment-enriched dominants rather than pure cell-type identities. PCa genetic association scores were aggregated within annotated compartments using Cauchy-combined p value and median spot-level p value as a complementary descriptive summary. Gene-level spatial correspondence was further assessed using Pearson correlation coefficients (PCCs) between gene-expression patterns and spot-level PCa gsMap signal. We annotated normal prostate tissue into four major compartments: basal epithelium-enriched, luminal epithelium-enriched, smooth muscle-enriched, and fibroblast stroma-enriched compartments. Basal epithelium-enriched compartments showed the strongest and most consistent enrichment for PCa association, with a Cauchy-combined p of 1.85 × 10^-8 and a median spot-level based p of 3.67 × 10^-7. Luminal epithelium-enriched compartments also showed strong enrichment, with a Cauchy p of 3.79 × 10^-7 and a median-based p of 1.45 × 10^-6. In contrast, fibroblast stroma-enriched compartments showed no evidence of enrichment (Cauchy p = 0.52; median p = 0.42), while smooth muscle-enriched compartments showed only modest evidence in the Cauchy-based analysis that was accompanied by a weaker median spot-level association signal (Cauchy p = 8.95 × 10^-4; median p = 0.18). PCC analysis showed that genes with the strongest spatial correspondence to PCa gsMap signal were predominantly epithelial-associated, including CDH1, VAMP8, GMNN, TSPAN1, and FOXA1. In normal human prostate tissue, inherited PCa association localizes preferentially to basal epithelium-enriched and luminal epithelial-enriched compartments rather than to smooth muscle-enriched or fibroblast stroma-enriched compartments. Gene-level PCC patterns further support spatial alignment between PCa-associated gsMap signal and epithelial transcriptional programs. These findings provide proof-of-concept evidence for the spatial localization of inherited PCa susceptibility through the integration of large-scale population-based genetic data and spatial transcriptomics.