Deciphering the Genomic architecture of three major Cancers in African-Ancestry Populations

Abstract

Genomic studies of cancer risk have disproportionately focused on populations of European ancestry, limiting biological insight and risk prediction in African-ancestry populations that experience a high burden of disease. Here, we analysed breast, colorectal, and prostate cancers in African-ancestry participants from the UK Biobank using ancestry-aware genome-wide association studies (GWAS), SNP-based heritability estimation, fine-mapping, transcriptome-wide association studies (TWAS), and polygenic risk scoring (PRS). SNP-based heritability analyses revealed a comparatively high point estimate of common-variant heritability for colorectal cancer risk in African-ancestry individuals, alongside more modest estimates for breast and prostate cancer. Five loci reached genome-wide significance (p < 5x10-{square}), including four colorectal cancer loci (notably rs111448231 in RYR2) and one novel breast cancer locus (rs78768133). Gene-based burden testing identified eight prostate cancer-associated genes (MRPL45, PSMD8, GGN, SPRED3, FAM98C, BCLAF1, MTFR2, and NELL2) with FDR-significant associations, clustering within biologically plausible chromosomal regions on chr19q13 and chr6q23. Transcriptome-wide association analysis identified CYTH2 (ENSG00000105443.13) as a significant gene for prostate cancer. Polygenic risk scores incorporating African-ancestry linkage disequilibrium demonstrated heterogeneous predictive performance across cancers, with modest discrimination for colorectal and breast cancer and substantially stronger performance for prostate cancer (AUC = 0.89). Together, these findings delineate ancestry-relevant cancer genetic architectures and demonstrate the importance of population-matched genomic approaches for equitable precision oncology.