BACKGROUND: Polygenic risk scores (PRSes) have emerged as promising tools for stratifying breast cancer risk on the basis of genomic data. However, PRSes developed in one ancestral population may vary in predictive performance when applied to a different population. This study aimed to evaluate the utility of six breast cancer PRSes developed via European and East Asian genome-wide association studies (GWASes) in a Japanese prospective cohort. METHODS: We analysed data from 7,965 women enrolled in the Japan Multi-Institutional Collaborative Cohort (J-MICC) study, with a mean follow-up of 11.3 years. Six published PRSes were calculated via genome-wide genotype data, and their associations with breast cancer incidence were assessed via Cox proportional hazards models adjusted for age, survey area, and population structure. Hazard ratios (HRs), 95% confidence intervals (CIs), and Harrell’s C statistics are reported. RESULTS: Five of the six PRSes were significantly associated with breast cancer risk. The European-derived PRS313_BC demonstrated the highest predictive performance (HR per standard deviation (SD) = 1.64; C statistic = 0.69; P < 0.001), particularly among women diagnosed with breast cancer before 50 years of age. CONCLUSIONS: Despite cross-ancestry differences, a European-derived PRS demonstrated strong predictive value for breast cancer risk in a Japanese population, likely reflecting the scale and optimization of the original GWAS. These findings indicate that GWAS scale and statistical power may currently outweigh ancestry matching in PRS performance. However, large-scale ancestry-specific GWASes remain essential to improve prediction accuracy in East Asian populations.
Keywords: Breast cancer; Polygenic risk score; cohort study.