Validation of a breast cancer polygenic risk score in 15,490 Brazilians using exome sequencing.

Abstract

10631 Background: Brazil has a highly admixed population with significant Southern European, African and Native American genetic ancestries. Approximately 75% of the population has two or more continental ancestries. Therefore, breast cancer polygenic risk scores (PRS) developed in predominantly European populations requires validation in the Brazilian population. Compared with genotyping arrays and whole genome sequencing, affordable and scalable whole exome sequencing (WES) allows genotyping of both rare and common variants in the targeted regions and imputation of common variants not directly captured. In this study we validate a previously developed 3820-variant PRS using WES in 15,490 Brazilians. Methods: The cohort consisted of 6,362 women with a history of breast cancer (mean age 49.6, stddev 11.6) and 9,128 unphenotyped adults as controls (mean age 41.6, stddev 13.3). WES were performed on germline DNA and variants were called and curated. Imputation was done using the 1000 Genomes Phase 3 reference panel for variants falling outside the WES captured regions. After excluding subjects with known pathogenic or likely pathogenic variants in BRCA1, BRCA2, PALB2, PTEN, or TP53, and first or second-degree relatives of the probands, 14,577 individuals remained in the final analysis (5,730 cases and 8,847 controls). We restricted PRS calculation to 2,893 of the original 3,820 variants with an estimated global imputation accuracy of 96%. Odds ratios (OR) were adjusted using the top 10 principal components. Results: The breast cancer PRS showed a highly significant correlation with cancer risk (p-value = 7.1e-81, OR = 1.43, 95% confidence interval (CI) = 1.38-1.48 per PRS standard deviation). Compared with women in the middle deciles, those in the highest 10% had an OR of 1.93, while those in the lowest 10% had an OR of 0.52 (Table). Conclusions: In this study we show that WES can capture 76% of the variants included in a previously developed and validated breast cancer PRS, and the calculated PRS is correlated with breast cancer risk in the admixed Brazilian population. This opens the door for clinical studies into the application of PRS in Brazil, particularly in women without mutations in breast cancer high-penetrance genes. In addition, our WES-based PRS strategy can be quickly, affordably, and seamlessly adopted by labs that already offer Hereditary Breast Cancer panels that use WES as a backbone. [Table: see text]