Abstract
AbstractPurposeBrazil has a highly admixed population. Polygenic Risk Scores (PRS) have been mostly developed from European population studies and applying them to other populations is challenging. To assess the use of PRS for breast cancer (BC) risk in Brazil, we validated four PRSs developed in the Brazilian population.Patients and MethodsWe analyzed 6,362 women with a history of breast cancer and 9,128 unphenotyped adults as controls in a sample obtained from a clinical laboratory. Genomic variants were imputed from exomes and scores were calculated for all samples.ResultsAfter excluding individuals with known pathogenic or likely pathogenic variants inBRCA1,BRCA2,PALB2,PTEN, orTP53,and first-degree relatives of the probands, 5,730 cases and 8,847 controls remained. Four PRS models were compared, and PRS 3820 from Mavaddatet al.2019 performed best, with an Odds Ratio (OR) of 1.41 per standard deviation (SD) increase (p-value: < 0.0001) and an OR of 1.94 (p-value: < 0.0001) for the individuals in the top risk decile. PRS 3820 also performed well for different ancestry groups: East Asian majority (Group 1), Non-European majority (Group 2), and European majority (Group 3), showing significant effect sizes for all groups: (Group 1: OR 1.54, p-value 0.006; Group 2: OR 1.44, p-value: <0.001; Group 3 OR: 1.43, p-value: <0.001). PRS 90% compares with monogenic moderate BC risk genes (PRS90 OR: 1.94; CHEK2 OR: 1.89; ATM OR: 1.99).ConclusionPRS 3820 can be accurately used in the Brazilian population. This will allow a more precise BC risk assessment of mutation-negative women in Brazil.
Publisher
Cold Spring Harbor Laboratory