Holly J. Pederson, MD, Director Medical Breast Services, Cleveland Clinic, Associate Professor of Medicine, CCLCM speaks about the ASCO 2021 Abstract – Ancestrally Unbiased Polygenic Breast Cancer Risk Assessment – new data validates the use of a polygenic breast cancer risk assessment in women across all ancestries.
Link to Abstract:
https://meetings.asco.org/abstracts-presentations/197017
Background information:
Single-nucleotide polymorphisms (SNPs) with minor effects that can be aggregated into polygenic risk scores affect BC risk (PRSs). PRSs is mainly created and validated for people of European ancestry. We created and validated a novel global PRS (gPRS) that uses human ancestral genetic makeup to make a PRS accessible to all women.
Methodologies:
An African PRS was developed using a cohort of 31,126 self-reported African American patients referred for hereditary cancer testing; an East Asian PRS was developed based on published data from the Asia Breast Cancer Consortium; and a European PRS was developed using a cohort of 31,126 self-reported European American patients referred for hereditary cancer testing. The fractional ethnicity due to each of the three continents was calculated using ancestry-informative SNPs for each patient. The gPRS was made up of the number of ancestry-specific PRSs that were weighted based on genetic ancestral composition. We analyzed sexism and calibration of gPRS in an independent validation cohort (N = 62,707), and compared results to a previously reported 86-SNP PRS for women of European ancestry. SNPs and PRSs were linked to BC using logistic regression, which was modified for personal and family cancer background, age, and ethnicity. Within the corresponding patient population, odds ratios (ORs) are stated per standard deviation. Two-sided P-values are published.
The below are the outcomes:
In both the complete validity cohort and sub-cohorts identified by self-reported ancestry, the gPRS was closely linked to BC (Table). Within both of the self-reported populations, 95 percent (88/93) of BC SNPs had a 1% prevalence of vulnerability alleles. With the exception of the Asian group, where the sample size was too insufficient to indicate dominance to any score, the gPRS demonstrated increased segregation overall and within each sub-cohort when compared to the 86-SNP PRS. The 86-SNP PRS was designed for white non-Hispanic women, but it was incorrectly optimized for people with non-European ancestry. For both women, the gPRS was correctly calibrated.
Final Thoughts:
For women of all ancestries, the 149-SNP gPRS has been validated and optimized. This technique, when combined with clinical and biological risk factors, could provide better risk stratification for all women, regardless of ethnicity.