Abstract B172: Comparing pathogenic variants (PV) and variants of uncertain significance (VUS) in breast cancer (BC) susceptibility genes by genetic similarity groups among WISDOM 2.0 participants
TL;DR
Findings provide population-scale evidence that current reference databases underserve non-European populations, highlighting an urgent need to diversify genomic datasets to ensure equitable variant classification in clinical risk-based screening.
Abstract
WISDOM (Women Informed to Screen Depending On Measures of risk), a study of risk-based breast cancer (BC) screening, is the first to deploy population-based germline testing for high- and intermediate-penetrance BC genes. Given that genomic reference databases remain disproportionately European in ancestry, VUS detection rates are known to be higher in non-European populations, undermining equitable, personalized risk assessment. We used single nucleotide variants (SNVs) to infer genetic similarity to reference populations and evaluated variant classification and frequency by genetic similarity groups across the WISDOM cohort. We analyzed sequencing data from 17,930 WISDOM 2.0 participants with qualifying test results across 12 BC susceptibility genes (ATM, BARD1, BRCA1, BRCA2, CDH1, CHEK2, PALB2, PTEN, RAD51C, RAD51D, STK11, TP53), classifying variants as pathogenic (P), likely pathogenic (LP), or of uncertain significance (VUS). Genetic similarity was estimated using 1000 Genomes dataset as a reference, including 5 meta-populations: African (AFR), East Asian (EAS), European (EUR), Mixed American (AMR), or South Asian (SAS). We used data ∼100,000 independent SNVs (R2<0.10) and ran principal components analysis (PCA). We then projected WISDOM participants into predetermined cutoffs to identify each participant as most like one of 5 populations. We compared P/LP frequencies across all groups with Fisher's exact test; VUS proportions with chi-squared tests. PCA analyses were conducted in PLINK and all statistical tests were run as heterogeneity tests including all groups using R. Of the 17,930 participants, 336 (1.9%) carried a P/LP variant and 1,323 (7.4%) carried a VUS in at least one gene, including 158 (1.0%) in CHEK2. VUS burden varied significantly (p=0.0005) by genetic similarity group: EAS participants had the highest (88/765, 11.5%), followed by AFR (93/891, 10.4%), AMR (94/917, 10.3%), and SAS (15/159, 9.4%) all higher than EUR (1,033/15,198, 6.8%). There was also significant heterogeneity (p=0.01) in the rate of P/LP variants by genetic similarity group. EUR participants had the highest rate of P/LP variants (305/15198, 2.0%), comparable to SAS (3/159, 1.9%) and AFR (15/891, 1.7%), and higher than EAS (8/765, 1.0%) and AMR (7/917, 0.8%). The P/LP variants in CHEK2 were particularly high in EUR participants (1.0%) and driven by 3 founder mutations compared to other genetic similarity groups. P/LP rates for other genes did not differ significantly by genetic similarity group. In this pragmatic screening trial, P/LP detection rates were highest in EUR genetic similarity participants, largely driven by CHEK2 founder variants, while VUS rates were higher across all non-EUR groups and most pronounced in EAS participants. These findings provide population-scale evidence that current reference databases underserve non-European populations, highlighting an urgent need to diversify genomic datasets to ensure equitable variant classification in clinical risk-based screening. Jackelyn J. Moya, Daisy Montoya, Allison Fiscalini, Irene Acerbi Soto, Yaruska Ordinola Perez, Kirk Fergus, Yiwey Shieh, Katherine Ross, Deborah Goodman, Dezheng Huo, Donglei Hu, Funmi Olopade, Jeff Tice, Diane Heditsian, Dolores Moorehead, Laura Esserman, Lisa Madlensky, Elad Ziv. Comparing pathogenic variants (PV) and variants of uncertain significance (VUS) in breast cancer (BC) susceptibility genes by genetic similarity groups among WISDOM 2.0 participants [abstract]. In: Proceedings of the 19th AACR Conference on the Science of Cancer Health Disparities; 2026 Oct 18-21; Atlanta, GA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2026;35(10 Suppl):Abstract nr B172.