Abstract Background: High-grade serous ovarian cancer (HGSC) is a deadly gynecologic cancer and is characterized by widespread copy number alterations (CNA). However, knowledge of racial differences in HGSC CNA is limited. Using methylation-inferred CNA, we defined and characterized CNA-based tumor clusters to assess patterns by self-reported race. Methods: Tumor CNA were calculated using Illumina EPIC v1/v2 methylation data for 678 cohort participants (African American Cancer Epidemiology Study, North Carolina Ovarian Cancer Study, Nurses’ Health Study) compared to 296 blood samples. Probes were binned (≥10; ≥100 kb), segmented using circular binary methods, autocorrected, and filtered (|mean log R ratio|≥0.3; 4 markers), to define cytoband-level copy number states (-1-loss /0-neutral /+1-gain). CNA were clustered with Ward’s method. For each cluster, we computed length-weighted cytoband burden (% genome in gain, loss, or neutral states) and region-level ORs to identify CNA-enriched regions. Models clustered by race were then fit to assess differences. Results: Combined analysis of HGSC samples from 403 White (59%) and 275 Black women (41%), identified three CNA clusters: C1 (27%), C2 (57%), and C3 (15%). Clusters showed distinct genomic burdens (Kruskal-Wallis p10 −55). C1 showed widespread losses, notably in 4q (∼75% vs 14% in C2/C3, q10-45), which has been reported in HGSOC, and 17p (71% vs 13% in C2/C3, q10-46), which includes TP53. C3 exhibited extensive chromosomal gains, including 20p/q (∼82%/72%, q 10-38), 1q/1p (∼64%/39%, q10-32), and 2p/2q (∼56%/42%, q 10-31). Gains on 2p/q have been linked to advanced stage and poor prognosis, potentially via overexpression of AURKA, GNAS, and TPD52L2. Amplified 1q may enhance DHX9 expression, a regulator of genomic stability, while BARD1 (2q) interacts with BRCA1 in DNA repair; aberrant BARD1 expression correlates with poor outcomes. Cluster membership was similar between Black and White women (C1: 30% vs 26%, C2: 53% vs 59%, C3: 17% vs 15%). However, arm-level copy number racial differences were observed. Overall, 21q loss was more frequent in tumors from Black women (28%) than White women (16%; OR=0.5, 95% CI=0.34-0.73, q=0.02). This difference was most pronounced in C2, where 21q loss occurred in 27% of Black women compared to 10% in White women (OR = 0.31, 95% CI=0.17-0.56, qcluster=0.002). Conclusion: Broad arm-level CNAs define molecularly distinct HGSC subgroups with characteristic gain/loss patterns. Race-associated genomic alterations, including 21q loss among Black women, appear to be cluster-dependent genomic events suggesting underlying biological heterogeneity related to focal genomic instability or selective pressures rather than the result of wide-spread changes in overall copy number burden. Citation Format: Irma M. Vlasac, Brett M. Reid, Courtney Johnson, Alicia R. Richards, Christelle M. Colin Leitzinger, Sean J. Yoder, Dana Roeber, Tania Mesa, Andrew Berchuck, Brooke Fridley, Jing-Yi Chern, Jennifer A. Doherty, Kristin Haller, Shelley Tworoger, Jeffrey R. Marks, Joellen M. Schildkrau, Brock C. Christensen, Lauren Cole Peres, Lucas A. Salas Diaz, on behalf of the African American Cancer Epidemiology Study. Characterization of high-grade serous ovarian cancer copy number alterations in Black and White Women abstract. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 6813.
Vlasac et al. (2026) studied this question.