Abstract PS4-09-25: Variation in diagnostic imaging features of breast cancer by race and ethnicity

Abstract BACKGROUND: Though racial and ethnic differences in tumor characteristics are well described, less is known about the potential variability in diagnostic and imaging features at presentation. Here, we examined variations in imaging features of breast cancer (BC) by race and ethnicity and explored their associations with clinical outcomes. METHODS: Diagnostic imaging features, clinicopathological characteristics and race and ethnicity information for adult females with a first diagnosis of stage I-III BC during 2016-2023 were obtained from our NCI-designated center’s prospectively maintained database, which captures all patients having breast surgery at our high-volume center. Occult disease was defined as a lack of detection on diagnostic mammography. Multifocal or multicentric disease was defined as having this feature on any imaging modality — mammogram, MRI, or ultrasound. Natural language processing (NLP) extracted breast density information from mammograms in the electronic health record (EHR). Additional demographics were extracted from the EHR including education and residential zip code, with zip codes used to calculate social deprivation index (SDI). Multivariable logistic regression was used to examine racial and ethnic differences in the presence of mammographically occult, multifocal, and multicentric disease as well as breast density, adjusting for age and tumor features. Cox proportional hazards model evaluated overall survival (OS), recurrence free survival (RFS), and distant recurrence-free survival (DRFS) by race and ethnicity, accounting for clinicopathological, diagnostic, and demographic characteristics. RESULTS: Among 10038 patients, mean age was 58.5 SD 13.3); 86.5% were White, 4.7% Black, 4.4% Asian, and 3.9% Hispanic. Most patients had clinical T1 (51.6%), node-negative (85.0%) and HR+HER2- (75.2%) disease. 5.5% had occult disease; 33.9% multifocal and 12.7% multicentric disease on imaging. 6.6% were in the most deprived SDI quintile. Occult disease was most prevalent among Hispanic patients (9.4%), while Asian patients had the highest rates of multifocal (43.4%), multicentric (22.0%) disease and extremely dense breast (24.7%). Black patients most frequently fell into the most deprived SDI quintile (34.6%). In multivariable analyses, after adjusting for clinicopathological features, Asian (vs. White) race was significantly associated with higher odds of multifocal (OR 1.24, 95% CI 1.01-1.52) and multicentric (OR 1.59, 95% CI 1.23-2.03) disease and extreme breast density (OR 1.81, 95% CI 1.39-2.34) (all p 0.001). Hispanic (vs. non-Hispanic) ethnicity was associated with higher odds of occult disease (OR 2.08, 95% CI 1.35-3.13, p=0.001). Black (vs. White) women had lower odds of having extremely dense breasts (OR 0.61, 95% CI 0.42-0.87, p=0.008). Occult and multicentric disease were associated with extremely dense breasts (all p .05). After adjusting for all diagnostic features, lymphovascular invasion (LVI), triple negative BC (TNBC), older age, higher stage and grade were associated with worse RFS, DRFS and OS (p 0.05), and residing in a more deprived area was associated with lower RFS and DRFS (HR 1.37, p 0.05). Race, ethnicity, breast density, multifocal, and multicentric disease were not associated with RFS, DRFS or OS. CONCLUSIONS: In this unique, large sample of women with breast cancer, we identified racial and ethnic differences in imaging features of breast cancer, particularly among Asian and Hispanic patients. Although we observed differences in breast density, multifocality and multicentricity, after adjusting for more traditional features (e.g., subtype), these dissimilarities were not associated with RFS, DRFS or OS and may not be prognostic. Longer follow-up is needed to see if accounting for more enhanced diagnostic characteristics impacts outcomes. Citation Format: A. Odai-Afotey, Q. Jin, N. Tayob, B. N. Durieux, C. Lindvall, J. Vincuilla, T. A. King, E. A. Mittendorf, R. A. Freedman. Variation in diagnostic imaging features of breast cancer by race and ethnicity abstract. In: Proceedings of the San Antonio Breast Cancer Symposium 2025; 2025 Dec 9-12; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2026;32(4 Suppl):Abstract nr PS4-09-25.