Scott, C

Abstract Artificial intelligence (AI)-scores estimated from digital mammograms predict future breast cancer (BC) risk. MIRAI is a deep learning BC risk model that provides a continuous 5-year risk of overall BC from four screening full field digital mammograms. Common germline genetic variation in the form of a polygenic risk score (BC-PRS) is associated with increased BC risk and may improve MIRAI’s 5-year risk prediction. Our goal was to develop an updated MIRAI 5-year risk model that incorporates the BC-PRS (MIRAI+PRS) and to evaluate its discriminatory accuracy and calibration for both overall and invasive BC compared to MIRAI alone. We developed the MIRAI+PRS model by multiplying each woman’s 5-year MIRAI risk estimate by their relative risk based on their BC-PRS relative to the population mean. We evaluated the models within the Mayo Clinic Biobank mammography cohort, comprised of 12,307 women without a prior history of BC; 176 invasive and 250 overall BC were diagnosed within 5 years. MIRAI was estimated on screening mammograms closest to enrollment but at least 6 months prior to BC. Discriminatory accuracy, assessed by C-statistic, was high and similar for MIRAI+PRS vs. MIRAI models, for overall BC and invasive BC (Table). Calibration assessed by observed to expected (O/E) ratios was also similar for MIRAI+PRS compared to MIRAI predictions for BC outcomes (Table), although there was improvement in decile-specific O/E ratios across the lowest risk deciles (<1.67%) for MIRAI+PRS. Calibration for invasive cancer was poor for MIRAI with or without PRS. In summary, the MIRAI+PRS risk model did not result in significant difference of discriminatory accuracy or overall calibration compared to the MIRAI model, but there was evidence for improved calibration for women with 5-year risk below 1.67%. For invasive BC, the model had poor calibration regardless of whether PRS was included, underscoring the importance of training AI models for BC outcomes that are associated with a clinical intervention. Citation Format: Christopher G. Scott, Peter Kraft, Imon Banerjee, Ramon Correa Medero, Aaron D. Norman, Fergus J. Couch, Karla Kerlikowske, Stacey J. Winham, Celine M. Vachon. Development and evaluation of a MIRAI 5-year risk model with a breast cancer polygenic risk score [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 2779.

Abstract Background Common genetic variation has been linked to an increased risk of developing dilated cardiomyopathy (DCM) in case-control studies. However, the phenotypic profile and biological mechanisms by which genetic background may predispose individuals to DCM and outcomes in the general population remain unclear. Aim The primary objectives of this study were twofold: (1) to validate a DCM polygenic score (PRS-1) based on a prior meta-analysis of case–control genetic association studies [Tadros R et al. Nat Genet (2021)]; and (2) To apply this PRS, along with two recently developed PRSs for DCM [PRS-2 from Zheng SL et al. Nat Genet (2024) and PRS-3 from Jurgens SJ et al. Nat Genet (2024)] to a general population that underwent comprehensive longitudinal phenotypic assessments and had long-term outcomes available. Methods While the PRS-1 was derived from only 13 SNPs, PRS-2 and PRS-3 were generated from 709,534 and 1,072,247 SNPs, respectively with significant overlap between them. These DCM-PRSs were applied to an unselected community-based cohort (n=1897) aged >45 years, who underwent a physical examination, echocardiogram, and blood sample at baseline. Subsequent clinical outcomes were collected over a median 16 year follow-up (IQR: 15-17). Results No significant differences between the quartiles were present in terms of age, sex (48% male), and cardiovascular risk factors (CV-RF) (such as hypertension, diabetes, and coronary artery disease). However, linear regression models demonstrated a significant correlation between all DCM-PRSs and left ventricular (LV) parameters, including LV ejection fraction, LV end-systolic diameter, and LV end-diastolic diameter (Table). Multivariable regression analysis confirmed that these correlations were independent of age, sex, CV-RF, and population stratification (adjusted for the first 10 principal components [PCs]). During follow-up, 47 individuals developed DCM. A logistic regression analysis, using a 3:1 matched cohort of non-DCM vs. DCM individuals based on age and sex and adjusted for the same 10 PCs, found a significant correlation between DCM development and both PRS-1 [OR=2.27 (1.06–4.85), p=0.03] and PRS-3 [OR=2.63 (1.20–5.75), p=0.02], while PRS-2 did not show a significant association [OR=1.85 (0.84–4.06), p=0.12]. No association was found between the DCM-PRSs and heart failure (HF) biomarkers (aldosterone, natriuretic peptides, ST2, Galectin-3) or HF outcomes (all-cause mortality/HF hospitalization). Conclusion PRS-1 for DCM despite being derived from only 13 risk alleles, performs at least as well as the newer and more extensive PRSs, in its correlation with LV function, dimensions, and DCM development (except for PRS-2) in a general population. The absence of correlation between these DCM-PRSs and the renin-angiotensin-aldosterone system or natriuretic peptides suggests that genetic background influences LV traits through mechanisms independent of these biological pathways.