Kerlikowske, Karla

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 Artificial intelligence (AI)–based, mammography breast cancer risk prediction models show improved discriminatory accuracy relative to clinical risk models. However, data on their calibration are limited. This study compared model performance of 3 clinical breast cancer risk models: Gail, Tyrer–Cuzick v8, and Breast Cancer Surveillance Consortium v3 with the MIRAI AI-risk model. Methods Digital mammograms were ascertained from a screening mammography cohort of 12 308 women within the Mayo Clinic Biobank with 250 incident breast cancers (176 invasive) within 5 years. We predicted 5-year breast cancer risk, estimated discriminatory accuracy (C-index), and calibration (observed-to-expected ratio) of both overall and invasive breast cancer and compared estimates using bootstrapping approaches. Results MIRAI demonstrated similar or improved discriminatory accuracy of overall breast cancer (C-index = 0.71, 95% confidence interval [CI] = 0.68 to 0.74) and invasive breast cancer (C-index = 0.71, 95% CI = 0.67 to 0.75) compared with clinical models (overall breast cancer: C-index = 0.59-0.68; invasive breast cancer: C-index = 0.60-0.68). MIRAI’s calibration for risk of overall breast cancer (observed-to-expected ratio = 0.96, 95% CI = 0.85 to 1.08) was improved compared with Gail (observed-to-expected ratio = 1.22, 95% CI = 1.07 to 1.38) and Breast Cancer Surveillance Consortium (observed-to-expected ratio = 1.38, 95% CI = 1.22 to 1.56) but similar to TC with volumetric percent density and polygenic risk score (observed-to-expected ratio = 0.99, 95% CI = 0.87 to 1.13). However, for low-risk women (approximately 50%), MIRAI overestimated risk of overall breast cancer. MIRAI also overestimated risk of invasive breast cancer across the risk spectrum (observed-to-expected ratio = 0.68, 95% CI = 0.58 to 0.78), whereas clinical models had good calibration (observed-to-expected ratio = 0.86–0.99). Conclusion MIRAI demonstrated stronger discriminatory accuracy than clinical models for 5-year overall and invasive breast cancer risk prediction but overestimated risk for both breast cancer endpoints. AI-based risk models should consider discriminatory accuracy and calibration for invasive cancer before implementation.

Abstract Background There is growing interest in artificial intelligence (AI) models for predicting future breast cancer. We performed a systematic review of studies of mammography-based AI models for future breast cancer risk prediction to summarize current evidence, identify knowledge gaps, and inform future research directions. Methods We searched 6 databases for studies from January 1, 2012, to February 28, 2025, that evaluated mammography-based AI models for future breast cancer risk prediction. We extracted study design, participants’ race and ethnicity, geographic origin, mammogram type, vendor, prediction time frame, breast cancer type predicted, external validation, and exclusion of cancers diagnosed on the index screening mammogram. Areas under the receiver operating curve (AUCs) were summarized overall and by study characteristics. Results A total of 41 studies met our inclusion criteria. All studies were retrospective, and most used 2D mammograms (n = 37 studies) acquired using Hologic equipment (n = 25) and performed in the United States (n = 17); White, non-Hispanic women were most represented. Nearly all (n = 40) studies assessed discrimination performance with a median AUC of 0.71 for no longer than 2-year risk prediction, 0.72 for 3-4 years, and 0.71 for 5 years or more prediction. Median AUC was 0.75 for studies including index cancers vs 0.68 when excluded. Six studies reported model calibration performance ranging from good to overestimation of risk. Conclusion Future studies should evaluate models using digital breast tomosynthesis, examine performance for aggressive or advanced breast cancer, include diverse populations, and evaluate both discrimination and model calibration. Prospective evaluations are needed to determine the clinical utility of mammography-based AI models for personalized risk-based breast cancer screening before implementation.

Radiomic parenchymal phenotypes on mammograms were used to predict breast cancer risk among both Black and White women, notably for false-negative findings and symptomatic interval cancer.

Abstract Artificial intelligence (AI) models based on deep learning of mammogram images have been developed for lesion detection and diagnosis as well as breast cancer (BC) risk. We and others have shown that these AI models, coupled with breast density, predict long-term breast cancer risk. Common germline genetic variation, in particular a polygenic risk score (PRS) that captures the cumulative impact of hundreds of common risk variants, has been shown associated with 3-fold increased BC risk. The BC-PRS may complement imaging risk factors to improve identification of women at high BC risk and better target prevention strategies. We evaluated the contribution of a BC-PRS to two imaging-based AI models (Transpara AI cancer detection system and Mirai 5-year risk) and breast density measures to long-term BC risk. We hypothesized that the BC-PRS will improve long-term risk prediction above AI and density measures. We examined this hypothesis within a cohort of 10,271 women enrolled to the Mayo Clinic Biobank between 2009 and 2015. Eligible women had no prior cancer, were ages 35 to 85, had a full-field digital screening mammogram within two years of enrollment and genotyping by Regeneron. Incident invasive and DCIS BC was identified through the Mayo tumor registry and follow-up questionnaires, and person years were calculated from date of mammogram through date of BC, last mammogram at Mayo or last questionnaire. We estimated Transpara AI malignancy score (1-10), 5-year Mirai risk and Volpara volumetric density measures (volumetric percent density, dense volume) on screening mammograms at enrollment and obtained clinical BI-RADS density. We calculated the BC-PRS using 268 SNPs (of the 313-SNP score) and the weighted sum of the SNPs using odds ratios from the largest BC GWAS of European ancestry to date. We performed cox proportional hazards regression adjusting for age and BMI to estimate hazard ratios (HRs), 95% confidence intervals (CI), and C-statistics (AUC) to describe the contribution of BC-PRS (per standard deviation (SD)) to the association of AI scores and breast density with BC risk. Likelihood ratio tests (LRT) and bootstrapping were used to compare model performance with vs. without inclusion of BC-PRS, at 10-years follow-up. Comparisons to the Tyrer-Cuzick model are ongoing. Over a median of 9.1 years (Interquartile range, 6.8 to 10.6), 446 incident BC were identified of which 285 invasive and 103 DCIS were confirmed by pathology report (60 were self-reported); 399 (256 invasive) occurred within 10 years and were used in primary analyses. Transpara score [HR per 1 unit=1.18 (1.14, 1.23), AUC=0.655 (0.628, 0.682)] and Mirai risk [HR per SD log risk=1.54 (1.43, 1.65), AUC=0.681 (0.655, 0.706)] were both associated with BC. All three breast density measures were also significantly associated with BC, including in models with AI scores (Table). Addition of the BC-PRS to models with AI scores and breast density improved both model fit (All PLRT < 0.001; Table) and discrimination, with increases in AUC ranging from 0.033 to 0.034 for models with Transpara and density and 0.025 to 0.026 for models with Mirai and density (all P-values < 0.001; Table). Analyses restricted to invasive BC showed identical results. But, the BC-PRS contribution at 5-year follow-up was attenuated. In this cohort study, we found BC-PRS contributed to long-term risk prediction of BC above imaging AI models and breast density. Imaging and genetic risk factors are complementary and may outperform existing risk models and provide more accurate risk assessment to aid in precision prevention strategies. Citation Format: Celine Vachon, Christopher Scott, Imon Banerjee, Ramon Correa Moderno, Dan Hursh, Matthew Jensen, Yvonne Wang, Kathleen Brandt, Karla Kerlikowske, Sandhya Pruthi, Fergus Couch, Stacey Winham. Contribution of a breast cancer polygenic risk score to mammography artificial intelligence models and breast density for long term breast cancer risk prediction [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PS10-06.

Abstract Artificial intelligence (AI) models applied to mammography have been shown to improve breast cancer (BC) detection and risk estimation. The Transpara AI detection system has been associated with short and longer-term risk of invasive BC. Whether Transpara can identify changes in the breast associated with tamoxifen (TAM) and aromatase inhibitor (AIs) treatment is unknown. We compare changes in the Transpara malignancy score in the unaffected breast of BC cases using TAM or AIs to those of untreated women without cancer. Cases and controls were sampled from the Mayo Clinic Rochester Breast Screening Practice. Eligible cases were women with breast cancer treated with TAM and/or AIs for at least 8 months who had full field digital (FFDM) mammograms within two years prior to start of treatment (pretreatment mammogram). Controls did not have BC, were not using TAM, AIs, or postmenopausal hormones, and had at least two FFDM screening mammograms. Transpara scores (1-10) were calculated for the contralateral breast in cases and a randomly selected side for controls. Two controls were matched to each case on menopausal status at both mammograms, time between mammograms (< 6 months) and baseline Transpara score (±2 units). Conditional logistic regression was used to assess the association of treatment (case (TAM/AIs) vs. control) with change in Transpara score between the pretreatment mammogram and the mammogram taken closest to 13 months later. Change was classified as having a decrease in Transpara score between the two mammograms vs. no change or an increase. Models were adjusted for age and Transpara score at pretreatment mammogram, and time between the two mammograms. We performed analyses on all women combined and stratified by time on treatment, menopausal status, and treatment type (TAM vs. AIs). A total of 134 cases were identified, with 53 on TAM (21 premenopausal and 32 postmenopausal) and 81 on AIs (postmenopausal); 268 controls were matched to the cases. Characteristics of the study population are summarized (Table). Most case mammograms were obtained within six months prior to treatment start [median 2.63 months (interquartile range (IQR), 1.62, 5.72)] and median time on TAM/AIs was 10.9 months (IQR, 8.03, 16.6) at time of second mammogram. The median absolute change in Transpara score between the two mammograms was similar in cases [0 (IQR, -2, 1)] vs. controls [0 (IQR, -1.25, 1)]; 40.3% of cases and 39.2% of controls had any decrease. Conditional logistic models found no evidence for an association between TAM/AIs and decrease in Transpara score [odds ratio (OR) = 1.07; 95% confidence interval (CI), 0.65, 1.78]. When stratified by median time on treatment (10.9 months), there was suggestion of a stronger association for those on treatment for a longer duration (OR = 1.21; 95% CI, 0.61, 2.40 for >10.9 months vs. OR = 0.96; 95% CI, 0.43, 2.13 for < 10.9 months) but these results were not statistically significant. Similarly, odds ratios were stronger in postmenopausal (OR = 1.18; 95% CI, 0.67, 2.07) compared to premenopausal (OR = 0.91; 95% CI, 0.24, 3.45) women, and for AIs (OR = 1.21, 95% CI, 0.64, 2.29) vs. TAM (OR = 0.93; 95% CI, 0.39, 2.22) but none of these results reached statistical significance. In summary, we found no association between treatment with TAM/AIs and change in Transpara score. However, our study had limited power. Also, longer time on treatment as well as investigation of AI-based scores trained directly to assess BC risk may be necessary to identify relevant changes. Citation Format: Yvonne Wang, Christopher Scott, Matthew Jensen, Dan Hursh, Kathleen Brandt, Aaron Norman, Fang Fang Wu, Stacey Winham, Karla Kerlikowske, Celine Vachon. Changes in an artificial intelligence model score for breast cancer detection (Transpara) with tamoxifen and aromatase inhibitor treatment [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO2-09-04.

PURPOSE Artificial intelligence (AI) algorithms improve breast cancer detection on mammography, but their contribution to long-term risk prediction for advanced and interval cancers is unknown. METHODS We identified 2,412 women with invasive breast cancer and 4,995 controls matched on age, race, and date of mammogram, from two US mammography cohorts, who had two-dimensional full-field digital mammograms performed 2-5.5 years before cancer diagnosis. We assessed Breast Imaging Reporting and Data System density, an AI malignancy score (1-10), and volumetric density measures. We used conditional logistic regression to estimate odds ratios (ORs), 95% CIs, adjusted for age and BMI, and C-statistics (AUC) to describe the association of AI score with invasive cancer and its contribution to models with breast density measures. Likelihood ratio tests (LRTs) and bootstrapping methods were used to compare model performance. RESULTS On mammograms between 2-5.5 years prior to cancer, a one unit increase in AI score was associated with 20% greater odds of invasive breast cancer (OR, 1.20; 95% CI, 1.17 to 1.22; AUC, 0.63; 95% CI, 0.62 to 0.64) and was similarly predictive of interval (OR, 1.20; 95% CI, 1.13 to 1.27; AUC, 0.63) and advanced cancers (OR, 1.23; 95% CI, 1.16 to 1.31; AUC, 0.64) and in dense (OR, 1.18; 95% CI, 1.15 to 1.22; AUC, 0.66) breasts. AI score improved prediction of all cancer types in models with density measures ( P LRT values < .001); discrimination improved for advanced cancer (ie, AUC for dense volume increased from 0.624 to 0.679, Δ AUC 0.065, P = .01) but did not reach statistical significance for interval cancer. CONCLUSION AI imaging algorithms coupled with breast density independently contribute to long-term risk prediction of invasive breast cancers, in particular, advanced cancer.

Abstract Breast parenchymal patterns on radiologic images are associated with breast cancer risk. Radiomic features have been proposed as quantitative measures of parenchymal patterns. We defined intrinsic imaging phenotypes of breast parenchymal patterns based on radiomic features extracted from full field digital mammography (FFDM) in breast screening populations and assessed whether these phenotypes are associated with breast cancer risk and masking. We selected 30,000 women with 4-view FFDM exams from Hologic machines from three institutions (Hospital of the University of Pennsylvania, Mayo Clinic, and San Francisco Mammography Registry), randomly split into a training (20,000 women) and test set (10,000 women). In total, 390 radiomic features were automatically extracted from each image using a validated software pipeline, standardized, and adjusted for site differences using ComBat. We used two methods, hierarchical clustering and Principal Components (PCs) analysis, to classify significant variation among the features in the training set and replicate among the test set. Next, we applied the replicated clusters and PCs to an independent nested case-control set [1082 invasive breast cancer (BC) cases (of which 151 were Black and 893 White women, 38 other race) matched to 2837 controls (411 Black and 2345 White women, 81 other race) on age, race, timing of images, and site]. We examined associations of the clusters and PCs with invasive breast cancer risk, as well as masking [defined as a false-negative (FN) screen (124 cases and 319 matched controls) and additionally the subset with symptomatic interval cancer (IC) within 12 months of negative screen (88 cases and 223 matched controls)] using conditional logistic regression. We evaluated their association with breast cancer alone, and with adjustment for age, body mass index (BMI) and breast density assessed by Breast Imaging Reporting and Data System (BI-RADS) using likelihood ratio tests. We estimated discrimination using area under the curve (AUC) and compared AUCs for models that included the radiomic clusters and PCs with the model that included only age, BMI and density. We also stratified analyses by race (Black/White). From hierarchical clustering, we defined six statistically significant phenotype clusters (each of at least 1000 women) in the training set which were replicated in the test set. For PC Analysis, we identified six PCs in the training set, explaining 85% of the variation in texture features and reproduced these in the test set. The six radiomic phenotype clusters (P&lt; 0.001) and six PCs (P&lt; 0.001) were both associated with invasive BC, including after adjusting for age, BMI, and density (cluster P=0.004; PCs P&lt; 0.001). Improvements in discrimination of invasive BC with inclusion of PCs or clusters were more pronounced among Black women (Table). Further, the PCs (P&lt; 0.001) and clusters (P&lt; 0.001) were significantly associated with FN overall and for symptomatic IC (PCs P&lt; 0.001; clusters P=0.001), but only PCs remained significant after adjusting for age, BMI and density (PCs P=0.004 for FN; PCs P=0.007 for symptomatic IC). Discrimination of masking also improved with inclusion of both clusters and PCs (Table). We identified reproducible radiomic phenotypes that are associated with invasive BC risk, above and beyond breast density with the strongest associations for invasive BC among Black women and symptomatic interval cancers. Citation Format: Stacey Winham, Anne Marie McCarthy, Aimilia Gastounioti, Christopher Scott, Aaron Norman, Walter C. Mankowski, Lauren Pantalone, Matthew Jensen, Eric A. Cohen, Hannah Horng, Kathleen Brandt, Emily F. Conant, Karla Kerlikowske, Despina Kontos, Celine Vachon. Radiomic phenotypes of breast texture and association with breast cancer risk and masking [abstract]. In: Proceedings of the 2022 San Antonio Breast Cancer Symposium; 2022 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2023;83(5 Suppl):Abstract nr GS4-06.

Digital mammography has seen an explosion in the number of radiomic features used for risk-assessment modeling. However, having more features is not necessarily beneficial, as some features may be overly sensitive to imaging physics (contrast, noise, and image sharpness). To measure the effects of imaging physics, we analyzed the feature variation across imaging acquisition settings (kV, mAs) using an anthropomorphic phantom. We also analyzed the intra-woman variation (IWV), a measure of how much a feature varies between breasts with similar parenchymal patterns—a woman’s left and right breasts. From 341 features, we identified “robust” features that minimized the effects of imaging physics and IWV. We also investigated whether robust features offered better case-control classification in an independent data set of 575 images, all with an overall BI-RADS® assessment of 1 (negative) or 2 (benign); 115 images (cases) were of women who developed cancer at least one year after that screening image, matched to 460 controls. We modeled cancer occurrence via logistic regression, using cross-validated area under the receiver-operating-characteristic curve (AUC) to measure model performance. Models using features from the most-robust quartile of features yielded an AUC = 0.59, versus 0.54 for the least-robust, with p < 0.005 for the difference among the quartiles.