Breast Cancer Risk Assessment Models Research Articles

Computational dissection of the regulatory mechanisms of aberrant metabolism in remodeling the microenvironment of breast cancer.

The composition of T cell subsets and tumor-specific T cell interactions within the tumor microenvironment (TME) contribute to the heterogeneity observed in breast cancer. Moreover, aberrant tumor metabolism is often intimately linked to dysregulated anti-tumor immune function of T cells. Identifying key metabolic genes that affect immune cell interactions thus holds promise for uncovering potential therapeutic targets in the treatment of breast cancer. This study leverages single-cell transcriptomic data from breast cancer to investigate tumor-specific T-cell subsets and their interacting subnetworks in the TME during cancer progression. We further assess the metabolic pathway activities of tumor-specifically activated T-cell subsets. The results reveal that metabolic pathways involved in insulin synthesis, secretion, degradation, as well as fructose catabolism, significantly influence multiple T cell interactions. By integrating the metabolic pathways that significantly up-regulate T cells in tumors and influence their interactions, we identify key abnormal metabolic genes associated with T-cell collaboration and further develop a breast cancer risk assessment model. Additionally, using gene expression profiles of prognosis-related genes significantly associated with aberrant metabolism and drug IC50 values, we predict targeted drugs, yielding potential candidates like GSK-J4 and PX-12. This study integrate the analysis of abnormal T-cell interactions and metabolic pathway abnormalities in the breast cancer TME, elucidating their roles in cancer progression and providing leads for novel breast cancer therapeutic strategies.

Abstract C150: Evaluating breast cancer risk by race and ethnicity using the Tyrer Cuzick and Gail models in the Komen Tissue Bank cohort

Abstract Introduction: Breast cancer is the second leading cause of cancer death in women after lung cancer in the United States. While early detection is critical to reducing morbidity and mortality, predication of individualized risk is important in prevention and could also reduce unnecessary costs related to screening. The Susan G. Komen Tissue Bank (KTB) at Indiana University is the only tissue bank in the world that provides breast tissue specimens donated by healthy individuals to further to breast cancer research; however, as donors are volunteers, it is not clear how representative they are with respect to breast cancer risk compared to the general population. Methods: In this study, we calculated Breast Cancer Risk Assessment scores using Tyrer Cuzick (TCZ) and the Breast Cancer Risk Assessment Tool (The Gail Model, GM) in the KTB cohort by race and ethnicity. Recruited between 2007 and 2023, the participants completed a questionnaire encompassing their demographics, personal health and reproductive history, family history, and modifiable risk factors like smoking and alcohol use. In both models, participants with a lifetime risk of breast cancer of more than 15% were classified as having higher than average risk, while those with a risk of less than 15% were classified as having an average risk. Data analysis was performed using SAS v.9.4 software. Statistical significance was determined by p-values <0.05. Results: Out of 5118 participants, 206 with prior breast cancer and 37 male donors were excluded, resulting in a final sample of 4875 participants (aged 18–90) for analysis. The cohort was predominantly non-Hispanic white (NHW, 69.73%), followed by non-Hispanic black (NHB, 16.7%), Hispanic (8. 6%), and Asian individuals (3.3%), with 1.8% reporting multiracial or other race. The GM consistently categorized fewer women as having higher than average risk compared to the TZM. Across both TCZ and GM, the percentages of individuals classified as higher than average risk was greatest for NHW women, with 29.2% being categorized as higher than average risk in the TCZ, compared to 21.0% in the GM. The greatest discordance between the models was seen in NHB women, with the TCZ estimating 25.2% were at greater than average risk, and the GM only reporting 3.7% to be at greater than average risk. Overall, risk assessment scores varied across racial and ethnic groups (p- value<0.001). Factors driving these differences will be presented in additional analyses. Conclusion: There are known limitations in breast cancer risk assessment models by race and ethnicity that were confirmed in the women who comprise the KTB cohort. Donors to the KTB are generally representative of the US population with respect to breast cancer risk. As the KTB cohort continues to grow and mature, it can be utilized to assess breast cancer risk models for various racial and ethnic groups to assist in the identification of high-risk individuals that may benefit from prevention strategies and targeted screening. Citation Format: Vidya Patil, Michele L. Cote, Jenny Cui. Evaluating breast cancer risk by race and ethnicity using the Tyrer Cuzick and Gail models in the Komen Tissue Bank cohort [abstract]. In: Proceedings of the 17th AACR Conference on the Science of Cancer Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; 2024 Sep 21-24; Los Angeles, CA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2024;33(9 Suppl):Abstract nr C150.

Feasibility of risk assessment for breast cancer molecular subtypes

PurposeFew breast cancer risk assessment models account for the risk profiles of different tumor subtypes. This study evaluated whether a subtype-specific approach improves discrimination.MethodsAmong 3389 women who had a screening mammogram and were later diagnosed with invasive breast cancer we performed multinomial logistic regression with tumor subtype as the outcome and known breast cancer risk factors as predictors. Tumor subtypes were defined by expression of estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) based on immunohistochemistry. Discrimination was assessed with the area under the receiver operating curve (AUC). Absolute risk of each subtype was estimated by proportioning Gail absolute risk estimates by the predicted probabilities for each subtype. We then compared risk factor distributions for women in the highest deciles of risk for each subtype.ResultsThere were 3,073 ER/PR+ HER2 − , 340 ER/PR +HER2 + , 126 ER/PR−ER2+, and 300 triple-negative breast cancers (TNBC). Discrimination differed by subtype; ER/PR−HER2+ (AUC: 0.64, 95% CI 0.59, 0.69) and TNBC (AUC: 0.64, 95% CI 0.61, 0.68) had better discrimination than ER/PR+HER2+ (AUC: 0.61, 95% CI 0.58, 0.64). Compared to other subtypes, patients at high absolute risk of TNBC were younger, mostly Black, had no family history of breast cancer, and higher BMI. Those at high absolute risk of HER2+ cancers were younger and had lower BMI.ConclusionOur study provides proof of concept that stratifying risk prediction for breast cancer subtypes may enable identification of patients with unique profiles conferring increased risk for tumor subtypes.

Deep learning vs traditional breast cancer risk assessment models: Are we offering supplemental services to those who benefit the most?

1556 Background: Mammography-based deep learning (DL) risk stratification models improve discriminatory accuracy in predicting future breast cancer compared to traditional risk models; however, decisions for and insurance payment of supplemental services such as MRI are driven by traditional scores. The purpose of this study was to compare risk scores and cancer detection rates in patients identified as increased risk by DL vs traditional models in a large screening mammography cohort. Methods: This multisite study included consecutive patients ≥40y undergoing routine bilateral screening mammography from 9/2017 to 1/2022 at five facilities. Tyrer-Cuzick version 8 (TC) and National Cancer Institute Breast Cancer Risk Assessment Tool (BCRAT) 5-year and lifetime models and a DL 5-year model were used to assess risk. The following thresholds were used to define intermediate risk: ≥1.67% for TC and BCRAT 5-year, 2.2 for DL model and high risk: ≥20% for TC and BCRAT lifetime, ≥6.0 for DL model. Patients were included if all risk scores were available. Patient demographics were retrieved from electronic medical records and cancer outcomes through regional tumor registry linkage. The proportion of increased-risk patients and cancer detection rates (CDR [cancers per 1000 women screened]) across models were compared using Pearson’s Chi-square test. Results: 148476 exams in 69464 patients (mean age 58.0y [IQR: 50.0-67.0]) were performed. 81.7% (119386/146075) of exams were in White patients and 18.3% (26689/146075) in races other than White. CDR of those classified as intermediate risk by DL was 7.1 (450/63109) vs 4.7 (320/68545; P<0.001) by TC and 4.2 (300/70831; P<0.001) by BCRAT, and as high risk was 20.6 (305/14772) by DL vs 3.9 (64/16442; P<0.001) by TC and 5.0 (29/5782; P<0.001) by BCRAT. 13.0% (19248/148476) were classified as intermediate risk by DL but not by TC or BCRAT, with a CDR of 6.0 in this cohort. 32.0% (47435/148476) were classified as intermediate risk by TC or BCRAT but not by DL with a CDR of 1.1. 8.5% (12577/148476) were classified as high risk by DL but not by TC or BCRAT with a CDR of 20.8. 11.1% (16451/148476) were classified as high risk by TC or BCRAT but not by DL with a CDR of 1.9. Conclusions: A significant proportion of patients identified as increased risk by DL model are not assessed as increased risk by commonly used risk models, effectively excluding them from accessing potentially life-saving supplemental services. DL scores met the American College of Radiology's acceptable CDR on screening mammography (≥2.5) regardless of traditional scores, however, elevated traditional scores in the absence of elevated DL scores did not meet acceptable standards. Risk assessment guidelines incorporating DL models are necessary to capture patients who benefit the most.

Personalized survivorship care: Routine breast cancer risk assessment in the gynecologic oncology clinic

IntroductionGynecologic and breast cancers share several risk factors. Breast cancer risk assessment tools can identify those at elevated risk and allow for enhanced breast surveillance and chemoprevention, however such tools are underutilized. We aim to evaluate the use of routine breast cancer risk assessment in a gynecologic oncology clinic. MethodsA patient-facing web-based tool was used to collect personal and family history and run four validated breast cancer risk assessment models (Tyrer-Cuzick (TC), Gail, BRCAPRO, and Claus) in a gynecologic oncology clinic. We evaluated completion of the tools and identification of patients at elevated risk for breast cancer using the four validated models. ResultsA total of 99 patients were included in this analysis. The BRCAPRO model had the highest completion rate (84.8%), followed by the TC model (74.7%), Gail model (74.7%), and the Claus model (52.1%). The TC model identified 21.6% of patients completing the model as having ≥20% lifetime risk of breast cancer, compared to 6.8% by the Gail model, and 0% for both the BRCAPRO and Claus models. The Gail model identified 52.5% of patients as having ≥1.67% 5-year risk of breast cancer. Among patients identified as high-risk for breast cancer and eligible for screening, 9/9 (100%) were referred to a high-risk breast clinic. ConclusionAmong patients that completed the TC breast cancer risk assessment in a gynecologic oncology clinic, approximately 1 in 5 were identified to be at significantly elevated lifetime risk for breast cancer. The gynecologic oncologist's office might offer a convenient and feasible setting to incorporate this risk assessment into routine patient care, as gynecologic oncologists often have long-term patient relationships and participate in survivorship care.

Next Top Model: An Overview of Breast Cancer Risk Assessment Models

Next Top Model: An Overview of Breast Cancer Risk Assessment Models

Deep Learning Approaches with Digital Mammography for Evaluating Breast Cancer Risk, a Narrative Review.

Breast cancer remains the leading cause of cancer-related deaths in women worldwide. Current screening regimens and clinical breast cancer risk assessment models use risk factors such as demographics and patient history to guide policy and assess risk. Applications of artificial intelligence methods (AI) such as deep learning (DL) and convolutional neural networks (CNNs) to evaluate individual patient information and imaging showed promise as personalized risk models. We reviewed the current literature for studies related to deep learning and convolutional neural networks with digital mammography for assessing breast cancer risk. We discussed the literature and examined the ongoing and future applications of deep learning techniques in breast cancer risk modeling.

22 (PB-022) Poster Spotlight - An individualized breast cancer risk assessment model for women attending screening in BreastScreen Norway

Background: The results of the effectiveness of population-based screening are controversial in terms of the balance between mortality reduction and adverse effects. In order to improve it, studies have proposed personalized screening strategies based on woman’s individual breast cancer (BC) risk. There is, therefore, a need to create individual risk prediction models through the analysis of large population-based databases. We developed a model that could be used to classify women targeted for mammography screening according to individual BC risk.

Clinical tools and counseling considerations for breast cancer risk assessment and evaluation for hereditary cancer risk.

Breast cancer is a complex and heterogeneous disease. Unfortunately, it is the most common malignancy diagnosed in women in the USA, with 281,550 new cases of invasive breast cancer and 49,290 new cases of noninvasive breast cancer are diagnosed per year. In England, it is currently estimated that approximately 1 in 7 (14%) women will be diagnosed with breast cancer in their lifetime. In the UK in 2017, 54,700 women and 390 men were diagnosed with breast cancer. The risk of breast cancer is influenced by many factors, including but not limited to age, family history, reproductive history, hormonal exposure, proliferative breast lesions, physical activity, alcohol use, tobacco use, breast density, and environmental exposures. Breast cancer risk assessment is a critical part of public health. By identifying women at high risk for breast cancer, personalized recommendations can be deployed with regards to modes of screening, the age to initiate breast screening, and the frequency for completing such screens. In addition, breast cancer risk assessment can assist in determining a woman's eligibility for interventions to reduce risk, either through the use of chemoprevention medications or through surgical means with risk-reducing bilateral mastectomy. This chapter summarizes breast cancer risk assessment models and discusses interventions to reduce breast cancer risk to aid in reducing morbidity and mortality from breast cancer.

Beyond the AJR: Validation of an Artificial Intelligence Breast Cancer Risk Assessment Model Across Diverse International Cohorts

Beyond the <i>AJR</i>: Validation of an Artificial Intelligence Breast Cancer Risk Assessment Model Across Diverse International Cohorts

Prediction of Short-Term Breast Cancer Risk with Fusion of CC- and MLO-Based Risk Models in Four-View Mammograms.

This study performed and assessed a novel program to improve the accuracy of short-term breast cancer risk prediction by using information from craniocaudal (CC) and mediolateral-oblique (MLO) views of two breasts. An age-matched dataset of 556 patients with at least two sequential full-field digital mammography examinations was applied. In the second examination, 278 cases were diagnosed and pathologically verified as cancer, and 278 were negative, while all cases in the first examination were negative (not recalled). Two generalized linear-model-based risk prediction models were established with global- and local-based bilateral asymmetry features for CC and MLO views first. Then, a new fusion risk model was developed by fusing prediction results of the CC- and MLO-based risk models with an adaptive alpha-integration-based fusion method. The AUC of the fusion risk model was 0.72 ± 0.02, which was significantly higher than the AUC of CC- or MLO-based risk model (P < 0.05). The maximum odds ratio for CC- and MLO-based risk models were 8.09 and 5.25, respectively, and increased to 11.99 for the fusion risk model. For subgroups of patients aged 37-49years, 50-65years, and 66-87years, the AUCs of 0.73, 0.71, and 0.75 for the fusion risk model were higher than AUC for CC- and MLO-based risk models. For the BIRADS 2 and 3 subgroups, the AUC values were 0.72 and 0.71 respectively for the fusion risk model which were higher than the AUC for the CC- and MLO-based risk models. This study demonstrated that the fusion risk model we established could effectively derive and integrate supplementary and useful information extracted from both CC and MLO view images and adaptively fuse them to increase the predictive power of the short-term breast cancer risk assessment model.

Abstract P2-10-06: Validation of abridged breast cancer risk assessment model for the general population

Abstract The ability to efficiently and accurately risk stratify women can lead to improved risk-reduction measures. Actionable risk thresholds for both screening and risk-reduction guidance exist; however, despite several clinical risk models currently in existence, the utilization of such models is underwhelming. We have developed and validated an abridged clinical risk model, COSMO (Clinical pOlygenic denSity faMily mOdel), incorporating three of the most relevant epidemiological risk factors associated with sporadic and familial breast cancer in a format that is designed for “ease of use”. Polygenic risk is recognized as a promising epidemiological risk factor however, its value is greater when integrated with other phenotypic risk factors including 1st and 2nd family history, breast density menopausal status, BMI and age. We used 200,009 active UK Biobank (UKB) members who were female, genetically Caucasian, had SNP data for 313 breast cancer associated variants and aged 40 to 69 years at baseline assessment date. We used age and date of baseline assessment to calculate censoring age for those who had died before completing 5 years of follow-up and to determine the date at which 5 years of follow-up was completed for the others. We used both self-reported and linked cancer registry data (ICD9 code 174 or ICD10 code C50) to determine the earliest diagnosis of invasive breast cancer for each affected woman. In terms of discrimination, COSMO (AUC=0.651; 95% CI 0.642, 0.661) performed better than the Gail model (AUC=0.567; 95% CI=0.557, 0.567); P value for difference &amp;lt;0.0001. Overall calibration showed the number of breast cancers expected using COSMO was slightly higher than the number observed (SIR=0.94; 95% CI=0.91, 0.97), while for the Gail model, the number of breast cancers was lower than the number observed (SIR=1.07; 95% CI=1.04, 1.11). The modest over estimation may be explained by the “healthy volunteer” selection bias of the UKB. Importantly, COSMO significantly improved predicted risk classification of women compared to the Gail when stratified according to actionable thresholds. (Table) Limitations to the validation include the lack of mammographic breast density data in the UKB—a critical factor in COSMO. To this end, we have an ongoing validation in a Nurses’ Health Study population (1,269 cases and 1,812 controls) that includes breast density data. Ongoing model comparison studies will confirm the equivalence of COSMO to standard clinical models. If performance is equivalent, COSMO has an advantage in clinical implementation because the abridged model can be less time consuming to implement in a general practice or screening clinic setting due to the truncated pedigree. Furthermore, as multiethnic cohort data access continues to improve, the epidemiological components can be recalibrated for application across any ethnicity. COSMO integrates the most impactful epidemiological risk factors, is well calibrated for the general population and provides improved discrimination over a standard risk model. General population risk assessment like COSMO can lead to the identification of more women who would benefit from clinical intervention based on current risk-reduction guidelines. Beyond that, incidental by products of a general population risk assessment include promoting breast cancer awareness, empowering women to understand their own risk, and encouraging all women to consider their lifestyle habits. Citation Format: Erika Spaeth, Gillian Dite, Richard Allman. Validation of abridged breast cancer risk assessment model for the general population [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P2-10-06.

Are English-language online patient education materials related to breast cancer risk assessment understandable, readable, and actionable?

PurposeTo evaluate the readability, understandability, and actionability of online patient education materials (OPEM) related to breast cancer risk assessment. Material and methodsWe queried seven English-language search terms related to breast cancer risk assessment: breast cancer high-risk, breast cancer risk factors, breast cancer family history, BRCA, breast cancer risk assessment, Tyrer-Cuzick, and Gail model. Websites were categorized as: academic/hospital-based, commercial, government, non-profit or academic based on the organization hosting the site. Grade-level readability of qualifying websites and categories was determined using readability metrics and generalized estimating equations based on written content only. Readability scores were compared to the recommended parameters set by the American Medical Association (AMA). Understandability and actionability of OPEM related to breast cancer high-risk were evaluated using the Patient Education Materials Assessment Tool (PEMAT) and compared to criteria set at ≥70%. Descriptive statistics and inter-rater reliability analysis were utilized. Results343 websites were identified, of which 162 met study inclusion criteria. The average grade readability score was 12.1 across all websites (range 10.8–13.4). No website met the AMA recommendation. Commercial websites demonstrated the highest overall average readability of 13.1. Of the 26 websites related to the search term breast cancer high-risk, the average understandability and actionability scores were 62% and 34% respectively, both below criteria. ConclusionsOPEM on breast cancer risk assessment available to the general public do not meet criteria for readability, understandability, or actionability. To ensure patient comprehension of medical information online, future information should be published in simpler, more appropriate terms.

Abstract PD10-09: Development of a breast cancer risk assessment model for ATM mutation carriers incorporating tyrer-cuzick and a polygenic risk score (PRS)

Abstract Background: Germline pathogenic variants (PVs) in the moderate penetrance ATM gene confer a roughly 2-fold increased risk for breast cancer. Currently, any woman with an ATM PV meets the &amp;gt;20% lifetime risk threshold for consideration of relatively aggressive screening recommendations, which include initiating screening at younger ages and consideration of breast MRI. We and others have previously shown that breast cancer risks for women with inherited PVs in many hereditary breast cancer genes can be adjusted using PRS data, breast cancer family history, and other clinical information. Herein we show the development of a comprehensive breast cancer risk model for ATM PV carriers incorporating a previously described 86-variant PRS along with family history/clinical information captured by Tyrer-Cuzick V7.02. Methods: This IRB-approved study included de-identified clinical records from 353,809 women of European ancestry who were tested clinically for hereditary cancer risk with a multi-gene panel. Model development analyzed ATM PV carriers (N=2,666) and women negative for breast cancer gene PVs (N=351,143) who were tested between September 2013 and November 2019 (July 2019 for non-carriers). Women with the unusually high risk ATM c.7271 allele were excluded from analysis. Risk estimates incorporating ATM, PRS, and Tyrer-Cuzick were calculated using a fixed-stratified method that accounted for correlations between risk factors in a manner equivalent to multivariable co-estimation. Risk stratification was assessed in an independent cohort of ATM carriers (N=272) who were tested after November 2019 and were not included in model development. Results: We detected significant positive correlation of ATM status with breast cancer family history (p=1.6×10−7). Within ATM PV carriers, we observed positive yet non-significant (at alpha &amp;lt;0.05) correlation between PRS and breast cancer family history (p=0.10); joint effects were co-estimated using the fixed-stratified method. After adjusting for multiple testing, we found no evidence of interaction of ATM status with clinical factors, or PRS with clinical factors within ATM carriers. In an independent cohort, 30.1% of ATM carriers were categorized as having low breast cancer risk (&amp;lt;20%), 58.5% as moderate risk (20-50%), and 11.4% as high risk (&amp;gt;50%). Conclusions: In ATM PV carriers, our comprehensive model allowed for differentiation of ATM PV carriers into low, moderate, and high breast cancer risk categories. Precision breast cancer risk estimation may inform individualized clinical screening and prevention strategies. Citation Format: Shannon Gallagher, Elisha Hughes, Eric Rosenthal, Allison W. Kurian, Susan Domchek, Judy Garber, Braden Probst, Brian Morris, Placede Tshiaba, Benjamin Roa, Thomas P. Slavin, Susanne Wagner, Jeffrey N. Weitzel, Alexander Gutin, Jerry S. Lanchbury, Mark E. Robson. Development of a breast cancer risk assessment model for ATM mutation carriers incorporating tyrer-cuzick and a polygenic risk score (PRS) [abstract]. In: Proceedings of the 2020 San Antonio Breast Cancer Virtual Symposium; 2020 Dec 8-11; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2021;81(4 Suppl):Abstract nr PD10-09.

Abstract PS7-02: The relationship of established breast cancer risk factors with breast cancer subtypes

Abstract Background: Molecular characterization of breast tumors has revealed four subtypes which differ in expression of estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2). Breast cancer subtypes have different prognosis and unique risk factors. Breast cancer risk assessment models mainly reflect risk of ER/PR+ HER2- tumors, the most common subtype, and may not reflect risk of other subtypes. The study objective was to compare associations of established breast cancer risk factors across invasive breast cancer subtypes. Methods: The study population included women aged 40-84 years who had a screening mammogram at Massachusetts General Hospital, Newton Wellesley Hospital, or the University of Pennsylvania from 2006-2015. Patients completed a risk factor questionnaire and additional risk factors were ascertained from clinical records. Women with prior breast cancer, breast implants, or BRCA1/2 mutations were excluded. Women diagnosed with breast cancer within 6 months were excluded to remove those with cancer at the time of risk assessment. Tumor characteristics were obtained from linkage with hospital and state cancer registries. For invasive tumors, subtype was defined based on immunohistochemistry as ER and/or PR+ HER2-, ER and/or PR+HER2+, ER and PR- HER2+, or ER and PR and HER2- (triple negative breast cancer, TNBC). Cox proportional hazards models were used to estimate hazard ratios (HR) and 95% confidence intervals (CI) for breast cancer using time from mammogram to cancer diagnosis or censoring. Separate Cox models were fit for each subtype, and patients who developed DCIS or another subtype, died, or were alive and cancer free on December 31, 2017 were censored. Age, biopsy history, atypical hyperplasia, age at menarche, age at first live birth, family history, race, BMI, and breast density at the time of screening were included in the models. Results: The study population (N=197,836) had a mean age of 54 years and 74% of patients were white, 15% were Black, and 11% were other races. During a mean follow-up of 6.3 years, 4,510 (2.3%) women developed breast cancer. Of these cancers, 1068 (24%) were DCIS. Of the invasive cancers, 2675 (77%) were ER/PR+HER2-, 290 (8%) were ER/PR+HER2+, 108 (3%) were ER/PR-HER2+, 264 (8%) were TNBC and 105 (3%) had missing subtype. For ER/PR+HER2- cancers, all risk factors were consistent with the literature and statistically significant. Breast density was associated with increased risk of all four subtypes compared to women with less dense breasts. Atypical hyperplasia was strongly associated with HER2+ cancers (HR=2.97 CI 1.63-5.40 p&amp;lt;0.01), less strongly associated with ER/PR+HER2- cancers, and not significantly associated with TNBC. Black women had higher risk of TNBC than white women (HR=2.61 CI 1.91-3.57 p&amp;lt;0.01). Conclusion Our results highlight both similarities and heterogeneity in risk factors across breast cancer subtypes. Prior diagnosis of atypical hyperplasia was more strongly associated with HER2+ compared to HER2- tumors. While it is well known that Black women have higher risk of TNBC, it is striking that the more than two-and-a-half-fold increased risk persisted even with comprehensive adjustment for breast cancer risk factors in a screened population. These results suggest that additional factors, such as genetics, biomarkers, and environmental exposures should be included in risk assessment to better capture risk of less common breast cancer subtypes such as TNBC. Risk factors for breast cancer subtypes among 197,836 women undergoing screening mammography*ER/PR+ HER2- N=2674ER/PR+ HER2+ N=290ER/PR-HER2+ N=108ER/PR/HER2- N=264HR,95% CIpHR,95% CIpHR,95% CIpHR,95% CIpBlack vs. white unadjusted0.67,0.58-0.77&amp;lt;0.010.73,0.48-1.090.120.91,0.49-1.680.772.61,1.96-3.46&amp;lt;0.01Black vs. white multivariate*0.72,0.63-0.83&amp;lt;0.010.75,0.49-1.150.191.23,0.65-2.330.532.61,1.91-3.57&amp;lt;0.01Atypical Hyperplasia*1.38,1.02-1.870.042.77,1.42-5.42&amp;lt;0.013.89,1.03-14.70.040.43,0.06-3.140.401 FDR** vs. none1.46,1.32-1.63&amp;lt;0.011.16,0.82-1.650.391.93,1.18-3.160.011.11,0.75-1.630.602 FDR** vs. none2.12,1.67-2.71&amp;lt;0.011.17,0.44-3.150.751.90,0.47-7.760.372.81,1.44-5.49&amp;lt;0.01BMI ≥25 vs. &amp;lt;25 kg/m2*1.37,1.25-1.50&amp;lt;0.011.35,1.04-1.750.031.07,0.71-1.640.741.29,0.97-1.730.08Dense vs. non-dense breasts1.55,1.42-1.69&amp;lt;0.011.75,1.34-2.29&amp;lt;0.011.97,1.25-3.10&amp;lt;0.011.65,1.26-2.17&amp;lt;0.01*Adjusted for all factors in the table and additionally age, biopsy, age at menarche, age at first live birth**FDR= first degree relative with breast cancer Citation Format: Anne Marie McCarthy, Tara Friebel, Wei He, Michaela Welch, Sarah Ehsan, Kevin Hughes, Alan Semine, Jinbo Chen, Despina Kontos, Susan Domchek, Emily Conant, Aditya Bardia, Constance Lehman, Katrina Armstrong. The relationship of established breast cancer risk factors with breast cancer subtypes [abstract]. In: Proceedings of the 2020 San Antonio Breast Cancer Virtual Symposium; 2020 Dec 8-11; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2021;81(4 Suppl):Abstract nr PS7-02.

Association of the Differences in Average Glandular Dose with Breast Cancer Risk

Objectives. To compare the differences in normalized average glandular dose (NAGD) between the breasts of healthy subjects and those of cancer patients and to determine if the NAGD difference is associated with breast cancer risk and improves breast cancer classification. Materials and Methods. Craniocaudal view and mediolateral view full-field digital mammography (FFDM) images were obtained from 1682 healthy subjects whose breasts were categorized as Breast Imaging-Reporting and Data System (BI-RADS) I or II and from 811 biopsy-confirmed unilateral breast cancer patients whose breasts on the contralateral side were category I or II. Both populations were randomized into training and test sets. Multivariate logistic regression analysis was used to build the breast cancer risk assessment model, and the area under the receiver operating characteristic curve ( A z ) was used to evaluate the model. Twenty-two breast cancer patients who were originally categorized as BI-RADS I or II for both breasts, but were diagnosed with unilateral biopsy-confirmed breast cancer subsequently, were included to validate the model. Results. The NAGD differences in both FFDM images between tumor-bearing breasts and the healthy breasts of patients were significantly higher than those in healthy subjects ( P &lt; 0.001 ). The model with NAGD differences had a higher A z value than the model without NAGD differences. While there was no NAGD differences between originally healthy breasts of breast cancer patients, significant NAGD differences between now tumor-bearing breasts and the then previously healthy breasts were found in both FFDM images. Conclusions. NAGD differences between both breasts can be included in the breast cancer risk assessment model to evaluate breast cancer risk.

Abstract P6-08-07: Polygenic breast cancer risk modification in carriers of high and intermediate risk gene mutations

Abstract Background: The incorporation of polygenic scores into breast cancer risk assessment models has led to improved risk management for women of European ancestry. However, few studies have examined the extent to which polygenic scores modify risk for carriers of breast cancer susceptibility gene mutations. Current literature suggests that a 77-SNP polygenic score developed from large, mainly non-carrier general population studies can stratify breast cancer risk for carriers of the CHEK2 founder mutation 1100delC with the same effect size observed for non-carriers. Consequently, several groups have modified risk estimates for carriers of other pathogenic variants by applying published effect estimates from non-carriers. In a recent study, a similar 88-SNP polygenic score showed reduced effectiveness in BRCA1 and BRCA2 carriers compared to non-carriers. This highlights the need for more information on polygenic score performance among women who carry pathogenic variants in ATM or PALB2, or in CHEK2 other than the 1100delC founder mutation. We previously described an 86-SNP polygenic residual risk score (RRS) that characterizes genetic risk which is not due to high- or moderate-penetrance breast cancer gene mutations. Here, we present pre-specified independent validation of the RRS for BRCA1, BRCA2, CHEK2, ATM and PALB2 carriers.Methods: This IRB-approved study included 152,172 women of European ancestry who were tested clinically for hereditary cancer risk with a multi-gene panel. The RRS was evaluated separately for carriers of BRCA1 (N=2,249), BRCA2 (N=2,638), CHEK2 (N=2,564), ATM (N=1,445) and PALB2 (N=906), and for mutation-free women (N=141,160). Clinical information was obtained from test requisition forms. Multivariable logistic regression was used to examine the association of RRS with invasive breast cancer after accounting for age and family cancer history. Effect sizes, expressed as standardized odds ratios (ORs) with 95% confidence intervals (CIs), were assessed for carriers of each gene and for non-carriers. Results: RRS was strongly associated with breast cancer risk in each of the BRCA1, BRCA2, CHEK2, ATM and PALB2 carrier populations (p&amp;lt;10−4). Risk modification in CHEK2 carriers (OR 1.49; CI 1.36-1.64) was consistent with that observed in non-carriers (OR 1.47; CI 1.45-1.49). CHEK2 carriers in the highest RRS quintile had approximately 1.7-fold higher risk than the middle quintile (OR 1.67; CI 1.26 - 2.20), and the lowest quintile had risk lowered by factor of 0.59 (OR 0.59; CI 0.44 - 0.80). We observed reduced modification similar to existing literature for carriers of BRCA1 (OR 1.20; CI 1.10-1.32) and BRCA2 (OR 1.23; CI 1.12-1.34). Odds ratios for ATM (1.37; CI 1.21-1.55) and PALB2 (1.34; CI 1.16-1.55) had wide confidence intervals that overlapped those for the other carrier populations. Discussion: The RRS significantly modifies risk for carriers of BRCA1, BRCA2, CHEK2, ATM and PALB2 mutations albeit with apparently different effect sizes. These results indicate that risk estimates employing polygenic scores may need to be derived separately for each gene. Given the modest effect on BRCA1 and BRCA2 risk and the substantial risks conferred by these genes even with lower RRS, the clinical implications of polygenic modification in this setting are yet to be determined. In CHEK2 carriers, we observed strong polygenic modification similar to that achieved for non-carriers. For ATM and PALB2, larger studies are needed to refine polygenic effect size estimates. Comprehensive risk assessment combining RRS with family cancer history and other risk factors has potential to refine risk estimates and facilitate decision-making for women with mutations in moderate penetrance genes. Citation Format: Shannon Gallagher, Elisha Hughes, Susanne Wagner, Placede Tshiaba, Eric Rosenthal, Benjamin B. Roa, Allison Kurian, Susan Domchek, Judy Garber, Johnathan M. Lancaster, Jeffrey Weitzel, Alexander Gutin, Jerry S. Lanchbury, Mark Robson. Polygenic breast cancer risk modification in carriers of high and intermediate risk gene mutations [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr P6-08-07.

Breast Cancer Risk Assessment and Chemoprevention Use Among Veterans Affairs Primary Care Providers: A National Online Survey.

Breast cancer is the most common cancer diagnosed among women and the second most common cause of cancer death among women. There are ways to reduce a woman's risk of breast cancer; however, most eligible women in the United States are neither offered personalized screening nor chemoprevention. Surveys have found that primary care providers are largely unaware of breast cancer risk assessment models or chemoprevention. This survey aims to investigate Veterans Health Administration primary care providers' comfort level, practice patterns, and knowledge of breast cancer risk assessment and chemoprevention. An online, Research Electronic Data Capture-generated survey was distributed to VHA providers in internal medicine, family medicine, and obstetrics/gynecology. Survey domains were provider demographics, women's health experience, comfort level, practice patterns, barriers to using risk models and chemoprevention, and knowledge of chemoprevention. Of the 167 respondents, 33.1% used the Gail model monthly or more often and only 2.4% prescribed chemoprevention in the past 2 years. Most VHA primary care providers did not answer chemoprevention knowledge questions correctly. Designated women's health providers were more comfortable with risk assessment (P<0.018) and chemoprevention (P<0.011) and used both breast cancer risk models (P<0.0045) and chemoprevention more often (P<0.153). Reported barriers to chemoprevention were lack of education and provider time. VHA providers and women Veterans would benefit from a system to ensure that women at increased risk of breast cancer are identified with risk modeling and that risk reduction options, such as chemoprevention, are offered when appropriate. VHA providers requested risk reduction education, which could improve primary care provider comfort level with chemoprevention.

Performance of Breast Cancer Risk-Assessment Models in a Large Mammography Cohort.

Several breast cancer risk-assessment models exist. Few studies have evaluated predictive accuracy of multiple models in large screening populations. We evaluated the performance of the BRCAPRO, Gail, Claus, Breast Cancer Surveillance Consortium (BCSC), and Tyrer-Cuzick models in predicting risk of breast cancer over 6 years among 35921 women aged 40-84 years who underwent mammography screening at Newton-Wellesley Hospital from 2007 to 2009. We assessed model discrimination using the area under the receiver operating characteristic curve (AUC) and assessed calibration by comparing the ratio of observed-to-expected (O/E) cases. We calculated the square root of the Brier score and positive and negative predictive values of each model. Our results confirmed the good calibration and comparable moderate discrimination of the BRCAPRO, Gail, Tyrer-Cuzick, and BCSC models. The Gail model had slightly better O/E ratio and AUC (O/E = 0.98, 95% confidence interval [CI] = 0.91 to 1.06, AUC = 0.64, 95% CI = 0.61 to 0.65) compared with BRCAPRO (O/E = 0.94, 95% CI = 0.88 to 1.02, AUC = 0.61, 95% CI = 0.59 to 0.63) and Tyrer-Cuzick (version 8, O/E = 0.84, 95% CI = 0.79 to 0.91, AUC = 0.62, 95% 0.60 to 0.64) in the full study population, and the BCSC model had the highest AUC among women with available breast density information (O/E = 0.97, 95% CI = 0.89 to 1.05, AUC = 0.64, 95% CI = 0.62 to 0.66). All models had poorer predictive accuracy for human epidermal growth factor receptor 2 positive and triple-negative breast cancers than hormone receptor positive human epidermal growth factor receptor 2 negative breast cancers. In a large cohort of patients undergoing mammography screening, existing risk prediction models had similar, moderate predictive accuracy and good calibration overall. Models that incorporate additional genetic and nongenetic risk factors and estimate risk of tumor subtypes may further improve breast cancer risk prediction.

A new automated method to evaluate 2D mammographic breast density according to BI-RADS® Atlas Fifth Edition recommendations.

Radiologists' visual assessment of breast mammographic density (BMD) is subject to inter-observer variability. We aimed to develop and validate a new automated software tool mimicking expert radiologists' consensus assessments of 2D BMD, as per BI-RADS V recommendations. The software algorithm was developed using a concept of Manhattan distance to compare a patient's mammographic image to reference mammograms with an assigned BMD category. Reference databases were built from a total of 2289 pairs (cranio-caudal and medio-lateral oblique views) of 2D full-field digital mammography (FFDM). Each image was independently assessed for BMD by a consensus of radiologists specialized in breast imaging. A validation set of additional 800 image pairs was evaluated for BMD both by the software and seven blinded radiologists specialized in breast imaging. The median score was used for consensus. Software reproducibility was assessed using FFDM image pairs from 214 patients in the validation set to compare BMD assessment between left and right breasts. The software showed a substantial agreement with the radiologists' consensus (unweighted κ = 0.68, 95% CI 0.64-0.72) when considering the four breast density categories, and an almost perfect agreement (unweighted κ = 0.84, 95% CI 0.80-0.88) when considering clinically significant non-dense (A-B) and dense (C-D) categories. Correlation between left and right breasts was high (rs = 0.87; 95% CI 0.84-0.90). BMD assessment by the software was strongly correlated to radiologists' consensus assessments of BMD. Its performance should be compared to other methods, and its clinical utility evaluated in a risk assessment model. • A new software tool assesses breast density in a standardized way. • The tool mimics radiologists' clinical assessment of breast density. • It may be incorporated in a breast cancer risk assessment model.