BI-RADS Density Categories Research Articles

Concordant and discordant breast density patterns by different approaches for assessing breast density and breast cancer risk.

To examine the discrepancy in breast density assessments by radiologists, LIBRA software, and AI algorithm and their association with breast cancer risk. Among 74,610 Korean women aged ≥ 34years, who underwent screening mammography, density estimates obtained from both LIBRA and the AI algorithm were compared to radiologists using BI-RADS density categories (A-D, designating C and D as dense breasts). The breast cancer risks were compared according to concordant or discordant dense breasts identified by radiologists, LIBRA, and AI. Cox-proportional hazards models were used to determine adjusted hazard ratios (aHRs) [95% confidence intervals (CIs)]. During a median follow-up of 9.9years, 479 breast cancer cases developed. Compared to the reference non-dense breast group, the aHRs (95% CIs) for breast cancer were 2.37 (1.68-3.36) for radiologist-classified dense breasts, 1.30 (1.05-1.62) for LIBRA, and 2.55 (1.84-3.56) for AI. For different combinations of breast density assessment, aHRs (95% CI) for breast cancer were 2.40 (1.69-3.41) for radiologist-dense/LIBRA-non-dense, 11.99 (1.64-87.62) for radiologist-non-dense/LIBRA-dense, and 2.99 (1.99-4.50) for both dense breasts, compared to concordant non-dense breasts. Similar trends were observed with radiologists/AI classification: the aHRs (95% CI) were 1.79 (1.02-3.12) for radiologist-dense/AI-non-dense, 2.43 (1.24-4.78) for radiologist-non-dense/AI-dense, and 3.23 (2.15-4.86) for both dense breasts. The risk of breast cancer was highest in concordant dense breasts. Discordant dense breast cases also had a significantly higher risk of breast cancer, especially when identified as dense by either AI or LIBRA, but not radiologists, compared to concordant non-dense breast cases.

Breast Density Status Changes: Frequency, Sequence, and Practice Implications.

Changes in a patient's reported breast density status (dense vs nondense) trigger modifications in their cancer risk profile and supplemental screening recommendations. This study tracked the frequency and longitudinal sequence of breast density status changes among patients who received serial mammograms. This IRB-approved, HIPAA-compliant retrospective cohort study tracked breast density changes among patients who received at least 2 mammograms over an 8-year study period. BI-RADS density assessment categories A through D, visually determined at the time of screening, were abstracted from electronic medical records and dichotomized into either nondense (categories A or B) or dense (categories C or D) status. A sequence analysis of longitudinal changes in density status was performed using Microsoft SQL. A total of 58 895 patients underwent 231 997 screening mammograms. Most patients maintained the same BI-RADS density category A through D (87.35% [51 444/58 895]) and density status (93.35% [54 978/58 859]) throughout the study period. Among patients whose density status changed, the majority (97% [3800/3917]) had either scattered or heterogeneously dense tissue, and over half (57% [2235/3917]) alternated between dense and nondense status multiple times. Our results suggest that many cases of density status change may be attributable to intra- and interradiologist variability rather than to true underlying changes in density. These results lend support to consideration of automated density assessment because breast density status changes can significantly impact cancer risk assessment and supplemental screening recommendations.

Clinical Significance of Combined Density and Deep-Learning-Based Texture Analysis for Stratifying the Risk of Short-Term and Long-Term Breast Cancer in Screening.

Assessing a woman's risk of breast cancer is important for personalized screening. Mammographic density is a strong risk factor for breast cancer, but parenchymal texture patterns offer additional information which cannot be captured by density. We aimed to combine BI-RADS density score 4th Edition and a deep-learning-based texture score to stratify women in screening and compare rates among the combinations. This retrospective study cohort study included 216,564 women from a Danish populations-based screening program. Baseline mammograms were evaluated using BI-RADS density scores (1-4) and a deep-learning texture risk model, with scores categorized into four quartiles (1-4). The incidence rate ratio (IRR) for screen-detected, interval, and long-term cancer were adjusted for age, year of screening and screening clinic. Compared with subgroup B1-T1, the highest IRR for screen-detected cancer were within the T4 category (3.44 (95% CI: 2.43-4.82)-4.57 (95% CI: 3.66-5.76)). IRR for interval cancer was highest in the BI-RADS 4 category (95% CI: 5.36 (1.77-13.45)-16.94 (95% CI: 9.93-30.15)). IRR for long-term cancer increased both with increasing BI-RADS and increasing texture reaching 5.15 (4.31-6.16) for the combination of B4-T4 compared with B1-T1. Deep-learning-based texture analysis combined with BI-RADS density categories can reveal subgroups with increased rates beyond what density alone can ascertain, suggesting the potential of combining texture and density to improve risk stratification in breast cancer screening.

Abstract B078: Artificial-intelligence-driven breast density assessment in the transition from full-field digital mammograms to digital breast tomosynthesis

Abstract Introduction: To enhance reproducibility and robustness in mammographic density assessment, various artificial intelligence (AI) models have been proposed to automatically classify mammographic images into BI-RADS density categories. Despite their promising performances, so far density AI models have been assessed primarily in traditional full-field digital mammography (FFDM) images. Our study aims to assess the potential of AI in breast density assessment in FFDM versus the newer synthetic mammography (SM) images acquired with digital breast tomosynthesis. Methods: We retrospectively analyzed negative (BI-RADS 1 or 2) routine mammographic screening exams (Selenia or Selenia Dimensions; Hologic) acquired at sites within the Barnes-Jewish/Christian (BJC) Healthcare network in St. Louis, MO from 2015 to 2018. BI-RADS breast density assessments of radiologists were obtained from BJC’s mammography reporting software (Magview 7.1). For each mammographic imaging modality, a balanced dataset of 4,000 women was selected so there were equal numbers of women in each of the four BI-RADS density categories, and each woman had at least one mediolateral oblique (MLO) and one craniocaudal (CC) view per breast in that mammographic imaging modality. Previously validated pre-processing steps were applied to all FFDM and SM images to standardize image orientation and intensity. Images were then split into training, validation, and test sets at ratios of 80%, 10%, and 10%, respectively, while maintaining the distribution of breast density categories and ensuring that all images of the same woman appear only in one set. Our AI model was based on the widely used ResNet50 architecture and was designed to accept as an input a mammographic image and predict the BI-RADS breast density category that the image belongs to. Our AI model was optimized, trained, and evaluated separately for each mammographic imaging modality. We report on the AI model’s predictive accuracy on the test set for each mammographic imaging modality, for both views as well as separately for CC and MLO; accuracy differences in FFDM versus SM were assessed via bootstrapping. Results: A batch size of 32, learning rate of e-6, and Adam optimizer were chosen as the optimal hyperparameters for our AI model. Using the same hyperparameters, the AI model demonstrated substantially higher accuracy on the test set for FFDM than for SM (FFDM: accuracy = 71% ± 4.5% versus SM: accuracy = 66% ± 4.2%; p-value<0.001 for comparison). Similar conclusion held when CC and MLO views were evaluated separately (accuracy = 72% ± 4.6% versus 66% ± 4.3% for CC; accuracy = 69% ± 4.5% versus 62% ± 4.3% for MLO; p-value<0.001 for both comparisons). Conclusions: AI performance in BI-RADS breast density assessment was significantly higher on FFDM versus SM, even under the same AI model design, dataset size and training process. Our preliminary findings suggest that further AI optimizations and adaptations may be needed as we translate AI models from FFDM to the newer SM format acquired with digital breast tomosynthesis. Citation Format: Krisha Anant, Juanita Hernandez Lopez, Debbie Bennett, Aimilia Gastounioti. Artificial-intelligence-driven breast density assessment in the transition from full-field digital mammograms to digital breast tomosynthesis [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Advances in Breast Cancer Research; 2023 Oct 19-22; San Diego, California. Philadelphia (PA): AACR; Cancer Res 2024;84(3 Suppl_1):Abstract nr B078.

Abstract 1899: Breast density comparison of Hispanic versus Non-Hispanic patients undergoing screening mammograms: Influence of age, parity and breast-feeding

Abstract Background: Higher mammographic breast density is associated with a 4 to 6-fold increased risk of breast cancer. Some single institution retrospective studies show higher breast densities in Hispanic patients (Oppong et al. Breast J. 2018). The prevalence of breast density based on Hispanic ethnicity has not been studied in large database studies. Methods: We performed a retrospective review of women > 40 years undergoing screening mammograms from 2016-2022. Patients with prior history of breast cancer were excluded. We reviewed demographic variables such as race, self-reported ethnicity, body mass index (BMI), parity, breastfeeding and menopausal status. Breast density was characterized as predominantly fatty (PF), scattered fibroglandular (SF), heterogeneously dense (HD) and extremely dense (ED). We used Chi-square and Kruskal-Wallace tests to compare breast density across measured variables, and Ordinal Logistic Regression to compare breast density levels between self-declared ethnicity while adjusting for other variables. Results: 15946 patients were studied; 7317 (46%) identified as Hispanic, the rest non-Hispanic (NH). 6131 (83%) of Hispanics identified as white. Mean age on 1st screening mammogram was 57.64 (SD 7.21). Compared to NH, Hispanics had mean earlier age of 1st mammogram (56.8 vs. 58.3 years), higher BMI (30.5 vs. 28.6) and higher parity (2.53 vs. 1.71). There were significant differences in BI-RADS density categories by Hispanic vs. NH (p<0.001), with differences in PF (20.2% vs. 18.5%), SF (41.7% vs. 39.0%), HD (35.8% vs. 37.2%), and ED (2.3% vs. 5.3%). The ordinal logistic regression model suggests that Hispanics have 17% lower odds than NH of 1 category higher density (OR, 95% CI: 0.83, 0.78-0.88). There were also significant ethnicity differences in age at first exam, BMI, parity, and number of children breast-fed (P<0.001). Adjusting for these ethnicity-associated differences explained the observed differences in density between Hispanic and NH individuals (ORadj, 95% CI: 0.98, 0.92-1.04); adjusting only for BMI served to explain the observed differences in density (ORadj, 95% CI: 0.99, 0.94-1.05). Conclusions: Hispanics were observed to have a lower age of first screening mammogram, higher BMI, higher parity, and breast-fed children compared to non-Hispanics. On univariate analysis, Hispanics were noted to have lower breast density compared to non-Hispanics, these differences do not persist on multivariate analysis. This suggests that BMI likely plays a predominant role in driving the ethnicity-related differences in breast density noted here. Further research efforts should focus on the influence of BMI and other individual-level factors on mammographic density, with a view towards understanding their potential roles in describing differential ethnicity-specific breast cancer risk. Citation Format: Sangeetha Prabhakaran, Vernon S. Pankratz, Jennifer L. Saline, Emily J. Reese. Breast density comparison of Hispanic versus Non-Hispanic patients undergoing screening mammograms: Influence of age, parity and breast-feeding [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 1899.

Abstract P3-04-13: Patient experience with automated SoftVue 3D whole breast tomographic ultrasound

Abstract Authors: Mary Yamashita, MD; Rachel Brem, MD; Lauren Baker, Ph.D; MD; Taylor Mahoney, PhD; Patrick Walker, PharmD, MPH; Rachel Treat, MA; Linda Hovanessian Larsen, MD. Title: Patient experience with automated SoftVue 3D whole breast tomographic ultrasound. Purpose: SoftVue (SV) is an automated, 3D whole breast ultrasound tomographic imaging device which is FDA PMA approved as adjunct to mammography for women with dense breasts. Screening with handheld US is labor intensive and with ABUS is associated with significant patient discomfort. A benefit of SV is that image acquisition is not operator dependent and does not require compression. Because acceptance by patients is crucial to implementation of US screening, we evaluated patients’ experience with SV, specifically, asymptomatic women with BI-RADS c or d density undergoing FFDM. Materials and Methods: As part of a prospective, 10-site study, 7,439 asymptomatic women with BI-RADS density category c or d were screened on the same day with FFDM and SV. Each patient’s experience was assessed for perceived pain, discomfort and anxiety, discretion and modesty, overall satisfaction, and whether they would recommend it to others. The responses were measured on a Likert scale with 5 choices from strongly agree to strongly disagree, then studied by Chi-Squared analysis. Results: The mean age was 53.9 ± 9.7 yo, mostly white women (87.7%). The median BMI was 24.4. Majority had no personal history of breast cancer (97.2%), but 24.8% had a previous biopsy and almost half (46.3%) had a family history of breast cancer. Almost all patients (99.6%) completed the survey. SV was perceived as significantly more comfortable than FFDM (83.7% vs 52.2%, p< 0.001), was painless (94.9% vs 53.1%, p< 0.001), and was associated with less anxiety during the procedure (95.1% vs 79.9, p< 0.001). Lastly, 99.3% felt the experience was private and discreet, and 95% would recommend the SV exam to other women. Conclusion: Pain, fear, anxiety, and modesty concerns are some of the barriers preventing widespread implementation of screening breast US. This data suggests that SV, an FDA PMA approved adjunctive screening exam for women with dense breast tissue, is painless, offers a private and discreet scan that limits anxiety, and is well accepted by patients. Clinical relevance statement: SV is a novel automated US tomographic screening technology that is comfortable, well-accepted, FDA PMA approved, and will likely result in improved implementation of screening breast ultrasound in women with dense breasts. Citation Format: Mary Yamashita, Rachel F. Brem, Lauren Baker, Taylor F. Mahoney, Patrick Walker, Rachel M. Treat, Linda Hovanessian Larsen. Patient experience with automated SoftVue 3D whole breast tomographic ultrasound [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 P3-04-13.

Mammographic Breast Density, Benign Breast Disease, and Subsequent Breast Cancer Risk in 3.9 Million Korean Women.

Background Mammographic breast density and benign breast disease are strong risk factors for breast cancer. Accordingly, women with both risk factors may have a markedly high risk for developing breast cancer. Purpose To investigate the risk of breast cancer associated with the combination of mammographic density and benign breast disease in Korean women, where population-based mammographic breast cancer screening is provided for all women aged at least 40 years. Materials and Methods This is a retrospective analysis of data from a nationwide breast cancer screening program linked with the national cancer registry. The study included Korean women between 40-74 years of age screened for breast cancer between January 2009 and December 2010 and observed up to December 2020 (median follow-up of 10.6 years). Benign breast disease and breast density were extracted from mammography screening results. Cox proportional hazard regression analysis was used to calculate adjusted hazard ratios (HRs) for breast cancer risk. Results In this study, 3 911 348 women (mean age, 53 years ± 9 [SD]) were analyzed. During follow-up (median, 10.6 years), 58 321 women developed breast cancer. At screening, 10 729 (18.4%) cases of benign breast disease were detected among women who developed breast cancer. Women with extremely dense breasts (Breast Imaging Reporting and Data System [BI-RADS] density category D) and benign breast disease had a greater risk of breast cancer when compared with women presenting with fatty breast (BI-RADS density category A) and those without benign breast disease (HR, 2.75; 95% CI: 2.63, 2.88; P < .001). Women with benign breast disease and fatty breasts (HR, 1.49; 95% CI: 1.40, 1.58; P < .001) and women with extremely dense breasts and without benign breast disease (HR, 2.28; 95% CI: 2.20, 2.35; P < .001) also had an elevated breast cancer risk compared with women with fatty breast tissue. Conclusion Women with dense breasts and benign breast disease at screening mammography had an elevated risk of future breast cancer. © RSNA, 2022 Online supplemental material is available for this article.

Spatial Distribution Analysis of Novel Texture Feature Descriptors for Accurate Breast Density Classification

Breast density has been recognised as an important biomarker that indicates the risk of developing breast cancer. Accurate classification of breast density plays a crucial role in developing a computer-aided detection (CADe) system for mammogram interpretation. This paper proposes a novel texture descriptor, namely, rotation invariant uniform local quinary patterns (RIU4-LQP), to describe texture patterns in mammograms and to improve the robustness of image features. In conventional processing schemes, image features are obtained by computing histograms from texture patterns. However, such processes ignore very important spatial information related to the texture features. This study designs a new feature vector, namely, K-spectrum, by using Baddeley’s K-inhom function to characterise the spatial distribution information of feature point sets. Texture features extracted by RIU4-LQP and K-spectrum are utilised to classify mammograms into BI-RADS density categories. Three feature selection methods are employed to optimise the feature set. In our experiment, two mammogram datasets, INbreast and MIAS, are used to test the proposed methods, and comparative analyses and statistical tests between different schemes are conducted. Experimental results show that our proposed method outperforms other approaches described in the literature, with the best classification accuracy of 92.76% (INbreast) and 86.96% (MIAS).

Interobserver Variability and Distribution of Mammographic Breast Density in Patients with Invasive Breast Carcinoma using the Fifth Edition of BI-RADS Guidelines

INTRODUCTION: Breast density is associated with an increased risk of developing breast cancer. The present study aims to determine the distribution and interobserver variability of mammographic breast density in patients with invasive breast carcinoma, using the fifth edition of BI-RADS guidelines. It is part of a larger study to ascertain the association between mammographic breast density and breast cancer characteristics. MATERIALS AND METHODS: Two radiologists independently assessed 122 mammograms of patients with histologically confirmed invasive breast carcinoma and assigned the breast density to categories A-D based on the fifth edition of BI-RADS guidelines. The interobserver variability was calculated using the weighted kappa coefficient and the level of agreement was determined using the Landis and Koch guidelines. RESULTS: In this study, 55.7% of patients with invasive breast carcinoma were assigned to category B, followed by category C with 36.1%. Only 4.1% of patients were assigned to categories A and D respectively. There was substantial agreement between the two readers’ judgement, k=0.610 (95% CI, 0.523-0.697), p &lt; 0.001 for specific BI-RADS categories. CONCLUSION: Among patients with invasive breast carcinoma, there were more patients with non-dense breasts than dense breasts. Overall, there is a substantial interobserver agreement when radiologists used the fifth edition of the BI-RADS guideline, which is in line with results found in the literature. This suggests that the BI-RADS density classification is an acceptable method and can be reliably used in clinical practice.

Abstract PS7-32: Age of first full term pregnancy and other reproductive factors affect mammographic breast density in postmenopausal women

Abstract The objective of this study is to determine the role of age at first full term pregnancy (FFTP), and mammographic breast density (MBD) in postmenopausal women. For this purpose women, age 50-69y, participating in the Flemish (Belgium) population based breast cancer screening program were invited to participate in the study during their visit at the mammography department in the University Hospital Leuven (UZLeuven) or at the mobile unit of their hometown. A self-administered questionnaire providing detailed information on lifestyle and environmental exposure was completed by the participants. In this analysis 1,034 women were included (mean age of 59.1 years (range: 49.2 to 69.8). The three parameters of MBD were the percentage glandular tissue of the total breast volume (GLAND), the ratio of the glandular tissue volume compared to the whole breast volume or Volumetric Breast Density (VBD) and the BI-RADS density classification (VolparaDataManager®, USA). The first two parameters (GLAND and VBD) were transformed according to the natural logarithm to improve normality. These two parameters were linked with the FFTP via a piecewise linear regression (PLR) model using the NLIN procedure in SAS (version 9.4, SAS Institute, Cary, NC, USA). For the BI-RADS, we used ordinal logistic regression analysis. All models were corrected for a priori chosen variables: age at menarche, the ever use of oral contraceptive pill (OC), hormonal use at menopause, age at MBD measurement, and the body mass index at participation. Estimates are provided as a % change (95% confidence intervals [CI]).The average age at FFTP was 26.1 years (5th-95th percentile: 20 to 34). The PLR model estimated the breakpoint in our analysis at a FFTP of 25.7 years (95% CI: 22.3 to 29.1). For the woman with a FFTP younger than 25.7 years, the association between GLAND or VBD and FFTP is statistically not significant. For woman with a FFTP above 25.7 years, each year increase in FFTP was associated with 1.3% increase in GLAND (95% CI: 0.0% to 2.5%) and 1.5% increase in VBD (95% CI: 0.2% to 2.8%). Analysis of the BI-RADS showed similar results, the odds of belonging to a higher BI-RADS classification (e.g. from class 1 to class 2) increased with 5.4% (95% CI: 0.0% to 11.0%) for each year increase in FFTP age after the age of 25.7 years. Among other reproductive factors such as age at menarche, for every year delay there is 3.5% higher GLAND (95% CI: 1.4 % to 5.6%) and 2.5% higher VBD (95% CI: 0.5% to 4.6%) and the ever use of OC resulted in 10.0 % decrease in GLAND (95% CI: -18.8% to -0.2%) and 10.2% decrease in VBD (95%CI:-19.1% to -0.3%).In conclusion, this study is to our knowledge the first one to show that mammographic breast density is significantly increased when pregnancy takes place after 25.7 years of age. Later age at menarche is increasing and use of oral contraceptive is reducing mammographic breast density at postmenopause. (This study was supported by a Grant of the Breast Cancer Organization Think Pink of Belgium, the Department of Breast Radiology and Mobile Unit of the University Hospital Leuven, the Centre of Environmental Sciences and the Centre of Biostatistics of the Hasselt University, and NIH Grant CA06927 to FCCC, PA, USA). Citation Format: Magda Johanna Vandeloo, Kristof Y. Neven, Liesbeth M. Bruckers, Jose Russo, Liesbeth Vancoillie, Esmee M. Bijnens, Eliane Kellen, Tim S. Nawrot, Chantal Van Ongeval. Age of first full term pregnancy and other reproductive factors affect mammographic breast density in postmenopausal women [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 PS7-32.

Evaluation of the association between mammographic density and the risk of breast cancer using Quantra software and the BI-RADS classification.

To determine the association between mammographic density (MD) and the risk of breast cancer (BC) in Chinese women and to investigate the role of fertility risk factors in regulating the relationship between MD and BC.We used Quantra software and the BI-RADS classification to assess MD in 466 patients and 932 controls. Conditional matched logistic multiple regression analysis was used to determine the relationship between MD and BC, and risk was evaluated with the odds ratio (OR) and 95% confidence interval (CI).The ORs for category 4 versus category 2 were 1.95 (95% confidence interval [95% CI] (1.42∼2.66)) and 1.76 (95% CI (1.28∼2.42)) for the BI-RADS and Quantra classifications, respectively. The ORs for category 5 volumetric breast density (VBD) versus category 2 VBD and 5 fibroglandular tissue volume (FGV) versus category 2 FGV were 1.63 (95% CI (1.20∼2.23)) and 1.92 (95% CI (1.40∼2.63)), respectively. Females with category 5 VBD whose age at menarche was ≤13 years had the highest risk of BC (OR = 2.16, 95% CI (1.24∼3.79)), and females with category 5 FGV whose age at menarche was = 15 years had the lowest risk of BC (OR = 1.65, 95% CI (1.05∼2.62)). Females with categories 3–5 VBD and categories 3–5 FGV had reduced risks of BC with increasing number of births. Females with category 5 VBD had an increased risk of BC with increasing age at first childbirth (the OR increased from 1.49 to 1.95). Those with category 5 VBD had a reduced risk of BC with increasing breastfeeding duration (the OR decreased from 2.08 to 1.55). Females with category 5 FGV had a reduced risk of BC with increasing breastfeeding duration (the OR decreased from 4.12 to 1.62).Both the BI-RADS density classification and Quantra measures indicated that MD is positively associated with the risk of BC in Chinese women and that associations between MD and BC risk differ by age at menarche, parity, age at first childbirth and breastfeeding duration.

Abstract 3661: Association of mammographic density with blood DNA methylation

Abstract Background: Mammographic density is a strong risk factor for breast cancer. While numerous genetic variations have been shown to be associated with mammographic density, an epigenetic signature of mammographic density has not yet been identified. The objective of this study was to identify a DNA methylation signature associated with mammographic density. Methods: Blood samples were collected from 197 postmenopausal women 45-65 years old with no personal history of breast cancer. For each participant, mammographic density according to the BI-RADs density classification was obtained from the most recent mammogram report. White blood cell DNA methylation at more than 850,000 CpG sites was measured using the Illumina HumanMethylationEPIC BeadChip. Differential methylation was assessed using genome-wide, probe-level, and regional approaches after filtering and normalization using the recommended steps for DNA methylation microarray data. All analyses were adjusted for appropriate confounding variables. Results: On average, genome-wide DNA methylation was higher in women with elevated mammographic density. Additionally, DNA methylation at 343 individual sites and in 34 regions was significantly associated with mammographic density after false discovery rate correction. Most density-associated sites were hypermethylated. The top differentially methylated sites were located near cell cycle regulatory genes (ICMT, DESI2, TCF21/TARID, ROBO3), as well as those encoding cytoskeletal or plasma membrane proteins (PALLD, ANO7, ATP8B3, MTUS2). We observed hypermethylation near the promoters of genes including RUFY1 and SLFN12 and within the bodies of genes including LCN6, SNED1, and LRFN1. Functional analysis of the differentially methylated sites revealed enrichment of gene ontology categories including angiogenesis and regulation of cell adhesion, as well as apoptotic signaling in response to DNA damage. Conclusions: White blood cell DNA methylation differed according to mammographic density in a sample of postmenopausal women. Probe-based and regional analysis identified differentially methylated regions located near genes implicated in cancer. These differentially methylated regions may serve as a signature of mammographic density, and future studies of the genes implicated may provide a deeper understanding of the connection between mammographic density and breast cancer risk. Citation Format: Rachel McFarland Lucia, Wei-Lin Huang, Andrea Alvarez, Irene Masunaka, Argyrios Ziogas, Deborah Goodman, Andrew O. Odegaard, Trina M. Norden-Krichmar, Hannah Lui Park. Association of mammographic density with blood DNA methylation [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 3661.

Abstract P2-08-10: Mammographic breast density changes with endocrine therapy (ET) in women with hormone receptor positive (HR+) breast cancer in a Utah population

Abstract Introduction: Women with extremely dense breast tissue on mammography have a higher risk of breast cancer compared to similarly-aged women with less dense breast tissue. Tamoxifen therapy has been observed to reduce breast density among both pre- and postmenopausal women with HR+ breast cancer and may predict benefit of therapy. Studies of aromatase inhibitor (AI) therapy and breast density change, however, are limited and results inconclusive. Factors associated with breast density change with ET, and effect of duration of ET on breast density change, have not been well studied. Methods: Using the Utah Population Database, women were identified who underwent screening mammography in 2005-2016 and who had no personal history of breast cancer. Those who were subsequently diagnosed with HR+ breast cancer and treated in one of the major health care systems in Utah (Intermountain Healthcare or University of Utah Health) were included in the analysis. ET prescription information and duration of ET were obtained from provider records and the Utah All-Payer Claims Database. Women with mammography results in the electronic medical record both before and within 3 years after the breast cancer diagnosis were included. BI-RADS density classification from mammogram reports 6-12 months, 13-24 months, and 25-36 months after ET initiation were compared to pre-diagnosis BI-RADS density classification. Breast density reduction was defined as a decrease of ≥1 BI-RADS category compared to baseline. Fisher’s exact test was used to examine ET and breast density reduction among women ≥55y and &amp;lt;55y. Results: In total, 1328 women with HR+ breast cancer and baseline BI-RADS breast density categories B-D were included. Women with BI-RADS density category A(almost entirely fatty) were excluded. A majority (61%) were age ≥55y at diagnosis, were non-Hispanic white (90%), and had BMI ≥25 kg/m2 (56%); 15% had a first-degree family history of breast cancer. At baseline, 602 (44%) women had scattered fibroglandular densities (BI-RADS category B), 697 (50%) had heterogeneously dense breasts (BI-RADS category C), and 83 (6%) had extremely dense breasts (BI-RADS category D). Half were treated with AI only (49%), 21% with tamoxifen only, and 29% with both during follow-up. In both the age ≥55y and &amp;lt;55y groups, women with BI-RADS category D at baseline were more likely to experience a reduction in breast density with ET compared to women with either BI-RADS categories B or C (Table 1). There were no significant differences in breast density reduction observed between women treated with AI only compared to those treated with tamoxifen only, irrespective of age and baseline density. Breast cancer family history also had no effect on breast density reduction with ET. Conclusion: Baseline breast density is associated with a demonstrable reduction in breast density during ET. Our findings suggest that AI therapy can reduce breast density similar to that seen with tamoxifen therapy. Table 1 Breast Density (BD) changes among women who underwent mammography 6-12 months, 13-24 months, and 25-36 months after ET initiationBaseline BDN≥55 years oldPTotal&amp;lt;55 years oldPNo BD reduction N(%)BD reduction N(%)No BD reduction N(%)BD reduction N(%)6-12 months after ET initiationScattered157152(97)5(3)5452(96)2(3)Heterogeneous148115(78)33(22)10477(74)27(26)Dense144(29)10(71)&amp;lt;0.0001248(33)16(67)&amp;lt;0.0001Total319271(85)48(15)182137 (75)45(25)13-24 months after ET initiationScattered232214(92)18(8)8473(87)11(13)Heterogeneous211144(68)67(32)155107(69)48(31)Dense172(12)15(88)&amp;lt;0.0001287(25)21(75)&amp;lt;0.0001Total460360(78)100(22)267187(70)80(30)25-36 months after ET initiationScattered248220(89)28(11)8374(89)9 (11)Heterogeneous203133(66)70(35)16097(61)63(39)Dense162(13)14(88)&amp;lt;0.00013213(41)19 (59)&amp;lt;0.0001Total467355(76)112(24)275184(67)91(33) Citation Format: Mei Wei, Lisa Pappas, Emily Guinto, Carol Sweeney, Nicola Camp, Phoebe Freer, Saundra Buys, N. Lynn Henry, Karen Curtin. Mammographic breast density changes with endocrine therapy (ET) in women with hormone receptor positive (HR+) breast cancer in a Utah population [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 P2-08-10.

Abstract 593: Mammographic breast density in postmenopausal women is affected by the age of first full term pregnancy

Abstract Mammographic breast density (MBD) is accepted as a risk factor in breast cancer, whereas age of first full term pregnancy (FFTP) has been demonstrated to be a protective one. The objective of this study is to investigate the association between reproductive history and MBD at menopause. Women, age 50-69y, participating in the Flemish (Belgium) governmental organised breast cancer screening program were invited to participate in the study during their visit at the mammography department in Leuven (hospital and the mobile unit). The study was approved by the ethical committee of the hospital. After signing the informed consent, a self-administered questionnaire (SAQ) providing information on lifestyle and environmental exposure was completed by the participants. In the present work we present the analysis of the first 640 women that completed the SAQ and underwent mammography. Questionnaire data were registered via the Qualtrics software. The three parameters of MBD (VolparaDataManager®, USA) included the percentage glandular tissue of the total breast volume (GLAND), the ratio of the glandular tissue volume compared to the whole breast volume or Volumetric Breast Density (VBD) and the BIRADS density classification. We applied linear regression models to investigate the association between the continuous breast density markers (GLAND and VBD, log transformed) and the following explanatory variables: age of woman, ever pregnant, number of pregnancies, age of FFTP, ever breastfed (BF), amount and duration of BF, and breastfeeding in FFTP. An ordinal logistic regression model was used for the breast density marker BIRADS. The average GLAND and VBD changes with a multiplicative factor of 1.073 (p = 0.013) and 1.072 (p = 0.013) for a 5-year increase of age of FFTP. Thus, an increase of 5 years in age of FFTP is associated with about 7% increase in geometric mean of GLAND and VBD. The odds ratio for a higher versus a lower score of the BIRADS classification is 1.24 (p = 0.023) greater for every 5-year increase in age of FFTP. These results are independent for the inverse association between the age of the woman and the MBD markers. For a 5 year increment in age, geometric GLAND (p = 0.019) and VBD (p = 0.024) are approximately 5% lower and the odds ratio for a higher BIRADS score is 12% higher (p = 0.079). There is no association between the MBD markers and breastfeeding, the number and duration BF, and the number of pregnancies after correcting for age and age of FFTP. In conclusion, this study is to our knowledge the first one to show that MBD is significantly reduced when pregnancy takes place at early age.( This study was supported by a Grant of the Breast Cancer Organization Think Pink of Belgium, the Department of Breast Radiology and Mobile Unit of the University Hospital Leuven, the Center of Environmental Sciences and the Center of Biostatistics of the Hasselt University, and NIH Grant CA06927 to FCCC, PA, USA.) Citation Format: Magda J. Vandeloo, Kristof Y. Neven, Liesbeth Bruckers, Jose Russo, Liesbeth Vancoullie, Esmée Bijnens, Eliane Kellen, Tim S. Nawrot, Chantal Van Ongeval. Mammographic breast density in postmenopausal women is affected by the age of first full term pregnancy [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 593.

Abstract C82: Breast density notification in a racially diverse and immigrant sample of women: Evidence for social disparities in awareness, knowledge, and follow-up

Abstract As mandated by legislation in 31 states, clinical mammographic breast density (MBD) information is currently disclosed to women identified as having heterogeneously or extremely dense breasts on routine mammograms (Breast Imaging-Reporting and Data System [BI-RADS] density categories of C/D, hereafter referred to as high MBD). The New York State legislation, enacted in 2013, informs women about the lower sensitivity of mammography and increased breast cancer risk associated with having dense breasts and advises them to seek risk assessment and additional screening; women with low MBD (BI-RADS density categories of A/B) do not receive density information. The legally mandated MBD information provided to women is ambiguous and written at a high literacy level. Additionally, any follow-up actions on this information taken by women require personal resources, including high educational attainment, health insurance coverage, and access to regular health care provider versed in breast cancer risk assessment and screening guidelines. We investigated the impact of this legislation on women's awareness of MBD, knowledge of its effect on breast cancer risk and detection, and communication with health care providers in a predominately Hispanic and foreign-born sample of 466 women, recruited from a mammography clinic in the Northern Manhattan community of New York City (age range: 40-60 years; 79% Hispanic, 12% African American; 81% foreign-born; 26% less than high school education). Overall 88% of the sample had obtained a prior mammogram since 2013 and 31% had heard about MBD. Awareness of MBD did not differ by risk factors for breast cancer (e.g., family history of breast cancer, breast biopsy, parity-related factors) or psychological factors related to breast cancer screening (breast cancer risk perception and worry). Non-Hispanic white women were significantly more likely to have heard of MBD (81%) than non-Hispanic African American (54%), and Hispanic women (13%). In multivariable logistic regression models, odds of MBD awareness were significantly lower in women who were Hispanic (OR=0.1, 95% CI: 0.04, 0.3 relative to white women), foreign-born (OR=0.3, 95% CI: 0.2, 0.6 relative to U.S.-born), or had lower educational attainment (e.g., OR=0.3, 95% CI: 0.2-0.7 for high school or less vs. college or higher degree). Women with personal high MBD were twice as likely as women with low MBD to be aware of MBD, but adjusting for personal MBD did not substantially alter the associations of race/ethnicity, education, and nativity with MBD awareness. In women who reported awareness of MBD, 47% and 76%, respectively, were able to correctly identify the higher risk and masking potential associated with high MBD, and 31% reported following on MBD information with their health care providers. Racial/ethnic and educational differences in knowledge of MBD impact and health care provider communication were smaller and less consistent. In conclusion, awareness of MBD remains low in women from racial/ethnic minority and lower socioeconomic backgrounds, even in a regularly screened population. These findings underscore the importance of understanding and averting potential unintended disparate consequences of breast density notification. Citation Format: Parisa Tehranifar, Carmen B. Rodriguez, Mariangela Agovino, Karen M. Schmitt. Breast density notification in a racially diverse and immigrant sample of women: Evidence for social disparities in awareness, knowledge, and follow-up [abstract]. In: Proceedings of the Tenth AACR Conference on the Science of Cancer Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; 2017 Sep 25-28; Atlanta, GA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2018;27(7 Suppl):Abstract nr C82.

Abstract A02: Comparative analysis of breast density among Black, White, and Hispanic women presenting for screening mammography

Abstract Background: Data on ethnic variations in breast density are limited and often not inclusive of underrepresented minorities, including Black and Hispanic women. As breast density is associated with elevated breast cancer risk, investigating racial and ethnic difference may elucidate the observed differences in breast cancer risk among different populations. Methods: We reviewed breast density recorded at the initial mammographic screening study in women presenting to the Capital Breast Care Center (CBCC) and Georgetown University Hospital (GUH) from 2010 to 2014. Patient demographics including race, age at screening, education and menopausal status were abstracted in addition to body mass index (BMI). From imaging reports, we recorded the BI-RADS density categories: 1-fatty, 2-scattered fibroglandular densities, 3-heterogeneously dense and 4-extremely dense. Multivariable unconditional logistic regression was used to identify predictors of breast density. Results: Density categorization was recorded for 2,146 women over the five-year period, with 940 (43.8%) Black, 893 (41.6%), Hispanic and 314 (14.6%) White. Analysis of subject characteristics by low density (categories 1 and 2) and high density (categories 3 and 4) show that high category is observed in younger, Hispanic, nulliparous, premenopausal and non-obese women (P-values &amp;lt;.0001). Obese women are 70% less likely to have high breast density compared to non-obese women. Being Hispanic, premenopausal, and non-obese were predictive of high mammographic density on logistic regression. Conclusion: In this analysis of density distribution in a diverse sample of women presenting for mammography, Hispanic women have the highest breast density, followed by Black women with Whites having the lowest. Unique in our findings is women who identify as Hispanic having the highest breast density and lower rates of obesity. Further investigation of the impact of obesity on breast density in minority women, especially in the understudied Hispanic group is needed. Citation Format: Bridget A. Oppong, Chiranjeev Dash, Suzanne Oneill, Kepher Makambi, Tesha Coleman, Lucile L. Adams-Campbell. Comparative analysis of breast density among Black, White, and Hispanic women presenting for screening mammography [abstract]. In: Proceedings of the AACR International Conference: New Frontiers in Cancer Research; 2017 Jan 18-22; Cape Town, South Africa. Philadelphia (PA): AACR; Cancer Res 2017;77(22 Suppl):Abstract nr A02.

Understanding Clinical Mammographic Breast Density Assessment: a Deep Learning Perspective.

Mammographic breast density has been established as an independent risk marker for developing breast cancer. Breast density assessment is a routine clinical need in breast cancer screening and current standard is using the Breast Imaging and Reporting Data System (BI-RADS) criteria including four qualitative categories (i.e., fatty, scattered density, heterogeneously dense, or extremely dense). In each mammogram examination, a breast is typically imaged with two different views, i.e., the mediolateral oblique (MLO) view and cranial caudal (CC) view. The BI-RADS-based breast density assessment is a qualitative process made by visual observation of both the MLO and CC views by radiologists, where there is a notable inter- and intra-reader variability. In order to maintain consistency and accuracy in BI-RADS-based breast density assessment, gaining understanding on radiologists' reading behaviors will be educational. In this study, we proposed to leverage the newly emerged deep learning approach to investigate how the MLO and CC view images of a mammogram examination may have been clinically used by radiologists in coming up with a BI-RADS density category. We implemented a convolutional neural network (CNN)-based deep learning model, aimed at distinguishing the breast density categories using a large (15,415 images) set of real-world clinical mammogram images. Our results showed that the classification of density categories (in terms of area under the receiver operating characteristic curve) using MLO view images is significantly higher than that using the CC view. This indicates that most likely it is the MLO view that the radiologists have predominately used to determine the breast density BI-RADS categories. Our study holds a potential to further interpret radiologists' reading characteristics, enhance personalized clinical training to radiologists, and ultimately reduce reader variations in breast density assessment.

Effects of Changes in BI-RADS Density Assessment Guidelines (Fourth Versus Fifth Edition) on Breast Density Assessment: Intra- and Interreader Agreements and Density Distribution.

The objective of our study was to determine intra- and interreader agreements for density assessment using the fifth edition of the BI-RADS guidelines and to compare with those for density assessment using the fourth edition of the BI-RADS guidelines. Five radiologists assessed breast density four times in 104 mammographic examinations: twice using the fourth edition of the BI-RADS guidelines and twice using the fifth edition. The intra- and interreader agreements for density assessment based on each guideline were determined and compared. The density distribution pattern under each of the four BI-RADS density categories using each guideline was also noted and compared. The intrareader agreement for density assessment using the fifth-edition criteria was lower than that using the fourth-edition criteria (p = 0.0179). The overall intrareader agreement (weighted kappa) using the old criteria was 0.84 (95% CI, 0.80-0.87), and the individual intrareader agreement values in five readers ranged from 0.78 (95% CI, 0.69-0.88) to 0.92 (95% CI, 0.87-0.97). The overall intrareader agreement using the new BI-RADS criteria was 0.77 (95% CI, 0.73-0.81), and the individual intrareader agreement values in five readers ranged from 0.74 (95% CI, 0.64-0.84) to 0.99 (95% CI, 0.98-1.00). The interreader agreement values obtained using the fifth-edition criteria were also lower than those obtained using the fourth-edition criteria (p = 0.006). The overall interreader agreement using the old BI-RADS criteria was 0.65 (95% CI, 0.61-0.69), whereas the overall interreader agreement using the new BI-RADS criteria was 0.57 (95% CI, 0.53-0.61). Overall a higher number of dense assessments were given when the fifth-edition guidelines were used (p < 0.0001). Compared with the intra- and interreader agreements obtained using the fourth edition of the BI-RADS guidelines, the intra- and interreader agreements were lower using the fifth-edition guidelines. An increased number of dense assessments were given when the fifth-edition guidelines were used.

Assessing the Relationship of Mammographic Breast Density and Proliferative Breast Disease.

Increased breast density and a history of benign breast biopsy are both considered risk factors for developing breast cancer. Understanding the specifics of these risk factors and their relationship to each other can lead to a better understanding of a patient's propensity for breast cancer development and improved surveillance strategies. We included 245 women who underwent a benign breast biopsy without atypia between October 2011 and June 2013. Biopsies were performed for suspicious calcifications as well as masses and architectural distortion. Lesions biopsied were divided into two groups: calcified and noncalcified lesions. The patient's breast density was assessed on most recent mammogram and was classified using the American College of Radiology BI-RADS density categories. Based on histologic diagnosis, each case was classified as proliferative or nonproliferative breast disease. The median age of the cohort (n = 245) was 55 years (range, 40-84 years). There were 162 (66%) postmenopausal women in the study. A core biopsy was performed for calcifications in 33.5% cases and for noncalcified lesions in 58% cases. In patients with dense breast tissue, an underlying proliferative histology was found significantly more frequently with calcifications (66.7%) as opposed to noncalcified lesions (35.9%) (RR = 2.3 (1.3-4.0); χ(2) = 8.7; p = 0.003). In nondense breast patients, there was no significant difference (RR = 1.1 (0.7-1.8); χ(2) = 0.1; p = 0.738). In the postmenopausal group, women with dense breasts had proliferative histology significantly more frequently than women with nondense breasts (55.3% versus 38.3%; p < 0.05), regardless of the underlying lesion type. Postmenopausal women with dense breasts who underwent a breast biopsy with benign histology had a significantly higher likelihood of having proliferative breast disease, regardless of underlying lesion type. Women with dense breasts also showed proliferative histology significantly more often for calcifications as opposed to noncalcified lesions.

Effect of Training on Qualitative Mammographic Density Assessment

PurposeThe aim of this study was to evaluate the accuracy of visual mammographic breast density assessment and determine if training can improve this assessment, to compare the accuracy of qualitative density assessment before and after training with a quantitative assessment tool, and to evaluate agreement between qualitative and quantitative density assessment methods. MethodsConsecutive screening mammograms performed over a 4-month period were visually assessed by two study breast radiologists (the leads), who selected 200 cases equally distributed among the four BI-RADS density categories. These 200 cases were shown to 20 other breast radiologists (the readers) before and after viewing a training module on visual density assessment. Agreement between reader assessment and lead radiologist assessment was calculated for both reading sessions. Quantitative volumetric density of the 200 mammograms, determined using a commercially available tool, was compared with both sets of reader assessment and with lead radiologist assessment. ResultsCompared with lead radiologist assessment, reader accuracy of breast density assessment increased from 65% before training to 72% after training (odds ratio, 1.41; P < .0001). Training specifically improved assignment to BI-RADS categories 1 (P < .0001) and 4 (P < .10). Compared with quantitative assessment, reader accuracy showed statistically nonsignificant improvement with training (odds ratio, 1.1; P = .26). Substantial agreement between qualitative and quantitative breast density assessment was demonstrated (κ = 0.78). ConclusionsTraining may improve the accuracy of mammographic breast density assessment. Substantial agreement between qualitative and quantitative breast density assessment exists.