Breast Density: More Than Meets the Eye

Wolfe first recognized mammographic density as an important risk factor for breast cancer in the 1970s. His pioneering observation of increased breast cancer risk with increased mammographic density has since been confirmed in over a hundred studies in multiple racial/ethnic populations and using both qualitative and quantitative measures of breast density ( 1 ). The Breast Imaging Reporting and Data System (BI-RADS) breast density categories, estimated subjectively by radiologists ( 2 ), is the standard for reporting breast density in clinical practice in the United States. More than half of U.S. states now legally require some level of notification about dense breasts (heterogeneously or extremely dense breasts based on BI-RADS) with some notifications advising they discuss with their providers whether supplemental screening is right for them ( 3 ). An estimated 27.6 million women (47%) age 40 to 74 years in the United States have dense breasts by the BI-RADS system, which means in the 25 states with density laws millions of woman who undergo mammography are being made aware of their breast density ( 4 ).

New evidence reported in this issue of the Journal suggests that BI-RADS breast density may not reflect the volume of breast density in black women ( 5 ). McCarthy et al. compared clinical estimates of BI-RADS density from a screening practice of digital mammography in a large academic hospital with quantitative volume and area-based measures of breast density assessed with commercial (Quantra) and publicly available (LIBRA) software among black and white women. Breast density was categorized as dense (heterogeneously or extremely dense) or nondense (almost entirely fat and scattered fibrogladular densities) based on BI-RADS categories, and quantitative measures (dense breast volume in cubic centimeters) as quartiles above and below the median. Consistent with other studies ( 6 ), McCarthy et al. report that the distribution of BI-RADS breast density categories are similar in black and white women but are the first to report that dense breast volume is higher in black than white women both for obese and nonobese women ( 5 ). Similar differences between black and white women were reported for quantitative two dimensional measures (dense area measured in square centimeters and area percent density) and volume percent density ( 5 ). Not considered in the study by McCarthy et al. is menopausal status, where lower absolute dense volume in postmenopausal compared with premenopausal women has been shown ( 7 ). Premenopausal black women may have higher dense breast volume compared with white women, but dense breast volume may be lower and similar in postmenopausal black and white women, mirroring their breast cancer risk. Future studies should take into account race/ethncity, menopausal status, and body mass index (BMI) when comparing density measures.

The correlation of BI-RADS breast density with quantitative dense volume measures (Quantra and Volpara, another commercailly available software) is modest ( 8–10 ). Thus, not surprisingly, similar to the McCarthy finding, we previously found a discordant association between BI-RADS breast density and dense breast volume ( 11 ). Compared with white women, Asian women had a higher proportion of dense breasts based on BI-RADS categories, but dense breast volume was lower in Asian than white women ( 11 ). Thus, clinical BI-RADS and quantitative breast density measures, such as area and volume of breast density, may be assessing different aspects of breast density. BI-RADS density categories are estimated visually and reflect density quantity, distribution, and parenchyma pattern while quantitative measures algorithmically assess absolute dense volume or area.

Notably, although density measures may be assessing different aspects of breast density, the association with breast cancer appears to be robust, with qualitative and quantitative measures and in all race/ethnicity groups examined ( 1 , 6 , 11 , 12 ). The only available clinical risk model that incorporates breast density is the Breast Cancer Surveillance Consortium (BCSC) risk calculator ( https://tools.bcsc-scc.org/bc5yearrisk/calculator.htm ). The BCSC five-year and 10-year risk prediction model includes age, race/ethnicity, family history of breast cancer, history of breast biopsy and benign biopsy result, and BI-RADS density. The BCSC risk model has better discrimination than a model with clinical factors only without breast density, and calibration is accurate and similar for white and black women and slightly underestimates risk for Asian and Hispanic women ( 6 , 13 ). Thus, BI-RADS breast density can be incorporated with clinical risk factors to calculate breast cancer risk with the BCSC risk calculator, and it accurately predicts breast cancer in all common racial/ethnic groups.

Studies have found that quantitative measures of breast density are strongly associated with breast cancer risk ( 6 , 11 , 12 , 14 ), but risk prediction models have yet to incorporate an automated volumetric breast density measure. An advantage of incorporating dense breast volume into risk prediction models is that it is independent of other clinical risk factors (family history of breast cancer and benign breast biopsy) and weakly confounded by BMI. BMI is modestly associated with breast cancer risk but prevalent, with 50% to 60% of women in the United States being overweight or obese. Including dense breast volume, BI-RADS density, and BMI with other clinical risk factors into risk prediction models could improve discriminatory accuracy. Also, dense breast volume may better reflect underlying breast density biology and the association with breast cancer risk. For example, dense breast volume is low in Asian women ( 11 ), which corresponds to a low incidence of breast cancer, while dense breast volume is high in black women ( 5 ), who have a high incidence of breast cancer. Increased quantitative density measures are also associated with advanced disease, large tumors, and lymph node–positive disease ( 15 ), which is more common in black women, particularly premenopausal women ( 16 ). Given that premenopausal black women are at high risk for advanced disease and estrogen-negative breast cancer ( 16 ), it will be important to determine if dense breast volume is higher in premenopausal black women compared with other race/ethnicities.

Breast cancer risk prediction models that incorporate BI-RADS breast density are being used to identify women with dense breasts at high risk of interval cancer ( 17 ), tailor screening including starting and stopping ages and screening interval ( 18 , 19 ), and identify those eligible for primary prevention. Thus, the different nature of breast density measurements, whether estimated subjectively by radiologists (BI-RADS) or automated by programs (volume), likely provides complementary information to inform breast cancer risk. It will be important to examine the independent contributions of dense breast volume and BI-RADS breast density to both rates of interval cancer and absolute breast cancer risk.

Funding

This work was supported by a National Cancer Institute-funded Program Project (P01 CA154292) and R01 CA177150.

Notes

The funders had no role in the writing of the editorial or the decision to submit it for publication.

The authors have no conflicts of interest to disclose.