Background: Women with high breast density have a 4-6 fold increased risk of breast cancer compared to women with low breast density. There is a strong inverse association between body mass index (BMI) and breast density, so we hypothesized that some of the racial/ethnic variation in mammographic density may reflect differences in the prevalence of high BMI or the strength of the association between BMI and mammographic density by race/ethnicity.
Methods: We leveraged data from the Utah Population Database (UPDB) mammography cohort, including 140,200 non-Hispanic White (NHW) women, 703 non-Hispanic Black (NHB) women, 15,560 Hispanic women, 713 American Indian/Alaskan Native (AIAN) women, 2,525 Asian women, and 434 Native Hawaiian/Pacific Islander (NHPI) women with a screening mammogram obtained 2005-2019. We estimated the association between BMI and binary mammographic density (BI-RADS A & B versus C & D) using logistic regression adjusted for age, education, and parity in Utah. Menopausal status was not available, so we used age 55 as a proxy. We considered effect modification by running stratified analyses and conducting a likelihood ratio test of models with and without an interaction between BMI and race/ethnicity. We also calculated population attributable risks (PAR%).
Results: The prevalence of high BMI differed by race/ethnicity with the highest BMI among NHPI women (29.4% overweight and 52.2% obese) and lowest BMI among Asian women (17.9% overweight and 5.7% obese). Results from multivariable models were consistent with a strong inverse association between BMI and mammographic density (ORBMI>30v≤25=0.21, 95% CI=0.21-0.22, p-trend <0.001) that did not differ by race/ethnicity (p=0.07). There was also no evidence of heterogeneity by racial/ethnic group among women aged <55 (p=0.15), but some evidence of heterogeneity among women ≥55 years (p=0.004). We jointly considered prevalence of high BMI and estimates of association between BMI and mammographic density by calculating PARs. For age <55, after accounting for HT use, 26.7% (95% CI 26.3-27.2) of high mammographic density was explained by BMI ≤25. PARs were similar for NHW, Asian, and AIAN women, but lower for other groups, especially NHPI women (PAR%=9.8, 95% CI=6.4-13.3). For women age ≥55, 22.2% (95% CI 21.6-22.9) of high mammographic density was explained by BMI ≤25. PARs were similar for NHW, Hispanic, AIAN and NHPI women, and highest in NHB women (PAR%=42.2, 95% CI=35.9-48.5).
Discussion: We observed the strongest evidence of racial/ethnic differences in BMI when comparing the two groups most commonly studied together: Asians and NHPIs. While we present only limited evidence to suggest that BMI is differentially associated with breast density by race/ethnicity, differences in the prevalence of high BMI were substantial. Overall, our findings suggest that risk factor prevalence should not be overlooked when evaluating potential contributors to cancer disparities.
Citation Format: Mollie E. Barnard, Tarun Martheswaran, Karen Curtin, Jennifer A. Doherty. Body mass index and mammographic density by race and ethnicity [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 32.