Comparisons of the reference percentiles of breast tissue densities among women between 40 and 60 years old

Authors: Huichen Zhu, Ying Wei, Parisa Tehranifar

Category: Early Detection & Risk Prediction, International Issues in Cancer
Conference Year: 2018

Abstract Body:
We estimated the reference distributions of mammographic breast density (MBD), a measure of fibroglandular breast tissue and a strong risk factor for breast cancer, in 483 women aged 40-60 years (44% obese body mass index [BMI], 51% postmenopausal, 68% foreign-born). We measured MBD using Cumulus software as percentage density (PD) and dense area (DA). We estimated the 10%, 50% and 90% reference percentiles of PD and DA over age using non-parametric B-spline quantile regression to estimate the trajectories of PD or DA reference percentiles in a flexible way without restricting to any a-prior shapes. We also stratified PD or DA reference percentiles by BMI, nativity, and menopause statuses. We found that the 10th and 50th PD and DA percentiles decrease slowly between ages 44 and 55 while the 90th PD and DA percentile decrease quickly between ages 40 and 45, become stable at ages 45-50 and continues to decrease slowly after 50. Overweight and obese women have rates of decline in PD are slower than that of the women with normal BMI, but the distributions of DA are comparable among women with different BMI status. US- and foreign-born women share the same median or 10th percentiles of PD and DA, but the 90th percentiles of PD and DA decline quickly between 40 and 45, and then after age 50 for the United States born women. The decline of the 90th percentile is only observed between 40 and 50 for foreign-born women. Results from age and interaction of risk factors show that BMI and menopausal status are significant risk factors for PD at all reference percentiles (e.g., for BMI: P- value = 9.53e-4 for 10th percentile, 1.07e-5 for 50th percentile and .0358 for 90th percentile), Similar results were observed for menopausal status and DA, but BMI was only a significant risk factor for DA at 50th percentile (.004). Nativity is a significant risk factor for PD at 90th percentile (1.58e-6), and for DA at 10th percentile and 50th percentile (4.47e-6 for 10th percentile, .028). In conclusion, the reference percentiles of MBD change over age with percentile-specific trajectory patterns that differ by BMI nativity and menopausal status. The commonly used linear regression models of MBD may not accurately estimate the associations of these risk factors for MBD.

Keywords: Breast cancer, risk factors