Characteristics of Mammographic Density for Women in China

Abstract

Objectives: The purpose of this study is to understand the characteristics of mammographic density for women in China. Methods: The potential factors of density were examined using two approaches - one qualitative (i.e. BIRADS classification) and one quantitative measurement (i.e. AutoDensity algorithm). The BIRADS study (with 4867 women) used Spearman tests and Mann-Whitney/Kruskal-Wallis tests to examine the associations between density and variables. Logistic regression was then performed to produce odds ratio. The AutoDensity study recruited 1071 (84 with and 987 without breast cancer) women and examined the associations using Pearson tests and t-tests. Multivariate models were then built. Results: The BIRADS study showed negative associations between BIRADS density and age (rho=-0.23) and BMI (rho=-0.18). Density was statistically different (p<0.05) across educational, province of residence and occupation groups. Women with a history of early age of menarche, pre-menopause, nulliparity, no breastfeeding and benign breast disease demonstrated increased density (p<0.05). The AutoDensity study found that, for women without cancer, weight and BMI (p<0.01) were negatively associated (r=-0.24, r=-0.27) with percentage density (PD) whereas positively associated (r=0.11, r=0.10) with dense area (DA). Lower PD and DA were found for secondary education compared to tertiary education (p<0.05). There is no association (p>0.05) for smoking history, alcohol consumption and family history of breast cancer. For women with breast cancer, PD demonstrated similar relationships with those of cancer-free women whilst breast area was the only factor associated with DA (r=0.74, p<0.001). Conclusions: For the first time in China, this work has shown distribution of Chinese mammographic density and demonstrated important associations for demographic, lifestyle, menstrual, reproductive and familial factors with density being assessed using both qualitative and quantitative methods.