Abstract B24: Lifecourse socioeconomic factors and mammographic breast density

Abstract

Abstract Breast cancer risk varies by socioeconomic status (SES), but it is unclear whether similar SES patterns exist for mammographic density, a strong intermediate marker of future breast cancer risk. Evidence on the associations between SES and mammographic density can further support the validity of mammographic density as an intermediate marker of breast cancer risk. Furthermore, given that breast cancer risk develops over the lifecourse, socioeconomic conditions in early life periods may be associated with density. We examined the associations between early life (infancy) and adult socioeconomic factors and mammographic density in 348 women from the Childhood Health and Development study, a birth cohort established in 1960–1967 in California. We further investigated the mediating role of adult SES, adult body mass index (BMI, Kg/m2) and age at first live birth in the associations between early life SES and density. Data on maternal education and family income were reported by participants’ mothers around the time of participants’ birth. Data on participants’ own education, and adult household income, ethnicity, BMI and reproductive events were collected directly from adult participants. We obtained participants’ most recent mammograms from the facilities where mammograms were performed and evaluated the proportion of dense area (dense area/breast area in percentage) and dense breast tissue (dense area in cm2) using Cumulus software (average age=43.0±2.1). We used linear regression analyses to investigate the associations between parental and own education and income and percent density and dense area. We also examined the associations of density with educational and income trajectories from early life to adulthood (persistent low, downward mobility, upward mobility vs. persistent high group). Maternal education, adult education and adult income had moderate to strong correlations (e.g., Spearman correlation coefficients for adult education with maternal education and adult income were 0.34 and 0.42, respectively), but early life family income had minimal correlation with other SES variables (range of 0.16 to 0.21). In separate models that adjusted for age and ethnicity, maternal education and adult education were each inversely associated with percent density (e.g., β=-10.4 95% confidence interval [CI]: −17.8, −3.0 & = −10.9, 95% CI: −17.7,−4.1 for lowest vs. highest groups for maternal education and adult education, respectively) and with dense area (e.g., β=−9.4, 95% CI: −18.0, −0.8 & β=−11.2, 95% CI: −19.7, −2.7 for lowest vs. highest groups for maternal education and adult education, respectively). Mutual adjustment for maternal and adult education attenuated the estimates of their associations with both percent density and dense area. The associations between maternal and adult education and percent density were substantially reduced by adjustment for BMI, with further small reduction from adjustment for age at first live birth. BMI and age at first live birth resulted in relatively small reductions in the associations between education and dense area but educational differences in dense area were no longer statistically significant in the final model. Results of social trajectory models suggested lower percent density and dense area for women with persistent low education and downward mobility that were substantially accounted for by adjustment for BMI and age at first birth for percent density, and to a lesser extent, for dense area. Early life and adult income, and income trajectories did not show any significant associations with either measure of density. In summary, the inverse educational patterns in mammographic density, observed for both maternal and own education, parallel SES variations in breast cancer risk. Educational differences in percent density, but not dense area, were mainly explained by adult BMI, which may be due to the inverse association between BMI and non-dense breast tissue. Citation Information: Cancer Epidemiol Biomarkers Prev 2011;20(10 Suppl):B24.