Mediation of black/white disparities in triple negative breast cancer by socioeconomic position, reproductive factors and diabetes in the Nurses’ Health Study Cohorts I & II.

Abstract

1565 Background: The higher rates of triple negative (TN) breast cancer (BC) observed among black compared to white women may be attributable to social and reproductive factors, as well as biological conditions like diabetes mellitus type 2. We examine the extent to which the black/white race disparity in TNBC is transmitted through these factors. Methods: Data are from the Nurses’ Health Study I and II prospective cohorts and include non-Hispanic (nH) black (n = 97) and nH white (n = 8,876) women aged 25-55 years at enrollment with invasive BC diagnoses. Participant characteristics are self-reported or drawn from medical records. We estimated average controlled direct associations (ACDA) using logistic regression with model-based standardization to evaluate the age/body mass index (BMI)-adjusted prevalence difference for TN versus luminal A/B type BC by race, and performed a series of ACDA (with comparison of rescaled coefficients (RC)) controlling for diabetes, mean family income before diagnosis (cont.), husband’s education ( < college, college, graduate school), parental ownership of participant’s childhood home (yes/no), age at menarche (cont.), age at first birth (cont.), parity (integer), and breastfeeding (BF) (parous/never BF, parous/ever BF). Results: Compared to nH whites, nh black women had higher prevalence of TNBC (24.5% v. 45.4%; p < 0.01), higher prevalence of diabetes (5.5% v. 11.3%; p = 0.01), lower prevalence of BF (59.2% v. 46.6%; p = 0.02), higher mean BMI (26.1 v. 29.9; p < 0.01), and lower mean family income ($17,304 v. 15,750; p < 0.01). BMI, age, and menopause were positively associated with prevalence of diabetes. In the age/BMI adjusted model, nH black women were 20%-points more likely than nH whites to have TNBC (95% Confidence Interval (CI): 0.11, 0.29). This disparity was reduced to 18.1% after also adjusting for BF (95% CI: 0.09, 0.28), to 13.7% after additionally adjusting for socioeconomic indicators (95% CI: 0.02, 0.25), and to 13.3% after adjusting for all potential mediators (95% CI: 0.01, 0.26). Based on the method of RC, socioeconomic indicators accounted for the largest fraction of mediated effects and BF accounted for most of main effect attenuation associated with reproductive factors; diabetes accounted for negligible effect. Conclusions: Our findings support the hypothesis that observed racial differences in TNBC diagnoses may be at least partially mediated by differences in socioeconomic position and reproductive patterns, namely breastfeeding.