African-American (AA) women are more likely than other US groups to be diagnosed with estrogen receptor negative (ER-) breast cancer, with poorer prognosis and higher mortality. Understanding the biological mechanisms underlying ER- breast cancer in AA women and developing effective preventive strategies represents a critical unmet need with major public health implications. We and others have shown that risk of ER- breast cancer is increased among AA women who are parous and do not breastfeed; factors that are more common among AA women, and may help to explain their higher incidence of ER- tumors. We hypothesize that specific reproductive exposures result in epigenetic silencing of pro-luminal differentiation genes via DNA methylation, leading to an expansion of aberrant, maturation-arrested luminal progenitor cells, which can give rise to ER- cancers. Our previous data showed distinct differences in tumor DNA methylation according to ER status. One strong candidate gene derived from these data is FOXA1, which promotes luminal cell differentiation by positively regulating a luminal gene expression signature in progenitor cells and repressing the basal cell phenotype. This gene was hyper-methylated in ER- versus ER+ tumors from AA women, particularly in those who were parous and did not breastfeed. Consistent with inhibitory effects of methylation on gene expression, FOXA1 protein levels were lower in ER- versus ER+ breast tumors, and lower in ER- tumors from parous vs. nulliparous women. Supporting this hypothesis, we recently showed that heterozygous deletion of Foxa1 in the mouse mammary gland results in a dramatic skewing of epithelial cell populations toward luminal progenitors. Building on these preliminary results, we propose a comprehensive genome-wide DNA methylation analysis of tumor samples from 1,621 AA women with breast cancer from the Black Women's Health Study) and the Women's Circle of Health Study using the IIllumina EPIC 850K array. Combining these profiles with existing 450K data from 383 AA cases after methylation imputation, we will examine FOXA1 and differentially methylated loci (DMLs) that distinguish ER subgroups. With epidemiologic data from these 2,004 AA cases, we will assess associations between reproductive risk factors and methylation of FOXA1 and top DMLs, using weighted gene correlation network analysis and structural equation modeling to evaluate complex relationships. We will evaluate the same relationships between parity, breastfeeding and methylation in normal breast tissue donated by healthy AA volunteers to the Komen Tissue Bank. Using mouse models, we will experimentally investigate if parity and breastfeeding influence the methylation level of Foxa1 and other pro-luminal candidate genes, as well as relative proportions of distinct mammary gland epithelial cell populations. This transdisciplinary, multi-pronged approach will enable us to understand the etiology of aggressive breast cancer in AA women, facilitate the development of novel markers for those at highest risk, and uncover promising molecular targets for precision prevention approaches.
African-American women have high rates of aggressive estrogen receptor negative (ER-) breast cancer. There is growing evidence that, while having children reduces risk of ER+ breast cancer, it actually increases risk of ER- tumors, especially without breastfeeding. We will use approaches from epidemiology and from the laboratory to better understand why having children has the opposite effect on breast cancer subtypes, and identify the gene pathways that are affected by these events and that lead to aggressive disease. Identification of these critical markers can make inroads into prevention of this deadly form of breast cancer and potentially provide therapeutic targets.