Abstract B47: Methylation differences in breast tumor DNA from African American and European women are predominant in estrogen receptor (ER) negative breast cancer and are associated with childbearing

Abstract

Background: Although European American (EA) women have higher incidence of breast cancer than African American (AA) women, AA women are more often diagnosed with aggressive estrogen receptor negative (ER-) tumors, the risk of which increases with parity, lack of breastfeeding, and an early age at menarche. DNA methylation could be affected by reproductive factors and may be a potential molecular mechanism driving differences in tumor etiology. In a small study using fresh frozen breast tissue (n=58 AA, 80 EA), we previously found more differentially methylated loci (DMLs) in ER- tumors from AA and EA women than in ER+ tumors. Methods: To follow up on these preliminary findings, we used the Illumina 450K platform to determine genome-wide DNA methylation profiles in formalin fixed paraffin embedded (FFPE) breast tumors from 383 AA and 350 EA women who participated in the Women's Circle of Health Study, a case control study conducted in NY and NJ. We identified DMLs by race and ER status, as well as differentially methylated regions (DMRs). DMLs in eight genes were validated using the Sequenom EpiTYPER platform. Recursively partitioned mixture modeling (RPMM) package in R was used to examine relationships between methylation clusters and reproductive factors, information available from in-person interviews. In addition, using fresh frozen tumor tissue from 50 patients treated at Roswell Park Cancer Institute, we performed RNA sequencing on samples with methylation data available from a prior study, and used Spearmen's correlation to compare methylation and gene expression of DMLs and DMRs. Results: In assessing average methylation by location relative to CpG-islands (CGIs), we found that CGIs and shores in ER- tumors were significantly hypomethylated compared to CGI and shores in ER+ tumors from AA women, but not in tumors from EA women. We also identified 410 DMLs (Δβ>0.10 & FDR<0.05) between AA and EA breast tumors (race-associated DMLs, raDMLs), the majority of which were unique to ER- tumors (n = 260) and hypomethylated in AAs. These loci were enriched for immune response genes and several cell adhesion and inflammatory pathways. RPMM showed that parity, but not breastfeeding or age at menarche, was significantly associated with methylation class of ER- breast tumors. Of the genes that had ER- specific raDMLs, FOXA1 and THSD4 DNA methylation was highly correlated with gene expression (rho=-0.80, rho=+0.87, respectively; pvalue< 2.2x10-16). Summary: ER- tumors from AA women, but not from EA women, exhibited global hypomethylation at CGI and shores compared to ER+ tumors and contained the majority of raDMLs, many of which were associated with genes involved in immune and inflammatory response. The FOXA1 gene, which encodes a pioneer factor previously implicated in suppressing the molecular phenotype of basal breast cancer cells, was found to be methylated and silenced in the majority of ER- tumors analyzed for expression, consistent with aggressiveness of these tumors. These results, and the novel finding of associations between parity and DNA methylation signature, bring us a step forward in understanding the heterogeneous population of ER- tumors and provide insight into mechanisms of racial disparities in breast cancer. Citation Format: Allyson C. Espinal, Dan Wang, Lara Sucheston-Campbell, Song Liu, Qiang Hu, Li Tang, Gary Zirpoli, Thaer Khoury, Song Yao, Kitaw Demissie, Elisa V. Bandera, Christine B. Ambrosone, Michael J. Higgins. Methylation differences in breast tumor DNA from African American and European women are predominant in estrogen receptor (ER) negative breast cancer and are associated with childbearing. [abstract]. In: Proceedings of the Eighth AACR Conference on The Science of Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; Nov 13-16, 2015; Atlanta, GA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2016;25(3 Suppl):Abstract nr B47.