Our laboratory is broadly interested in both genetic and epigenetic modifications associated with exposure and cancer. Epigenetic modifications, including DNA methylation, are being increasingly recognized as is being recognized as important determinants of gene transcriptional regulation that have both heritable and acquired characteristics. Aberrant DNA methylation patterns are among the earliest and most common events in carcinogenesis and recent studies suggest that the epigenetic profile of DNA from a surrogate tissue, peripheral blood, may differ between women with active ovarian cancer compared to women without disease. We have employed genome-wide profiling of DNA methylation in peripheral blood samples from more than 900 women in order to investigate whether the pattern of DNA methylation is associated with breast cancer risk. Using case-cohort proportional hazard regression we identified 250 CpGs that significantly differed (at false discovery rate (FDR) q=0.05) between cases and non-cases. These 250 differentially methylated CpGs were significantly more likely to be undermethylated in cases relative to non-cases, and the difference was consistently greater in women who developed breast cancer in the year following their blood draw, compared to women with longer intervals. CpGs near known breast cancer susceptibility genes were twice as likely to have unadjusted p values <0.05 (including ATM, BRCA1, CHEK2, FAM84B, FGFR2, MLH1, MSH2, PTEN, TNP1);one of these, located in FGFR2, reached study-wide significance at FDR q <0.05. Pathway analysis showed statistically significant enrichment for 8 of 14 cancer pathways in the Kyoto Encyclopedia of Genes and Genomes (KEGG), suggesting that the CpGs are located in genes involved in cancer. We demonstrated that the methylation differences were unlikely to be caused by circulating tumor DNA or due to shifts in leukocyte subpopulations. Although the methylation differences that we observed were small, they might be useful in breast cancer risk prediction. We first examined existing predictors (the Gail Model based on known risk factors, and common SNPs identified from GWAS studies) and computed predictive accuracy. Our estimates were remarkably consistent with general population estimates and ranged from 56% for the Gail Model alone, to 61% for the Gail Model plus GWAS SNPs. In order to select CpGs and evaluate their performance, we used a double-loop cross-validation method and found that a set of 57 CpGs achieved significantly better classification performance in independent test sets (66%) than either the Gail Model or GWAS SNPs or both. A small set of only 5 CpGs performed almost as well (64%). We replicated the 5 CpG result in an independent set of 25 cases and 56 non-cases of different ethnicities from the Sister Study cohort, and found similar classification performance of 63%. Although the level of classification performance is not yet high enough for clinical application, this is the first epigenome-wide association study of any cancer to use prospectively collected samples, and provides support for the hypothesis that methylation profiles in blood may be useful for cancer prediction. A second form of epigenetic modification is miRNA expression. Altered miRNA expression is a central feature of cancer and miRNA expression signatures have been shown to be associated with diagnosis, stage, prognosis, and response to treatment. Expression patterns for cancer show high tissue specificity making them potential markers for cancer screening. Breast cancer specific miRNAs have been shown to correlate with stage, vascular invasion, proliferative index, and ER/PR status. Recently, sufficient levels of miRNAs have been found in human plasma and serum to permit profiling, with sufficient power to distinguish men with metastatic prostate cancer from men without cancer and women with ovarian cancer. We tested the hypothesis that serum levels of miRNAs are markers of susceptibility to breast cancer. We used prospectively collected samples from the Sister Study in a matched-pair nested case-control design: We identified 205 women who developed invasive breast cancer and selected controls who remained cancer-free, matching on race, age, and date of blood draw. My laboratory extracted miRNAs from serum samples and measured miRNA levels using Affymetrix GeneChip miRNA 2.0 arrays, that contains probes for 1,105 known human miRNAs. We found a set of 21 miRNAs that were differentially expressed between cases and controls. Using qRT-PCR in a small independent set of samples, we validated the three miRNAs (miR- 18a, miR-181a, and miR-222 ) that had the largest case-control differences. Our most noteworthy finding was increased case levels of miR-222, a miRNA that is known to directly target ER alpha. Overexpression of miR-222 has been reported to lead to breast cancer with more aggressive ER-negative phenotype 10, and miR-222 is in turn regulated by ER alpha, via a negative-regulatory loop 11. In addition this miRNA acts as an oncogene by repressing p27/Kip1 and p57, and thus promotes cell proliferation and self-renewal 12. Two smaller studies have also suggested that circulating miR-222 is associated with breast cancer13, 14 and, coupled with our prospective data, strengthens the case that circulating mirR-222 is associated with breast cancer risk.

Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
Zip Code
Xu, Zongli; Taylor, Jack A (2014) Genome-wide age-related DNA methylation changes in blood and other tissues relate to histone modification, expression and cancer. Carcinogenesis 35:356-64
Markunas, Christina A; Xu, Zongli; Harlid, Sophia et al. (2014) Identification of DNA Methylation Changes in Newborns Related to Maternal Smoking during Pregnancy. Environ Health Perspect :
Cooke, Susanna L; Shlien, Adam; Marshall, John et al. (2014) Processed pseudogenes acquired somatically during cancer development. Nat Commun 5:3644
Harlid, Sophia; Xu, Zongli; Panduri, Vijayalakshmi et al. (2014) CpG sites associated with cigarette smoking: analysis of epigenome-wide data from the Sister Study. Environ Health Perspect 122:673-8
Deroo, Lisa A; Bolick, Sophia C E; Xu, Zongli et al. (2014) Global DNA methylation and one-carbon metabolism gene polymorphisms and the risk of breast cancer in the Sister Study. Carcinogenesis 35:333-8
White, Alexandra J; Sandler, Dale P; Bolick, Sophia C E et al. (2013) Recreational and household physical activity at different time points and DNA global methylation. Eur J Cancer 49:2199-206
Godfrey, Ashley C; Xu, Zongli; Weinberg, Clarice R et al. (2013) Serum microRNA expression as an early marker for breast cancer risk in prospectively collected samples from the Sister Study cohort. Breast Cancer Res 15:R42
Kim, Sangmi; Parks, Christine G; DeRoo, Lisa A et al. (2009) Obesity and weight gain in adulthood and telomere length. Cancer Epidemiol Biomarkers Prev 18:816-20
Horton, Julie K; Watson, Mary; Stefanick, Donna F et al. (2008) XRCC1 and DNA polymerase beta in cellular protection against cytotoxic DNA single-strand breaks. Cell Res 18:48-63