Blood DNA methylation biomarkers for breast cancer risk

Abstract

Breast cancer is the most common malignancy affecting women worldwide with an average lifetime risk of 12%. Risk is affected by age, family history, genetics, reproductive factors and environmental exposures, however many unknown risk factors may exist. Regular screening, lifestyle advice and preventative therapy may be offered to women at highest risk; however in the absence of high-penetrance mutations, personal breast cancer risk cannot be accurately estimated. Risk biomarkers are therefore required to help improve current risk prediction models. Epigenetic mechanisms control gene expression and genome function, and are influenced by both heritable and environmental factors. DNA methylation, the most widely studied epigenetic mark, is widely deregulated in cancer and cancer precursor lesions; however the contribution to disease risk of DNA methylation variability in normal tissue prior to disease is poorly understood. Several blood DNA methylation markers associated with cancer have been reported, including genome-wide hypomethylation and hypermethylation of the ATM gene associated with breast cancer, however, validation of these associations in samples collected prior to diagnosis (prospectively collected) are required to determine association with breast cancer risk. The aims of this thesis were to 1) validate ATM methylation as a breast cancer risk marker in three nested case-control studies from prospective cohorts; 2) To investigate hypomethylation of LINE1 in the same prospective cohorts and compare this in a metaanalysis with all other published LINE1 data; 3) To investigate potential mechanisms or modifiers of ATM methylation; 4) To perform discovery microarray studies to identify novel DNA methylation markers of breast cancer risk. Herein, we show that ATM hypermethylation showed a 1.9 fold increased risk of breast cancer limited to women in the highest quintile of methylation (OR =1.89 (1.36-2.64), p= 1.64x10-4). There was no evidence of LINE1 methylation associated with cancer risk in the prospective cohort studies. The meta-analysis 3 of LINE1 and other global methylation markers showed little evidence of association with cancer risk for surrogate assays of repetitive elements, but relatively consistent association with cancer risk using HPLC based total methyl-cytosine levels. Investigation of potential modifiers of ATM methylation revealed that methylation was independent of genetic haplotype, but independently associated with age, genotype of the one-carbon metabolism enzyme MTHFR, and serum levels of serum kynurenic acid levels in controls (p=0.02). Surprisingly, ATM methylation was reduced in controls (p= 5.707e-06) and cases (p= 0.008) that had fasted compared to those that had not. The effect of fasting on ATM methylation could be recapitulated by glucose restriction in ex-vivo PBMCs (p=0.046), independent of cell proliferation. Discovery studies to identify novel DNA methylation risk markers were conducted using differential methylation hybridisation and Illumina Infinium HumanMethylayion450 BeadChip microarrays; however, significant associations were not reproducible in validation sample sets. Discussed are prospects and caveats for epigenetics association studies, including the implications of temporal alteration of DNA methylation by environmental exposures, biases associated with genetic influences on DNA methylation, and the potential for investigation of these interactions to better understand the contribution of epigenetics to gene expression and cancer risk.