Investigation of an epigenetic cancer risk biomarker reveals glucose metabolism as a key determinant of DNA methylation


Session type:

Kevin Brennan1, Charlotte Wilhelm-Benartzi1, Kirsty Flower1, Karin van Veldhoven1, Silvia Polidoro2, Carlotta Sacerdote2, Alessio Nacarrati2, Paolo Vineis1,2, Robert Brown1,3, James Flanagan1
1Imperial College London, London, UK, 2HuGeF Foundation, Torino, Italy, 3Institute of Cancer Research, London, UK


In addition to germline mutations and common polymorphisms, variability in epigenetic marks such as DNA methylation, may affect cancer risk, perhaps through alteration by age and environmental factors such as alcohol, diet or exercise. We have previously identified a DNA methylation variable position (MVP) at the ATM gene, at which hypermethylation detectable in blood was associated with increased breast cancer risk1.


In order to determine the factors driving this association, we have investigated age, BMI, genetic factors, and dietary factors, including serum metabolites and fasting status. ATM methylation in peripheral blood DNA samples was measured by bisulphite pyrosequencing in six sample sets from the European Prospective Investigation into Nutrition and Cancer (EPIC) cohort including three sets of healthy control subjects (combined n=485), and three nested breast cancer case-control studies (combined cases n=490, controls n=497). ATM methylation and haplotype were measured in a high risk familial case-control study from the kConFab study (cases n=166, controls n=225).


We show that ATM methylation was not affected by genetic factors or measured serum metabolites. Surprisingly, fasting status was associated with reduced ATM methylation and replicated in each of the six sample sets, with a 2.38-fold increased risk of low ATM methylation (< median) in fasting individuals compared with non-fasting individuals (95% CI 1.85-3.03). Furthermore, we show that ATM methylation was significantly reduced by glucose restriction in ex-vivo PBMCs derived from six healthy females, cultured for 24 hours in low glucose conditions, compared with cells cultured in full glucose (p=0.046). Using the Illumina 450k methylation beadchip array, we investigated the effect of fasting on DNA methylation at >480,000 CpG sites genome-wide (n=254), we have identified 8658 loci at which methylation is significantly altered by fasting (FDR q<0.05).


These data suggest that glucose metabolism appears to display a widespread effect on DNA methylation in WBC DNA.