Abstract 2763: Maternal folate intake at periconception and genome-wide DNA methylation modifications at birth in children

Abstract

Abstract Introduction Folate supplementation during pregnancy was previously associated with a lower risk of acute lymphoblastic childhood leukemia (ALL). Folate is involved in the “one-carbon” metabolic cycle necessary for shuttling methyl groups for DNA methylation. Given the observation that aberrant DNA methylation patterns are a characteristic of leukemic cells during ALL oncogenesis, we hypothesized that (1) low folate exposure during fetal development might modify the neonate's DNA genome-wide methylation pattern, and that (2) such a modified pattern can lead to ALL development. Here we investigate the first part of this hypothesis. Methods We analyzed healthy control participants of the California Childhood Leukemia case-control study. Participants’ peri-conception folate exposure was estimated from self-reported maternal folate intake and supplements, which was derived from a food frequency questionnaire on the diet in the year prior to pregnancy, as a surrogate for nutrient “environment” during the peri-conception period. DNA obtained from archived neonatal blood spots was purified, treated with bisulfite and assayed on the Illumina Infinium 450K genome-wide DNA methylation array. After removing cross-reacting probes, SNP-related and polymorphic CpGs, we analyzed the association of folate with methylation intensity of 319,265 CpGs in two independent datasets (n set 1 = 167 and n set 2 = 176). Models were adjusted for cell mixture, sex, gestational age, and race. We used a false discovery rate with random resampling to reduce the number of false-positive findings and we cross-validated the results between sets. Results Maternal folate intake was broadly associated with less child's DNA methylation: with a q-value <0.05, we found more than 122,000 CpG sites associated with folate intake in common in both sets. Because a fraction of these CpGs were sign discordant between sets, we constrained our analysis to the most significant CpGs, in order to avoid false positive associations. The three most significant CpGs found in common in both sets were negatively associated with folate. First, cg21039708 (chr.14) in the gene OTX2OS1 (q <10-14 in both sets) is implicated in facial and eye defects. Second, cg13499966 (chr.2) in the gene CYS1 (q <10-14 in both sets) is implicated in cystic kidney disease and biliary liver fibrosis. Third, cg22664307 (chr.6) in the gene STX11 (q <10-14 in both sets) is associated with lymphohistiocytosis. Conclusions This study was the first to investigate the association between maternal folate intake during the peri-conception period and DNA methylation at birth with a genome-wide approach and it found numerous associations throughout the genome. The relationships between folate-sensitive CpG sites and DNA methylation events critical to childhood leukemogenesis will be considered in our subsequent research. Citation Format: Semira Gonseth, Ritu Roy, E. Andres Houseman, Adam de Smith, Mi Zhou, Seung-Tae Lee, Sébastien Nusslé, Amanda W. Singer, Margaret R. Wrensch, Catherine Metayer, Joseph L. Wiemels. Maternal folate intake at periconception and genome-wide DNA methylation modifications at birth in children. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 2763. doi:10.1158/1538-7445.AM2015-2763