Methylation signatures of prenatal exposures for use in etiologic studies of prenatal cancer
Mentor: Dr. Logan Spector
Most pediatric cancers begin in utero, thus many etiologic investigations have examined exposure during pregnancy. However, due to the rarity of pediatric cancer many studies use the case-control design with maternal interview, which is subject to recall and social desirability biases. For instance, there are numerous reports of the association between maternal smoking and childhood cancers, but little confidence in the accuracy of results. Social desirability bias is even present in prospective or cross-sectional studies when it comes to a stigmatized behavior such as maternal smoking during pregnancy. We propose that an objective measurement of prenatal exposures through the examination of DNA methylation signatures may be superior to self-report. Briefly, DNA methylation is a process by which molecules are added to DNA, turning off gene transcription and thereby modifying gene expression. Methylation marks can be responsive to the environment and are semi-permanent. Since the introduction of genomewide methylation arrays in 2006, there is a growing literature on methylation signatures of prenatal exposures such as maternal smoking, gestational diabetes, and maternal stress. In order to inform future pediatric cancer studies, I propose to conduct a systematic review of the literature on methylation signatures of prenatal exposures. The review will include a ranking the quality of evidence supporting each signature, description of variability in signatures by tissue, and examination of whether signatures persist into older ages. Given eight weeks of focused effort, I will submit the review for publication by the end of my ALSF POST position. Additionally, a full list of genomic coordinates of methylation signature sites (CpGs) will be published as supplemental material for the ease of use by other researchers.
