Epigenetic changes in developing fetuses linked to smoking during pregnancy
Michael Cowley, Brian Abraham, Christopher Gregg, Peng Jin| Epigenetics Experts | March 31, 2016

Smoking while pregnant can cause epigenetic changes to the DNA of newborns, according to a large scale meta-analysis of thirteen previous studies. Analyzing data from 6,685 mothers and newborns, researchers found significant epigenetic differences between children whose mothers were sustained smokers and those that were non-smokers. The authors say their results could help explain the higher risks of health problems such as birth defects and cancer in children whose mothers smoked during pregnancy.


Dr. Michael Cowley, Assistant Professor, Department of Biological Sciences, and Center for Human Health and the Environment, North Carolina State University (webpage):

Expertise: effects of environmental exposures during early life on the epigenome and its impact on adult health

“This paper by Joubert and colleagues identifies thousands of epigenetic changes in newborns associated with maternal smoking during pregnancy. These epigenetic changes to the offspring’s DNA could explain how children born to mothers who smoke are more likely to have reduced birth weight, reduced lung function, and increased risk of several diseases.

“While previous studies have demonstrated links between maternal smoking and epigenetic changes, a key limitation is the small numbers of participants in each. It is challenging, logistically and financially, to recruit sufficient human subjects to enable associations between exposures and epigenetic changes to be determined with confidence. The power of this new study, which represents significant progress in this field, lies in the integration of data from thirteen separate studies (a ‘meta-analysis’), greatly increasing the number of samples studied.

“Another challenge of such meta-analyses is ensuring reproducibility between the laboratories collecting and analyzing the data. This new study uses multiple, robust approaches to reduce errors and standardize techniques, and thus we can be confident about their findings.

“Could smoking-associated epigenetic changes be impacting gene expression, potentially explaining some clinical outcomes such as reduced birth weight? Joubert and colleagues have made a significant, albeit preliminary, attempt at addressing this question, finding that several genes near a subset of these epigenetic changes show altered expression. Whether these effects actually contribute to increased disease risk cannot be determined from this study, but this demonstration that smoking-associated epigenetic changes have functional relevance is important. Future work will be required to demonstrate causative links between smoking-associated epigenetic changes and offspring disease, but this study provides a reliable foundation on which to build.”


Dr. Brian Abraham, Postdoctoral Fellow, Whitehead Institute for Biomedical Research, Massachusetts Institute of Technology (webpage):

Expertise: epigenetics and cancer, bioinformatics

“When a mother smokes during pregnancy, the risk of her child developing one of a number of health problems rises, but how maternal smoking leads to diseases is not understood. A consortium of multiple labs asked if DNA methylation – chemical flags added to specific genes that indicate that a protein should or should not be made – could explain how maternal smoking affects the fetus. The consortium statistically linked methylation at a few thousand spots on the genome to whether or not the mothers smoked during pregnancy. They further proposed that gene expression changes caused by DNA methylation could contribute to a handful of diseases.

“Genome-scale statistical studies, no matter how big they are, can generally only suggest hypotheses that need to be validated with biological and chemical experiments. This study proposes that some of the thousands of methylation events in a genome link maternal smoking to childrens’ diseases. However, the links between methylation events, gene expression changes and specific diseases needs experimental confirmation before causality and clinical relevance can be claimed.

“It is unsurprising that the study confirmed and discovered DNA methylation events that are linked with such a profound environmental factor, especially in such a large sample group. But all genome-scale analyses have the potential to falsely identify incorrect hits while overlooking true ones, and there are several models for how these methylation events happen that may be mere coincidence. The use of multiple labs’ methods does suggest at least the link between certain methylation events and maternal smoking are consistently true. Further work will be needed to show that the methylation events that differ between the children of smoking and non-smoking mothers 1) alter gene expression and 2) actually matter for the diseases proposed here.”


Dr. Christopher Gregg, Assistant Professor of Neurobiology & Anatomy and Human Genetics, University of Utah (webpage):

Expertise: genome analysis and epigenetics with applications to neuroscience and behavior

“The study uses a very well established platform to measure DNA methylation changes in the blood and the statistical methods are also reasonably well established in the field. Importantly however, by analyzing data from over six thousand people, the authors are able to more effectively rule out potentially confounding factors, such as maternal age and socioeconomic status. It is difficult to control for all potential confounding factors, which is why such large studies are required to be confident in the results. Therefore the methodology in this study is not novel, but the size of the study and the striking impact of maternal smoking on DNA methylation in offspring underlies the importance and high impact of the work.

“The clinical relevance of these findings is that they provide insights into how maternal smoking impacts the fetus and provide initial insights into the mechanisms. The authors determined that sustained smoking during pregnancy has the most substantial effects on the epigenome of the fetus. It is well established that pregnant women should not smoke, but these new results reveal that smoking during pregnancy leaves a lasting mark on the genome that persists into childhood, and identifies the sites and genes in the genome that are especially susceptible to these effects.

“Large association studies of this kind sometimes reveal that previous findings from smaller studies could not be reproduced. However, this was not the case here and strong evidence is provided for reproducible DNA methylation changes due to maternal smoking at a gene called AHRR or the aryl hydrocarbon receptor repressor, which plays a role in cellular responses to dioxins, toxic compounds found in cigarette smoke. In addition, thousands of new sites in the genome of newborns that exhibit DNA methylation changes in response to maternal smoking are identified. A smaller number of these sites are still impacted in older children.”


Dr. Peng Jin, Professor of Human Genetics, Emory University (webpage):

Expertise: functions and mechanisms of epigenetics and noncoding RNAs in neurodevelopment and neurodegenerative disorders.

“I am very impressed by this study done by the PACE consortium. The authors performed a meta analysis of DNA methylation studies using thirteen distinct study groups, and showed the association between maternal smoking in pregnancy and newborn blood DNA methylation. They identified over 6,000 regions that showed altered methylation, some of which are relevant to diseases later on in life. The degree of methylation changes for each gene region is small, but statistically significant. The authors performed many detailed analyses and their conclusions are quite strong.

“The caveats of this study could include the following: they used a specific platform (Illumina 450K array) and the information obtained through this could be limited since the genome was only partially surveyed. Also blood DNA was used which has another epigenetic mark called 5-hydroxylmethocytosine (5hmC). This is different from 5-methylcytosine (5mC) which is the more common epigenetic mark on genomic DNA. The method used in this paper would not be able to distinguish 5hmC from 5mC and this is important as they two marks have distinctly different effects on gene expression. Thus the changes they observed might be due to the change of 5hmC as well.

“Overall, I think this meta-analysis provides a much stronger link between smoking in mothers and epigenetic changes. However, how the epigenetic changes (which in many regions are quite small) could lead to physiological alterations or diseases remains to be determined. The authors did examine older kids and observed similar changes, which potentially support the notion of “Fetal-Origin of Adult Diseases”.”


Declared interests (see GENeS register of interests policy):

No interests declared.



DNA Methylation in Newborns and Maternal Smoking in Pregnancy: Genome-wide Consortium Meta-analysis” by Bonnie R. Joubert et al, published in American Journal of Human Genetics on Thursday, 31 March 2016