Medications during pregnancy: effects on neurodevelopmental disorders in childhood and epigenetic outcomes

Medications during pregnancy: effects on neurodevelopmental disorders in childhood and epigenetic outcomes Medications are marketed without sufficient information on safety during pregnancy, with potentially harmful consequences for the child. Recent epidemiological studies have shown that exposure to common drugs like paracetamol and antidepressants during pregnancy is associated with neurodevelopmental outcomes (e.g. ADHD). In this project, which is part of PharmaTox (www.mn.uio.no/farmasi/english/research/groups/pharmatox/), we study the mechanisms that might be involved. Specifically, whether exposure common drugs during pregnancy involves epigenetic changes (i.e. DNA methylation). Epigenetic modifications are essential during normal development and involved in regulation of genes. The prenatal period is a critical time for establishment of DNA methylation patterns and particularly prone to environmental influence by medications. Recently, we showed for the first time that long-term exposure during pregnancy was associated with differences in DNA methylation in children with ADHD included in the Norwegian mother and child cohort study (Moba). At this stage, it is not possible reveal the causal relationship between differences in DNA methylation and drug exposure in pregnancy. Hence, a central part of this study involves establishment of a human embryonic stem cell (hESCs) and neural differentiation model. This model will be valuable for the causal interpretation of the associations found in MoBa. Differentiation of hESCs involves development of cells at different stages, with different maturity displaying different DNA methylation and gene expression patterns. This will potentially confound analyses of DNA methylation and gene expression using this model. Hence, an important part in the establishment of this model has been to fully characterize the cellular composition at different developmental stages during differentiation. To do this, we are sequencing single cells (so-called single cell sequencing, 10x Genomics). These data will be crucial to elucidate and account for differences in cellular composition in the DNA methylation and gene expression analyses of drug exposure using this model. Currently, there is no publicly available protocol for a hESC model to study drug neurotoxicity and this work will be published as a method paper. In addition, the hESC differentiation model will be used to study the effect of paracetamol and antidepressants. It is well known that DNA methylation at neighbouring positions in the genome is correlated. This correlation structure is often neglected in analysis when identifying differentially methylated positions. Including the correlation structure in the analysis can increase the power of the study. To model the correlation structure, we use an existing statistical framework (Integrated Nested Laplace Approximations, Bayesian regression tool) previously not used for such analysis. To better understand the mechanisms involved, we are performing an exploratory analysis where we model the correlation structure and investigate the association of different covariates. This is done in a large Finnish data set (1200 samples). Since covariates such as age, BMI and smoking are associated with DNA methylation, these analyses are important to identify covariates to control for when investigating differences in correlation structure. Preliminary results show that the correlation structure is associated with age, and further analysis will investigate other covariates and whether we can replicate this in an independent data set. Further, we plan to use this model to identify differences between children exposed to paracetamol with ADHD and unexposed healthy children in MoBa.

Contribution to the research field 1. Developed a reference data set for cord blood DNAm. This is available as a package in R and used widely by the community. In addition, paper describing advice for the community for using reference data sets. 2. Demonstrated a link between newborn DNA methylation, ADHD and paracetamol intake, the first time this has been shown. 3. Development of a new statistical method for analysing DNAm data. 4. Review and recommendations for EWAS studies. 5. Paper describing the challenges with replication in epignome-wide association studies. 6. Development of a protocol to differentiate human embryonic stem cells (hESCs) to neuronal cells. Provide open access web tools for data analysis. This protocol will be valuable for the assessment of drug toxicity and basic research in early neuronal development. Interdisciplinary research The project was very interdisciplinary, consisting of 13 directly involved researchers with diverse backgrounds including genetics, epigenetics, pharmacy, cell biology, neuroscience, data science, and clinical fields. These researchers are based at 8 different departments in Norway (Oslo University Hospital, University of Oslo and Norwegian Institute of Public Health. It also included two collaborators based at Karolinska in Sweden. Societal impact 1. Pregnant women are not included in randomized clinical trials due to ethical considerations and thus medications are marketed without sufficient information on safety during pregnancy. Our demonstration of a link between DNA methylation, paracetamol use during pregnancy and ADHD, has clear implications for women and children. We are currently analysing data for antidepressants in a similar study design. It should be noted that studies such as these are a very active area of research with the scientific community. 2. Development of an ES cell protocol for the study of neuronal development. This hESC model can be used to screen other pharmaceuticals commonly used during pregnancy. The development of this methods has great potential for to extend the use of in vitro methods (ie not animal-based) to assess drug toxicity in neurodevelopment.

Pregnant women are not included in randomized clinical trials due to ethical considerations and thus medications are marketed without sufficient information on safety during pregnancy. Currently, more than 250 000 children are diagnosed with neurodevelopment delays in the EU region each year. Why? Is there a link? We hypothesize that medications prescribed during pregnancy may play an unrecognized role in the development of neurodevelopmental disorders, such as delayed psychomotor development and attention deficit hyperactivity disorder (ADHD). Several studies provide the proof of concept that DNA methylation patterns from blood samples in newborns may serve as early markers for long-term neurobehavioral disorders. Furthermore, lower DNA methylation in newborns was recently linked to more ADHD symptoms in 6-year-old children. These initial human findings requires replication, but if proven correct, suggest that exposure to psychotropic medications may be associated with specific measurable DNA methylation profiles in newborns and that these may be used as early biomarkers of neurodevelopmental disorders, enabling early interventions. Using data and samples from the Norwegian Mother and Child cohort (MoBa) we will investigate the link between medications taken during pregnancy, epigenetics and neurobiological outcomes.The size and richness of the available questionnaire data and biological samples from more than a hundred thousand pregnancies, and linkage to large patient registries, makes this project unique. Only large prospective cohort studies like this can measure the small effect sizes associated with epigenetic changes. The final outcome will be fundamentally new knowledge about how medications affect the developing unborn child and provide opportunities for future research in a new field of pharmacoepigenetics. The implications will not only be for how we treat pregnant women, but how we use pharmaceuticals in any patient population.