Autoimmune diseases represent a significant source of morbidity and mortality and are a major financial burden to the economy. Evidence has emerged from cohort studies and animal models of disease of a link between viruses and many autoimmune conditions. The development of several promising vaccines and therapies targeting viral infection affords the tantalising possibility that new agents and existing antiviral treatments could be used to treat or even prevent autoimmune disease, and indeed some are already in Phase 1 clinical trials. However before embarking on large and expensive trials evaluating the effectiveness of such agents, the issue of whether viral pathogens trigger autoimmune disease needs to be established convincingly. The overall aim of this project is to investigate a possible causal link between six ubiquitous human viruses and the development of four autoimmune diseases using a statistical genetics methodology that is robust to confounding and reverse causality, and will be able to provide evidence in favour or against a role of viral infection in disease aetiology. Our approach involves finding genetic variants associated with antibody response to viral infection and determining whether the same variants also affect risk of autoimmune disease using a technique called Mendelian randomization. Should our results be consistent with a causal relationship, we expect that approaches aimed at controlling viral infection through vaccination, antiviral drugs or treatment with virus-specific T cell
infusions may become effective treatments or preventative strategies against autoimmune diseases in the future. Equally important, should we find no evidence for a causal relationship, then our results would suggest that expensive clinical trials involving anti-viral agents and/or vaccines to these pathogens are unlikely to succeed and shouldn’t be conducted- potentially saving hundreds of millions of dollars by avoiding costly studies likely to fail.

Information derived from social media platforms such as Twitter and Facebook has the potential to greatly increase our understanding of the origins and consequences of mental health and disorder. However, we need to make sure that when researchers link social media data in cohorts such as Children of the 90s this is done in a secure and ethical way that takes into account the preferences of the participants. We will engage with cohort participants and leaders to work out what is and what is not acceptable in linking social media data, and develop an ethical and technical system for doing this that can be replicated across UK cohort studies.

In the UK, surveillance by frontline services aim to identify children at risk of abuse or neglect, or of not having their needs met by their birth families. If a child is defined as being ‘in need’, their family will have social services involvement, and be provided with additional support. If this is inadequate to mitigate risk or to enable the child’s needs to be met, the child may be taken into care, usually to live with foster carers. A challenge when trying to safe-guard vulnerable children is how to identify them at an early stage, so as to expedite their removal from an abusive environment, or provide extra support to set the family on a more positive trajectory.

It is therefore important that early-life predictors of a child becoming in-need or looked-after are well understood. Patterns of health service usage may differ for vulnerable children relative to their peers, and as such could be a useful early marker of a child being at risk. We hypothesise that parents of children who become in-need or looked-after will be more likely to miss or cancel routine appointments for their child, to attend emergency out-of-hours appointments and A&E, and to not start or complete child vaccinations or other routine child health checks.

In our project we will use two birth cohort studies. We will identify which children are in-need or looked-after, and then we will examine whether their health service usage differed from their peers in early life by using record linkage.

The Deciphering Developmental Disorders Study has recruited a large cohort of ~13,000 patients with severe developmental disorders (DDs) and is conducting exome sequencing on them and their parents in order to try to uncover the causal genetic mutations (Wright et al., 2015; Deciphering Developmental Disorders Study, 2017). This is a highly heterogeneous cohort, and the most common phenotypes include intellectual disability (79%), seizures (19%) and autism spectrum disorders ASDs) (12%). We have discovered a damaging mutation in approximately 1/3 of the cohort that is thought to account for all or most of their phenotype.

However, it is becoming increasingly clear that common diseases, and in particular, psychiatric phenotypes, can have both common and rare genetic variants contributing to aetiology (Gaugler et al., 2014; Singh et al., 2016; Pardiñas et al., 2016). Thus, we have been investigating the potential role of common variation in DD, using genome-wide genotype data on ~8,000 DDD probands and ~13,000 ancestry-matched controls. We have shown, using these data, that common variants contribute a small proportion of the variation in risk of DD (heritability=9.3%; standard error=4.0%), and also that there is a significant genetic correlation (rg=-0.77; p=5.1E-6) between DD risk and educational attainment (EA) in the general population (i.e. correlation in the phenotypes due to shared polygenic background). In order to increase our power to perform further common variant analyses, we would like to add the ALSPAC data as additional controls, so we are seeking access to the genotype data as well as data on EA and IQ measures.

References:
Deciphering Developmental Disorders Study. Prevalence and architecture of de novo mutations in developmental disorders. Nature, 542, 433–438 (2017).

Gaugler et al. Most genetic risk for autism resides with common variation. Nat. Genet. 46(8) 881-885 (2014).

Pardiñas et al. Common schizophrenia alleles are enriched in mutation-intolerant genes and maintained by background selection. Biorxiv. (2016).

Singh et al. The contribution of rare variants to risk of schizophrenia in individuals with and without intellectual disability. Nat. Genet. 49(8):1167-1173 (2016).

Wright,C.F., Fitzgerald,T.W., Jones,W.D., Clayton,S., McRae,J.F., van Kogelenberg,M., King,D.A., Ambridge,K., Barrett,D.M., Bayzetinova,T., et al. Genetic diagnosis of developmental disorders in the DDD study: a scalable analysis of genome-wide research data. Lancet, 385, 1305–1314 (2015).

If a mother is obese during pregnancy, her offspring will also be at increased obesity risk. This might be explained by the hypothesis that environmental factors in the womb program the fetus for increased obesity risk later. However, the correlation between the mother’s and offspring’s body mass index (BMI) could also be due to genetic variation that is shared by the mother and offspring. If this is the case then interventions to reduce the BMI of mothers before pregnancy in order to reduce the obesity risk of the offspring may be less effective. Preliminary data from the Northern Finland Birth Cohorts suggest that genetics may explain a large part of the correlation between mother’s and offspring’s BMI. This project aims to use ALSPAC data to i) provide an independent replication of the results from the Northern Finland Birth Cohorts (using the Bivariate Genome Wide Complex Traits Analysis (GCTA) statistical model) and ii) account for maternal genetic factors by offering the opportunity to perform maternal genome wide complex trait analysis.

Polygenic risk scores (PGS) are being used more frequently with the emergence of new genome wide association study (GWAS) findings, opening new possibilities for understanding how genes associate with health and socioeconomic traits. However, much of the research using PGS may be underpowered, resulting in a failure to identify associations with traits or even spurious associations. Previous work conducted in similar projects has explored differences in the predictive power of a PGS when modelled using various strategies such as a phenotype measured at single or multiple occasions. We wish to extend this work in three ways. Firstly, as a proof of principle, we will compare the predictive power of different PGS using single and repeated measure phenotypes including mood (depression), substance use (cannabis use) and anthropometric measures (blood pressure and height). Secondly, we will extend upon previous Mendelian Randomisation analyses to determine the statistical benefits of modelling an exposure and/or an outcome in a repeat measure framework. Finally, we will investigate the potential for increasing power in GWAS’ by using a repeated measures framework to determine if power can be maximised by using repeated measures rather than increased sample size. To meet these ends, we require phenotypic and genomic data at multiple occasions for mood, substance use and anthropometric measures. ALSPAC is the perfect resource for conducting this research given its impressive archive of data and sample.

Previous studies in adults have identified genetic variants associated with a higher body mass index (BMI), but with a lower risk of adverse outcomes - type II diabetes, hypertension and heart disease. We aim to determine whether these associations, identified in adults, are also found in children. We will also search for the causal effects of this "favourable" adiposity across a wide range of traits and disease.

Cardiovascular disease (CVD) is a leading cause of morbidity and mortality. Obesity, a risk factor for CVD, is a growing public health problem even in young children, highlighting early childhood as a potential point of prevention for later life disease. Psychosocial stressors in early childhood, such as experiencing parental substance use or depression, violence, household chaos, or negative life events, have been linked to health problems such as CVD and obesity in adulthood. Changes in biology and behavior have both been proposed to link early stress and later health problems. One proposed mechanism between early stress and later CVD risk is dysregulation in cortisol, a hormone responsible for responding to stress and regulating daily processes like metabolism. It is currently unknown whether cortisol dysregulation mediates the association between early psychosocial stress and adolescent biological and behavioral risk for CVD and obesity. Risk factors for CVD include more inflammatory/metabolic markers in the blood, obesity, unhealthy diet and eating behaviors, lower physical activity, and greater sedentary behavior. The goal of this project is to understand whether cortisol dysregulation is a pathway between early childhood stress and risk for CVD as early as adolescence in order to identify targets for interventions that will reduce the burden of CVD and obesity on populations who have experienced early stress.

Exposure to maternal depression and anxiety during pregnancy and the postnatal period increases the risk of behavioral, emotional and cognitive difficulties in childhood, and also increases risk of mental health disorders, such as depression and anxiety, in adolescence. However, due to the complex nature of genetic and environmental contributions to the onset of mental health disorders, the process by which perinatal depression increases risk for offspring psychopathology remains unclear.

Recent research from our group and others suggests that the interplay between prenatal and postnatal depression may be important when considering childhood outcomes, and also that there may be specific gender differences in the development of childhood and adolescent mental health difficulties.

This project will use data from a large sample of children born in the early 1990s and their families. Measures of maternal depression were taken regularly during pregnancy and after birth, as were measures of childhood behavior, emotionality and cognitive function. Further, mental health was assessed at a number of time points during adolescence, when depression and anxiety are at peak onset. This information will allow us to question whether combinations of prenatal and postnatal depression are important in predicting offspring mental health, and whether such associations may be different for male and female infants. If this project does identify sex-specific associations between perinatal depression and offspring mental health, this may lead to targeted gender-specific interventions to avert the onset of mental health difficulties.