Proposal summaries
B2403 - Unravelling the link between depression and violence in adolescents - 05/03/2015
Background
Depression is one of the most widespread mental health concerns in the world (Kessler & Bromet, 2013). It is often linked to inwardly directed harm such as self-injuries and suicide (Cuijpers et al., 2014; Hawton et al., 2012; Hawton et al., 2003). The association between depression and violence against others remains uncertain. Emerging evidence is showing that depression and violent behaviours are associated. Positive depression-violence associations have been reported in cross-sectional community survey and register-based investigations (Coid et al., 2006; Wallence et al., 1998; Swanson, 1990), and longitudinal cohort studies (Arseneault et al., 2000; Monahan et al., 1983). Recent research on a total population cohort has shown that risk of violent crime was increased in individuals with depression even after adjustment for familial, socio-demographic and individual factors (Fazel et al., 2015). To date, however, evidence regarding this link has been largely limited to adult and selected samples of individuals in secondary or tertiary care. It is not clear whether the magnitude of depression-violence link is different according to the developmental stages of an individual and whether it is related to more sensitive markers of the illness course. Further, it is not known the mechanisms of any association.
Aim: This project sets out to dissect the depression-violence link with a developmental perspective. Subsequently, this project proposes to test the moderating role of situational factors (i.e., stressful life events) and the mediating role of self-regulation abilities (i.e., inhibition, locus of control), self-harm, and peer victimization. The gained knowledge will be used to facilitate the development of tailor-made prevention and intervention strategies to mitigate the link between depression and violence.
B2402 - An examination of the association between school engagement at 10-11 yrs and sexual health and behaviours at 21 yrs - 05/03/2015
Background
Identifying factors in childhood that are associated with later poor sexual health (e.g. early sexual behaviour, sexually transmitted infections, and teenage pregnancy) may be valuable in recognising and targeting populations at risk (Parkes et al., 2014). Lower school grades and school adjustment problems in childhood have been proposed as potentially important determinants of early sexual initiation (Zimmer-Gembeck et al., 2008), and recent studies using ALSPAC data found persistent school dislike in primary school to be associated with early sexual behaviour and sexual risk taking, reported at age 15 years (Parkes et al., 2014) and low educational attainment at 11 years to be associated with Chlamydia infection at 18 years (Crichton et al., 2014). We wish to extend this work, to examine the association between school academic achievement and school engagement in childhood with sexual health outcomes reported at age 21 years.
Aim
To determine if academic achievement or school engagement in Year 6 (age 10-11 years, final year of primary school) predict self-reported sexual health/behaviours at age 21 years.
B2401 - Maternal substance use in pregnancy and childs educational attainment at age 16 years - 05/03/2015
Background
Two previous studies using ALSPAC data have examined the association between prenatal alcohol exposure and educational perfomance at Key Stage 2 (the last year of primary school, aged 11 years approximately). One study contrasted maternal and paternal alcohol intake in pregnancy, and found that maternal binge drinking was associated with lower KS2 scores but paternal binge drinking was not, thus suggesting an intrauterine mechanism (Alati, 2013). In support of this, a second study found children of mothers whose genotype predisposed them to lower alcohol consumption during pregnancy had a better performance at KS2 than the children of mothers whose genotype predisposed to heavier drinking; a finding which suggests positive associations in observational studies between self-reported moderate alcohol intake in pregnancy and child's educational attainment are explained by residual confounding by factors associated with socio-economic position (Zuccolo, 2013).
Prenatal cannabis exposure has also been shown to have a negative effect on intellectual development at age 6 (Goldschmidt, 2008), resulting in lower educational attainment at age 14 years (Goldschmidt, 2012). This association has not previously been explored using ALSPAC data. Intrauterine exposure to tobacco smoke and child's IQ at 8 has been studies using ALSPAC; associations between maternal and paternal smoking and child's IQ were similar, suggesting any relationship between smoking in pregnancy and child's IQ may not explained by intrauterine exposure (Alati, 2008).
Aims
This proposed study would extend previous work to determine if any negative impact of intrauterine exposure to alcohol, cannabis or tobacco on educational attainment persists to Key Stage 4 (GCSEs, taken when aged 15-16 years). [The inclusion of cannabis as an exposure will depend on whether numbers are adequate].
B2400 - Lookup of methQTL signals for variants associated with anthropometric traits and identified in the UK10K project - 05/03/2015
The UK10K project represents one of the first large scale applications of next generation sequencing to population based epidemiological samples and the examination of complex phenotypes. The objectives of this work are to record whole genome sequence variation at and below 1% minor allele frequency, to provide an imputation reference and to use this, not only to provide a resource for the scientific community (both genotypes and phenotypes), but also to examine genetic associations across a spectrum of genetic variation.
The study is composed of two samples drawn from European population based epidemiological studies (The Avon Longitudinal Study of Parents and children and Twins UK) and forms a collection of nearly 4000 participants now with whole genome sequence data and phenotypes (along with complementary imputed internal replication data sets). A total of 1,990 individuals from TwinsUK and 2,040 individuals from ALSPAC were consented for sequencing. Variant sites and genotype likelihoods were called using samtools and genotypes were refined and phased using BEAGLE, with similar procedures to the 1000 Genomes Project.
In addition, both ALSPAC and TwinsUK consented to release phenotype data related to cardiovascular disease as a public resource for the association analyses. Twelve anthropometric traits including height, weight, BMI, waist-hip-ratio, waist-hip-ratio (BMIadj), waist circumference, waist circumference (BMIadj), hip circumference and hip circumference (BMI adj) and DXA measures (total fat mass, total lean mass and trunk fat mass) phenotypes were released. The association analyses for the anthropometric traits are currently reaching a critical phase for the singlepoint associations, and also for those more rare variants with analyses being undertaken using variant aggregation techniques such as SKAT. Consequently, we are following up possible avenues for unraveling the underlying molecular mechanism of the identified genetic associations. To this end, we are writing to seek a collaborative arrangement efforts to continue our analysis of both single point and rare variant associations using expression (already part of UK10K agreement) and methylation data.
B2399 - Association between brain anatomy and genetic risk for mental disorders in healthy volunteers - 05/03/2015
This project aims at combining multiple MRI datasets obtained in the same magnet (scanner) through very similar acquisiton procedures in order to investigate the association between anatomical markers and genetic risk factors for major mental disorders (psychosis, bipolar disorder and Alzheimer's disease) and associated polygenic pathway scores (e.g. dopaminergic transmission) in healthy participants. The combination of ALSPAC's and our own samples will constitute a unique resource to investigate neuroimaging and genetic associations, since it will include data from potentially up to 1,000 participants avoiding the counfounds of previous similar efforts from other groups using different magnets and not always equivalent data acquisition methods.
B2398 - Genomics of Neuropsychiatric Traits in Children and Adolescents MRC IEU project - 05/03/2015
Aims & Hypotheses
The overall aim of our research is to identify genetic variants associated with a neuropsychiatric traits in children using community-based samples. We are interested in three specific traits: obsessive-compulsive traits, ADHD traits and response inhibition as measured by the Stop-Signal Task. Given that response inhibition is an endophenotype for ADHD, we expect that identifying genetic variants associated with response inhibition will also help us identify genetic variants associated with ADHD.
B2397 - Metabolomic profile of Familial hypercholesterolaemia - 05/03/2015
Familial hypercholesterolaemia (FH) is an inherited condition affecting at least one in 500, although current studies suggest that the frequency is much higher (1 in 250). Affected individuals are characterised by a significant increase of LDL cholesterol levels from birth. It is ussually caused by mutations in one of three genes (LDLR, APOB or PCSK9), however recently a polygenic cause of high LDL-C has been found to mimic the clinical FH phenotype. Our preliminary results showed that individuals with the polygenic form of FH have significantly higher triglyceride levels in comparison to the monogenic FH patients, which suggest that different metabolic pathways may be involved in the development of hypercholesterolaemia in these two forms of FH. We therefore would like to:
1) Compare the metabolomic profiles of the polygenic vs the monogenic FH to see if there are different pathways involved in the two groups
2) Focus on the known FH associated loci (LDLR and APOB) and compare the FH mutations versus common occuring polymorphisms in the same locus which have previously been associated with lipids in large scale association studies. This will help us test if the FH mutations are associated with metabolic changes different to those attributed to their genes in studies using healthy population samples.
Replication will be available through UCLEB where similar information is available.
B2395 - External validation of two studies on risk factors and prediction of childhood asthma - 26/02/2015
Background: We recently published two studies in which we developed a novel and robust tool for predicting asthma at school age in preschool children with wheeze or cough (REF1), and determined the association between breastfeeding and school-age lung function (REF2). The two studies made use of population-based cohorts from children of Leicestershire, United Kingdom.
Aim: We want to externally validate the findings of those two studies using the ALSPAC data set.
B2394 - Socioeconomic gradient of smoking comparing families where one or two parents smoke - 26/02/2015
We seek approval to repeat this analysis among the four English STELAR cohorts. Data requested are (at a given timepoint) socioeconomic status, number of smoking adults in the child's house, number of adults in the house (this addresses the issue of single parent families). In some of the cohorts, these outcomes may be available at more than one time point and this would allow us to see whether these observations are static or change over time.
B2393 - Using a genetic risk score for Coronary Artery Disease to investigate casusal influences on the metabolome - 19/02/2015
Multi-locus genetic risk profiles, or genetic risk scores (GRS) aggregate data from multiple SNPs to provide a robust genetic instrument for traits of interest. They are more powerful than single SNP GWAS scores as the use of multiple sites reduces the effects of pleitropy and minor allele frequencies. GRS have been developed for many conditions including CAD and obesity. They provide a useful genetic tool for probing the causal pathways of multifactorial diseases.
In this study we intend to use multiple GRSs developed from CARDioGRAMplusC4D SNPs as a genetic instrument of CAD to investigate associations between predictors of CAD and differences in metabolite profile in ALSPAC mothers and children. GRSs are the exposure variable in this case, and altered metabolite profile is the exposure. We will use a range of GRSs developed using varied SNP significance threshlds to demnstrate the robustness of GRSs a genetic instrument, while investigating the genetic influence on the metabolism at a pathway specific and global level.
Three GRSs for CAD will be produced based on highly strict, moderate, and lenient SNP significance thresholds using CARDioGRAMplusC4D data. A selection of the most significant SNPs will also be analysed seperately. The individual SNPs and strict GRS should reveal how specific SNPs or groups of SNPs influence isolated metabolic pathways related to CAD, while the more lenient GRSs demostrate the wider influence of genetic variants on global metabolism. These GRSs will be regressed against the top ten principle components of NMR metabolite data from mothers and children. Principle components that are significantly associated can then be deconstructed to reveal specific changes in metabolite levels. By using principle components instead of individual metabolite concentrations, the number of analyses is significantly reduced, and the components themselves may represent biologically meaningful metabolic pathways.
B2392 - Long-term outcomes of children with borderline personality disorder traits at 11-12 years - 19/02/2015
BACKGROUND: Recently there has been a growing body of research examining Borderline Personality Disorder in youth, i.e., childhood and adolescence (Hawes, 2014). Nevertheless, diagnosis in this age group remains controversial, partly due to concerns regarding the validity of the construct (Griffiths, 2011). Predictive validity reflects the degree to which BPD in youth is prognostic of future impairment, and is considered a crucial aspect of the validity of a construct (Van Os et al., 2009). While, a small body of literature has considered the stability of BPD symptoms over time, there are few studies that have examined the long-term clinical and psychosocial impacts of the disorder. The few studies that have reported negative outcomes in youth with BPD symptoms in community populations (e.g., Cohen et al., 2007; Winograd et al., 2008) utilised ad hoc, self-created assessments of BPD rather than established, validated tools. Furthermore, these studies did not consider the mechanistic pathways from early BPD symptoms to outcomes in late adolescence/early adulthood (e.g., do BPD symptoms increase risk of bullying exposure, subsequently heightening risk of depression symptoms?).
AIMS: To explore to what extent BPD symptoms at 11-12 years of age are predictive of future psychopathology and negative psychosocial outcomes, and to examine mechanisms via which increased risk manifests.
B2391 - Investigating putative risk factors for Alzheimers disease using Mendelian Randomization - 19/02/2015
The aim of this study to investigate the role of potential risk factors for Alzheimer's (AD) using Mendelian Randomization. Smoking, hypertension, increased body weight, dyslipidemia and type 2 diabetes have all been suggested as risk factors for AD but the results of epidemiological studies on them have been inconclusive.
In this project we are planning to use genetic variants that are robustly associated with the above exposures to test if they have a causal effect on AD through Mendelian randomization (MR). For blood pressure, BMI, lipids and type 2 diabetes the method of analysis will be a 2-sample MR in which the "reduced-form" estimate (the coefficient for the association between the IV and the outcome) and the "first-stage" estimate (the coefficient for the association between the IV and the exposure) are obtained from non-overlapping samples (Pierce and Burgess 2013). The instrumental variables for the MR analysis will consist of an allelic score calculated based on the results of independent genome-wide association studies of the above traits.
Polygenic risk scores will be calculated based on the results of published GWAS studies. These identified risk alleles will be used to calculate a polygenic score for each individual in an independent (target) sample (ALSPAC), corresponding to the mean number of score alleles (weighted by the logarithm of the odds ratio) across the set of SNPs, using PLINK. Standard linear regression models will be used to calculate coefficients for the association of the polygenic scores with the trait of interest in the ALSPAC target sample (first-stage coefficients). The reduced-form coefficients of the genetic variants/polygenic scores will be calculated in the International Genetics of Alzheimer's Project (IGAP) sample, which is independent of the target sample where the first-stage coefficients would be calculated. The IGAP sample is a collaboration of 4 groups in Europe and USA with approximately 60,000 individuals (either with AD or controls). Genetic data are available on all of them. We have obtained permission to perform this analysis in the IGAP sample.
B2390 - A study of origins correlates and determinants of locus of control - 12/02/2015
Locus of control (LOC) refers to the connections individuals perceive between their behavior and what happens to them. If they perceive such connections they are internally controlled; but if they see their outcomes as due to luck, fate or chance they are externally controlled. Even though researchers have found an internal LOC to be related positively to important aspects of human life including academic achievement, business success, physical and mental health and, in a United Nations study of 84 countries, to happiness, little is known about how internal and external LOC develop, their stability over time or whether they can be changed. Such information is important because over the past 30 years the average locus of control score for adults and children has become more external. We need to know why, and how to change that negative trend. The data to be used in this project include measures of LOC of parents during pregnancy and 16 years later, and of their children at ages 8 and 18 on over 12000 families. It is the only large study in the world that includes multiple assessments of LOC for children and their parents linked to relevant outcome measures. Statistical analyses will provide crucial information on how to foster internality and hence develop interventions.
B2389 - Coventry Reading Engagement Self-Belief and Teaching Project - 12/02/2015
The aims of this project are to examine the links between attitudes, self-belief and literacy behaviour and attainments. It is likely that these factors form a reciprocal relationship - i.e. success at reading means that a child believes that they can do well at reading, and they are therefore motivated to a) try harder when decoding unknown words and b) read for pleasure. In turn, both of these activities are likely to improve reading attainments. Previous studies have indicated a correlation between these areas but it is difficult to establish the cause of this correlation. Large-scale longitudinal work is required to assess whether these relationships exist and whether they vary with ability or age (for example, it could be that self-belief is more important for poor readers or in adolescence)
B2382 - Defining the mitochondrial DNA genetic bottleneck by studying the inheritance of low-level heteroplasmy between mothers and offspring - 12/02/2015
Our aim is to study low-level heteroplasmy in a large number of mother-child pairs using DNA samples already extracted for ALSPAC genotyping studies. We estimate that greater than 1000 pairs will be required to carry out a meaningful experiment, based on our pilot data from existing data sets. We will use ultra high depth next generation sequencing (IlluminaMiSeq) to carry out this experiment, allowing the detection of greater than 0.5% heteroplasmy from a standard genomic DNA sample. Statistical analysis will be carried out by Dr Ian Wilson at the Institute of Genetic Medicine. We will model the genetic bottleneck as done previously,7 and determine whether the background mtDNA sequence influences the rate of segregation of mtDNA heteroplasmy in healthy controls. We will then compare this to data acquired from humans transmitting pathogenic mtDNA mutations to determine whether they behave differently.8 This will advance our understanding of the underlying biology, and will be directly relevant to current work aimed at preventing the transmission of mtDNA mutations.
B2386 - The development of an integrated OMICS signature that links in utero air pollution to growth and cardiovascular health - 05/02/2015
Hypotheses:
In utero particulate matter exposure has been linked to adverse pregnancy outcomes such as low birth weight and intrauterine growth retardation, which in turn have been linked to adult cardiovascular disease and metabolic abnormalities such as obesity, hypertension, insulin resistance and glucose intolerance. These findings suggest that adverse intrauterine conditions that promote reduced birthweight, such as air pollution exposure, may also promote lifelong susceptibility to cardiovascular diseases. Although it is becoming clear that in utero exposure to environmental stressors plays a role in fetal metabolic programming, the mechanisms are still unclear. The investigation of OMICS measurements at different molecular levels provides an insight into the biological pathways involved in the process linking in utero air pollution exposure to early life growth and cardiovascular development.
Aims:
The objective of this project is to derive a molecular pathway linking in utero air pollution exposure to fetal and childhood growth and to study potential consequences on the development of the cardiovascular system. I address four main objectives:
1) The investigation of prenatal and postnatal growth in association with in utero air pollution exposure.
2) The derivation of biomarkers (within in each OMICS level) linking prenatal air pollution exposure to growth trajectories in the first years of life.
3) The identification of common patterns across different OMICS levels (cross-omic analyses) in order to get an insight into the biological pathways involved in the association between in utero air pollution exposure and growth.
4) The investigation of the role of early life growth and the derived OMICS signature in the association between prenatal air pollution exposure and cardiovascular endpoints in childhood and early adulthood.
B2385 - Genome wide association study for physical activity and sedentary behaviour - 05/02/2015
This analysis will form part of an expanded meta-analysis for self-reported moderate and vigorous leisure time activity and sedentary behaviour. Our first goal is to capture a dichotomous trait corresponding to moderate and vigorous leisure activity. Such activity would not include occupation-related activity (i.e. shoveling, heavy-lifting) and/or light leisure activity (i.e. walking, gardening). The second goal is to capture all aspects of sedentary behaviour for which we defined four traits; three dichotomous traits, reflecting sedentary behaviour at work, home and during commuting, and one continuous trait: TV viewing.Separate GWAS analyses will be run for male and female offspring and the mothers. Covariates included in the models are age and body mass index.
B2384 - Novel Epidemiological Methods to Infer Causal Effects of Risk Factors on Neuropsychiatric Cardiovascular Disorder - 29/01/2015
BACKGROUND
This project focuses on developing methodological applications for analysing "omics" data resources (genome-wide genotypes, metabolomics, the epigenome), an area that has exciting prospects for observational epidemiology (Brion et al., Curr Epidem Rep 2014). We have previously published work indicating that genome-wide allelic scores can be used to data-mine large numbers of associations between biological intermediates and disease-related outcomes and screen for potentially causal relationships (Evans, Brion et al., PLoS Genetics 2013). We would like to build on this work and investigate the use of similar allelic scores in studies based on metabolomic and epigenetic measures. In addition, we would like to develop and implement novel applications of Mendelian Randomization to these genome-wide genotype, epigenetic and metabolomic measures, such as by implementing bidirectional MR (Welsh et al., J Clin Endocrin Metab, 2010) and MR for mediation (Relton & Davey Smith, IJE 2012). We would then implement these methods to test causal relationships involving blood methylation (epigenetic) markers and metabolomic measures.
AIMS
1.To develop and test novel epidemiological approaches, such as Mendelian Randomization and data-mining approaches, that exploit the availability of high throughput biological data (genome-wide single nucleotide polymorphism (SNP) data, epigenetic, metabolomic).
2.To apply these novel methods to infer the causality between risk factors, such as environmental exposures and biomarkers, with cardiovascular and psychological outcomes, through potentially mediating epigenomic and metabolomic pathways.
B2383 - MR of blood pressure and NMR metabolites - 29/01/2015
Higher blood pressure is associated with increased cardiovascular and renal disease and premature mortality. Although some of the biological pathways and disease endpoints linked to blood pressure changes are well understood, a large number of them remain unexplained. Higher blood pressure is associated with other risk factors for cardiovascular and renal disease, including dyslipideamia, hyperglycaemia and insulin resistance (referred to collectively as metabolites) and higher levels of inflammatory markers, which might mediate any causal effect of higher blood pressure on later disease. But associations with these other risk factors and their joint role in causing later disease could reflect several different alternatives:
1.Blood pressure causes variation in these other risk factors and via these effects causes later CVD and other diseases.
2.These other risk factors cause variation in blood pressure and the primary causal agents are the metabolites/inflammatory markers with blood pressure being more downstream and proximal to disease endpoints.
3.There is a bidirectional causal relationship between BP and metabolites/inflammatory markers
4.There is no causal relationship between blood pressure and metabolites/inflammatory markers, but rather the association is confounded, for example by socioeconomic position, lifestyle characteristics and obesity. .
In preliminary analyses, we have already identified a number of metabolites from the NMR platform that are strongly correlated with BP. We want to extend this work by using Mendelian randomization (MR) to determine whether the relationships are causal and if so which direction they are in (i.e. to distinguish which of the 3 alternatives listed above is most likely).
With UCLEB we can increase the number of available individuals to more than 30 000, all with blood pressure measurements, GWAS or other array genotyping, and NMR metabolites and inflammatory markers.
We will use the known BP genetic instruments to assess whether BP is causally related to the metabolites and inflammatory markers (alternative 1 or 3 above vs 4) and metabolite instruments from recent GWAS to test whether metabolites are causally related to BP (alternative 2 or 3 vs 4).
B2381 - Trajectories of Brain Injury in the ALSPAC cohort - 22/01/2015
Aims:
The aim of the study is to investigate the trajectories of traumatic brain injury (TBI) from childhood into adolescence using the ALSPAC cohort. In particular the study will explore the impact that a single TBI event and recurrent TBI events will have on aspects of social cognition, risk behaviour engagement and general cognition. The control groups will include participants who have had a broken bone/fracture and also participants who have never had an injury. There will be exploratory analyses conducted to see whether the detrimental effects of recurrent TBI events are greater than for a single TBI event.
Social cognition can be broadly conceived as the ability to understand other people through emotion perception and empathy and also by inferring the beliefs, intentions and feelings of others 1 Following a TBI event, many facets of social cognition have been shown to be impaired, such as the recognition of facial emotional expressions 1-4; however, current evidence is hampered by small, heterogeneous sample sizes 1.
Engagement in risky behaviour such as substance use has been implicated as an outcome of TBI 5-8. One large cohort study demonstrated that a TBI injury requiring hospitalisation between the ages of 0-5 years old increases the risk for adolescent substance use 5; while findings from another birth cohort indicate that drinking to intoxication is more common among 14 year olds who have experienced a TBI compared to those who have not 6. Ilie and colleagues 7 found that high school students who had experienced a TBI had higher odds of using illicit substances, were twice as likely to consume hazardous amounts of alcohol and were at greater risk for cannabis dependence. Likewise, in a group of incarcerated youths, those with a TBI history were found to be more likely to engage in risky alcohol and cannabis use as well as weekly use of another illicit drug prior to incarceration; this was also the case for participants reporting multiple TBI events9. In comparison to a single TBI event, individuals who have experience two or more TBI events have reported more frequent alcohol, tobacco and illicit drug use 8.
The sequelae of TBI is characterised by cognitive problems 10. Corrigan and colleagues 11 compared different characteristic profiles of TBI histories in persons with substance use disorders and found that a TBI event occurring between the ages of 6-10 may lead to slower processing speed and a greater number of cognitive complaints, while working memory was more impaired in those with a severe adult TBI, mild adolescent injury or multiple mild injuries. Elsewhere, processing speed has been found to be more impaired in participants with more than one TBI resulting in loss of consciousness than in participants who have had just one TBI 8.
Predictors of TBI from birth cohort studies include male gender 6,12, parental alcohol misuse 6, a punitive parenting style and the number of adverse life events experienced by the family 12. These will be controlled for in the proposed study.