Proposal summaries
B2051 - Identifying common genetic variants and putative genes associated with facial traits - 29/08/2013
Dysmorphic facial features play a prominent role in the diagnosis of a variety of classical neurodevelopmental disorders. For example, Down syndrome1, Prader-Willi syndrome2 and Foetal Alcohol Syndrome3 are all associated with characteristic facial features that are used as part of the diagnostic criteria. In recent years, the development of 3D geometric morphometrics (the statistical analysis of face shape) allowed researchers to identify more subtle facial dysmorphologies4. Hennessy et al5 found subtle differences in the 3D facial shape of schizophrenia patients compared to controls, such as a lengthened lower mid-facial height in schizophrenia patients. Hennessy et al6 also found that bipolar patients exhibited several facial dysmorphologies relative to controls and schizophrenia patients. The development of the face and the brain is closely connected7. Facial dysmorphologies therefore provide insight into the developmental origins of neurodevelopmental disorders, since the developmental biology of facial features is better understood than that of the brain5.
Craniofacial morphology is highly heritable8-11. Yet, fairly little is known about the genetic variants that influence normal craniofacial development in the general population. 3D geometric morphometric techniques can be used to identify these genetic variants, which could provide a valuable resource for future studies on craniofacial development and a variety of genetically determined neurodevelopmental disorders. For example, Paternoster et al12 found a significant association between a variant of the PAX3 gene and nasion position (the point in the skull where the nasal and frontal bones unite) in the Avon Longitudinal Study of Parents and their Children (ALSPAC) data set. Mutations in the PAX3 gene has previously been associated with Waardenburg syndrome13.
Evolutionary approaches to the study of facial traits provide three general facial traits that could be useful in the identification of genetic variants associated with craniofacial development: symmetry, averageness and sexual dimorphism. Symmetry indicates an individual's ability to resist environmental and genetic stresses during development14-16. It follows that fluctuating asymmetry (a deviation from bilateral symmetry in normally bilaterally symmetrical traits) is associated with a variety of genetic disorders and some indicators of poor health during development16-20. For example, Hammond et al19 found significantly higher facial asymmetry in young boys with autism spectrum disorders, compared to age/sex/ethnicity matched controls.
Facial averageness is defined as the proximity of a face to the sex-specific average face for that population21. In other words, how closely the face resembles the majority of same sex faces in the population. Symons22 hypothesised that average features could be functionally optimal as a result of the stabilising effect of natural selection. Theoretically, averageness might also denote genetic heterozygosity23. Few studies have investigated the relationship between facial averageness, genetic disorders and more general health. Nevertheless, genetic diseases are often associated with multiple facial dysmorphologies1,2,5,6, which do make these faces fairly distinctive compared to the population mean. Rhodes et al24 also found a weak positive association between facial averageness and general health scores.
Sexual dimorphism is defined as the phenotypic difference between males and females of the same species. In other words, how masculine or feminine a person is. Male and female faces diverge at puberty due to the action of sex hormones16,25. The age of onset, rate and extent to which faces become sexually dimorphic differ between individuals and are partly determined by genes26. Masculinity in male faces is generally assumed to serve as an indicator of immunocompetence21,27,28, although recent work has called the relationship into question28-30. Femininity in female faces does not seem to be associated with immunocompetence31,32, but is significantly associated with late follicular oestrogen levels33. Abnormal sexual dimorphism is also a characteristic of certain genetic diseases, such as Klinefelter syndrome34.
The primary aim of this study is to identify common genetic variants, and ultimately putative genes, that are associated with facial symmetry, averageness and sexual dimorphism using Genome-Wide Association (GWA) methodologies. We specifically chose the ALSPAC dataset because it is, to our knowledge, the largest dataset with both facial images and GWA data. To accomplish this aim, the 3D facial images will be manually delineated by defining 38 feature points in the software program MorphAnalyser (developed by Dr Tiddeman, Aberystwyth University, UK); this method is identical to the method used in Coetzee et al35. The delineated 3D images will then be used to calculate indexes for symmetry, averageness and sexual dimorphism in MorphAnalyser. The ALSPAC team have cleaned and imputed a GWA study dataset consisting of 8365 individuals with genotype calls for ~2.5 million common variants spread across the genome. We will use this resource to conduct a 2 stage (discovery and replication) genome-wide association study. Initially the discovery phase will include analysing ~ 5000 individuals that have both genotypic data and facial images. Power analysis (PowerGwas/QT version 1.0)36 indicate a sample size of 5000 is adequate to provide 80% statistical power to detect single nucleotide polymorphisms (SNPs) that explain as little as 0.8% of the variance in facial traits. The association between each of the ~2.5 million SNPs (exposure variables) and facial symmetry, averageness and sexual dimorphism (outcome variables) will be independently tested in the ALSPAC cohort using linear additive regression. SNP associations that exceed the standard significance threshold for genome-wide significance (p less than 5 x 10 -8)37, will be identified and replicated independently in additional cohorts, making up the second replication phase for the GWA study.
To determine which genes (and pathways) are most likely associated with these facial traits we will conduct a range of post-hoc analyses including (a) assigning SNPs to genes, (b) epistasis modelling and (c) pathway analyses. Briefly, multiple genes are ascribed to each SNP and these genes are then prioritised using epistasis modelling and pathway analysis38, allowing us to further identify which biological processes regulate these facial traits. In addition, we will calculate heritability estimates for each of the three facial traits; do a GCTA analysis to estimate how much of the phenotypic variance in the facial traits are explained by common SNPs; test the relationship between admixture and the facial traits (especially averageness); test the relationship between genome-wide heterozygosity and the facial traits (especially symmetry and averageness) using the ~2000 ALSPAC individuals who have whole genome sequencing data; and test the association between previously imputed classical HLA alleles and these facial traits separately for males and females. All GWA analyses will be conducted at the University of Bristol.
The second aim is to determine the association between health measures (exposure variables) and facial symmetry, averageness and sexual dimorphism (outcome variables). The health measures will be divided into prenatal risk factors (e.g. parental age, presence of gestational diabetes) and childhood health measures (e.g. body mass index, blood pressure and self-reported health). All three facial traits are generally assumed to indicate good health27, but studies testing these assumptions have suffered from various methodological drawbacks, including small sample sizes (~N=40-200). The size and quality of the ALSPAC data set, especially the wide range of physiological measurements, provides us with the ideal opportunity to test the association between health indices and these three facial traits in male and female faces respectively.
The third aim is to determine whether SNPs associated with these facial traits are also associated with traits proposed to indicate overall condition, specifically increased height and body mass index (BMI; within the healthy BMI range). To do so we will test the relationship between allelic scores of SNPs for the three facial traits, height and BMI.
To accomplish these aims we kindly request access to the following data in the ALSPAC dataset
Concept
Specific measure
Person
Source
Time point
Prenatal risk
Maternal age
Mother
Q
8-42 wks gestation
Prenatal risk
Paternal age
Mother
Q
12 wks gestation
Prenatal risk
Height & weight
Mother
C
1st trimester
Prenatal risk
Gestational diabetes
Mother
C
1st-3rd trimester
Prenatal risk
Blood pressure
Mother
C
1st-3rd trimester
Prenatal risk
Smoking
Mother
Q
18 wks gestation
Prenatal risk
Alcohol use
Mother
Q
18 wks gestation
Prenatal risk
General health
Mother
Q
32 wks gestation
Childhood health
Hospital admittance
Child
Q
4 wks-69 mths
Childhood health
General health
Child
Q
6-91 mths
Childhood health
Specific health problems (e.g. coughing)
Child
Q
4 wks-91 mths
Childhood health
Height & weight
Child
C
4 mths-17 yrs
Childhood health
Fat and lean mass from DXA scan
Child
C
9-17 yrs
Childhood health
Lung function
Child
C
61 mths-17 yrs
Childhood health
Grip strength
Child
C
11 yrs
Childhood health
Blood pressure & pulse rate
Child
C
37 mths-17 yrs
Childhood health
Blood pressure after exercise
Child
C
9-17 yrs
Confounding
Pubertal development
Child
Q
175 mths (or PUB7)
Images
Facial scans
Child
C
15 yrs
Genetic
Genome wide association data
Child
C
-
questionnaire=Q; clinic=C; weeks=wks; months=mths; years=yrs
References
1. Farkas LG, Katic MJ, Forrest CR, et al (2001) J Craniofac Surg, 12, 373-379.
2. Holm VA, Cassidy SB, Butler MG, et al (1993) Pediatrics, 91, 398-402.
3. Fang S, McLaughlin J, Fang J, et al (2008) Orthod Craniofac Res, 11, 162-171.
4. Hammond P. (2007) Arch Dis Child, 92, 1120-1126.
5. Hennessy RJ, Lane A, Kinsella A, et al (2004) Schizophr Res, 67, 261-268.
6. Hennessy RJ, Baldwin PA, Browne DJ, et al (2010) Schizophr Res, 122, 63-71.
7. Diewert V, Lozanoff S & Choy V. (1993) J Cran Genet Dev Bio, 13, 193.
8. Kohn L. (1991) Annu Rev Anthropol, 20, 261-278.
9. Hunter WS, Balbach DR & Lamphiear DE. (1970) Am J Orthod, 58, 128-134.
10. Nakata M, Yu P-I, Davis B, et al (1973) Am J Orthod, 63, 471-480.
11. Johannsdottir B, Thorarinsson F, Thordarson A, et al (2005) Am J Orthod Dentofac, 127, 200-207.
12. Paternoster L, Zhurov AI, Toma AM, et al (2012) Am J Hum Genet, 90, 478-485.
13. Tassabehji M, Read AP, Newton VE, et al (1993) Nat Genet, 3, 26-30.
14. Mather K. (1953) Heredity, 7, 297-336.
15. Van Valen L. (1962) Evolution, 16, 125-142.
16. Thornhill R & Moller AP. (1997) Biol Rev, 72, 497-548.
17. Sforza C, Dellavia C, Tartaglia GM, et al (2005) Int J Oral Max Surg, 34, 480-486.
18. Markow TA & Wandler K. (1986) Psychiat Res, 19, 323-328.
19. Hammond P, Forster-Gibson C, Chudley AE, et al (2008) Mol Psychiatr 13, 614-623.
20. Livshits G & Kobyliansky E. (1991) Hum Biol, 63, 441-446.
21. Rhodes G. (2006) Annu Rev Psychol, 57, 199-226.
22. Symons D. (1979) The evolution of human sexuality, Oxford University Press.
23. Thornhill R & Gangestad SW. (1993) Hum Nat, 4, 237-269.
24. Rhodes G, Zebrowitz LA, Clark A, et al (2001) Evol Hum Behav, 22, 31-46.
25. Farkas LG & Munro, IR (1988) Anthropometric Proportions in Medicine, Thomas, p29-56.
26. Mustanski BS, Viken RJ, Kaprio J, et al (2004) Dev Psychol, 40, 1188-1198.
27. Thornhill R & Gangestad SW. (1999) Trends Cogn Sci, 3, 452-460.
28. Scott IML, Clark AP, Boothroyd LG, et al (2012) Behav Ecol, 24, 579-589.
29. Rantala MJ, Coetzee V, Moore FR, et al (2013) P Roy Soc Lond B Bio, 280, 1471-2954.
30. Lie HC, Rhodes G & Simmons LW (2008) Evolution, 62, 2473-2486.
31. Rhodes G, Chan J, Zebrowitz LA, et al (2003) P Roy Soc Lond B Bio, 270, S93-S95.
32. Thornhill R & Gangestad SW. (2006) Evol Hum Behav, 27, 131-144.
33. Law Smith MJ, Perrett DI, Jones BC, et al (2006) P Roy Soc Lond B Bio, 273, 135-140.
34. Kamischke A, Baumgardt A, Horst J, et al (2003) J Androl, 24, 41-48.
35. Coetzee V, Re DE, Perrett DI, et al (2011) Body Image, 8, 190-193.
36. Feng S, Wang S, Chen C-C, et al (2011) BMC Genet, 12, 12.
37. Corvin A, Craddock N & Sullivan PF. (2010) Psychol Med, 40, 1063-1077.
38. Cantor RM, Lange K & Sinsheimer JS. (2010) Am J Hum Genet, 86, 6-22.
B2066 - Antenatal interpersonal sensitivity depression and subsequent mother-infant relationship quality - 23/08/2013
The aim of this study is to examine the precision of perinatal depression screening, including interpersonal sensitivity, for identifying mother-infant relationship and maternal mental health problems that continue into or emerge in the second year of life.
The study will answer the research question: how much of the variance in mother-infant relational quality is explained by interpersonal sensitivity (IPSM) and perinatal depression secreening (EPDS) scores measured during pregnancy?
B2068 - Investigating the effects of disclosure control - 15/08/2013
Aims: To develop an anonymisation policy for ALSPAC to be implemented for use of any release which may include linked NHS data. The policy needs to be submitted with the application of the Information Goverance Toolkit which needs to be in place to continue receiving and gaining further section 251 support. Further to this policy, we aim to explore anonymisation techniques determining the most secure level of anonymisation with the least effect upon the quality of the data for research. Using previous research on datasets which have been anoymised to different levels and using different techniques.
The anonymisation policy, for the IG Toolkit, will need to include the minimum level of anonymisation needed for release files of linked NHS data, k-anonymisation on quasi-identifiers.
Hyptheses: To determine the most secure method of anonymisation with the least impact on the quality of the data for research.
The hypotheses will be explored using a previous research question published by Kate Northstone. The research used to test the impact is as follows:
Maternal fish intake and child dietary patterns and associations with educational outcome at 16 years of age.
Outcomes: Child achieved 5 or more GCSE's at grade C or above, including maths and english.
Primary exposures:
1) Fish intake during pregnancy (none, 1-340g and greater than 340g)
2) Omega-3 intake (none, any actual intake split into tertiles)
3) Child Dietary patterns derived from PCA at 3,4,7,9 and 13 years of age
Confounders:
Gender
Maternal age
Maternal education
Housing
Maternal life events
Birthweight
Gestation
Maternal Smoking
Parity
Maternal and paternal occupation
IMD
Geography.
B2067 - Utilising multiple independent combinations of genetic variants to evaluate potential pleiotropy - 15/08/2013
We aim to develop a method to investigate potential pleiotropy in causal estimates derived using the Mendelian randomization approach. The ALSPAC data will be used as an illustrative example.
Exposure: height, Outcome: FVC lung function (unadjusted for height), instrumental variables: 20 genotypes listed below.
B2065 - Maternal obesity associated metabolic conditions and neurodevelopmental and psychiatric disorders in children - 15/08/2013
AIMS
Our goal is to contribute to an emerging body of literature regarding the association between maternal obesity in pregnancy and neurodevelopmental and psychiatric disorders in children. To achieve this goal, our primary objective is to test theoretical models of the relationship between pre-pregnancy body mass index and other clinical manifestations of maternal insulin resistance (hypertension, diabetes) and three outcomes in children: autism; attention deficit hyperactivity disorder (ADHD); and schizophrenia.
HYPOTHESIS
We hypothesize that a high pre-pregnancy body mass index will be associated with an increased risk of adverse neurodevelopmental and psychiatric outcomes among children.
VARIABLES
Exposures (maternal): Pre-pregnancy body mass index, weight gain during pregnancy, hypertensive disorders of pregnancy, diabetes
Outcomes (children): Autism, ADHD, symptoms of schizophrenia
Covariates: Various factors will be considered as potential confounders or mediators of the hypothesized associations. These include cigarette smoking, substance use, alcohol consumption, infection, and depression and anxiety during pregnancy; postpartum depression and anxiety; maternal and paternal history of mental health conditions; socioeconomic status; race/ethnicity; maternal age at delivery; parity; prior pregnancy loss; child's sex; multiple birth; preterm birth; Caesarean section (planned/emergency); birth weight greater than 4000g; birth weight greater than 4500g; and intellectual disability in the child.
B2064 - Genome-wide and candidate gene association studies of visual and cognitive phenotypes and their inter-relationships - 15/08/2013
Aims:
1. To determine whether visual and cognitive phenotypes cluster together within a normal population cohort of children.
2. To determine the underlying genetic associations of visual and cognitive phenotypes on a genome-wide basis in a normal population cohort.
3. To extrapolate datasets from animal models of associated visual and cognitive disorders, e.g. genes expressed in interneuron populations both in the retina and CNS, and investigate these candidate genes (within the datasets obtained as above) for associations with visual and cognitive phenotypes in a normal population cohort (this mouse data is already available from my lab).
Hypotheses:
Eye development in different organisms produces dramatically different structures, like the compound eye of insects and the camera-like eye of vertebrates. Nevertheless, the molecular mechanisms underlying eye specification are highly conserved (1), and the study of eye development in animal models has proven to be highly informative of the regulatory events that control human eye formation. For example, Pax6 was identified as a 'master' regulator, at the top of the hierarchical network of transcription factors (TFs) involved in eye development, since loss-of-function mutations of the eyeless gene (the Pax6 Drosophila homologue) lead to an eyeless phenotype (2) and over-expression can direct the formation of histologically normal ectopic eyes in flies (2) and in some vertebrates. Furthermore, recent work has established that the evolutionary conservation of the visual system extends beyond eye specification (by genes like Pax6) to include the visual system circuitry that connects the eye to the brain (1). Pax6 mutations in mice cause the small eye phenotype, and human PAX6 mutations lead to eye malformations including aniridia and other anomalies, while homozygotes demonstrate malformations of the central nervous system (CNS)(3). More recent work has shown that heterozygote PAX6 mutations are also associated with previously unrecognised and subtle structural brain abnormalities and cognitive deficits in humans (4). Furthermore, distant regulatory enhancer sequences influence transcription of Pax6 in mice, and consequently lead to eye abnormalities without any associated mutations of the Pax6 gene itself (5). While Pax6 is the most well-studied gene influencing eye development, several others are known to regulate interneuron development in the retina and CNS in an increasing number of animal models. Based on these data, we would predict that children carrying genetic polymorphisms and/or mutations in genomic regions encompassing so-called retinal and CNS "interneuron genes" would manifest deficits in vision and/or cognition.
B2063 - Assessing linkage error and bias between ALSPAC and HES - 15/08/2013
Longitudinal studies are making increased use of routine health and administrative data as a means of informing missing data techniques and sustainable data collection. These advantages are dependent on the accurate interpretation of the linkage. Links between an individual and their routine records are established by comparing personal identifiers common to both datasets. The potential to do this accurately is impacted by the choice and application of the linkage algorithms and the quality and discriminatory potential of the available identifiers. Recent work by Goldstein, Harron and Wade (2012) demonstrated new methods to enhance the efficiency of the linkage process using multiple imputation (MI) techniques. Once linked, the onus is on the study team to provide the provenance of the data; describing the linkage methodology and assessing the quality of the linkage at an individual level.
Through the Project to Enhance ALSPAC through Record Linkage (PEARL) we are linking the study index children to their secondary health care records, held within the Hospital Episodes Statistics (HES) dataset. The accuracy of this linkage is of concern as the personal identifiers held in early HES data (pre 1997) will in some cases lack the discriminatory power to identify a single individual. The NHS Data Linkage Service linked a pilot sample of 3,198 study participants to their 1991-2012 HES records. The linkage algorithm varied depending on the ability of the identifiers to establish a 'true match'.
AIM: To provide evidence of the quality of the linkage between ALSPAC and HES, particularly in terms of population coverage. Ultimately to use the evidence (if the hypothesis is true) to seek HES permissions to alter the linkage methodology, specifically to use the prior informed imputation techniques proposed by Goldstein. To use ALSPAC data obtained through different channels (data abstraction) as a 'gold standard' to 'validate' the imputed output through replicating a known ALPSAC finding (this will be subject to a new research proposal with standard data access conditions).
HYPOTHESIS: That the linkage variables used by HES to conduct the match are insufficient to identify all possible ALSPAC records.
EXPOSURE VARIABLES: The administrative linkage data supplied to the NHS, residential address history (specifically if they lived within England & Wales or not), self-reported hospital admissions (including cause and date and length of stay).
OUTCOME VARIABLES: Linkage status, delivery location and date, birth weight, gestational age.
CONFOUNDING VARIABLES: These relate to the accuracy of the linkage variables: enrolment details (i.e. for new cases we won't have postcode at delivery) and indicators of address quality (participation status at delivery, home movement,house tenure, known birth outcome).
B2062 - Mining for complex patterns in epidemiological data - 15/08/2013
The aim of this project is to use advanced machine learning and data mining techniques to extract features and patterns from the ALSPAC data in an unsupervised manner. Techniques to be used include subgroup discovery and exceptional model mining, which find subpopulations that are statistically robustly different wrt. properties of interest compared to the overall population. Epidemiological properties of interest include very dense genetic data (genome wide single nucleotide polymorphism, copy number variation and sequence data), gene expression and gene methylation data and outcomes related to body composition, obesity, physical activity, health related behaviours such as diet and smoking and cognitive function.
B2061 - Effects of maternal alcohol consumption during pregnancy on childhood behaviour in the ALSPAC cohort - 15/08/2013
Development of behavioural characteristics is dependent on a complex interaction of genetic and environmental factors. It is also known that neonatal development can be altered by a variety of maternally derived exposures.
Alcohol consumption in pregnancy is an important health issue and the DOH currently advise that 'Pregnant women or women trying to conceive should avoid drinking alcohol' (http://www.dh.gov.uk/en/Publichealth/Healthimprovement/Alcoholmisuse/DH_...). This is due to significant evidence that maternal alcohol consumption has deleterious effects on the development of the unborn infant.
Fetal alcohol syndrome (FAS) is thought to be a leading cause of learning disability in the western world. Characteristics of FAS include; lower IQ, abnormal facial features and behavioural and mental health problems. Whilst FAS is caused by very heavy alcohol consumption for prolonged periods during pregnancy, it is thought that exposure to lower doses of alcohol in utero may cause less severe sub-optimal development, which is largely undetected. However, detecting these effects is problematic since alcohol consumption in the mother is heavily confounded by other lifestyle factors.
Proposed study
ALSPAC has collected information on maternal alcohol consumption through questionnaires at 12, 18 and 32 weeks of gestation as well as data regarding the drinking consumption of the mothers' parents. In addition quantity of maternal alcohol intake data is available related to the timing of fetal development; before pregnancy, first 3 months of pregnancy, at around the time first felt the baby move.
We also have genetic data on these mothers and children, and have shown that a common genetic variant in the ADH1B gene in mothers is associated with alcohol consumption levels during pregnancy. In addition we have shown that 4 genetic variants in ALSPAC children are related to their IQ at age 8, but only among children born to mothers who drank some alcohol during pregnancy. We plan to use these genotypes in an analysis of alcohol and children's behaviour to determine whether maternal alcohol consumption during pregnancy is likely to be causally related to problem behaviour in her offspring.
Children's behaviour problems (conduct problems and hyperactivity) have been assessed repeatedly from ages 4 to 16 years in ALSPAC, using the Strengths and Difficulties Questionnaire (SDQ) and the Development and Wellbeing Assessment (DAWBA). Based on these measures, three types of outcome are available for analyses: symptom scores (on the SDQ), diagnoses (using DAWBA), and longitudinal trajectories of conduct problems between ages 4-13 years (previously created by Barker and Maughan, 2009, using SDQ scores).
We will look at main effects of genotype on children's behaviour, and carry-out a stratified analysis to determine whether effects are due to exposure to alcohol rather than pleiotrophy.
References
Barker ED, Maughan B. Differentiating Early-Onset Persistent Versus Childhood-Limited Conduct Problem Youth. Am J Psychiatry. 2009;166(8):900-8.
Lewis SJ, Zuccolo L, Davey Smith G, Macleod J, Rodriguez S, Draper ES, Barrow M, Alati R, Sayal K, Ring S, Golding J, Gray R.Fetal Alcohol Exposure and IQ at Age 8: Evidence from a Population-Based Birth-Cohort Study.PLoS One. 2012;7(11):e49407. doi: 10.1371/journal.pone.0049407. Epub 2012 Nov 14.
Zuccolo L, Fitz-Simon N, Gray R, Ring SM, Sayal K, Davey Smith G, Lewis SJ. A non-synonymous variant in ADH1B is strongly associated with prenatal alcohol use in a European sample of pregnant women. Hum Mol Genet.2009;18:4457-66.
B2059 - Dual trajectories of adolescent smoking and depression - 15/08/2013
Project outline:
Smoking and depression commonly co-occur in the general population. That is, individuals with depression are more likely to be smokers and smoke more heavily compared to individuals without depression, and smokers are more likely to report depressive symptoms compared to nonsmokers. In adolescents, there is also evidence of this relationship. It has been shown that depression and other mood disorders (e.g., anxiety) in adolescents increase the likelihood of experimental smoking and smoking initiation. Others have demonstrated that tobacco smoking among adolescents leads to increase depressive symptoms, which remit following cessation. Regardless of directionality, it is well-known that smokers with elevated depressive symptoms experience more difficulties when quitting smoking. Furthermore, concurrent depressive symptoms and tobacco use may interact synergistically to produce greater health risks than either disorder alone, especially for heart disease.
Better characterizing the developmental (longitudinal) relationship between these variables, particularly among adolescents, would be a significant contribution to the existing literature and provide evidence for the diversity in comorbidity of adolescent smoking and depression. This is an important area as both smoking and depression are related to increased use of other drugs and illicit substances, mental health distress, and physical health problems. Recently, much research has focused on modelling patterns in substance abuse and mood disorders across early developmental periods, but smoking and depression among adolescents have yet to be modelled in a dual trajectory analysis. Smoking trajectories for this cohort have been previously established and the data lend themselves to modelling trajectories of the development and early course of depression as well. The ALSPAC cohort presents an ideal dataset to be analysed as dual trajectories to model the comorbid development of smoking and depression in adolescents. Further, we can examine sociodemographic predictors and mental, behavioural, and physical outcomes of group membership.
Aims:
1. Examine the co-occurrence of smoking and depression among adolescents using dual trajectory modelling.
2. Explore outcomes between different trajectory groups, including:
a. mood symptoms (e.g., anxiety),
b. substance use (e.g., cannabis), and
c. biological markers (e.g., lung function).
We will use longitudinal latent class analysis to generate trajectories for adolescent smoking and depression, both separately and then in conjunction. Adolescent smoking will be based on self-reported smoking status. Adolescent depression will be based on the self-report MFQ.
Hypotheses:
We hypothesize that severity of tobacco use will be related to severity of depression (to be examined by cross tabulation). More specifically, we expect that adolescents who smoke will endorse higher levels of depressive symptoms, and that logical dual trajectories of adolescent smoking and depression will emerge. We also expect that there will be significant group differences, both in predictor variables (sociodemographics) and outcome variables (substance use, mood symptoms, and biological markers).
Variables:
Exposure variables. Smoking status will be determined by self-report during the clinical interview and depression will be determined by self-reported on the MFQ. These variables will then be modelled by dual trajectories which will elicit specific smoking x depression groups.
Outcome variables. Other substance use (e.g., cannabis), mood symptoms (e.g., anxiety), and biological markers (e.g., lung function). These outcome variables will be applied to both the groups that emerge from the individual trajectory models (smoking and depression) as well as those from the smoking x depression dual trajectory model.
Confounding variables. Smoking status and depression have also been shown to be differentially associated with gender, race, age, education and SES (housing tenure, crowding status and maternal educational attainment). Therefore these covariates will be included in our models. Another variable that has garnered attention recently is traumatic events in childhood, so we will explore the effect of early life stressors (e.g., abuse, deaths) on both smoking and depression.
B2056 - Assessing the impact of partner smoking on cotinine levels in the ALSPAC mothers - 01/08/2013
Aim
To investigate the impact of partner smoking on maternal cotinine levels during pregnancy.
Hypotheses
Comparison of the magnitude of association of maternal and paternal smoking during pregnancy is a useful method for assessing whether smoking during pregnancy may have an intrauterine effect on offspring outcomes. (1) This has been used in ALSPAC to investigate the impact of smoking during pregnancy on offspring birthweight, blood pressure, trajectories of height and adiposity and attention deficit hyperactivity disorder. (2-4) This method assumes that the effect of passive smoking in utero on offspring of mothers who do not smoke but have partners who smoke is minimal. However, recent work in ALSPAC has demonstrated that maternal smoking is strongly associated with cotinine (a metabolite of nicotine) levels in non-smoking offspring during childhood and adolescence. (5) Therefore, it is possible that exposure to household smoking may be an important determinant of health outcomes. If this is the case, it may be necessary to control for this in comparisons of the effect of maternal and paternal smoking on offspring outcomes.
We aim to investigate the extent to which partner smoking affects maternal cotinine levels during pregnancy. Associations between partner smoking and maternal cotinine will be investigated using linear regression, stratified by self-reported smoking status of the mother. In addition, we will look at associations of partner smoking with cotinine in all mothers, adjusted for maternal smoking status and heaviness. It may be necessary to use cotinine cut offs to validate maternal self-reported smoking status.
Exposure variables
Paternal smoking status and heaviness
Stratify by maternal smoking status
Outcome variables
Maternal cotinine measured during pregnancy
Confounding variables
Age
Maternal BMI
1. Smith GD. Assessing intrauterine influences on offspring health outcomes: can epidemiological studies yield robust findings? Basic & clinical pharmacology & toxicology. 2008;102(2):245-56. Epub 2008/01/30.
2. Howe LD, Matijasevich A, Tilling K, Brion MJ, Leary SD, Smith GD, et al. Maternal smoking during pregnancy and offspring trajectories of height and adiposity: comparing maternal and paternal associations. International journal of epidemiology. 2012;41(3):722-32. Epub 2012/03/13.
3. Langley K, Heron J, Smith GD, Thapar A. Maternal and Paternal Smoking During Pregnancy and Risk of ADHD Symptoms in Offspring: Testing for Intrauterine Effects. American Journal of Epidemiology. 2012;176(3):261-8.
4. Brion MJ, Leary SD, Smith GD, Ness AR. Similar associations of parental prenatal smoking suggest child blood pressure is not influenced by intrauterine effects. Hypertension. 2007;49(6):1422-8. Epub 2007/04/04.
5. Stiby AL, Macleod J, Hickman M, Yip V, Timpson N, Munafo M. Association of Maternal Smoking With Child Cotinine Levels. Nicotine & tobacco research : official journal of the Society for Research on Nicotine and Tobacco. 2013.
B2055 - Injection drug use by the age of 21 socio-economic patterning parental substance use and early childhood adversity - 01/08/2013
Aims: The aim of this project is to investigate early risk factors of IDU in ALSPAC. Previous research has suggested that the onset of IDU is associated with family structure (not living with both parents, or in care or a foster home at any point) as well as early conduct problems, particularly school exclusion and childhood contact with the criminal justice system. Moreover, violence, criminality and financial problems in the family have been shown to be associated with increased risk as well as any types of carer substance use (Macleod et al 2012). However, these association have to date only been shown within the context of case control studies (Macleod et al 2012; Conroy et al., 2009; Tomas et al., 1990; Obot et al, 1999). To our knowledge there is no prospective evidence investigating the association between these early life risk factors and IDU in early adulthood.
Outcome variables: self-reported injecting drug use at ages 17 and 21
We will investigate IDU by age 17 and by age 21 and also create an outcome variable indicating IDU by age 21 (combining TF4 data with data from the 21yr questionnaire)
We propose to examine the following co-variates in relation to the outcome:
Indicators of socio-economic positioning
Indicators of early childhood adversity
- Contact with social services (mother self-report)
- "Child at risk" Register (linkage)
- Experience of victimisation (physical, emotional and sexual abuse during childhood)
- Antisocial behaviour
- Conduct disorder trajectories and early measures of conduct problems
- Alcohol, Smoking and substance use up to young adulthood
- Stessful life events in the child
- Post traumatic stress in the child
Educational measures:
- KS1 & KS2
- School exclusion and data on absenteeism (teacher reported, YP self-reported and linkage data)
Parental factors
- Parental substance use including smoking, alcohol problems, drug taking and injection drug use
- Indicators of family involvement with the police/court, criminal convictions
- Cruelty within the family (towards the carer, towards child), financial hardship
References:
Macleod J, Hickman M, Jones J, Copeland L, McKenzie J, De Angelis D, Kimber J & Robertson JR (2012): Early life influences on the risk of injecting drug use: case control study based on the Edinburgh Addiction Cohort. Addiction. Vol.108(4). pp. 743 - 750
Conroy E, Degenhardt L, Mattick RP & Nelson E (2009): Child maltreatment as a risk factor for opiod dependence:comparison of family characteristics and type and severity of child maltreatment with a matched control group. Child Abuse & Neglect. Vol.33. pp. 343 - 352.
Tomas JM, Vlahov D, Anthony JC (1990): Association between intravenous drug use and early misbehaviour. Drug & Alcohol Dependency. Vol.25. pp. 79 - 89.
Obot IS & Anthony JC (1999): Association of school dropout with recent and past injecting drug use among African American adults. Addicitive Behaviour. Vol. 24. pp. 701 - 705.
B2054 - Antisocial and callous behaviour the role of fish intake and their associated fatty acids - 01/08/2013
Background: Antisocial behaviour in childhood and adolescence has been associated with more serious offending in adulthood [1]. Recent evidence suggests that the consideration of calluous and unemotional traits may further define the psychopathology [2]. It is clear that individuals can have different trajectories with early onset persistent problems identifying a particularly high risk group [3]. Other research has also shown a number of antecedents reflecting child characteristics (such as uncontrolled temperament and low self-esteem), parenting skills and the home environment including inter-parental conflict [4, 5]. Studies have also shown associations between exposures during pregnancy and behaviour many years later [5]. But to date research on dietary influences and in particular the role of omega-3 fatty acids has been limited. What evidence exists warrants further investigation of these issues [6].
Aims: To examine and further refine the phenotype based upon potential comorbid characteristics such as particular facets of behaviour, motor skills, language impairments and IQ. To explore the associations of varying risk groups with fish intake and fatty acids.
Hypotheses: (a) That high intakes of omega-3 as reflected in high fish intakes and high blood levels will lower risk for antisocial and callous behaviour. Such effects may be more strongly associated with DHA.
(b) That the ratio of omega-6 to omega-3 fatty acids reflecting an inflammatory response index may independently increase risk of outcomes.
(c) That the fetal environment as reflected by maternal fatty acids and fish intake may 'program' the child to be more susceptible or more resilient to potential risk factors.
Outcomes: Antisocial and callous behaviour taking account of other comorbid conditions.
Exposures: Fish intake and fatty acids obtained from blood samples.
Confounders: Socio-economic, maternal psychopathology, parenting and the home environment.
Analyses are likely to be untaken in parallel both in the USA and Bristol.
References
1. Benda BB, Corwyn RF, Toombs NJ. Recidivism among adolescent serious offenders: Prediction of entry into correctional system for adults. Criminal Justice and Behavior 28:5 (2001), pp 588-613
2. Rowe R, Maughan B, Moran P, Ford T, Briskman J, Goodman R. The role of callous and unemotional traits in the diagnosis of conduct disorder. Journal of Child Psychology and Psychiatry 51:6 (2010), pp 688-695
3. Barker ED, Oliver BR, Maughan B. Co-occurring problems of early onset persistent, childhood limited, and adolescent onset conduct problem youth. Journal of Child Psychology and Psychiatry 51:11 (2010), pp 1217-1226
4. Bowen E, Heron J, and Steer C. Anti-Social and Other Problem Behaviours Among Young Children: Findings From the Avon Longitudinal Study of Parents and Children. London: Home Office; 2008. Report 02/08.
5. Barker ED, Maughan B. Differentiating Early-Onset Persistent Versus Childhood-Limited Conduct Problem Youth. American Journal of Psychiatry 166:8 (2009), pp 900-908
6. Kohlboeck G, Glaser C, Tiesler C, Demmelmair H, Standl M, Romanos M, Koletzko B, Lehmann I, Heinrich J, for the LISAplus Study Group. Effect of fatty acid status in cord blood serum on children's behavioral difficulties at 10 y of age: results from the LISAplus Study. American Journal of Clinical Nutrition 94:6 (2011), pp1592-1599.
B2050 - Meta-analysis of maternal smoking during pregnancy and methylation in offspring - 18/07/2013
AIMS
To investigate the relationship between maternal smoking habits during pregnancy (i.e. smoking status, sustained smoking/smoking duration, smoking quantity) and DNA methylation levels in cord blood samples from newborn offspring utilising the ALSPAC-ARIES HM450 dataset. This analysis will form part of a meta-analysis across multiple study cohorts.
Overall research question: Is maternal smoking during pregnancy related to CpG site-specific methylation in newborns?
ANALYSIS PLAN
Exposure variable: Four questions have been drawn up to address issues relating to smoking habits throughout pregnancy, timing, dosage and paternal effects. It is acknowledged that not all individual study cohorts will have the relevant data to address all these questions. Hence, each study cohort should address those applicable.
1. Active smoking
a. Sustained active smoking versus no smoking (dichotomous): mothers who smoked during most of the pregnancy/into late pregnancy (2nd/3rd trimester) versus those who did not smoke at all during pregnancy (including those who quit prior to pregnancy).
b. Early pregnancy smoking versus no smoking (dichotomous): mothers who smoked during early pregnancy and quit later versus those who did not smoke at all during pregnancy (including those who quit prior to pregnancy).
c. Ever smoked versus no smoke (dichotomous): mothers who reported smoking at anytime during pregnancy versus those who did not smoke at all during pregnancy (including those who quit prior to pregnancy).
2. Passive smoking
Definition: any indication that mothers were exposed to passive smoking (i.e. partner smoked, other relatives/household members smoked, exposed at home, exposed at work, quantified e.g. greater than 1 hour per day). Perform analysis in non-smokers only, split into passive and non-passive smoking as appropriate (Dichotomous). If possible, perform analyses in the three sets as above.
3. Smoking quantity
Definition: if possible split mothers by 1-9 cigarettes per day, 10+ cigarettes per day, non-smokers (trichotomous). If possible, perform analyses in the three sets as above.
4. Smoking in biological father
If smoking status of biological father is known perform analyses on paternal smoking prior to pregnancy (yes/no, dichotomous).
Outcome variable: DNA methylation utilising the HM450 ARIES data on cord blood samples. If possible, use the raw beta values for all probes i.e. with no normalisations or transformations. Alternatively (or in addition) perform preferred QC and pre-processing analyses as necessary.
Statistical Analysis: Perform robust linear regression with individual CpG site methylation levels as the outcome variable and smoking status as the exposure of interest. Include any potential confounders on a cohort-specific manor. Summary statistics will be provided to Dr Jourbert at NIEHS enabling mixed/random effects modelling in downstream meta-analyses.
Confounders for ALSPAC-ARIES: Definition - any factor associated with the exposure variable (i.e. smoking variable) and plausibly associated with DNA methylation. Assess the potential confounding effects of the following variables and include within the statistical model as necessary: sex, genetic ancestry/ethnical background, social-economical background, maternal age, pre-pregnancy BMI, parity.
Possible Sensitivity analyses: Perform the primary model (sustained vs non-smoke) adjusting for cell composition if possible and adjusting/stratified for preterm birth.
Other stipulations: restrict analyses to singleton births. Do not adjust for gestational age or birth weight in the first instance as these may be on the causal pathway linking smoking, methylation and health outcomes.
B2049 - Socioeconomic distribution of excess weight in children - 18/07/2013
Aim:
We propose to go beyond prevalence-based methods to test the relationship between obesity and parental income amongst children and investigate the differences by race/ethnicity and gender.
Hypotheses:
1. a)Using Unconditional Quantile Regression (Joliffee 2011) with BMI Z-scores the independent variable, low socio-economic status will be associated with greater weight at the overweight and obese BMI Z-score thresholds.
b)When accounting for parental weight status (overweight or obese), the coefficient for the effect of income on child BMI Z-scores at the overweight and obese thresholds will be reduced
c) Comparing Ordinary Least Squares estimates for the effect of parental weight status and income on child BMI Z-scores will underestimate the effect at the overweight and obese thresholds.
2. a) The net concentration index for measures of child obesity will be negative
b) A significant proportion of the concentration index for measures of child obesity will be explained by parental and weight status. Including parental obesity status will reduce the independent effect of income on obesity status for children.
The results of this analysis has policy implications. If the coefficient is significant after controlling for parental obesity status, this may indicate that healthy weight initiatives should target poor children. If, however, the coefficient becomes insignificant, policies should target all children and particularly children with overweight or obese parents.
Exposure variables:
Income, parental weight status
Outcome variable:
BMI z-score (calculated using weight, height, gender and age)
Confounding variables:
Socioeconomic measures such as ethnicity, lifestyle behaviours including diet, TV viewing, physical activity, tobacco use, and sleep duration.
Reference List:
Jolliffe, D. (2011). Overweight and poor? On the relationship between income and the body mass index. Economics & Human Biology, 9, 342-355.
B2048 - CLOSER work package one Data harmonisation of measures of biological function and structure across the cohorts - 18/07/2013
Aims:
To facilitate cross cohort work by harmonising measures of biological structure and function across the UK birth cohort studies, starting with weight and height in the 1946, 1958, 1970, 2000 birth cohort studies, plus ALSPAC. To sue the harmonised measures in two demonstration papers that:
1) compare body size distributions and mean trajectories, across different phases of the life course, between cohorts, and
2) investigate how SEP inequalities (using measures harmonised in CLOSER work package 2) in body size trajectories, across different phases of the life course, differ between cohorts. To extend the harmonised dataset to include data from other birth cohort studies (e.g. HCS, SWS, Biobank) and other measures of biological structure and function (e.g. blood pressure and grip strength), if necessary.
B2047 - Modifiable risk factors for depression in adolescents - understanding the role of physical activity and obesity - 18/07/2013
We aim to investigate the association between objective measures of obesity and physical activity and depression in adolescents.
1. To investigate the association between obesity, measured objectively, and depression in adolescents, including the possibility of a bi-directional association
2. To investigate the association between physical activity, measured objectively, and depression in adolescents, including the possibility of a bi-directional association
3. To investigate the inter-relationship of these modifiable risk factors.
B2046 - Social and genetic trajectories from motor development to academic attainment via energy balance-related behaviour - 18/07/2013
The aim of this project is to examine the effects of infant motor development on adolescent energy balance-related behaviour (EBRB) (physical activity and sedentary behaviour), and to identify learning and development, behavioural and health-related factors that mediate these effects (study I). In addition, we aim to examine the effects of self-reported and objectively measured EBRB on adolescent self-reported and teacher-rated academic achievement, and to identify factors/processes that mediate these effects (study II).
We hypothesise that early infant motor development predicts adolescent EBRB via developmental, behavioural and health-related factors like learning, behaviour/psychopathology, language development, personality, motor abilities and behaviour, anthropometry and various physical and mental health measures. We also hypothesise that EBRB predicts academic attainment via social, psychological and health-related factors like cognitive function, self-esteem, academic motivation and goal setting, interpersonal and social skills, psychopathology and school enjoyment, sleep, stress, obesity and physical fitness.Because of partly explorative nature of the study, mediating variables will be specified during structural equation model (SEM) building.
In study I, infant motor development at the age of 6 and 18 months is the main independent factor. Dependent variables include objectively and subjectively measured physical activity and sedentary behaviour/time. Possible mediating factors include parent-reported, teacher assessed and/or clinically measured learning (e.g. learning abilities and performance), behaviour/psychopathology (e.g. emotional and behavioural problems), language development, personality (e.g. self-concept, sensation seeking and locus of control), motor abilities and behaviour (e.g. gross and fine motor skills), anthropometry (height, weight and fat mass) and physical and mental/psycho-social health measures.
In study II, independent variables include objectively and subjectively measured physical activity and sedentary behaviour/time from age 7 to 16 years. Dependent variables are academic achievement at age 16 years, including spelling, understanding mathematics and science and school experiences and aspirations. Possible mediating factors include teacher assessed and clinically measured learning (e.g. learning abilities and performance), cognitive function, self-concept and academic self-confidence, interpersonal and social skills and relationships, behaviour/psychopathology (e.g. emotional and behavioural problems), school enjoyment, anthropometry (height, weight and fat mass), physical fitness, as well as physical and mental health measures.
B2044 - Association between mitrochondrial copy number and chronic stress - 18/07/2013
Aims . The aim of this proposal is to determine whether there is an association between mitochondrial (MT) copy number and measures of chronic stress (particularly early childhood chronic adversity). This analysis will only be possible using the ALSPAC sample included in the UK 10K project as I need sequence data to answer this question.
Hypotheses
Based on preliminary data from a case control study of major depression I think that early childhood chronic adversity elevates MT copy number, through an unknown mechanism. From DNA sequence data I have observed that in both cases and controls there is correlation between the following measures and MT copy number: neuroticism, number of stressful life events, childhood sexual abuse.
Variables
A meausure of susceptibility to stress (neuroticism), number of stressful life events (we used the 16 item stressful life events questionnaires), and measures of childhood abuse (we interrogate specifically non-genital, genetal and intercourse). Meaures of MT copy number are obtained from the numbers of reads mapping to the reference MT genome, normalized by the reads mapping to the autosomes (I use read data only from one chromosome to get this measure).
Outcomes
A linear model including age and weight, is used to predict MT copy number.
Confounds
As far as we know, age and weight are the only relevant confounds.
B2045 - UK10K Secondary proposal age at menarche - 04/07/2013
As part of an approved UK10K secondary proposal, we aim to perform an age of menarche whole-genome meta-analysis between ALSPAC and Twins UK. A UK10K secondary proposal covers any phenotype not included as a 'core' trait intended to be published as part of UK10K. We aim to analyse menarche and publish as a separate satelite paper at a similar time to the main UK10K effort.
There are two genetic datasets used in this project (that are already available within UK10K):
The ALSPAC UK10K sequence genotypes
Imputed UK10K genotypes from the wider ALSPAC GWAS sample
(We will perform two models, using the same statistical parameters used by the ReproGen consortium.
Model A: Menarche ~ SNP + Birth year
Model B: Menarche ~ SNP + Birth year + BMI (closest to age at menarche)
Model C: Menarche ~ SNP + Birth year + BMI (furthest time point from menarche available)
Similar to the core analyses in UK10K, we aim to perform a 4-way meta-analysis of the following strata:
Discovery sequence genotypes in ALSPAC
Discovery sequence genotypes in Twins UK
Imputed UK10K sequence genotypes from wider GWAS in ALSPAC
Imputed UK10K sequence genotypes from wider GWAS in Twins UK
We plan to share both individual level data (where possible) and summary level SNP results for meta-analysis. Only analysts registered on a UK10K data access agreement will be using the data - currently covers John Perry (KCL side) and Carolina Bonilla (ALSPAC side) for this project.