The neurobiological underpinnings of human
personality: the intersection of genetics and DTI in
population-based school-aged children
Thirza Dado 10492682 E: tm.dado93@gmail.com
BSc Biomedical sciences, Neurobiology
Faculty of natural sciences, mathematics and computer science University of Amsterdam
30 June 2016
Internship organisation:
Erasmus MC: University Medical Centre Rotterdam Complex Trait Genetics Lab, Vrije Universiteit Amsterdam Internship supervisor:
Philip Jansen, MD, MSc E: p.r.jansen@erasmusmc.nl University supervisor:
Dr. J.A. van Hooft Second Reader: Dr. R.J. Dekker
2 Abstract
Little is known about how identified DNA variations for personality cause differences in behaviour through variation in the brain. Neuroimaging data can be considered as the missing link between genetics and behaviour. In this exploratory study, polygenic architectures of complex personality traits are investigated by genomic profile risk scoring (GPRS), which uses results from genome-wide association studies (GWAS) (discovery sample) to assign polygenic risk scores to school-aged children from the ongoing population-based Generation R cohort study (independent target sample). Next, imaging genetics is used to examine the association between these risk scores and measured white matter development, obtained by diffusion tensor imaging (DTI). Significant results are found between higher agreeableness and lower fractional anisotropy (FA) in the inferior frontooccipital fasciculus (β=-0.0725; P=0.0255*) and parahippocampal cingulum (β=-0.0668; P=0.0365*), between higher neuroticism and lower FA in the forceps minor (β=-0.0980; P=0.0028**), and between higher openness and lower FA in the superior longitudinal fasciculus (β=-0.0751; P=0.0202*). In conclusion, findings of this proof-of-concept study for the evaluation of genetic architectures of complex traits have identified genetic overlap, and, therefore, this approach of imaging genetics has proven to be a useful tool in investigating neurobiological factors associated with personality traits.
3 Contents
1. Introduction ……….………... 4
2. Theoretical background……….………... 6
2.1 Missing heritability in GWAS………. 6
2.2 Polygenic risk score method……….……….. 7
2.3 Imaging genetics ………..……… 8
3. Materials and method………..………... 9
3.1 Discovery sample………..………... 9
3.2 Target sample………... 9
3.3 GPRS………... 10
3.4 Imaging genetics and DTI.………..…… 10
3.5 Possible confounding factors……….……….……. 12
4. Results ……….………... 12
4.1 Personality and white matter integrity at nine-years-old………. 12
4.2 Personality and white matter integrity between five- and nine-years-old………..……… 14
4.3 Personality and white matter integrity, and CBCL………. 15
5. Discussion………... 15 5.1 Neuroticism………..……… 16 5.2 Global FA values………. 16 6. Acknowledgements……….. 17 7. Literature………... 18 8. Supplementary materials………. 24
8.1 Effect sizes of different thresholds for SNP inclusion during GPRS……… 24
8.2 Imaging genetics for white matter tracts at nine-years-old………... 24
8.3 Longitudinal effect sizes of different thresholds for SNP inclusion during GPRS.……… 26
8.4 Polygenic risk scores as predictor of CBCL scores at nine-years-old………. 27
4 1. Introduction
Twin studies have made enormous contributions to clarifying how innate, biological factors interact with events experienced in particular environments. In this way, it is investigated what proportion of complex traits can be explained by genetics. There have been impressive examples of identical twins who are separated shortly after birth, and brought back together again after decades, to find almost completely matching personality profiles and remarkable behavioural similarities (Dahlstrom, Welsh, & Dahlstrom, 1975; Segal, 2012; Smith, et al., 2014). Such similarities reflect genetic components of more basic personality traits, which can in turn lead to preferences for particular kinds of behaviours (Smith, et al., 2014). Therefore, personality can be thought of as a set of characteristics or traits that influences people’s thoughts, feelings, and behaviours across a variety of settings (De Moor, et al., 2012). Scientific consensus has converged on the Five Factor Model (FFM) as one of the most widely accepted taxonomies of personality traits, which is based on five higher-order dimensions of openness to experience (O), conscientiousness or will to achieve (C), extraversion (E), agreeableness (A), and neuroticism or emotional stability (N) (table 1) (Costa & McCrae, 1992a; Costa & McCrae, 1992b). Results from peer rating scales, self-reports on trait descriptive adjectives, questionnaires measures of needs and motives, in expert ratings on the California Q-set, and in personality disorder symptom clusters suggest that these five core personality traits would organize the myriad of personality traits that have been discussed by other researchers (McCrae & Costa Jr, 1997; McCrae & Costa Jr, 1999). The Revised NEO Personality Inventory (NEO-PI-R) was developed to assess these personality traits, of which each higher-order trait includes six facets. The observable phenotypic traits of personality have been well described, but underlying causality remains unclear to very high extent. Several studies to twins and families have proven these five traits as heritable (Bouchard Jr & Loehlin, 2001; Distel, et al., 2009; Jang, et al., 1996; Pilia, et al., 2006; Yamagata, et al., 2006), which implies that inherited genetic variants play an important role in explaining personality. Estimates of heritability ranged between 33 and 65%.
Table 1 Five trait factors. Each trait factor captures a wide range of behaviours (Smith, Nolen-Hoeksema, Fredrickson, & Loftus, 2014).
Little is known about how identified SNPs cause differences in behaviour through variation in the brain. The theory of the genome-brain-behaviour axis states that genetics influence brain development, which then, in turn, influences behaviour (figure 1). Earlier studies have found associations between DNA variations and observed personality traits (De Moor, et al., 2012; De Moor, et al., 2015; Van Den
Higher-order dimension Lower-order dimension
Openness
(focus on the new)
Conventional-Original Unadventurous-Daring Conservative-Liberal Conscientiousness (focus on results) Careless-Careful Undependable-Reliable Negligent-Conscientious Extraversion
(focus on the outside world)
Retiring-Sociable Quiet-Talkative Inhibited-Spontaneous
Agreeableness
(focus on the other person)
Irritable-Good natured Ruthless-Soft hearted Selfish-Selfless Neuroticism (emotional stability) Calm-Worrying Hardy-Vulnerable Secure-Insecure
5 Berg, et al., 2014; Van Den Berg, et al., 2015). According to the genome-brain-behaviour axis, neuroimaging data can be considered as the missing link between genetics and behaviour. These endophenotypes are related to genes on closer level than complex traits, such as personality. Therefore, this survey addresses associations between earlier found DNA variations for personality and neuroimaging data of white matter integrity. Findings of previous imaging studies support the hypothesis that the association between white matter integrity and personality exist (Bjørnebekk, et al., 2013; McIntosh, et al., 2013; Xu & Potenza, 2012). Based on this knowledge, it is expected that individuals with high genetic susceptibility to deviant personalities have deviant developed white matter tracts. In other words, there exists a relationship between behaviour and white matter integrity: when behaviour deviates from the norm, white matter will similarly differ from the norm, and its occurrence can be led back to specific genetic architectures. Importantly, this research implies strong causality: effects of genetic architectures on the brain. The usual reverse causality does not play a role.
Knowledge of neurobiological underpinnings of personality traits can offer benefit in understanding the neurobiological factors related to behavioural tendencies and subjective well-being (Xu & Potenza, 2012). Gaining such understanding is of profound public health significance as well, since these traits, and neuroticism in particular, are thought to be able to negatively affect public health (Chapman, Roberts, & Duberstein, 2011; Lahey, 2009). Importantly, identified genetic variants associated with personality overlap with genetic influences on complex psychiatric diseases (Distel et al., 2009; Hettema et al., 2006; Kendler et al., 1993). Therefore, insights of relationships between genetic variants and personality are important for unmasking and further comprehension of genetic aetiology of mental disorders. Eventually, such aetiological knowledge could contribute to reliability of diagnosis, which is currently only based on clinical observations (American Psychiatric Association, 2000).
Figure 1 Genome-brain-behaviour axis. According to this theory, genetic variations on the genome lead to specific brain development, which in turn lead to the occurrence of specific behaviour. Earlier studies have investigated which variations corresponded with particular traits. Neuroimaging data can be considered as the missing link between genetics and behaviour.
6 2. Theoretical background
2.1 Missing heritability in GWAS
Genome-wide association studies (GWAS) is a scientific method that allows researchers to identify common genetic variations associated with small effects on complex traits and diseases by means of systematic testing across the genome. This opportunity has arisen as technological developments have led to genotyping technologies that measure all variants in the genome simultaneously. As a result, researchers have been able to probe and compare genetic data from unrelated individuals as well, in addition to twin and family studies. Subsequently, specific DNA variations could be identified, which was not possible with earlier twin studies. Involved variants include single-nucleotide polymorphisms (SNPs) and copy-number variations (CNVs; insertions or deletions, usually >100 kb). The current research to biological determinants of personality converges its analysis on SNPs associated with personality traits. GWAS have tested these ‘personality SNPs’ by computing differences in mean score of a trait for the alternative SNP alleles, or differences in allele frequencies between cases and controls (De Moor, et al., 2012). The minor allele frequency (MAF) contains the occurence of the least common allele in a given population, and for measured alleles this was at lease 0.01 and mostly higher (Wray, et al., 2014). Therefore, the list of identified SNPs that correlate to personality traits consists of
common genetic variants only. The threshold for significant association is stringent: 5 x 10-8 (Chanock, et al., 2007), to avoid false positives. Eventually, effects of all genome-wide significant (GWS) SNPs could be summed to determine the proportion of variance in liability explained by these SNPs together (Wray, et al., 2014). Basically, GWAS is based on the ‘common disease, common variant’ hypothesis, which states that allelic variants present in more than one to five percent of the population would explain common diseases (Pritchard, 2001; Reich & Lander, 2001). However, when association studies were applied on identification of SNPs in complex genetic traits, estimates of SNP-heritability (h2SNP) were much lower than earlier estimates of heritability on the liability scale based on family studies (h2) (figure 2). This phenomenon has been termed ‘missing heritability’, and consensus is lacking on its causality and research strategies to resolve the problem of finding the liable genetics of complex traits (Manolio, et al., 2009).
Figure 2 Estimates of heritability on the liability scale. Family studies (h2): proportion of variance of liability to complex traits attributed to inherited genetic factors. Bars indicate a mixture of 95% confidence intervals from meta-analysis and of reported ranges. SNP-heritability (h2SNP): proportion of variance of liability to complex traits explained by all SNPs. Bars indicate 95% intervals as an estimate ± 1.96 SE (OCD; ADHD). It is notable that values of h2 are much greater h2SNP, which has been termed ‘missing heritability’ (Wray, et al., 2014).
7 As earlier linkage studies have indicated that genetic variability of complex traits cannot be explained according to a small amount of rare variants with large effects, GWAS have demonstrated that it can neither be explained by a limited number of common variants with moderate effects (Manolio, et al., 2009). At this moment, the prevalent hypothesis states that complex characteristics result from operations of many genetic variations combined. Correspondingly, complex traits are polygenic and result from underlying genetic architectures that include both common and rare risk variants (Wray, et al., 2014). Correspondingly, missing heritability could be explained according to difficulties with measuring rare SNPs, of which contributions are nonetheless unneglectable. Moreover, it has been proven that an increase in sample size (and therefore effect size) leads to an increase of identified common GWS variants (Visscher et al., 2012). This would imply that common SNPs of intermediate effects do indeed contribute to observable phenotypic traits, but earlier studies were simply underpowered to detect them. Data from large meta-analyses of multiple GWAS could offer the solution in investigating the polygenic architectures of complex genetic traits, and subsequently finding the missing heritability.
2.2 Polygenic risk score method
Previous GWAS studies have shown that more disease variance can be explained by involving marginally significant SNPs as well, than by use of GWS (p < 10-8) SNPs alone. This implicates that the ensemble of sub-significant SNPs may conceal a ‘genetic signal’ that can be captured by genomic profile risk scoring (GPRS). This method utilizes effect sizes from GWAS from a ‘discovery sample’ to generate genomic profile risk scores (GPRS), consisting of the weighted sum of its trait-associated alleles, in an independent ‘target sample’ (Wray, et al., 2014). Importantly, SNPs that are closely located to each other are more likely to be inherited together since the probability of recombination during meiosis is lower than within more distant genetic regions. Such correlated alleles are said to be in linkage disequilibrium (LD) (figure 3). Specifically, two alleles at different loci on a chromosome are in LD if they occur together more often than would be predicted by random chance. Clumping is a procedure in which SNPs are selected based on the association p-value and stringent LD threshold (r2 < .2 across 500 kb) between SNPs (Wray, et al., 2014). The resulting set of variants includes the most associated SNPs within LD regions. Moreover, the discovery and target samples are assembled independently, and individuals are conventionally unrelated (more distantly related than second cousins). This prevents reflection of a higher LD than the real population LD by causal variants (false positives), as genetically related individuals share more genetic variants and often share environments. Eventually, profile risk scores can be computed as the sum of risk alleles times their effect sizes (SNP effect) per individual from an independent target sample. A higher profile score means an increased genetic susceptibility for a certain trait.
Figure 3 Linkage disequilibrium (LD). This heatmap displays high correlations between SNPs that are located in the same genomic region, and associations are likely to be caused by the same causal variant. More distant SNPs are less correlated, and associations are more likely to result from an independent causal variant.
(http://www.eurac.edu/en/research/health/biomed/ services/Pages/LDExplorer.aspx)
8 The main aim of association analysis is to identify specific associated variants, which makes a stringent association threshold necessary. However, the intention of GPRS analysis is to gain knowledge about genetic architectures, so that associations between variants beneath this threshold are of interest as well. Complex traits can differ in the extent to which they are polygenic: some are influenced by more SNPs than others. In the most extreme situation, a certain phenotype is caused by one single gene, at which the appropriate threshold would correspond with one SNP. On the other extreme side, a phenotype is determined by thousands of SNPs, at which the proper threshold should be less strict in order to include and discover all the SNPs which cause the effect. Consequently, the increase of total SNP inclusion raises the ratio false positives : true positives. This is tolerable since profile score analyses can cope with type I errors, while valuable information from true positives may still contribute (Wray, et al., 2014). Lastly, polygenic scoring can be used to test association with several kinds of phenotypes, including brain measurements such as white matter integrity or brain volumes.
2.3 Imaging genetics and DTI
Imaging genetics is an analytic approach that combines genetics and imaging datasets. In this way, SNP effects can be investigated based on their association with particular phenotypes (Linden, 2012). Knowledge of interconnectedness and interactions between different brain parts and regions could make a significant contribution to the understanding of the underlying, biological mechanisms of personality profiles. White matter integrity of the brain can be measured by diffusion tensor imaging (DTI): a magnetic resonance imaging (MRI) method for imaging fiber tracts by measuring the preferred direction of proton diffusion within tissues. The major part of axonal brain fibers are coated with particularly hydrophobic myelin, which forms an electrically insulating layer. As a result, neural transmission of impulses along fibers is speeded. Protons diffuse more easily along the axonal bundles than perpendicular because there are fewer obstacles to prevent movement (Stejskal, 1965). Therefore, white matter tracts display prominent anisotropy in its diffusion characteristics while other brain areas tend to be isotropic (no preferred direction) (Mori & Zhang, 2006). The relative diffusivity of water molecules in each voxel can be quantified into directional components, whereafter fractional anisotropy (FA), as measure of white matter integrity, can be computed for each voxel to express the degree of anisotropy. FA values can range from 0 (completely isotropic) to 1 (completely anisotropic; diffusion along one axis). This scalar quantity can identify differences in white matter integrity between individuals, and thereby provide information about diversity in cognitive function and behaviours (Purves, et al., 2008; Muetzel, et al., 2015).
Myelination is the production of the myelin sheath around axons. In the central nervous system, this is supplied by oligodendrocytes. The process of myelination starts during fetal development and occurrence continues up to the time of adolescence (Huttenlocher, 1990). Maturational schedules for different brain regions vary among different individuals (Hermoye et al., 2006; Lebel et al., 2008), which reinforces the hypothesis that a unique development of white matter maturation contributes to the unique components of behaviour and characteristics during the stage of adolescence. Further DTI-based research found that, during adolescence, maturation was attained in broadly distributed association- and projection fibers (Asato et al., 2010; Eluvathingal et al., 2007; Lebel et al., 2008). Age-related changes in radial diffusion values are associated with improvements in cognitive control (Asato et al., 2010). Therefore, improvements in cognitive control could be observed as function of age as well. Moreover, previous research has demonstrated developmental differences in white matter tracts between boys and girls (Erus, et al., 2015; Simmonds et al., 2014), which could possibly underlie differences in cognition and behaviour.
9 3. Material and methods
In this exploratory study, genetic variants identified by GWAS to be associated with personality traits are probed by GPRS to investigate polygenic architectures of the five personality traits. Next, the method of imaging genetics examines the association between polygenic risk scores and FA values of white matter tracts through DTI. The scientific relevance of this research lies in closing the gap between two bodies of literature: linking possible polygenic architectures of personality traits with endophenotypic fiber bundles.
3.1 Discovery sample
For GPRS, the discovery sample consisted of genetics data from the Genetics of Personality Consortium (GPC) (De Moor, Van den Berg, & Boomsma, 2012-2015). The GPC comprises large GWAS, which have found SNPs with effects on the personality five traits (De Moor, et al., 2012; De Moor, et al., 2015; Van Den Berg, et al., 2014; Van Den Berg, et al., 2015). A meta-analysis of GWAS in ten discovery cohorts (n=17,375) and five replication cohorts (n=3,294) has been conducted for each of the five personality traits assessed with the 60-item NEO-FFI, the 240-item NEO-PI-R, or samples or combinations of these tests (De Moor, et al., 2012). DNA was extracted from blood samples of participants. Furthermore, basic checks have been performed concerning European ancestry, Mendelian errors, gender inconsistencies, and high genome-wide homozygosity (De Moor, et al., 2012). The trait-associated alleles for agreeableness, consciousness, and openness are based on results of this meta-analysis. For neuroticism, a meta-analysis is executed of GWAS in 29 discovery cohorts (n=63,661) and one replication cohort (n=9,786) (De Moor, et al., 2015). Lastly, the largest meta-analysis for extraversion so far is used, consisting of 29 cohorts (n=63,030) and a 30th cohort for replication (9,783) (Van Den Berg, et al., 2015).
3.2 Target sample
The independent target sample included genome-wide genotypes of the ongoing prenatal Generation
R Cohort study in Rotterdam, the Netherlands (n=9,778) (Jaddoe, et al., 2007; Jaddoe, et al., 2012;
White, et al., 2013). Subjects in this study are population-based school-aged children who were recruited during fetal life. Therefore, subject recruitment into Generation R is non-biased and the study represents a reliable reflection of the true population in Rotterdam, including the ethnical diversity. However, for the current study, only European individuals are included, so that an unambiguous statement can be generated about one homogeneous group. The SNPs associated with personality that are found by the GPC are analysed in the target sample, and subsequently its effects on the brain. In this way, the effect of GWAS results on the general population is examined. A sub-sample of Generation R children underwent neuroimaging sessions, including high-resolution structural, diffusion tensor, and resting-state functional MRI sequences (White, et al., 2013). Few subjects had to be excluded due to excessive motion or artefacts in the data. DTI sequences of white matter integrity at two different points in time are used: around five- and nine-years-old. Eventually, genetic data of 559 6- to 10-year-old children (the ‘at-five sample’), and 943 9- to 12-year olds (the ‘at-nine sample’) was available. 159 of these children were scanned at both ages, and this data is analysed in order to find longitudinal differences.
10
3.3 GPRS
First, profile risk scores are computed for every individual of the target sample by means of GPRS to determine genetic susceptibility for each trait (figure 5). These risk scores are based on SNP effect sizes and p-values from the discovery sample. The used genetics software in Linux contained a risk score function that multiplied the amount of alleles per individual (0, 1, or 2) with the effect size for every SNP above a threshold of either 0.1, 0.01, 0.05, or 0.001. This function is expressed as:
By testing different threshold values, genetic architectures of personality are explored: if a particular trait is very polygenic, then the optimum effect size will be found at a less stringent threshold value. Accordingly, p-value thresholds do not necessarily say anything about the outcome, but about the amount of SNPs in the risk scores. In this way, genetics of personality is quantified and modelled for every subject in the independent target sample.
Figure 4 Genomic profile risk scoring (GPRS) method. (1) A discovery sample is identified based on GWAS, which provides summary statistics. (2) An independent target sample is identified based on (another) GWAS. (3) Which SNPs do both samples share? (4) Clumping: the SNP list is pruned based on the LD threshold (r2 < .2 across 500 kb). (5) Clumping: the SNP list is limited further to SNPs with association p-value less than defined threshold. (6) Generation of GPRSs in the target sample (score = Σ (risk allele x effect size)). (7) Regression analysis will be performed (Wray, et al., 2014).
3.4 Imaging genetics
Multiple linear regression analysis has estimated relationships between standardized risk scores for personality traits as independent variable (predictor) and standardized FA values of white matter fiber tracts as dependent variable (figure 5). In this way, imaging genetics uses this imaging technology to evaluate genetic variation. Since evidence is lacking to suggest that specific tracts or sets of tracts are responsible for the generation of complex traits, this study expected such traits to result from contributions of widespread brain regions together. Presumably, deviations in FA values in a certain white matter tract could have effects on other fiber bundles which are functionally connected. In order to investigate involvement of multiple brain regions in personality, global associations across multiple white matter tracts are researched by means of global FA measures.
Subsequently, relationships are investigated between polygenic risk scores and seven separate white matter tracts that are thought to play a role in personality: the anterior thalamic radiation (ATR), uncinate fasciculus (UF), inferior frontooccipital fasciculus (IFF), superior longitudinal fasciculus (SLF), cingulate gyrus cingulum (CGC), parahippocampal cingulum (PC), and forceps minor (FM). The ATR connects thalamic nuclei with the prefrontal cortex, which is known to be responsible for regulation
11 of executive function and planning of complex behaviours (Floresco & Grace, 2003; Van Der Werf, et al., 2003). The UF links components of the limbic system in the anterior temporal lobe with the orbitofrontal cortex (Kier, et al., 2004; Von Der Heide, et al., 2013), and disruption results in severe memory impairment (Gaffan, Easton, & Parker, 2002). The IFF integrates auditory and visual association cortices with the prefrontal cortex, and plays an important role in language semantics (Martino, et al., 2010). The SLF connects the dorsolateral prefrontal cortex with the supramarginal gyrus in the temporal lobe (Petrides & Pandya, 2002), and maturation of white matter in this tract correlates with working memory performance (Karlsgodt, et al., 2008; Klingberg, 2006). Findings of De Schotten et al. (2011) suggest that the SLF represents a direct communication between dorsal and ventral networks during orienting attention. Moreover, the left SLF is thought to connect receptive (auditory nuclei) with expressive (speech nuclei) language areas (Bernal & Altman, 2010). The CGC brings projections from the cingulate gyrus to the cingulum (Wakana, et al., 2004), and therefore allows limbic structures for emotion and pain (Hadland, et al., 2003; Vogt, 2005) and learning and memory (Frankland, et al., 2004; Teixeira, et al., 2006) to communicate. The PC is part of the cingulum bundle as well, and consists mainly of projections of the posterior parietal cortex (Jones, Christiansen, Chapman, & Aggleton, 2013). The FM consists of projections of bilateral frontal lobes, and crosses the midline via the genu of the corpus callosum Wakana, et al., 2004). For each trait, polygenic risk scores are used at the threshold that displayed the highest effect size during regression analysis with global FA values.
Figure 5 White matter tractography. FA values of fiber bundles are determined by diffusion tensor imaging (DTI). Every colour in this template represents a distinct fiber bundle. Visualisation by FSLView..
(http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/AutoPtx) (Fiber bundles: Uncinate Fasciculus, Anterior Thalamic Radiation, Medial Lemniscus, Inferior Frontooccipital Fasciculus, Superior Thalamic Radiation, Cingulate Gyrus Cingulum, Posterior Thalamic Radiation, Superior
Longitudinal Fasciculus, Acoustic Radiation, Forceps Minor, Forceps Major, Inferior Longitudinal Fasciculus, Middle Cerebellar Peduncle, Parahippocampal Cingulum, Corticospinal Tract)
First, there is searched for associations between genetic risk scores of nine-year-old children (n=943) and white matter integrity. Multiple linear regression analyses are accomplished to compare the variance explained by the discovery sample (x = profile score) to phenotypic white matter tracts of individuals from the target sample (y = FA value). Data analysis is carried out in RStudio (R version 3.2.5). The model is defined as:
12 Second, there is searched for possible longitudinal differences between children at five-years-old and nine-years-old (n=159) by calculating the difference in FA value at these two time points for every child. In this way, it is examined if risk scores serve as predictor of these longitudinal differences by means of multiple regression analysis. There has been corrected for differences in growth and development between individuals as well: a child who already takes a high position on the growth curve at a relatively young age will display a less rapid increase than a child who develops relatively late. This model is defined as:
The children of Generation R did not make the NEO-PI-R to directly assess the five personality traits, so that representativeness of the polygenic risk scores on the target sample could not be examined. Instead, it is assumed that the identified SNPs contribute to the development of certain personality traits, since the very large discovery sample contains much statistical power and every GWAS contains at least one replication cohort to guarantee the goodness of the fit. Yet, in attempt to still assess behaviour, associations of externalizing (attention problems and aggressive behaviour) and internalizing (anxiety, sadness, and withdrawn) scores of the Child Behaviour Checklist (CBCL) (Roza, et al., 2009) with personality and white matter integrity are studied. The CBCL is a reliable and valid questionnaire (Achenbach & Rescorla, 2000) to quantify behavioural and emotional problems of children in a standardized way. Questions are answered by the mother. The two multiple regression model are defined as:
3.5 Possible confounding factors
First, two alleles at different loci on a chromosome are in LD if they occur more often than would be predicted by random chance. During assembly of the discovery sample, clumping intended to select the most associated SNP per LD region to exclude SNPs associated with the same causal variant. In this way, SNPs associated with independent causal variants had to remain. Second, subjects in the discovery and target sample are conventionally unrelated to prevent reflection of a higher LD than the real population LD. After all, genetically related individuals share more genetic variants. A third confound could occur by population stratification effects, at which subtle differences in origin can lead to higher or lower profile risk scores. For such effects is corrected by means of principal components. 4. Results
4.1 Personality and white matter integrity at nine-years-old
For agreeableness, the highest effect size on global FA values is predicted by polygenic risk scores that consist of SNPs below the threshold of p=0.001. For consciousness, most effect is found by p=0.01 risk scores, and for openness by p=0.05 (supplementary materials: figure 8, table 3). The threshold value
Δ(FA t2 (@9) – FA t1 (@5)) ~ risk score + PC1-4 + age + gender + FA t1 (@5)
CBCL score (ext or int) ~ risk score + PC1-4 + age + gender CBCL score (ext or int) ~ FA (@9) + age + gender
13 of agreeableness is the most stringent, and therefore this trait is thought to be the least polygenic. In other words, fewer SNPs cause agreeableness, and, hence, fewer SNPs should be included in its analysis. Likewise, openness is thought to be the most polygenic. Accordingly, these specific thresholds for SNP inclusion are considered characteristic for each individual trait, and used for further analyses during this research.
Subsequently, it is examined if any relationship exists between polygenic risk scores and seven individual white matter tracts which are thought to play a role in personality formation (table 2, figure 6). Table 2 presents standardized effect sizes (β) and corresponding p-values for the strength of polygenic risk scores as predictor of white matter integrity. Beta coefficients are considered part of a trend, when they lay further away from zero than +/- 0.05 times the standard deviation. Findings suggest that individuals with more SNPs for agreeableness, and therefore a higher polygenic risk score for this trait, have, on average, lower FA values in the IFF 0.07252613; P=0.0255*), PC (β=-0.06684235; P=0.0365*), UF (β=-0.05953353; P=0.0680), and FM (β=-0.05132363; P=0.1168). For consciousness, a higher risk score leads to a lower FA value in the FM (β=-0.05152503; P=0.1141), and those with an increased risk for extraversion have a reduced FA value in the PC (β=-0.05814054; P=0.0683). Individuals with higher risk scores for neuroticism have more negative FA values in the FM (β=-0.09796165; P=0.0028**) and UF (β=-0.06290685; P=0.0542), and a more positive FA value in the PC (β=0.05160381; P=0.1071). For openness, a higher risk score leads to a lower FA value in the SLF (β=-0.07513985; P=0.0202*).
The most significant result suggests a negative relationship between neuroticism and FA values in the forceps minor of the corpus callosum, which connects the frontal lobes of the brain. Correspondingly, individuals with high risk for neuroticism have less integration in frontal functions. Moreover, effect sizes of risk scores for agreeableness on the IFF for visual and auditory integration and on the limbic PC, and of risk scores for openness on the SLF for working memory and attention were significant as well.
Figure 6 Effect sizes (beta) of polygenic risk scores on FA values of white matter tracts in nine-year-old subjects (n=943). The seven separate tracts are displayed on the x-axis, and effect sizes (beta) of risk scores on FA values are displayed on the y-axis. Beta coefficients are considered part of a trend when they lay further away from zero than +/- 0.05 times the standard deviation, and this location is denoted by dashed lines.
14 Table 2 Effect sizes and p-values of polygenic risk scores of personality traits on FA values of single white matter tracts in the nine-year-old sample. Regression coefficient (β) and P-value of polygenic risk scores for every personality trait. The green cells contain results with effect sizes above or below 0.05, and are considered part of a trend.
Agreeableness (p = 0.001) Consciousness (p = 0.01) Extraversion (p = 0.01) Neuroticism (p = 0.01) Openness (p = 0.05) Tract β P β P β P β P β P ATR -0.0372 0.2490 -0.0397 0.2166 -0.0132 0.6826 -0.0495 0.1258 -0.0401 0.2129 CGC 0.0090 0.7754 0.0037 0.9055 -0.0129 0.6811 -0.0040 0.8991 -0.0248 0.4302 FM -0.0513 0.1168 -0.0515 0.1141 0.0105 0.7479 -0.0980 0.0028** -0.0100 0.7585 IFF -0.0725 0.0255* 0.0038 0.9079 -0.0391 0.2276 -0.0403 0.2154 -0.0230 0.4780 PC -0.0668 0.0365* -0.0269 0.3982 -0.0581 0.0683 0.0516 0.1071 -0.0299 0.3488 SLF -0.0451 0.1653 -0.0152 0.6378 -0.0326 0.3151 -0.0340 0.2954 -0.0751 0.0202* UF -0.0595 0.0680 -0.0311 0.3393 -0.0121 0.7100 -0.0629 0.0542 -0.0174 0.5931
4.2 Personality and white matter integrity between five- and nine-years-old
Longitudinal differences in white matter integrity are found between the five- and nine-year-old sample (figure 7; table 3). Even though the sample size was very low (n=159), thereby allowing the power to drop considerably, significant effect sizes are discovered of polygenic risk scores for neuroticism as predictor of longitudinal differences in FA values in the FM (β = -0.2477; P = 0.0007***) of the corpus callosum and the UF (β = -0.1518; P = 0.0114*) which connects the limbic system with the orbitofrontal cortex for decision making and emotional processing (Bechara, Damasio, & Damasio, 2000). Accordingly, FA values in the FM and UF are less developed (negative slope on the absolute difference) between the age of five and nine in subjects with high polygenic risk scores for neuroticism. In other words, high neuroticism scores corresponded with a less steep increase in white matter integrity. Interestingly, for higher neuroticism scores, a (non-significant) trend displays lower white matter integrity in all seven measured tracts. Interestingly, for neuroticism scores, a (non-significant) trend displays high effect sizes, and therefore, deviant white matter development in all seven tracts.
Figure 7 Effect sizes (beta) of polygenic risk scores on the difference in FA values of white matter tracts between two time points (n=159). The seven separate tracts are displayed on the x-axis, and effect sizes (beta) of risk scores on FA values are displayed on the y-axis. Beta coefficients are considered part of a trend when they lay further away from zero than +/- 0.05 times the standard deviation, and this location is denoted by dashed lines.
15 Table 3 Effect sizes and p-values of polygenic risk scores of personality traits on the difference in FA values of separate white matter tracts between @9-@5.
Openness (p = 0.05) Consciousness (p = 0.01) Extraversion (p = 0.01) Agreeableness (p = 0.001) Neuroticism (p = 0.01) Tract β P β P β P β P β P ATR 0.0342 0.5866 -0.0050 0.9335 -0.0516 0.3850 -0.0574 0.3529 -0.1054 0.1086 CGC 0.0182 0.7752 -0.0276 0.6599 0.0398 0.5206 -0.0425 0.5075 -0.1118 0.0997 FM 0.0394 0.5774 -0.0126 0.8542 0.0859 0.2015 0.0092 0.8950 -0.2477 0.0007*** IFF -0.0078 0.8979 -0.0333 0.5739 0.0182 0.7564 -0.0106 0.8616 -0.0842 0.1915 PC 0.0171 0.7567 0.0596 0.2625 -0.0205 0.6963 -0.0078 0.8864 0.0897 0.1225 SLF 0.0165 0.8036 0.0006 0.9915 -0.0234 0.7141 0.0343 0.6019 -0.1080 0.1206 UF 0.0473 0.4122 0.0110 0.8443 0.0145 0.7919 -0.0057 0.9204 -0.1518 0.0114*
4.3 Personality and white matter integrity, and CBCL
No significant results are found that support the association between white matter integrity and polygenic risk scores with CBCL scores. This suggests that polygenic risk scores for personality traits do not predict externalizing or internalizing behaviour of children according to the CBCL. Results can be viewed in the supplementary materials.
5. Discussion
In this exploratory research, the relationship between genetics for personality and white matter integrity is examined in population-based children. By using polygenic risk scores that are based on a large discovery sample, this study distinguishes itself from earlier imaging genetics studies that only looked at one or few genetic variants, such as Bis, et al. (2012), Hariri et al. (2006), and Martin et al. (2009). The current method is more adequate since complex traits, such as personality, are influenced by thousands of SNPs combined. Furthermore, this research implies strong causality: effects of genetic architectures on the brain. The usual reverse causality does not play a role. Therefore, it can be concluded that final results propose deviant white matter integrity as a result of SNP effects that are associated with agreeableness, neuroticism, and openness.
First, for agreeableness, a significant association is found between increased risk scores and lower FA values in the IFF, and PC. This implies that individuals who have higher risk for this trait are less capable of integrating visuals and sounds, and have a less developed cingulum to the hippocampus for memory encoding and retrieval. Second, a significant relationship between higher openness and lower FA values in the SLF is found. According to the previously discussed tract functions, people who are more open for experience are likely to have worse working memory, to be worse in orienting attention, and worse in connecting receptive with expressive language areas. All above statements can be tested for association with behaviour during the course of Generation R. Third, the most significant finding demonstrates that higher risk scores for neuroticism are associated with lower FA values in the FM. This implies that a lack of frontal integration leads to neuroticism, and its occurrence can be led back to specific genetic architectures. Lastly, when the Generation R children reach adolescence, it is recommended that they complete the NEO-PI-R, in order to see how this corresponds with their risk scores.
16
5.1 Neuroticism
Results associated with neuroticism displayed the most significant effect sizes. Besides the significant finding of lower FA in the FM, FA values in the UF seem to decrease as well when risk scores for neuroticism increase. This indicates a weaker connection between the limbic system and the orbitofrontal cortex for decision making and emotional processing. Previous studies have found facets of neuroticism, such as anxiety, angry hostility, depression, self-consciousness, impulsiveness, and vulnerability, to be extensively correlated (Costa & McCrae, 1992a). Individuals with high neuroticism scores tend to have negative emotional responses to challenges on a frequent basis, which are out of proportion to the circumstances (McCrae & Costa, 2003, cited in Lahey, 2009). This knowledge can be explained by weakened FA value of the UF. Moreover, a previous imaging study found a significant association between neuroticism and lower white matter integrity in the UF (McIntosh, et al., 2013). Even though, in this research, the found effect size of risk scores of neuroticism on the UF is not statistically significant, it should not be overlooked, because the possibility exists that the current sample size (n=943) is not powerful enough yet to detect significance. The larger imaging dataset of Generation R study, which consists of 4,000 nine-year-old children, will be released at any time from now. It is expected that the large extent of this dataset will reveal more powerful relationships, since statistical power depends on three factors: statistical significance, effect size, and sample size. An increase in sample size will lead to an increase in power. However, when the found effect size on the UF is coincidence after all, and no association between these variables exists, the p-value will remain not significant. In an experiment to false positives in imaging genetics, Meyer-Lindenberg et al. (2008) used genetic variants that were unlikely to impact relevant brain function. The observed rate of positives was low, which indicates conservative statistical thresholds. Furthermore, longitudinal findings of subjects with higher neuroticism scores displayed less developed FA values in the FM and UF. The effect sizes on these tracts were both significant, supporting the hypothesis that the association between higher neuroticism and lower FA values in the UF exists. Therefore, it should be studied by the large imaging dataset of Generation R.
Earlier research (both twin studies and genetic analyses), has shown that heritability of neuroticism peaks around early adolescence and early adulthood (Gillespie, et al., 2004; Lake, et al., 2000; Rettew, et al., 2006; Viken, et al., 1994). Correspondingly, the magnitude of influence of genetic architectures for neuroticism will increase when the Generation R children reach this phase in age. As a consequence, effect sizes of neuroticism scores will increase, and results are expected to have high potential in clarifying aetiology of neuroticism. In his paper, Lahey (2009) extensively explains that neuroticism is of profound public health significance. Knowledge expansion of neuroticism and its aetiology is desirable, since it is highly correlated with broad range of mental and physical health problems (more than the other traits) (Malouff, Thorsteinsson, & Schutte, 2005; Malouff, Thorsteinsson, & Schutte, 2006; Saulsman & Page, 2004). Understanding of the relationship between neuroticism and the diverse variety of health outcomes is thought to reveal what these outcomes have in common, and, subsequently, could support improved strategies for prevention (Lahey, 2009). During the course of this longitudinal research, Generation R can examine the predictive utility of neuroticism in predicting mental and physical health outcomes.
5.2 Global FA values
This research has chosen to continue data analysis with the polygenic risk scores for each personality trait of which the threshold for SNP inclusion gave the highest effect size on global FA values. This p-value threshold for SNP inclusion is characteristic for each individual trait, since its stringency represents the extent to which a particular trait is polygenic. The idea behind this approach was that
17 personality traits result from contributions of widespread brain regions together, as no evidence exists that responsibility can be assigned to specific white matter tracts. Since this study was exploratory, the main goal was to find any main effects in order to direct future studies where to continue research. Therefore, the use of global FA values as first step in the good direction seemed convenient. However, effect sizes of risk scores on global FA values were never significant (supplementary materials: figure 8, table 3), but those on FA in separate white matters were. Next, data analysis has revealed that effect sizes of risk scores on FA values of separate tracts were sometimes higher at the other thresholds for SNP inclusion than those with highest effects on global FA values. For example, risk scores for consciousness has displayed most effect on global FA at the SNP inclusion threshold of p=0.01. Unfortunately, no significant effect sizes of these risk scores on separate white matter tracts are found. Yet, risk scores at the threshold of p=0.001 do appear to display a significant effect size on FA of the PC (supplementary materials: figure 9; β=-0.0695, P=0.029477*). Perhaps, less SNPs contribute to the white matter integrity of the PC (SNP inclusion threshold p=0.001), whereas more consciousness SNPs are involved in the global picture of brain fibers (SNP inclusion threshold p=0.01). This is not unexpected, since there is a scale difference between global measurements of white matter integrity and separate white matter tracts. Therefore, conclusions that are drawn from global associations cannot completely reliable be extended to properties of fiber tracts. Nevertheless, global FA values are useful exploratory variables, and have, in this research, proven to be helpful providers of preliminary information. Yet, based on the results, future research is recommended to directly look at separate fiber bundles, instead of using a global measure of white matter integrity.
In conclusion, this study should be considered a proof-of-concept for the evaluation of genetic architectures of complex traits, rather than decisively addressing the hypotheses concerning the genetic overlap of brain imaging measures with risk for personality traits. Since results identified evidence of genetic overlap, this approach of imaging genetics has proven to be a useful tool in investigating the neurobiological factors associated with personality traits and is, therefore, highly recommended to more definitive and larger future studies. Furthermore, future research is recommended to involve other imaging techniques as well, such as structural or functional research, besides DTI. Overlap of different endophenotypes would provide valuable directions to the neurobiological underpinnings of complex traits. Along these lines, a more and more complete picture of the establishment of personality could systematically be clarified.
6. Acknowledgements
Grateful acknowledgement is given to the following:
Philip Jansen, MD, MSc, PhD student, for his attentive supervision and tips regarding the data and how to analyse it. Dr. J. A. Van Hooft for his feedback on structure and content. The contribution of children and parents to Generation R. Neuroimaging studies within Generation R is made possible by support of the Netherlands Organization for Health Research and Development (NWO), the European Community’s 7th Framework Programme, the Stichting Sophia Kinderziekenhuis Fonds, and General Electric Healthcare. Financial support is provided by the Erasmus Medical Center, Rotterdam, the Erasmus University Rotterdam, the Netherlands Organization for Scientific Research (NWO), the Ministry of Health, Welfare and Sport, and the Ministry of Youth and Families. The Genetic Personality Consortium for publishing their genetics data. Members of the methodology store at the psychology department of the University of Amsterdam for their help with the language R (software environment for statistical computing and graphics).
18 7. Literature
Achenbach, T. M., & Rescorla, L. A. (2000). Manual for the ASEBA Preschool Forms & Profiles: Child
Behavior Checklist for Ages 1/2-5, Language Development Survey, Caregiver-Teacher Report Form; ASEBA., an Integrated System of Multi-informant Assessment. ASEBA.
American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders
(revised 4th ed.). Washington, DC: American Psychiatric Association.
Asato, M. R., Terwilliger, R., Woo, J., & Luna, B. (2010). White matter development in adolescence: a DTI study. Cerebral Cortex, 20(9), 2122-2131.
Bechara, A., Damasio, H., & Damasio, A. R. (2000). Emotion, decision making and the orbitofrontal cortex. Cerebral cortex, 10(3), 295-307.
Bernal, B., & Altman, N. (2010). The connectivity of the superior longitudinal fasciculus: a tractography DTI study. Magnetic resonance imaging, 28(2), 217-225.
Bis, J. C., DeCarli, C., Smith, A. V., Van Der Lijn, V., ..., & Seshadri, S. (2012). Common variants at 12q14 and 12q24 are associated with hippocampal volume. Nature genetics, 44(5), 545-551. Bjørnebekk, A., Fjell, A. M., Walhovd, K. B., Grydeland, H., Torgersen, S., & Westlye, L. T. (2013).
Neuronal correlates of the five factor model (FFM) of human personality: Multimodal imaging in a large healthy sample. Neuroimage, 65, 194-208.
Bouchard Jr, T. J., & Loehlin, J. C. (2001). Genes, evolution, and personality. Behavior genetics, 31(3), 243-273.
Cattell, R. B., Eber, H. W., & Tatsuoka, M. M. (1970). Handbook for the sixteen personality factor
questionnaire (16 PF): In clinical, educational, industrial, and research psychology, for use with all forms of the test. Institute for Personality and Ability Testing.
Chanock, S. J., Manolio, T., Boehnke, M., Boerwinkle, E., Hunter, D. J., Thomas, G., . . . Brooks, L. D. (2007). Replicating genotype–phenotype associations. Nature, 447(7145), 655-660.
Chapman, B. P., Roberts, B., & Duberstein, P. (2011). Personality and longevity: knowns, unknowns, and implications for public health and personalized medicine. Journal of aging research,
2011.
Comrey, A. L. (1970). Manual for the Comrey personality scales. San Diego: CA: Educational and Industrial Testing Service.
Costa, P. T., & McCrae, R. R. (1992a). Four ways five factors are basic. Personality and individual
differences, 13(6), 653-665.
Costa, P. T., & McCrae, R. R. (1992b). Professional manual: revised NEO personality inventory (NEO-PI-R) and NEO five-factor inventory (NEO-FFI). Odessa, FL: Psychological Assessment Resources, 61.
Dahlstrom, W. G., Welsh, G. S., & Dahlstrom, L. E. (1975). An MMPI handbook: Research applications
(Vol. 2). University of Minnesota Press.
De Moor, M. H., Costa, P. T., Terracciano, A., Krueger, R. F., De Geus, E. J., Toshiko, T., . . . Amin, N. (2012). Meta-analysis of genome-wide association studies for personality. Molecular
19 De Moor, M. H., Van Den Berg, S. M., Verweij, K. J., Krueger, R. F., Luciano, M., Vasquez, A. A., . . .
Gordon, S. D. (2015). Meta-analysis of genome-wide association studies for neuroticism, and the polygenic association with major depressive disorder. JAMA psychiatry, 72(7), 642-650. De Moor, M., Van den Berg, S., & Boomsma, D. (2012-2015). The Genetics of Personality Consortium.
Retrieved from Nederlands Tweelingen Register: http://www.tweelingenregister.org/GPC/ De Schotten, M. T., Dell'Acqua, F., Forkel, S. J., Simmons, A., Vergani, F., Murphy, D. G., & Catani, M.
(2011). A lateralized brain network for visuospatial attention. Nature neuroscience, 14(10), 1245-1246.
Distel, M. A., Trull, T. J., Willemsen, G., Vink, J. M., Derom, C. A., Lynskey, M., . . . Boomsma, D. I. (2009). The five-factor model of personality and borderline personality disorder: a genetic analysis of comorbidity. Biological psychiatry, 66(12), 11.
Eluvathingal, T. J., Hasan, K. M., Kramer, L., Fletcher, J. M., & Ewing-Cobbs, L. (2007). Quantitative diffusion tensor tractography of association and projection fibers in normally developing children and adolescents. Cerebral Cortex, 17(12), 2760-2768.
Erus, G., Battapady, H., Satterthwaite, T. D., Hakonarson, H., Gur, R. E., Davatzikos, C., & Gur, R. C. (2015). Imaging patterns of brain development and their relationship to cognition. Cerebral
Cortex, 25, 1676-1684.
Euesden, J., Lewis, C., & O'Reilly, P. (2016). PRSice: Polygenic Risk Score software. Retrieved from http://prsice.info/
Eysenck, S. B. (1983). One approach to cross‐cultural studies of personality. Australian Journal of
Psychology, 35(3), 381-391.
Floresco, S. B., & Grace, A. A. (2003). Gating of hippocampal-evoked activity in prefrontal cortical neurons by inputs from the mediodorsal thalamus and ventral tegmental area. The Journal of
neuroscience, 23(9), 3930-3943.
Frankland, P. W., Bontempi, B., Talton, L. E., Kaczmarek, L., & Silva, A. J. (2004). The involvement of the anterior cingulate cortex in remote contextual fear memory. Science, 304(5672), 881-883.
Gaffan, D., Easton, A., & Parker, A. (2002). Interaction of inferior temporal cortex with frontal cortex and basal forebrain: double dissociation in strategy implementation and associative learning.
The Journal of Neuroscience, 22(16), 7288-7296.
Gillespie, N. A., Evans, D. E., Wright, M. M., & Martin, N. G. (2004). Genetic simplex modeling of Eysenck's dimensions of personality in a sample of young Australian twins. Twin Research,
7(06), 637-648.
Hadland, K. A., Rushworth, M. F., Gaffan, D., & Passingham, R. E. (2003). The effect of cingulate lesions on social behaviour and emotion. Neuropsychologia, 41(8), 919-931.
Hariri, A. R., Drabant, E. M., & Weinberger, D. R. (2006). Imaging genetics: perspectives from studies of genetically driven variation in serotonin function and corticolimbic affective processing.
20 Hermoye, L., Saint-Martin, C., Cosnard, G., Lee, S. K., Kim, J., Nassogne, M. C., & Wakana, S. (2006).
Pediatric diffusion tensor imaging: normal database and observation of the white matter maturation in early childhood. Neuroimage, 29(2), 493-504.
Hettema, J. M., Neale, M. C., Myers, J. M., Prescott, C. A., & Kendler, K. S. (2006). A population-based twin study of the relationship between neuroticism and internalizing disorders. American
journal of Psychiatry.
Huttenlocher, P. R. (1990). Morphometric study of human cerebral cortex development.
Neuropsychologia, 28(6), 517-527.
Jaddoe, V. W., Bakker, R., Van Duijn, C. M., van der Heijden, A. J., Lindemans, J., Mackenbach, J. P., . . . Verhulst, F. C. (2007). The Generation R Study Biobank: a resource for epidemiological studies in children and their parents. European journal of epidemiology, 22(12), 917-923. Jaddoe, V. W., van Duijn, C. M., Franco, O. H., van der Heijden, A. J., van IIzendoorn, M. H., de
Jongste, J. C., . . . Rivadeneira, F. (2012). The Generation R Study: design and cohort update 2012. European journal of epidemiology, 27(9), 739-756.
Jang, K. L., Livesley, W. J., & Vemon, P. A. (1996). Heritability of the big five personality dimensions and their facets: A twin study. Journal of personality, 64(3), 577-592.
Jang, K. L., Livesley, W. J., Angleitner, A., Riemann, R., & Vernon, P. A. (2002). Genetic and
environmental influences on the covariance of facets defining the domains of the five-factor model of personality. Personality and individual Differences, 33(1), 83-101.
Johnson, W., & Krueger, R. F. (2004). Genetic and environmental structure of adjectives describing the domains of the Big Five Model of personality: A nationwide US twin study. Journal of
Research in Personality, 38(5), 448-472.
Jones, D. K., Christiansen, K. F., Chapman, R. J., & Aggleton, J. P. (2013). Distinct subdivisions of the cingulum bundle revealed by diffusion MRI fibre tracking: implications for neuropsychological investigations. Neuropsychologia, 51(1), 67-78.
Karlsgodt, K. H., van Erp, T. G., Poldrack, R. A., Bearden, C. E., Nuechterlein, K. H., & Cannon, T. D. (2008). Diffusion tensor imaging of the superior longitudinal fasciculus and working memory in recent-onset schizophrenia. Biological psychiatry, 63(5), 512-518.
Kendler, K. S., Neale, M. C., Kessler, R. C., Heath, A. C., & Eaves, L. J. (1993). A longitudinal twin study of personality and major depression in women. Archives of general psychiatry, 50(11), 853-862.
Kier, E. L., Staib, L. H., Davis, L. M., & Bronen, R. A. (2004). MR imaging of the temporal stem: anatomic dissection tractography of the uncinate fasciculus, inferior occipitofrontal fasciculus, and Meyer’s loop of the optic radiation. American Journal of Neuroradiology,
25(5), 677-691.
Klingberg, T. (2006). Development of a superior frontal–intraparietal network for visuo-spatial working memory. Neuropsychologia, 44(11), 2171-2177.
21 Lake, R. I., Eaves, L. J., Maes, H. H., Heath, A. C., & Martin, N. G. (2000). Further evidence against the
environmental transmission of individual differences in neuroticism from a collaborative study of 45,850 twins and relatives on two continents. Behavior genetics, 30(3), 223-233. Lebel, C., Walker, L., Leemans, A., Phillips, L., & Beaulieu, C. (2008). Microstructural maturation of the
human brain from childhood to adulthood. Neuroimage, 40(3), 1044-1055.
Linden, D. E. (2012). The challenges and promise of neuroimaging in psychiatry. Neuron, 73(1), 8-22. Malouff, J. M., Thorsteinsson, E. B., & Schutte, N. S. (2005). The relationship between the five-factor
model of personality and symptoms of clinical disorders: a meta-analysis. Journal of
Psychopathology and Behavioral Assessment, 27(2), 101-114.
Malouff, J. M., Thorsteinsson, E. B., & Schutte, N. S. (2006). The five-factor model of personality and smoking: A meta-analysis. Journal of drug education, 36(1), 47-58.
Manolio, T. A., Collins, F. S., Cox, N. J., Goldstein, D. B., Hindorff, L. A., Hunter, D. J., . . . Cho, J. H. (2009). Finding the missing heritability of complex diseases. Nature, 461(7265), 747-753. Martin, E. I., Ressler, K. J., Binder, E., & Nemeroff, C. B. (2009). The neurobiology of anxiety disorders:
brain imaging, genetics, and psychoneuroendocrinology. Psychiatric Clinics of North America,
32(3), 549-575.
Martino, J., Brogna, C., Robles, S. G., Vergani, F., & Duffau, H. (2010). Anatomic dissection of the inferior fronto-occipital fasciculus revisited in the lights of brain stimulation data. Cortex,
46(5), 691-699.
McCrae, R. R., & Costa Jr, P. T. (1997). Personality trait structure as a human universal. American
psychologist, 52(5), 509-516.
McCrae, R. R., & Costa Jr, P. T. (1999). A five-factor theory of personality. Handbook of personality:
Theory and research, 2, 139-153.
McIntosh, A. M., Bastin, M. E., Luciano, M., Maniega, S. M., Hernández, M. D., Royle, N. A., . . . Wardlaw, J. M. (2013). Neuroticism, depressive symptoms and white-matter integrity in the Lothian Birth Cohort 1936. . Psychological medicine, 43(06), 1197-1206.
Meyer-Lindenberg, A., Nicodemus, K. K., Egan, M. F., Callicott, J. H., Mattay, V., & Weinberger, D. R. (2008). False positives in imaging genetics. Neuroimage, 40(2), 655-661.
Mori, S., & Zhang, J. (2006). Principles of diffusion tensor imaging and its applications to basic neuroscience research. Neuron, 51(5), 527-539.
Muetzel, R. L., Mous, E., S., van der Ende, J., Blanken, L. M., van der Lugt, A., . . . White, T. (2015). White matter integrity and cognitive performance in school-age children: A population-based neuroimaging study. NeuroImage, 119, 119-12.
Petrides, M., & Pandya, D. N. (2002). Association pathways of the prefrontal cortex and functional observations. Principles of frontal lobe function, 1, 31-50.
Pilia, G., Chen, W. M., Scuteri, A., Orrú, M., Albai, G., Dei, M., . . . Mameli, C. (2006). Heritability of cardiovascular and personality traits in 6,148 Sardinians. PLoS Genet, 2(8), e132.
Pritchard, J. K. (2001). Are rare variants responsible for susceptibility to complex diseases? The
22 Purves, D., Brannon, E. M., Cabeza, R., Huettel, S. A., LaBar, K. S., Platt, M. L., & Woldorff, M. G.
(2008). Principles of cognitive neuroscience (Vol. 83). Sunderland: MA: Sinauer Associates. Reich, D. E., & Lander, E. S. (2001). On the allelic spectrum of human disease. TRENDS in Genetics,
17(9), 502-510.
Rettew, D. C., Vink, J. M., Willemsen, G., Doyle, A., Hudziak, J. J., & Boomsma, D. I. (2006). The genetic architecture of neuroticism in 3301 Dutch adolescent twins as a function of age and sex: a study from the Dutch twin register. Twin Research and Human Genetics, 9(01), 24-29. Roza, S. J., Verhulst, F. C., Jaddoe, V. W., Steegers, E. A., Mackenbach, J. P., Hofman, A., & Tiemeier,
H. (2009). Maternal smoking during pregnancy and child behaviour problems: the Generation R Study. International Journal of Epidemiology, 38(3), 680-689.
Saulsman, L. M., & Page, A. C. (2004). The five-factor model and personality disorder empirical literature: A meta-analytic review. Clinical psychology review, 23(8), 1055-1085. Segal, N. L. (2012). Born together—reared apart: The landmark Minnesota twin study. Harvard
University Press.
Simmonds, D. J., Hallquist, M. N., Asato, M., & Luna, B. (2014). Developmental stages and sex
differences of white matter and behavioral development through adolescence: a longitudinal diffusion tensor imaging (DTI) study. Neuroimage, 92, , 356-368.
Smith, E. E., Nolen-Hoeksema, S., Fredrickson, B., & Loftus, G. R. (2014). Atkinson & Hilgards
Introduction to Psychology (16th edition). Belmont: CA: Wadsworth/Thomson Learning.
Stejskal, E. O. (1965). Use of Spin Echoes in a Pulsed Magnetic‐Field Gradient to Study Anisotropic, Restricted Diffusion and Flow. The Journal of Chemical Physics, 43(10), 3597-3603.
Teixeira, C. M., Pomedli, S. R., Maei, H. R., Kee, N., & Frankland, P. W. (2006). Involvement of the anterior cingulate cortex in the expression of remote spatial memory. The Journal of
neuroscience, 26(29), 7555-7564.
Van Den Berg, S. M., De Moor, M. H., McGue, M., Pettersson, E., Terracciano, A., Verweij, K. J., . . . Hansell, N. K. (2014). Harmonization of Neuroticism and Extraversion phenotypes across inventories and cohorts in the Genetics of Personality Consortium: an application of Item Response Theory. Behavior genetics, 44(4), 295-313.
Van Den Berg, S. M., De Moor, M. H., Verweij, K. J., Krueger, R. F., Luciano, M., Vasquez, A. A., . . . Gordon, S. D. (2015). Meta-analysis of Genome-Wide Association Studies for Extraversion: Findings from the Genetics of Personality Consortium. Behavior genetics, 1-13.
Van Der Werf, Y. D., Jolles, J., Witter, M. P., & Uylings, H. B. (2003). Contributions of thalamic nuclei to declarative memory functioning. Cortex, 39(4), 1047-1062.
Viken, R. J., Rose, R. J., Kaprio, J., & Koskenvuo, M. (1994). A developmental genetic analysis of adult personality: extraversion and neuroticism from 18 to 59 years of age. Journal of personality
and social psychology, 66(4), 722.
Visscher, P. M., Brown, M. A., McCarthy, M. I., & Yang, J. (2012). Five years of GWAS discovery. The
American Journal of Human Genetics, 90(1), 7-24.
Vogt, B. A. (2005). Pain and emotion interactions in subregions of the cingulate gyrus. Nature
23 Von Der Heide, R. J., Skipper, L. M., Klobusicky, E., & Olson, I. R. (2013). Dissecting the uncinate
fasciculus: disorders, controversies and a hypothesis. Brain, 136(6), 1692-1707.
Wakana, S., Jiang, H., Nagae-Poetscher, L. M., Van Zijl, P. C., & Mori, S. (2004). Fiber tract–based atlas of human white matter anatomy 1. Radiology, 230(1), 77-87.
White, T., Marroun, E., H., N. I., Schmidt, M., van der Lugt, A., Wielopolki, P. A., . . . Verhulst, F. C. (2013). Pediatric population-based neuroimaging and the Generation R Study: the intersection of developmental neuroscience and epidemiology. European journal of
epidemiology, 28(1), 99-111.
Wray, N. R., Lee, S. H., Mehta, D., Vinkhuyzen, A. A., Dudbridge, F., & Middeldorp, C. M. (2014). Research review: polygenic methods and their application to psychiatric traits. Journal of
Child Psychology and Psychiatry, 55(10), 1068-1087.
Xu, J., & Potenza, M. N. (2012). White matter integrity and five-factor personality measures in healthy adults. Neuroimage, 59(1), 800-807.
Yamagata, S., Suzuki, A., Ando, J., Ono, Y., Kijima, N., Yoshimura, K., . . . Livesley, W. J. (2006). Is the genetic structure of human personality universal? A cross-cultural twin study from North America, Europe, and Asia. . Journal of personality and social psychology, 90(6), 987.
24 8. Supplementary materials
8.1 Effect sizes of different thresholds for SNP inclusion during GPRS
Results of multiple regression analysis (imaging genetics) between global FA values (independent variable) and polygenic risk scores (dependent variable) in the nine-year-old sample display most effect for consciousness, extraversion, and neuroticism at the threshold of 0.01 (figure 8, table 4). Agreeableness scores have the highest effect on FA at the threshold of 0.001, and openness scores at 0.05. From this point, this research has converged its analyses on these specific thresholds for risk score generation during GPRS in attempt to find a genetic signal between genetic risks for personality and white matter integrity.
Figure 8 Effect sizes (beta) of risk scores on global FA values of nine-year-old subjects. SNPs are selected at four different thresholds to investigate which would display the most effect. In this graph, it can be noted that the highest effect size for agreeableness is found at p=0.001, for consciousness, extraversion, and neuroticism at p=0.01, and for openness at p=0.05.
Table 4 Effect sizes of risk scores on global FA values at nine-years-old. Regression coefficient (β) and P-value of polygenic risk scores at different thresholds for every personality trait. The cells of which the edges are bold contain the results with the highest effect size. These thresholds are used for further examination.
Openness Consciousness Extraversion Agreeableness Neuroticism
Threshold β P β P β P β P β P
0.1 -0.09423 0.1055 -0.02102 0.7182 -0.007037 0.9039 -0.06504 0.2643 -0.04847 0.4084 0.05 -0.12367 0.0337 0.008586 0.8829 -0.01518 0.7946 -0.07654 0.1888 -0.06062 0.3021 0.01 -0.02935 0.6145 -0.03591 0.5375 -0.04114 0.4804 -0.01578 0.787 -0.09573 0.1017 0.001 -0.05668 0.3314 -0.03447 0.5546 0.001884 0.9743 -0.07858 0.1785 -0.08284 0.1566
8.2 Imaging genetics for white matter tracts at nine-years-old
Effect sizes of polygenic risk scores of SNP inclusion at p=0.001 (figure 9), p=0.01 (figure 10), and p=0.05 (figure 11) are visualised. All five personality traits are included in these graphs. However, only the traits with the highest effect size on global FA will be discussed per threshold for SNP inclusion. For agreeableness, the highest effect size is found at p=0.001. Results display that higher risk scores for agreeableness lead to more negative FA values in the IFF 0.07252613; P=0.025512*), PC
