Advertisement

Brain structure and function show distinct relations with genetic predispositions to mental health and cognition

  • Shu Liu
    Correspondence
    Corresponding authors: Shu Liu, MSc and Guido van Wingen, PhD, Department of Psychiatry, Amsterdam UMC, University of Amsterdam, Meibergdreef 5, 1100 DD Amsterdam The Netherlands or
    Affiliations
    Amsterdam UMC, University of Amsterdam, Department of Psychiatry, Amsterdam Neuroscience, Amsterdam, the Netherlands
    Search for articles by this author
  • Dirk J.A. Smit
    Affiliations
    Amsterdam UMC, University of Amsterdam, Department of Psychiatry, Amsterdam Neuroscience, Amsterdam, the Netherlands
    Search for articles by this author
  • Author Footnotes
    ∗ These authors jointly supervised the work.
    Abdel Abdellaoui
    Footnotes
    ∗ These authors jointly supervised the work.
    Affiliations
    Amsterdam UMC, University of Amsterdam, Department of Psychiatry, Amsterdam Neuroscience, Amsterdam, the Netherlands
    Search for articles by this author
  • Author Footnotes
    ∗ These authors jointly supervised the work.
    Guido A. van Wingen
    Correspondence
    Corresponding authors: Shu Liu, MSc and Guido van Wingen, PhD, Department of Psychiatry, Amsterdam UMC, University of Amsterdam, Meibergdreef 5, 1100 DD Amsterdam The Netherlands or
    Footnotes
    ∗ These authors jointly supervised the work.
    Affiliations
    Amsterdam UMC, University of Amsterdam, Department of Psychiatry, Amsterdam Neuroscience, Amsterdam, the Netherlands
    Search for articles by this author
  • Author Footnotes
    ∗ These authors jointly supervised the work.
    Karin J.H. Verweij
    Footnotes
    ∗ These authors jointly supervised the work.
    Affiliations
    Amsterdam UMC, University of Amsterdam, Department of Psychiatry, Amsterdam Neuroscience, Amsterdam, the Netherlands
    Search for articles by this author
  • Author Footnotes
    ∗ These authors jointly supervised the work.
Open AccessPublished:August 09, 2022DOI:https://doi.org/10.1016/j.bpsc.2022.08.003

      Abstract

      Background

      Mental health and cognitive achievement are partly heritable, highly polygenic, and associated with brain variations in structure and function. However, the underlying neural mechanisms remains unclear.

      Methods

      We investigated the association between genetic predispositions to various mental health and cognitive traits and a large set of structural and functional brain measures from the UK Biobank (N=36,799). We also applied linkage disequilibrium score regression to estimate the genetic correlations between various traits and brain measures based on genome-wide data. To decompose the complex association patterns, we performed a multivariate partial least squares model of the genetic and imaging modalities.

      Results

      The univariate analyses showed that certain traits were related to brain structure (significant genetic correlations with total cortical surface area from rg= -0.101 for smoking initiation to rg= 0.230 for cognitive ability), while other traits were related to brain function (significant genetic correlations with functional connectivity from rg= -0.161 for educational attainment to rg= 0.318 for schizophrenia). The multivariate analysis showed that genetic predispositions to attention deficit hyperactivity disorder, smoking initiation, and cognitive traits have stronger associations with brain structure than with brain function, whereas genetic predispositions to most other psychiatric disorders have stronger associations with brain function than with brain structure.

      Conclusions

      These results reveal that genetic predispositions to mental health and cognitive traits have distinct brain profiles.

      Keywords

      Introduction

      Twin and family studies have revealed that mental health and cognitive traits are partly heritable (
      • Polderman T.J.C.
      • Benyamin B.
      • De Leeuw C.A.
      • Sullivan P.F.
      • Van Bochoven A.
      • Visscher P.M.
      • Posthuma D.
      Meta-analysis of the heritability of human traits based on fifty years of twin studies.
      ). Genome-wide association studies (GWAS) have identified many single-nucleotide polymorphisms (SNPs) across the whole genome associated with these traits(
      • Selzam S.
      • Ritchie S.J.
      • Pingault J.-B.
      • Reynolds C.A.
      • O’Reilly P.F.
      • Plomin R.
      Comparing within-and between-family polygenic score prediction.
      ). These complex traits are influenced by many genes that individually explain a very small part of the variance, as well as the environment. These genetic effects influence other underlying traits and may also affect which environments people get exposed to(
      • Abdellaoui A.
      • Verweij K.J.H.
      Dissecting polygenic signals from genome-wide association studies on human behaviour.
      ). The polygenic nature poses a challenge to understanding the genetic architecture and biological mechanisms underlying specific traits(
      • Krapohl E.
      • Patel H.
      • Newhouse S.
      • Curtis C.J.
      • Von Stumm S.
      • Dale P.S.
      • et al.
      Multi-polygenic score approach to trait prediction.
      ). A promising approach that we will explore here is the use of imaging genetics to examine brain-based intermediate phenotypes, also known as endophenotypes(
      • Liu B.
      • Zhang X.
      • Cui Y.
      • Qin W.
      • Tao Y.
      • Li J.
      • et al.
      Polygenic risk for schizophrenia influences cortical gyrification in 2 independent general populations.
      ,
      • Kauppi K.
      • Westlye L.T.
      • Tesli M.
      • Bettella F.
      • Brandt C.L.
      • Mattingsdal M.
      • et al.
      Polygenic risk for schizophrenia associated with working memory-related prefrontal brain activation in patients with schizophrenia and healthy controls.
      ). The effects of genes are not expressed directly at the level of behavior, but are more likely mediated by their molecular and cellular effects on brain development and information processing(
      • Rose E.J.
      • Donohoe G.
      Brain vs behavior: An effect size comparison of neuroimaging and cognitive studies of genetic risk for schizophrenia.
      ,
      • Mattay V.S.
      • Goldberg T.E.
      • Sambataro F.
      • Weinberger D.R.
      Neurobiology of cognitive aging: Insights from imaging genetics.
      ). The neural mechanisms through which genetic variants travel to influence one’s cognitive capacity and mental health remain largely unknown.
      The cerebral cortex is the outer layer of the brain that contains the neuronal cell bodies, also known as gray matter. It forms a folded sheet that plays a major role in multiple aspects of higher cognitive function. A recent study by Grasby et al. reported that variation in total cortical surface area (CSA) was genetically correlated with cognitive function, depression, and attention deficit hyperactivity disorder (ADHD), but not with many other psychiatric disorders(
      • Grasby K.L.
      • Jahanshad N.
      • Painter J.N.
      • Colodro-Conde L.
      • Bralten J.
      • Hibar D.P.
      • et al.
      The genetic architecture of the human cerebral cortex.
      ). Other studies also found significant genetic associations between total brain volume and cognitive traits including general cognitive function, educational outcomes, intelligence, and numerical reasoning(
      • Zhao B.
      • Luo T.
      • Li T.
      • Li Y.
      • Zhang J.
      • Shan Y.
      • et al.
      Genome-wide association analysis of 19,629 individuals identifies variants influencing regional brain volumes and refines their genetic co-architecture with cognitive and mental health traits.
      ,
      • Hofer E.
      • Roshchupkin G.V.
      • Adams H.H.H.
      • Knol M.J.
      • Lin H.
      • Li S.
      • et al.
      Genetic correlations and genome-wide associations of cortical structure in general population samples of 22,824 adults.
      ). For brain function, a recent study detected significant genetic correlations with many complex traits, such as education, cognitive performance, ADHD, major depressive disorder, schizophrenia, sleep, and neuroticism(
      • Zhao B.
      • Li T.
      • Smith S.M.
      • Xiong D.
      • Wang X.
      • Yang Y.
      • et al.
      Common variants contribute to intrinsic functional architecture of human brain.
      ). In addition, psychiatric disorders, especially schizophrenia and bipolar disorder, have been consistently found to be genetically associated with variation in brain function(
      • Cao H.
      • Zhou H.
      • Cannon T.D.
      Functional connectome-wide associations of schizophrenia polygenic risk.
      ,

      Dezhina Z, Ranlund S, Kyriakopoulos M, Williams SCR, Dima D (2018): A systematic review of associations between functional MRI activity and polygenic risk for schizophrenia and bipolar disorder. https://doi.org/10.1007/s11682-018-9879-z

      ,
      • Whalley H.C.
      • Papmeyer M.
      • Sprooten E.
      • Romaniuk L.
      • Blackwood D.H.
      • Glahn D.C.
      • et al.
      The influence of polygenic risk for bipolar disorder on neural activation assessed using fMRI.
      ), but inconsistent results regarding the association between polygenic risk for psychiatric disorders and brain structure were observed in previous studies(
      • Van Der Auwera S.
      • Wittfeld K.
      • Homuth G.
      • Teumer A.
      • Hegenscheid K.
      • Grabe H.J.
      No Association between Polygenic Risk for Schizophrenia and Brain Volume in the General Population.
      ,
      • Neilson E.
      • Shen X.
      • Cox S.R.
      • Clarke T.K.
      • Wigmore E.M.
      • Gibson J.
      • et al.
      Impact of Polygenic Risk for Schizophrenia on Cortical Structure in UK Biobank.
      ,
      • Ranlund S.
      • Rosa M.J.
      • de Jong S.
      • Cole J.H.
      • Kyriakopoulos M.
      • Fu C.H.Y.
      • et al.
      Associations between polygenic risk scores for four psychiatric illnesses and brain structure using multivariate pattern recognition.
      ). Besides psychiatric disorders and cognitive traits, substance use traits have also been reported to have shared genetic etiology with cortical brain morphology(
      • Rabinowitz J.A.
      • Campos A.I.
      • Ong J.S.
      • García-Marín L.M.
      • Alcauter S.
      • Mitchell B.L.
      • et al.
      Shared Genetic Etiology between Cortical Brain Morphology and Tobacco, Alcohol, and Cannabis Use.
      ).
      While studies combining genetic and neuroimaging data have made considerable progress in revealing neural correlates of various traits, there are still some limitations: 1, the sample sizes of previous studies have been relatively limited, resulting in insufficient statistical power; 2, genetic correlations have been used to estimate the magnitude of the overlap between brain measures and complex traits in genetic influences, but have been shown to be less powerful in detecting a genetic association between two traits than methods based on individual-level genotype data(
      • Ni G.
      • Moser G.
      • Ripke S.
      • Neale B.M.
      • Corvin A.
      • Walters J.T.R.
      • et al.
      Estimation of Genetic Correlation via Linkage Disequilibrium Score Regression and Genomic Restricted Maximum Likelihood.
      ,
      • Ni G.
      • Amare A.T.
      • Zhou X.
      • Mills N.
      • Gratten J.
      • Lee S.H.
      The genetic relationship between female reproductive traits and six psychiatric disorders.
      ); 3, and importantly, to our knowledge, no studies have combined both structural and functional neuroimaging information jointly with multiple mental health outcomes in multivariate analyses, leaving open whether certain predispositions are predominantly linked to brain structure or function.
      In the present study, we systematically investigated the genetic associations of multiple mental health and cognitive traits with brain structure and function. Specifically, we included seven mental health disorders (ADHD, Alzheimer’s disease, autism spectrum disorder (ASD), bipolar disorder, eating disorder, major depression, schizophrenia), five substance use traits (cigarettes per day, smoking initiation, smoking cessation, caffeine consumption, alcoholic drinks per week), and two cognitive traits (cognitive ability and educational attainment). The 14 traits were highly heritable and had large enough GWAS discovery sample(
      • Abdellaoui A.
      • Hugh-Jones D.
      • Yengo L.
      • Kemper K.E.
      • Nivard M.G.
      • Veul L.
      • et al.
      Genetic correlates of social stratification in Great Britain.
      ). Measures of brain structure and function were derived from T1-weighted and resting-state functional MRI. We used both PGS(
      • Purcell S.M.
      • Wray N.R.
      • Stone J.L.
      • Visscher P.M.
      • O’Donovan M.C.
      • Sullivan P.F.
      • et al.
      Common polygenic variation contributes to risk of schizophrenia and bipolar disorder.
      ) analyses and genome-wide genetic correlations based on GWAS summary statistics(
      • Bulik-Sullivan B.
      • Finucane H.K.
      • Anttila V.
      • Gusev A.
      • Day F.R.
      • Loh P.R.
      • et al.
      An atlas of genetic correlations across human diseases and traits.
      ). In detail, we calculated PGSs for 14 traits which represent the aggregated genetic effects across the genome, and assessed their associations with structural and functional neuroimaging measures. In addition, we also estimated genetic correlations of the mental health and cognitive traits with neuroimaging features based on genome-wide SNP data using linkage disequilibrium score (LDSC) regression. Finally, we modeled the joint genetic associations of the mental health and cognitive traits with brain structure and function, using a robust partial least squares (PLS) regression(
      • Lorenzi M.
      • Altmann A.
      • Gutman B.
      • Wray S.
      • Arber C.
      • Hibar D.P.
      • et al.
      Susceptibility of brain atrophy to TRIB3 in Alzheimer’s disease, evidence from functional prioritization in imaging genetics.
      ). To decompose the association patterns of various traits with brain structure and function, PLS regression is used to identify a limited number of latent variables that link the many genotype variables to the many imaging phenotypes.

      Methods

      Participants

      The data used in this study are from UK Biobank(
      • Bycroft C.
      • Freeman C.
      • Petkova D.
      • Band G.
      • Elliott L.T.
      • Sharp K.
      • et al.
      The UK Biobank resource with deep phenotyping and genomic data.
      ). We used the latest available imaging release (on January 2020) including 36,799 unrelated European ancestry participants from across the United Kingdom with both genetic and imaging data (19,512 females and 17,287 males, mean age = 54.90 years, SD = 7.43 years, range from 40 to 70 years). The National Health Service North West Centre for Research Ethics Committee provided UK Biobank project with ethical approval (reference: 11/NW/0382) (http://biobank.ctsu.ox.ac.uk/crystal/field.cgi?id=200).

      Imaging procedures

      Detailed information about the image acquisition protocols and processing pipeline is provided at: http://biobank.ctsu.ox.ac.uk/crystal/crystal/docs/brain_mri.pdf. For brain structure, there were three global measures (total cortical surface area (CSA), intra-cranial volume (ICV) and average cortical thickness (CT)). Apart from them, we also focused on the regional measures including the volumes of 14 subcortical regions, 66 surface area of parcels identified on the cortical surface, and cortical thickness within these areas. The cortical parcellations were based on the Desikan-Killiany (DK) atlas(
      • Desikan R.S.
      • Ségonne F.
      • Fischl B.
      • Quinn B.T.
      • Dickerson B.C.
      • Blacker D.
      • et al.
      An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest.
      ). For brain function, there were amplitude within 21 functional networks and functional connectivities (FCs) between these networks derived from rsfMRI. Additional details on structural and functional measures are reported in the Supplementary Information (SI).

      Polygenic scores (PGSs)

      We conducted the quality control procedures on genotype data (SI Methods). PGSs were then calculated by summing the number of risk alleles an individual carries weighted by the strength of the association of each allele with a specific trait. The strength of association for each SNP is based on the effect size estimate from existing GWAS summary statistics. We used the summary-data based best linear unbiased prediction (SBLUP) approach(
      • Robinson M.R.
      • Kleinman A.
      • Graff M.
      • Vinkhuyzen A.A.E.
      • Couper D.
      • Miller M.B.
      • et al.
      Genetic evidence of assortative mating in humans.
      ) to create PGSs (SI Methods and Table S1). The scores were computed on a set of 1,312,100 autosomal HapMap 3 SNPs with a minor allele count of >5, an info score of >0.3, Hardy–Weinberg equilibrium (HWE) P-value (PHWE) < 10−6 and missingness < 0.05.

      Genetic relationships analysis

      We used the SciPy Stats package in python to calculate the Pearson correlation coefficients of PGSs for the 14 traits with global brain measures, the volumes of 14 subcortical regions as well as CSA and CT for the 66 DK atlas regions, and brain functional measures which includes amplitude of 21 functional networks and 210 FCs (Figure S1). We regressed out the covariates of the 14 PGSs and all brain measures by the estimation of residuals (Table S2). There were age, sex, age*sex, age2, age2*sex, head positions, the top 25 genetic principal components (PCs), rsfMRI head motions, ICV, total CSA, and average CT, which were consistent with the main covariates used in the GWASs by Smith(
      • Smith S.M.
      • Douaud G.
      • Chen W.
      • Hanayik T.
      • Alfaro-Almagro F.
      • Sharp K.
      • Elliott L.T.
      An expanded set of genome-wide association studies of brain imaging phenotypes in UK Biobank.
      ) and Elliott(
      • Elliott L.T.
      • Sharp K.
      • Alfaro-Almagro F.
      • Shi S.
      • Miller K.L.
      • Douaud G.
      • et al.
      Genome-wide association studies of brain imaging phenotypes in UK Biobank.
      ). We applied false discovery rate (FDR) multiple testing correction with a significance threshold of 0.0167 (0.05/3) to account for the above three sets of analyses: global structural measures, regional structural measures, or functional measures.
      In addition, we conducted the GWASs for brain imaging measures in the UK Biobank sample, and used LDSC regression(
      • Bulik-Sullivan B.
      • Finucane H.K.
      • Anttila V.
      • Gusev A.
      • Day F.R.
      • Loh P.R.
      • et al.
      An atlas of genetic correlations across human diseases and traits.
      ) to calculate their genetic correlations with mental health disorders, substance use, or cognitive traits (SI Methods). Covariates like those applied in the Pearson correlation analysis were regressed out.

      Joint partial least squares modeling analysis

      To test for differences in relationships of 14 mental health and cognitive traits with brain structure and function, we applied joint partial least squares (PLS) modeling of the genetic and imaging modalities, which find the multidimensional direction in the genetic space that explains the maximum multidimensional variance direction in the imaging space(
      • Wold H.
      Estimation of principal components and related models by iterative least squares.
      ). The detailed procedures of PLS method are represented in Figure S2. We performed a canonical symmetric version of the PLS regression from the scikit-learn package in python(
      • Pedregosa F.
      • Varoquaux G.
      • Gramfort A.
      • Michel V.
      • Thirion B.
      • Grisel O.
      • et al.
      Scikit-learn: Machine learning in Python.
      ) to find linear relations between two multivariate datasets (Figure 1A). In short, the input features included PGSs for the 14 mental health and cognitive traits as one set of variables and 377 structural and functional measures as the other. PLS modeling produces a weight for each input feature that represents its relative importance for describing the global joint multimodal relationship(
      • Lorenzi M.
      • Altmann A.
      • Gutman B.
      • Wray S.
      • Arber C.
      • Hibar D.P.
      • et al.
      Susceptibility of brain atrophy to TRIB3 in Alzheimer’s disease, evidence from functional prioritization in imaging genetics.
      ). Comparing the weights helps identify PGSs that are linked to the patterns of brain structure or function. Age, sex, age*sex, age2, age2*sex, head positions, and the top 25 genetic PCs were added as covariates.
      Figure thumbnail gr1
      Figure 1Flow chart of data-driven multivariate statistical analysis. A, input features of PLS regression. wi indicated the weight of i-th input feature, which was served as the importance of relative importance for describing the multimodal relationship. Ti and Ui respectively indicated the i-th genotype and phenotype PLS component transformed from genetic and imaging data. B, the dataset was randomly divided into two groups for validation in which PLS analyses were performed independently. This procedure was repeated 5000 times. The averaged results from the 5000 iterations identified the robust and consistently selected PGSs and brain imaging measures that received the largest weights in both groups for each iteration. C, for each repetition, the absolute weights of 14 PGSs were sorted, and the top 5 PGSs were selected in two random groups. The selected PGSs were labeled one or zero otherwise. The resulting binary arrays independently estimated in the two groups were then merged, such that only replicated PGSs receiving the largest PLS weights were selected. D, for each repetition, the absolute weights of brain measures were sorted, and the top 10% brain measures are selected in two random groups. Similarly, only replicated brain measures receiving the largest weights were selected.
      A robust approach for the stable estimation and interpretation of PLS weights was applied in this study (Figure 1B-D), following a previously described procedure(
      • Lorenzi M.
      • Altmann A.
      • Gutman B.
      • Wray S.
      • Arber C.
      • Hibar D.P.
      • et al.
      Susceptibility of brain atrophy to TRIB3 in Alzheimer’s disease, evidence from functional prioritization in imaging genetics.
      ). Briefly, a stability selection procedure was conducted to assess the reproducibility and robustness of the PLS parameters. The whole sample was randomly split in half and the PLS model was run on each sampled subgroup independently. Through 5000 repetitions of this procedure, we estimated a confidence measure for each input feature ranging between 0.0 and 1.0, indicating its probability containing highly reproducible PLS weights. Therefore, the measure can be regarded as the importance of input feature. Similar to the study by Lorenzi and Altmann(
      • Lorenzi M.
      • Altmann A.
      • Gutman B.
      • Wray S.
      • Arber C.
      • Hibar D.P.
      • et al.
      Susceptibility of brain atrophy to TRIB3 in Alzheimer’s disease, evidence from functional prioritization in imaging genetics.
      ), the number of components was also initially set as five. In addition, we used the permutation test (n = 1000) to assess the significance of the estimated selection probability of PGSs and neuroimaging measures through randomly shuffling the PGSs of individuals to conduct the PLS model for 1000 times.

      Results

      Genetic relationships with global brain structure

      The PGSs showed many significant associations with total CSA and ICV, but none with average CT (Table 1, Table S3). Higher polygenic risks for ADHD, Alzheimer’s disease, major depression, cigarettes per day, and smoking initiation were strongly associated with lower total CSA. Conversely, higher PGSs for smoking cessation, alcoholic drinks per week, cognitive ability, and educational attainment were strongly associated with higher total CSA. Similarly, higher polygenic risks for ADHD, smoking initiation, and Alzheimer’s disease were significantly associated with lower ICV, whereas higher PGSs for cognitive ability, educational attainment, and eating disorder were significantly associated with higher ICV. Polygenic risks for ASD, bipolar disorder, schizophrenia and caffeine consumption were not significantly associated with any global measure of brain structure.
      Table 1Correlations of 14 PGSs for mental health and cognitive traits with three global structural brain measures
      TraitsTotal cortical surface area (CSA)Average cortical thickness (CT)Intra-cranial volume (ICV)
      rppFDRrppFDRrppFDR
      ADHD-0.05871.68E-292.36E-280.00260.62440.7285-0.0543.67E-253.85E-24
      Alzheimer’s disease-0.0222.41E-051.01E-040.0060.2470.3705-0.02346.92E-063.23E-05
      ASD0.00350.4980.6750.00270.59820.71780.0030.5620.703
      Bipolar disorder0.01369.33E-032.24E-02-0.00640.21730.3510.00580.2630.381
      Eating disorder0.00810.1190.20.00610.24230.37050.01484.50E-031.26E-02
      Major depression-0.01961.74E-046.08E-040.00140.79090.8727-0.01251.69E-023.55E-02
      Schizophrenia-0.00927.76E-020.136-0.010.05620.1027-0.01025.12E-029.78E-02
      Cigarettes per day-0.01834.40E-041.32E-03-0.0050.33650.4711-0.01359.62E-032.24E-02
      Smoking initiation-0.03784.22E-133.55E-12-0.00130.81040.8727-0.0312.75E-091.92E-08
      Smoking cessation0.02173.07E-051.17E-040.00030.96150.99660.0147.20E-031.89E-02
      Caffeine consumption-0.0030.569240.703-0.00310.54740.70320.00220.6760.768
      Alcoholic drinks per week0.01912.48E-048.02E-04-0.01320.01150.02540.01231.79E-023.58E-02
      Cognitive ability0.02892.97E-081.56E-07-0.00020.97290.99660.02951.51E-089.07E-08
      Educational attainment0.08355.43E-581.14E-56<0.00010.99770.99770.08435.31E-592.23E-57
      r = Pearson’s correlation, p = p-value; pFDR = p-value after FDR multiple correction. Bold values indicate significant associations after FDR multiple correction testing (pFDR < 0.0167).
      LDSC regression analysis showed the highest SNP-based heritability estimates for total CSA (37.8%) and ICV (35.4%) (Table S4, Figure S3). Moreover, the results of the genetic correlation analysis for global structural measures were largely in line with those of the PGS analysis (Table S5). Total CSA and ICV showed highly significant genetic correlations with ADHD (rg = -0.170 and rg = -0.159), cognitive ability (rg = 0.230 and rg = 0.228) and educational attainment (rg = 0.211 and rg = 0.159), but no significant correlations were observed with average CT. Most psychiatric disorders (ASD, bipolar disorder, eating disorder, schizophrenia) were not genetically correlated with measures of global brain structure.

      Genetic relationships with regional brain structure

      We found six significant associations for subcortical volumes controlling for ICV (Figure S4, Table S6). Notably, volume of the thalamus was negatively associated with genetic vulnerability to schizophrenia and positively with genetic predisposition to educational attainment. In addition, the PGS for major depression was positively associated with volume of the left caudate nucleus, the PGS for Alzheimer’s disease was negatively associated with volume of the left putamen, and the PGS for cigarettes per day showed negative associations with volume of the right putamen. As for genetic correlation analysis, we only found a significant correlation between volume of the thalamus and schizophrenia (rg = -0.135) (Figure S5, Table S7).
      For CSA of brain regions, a total of 27 significant associations were found with the 14 PGSs controlling for total CSA, which were mainly identified on educational attainment, smoking initiation and cognitive ability (Figure 2A-B, Table S8, SI Results). The genetic correlation analysis found largely comparable results though with fewer significant associations, with significant genetic correlations of cognitive ability and educational attainment with CSA of the pars orbitalis of the inferior frontal gyrus (rg = 0.147 and rg = 0.170) (Figure S6, Table S9). Although the global measure of CT did not show any significant associations with PGSs, we found 27 significant associations with regional CT controlling for average CT (Figure 2C-D, Table S10, SI Results). Similarly, the genetic correlation analysis showed a negative correlation of educational attainment with CT of the inferior parietal cortex and two positive correlations with CT of the superior temporal regions. (Figure S7, Table S11).
      Figure thumbnail gr2
      Figure 2Associations between the 14 PGSs and the CSA and CT of 66 cortical regions, corrected for total CSA and average CT, respectively. The horizontal axis shows the different brain regions and the vertical axis shows the various traits. The size and color of each box indicate the magnitude of the association. Blue squares indicate positive associations, and red squares indicate negative associations. A, the PGS-CSA correlation map across the left hemisphere. B, the PGS-CSA correlation map across the right hemisphere. C, the PGS-CT correlation map across the left hemisphere. D, the PGS-CT correlation map across the right hemisphere.
      Associations without correction for global brain structure are presented in the supplementary materials (Figures S8-10, Tables S12-14). Those results showed considerable overlap with those for global brain structure.

      Genetic relationships with brain function

      We examined the associations between the 14 PGSs and these 231 functional measures. We found 22 significant associations between PGSs and amplitude within 21 networks (Figure 3, Figure S11, Table S15, SI Results). Most of them were identified on bipolar disorder, schizophrenia, major depression, and educational attainment. Although the pattern of results was comparable, the genetic correlation analysis showed no significant associations between network amplitudes and the 14 mental health and cognitive traits (Figure S12, Table S16). In addition, a total of 133 significant associations between PGSs and FCs were observed (Figure 3, Table S17, SI Results), which were also mainly identified on educational attainment, schizophrenia, major depression and ASD. The genetic correlation analysis only showed a few significant genetic correlations between FCs and schizophrenia, cognitive ability, and educational attainment (Figure S13, Table S18).
      Figure thumbnail gr3
      Figure 3Significant associations between the 14 PGSs and functional measures. The nodes in each circle show the 21 resting-state networks. Asterisks indicate significant associations with network amplitude (temporal fluctuation) of the 21 networks after FDR multiple testing correction. The lines between the nodes indicate significant FCs between the 21 functional networks after FDR multiple testing correction. The color of each line and node indicates the magnitude of the associations. Blue squares indicate positive associations, and red squares indicate negative associations.

      Multivariate modeling of joint relationships between PGSs and brain measures

      The univariate analyses reported above showed complex genetic relationships of multiple traits with brain structure and function. In order to decompose the association patterns of various traits with brain structure and function, we performed a data-driven multivariate statistical analysis of the genetics and imaging data with split-half cross-validation on basis of partial least squares (PLS) regression(
      • Lorenzi M.
      • Altmann A.
      • Gutman B.
      • Wray S.
      • Arber C.
      • Hibar D.P.
      • et al.
      Susceptibility of brain atrophy to TRIB3 in Alzheimer’s disease, evidence from functional prioritization in imaging genetics.
      ) (Figure 1). We focused on the first five components which ensured that we can get stable and robust latent variables(
      • Lorenzi M.
      • Altmann A.
      • Gutman B.
      • Wray S.
      • Arber C.
      • Hibar D.P.
      • et al.
      Susceptibility of brain atrophy to TRIB3 in Alzheimer’s disease, evidence from functional prioritization in imaging genetics.
      ). This method with inherent replication identifies multivariate components that link multiple traits with multiple brain imaging variables.
      Figure 4Aand 4B respectively show the mean weights for the first PLS genotype and phenotype component, which reflect the importance of each input variable for the PGSs-imaging relationships in the PLS model. Figure 4C and 4D show the probabilities for PGSs and imaging measures receiving the largest PLS weights in two random groups across 5000 repetitions. PGSs for ADHD, educational attainment, smoking initiation, and cognitive ability had higher weights and were consistently selected for the first genotype component (selection probability > 50%). The corresponding imaging component was mainly associated with larger CSA of many cortical regions at the threshold of 50% selection probability, especially the frontal and temporal regions, as well as larger volume of the thalamus, whereas regional CT and functional measures had low weights and were not consistently selected. Similarly, Figure 4E-H show the weights and probabilities for the second PLS genotype and phenotype component. The genetic patterns of PGSs for psychiatric disorders (schizophrenia, major depression, bipolar disorder, and ASD) corresponded consistently to lower amplitude of 5 functional networks and to 10 both stronger and weaker FCs at the threshold of 50% selection probability. All structural measures had low weights and were not consistently selected. The selected nodes (resting-state networks) included the visual system (node 2 and 4), somatosensory system (node 10), cerebellum (node 15), and basal ganglia (node 18). FCs included 5 positive and 5 negative associations across the entire brain, including networks often associated with psychiatric disorders such as the default mode network, salience network, and ventral and dorsal attention networks. Through the permutation test (n = 1000) on selection probabilities, the selected PGSs and brain measures in the first two components were significantly different from chance (Figure S14 and Table S19).
      Figure thumbnail gr4
      Figure 4First two components of the PLS model. A, the mean weights across 5000 repetitions for 14 PGSs in the first genotype component. B, the mean weights across 5000 repetitions for the CSA of cortical regions and volumes of subcortical regions in the first phenotype component. C, probabilities of the 14 PGSs of being selected across 5000 repetitions in the first genotype component. D, probabilities of regional CSA and subcortical volumes of being selected across 5000 repetitions in the first phenotype component. E, the mean weights across 5000 repetitions for 14 PGSs in the second genotype component. F, the mean weights across 5000 repetitions for amplitude of 21 functional networks and 210 FCs between these networks in the second phenotype component. G, probabilities of the 14 PGSs of being selected across 5000 repetitions in the second genotype component. H, probabilities of amplitude of 21 functional networks and 210 FCs between these networks of being selected across 5000 repetitions in the second phenotype component. In the bar plots, orange bars indicate PGSs with negative PLS weights, and green bars indicate PGSs with positive PLS weights.
      The first two components explained most of the total variance, 52.4% and 11% respectively, and the PGSs and brain measures could be stably and consistently selected across 5000 repetitions compared to other components, and were stable across a range of thresholds (SI Results).

      Discussion

      We systematically investigated the genetic relationships of the 14 mental health and cognitive traits with 380 structural and functional brain measures using neuroimaging and genetic data from 36,799 participants. The results show that genetic predispositions to ADHD, educational attainment, smoking initiation, and cognitive ability are more strongly associated with brain structure than with brain function, whereas genetic predispositions to most psychiatric disorders (schizophrenia, major depression, bipolar disorder, and ASD) are more strongly associated with brain function than with brain structure. To our knowledge, this is the first study providing evidence for these differential associations, and suggests that there are distinct neural pathways through which genes eventually influence different mental health and cognition traits.
      The univariate correlation analyses showed various and complex associations between the14 PGSs and 380 brain measures. We also estimated the SNP-based genetic correlations between these traits and neuroimaging features and found generally consistent results, although fewer associations were significant due to lower power of these analyses. In order to decompose the joint genetic association patterns into underlying components, we performed a data-driven multivariate statistical analysis (PLS regression) of the genetics and imaging data. The first PLS component confirmed that ADHD was more strongly associated with brain structure than with brain function. In fact, previous studies consistently reported significant genetic sharing between ADHD and total CSA and ICV(
      • Grasby K.L.
      • Jahanshad N.
      • Painter J.N.
      • Colodro-Conde L.
      • Bralten J.
      • Hibar D.P.
      • et al.
      The genetic architecture of the human cerebral cortex.
      ,
      • Klein M.
      • Walters R.K.
      • Demontis D.
      • Stein J.L.
      • Hibar D.P.
      • Adams H.H.
      • et al.
      Genetic markers of ADHD-related variations in intracranial volume.
      ,
      • Mooney M.A.
      • Bhatt P.
      • Hermosillo R.J.M.
      • Ryabinin P.
      • Nikolas M.
      • Faraone S.V.
      • et al.
      Smaller total brain volume but not subcortical structure volume related to common genetic risk for ADHD.
      ), whereas only a few FCs were significantly associated with the PGS for ADHD(
      • Hermosillo R.J.M.
      • Mooney M.A.
      • Fezcko E.
      • Earl E.
      • Marr M.
      • Sturgeon D.
      • et al.
      Polygenic Risk Score–Derived Subcortical Connectivity Mediates Attention-Deficit/Hyperactivity Disorder Diagnosis.
      ). Epidemiologic and clinical studies have found that genetic risk factors that affect the structure and functional capacity of brain networks are involved in the etiology of ADHD(
      • Demontis D.
      • Walters R.K.
      • Martin J.
      • Mattheisen M.
      • Als T.D.
      • Agerbo E.
      • et al.
      Discovery of the first genome-wide significant risk loci for attention deficit/hyperactivity disorder.
      ,
      • Faraone S.V.
      • Asherson P.
      • Banaschewski T.
      • Biederman J.
      • Buitelaar J.K.
      • Ramos-Quiroga J.A.
      • et al.
      Attention-deficit/hyperactivity disorder.
      ). Many genes that have been associated with ADHD, such as FOXP2, SORCS3, and DUSP6, are highly expressed in the brain and implicated in neurodevelopmental processes(
      • Demontis D.
      • Walters R.K.
      • Martin J.
      • Mattheisen M.
      • Als T.D.
      • Agerbo E.
      • et al.
      Discovery of the first genome-wide significant risk loci for attention deficit/hyperactivity disorder.
      ).
      Comparable results were obtained for smoking initiation, which showed a similar association pattern with the brain measures as ADHD. Although the molecular mechanisms underlying these associations have yet to be elucidated, a previous gene set analysis of enriched biological pathways provided some possible clues that brain dopamine receptor signaling is associated with smoking initiation(
      • Wang J.
      • Li M.D.
      Common and unique biological pathways associated with smoking initiation/progression, nicotine dependence, and smoking cessation.
      ). For example, BDNF regulates synaptic plasticity and survival of cholinergic and dopaminergic neurons(
      • Zhang L.I.
      • Poo M.M.
      Electrical activity and development of neural circuits.
      ). Genetic variants within the BDNF gene were found to be genome-wide significantly associated with smoking initiation(
      • Furberg H.
      • Kim Y.
      • Dackor J.
      • Boerwinkle E.
      • Franceschini N.
      • Ardissino D.
      • et al.
      Genome-wide meta-analyses identify multiple loci associated with smoking behavior.
      ) and the gene was highly expressed in the prefrontal cortex, which might be implicated in the cognitive-enhancing effects of nicotine(
      • Levin E.D.
      • McClernon F.J.
      • Rezvani A.H.
      Nicotinic effects on cognitive function: Behavioral characterization, pharmacological specification, and anatomic localization.
      ). This is in line with our finding that the PGS for smoking initiation was especially associated with the CSA of several frontal regions.
      In addition, genetic predispositions to cognitive traits were significantly associated with both brain structure and function, though the first PLS component showed that the association with brain structure was stronger. Our results for cognitive traits are in line with previous studies in which the PGS for educational attainment was associated with larger brains(
      • Elliott M.L.
      • Belsky D.W.
      • Anderson K.
      • Corcoran D.L.
      • Ge T.
      • Knodt A.
      • et al.
      A Polygenic Score for Higher Educational Attainment is Associated with Larger Brains.
      ,
      • Mitchell B.L.
      • Cuéllar-Partida G.
      • Grasby K.L.
      • Campos A.I.
      • Strike L.T.
      • Hwang L.D.
      • et al.
      Educational attainment polygenic scores are associated with cortical total surface area and regions important for language and memory.
      ) and CSA and CT in some frontal and temporal regions mediated the association between the PGS for intelligence and the g-factor(
      • Lett T.A.
      • Vogel B.O.
      • Ripke S.
      • Wackerhagen C.
      • Erk S.
      • Awasthi S.
      • et al.
      Cortical Surfaces Mediate the Relationship Between Polygenic Scores for Intelligence and General Intelligence.
      ), but no large-scale study provided credible and sufficient evidence on relationships of PGSs for cognitive traits with brain function. Previous studies have presented some possible explanations for the associations, describing that biological annotation of genes associated with educational attainment were involved in brain-development processes and neuron-neuron communication(
      • Lee J.J.
      • Wedow R.
      • Okbay A.
      • Kong E.
      • Maghzian O.
      • Zacher M.
      • et al.
      Gene discovery and polygenic prediction from a genome-wide association study of educational attainment in 1.1 million individuals.
      ,
      • Okbay A.
      • Beauchamp J.P.
      • Fontana M.A.
      • Lee J.J.
      • Pers T.H.
      • Rietveld C.A.
      • et al.
      Genome-wide association study identifies 74 loci associated with educational attainment.
      ).
      The second PLS component shows that PGSs for most psychiatric disorders (schizophrenia, major depression, bipolar disorder, and ASD) were more strongly associated with brain function than with brain structure. Previous studies for the genetic relationships between psychiatric disorders and global brain structure produced mixed results, likely because of limited sample sizes. For example, some studies reported that a decrease in total brain volume or cortical thickness was significantly associated with an increased polygenic risk for schizophrenia(
      • Neilson E.
      • Shen X.
      • Cox S.R.
      • Clarke T.K.
      • Wigmore E.M.
      • Gibson J.
      • et al.
      Impact of Polygenic Risk for Schizophrenia on Cortical Structure in UK Biobank.
      ,
      • Terwisscha Van Scheltinga A.F.
      • Bakker S.C.
      • Van Haren N.E.M.
      • Derks E.M.
      • Buizer-Voskamp J.E.
      • Boos H.B.M.
      • et al.
      Genetic schizophrenia risk variants jointly modulate total brain and white matter volume.
      ), whereas another study with a relatively large sample found no significant associations(
      • Van Der Auwera S.
      • Wittfeld K.
      • Homuth G.
      • Teumer A.
      • Hegenscheid K.
      • Grabe H.J.
      No Association between Polygenic Risk for Schizophrenia and Brain Volume in the General Population.
      ). With respect to brain function, previous studies indicated that PGSs for ASD, bipolar disorder, major depression, and schizophrenia were associated with functional measures(
      • Kauppi K.
      • Westlye L.T.
      • Tesli M.
      • Bettella F.
      • Brandt C.L.
      • Mattingsdal M.
      • et al.
      Polygenic risk for schizophrenia associated with working memory-related prefrontal brain activation in patients with schizophrenia and healthy controls.
      ,
      • Wang T.
      • Zhang X.
      • Li A.
      • Zhu M.
      • Liu S.
      • Qin W.
      • et al.
      Polygenic risk for five psychiatric disorders and cross-disorder and disorder-specific neural connectivity in two independent populations.
      ,
      • Liu S.
      • Li A.
      • Liu Y.
      • Yan H.
      • Wang M.
      • Sun Y.
      • et al.
      Polygenic effects of schizophrenia on hippocampal grey matter volume and hippocampus-medial prefrontal cortex functional connectivity.
      ). Overall, although our results do support a limited link between PGSs for psychiatric disorders and brain structure, results from our multivariate analysis highlights that the association with brain function is stronger when both are considered together. The stronger associations between PGSs and brain function may indicate that genetic predispositions to psychiatric disorders are likely to cause difficulties in information integration across the brain, which in turn makes individuals more prone to developing a psychiatric disorder(
      • Cao H.
      • Zhou H.
      • Cannon T.D.
      Functional connectome-wide associations of schizophrenia polygenic risk.
      ).
      Several interesting genetic relationships with different brain measures were identified in this study. For example, higher PGS for schizophrenia was significantly associated with lower CT of orbitofrontal regions. We also found that PGSs for ASD, bipolar disorder, and major depression showed positive associations with CSA of cingulate regions. The orbitofrontal cortex and cingulate cortex are involved in decision making, emotion regulation, response to reward, learning and memory(
      • Mızrak E.
      • Bouffard N.R.
      • Libby L.A.
      • Boorman E.D.
      • Ranganath C.
      The hippocampus and orbitofrontal cortex jointly represent task structure during memory-guided decision making.
      ,
      • Rudebeck P.H.
      • Rich E.L.
      Orbitofrontal cortex.
      ,
      • van Heukelum S.
      • Mars R.B.
      • Guthrie M.
      • Buitelaar J.K.
      • Beckmann C.F.
      • Tiesinga P.H.E.
      • et al.
      Where is Cingulate Cortex? A Cross-Species View.
      ). For substance use traits, we found that smoking and drinking behaviors were genetically associated with prefrontal brain structure, which are involved in self-control and risk-taking (
      • Morawetz C.
      • Mohr P.N.C.
      • Heekeren H.R.
      • Bode S.
      The effect of emotion regulation on risk-taking and decision-related activity in prefrontal cortex.
      ).
      The heritability estimates (see Figure S3) differed for the various brain measures. It was highest for total CSA and ICV and somewhat lower for average CT, which may have contributed to the significant associations of PGSs with total CSA and ICV but not average CT. Also the regional measures showed differences in heritability, but there was no apparent relationship between these heritability estimates and the likelihood for significant associations of PGSs. Instead, the distinct PLS components suggest that different PGSs are related to distinct brain measures.
      We note several limitations to our study. First, the predictive power of PGSs and the significance of the genetic correlations (LDSC regression) depend on the sample size of the discovery GWAS summary statistics. Thus, the statistical power differs substantially between different traits, which may influence the comparisons of association patterns. Second, our results point to genetic associations between different traits and brain measures, but the nature of the association is not yet known. We cannot distinguish genetic pleiotropy versus causal relationships from mental health and cognitive traits to brain structure and function or vice versa. The general assumption is that the brain is the mediator through which genes exert their effects on behavior. But it is also possible that genetic variants simultaneously influence brain measures and behaviour (i.e. pleiotropy) and there are also genetic influences on the environment and continuous interactions between the brain, behavior and the environment. The relation between genes and the brain may therefore also be influenced by genetic effects on the environment that subsequently shape the brain(
      • Sprooten E.
      • Franke B.
      • Greven C.U.
      The P-factor and its genomic and neural equivalents: an integrated perspective.
      ,
      • Avinun R.
      The E Is in the G: Gene–Environment–Trait Correlations and Findings From Genome-Wide Association Studies.
      ). Third, the polygenic scores may contain some environmental effects, because genetic effects may possibly act via the environment to influence the brain. These environmental effects can include the influence of parents who provide the offspring with their genome as well as with their rearing environment(
      • Kong A.
      • Thorleifsson G.
      • Frigge M.L.
      • Vilhjalmsson B.J.
      • Young A.I.
      • Thorgeirsson T.E.
      • et al.
      The nature of nurture: Effects of parental genotypes.
      ), which can result in an inflation of estimated genetic effects. Gene-environment correlations can also occur on a regional level: the educational attainment polygenic score, for example, has been shown to influence where people live (individuals with a higher polygenic score are more likely to migrate to a richer neighborhood), which may have an effect on mental health and cognitive outcomes(
      • Abdellaoui A.
      • Hugh-Jones D.
      • Yengo L.
      • Kemper K.E.
      • Nivard M.G.
      • Veul L.
      • et al.
      Genetic correlates of social stratification in Great Britain.
      ). Fourth, all participants in UK biobank are 40+ years old and most of them are healthy, so whether our findings can be generalized to younger population as well as individuals with mental disorders needs further verification. Fifth, the functional measures were derived from resting-state fMRI, and the genetic associations may be different for task-based fMRI that can probe specific brain functions.
      In summary, this study provides evidence for distinct association patterns of genetic predispositions to mental health and cognition with brain structure and function, with genetic predispositions to ADHD, smoking initiation, and cognitive traits having stronger associations with brain structure than with brain function, whereas genetic predispositions to most other psychiatric disorders having stronger associations with brain function than with brain structure. These results bring us closer to understanding the neurobiological basis of these traits.

      Acknowledgments and Disclosures

      This research has been conducted using data from UK Biobank under application numbers 40310 and 30091. S.L.is supported by the China Scholarship Council. A.A. & K.J.H.V. are supported by the Foundation Volksbond Rotterdam.
      A previous version of this article was published as a preprint on medRxiv: https://www.medrxiv.org/content/10.1101/2021.03.07.21252728v1.
      The authors declare no biomedical financial interests or potential conflicts of interest.

      Supplementary Material

      References

        • Polderman T.J.C.
        • Benyamin B.
        • De Leeuw C.A.
        • Sullivan P.F.
        • Van Bochoven A.
        • Visscher P.M.
        • Posthuma D.
        Meta-analysis of the heritability of human traits based on fifty years of twin studies.
        Nat Genet. 2015; 47: 702-709
        • Selzam S.
        • Ritchie S.J.
        • Pingault J.-B.
        • Reynolds C.A.
        • O’Reilly P.F.
        • Plomin R.
        Comparing within-and between-family polygenic score prediction.
        Am J Hum Genet. 2019; 105: 351-363
        • Abdellaoui A.
        • Verweij K.J.H.
        Dissecting polygenic signals from genome-wide association studies on human behaviour.
        Nature Human Behaviour. 2021; https://doi.org/10.1038/s41562-021-01110-y
        • Krapohl E.
        • Patel H.
        • Newhouse S.
        • Curtis C.J.
        • Von Stumm S.
        • Dale P.S.
        • et al.
        Multi-polygenic score approach to trait prediction.
        Mol Psychiatry. 2018; 23: 1368-1374
        • Liu B.
        • Zhang X.
        • Cui Y.
        • Qin W.
        • Tao Y.
        • Li J.
        • et al.
        Polygenic risk for schizophrenia influences cortical gyrification in 2 independent general populations.
        Schizophr Bull. 2017; 43: 673-680
        • Kauppi K.
        • Westlye L.T.
        • Tesli M.
        • Bettella F.
        • Brandt C.L.
        • Mattingsdal M.
        • et al.
        Polygenic risk for schizophrenia associated with working memory-related prefrontal brain activation in patients with schizophrenia and healthy controls.
        Schizophr Bull. 2015; 41: 736-743
        • Rose E.J.
        • Donohoe G.
        Brain vs behavior: An effect size comparison of neuroimaging and cognitive studies of genetic risk for schizophrenia.
        Schizophr Bull. 2013; 39: 518-526
        • Mattay V.S.
        • Goldberg T.E.
        • Sambataro F.
        • Weinberger D.R.
        Neurobiology of cognitive aging: Insights from imaging genetics.
        Biol Psychol. 2008; 79: 9-22
        • Grasby K.L.
        • Jahanshad N.
        • Painter J.N.
        • Colodro-Conde L.
        • Bralten J.
        • Hibar D.P.
        • et al.
        The genetic architecture of the human cerebral cortex.
        Science. 2020; 80https://doi.org/10.1126/science.aay6690
        • Zhao B.
        • Luo T.
        • Li T.
        • Li Y.
        • Zhang J.
        • Shan Y.
        • et al.
        Genome-wide association analysis of 19,629 individuals identifies variants influencing regional brain volumes and refines their genetic co-architecture with cognitive and mental health traits.
        Nat Genet. 2019; https://doi.org/10.1038/s41588-019-0516-6
        • Hofer E.
        • Roshchupkin G.V.
        • Adams H.H.H.
        • Knol M.J.
        • Lin H.
        • Li S.
        • et al.
        Genetic correlations and genome-wide associations of cortical structure in general population samples of 22,824 adults.
        Nat Commun. 2020; https://doi.org/10.1038/s41467-020-18367-y
        • Zhao B.
        • Li T.
        • Smith S.M.
        • Xiong D.
        • Wang X.
        • Yang Y.
        • et al.
        Common variants contribute to intrinsic functional architecture of human brain.
        BioRxiv. 2020; https://doi.org/10.1101/2020.07.30.229914
        • Cao H.
        • Zhou H.
        • Cannon T.D.
        Functional connectome-wide associations of schizophrenia polygenic risk.
        Mol Psychiatry. 2020; https://doi.org/10.1038/s41380-020-0699-3
      1. Dezhina Z, Ranlund S, Kyriakopoulos M, Williams SCR, Dima D (2018): A systematic review of associations between functional MRI activity and polygenic risk for schizophrenia and bipolar disorder. https://doi.org/10.1007/s11682-018-9879-z

        • Whalley H.C.
        • Papmeyer M.
        • Sprooten E.
        • Romaniuk L.
        • Blackwood D.H.
        • Glahn D.C.
        • et al.
        The influence of polygenic risk for bipolar disorder on neural activation assessed using fMRI.
        Transl Psychiatry. 2012; https://doi.org/10.1038/tp.2012.60
        • Van Der Auwera S.
        • Wittfeld K.
        • Homuth G.
        • Teumer A.
        • Hegenscheid K.
        • Grabe H.J.
        No Association between Polygenic Risk for Schizophrenia and Brain Volume in the General Population.
        Biol Psychiatry. 2015; 78: e41-e42
        • Neilson E.
        • Shen X.
        • Cox S.R.
        • Clarke T.K.
        • Wigmore E.M.
        • Gibson J.
        • et al.
        Impact of Polygenic Risk for Schizophrenia on Cortical Structure in UK Biobank.
        Biol Psychiatry. 2019; 86: 536-544
        • Ranlund S.
        • Rosa M.J.
        • de Jong S.
        • Cole J.H.
        • Kyriakopoulos M.
        • Fu C.H.Y.
        • et al.
        Associations between polygenic risk scores for four psychiatric illnesses and brain structure using multivariate pattern recognition.
        NeuroImage Clin. 2018; https://doi.org/10.1016/j.nicl.2018.10.008
        • Rabinowitz J.A.
        • Campos A.I.
        • Ong J.S.
        • García-Marín L.M.
        • Alcauter S.
        • Mitchell B.L.
        • et al.
        Shared Genetic Etiology between Cortical Brain Morphology and Tobacco, Alcohol, and Cannabis Use.
        Cereb Cortex. 2022; 32: 796-807
        • Ni G.
        • Moser G.
        • Ripke S.
        • Neale B.M.
        • Corvin A.
        • Walters J.T.R.
        • et al.
        Estimation of Genetic Correlation via Linkage Disequilibrium Score Regression and Genomic Restricted Maximum Likelihood.
        Am J Hum Genet. 2018; https://doi.org/10.1016/j.ajhg.2018.03.021
        • Ni G.
        • Amare A.T.
        • Zhou X.
        • Mills N.
        • Gratten J.
        • Lee S.H.
        The genetic relationship between female reproductive traits and six psychiatric disorders.
        Sci Rep. 2019; https://doi.org/10.1038/s41598-019-48403-x
        • Abdellaoui A.
        • Hugh-Jones D.
        • Yengo L.
        • Kemper K.E.
        • Nivard M.G.
        • Veul L.
        • et al.
        Genetic correlates of social stratification in Great Britain.
        Nat Hum Behav. 2019; https://doi.org/10.1038/s41562-019-0757-5
        • Purcell S.M.
        • Wray N.R.
        • Stone J.L.
        • Visscher P.M.
        • O’Donovan M.C.
        • Sullivan P.F.
        • et al.
        Common polygenic variation contributes to risk of schizophrenia and bipolar disorder.
        Nature. 2009; https://doi.org/10.1038/nature08185
        • Bulik-Sullivan B.
        • Finucane H.K.
        • Anttila V.
        • Gusev A.
        • Day F.R.
        • Loh P.R.
        • et al.
        An atlas of genetic correlations across human diseases and traits.
        Nat Genet. 2015; https://doi.org/10.1038/ng.3406
        • Lorenzi M.
        • Altmann A.
        • Gutman B.
        • Wray S.
        • Arber C.
        • Hibar D.P.
        • et al.
        Susceptibility of brain atrophy to TRIB3 in Alzheimer’s disease, evidence from functional prioritization in imaging genetics.
        Proc Natl Acad Sci U S A. 2018; https://doi.org/10.1073/pnas.1706100115
        • Bycroft C.
        • Freeman C.
        • Petkova D.
        • Band G.
        • Elliott L.T.
        • Sharp K.
        • et al.
        The UK Biobank resource with deep phenotyping and genomic data.
        Nature. 2018; https://doi.org/10.1038/s41586-018-0579-z
        • Desikan R.S.
        • Ségonne F.
        • Fischl B.
        • Quinn B.T.
        • Dickerson B.C.
        • Blacker D.
        • et al.
        An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest.
        Neuroimage. 2006; https://doi.org/10.1016/j.neuroimage.2006.01.021
        • Robinson M.R.
        • Kleinman A.
        • Graff M.
        • Vinkhuyzen A.A.E.
        • Couper D.
        • Miller M.B.
        • et al.
        Genetic evidence of assortative mating in humans.
        Nat Hum Behav. 2017; https://doi.org/10.1038/s41562-016-0016
        • Smith S.M.
        • Douaud G.
        • Chen W.
        • Hanayik T.
        • Alfaro-Almagro F.
        • Sharp K.
        • Elliott L.T.
        An expanded set of genome-wide association studies of brain imaging phenotypes in UK Biobank.
        Nat Neurosci. 2021; https://doi.org/10.1038/s41593-021-00826-4
        • Elliott L.T.
        • Sharp K.
        • Alfaro-Almagro F.
        • Shi S.
        • Miller K.L.
        • Douaud G.
        • et al.
        Genome-wide association studies of brain imaging phenotypes in UK Biobank.
        Nature. 2018; 562: 210-216
        • Wold H.
        Estimation of principal components and related models by iterative least squares.
        Multivariate Analysis. 1966;
        • Pedregosa F.
        • Varoquaux G.
        • Gramfort A.
        • Michel V.
        • Thirion B.
        • Grisel O.
        • et al.
        Scikit-learn: Machine learning in Python.
        J Mach Learn Res. 2011;
        • Klein M.
        • Walters R.K.
        • Demontis D.
        • Stein J.L.
        • Hibar D.P.
        • Adams H.H.
        • et al.
        Genetic markers of ADHD-related variations in intracranial volume.
        Am J Psychiatry. 2019; https://doi.org/10.1176/appi.ajp.2018.18020149
        • Mooney M.A.
        • Bhatt P.
        • Hermosillo R.J.M.
        • Ryabinin P.
        • Nikolas M.
        • Faraone S.V.
        • et al.
        Smaller total brain volume but not subcortical structure volume related to common genetic risk for ADHD.
        Psychol Med. 2020; https://doi.org/10.1017/S0033291719004148
        • Hermosillo R.J.M.
        • Mooney M.A.
        • Fezcko E.
        • Earl E.
        • Marr M.
        • Sturgeon D.
        • et al.
        Polygenic Risk Score–Derived Subcortical Connectivity Mediates Attention-Deficit/Hyperactivity Disorder Diagnosis.
        Biol Psychiatry Cogn Neurosci Neuroimaging. 2020; https://doi.org/10.1016/j.bpsc.2019.11.014
        • Demontis D.
        • Walters R.K.
        • Martin J.
        • Mattheisen M.
        • Als T.D.
        • Agerbo E.
        • et al.
        Discovery of the first genome-wide significant risk loci for attention deficit/hyperactivity disorder.
        Nat Genet. 2019; https://doi.org/10.1038/s41588-018-0269-7
        • Faraone S.V.
        • Asherson P.
        • Banaschewski T.
        • Biederman J.
        • Buitelaar J.K.
        • Ramos-Quiroga J.A.
        • et al.
        Attention-deficit/hyperactivity disorder.
        Nat Rev Dis Prim. 2015; https://doi.org/10.1038/nrdp.2015.20
        • Wang J.
        • Li M.D.
        Common and unique biological pathways associated with smoking initiation/progression, nicotine dependence, and smoking cessation.
        Neuropsychopharmacology. 2010; https://doi.org/10.1038/npp.2009.178
        • Zhang L.I.
        • Poo M.M.
        Electrical activity and development of neural circuits.
        Nat Neurosci. 2001;
        • Furberg H.
        • Kim Y.
        • Dackor J.
        • Boerwinkle E.
        • Franceschini N.
        • Ardissino D.
        • et al.
        Genome-wide meta-analyses identify multiple loci associated with smoking behavior.
        Nat Genet. 2010; https://doi.org/10.1038/ng.571
        • Levin E.D.
        • McClernon F.J.
        • Rezvani A.H.
        Nicotinic effects on cognitive function: Behavioral characterization, pharmacological specification, and anatomic localization.
        Psychopharmacology. 2006; https://doi.org/10.1007/s00213-005-0164-7
        • Elliott M.L.
        • Belsky D.W.
        • Anderson K.
        • Corcoran D.L.
        • Ge T.
        • Knodt A.
        • et al.
        A Polygenic Score for Higher Educational Attainment is Associated with Larger Brains.
        Cereb Cortex. 2019; https://doi.org/10.1093/cercor/bhy219
        • Mitchell B.L.
        • Cuéllar-Partida G.
        • Grasby K.L.
        • Campos A.I.
        • Strike L.T.
        • Hwang L.D.
        • et al.
        Educational attainment polygenic scores are associated with cortical total surface area and regions important for language and memory.
        Neuroimage. 2020; 212https://doi.org/10.1016/j.neuroimage.2020.116691
        • Lett T.A.
        • Vogel B.O.
        • Ripke S.
        • Wackerhagen C.
        • Erk S.
        • Awasthi S.
        • et al.
        Cortical Surfaces Mediate the Relationship Between Polygenic Scores for Intelligence and General Intelligence.
        Cereb Cortex. 2020; 30: 2708-2719
        • Lee J.J.
        • Wedow R.
        • Okbay A.
        • Kong E.
        • Maghzian O.
        • Zacher M.
        • et al.
        Gene discovery and polygenic prediction from a genome-wide association study of educational attainment in 1.1 million individuals.
        Nat Genet. 2018; https://doi.org/10.1038/s41588-018-0147-3
        • Okbay A.
        • Beauchamp J.P.
        • Fontana M.A.
        • Lee J.J.
        • Pers T.H.
        • Rietveld C.A.
        • et al.
        Genome-wide association study identifies 74 loci associated with educational attainment.
        Nature. 2016; https://doi.org/10.1038/nature17671
        • Terwisscha Van Scheltinga A.F.
        • Bakker S.C.
        • Van Haren N.E.M.
        • Derks E.M.
        • Buizer-Voskamp J.E.
        • Boos H.B.M.
        • et al.
        Genetic schizophrenia risk variants jointly modulate total brain and white matter volume.
        Biol Psychiatry. 2013; https://doi.org/10.1016/j.biopsych.2012.08.017
        • Wang T.
        • Zhang X.
        • Li A.
        • Zhu M.
        • Liu S.
        • Qin W.
        • et al.
        Polygenic risk for five psychiatric disorders and cross-disorder and disorder-specific neural connectivity in two independent populations.
        NeuroImage Clin. 2017; https://doi.org/10.1016/j.nicl.2017.02.011
        • Liu S.
        • Li A.
        • Liu Y.
        • Yan H.
        • Wang M.
        • Sun Y.
        • et al.
        Polygenic effects of schizophrenia on hippocampal grey matter volume and hippocampus-medial prefrontal cortex functional connectivity.
        Br J Psychiatry. 2020; https://doi.org/10.1192/bjp.2019.127
        • Mızrak E.
        • Bouffard N.R.
        • Libby L.A.
        • Boorman E.D.
        • Ranganath C.
        The hippocampus and orbitofrontal cortex jointly represent task structure during memory-guided decision making.
        Cell Rep. 2021; https://doi.org/10.1016/j.celrep.2021.110065
        • Rudebeck P.H.
        • Rich E.L.
        Orbitofrontal cortex.
        Current Biology. 2018; https://doi.org/10.1016/j.cub.2018.07.018
        • van Heukelum S.
        • Mars R.B.
        • Guthrie M.
        • Buitelaar J.K.
        • Beckmann C.F.
        • Tiesinga P.H.E.
        • et al.
        Where is Cingulate Cortex? A Cross-Species View.
        Trends in Neurosciences. 2020; https://doi.org/10.1016/j.tins.2020.03.007
        • Morawetz C.
        • Mohr P.N.C.
        • Heekeren H.R.
        • Bode S.
        The effect of emotion regulation on risk-taking and decision-related activity in prefrontal cortex.
        Soc Cogn Affect Neurosci. 2020; https://doi.org/10.1093/scan/nsz078
        • Sprooten E.
        • Franke B.
        • Greven C.U.
        The P-factor and its genomic and neural equivalents: an integrated perspective.
        Mol Psychiatry. 2021; https://doi.org/10.1038/s41380-021-01031-2
        • Avinun R.
        The E Is in the G: Gene–Environment–Trait Correlations and Findings From Genome-Wide Association Studies.
        Perspect Psychol Sci. 2020; https://doi.org/10.1177/1745691619867107
        • Kong A.
        • Thorleifsson G.
        • Frigge M.L.
        • Vilhjalmsson B.J.
        • Young A.I.
        • Thorgeirsson T.E.
        • et al.
        The nature of nurture: Effects of parental genotypes.
        Science. 2018; 80https://doi.org/10.1126/science.aan6877