A Multivariate Genome-Wide Association Study Reveals Neural Correlates and Common Biological Mechanisms of Psychopathology Spectra

A Multivariate Genome-Wide Association Study Reveals Neural Correlates and Common Biological Mechanisms of Psychopathology Spectra | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article A Multivariate Genome-Wide Association Study Reveals Neural Correlates and Common Biological Mechanisms of Psychopathology Spectra Christal Davis, Yousef Khan, Sylvanus Toikumo, Zeal Jinwala, D Boomsma, and 4 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4228593/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract There is considerable comorbidity across externalizing and internalizing behavior dimensions of psychopathology. We applied genomic structural equation modeling (gSEM) to genome-wide association study (GWAS) summary statistics to evaluate the factor structure of externalizing and internalizing psychopathology across 16 traits and disorders among European-ancestry individuals (n’s = 16,400 to 1,074,629). We conducted GWAS on factors derived from well-fitting models. Downstream analyses served to identify biological mechanisms, explore drug repurposing targets, estimate genetic overlap between the externalizing and internalizing spectra, and evaluate causal effects of psychopathology liability on physical health. Both a correlated factors model, comprising two factors of externalizing and internalizing risk, and a higher-order single-factor model of genetic effects contributing to both spectra demonstrated acceptable fit. GWAS identified 409 lead single nucleotide polymorphisms (SNPs) associated with externalizing and 85 lead SNPs associated with internalizing, while the second-order GWAS identified 256 lead SNPs contributing to broad psychopathology risk. In bivariate causal mixture models, nearly all externalizing and internalizing causal variants overlapped, despite a genetic correlation of only 0.37 (SE = 0.02) between them. Externalizing genes showed cell-type specific expression in GABAergic, cortical, and hippocampal neurons, and internalizing genes were associated with reduced subcallosal cortical volume, providing insight into the neurobiological underpinnings of psychopathology. Genetic liability for externalizing, internalizing, and broad psychopathology exerted causal effects on pain, general health, cardiovascular diseases, and chronic illnesses. These findings underscore the complex genetic architecture of psychopathology, identify potential biological pathways for the externalizing and internalizing spectra, and highlight the physical health burden of psychiatric comorbidity. Social science/Psychology/Human behaviour Biological sciences/Genetics/Genetic association study/Genome-wide association studies Biological sciences/Psychology/Human behaviour Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction Traditional categorical classifications of psychopathology suffer from significant limitations. In epidemiological studies, psychiatric disorders consistently co-occur more often than expected, 1,2 suggesting overlapping genetic underpinnings. 3 Furthermore, largely arbitrary thresholds and polythetic criterion sets yield thousands of unique symptom combinations that lead to the same diagnosis. 4 Along with the challenges these limitations present for clinical care, they hinder progress in psychiatric genetics and neuroscience research, where efforts to identify biological mechanisms that underlie psychiatric illness have had limited success. 5,6 Recent attempts to address these limitations have included alternative approaches to understanding psychopathology, most notably the Hierarchical Taxonomy of Psychopathology (HiTOP) and the National Institute of Mental Health’s Research Domain Criteria (RDoC) initiative. 5,7–9 HiTOP proposes a dimensional structure of psychopathology that progresses hierarchically from symptoms to an overarching psychopathology factor. In contrast, RDoC aims to identify the underlying mechanisms of psychopathology by focusing on domains of functioning rather than diagnostic categories. Despite differences in the units of analysis and the dimensions they identify, these systems’ constructs align well in a model of psychopathology. 10 Beginning in the 1990s, twin and family studies showed that dimensions of psychopathology had a shared genetic basis, 11,12 with externalizing and internalizing psychopathology being the subject of much of this research. Whereas externalizing behaviors involve interaction with the social environment (e.g., aggression, impulsivity), internalizing behaviors are directed inwards (e.g., anxiety, depression). 13 With statistical and methodological advances, molecular genetic research has also identified common externalizing 14 and internalizing 15,16 genetic factors that underlie each spectrum of psychopathology. Twin and family studies 17–19 and principal component analyses 17,20 have also examined genetic factors shared by both externalizing and internalizing psychopathology. Genome-wide association studies (GWAS) of childhood behavior problems, which encompass externalizing and internalizing psychopathology, identified two genome-wide significant loci. 21,22 Romero and colleagues recently used a cross-trait GWAS meta-analysis to identify pleiotropic genetic effects across 12 psychiatric disorders. 23 Because the meta-analytic signal in that study was driven by schizophrenia, the interpretation and joint biological characterization of the cross-trait signal was limited. Genomic structural equation modeling (gSEM) offers several advantages over cross-trait meta-analysis for identifying the shared genetic architecture that underlies psychopathology. First, gSEM enables specific hypotheses about the factor structure of psychopathology to be tested, with explicit comparison of proposed models that could account for the overlap observed across externalizing and internalizing spectra. Second, the use of latent variables helps to identify the common genetic effects across externalizing and internalizing spectra, minimizing the capture of genetic signals associated with the most dominant trait, as in the meta-analytic study of Romero et al. 23 Other gSEM studies have investigated the factor structure of psychiatric disorders and identified one to four factors that underlie their shared liability. 16,24,25 A previous GWAS identified two genome-wide hits for a higher-order p -factor encompassing compulsive, psychotic, internalizing, and neurodevelopmental disorders, and 66 significant hits upon post hoc examination of a bifactor model p -factor. Because the study included only psychiatric disorders, it did not capture a broad spectrum of psychopathology consistent with dimensional models like HiTOP. It also included only two internalizing (anxiety and major depressive disorder) and two externalizing (attention-deficit hyperactivity disorder and problematic alcohol use) conditions. To conduct a detailed examination of the shared genetic architecture of externalizing and internalizing psychopathology, we applied gSEM to large GWAS summary statistics. Adopting a dimensional, transdiagnostic approach, we first evaluated models of psychopathology to determine which factor structure provided the best fit to the pattern of genetic covariance across 16 externalizing and internalizing traits and disorders. To identify genetic effects for the externalizing spectrum, internalizing spectrum, and across the externalizing and internalizing spectra, we conducted GWAS on the latent factors derived from models with acceptable fit. Next, we performed downstream analyses to characterize biological mechanisms underlying the shared genetic liability to psychopathology and to examine potential causal effects on physical health. Identifying mechanisms that account for vulnerability across levels of psychopathology can yield insights into the genetic basis of psychopathology, which could lead to advancements in treatment, diagnosis, and disorder classification. Subjects and Methods Summary Statistics Externalizing. Ten sets of summary statistics in European-ancestry (EUR) individuals were selected based on existing theory regarding the externalizing spectrum (Supplementary Table 1). We included summary statistics from the largest available GWAS of the following externalizing disorders: attention deficit hyperactivity disorder (ADHD; n = 225,534), 26 four substance use disorders [SUDs; i.e., alcohol (AUD; n = 753,248), 27 cannabis (CanUD; n = 886,025), 28 opioid (OUD; n = 425,944), 29 and tobacco (TUD; n = 495,005) 30 ]. We also included broader measures of externalizing psychopathology [age of first sexual intercourse (AgeSex; reverse-coded; n = 317,694), 14 general risk tolerance (Risk; n = 431,126), 14,31 number of sexual partners (NumSex; n = 370,711), 14,31 antisocial behavior (ASB; n = 16,400), 32 and automobile speeding propensity (n = 404,291) 31 ]. Internalizing. Summary statistics from eight GWAS in EUR individuals were selected to capture the internalizing spectrum (Supplementary Table 1). Three were the largest available GWAS of internalizing disorders: ( 1 ) anorexia nervosa (AN; n = 72,517), 33 ( 2 ) major depressive disorder (MDD; n = 1,074,629), 34 and ( 3 ) posttraumatic stress disorder (PTSD; n = 214,408). 35 To reflect a broad liability to internalizing, we included irritability ( http://www.nealelab.is/uk-biobank/ ; n = 345,231), loneliness (n = 490,689), 36 subjective wellbeing (SWB; reverse-coded; n = 298,420), 37 miserableness ( http://www.nealelab.is/uk-biobank/ ; n = 355,182), and anxiety (ANX; n = 280,490). 38–40 38–41 To boost power to detect variants associated with both anxiety disorders and subclinical anxiety, we combined three anxiety GWAS 38–40 using multi-trait analysis of GWAS. 41 Exploratory Analysis Genetic Correlations. Using linkage disequilibrium score regression (LDSC) 42 in GenomicSEM, 24 we calculated genetic correlations ( r g ) between the input traits. Single nucleotide polymorphisms (SNPs) were filtered using EUR HapMap3 reference panels, 43 and SNPs with MAF < 0.01 were removed. LDSC was performed using ancestry-matched 1000 Genomes Phase 3 linkage disequilibrium (LD) scores. 44 When available in the summary statistics, SNP-level sample sizes were specified. Otherwise, the effective sample size was calculated by summing effective sample sizes across the input GWAS cohorts. 45 After conducting LDSC, genetic correlations were inspected to identify traits having weak associations with the other input traits prior to fitting structural equation models. Traits with average r g < 0.20 were excluded from gSEM models. Hierarchical Cluster Analysis. To evaluate whether traits clustered with their predicted spectrum, we conducted hierarchical cluster analysis of the r g matrix using the hclust() function in RStudio. 46 We calculated a Euclidean distance matrix and used Ward’s agglomerative clustering algorithm 47 to identify clusters. A plot of the within-cluster sum of squares was used to determine the optimal number of clusters. Network Analysis. A network analysis of the r g matrix was performed using the igraph package in RStudio. 48 The matrix was transformed into an undirected and weighted network graph, in which nodes represent each trait and the weights of the links between traits represent the magnitude of their genetic correlation. The optimal network community structure was determined by maximizing modularity, a measure of the quality of a clustering solution. Genomic Structural Equation Modeling We fit four confirmatory factor analyses (CFAs) based on existing theories of psychopathology. 49–51 First, we evaluated a correlated factors model with two factors representing externalizing and internalizing psychopathology. Next, we fit a bifactor model consisting of a general psychopathology factor on which all traits loaded, and two narrower externalizing and internalizing psychopathology factors. A higher-order model was also fit with two first-order factors representing externalizing and internalizing, and a second-order factor (EXT + INT) representing genetic effects shared by the two spectra. Finally, we fit a unidimensional, or p -factor, model where all traits loaded onto a single latent factor. In all CFA models, the residuals of the four SUDs were allowed to correlate; all other residuals were uncorrelated. We evaluated the models with chi-square, the Akaike information criterion (AIC), comparative fit index (CFI), and standardized root mean squared residual (SRMR) fit statistics. We also inspected the results for low (< 0.35) or negative loadings as an indicator of each model’s appropriateness. When preparing the data for GWAS, we excluded SNPs with MAF < 0.01 and imputation scores < 0.6. Coefficients and standard error values were standardized across summary statistics to ensure that effects were scaled similarly for all traits. Each SNP was regressed on the model latent variable(s) using diagonally weighted least squares estimation. After performing GWAS, we calculated factor-specific Q SNP values by comparing the fit of a common pathway model to an independent pathway model. 16 Q SNP provides a measure of SNP heterogeneity, reflecting the extent to which a SNP exerts effects entirely through the common factor (i.e., common pathway model) or, instead, exerts effects differentially across a factor’s indicators (i.e., independent pathway model). SNPs with a Q chi-square p-value 0.10 with the EUR 1000 Genomes Phase 3 reference panel. 44 Biological Characterization Gene-based tests, gene-set enrichment, and gene-tissue expression analyses were conducted using MAGMA 53 in FUMA v1.6.0 54 . We examined gene expression in BrainSpan 55 and GTEx v8 56 tissue samples. In FUMA, gene associations were identified based on their: ( 1 ) position, ( 2 ) expression quantitative trait loci (eQTLs) from PsychENCODE 57 and GTEx v8 56 brain tissue samples, and ( 3 ) chromatin interactions using Hi-C data from the dorsolateral prefrontal cortex, hippocampus, ventricles, and neural progenitor cells. We also analyzed gene expression at the cellular level in single-cell RNA sequencing (scRNA-seq) datasets from 15 human brain cell expression profiles. 58–61 For these analyses, we used a three-step approach: ( 1 ) conducting gene-property analyses within each dataset, ( 2 ) identifying independent associations using within-dataset conditional analyses, and ( 3 ) identifying independent clusters of signals using cross-datasets conditional analyses. 58 Transcriptome Wide Association Studies We conducted two transcriptome-wide association studies (TWAS) on each factor. 62 First, we used S-MultiXcan 62 , which prioritizes likely causal genes by jointly predicting gene expression across multiple tissues. S-MultiXcan produces an overall Z-score and p-value across all tissues, as well as values for the most and least associated tissues. We examined expression across the 13 brain tissues in GTEx v8 56 and identified the most associated tissue for each gene. To complement this approach, we used S-PrediXcan 63 and weights trained on transcriptional differences in the frontal and temporal cortices of psychiatric cases and controls 64 from PsychENCODE. 65 A Bonferroni correction was applied to identify significant associations. Associations with Brain Phenotypes We used BrainXcan 66 to examine associations between the psychopathology spectra and 327 brain image-derived phenotypes (IDPs) from structural (T1-weighted) and diffusion magnetic resonance images (dMRIs) using ridge regression. Effect sizes and p-values were adjusted using LD block-based permutation, and Bonferroni correction was used to account for multiple testing (T1: 0.05/109 = 4.59 x 10 − 4 ; dMRI = 0.05/218 = 2.29 x 10 − 4 ). We also conducted bidirectional Mendelian randomization (MR) analyses for the most significantly associated brain IDPs. Because the significance of the IDP-factor association was used to identify pairs on which to perform MR, the resulting MR p-values were used to discern the possible direction of association, rather than to evaluate significance. 66 Drug Repurposing To identify gene targets for drug repurposing, we used five different gene annotation approaches: ( 1 ) MAGMA, 53 ( 2 ) chromatin interactions, ( 3 ) eQTL, ( 4 ) S-MultiXcan, and ( 5 ) S-PrediXcan. 62,63 To avoid unreliable associations, we queried the subset of druggable genes 67 identified by multiple biological annotation sources using the Drug-Gene-Interaction Database (DGIdb). 68 For the first-order factors, we limited drug repurposing analyses to genes that showed specificity of association with either externalizing or internalizing. Causal Mixture Models (MiXeR) Univariate MiXeR analyses 69 were conducted to estimate each factor’s polygenicity (i.e., the number of causal variants required to explain 90% SNP heritability) and discoverability (i.e., the average effect size of causal variants). 70 Next, bivariate models were used to identify the proportion of unique and shared causal variants for the externalizing and internalizing spectra. In contrast to genetic correlations, MiXeR accounts for polygenic overlap regardless of whether causal variants have the same or opposite direction of effect. Genetic Correlations Using the Complex Trait Genetics Virtual Lab 71 , we calculated batch genetic correlations between each factor GWAS and 1,437 phenotypes from publicly available GWAS. GWAS that were used as an input for the gSEM models were excluded. Genetic correlations were calculated using LDSC 42 and EUR 1000 Genomes Phase 3 data 44 as LD references. To account for multiple testing, a Benjamini-Hochberg false discovery rate (FDR) correction was applied to each set of genetic correlations. Generalized Summary-data-based Mendelian Randomization To evaluate potentially causal impacts of externalizing and internalizing genetic risk on physical health, we conducted Generalized Summary-data-based Mendelian Randomization (GSMR) 72 using 15 health traits as outcomes. Traits were chosen across four domains—( 1 ) pain, ( 2 ) general health, ( 3 ) cardiovascular disease, and ( 4 ) chronic illness—each of which has demonstrated associations with psychopathology. 73–75 We used summary statistics from GWAS of pain intensity, 76 multisite chronic pain, 77 and back pain. 78 General health indices included GWAS summary statistics from the UK Biobank for longstanding illness, disability, or infirmity; hospitalization; and age at death. We selected five cardiovascular GWAS: ( 1 ) heart failure, 79 ( 2 ) stroke, 79 ( 3 ) myocardial infarction, 80 ( 4 ) hypertension ( http://www.nealelab.is/uk-biobank/ ), and ( 5 ) abdominal aortic aneurysm. 79 Finally, we selected four GWAS of chronic illnesses: ( 1 ) type 2 diabetes, 81 ( 2 ) inflammatory bowel disease (IBD), 82 ( 3 ) chronic obstructive pulmonary disease (COPD), 79 and ( 4 ) asthma. 79 Genetic instruments with significant pleiotropic effects on both the exposure and outcome were removed using the heterogeneity in dependent instruments outlier (HEIDI) method. 72 A Bonferroni adjusted p-value was applied to identify significant effects (0.05/45 = 0.001). Results Based on previous GWAS of externalizing and internalizing 14,15,83 and existing theory, 7,9 we considered a total of 18 externalizing and internalizing traits for inclusion in the analyses (Supplementary Fig. 1). We excluded two traits (automobile speeding propensity and anorexia nervosa) that were weakly associated with the others in the model (mean r g < 0.20). Network analysis and hierarchical agglomerative cluster analysis both revealed two clusters that correspond to externalizing and internalizing spectra (see Fig. 1 ). Using the 16 traits, we tested several CFA models (see Fig. 2 and Supplementary Figs. 2 and 3). A general psychopathology factor model did not provide adequate fit to the data ( \(\chi\) 2 (98) = 8965.28, AIC = 9041.28, CFI = 0.79, and SRMR = 0.15), although standardized loadings were all significant and > 0.35. A bifactor model comprising a general psychopathology factor and two specific factors representing externalizing and internalizing spectra fit the data well ( \(\chi\) 2 ( 81 ) = 2527.19, AIC = 2637.19, CFI = 0.94, and SRMR = 0.05). However, the model led to several weak (< 0.35) and one negative standardized loading, possibly from overfitting the data. Thus, despite its good fit, we did not perform GWAS on factors from this model because they would be difficult to interpret. Two CFA models provided adequate fit and had strong factor loadings: ( 1 ) a correlated-factors model and ( 2 ) a higher-order factor model. Fit statistics for both models were \(\chi\) 2 (97) = 3877.82, AIC = 3955.82, CFI = 0.91, and SRMR = 0.09. To ensure identification in the higher order model, the loadings of externalizing and internalizing onto the second-order factor were constrained equal to the square root of the genetic correlation between the externalizing and internalizing factors. 84 Multivariate GWAS of Psychopathology Spectra Using a Q SNP analysis, 228 independent SNPs exhibited heterogeneous effects across the externalizing spectrum (Supplementary Fig. 4). Among the associations of these SNPs within the input GWAS (Supplementary Table 2), a plurality was most strongly associated with age at first sexual intercourse (37.23%), followed by TUD (23.38%). After filtering heterogenous SNPs, a multivariate GWAS of externalizing identified 409 GWS independent lead SNPs (Supplementary Table 3). Of these, 92 (22.49%) were not identified or within \(\pm\) 1000 kb of SNPs identified by any of the input GWAS, and four were not previously associated with any externalizing trait using the same threshold. Three of the four novel SNPs were on chromosome 4 (rs1961547, rs9316, and rs7682762), with the fourth on chromosome 22 (rs1473811). These SNPs showed phenotypic associations with chronotype, schizophrenia, and social support, among other traits (Supplementary Table 4). For internalizing, 222 independent SNPs exhibited significant heterogeneity (Supplementary Fig. 5), with most (86.49%) showing the strongest associations with MDD (Supplementary Table 5). After filtering heterogeneous SNPs, there were 85 GWS independent lead SNPs (Supplementary Table 6). Of these, 23 (27.06%) were not identified or within \(\pm\) 1000 kb of SNPs identified by the input GWAS, and two were not previously associated with any internalizing trait. The two novel associations were on chromosomes 3 and 4 (rs1381763 and rs4698408, respectively). Novel SNPs had phenotypic associations with neuroticism and depression (Supplementary Table 7). In the GWAS of genetic effects shared across externalizing and internalizing (EXT + INT factor), there were 130 lead SNPs that exhibited heterogeneous effects (Supplementary Fig. 6). Of these (Supplementary Table 8), a plurality (47.69%) was most strongly associated with age at first sexual intercourse, followed by AUD (17.69%). There were 256 GWS independent lead SNPs associated with EXT + INT (Supplementary Table 9), 38 of which (14.84%) were not identified or within \(\pm\) 1000 kb of SNPs identified by any of the input GWAS. All significant loci were previously associated with at least one externalizing or internalizing phenotype (Fig. 3 ). Biological Characterization MAGMA identified 493 genes significantly associated with externalizing, including CADM2 and DRD2 (Supplementary Table 10). Gene-property analysis showed enriched expression during prenatal brain development (Supplementary Fig. 7). Gene expression was significantly enhanced in almost all brain tissues, with the top associations being with the cerebellar hemisphere, cerebellum, and frontal cortex (Supplementary Fig. 8). A gene-set related to mRNA binding was the only significant association with externalizing ( p bon = 0.04; Supplementary Table 11). Using scRNA-seq datasets, externalizing was significantly associated with dopaminergic and GABAergic neurons and neuroblasts from embryonic brain samples (GSE76381), human cortical neurons and hybrid cells that display characteristics of neurons and astrocytes (GSE67835), and pyramidal neurons from the cornu ammonis (CA1) hippocampal region 60 . After conditional analyses, independent significant associations remained for GABAergic neurons, cortical neurons, and hippocampal neurons (Supplementary Fig. 9). There were 146 genes significantly associated with internalizing, including several on chromosome 8 ( BLK , XKR6 , and C8orf12 ) that were previously associated with neuroticism (Supplementary Table 12). 85 Gene expression was not significantly enhanced at any developmental stage (Supplementary Fig. 7), but predominated in the brain, with the frontal cortex and anterior cingulate cortex most strongly associated (Supplementary Fig. 8). Although no gene-sets were significant after Bonferroni correction, the top associations were with genes involved in synaptic assembly and transmission (Supplementary Table 11). In scRNA-seq analyses, the only significant cell-type association was with GABAergic neurons (GSE76381), which was not independently significant after conditional analyses. There were 321 genes significantly associated with EXT + INT (Supplementary Table 13). The top hits were for FAM120AOS , DCC , and P4HTM , all of which have previously been associated with both internalizing and externalizing traits. Gene expression was enhanced in brain tissue during prenatal developmental (Supplementary Fig. 7), and genes associated with the broad spectrum of psychopathology were predominantly expressed in the brain (Supplementary Fig. 8). No gene sets were significant after Bonferroni correction, but like internalizing, the top sets comprised genes involved in synaptic activity (Supplementary Table 11). Using scRNA-seq, EXT + INT showed significant associations with dopaminergic neurons (GSE76381), GABAergic neurons (GSE76381), and cortical neurons (GSE67835), though these associations were not independently significant. Transcriptome-Wide Association Analysis Using S-MultiXcan to predict the effects of SNP variation on gene expression across 13 brain tissues revealed 352 significant gene associations for externalizing, 141 for internalizing, and 238 for EXT + INT (Fig. 4 , Supplementary Tables 14–16, and Supplementary Fig. 10). TWAS using PsychENCODE data for S-PrediXcan identified 207 significant genes for externalizing, 52 for internalizing, and 124 for EXT + INT (Supplementary Tables 17–19 and Supplementary Fig. 11). Forty-five genes were identified by both S-MultiXcan and S-PrediXcan for externalizing, 21 for internalizing, and 36 for EXT + INT, with gene-property analyses showing these genes to be consistently upregulated across brain tissues (Supplementary Figs. 12–14), and three ( C1QTNF4 , DPYSL5 , and SLC12A5 ) were almost exclusively upregulated in brain tissues. Associations with Brain Phenotypes After Bonferroni correction of the LD-adjusted p-values, 8 T1 (Supplementary Fig. 15) and 12 dMRI IDPs (Supplementary Fig. 16) were significantly associated with externalizing (Supplementary Table 20), including positive associations with gray matter volume in the thalamus, caudate nuclei, and occipital pole, and negative associations with the right ventral striatum and left amygdala. From dMRIs, there were significant associations with intra-cellular volume fraction or orientation dispersion indices (ODI) in the medial lemniscus, cerebral peduncle, and middle cerebellar peduncle, among others. The only significant association for internalizing was with lower gray matter volume in the left subcallosal cortex (Supplementary Table 21 and Supplementary Figs. 17–18). Genetic factors shared across externalizing and internalizing spectra were significantly negatively associated with gray matter volume in the left amygdala and left subcallosal cortex (Fig. 5 ), positively associated with ODI in the medial lemniscus and left cerebellar peduncle, and negatively associated with ODI in the right external capsule (Supplementary Table 22 and Supplementary Figs. 19–20). MR analyses showed potential bidirectional relationships between externalizing and gray matter volume in the right ventral striatum and left thalamus (Supplementary Table 23). There was greater evidence that reduced gray matter volume in the left subcallosal cortex was causally related to internalizing than vice versa ( p = 0.008 vs. 0.401; Supplementary Table 24). Evidence was mixed regarding the direction of causal effects for the second-order externalizing and internalizing factor (Supplementary Table 25). Drug Repurposing Among the 1,759 unique genes identified for externalizing using biological annotation tools, 308 were druggable targets. 67 Sixty of these genes were identified by at least two biological annotation tools, and 52 exhibited specificity for externalizing (i.e., were not associated with internalizing). When queried in DGIdb, these genes yielded 492 drug-gene interactions (Supplementary Table 26), including with dextroamphetamine (used to treat ADHD), phenobarbital (used to prevent withdrawal symptoms from benzodiazepines and alcohol), baclofen (used to treat AUD), naltrexone (used to treat AUD and OUD), naloxone (used to reverse opioid overdose), and methadone (used to treat OUD). Gene interactions with antimigraine, anti-inflammatory, and anticonvulsant drugs (e.g., topiramate and lamotrigine) were also identified. Most of the identified drugs had regulatory approval (64.84%). Biological annotation identified 454 unique genes associated with internalizing, 60 of which were druggable targets. 67 Fifteen of these were identified by at least two biological annotation tools and seven exhibited specificity for internalizing, yielding 292 drug-gene interactions (Supplementary Table 27). Drug targets included antidepressants and antipsychotics. Unlike externalizing, most identified drugs (82.33%) were not currently approved, suggesting potential candidates for use in treating internalizing psychopathology. For EXT + INT 1,138 unique genes were identified using the five biological annotation tools, nearly one-fifth (17.93%, n = 204) of which were druggable targets, with 47 of those implicated by more than one biological annotation method. Using DGIdb, we identified 460 unique drug-gene interactions (Supplementary Table 28), many of which were also present in the internalizing or externalizing results. As with internalizing, most of these drugs (75.52%) were not currently approved. Causal Mixture Models (MiXeR) The externalizing and internalizing spectra displayed similar levels of polygenicity, with an estimated 12,600 and 13,200 causal variants, respectively. However, internalizing had lower discoverability ( \({\widehat{\sigma }}_{\beta }\) 2 = 1.40 x 10 − 5 ) than externalizing ( \({\widehat{\sigma }}_{\beta }\) 2 = 1.44 x 10 − 4 ), suggesting that SNPs that influence internalizing traits may exert smaller effects and thus require larger samples to detect. Despite a MiXeR-estimated genetic correlation of 0.37, almost all causal variants (96.83% of externalizing and 92.42% of internalizing; Supplementary Fig. 21) overlapped across the two spectra, with more overlap than predicted by genetic correlation alone (AIC = 12.30, BIC = 4.06, where positive values indicate that the predicted model explains the data better than the genetic correlation alone). In fact, the models do not exclude the possibility that causal variants for externalizing were a subset of those for internalizing. Of the shared causal variants, 62.92% were estimated to be concordant in direction of effect. Genetic Correlations Applying a Bonferroni-adjusted p-value (0.05/1368 = 3.65 x 10 − 5 ), there were 413 significant genetic correlations with externalizing (Supplementary Table 29 and Supplementary Fig. 22). Tobacco phenotypes were among the most strongly correlated (current smoking: r g = 0.79, SE = 0.02; maternal smoking around birth: r g = 0.71, SE = 0.03; and ever smoked: r g = 0.62, SE = 0.02), along with lower socioeconomic status, including Townsend deprivation index ( r g = 0.68, SE = 0.03), living in housing supplied by a local authority or housing association ( r g = 0.66, SE = 0.03), experiencing financial difficulties ( r g = 0.58, SE = 0.03), and lower educational attainment ( r g = -0.44, SE = 0.02). After Bonferroni correction, 311 phenotypes were significantly genetically correlated with internalizing (Supplementary Table 30 and Supplementary Fig. 23). Among the strongest correlations were psychiatric phenotypes, such as mood swings ( r g = 0.90, SE = 0.01), neuroticism ( r g = 0.89, SE = 0.01), and feeling fed-up ( r g = 0.82, SE = 0.01). Internalizing was also significantly genetically correlated with several types of pain (abdominal: r g = 0.60, SE = 0.04; facial: r g = 0.51, SE = 0.08; chest: r g = 0.49, SE = 0.03; and multisite chronic pain: r g = 0.49, SE = 0.03, among others). There were 474 significant genetic correlations with EXT + INT, with most being like the first-order factors (Supplementary Table 31 and Supplementary Fig. 24). Generalized Summary-data-based Mendelian Randomization Externalizing had significant positive causal effects on all physical health traits examined, except age at death and IBD. Internalizing was significantly causally associated with all traits except age at death and abdominal aortic aneurysm. Additionally, internalizing had protective effects on IBD ( b xy = -0.32, SE = 0.09, p = 0.0004) and stronger positive associations than externalizing with all three pain phenotypes, all five cardiovascular diseases, and three of four chronic illnesses (Fig. 6 ). Like externalizing, EXT + INT had significant positive effects on all physical health traits except age at death and IBD (Supplementary Table 32). Discussion Comparing candidate factor structures of psychopathology, we found support for hierarchical models consistent with the HiTOP framework. Our models, which included symptom-level (e.g., risk tolerance and irritability) and disorder-level (e.g., TUD and MDD) manifestations of psychopathology, indicated that these traits could be organized onto higher dimensions representing externalizing and internalizing spectra, which were themselves subsumed under a broader umbrella of psychopathology genetic risk. Leveraging GWAS summary statistics that included over 1.5 million individuals, our findings also show that although there is shared variance across forms of psychopathology, the commonality does not manifest as a single overarching dimension (i.e., p -factor). Rather, the genetic architecture of psychopathology was better captured by a model that distinguished between specific dimensions while recognizing their interrelatedness. Our findings also demonstrated connections between psychopathology and RDoC domains 10 in downstream analyses that encompassed multiple RDoC units of analysis. For example, at the cellular level, externalizing-related variants were associated with RNA expression in pyramidal hippocampal neurons, which are implicated in RDoC’s cognitive control construct. For internalizing, analyses revealed molecular-level associations with drugs targeting serotonin and dopamine, which align with RDoC’s negative valence systems domain. Integrating HiTOP’s dimensional framework with RDoC’s etiological approach makes it possible to evaluate the extent to which psychopathology spectra show etiologically consistent associations across domains of functioning and units of analysis. Although externalizing and internalizing were better represented as correlated but distinct factors and had only a small-to-moderate genetic correlation, bivariate causal mixture models estimated extensive overlap in causal variants between the two spectra. This pleiotropy suggests that there are shared biological pathways that underpin externalizing and internalizing psychopathology. Through GWAS and downstream analyses, we identified potential neural mechanisms of this shared risk. Broad psychopathology genetic risk was associated with structural and functional differences in the brain, including reduced gray matter volume in the left amygdala and subcallosal cortex. Reduced left amygdala volume has been shown to mediate the relation between childhood threat exposure and the development of externalizing and internalizing symptoms. 86 Similarly, reduced left subcallosal cortical volume is a potential mediator of associations between personality and various emotional states (e.g., subjective well-being) and psychiatric disorders (e.g., alcohol dependence). 87 Psychopathology liability was also related to alterations in the organization and microstructure of white matter fibers, reflected by changes in orientation dispersion indices, which may signify disruptions in neural communication that contribute to various manifestations of psychopathology. 88 In summary, we identified neural mechanisms that operate at varying levels of specificity, 89 with some correlates linked specifically to externalizing or internalizing and others linked to broad psychopathology liability. Extending these findings beyond mental health, MR analyses showed that liability to psychopathology exerted potentially causal effects on adverse physical health outcomes. Of note, internalizing generally showed stronger associations with pain, general health, cardiovascular disease, and chronic illness than externalizing. Previous research also supports likely causal effects of genetic risk for internalizing traits/disorders on localized pain 90,91 and various disease outcomes. 92 An unexpected finding here was that internalizing was protective for IBD after removing pleiotropic variants. Although some studies demonstrated causal effects of MDD on risk for IBD, 93 studies of other internalizing disorders (e.g., anxiety) have not. 94 Thus, more research is needed to disentangle the complex relations of internalizing and inflammatory, autoimmune conditions such as IBD. Finally, although the impact of internalizing liability on physical health was particularly pronounced, both externalizing and broad psychopathology liability also showed potentially causal effects across physical health domains. These findings underscore the contribution of psychopathology liability to the emergence of co-occurring physical health conditions. A limitation of the present study is its inclusion of only European-ancestry individuals. Data adequate to explore a broad liability to internalizing and externalizing disorders and subclinical features in other ancestry groups were not available. Although data from GWAS of externalizing and internalizing-related disorders are available for some other non-European ancestry groups, more precise, non-disorder psychiatric phenotypes are limited in these populations. Nonetheless, some research suggests that a similar factor structure applies in African-ancestry individuals. At the disorder level, a gSEM of African-ancestry individuals (unpublished data) identified substance use and psychiatric disorder factors that roughly aligned with externalizing and internalizing, respectively. Additionally, the analyses showed that a higher-order factor accounted for genetic variance shared by substance use and psychiatric disorders, which to a degree corresponds to our EXT + INT factor. With the growth of diverse biobanks 95 and deep phenotyping 96 , more direct replication in other ancestries should soon be possible and given high priority. In conclusion, our findings supported a hierarchical structure of psychopathology, recognizing correlated externalizing and internalizing dimensions subsumed under a broader psychopathology liability. By discerning genetic variants and neural mechanisms operating at varying levels of specificity, the findings revealed the utility of applying a dimensional, hierarchical genetic approach to investigate psychopathology, which could augment existing categorical frameworks. Developing a more nuanced understanding of underlying biological mechanisms across forms of psychopathology could aid in constructing a more precise and comprehensive psychiatric nosology, providing a foundation on which to improve treatment and clinical outcomes. Declarations Acknowledgments: This work was supported by the Veterans Integrated Service Network 4 Mental Illness Research, Education and Clinical Center and by Department of Veterans Affairs grants I01 BX004820 to H.R.K., National Institute on Alcohol Abuse and Alcoholism grant AA028292 to R.L.K, and KNAW (Royal Netherlands Academy of Arts and Sciences) Academy Professor Award (PAH/6635) to D.I.B. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Conflicts of Interest: Dr. Kranzler is a member of advisory boards for Dicerna Pharmaceuticals, Sophrosyne Pharmaceuticals, Enthion Pharmaceuticals, and Clearmind Medicine; a consultant to Sobrera Pharmaceuticals; the recipient of research funding and medication supplies for an investigator-initiated study from Alkermes; and a member of the American Society of Clinical Psychopharmacology’s Alcohol Clinical Trials Initiative, which was supported in the last three years by Alkermes, Dicerna, Ethypharm, Lundbeck, Mitsubishi, Otsuka, and Pear Therapeutics. Drs. Kranzler and Gelernter hold U.S. patent 10,900,082 titled: "Genotype-guided dosing of opioid agonists," issued 26 January 2021. The other authors have no disclosures to make. References Kessler, R.C., Chiu, W.T., Demler, O. & Walters, E.E. Prevalence, severity, and comorbidity of 12-month DSM-IV disorders in the National Comorbidity Survey replication. Archives of General Psychiatry 62, 617–627 (2005). Kessler, R.C. et al. Lifetime and 12-month prevalence of DSM-III-R psychiatric disorders in the United States: Results from the National Comorbidity Survey. Archives of General Psychiatry 51, 8–19 (1994). Lee, P.H., Feng, Y.-C.A. & Smoller, J.W. Pleiotropy and cross-disorder genetics among psychiatric disorders. Biological Psychiatry 89, 20–31 (2021). Galatzer-Levy, I.R. & Bryant, R.A. 636,120 ways to have posttraumatic stress disorder. Perspectives on Psychological Science 8, 651–62 (2013). Kozak, M.J. & Cuthbert, B.N. The NIMH Research Domain Criteria Initiative: Background, issues, and pragmatics. Psychophysiology 53, 286–297 (2016). Waszczuk, M.A. et al. Redefining phenotypes to advance psychiatric genetics: Implications from hierarchical taxonomy of psychopathology. Journal of Abnormal Psychology 129, 143–161 (2020). Kotov, R. et al. The Hierarchical Taxonomy of Psychopathology (HiTOP): A dimensional alternative to traditional nosologies. Journal of Abnormal Psychology 126, 454–477 (2017). Cuthbert, B.N. Research Domain Criteria: Toward future psychiatric nosologies. Dialogues in Clinical Neuroscience 17, 89–97 (2015). Kotov, R. et al. The Hierarchical Taxonomy of Psychopathology (HiTOP): A quantitative nosology based on consensus of evidence. Annual Review of Clinical Psychology 17, 83–108 (2021). Michelini, G., Palumbo, I.M., DeYoung, C.G., Latzman, R.D. & Kotov, R. Linking RDoC and HiTOP: A new interface for advancing psychiatric nosology and neuroscience. Clinical Psychology Review 86, 102025 (2021). Hewitt, J.K., Silberg, J.L., Neale, M.C., Eaves, L.J. & Erickson, M. The analysis of parental ratings of children's behavior using LISREL. Behavior Genetics 22, 293–317 (1992). Silberg, J.L. et al. The application of structural equation modeling to maternal ratings of twins' behavioral and emotional problems. Journal of Consulting and Clinical Psychology 62, 510–21 (1994). Nikstat, A. & Riemann, R. On the etiology of internalizing and externalizing problem behavior: A twin-family study. PLoS One 15, e0230626 (2020). Karlsson Linnér, R. et al. Multivariate analysis of 1.5 million people identifies genetic associations with traits related to self-regulation and addiction. Nature Neuroscience 24, 1367–1376 (2021). Brick, L.A. et al. Genetic associations among internalizing and externalizing traits with polysubstance use among young adults. medRxiv (2023). Grotzinger, A.D. et al. Genetic architecture of 11 major psychiatric disorders at biobehavioral, functional genomic and molecular genetic levels of analysis. Nature Genetics 54, 548–559 (2022). Allegrini, A.G. et al. The p factor: genetic analyses support a general dimension of psychopathology in childhood and adolescence. Journal of Child Psychology and Psychiatry 61, 30–39 (2020). Lahey, B.B., Van Hulle, C.A., Singh, A.L., Waldman, I.D. & Rathouz, P.J. Higher-order genetic and environmental structure of prevalent forms of child and adolescent psychopathology. Archives of General Psychiatry 68, 181–189 (2011). Pettersson, E., Larsson, H. & Lichtenstein, P. Common psychiatric disorders share the same genetic origin: A multivariate sibling study of the Swedish population. Molecular Psychiatry 21, 717–721 (2016). Selzam, S., Coleman, J.R.I., Caspi, A., Moffitt, T.E. & Plomin, R. A polygenic p factor for major psychiatric disorders. Translational Psychiatry 8, 205 (2018). Neumann, A. et al. A genome-wide association study of total child psychiatric problems scores. PLOS ONE 17, e0273116 (2022). Pappa, I. et al. Single nucleotide polymorphism heritability of behavior problems in childhood: Genome-wide complex trait analysis. Journal of the American Academy of Child and Adolescent Psychiatry 54, 737–44 (2015). Romero, C. et al. Exploring the genetic overlap between twelve psychiatric disorders. Nature Genetics 54, 1795–1802 (2022). Grotzinger, A.D. et al. Genomic structural equation modelling provides insights into the multivariate genetic architecture of complex traits. Nature Human Behaviour 3, 513–525 (2019). Lee, P.H. et al. Genomic relationships, novel loci, and pleiotropic mechanisms across eight psychiatric disorders. Cell 179, 1469–1482.e11 (2019). Demontis, D. et al. Genome-wide analyses of ADHD identify 27 risk loci, refine the genetic architecture and implicate several cognitive domains. Nature Genetics 55, 198–208 (2023). Zhou, H. et al. Multi-ancestry study of the genetics of problematic alcohol use in over 1 million individuals. Nature Medicine (2023). Levey, D.F. et al. Multi-ancestry genome-wide association study of cannabis use disorder yields insight into disease biology and public health implications. Nature Genetics 55, 2094–2103 (2023). Kember, R.L. et al. Cross-ancestry meta-analysis of opioid use disorder uncovers novel loci with predominant effects in brain regions associated with addiction. Nature Neuroscience 25, 1279–1287 (2022). Toikumo, S. et al. Multi-ancestry meta-analysis of tobacco use disorder prioritizes novel candidate risk genes and reveals associations with numerous health outcomes. medRxiv (2023). Karlsson Linnér, R. et al. Genome-wide association analyses of risk tolerance and risky behaviors in over 1 million individuals identify hundreds of loci and shared genetic influences. Nature Genetics 51, 245–257 (2019). Tielbeek, J.J. et al. Genome-wide association studies of a broad spectrum of antisocial behavior. JAMA Psychiatry 74, 1242–1250 (2017). Watson, H.J. et al. Genome-wide association study identifies eight risk loci and implicates metabo-psychiatric origins for anorexia nervosa. Nature Genetics 51, 1207–1214 (2019). Als, T.D. et al. Depression pathophysiology, risk prediction of recurrence and comorbid psychiatric disorders using genome-wide analyses. Nature Medicine 29, 1832–1844 (2023). Stein, M.B. et al. Genome-wide association analyses of post-traumatic stress disorder and its symptom subdomains in the Million Veteran Program. Nature Genetics 53, 174–184 (2021). Abdellaoui, A. et al. Phenome-wide investigation of health outcomes associated with genetic predisposition to loneliness. Human Molecular Genetics 28, 3853–3865 (2019). Okbay, A. et al. Genetic variants associated with subjective well-being, depressive symptoms, and neuroticism identified through genome-wide analyses. Nature Genetics 48, 624–633 (2016). Otowa, T. et al. Meta-analysis of genome-wide association studies of anxiety disorders. Molecular Psychiatry 21, 1391–1399 (2016). Levey, D.F. et al. Reproducible genetic risk loci for anxiety: Results from ∼200,000 participants in the Million Veteran Program. American Journal of Psychiatry 177, 223–232 (2020). Purves, K.L. et al. A major role for common genetic variation in anxiety disorders. Molecular Psychiatry 25, 3292–3303 (2020). Turley, P. et al. Multi-trait analysis of genome-wide association summary statistics using MTAG. Nature Genetics 50, 229–237 (2018). Bulik-Sullivan, B.K. et al. LD Score regression distinguishes confounding from polygenicity in genome-wide association studies. Nature Genetics 47, 291–295 (2015). The International HapMap 3 Consortium. Integrating common and rare genetic variation in diverse human populations. Nature 467, 52 – 8 (2010). The 1000 Genomes Project Consortium. A global reference for human genetic variation. Nature 526, 68–74 (2015). Grotzinger, A.D., Fuente, J.d.l., Privé, F., Nivard, M.G. & Tucker-Drob, E.M. Pervasive downward bias in estimates of liability-scale heritability in genome-wide association study meta-analysis: A simple solution. Biological Psychiatry 93, 29–36 (2023). R Core Team. R: A language and environment for statistical computing. (R Foundation for Statistical Computing, Vienna, Austria, 2023). Ward, J.H. Hierarchical grouping to optimize an objective function. Journal of the American Statistical Association 58, 236–244 (1963). Csardi, G. & Nepusz, T. The igraph software package for complex network research. Interjournal, Complex Systems 1695(2006). Caspi, A. et al. The p Factor: One general psychopathology factor in the structure of psychiatric disorders? Clinical Psychological Science 2, 119–137 (2013). Markon, K.E. Bifactor and hierarchical models: Specification, inference, and interpretation. Annual Review of Clinical Psychology 15, 51–69 (2019). Krueger, R.F. & Markon, K.E. Reinterpreting comorbidity: A model-based approach to understanding and classifying psychopathology. Annual Review of Clinical Psychology 2, 111–133 (2006). Purcell, S. et al. PLINK: a tool set for whole-genome association and population-based linkage analyses. American Journal of Human Genetics 81, 559–575 (2007). de Leeuw, C.A., Mooij, J.M., Heskes, T. & Posthuma, D. MAGMA: Generalized gene set analysis of GWAS data. PLOS Computational Biology 11, e1004219 (2015). Watanabe, K., Taskesen, E., van Bochoven, A. & Posthuma, D. Functional mapping and annotation of genetic associations with FUMA. Nature Communications 8, 1826 (2017). Li, M. et al. Integrative functional genomic analysis of human brain development and neuropsychiatric risks. Science 362, eaat7615 (2018). The GTEx Consortium et al. The GTEx Consortium atlas of genetic regulatory effects across human tissues. Science 369, 1318–1330 (2020). Wang, D. et al. Comprehensive functional genomic resource and integrative model for the human brain. Science 362(2018). Watanabe, K., Umićević Mirkov, M., de Leeuw, C.A., van den Heuvel, M.P. & Posthuma, D. Genetic mapping of cell type specificity for complex traits. Nature Communications 10, 3222 (2019). La Manno, G. et al. Molecular diversity of midbrain development in mouse, human, and stem cells. Cell 167, 566–580.e19 (2016). Habib, N. et al. Massively parallel single-nucleus RNA-seq with DroNc-seq. Nature Methods 14, 955–958 (2017). Darmanis, S. et al. A survey of human brain transcriptome diversity at the single cell level. Proceeding of the National Academy of Science USA 112, 7285-90 (2015). Barbeira, A.N. et al. Integrating predicted transcriptome from multiple tissues improves association detection. PLOS Genetics 15, e1007889 (2019). Barbeira, A.N. et al. Exploring the phenotypic consequences of tissue specific gene expression variation inferred from GWAS summary statistics. Nature Communications 9, 1825 (2018). Gandal, M.J. et al. Transcriptome-wide isoform-level dysregulation in ASD, schizophrenia, and bipolar disorder. Science 362, eaat8127 (2018). Jourdon, A., Scuderi, S., Capauto, D., Abyzov, A. & Vaccarino, F.M. PsychENCODE and beyond: transcriptomics and epigenomics of brain development and organoids. Neuropsychopharmacology 46, 70–85 (2021). Liang, Y. et al. BrainXcan identifies brain features associated with behavioral and psychiatric traits using large scale genetic and imaging data. medRxiv , 2021.06.01.21258159 (2022). Finan, C. et al. The druggable genome and support for target identification and validation in drug development. Science Translational Medicine 9, eaag1166 (2017). Freshour, S.L. et al. Integration of the Drug–Gene Interaction Database (DGIdb 4.0) with open crowdsource efforts. Nucleic Acids Research 49, D1144-D1151 (2021). Frei, O. et al. Bivariate causal mixture model quantifies polygenic overlap between complex traits beyond genetic correlation. Nature Communications 10, 2417 (2019). Holland, D. et al. Beyond SNP heritability: Polygenicity and discoverability of phenotypes estimated with a univariate Gaussian mixture model. PLOS Genetics 16, e1008612 (2020). Cuéllar-Partida, G. et al. Complex-Traits Genetics Virtual Lab: A community-driven web platform for post-GWAS analyses. bioRxiv, 518027 (2019). Zhu, Z. et al. Causal associations between risk factors and common diseases inferred from GWAS summary data. Nature Communications 9, 224 (2018). Waszczuk, M.A. et al. General v. specific vulnerabilities: polygenic risk scores and higher-order psychopathology dimensions in the Adolescent Brain Cognitive Development (ABCD) Study. Psychological Medicine 53, 1937–1946 (2023). Isvoranu, A.M. et al. Extended network analysis: From psychopathology to chronic illness. BMC Psychiatry 21, 119 (2021). Zhang, F. et al. Causal influences of neuroticism on mental health and cardiovascular disease. Human Genetics 140, 1267–1281 (2021). Toikumo, S. et al. The genetic architecture of pain intensity in a sample of 598,339 U.S. veterans. medRxiv (2023). Johnston, K.J.A. et al. Genome-wide association study of multisite chronic pain in UK Biobank. PLOS Genetics 15, e1008164 (2019). Freidin, M.B. et al. Insight into the genetic architecture of back pain and its risk factors from a study of 509,000 individuals. PAIN 160(2019). Zhou, W. et al. Global Biobank Meta-analysis Initiative: Powering genetic discovery across human disease. Cell Genomics 2, 100192 (2022). Hartiala, J.A. et al. Genome-wide analysis identifies novel susceptibility loci for myocardial infarction. European Heart Journal 42, 919–933 (2021). Mahajan, A. et al. Multi-ancestry genetic study of type 2 diabetes highlights the power of diverse populations for discovery and translation. Nature Genetics 54, 560–572 (2022). Liu, Z. et al. Genetic architecture of the inflammatory bowel diseases across East Asian and European ancestries. Nature Genetics 55, 796–806 (2023). Paul, S.E. et al. Phenome-wide investigation of behavioral, environmental, and neural associations with cross-disorder genetic liability in youth of European ancestry. medRxiv , 2023.02.10.23285783 (2023). Loehlin, J.C. The Cholesky approach: A cautionary note. Behavior Genetics 26, 65–69 (1996). Nagel, M. et al. Meta-analysis of genome-wide association studies for neuroticism in 449,484 individuals identifies novel genetic loci and pathways. Nature Genetics 50, 920–927 (2018). Picci, G. et al. Left amygdala structure mediates longitudinal associations between exposure to threat and long-term psychiatric symptomatology in youth. Human Brain Mapping 43, 4091–4102 (2022). Wendt, F.R. et al. Multivariate genome-wide analysis of education, socioeconomic status and brain phenome. Nature Human Behaviour 5, 482–496 (2021). Kraguljac, N.V., Guerreri, M., Strickland, M.J. & Zhang, H. Neurite orientation dispersion and density imaging in psychiatric disorders: A systematic literature review and a technical note. Biological Psychiatry Global Open Science 3, 10–21 (2023). Zald, D.H. & Lahey, B.B. Implications of the hierarchical structure of psychopathology for psychiatric neuroimaging. Bioligical Psychiatry: Cognitive Neuroscience and Neuroimaging 2, 310–317 (2017). Yao, C. et al. Exploring the bidirectional relationship between pain and mental disorders: a comprehensive Mendelian randomization study. The Journal of Headache and Pain 24, 82 (2023). Williams, F.M.K. et al. Causal effects of psychosocial factors on chronic back pain: a bidirectional Mendelian randomisation study. European Spine Journal 31, 1906–1915 (2022). Mulugeta, A., Zhou, A., King, C. & Hyppönen, E. Association between major depressive disorder and multiple disease outcomes: a phenome-wide Mendelian randomisation study in the UK Biobank. Molecular Psychiatry 25, 1469–1476 (2020). Luo, J., Xu, Z., Noordam, R., van Heemst, D. & Li-Gao, R. Depression and inflammatory bowel disease: A bidirectional two-sample Mendelian randomization study. Journal of Crohn's and Colitis 16, 633–642 (2022). He, Y., Chen, C.L., He, J. & Liu, S.D. Causal associations between inflammatory bowel disease and anxiety: A bidirectional Mendelian randomization study. World Journal of Gastroenterology 29, 5872–5881 (2023). Bianchi, D.W. et al. The All of Us Research Program is an opportunity to enhance the diversity of US biomedical research. Nature Medicine 30, 330–333 (2024). Sanchez-Roige, S. & Palmer, A.A. Emerging phenotyping strategies will advance our understanding of psychiatric genetics. Nature Neuroscience 23, 475–480 (2020). Additional Declarations Yes there is potential Competing Interest. Dr. Kranzler is a member of advisory boards for Dicerna Pharmaceuticals, Sophrosyne Pharmaceuticals, Enthion Pharmaceuticals, and Clearmind Medicine; a consultant to Sobrera Pharmaceuticals; the recipient of research funding and medication supplies for an investigator-initiated study from Alkermes; and a member of the American Society of Clinical Psychopharmacology’s Alcohol Clinical Trials Initiative, which was supported in the last three years by Alkermes, Dicerna, Ethypharm, Lundbeck, Mitsubishi, Otsuka, and Pear Therapeutics. Drs. Kranzler and Gelernter hold U.S. patent 10,900,082 titled: "Genotype-guided dosing of opioid agonists," issued 26 January 2021. The other authors have no disclosures to make. Supplementary Files SupplementaryFiguressubmission.docx Supplementary_Figures SupplementaryTablessubmission.xlsx Supplementary_Tables Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-4228593","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":288885885,"identity":"fcc90b9e-7921-4199-b3b3-d09f771b7626","order_by":0,"name":"Christal Davis","email":"data:image/png;base64,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","orcid":"https://orcid.org/0000-0003-3974-5598","institution":"Crescenz VA Medical Center","correspondingAuthor":true,"prefix":"","firstName":"Christal","middleName":"","lastName":"Davis","suffix":""},{"id":288885886,"identity":"e780437c-d648-4458-afda-358a077458af","order_by":1,"name":"Yousef Khan","email":"","orcid":"https://orcid.org/0009-0002-3093-3947","institution":"University of Pennsylvania Perelman School of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Yousef","middleName":"","lastName":"Khan","suffix":""},{"id":288885887,"identity":"c643fb0f-26ed-4dcc-a148-7334bad8c862","order_by":2,"name":"Sylvanus Toikumo","email":"","orcid":"https://orcid.org/0000-0002-6024-0693","institution":"University of Pennsylvania","correspondingAuthor":false,"prefix":"","firstName":"Sylvanus","middleName":"","lastName":"Toikumo","suffix":""},{"id":288885888,"identity":"d926024d-d7b4-49d9-a800-2a075f8ccd17","order_by":3,"name":"Zeal Jinwala","email":"","orcid":"https://orcid.org/0009-0005-5062-0827","institution":"University of Pennsylvania Perelman School of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Zeal","middleName":"","lastName":"Jinwala","suffix":""},{"id":288885889,"identity":"6294ddfb-d7ca-44da-b466-649cee753558","order_by":4,"name":"D Boomsma","email":"","orcid":"https://orcid.org/0000-0002-7099-7972","institution":"Vrije Universiteit Amsterdam, The Netherlands","correspondingAuthor":false,"prefix":"","firstName":"D","middleName":"","lastName":"Boomsma","suffix":""},{"id":288885890,"identity":"8b7f3b10-fc3e-45bc-8f45-2b883d0e129e","order_by":5,"name":"Daniel Levey","email":"","orcid":"https://orcid.org/0000-0001-8431-9569","institution":"Yale School of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Daniel","middleName":"","lastName":"Levey","suffix":""},{"id":288885891,"identity":"ef662dd4-bf13-466b-ab0f-63cc189d4516","order_by":6,"name":"Joel Gelernter","email":"","orcid":"https://orcid.org/0000-0002-4067-1859","institution":"Yale University","correspondingAuthor":false,"prefix":"","firstName":"Joel","middleName":"","lastName":"Gelernter","suffix":""},{"id":288885892,"identity":"cb440573-b960-4b27-9f0d-a0de926888b1","order_by":7,"name":"Rachel Kember","email":"","orcid":"https://orcid.org/0000-0001-8820-2659","institution":"University of Pennsylvania","correspondingAuthor":false,"prefix":"","firstName":"Rachel","middleName":"","lastName":"Kember","suffix":""},{"id":288885893,"identity":"b09f6070-0149-4425-9176-d5da71fe6cc5","order_by":8,"name":"Henry Kranzler","email":"","orcid":"https://orcid.org/0000-0002-1018-0450","institution":"University of Pennsylvania Perelman School of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Henry","middleName":"","lastName":"Kranzler","suffix":""}],"badges":[],"createdAt":"2024-04-06 19:40:10","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4228593/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4228593/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":54316006,"identity":"b4608acf-cbdb-40c1-977e-298cbde01f45","added_by":"auto","created_at":"2024-04-08 17:46:15","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":112312,"visible":true,"origin":"","legend":"\u003cp\u003eExploratory analyses of internalizing and externalizing traits.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ea)\u003c/strong\u003e network analysis results where colors represent clusters, the size of nodes indicates centrality, and the width of the edges between nodes represents the genetic correlation between traits, \u003cstrong\u003eb) \u003c/strong\u003eagglomerative hierarchical clustering using Ward’s criterion to identify clusters, and \u003cstrong\u003ec) \u003c/strong\u003egenetic correlations for the sixteen traits that were retained for Genomic Structural Equation Modeling. PTSD = posttraumatic stress disorder, TUD = tobacco use disorder, CanUD = cannabis use disorder, AUD = alcohol use disorder, OUD = opioid use disorder, ADHD = attention deficit hyperactivity disorder, AgeSex = age at first sexual intercourse (reverse coded), ASB = antisocial behavior, NumSex = number of sexual partners. Wellbeing is reverse coded.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eFigure 2. \u003c/em\u003eConfirmatory factor analysis models used for Genomic Structural Equation Modeling.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-4228593/v1/7ad66de79c739f93ad6fb582.png"},{"id":54316003,"identity":"b4985c49-035f-447c-ae92-11b883aaef19","added_by":"auto","created_at":"2024-04-08 17:46:15","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":127729,"visible":true,"origin":"","legend":"\u003cp\u003eConfirmatory factor analysis models used for Genomic Structural Equation Modeling.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ea)\u003c/strong\u003e results of the correlated factors model. Model fit: X\u003csup\u003e2\u003c/sup\u003e(97) = 3877.82, AIC = 3955.82, CFI = 0.91, SRMR = 0.09, \u003cstrong\u003eb\u003c/strong\u003e) results of the higher order factor model. Model fit: X\u003csup\u003e2\u003c/sup\u003e(97) = 3877.82, AIC = 3955.82, CFI = 0.91, SRMR = 0.09. EXT = externalizing, INT = internalizing, ADHD = attention deficit hyperactivity disorder, AgeSex = age at first sexual intercourse (reverse coded), NumSex = number of sexual partners, ASB = antisocial behavior, AUD = alcohol use disorder, CanUD = cannabis use disorder, OUD = opioid use disorder, TUD = tobacco use disorder, SWB = subjective wellbeing, PTSD = posttraumatic stress disorder, MDD = major depressive disorder, ANX = anxiety. Alternative models considered can be found in Supplementary Figures 2 and 3.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-4228593/v1/961d4de1bb32ce7792949c69.png"},{"id":54316509,"identity":"db9967f7-4fce-40b3-956f-6d0ababb5fe2","added_by":"auto","created_at":"2024-04-08 17:54:15","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":550074,"visible":true,"origin":"","legend":"\u003cp\u003eManhattan plots of GWAS results.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ea)\u003c/strong\u003e depicts results for the externalizing (EXT) GWAS, \u003cstrong\u003eb) \u003c/strong\u003efor the internalizing (INT) GWAS, and \u003cstrong\u003ec) \u003c/strong\u003efor the second order EXT + INT GWAS. Green diamonds denote lead SNPs not identified in any of the input GWAS for the spectrum, and yellow diamonds denote lead SNPs not previously identified in association with any of the input traits for a spectrum based on a GWAS Catalog search.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-4228593/v1/5aacfde76b19ffaf4e816f07.png"},{"id":54316008,"identity":"5edb3141-0318-4a5f-a4aa-49c6996b59cf","added_by":"auto","created_at":"2024-04-08 17:46:15","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":62624,"visible":true,"origin":"","legend":"\u003cp\u003eTranscriptome wide association study (TWAS) results for the second order externalizing and internalizing factor across 13 brain tissues using S-MultiXcan.\u003c/p\u003e\n\u003cp\u003eThe gene names for the top 25 significant associations are annotated. Significance was determined using a Bonferroni-adjusted p-value of 3.73 × 10-6 (0.05/13,406 tests). The dashed line at 5.43 indicates the significance level (-log10(3.73 x 10-6). A total of 236 associations were significant after multiple testing correction. Full TWAS results for all the factors can be found in Supplementary Tables 14-19.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-4228593/v1/af1815ff02bd6d6adb588b5b.png"},{"id":54316005,"identity":"e2aec80a-4bdf-4d98-915e-40fe2e22e754","added_by":"auto","created_at":"2024-04-08 17:46:15","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":502793,"visible":true,"origin":"","legend":"\u003cp\u003ePartial results of BrainXcan association analysis for the second order externalizing + internalizing factor.\u003c/p\u003e\n\u003cp\u003eAssociations shown are for image-derived phenotypes from structural (T1) magnetic resonance imaging. Full results from BrainXcan are in Supplementary Figures 15-20 and Supplementary Tables 20-22. Blue colors represent reduced volume, and orange represents increased volume.\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-4228593/v1/9c2d8396032b3ee6c2d86ceb.png"},{"id":54316007,"identity":"c6f8e9f4-ce10-4c28-8aac-517d13744125","added_by":"auto","created_at":"2024-04-08 17:46:15","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":369855,"visible":true,"origin":"","legend":"\u003cp\u003eRepresentative results of generalized summary-data-based Mendelian Randomization (GSMR) across four domains of physical health.\u003c/p\u003e\n\u003cp\u003ePain intensity results: INT: \u003cem\u003eb\u003c/em\u003e\u003csub\u003e\u003cem\u003exy \u003c/em\u003e\u003c/sub\u003e= 0.48, SE = 0.04, \u003cem\u003ep \u003c/em\u003e= 1.11E-28; EXT: \u003cem\u003eb\u003c/em\u003e\u003csub\u003e\u003cem\u003exy \u003c/em\u003e\u003c/sub\u003e= 0.31, SE = 0.01, \u003cem\u003ep \u003c/em\u003e= 1.44E-155. Longstanding illness, disability, or infirmity results: INT: OR = 1.12, SE = 0.01, \u003cem\u003ep \u003c/em\u003e= 9.83E-33; EXT: OR = 1.05, SE = 0.002, \u003cem\u003ep \u003c/em\u003e= 2.27E-75. Myocardial infarction results: INT: OR = 1.27, SE = 0.07, \u003cem\u003ep \u003c/em\u003e= 0.0003; EXT: OR = 1.24, SE = 0.02, \u003cem\u003ep \u003c/em\u003e= 6.32E-38. Type 2 Diabetes results: INT: OR = 1.72, SE = 0.06, \u003cem\u003ep \u003c/em\u003e= 7.28E-20; EXT: OR = 1.19, SE = 0.02, \u003cem\u003ep \u003c/em\u003e= 6.94E-28. All results depicted are significant at a Bonferroni corrected p-value of 0.001. In all analyses, INT/EXT is the exposure, and the physical health trait is the outcome. INT = internalizing, EXT = externalizing, OR = odds ratio. Full results of GSMR analyses, including results for EXT+INT, can be found in Supplementary Table 32.\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-4228593/v1/9b237fb7133e1bfe774a6fa3.png"},{"id":62661871,"identity":"3dd18528-6445-4ecc-a75a-e3364e7f1d90","added_by":"auto","created_at":"2024-08-17 02:53:43","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2427097,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4228593/v1/c4a20a94-c3ff-487e-a9ae-8c3933e69008.pdf"},{"id":54316009,"identity":"cf0ed1da-b768-4c0c-8dab-c9da7a19031e","added_by":"auto","created_at":"2024-04-08 17:46:15","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":40228557,"visible":true,"origin":"","legend":"Supplementary_Figures","description":"","filename":"SupplementaryFiguressubmission.docx","url":"https://assets-eu.researchsquare.com/files/rs-4228593/v1/d88a934a022a3875d834a1ba.docx"},{"id":54316012,"identity":"7875123e-db8d-4528-a857-aec272a7d3fc","added_by":"auto","created_at":"2024-04-08 17:46:21","extension":"xlsx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":46778745,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary_Tables\u003c/p\u003e","description":"","filename":"SupplementaryTablessubmission.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-4228593/v1/1e15ba6446af69d29033ef0b.xlsx"}],"financialInterests":"\u003cb\u003eYes\u003c/b\u003e there is potential Competing Interest.\nDr. Kranzler is a member of advisory boards for Dicerna Pharmaceuticals, Sophrosyne Pharmaceuticals, Enthion Pharmaceuticals, and Clearmind Medicine; a consultant to Sobrera Pharmaceuticals; the recipient of research funding and medication supplies for an investigator-initiated study from Alkermes; and a member of the American Society of Clinical Psychopharmacology’s Alcohol Clinical Trials Initiative, which was supported in the last three years by Alkermes, Dicerna, Ethypharm, Lundbeck, Mitsubishi, Otsuka, and Pear Therapeutics. Drs. Kranzler and Gelernter hold U.S. patent 10,900,082 titled: \"Genotype-guided dosing of opioid agonists,\" issued 26 January 2021. The other authors have no disclosures to make.","formattedTitle":"A Multivariate Genome-Wide Association Study Reveals Neural Correlates and Common Biological Mechanisms of Psychopathology Spectra","fulltext":[{"header":"Introduction","content":"\u003cp\u003eTraditional categorical classifications of psychopathology suffer from significant limitations. In epidemiological studies, psychiatric disorders consistently co-occur more often than expected,\u003csup\u003e1,2\u003c/sup\u003e suggesting overlapping genetic underpinnings.\u003csup\u003e3\u003c/sup\u003e Furthermore, largely arbitrary thresholds and polythetic criterion sets yield thousands of unique symptom combinations that lead to the same diagnosis.\u003csup\u003e4\u003c/sup\u003e Along with the challenges these limitations present for clinical care, they hinder progress in psychiatric genetics and neuroscience research, where efforts to identify biological mechanisms that underlie psychiatric illness have had limited success.\u003csup\u003e5,6\u003c/sup\u003e Recent attempts to address these limitations have included alternative approaches to understanding psychopathology, most notably the Hierarchical Taxonomy of Psychopathology (HiTOP) and the National Institute of Mental Health\u0026rsquo;s Research Domain Criteria (RDoC) initiative.\u003csup\u003e5,7\u0026ndash;9\u003c/sup\u003e HiTOP proposes a dimensional structure of psychopathology that progresses hierarchically from symptoms to an overarching psychopathology factor. In contrast, RDoC aims to identify the underlying mechanisms of psychopathology by focusing on domains of functioning rather than diagnostic categories. Despite differences in the units of analysis and the dimensions they identify, these systems\u0026rsquo; constructs align well in a model of psychopathology.\u003csup\u003e10\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eBeginning in the 1990s, twin and family studies showed that dimensions of psychopathology had a shared genetic basis,\u003csup\u003e11,12\u003c/sup\u003e with externalizing and internalizing psychopathology being the subject of much of this research. Whereas externalizing behaviors involve interaction with the social environment (e.g., aggression, impulsivity), internalizing behaviors are directed inwards (e.g., anxiety, depression).\u003csup\u003e13\u003c/sup\u003e With statistical and methodological advances, molecular genetic research has also identified common externalizing\u003csup\u003e14\u003c/sup\u003e and internalizing\u003csup\u003e15,16\u003c/sup\u003e genetic factors that underlie each spectrum of psychopathology. Twin and family studies\u003csup\u003e17\u0026ndash;19\u003c/sup\u003e and principal component analyses\u003csup\u003e17,20\u003c/sup\u003e have also examined genetic factors shared by \u003cem\u003eboth\u003c/em\u003e externalizing and internalizing psychopathology. Genome-wide association studies (GWAS) of childhood behavior problems, which encompass externalizing and internalizing psychopathology, identified two genome-wide significant loci.\u003csup\u003e21,22\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eRomero and colleagues recently used a cross-trait GWAS meta-analysis to identify pleiotropic genetic effects across 12 psychiatric disorders.\u003csup\u003e23\u003c/sup\u003e Because the meta-analytic signal in that study was driven by schizophrenia, the interpretation and joint biological characterization of the cross-trait signal was limited. Genomic structural equation modeling (gSEM) offers several advantages over cross-trait meta-analysis for identifying the shared genetic architecture that underlies psychopathology. First, gSEM enables specific hypotheses about the factor structure of psychopathology to be tested, with explicit comparison of proposed models that could account for the overlap observed across externalizing and internalizing spectra. Second, the use of latent variables helps to identify the common genetic effects across externalizing and internalizing spectra, minimizing the capture of genetic signals associated with the most dominant trait, as in the meta-analytic study of Romero et al.\u003csup\u003e23\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eOther gSEM studies have investigated the factor structure of psychiatric disorders and identified one to four factors that underlie their shared liability.\u003csup\u003e16,24,25\u003c/sup\u003e A previous GWAS identified two genome-wide hits for a higher-order \u003cem\u003ep\u003c/em\u003e-factor encompassing compulsive, psychotic, internalizing, and neurodevelopmental disorders, and 66 significant hits upon \u003cem\u003epost hoc\u003c/em\u003e examination of a bifactor model \u003cem\u003ep\u003c/em\u003e-factor. Because the study included only psychiatric disorders, it did not capture a broad spectrum of psychopathology consistent with dimensional models like HiTOP. It also included only two internalizing (anxiety and major depressive disorder) and two externalizing (attention-deficit hyperactivity disorder and problematic alcohol use) conditions.\u003c/p\u003e \u003cp\u003eTo conduct a detailed examination of the shared genetic architecture of externalizing and internalizing psychopathology, we applied gSEM to large GWAS summary statistics. Adopting a dimensional, transdiagnostic approach, we first evaluated models of psychopathology to determine which factor structure provided the best fit to the pattern of genetic covariance across 16 externalizing and internalizing traits and disorders. To identify genetic effects for the externalizing spectrum, internalizing spectrum, and \u003cem\u003eacross\u003c/em\u003e the externalizing and internalizing spectra, we conducted GWAS on the latent factors derived from models with acceptable fit. Next, we performed downstream analyses to characterize biological mechanisms underlying the shared genetic liability to psychopathology and to examine potential causal effects on physical health. Identifying mechanisms that account for vulnerability across levels of psychopathology can yield insights into the genetic basis of psychopathology, which could lead to advancements in treatment, diagnosis, and disorder classification.\u003c/p\u003e"},{"header":"Subjects and Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eSummary Statistics\u003c/h2\u003e \u003cp\u003e \u003cem\u003eExternalizing.\u003c/em\u003e Ten sets of summary statistics in European-ancestry (EUR) individuals were selected based on existing theory regarding the externalizing spectrum (Supplementary Table\u0026nbsp;1). We included summary statistics from the largest available GWAS of the following externalizing disorders: attention deficit hyperactivity disorder (ADHD; n\u0026thinsp;=\u0026thinsp;225,534),\u003csup\u003e26\u003c/sup\u003e four substance use disorders [SUDs; i.e., alcohol (AUD; n\u0026thinsp;=\u0026thinsp;753,248),\u003csup\u003e27\u003c/sup\u003e cannabis (CanUD; n\u0026thinsp;=\u0026thinsp;886,025),\u003csup\u003e28\u003c/sup\u003e opioid (OUD; n\u0026thinsp;=\u0026thinsp;425,944),\u003csup\u003e29\u003c/sup\u003e and tobacco (TUD; n\u0026thinsp;=\u0026thinsp;495,005)\u003csup\u003e30\u003c/sup\u003e]. We also included broader measures of externalizing psychopathology [age of first sexual intercourse (AgeSex; reverse-coded; n\u0026thinsp;=\u0026thinsp;317,694),\u003csup\u003e14\u003c/sup\u003e general risk tolerance (Risk; n\u0026thinsp;=\u0026thinsp;431,126),\u003csup\u003e14,31\u003c/sup\u003e number of sexual partners (NumSex; n\u0026thinsp;=\u0026thinsp;370,711),\u003csup\u003e14,31\u003c/sup\u003e antisocial behavior (ASB; n\u0026thinsp;=\u0026thinsp;16,400),\u003csup\u003e32\u003c/sup\u003e and automobile speeding propensity (n\u0026thinsp;=\u0026thinsp;404,291)\u003csup\u003e31\u003c/sup\u003e].\u003c/p\u003e \u003cp\u003e \u003cem\u003eInternalizing.\u003c/em\u003e Summary statistics from eight GWAS in EUR individuals were selected to capture the internalizing spectrum (Supplementary Table\u0026nbsp;1). Three were the largest available GWAS of internalizing disorders: (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e) anorexia nervosa (AN; n\u0026thinsp;=\u0026thinsp;72,517),\u003csup\u003e33\u003c/sup\u003e (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e) major depressive disorder (MDD; n\u0026thinsp;=\u0026thinsp;1,074,629),\u003csup\u003e34\u003c/sup\u003e and (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e) posttraumatic stress disorder (PTSD; n\u0026thinsp;=\u0026thinsp;214,408).\u003csup\u003e35\u003c/sup\u003e To reflect a broad liability to internalizing, we included irritability (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.nealelab.is/uk-biobank/\u003c/span\u003e\u003cspan address=\"http://www.nealelab.is/uk-biobank/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e; n\u0026thinsp;=\u0026thinsp;345,231), loneliness (n\u0026thinsp;=\u0026thinsp;490,689),\u003csup\u003e36\u003c/sup\u003e subjective wellbeing (SWB; reverse-coded; n\u0026thinsp;=\u0026thinsp;298,420),\u003csup\u003e37\u003c/sup\u003e miserableness (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.nealelab.is/uk-biobank/\u003c/span\u003e\u003cspan address=\"http://www.nealelab.is/uk-biobank/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e; n\u0026thinsp;=\u0026thinsp;355,182), and anxiety (ANX; n\u0026thinsp;=\u0026thinsp;280,490).\u003csup\u003e38\u0026ndash;40 38\u0026ndash;41\u003c/sup\u003e To boost power to detect variants associated with both anxiety disorders and subclinical anxiety, we combined three anxiety GWAS\u003csup\u003e38\u0026ndash;40\u003c/sup\u003e using multi-trait analysis of GWAS.\u003csup\u003e41\u003c/sup\u003e\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003eExploratory Analysis\u003c/h2\u003e \u003cp\u003e \u003cem\u003eGenetic Correlations.\u003c/em\u003e Using linkage disequilibrium score regression (LDSC)\u003csup\u003e42\u003c/sup\u003e in GenomicSEM,\u003csup\u003e24\u003c/sup\u003e we calculated genetic correlations (\u003cem\u003er\u003c/em\u003e\u003csub\u003eg\u003c/sub\u003e) between the input traits. Single nucleotide polymorphisms (SNPs) were filtered using EUR HapMap3 reference panels,\u003csup\u003e43\u003c/sup\u003e and SNPs with MAF\u0026thinsp;\u0026lt;\u0026thinsp;0.01 were removed. LDSC was performed using ancestry-matched 1000 Genomes Phase 3 linkage disequilibrium (LD) scores.\u003csup\u003e44\u003c/sup\u003e When available in the summary statistics, SNP-level sample sizes were specified. Otherwise, the effective sample size was calculated by summing effective sample sizes across the input GWAS cohorts.\u003csup\u003e45\u003c/sup\u003e After conducting LDSC, genetic correlations were inspected to identify traits having weak associations with the other input traits prior to fitting structural equation models. Traits with average \u003cem\u003er\u003c/em\u003e\u003csub\u003eg\u003c/sub\u003e \u0026lt; 0.20 were excluded from gSEM models.\u003c/p\u003e \u003cp\u003e \u003cem\u003eHierarchical Cluster Analysis.\u003c/em\u003e To evaluate whether traits clustered with their predicted spectrum, we conducted hierarchical cluster analysis of the \u003cem\u003er\u003c/em\u003e\u003csub\u003eg\u003c/sub\u003e matrix using the hclust() function in RStudio.\u003csup\u003e46\u003c/sup\u003e We calculated a Euclidean distance matrix and used Ward\u0026rsquo;s agglomerative clustering algorithm\u003csup\u003e47\u003c/sup\u003e to identify clusters. A plot of the within-cluster sum of squares was used to determine the optimal number of clusters.\u003c/p\u003e \u003cp\u003e \u003cem\u003eNetwork Analysis.\u003c/em\u003e A network analysis of the \u003cem\u003er\u003c/em\u003e\u003csub\u003eg\u003c/sub\u003e matrix was performed using the \u003cem\u003eigraph\u003c/em\u003e package in RStudio.\u003csup\u003e48\u003c/sup\u003e The matrix was transformed into an undirected and weighted network graph, in which nodes represent each trait and the weights of the links between traits represent the magnitude of their genetic correlation. The optimal network community structure was determined by maximizing modularity, a measure of the quality of a clustering solution.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003eGenomic Structural Equation Modeling\u003c/h2\u003e \u003cp\u003eWe fit four confirmatory factor analyses (CFAs) based on existing theories of psychopathology.\u003csup\u003e49\u0026ndash;51\u003c/sup\u003e First, we evaluated a correlated factors model with two factors representing externalizing and internalizing psychopathology. Next, we fit a bifactor model consisting of a general psychopathology factor on which all traits loaded, and two narrower externalizing and internalizing psychopathology factors. A higher-order model was also fit with two first-order factors representing externalizing and internalizing, and a second-order factor (EXT\u0026thinsp;+\u0026thinsp;INT) representing genetic effects shared by the two spectra. Finally, we fit a unidimensional, or \u003cem\u003ep\u003c/em\u003e-factor, model where all traits loaded onto a single latent factor. In all CFA models, the residuals of the four SUDs were allowed to correlate; all other residuals were uncorrelated. We evaluated the models with chi-square, the Akaike information criterion (AIC), comparative fit index (CFI), and standardized root mean squared residual (SRMR) fit statistics. We also inspected the results for low (\u0026lt;\u0026thinsp;0.35) or negative loadings as an indicator of each model\u0026rsquo;s appropriateness.\u003c/p\u003e \u003cp\u003eWhen preparing the data for GWAS, we excluded SNPs with MAF\u0026thinsp;\u0026lt;\u0026thinsp;0.01 and imputation scores\u0026thinsp;\u0026lt;\u0026thinsp;0.6. Coefficients and standard error values were standardized across summary statistics to ensure that effects were scaled similarly for all traits. Each SNP was regressed on the model latent variable(s) using diagonally weighted least squares estimation. After performing GWAS, we calculated factor-specific Q\u003csub\u003eSNP\u003c/sub\u003e values by comparing the fit of a common pathway model to an independent pathway model.\u003csup\u003e16\u003c/sup\u003e Q\u003csub\u003eSNP\u003c/sub\u003e provides a measure of SNP heterogeneity, reflecting the extent to which a SNP exerts effects entirely through the common factor (i.e., common pathway model) or, instead, exerts effects differentially across a factor\u0026rsquo;s indicators (i.e., independent pathway model). SNPs with a Q chi-square p-value\u0026thinsp;\u0026lt;\u0026thinsp;5x10\u003csup\u003e\u0026minus;\u0026thinsp;8\u003c/sup\u003e were filtered prior to conducting all subsequent analyses. Finally, to identify lead SNPs for each factor, we performed LD clumping in PLINK 1.9\u003csup\u003e52\u003c/sup\u003e using a range of 3 Mb and \u003cem\u003er\u003c/em\u003e\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;\u0026gt;\u0026thinsp;0.10 with the EUR 1000 Genomes Phase 3 reference panel.\u003csup\u003e44\u003c/sup\u003e\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003eBiological Characterization\u003c/h2\u003e \u003cp\u003eGene-based tests, gene-set enrichment, and gene-tissue expression analyses were conducted using MAGMA\u003csup\u003e53\u003c/sup\u003e in FUMA v1.6.0\u003csup\u003e54\u003c/sup\u003e. We examined gene expression in BrainSpan\u003csup\u003e55\u003c/sup\u003e and GTEx v8\u003csup\u003e56\u003c/sup\u003e tissue samples. In FUMA, gene associations were identified based on their: (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e) position, (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e) expression quantitative trait loci (eQTLs) from PsychENCODE\u003csup\u003e57\u003c/sup\u003e and GTEx v8\u003csup\u003e56\u003c/sup\u003e brain tissue samples, and (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e) chromatin interactions using Hi-C data from the dorsolateral prefrontal cortex, hippocampus, ventricles, and neural progenitor cells. We also analyzed gene expression at the cellular level in single-cell RNA sequencing (scRNA-seq) datasets from 15 human brain cell expression profiles.\u003csup\u003e58\u0026ndash;61\u003c/sup\u003e For these analyses, we used a three-step approach: (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e) conducting gene-property analyses within each dataset, (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e) identifying independent associations using within-dataset conditional analyses, and (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e) identifying independent clusters of signals using cross-datasets conditional analyses.\u003csup\u003e58\u003c/sup\u003e\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003eTranscriptome Wide Association Studies\u003c/h2\u003e \u003cp\u003eWe conducted two transcriptome-wide association studies (TWAS) on each factor.\u003csup\u003e62\u003c/sup\u003e First, we used S-MultiXcan\u003csup\u003e62\u003c/sup\u003e, which prioritizes likely causal genes by jointly predicting gene expression across multiple tissues. S-MultiXcan produces an overall Z-score and p-value across all tissues, as well as values for the most and least associated tissues. We examined expression across the 13 brain tissues in GTEx v8\u003csup\u003e56\u003c/sup\u003e and identified the most associated tissue for each gene. To complement this approach, we used S-PrediXcan\u003csup\u003e63\u003c/sup\u003e and weights trained on transcriptional differences in the frontal and temporal cortices of psychiatric cases and controls\u003csup\u003e64\u003c/sup\u003e from PsychENCODE.\u003csup\u003e65\u003c/sup\u003e A Bonferroni correction was applied to identify significant associations.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eAssociations with Brain Phenotypes\u003c/h2\u003e \u003cp\u003eWe used BrainXcan\u003csup\u003e66\u003c/sup\u003e to examine associations between the psychopathology spectra and 327 brain image-derived phenotypes (IDPs) from structural (T1-weighted) and diffusion magnetic resonance images (dMRIs) using ridge regression. Effect sizes and p-values were adjusted using LD block-based permutation, and Bonferroni correction was used to account for multiple testing (T1: 0.05/109\u0026thinsp;=\u0026thinsp;4.59 x 10\u003csup\u003e\u0026minus;\u0026thinsp;4\u003c/sup\u003e; dMRI\u0026thinsp;=\u0026thinsp;0.05/218\u0026thinsp;=\u0026thinsp;2.29 x 10\u003csup\u003e\u0026minus;\u0026thinsp;4\u003c/sup\u003e). We also conducted bidirectional Mendelian randomization (MR) analyses for the most significantly associated brain IDPs. Because the significance of the IDP-factor association was used to identify pairs on which to perform MR, the resulting MR p-values were used to discern the possible direction of association, rather than to evaluate significance.\u003csup\u003e66\u003c/sup\u003e\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003eDrug Repurposing\u003c/h2\u003e \u003cp\u003eTo identify gene targets for drug repurposing, we used five different gene annotation approaches: (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e) MAGMA,\u003csup\u003e53\u003c/sup\u003e (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e) chromatin interactions, (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e) eQTL, (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e) S-MultiXcan, and (\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e) S-PrediXcan.\u003csup\u003e62,63\u003c/sup\u003e To avoid unreliable associations, we queried the subset of druggable genes\u003csup\u003e67\u003c/sup\u003e identified by multiple biological annotation sources using the Drug-Gene-Interaction Database (DGIdb).\u003csup\u003e68\u003c/sup\u003e For the first-order factors, we limited drug repurposing analyses to genes that showed specificity of association with \u003cem\u003eeither\u003c/em\u003e externalizing or internalizing.\u003c/p\u003e \u003cdiv id=\"Sec10\" class=\"Section3\"\u003e \u003ch2\u003eCausal Mixture Models (MiXeR)\u003c/h2\u003e \u003cp\u003eUnivariate MiXeR analyses\u003csup\u003e69\u003c/sup\u003e were conducted to estimate each factor\u0026rsquo;s polygenicity (i.e., the number of causal variants required to explain 90% SNP heritability) and discoverability (i.e., the average effect size of causal variants).\u003csup\u003e70\u003c/sup\u003e Next, bivariate models were used to identify the proportion of unique and shared causal variants for the externalizing and internalizing spectra. In contrast to genetic correlations, MiXeR accounts for polygenic overlap regardless of whether causal variants have the same or opposite direction of effect.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eGenetic Correlations\u003c/h2\u003e \u003cp\u003eUsing the Complex Trait Genetics Virtual Lab\u003csup\u003e71\u003c/sup\u003e, we calculated batch genetic correlations between each factor GWAS and 1,437 phenotypes from publicly available GWAS. GWAS that were used as an input for the gSEM models were excluded. Genetic correlations were calculated using LDSC\u003csup\u003e42\u003c/sup\u003e and EUR 1000 Genomes Phase 3 data\u003csup\u003e44\u003c/sup\u003e as LD references. To account for multiple testing, a Benjamini-Hochberg false discovery rate (FDR) correction was applied to each set of genetic correlations.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eGeneralized Summary-data-based Mendelian Randomization\u003c/h2\u003e \u003cp\u003eTo evaluate potentially causal impacts of externalizing and internalizing genetic risk on physical health, we conducted Generalized Summary-data-based Mendelian Randomization (GSMR)\u003csup\u003e72\u003c/sup\u003e using 15 health traits as outcomes. Traits were chosen across four domains\u0026mdash;(\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e) pain, (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e) general health, (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e) cardiovascular disease, and (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e) chronic illness\u0026mdash;each of which has demonstrated associations with psychopathology.\u003csup\u003e73\u0026ndash;75\u003c/sup\u003e We used summary statistics from GWAS of pain intensity,\u003csup\u003e76\u003c/sup\u003e multisite chronic pain,\u003csup\u003e77\u003c/sup\u003e and back pain.\u003csup\u003e78\u003c/sup\u003e General health indices included GWAS summary statistics from the UK Biobank for longstanding illness, disability, or infirmity; hospitalization; and age at death. We selected five cardiovascular GWAS: (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e) heart failure,\u003csup\u003e79\u003c/sup\u003e (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e) stroke,\u003csup\u003e79\u003c/sup\u003e (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e) myocardial infarction,\u003csup\u003e80\u003c/sup\u003e (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e) hypertension (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.nealelab.is/uk-biobank/\u003c/span\u003e\u003cspan address=\"http://www.nealelab.is/uk-biobank/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e), and (\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e) abdominal aortic aneurysm.\u003csup\u003e79\u003c/sup\u003e Finally, we selected four GWAS of chronic illnesses: (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e) type 2 diabetes,\u003csup\u003e81\u003c/sup\u003e (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e) inflammatory bowel disease (IBD),\u003csup\u003e82\u003c/sup\u003e (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e) chronic obstructive pulmonary disease (COPD),\u003csup\u003e79\u003c/sup\u003e and (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e) asthma.\u003csup\u003e79\u003c/sup\u003e Genetic instruments with significant pleiotropic effects on both the exposure and outcome were removed using the heterogeneity in dependent instruments outlier (HEIDI) method.\u003csup\u003e72\u003c/sup\u003e A Bonferroni adjusted p-value was applied to identify significant effects (0.05/45\u0026thinsp;=\u0026thinsp;0.001).\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cp\u003eBased on previous GWAS of externalizing and internalizing\u003csup\u003e14,15,83\u003c/sup\u003e and existing theory,\u003csup\u003e7,9\u003c/sup\u003e we considered a total of 18 externalizing and internalizing traits for inclusion in the analyses (Supplementary Fig.\u0026nbsp;1). We excluded two traits (automobile speeding propensity and anorexia nervosa) that were weakly associated with the others in the model (mean \u003cem\u003er\u003c/em\u003e\u003csub\u003eg\u003c/sub\u003e \u0026lt; 0.20). Network analysis and hierarchical agglomerative cluster analysis both revealed two clusters that correspond to externalizing and internalizing spectra (see Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eUsing the 16 traits, we tested several CFA models (see Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e and Supplementary Figs.\u0026nbsp;2 and 3). A general psychopathology factor model did not provide adequate fit to the data (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\chi\\)\u003c/span\u003e\u003c/span\u003e\u003csup\u003e2\u003c/sup\u003e(98)\u0026thinsp;=\u0026thinsp;8965.28, AIC\u0026thinsp;=\u0026thinsp;9041.28, CFI\u0026thinsp;=\u0026thinsp;0.79, and SRMR\u0026thinsp;=\u0026thinsp;0.15), although standardized loadings were all significant and \u0026gt;\u0026thinsp;0.35. A bifactor model comprising a general psychopathology factor and two specific factors representing externalizing and internalizing spectra fit the data well (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\chi\\)\u003c/span\u003e\u003c/span\u003e\u003csup\u003e2\u003c/sup\u003e(\u003cspan citationid=\"CR81\" class=\"CitationRef\"\u003e81\u003c/span\u003e)\u0026thinsp;=\u0026thinsp;2527.19, AIC\u0026thinsp;=\u0026thinsp;2637.19, CFI\u0026thinsp;=\u0026thinsp;0.94, and SRMR\u0026thinsp;=\u0026thinsp;0.05). However, the model led to several weak (\u0026lt;\u0026thinsp;0.35) and one negative standardized loading, possibly from overfitting the data. Thus, despite its good fit, we did not perform GWAS on factors from this model because they would be difficult to interpret. Two CFA models provided adequate fit and had strong factor loadings: (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e) a correlated-factors model and (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e) a higher-order factor model. Fit statistics for both models were \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\chi\\)\u003c/span\u003e\u003c/span\u003e\u003csup\u003e2\u003c/sup\u003e(97)\u0026thinsp;=\u0026thinsp;3877.82, AIC\u0026thinsp;=\u0026thinsp;3955.82, CFI\u0026thinsp;=\u0026thinsp;0.91, and SRMR\u0026thinsp;=\u0026thinsp;0.09. To ensure identification in the higher order model, the loadings of externalizing and internalizing onto the second-order factor were constrained equal to the square root of the genetic correlation between the externalizing and internalizing factors.\u003csup\u003e84\u003c/sup\u003e\u003c/p\u003e\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eMultivariate GWAS of Psychopathology Spectra\u003c/h2\u003e \u003cp\u003eUsing a Q\u003csub\u003eSNP\u003c/sub\u003e analysis, 228 independent SNPs exhibited heterogeneous effects across the externalizing spectrum (Supplementary Fig.\u0026nbsp;4). Among the associations of these SNPs within the input GWAS (Supplementary Table\u0026nbsp;2), a plurality was most strongly associated with age at first sexual intercourse (37.23%), followed by TUD (23.38%). After filtering heterogenous SNPs, a multivariate GWAS of externalizing identified 409 GWS independent lead SNPs (Supplementary Table\u0026nbsp;3). Of these, 92 (22.49%) were not identified or within \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\pm\\)\u003c/span\u003e\u003c/span\u003e1000 kb of SNPs identified by any of the input GWAS, and four were not previously associated with any externalizing trait using the same threshold. Three of the four novel SNPs were on chromosome 4 (rs1961547, rs9316, and rs7682762), with the fourth on chromosome 22 (rs1473811). These SNPs showed phenotypic associations with chronotype, schizophrenia, and social support, among other traits (Supplementary Table\u0026nbsp;4).\u003c/p\u003e \u003cp\u003eFor internalizing, 222 independent SNPs exhibited significant heterogeneity (Supplementary Fig.\u0026nbsp;5), with most (86.49%) showing the strongest associations with MDD (Supplementary Table\u0026nbsp;5). After filtering heterogeneous SNPs, there were 85 GWS independent lead SNPs (Supplementary Table\u0026nbsp;6). Of these, 23 (27.06%) were not identified or within \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\pm\\)\u003c/span\u003e\u003c/span\u003e1000 kb of SNPs identified by the input GWAS, and two were not previously associated with any internalizing trait. The two novel associations were on chromosomes 3 and 4 (rs1381763 and rs4698408, respectively). Novel SNPs had phenotypic associations with neuroticism and depression (Supplementary Table\u0026nbsp;7).\u003c/p\u003e \u003cp\u003eIn the GWAS of genetic effects shared across externalizing and internalizing (EXT\u0026thinsp;+\u0026thinsp;INT factor), there were 130 lead SNPs that exhibited heterogeneous effects (Supplementary Fig.\u0026nbsp;6). Of these (Supplementary Table\u0026nbsp;8), a plurality (47.69%) was most strongly associated with age at first sexual intercourse, followed by AUD (17.69%). There were 256 GWS independent lead SNPs associated with EXT\u0026thinsp;+\u0026thinsp;INT (Supplementary Table\u0026nbsp;9), 38 of which (14.84%) were not identified or within \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\pm\\)\u003c/span\u003e\u003c/span\u003e1000 kb of SNPs identified by any of the input GWAS. All significant loci were previously associated with at least one externalizing or internalizing phenotype (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eBiological Characterization\u003c/h2\u003e \u003cp\u003eMAGMA identified 493 genes significantly associated with externalizing, including \u003cem\u003eCADM2\u003c/em\u003e and \u003cem\u003eDRD2\u003c/em\u003e (Supplementary Table\u0026nbsp;10). Gene-property analysis showed enriched expression during prenatal brain development (Supplementary Fig.\u0026nbsp;7). Gene expression was significantly enhanced in almost all brain tissues, with the top associations being with the cerebellar hemisphere, cerebellum, and frontal cortex (Supplementary Fig.\u0026nbsp;8). A gene-set related to mRNA binding was the only significant association with externalizing (\u003cem\u003ep\u003c/em\u003e\u003csub\u003ebon\u003c/sub\u003e = 0.04; Supplementary Table\u0026nbsp;11). Using scRNA-seq datasets, externalizing was significantly associated with dopaminergic and GABAergic neurons and neuroblasts from embryonic brain samples (GSE76381), human cortical neurons and hybrid cells that display characteristics of neurons and astrocytes (GSE67835), and pyramidal neurons from the cornu ammonis (CA1) hippocampal region\u003csup\u003e60\u003c/sup\u003e. After conditional analyses, independent significant associations remained for GABAergic neurons, cortical neurons, and hippocampal neurons (Supplementary Fig.\u0026nbsp;9).\u003c/p\u003e \u003cp\u003eThere were 146 genes significantly associated with internalizing, including several on chromosome 8 (\u003cem\u003eBLK\u003c/em\u003e, \u003cem\u003eXKR6\u003c/em\u003e, and \u003cem\u003eC8orf12\u003c/em\u003e) that were previously associated with neuroticism (Supplementary Table\u0026nbsp;12).\u003csup\u003e85\u003c/sup\u003e Gene expression was not significantly enhanced at any developmental stage (Supplementary Fig.\u0026nbsp;7), but predominated in the brain, with the frontal cortex and anterior cingulate cortex most strongly associated (Supplementary Fig.\u0026nbsp;8). Although no gene-sets were significant after Bonferroni correction, the top associations were with genes involved in synaptic assembly and transmission (Supplementary Table\u0026nbsp;11). In scRNA-seq analyses, the only significant cell-type association was with GABAergic neurons (GSE76381), which was not independently significant after conditional analyses.\u003c/p\u003e \u003cp\u003eThere were 321 genes significantly associated with EXT\u0026thinsp;+\u0026thinsp;INT (Supplementary Table\u0026nbsp;13). The top hits were for \u003cem\u003eFAM120AOS\u003c/em\u003e, \u003cem\u003eDCC\u003c/em\u003e, and \u003cem\u003eP4HTM\u003c/em\u003e, all of which have previously been associated with both internalizing and externalizing traits. Gene expression was enhanced in brain tissue during prenatal developmental (Supplementary Fig.\u0026nbsp;7), and genes associated with the broad spectrum of psychopathology were predominantly expressed in the brain (Supplementary Fig.\u0026nbsp;8). No gene sets were significant after Bonferroni correction, but like internalizing, the top sets comprised genes involved in synaptic activity (Supplementary Table\u0026nbsp;11). Using scRNA-seq, EXT\u0026thinsp;+\u0026thinsp;INT showed significant associations with dopaminergic neurons (GSE76381), GABAergic neurons (GSE76381), and cortical neurons (GSE67835), though these associations were not independently significant.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eTranscriptome-Wide Association Analysis\u003c/h2\u003e \u003cp\u003eUsing S-MultiXcan to predict the effects of SNP variation on gene expression across 13 brain tissues revealed 352 significant gene associations for externalizing, 141 for internalizing, and 238 for EXT\u0026thinsp;+\u0026thinsp;INT (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e, Supplementary Tables\u0026nbsp;14\u0026ndash;16, and Supplementary Fig.\u0026nbsp;10). TWAS using PsychENCODE data for S-PrediXcan identified 207 significant genes for externalizing, 52 for internalizing, and 124 for EXT\u0026thinsp;+\u0026thinsp;INT (Supplementary Tables\u0026nbsp;17\u0026ndash;19 and Supplementary Fig.\u0026nbsp;11). Forty-five genes were identified by both S-MultiXcan and S-PrediXcan for externalizing, 21 for internalizing, and 36 for EXT\u0026thinsp;+\u0026thinsp;INT, with gene-property analyses showing these genes to be consistently upregulated across brain tissues (Supplementary Figs.\u0026nbsp;12\u0026ndash;14), and three (\u003cem\u003eC1QTNF4\u003c/em\u003e, \u003cem\u003eDPYSL5\u003c/em\u003e, and \u003cem\u003eSLC12A5\u003c/em\u003e) were almost exclusively upregulated in brain tissues.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eAssociations with Brain Phenotypes\u003c/h2\u003e \u003cp\u003eAfter Bonferroni correction of the LD-adjusted p-values, 8 T1 (Supplementary Fig.\u0026nbsp;15) and 12 dMRI IDPs (Supplementary Fig.\u0026nbsp;16) were significantly associated with externalizing (Supplementary Table\u0026nbsp;20), including positive associations with gray matter volume in the thalamus, caudate nuclei, and occipital pole, and negative associations with the right ventral striatum and left amygdala. From dMRIs, there were significant associations with intra-cellular volume fraction or orientation dispersion indices (ODI) in the medial lemniscus, cerebral peduncle, and middle cerebellar peduncle, among others. The only significant association for internalizing was with lower gray matter volume in the left subcallosal cortex (Supplementary Table\u0026nbsp;21 and Supplementary Figs.\u0026nbsp;17\u0026ndash;18). Genetic factors shared across externalizing and internalizing spectra were significantly negatively associated with gray matter volume in the left amygdala and left subcallosal cortex (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e), positively associated with ODI in the medial lemniscus and left cerebellar peduncle, and negatively associated with ODI in the right external capsule (Supplementary Table\u0026nbsp;22 and Supplementary Figs.\u0026nbsp;19\u0026ndash;20).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eMR analyses showed potential bidirectional relationships between externalizing and gray matter volume in the right ventral striatum and left thalamus (Supplementary Table\u0026nbsp;23). There was greater evidence that reduced gray matter volume in the left subcallosal cortex was causally related to internalizing than vice versa (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.008 vs. 0.401; Supplementary Table\u0026nbsp;24). Evidence was mixed regarding the direction of causal effects for the second-order externalizing and internalizing factor (Supplementary Table\u0026nbsp;25).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003eDrug Repurposing\u003c/h2\u003e \u003cp\u003eAmong the 1,759 unique genes identified for externalizing using biological annotation tools, 308 were druggable targets.\u003csup\u003e67\u003c/sup\u003e Sixty of these genes were identified by at least two biological annotation tools, and 52 exhibited specificity for externalizing (i.e., were not associated with internalizing). When queried in DGIdb, these genes yielded 492 drug-gene interactions (Supplementary Table\u0026nbsp;26), including with dextroamphetamine (used to treat ADHD), phenobarbital (used to prevent withdrawal symptoms from benzodiazepines and alcohol), baclofen (used to treat AUD), naltrexone (used to treat AUD and OUD), naloxone (used to reverse opioid overdose), and methadone (used to treat OUD). Gene interactions with antimigraine, anti-inflammatory, and anticonvulsant drugs (e.g., topiramate and lamotrigine) were also identified. Most of the identified drugs had regulatory approval (64.84%).\u003c/p\u003e \u003cp\u003eBiological annotation identified 454 unique genes associated with internalizing, 60 of which were druggable targets.\u003csup\u003e67\u003c/sup\u003e Fifteen of these were identified by at least two biological annotation tools and seven exhibited specificity for internalizing, yielding 292 drug-gene interactions (Supplementary Table\u0026nbsp;27). Drug targets included antidepressants and antipsychotics. Unlike externalizing, most identified drugs (82.33%) were not currently approved, suggesting potential candidates for use in treating internalizing psychopathology.\u003c/p\u003e \u003cp\u003eFor EXT\u0026thinsp;+\u0026thinsp;INT 1,138 unique genes were identified using the five biological annotation tools, nearly one-fifth (17.93%, n\u0026thinsp;=\u0026thinsp;204) of which were druggable targets, with 47 of those implicated by more than one biological annotation method. Using DGIdb, we identified 460 unique drug-gene interactions (Supplementary Table\u0026nbsp;28), many of which were also present in the internalizing or externalizing results. As with internalizing, most of these drugs (75.52%) were not currently approved.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003eCausal Mixture Models (MiXeR)\u003c/h2\u003e \u003cp\u003eThe externalizing and internalizing spectra displayed similar levels of polygenicity, with an estimated 12,600 and 13,200 causal variants, respectively. However, internalizing had lower discoverability (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\({\\widehat{\\sigma }}_{\\beta }\\)\u003c/span\u003e\u003c/span\u003e\u003csup\u003e2\u003c/sup\u003e = 1.40 x 10\u003csup\u003e\u0026minus;\u0026thinsp;5\u003c/sup\u003e) than externalizing (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\({\\widehat{\\sigma }}_{\\beta }\\)\u003c/span\u003e\u003c/span\u003e\u003csup\u003e2\u003c/sup\u003e = 1.44 x 10\u003csup\u003e\u0026minus;\u0026thinsp;4\u003c/sup\u003e), suggesting that SNPs that influence internalizing traits may exert smaller effects and thus require larger samples to detect. Despite a MiXeR-estimated genetic correlation of 0.37, almost all causal variants (96.83% of externalizing and 92.42% of internalizing; Supplementary Fig.\u0026nbsp;21) overlapped across the two spectra, with more overlap than predicted by genetic correlation alone (AIC\u0026thinsp;=\u0026thinsp;12.30, BIC\u0026thinsp;=\u0026thinsp;4.06, where positive values indicate that the predicted model explains the data better than the genetic correlation alone). In fact, the models do not exclude the possibility that causal variants for externalizing were a subset of those for internalizing. Of the shared causal variants, 62.92% were estimated to be concordant in direction of effect.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003eGenetic Correlations\u003c/h2\u003e \u003cp\u003eApplying a Bonferroni-adjusted p-value (0.05/1368\u0026thinsp;=\u0026thinsp;3.65 x 10\u003csup\u003e\u0026minus;\u0026thinsp;5\u003c/sup\u003e), there were 413 significant genetic correlations with externalizing (Supplementary Table\u0026nbsp;29 and Supplementary Fig.\u0026nbsp;22). Tobacco phenotypes were among the most strongly correlated (current smoking: \u003cem\u003er\u003c/em\u003e\u003csub\u003eg\u003c/sub\u003e = 0.79, SE\u0026thinsp;=\u0026thinsp;0.02; maternal smoking around birth: \u003cem\u003er\u003c/em\u003e\u003csub\u003eg\u003c/sub\u003e = 0.71, SE\u0026thinsp;=\u0026thinsp;0.03; and ever smoked: \u003cem\u003er\u003c/em\u003e\u003csub\u003eg\u003c/sub\u003e = 0.62, SE\u0026thinsp;=\u0026thinsp;0.02), along with lower socioeconomic status, including Townsend deprivation index (\u003cem\u003er\u003c/em\u003e\u003csub\u003eg\u003c/sub\u003e = 0.68, SE\u0026thinsp;=\u0026thinsp;0.03), living in housing supplied by a local authority or housing association (\u003cem\u003er\u003c/em\u003e\u003csub\u003eg\u003c/sub\u003e = 0.66, SE\u0026thinsp;=\u0026thinsp;0.03), experiencing financial difficulties (\u003cem\u003er\u003c/em\u003e\u003csub\u003eg\u003c/sub\u003e = 0.58, SE\u0026thinsp;=\u0026thinsp;0.03), and lower educational attainment (\u003cem\u003er\u003c/em\u003e\u003csub\u003eg\u003c/sub\u003e = -0.44, SE\u0026thinsp;=\u0026thinsp;0.02). After Bonferroni correction, 311 phenotypes were significantly genetically correlated with internalizing (Supplementary Table\u0026nbsp;30 and Supplementary Fig.\u0026nbsp;23). Among the strongest correlations were psychiatric phenotypes, such as mood swings (\u003cem\u003er\u003c/em\u003e\u003csub\u003eg\u003c/sub\u003e = 0.90, SE\u0026thinsp;=\u0026thinsp;0.01), neuroticism (\u003cem\u003er\u003c/em\u003e\u003csub\u003eg\u003c/sub\u003e = 0.89, SE\u0026thinsp;=\u0026thinsp;0.01), and feeling fed-up (\u003cem\u003er\u003c/em\u003e\u003csub\u003eg\u003c/sub\u003e = 0.82, SE\u0026thinsp;=\u0026thinsp;0.01). Internalizing was also significantly genetically correlated with several types of pain (abdominal: \u003cem\u003er\u003c/em\u003e\u003csub\u003eg\u003c/sub\u003e = 0.60, SE\u0026thinsp;=\u0026thinsp;0.04; facial: \u003cem\u003er\u003c/em\u003e\u003csub\u003eg\u003c/sub\u003e = 0.51, SE\u0026thinsp;=\u0026thinsp;0.08; chest: \u003cem\u003er\u003c/em\u003e\u003csub\u003eg\u003c/sub\u003e = 0.49, SE\u0026thinsp;=\u0026thinsp;0.03; and multisite chronic pain: \u003cem\u003er\u003c/em\u003e\u003csub\u003eg\u003c/sub\u003e = 0.49, SE\u0026thinsp;=\u0026thinsp;0.03, among others). There were 474 significant genetic correlations with EXT\u0026thinsp;+\u0026thinsp;INT, with most being like the first-order factors (Supplementary Table\u0026nbsp;31 and Supplementary Fig.\u0026nbsp;24).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003eGeneralized Summary-data-based Mendelian Randomization\u003c/h2\u003e \u003cp\u003eExternalizing had significant positive causal effects on all physical health traits examined, except age at death and IBD. Internalizing was significantly causally associated with all traits except age at death and abdominal aortic aneurysm. Additionally, internalizing had protective effects on IBD (\u003cem\u003eb\u003c/em\u003e\u003csub\u003e\u003cem\u003exy\u003c/em\u003e\u003c/sub\u003e = -0.32, SE\u0026thinsp;=\u0026thinsp;0.09, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.0004) and stronger positive associations than externalizing with all three pain phenotypes, all five cardiovascular diseases, and three of four chronic illnesses (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e). Like externalizing, EXT\u0026thinsp;+\u0026thinsp;INT had significant positive effects on all physical health traits except age at death and IBD (Supplementary Table\u0026nbsp;32).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eComparing candidate factor structures of psychopathology, we found support for hierarchical models consistent with the HiTOP framework. Our models, which included symptom-level (e.g., risk tolerance and irritability) and disorder-level (e.g., TUD and MDD) manifestations of psychopathology, indicated that these traits could be organized onto higher dimensions representing externalizing and internalizing spectra, which were themselves subsumed under a broader umbrella of psychopathology genetic risk. Leveraging GWAS summary statistics that included over 1.5\u0026nbsp;million individuals, our findings also show that although there is shared variance across forms of psychopathology, the commonality does not manifest as a single overarching dimension (i.e., \u003cem\u003ep\u003c/em\u003e-factor). Rather, the genetic architecture of psychopathology was better captured by a model that distinguished between specific dimensions while recognizing their interrelatedness.\u003c/p\u003e \u003cp\u003eOur findings also demonstrated connections between psychopathology and RDoC domains\u003csup\u003e10\u003c/sup\u003e in downstream analyses that encompassed multiple RDoC units of analysis. For example, at the cellular level, externalizing-related variants were associated with RNA expression in pyramidal hippocampal neurons, which are implicated in RDoC\u0026rsquo;s cognitive control construct. For internalizing, analyses revealed molecular-level associations with drugs targeting serotonin and dopamine, which align with RDoC\u0026rsquo;s negative valence systems domain. Integrating HiTOP\u0026rsquo;s dimensional framework with RDoC\u0026rsquo;s etiological approach makes it possible to evaluate the extent to which psychopathology spectra show etiologically consistent associations across domains of functioning and units of analysis.\u003c/p\u003e \u003cp\u003eAlthough externalizing and internalizing were better represented as correlated but distinct factors and had only a small-to-moderate genetic correlation, bivariate causal mixture models estimated extensive overlap in causal variants between the two spectra. This pleiotropy suggests that there are shared biological pathways that underpin externalizing and internalizing psychopathology. Through GWAS and downstream analyses, we identified potential neural mechanisms of this shared risk. Broad psychopathology genetic risk was associated with structural and functional differences in the brain, including reduced gray matter volume in the left amygdala and subcallosal cortex. Reduced left amygdala volume has been shown to mediate the relation between childhood threat exposure and the development of externalizing and internalizing symptoms.\u003csup\u003e86\u003c/sup\u003e Similarly, reduced left subcallosal cortical volume is a potential mediator of associations between personality and various emotional states (e.g., subjective well-being) and psychiatric disorders (e.g., alcohol dependence).\u003csup\u003e87\u003c/sup\u003e Psychopathology liability was also related to alterations in the organization and microstructure of white matter fibers, reflected by changes in orientation dispersion indices, which may signify disruptions in neural communication that contribute to various manifestations of psychopathology.\u003csup\u003e88\u003c/sup\u003e In summary, we identified neural mechanisms that operate at varying levels of specificity,\u003csup\u003e89\u003c/sup\u003e with some correlates linked specifically to externalizing or internalizing and others linked to broad psychopathology liability.\u003c/p\u003e \u003cp\u003eExtending these findings beyond mental health, MR analyses showed that liability to psychopathology exerted potentially causal effects on adverse physical health outcomes. Of note, internalizing generally showed stronger associations with pain, general health, cardiovascular disease, and chronic illness than externalizing. Previous research also supports likely causal effects of genetic risk for internalizing traits/disorders on localized pain\u003csup\u003e90,91\u003c/sup\u003e and various disease outcomes.\u003csup\u003e92\u003c/sup\u003e An unexpected finding here was that internalizing was protective for IBD after removing pleiotropic variants. Although some studies demonstrated causal effects of MDD on risk for IBD,\u003csup\u003e93\u003c/sup\u003e studies of other internalizing disorders (e.g., anxiety) have not.\u003csup\u003e94\u003c/sup\u003e Thus, more research is needed to disentangle the complex relations of internalizing and inflammatory, autoimmune conditions such as IBD. Finally, although the impact of internalizing liability on physical health was particularly pronounced, both externalizing and broad psychopathology liability also showed potentially causal effects across physical health domains. These findings underscore the contribution of psychopathology liability to the emergence of co-occurring physical health conditions.\u003c/p\u003e \u003cp\u003eA limitation of the present study is its inclusion of only European-ancestry individuals. Data adequate to explore a broad liability to internalizing and externalizing disorders and subclinical features in other ancestry groups were not available. Although data from GWAS of externalizing and internalizing-related \u003cem\u003edisorders\u003c/em\u003e are available for some other non-European ancestry groups, more precise, \u003cem\u003enon-disorder\u003c/em\u003e psychiatric phenotypes are limited in these populations. Nonetheless, some research suggests that a similar factor structure applies in African-ancestry individuals. At the disorder level, a gSEM of African-ancestry individuals (unpublished data) identified substance use and psychiatric disorder factors that roughly aligned with externalizing and internalizing, respectively. Additionally, the analyses showed that a higher-order factor accounted for genetic variance shared by substance use and psychiatric disorders, which to a degree corresponds to our EXT\u0026thinsp;+\u0026thinsp;INT factor. With the growth of diverse biobanks\u003csup\u003e95\u003c/sup\u003e and deep phenotyping\u003csup\u003e96\u003c/sup\u003e, more direct replication in other ancestries should soon be possible and given high priority.\u003c/p\u003e \u003cp\u003eIn conclusion, our findings supported a hierarchical structure of psychopathology, recognizing correlated externalizing and internalizing dimensions subsumed under a broader psychopathology liability. By discerning genetic variants and neural mechanisms operating at varying levels of specificity, the findings revealed the utility of applying a dimensional, hierarchical genetic approach to investigate psychopathology, which could augment existing categorical frameworks. Developing a more nuanced understanding of underlying biological mechanisms across forms of psychopathology could aid in constructing a more precise and comprehensive psychiatric nosology, providing a foundation on which to improve treatment and clinical outcomes.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgments:\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by the Veterans Integrated Service Network 4 Mental Illness Research, Education and Clinical Center and by Department of Veterans Affairs grants I01 BX004820 to H.R.K., National Institute on Alcohol Abuse and Alcoholism grant AA028292 to R.L.K, and KNAW (Royal Netherlands Academy of Arts and Sciences) Academy Professor Award (PAH/6635) to D.I.B. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflicts of Interest:\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eDr. Kranzler is a member of advisory boards for Dicerna Pharmaceuticals, Sophrosyne Pharmaceuticals, Enthion Pharmaceuticals, and Clearmind Medicine; a consultant to Sobrera Pharmaceuticals; the recipient of research funding and medication supplies for an investigator-initiated study from Alkermes; and a member of the American Society of Clinical Psychopharmacology\u0026rsquo;s Alcohol Clinical Trials Initiative, which was supported in the last three years by Alkermes, Dicerna, Ethypharm, Lundbeck, Mitsubishi, Otsuka, and Pear Therapeutics. Drs. Kranzler and Gelernter hold U.S. patent 10,900,082 titled: \u0026quot;Genotype-guided dosing of opioid agonists,\u0026quot; issued 26 January 2021. The other authors have no disclosures to make.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eKessler, R.C., Chiu, W.T., Demler, O. \u0026amp; Walters, E.E. Prevalence, severity, and comorbidity of 12-month DSM-IV disorders in the National Comorbidity Survey replication. Archives of General Psychiatry 62, 617\u0026ndash;627 (2005).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKessler, R.C. \u003cem\u003eet al.\u003c/em\u003e Lifetime and 12-month prevalence of DSM-III-R psychiatric disorders in the United States: Results from the National Comorbidity Survey. Archives of General Psychiatry 51, 8\u0026ndash;19 (1994).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLee, P.H., Feng, Y.-C.A. \u0026amp; Smoller, J.W. Pleiotropy and cross-disorder genetics among psychiatric disorders. Biological Psychiatry 89, 20\u0026ndash;31 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGalatzer-Levy, I.R. \u0026amp; Bryant, R.A. 636,120 ways to have posttraumatic stress disorder. Perspectives on Psychological Science 8, 651\u0026ndash;62 (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKozak, M.J. \u0026amp; Cuthbert, B.N. The NIMH Research Domain Criteria Initiative: Background, issues, and pragmatics. Psychophysiology 53, 286\u0026ndash;297 (2016).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWaszczuk, M.A. \u003cem\u003eet al.\u003c/em\u003e Redefining phenotypes to advance psychiatric genetics: Implications from hierarchical taxonomy of psychopathology. Journal of Abnormal Psychology 129, 143\u0026ndash;161 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKotov, R. \u003cem\u003eet al.\u003c/em\u003e The Hierarchical Taxonomy of Psychopathology (HiTOP): A dimensional alternative to traditional nosologies. Journal of Abnormal Psychology 126, 454\u0026ndash;477 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCuthbert, B.N. Research Domain Criteria: Toward future psychiatric nosologies. Dialogues in Clinical Neuroscience 17, 89\u0026ndash;97 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKotov, R. \u003cem\u003eet al.\u003c/em\u003e The Hierarchical Taxonomy of Psychopathology (HiTOP): A quantitative nosology based on consensus of evidence. Annual Review of Clinical Psychology 17, 83\u0026ndash;108 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMichelini, G., Palumbo, I.M., DeYoung, C.G., Latzman, R.D. \u0026amp; Kotov, R. Linking RDoC and HiTOP: A new interface for advancing psychiatric nosology and neuroscience. Clinical Psychology Review 86, 102025 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHewitt, J.K., Silberg, J.L., Neale, M.C., Eaves, L.J. \u0026amp; Erickson, M. The analysis of parental ratings of children's behavior using LISREL. Behavior Genetics 22, 293\u0026ndash;317 (1992).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSilberg, J.L. \u003cem\u003eet al.\u003c/em\u003e The application of structural equation modeling to maternal ratings of twins' behavioral and emotional problems. Journal of Consulting and Clinical Psychology 62, 510\u0026ndash;21 (1994).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNikstat, A. \u0026amp; Riemann, R. On the etiology of internalizing and externalizing problem behavior: A twin-family study. PLoS One 15, e0230626 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKarlsson Linn\u0026eacute;r, R. \u003cem\u003eet al.\u003c/em\u003e Multivariate analysis of 1.5 million people identifies genetic associations with traits related to self-regulation and addiction. Nature Neuroscience 24, 1367\u0026ndash;1376 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBrick, L.A. \u003cem\u003eet al.\u003c/em\u003e Genetic associations among internalizing and externalizing traits with polysubstance use among young adults. medRxiv (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGrotzinger, A.D. \u003cem\u003eet al.\u003c/em\u003e Genetic architecture of 11 major psychiatric disorders at biobehavioral, functional genomic and molecular genetic levels of analysis. Nature Genetics 54, 548\u0026ndash;559 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAllegrini, A.G. \u003cem\u003eet al.\u003c/em\u003e The p factor: genetic analyses support a general dimension of psychopathology in childhood and adolescence. Journal of Child Psychology and Psychiatry 61, 30\u0026ndash;39 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLahey, B.B., Van Hulle, C.A., Singh, A.L., Waldman, I.D. \u0026amp; Rathouz, P.J. Higher-order genetic and environmental structure of prevalent forms of child and adolescent psychopathology. Archives of General Psychiatry 68, 181\u0026ndash;189 (2011).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePettersson, E., Larsson, H. \u0026amp; Lichtenstein, P. Common psychiatric disorders share the same genetic origin: A multivariate sibling study of the Swedish population. Molecular Psychiatry 21, 717\u0026ndash;721 (2016).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSelzam, S., Coleman, J.R.I., Caspi, A., Moffitt, T.E. \u0026amp; Plomin, R. A polygenic p factor for major psychiatric disorders. Translational Psychiatry 8, 205 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNeumann, A. \u003cem\u003eet al.\u003c/em\u003e A genome-wide association study of total child psychiatric problems scores. PLOS ONE 17, e0273116 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePappa, I. \u003cem\u003eet al.\u003c/em\u003e Single nucleotide polymorphism heritability of behavior problems in childhood: Genome-wide complex trait analysis. Journal of the American Academy of Child and Adolescent Psychiatry 54, 737\u0026ndash;44 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRomero, C. \u003cem\u003eet al.\u003c/em\u003e Exploring the genetic overlap between twelve psychiatric disorders. Nature Genetics 54, 1795\u0026ndash;1802 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGrotzinger, A.D. \u003cem\u003eet al.\u003c/em\u003e Genomic structural equation modelling provides insights into the multivariate genetic architecture of complex traits. Nature Human Behaviour 3, 513\u0026ndash;525 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLee, P.H. \u003cem\u003eet al.\u003c/em\u003e Genomic relationships, novel loci, and pleiotropic mechanisms across eight psychiatric disorders. Cell 179, 1469\u0026ndash;1482.e11 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDemontis, D. \u003cem\u003eet al.\u003c/em\u003e Genome-wide analyses of ADHD identify 27 risk loci, refine the genetic architecture and implicate several cognitive domains. Nature Genetics 55, 198\u0026ndash;208 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhou, H. \u003cem\u003eet al.\u003c/em\u003e Multi-ancestry study of the genetics of problematic alcohol use in over 1 million individuals. Nature Medicine (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLevey, D.F. \u003cem\u003eet al.\u003c/em\u003e Multi-ancestry genome-wide association study of cannabis use disorder yields insight into disease biology and public health implications. Nature Genetics 55, 2094\u0026ndash;2103 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKember, R.L. \u003cem\u003eet al.\u003c/em\u003e Cross-ancestry meta-analysis of opioid use disorder uncovers novel loci with predominant effects in brain regions associated with addiction. Nature Neuroscience 25, 1279\u0026ndash;1287 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eToikumo, S. \u003cem\u003eet al.\u003c/em\u003e Multi-ancestry meta-analysis of tobacco use disorder prioritizes novel candidate risk genes and reveals associations with numerous health outcomes. medRxiv (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKarlsson Linn\u0026eacute;r, R. \u003cem\u003eet al.\u003c/em\u003e Genome-wide association analyses of risk tolerance and risky behaviors in over 1 million individuals identify hundreds of loci and shared genetic influences. Nature Genetics 51, 245\u0026ndash;257 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTielbeek, J.J. \u003cem\u003eet al.\u003c/em\u003e Genome-wide association studies of a broad spectrum of antisocial behavior. JAMA Psychiatry 74, 1242\u0026ndash;1250 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWatson, H.J. \u003cem\u003eet al.\u003c/em\u003e Genome-wide association study identifies eight risk loci and implicates metabo-psychiatric origins for anorexia nervosa. Nature Genetics 51, 1207\u0026ndash;1214 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAls, T.D. \u003cem\u003eet al.\u003c/em\u003e Depression pathophysiology, risk prediction of recurrence and comorbid psychiatric disorders using genome-wide analyses. Nature Medicine 29, 1832\u0026ndash;1844 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStein, M.B. \u003cem\u003eet al.\u003c/em\u003e Genome-wide association analyses of post-traumatic stress disorder and its symptom subdomains in the Million Veteran Program. Nature Genetics 53, 174\u0026ndash;184 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAbdellaoui, A. \u003cem\u003eet al.\u003c/em\u003e Phenome-wide investigation of health outcomes associated with genetic predisposition to loneliness. Human Molecular Genetics 28, 3853\u0026ndash;3865 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOkbay, A. \u003cem\u003eet al.\u003c/em\u003e Genetic variants associated with subjective well-being, depressive symptoms, and neuroticism identified through genome-wide analyses. Nature Genetics 48, 624\u0026ndash;633 (2016).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOtowa, T. \u003cem\u003eet al.\u003c/em\u003e Meta-analysis of genome-wide association studies of anxiety disorders. Molecular Psychiatry 21, 1391\u0026ndash;1399 (2016).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLevey, D.F. \u003cem\u003eet al.\u003c/em\u003e Reproducible genetic risk loci for anxiety: Results from \u0026sim;200,000 participants in the Million Veteran Program. American Journal of Psychiatry 177, 223\u0026ndash;232 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePurves, K.L. \u003cem\u003eet al.\u003c/em\u003e A major role for common genetic variation in anxiety disorders. Molecular Psychiatry 25, 3292\u0026ndash;3303 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTurley, P. \u003cem\u003eet al.\u003c/em\u003e Multi-trait analysis of genome-wide association summary statistics using MTAG. Nature Genetics 50, 229\u0026ndash;237 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBulik-Sullivan, B.K. \u003cem\u003eet al.\u003c/em\u003e LD Score regression distinguishes confounding from polygenicity in genome-wide association studies. Nature Genetics 47, 291\u0026ndash;295 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eThe International HapMap 3 Consortium. Integrating common and rare genetic variation in diverse human populations. \u003cem\u003eNature\u003c/em\u003e 467, 52\u0026thinsp;\u0026ndash;\u0026thinsp;8 (2010).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eThe 1000 Genomes Project Consortium. A global reference for human genetic variation. Nature 526, 68\u0026ndash;74 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGrotzinger, A.D., Fuente, J.d.l., Priv\u0026eacute;, F., Nivard, M.G. \u0026amp; Tucker-Drob, E.M. Pervasive downward bias in estimates of liability-scale heritability in genome-wide association study meta-analysis: A simple solution. Biological Psychiatry 93, 29\u0026ndash;36 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eR Core Team. R: A language and environment for statistical computing. (R Foundation for Statistical Computing, Vienna, Austria, 2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWard, J.H. Hierarchical grouping to optimize an objective function. Journal of the American Statistical Association 58, 236\u0026ndash;244 (1963).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCsardi, G. \u0026amp; Nepusz, T. The igraph software package for complex network research. Interjournal, Complex Systems 1695(2006).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCaspi, A. \u003cem\u003eet al.\u003c/em\u003e The p Factor: One general psychopathology factor in the structure of psychiatric disorders? Clinical Psychological Science 2, 119\u0026ndash;137 (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMarkon, K.E. Bifactor and hierarchical models: Specification, inference, and interpretation. Annual Review of Clinical Psychology 15, 51\u0026ndash;69 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKrueger, R.F. \u0026amp; Markon, K.E. Reinterpreting comorbidity: A model-based approach to understanding and classifying psychopathology. Annual Review of Clinical Psychology 2, 111\u0026ndash;133 (2006).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePurcell, S. \u003cem\u003eet al.\u003c/em\u003e PLINK: a tool set for whole-genome association and population-based linkage analyses. American Journal of Human Genetics 81, 559\u0026ndash;575 (2007).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ede Leeuw, C.A., Mooij, J.M., Heskes, T. \u0026amp; Posthuma, D. MAGMA: Generalized gene set analysis of GWAS data. \u003cem\u003ePLOS Computational Biology\u003c/em\u003e 11, e1004219 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWatanabe, K., Taskesen, E., van Bochoven, A. \u0026amp; Posthuma, D. Functional mapping and annotation of genetic associations with FUMA. Nature Communications 8, 1826 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi, M. \u003cem\u003eet al.\u003c/em\u003e Integrative functional genomic analysis of human brain development and neuropsychiatric risks. Science 362, eaat7615 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eThe GTEx Consortium \u003cem\u003eet al.\u003c/em\u003e The GTEx Consortium atlas of genetic regulatory effects across human tissues. Science 369, 1318\u0026ndash;1330 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang, D. \u003cem\u003eet al.\u003c/em\u003e Comprehensive functional genomic resource and integrative model for the human brain. Science 362(2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWatanabe, K., Umićević Mirkov, M., de Leeuw, C.A., van den Heuvel, M.P. \u0026amp; Posthuma, D. Genetic mapping of cell type specificity for complex traits. Nature Communications 10, 3222 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLa Manno, G. \u003cem\u003eet al.\u003c/em\u003e Molecular diversity of midbrain development in mouse, human, and stem cells. Cell 167, 566\u0026ndash;580.e19 (2016).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHabib, N. \u003cem\u003eet al.\u003c/em\u003e Massively parallel single-nucleus RNA-seq with DroNc-seq. Nature Methods 14, 955\u0026ndash;958 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDarmanis, S. \u003cem\u003eet al.\u003c/em\u003e A survey of human brain transcriptome diversity at the single cell level. \u003cem\u003eProceeding of the National Academy of Science USA\u003c/em\u003e 112, 7285-90 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBarbeira, A.N. \u003cem\u003eet al.\u003c/em\u003e Integrating predicted transcriptome from multiple tissues improves association detection. PLOS Genetics 15, e1007889 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBarbeira, A.N. \u003cem\u003eet al.\u003c/em\u003e Exploring the phenotypic consequences of tissue specific gene expression variation inferred from GWAS summary statistics. Nature Communications 9, 1825 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGandal, M.J. \u003cem\u003eet al.\u003c/em\u003e Transcriptome-wide isoform-level dysregulation in ASD, schizophrenia, and bipolar disorder. Science 362, eaat8127 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJourdon, A., Scuderi, S., Capauto, D., Abyzov, A. \u0026amp; Vaccarino, F.M. PsychENCODE and beyond: transcriptomics and epigenomics of brain development and organoids. Neuropsychopharmacology 46, 70\u0026ndash;85 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiang, Y. \u003cem\u003eet al.\u003c/em\u003e BrainXcan identifies brain features associated with behavioral and psychiatric traits using large scale genetic and imaging data. \u003cem\u003emedRxiv\u003c/em\u003e, 2021.06.01.21258159 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFinan, C. \u003cem\u003eet al.\u003c/em\u003e The druggable genome and support for target identification and validation in drug development. Science Translational Medicine 9, eaag1166 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFreshour, S.L. \u003cem\u003eet al.\u003c/em\u003e Integration of the Drug\u0026ndash;Gene Interaction Database (DGIdb 4.0) with open crowdsource efforts. Nucleic Acids Research 49, D1144-D1151 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFrei, O. \u003cem\u003eet al.\u003c/em\u003e Bivariate causal mixture model quantifies polygenic overlap between complex traits beyond genetic correlation. Nature Communications 10, 2417 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHolland, D. \u003cem\u003eet al.\u003c/em\u003e Beyond SNP heritability: Polygenicity and discoverability of phenotypes estimated with a univariate Gaussian mixture model. PLOS Genetics 16, e1008612 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCu\u0026eacute;llar-Partida, G. \u003cem\u003eet al.\u003c/em\u003e Complex-Traits Genetics Virtual Lab: A community-driven web platform for post-GWAS analyses. bioRxiv, 518027 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhu, Z. \u003cem\u003eet al.\u003c/em\u003e Causal associations between risk factors and common diseases inferred from GWAS summary data. Nature Communications 9, 224 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWaszczuk, M.A. \u003cem\u003eet al.\u003c/em\u003e General v. specific vulnerabilities: polygenic risk scores and higher-order psychopathology dimensions in the Adolescent Brain Cognitive Development (ABCD) Study. Psychological Medicine 53, 1937\u0026ndash;1946 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIsvoranu, A.M. \u003cem\u003eet al.\u003c/em\u003e Extended network analysis: From psychopathology to chronic illness. BMC Psychiatry 21, 119 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang, F. \u003cem\u003eet al.\u003c/em\u003e Causal influences of neuroticism on mental health and cardiovascular disease. Human Genetics 140, 1267\u0026ndash;1281 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eToikumo, S. \u003cem\u003eet al.\u003c/em\u003e The genetic architecture of pain intensity in a sample of 598,339 U.S. veterans. \u003cem\u003emedRxiv\u003c/em\u003e (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJohnston, K.J.A. \u003cem\u003eet al.\u003c/em\u003e Genome-wide association study of multisite chronic pain in UK Biobank. PLOS Genetics 15, e1008164 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFreidin, M.B. \u003cem\u003eet al.\u003c/em\u003e Insight into the genetic architecture of back pain and its risk factors from a study of 509,000 individuals. PAIN 160(2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhou, W. \u003cem\u003eet al.\u003c/em\u003e Global Biobank Meta-analysis Initiative: Powering genetic discovery across human disease. Cell Genomics 2, 100192 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHartiala, J.A. \u003cem\u003eet al.\u003c/em\u003e Genome-wide analysis identifies novel susceptibility loci for myocardial infarction. European Heart Journal 42, 919\u0026ndash;933 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMahajan, A. \u003cem\u003eet al.\u003c/em\u003e Multi-ancestry genetic study of type 2 diabetes highlights the power of diverse populations for discovery and translation. Nature Genetics 54, 560\u0026ndash;572 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu, Z. \u003cem\u003eet al.\u003c/em\u003e Genetic architecture of the inflammatory bowel diseases across East Asian and European ancestries. Nature Genetics 55, 796\u0026ndash;806 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePaul, S.E. \u003cem\u003eet al.\u003c/em\u003e Phenome-wide investigation of behavioral, environmental, and neural associations with cross-disorder genetic liability in youth of European ancestry. \u003cem\u003emedRxiv\u003c/em\u003e, 2023.02.10.23285783 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLoehlin, J.C. The Cholesky approach: A cautionary note. Behavior Genetics 26, 65\u0026ndash;69 (1996).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNagel, M. \u003cem\u003eet al.\u003c/em\u003e Meta-analysis of genome-wide association studies for neuroticism in 449,484 individuals identifies novel genetic loci and pathways. Nature Genetics 50, 920\u0026ndash;927 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePicci, G. \u003cem\u003eet al.\u003c/em\u003e Left amygdala structure mediates longitudinal associations between exposure to threat and long-term psychiatric symptomatology in youth. Human Brain Mapping 43, 4091\u0026ndash;4102 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWendt, F.R. \u003cem\u003eet al.\u003c/em\u003e Multivariate genome-wide analysis of education, socioeconomic status and brain phenome. Nature Human Behaviour 5, 482\u0026ndash;496 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKraguljac, N.V., Guerreri, M., Strickland, M.J. \u0026amp; Zhang, H. Neurite orientation dispersion and density imaging in psychiatric disorders: A systematic literature review and a technical note. Biological Psychiatry Global Open Science 3, 10\u0026ndash;21 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZald, D.H. \u0026amp; Lahey, B.B. Implications of the hierarchical structure of psychopathology for psychiatric neuroimaging. Bioligical Psychiatry: Cognitive Neuroscience and Neuroimaging 2, 310\u0026ndash;317 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYao, C. \u003cem\u003eet al.\u003c/em\u003e Exploring the bidirectional relationship between pain and mental disorders: a comprehensive Mendelian randomization study. The Journal of Headache and Pain 24, 82 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWilliams, F.M.K. \u003cem\u003eet al.\u003c/em\u003e Causal effects of psychosocial factors on chronic back pain: a bidirectional Mendelian randomisation study. European Spine Journal 31, 1906\u0026ndash;1915 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMulugeta, A., Zhou, A., King, C. \u0026amp; Hypp\u0026ouml;nen, E. Association between major depressive disorder and multiple disease outcomes: a phenome-wide Mendelian randomisation study in the UK Biobank. Molecular Psychiatry 25, 1469\u0026ndash;1476 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLuo, J., Xu, Z., Noordam, R., van Heemst, D. \u0026amp; Li-Gao, R. Depression and inflammatory bowel disease: A bidirectional two-sample Mendelian randomization study. Journal of Crohn's and Colitis 16, 633\u0026ndash;642 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHe, Y., Chen, C.L., He, J. \u0026amp; Liu, S.D. Causal associations between inflammatory bowel disease and anxiety: A bidirectional Mendelian randomization study. World Journal of Gastroenterology 29, 5872\u0026ndash;5881 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBianchi, D.W. \u003cem\u003eet al.\u003c/em\u003e The All of Us Research Program is an opportunity to enhance the diversity of US biomedical research. Nature Medicine 30, 330\u0026ndash;333 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSanchez-Roige, S. \u0026amp; Palmer, A.A. Emerging phenotyping strategies will advance our understanding of psychiatric genetics. Nature Neuroscience 23, 475\u0026ndash;480 (2020).\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-4228593/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4228593/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThere is considerable comorbidity across externalizing and internalizing behavior dimensions of psychopathology. We applied genomic structural equation modeling (gSEM) to genome-wide association study (GWAS) summary statistics to evaluate the factor structure of externalizing and internalizing psychopathology across 16 traits and disorders among European-ancestry individuals (n\u0026rsquo;s\u0026thinsp;=\u0026thinsp;16,400 to 1,074,629). We conducted GWAS on factors derived from well-fitting models. Downstream analyses served to identify biological mechanisms, explore drug repurposing targets, estimate genetic overlap between the externalizing and internalizing spectra, and evaluate causal effects of psychopathology liability on physical health. Both a correlated factors model, comprising two factors of externalizing and internalizing risk, and a higher-order single-factor model of genetic effects contributing to both spectra demonstrated acceptable fit. GWAS identified 409 lead single nucleotide polymorphisms (SNPs) associated with externalizing and 85 lead SNPs associated with internalizing, while the second-order GWAS identified 256 lead SNPs contributing to broad psychopathology risk. In bivariate causal mixture models, nearly all externalizing and internalizing causal variants overlapped, despite a genetic correlation of only 0.37 (SE\u0026thinsp;=\u0026thinsp;0.02) between them. Externalizing genes showed cell-type specific expression in GABAergic, cortical, and hippocampal neurons, and internalizing genes were associated with reduced subcallosal cortical volume, providing insight into the neurobiological underpinnings of psychopathology. Genetic liability for externalizing, internalizing, and broad psychopathology exerted causal effects on pain, general health, cardiovascular diseases, and chronic illnesses. These findings underscore the complex genetic architecture of psychopathology, identify potential biological pathways for the externalizing and internalizing spectra, and highlight the physical health burden of psychiatric comorbidity.\u003c/p\u003e","manuscriptTitle":"A Multivariate Genome-Wide Association Study Reveals Neural Correlates and Common Biological Mechanisms of Psychopathology Spectra","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-04-08 17:46:10","doi":"10.21203/rs.3.rs-4228593/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"3b33ac70-0d39-4975-b0b7-2a106a18730b","owner":[],"postedDate":"April 8th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":30416209,"name":"Social science/Psychology/Human behaviour"},{"id":30416210,"name":"Biological sciences/Genetics/Genetic association study/Genome-wide association studies"},{"id":30416211,"name":"Biological sciences/Psychology/Human behaviour"}],"tags":[],"updatedAt":"2024-09-24T14:40:35+00:00","versionOfRecord":[],"versionCreatedAt":"2024-04-08 17:46:10","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-4228593","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4228593","identity":"rs-4228593","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}