Abstract
Over 90% of drug candidates fail in clinical trials, while it takes 10–15 years and one billion US dollars to develop a single successful drug. Drug development is more challenging for psychiatric disorders, where disease comorbidity and complex symptom profiles obscure the identification of causal mechanisms for therapeutic intervention. One promising approach for determining more suitable drug candidates in clinical trials is integrating human genetic data into the selection process. Genome-wide association studies have identified thousands of replicable risk loci for psychiatric disorders, and sophisticated statistical tools are increasingly effective at using these data to pinpoint likely causal genes. These studies have also uncovered shared or pleiotropic genetic risk factors underlying comorbid psychiatric disorders. In this article, we argue that leveraging pleiotropic effects will provide opportunities to discover novel drug targets and identify more effective treatments for psychiatric disorders by targeting a common mechanism rather than treating each disease separately.
Similar content being viewed by others
Introduction
There are significant challenges in identifying and developing new medications for human diseases. Less than 10% of drugs that begin phase I clinical trials are estimated to gain regulatory approval [1, 2]. Failure rates are notably worse in psychiatry, with the field having the second lowest likelihood of approval at 6.2% (the lowest being oncology). There are several reasons why a clinical trial might fail, such as safety, commercial, or operational factors. However, a significant contributor to trial failure is due to a lack of efficacy, particularly in the later stages of drug development [3].
An approach to increase trial success rates involves incorporating human genetic data to reveal novel biological insights [4]. Studies have demonstrated that drug compounds are more likely to be approved if their genetic targets were previously identified by human genetic evidence, with the likelihood increasing if exact causal genes are known [4, 5]. Large-scale genome-wide association studies (GWAS) have identified hundreds of replicable genetic risk factors for complex diseases, including candidate causal genes [6]. These studies can establish the genetic evidence (i.e., interactions between drugs and disease-associated genes) needed for improved drug discovery and treatment [7].
The time and cost associated with traditional drug development pipelines are a significant concern in the pharmaceutical industry [8]. An alternative method to identify novel drugs for psychiatric disorders is the use of existing drugs for a new clinical indication(s), known as drug repurposing [9]. This process can bypass some de novo drug discovery steps by identifying an already approved or investigational drug that may have been abandoned due to lack of efficacy yet passed safety assessment [10]. Due to existing research on drug safety and pharmacokinetics in humans, drug repurposing can decrease the timeframe and cost of de novo drug development with a lower risk of failure.
Psychiatric disorders co-occur in the same individual more often than expected by chance (known as comorbidity) (Fig. 1A) [11, 12]. Large positive genetic correlations are also frequently observed between pairs of psychiatric disorders [7]. One explanation behind these genetic correlations is pleiotropy [13,14,15], which occurs when a genetic factor is associated with multiple phenotypes (Fig. 1B). Investigating these shared genetic risk factors may uncover potential aetiological overlap between disorders for drug targeting.
A The increased risk of psychiatric disorder diagnosis from a prior disorder reflects widespread genetic correlations. B Targeting pleiotropic mechanisms for the treatment of multiple conditions. Hazard ratios between comorbidity pairs were obtained from McGrath et al. [12]. Genetic correlations adapted from Derks et al. [7]. Acronyms: generalised anxiety disorder (GAD), anxiety disorders (ANX), major depressive disorder (MDD), obsessive-compulsive disorder (OCD), bipolar disorder (BD), post-traumatic stress disorder (PTSD).
Here, we will argue that shifting drug development towards targeting pleiotropic mechanisms is a promising avenue for determining effective drug targets in psychiatry. A robust pleiotropic mechanism can be identified using GWAS summary statistics and appropriate statistical methods. By integrating transcriptomic, proteomic, and drug datasets, these mechanisms can elucidate shared downstream molecular effects for drug repurposing opportunities. Further network medicine approaches can identify additional drug targets and reveal latent connections between disorders or between disorders and drug compounds.
Using pleiotropy for drug target identification has several potential benefits. Psychiatric disorders are highly diverse in their aetiology, symptoms, onset, and course, which limits our understanding of specific disease mechanisms and the development of personalised medicine [7, 16]. Narrowing down drug target identification to more homogenous phenotypes, such as co-occurring symptoms or conditions, can restrict this heterogeneity and improve power to distinguish the underlying mechanisms. Additionally, a single drug could treat multiple conditions by targeting a shared mechanism to provide a broader therapeutic impact.
This review will show how leveraging psychiatric pleiotropy can be an encouraging approach to identifying more effective drug targets. Methods and research investigating the nature of pleiotropy in psychiatric disorders and strategies to identify and prioritise drug candidates for repurposing will be examined.
Understanding pleiotropy in psychiatric disorders
There is significant genetic overlap between psychiatric disorders. Family and twin studies first identified positive correlations and patterns of inheritance between psychiatric disorders [17,18,19]. The introduction of methods for estimating heritability and genetic correlations using GWAS summary statistics, such as linkage disequilibrium score regression (LDSC) [20] later identified genetic correlations across numerous psychiatric phenotypes [15, 21, 22]. Genetic correlations for psychiatric disorders range between –0.17 for ADHD and OCD to 0.9 for depression and anxiety [7]. Disentangling the pleiotropic effects that underlie these correlations is a key challenge of psychiatric genetics.
Pleiotropy can be studied across different levels. Firstly, pleiotropy is observed at a regional or loci level, with a comprehensive study analysing eight psychiatric disorders identifying 146 loci associated with a psychiatric phenotype, with 109 displaying cross-phenotype associations [23]. Secondly, transcriptome and proteome studies have shown extensive correlations in (imputed) gene expression levels [24] and shared causal and interacting proteins [25], demonstrating pleiotropy at a molecular level. Finally, the downstream biological effects of pleiotropy can be elucidated at a network and pathway level. A study identified 49 pathways associated with three psychiatric traits, which clustered into processes involving histone, synaptic biology, neuronal, and immune pathways [26]. Figure 2 outlines methodologies investigating pleiotropy at these various levels to be integrated for drug repurposing.
Acronyms: Colocalisation (Coloc), mendelian randomisation (MR), genomic structural equation modelling (genomic SEM), genome-wide association study (GWAS), transcriptome-wide association study (TWAS), protein-wide association study (PWAS), quantitative trait loci (QTL), instrumental variables (IVs), phenome-wide association study (PheWAS).
Targeting pleiotropic mechanisms may advance transdiagnostic treatments in psychiatry, i.e., using a single treatment for symptoms or conditions across multiple diagnoses. Evidence in clinical practice shows that psychiatric transdiagnostic treatments are an already recognised concept, e.g., selective serotonin reuptake inhibitors for the treatment of major depressive disorder and generalised anxiety disorder [27]. As most drug targets are proteins [28], improving our understanding of the shared molecular factors and pathways between disorders can further transdiagnostic treatments by targeting the potential shared consequence of genetic variation.
Identifying variant-level pleiotropy between disorders
A cross-phenotype association is a statistical discovery of a single locus displaying an association with multiple phenotypes. Identifying a shared genetic variant/s between comorbid disorders from this association can help us understand the molecular causes underlying psychiatric comorbidity. This, in turn, provides opportunities for therapeutic interventions that target the molecular consequence of a shared causal variant rather than treating the conditions separately. However, several considerations are needed to establish a pleiotropic mechanism from a cross-phenotype association.
Initial considerations
Cross-phenotype associations
The first consideration is using appropriate statistical methods to discover cross-phenotype associations. Numerous analytic approaches exist for the joint evaluation of multiple traits (see [29] for a full review). Briefly, method choice depends on data availability, phenotype distribution, number of traits, and if there are overlapping samples between datasets. While methods that jointly analyse traits directly are typically more powerful [29], they are substantially limited in scope by requiring individual-level genotype data and all traits to be measured in the same individuals. These conditions can be challenging to obtain due to privacy and data restrictions and many traits going unmeasured (e.g., low prevalence disorders). More recently, methods have been developed that require only univariate GWAS summary statistics and account for sample overlap, e.g., LDSC [20], MTAG [30] and genomic SEM [31]. This has enabled the identification of cross-phenotype associations at a larger scale and across a wide range of traits, which can be performed at various levels, from genome-wide to regional or a variant level.
Pleiotropic mechanism
The second consideration is whether the variant has an actual biological effect on each phenotype. This uncertainty stems from the different mechanisms of pleiotropy that underly cross-phenotype associations. These mechanisms include biological, mediated, and spurious pleiotropy (Fig. 3A) [32]. Biological pleiotropy (horizontal pleiotropy) is when a variant affects two or more phenotypes through a direct biological effect. Biological pleiotropy can be broken down further into single-gene pleiotropy (the variant impacts one gene that affects two or more phenotypes) and multi-gene pleiotropy (the variant impacts multiple genes affecting multiple phenotypes) [13]. Figure 3B demonstrates biological pleiotropy via pleiotropic noncoding variants associated with gene/protein expression changes. Mediated pleiotropy (or vertical pleiotropy) arises when one phenotype causally affects a second phenotype, resulting in a variant associated with phenotype one being indirectly associated with phenotype two. Finally, spurious pleiotropy involves variants associated with multiple phenotypes from sources of bias, including phenotype misclassifications, study design artefacts, or strong LD patterns.
A The three divisions of pleiotropy. Biological: when a variant is associated with multiple phenotypes owing to true biological effects. Mediated: a phenotype causally influences another phenotype, resulting in a variant to be indirectly associated with the second phenotype. Spurious: when a variant is identified as pleiotropic due to sources of bias, linkage disequilibrium (LD) patterns, or study design artefact. B Additional types of biological pleiotropic mechanisms. Causal noncoding variants in cis, regulate gene and protein expression (single and multigene) with biological products affecting multiple phenotypes. Acronyms: expression quantitative trait loci (eQTLs), protein quantitative trait loci (pQTLs).
Methodological approaches
When identifying pleiotropic variants from GWAS summary statistics, taking the intersection of variants associated with each trait is less optimal than methods that directly model pleiotropic relationships, as distinguishing pleiotropic variants is limited by the power of both GWAS to detect the variant.
Several resources are available to address this issue and estimate the probability of a pleiotropic variant/s across phenotypes [33,34,35,36,37]. For example, PolarMorphism can assess two or more phenotypes by converting trait-specific effect sizes into polar coordinates to evaluate variant sharedness while distinguishing biological pleiotropic variants from mediated variants [33]. Another method, PLACO, has been shown to identify an equal amount of pleiotropic variants as PolarMorphism; however, it can only examine two traits at a time [34]. Colocalisation is another approach using Bayesian statistical methods to estimate the probability that a single variant in a shared locus is causal in both GWAS [38]. Other colocalisation methods have been developed to examine more than two traits simultaneously and have different assumptions regarding the number of causal variants in a locus [37, 39,40,41].
The improved accessibility of phenotypic data from large-scale biobanks has enabled researchers to combine phenotypic and genetic data to identify pleiotropic variants. Phenome-wide association studies (PheWAS) analyse variants or groups of variants against a diverse set of phenotypes [42]. This method identifies variant associations with multiple diseases, thereby identifying potential pleiotropic effects, although the underlying mechanism behind these associations still needs to be clarified. PheWAS is highly versatile, demonstrating use in validating drug targets [43], studying risk variants associated with polygenic risk scores [44], uncovering pleiotropic biological function [45], predicting drug adverse effects [46], and drug repurposing [47].
Variant pleiotropy and drug repurposing
The mechanism underlying a pleiotropic variant can inform therapeutic interventions. If a drug target is associated with two disorders via mediated pleiotropy, treating one disorder would either alleviate or exacerbate the symptoms of the second disorder, depending on their relationship. Additionally, targeting a mediated pleiotropic mechanism would only benefit the second disorder if the first disorder is present. Targeting a biological pleiotropic mechanism can help or worsen each disorder or benefit one disorder and worsen the other, depending on the variant direction of effect. As it is likely that a combination of these mechanisms is contributing to psychiatric comorbidity, investigating these various models is required to distinguish robust drug targets.
A variant’s effect direction has implications in treating multiple disorders. A drug compound targeting shared mechanisms with opposite effects may help one phenotype but inadvertently negatively influence the other. Additionally, variant direction influences the ideal drug mechanism. For example, a gain-of-function variant would increase the function of its downstream effectors, thus requiring an inhibitor or antagonist for effective treatment. A recent method, genetic priority scores, prioritises genetic targets and drug indications derived from an integration of genetic evidence [48]. Genetic priority scores also incorporate the direction of genetic effect to determine the drug mechanism for repurposing opportunities.
Limitations and challenges
The biggest challenge when identifying a robust pleiotropic variant for drug repurposing is discerning the underlying mechanisms. One approach, Mendelian Randomisation (MR), can help distinguish between mediated and biological pleiotropy. MR is a frequently used method for inferring causation using genetic associations from GWAS data [49]. The foundation of MR relies on Mendel’s laws of segregation and independent assortment, where variants are inherited randomly. This method utilises variants as instrumental variables (IVs) to enable a practical approach for causal inference between an exposure variable and an outcome phenotype (if key assumptions are satisfied). The assumptions of MR include 1) the IV is strongly associated with the exposure, 2) the IV is not associated with confounders between exposure and outcome, and 3) the IV is only associated with the outcome through the exposure.
In addition to determining causal exposure-outcome relationships, MR can serve as a test for mediated pleiotropy [13]. Notably, due to psychiatric disorders’ unknown biology, bidirectional relationships, heterogeneity, and high levels of pleiotropy and polygenicity, thorough sensitivity analyses and caution in interpretations are essential [50]. In light of these limitations, several methodologies are available to clarify these complex relationships and rule out alternative hypotheses [50, 51]. Other mediation analyses have been created to break down the total effect of the variant on a phenotype into direct and indirect effects [29, 52]. Another tool, BUHMBOX, can identify whether observed associations between two diseases are caused by biological pleiotropy or subgroup heterogeneity in disease cases, determining potential spurious pleiotropy [53].
Another challenge in determining a pleiotropic variant is caused by linkage disequilibrium (LD). A variant identified by GWAS may not be truly causal but is in linkage with the actual causal variant. Fine-mapping resources use Bayesian frameworks to prioritise variants by integrating LD data, functional information, and association patterns to estimate the posterior probability of each variant being causal [54, 55]. fastPAINTOR is an additional method that aims to increase fine-mapping accuracy at pleiotropic risk loci by leveraging pleiotropy to strengthen the causal variant signal [56]. Using fine mapping to distinguish the likely causal variant affecting multiple phenotypes can ensure robust downstream drug targets.
Modelling higher-order factors of psychopathology
Certain psychiatric diseases co-occur more frequently than others and group together at clinical and subclinical levels [11, 23]. Similarly, modelling of the genetic correlations between disorders has revealed that subsets of disorders cluster together (e.g. internalising disorders or compulsive disorders) [57], suggesting the presence of a higher-order structure of psychopathology at a genetic level. Shifting focus from individual disorders toward these higher-order components can reveal key pleiotropic mechanisms for therapeutic intervention.
Multivariate approaches can identify higher-order dimensions of broad liability, known as latent factors, that explain the observed genetic or phenotypic variance between multiple traits. These higher-order latent factors are not directly measured but inferred using various statistical methods such as exploratory factor analysis [58] or structural equation modelling (SEM) [59]. The identification of latent factors of genetic liability across disorders can inform the aetiological structure of psychopathology and subsequently inform pleiotropic mechanisms for drug targets.
Initial considerations
Latent Factor Structure
Several transdiagnostic latent factor structures have been proposed to model the comorbidity patterns of psychiatric disorders identified using phenotypic, twin or family data. Key examples include a single p factor model [60] that captures the general psychopathology across all disorders, a bi-factor model [58, 61] that divides traits by internalising vs externalising symptoms and behaviours, or a three-factor model including internalising, externalising, and a thought disorder factor [60]. The p factor is a frequent feature of dimensional approaches and may reflect the overall pleiotropic biology underlying psychiatric risk. However, breaking down the genomic p factor into intermediate components to capture more specific (i.e., less heterogeneous) disease clusters may be more beneficial for biological insight [21]. A recently developed psychiatric classification system, the Hierarchical Taxonomy of Psychopathology (HiTOP) [62], proposes a hierarchical system to capture different levels of specificity, including higher-order dimensions of psychopathology, which can account for pleiotropy/comorbidity between disorders directly [7]. The discovery of pleiotropy drug targets can be improved by incorporating these higher-dimensional frameworks.
Methodological approaches
Genomic SEM analyses the joint genetic architecture of numerous phenotypes using only GWAS summary statistics (Fig. 4A) [31]. Genomic SEM has been used to identify a four-factor model capturing genetic overlap between 11 major psychiatric disorders, including compulsive, psychotic, neurodevelopmental, and internalising factors [21]. An additional substance-use factor has been determined using a common factor model of genetic liability between four substance-use disorders [63]. These disease factors could provide a promising starting point for identifying pleiotropic drug targets within these clusters.
A Example of a two-factor, four phenotype model fitted to the sample covariance/genetic covariance matrices calculated by linkage disequilibrium score regression (LDSC). Factor 1 (F1) captures the broad liability between phenotypes 1 and 2 (P1, P2). Factor 2 (F2) captures phenotype 3 and 4 (P3-4). B Identifying significant variants (SNPs) associated with a latent factor. Example demonstrates adding SNP effect to the matrices to estimate a given SNPs effect on factor 1. Multivariate genome-wide association study (GWAS) identifies loci operating through the factor, i.e., associated with P1 and P2.
Genome-wide SEM is another multivariate method that uses phenotypic covariance to construct an SEM, incorporating raw genotype data to identify variants associated with the modelled factors [64, 65]. This method models phenotypic with genetic information; however, it can be limited by the accessibility of raw genotype data and what phenotypic information is available across all samples.
Pleiotropic decomposition regression (PDR) is an alternative approach to decomposing the covariance of multiple traits into factors or components [66]. This method deviates from other decomposition methods by modelling the distribution of variant effect sizes to separate variants into components they affect the most, thereby limiting the number of components a variant is associated with. It is argued that this is more mechanistically informative, as knowing the traits an individual SNP affects informs a single mechanism or component [66]. PDR is limited by the number of traits that can be examined simultaneously due to computational demands and does not produce standard errors.
Latent modelling for pleiotropy and drug repurposing
Identifying genetic latent factor models can provide insight into shared aetiology and identify drug targets for treating multiple disorders. An application of genomic SEM, multivariate GWAS, can determine variants impacting disorders through the inferred factors (Fig. 4B) [31]. Another addition, stratified genomic SEM, identifies gene sets and functional categories that disproportionately contribute to shared genetic risk, revealing mechanistic and dynamic insights between disease clusters [21]. The downstream effects on shared molecular mechanisms and pathways influencing lower-order disorders can be determined by identifying these variants or gene sets associated with disorder clusters or pairs.
Transcriptomic data can be integrated with genomic SEM to examine gene expression patterns associated with genetically overlapping psychiatric disorders, known as transcriptome-wide structural equation modelling (T-SEM) [67]. Grotzinger et al. [67] used T-SEM to identify 466 genes with expression levels associated with a five-factor model across 13 major psychiatric disorders. Additionally, the study investigated existing drug compounds to target the shared risk pathways, identifying 35 drug-gene pairs for the thought disorders factor disorder, one for the internalising disorder factor, and five for the p factor [67]. The results suggest these drugs have a transdiagnostic effect targeting the psychiatric disorders clustered within these factors.
Limitations and challenges
Latent factor analysis has many diverse analytical applications; however, several limitations exist. Primarily, the limitations of the statistical analysis apply to the method used, i.e., the same limitations as SEM apply to genomic SEM. Some notable limitations of SEM and genomic SEM include:
-
1.
SEM relies on the user to specify the initial model of relationships between latent variables and the observed variables. Bias estimates and inaccurate conclusions can occur if the model is misspecified or lacks justification.
-
2.
Various models may fit the data with similar effectiveness or better, and determining the one perfect model is not practical [68]. A model fitting the data too well can risk overfitting, reducing generalisability performance, decreasing model robustness, and introduce noise.
-
3.
Including traits with extreme correlations (i.e., highly multicollinear traits) will introduce a bias to a particular model owing to their substantial genetic overlap [69].
-
4.
Genomic SEM only uses summary-level data, limiting the investigation of phenotypic or environmental causal pathways.
-
5.
The power to detect a genetic drug target is determined by the power of the inputted GWAS in a given model, as seen in [67], which identified the majority of drug-gene interactions with well-powered GWAS (i.e., thought disorders).
Elucidating molecular consequences for drug targeting
The identification of pleiotropic variants for psychiatric disorders may provide new prospects for the development of drugs that target a shared biological mechanism. To characterise the potential biological effects of a pleiotropic variant, we must integrate diverse functional genomic data from different cellular contexts. The recent availability of omics data from various tissues and cell types and improvements in integrative, systems biology-based approaches offer opportunities to establish the functional consequences of DNA sequence variation.
Initial considerations
Data type
The type of molecular data is the first consideration when integrating functional data for drug repurposing. GWAS-identified variants are enriched in noncoding regions [70] and are believed to have an intermediate effect on disease via various regulatory mechanisms. These effects can be examined by integrating molecular quantitative trait loci (molQTL) datasets, with these variants associated with variation in molecular traits. Expression quantitative trait loci (eQTL) are a type of molQTL associated with changes in gene expression. eQTLs can be further divided into cis-eQTLs, variants associated with the expression of nearby genes, or trans-eQTLs, variants associated with changes in gene expression from a further distance or a different chromosome. Compared to cis-eQTLs, trans-eQTLs typically have smaller effects and thus require larger samples to detect and tend to be more tissue-specific [71]. Another molQTL, protein quantitative trait loci (pQTL), are variants associated with changes in protein expression. Many pQTLs are often recognised as eQTLs, with one study estimating 61% of pQTLs also identified as eQTLs via data obtained from post-mortem brain samples [72].
Using gene expression levels (eQTLs) and protein abundance (pQTLs) to determine genetic targets for therapeutics have their strengths and weaknesses (see Table 1). The primary argument is whilst eQTLs capture the more immediate molecular effects of disease-associated variants, variation in the post-translational stages is more therapeutically actionable, and these do not always correlate with each other [73]. Incorporating both data layers would be preferable for a comprehensive understanding of a variant’s functional consequence and provide robust evidence for any identified drug targets.
Dataset availability
The second consideration for the functional interpretation of a pleiotropic variant is dataset availability. molQTL will differ between tissues, cell types, developmental stages, and environmental stimuli. Therefore, incorporating disease-relevant tissue data and exploring molQTLs across varying contexts when possible is essential. For psychiatric disorders, there are several publicly available brain eQTL datasets, including GTEx, a resource of eQTLs across 13 brain tissues collected from post-mortem donors [74], PsychENCODE a large dataset of 1,866 individuals [75], and MetaBrain, a meta-analysis conducted on 14 brain datasets (including GTEx and PsychENCODE) [76]. While less advanced than RNA sequencing, profiling the human proteome has made progress, resulting in pQTL identification in blood tissue [77] or brain tissue [78].
Conventional eQTL resources utilise bulk tissue expression samples, restricting insights into specific cell types, development stages, and cell states. A promising new approach is using single-cell transcriptomic assays to analyse dynamic contexts that bulk expression cannot capture by calculating gene expression at a single-cell resolution. Studies analysing brain cell type-specific eQTLs have demonstrated that many affected genes have differing expression levels in various cell types [76, 79]. While single-cell datasets are less available, significant advances have been made. Some examples of available datasets include the development stages of the human prefrontal cortex [80], brain cell types [79], and databases/portals sharing single-cell findings, e.g., scQTLbase [81] or Single Cell Portal [82].
Methodological approaches
There are three primary methods to incorporate molQTL to identify functional mechanisms from GWAS data. The first, transcriptome-wide association studies (TWAS), uses cis-eQTL datasets to impute gene expression from genetic variants via multivariate models [9, 83]. This method generates genetically regulated expression prediction models by modelling eQTL information. These models predict gene expression from GWAS summary statistics to indicate a gene’s expression level and association with the trait of interest, i.e., if a gene is upregulated or downregulated in the phenotype. Reflecting TWAS, proteome-wide association studies (PWAS) can integrate pQTL and GWAS data to impute protein expression and identify potential drug-repurposing targets by utilising the same tools available from TWAS [83, 84].
The second method, Mendelian Randomisation (MR), also integrates molQTL and GWAS data. As outlined previously, MR is typically used for causal inference of an exposure to an outcome using genetic variants as instrumental variables (IVs). MR, for the application of drug repurposing, utilises cis variants associated with targetable proteins as IVs, i.e., pQTLs or eQTLs [85]. These IVs are used for drug repurposing based on a drug compound having therapeutic potential if its protein target (the exposure) causally impacts an outcome in the right direction as the intended pharmacological mechanism (Fig. 5). As proteins are proximal effectors of biological processes, it has been argued that there is less chance to break the no horizontal pleiotropy assumption compared to downstream exposures [86]. Additional MR approaches, such as multi-response MR [87] or two-stage multivariate MR [88], enable the modelling of exposures on multiple outcomes (e.g., two comorbid phenotypes) rather than performing MR for each outcome independently.
Credibly causal molecular quantitative trait loci (molQTLs) as instrumental variables (IVs) to validate that the expression of the target is causal with disease phenotype (outcome). If a drug's mechanism opposes the direction of the estimated exposure-outcome effect (β), an approved drug is prioritised for repurposing. The example illustrates that this protein X expression increases disease risk; therefore, a drug with a mechanism opposing the target, e.g., an antagonist, can be a repurposing candidate.
Colocalisation is the third method and is frequently used with TWAS/PWAS/MR to further prioritise genes or proteins most likely to influence disease [38, 84, 89]. As previously discussed, colocalisation can identify pleiotropic variants across multiple phenotypes. However, it can also include molecular phenotypes through molQTL datasets. Applying this method with TWAS/PWAS can identify putatively causal genes or proteins associated with multiple diseases via changes in expression. Many other approaches have expanded upon MR and TWAS methods to improve accuracy or address limitations (Table 2).
Molecular pleiotropy and drug repurposing
Integrating functional and genetic data can improve our understanding of the shared aetiological causes between comorbid phenotypes. This combination can identify pleiotropic genes related to specific biological processes. For instance, Antón-Galindo et al. [90] determined which genes from dopaminergic and serotonergic processes were significantly associated with multiple or individual addiction-related phenotypes through predicted expression levels [90].
Drug repurposing candidates can be prioritised using functional data to provide direction for the pleiotropic targets. Firstly, a drug can be prioritised if it directly modulates a disease genetic target in the opposite direction to its expression, e.g., an antagonist drug targeting an overexpressed gene. Secondly, signature mapping is a computational method that compares the disease gene expression signature to a drug-gene expression signature [91]. A signature describes the differential gene expression levels from perturbation caused by a drug compound or disease. This method prioritises drug compounds opposing the disease signature to theoretically normalise gene expression levels.
Many studies have identified drug repurposing candidates in psychiatric disorders based on genetically regulated expression levels in human brain tissue using the described methods (see Table 3). Developing upon this idea, leveraging pleiotropic variants or latent factors to discover genetically regulated gene or protein expression targets could identify transdiagnostic treatments for psychiatric disorders. Hatoum et al. [63] integrated multiple methodologies, including signature mapping, latent factor analysis, and TWAS, to identify 104 drug-repurposing candidates targeting an addiction latent factor. These drugs are predicted to reverse the addiction disease-expression profile across multiple lower-order conditions [63].
Limitations and challenges
Despite the critical utility of TWAS/PWAS to integrate QTL data with GWAS for new genetic discoveries in psychiatry, there are several limitations (see [92] for a comprehensive outline). Some notable limitations include: 1) LD and gene co-regulation can obscure which gene is causal at a given locus; 2) genetically regulated expression models are limited by difficult-to-acquire post-mortem tissue sample sizes for psychiatric disorders; 3) TWAS does not factor in other modes of gene expression modulation, such as environmental factors or transcription factor regulation (this is also a strength of TWAS as associations cannot be explained by confounding).
Using MR to validate pleiotropic genes/proteins as drug targets for treating multiple conditions can be challenging. MR requires that the IV impacts the outcome only via the exposure (mediated/vertical pleiotropy) and not by other processes (horizontal pleiotropy). As these targets are identified as highly pleiotropic, careful selection of IVs associated with the phenotypes of interest is critical for MR analysis. However, the biological function of variants is often unknown, and the underlying mechanism of how a pleiotropic drug target is associated with each phenotype is difficult to pinpoint. Sensitivity analysis robust to pleiotropy must be performed (see [51, 93] for details), and IVs strongly associated with each outcome only through the exposure must be identified, e.g., via coloc [38] or HEIDI [94].
Network computational drug repurposing
Network approaches can provide a broader understanding of the biological impact of pleiotropy. A pleiotropic component, such as a variant, gene, or protein, can affect many cellular functions through complex molecular interactions, causing possible pleiotropic effects [95]. Investigating these interactions using multiple layers of large-scale molecular information can inform the biological implications of pleiotropy in psychiatric disorders. Notably, using the appropriate methods to implicate genuine biological pleiotropy is essential. Like previous methods, molecular networks can uncover the physiological relevance of disease-associated variants and examine genotype-phenotype relationships to identify genetically informed drug targets for repurposing.
Initial considerations
Data type
The central consideration for network methods is the data type used to build a network. Data can be sourced from several resources, including experimental data, text mining or databases. Drug repurposing studies often integrate multiple data sources to complete the missingness of each data level, an advantage of network approaches (refer to [96] for a comprehensive review of data sources and network-based drug repurposing for psychiatric disorders). See Table 4 for common data types and their uses.
Methodological approaches
The construction of a network (also known as network inference) involves using biomedical data to simplify the complexity of biological systems (Fig. 6A). The network nodes represent a system’s components, including genes, proteins, drug compounds, or diseases [96]. The interactions or relationships between nodes are represented as edges and inferred using physical connections, correlations, machine learning, or conditional dependencies (Fig. 6A).
A Using networks to explore biological connections. Constructing a network (network inference) involves integrating multiple layers of data to interpret relationships between biological components. The example demonstrates an ABC network model, connecting a disease gene to drug target. B Graphical representation of the process and applications of network propagation. Propagation can begin with integrating the known interactome with genome-wide association study (GWAS)-linked genes (seeds). Two time points of the propagation process are illustrated, with t = 1 demonstrating the initial state, and t = ∞ showing the final steady state (convergence). Network propagation can identify gene modules/clusters, additional drug targets, and overlapping communities between datasets. C Common pleiotropic properties of a network that may act as shared therapeutic targets.
ABC model
The ABC model is a classic network approach for identifying new links between components. The model assumes that if component A is linked to B and B is linked to C, components A and C are also linked. In this model, several steps (Bn) can link A and C, and A and C must come from different data layers to generate latent connections. For example, used drug-gene networks (linking A to B) and disease-associated expression networks (linking B to C) to identify drug repurposing candidates.
GBA model
Guilt-by-association (GBA) is a method that relies on the connection between two entities in a network. It assumes that if two components are linked, they are more likely to share similar functions or be involved in related diseases. This approach uses similarity metrics to estimate new drug indications. If two diseases share significant characteristics, a drug for one disorder could be used for the other. Alternatively, if two drugs share the same properties, they are assumed to be useful for similar indications.
Network propagation is an application of guilt-by-association to predict additional disease-associated genes using GWAS and protein-interaction-based approaches (Fig. 6B). This is founded on the principle that interacting proteins contribute to similar biological functions, affecting the same trait [97]. Different algorithms can rank proteins as disease-associated based on their interaction profile with the disease-associated proteins identified via GWAS (seed proteins). Genes with a high frequency of interactions and thus shared function with a seed gene would be prioritised compared to those with low seed-gene interactions.
Network pleiotropy and drug repurposing
Network propagation for unravelling pleiotropy and drug repurposing
Analysis of network-propagated gene interactions can determine highly connected modules enriched in disease association. These groups can infer biological mechanisms and shared pleiotropic cellular processes influencing disease. Network propagation has previously identified pleiotropic gene modules across GWAS traits [98] and pleiotropic drug targets shared between schizophrenia and cognitive performance [99].
Network propagation can also determine drug targets without genetic association to a disease. Many drug compounds with proven efficacy in clinical trials have targets without direct genetic evidence connecting them to the treated disease [4]. Many of these unsupported drug targets can be identified through network-based rankings, with studies demonstrating the enrichment of approved drug targets in genes that interact with disease-associated genes [100]. Expanding the interactome from a seed may reveal additional novel drug targets.
Targeting pleiotropic network components
Drug repurposing can leverage network-based methods to identify drug candidates targeting pleiotropic disease-associated components, such as specific nodes or clusters within a network (Fig. 6C). Investigating disease-network overlap from GWAS data can pinpoint shared disease components to provide a broader therapeutic effect. Common pleiotropic network components include:
Hubs
Pleiotropy has been linked to gene/protein network centrality [45]. The more central a gene or protein is within a network, the more likely it is to be involved in multiple processes [101, 102]. A central protein will have an increase in protein-protein interactions and higher distribution in cellular components, causing a protein’s molecular function to have a broader influence on biological processes [103]. Hubs are defined as having a high degree (i.e., number of connections), often in central network positions, and are integral topological components of a network.
Hubs are a common drug target in network medicine, as modulating the activity will have a more significant impact when compared to peripheral nodes [104]. Targeting shared hub nodes between diseases may have broad therapeutic benefits and treat multiple conditions due to their connectedness. For instance, one study identified five hub gene targets for potential drug targeting that were associated with both Alzheimer’s disease and major depressive disorder, two diseases with a high risk of comorbidity [105].
Bottlenecks
A pure bottleneck node is characterised by a low number of interactions, a high level of betweenness centrality (i.e., how central a node is), and often links subnetworks together [45]. Due to their strategic network positioning, these proteins can influence multiple diseases simultaneously, exerting their effects on distinct disease-related pathways. Many bottlenecks can also be classified as hubs if the node has many interactions and betweenness. These hub-bottleneck nodes are often evolutionary-constrained and associated with multiple pathways [106].
Bottlenecks and hub-bottleneck nodes are enriched in approved and experimental drug genetic targets, indicating betweenness as a promising measure for network drug repurposing [106]. Mahboubi et al. [107] identified a critical hub-bottleneck gene (ESR1) associated with both schizophrenia and obsessive-compulsive disorder (OCD) as an encouraging drug target for OCD-schizophrenia comorbidity [107].
Modules
Network modules are densely connected nodes with limited connections to outside nodes that may conduct a specific biological function [108]. Disease-associated genes tend to cluster together in well-defined neighbourhoods, known as disease modules, with a drug’s efficacy and potential adverse effects related to the proximity to these small network neighbourhoods [109, 110].
Menche et al. [111] observed overlapping network modules between disease pairs, with the extent of overlap strongly indicative of the pathobiological similarity between diseases. Shared modules between disease pairs are linked to genes demonstrating increased co-expression, common symptoms, and high rates of comorbidity [111]. Investigating module overlap between disorders can predict shared pathobiological functions for subsequent drug targets.
Limitations and challenges
Network-based approaches in identifying novel drug candidates in psychiatry show much promise. However, there are still many challenges that must be addressed. First, current network-repurposing depictions only capture a fixed setting of the biological system, regardless of the dynamic nature of biological systems or the prevalence of transient interactions. Second, our knowledge of protein interactions is incomplete. The complexity of the interactome, including the vast number of proteins, their isoforms, and post-translational states, makes completing the interactome challenging. Third, networks are often based on known biological interactions and will be skewed towards well-researched mechanisms. Fourth, the direction of effect between network-associated gene interactions is usually unknown and requires further functional investigation. Fifth, establishing a causal relationship between disparities in gene expression and psychiatric disorders via analyses of transcriptomic networks can be difficult, as the disorder itself may induce changes in gene expression levels [13]. Lastly, integrating diverse data sources and managing heterogeneity can be complex due to sparse, incompatible, or absent data [112].
Pleiotropy and adverse effects
Targeting a pleiotropic mechanism may result in a broader spectrum of therapeutic potential, but this will also increase the risk of adverse effects. Resources such as SIDER [113], a vast database on adverse drug reactions, would need to be included in any drug repurposing pipeline as a prudent measure. The genetic priority scores resource integrated SIDER, linking drug indications to genetic targets to develop a valuable tool for side effect prediction [48].
Genetic data can also be used to identify possible adverse effects when modulating a given target (see [114] for a full review). As previously described, PheWAS identifies multiple phenotypes associated with a variant. Alternatively, this method can predict potential adverse effects if a drug target is associated with other disease phenotypes in the corresponding direction of the drug mechanism, e.g., an agonist drug for a protein target that has a positive association with a disease phenotype. PheWAS has also been integrated into MR (MR-PheWAS) to examine the impact of exposure on numerous outcomes [115]. Similarly, this method can identify novel causal proteins/prioritising drug targets [116] and examine potential adverse effects [117]. This inference is based on an approved drug target (the exposure) being causal with outcomes other than its primary indication [118].
From a network perspective, targeting pleiotropic components may have safety-related limitations due to their broad influences. For example, targeting highly connected hubs may cause an increase in adverse effects or undermine network integrity [104]. Therefore, prioritising hubs with an intermediate degree can limit the risk of adverse effects. Alternatively, perturbing an interacting protein may modulate the effector protein to a lesser effect than a direct interaction or instead target the shared disease module the protein resides in. Drug-disease proximity is another important measure to predict a drug compound’s efficacy and detection of adverse effects, with a drug target required to be proximal to the disease module [110, 119].
Advances in technology and adverse effect datasets have enabled artificial intelligence, machine and deep learning methods to extract drug information and predict adverse reactions, e.g., via drug-to-drug networks [120] or neural networks and gene expression profiles [121]. Due to the expansive nature of pleiotropy, combining multiple approaches to predict adverse reactions when targeting such mechanisms is essential.
Conclusions
Leveraging pleiotropy among psychiatric disorders may identify new treatments for comorbidities. This review outlines current research and methods used to investigate pleiotropy and identify drug repurposing opportunities for psychiatric disorders. Increasing sample sizes across psychiatric phenotypes will progress the identification of both shared and unique genetic risk factors. Alternatively, developing smaller but well-defined cohorts may be more helpful for drug repurposing, e.g., individuals with strict comorbidity diagnosis. This avenue of research would reduce the heterogeneity surrounding psychiatric disorders to identify specific biological mechanisms for precision medicine.
Future work should focus on combining evidence from multiple resources and methods to identify robust repurposing candidates for future validation studies. Multi-omic approaches are needed to gain a comprehensive biological understanding of genetic risk factors in psychiatric disorders for drug targeting. Developments in technologies and decreasing costs have improved the generation and feasibility of large-scale omics data, resulting in advances and prospects in multi-omic integration methods, such as machine learning models [122]. Additionally, electronic health records provide a valuable longitudinal and pathological patient data resource. Future access and incorporation of deidentified datasets into the drug repurposing process can assist in predicting and validating drug candidates [123].
In conclusion, understanding the underlying pleiotropic mechanisms across psychiatric disorders is a challenging task. However, improvements in data access, larger sample sizes, and advances in methodologies provide an excellent opportunity to continue our developments in unravelling the complex interplay across psychiatric genetic risk factors. The parallel between comorbidity and pleiotropy prevalence suggests shared genetic mechanisms that drug candidates could target to improve therapeutics for patients.
References
Thomas D, Burns J, Audette J, Carroll A, Dow-Hygelund C, Hay M. Clinical development success rates 2006–2015. Biotechnology Innovation Organization (BIO) industry analysis; 2016.
Sun D, Gao W, Hu H, Zhou S. Why 90% of clinical drug development fails and how to improve it? Acta Pharm Sin B. 2022;12:3049–62.
Cook D, Brown D, Alexander R, March R, Morgan P, Satterthwaite G, et al. Lessons learned from the fate of AstraZeneca’s drug pipeline: a five-dimensional framework. Nat Rev Drug Discov. 2014;13:419–31.
Nelson MR, Tipney H, Painter JL, Shen J, Nicoletti P, Shen Y, et al. The support of human genetic evidence for approved drug indications. Nat Genet. 2015;47:856–60.
Ochoa D, Karim M, Ghoussaini M, Hulcoop DG, McDonagh EM, Dunham I. Human genetics evidence supports two-thirds of the 2021 FDA-approved drugs. Nat Rev Drug Discov. 2022;21:551.
Uffelmann E, Posthuma D. Emerging methods and resources for biological interrogation of neuropsychiatric polygenic signal. Biol Psychiatry. 2021;89:41–53.
Derks EM, Thorp JG, Gerring ZF. Ten challenges for clinical translation in psychiatric genetics. Nat Genet. 2022;54:1457–65.
Crismon ML, Walkow J, Sommi RW Drug Development for New Psychiatric Drug Therapies. In: Macaluso M, Preskorn SH, Shelton RC, editors. Drug Development in Psychiatry [Internet]. Cham: Springer International Publishing; 2023 [cited 2023 Jul 11]. p. 131–67. (Advances in Neurobiology; vol. 30). Available from: https://link.springer.com/10.1007/978-3-031-21054-9_5.
Reay WR, Cairns MJ. Advancing the use of genome-wide association studies for drug repurposing. Nat Rev Genet. 2021;22:658–71.
Pushpakom S, Iorio F, Eyers PA, Escott KJ, Hopper S, Wells A, et al. Drug repurposing: progress, challenges and recommendations. Nat Rev Drug Discov. 2019;18:41–58. Jan
Plana-Ripoll O, Pedersen CB, Holtz Y, Benros ME, Dalsgaard S, de Jonge P, et al. Exploring comorbidity within mental disorders among a Danish national population. JAMA Psychiatry. 2019;76:259–70.
McGrath JJ, Lim CCW, Plana-Ripoll O, Holtz Y, Agerbo E, Momen NC, et al. Comorbidity within mental disorders: a comprehensive analysis based on 145 990 survey respondents from 27 countries. Epidemiol Psychiatr. Sci. 2020;29:e153.
Lee PH, Feng YCA, Smoller JW. Pleiotropy and cross-disorder genetics among psychiatric disorders. Biol Psychiatry. 2021;89:20–31.
Cross-Disorder Group of the Psychiatric Genomics Consortium. Identification of risk loci with shared effects on five major psychiatric disorders: a genome-wide analysis. Lancet Lond Engl. 2013;381:1371–9.
Romero C, Werme J, Jansen PR, Gelernter J, Stein MB, Levey D, et al. Exploring the genetic overlap between twelve psychiatric disorders. Nat Genet. 2022;54:1795–802.
Feczko E, Miranda-Dominguez O, Marr M, Graham AM, Nigg JT, Fair DA. The heterogeneity problem: approaches to identify psychiatric subtypes. Trends Cogn Sci. 2019;23:584–601.
Roy MA, Neale MC, Pedersen NL, Mathé AA, Kendler KS. A twin study of generalized anxiety disorder and major depression. Psychol Med. 1995;25:1037–49.
Kendler KS, Gardner CO, Gatz M, Pedersen NL. The sources of co-morbidity between major depression and generalized anxiety disorder in a Swedish national twin sample. Psychol Med. 2007;37:453–62.
Shih RA, Belmonte PL, Zandi PP. A review of the evidence from family, twin and adoption studies for a genetic contribution to adult psychiatric disorders. Int Rev Psychiatry. 2004;16:260–83.
ReproGen Consortium, Psychiatric Genomics Consortium. Genetic Consortium for Anorexia Nervosa of the Wellcome Trust Case Control Consortium 3, Bulik-Sullivan B, Finucane HK, Anttila V, et al. An atlas of genetic correlations across human diseases and traits. Nat Genet. 2015;47:1236–41.
Grotzinger AD, Mallard TT, Akingbuwa WA, Ip HF, Adams MJ, Lewis CM, et al. Genetic architecture of 11 major psychiatric disorders at biobehavioral, functional genomic and molecular genetic levels of analysis. Nat Genet. 2022;54:548–59.
Brainstorm Consortium, Anttila V, Bulik-Sullivan B, Finucane HK, Walters RK, Bras J, et al. Analysis of shared heritability in common disorders of the brain. Science. 2018;360:eaap8757.
Cross-Disorder Group. of the Psychiatric Genomics Consortium. Electronic address: plee0@mgh.harvard.edu, Cross-Disorder Group of the Psychiatric Genomics Consortium. Genomic Relationships, Novel Loci, and Pleiotropic Mechanisms across Eight Psychiatric Disorders. Cell. 2019;179:1469–1482.e11.
Gerring ZF, Thorp JG, Gamazon ER, Derks EM. An analysis of genetically regulated gene expression and the role of co-expression networks across 16 psychiatric and substance use phenotypes. Eur J Hum Genet. 2022;30:560–6.
Wingo TS, Liu Y, Gerasimov ES, Vattathil SM, Wynne ME, Liu J, et al. Shared mechanisms across the major psychiatric and neurodegenerative diseases. Nat Commun. 2022;13:4314.
The Network and Pathway Analysis Subgroup of the Psychiatric Genomics Consortium. Psychiatric genome-wide association study analyses implicate neuronal, immune and histone pathways. Nat Neurosci. 2015;18:199–209.
Chu A, Wadhwa R. Selective Serotonin Reuptake Inhibitors. In: StatPearls [Internet]. Treasure Island (FL): StatPearls Publishing; 2024 [cited 2024 Apr 25]. Available from: http://www.ncbi.nlm.nih.gov/books/NBK554406/.
Santos R, Ursu O, Gaulton A, Bento AP, Donadi RS, Bologa CG, et al. A comprehensive map of molecular drug targets. Nat Rev Drug Discov. 2017;16:19–34.
Hackinger S, Zeggini E. Statistical methods to detect pleiotropy in human complex traits. Open Biol. 2017;7:170125.
23andMe Research Team, Social Science Genetic Association Consortium, Turley P, Walters RK, Maghzian O, Okbay A, et al. Multi-trait analysis of genome-wide association summary statistics using MTAG. Nat Genet. 2018;50:229–37.
Grotzinger AD, Rhemtulla M, de Vlaming R, Ritchie SJ, Mallard TT, Hill WD, et al. Genomic structural equation modelling provides insights into the multivariate genetic architecture of complex traits. Nat Hum Behav. 2019;3:513–25.
Solovieff N, Cotsapas C, Lee PH, Purcell SM, Smoller JW. Pleiotropy in complex traits: challenges and strategies. Nat Rev Genet. 2013;14:483–95.
Von Berg J, Ten Dam M, Van Der Laan SW, De Ridder J. PolarMorphism enables discovery of shared genetic variants across multiple traits from GWAS summary statistics. Bioinformatics. 2022;38:i212–9.
Ray D, Chatterjee N. A powerful method for pleiotropic analysis under composite null hypothesis identifies novel shared loci between Type 2 Diabetes and Prostate Cancer. Epstein MP, editor. PLOS Genet. 2020 Dec 8;16:e1009218.
Lee CH, Shi H, Pasaniuc B, Eskin E, Han B. PLEIO: a method to map and interpret pleiotropic loci with GWAS summary statistics. Am J Hum Genet. 2021;108:36–48.
Jordan DM, Verbanck M, Do R. HOPS: a quantitative score reveals pervasive horizontal pleiotropy in human genetic variation is driven by extreme polygenicity of human traits and diseases. Genome Biol. 2019;20:222.
Gleason KJ, Yang F, Pierce BL, He X, Chen LS. Primo: integration of multiple GWAS and omics QTL summary statistics for elucidation of molecular mechanisms of trait-associated SNPs and detection of pleiotropy in complex traits. Genome Biol. 2020;21:236.
Giambartolomei C, Vukcevic D, Schadt EE, Franke L, Hingorani AD, Wallace C, et al. Bayesian Test for Colocalisation between Pairs of Genetic Association Studies Using Summary Statistics. Williams SM, editor. PLoS Genet. 2014 May 15;10:e1004383.
Foley CN, Staley JR, Breen PG, Sun BB, Kirk PDW, Burgess S, et al. A fast and efficient colocalization algorithm for identifying shared genetic risk factors across multiple traits. Nat Commun. 2021;12:764.
Giambartolomei C, Zhenli Liu J, Zhang W, Hauberg M, Shi H, Boocock J, et al. A Bayesian framework for multiple trait colocalization from summary association statistics. Berger B, editor. Bioinformatics. 2018;34:2538–45.
Wu Y, Qi T, Wray NR, Visscher PM, Zeng J, Yang J. Joint analysis of GWAS and multi-omics QTL summary statistics reveals a large fraction of GWAS signals shared with molecular phenotypes. Cell Genomics. 2023;100344.
Bush WS, Oetjens MT, Crawford DC. Unravelling the human genome–phenome relationship using phenome-wide association studies. Nat Rev Genet. 2016;17:129–45.
Diogo D, Tian C, Franklin CS, Alanne-Kinnunen M, March M, Spencer CCA, et al. Phenome-wide association studies across large population cohorts support drug target validation. Nat Commun. 2018;9:4285.
Leppert B, Millard LAC, Riglin L, Davey Smith G, Thapar A, Tilling K et al. A cross-disorder PRS-pheWAS of 5 major psychiatric disorders in UK Biobank. Zhu X, editor. PLOS Genet. 2020;16:e1008185.
Hu JX, Thomas CE, Brunak S. Network biology concepts in complex disease comorbidities. Nat Rev Genet. 2016;17:615–29.
Pulley JM, Shirey-Rice JK, Lavieri RR, Jerome RN, Zaleski NM, Aronoff DM, et al. Accelerating precision drug development and drug repurposing by leveraging human genetics. ASSAY Drug Dev Technol. 2017;15:113–9.
Rastegar-Mojarad M, Ye Z, Kolesar JM, Hebbring SJ, Lin SM. Opportunities for drug repositioning from phenome-wide association studies. Nat Biotechnol. 2015;33:342–5.
Duffy Á, Petrazzini BO, Stein D, Park JK, Forrest IS, Gibson K, et al. Development of a human genetics-guided priority score for 19,365 genes and 399 drug indications. Nat Genet. 2024;56:51–9.
Burgess S, Butterworth A, Thompson SG. Mendelian randomization analysis with multiple genetic variants using summarized data. Genet Epidemiol. 2013;37:658–65.
Wootton RE, Jones HJ, Sallis HM. Mendelian randomisation for psychiatry: how does it work, and what can it tell us? Mol Psychiatry. 2022;27:53–7.
Zheng J, Baird D, Borges MC, Bowden J, Hemani G, Haycock P, et al. Recent developments in mendelian randomization studies. Curr Epidemiol Rep. 2017;4:330–45.
Pickrell JK, Berisa T, Liu JZ, Ségurel L, Tung JY, Hinds DA. Detection and interpretation of shared genetic influences on 42 human traits. Nat Genet. 2016;48:709–17.
Han B, Pouget JG, Slowikowski K, Stahl E, Lee CH, Diogo D, et al. A method to decipher pleiotropy by detecting underlying heterogeneity driven by hidden subgroups applied to autoimmune and neuropsychiatric diseases. Nat Genet. 2016;48:803–10.
Weissbrod O, Hormozdiari F, Benner C, Cui R, Ulirsch J, Gazal S, et al. Functionally informed fine-mapping and polygenic localization of complex trait heritability. Nat Genet. 2020;52:1355–63.
Wang G, Sarkar A, Carbonetto P, Stephens M. A simple new approach to variable selection in regression, with application to genetic fine mapping. J R Stat Soc Ser B Stat Methodol. 2020;82:1273–300.
Kichaev G, Roytman M, Johnson R, Eskin E, Lindström S, Kraft P, et al. Improved methods for multi-trait fine mapping of pleiotropic risk loci. Bioinformatics. 2017;33:248–55.
Smoller JW, Andreassen OA, Edenberg HJ, Faraone SV, Glatt SJ, Kendler KS. Psychiatric genetics and the structure of psychopathology. Mol Psychiatry. 2019;24:409–20.
Krueger RF, Caspi A, Moffitt TE, Silva PA. The structure and stability of common mental disorders (DSM-III-R): A longitudinal-epidemiological study. J Abnorm Psychol. 1998;107:216–27.
MacCallum RC, Austin JT. Applications of Structural Equation Modeling in Psychological Research. Annu Rev Psychol. 2000;51:201–26.
Caspi A, Houts RM, Belsky DW, Goldman-Mellor SJ, Harrington H, Israel S, et al. The p Factor: One general psychopathology factor in the structure of psychiatric disorders? Clin Psychol Sci. 2014;2:119–37.
Kendler KS, Prescott CA, Myers J, Neale MC. The structure of genetic and environmental risk factors for common psychiatric and substance use disorders in men and women. Arch Gen Psychiatry. 2003;60:929–37.
Waszczuk MA, Eaton NR, Krueger RF, Shackman AJ, Waldman ID, Zald DH, et al. Redefining phenotypes to advance psychiatric genetics: Implications from hierarchical taxonomy of psychopathology. J Abnorm Psychol. 2020;129:143–61.
Hatoum AS, Colbert SMC, Johnson EC, Huggett SB, Deak JD, Pathak GA, et al. Multivariate genome-wide association meta-analysis of over 1 million subjects identifies loci underlying multiple substance use disorders. Nat Ment Health. 2023;1:210–23.
Verhulst B, Maes HH, Neale MC. GW-SEM: A statistical package to conduct genome-wide structural equation modeling. Behav Genet. 2017;47:345–59.
Pritikin JN, Neale MC, Prom-Wormley EC, Clark SL, Verhulst B. GW-SEM 2.0: Efficient, flexible, and accessible multivariate GWAS. Behav Genet. 2021;51:343–57.
Ballard JL, O’Connor LJ. Shared components of heritability across genetically correlated traits. Am J Hum Genet. 2022;109:989–1006.
Grotzinger AD, Singh K, Miller-Fleming TW, Lam M, Mallard TT, Chen Y, et al. Transcriptome-wide structural equation modeling of 13 major psychiatric disorders for cross-disorder risk and drug repurposing. JAMA Psychiatry. 2023 Jun 14 [cited 2023 Jun 24]; Available from: https://jamanetwork.com/journals/jamapsychiatry/fullarticle/2806010.
Tomarken AJ, Waller NG. Structural equation modeling: strengths, limitations, and misconceptions. Annu Rev Clin Psychol. 2005;1:31–65.
Tarka P. An overview of structural equation modeling: its beginnings, historical development, usefulness and controversies in the social sciences. Qual Quant. 2018;52:313–54.
Hindorff LA, Sethupathy P, Junkins HA, Ramos EM, Mehta JP, Collins FS, et al. Potential etiologic and functional implications of genome-wide association loci for human diseases and traits. Proc Natl Acad Sci. 2009;106:9362–7.
Võsa U, Claringbould A, Westra HJ, Bonder MJ, Deelen P, Zeng B, et al. Large-scale cis- and trans-eQTL analyses identify thousands of genetic loci and polygenic scores that regulate blood gene expression. Nat Genet. 2021;53:1300–10.
Robins C, Liu Y, Fan W, Duong DM, Meigs J, Harerimana NV, et al. Genetic control of the human brain proteome. Am J Hum Genet. 2021;108:400–10.
Johnson ECB, Carter EK, Dammer EB, Duong DM, Gerasimov ES, Liu Y, et al. Large-scale deep multi-layer analysis of Alzheimer’s disease brain reveals strong proteomic disease-related changes not observed at the RNA level. Nat Neurosci. 2022;25:213–25.
Aguet F, Barbeira AN, Bonazzola R, Brown A, Castel SE, Jo B, et al. The GTEx Consortium atlas of genetic regulatory effects across human tissues [Internet]. Genetics; 2019 Oct [cited 2023 Apr 13]. Available from: http://biorxiv.org/lookup/doi/10.1101/787903.
Akbarian S, Liu C, Knowles JA, Vaccarino FM, Farnham PJ, Crawford GE, et al. The PsychENCODE project. Nat Neurosci. 2015;18:1707–12.
de Klein N, Tsai EA, Vochteloo M, Baird D, Huang Y, Chen CY, et al. Brain expression quantitative trait locus and network analyses reveal downstream effects and putative drivers for brain-related diseases. Nat Genet. 2023;55:377–88.
Ferkingstad E, Sulem P, Atlason BA, Sveinbjornsson G, Magnusson MI, Styrmisdottir EL, et al. Large-scale integration of the plasma proteome with genetics and disease. Nat Genet. 2021;53:1712–21.
Bennett DA, Buchman AS, Boyle PA, Barnes LL, Wilson RS, Schneider JA. Religious orders study and rush memory and aging project. J Alzheimers Dis JAD. 2018;64:S161–89.
Bryois J, Calini D, Macnair W, Foo L, Urich E, Ortmann W, et al. Cell-type-specific cis-eQTLs in eight human brain cell types identify novel risk genes for psychiatric and neurological disorders. Nat Neurosci. 2022;25:1104–12.
Herring CA, Simmons RK, Freytag S, Poppe D, Moffet JJD, Pflueger J, et al. Human prefrontal cortex gene regulatory dynamics from gestation to adulthood at single-cell resolution. Cell. 2022;185:4428–4447.e28.
Ding R, Wang Q, Gong L, Zhang T, Zou X, Xiong K, et al. scQTLbase: an integrated human single-cell eQTL database. Nucleic Acids Res. 2024;52:D1010–7.
Tarhan L, Bistline J, Chang J, Galloway B, Hanna E, Weitz E. Single Cell Portal: an interactive home for single-cell genomics data. 2023 [cited 2024 Apr 25]. Available from: http://biorxiv.org/lookup/doi/10.1101/2023.07.13.548886.
Gusev A, Ko A, Shi H, Bhatia G, Chung W, Penninx BWJH, et al. Integrative approaches for large-scale transcriptome-wide association studies. Nat Genet. 2016;48:245–52.
Reay WR, Geaghan MP, Atkins JR, Carr VJ, Green MJ, Cairns MJ. Genetics-informed precision treatment formulation in schizophrenia and bipolar disorder. Am J Hum Genet. 2022;109:1620–37.
Lu T, Forgetta V, Greenwood CMT, Zhou S, Richards JB. Circulating proteins influencing psychiatric disease: A Mendelian randomization study. Biol Psychiatry. 2023;93:82–91.
Schmidt AF, Finan C, Gordillo-Marañón M, Asselbergs FW, Freitag DF, Patel RS, et al. Genetic drug target validation using Mendelian randomisation. Nat Commun. 2020;11:3255.
Zuber V, Lewin A, Levin MG, Haglund A, Ben-Aicha S, Emanueli C, et al. Multi-response Mendelian randomization: Identification of shared and distinct exposures for multimorbidity and multiple related disease outcomes. Am J Hum Genet. 2023;110:1177–99.
Deng Y, Tu D, O’Callaghan CJ, Liu G, Xu W. Two-stage multivariate Mendelian randomization on multiple outcomes with mixed distributions. Stat Methods Med Res. 2023;32:1543–58.
Greco LA, Reay WR, Dayas CV, Cairns MJ. Exploring opportunities for drug repurposing and precision medicine in cannabis use disorder using genetics. Addict Biol. 2023;28:e13313. Aug
Antón-Galindo E, Cabana-Domínguez J, Torrico B, Corominas R, Cormand B, Fernàndez-Castillo N. The pleiotropic contribution of genes in dopaminergic and serotonergic pathways to addiction and related behavioral traits. Front Psychiatry. 2023;14:1293663.
Subramanian A, Narayan R, Corsello SM, Peck DD, Natoli TE, Lu X, et al. A next generation connectivity map: L1000 platform and the first 1,000,000 profiles. Cell. 2017;171:1437–1452.e17.
Wainberg M, Sinnott-Armstrong N, Mancuso N, Barbeira AN, Knowles DA, Golan D, et al. Opportunities and challenges for transcriptome-wide association studies. Nat Genet. 2019;51:592–9.
Burgess S, Davey Smith G, Davies NM, Dudbridge F, Gill D, Glymour MM, et al. Guidelines for performing Mendelian randomization investigations: update for summer 2023 [version 3; peer review: 2 approved]. Wellcome Open Res. 2023;4:186.
Zhu Z, Zhang F, Hu H, Bakshi A, Robinson MR, Powell JE, et al. Integration of summary data from GWAS and eQTL studies predicts complex trait gene targets. Nat Genet. 2016;48:481–7.
Park J, Lee D, Christakis NA, Barabási A. The impact of cellular networks on disease comorbidity. Mol Syst Biol. 2009;5:262.
Truong TTT, Panizzutti B, Kim JH, Walder K. Repurposing drugs via network analysis: opportunities for psychiatric disorders. Pharmaceutics. 2022;14:1464.
Barrio-Hernandez I, Beltrao P. Network analysis of genome-wide association studies for drug target prioritisation. Curr Opin Chem Biol. 2022;71:102206.
Barrio-Hernandez I, Schwartzentruber J, Shrivastava A, del-Toro N, Gonzalez A, Zhang Q, et al. Network expansion of genetic associations defines a pleiotropy map of human cell biology. Nat Genet. 2023;55:389–98.
Koch E, Kauppi K, Chen CH. Candidates for drug repurposing to address the cognitive symptoms in schizophrenia. Prog Neuropsychopharmacol Biol Psychiatry. 2023;120:110637.
MacNamara A, Nakic N, Amin Al Olama A, Guo C, Sieber KB, Hurle MR, et al. Network and pathway expansion of genetic disease associations identifies successful drug targets. Sci Rep. 2020;10:20970.
Mazaya M, Kwon YK. In Silico Pleiotropy Analysis in KEGG signaling networks using a Boolean network model. Biomolecules. 2022;12:1139.
Chavali S, Barrenas F, Kanduri K, Benson M. Network properties of human disease genes with pleiotropic effects. BMC Syst Biol. 2010;4:78.
He X, Zhang J. Toward a molecular understanding of pleiotropy. Genetics. 2006;173:1885–91.
Penrod NM, Cowper-Sal-lari R, Moore JH. Systems genetics for drug target discovery. Trends Pharmacol Sci. 2011;32:623–30.
Cheng Y, Sun M, Wang F, Geng X, Wang F. Identification of hub genes related to Alzheimer’s disease and major depressive disorder. Am J Alzheimers Dis Dementias®. 2021;36:15333175211046123.
Nithya C, Kiran M, Nagarajaram HA. Dissection of hubs and bottlenecks in a protein-protein interaction network. Comput Biol Chem. 2023;102:107802.
Mahboubi M, Azodi MZ, Tavirani MR, Mansouri V, Ahmadi NA, Hamdieh M, et al. Protein-Protein Interaction Analysis of Common Top Genes in Obsessive-Compulsive Disorder (OCD) and Schizophrenia: Towards New Drug Approach Obsessive-Compulsive disorder (OCD) and Schizophrenia Comorbidity Gene Analysis. Iran J Pharm Res IJPR. 2018;17:173.
Loscalzo J, Barabási AL, Silverman EK. Network Medicine: Complex Systems in Human Disease and Therapeutics. Harvard University Press; 2017.
Cheng F, Desai RJ, Handy DE, Wang R, Schneeweiss S, Barabási AL, et al. Network-based approach to prediction and population-based validation of in silico drug repurposing. Nat Commun. 2018;9:2691.
Guney E, Menche J, Vidal M, Barábasi AL. Network-based in silico drug efficacy screening. Nat Commun. 2016;7:10331.
Menche J, Sharma A, Kitsak M, Ghiassian SD, Vidal M, Loscalzo J, et al. Uncovering disease-disease relationships through the incomplete interactome. Science. 2015;347:1257601–1257601.
Guo MG, Sosa DN, Altman RB. Challenges and opportunities in network-based solutions for biological questions. Brief Bioinform. 2022;23:bbab437.
Kuhn M, Letunic I, Jensen LJ, Bork P. The SIDER database of drugs and side effects. Nucleic Acids Res. 2016;44:D1075–9.
Carss KJ, Deaton AM, Del Rio-Espinola A, Diogo D, Fielden M, Kulkarni DA, et al. Using human genetics to improve safety assessment of therapeutics. Nat Rev Drug Discov. 2023;22:145–62.
Millard LAC, Davies NM, Timpson NJ, Tilling K, Flach PA, Smith GD. MR-PheWAS: hypothesis prioritization among potential causal effects of body mass index on many outcomes, using Mendelian randomization. Sci Rep. 2015;5:16645.
Zheng J, Haberland V, Baird D, Walker V, Haycock PC, Hurle MR, et al. Phenome-wide Mendelian randomization mapping the influence of the plasma proteome on complex diseases. Nat Genet. 2020;52:1122–31.
Walker VM, Davey Smith G, Davies NM, Martin RM. Mendelian randomization: a novel approach for the prediction of adverse drug events and drug repurposing opportunities. Int J Epidemiol. 2017;46:2078–89.
Yan Q, editor. Pharmacogenomics in Drug Discovery and Development [Internet]. New York, NY: Springer US; 2022 [cited 2023 Nov 10]. (Methods in Molecular Biology; vol. 2547). Available from: https://link.springer.com/10.1007/978-1-0716-2573-6.
Fiscon G, Conte F, Farina L, Paci P. A comparison of network-based methods for drug repurposing along with an application to human complex diseases. Int J Mol Sci. 2022;23:3703.
Zhou F, Uddin S. Interpretable drug-to-drug network features for predicting adverse drug reactions. Healthcare. 2023;11:610.
Mohsen A, Tripathi LP, Mizuguchi K. Deep learning prediction of adverse drug reactions in drug discovery using open TG–GATEs and FAERS Databases. Front Drug Discov. 2021 [cited 2024 Jan 30];1. Available from: https://www.frontiersin.org/articles/10.3389/fddsv.2021.768792.
Sathyanarayanan A, Mueller TT, Ali Moni M, Schueler K, ECNP TWG Network members, Baune BT, et al. Multi-omics data integration methods and their applications in psychiatric disorders. Eur Neuropsychopharmacol J Eur Coll Neuropsychopharmacol. 2023;69:26–46.
Zong N, Wen A, Moon S, Fu S, Wang L, Zhao Y, et al. Computational drug repurposing based on electronic health records: a scoping review. Npj Digit Med. 2022;5:77.
Zhao S, Crouse W, Qian S, Luo K, Stephens M, He X. Adjusting for genetic confounders in transcriptome-wide association studies improves discovery of risk genes of complex traits. Nat Genet. 2024;56:336–47.
Thompson M, Gordon MG, Lu A, Tandon A, Halperin E, Gusev A, et al. Multi-context genetic modeling of transcriptional regulation resolves novel disease loci. Nat Commun. 2022;13:5704.
Mancuso N, Freund MK, Johnson R, Shi H, Kichaev G, Gusev A, et al. Probabilistic fine-mapping of transcriptome-wide association studies. Nat Genet. 2019;51:675–82.
Wu C, Pan W. A powerful fine-mapping method for transcriptome-wide association studies. Hum Genet. 2020;139:199–213.
Liu L, Yan R, Guo P, Ji J, Gong W, Xue F, et al. Conditional transcriptome-wide association study for fine-mapping candidate causal genes. Nat Genet. 2024; 26;1–9.
Zhou D, Jiang Y, Zhong X, Cox NJ, Liu C, Gamazon ER. A unified framework for joint-tissue transcriptome-wide association and Mendelian randomization analysis. Nat Genet. 2020;52:1239–46.
Burgess S, Thompson SG. Multivariable Mendelian randomization: the use of pleiotropic genetic variants to estimate causal effects. Am J Epidemiol. 2015;181:251–60.
Yuan Z, Zhu H, Zeng P, Yang S, Sun S, Yang C, et al. Testing and controlling for horizontal pleiotropy with probabilistic Mendelian randomization in transcriptome-wide association studies. Nat Commun. 2020;11:3861.
Porcu E, Rüeger S, Lepik K, eQTLGen Consortium, Agbessi M, Ahsan H, et al. Mendelian randomization integrating GWAS and eQTL data reveals genetic determinants of complex and clinical traits. Nat Commun. 2019;10:3300.
Baird DA, Liu JZ, Zheng J, Sieberts SK, Perumal T, Elsworth B, et al. Identifying drug targets for neurological and psychiatric disease via genetics and the brain transcriptome. Myers AJ, editor. PLOS Genet. 2021;17:e1009224.
Fabbri C, Pain O, Hagenaars SP, Lewis CM, Serretti A. Transcriptome-wide association study of treatment-resistant depression and depression subtypes for drug repurposing. Neuropsychopharmacology. 2021;46:1821–9.
Gaspar HA, Gerring Z, Hübel C, Major Depressive Disorder Working Group of the Psychiatric Genomics Consortium, Middeldorp CM, Derks EM, et al. Using genetic drug-target networks to develop new drug hypotheses for major depressive disorder. Transl Psychiatry. 2019;9:117.
Gedik H, Nguyen TH, Peterson RE, Chatzinakos C, Vladimirov VI, Riley BP, et al. Identifying potential risk genes and pathways for neuropsychiatric and substance use disorders using intermediate molecular mediator information. Front Genet. 2023;14:1191264.
Jiang JC, Hu C, McIntosh AM, Shah S. Investigating the potential anti-depressive mechanisms of statins: a transcriptomic and Mendelian randomization analysis. Transl Psychiatry. 2023;13:110.
Koch E, Demontis D. Drug repurposing candidates to treat core symptoms in autism spectrum disorder. Front Pharmacol. 2022;13:995439.
Li X, Shen A, Zhao Y, Xia J. Mendelian randomization using the druggable genome reveals genetically supported drug targets for psychiatric disorders. Schizophr Bull. 2023;49:1305–15.
Li W, Chen R, Feng L, Dang X, Liu J, Chen T, et al. Genome-wide meta-analysis, functional genomics and integrative analyses implicate new risk genes and therapeutic targets for anxiety disorders. Nat Hum Behav. 2023;8:361–79.
Liu J, Cheng Y, Li M, Zhang Z, Li T, Luo XJ. Genome-wide Mendelian randomization identifies actionable novel drug targets for psychiatric disorders. Neuropsychopharmacol Off Publ Am Coll Neuropsychopharmacol. 2023;48:270–80.
Liu J, Li X, Luo XJ. Proteome-wide association study provides insights into the genetic component of protein abundance in psychiatric disorders. Biol Psychiatry. 2021;90:781–9.
Rodriguez-López J, Arrojo M, Paz E, Páramo M, Costas J. Identification of relevant hub genes for early intervention at gene coexpression modules with altered predicted expression in schizophrenia. Prog Neuropsychopharmacol Biol Psychiatry. 2020;98:109815.
Wingo TS, Liu Y, Gerasimov ES, Gockley J, Logsdon BA, Duong DM, et al. Brain proteome-wide association study implicates novel proteins in depression pathogenesis. Nat Neurosci. 2021;24:810–7.
Wingo TS, Gerasimov ES, Liu Y, Duong DM, Vattathil SM, Lori A, et al. Integrating human brain proteomes with genome-wide association data implicates novel proteins in post-traumatic stress disorder. Mol Psychiatry. 2022;27:3075–84.
Woodward DJ, Thorp JG, Akosile W, Ong JS, Gamazon ER, Derks EM, et al. Identification of drug repurposing candidates for the treatment of anxiety: A genetic approach. Psychiatry Res. 2023;326:115343.
Zhou H, Kember RL, Deak JD, Xu H, Toikumo S, Yuan K, et al. Multi-ancestry study of the genetics of problematic alcohol use in over 1 million individuals. Nat Med. 2023;29:3184–92.
Gao H, Ni Y, Mo X, Li D, Teng S, Huang Q, et al. Drug repositioning based on network-specific core genes identifies potential drugs for the treatment of autism spectrum disorder in children. Comput Struct Biotechnol J. 2021;19:3908–21.
Truong TTT, Liu ZSJ, Panizzutti B, Kim JH, Dean OM, Berk M, et al. Network-based drug repurposing for schizophrenia. Neuropsychopharmacology. 2024; 6:1–10.
Zhou M, Wang Q, Zheng C, John Rush A, Volkow ND, Xu R. Drug repurposing for opioid use disorders: integration of computational prediction, clinical corroboration, and mechanism of action analyses. Mol Psychiatry. 2021;26:5286–96.
Funding
EMD is supported by an NHMRC Investigator grant (2026364). JGT is supported by an NHMRC Investigator Grant (2027002). EMD and ZFG are supported by NIH/NIA AG068026.
Author information
Authors and Affiliations
Contributions
DJW: Writing original draft; Writing – review and editing, literature review; Figure creation. ZFG, EMD & JGT: Conceptualisation; Supervision; Writing – review and editing. CMM & WA: Writing – review and editing.
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Woodward, D.J., Thorp, J.G., Middeldorp, C.M. et al. Leveraging pleiotropy for the improved treatment of psychiatric disorders. Mol Psychiatry (2024). https://doi.org/10.1038/s41380-024-02771-7
Received:
Revised:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41380-024-02771-7