Mental disorders share clinical features, cognitive deficits, and drug responses, suggesting a potential for treating them along symptom spectrums rather than as distinct categories. High heritability and influence from many genetic variants of small effect characterize the genetic architecture of many mental illnesses, cognitive traits, and brain morphology traits. Recent research has revealed shared genetic risk factors among mental disorders and associated traits, often identified at the single nucleotide polymorphism (SNP) level. However, there's a lack of systematic studies investigating genetic overlaps at the gene level. This study addresses this gap by examining pleiotropy at the gene level, focusing on the association of multiple genetic variants within a locus with different traits. Allelic heterogeneity, where different variants within a gene are independently associated with the same trait, is also considered. The study uses GWAS summary statistics for various mental disorders, brain morphometric traits (subcortical brain volumes), cognitive traits (educational attainment, general cognitive function), and personality traits (subjective well-being, depressive symptoms, neuroticism, extraversion, openness to experience, agreeableness, conscientiousness, children’s aggressive behavior, loneliness) to identify pleiotropy at the gene level. The rationale for selecting these traits is based on their reported associations with mental disorders and potential roles as endophenotypes.

Previous studies have reported genetic overlaps within mental disorders and between psychiatric and cognitive traits, typically at the SNP level using methods that account for polygenic effects and linkage disequilibrium. The Brainstorm consortium's atlas of genetic correlation, for example, revealed genetic overlap at the single-marker level between several of the traits included in this study, showing correlations among ADHD, BPD, MDD, and schizophrenia, along with some personality and cognitive traits. However, these studies didn't examine genetic overlaps at the gene level. Studies on cognitive functions, brain volume imaging traits, and personality traits were also reviewed to establish their established links to mental disorders and their potential role as endophenotypes. The literature indicates consistent cognitive deficits in disorders like schizophrenia, bipolar disorder, and autism. GWAS studies have successfully identified genetic factors for educational attainment, general cognitive function, and intelligence, with demonstrated genetic overlap between intelligence and schizophrenia/bipolar disorder, and between educational attainment and schizophrenia. Brain volume imaging research from the ENIGMA consortium has also shown differences in brain volumes between patients with schizophrenia and healthy controls. Finally, the role of personality traits as risk factors for, and modifiers of, mental disorders and their known genetic associations were discussed.

Summary statistics from GWASs for 27 traits (7 mental disorders – schizophrenia, autism spectrum disorder, major depressive disorder, anorexia nervosa, ADHD, bipolar disorder, and anxiety; 8 brain morphometric traits; 2 cognitive traits – educational attainment and general cognitive function; 9 personality traits) were obtained and subjected to quality control procedures. These procedures involved filtering out poorly imputed SNPs (imputation score < 0.9), ambiguous SNPs, markers with minor allele frequencies < 0.1, and insertions/deletions. The genome mapping was based on the human genome reference hg19. Independent signals of association were selected for each trait using conditional stepwise regression with the cojo-GCTA tool, which is more robust than LD pruning for identifying independent and conjunctional association signals. Selected independent SNPs were then annotated to known RefSeq genes (±10 kb). Genes with overlapping SNPs were merged into gene_blocks. The analysis identified genetic overlaps using two approaches: a SNP-based analysis (selecting genes based on the single SNP minimal *p*-value) and a gene-based analysis (using Brown scores to evaluate the combined effect of markers within a gene). Three scenarios were considered for genetic overlaps: Scenario I (pleiotropy at the gene level – multiple associations to LD-independent genetic variants for at least two traits); Scenario II (allelic heterogeneity – multiple associations to LD-independent genetic variants within a single trait); and Scenario III (pleiotropy at the SNP level – multiple associations to LD-dependent genetic variants for more than one trait). A locus was considered to contain dependent markers if at least one pair of SNPs associated with different traits had r² ≥ 0.2. For the gene-based analysis, Brown scores were calculated for each gene and each trait, and those passing a study-wide FDR threshold of 0.0018 were selected. Haplotype analysis was performed using Haploview to examine potential haplotype associations in independent associations, using the MAN2A1 gene as an example. Finally, a comparison was made between genetic overlaps at the gene level and at the SNP level.

The analysis identified 2190 unique associated SNPs after conditional regression. Of these, 1415 were annotated to 1410 genes, which were then grouped into 1161 gene groups. Among these gene groups, 226 showed significant association with at least two traits. Applying the experiment-wide significance threshold (5.70 x 10⁻⁶), the study observed three scenarios: Scenario I (Pleiotropy at the gene level): 25 genes displayed independent associations with two traits. Examples include *RERE*, *LOC102724552*, *GPM6A* (SCZ and educational attainment); *RBKS* (SCZ and BPD); and *GLIS3* (SCZ and neuroticism). Scenario II (Allelic heterogeneity): 9 genes showed multiple independent signals associated with one trait, including *BSN* and *CTNNA3* for educational attainment, and *RPS6KA2* for BPD. Scenario III (Pleiotropy at the SNP level): 47 loci (53 genes) showed associations where the same SNP, or SNPs in LD (r² ≥ 0.2), were associated with different traits. The gene-based analysis identified 379 genes significantly associated with at least two traits, with 23 genes showing additional pleiotropy (e.g., *LRRN2*, *MAN2A1*, *EYS* (SCZ and educational attainment); *CLU*, *MIR6843* (SCZ and Alzheimer's disease); *SHANK3* (educational attainment and gF)). Haplotype analysis of *MAN2A1* revealed that the alleles associated with schizophrenia and educational attainment were located on different haplotypes. Comparison of genetic overlaps at the gene and SNP levels revealed more pleiotropy at the SNP level for most trait pairs, but also showed instances of gene-level pleiotropy not observed at the SNP level. The study also identified 104 additional genes with LD-linked markers and 48 genes with a mixed case scenario.

This study's main contribution is the identification of 48 genes exhibiting pleiotropy at the gene level across a range of mental disorders and related traits. This finding expands on previous research that has primarily focused on pleiotropy at the SNP level, highlighting the importance of considering gene-level pleiotropy for functional studies and drug targeting. The observed variety of genetic overlaps, including pleiotropy at the gene and SNP levels and allelic heterogeneity, reflects the complexity of the genomic landscape and the intricate relationships among these traits. The identification of pleiotropic genes provides valuable insights into the shared biological mechanisms underlying these diverse phenotypes. The authors discuss potential explanations for the independent associations observed, including associations in different parts of a gene, different genes within a gene group, or different haplotypes within the same gene region. They highlight the potential importance of considering synthetic associations with rare variants, which are difficult to capture with GWAS but may contribute to the observed pleiotropy. The study also notes that some associations may be due to effects on the regulation of a nearby gene rather than the gene itself. The findings suggest prioritizing these genes for deep genotyping or sequencing to identify rare variants and to understand which isoforms or functional variants are associated with the different independent associations.

This study demonstrates the presence of gene-based pleiotropy between psychiatric disorders and related traits, expanding on previous findings of SNP-level pleiotropy. The identification of 48 genes with at least two independent associations across different traits provides crucial information for functional analyses and drug target identification. Future research should focus on deep sequencing of these genes to identify rare variants and to determine if independent signals are due to different functional variants or isoforms. Understanding the relationship between pleiotropic genes in different disorders is crucial to avoid targeting genetic pathways that might have unintended consequences. This study emphasizes the need for a systematic approach that incorporates different layers of genetic evidence across related phenotypes to enhance our understanding of the genetic architecture of mental disorders.

The study used publicly available summary statistics, raising the possibility of biases due to overlapping cohorts within meta-analyses. While efforts were made to reduce overlap, caution is warranted when interpreting overlaps between traits sharing many cohorts. The analysis was limited to gene-level annotations, neglecting potential effects of regulatory regions. Differences in sample sizes and trait heritability across GWASs may have affected the power to detect pleiotropy. Finally, using a merged Norwegian and German sample as a European population proxy for the cojo-GCTA step might have introduced some noise into the initial signal selection and haplotype detection.