Approximately half of all people experience at least one psychiatric disorder in their lifetime, and many develop multiple conditions, which makes diagnosis difficult. These disorders are defined by symptoms rather than clear biological mechanisms, which creates further challenges in distinguishing them. Genomic research has identified hundreds of associated genetic loci. Many of these are shared across multiple disorders. This demonstrates the strong genetic link between conditions. The current research is the major project of the PGC cross-disorder group. This study aimed to examine the unique and shared effects of common genetic variants across 14 psychiatric disorders.
All genome-wide associated studies (GWAS) summary statistics underwent standard quality control, including removing ambiguous single-nucleotide polymorphisms (SNPs), low-frequency variants, and those with poor imputations. SNPs with limited sample and the major histocompatibility complex (MHC) region were excluded from this study. A five-factor model was used for downstream analysis, and a stratified genomic structural equation modelling (SEM) study assessed enrichment across 162 functional annotations. Multivariate GWAS statistics were utilized for the estimation of heterogeneity and factor-level associations. Additional methods like mixture of regression (MiXeR), cross-ancestry tests, case-case genome-wide association study (CC-GWAS), and local analysis of covariant association (LAVA) were used for the evaluation of case-care differences, polygenicity, and local correlations. Functional annotation mapped variants to genes and tested developmental and cell-type enrichment.
Researchers analyzed GWAS data for 14 psychiatric disorders using updated datasets with about 165% case enhancement over cross-disorder group 2 (CDG2) and added six new conditions, including three substance use disorders (SUDs). Sample sizes varied. Analysis focused on European (EUR)-like ancestry with additional bivariate tests in East Asian (EAS)-like and African (AFR)-like groups. Linkage disequilibrium score regression (LDSC) analyzed widespread genetic associations, including major depression-schizophrenia (MD-SCZ) genetic collection (rg) of 0.22 and standard error (s.e.) of 0.04 in the EUR-like group, whereas rg of 0.45 and s.e. of 0.09 were observed in the EAS-like group. Popcorn indicated cross-ancestry similarity, highest for SCZ (genetic effect correlation across populations [ρgi] = 0.85, s.e. = 0.04) with lower association for post-traumatic stress disorder (PTSD [ρgi = 0.59, s.e. = 0.27]) and MD (ρgi = 0.67, s.e. = 0.16).
MiXeR demonstrated strong genetic overlap across all psychiatric disorders compared to LDSC. It revealed that most disorders shared many causal variants, even when effect directions differed. Shared risk mainly comes from concordant variants, while disorder-specific risk comes from fewer discordant or unique variants. Analyses included eight disorders, such as anorexia nervosa (AN), attention-deficit/hyperactivity disorder (ADHD), SCZ, alcohol use disorder (AUD), PTSD, MD, bipolar disorder (BIP), and anxiety disorders (ANX). Genomic SEM analysis detected five latent genetic factors like SUD, internalizing, suicidality/behavior (SB), compulsive, and neurodevelopmental with a standardized root mean square residual (SRMR) of 0.063 and a comparative fit index (CFI) of 0.971. Median unexplained variance was found to be 33.5%. Tourette syndrome (TS) had 87% unique genetic signal. SUD and internalizing showed the strongest correlation with rg = 0.60. A higher-order p-factor also fits well with the CFI of 0.959.
Internalizing and SUD were linked to lower income (rg_Internalizing = −0.40, s.e. = 0.02; rg_SUD = −0.41, s.e. = 0.03) and weaker cognitive scores. SUD uniquely correlated with verbal numerical reasoning thinking with rg_SUD = −0.41, s.e. = 0.03, and adult intelligence with rg_SUD = −0.40, s.e. = 0.03. Neurodevelopmental factors strongly associated with childhood aggression (rg = 0.94, s.e. = 0.10). The p-factor strongly correlated with suicide attempts (rg_p = 0.87, s.e. = 0.03), loneliness (rg_p = 0.62, s.e. = 0.02), self-harm (rg_p = 0.74, s.e. = 0.04), stress (rg_p = 0.50, s.e. = 0.02), and neuroticism (rg_p = 0.64, s.e. = 0.02).
LAVA found 458 significant local regions and 101 hotspots, including a major region on chromosome 11 containing neural cell adhesion molecule 1 (NCAM1)- tetratricopeptide repeat domain 12 (TTC12)- ankyrin repeat and kinase domain containing 1 (ANKK1)- dopamine receptor D2 (DRD2), affecting eight disorders. Multivariate GWAS identified 295 loci, including 48 novel hits across factors.
This study’s limitations include a focus on EUR-like populations due to limited non-EUR GWAS data. Cross-ancestry correlations for PTSD and MD may be affected by variable sample sizes, heterogeneity, assortative mating, and diagnostic misclassification.
In conclusion, this study mapped cross-disorder psychiatric genetics and revealed high-overlap disorder subtypes, shared biological pathways, and implications for therapeutics and diagnostics. Future research is needed to study repurposed or novel interventions.
Reference: Grotzinger AD, Werme J, Peyrot WJ, et al. Mapping the genetic landscape across 14 psychiatric disorders. Nature. 2025. doi:10.1038/s41586-025-09820-3
