Multivariate genome-wide association study dissects shared biology and disorder-specific loci across internalizing spectrum in millions of ancestrally diverse participants

This study leverages a large-scale, multi-ancestry multivariate genome-wide association study of millions of participants to identify thousands of genetic loci, revealing that while generalized anxiety disorder and posttraumatic stress disorder share broad genetic mechanisms with a latent internalizing factor, major depressive disorder exhibits a significant proportion of disorder-specific genetic signals linked to distinct neurodevelopmental and stress-responsive pathways.

The Gist

Imagine your brain's mental health as a vast, bustling city. In this city, there are three major districts that often get crowded and chaotic: Generalized Anxiety (GAD), Major Depression (MDD), and Post-Traumatic Stress (PTSD).

For a long time, scientists knew these districts shared some of the same streets and buildings (they often happen together and share genetic roots), but they didn't know exactly which parts were shared and which parts were unique to each district.

This new study is like a massive, high-tech city planner surveying millions of residents from all over the world (including diverse ancestry groups) to draw a perfect map of this city. Here is what they found, broken down simply:

1. The "Common Ground" vs. The "Unique Neighborhoods"

The researchers discovered that these three disorders are like three houses built on the same foundation.

• The Shared Foundation (The "INT" Factor): A huge chunk of the genetic risk for all three disorders is the same. It's like a shared plumbing system or a common electrical grid that runs through all three houses. If this system has a glitch, it can cause problems in any of the houses. The study found that for Anxiety (GAD) and PTSD, almost their entire structure is built on this shared foundation.
• The Unique Additions: However, Depression (MDD) is different. While it shares the foundation, about 38% of its structure is unique. It has its own special rooms, wiring, and decorations that the other two don't have. This explains why depression can sometimes feel distinct from anxiety or PTSD, even though they are cousins.

2. Finding the "Broken Bricks" (Genetic Loci)

The team looked at the DNA of over 6 million people (a massive crowd!). They found thousands of specific "bricks" in our genetic code (called loci) that are associated with these conditions.

• They found 248 new shared bricks that affect the common foundation.
• They found hundreds of new bricks specific to each disorder.
• The Diversity Win: Unlike many past studies that only looked at people of European descent, this team included people of African, Asian, and Hispanic ancestry. This is like checking the blueprints for the city in different neighborhoods to make sure the map is accurate for everyone, not just one group. They found new genetic clues that were previously invisible because they were specific to these diverse groups.

3. The "Factory" Analogy (How Genes Work)

Once they found the broken bricks, they asked: What factory inside our cells is malfunctioning because of these bricks?

• The Shared Malfunction: The shared genetic risk seems to come from "broad" factory issues. Imagine a factory that makes the basic building blocks for neurons (brain cells). If the supply chain for these basic blocks is slow or messy, it affects the whole city. This involves general cell regulation and how cells talk to each other.
• The Specific Malfunction: The unique parts of Depression, for example, seem to come from "specialized" factories. These are like factories that build specific tools for handling stress, memory, or synaptic connections (the bridges between brain cells). When these specific tools break, you get the unique symptoms of that specific disorder.

4. The "Gut-Brain" Connection

The study also looked at how these mental health districts connect to the rest of the body. They found a strong "highway" connecting the brain to the gut.

• People with these genetic risks were also more likely to have issues like constipation, stomach pain, or digestive disorders.
• Analogy: Think of the gut and the brain as two cities connected by a super-highway. If there's a traffic jam in the mental health city, it often causes a traffic jam in the gut city, and vice versa. This confirms the idea that "your gut feeling" is literally part of your genetic makeup.

5. New Keys for Old Locks (Drug Repurposing)

Finally, the researchers asked: Can we use existing keys (drugs) to fix these broken locks?

• They found that some drugs currently used for other things (like painkillers, hormones, or even antipsychotics) might target the specific biological pathways they discovered.
• For example, they found that drugs affecting hormones might help with depression, and drugs that target pain pathways might help with anxiety. This is like realizing a wrench used to fix a car engine might also be the perfect tool to fix a leaky faucet in the brain.

The Big Takeaway

This study is a giant leap forward because it stops treating these disorders as completely separate islands. Instead, it shows they are a spectrum.

• Anxiety and PTSD are almost entirely about the "shared" genetic risk.
• Depression is a mix of shared risk and its own unique, specialized genetic quirks.

By understanding exactly which parts of the genetic "city" are shared and which are unique, doctors can eventually move away from a "one-size-fits-all" approach. In the future, treatments could be tailored to fix the specific "broken brick" or "malfunctioning factory" causing a person's specific symptoms, rather than just treating the label on the box.

Technical Summary

1. Problem Statement

Generalized Anxiety Disorder (GAD), Major Depressive Disorder (MDD), and Posttraumatic Stress Disorder (PTSD) are highly prevalent psychiatric conditions that frequently co-occur and share substantial genetic correlations. While previous large-scale studies (e.g., by the Psychiatric Genomics Consortium) have identified a shared "internalizing" genetic factor (INT), critical gaps remain:

• Lack of Specificity: It is unclear to what extent the genetic architecture is driven by shared mechanisms versus disorder-specific pathways.
• Methodological Limitations: Previous cross-disorder studies often used factor-analytic models that capture shared variance but fail to delineate the molecular pathways differentiating individual disorders.
• Ancestry Bias: Most prior studies were limited to European ancestry, hindering the understanding of genetic architecture across diverse populations.
• Molecular Mechanisms: The specific regulatory, cellular, and molecular signatures distinguishing these disorders remain poorly characterized.

2. Methodology

The authors conducted a massive, multivariate, and multi-ancestry Genome-Wide Association Study (GWAS) integrating data from multiple cohorts (UK Biobank, All of Us, MVP, PGC, 23andMe, FinnGen, iPSYCH).

• Sample Size:
  • GAD: 1,358,762 participants (123,006 cases).
  • MDD: 3,601,629 participants (685,929 cases).
  • PTSD: 1,617,876 participants (179,729 cases).
  • Ancestry: Included European (EUR), African (AFR), Admixed-American (AMR), and East Asian (EAS) populations.
• Statistical Frameworks:
  • Genomic Structural Equation Modeling (gSEM): Used to model a latent internalizing factor (INT) shared across GAD, MDD, and PTSD.
  • GWAS-by-Subtraction: A method to isolate disorder-specific genetic variance (e.g., MDD_SUB, PTSD_SUB) by regressing out the shared INT factor.
  • Cross-Ancestry Meta-Analysis: Combined results across diverse populations using METAL.
  • Fine-Mapping: Utilized SuSiEx to identify putative causal variants (Posterior Inclusion Probability, PIP > 0.8) across ancestries.
  • Multi-Omics Integration:
    • Colocalization (HyPrColoc): Integrated GWAS with brain-specific eQTLs, mQTLs, pQTLs, and scQTLs.
    • Gene-Based Analysis: MAGMA and PASCAL for gene prioritization.
    • TWAS/PWAS: Transcriptome-wide (S-PrediXcan) and Proteome-wide (FUSION) association studies using GTEx and ROSMAP/BSHRI data.
    • Mendelian Randomization: Summary-based MR (SMR) and Latent Causal Variable (LCV) models to infer causality.
  • Functional Enrichment: GSA-MiXeR for Gene Ontology (GO) enrichment and drug repurposing (gene2drug, DRUGSETS).

3. Key Contributions

• Scale and Diversity: The largest multi-ancestry GWAS to date for the internalizing spectrum, significantly increasing the representation of non-European populations (AFR, AMR, EAS).
• Disentangling Shared vs. Specific Risk: Successfully quantified the proportion of genetic risk that is shared (INT) versus disorder-specific, revealing that MDD has a significantly larger unique component compared to GAD and PTSD.
• Causal Variant Discovery: Identified >450 high-confidence causal variants through cross-ancestry fine-mapping, linking them to >1,250 genes.
• Hierarchical Genetic Model: Proposed a model where shared liability is driven by broad cellular/regulatory mechanisms, while disorder-specific risk involves specialized neurodevelopmental and synaptic pathways.

4. Key Results

A. Gene Discovery and Loci

• Total Loci Identified:
  • INT Factor: 248 loci (96 novel).
  • MDD: 591 loci (241 novel).
  • PTSD: 237 loci (113 novel).
  • GAD: 109 loci (63 novel).
• Overlap: GAD and PTSD genetic risks showed strong overlap with the INT factor. In contrast, ~38% of MDD genetic signals were disorder-specific, independent of the shared internalizing liability.
• Fine-Mapping: Identified 455 putative causal variants. Notable shared causal loci included TSNARE1 and MACROD2 (INT/PTSD), FURIN (GAD/MDD), and SOX5, EXOC4, LINGO1 (INT).
• Population-Specific Findings: Novel loci were discovered in AFR (GAD), AMR (MDD), and EAS (MDD, PTSD), highlighting the importance of diverse cohorts.

B. Multi-Omics Characterization

• Shared Genes: 193 genes were associated with at least one phenotype. High-confidence shared genes included LRFN5, KPNA2, DRD2, and RNF123, implicating synaptic organization, nuclear transport, and dopaminergic signaling.
• Disorder-Specific Genes:
  • MDD: Enriched for genes involved in translational regulation (TRMT61A), kinase signaling (MARK3), and mitochondrial metabolism (PTPMT1).
  • GAD: Linked to epigenetic regulation (SFMBT1) and immune-brain interactions (TLR9).
  • PTSD: Associated with mitochondrial trafficking (RHOT1) and activity-dependent mRNA translation (CPEB3).
• TWAS/PWAS: Identified 90 significant gene associations for INT and 71 protein-coding genes. CTNND1 was the most significant protein for INT; GPX1 was significant for MDD and PTSD.

C. Functional Enrichment and Pathways

• Shared Biology: The INT factor was strongly enriched for neurogenesis, neuron differentiation, and synapse formation.
• Disorder-Specific Biology:
  • GAD: Unique enrichment for the Node of Ranvier (GO:0033268).
  • MDD_SUB & PTSD_SUB: Unique enrichment for microtubule bundle formation.
• Drug Repurposing:
  • Shared: Fluspirilene (antipsychotic) and Abamectin showed potential for multiple internalizing disorders.
  • Specific: Lovastatin (GAD), hormonal regulators (e.g., Clomifene, Desogestrel), and pain medications (Gabapentin) were identified as candidates.

D. Pleiotropy and Causal Inference

• Psychiatric Overlap: High genetic correlations were found with Bipolar Disorder (BIP) and Schizophrenia (SCZ). Notably, the disorder-specific component of PTSD (PTSD_SUB) showed much lower correlation with BIP than the overall PTSD phenotype, suggesting BIP-PTSD comorbidity is driven by shared external factors (trauma) rather than specific genetic mechanisms.
• Physical Health: Strong genetic correlations with gastrointestinal conditions (constipation, gastritis), supporting the gut-brain axis.
• Causal Relationships:
  • GAD: Putative causal effect on abnormal neurological movements, diaphragmatic hernia, and spinal stenosis.
  • MDD: Putative causal effect on acute respiratory infections and back pain.

5. Significance and Implications

• Refined Nosology: The study provides empirical support for a hierarchical model of internalizing psychopathology, distinguishing a broad "internalizing" liability from specific disorder mechanisms.
• Therapeutic Targets: By identifying both shared and specific pathways, the study offers a roadmap for drug repurposing. Shared targets (e.g., synaptic regulation) may treat the broad internalizing spectrum, while specific targets (e.g., mitochondrial metabolism for MDD) could lead to more precise treatments.
• Equity in Genetics: The inclusion of diverse ancestries uncovered novel loci missed in European-only studies, reducing health disparities in genetic research.
• Clinical Insight: The identification of causal links between internalizing disorders and physical conditions (e.g., spinal stenosis, gut health) suggests that treating these disorders may require a holistic approach addressing both mental and physical comorbidities.

Limitations Noted: The study acknowledges potential misclassification due to self-reported data, the continued overrepresentation of European ancestry despite improvements, and the possibility that the large MDD sample size inflated the detection of disorder-specific signals compared to GAD and PTSD.

Conclusion: This study represents a major advance in understanding the genetic architecture of internalizing disorders, moving beyond simple correlation to dissect the specific molecular and cellular mechanisms that define shared risk versus disorder-specific pathology.