Genomic Structural Equation Modeling Reveals Latent Phenotypes in the Human Cortex with Distinct Genetic Architecture

Area of Science:

• Neurogenetics
• Brain Imaging Genetics
• Quantitative Genetics

Background:

• Genetic contributions to human cortical structure exhibit pleiotropy.
• This pleiotropy can define unique, neurobiologically distinct cortical parcellations.
• Existing parcellation schemes may not fully capture genetic influences.

Approach:

• Applied genomic structural equation modeling (SEM) to map genetic architecture of cortical surface area (SA) and thickness (CT).
• Utilized LD score regression (LDSC) on GWAS summary statistics to estimate genetic covariance and identify genetically informed brain networks (GIBNs).
• Performed multivariate genome-wide association studies (GWAS) on identified GIBNs.

Key Points:

• Identified 6 GIBNs for cortical SA and 4 GIBNs for CT.
• Discovered 74 genome-wide significant loci associated with GIBNs, many linked to neuroimaging, behavioral, and psychiatric traits.
• SA-derived GIBNs showed genetic correlations with bipolar disorder, cannabis use disorder, ADHD, MDD, and insomnia.
• CT-derived GIBNs showed a negative genetic correlation with alcohol dependence.

Conclusions:

• Genomic SEM effectively identifies GIBNs, offering a novel approach to cortical parcellation.
• GIBNs provide insights into the genetic underpinnings of complex brain disorders and traits.
• Jointly modeling genetic architecture across phenotypes enhances understanding of brain-region-specific genetic influences.