31 - GENETIC FACTORS INFLUENCING THE COMMON NEUROBIOLOGICAL SUBSTRATE FOR MENTAL DISORDERS

Abstract

Background A multitiered imaging study found grey matter loss to converge across diagnoses in three brain regions within the right and left anterior insula and in part of the dorsal anterior cingulate (Goodkind, Eickhoff et al. 2015, JAMA Psychiatry 72, 305–315). These regions thus form a common neurobiological substrate of mental disorders. The aim of our study was the identification of genetic factors contributing to the volumes of these regions in healthy individuals to obtain insights into the biological processes involved. Evidence from structural and functional connectivity analyses indicates that the common neurobiological substrate is a closely connected network. Thus, we analyzed the regions jointly. Methods Analyses were carried out in healthy subjects of five population-based cohorts: 1000BRAINS (n=539), CONNECT100 (n=93), BiDirect (n=589), SHIP-2 (n=1142), and SHIP-Trend (n=896). Voxel-based volumes were corrected for covariates (head/total brain size, sex, age, handedness). Principal component analysis on the corrected volumes was used to generate a common measure, the Component of the Common Substrate (CCS). Heritability of the CCS was estimated with GCTA. A genome-wide association study (GWAS) on the CCS was conducted in each cohort using imputed variants (1000 Genomes Phase I panel). GWAS was carried out in a two-stage design with the first four cohorts as discovery sample and SHIP-Trend for replication. GWAS were combined using a fixed-effects pooled analysis. Pathway analysis was conducted in MAGENTA using a narrow gene definition (0 kb flanking regions) and Reactome pathways; cutoff was the 75th percentile of gene p-values, p-values were corrected by FDR. Results The significant pooled estimate of the SNP-based heritability of the CCS across all cohorts was 25%. In the discovery-stage GWAS, three blocks of highly correlated variants at chromosome 5q35.2 reached genome-wide significance. Association of the lead variant in one block was significant in the replication cohort and was thus confirmed. Follow-up analyses showed that the association was driven by the insulae and supported in all three regions. Association with total grey matter volume was lower than with any CCS region. Associated variants did not overlap with protein-coding genes, yet poorly characterized non-coding RNAs map to the area. Further analyses identified six associated pathways that intersected with either neurodevelopment or with neurodegeneration. Finally, we examined whether genetic variants shaping the CCS overlapped with variants influencing the risk for mental disorders using LD score regression, polygenic risk scores and sign tests. No overlap of variants associated in recent publications with either schizophrenia, bipolar disorder or MDD was found. Discussion We discovered and replicated a significant association of intergenic SNPs with brain regions representing a common neurobiological substrate for mental disorders. No genetic correlation between variants associated with these brain volumes and variants associated with mental disorders was observed, in line with a previous publication that did not find a correlation between schizophrenia-associated and volume-associated variants from subcortical brain regions and intracranial volume (Franke et al. 2016, Nat Neurosci 19, 420–431). These regions were distinct from the regions studied here. The associations should be explored in future studies using other measures as well as functional imaging and neuropsychological approaches.