Abstract
Abstract
Background
Suicide is a major problem in severe mental illness (SMI), with increased risk for lifetime suicidal ideation and behavior compounded in veterans. Previous research has suggested that genomic risk factors exist for suicidal ideation and behavior, and we examined the genomic correlates of suicidal ideation and behavior in 8,049 male and 1,290 female veterans with schizophrenia or bipolar disorder.
Methods
Data were collected previously on demographic, clinical, and functional status factors, as well as self-reported suicidal ideation and behavior assessed with the Columbia Suicide Severity Rating Scale (lifetime version). For this study, we defined an ideation and behavior as a single latent trait, excluding patients with non-suicidal self-injurious behavior. A customized Affymetrix Axiom Biobank array was used to obtain genotype data. After sample and variant quality control steps and ancestry inference, analyses included 657,459 SNPs and 4,012 European American samples (with 3,223 cases and 789 controls, which include 3,507 males and 505 females), as well as 2,651 African American samples (with 1,955 cases and 696 controls, which include 2,249 males and 402 females). The results were then filtered with the imputation quality score R2>0.8, MAF>0.01, and Hardy-Weinberg equilibrium test p-value >1 x 10–6; a GWAS p-value threshold of 5 x 10–8 was adopted as genome-wide statistical significance. We performed both SNP-based analysis and gene-based analysis (using the UTMOST approach that leverages eQTL information across tissues from the GTEx project). To investigate the genetic correlation between the ideation and behavior score in the combined sample and other complex traits, we applied GNOVA, a framework to estimate the genetic overlap of ideation and behavior scores in schizophrenia or bipolar samples with respect to 2,626 traits from the UK Biobank (UKB) and other GWAS results. FDR was used to correct for multiple testing, and a q value<0.05 was set as the threshold for statistical significance.
Results
For European American (EA) and African American (AA) ancestry patients, no SNP passed the 5 x 10–8 the genome wide statistical significance; 4 SNPs were significant at 10–6 for the EA group, and 8 SNPs for the AA group. Gene-level analysis (UTMOST) identified 5 genes (GAMT 1.74×10–11, FTL 1.84×10–11, DCTN4 2.97×10–9, IYVE1 4.38×10–9, IQCC 7.07×10–9) from the EA group with suicide tendency; several had previously-noted clinical significance. For the GNOVA results, 25 traits showed statistically significant correlation at 10–3, and three survived FDR correction: depression in the past 2 weeks, miserableness, and tenseness in the past 2 weeks. All of the 25 traits described a mood or physical condition, but none of the traits included schizophrenia or bipolar disorder.
Discussion
Genomic correlates of suicidal ideation and behavior relate to traits in the general population that are linked to physical illness, mood, or anxiety. These findings suggest a commonality of genomic risk factors for suicidal ideation and behavior across SMI and the general population.
