Identification of 64 new risk loci for major depression, refinement of the genetic architecture and risk prediction of recurrence and comorbidities

Abstract

Major depression (MD) is a common mental disorder and a leading cause of disability worldwide. We conducted a GWAS meta-analysis of more than 1.3 million individuals, including 371,184 with MD, identifying 243 risk loci. Sixty-four loci are novel, including glutamate and GABA receptors that are targets for antidepressant drugs. Several biological pathways and components were enriched for genetic MD risk, implicating neuronal development and function. Intersection with functional genomics data prioritized likely causal genes and revealed novel enrichment of prenatal GABAergic neurons, astrocytes and oligodendrocyte lineages. We found MD to be highly polygenic, with around 11,700 variants explaining 90% of the SNP heritability. Bivariate Gaussian mixture modeling estimated that > 97% of risk variants for other psychiatric disorders (anxiety, schizophrenia, bipolar disorder and ADHD) are influencing MD risk when both concordant and discordant variants are considered, and nearly all MD risk variants influence educational attainment. Additionally, we demonstrated that MD genetic risk is associated with impaired complex cognition, including verbal reasoning, attention, abstraction and mental flexibility. Analyzing Danish nation-wide longitudinal data, we dissected the genetic and clinical heterogeneity, revealing distinct polygenic architectures across case subgroups of MD recurrency and psychiatric comorbidity and demonstrating two- to six-fold increases in absolute risks for developing comorbid psychiatric disorders among MD cases with the highest versus the lowest polygenic burden. The results deepen the understanding of the biology underlying MD and its progression and inform precision medicine approaches in MD.

Differential Chromatin Architecture and Risk Variants in Deep Layer Excitatory Neurons and Grey Matter Microglia Contribute to Major Depressive Disorder

“…Similarly to ExN1, ExN2 showed similarities to deep-layer (layer 5-6) excitatory neurons (Extended Fig 2A, 3C). In fact, enrichments calculated from data obtained from both MDD GWAS used in our study 11,13 were coherently highest in ExN1 cluster (Fig 4A; p=0.003 (Howard), p=0.0005 (Als), Supplementary Table 9). We also calculated heritability enrichments in MDD-associated chromatin in each cluster (candidate DARs), showing that the enrichment was likewise highest in ExN1 for MDD (Fig 4B; p=0.01; nominal) and insomnia (p=0.02).…”

“…Of these genes, the highest psychiatric disease association was found for depression (Fig 2C), whereby 80% of these associations showed consistent evidence in PsyGeNet 26 and 64% for schizophrenia. In fact, there was an overrepresentation among DAR-linked genes (FDR < 0.05; one-sided Fisher’s exact test; Fig 2D) for PsyGeNet genes linked to depression, MDD-associated genetic variation 11,13 and genes previously observed to be differentially expressed in MDD 20 . We did not observe similar enrichments for genes associated with cell-type DARs observed at broad level.…”

“…To examine the functional impact of MDD-associated genetic variation within each cluster, we assessed allelic-effects of MDD-associated SNPs on accessibility of cCREs and TF binding sites and identified likely target risk genes. We used MDD-associated SNPs in GWAS 11–13 , those in high LD with them (r2 value ≥0.8) and those identified by finemapping 11,44 . To predict accessibility in each cluster, we used three complementary models; in-silico mutagenesis (ISM), delta-support vector machine (delta-SVM) 45 and gapped-kmer (gkm-explain) 46 based variant-effect scores derived from cluster-specific gkmSVM classifiers 47,48 (Supplementary Fig 6-7).…”

“…The nodes in green represent DARs enriched for individual TF binding sites directed towards putative target-genes colored according to MDD associated gene or MDD DEG status. For MDD-associated genes, we compiled list of potential risk-genes in MDD GWAS 11–13,72,113,114 , Hi-C based analysis (H-MAGMA; FDR<0.1) 115 and PsyGeNET depressive disorders genes 26 . For MDD DEGs, we used bulk (FDR<0.1) 116 and snRNA-seq DEGs 20 obtained from the meta-analysis of both sexes (FDR<0.1, abs(log2FC) > log2(1.1), signs > 0).…”

“…Genome-wide associations studies (GWAS) have identified over 200 MDD risk loci 11–13 ; however, similarly to other psychiatric phenotypes, characterizing the functional impact of such variants has remained challenging as they mostly map outside protein-coding regions. In addition to genetic variation, the adverse effects of environmental factors, such as early-life stress, on increasing the risk for depression 14,15 , are mediated by regulatory changes 16 .…”