Genome-wide mega-analysis identifies 16 loci and highlights diverse biological mechanisms in the common epilepsies

Abstract

The epilepsies affect around 65 million people worldwide and have a substantial missing heritability component. We report a genome-wide mega-analysis involving 15,212 individuals with epilepsy and 29,677 controls, which reveals 16 genome-wide significant loci, of which 11 are novel. Using various prioritization criteria, we pinpoint the 21 most likely epilepsy genes at these loci, with the majority in genetic generalized epilepsies. These genes have diverse biological functions, including coding for ion-channel subunits, transcription factors and a vitamin-B6 metabolism enzyme. Converging evidence shows that the common variants associated with epilepsy play a role in epigenetic regulation of gene expression in the brain. The results show an enrichment for monogenic epilepsy genes as well as known targets of antiepileptic drugs. Using SNP-based heritability analyses we disentangle both the unique and overlapping genetic basis to seven different epilepsy subtypes. Together, these findings provide leads for epilepsy therapies based on underlying pathophysiology.

Fig. 1

Manhattan plots for epilepsy genome-wide association analyses. Genome-wide association analyses of a all epilepsy, b focal epilepsy, and c genetic generalized epilepsy. Negative log₁₀-transformed P-values (Y-axis) are plotted against chromosomal position (x-axis). P-values were calculated with METAL using fixed-effects trans-ethnic meta-analyses. The red line represents the genome-wide significance threshold (p < 5 × 10⁻⁸). Previously known loci are indicated in black; novel loci in red. The names above each locus represent the prioritized gene in the locus (see Fig. 2) or the name of the locus itself in case of multiple prioritized genes in the locus

Fig. 2

Genome-wide significant loci of all analyses and prioritized biological epilepsy genes. Genes were prioritized based on 6 criteria and scored based on the number of criteria met per gene (filled red boxes). The highest scoring gene, or multiple if they have the same score, in each locus is reported as ‘prioritized biological epilepsy gene(s)’. Similar to previous studies^,, we used a minimum score of 2 to define these genes and we noted ‘none’ if no gene in the locus reached this score. Filled blue boxes indicate overlap with known targets of anti-epileptic drugs and established monogenic epilepsy genes. The lead SNP is defined as the SNP with the lowest P-value in the locus and the minor allele is displayed in brackets. P-values and Z-scores for All epilepsy, Focal epilepsy and Generalized epilepsy were calculated with fixed-effects trans-ethnic meta-analyses. P-values and Z-scores for JME, CAE, and Focal HS were calculated with BOLT-LMM. MAF minor allele frequency in the Human Reference Consortium reference panel. The direction of the Z-score is signed with respect to the minor allele. TWAS: significant TWAS association (based on data from the CommonMind Consortium), eQTL: significant eQTL within locus (based on data from the ROS/MAP projects), Brain exp: the gene is preferentially expressed in the brain, Missense: epilepsy GWAS missense variant in locus, PPI: gene prioritized by protein-protein interaction, KO mouse: relevant knockout mouse phenotype

Fig. 3

Functional annotation and heritability enrichment of epilepsy GWAS results. a functional categories of all genome-wide significant SNPs in all phenotypes. b Minimum (most active) chromatin state across 127 tissues for all genome-wide significant SNP in all phenotypes; TSS - transcription start site. c The RegulomeDB score for all genome-wide significant SNPs in all phenotypes, where 7 represents no evidence for affecting regulation and lower scores represent increasing evidence; NA - the variant does not exist in RegulomeDB. d Heritability enrichment for genetic generalized epilepsy with 6 different chromatin markers in 88 tissues, calculated with stratified LD-score regression using data from the Roadmap Epigenomics Project. The main bar chart represent the 10 tissues with the strongest heritability enrichment and the inset shows the full distribution of all chromatin markers in all tissues. e Heritability enrichment of genes expressed in 53 tissues, calculated with stratified LD-score regression using data from the gene-tissue expression (GTEx) Consortium

Fig. 4

Heritability estimates and genetic correlations between epilepsy syndromes, calculated using LDAK. Subjects with a diagnosis of both CAE and JAE were excluded from both phenotypes. The genetic correlation coefficient was calculated with LDAK and is denoted with a color scale ranging from 0% (white) to 100% (red). #P < 0.05; *P < 0.0024 (Bonferroni threshold); $h_L^2$2 SNP-based heritability on liability scale (95% CI); †heritability estimate exceeded 100%, possibly due to small sample size and large SD; CAE - childhood absence epilepsy, JAE - juvenile absence epilepsy, JME - juvenile myoclonic epilepsy, GTCS alone - generalized tonic-clonic seizures alone, focal HS - focal epilepsy with hippocampal sclerosis

Fig. 5

Genetic correlations of epilepsy with other phenotypes. The genetic correlation coefficient, calculated using LD-score regression, is denoted with a color scale ranging from -100% (blue) to 100% (red). #: P < 0.05 *P < 0.001 (Bonferroni threshold; 0.05/48)