A Myasthenia Gravis genomewide association study of three cohorts identifies Agrin as a novel risk locus

Abstract

Background Myasthenia Gravis (MG) is a rare autoimmune disorder affecting the neuromuscular junction. Here, we investigate the genetic architecture of MG performing a genomewide association study (GWAS) of the largest MG dataset analyzed to date. Methods We integrated GWAS from three different datasets (1,401 cases, 3,508 controls) and performed MG GWAS and onset-specific analyses. We also carried out HLA fine-mapping, gene-based, gene ontology and tissue enrichment analyses and investigated genetic correlation to other autoimmune disorders. Findings We observed the strongest MG association to TNFRSF11A (rs4369774, p=1.09x10-13; OR=1.4). Gene-based analysis revealed AGRN as a novel MG susceptibility gene. HLA fine-mapping pointed to two independent loci significantly associated with MG: HLA-DRB1 (with a protective role) and HLA-B. MG onset-specific analysis, reveals differences in the genetic architecture of Early-Onset vs Late-Onset MG. Furthermore, we find MG to be genetically correlated with Type 1 Diabetes, Rheumatoid Arthritis and late-onset Vitiligo. Interpretation Overall, our results are consistent with previous studies highlighting the role of the HLA and TNFRSF11A in MG etiology and different risk genes in EOMG vs LOMG. Furthermore, our gene-based analysis implicates, for the first time, AGRN as a MG susceptibility locus. AGRN encodes agrin, which is involved in neuromuscular junction formation. Mutations in AGRN have been found to underlie congenital myasthenic syndrome. Gene ontology analysis suggests an intriguing role for symbiotic processes in MG etiology. We also uncover genetic correlation of MG to Type 1 Diabetes, Rheumatoid Arthritis and late-onset Vitiligo, pointing to shared underlying genetic mechanisms. Funding This work was supported by NSF award #1715202, the European Social Fund and Greek funds through the National Strategic Reference Framework (NSRF) THALES Programme 2012-2015 and the NSRF ARISTEIA II Programme 2007-2013 to PP, and grants from the Association Francaise contre les Myopathies (AFM, Grant No. 80077) to ST.

Reconstructing SNP Allele and Genotype Frequencies from GWAS Summary Statistics

“…We further tested our method on real GWAS data, using GWAS summary statistics for myasthenia gravis samples from dbGaP as the base study and assessing its predicting power on 196 independent myasthenia gravis cases and 1,057 ancestry-matched controls from [25] for which we had individual level genotypes available. We generated GWAS summary statistics for the base study using standard quality controls and computed GWAS summary statistics for the target dataset as described.…”

“…This dataset has a total sample size of 196 cases and 1,057 controls, with 6,276,739 SNPs included after quality control. This dataset was described in detail in [25].…”

“…Next, we evaluated the performance of ReACt on real GWAS datasets, where the individual level genotype of the target study was available. For this experiment, we used GWAS summary statistics of myasthenia gravis samples from dbgap as the base study (see Section 4.4.1 for details) and an independent group of myasthenia gravis cases and matching controls as the target population [25]. We clumped the base summary statistics using the European samples from 1000 Genome Project as reference, under parameters --clump-p1 1 --clump-kb 250 --clump-r2 0.1.…”

“…wxyz) 2 − 4(wxz + (1 − z) 2 )(yz(x + yz)) 2(wxz + (1 − z) 2 ).Given d, we can immediately compute a, b, and c using eqns. (23),(24), and(25). In order to determine whether d is equal tod 1 or d 2 , we first check whether d 1 or d 2 guarantee that a, b, c, and d are all positive numbers.…”

Herpesvirus entry mediator on T cells as a protective factor for myasthenia gravis: A Mendelian randomization study

“…Mendelian randomization (MR) uses genetic variants as the exposure proxy of the exposure to examine the causal effect of that exposure on the outcome (8). The correlations between genetic variants and MG have been explored in several genomewide association studies (GWASs) and human leukocyte antigen (HLA) haplotype analysis, by which T-cell relevant genes, including CTLA4, TNFRSF11A, PTPN22, and the HLA haplotypes, have been implicated in the pathogenesis of MG (9)(10)(11)(12). With now available large data sets, MR analysis may be an elegant tool to explore the novel biomarkers from T cells with causal impacts on MG risk, which has rarely been performed in this field.…”