Causal Relationships Between Myasthenia Gravis and Gut Microbiota by Mendelian Randomization Analysis

Abstract

Abstract OBJECTIVE Evidence from observational studies suggests a possible association between gastrointestinal microbiota (GM) and myasthenia gravis (MG). The association between GM and MG remains to be determined because observational studies are confounded by external confounders and reverse causality. The present study determined the causal association between specific GM and MG by Mendelian randomization (MR). METHODS Genome-wide association studies (GWAS) explore the associations between complex diseases, shapes, and genes. Extraction of single nucleotide polymorphisms (SNPs) from GWAS as instrumental variables (IVs) for Mendelian randomization studies reveals causal associations between disease and exposure at the genetic level. We performed Mendelian randomization by performing a MibioGen consortium for 211 different species of gastrointestinal microbiota (from 18,340 individuals, 24 cohorts) with MG (ncase = 426, ncontrol = 373848) from the Finngen database. Inverse variance weighting (IVW) was used as the primary method of analysis to compare the causal association between the two after false discovery rate (FDR) correction, with P < 0.05 as the difference being significant. RESULTS A MR study identified eight gastrointestinal microbiota, weighted by inverse variance, that may be negatively associated with the MG family.Clostridiaceae1.id.1869 (OR:0.424, 95%CI:0.202–0.889, P = 0.023), family.Defluviitaleaceae.id.1924 (OR:0.537, 95%CI:0.290–0.995, P = 0.048), family.Enterobacteriaceae.id.3469 (OR:0.341, 95%CI:0.135–0.865, P = 0.023), genus.Actinomyces.id.423 (OR:0.520, 95%CI:0.271–0.999, P = 0.049), genus.Victivallis.id.2256 (OR:0.627, 95%CI:0.426–0.923, P = 0.018), genus.Unknown genus. id.826 (OR:0.407, 95%CI:0.209–0.793, P = 0.008), order.Enterobacteriales.id.3468 (OR:0.341, 95%CI:0.135–0.865, P = 0.023). Genus.Lachnoclostridium.id.11308 (OR:2.431, 95%CI:1.047–5.647, P = 0.039) showed a possible positive association with MG after inverse variance weighting. However, no GM showed a causal association with MG after FDR correction. Reverse MR likewise did not find a causal association between MG and GM. CONCLUSION Although the causal association of MG and GM was negative by MR Analysis, the extended database or new microbiome data needs to be further validated.