Alterations of Gut Microbiome, Metabolome, and Lipidome in Takayasu Arteritis

Mounting evidence has linked microbiome and metabolome to systemic autoimmunity and cardiovascular diseases (CVDs). Takayasu arteritis (TAK) is a rare disease that shares features of immune-related inflammatory diseases and CVDs, about which there is relatively limited information. This study was undertaken to characterize gut microbial dysbiosis and its crosstalk with phenotypes in TAK.To address the discriminatory signatures, we performed shotgun sequencing of fecal metagenome across a discovery cohort (n = 97) and an independent validation cohort (n = 75) including TAK patients, healthy controls, and controls with Behçet's disease (BD). Interrogation of untargeted metabolomics and lipidomics profiling of plasma and fecal samples were also used to refine features mediating associations between microorganisms and TAK phenotypes.A combined model of bacterial species, including unclassified Escherichia, Veillonella parvula, Streptococcus parasanguinis, Dorea formicigenerans, Bifidobacterium adolescentis, Lachnospiraceae bacterium 7 1 58FAA, Escherichia coli, Streptococcus salivarius, Klebsiella pneumoniae, Bifidobacterium longum, and Lachnospiraceae Bacterium 5 1 63FAA, distinguished TAK patients from controls with areas under the curve (AUCs) of 87.8%, 85.9%, 81.1%, and 71.1% in training, test, and validation sets including healthy or BD controls, respectively. Diagnostic species were directly or indirectly (via metabolites or lipids) correlated with TAK phenotypes of vascular involvement, inflammation, discharge medication, and prognosis. External validation against publicly metagenomic studies (n = 184) on hypertension, atrial fibrillation, and healthy controls, confirmed the diagnostic accuracy of the model for TAK.This study first identifies the discriminatory gut microbes in TAK. Dysbiotic microbes are also linked to TAK phenotypes directly or indirectly via metabolic and lipid modules. Further explorations of the microbiome-metagenome interface in TAK subtype prediction and pathogenesis are suggested.