Metabolomics and microbiomes for discovering biomarkers of antituberculosis drugs-induced hepatotoxicity.

Anti-tuberculosis (TB) drug-induced hepatotoxicity (ATDH) was related to metabolic and microbial dysregulation, but only limited data was available about the metabolomes and microbiomes in ATDH. We aimed at detecting the metabolic and microbial signatures of ATDH. Urine samples were obtained from ATDH (n = 33) and non-ATDH control (n = 41) and analyzed by untargeted gas chromatography time-of-flight mass spectrometry (GC-TOF-MS). Metabolites were analyzed by orthogonal projections to latent structures-discriminate analysis (OPLS-DA) and pathway analysis. Eight ATDH and eight non-ATDH control were evaluated by sequencing of 16S rRNA genes, and the Clusters of Orthologous Groups of proteins (COG) database were used for function prediction. Linear discriminant analysis (LDA) effect size (LEfSe) was applied to detect the differential microbiotas between the two groups. The differential microbiotas were further validated by correlation analysis with differential metabolites. OPLS-DA analysis suggested 11 metabolites that differed ATDH from non-ATDH control. Pathway analysis demonstrated that metabolism of arginine and proline, metabolism of d-arginine and d-ornithine, glutathione glycine metabolism, galactose metabolism, niacin and nicotinamide metabolism, and glycine, serine and threonine metabolism were related to ATDH. LEfSe suggested significant differences in microbiotas between the two groups. The o_ Bacteroidales, f_Prevotellaceae, and g_Prevotella were significantly increased in ATDH. In contrast, the f_Chitinophagaceae, c_Gammaproteobacteria, and p_Proteobacteria were significantly increased in non-ATDH group. The biological functions of the sequenced microbiota in this study were related to amino acid transport and metabolism and defense mechanisms. Finally, we detected strong association between urine metabolites and specific urine bacteria (|r| > 0.8). d-glucoheptose showed a strong relationship to Symbiobacterium. Creatine (r = -0.901; P < 0.001) and diglycerol were strongly associated with Alishewanella. Metabolomics and microbiomes indicate ATDH characterized by metabolic and microbial profiles may differ from non-ATDH control.