The landscape and interplay of gut microbiome and metabolites in drug-induced liver injury

Drug-induced liver injury (DILI) is an important cause of acute liver disease. Although the gut-liver axis plays a pivotal role in hepatic homeostasis, the functional links between the gut microbiome and systemic metabolites in DILI remain ill-defined. This study aimed to characterize the multi-omics signatures in DILI patients, while evaluating the predictive potential of key microbiota-associated metabolites for disease occurrence and clinical severity. We enrolled 56 participants, including healthy controls (HC) and DILI patients (stratified into Mild and Moderate-to-Severe). Plasma untargeted metabolomics (LC–MS/MS) and fecal 16S rRNA gene sequencing were integrated to map systemic metabolic profiles and gut microbial landscapes. Differential features were identified using OPLS-DA/PLS-DA (VIP > 1.0 combined with FDR < 0.05 for DILI vs. HC, or unadjusted p < 0.05 for severity comparisons) and diversity/rank-sum analyses. Spearman correlation analysis was performed between differential bacterial taxa and altered metabolites to identify key microbiota-associated metabolites. The discriminatory performance of these key metabolites for distinguishing DILI from HC was evaluated using receiver operating characteristic (ROC) curves and area under the curve (AUC) metrics. DILI patients exhibited a markedly altered multi-omics landscape, characterized by gut microbial dysbiosis and metabolic reprogramming (77 upregulated and 119 downregulated plasma metabolites). Correlation analysis identified key metabolites (glycocholic acid, carnosine, and Val-Phe) strongly associated with microbial dysbiosis, with AUC values of 0.900 (95%CI [0.819, 0.981]), 0.931 (95%CI [0.864, 0.998]), and 0.992 (95%CI [0.970, 1.000]), respectively. Importantly, the combined metabolite model further improved apparent discriminatory performance compared with individual metabolites, achieving an apparent AUC of 1.000 (95% CI [1.000, 1.000]). Severity-stratified analysis revealed differences in specific microbial taxa and bile acid/bilirubin-related metabolites between mild and moderate-to-severe DILI, with an apparent AUC of 0.885 (95% CI [0.762, 1.000]) for a combined metabolite model. Our multi-omics analysis reveals that DILI is characterized by gut microbial dysbiosis and systemic metabolic reprogramming. Key microbiota-associated metabolites showed exploratory associations with disease onset and severity grading, but their diagnostic performance was not validated and requires independent confirmation. Trial registration ClinicalTrials.gov-NCT05465642, July 20, 2022.