Baseline immune and gut microbial–metabolic profiles associated with hepatic immune-related adverse events in lung cancer.

e20605 Background: Hepatic immune-related adverse events (HRAEs) are clinically significant toxicities of immune checkpoint inhibitors in lung cancer with substantial inter-individual heterogeneity. Reliable baseline biomarkers for susceptibility are lacking. The contribution of pre-treatment host immune status and gut microbial and metabolic background to HRAE risk remains insufficiently defined. Methods: We conducted a single-center retrospective observational cohort of n = 70 lung cancer patients receiving PD-1 or PD-L1 inhibitors between January 2022 and June 2025, with biospecimens collected at baseline and at HRAE onset. Patients were followed from treatment initiation until HRAE onset, treatment discontinuation, or end of follow-up, with clinically significant HRAEs defined as Common Terminology Criteria for Adverse Events version 5.0 grade 2 or higher liver injury. Peripheral immune cell subsets, liver function indices, and fecal samples were assessed at baseline. In patients who developed HRAEs, paired immune and fecal samples were analyzed at HRAE onset for immune phenotyping, shotgun metagenomic sequencing, and untargeted metabolomic profiling. Associations with HRAEs were evaluated using logistic regression. Results: Baseline demographic and treatment characteristics were comparable between groups. Prior to treatment, reduced B-cell components and increased helper and naive CD4-positive T-cell proportions, together with intermediate activated T-cell proportions, defined a high-risk baseline immune phenotype for HRAEs. Baseline metagenomic profiling revealed distinct gut microbial community structures between risk states, including differential abundance of Phocaeicola coprocola and Ruminococcus, supporting a pre-existing microbial background for HRAE susceptibility. Untargeted metabolomics demonstrated enrichment of nucleoside-related, fatty acid metabolism–related, and microbiota-derived metabolites in HRAE-susceptible patients, with higher baseline adenosine, palmitoylcarnitine, and phenylacetylglutamine, indicating a coordinated immune–microbial–metabolic background linked to toxicity risk. At HRAE onset, immune changes were dominated by functional activation, with increased activated CD4-positive and CD8-positive T cells and natural killer cell activation, without evidence of systemic immune restructuring. Astilbin exhibited a stage-dependent pattern, with higher baseline levels followed by a decrease at HRAE onset. Conclusions: HRAE susceptibility in lung cancer reflects a pre-existing immune–microbial–metabolic background rather than immune checkpoint inhibitor exposure. Reduced B-cell components and increased helper and naive CD4-positive T-cell features define an immune-susceptible state linked to toxicity risk and may inform pre-treatment risk stratification.