Background Immune-related adverse events (irAEs) represent an urgent clinical challenge. Although accumulating evidence suggests that irAEs are associated with the gut microbiota and its metabolites, our understanding of the dynamic alterations in the gut microbiota and related metabolic profiles throughout the onset and progression of irAEs remains limited. Methods A total of 48 fecal samples were collected from 32 lung cancer patients treated with immune checkpoint inhibitors, including 16 patients who developed irAEs and 16 who did not. Fecal samples were collected at baseline and, in patients with irAEs, at the time of irAEs onset. Metagenomic sequencing and untargeted metabolomics analyses were performed to identify baseline differences in gut microbiota and metabolites, characterize longitudinal dynamic changes in gut microbiota and metabolite profiles in patients with irAEs, and construct a machine learning based random forest model to predict the occurrence of irAEs. Results There were baseline differences in microbial communities and metabolites between the two groups. In the non-irAEs group, Phocaeicola coprocola was enriched and Micrococales decreased. At baseline, viomycin was positively correlated with irAEs, while metabolites such as calcitriol and L-isoleucine were negatively correlated with irAEs. The roles of valine, leucine and isoleucine metabolism and vitamin B6 metabolism pathways were downregulated in the irAEs group. Compared to baseline, there were significant changes in gut microbiota and metabolites during the onset of irAEs, and the abundance of Veillonella increased during irAEs onset. Dynamic monitoring of metabolic changes in irAEs revealed decreased levels of trypsin butylester, BQ 123, DL-o-tyrosine, and nicotinamide-beta-riboside during irAEs attacks. Lysine degradation, arachidonic acid metabolism, folate biosynthesis, nicotinate and nicotinamide metabolism, and C5-branched dibasic acid metabolism were downregulated during the progression of irAEs. A model for predicting the occurrence of irAEs based on differential microbiota and metabolites was constructed, and after robust validation, the model showed good performance and excellent discriminative power. Conclusions The occurrence and development of irAEs are associated with the composition of the gut microbiota and metabolites, as well as their dynamic changes over time. These findings highlight the potential of gut microbiota and metabolites as biomarkers for predicting the occurrence and progression of irAEs.
Han et al. (Thu,) studied this question.