Interaction of bacterial genera associated with therapeutic response to immune checkpoint PD-1 blockade in a United States cohort

Abstract

Background: Recent studies show that human gut microbial composition can determine whether a patient is a responder or non-responder to immunotherapy but have not identified a common microbial signal shared by responding patients. The functional relationship between immunity, intestinal microbiota, and NSCLC response to immune checkpoint inhibitor/inhibition (ICI) in an American cohort remains unexplored. Methods: RNAlater-preserved fecal samples were collected from 65 pre-treatment (baseline) and post-treatment stage III/IV NSCLC patients undergoing ICI therapy, categorized as responders or non-responders according to RECIST criteria. Pooled and individual responder and non-responder microbiota were transplanted into a gnotobiotic mouse model of lung cancer and treated with ICIs. 16S rDNA and RNA sequencing was performed on patient fecal samples, 16S rDNA sequencing on mouse fecal samples, and flow cytometric analysis on mouse tumor tissue. Results: Responder patients have both a different microbial community structure than non-responders (P = 0.004) and a different bacterial transcriptome (PC2 = 0.03) at baseline. Taxa significantly enriched in responders include amplicon sequence variants (ASVs) belonging to the genera Ruminococcus, Akkermansia, and Faecalibacterium. Pooled and individual responder microbiota transplantation into gnotobiotic mice decreased tumor growth compared to non-responder colonized mice following ICI (P = 0.023, P = 0.019, P = 0.008, respectively). Responder tumors showed an increased anti-tumor cellular phenotype following ICI treatment. Responder mice are enriched with ASVs belonging to the genera Bacteroides, Blautia, Akkermansia, and Faecalibacterium. Overlapping taxa mapping between human and mouse cohorts correlated with tumor size and weight revealed a network highlighting responder-associated ASVs belonging to the genera Colidextribacter, Frisingicoccus, Marvinbryantia, and Blautia which have not yet been reported. Conclusions: The role of isolate-specific function and bacterial gene expression in gut microbial-driven responsiveness to ICI has been underappreciated. This work supports further investigation using isolate-driven models to characterize the mechanisms underlying this phenomenon.