Identification and characterization of four bacteriome- and mycobiome-derived subtypes in tumour and adjacent mucosa tissue of colorectal cancer patients.

Abstract

214 Background: The bacteriome and mycobiome, collectively referred to as the microbiome, is a key player in CRC pathogenesis, progression and response to chemotherapy, radiotherapy and immunotherapy. Microbial modulation and exacerbation in CRC is likely attributable to alterations of composition and diversity of the microbiome rather than to a single microbial species. We therefore systematically investigated alterations in the bacteriome and mycobiome of CRC patients in tumours and matched adjacent mucosa resulting in the identification of microbiome-based subtypes associated with host clinico-pathological and molecular characteristics. Methods: We estimated bacterial and fungal composition from RNA sequencing experiments using a validated subtractive method ( PathSeq) in in-house and publicly available cohorts. We evaluated the quality of the microbial estimates with two sets of orthogonal analysis determining i) the agreement with composition from an independent method and ii) retrieval of expected clinically-relevant microbial signatures. Diversity and composition of bacteriome and mycobiome of tumour and adjacent mucosa, and resulting subtypes were computationally deconvoluted using > 10,000 samples. Results: The bacteriome of tumours had higher dominance and lower ɑ-diversity compared to matched adjacent local and distant mucosa. Tumours were enriched with Proteobacteria (Gammaproteobacteria class), Fusobacteria (including Fusobacterium Nucleatum species) and Basidiomycota fungi (Malasseziaceae family). Tumours were depleted of Bacteroidetes (Bacteroidia class), Firmicutes (Clostridia class) and Ascomycota (Sordariomycetes and Saccharomycotina). Tumours and adjacent mucosa samples were classified into 4 microbial subtypes, termed C1 to C4, based on the bacteriome and mycobiome composition. The bacterial Propionibacteriaceae, Enterobacteriaceae, Fusobacteriaceae, Bacteroidaceae and Ruminococcaceae along with the fungal Malasseziaceae, Saccharomycetaceae and Aspergillaceae were among the key families driving the microbial subtyping. Microbial subtypes were associated with distinct tumour histology and patient phenotypes and served as independent prognostic marker for disease-free survival. Key associations between microbial subtypes and alterations in host immune response and signalling pathways were validated in the TCGA pan-cancer cohort. The microbial subtyping demonstrated stratification value in the pan-cancer settings beyond merely representing differences in survival by cancer type. Conclusions: This study demonstrates the translational potential of microbial subtyping in CRC patient stratification, and provides avenues to design tailored microbiota modulation therapy to further precision oncology.