Objective: Transcriptomic-based subtyping, Consensus Molecular Subtyping (CMS) and CRC Intrinsic Subtyping (CRIS), identify a patient subpopulation with mesenchymal traits (CMS4/CRIS-B) and poorer outcome. Here, we investigated the relationship between prevalence of Fusobacterium nucleatum (Fn) and Fusobacteriales, CMS/CRIS subtyping, cell type composition, immune infiltrates and host contexture to refine patients stratification and identify druggable context-specific vulnerabilities.
Design: We coupled cell culture experiments with characterization of Fn/Fusobacteriales prevalence and host biology/microenviroment in tumours from 2 independent CRC patient cohorts (Taxonomy: n=140; TCGA-COAD-READ: n=605).
Results: In vitro, Fn infection induced inflammation via NFκB/TNFα in HCT116 and HT29 cancer cell lines. In patients, high Fn/Fusobacteriales were found in CMS1, MSI tumours, with infiltration of macrophages M1, reduced macrophages M2, and high IL6/IL8/IL1β signaling. Analysis of the Taxonomy cohort suggested that Fn was prognostic for CMS4/CRIS-B patients, despite having lower Fn load than CMS1 patients. In the TCGA-COAD-READ cohort, we likewise identified a differential association between Fusobacteriales relative abundance and outcome when stratifying patients in mesenchymal (either CMS4 and/or CRIS-B) vs. non-mesenchymal (neither CMS4 nor CRIS-B). Patients with mesenchymal tumours and high Fusobacteriales had approximately 2-fold higher risk of worse outcome. These associations were null in non-mesenchymal patients. Modelling the 3-way association between Fusobacteriales prevalence, molecular subtyping, and host contexture with logistic models with an interaction term disentangled the pathogen/host-signaling relationship and identified aberrations (including EMT/WNT/NOTCH) as candidate targets.
Conclusion: This study identifies CMS4/CRIS-B patients with high Fn/Fusobacteriales prevalence as a high-risk subpopulation that may benefit from therapeutics targeting mesenchymal biology.