ROLIVER - Prospective Cohort for the Identification of Liver Microbiota

Liver diseases and notably fibrosis are featured by a gut microbiota dysbiosis with a large diversity. Classical statistical analyses does not allow to discriminate between the different low score of liver fibrosis F0,F1,F2. The existence of an adipose tissue microbiota causally involved in the triggering of a low grade inflammation could resemble what observed in liver fibrosis. To generate microbial hypotheses putatively responsible for the onset of liver fibrosis we sequenced the 16SrDNA gene from liver biopsies from 36 obese patients (ROLIVER cohort) and describe an original mathematical approach to decipher signatures of early stage of liver fibrosis F0, F1, F2. We identified that Protebacteria as the main phyla in liver with Pseudomonadaceae and Proteobacteriaceae families representing ~60% of the 16SrDNA gene diversity. While primary component analyses were unable to discriminate between the three groups the partial least square discriminant analysis appears as a powerful tool to identify signatures predictive for each group. We further constructed a matrix of interacting OTU clusters surrounding early scores of liver fibrosis. Eventually, we applied the tfidf approach to exemplify the rare variables which could be carrying some information suitable to refine the diagnosis. Altogether, we here propose a mathematical approach suitable for precise identification of bacteria putatively involve in the progressive development of liver fibrosis that represent hence a new therapeutic opportunity.