Human liver microbiota modeling strategy at the early onset of fibrosis

Aim: To understand the pathological impact of gut microbiota on the early events of liver fibrosis we studied liver microbiota sequences as a model of the translocation of bacteria from the gut to the liver. To overcome the impact of different group size, country, and of set analyses we developed innovative and adapted statistical approaches. Methods: The liver biopsies were from patients with low liver fibrosis scores 0,1,2; Romania (n=36), Austria (n=10), Italy (n=19), and Spain (n=17). The bacterial V3-V4 hypervariable regions of the 16SrDNA were amplified, sequenced and the OTUs and taxa identified. To identify specific taxonomic profiles and statistical differences between groups we took into account the frequency, sparsity, unbalanced sample size, the country and the set of analyses. We performed first primary multivariate techniques, then we used fair-discriminant analyses, and an adapted spectral clustering strategy to identify some classifying signatures. We further used predicted metagenomics to propose microbial metabolic pathways. Results: We first identified that the Proteobacteriaceae was the dominant family of the liver biopsies. Secondly we identified that the country origin is a major confounding factor masking specific signatures associated to the liver fibrosis score. Therefore, we thirdly developed “fairness” mathematical strategies and adapted a spectral clustering approach with L1 penalty that identifies highly refined clusters of variables independently from the country. The results obtained were then cross-checked and enabled us to identify a specific signature for each low fibrosis score and correspondingly the predicted metagenomics pathways. Conclusion: We propose a useful statistical approach to identify microbial signatures and overcome issues regarding sample size differences, the impact of country and sets of analyses. The taxonomic results suggest a role of bacterial translocation to the liver in the progression of the liver fibrosis at its early onset.