Answer to Ling et al “Predictive roles of gut dysbiosis on the severity of NAFLD”

We thank Ling and colleagues for their comments about our study. As they stated, there are numerous indexes for the evaluation of gut microbiota structure. In our work, overall structure of the gut microbiota was assessed with richness (chao1 estimator) and phylogenetic diversity (Faith's phylogenetic diversity), and these indexes showed no significant difference according to increasing NAFLD severity. We agree that non-abundant gut bacteria may act in the pathophysiology of NAFLD. Nevertheless, highlighting a potential role of non-abundant bacteria is difficult in a small subset of patients and requires large samples to produce significant and robust results. For the same reason, we didn't intend to evaluate gut dysbiosis as new biomarker for the non-invasive discrimination of NAFLD severity. Our experience in non-invasive testing of liver lesions demonstrates that large samples of several hundred or thousand patients are crucial to generate data with good external validity (1). Ling and colleagues specified that a control group (i.e., obese patients without liver steatosis) would have been required in our work. This additional group would have only allowed us to assess the association between gut dysbiosis and the presence or the absence of NAFLD. Consequently, as we hypothesised that gut dysbiosis was associated with NAFLD severity, this control group would have added no relevant information for our study goal.

At the phylum level, there was no association between the Firmicutes/Bacteroidetes ratio and increasing NAFLD severity (see Figure 1a in the article). At the genus level, the relative abundance of Bacteroides and Ruminococcus was independently linked to liver disease severity. To go beyond the simple description of a statistical association, we used PICRUSt to find some pathophysiological hypotheses. By accurately estimating the whole functional profile rather than the taxonomic composition of the gut microbiota, this approach circumvents the consideration of the ecosystem structure/diversity. Moreover, genes and subsequent functions associated with abundant and non-abundant gut bacteria are taken into account using this method. By showing that gut dysbiosis associated with NAFLD severity is accompanied by a shift in the metabolic functions of the gut microbiota, our study adds new hypotheses about how gut microbiota could promote steatohepatitis or fibrosis in NAFLD (2). Indeed, we can make the hypothesis that gut metabolic pathways could promote liver inflammation or fibrosis by their production of active metabolites then directed to the liver (3). Knowing that gut metabolic pathways and their products are closely linked to nutrition, our study provides rationale for future works which will need new technologies such as metagenomics and metabolomics to decipher the mechanisms linking nutrition, gut microbiota and NAFLD severity.