Effect of Antibiotic Therapy on Human Fecal Microbiota and the Relation to the Development of Clostridium difficile

M. F. De La Cochetière, T. Durand, V. Lalande, J. C. Petit, G. Potel, L. Beaugerie

The gastrointestinal tract is a complex ecosystem. Recent studies have shown that the human fecal microbiota is composed of a consortium of microorganism. It is known that antibiotic treatment alters the microbiota, facilitating the proliferation of opportunists that may occupy ecological niches previously unavailable to them. It is therefore important to characterize resident microbiota to evaluate its latent ability to permit the development of pathogens such as Clostridium difficile. Using samples from 260 subjects enrolled in a previously published clinical study on antibiotic-associated diarrhea, we investigated the possible relationship between the fecal dominant resident microbiota and the subsequent development of C. difficile. We used molecular profiling of bacterial 16S rDNA coupled with partial least square (PLS) regression analysis. Fecal samples were collected on day 0 (D0) before antibiotic treatment and on day 14 (D14) after the beginning of the treatment. Fecal DNA was isolated, and V6-to-V8 regions of the 16S rDNA were amplified by polymerase chain reaction with general primers and analyzed by temporal temperature gradient gel electrophoresis (TTGE). Main bacteria profiles were compared on the basis of similarity (Pearson correlation coefficient). The characteristics of the microbiota were determined using PLS discriminant analysis model. Eighty-seven TTGE profiles on D0 have been analyzed. The banding pattern was complex in all cases. The subsequent onset of C. difficile was not revealed by any clustering of TTGE profiles, but was explained up to 46% by the corresponding PLS model. Furthermore, 6 zones out of the 438 dispatched from the TTGE profiles by the software happened to be specific for the group of patients who acquired C. difficile. The first approach in the molecular phylogenetic analysis showed related sequences to uncultured clones. As for the 87 TTGE profiles on D14, no clustering could be found either, but the subsequent onset of C. difficile was explained up to 74.5% by the corresponding PLS model, thus corroborating the results found on D0. The non exhaustive data of the microbiota we found should be taken as the first step to assess the hypothesis of permissive microbiota. The PLS model was used successfully to predict C. difficile development. We found that important criteria in terms of main bacteria could be markedly considered as predisposing factors for C. difficile development. Yet, the resident microbiota in case of antibiotic-associated diarrhea has still to be analyzed. Furthermore, these findings suggest that strategies reinforcing the ability of the fecal microbiota to resist to modifications would be of clinical relevance.