Microbiome predictors of dysbiosis and VRE decolonization in patients with recurrent C. difficile infections in a multi-center retrospective study

Abstract

The gastrointestinal microbiome is intrinsically linked to the spread of antibiotic resistance. Antibiotic treatment puts patients at risk for colonization by opportunistic pathogens like vancomycin resistant Enterococcus and Clostridioides difficile by destroying the colonization resistance provided by the commensal microbiota. Once colonized, the host is at a much higher risk for infection by that pathogen. Furthermore, we know that microbiome community differences are associated with disease states, but we do not have a good understanding of how we can use these changes to classify different patient populations. To that end, we have performed a multicenter retrospective analysis on patients who received fecal microbiota transplants to treat recurrent Clostridioides difficile infection. We performed 16S rRNA gene sequencing on fecal samples collected as part of this study and used these data to develop a microbiome disruption index. Our microbiome disruption index is a simple index that is predictive across cohorts, indications, and batch effects. We are able to classify pre-fecal transplant vs post-fecal transplant samples in patients with recurrent C. difficile infection, and we are able to predict, using previously-published data from a cohort of patients receiving hematopoietic stem cell transplants, which patients would go on to develop bloodstream infections. Finally, we also identified patients in this cohort that were initially colonized with vancomycin resistant Enterococcus and that 92% (11/12) were decolonized after the transplant, but the microbiome disruption index was unable to predict such decolonization. We, however, were able to compare the relative abundance of different taxa between the two groups, and we found that increased abundance of Enterobacteriaceae predicts whether patients were colonized with vancomycin resistant Enterococcus. This work is an early step towards a better understanding of how microbiome predictors can be used to help improve patient care and patient outcomes.