Microbiota Composition and Predicted Function in Severe Endodontic Infections

Objectives: The aim of this study was to characterize differences in bacterial community composition and predicted functions between hospital inpatients and outpatients diagnosed with acute apical abscess.
Methods: Aspiration samples from abscesses were collected from 34 subjects diagnosed with acute apical abscesses. 15 subjects were hospitalized patients at the University of Maryland Medical Center and 19 patients were treated as outpatients. Bacterial community composition from apical abscess samples was characterized using 16S rRNA gene sequencing with llumina MiSeq. Statistical analyses were performed using LEfSe (LDA [Linear discriminant analysis] Effect Size) and Metastats to identify bacterial taxa associations with each patient group. Detailed metabolic and functional profiles of these bacterial communities were predicted from the 16S data using the PICRUSt (Phylogenetic Investigation of Communities by Reconstruction of Unobserved States) analysis software, which is a novel approach in oral microbiology.
Results: After taxonomic assignments, a total of 3,421 OTUs were assigned to 481 bacterial species across the 34 samples. Statistical analyses identified bacterial taxa significantly associated with each patient group. Fusobacterium necrophorum, Bacillus sp., Megasphaera sp., Alloprevotella sp., Leptothrix sp., Lachnospiraceae sp. and Dolosigranulum pigrum were found to be uniquely associated (p<0.05) with the inpatients group. In silico functional analysis using PICRUSt predicted a total of 328 KEGG modules across all the samples. At a functional level, samples clustered in two groups. 76 KEGG modules were found to be significantly different between the two functional groups (ANOVA test, p<0.05). Metabolism of xenobiotics by cytochrome P450 was the most significantly different (p<0.05) KEGG pathway between both functional groups. Bivariate associations between the functional clusters and the following variables; diabetes, fever and hospitalization days, were performed but did not reach statistical significance at p<0.05.
Conclusions: Bioinformatic analyses revealed differences in microbial diversity and predicted functions between the two groups of patients.