Use of Shotgun Metagenome Sequencing To Detect Fecal Colonization with Multidrug-Resistant Bacteria in Children

Abstract

Prevention of multidrug-resistant (MDR) bacterial infections relies on accurate detection of these organisms. We investigated shotgun metagenome sequencing for the detection of methicillin-resistant Staphylococcus aureus (MRSA), vancomycin-resistant Enterococcus (VRE), and MDR Enterobacteriaceae Fecal metagenomes were analyzed from high-risk inpatients and compared to those of low-risk outpatients and controls with minimal risk for a MDR bacterial infection. Principal-component analysis clustered patient samples into distinct cohorts, confirming that the microbiome composition was significantly different between cohorts (P = 0.006). Microbial diversity and relative anaerobe abundance were preserved in outpatients compared to those in controls. Relative anaerobe abundance was significantly reduced in inpatients compared to that in outpatients (P = 0.006). Although the potential for MDR bacteria was increased in inpatients and outpatients compared to that in controls (P < 0.001), there was no difference between inpatients and outpatients. However, 9 (53%) inpatients had colonization with a MDR bacterium that was not identified by culture. Unlike culture, shotgun sequencing quantitatively characterizes the burdens of multiple MDR bacteria relative to all of the microbiota within the intestinal community. We propose consideration of key microbiome features, such as diversity and relative anaerobe abundance, in addition to the detection of MDR bacteria by shotgun metagenome sequencing as a novel method that might better identify patients who are at increased risk of a MDR infection.

FIG 1

Principal-component analysis (PCA) of the microbiome (a) and resistome (b). Species and antibiotic resistance gene compositions of all samples were clustered using the Euclidean distances for the abundance of each species (antibiotic resistance gene) between samples. The dots at the end of each radial vector represent the spatial orientation of each patient sample, with the distance between dots representing the maximum variance between samples. The error sum of squares is the center (centroid) of each cluster. Species composition was statistically unique between all three patient cohorts by MRPP (P = 0.006; A = 0.077). There was no significant difference in the resistome compositions between groups (P = 0.126; A = 0.012).

FIG 2

Box and whisker plot of microbiome diversity (a) and resistome diversity (b) by Shannon Index using the median and interquartile ranges for each cohort. There was significantly reduced microbial diversity in high-risk inpatients compared to that in controls (P = 0.005) and a nonsignificant decreased microbial diversity in high-risk inpatients compared to that in low-risk outpatients (P = 0.163). Preservation of microbial diversity was clearly demonstrated in low-risk outpatients compared to that in controls (P = 0.631). However, resistome diversity was not significantly different between groups (P = 0.110).

FIG 3

Bubble graphs showing relative changes in the relative abundances of all species in the microbiome (a) and all antibiotic resistance genes in the resistome (b) after classification into clinically relevant categories between cohorts. There were significant differences between all groups after the classification of species (P = 0.006; A = 0.319) and antibiotic resistance genes (P = 0.048; A = 0.275) into categories by MRPP. Relative anaerobe abundance was statistically different between all groups (P < 0.001). High-risk inpatients had significantly reduced anaerobe abundance compared to outpatients (P = 0.006), but there was no difference between low-risk outpatients and controls (P = 0.190).

FIG 4

Radar graphs comparing the abundances of clinically important species and antibiotic resistance genes for the detection of MDR bacterium potential in high-risk inpatients (red shaded upper left quadrant) and low-risk outpatients (blue shaded upper right quadrant) relative to the abundances in controls (off white shaded lower part of circle) for MRSA (a), VRE (b), AmpC (c), ESBL (d), and CRE (e). The Z-scores for all relevant species and antibiotic resistance genes are plotted on the vertical axes individually in color curated categories for broad visualization. Each stool sample is represented by the radial lines. The abundances of the species and antibiotic resistance genes for each sample relative to controls is depicted by the distance the colored line extends from the core. The central core (dark gray) represents the abundances in controls within 5 standard deviations (Z-score < 5.0). Samples with a relative abundance of at least one relevant species and antibiotic resistance gene with a Z-score of >5.0 were considered positive for MDR bacterium potential by shotgun metagenome sequencing.