Compositional Flux Within the Intestinal Microbiota and Risk for Bloodstream Infection With Gram-negative Bacteria

Abstract

Background: Gram-negative bloodstream infections (BSIs) represent a significant complication facing allogeneic hematopoietic cell transplant (allo-HCT) recipients, as a result of intestinal translocation during neutropenia. In this study we sought to better understand how the composition of the intestinal microbiota is connected to risk of gram-negative BSIs, expanding on our prior work in these patients.

Methods: Fecal specimens were collected from recipients of allo-HCT and analyzed using 16S ribosomal RNA gene sequencing. Samples and clinical data extending from the pretransplant conditioning period through stem cell engraftment were used in the analysis. Intestinal domination (relative abundance ≥ 30%) by gram-negative bacteria was used as predictor of gram-negative BSI using Cox proportional hazards modeling. Further analysis of microbiota composition was performed at the genus level.

Results: Seven hundred eight allo-HCT subjects were studied (7.5% developed gram-negative infection), with 4768 fecal samples for analysis. Gram-negative intestinal domination was associated with subsequent BSI, which was observed overall and individually at the genus level: Escherichia, Klebsiella, Enterobacter, Pseudomonas, and Stenotrophomonas. Fluoroquinolone prophylaxis was associated with decreased BSI and intestinal colonization by gram-negative microbes. In fluoroquinolone-prophylaxed patients, Escherichia coli was more frequently observed as breakthrough, both in terms of intestinal colonization and BSIs, compared with nonprophylaxed patients. Initial colonization by members of Ruminococcaceae and Bacteroidetes were associated with protection against gram-negative BSI.

Conclusions: Gram-negative intestinal colonization is highly predictive of BSI in the setting of allo-HCT. Fluoroquinolones appear to reduce these infections by influencing gut colonization.

Keywords: allogeneic hematopoietic cell transplantation; gram-negative bloodstream infections; intestinal microbiota.

© The Author(s) 2020. Published by Oxford University Press for the Infectious Diseases Society of America. All rights reserved. For permissions, e-mail: journals.permissions@oup.com.

Figures

Figure 1.

Peak relative abundances and concomitant bloodstream infection, by genus level. The maximum relative abundance (small black dot) achieved for each patient is shown for each genus. The primary middle plot shows genera within Gammaproteobacteria. In patients who developed a corresponding bloodstream infection with that strain, the peak abundance is represented by a colored circle. The top plot shows the peak abundances for all detectable genera. The bottom plot shows percentage of patients who had detectable colonization for each bacterial genus. Zero abundances are not shown. Forty-six of 54 bloodstream infections are shown here; for 8 infections, a corresponding microbiome signature was not found.

Figure 2.

Timing of neutropenia, intestinal domination, and bloodstream infection. The plot and inset graph show the overall timing of inpatient hospitalization, neutropenia, intestinal domination by Proteobacteria, and gram-negative bloodstream infection, over the course of transplantation. Intestinal domination is depicted based on nondaily stool collection. Abbreviation: allo-HCT, allogeneic hematopoietic cell transplantation.

Figure 3.

Differences in gram-negative abundance between subjects with and without fluoroquinolone prophylaxis. Boxplots show mean relative abundances of various gram-negative genera, comparing patients receiving fluoroquinolone prophylaxis to those who did not. These groups were tested for differences within each genus using Mann-Whitney U test (***P < .001, **P .001–.01, *P .01–.05). Gram-negative genera with highest abundances are shown; colored boxes represent gram-negative genera with observed potential for bloodstream infection. The upper and lower hinges denote 1 standard error from the mean, and the upper and lower whiskers denote 2 standard errors from the mean. Abbreviations: FQ, fluoroquinolone; ns, not significant; ppx, prophylaxis.