ABSTRACT
The use of oral vancomycin or metronidazole for treatment of Clostridium difficile infection (CDI) may promote colonization by health care-associated pathogens due to disruption of the intestinal microbiota. Because the macrocyclic antibiotic fidaxomicin causes less alteration of the intestinal microbiota than vancomycin, we hypothesized that it would not lead to a loss of colonization resistance to vancomycin-resistant enterococci (VRE) and extended-spectrum-β-lactamase-producing Klebsiella pneumoniae (ESBL-Kp). Mice (8 per group) received orogastric saline, vancomycin, or fidaxomicin daily for 5 days at doses resulting in stool concentrations in mice similar to those measured in humans. The mice were challenged with 105 CFU of orogastric VRE or ESBL-Kp on day 2 of treatment and concentrations of the pathogens in stool were monitored. The impact of drug exposure on the microbiome was measured by cultures, real-time PCR for selected anaerobic bacteria, and deep sequencing. In comparison to saline controls, oral vancomycin promoted establishment of high-density colonization by VRE and ESBL-Kp in stool (8 to 10 log10 CFU/g; P < 0.001), whereas fidaxomicin did not (<4 log10 CFU; P > 0.5). Vancomycin treatment resulted in significant reductions in enterococci, Bacteroides spp., and Clostridium leptum, whereas the population of aerobic and facultative Gram-negative bacilli increased; deep-sequencing analysis demonstrated suppression of Firmicutes and expansion of Proteobacteria during vancomycin treatment. Fidaxomicin did not cause significant alteration of the microbiota. In summary, in contrast to vancomycin, fidaxomicin treatment caused minimal disruption of the intestinal microbiota and did not render the microbiota susceptible to VRE and ESBL-Kp colonization.
INTRODUCTION
Oral vancomycin and oral metronidazole are the most commonly used antibiotics for treatment of Clostridium difficile infection (CDI). One limitation of these agents is that they are nonselective (i.e., they inhibit normal anaerobic intestinal microbiota in addition to C. difficile) (1–4). For example, oral vancomycin treatment may result in suppression of Bacteroides/Prevotella, Clostridium coccoides, and Clostridium leptum group organisms in stool (2, 3). Inhibition of the anaerobic microbiota by vancomycin and metronidazole during CDI treatment may contribute to recurrences of CDI and to colonization by health care-associated pathogens such as vancomycin-resistant enterococci (VRE) (4, 5).
Fidaxomicin is a macrocycle antibiotic approved by the Food and Drug Administration for the treatment of CDI (1). In comparison to vancomycin, fidaxomicin causes minimal disruption of the anaerobic microbiota and in clinical studies was associated with fewer recurrences of CDI and less frequent acquisition of VRE and Candida spp. during CDI treatment (1, 6). Given the relative sparing of the microbiota during fidaxomicin treatment, we hypothesized that this agent would not lead to a loss of colonization resistance to VRE and extended-spectrum-β-lactamase-producing Klebsiella pneumoniae (ESBL-Kp). Here, we used a mouse model to compare the effect of fidaxomicin versus vancomycin on establishment of intestinal colonization by VRE and ESBL-Kp.
MATERIALS AND METHODS
Pathogens.Several pathogens were studied here, including E. faecium C68, a previously described VanB-type clinical VRE isolate (7). K. pneumoniae P62 is a clinical isolate that produces an SHV type extended-spectrum β-lactamase (ESBL). Both organisms have been used in previous mouse model studies (7, 8).
Susceptibility testing.Broth dilution MICs of the test antibiotics for the test organisms were determined using standard methods for susceptibility testing of aerobic bacteria (9).
Quantification of stool pathogens.Fresh stool specimens were processed as described elsewhere (7, 8). In order to quantify VRE and ESBL-Kp, diluted samples were plated onto Enterococcosel agar (Becton Dickinson, Cockeysville, MD) containing vancomycin at 20 μg/ml and MacConkey agar (Becton Dickinson) containing ceftazidime at 10 μg/ml, respectively. The plates were incubated in room air at 37°C for 48 h, and the number of CFU of each pathogen per gram of sample was calculated.
Antibiotic dose selection.Dose finding experiments were run to determine the amount of vancomycin and fidaxomicin needed to be dosed to result in stool concentrations in mice similar to those measured in humans (i.e., 1,000 to 2,000 μg of vancomycin/g and 1,000 to 3,000 μg of fidaxomicin/g in stool) (see references 10, 11, and 12 and Merck data on file). Mice (5 per group) received a single oral administration of vancomycin or fidaxomicin. Fecal pellets were collected within three intervals of 0 to 4, 4 to 8, and 8 to 24 h after dosing. Fecal levels of vancomycin, fidaxomicin, and OP-1118 were measured by liquid chromatography-mass spectrometry and confirmed using satellite animals dosed at 1.125 mg/day or 37.5 mg/kg for vancomycin and at 0.9 mg/day or 30 mg/kg and at 2.3 mg/day or 75 mg/kg for fidaxomicin. These dosing regimens resulted in measured maximal fecal peak levels of 1,826 μg of vancomycin/g and of 920 and 1,600 μg of fidaxomicin+OP-1118/g for the 30- and 75-mg/kg fidaxomicin doses, respectively. For the majority of experiments, the lower dose of fidaxomicin was used based upon the fact that the human dose of fidaxomicin is 80% of the usual daily dose of vancomycin (i.e., 400 mg per day versus 500 mg per day). Additional experiments were conducted using the higher dose of fidaxomicin because this dose resulted in a measured peak concentration that was equivalent to the peak concentration of vancomycin and that was equivalent to concentrations measured in humans receiving fidaxomicin (10).
Effect of the antibiotics on intestinal microbiota.The Animal Care Committee of the Cleveland Veterans Affairs Medical Center approved the experimental protocol. Initial experiments were conducted to assess the effects of treatment with the test antibiotics or saline on the intestinal microbiota of mice. Female CF-1 mice (6 per group) weighing ∼30 g (Harlan Sprague-Dawley, Indianapolis, IN) were housed in individual cages. Mice received daily oroesophageal instillation of the test antibiotics (0.2 ml, total volume) for 5 days using a stainless steel feeding tube (Perfektum, Popper & Sons, New Hyde Park, NY).
Quantitative culture of stool microbiota.Stool samples were collected at baseline, on days 2 and 5 of treatment, and at 3, 5, and 10 days after treatment for the evaluation of the effect of the antibiotics on the microbiota. Quantitative cultures for facultative and aerobic Gram-negative bacilli and enterococci were performed by plating serially diluted specimens onto MacConkey agar (Difco Laboratories, Detroit, MI) and Enterococcosel agar (Becton Dickinson), respectively.
Deep-sequencing analysis of stool microbiota.Deep-sequencing analysis was completed for mice treated with vancomycin and the lower dose of fidaxomicin. Fecal bacterial DNA was extracted from ∼500 mg of feces using the QIAmp DNA stool minikit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. Sequencing and analysis was carried out by Second Genome (San Bruno, CA). To enrich the samples for the bacterial 16S V4 rRNA gene region, DNA was PCR amplified using fusion primers designed against surrounding conserved regions, which are tailed with sequences to incorporate Illumina (San Diego, CA) adapter and indexing barcodes. After Illumina library construction, amplicons were sequenced using a MiSeq benchtop sequencer instrument (Illumina). Using QIIME and custom scripts, sequences were quality filtered and demultiplexed using exact matches to the supplied DNA barcodes. The resulting sequences were searched against the Greengenes reference database of 16S sequences, clustered at 97% by uclust (closed-reference OTU picking). The longest sequence from each operation taxonomic unit (OTU) thus formed was used as the OTU representative sequence and assigned taxonomic classification via MOTHUR's Bayesian classifier, trained against the Greengenes database clustered at 98%. Principal coordinate analysis using weighted Unifrac as the distance metric was carried out to visualize complex relationships between samples. A permutation-based multivariate analysis of variance test using distance metrics, as implemented in the Adonis function in the Vegan Package for R, was used to assess whole microbiome differences among groups (13, 14). Bar plot representations were generated to show the top eight microbial groups at the phylum level.
Analysis of Bacteroides spp. and Clostridium leptum by qPCR.Real-time PCR (qPCR) analysis was completed for mice treated with vancomycin and the lower dose of fidaxomicin. To determine the effect of antibiotic treatment on the concentration of Bacteroides spp. and C. leptum, a representative Firmicutes organism, qPCR was performed using the methods and primers of Louie et al. (2). Fecal bacterial DNA was extracted from 100 mg of feces using the QIAmp DNA stool minikit (Qiagen) according to the manufacturer's instructions. Purified template DNA from Bacteroides fragilis and C. leptum was used for melting-curve analysis and to generate standard curves for each primer set using 10-fold serial dilutions of DNA ranging from 10 to 10−6 ng. qPCR was performed using the CFX96 detection system (Bio-Rad, Hercules, CA). Amplification and detection were conducted in 96-well plates with SYBR green 2× qPCR master mix (Bio-Rad). Each sample was run in triplicate in a final volume of 20 μl containing a final concentration of 0.3 μM for each primer and 5 μl of 2-ng/μl template DNA using the following parameters: 1 cycle at 94°C for 5 min, followed by 49 cycles at 94°C for 20 s, 56 to 58°C for 20 s, and 72°C for 20 s.
Effect of the antibiotics on establishment of colonization by VRE and ESBL-Kp.To assess the effects of treatment on initial establishment of colonization, mice (8 per group) received oroesophageal instillation of 10,000 CFU of VRE or ESBL-Kp on day 2 of 5 of daily treatment with vancomycin or the lower dose of fidaxomicin or saline as described previously. The concentration of VRE and ESBL-Kp in stool was measured on day 5 of antibiotic treatment and 3, 5, and 10 days after completion of antibiotics.
Effect of the higher dose of fidaxomicin (75 mg/kg) on the microbiota and establishment of colonization by VRE and ESBL-Kp.To assess the impact of the higher dose of fidaxomicin on the microbiota, quantitative cultures for facultative and aerobic Gram-negative bacilli and enterococci were performed as described previously for mice treated with fidaxomicin or saline for 5 days. To assess the effect of the higher dose of fidaxomicin on establishment of colonization by VRE and ESBL-Kp, mice (8 per group) treated for 5 days with (i) oral saline, (ii) fidaxomicin at 2.3 mg/day (75 mg/kg), (iii) clindamycin at 1.4 mg/day, or (iv) both fidaxomicin and clindamycin received 10,000 CFU of oral VRE or ESBL-Kp on day 2 of treatment. The concentration of VRE and ESBL-Kp in stool was measured at baseline and 3 and 6 days after pathogen inoculation. The purpose of including a group receiving fidaxomicin plus clindamycin was to assess whether fidaxomicin has sufficient inhibitory activity to prevent clindamycin-associated promotion of VRE overgrowth (7).
Statistical analysis.One-way analysis of variance was performed to compare concentrations of organisms among the treatment groups. P values were adjusted for multiple comparisons using the Scheffe correction. Computations were performed with the use of Stata (version 5.0; Stata, College Station, TX) and Origin (version 9; OriginLab, Northampton, MA).
RESULTS
Susceptibility testing.MICs for ESBL-Kp were >256 μg/ml for vancomycin, metronidazole, and fidaxomicin. MICs for VRE were 256, >256, and 2 μg/ml for vancomycin, metronidazole, and fidaxomicin, respectively.
Effect of the antibiotics on indigenous enterococci and facultative Gram-negative bacilli by quantitative culture.Figure 1 shows the effect of antibiotic treatment on the concentrations of enterococci (A) and aerobic and facultative Gram-negative bacilli (B) by culture. Vancomycin significantly reduced levels of enterococci during treatment, whereas fidaxomicin did not. Levels of enterococci returned to baseline concentrations by 3 days after discontinuation of vancomycin. In comparison to saline controls, vancomycin exposure resulted in a 4-log increase in Gram-negative bacilli, whereas fidaxomicin did not. By 10 days after discontinuation of vancomycin, levels of Gram-negative bacilli were not significantly elevated in comparison to baseline levels.
Effect of antibiotic treatment on the concentrations of enterococci (A) and aerobic and facultative Gram-negative bacilli (B) in stool by culture. Mice received daily oral antibiotic treatment for 5 days. The error bars represent the standard error. *, P < 0.05.
Effect of the antibiotics on indigenous microbiota by deep sequencing and qPCR.Figure 2 shows the relative proportions of different bacterial phyla on day 5 of antibiotic exposure in comparison to the saline control group, including the summed total for each treatment group and data for individual mice. In control mice, Bacteroidetes and Firmicutes were predominant, with Proteobacteria making up only <2% of the indigenous microbiota. Fidaxomicin exposure was associated with a reduction in Firmicutes from ∼40 to ∼20% with no increase in Proteobacteria. In contrast, vancomycin treatment was associated with suppression of Firmicutes from ∼40% to <10% of the microbiota and expansion of Proteobacteria.
Comparison of stool microbiota of mice by 16S deep-sequencing analysis after 5 days of antibiotic treatment. The relative abundances of the major bacterial phyla are shown. Numbers indicate data for individual mice in each group.
Figure 3 shows the relative proportions of the different taxa in the vancomycin and fidaxomicin groups before, during, and after treatment. For the vancomycin group, there was an increased proportion of Proteobacteria at baseline in comparison to the other groups that was attributable to the presence of one outlier mouse; however, the differences between the groups at baseline were not statistically significant. For the vancomycin group, the proportion of Firmicutes increased from the end of treatment (day 5) to 10 days posttreatment (day 15), whereas the proportion of Proteobacteria decreased.
Comparison of stool microbiota of mice by 16S deep-sequencing analysis before, during, and after treatment with oral fidaxomicin or vancomycin. Mice received daily oral antibiotic treatment for 5 days (day 0 to day 5). Numbers indicate day of sample collection: day 0, prior to treatment; day 5, after 5 days of antibiotic treatment; day 10, 5 days after last antibiotic dose; and day 15, 10 days after last antibiotic dose. The relative abundances of the major bacterial phyla are shown as a composite of five total mice in each group at each time point.
Real-time PCR analysis demonstrated that vancomycin significantly reduced the concentrations of Bacteroides spp. (8.7 versus 5.6 log10CFU/g stool) and C. leptum (6.2 versus 5.6 log10CFU/g stool) on day 5 of treatment (P < 0.001 for each comparison), whereas fidaxomicin did not (P > 0.5).
Effect of antibiotic exposure on establishment of colonization by VRE and ESBL-Kp.Figure 4 shows the effect of exposure to vancomycin and the lower dose of fidaxomicin on establishment of colonization by VRE (Fig. 4A) and ESBL-Kp (Fig. 4B). In comparison to controls, oral vancomycin promoted overgrowth of both pathogens (P < 0.001), whereas fidaxomicin did not promote overgrowth of either pathogen. None of the control or fidaxomicin-treated mice had detectable VRE at any time point.
Effect of antibiotic treatment on establishment of colonization by VRE (A) and ESBL-Kp (B) in mice. Mice received daily oral antibiotic treatment for 5 days. The pathogens were administered orally on day 2 of antibiotic treatment. The error bars represent the standard error.
Effect of the higher dose of fidaxomicin (75 mg/kg) on the microbiota and establishment of colonization by VRE and ESBL-Kp.In comparison to saline controls, the higher dose of fidaxomicin significantly reduced concentrations of enterococci on day 5 of treatment (4.3 versus 6.1 log10CFU/g stool; P < 0.01), with levels returning to baseline by 3 days after treatment. Concentrations of aerobic and facultative Gram-negative bacilli did not differ between the fidaxomicin-treated mice and saline controls at any time point. As shown in Fig. 5, in comparison to saline controls, the higher dose of fidaxomicin did not promote overgrowth of VRE when challenged with oral VRE during treatment, whereas clindamycin alone or in combination with fidaxomicin did (P < 0.001); the concentrations of VRE were significantly higher in the clindamycin versus the clindamycin plus fidaxomicin group (P < 0.01). In comparison to saline controls, the higher dose of fidaxomicin also did not promote overgrowth of ESBL-Kp (peak concentration, 3.8 and 3.9 log10CFU/g stool; P = 1).
Effect of antibiotic treatment on establishment of colonization by VRE in mice. Mice received daily oral antibiotic treatment for 5 days. The pathogens were administered orally on day 2 of antibiotic treatment. The error bars represent the standard error.
DISCUSSION
In contrast to oral vancomycin, we found that oral fidaxomicin did not promote overgrowth of VRE and ESBL-Kp in mice. Vancomycin promoted overgrowth of aerobic and facultative Gram-negative bacilli, whereas fidaxomicin did not. By deep-sequencing analysis, vancomycin treatment resulted in marked suppression of Firmicutes and expansion of Proteobacteria, whereas fidaxomicin was associated with only a minor reduction in Firmicutes with no increase in Proteobacteria. By qPCR analysis, vancomycin suppressed levels of Bacteroides spp., and Clostridium leptum, whereas fidaxomicin did not. These findings add to the body of literature suggesting that the relative preservation of the intestinal microbiota during fidaxomicin treatment may be beneficial in reducing the risk for acquisition and overgrowth of health care-associated pathogens during CDI treatment.
Because fidaxomicin has minimal activity against Gram-negative bacilli, the lack of promotion of overgrowth of indigenous Gram-negative bacilli and ESBL-Kp is attributable entirely to relative preservation of the intestinal microbiota. However, fidaxomicin does have activity against enterococci (MIC for VRE test strain, 2 μg/ml). Therefore, lack of promotion of VRE overgrowth could be attributable to inhibitory activity against enterococci. The fact that fidaxomicin did not completely prevent overgrowth of VRE induced by disruption of the microbiota by clindamycin, it is likely that the reduced VRE expansion is due to both inhibitory activity and relative preservation of the microbiota.
Our findings for fidaxomicin and vancomycin are consistent with previous studies (4–6). Fidaxomicin treatment of CDI was associated with infrequent acquisition of VRE and Candida spp. colonization in comparison to oral vancomycin (6). Fidaxomicin may represent a good alternative to metronidazole use when vancomycin is not being considered. The finding that fidaxomicin exposure did not promote colonization by ESBL-Kp is significant given the increasing importance of emerging multiresistant Gram-negative pathogens (15).
Our study has some limitations. The study was conducted using a mouse model with healthy mice. Additional studies will be required to confirm that the findings are applicable to patients with CDI. We studied only one strain each of VRE and K. pneumoniae. However, we have previously shown that multiple VRE and K. pneumoniae strains gave similar results in the mouse model (7, 8). We studied only one species of antimicrobial-resistant Gram-negative bacilli. Future studies are needed that include other species such as Acinetobacter spp. Although the lower dose of fidaxomicin was 80% of the vancomycin dose (i.e., the same the ratio as in human dosing), the fecal concentration of fidaxomicin plus OP-1118 measured in mouse fecal pellets was lower than levels measured in human feces (10) and lower than the fecal concentration of vancomycin in mice. The lower fecal fidaxomicin levels measured in mice could potentially be due to lower technical extraction and recovery of fidaxomicin and OP-1118 from mouse versus human samples or due to differences between excretion or metabolism of the drug in mice and humans. The higher dose of fidaxomicin did result in a measured fecal fidaxomicin concentration that was similar to the concentration of vancomycin, and the higher dose did not promote colonization by VRE or ESBL-Kp. Finally, we did not include metronidazole in our evaluation. However, Lewis et al. (16) recently demonstrated that oral metronidazole promotes colonization by VRE and antibiotic-resistant Gram-negative bacilli in mice, although to a lesser degree than oral vancomycin.
ACKNOWLEDGMENTS
This study was supported by the Department of Veterans Affairs and by Merck and Co, Inc., Kenilworth, NJ.
FOOTNOTES
- Received 11 November 2015.
- Returned for modification 20 January 2016.
- Accepted 11 April 2016.
- Accepted manuscript posted online 18 April 2016.
- Copyright © 2016, American Society for Microbiology. All Rights Reserved.
