Effect of Parenteral Antibiotic Administration on ...

384

Effect of Parenteral Antibiotic Administration on Persistence of VancomycinResistant Enterococcus faecium in the Mouse Gastrointestinal Tract Curtis J. Donskey, Jennifer A. Hanrahan, Rebecca A. Hutton, and Louis B. Rice

Division of Infectious Diseases, Louis Stokes Cleveland Department of Veterans Affairs Medical Center, and Case Western Reserve University School of Medicine, Cleveland, Ohio

Enterococci are part of the normal intestinal flora of humans and animals [1]. They have emerged as important nosocomial pathogens over the past 2 decades, at least in part because of their intrinsic and acquired resistance to many antimicrobial agents [2]. Enterococci are intrinsically resistant to aztreonam, cephalosporins, clindamycin, penicillinase-resistant penicillins, trimethoprim-sulfamethoxazole, and low levels of aminoglycosides. In addition, they have acquired resistance to penicillin (both by altered penicillin-binding proteins and by production of b-lactamase), erythromycin, gentamicin, streptomycin, tetracycline, and vancomycin. Studies in both humans and mice have demonstrated that selective pressure from antibiotic use can result in suppression of other flora and overgrowth of enterococci in the intestinal tract [3–6]. This overgrowth may predispose to the development of clinical infection [7–9]. Administration of cephalosporins, in particular, has been associated with enterococcal colonization and infection [6, 7, 10–12]. Vancomycin-resistant Enterococcus faecium (VRE) emerged as important nosocomial pathogens in the past decade. The Received 30 October 1998; revised 1 March 1999; electronically published 9 July 1999. Presented in part: 36th annual meeting of the Infectious Diseases Society of America, Denver, September 1998. The experimental protocol was approved by the Animal Research Review Board, Wade Park VA Hospital. Grant support: Wyeth-Ayerst Pharmaceuticals. Reprints or correspondence: Dr. Louis Rice, Infectious Diseases Section 1110(W), VA Medical Center, 10701 East Blvd., Cleveland, OH 44106 ([email protected]). The Journal of Infectious Diseases 1999; 180:384–90 q 1999 by the Infectious Diseases Society of America. All rights reserved. 0022-1899/99/8002-0018$02.00

risk of colonization or infection with VRE has been associated with use of various antibiotics, including oral and intravenous glycopeptides [13–16], third-generation cephalosporins [16–18], and antibiotics with potent antianaerobic activity [19, 20]. Because of these associations, alteration of hospital antimicrobial formularies has been suggested as a means of limiting the spread of VRE [21, 22]. Restriction of vancomycin use was not effective in reducing the incidence of VRE colonization in large polyclonal outbreaks [23]; however, Quale et al. [22] reported that VRE colonization was reduced from 47% to 15% over 9 months, when use of cefotaxime and clindamycin was significantly reduced and use of ampicillin-sulbactam and piperacillintazobactam were increased. They concluded that the change in antibiotics resulted in decreased VRE colonization, although concurrent control of a Clostridium difficile outbreak and increased infection control measures may also have played a role. In Cleveland, we have noted an association between hospital purchases of certain classes of antibiotics and the prevalence of clinical VRE isolates [24]. Specifically, overall purchases of third-generation cephalosporins, ticarcillin-clavulanic acid, and clindamycin correlated positively with rates of VRE isolation, whereas purchases of ampicillin-sulbactam, piperacillin, and piperacillin-tazobactam correlated negatively. MICs for several clinical VRE isolates from Cleveland were very high for cephalosporins and ticarcillin (110,000) compared with MICs for ampicillin and piperacillin (range, 312–1250). Therefore, we hypothesized that ampicillin-sulbactam and piperacillin-tazobactam result in lower rates of clinical VRE isolation because of partial inhibition of intestinal VRE. Here we report the results of experiments designed to compare the effects of different antibiotics on the persistence of VRE stool colonization in a mouse model.

Downloaded from http://jid.oxfordjournals.org/ by guest on June 25, 2014

A mouse model of vancomycin-resistant Enterococcus faecium (VRE) intestinal colonization was used to study the effect of different subcutaneous antibiotics on persistence and density of VRE colonization. Gastric inoculation of a clinical VanB VRE isolate, in conjunction with oral vancomycin in drinking water (250 mg/mL), resulted in high-level VRE colonization (mean, 9.5 log10 cfu/g) in all 169 experimental mice. After discontinuation of oral vancomycin, the level of VRE in the stool specimens of mice receiving subcutaneous saline steadily decreased (mean, 3.59 log10 cfu/g at day 19). Subcutaneous vancomycin, clindamycin, piperacillintazobactam, ticarcillin-clavulanic acid, metronidazole, cefotetan, ampicillin, and ampicillinsulbactam all promoted persistent high levels of stool VRE. Subcutaneous ceftriaxone, cefepime, ciprofloxacin, and aztreonam promoted increased VRE density to a lesser degree or not at all. Thus, in a mouse model, vancomycin and antibiotics with potent antianaerobic activity promoted persistent high-density intestinal VRE colonization, whereas antibiotics lacking potent antianaerobic activity did not.

JID 1999;180 (August)

Vancomycin-Resistant E. faecium in Mice

385

Materials and Methods

Figure 1. Experimental time line for mouse model of vancomycinresistant Enterococcus faecium (VRE) intestinal colonization. For initial colonization, mice were inoculated with 4 3 108 or 1 3 109 cfu of clinical VanB-type VRE isolate by gavage on day 27, while receiving oral vancomycin (250 mg/mL) from days 212 to 0. On days 0–19, mice were given subcutaneous antibiotics every 12 h. Stool specimens were collected on days 4, 9, 14, and 19, to assess effect of different antibiotics on persistence of VRE colonization. *, sample stool for VRE quantification.

antibiotic administration for determination of VRE colony counts. To confirm that isolates recovered during the course of the experiment were identical to the initial C68 inoculum, agar dilution MICs were done on randomly selected stool isolates during the course of the experiment, and PFGE was performed on 5 stool isolates. Initial antibiotic comparisons. In initial experiments, antimicrobial agents administered to individual groups of animals included sterile normal saline (negative control), cefepime (2.4 mg/ day), ceftriaxone (2.4 mg/day), clindamycin (1.4 mg/day), piperacillin-tazobactam (8 mg/day), ticarcillin-clavulanic acid (12 mg/ day), vancomycin (1.4 mg/day), and continued oral vancomycin (positive control). These antibiotics were chosen on the basis of clinical studies, suggesting a positive or negative (for piperacillintazobactam) association with VRE. Cefepime has not specifically been identified as a risk factor for VRE; however, this cephalosporin is similar to ceftazidime (which has been associated with VRE), in that it has limited activity against anaerobes and is primarily renally excreted. These experiments were performed twice. The first run of experiments consisted of 30 mice (4 in each treatment group, except for the oral vancomycin control group, which had 2 mice). The second run consisted of 50 mice (6 in each treatment group, except for the oral vancomycin control group, which had 8 mice). Comparison of antibiotics with various levels of antianaerobic activity. A second set of experiments was performed to further examine the role of antianaerobic activity in the promotion of persistent VRE colonization in this model. The experimental model followed was identical to the initial experiments, except that stool samples were collected on experimental days 5, 10, 15, and 20. Antimicrobial agents administered to individual groups of animals included sterile normal saline (negative control), cefepime (2.4 mg/ day), cefotetan (3.0 mg/day), aztreonam (3.0 mg/day), metronidazole (8.0 mg/day), ciprofloxacin (0.4 mg/day), ampicillin (4 mg/ day), and ampicillin-sulbactam (6 mg/day). These experiments were also done twice. The first run of experiments consisted of 47 mice (5 in each treatment group, except for the oral vancomycin group, which had 7 mice, and the saline group, which had 4 mice). The


Characterization of colonizing VRE strain. Enterococcus faecium C68 is a clinical wound isolate representing the most common pulsed-field gel electrophoresis (PFGE) genotype seen in a 1996 Cleveland area VRE outbreak [25]. It is phenotypically and genotypically VanB and expresses high levels of resistance to vancomycin. MICs for vancomycin and ampicillin were determined by standard agar dilution techniques using two-fold serial dilutions of antibiotic diluted in brain-heart infusion (BHI) agar. Single concentrations of antibiotics were used for agar susceptibility testing of chloramphenicol (10 mg/mL), erythromycin (10 mg/mL), tetracycline (10 mg/mL), gentamicin (500 mg/mL), and streptomycin (2000 mg/mL). MICs for high concentrations of ceftriaxone, cefepime, ticarcillin-clavulanic acid, piperacillin-tazobactam, and clindamycin were determined by a macrodilution technique in BHI broth [26]. Quantification of stool organisms. Fresh stool specimens (1–2 pellets) were weighed, diluted in 800 mL of sterile normal saline, homogenized with a pestle, serially diluted in sterile saline, and plated on Enterococcosel agar (Becton Dickinson, Cockeysville, MD), supplemented with 6 mg/mL vancomycin for quantification of VRE. The number of VRE colonies was read at 72 h, and the number of colony-forming units/gram of stool was calculated. Selected stool samples were plated onto Enterococcosel agar without vancomycin for quantification of total enterococci and onto MacConkey agar (Difco Laboratories, Detroit) for quantification of total aerobic and facultative gram-negative rods. Experimental model. A mouse model of VRE gastrointestinal colonization was developed based on the methods of Whitman et al. [14]. Female CF1 mice (Harlan Sprague-Dawley, Indianapolis), weighing 25–30 g, were used in all experiments. Mice were housed in individual cages and fed rodent chow and water ad libitum. Bedding was changed twice each week. VRE colonization was established in all animals by administering oral vancomycin (250 mg/ mL) in drinking water for 5 days before and 7 days after gastric inoculation of E. faecium C68 (figure 1). During oral vancomycin administration, water bottles were refilled with freshly prepared vancomycin solution every 2 days. On experimental day 27, gastric gavage of 0.5 mL of overnight culture, prepared by inoculating a single colony of E. faecium C68 into sterile BHI broth, was accomplished by a stainless steel feeding tube (Perfektum; Popper & Sons, New Hyde Park, NY). The bacterial inoculum was determined by serial dilution of the overnight culture. Stool colony counts of VRE and total enterococci were determined for a subset of mice on day 213 (prior to any intervention) and again on day 28 (after 4 days of oral vancomycin, prior to inoculation with oral VRE). On day 24, stool colony counts of VRE were determined for all animals, to confirm colonization with VRE. On experimental day 0, oral vancomycin was discontinued in all but 1 experimental group (positive control), and subcutaneous injection of antibiotics was initiated at 12-h intervals. Each antibiotic dose was diluted in 0.2 mL of sterile normal saline. Total daily doses of antibiotics were equivalent to the total daily dose (per kilogram) recommended for healthy human adults (calculated for a mouse weight of 30 g). Antibiotic stock solutions were prepared every 24–48 h and stored at 47C. Fresh stool (1–2 pellets) was collected from each mouse on days 4, 9, 14, and 19 of subcutaneous

386

Donskey et al.

Results MICs for E. faecium C68 were 512 mg/mL for vancomycin, 1 mg/mL for teicoplanin, !10 mg/mL for chloramphenicol, 110 mg/mL for erythromycin, 110 mg/mL for tetracycline, 12000 for streptomycin, 1500 for gentamicin, 256 for ampicillin, 1250 for piperacillin-tazobactam, 110,000 for ceftriaxone, 110,000 for cefepime, 110,000 for ticarcillin-clavulanic acid, and 110,000 for clindamycin. Antibiotic susceptibilities of 14 selected isolates, recovered from the stool of mice during antimicrobial administration, were identical to those determined for the original strain. The 5 stool isolates subjected to PFGE analysis had restriction patterns identical to E. faecium C68. Initial antibiotic comparisons. At day 213, prior to any intervention, 17/17 mice tested were colonized with vancomycin-susceptible enterococci, with counts of 4.16–8.95 log10 cfu/ g of stool (mean, 5.48). None of these 17 animals were colonized with detectable levels of VRE. On experimental day 28 (after 4 days of oral vancomycin, prior to inoculation of VRE), 11/ 12 mice had no detectable enterococci (lower limit of detection !3.02 to !3.43 log10 cfu/g) and 1/12 had 3.11 log10 cfu/g. None of these 12 mice had detectable VRE.

The VRE inoculum administered by gavage was 1 3 10 9 cfu for the first group of mice and 4 3 10 8 cfu for the second group. Three days after VRE inoculation (experimental day 24), all 80 mice were colonized with high levels of VRE (8.84–11.15 log10 cfu/g). The effect of subcutaneous antibiotics on persistence of VRE colonization is shown in figure 2. For mice receiving saline injections (negative controls), the level of VRE stool colonization steadily decreased during the 19 days of the experiment. For mice receiving continued oral vancomycin (positive controls), persistent high levels of VRE colonization were seen throughout the 19 days of the experiment. The level of VRE colonization in negative and positive controls over time was similar to the results reported by Whitman et al. [14]. On experimental day 4, all experimental groups and saline controls had high levels of stool VRE (all means, 18.0 log10 cfu/g). On days 9, 14, and 19, all antibiotics, except cefepime, promoted higher density of VRE colonization than did saline controls (P ! .005). On days 9, 14, and 19, subcutaneous clindamycin, piperacillin-tazobactam, ticarcillin-clavulanic acid, and vancomycin promoted more high-density VRE colonization than did subcutaneous cefepime or ceftriaxone (P ! .005). The level of VRE colonization maintained by treatment with subcutaneous clindamycin, piperacillin-tazobactam, ticarcillinclavulanic acid, and vancomycin was similar to that observed with continued administration of oral vancomycin. Variation in VRE counts within treatment groups was low at all time points (SD, !2.0 log10 cfu/g for all groups except the negative control group on day 9; SD, !1.0 log10 cfu/g for 66% [21/32] of all groups). No significant or persistent diarrhea was observed in any treatment group. For the 50 mice in the second run of experiments, assessment of total enterococci on day 19 revealed significant recolonization with vancomycin-susceptible enterococci in the saline (P ! .001), cefepime (P 5 .029 ), and clindamycin (P 5 .004) treatment groups. Assessment of total lactose-fermenting aerobic or facultative gram-negative organisms in these same animals revealed significant suppression of gram-negative organisms in the cefepime (P 5 .003) and piperacillin-tazobactam (P 5 .004) groups, compared with saline controls, and increased growth of gram-negative organisms in the clindamycin (P 5 .001) and oral vancomycin (P 5 .005) groups. Comparison of antibiotics with varying antianaerobic activity. Prior to any intervention, 24/24 mice tested were colonized with vancomycin-susceptible enterococci (3.5–7.5 log10 cfu/g of stool), and none were colonized with VRE. On day 28 (after 4 days of oral vancomycin, prior to inoculation of VRE), 24/24 mice tested had no detectable enterococci. The inoculum of VRE C68 administered by gavage was 109 cfu for both runs of these experiments. Three days after VRE inoculation (day 24), all 89 mice were colonized with VRE (4.83–10.16 log10 cfu/g). The effect of these subcutaneous antibiotics on VRE C68 colonization is shown in figure 3. The density of VRE colonization in saline controls and in oral van-


second run of experiments consisted of 42 mice (5 in each treatment group, except for the metronidazole group, which had 8 mice, and the saline group, which had 4 mice; no oral vancomycin control group was included). Metronidazole was given in a higher dose than the other antibiotics, because we observed during the first run of experiments that the human equivalent dose (0.9 mg/day) did not promote high density of VRE colonization. However, when metronidazole was given at 103 the initial dose after experimental day 20, the density of VRE markedly increased (data not shown). The higher dose of metronidazole (4 mg/day) selectively eliminates strictly anaerobic cecal bacteria in mice, with subsequent increased growth of aerobic and facultative species, including enterococci [4]. Promotion of VRE after day 20 did not occur when ciprofloxacin, aztreonam, or cefepime were given in similar high doses (data not shown). Statistical analysis. Statistical analysis for the presence of significant differences in VRE counts among groups at each experimental time point was done by analysis of variance. Linear regression was performed to determine which experimental groups were significantly different from negative controls. For the first set of experiments, linear regression was used to compare clindamycin, piperacillin-tazobactam, ticarcillin-clavulanic acid, and vancomycin groups with cefepime and ceftriaxone groups. For the second set of experiments, paired t tests (P ! .05 ) were used to compare antibiotic groups with greater antianaerobic activity (cefotetan, metronidazole, and ampicillin-sulbactam) with antibiotic groups with less antianaerobic activity (cefepime, ciprofloxacin, aztreonam, and ampicillin). Paired t tests (P ! .05 ) were done to compare levels of stool gram-negative organisms from experimental groups with those of the saline control group at experimental days 19 or 20 and to compare the level of total enterococci (VRE plus vancomycin-susceptible enterococci) with the level of VRE for each group at day 19 or 20.




387

comycin– (positive controls) and cefepime-treated animals was similar at each time point to the results of the initial experiments. On day 5, all mice had high levels of stool VRE (saline control and all treatment group means, 18.0 log10 cfu/g). On days 10, 15, and 20, subcutaneous metronidazole, cefotetan, ampicillinsulbactam, and ampicillin promoted persistent high-density VRE colonization, compared with that in saline controls, while aztreonam and cefepime groups did not differ from saline controls. The ciprofloxacin group did not differ from the saline group on days 10 or 15 but had higher VRE density on day 20 (P 5 .014). On days 10, 15, and 20, subcutaneous metronidazole, cefotetan, ampicillin-sulbactam, and ampicillin promoted VRE colonization more than ciprofloxacin, cefepime, or aztreonam (P ! .005 ). Variation in VRE counts within treatment groups at all time points was low (SD, !2.0 log10 cfu/g for all groups). Assessment of total lactose-fermenting aerobic or facultative

gram-negative organisms on day 20 revealed significant suppression of gram-negative organisms in the cefotetan, ciprofloxacin, and aztreonam groups (P ! .05 ). A trend toward decreased gram-negative organisms was observed in the cefepime, ampicillin, and ampicillin-sulbactam groups, and a trend toward increased gram-negative organisms was observed in the metronidazole group, but these changes were not statistically significant. Assessment of total enterococci on day 20 revealed significant recolonization with vancomycin-susceptible enterococci in the saline (P 5 .03), cefepime (P 5 .01), aztreonam (P 5 .03), metronidazole (P ! .005 ), and ciprofloxacin (P 5 .049) treatment groups, but not in the oral vancomycin, ampicillin, or ampicillin-sulbactam groups. Discussion Case-control studies have shown an association between VRE colonization or infection and certain classes of antibiotics,


Figure 2. Persistence and density of vancomycin-resistant Enterococcus faecium (VRE) intestinal colonization in mice. Mice were given different antibiotics subcutaneously (sc), as shown in legend (10 mice/group), every 12 h. All 80 mice were colonized with VRE on day 0 (mean, 9.5 log10 cfu/g stool). Negative control mice received sc saline; positive controls received oral vancomycin (250 mg/mL) in drinking water. Stool VRE levels were quantified on days 4, 9, 14, and 19. For comparisons between groups, analysis of variance and linear regression were done. pip, piperacillin; tic, ticarcillin.

388

Donskey et al.


including vancomycin [13–16], third-generation cephalosporins [16–18, 22], and antibiotics with antianaerobic activity [19, 20]. To our knowledge, this is the first published study to directly compare the effects of different antibiotics on persistence and density of VRE intestinal colonization. A higher density of VRE fecal colonization may significantly increase the risk of transmission of VRE strains to the environment and to other patients [27]. In this mouse model of VRE intestinal colonization, subcutaneous vancomycin, clindamycin, piperacillintazobactam, ticarcillin-clavulanic acid, metronidazole, cefotetan, ampicillin, and ampicillin-sulbactam promoted persistent high-density VRE colonization. Cefepime, ceftriaxone, aztreonam, and ciprofloxacin promoted VRE to a lesser degree or not at all. Our results support the clinical studies that have shown an association between VRE colonization or infection and antibiotics with potent antianaerobic activity. Antibiotics may influence VRE intestinal colonization directly by inhibiting VRE or indirectly by inhibiting other in-

testinal flora that compete with VRE, thereby promoting VRE overgrowth. It is not known whether inhibition of certain components of the competing bowel flora, such as anaerobes or gram-negative aerobes, results in greater promotion of VRE overgrowth than suppression of other components. Previous studies in mice and humans have suggested that the intestinal anaerobic flora is most important in providing colonization resistance against colonization of the bowel by potentially pathogenic microorganisms [5]. In addition to the intrinsic antimicrobial activity of antibiotics, factors such as the amount of biliary excretion and intestinal degradation of an antibiotic are important in determining the effect of an antibiotic on intestinal flora. The results of this study suggest that antimicrobial activity against anaerobes is the most important factor for promoting persistent high-density VRE stool colonization in mice. Piperacillin-tazobactam, ticarcillin-clavulanic acid, clindamycin, metronidazole, ampicillin-sulbactam, and cefotetan all have potent


Figure 3. Persistence and density of vancomycin-resistant Enteroccus faecium (VRE) intestinal colonization in mice treated with subcutaneous (sc) antibiotics with variable activity against anaerobes. All 74 mice were colonized with VRE on day 0 (mean, 9.5 log10 cfu/g stool). Mice received sc injections of antibiotics, as shown in legend, every 12 h (no. of mice/group in parentheses). Negative control mice received sc saline; positive control mice received oral vancomycin. Stool VRE levels were quantified on days 5, 10, 15, and 20. For comparisons between groups, analysis of variance and linear regression were done. amp, ampicillin.



hypothesize that these small inocula may be inhibited by the level of piperacillin achievable in the human small intestine, resulting in less VRE colonization in patients treated with this antibiotic. Further studies are planned, to determine whether administration of piperacillin will prevent establishment of VRE colonization in mice when small inocula are ingested. In contrast to our results, several case-control studies have demonstrated an association between VRE colonization or infection and third-generation cephalosporins that lack potent in vitro activity against anaerobes [16–18]. This discrepancy may be due to differences between the model we used and the clinical setting or to differences between mice and humans. Cephalosporins lacking potent activity against anaerobes may promote establishment of VRE colonization in clinical settings without causing high-density colonization. Fecal overgrowth with E. faecium in oncology patients, however, has been associated with preceding treatment with third-generation cephalosporins, including ceftriaxone, ceftizoxime, ceftazidime, and cefotaxime [6]. Some cephalosporins that lack potent antianaerobic activity in vitro can inhibit intestinal anaerobes in humans, with significant variation between individual antibiotics [31]. Although many case-control studies have grouped third-generation cephalosporins together as a risk factor for VRE, these antibiotics may promote VRE to different degrees in humans, on the basis of their widely varying levels of biliary excretion and activity levels against anaerobes. The applicability of this mouse model to human VRE colonization is potentially limited by several factors. Like human intestinal flora, rodent flora is predominantly anaerobic, but individual species differ significantly [32]. It is unclear whether these differences in flora affect the relationship between antibiotic therapy and VRE colonization. Also, although the strain of VRE used in these experiments is representative of the majority of VRE isolates in the Cleveland area, many outbreaks in the United States have been polyclonal, with a wide range of levels of resistance to various antibiotics. The effect of antibiotics on VRE colonization may vary for strains with differing levels of antibiotic resistance. In addition, we assessed the effect of antibiotics after establishing colonization with high levels of VRE. As noted previously, some antibiotics may influence the establishment of VRE colonization. The pharmacokinetics of many antibiotics differ significantly in mice and humans. In mouse treatment models, more frequent and higher relative doses of antibiotics are required to achieve tissue levels similar to those in humans. In this model, we gave mice antibiotic doses that were equivalent to human doses on the basis of weight. Our goal was to approximate (on a milligram per kilogram basis) the daily quantity of antibiotic seen in the gastrointestinal tracts of humans. Percentage of biliary excretion of ticarcillin-clavulanic acid (SmithKline Beecham Pharmaceuticals, Van Nuys, CA) is similar for mice and humans; however, no data were available on biliary excretion of the remaining antibiotics in mice. After parenteral administra-


activity against anaerobes and resulted in sustained high levels of stool VRE. Ampicillin has activity against intestinal anaerobic species that do not produce b-lactamases and promoted high-density VRE colonization. Ceftriaxone has less-potent antianaerobic activity than cefotetan and promoted increased VRE density to a lesser degree. Cefepime, aztreonam, and ciprofloxacin have minimal activity against anaerobes and affected VRE density minimally or not at all. Inhibition of aerobic gramnegative bacilli was neither necessary (clindamycin-treated mice had increased levels of aerobic gram-negative rods) nor sufficient (cefepime-, ciprofloxacin-, and aztreonam-treated mice had decreased levels of aerobic gram-negative rods) for the promotion of increased VRE density. Antimicrobial activity against vancomycin-susceptible enterococci was not necessary for the promotion of increased VRE density (clindamycin- and metronidazole-treated mice had significant regrowth of vancomycin-susceptible enterococci). Unlike the other antibiotics that promoted stool VRE, vancomycin does not have in vitro activity against Bacteroides fragilis or other gram-negative anaerobes. Although Edlund et al. [28] reported that oral vancomycin treatment markedly decreased the level of Bacteroides species in human volunteers through unclear mechanisms, our study of oral vancomycin treatment of mice did not suppress Bacteroides organisms in stool (data not shown). Vancomycin has activity against grampositive anaerobic organisms (e.g., clostridia) and aerobic grampositive organisms. Antimicrobial activity against gram-positive anaerobic organisms is a common feature of all antibiotics in this study, which promoted high-density VRE colonization. Also, it should be noted that subcutaneous administration of vancomycin promoted intestinal VRE as much as oral vancomycin. In humans, intestinal levels of vancomycin are relatively low after intravenous administration (6–10 mg/mL), but elimination of Clostridium species and elimination or reduction of E. faecalis in feces have been demonstrated [29]. The level of vancomycin present in the intestinal tract of mice after parenteral administration is not known. The absence of regrowth of vancomycin-susceptible enterococci in mice treated with subcutaneous vancomycin suggests that the active drug was present in the intestinal tract. Our hypothesis that piperacillin-tazobactam is associated with less clinical VRE isolation in the Cleveland area, because of its intrinsic activity against intestinal VRE, was not supported by this study. This may be because the model we used examined the effect of antibiotics on established VRE colonization rather than on initial development of colonization. After high levels of intestinal colonization have been established, promotion of VRE by inhibition of anaerobic flora may outweigh any direct suppressive effect of piperacillin-tazobactam on VRE itself. In humans, piperacillin achieves levels in bile (11000 mg/ mL after a 4-g intravenous dose [30]) that are above the MICs of many VRE isolates. In clinical settings, intestinal colonization may result from ingestion of small inocula of VRE. We

389

390

Donskey et al.

References 1. Finegold SM, Flora DJ, Attebery HR, Sutter VL. Fecal bacteriology of colonic polyp patients and control patients. Cancer Res 1975; 35:3407–17. 2. Murray BE. The life and times of the enterococcus. Clin Microbiol Rev 1990; 3:46–65. 3. Chandrasekar PH, Smith BR, LeFrock JL, Carr B. Enterococcal superinfection and colonization with aztreonam therapy. Antimicrob Agents Chemother 1984; 26:280–2. 4. Wells CL, Jechorek RP, Maddaus MA, Simmons RL. Effects of clindamycin and metronidazole on the intestinal colonization and translocation of enterococci in mice. Antimicrob Agents Chemother 1988; 32:1769–75. 5. Vollaard EJ, Clasener HAL. Colonization resistance. Antimicrob Agents Chemother 1994; 38:409–14. 6. Suppola JP, Volin L, Valtonen VV, Vaara M. Overgrowth of Enterococcus faecium in the feces of patients with hematologic malignancies. Clin Infect Dis 1996; 23:694–7. 7. Chirurgi VA, Oster SE, Goldberg AA, McCabe RE. Nosocomial acquisition of b-lactamase–negative, ampicillin-resistant enterococcus. Arch Intern Med 1992; 152:1457–61. 8. Montecalvo MA, Horowitz H, Gedris C, et al. Outbreak of vancomycin-, ampicillin-, and aminoglycoside-resistant Enterococcus faecium bacteremia in an adult oncology unit. Antimicrob Agents Chemother 1994; 38:1363–7. 9. Henning KJ, DeLancastre H, Eagan J, et al. Vancomycin-resistant Enterococcus faecium on a pediatric oncology ward: duration of stool shedding and incidence of clinical infection. Pediatr Infect Dis J 1996; 15:848–54. 10. Yu V. Enterococcal superinfection and colonization after therapy with moxalactam, a new broad-spectrum antibiotic. Ann Intern Med 1981; 94: 784–5. 11. Moellering RC Jr. Enterococcal infections in patients treated with moxalactam. Rev Infect Dis 1982; 4(Suppl):S708–11. 12. Pallares R, Pujol M, Pena C, Ariza J, Martin R, Gudiol F. Cephalosporins as a risk factor for nosocomial Enterococcus faecalis bacteremia. Arch Intern Med 1993; 153:1581–6. 13. Rice LB. The theoretical origin of vancomycin-resistant enterococci. Clin Microbiol Newslett 1995; 17:189–92.

14. Whitman MS, Pitsakis PG, DeJesus E, Osborne AJ, Levison ME, Johnson CC. Gastrointestinal tract colonization with vancomycin-resistant Enterococcus faecium in an animal model. Antimicrob Agents Chemother 1996; 40:1526–30. 15. van der Auwera P, Pensart N, Korten V, Murray BE, Leclercq R. Influence of oral glycopeptides on the fecal flora of human volunteers: selection of highly glycopeptide-resistant enterococci. J Infect Dis 1996; 173:1129–36. 16. Moreno F, Grota P, Crisp C, et al. Clinical and molecular epidemiology of vancomycin-resistant Enterococcus faecium during its emergence in a city in southern Texas. Clin Infect Dis 1995; 21:1234–7. 17. Tornieporth NG, Roberts RB, John J, Hafner A, Riley LW. Risk factors associated with vancomycin-resistant Enterococcus faecium infection or colonization in 145 matched case patients and control patients. Clin Infect Dis 1996; 23:767–72. 18. Livornese LLJ, Dias S, Samel C. Hospital-acquired infection with vancomycin-resistant Enterococcus faecium transmitted by electronic thermometers. Ann Intern Med 1992; 117:112–6. 19. Edmond MB, Ober JF, Weinbaum DL, et al. Vancomycin-resistant Enterococcus faecium bacteremia: risk factors for infection. Clin Infect Dis 1995; 20:1126–33. 20. Lucas GM, Lechtzin N, Puryear DW, Yau LL, Flexner CW, Moore RD. Vancomycin-resistant and vancomycin-susceptible enterococcal bacteremia: comparison of clinical features and outcomes. Clin Infect Dis 1998; 26:1127–33. 21. Boyce JM. Vancomycin-resistant enterococcus: detection, epidemiology and control measures. Infect Dis Clin North Am 1997; 11:367–84. 22. Quale J, Landman D, Saurina G, Atwood E, DiTore V, Patel K. Manipulation of a hospital antimicrobial formulary to control an outbreak of vancomycin-resistant enterococci. Clin Infect Dis 1996; 23:1020–5. 23. Morris JG, Shay DK, Hebden JN, et al. Enterococci resistant to multiple antimicrobial agents, including vancomycin: establishment of endemicity in a university medical center. Ann Intern Med 1995; 123:250–9. 24. Donsky CJ, Schreiber J, Jacobs, MR, et al. A polyclonal outbreak of predominantly vanB vancomycin-resistant enterococci in Northeast Ohio. Clin Infect Dis, in press. 25. Carias LL, Rudin SD, Donskey CJ, Rice LB. Genetic linkage and co-transfer of a novel, vanB-encoding transposon (Tn5382) and a low-affinity penicillin-binding protein 5 gene in a clinical vancomycin-resistant Enterococcus faecium isolate. J Bacteriol 1998; 180:4426–34. 26. National Committee for Clinical Laboratory Standards. Methods for dilution antimicrobial susceptibility tests for bacteria that grow aerobically. Wayne, PA: NCCLS, 1993:M7–A3 27. Murray BE. Editorial response: what can we do about vancomycin-resistant enterococci? Clin Infect Dis 1995; 20:1134–6. 28. Edlund C, Barkholt L, Olsson-Liljequist B, Nord CE. Effect of vancomycin on intestinal flora of patients who previously received antimicrobial therapy. Clin Infect Dis 1997; 25:729–32. 29. Geraci JE, Heilman FR, Nichols DR, Wellman WE, Ross GT. Some laboratory and clinical experiences with a new antibiotic, vancomycin. Proc Staff Meet Mayo Clin 1956; 31:564–82. 30. Taylor EW, Poxon V, Alexander-Williams J, Jackson D. Biliary excretion of piperacillin. J Int Med Res 1983; 11:28–31. 31. Bodey GP, Fainstein V, Garcia I, Rosenbaum B, Wong Y. Effect of broadspectrum cephalosporins on the microbial flora of recipients. J Infect Dis 1983; 148:892–7. 32. Wilson KH. The microecology of Clostridium difficile. Clin Infect Dis 1993; 16(Suppl 4):S214–8. 33. van Ogtrop ML, Guiot HFL, Mattie H, van Furth R. Modulation of the intestinal flora of mice by parenteral treatment with broad-spectrum cephalosporins. Antimicrob Agents Chemother 1991; 35:976–82.


tion, the fraction of ceftriaxone that is present in the intestinal tract of mice and humans is similar [33]. The fraction of cefepime that is excreted into the intestinal tract of mice and humans also appears to be similar, although precise information about the concentration of cefepime in human feces is not known [33]. The demonstrated effect of several antibiotics on the mouse colonic flora suggests that significant levels of antibiotics were achieved in this study. In conclusion, our data demonstrate that antibiotics with potent activity against anaerobes promote sustained high-density VRE stool colonization in mice. These findings suggest that once persons are colonized with VRE, administration of antibiotics with potent antianaerobic activity may promote sustained high-density colonization. Further research is necessary to identify the mechanisms by which antibiotics influence VRE intestinal colonization. If some classes of antibiotics promote VRE intestinal colonization more than other equally effective alternative antibiotics, then alteration of antimicrobial formularies may provide a means of limiting the spread of these organisms.
