Emergence and Acquisition of Fluoroquinolone-Resistant Gram-Negative Bacilli in the Intestinal Tracts of Mice Treated with Fluoroquinolone Antimicrobial Agents

Abstract

After mice received orogastric administration of a fluoroquinolone-resistant Klebsiella pneumoniae strain, subcutaneous treatment with ciprofloxacin, levofloxacin, and moxifloxacin promoted persistent low-density colonization in 10% to 40% of the mice, whereas treatment with clindamycin consistently promoted high-density colonization. No emergence of fluoroquinolone-resistant gram-negative bacilli was detected in the mice during or after treatment with the fluoroquinolone antimicrobial agents.

Antimicrobial agents that are excreted into the intestinal tract may facilitate the emergence and acquisition of antimicrobial-resistant organisms. Antimicrobial-resistant strains may emerge as a consequence of the genetic alteration of susceptible strains in the colon, or preexisting subpopulations of resistant organisms may expand due to selective pressure (i.e., inhibition of competing microflora but not of resistant organisms) (5, 6). Antimicrobial-resistant organisms that are ingested may also establish colonization due to antimicrobial selective pressure (5, 6). Factors such as the concentration of the antimicrobial achieved in the intestinal tract, the duration of therapy, and the degree of disruption of the anaerobic microflora may influence the likelihood that resistant strains will emerge or be acquired (5, 21).

Numerous studies have demonstrated that fluoroquinolone-resistant gram-negative bacilli may emerge during treatment with fluoroquinolone antimicrobial agents (2, 3, 11, 12, 15-18, 20-22). Because the fluoroquinolones differ significantly with regard to the amount of intestinal excretion and in vitro activity against anaerobic bacteria (8, 9, 21), it is plausible that they may differ in their propensity to promote colonization with fluoroquinolone-resistant gram-negative bacilli. For example, ciprofloxacin achieves relatively high concentrations in the intestinal tract (185 to 2,220 μg/ml) in comparison with that of other fluoroquinolones, and some newer agents (i.e., moxifloxacin and gatifloxacin) have enhanced activity against anaerobes (8, 9, 21). Because levofloxacin achieves relatively low concentrations in the intestinal tract (0 to 163 μg/ml) and has modest activity against intestinal anaerobes (21), it might theoretically be less likely to promote emergence of resistant gram-negative bacilli. In addition, recent studies indicate that “high-dose, short-course” (i.e., 750 mg daily for 5 days) therapy with levofloxacin is as effective as 10 days of therapy at a lower dose for the treatment of community-acquired pneumonia and sinusitis (7, 19). This treatment strategy decreases total drug exposure by 25%, but it is not known if “high-dose, short-course” regimens are associated with a lower risk that resistant organisms will be selected. The objective of this study was to examine the relative risk of emergence and acquisition of colonization with fluoroquinolone-resistant gram-negative bacilli during treatment with different fluoroquinolone antimicrobial agents in mice.

The experimental protocol was approved by the Cleveland Veterans Affairs Medical Center's animal care committee. Female CF1 mice (Harlan Sprague-Dawley, Indianapolis, IN) weighing 25 to 30 g were housed individually. Three fluoroquinolones (ciprofloxacin, levofloxacin, and moxifloxacin) were studied; clindamycin was included as a positive control because we have previously demonstrated that it promotes overgrowth of antimicrobial-resistant gram-negative bacilli. Two different doses of antimicrobials were studied, a lower dose equal to the usual human dose administered over a 24-h period (milligrams of antibiotic per gram of body weight) and a dose that was 12-fold higher than that, calculated based on the method of Freireich et al. (10). All antimicrobials were administered subcutaneously once each day in 0.2 ml of phosphate-buffered saline for 10 days. For levofloxacin, one group received 10 days of treatment with a dose equivalent to 500 mg per day, and an additional group received treatment for 5 days with a dose equivalent to 750 mg per day. The dose of each antimicrobial was as follows: 0.5 or 6 mg/day for ciprofloxacin; 0.25 or 3 mg/day for levofloxacin; 0.375 or 4.5 mg/day for high-dose, short-course levofloxacin; 0.2 or 2.4 mg/day for gatifloxacin; 0.2 or 2.4 mg/day for moxifloxacin; and 1.4 mg/day for clindamycin.

An initial set of experiments was performed to determine whether the antimicrobial agents promoted acquisition of an exogenously administered fluoroquinolone-resistant gram-negative bacillus, Klebsiella pneumoniae KPC3. KPC3 is a clinical bloodstream infection isolate from the Cleveland VA Medical Center that produces a carbapenemase (Robert Bonomo, personal communication). The broth dilution MICs of ciprofloxacin and ceftriaxone for KPC3 were 256 and >64 μg/ml, respectively. Mice (10 per group) received subcutaneous antimicrobial treatments as described above, and 1,000 CFU of KPC3 was administered by orogastric gavage on day 2 of treatment in 0.5 ml of phosphate-buffered saline, using a stainless steel feeding tube (Perfektum; Popper & Sons, New Hyde Park, NY). Stool samples were collected prior to gavage with KPC3 and on days 1, 3, 6, and 10 postgavage, and the density of colonization was measured by plating serially diluted samples onto MacConkey agar with and without 1 μg/ml of ciprofloxacin. For these experiments, the limit of detection was 2.5 log₁₀ CFU/g of stool. To confirm that the fluoroquinolone-resistant gram-negative bacilli recovered were KPC3, a subset of 20 stool isolates was subjected to identification and susceptibility testing.

The concentrations of antimicrobials in stool were determined on day 5 of treatment using an agar diffusion assay with Escherichia coli as the indicator strain (13). Because the fluoroquinolones did not consistently promote colonization by KPC3, mice were challenged with clindamycin, subcutaneously, daily for 3 days after fluoroquinolone treatment was completed to assess whether low levels of KPC3 (i.e., below the limit of detection) could be augmented by disruption of the anaerobic microflora. An additional experiment was conducted in which the lower dose of the fluoroquinolones was given in combination with clindamycin or ceftriaxone to determine whether the absence of overgrowth was attributable to inhibitory activity of the fluoroquinolones. Both clindamycin and ceftriaxone promoted colonization by K. pneumoniae and vancomycin-resistant enterococci in mice in previous studies (13). Finally, mice receiving the higher doses of fluoroquinolone or clindamycin were challenged with 1,000 CFU of KPC3 by orogastric gavage 2 days after antimicrobial treatment was completed.

A second set of experiments was conducted to examine the effects of the antimicrobial agents on the emergence of fluoroquinolone-resistant gram-negative bacilli in stool. Twenty mice per group received subcutaneous treatment with the lower dose of fluoroquinolones daily as described above, and 15 mice per group received treatment with the higher dose of fluoroquinolones. Stool samples were collected prior to treatment, after 5 and 10 days of treatment, and at 7 days after treatment was completed and were assessed for emergence of fluoroquinolone-resistant gram-negative bacilli by plating onto MacConkey agar containing 0.125 μg/ml of levofloxacin. For this analysis, the limit of detection was 1.7 log₁₀ CFU/g of stool (i.e., ∼100 mg of stool was mixed thoroughly with a pipette tip in 800 μl of sterile saline, and 200 μl of the suspension was plated onto the selective medium). Stool samples were also plated onto MacConkey agar with no antibiotics to determine the effect of treatment on Enterobacteriaceae.

Data analyses were performed with Stata software (version 6.0; Stata, College Station, TX). A one-way analysis of variance was performed to compare the groups, with P values adjusted for multiple comparisons using the Scheffe correction.

The effect of antibiotic treatment on the establishment of colonization by KPC3 is shown in Fig. 1. At baseline, all of the mice were colonized with lactose-positive, fluoroquinolone-susceptible gram-negative bacilli (range, 4.9 to 8.0 log₁₀ CFU/g stool), and none had detectable fluoroquinolone-resistant gram-negative bacilli in the stool (level of detection, ∼2.5 log₁₀ CFU/g). Clindamycin promoted high-density overgrowth of KPC3 in comparison to that of saline controls (P < 0.001), whereas the low and high doses of the fluoroquinolones did not (P > 0.31). However, from 20% to 40% of the mice in each of the fluoroquinolone treatment groups (ciprofloxacin, 20%; levofloxacin, 30%; levofloxacin short course, 40%; moxifloxacin, 30%) maintained low levels of detectable KPC3, whereas none of the saline-treated control mice had detectable levels of KPC3. When the saline- and fluoroquinolone-treated mice were challenged with clindamycin for 3 days after fluoroquinolone treatment was completed, the mice that had detectable levels of KPC3 developed high-density colonization (∼9 log₁₀ CFU/g); none of the mice with undetectable levels of KPC3 developed detectable levels in response to clindamycin. Mice that received fluoroquinolones in combination with clindamycin or ceftriaxone had high-density overgrowth of KPC3 in comparison to that of the saline controls (P < 0.001) (Fig. 1C and D), suggesting that the absence of overgrowth during fluoroquinolone monotherapy was due to preservation of inhibitory anaerobic microflora rather than inhibitory activity of the fluoroquinolones against KPC3. When the mice were challenged with KPC3 2 days after they completed treatment, none of the fluoroquinolone-treated mice developed high-density colonization, whereas all of the clindamycin-treated mice did (data not shown). All 20 of the fluoroquinolone-resistant gram-negative bacillus isolates from stool samples that were subjected to testing were confirmed to be K. pneumoniae, with susceptibility patterns identical to KPC3.

FIG. 1.

Effect of subcutaneous antimicrobial treatment on the establishment of intestinal colonization by a fluoroquinolone-resistant Klebsiella pneumoniae strain (KPC3) in mice. (A) Lower-dose fluoroquinolones (i.e., equivalent to the usual human dose on a mg/kg basis); (B) higher-dose fluoroquinolones (i.e., 12-fold higher than the usual human dose on a mg/kg basis); (C) lower-dose fluoroquinolones in combination with clindamycin; and (D) lower-dose fluoroquinolones in combination with ceftriaxone. Mice received antimicrobial treatment once daily for 10 days, and 1,000 CFU of KPC3 was administered by orogastric gavage on day 2 of treatment; the “short-course, high-dose” levofloxacin group received treatment equivalent to the 750-mg dose for 5 days. If the pathogens were not detected in stool, the lower limit of detection (∼2.5 log₁₀ CFU/g) was assigned. Error bars indicate standard error.

For the lower-dose treatment regimen, the concentrations (mean ± standard deviation [SD]) of ciprofloxacin, levofloxacin, levofloxacin (high-dose, short-course), and moxifloxacin in stool were 215.1 ± 29.1 μg/g, 48.4 ± 46.5, 77.4 ± 69.2, and 64.0 ± 37.6, respectively. For the higher-dose regimen, the concentrations (mean ± SD) of ciprofloxacin, levofloxacin, levofloxacin (high-dose, short-course), and moxifloxacin in stool were 1,319.5 ± 139.2 μg/g, 262.5 ± 75.5, 285.8 ± 290.3, and 82.5.0 ± 48.1, respectively.

In the second set of experiments to determine the emergence of resistant organisms from within the indigenous intestinal microflora, no emergence of gram-negative bacilli with even low-level fluoroquinolone resistance (i.e., levofloxacin MIC of ≥0.125 μg/ml) was detected during or after treatment with the lower or higher dose of the fluoroquinolones. Each of the fluoroquinolones at either dosage consistently suppressed total Enterobacteriaceae bacteria to undetectable levels.

In summary, we found that the treatment of mice with fluoroquinolone antimicrobial agents suppressed indigenous Enterobacteriaceae bacteria but was not associated with emergence of detectable levels of fluoroquinolone-resistant gram-negative bacilli. After mice received exogenous administration of KPC3, a fluoroquinolone-resistant K. pneumoniae strain, subcutaneous administration of ciprofloxacin, levofloxacin, and moxifloxacin promoted persistent low-density colonization in 10% to 40% of the treated mice, whereas treatment with clindamycin alone and clindamycin or ceftriaxone in combination with the fluoroquinolones consistently promoted high-density colonization. These data are consistent with several studies of healthy human volunteers in the community that demonstrated that de novo emergence of resistant gram-negative bacilli may be uncommon during fluoroquinolone monotherapy regimens (8, 9, 21). However, frequent emergence of fluoroquinolone-resistant gram-negative bacilli has been observed in some community studies and in many studies in health care facilities (2, 3, 11, 12, 14-18, 20-22). Factors such as repeated exposures or preexisting colonization with resistant gram-negative organisms could contribute to more frequent acquisition of resistant gram-negative bacilli in these studies. In some regions, repeated ingestion of fluoroquinolone-resistant E. coli from contaminated chicken products could be an important contributor to human colonization (14). In healthcare facilities, our findings suggest that more frequent acquisition of fluoroquinolone-resistant gram-negative bacilli may be related in part to concurrent or sequential use of other antimicrobials that cause greater disruption of the indigenous microflora.

Although the fluoroquinolones differ with regard to the amount of intestinal excretion and to the in vitro activity against anaerobic bacteria (8, 21), we did not find significant differences among the three agents with regard to promotion of KPC3 or emergence of fluoroquinolone-resistant gram-negative bacilli. The failure to detect differences among the agents in the promotion of KPC3 may be related to the fact that these agents, including moxifloxacin, have much less effect on intestinal anaerobes than would be predicted based on the drug levels achieved (21). The minimal impact of the fluoroquinolone antibiotics on intestinal anaerobes has been attributed to high degrees of reversible binding of these agents to fecal matter, to reduced susceptibility of anaerobic bacteria to fluoroquinolones under strictly anaerobic conditions, and to an inoculum effect for inhibition of anaerobes (21). In this model, we did not find that the “short-course, high-dose” regimen of levofloxacin was associated with a reduced risk for promotion of fluoroquinolone-resistant gram-negative bacilli.

Our study has some limitations. First, although the levels of the study drugs in the stool of mice receiving equal dosages on a mg/kg basis were similar to stool concentrations measured in human volunteers (21), there are pharmacokinetic differences between mice and humans, and we cannot exclude the possibility that total antimicrobial exposure in mice differs significantly from exposure in humans. Second, a carbapenemase-producing K. pneumoniae strain was the only exogenously administered organism studied, and it is possible that other gram-negative organisms such as Pseudomonas aeruginosa and E. coli might yield different results. Although E. coli organisms are the fluoroquinolone-resistant gram-negative bacilli most frequently isolated among outpatients (12, 14), recent reports suggest that fluoroquinolone-resistant K. pneumoniae isolates are common in many long-term care facilities and hospitals and are often associated with multidrug-resistant phenotypes (17, 22). Third, the KPC3 test strain had high-level resistance to fluoroquinolones (ciprofloxacin MIC, 256 μg/ml) and was not inhibited by the fluoroquinolone levels achieved in the intestinal tract. Inhibition might occur if strains with lower levels of fluoroquinolone resistance are ingested. Finally, as noted previously, the mice received a single ingestion of KPC3, and we cannot exclude the possibility that repeated exposure might result in the increased likelihood of establishment of colonization.