|Home | About | Journals | Submit | Contact Us | Français|
Antibiotic treatment of lung infections may lead to the emergence of resistance in the gut flora. Appropriate dosing regimens could mitigate this adverse effect. In gnotobiotic rats harboring intestinal Escherichia coli and Enterococcus faecium populations, a lung infection by Klebsiella pneumoniae was instigated with two different sizes of inoculum to represent an early or a late initiation of antibiotic treatment. The rats were treated with marbofloxacin, an expanded-spectrum fluoroquinolone, by a single-shot administration or a fractionated regimen over 4 days. Intestinal bacterial populations were monitored during and after treatment. At the infection site, bacterial cure without any selection of resistance was observed. Whatever the dosage regimen, fluoroquinolone treatment had a transient negative impact on the E. coli gut population but not on that of E. faecium. The intestinal flora was colonized by the pathogenic lung bacteria, and there was the emergence of intestine-resistant K. pneumoniae, occurring more often in animals treated with a single marbofloxacin dose than with the fractionated dose. Bacterial cure without resistance selection at the infection site with fluoroquinolone treatment can be linked to colonization of the digestive tract by targeted pulmonary bacteria, followed by the emergence of resistance.
The emergence of antimicrobial resistance during antibiotic treatment can occur in the infected organ system and/or in the endogenous normal gut flora (3). Antimicrobial agents, including fluoroquinolones, can be extensively excreted into the intestinal tract, exposing the normal host flora to antimicrobial selective pressure (i.e., inhibition of competing microflora). This may lead to a secondary development of antibiotic-resistant gut organisms (2, 3, 17). Klebsiella pneumoniae is an important opportunistic pathogen implicated in nosocomial bacterial infections (19). Epidemiological studies have shown that the majority of K. pneumoniae infections are often preceded by colonization of the patient's gastrointestinal tract by the bacteria (9, 10). Recent reports suggest that fluoroquinolone-resistant Klebsiella pneumoniae isolates are common in many long-term care facilities and hospitals and are often associated with multidrug-resistant phenotypes (11, 22). The origins of this resistant K. pneumoniae gut subpopulation deserve attention. A possible factor contributing to the emergence of a resistant subpopulation of K. pneumoniae in the gut could be inadequate treatment of a prior K. pneumoniae infection, with secondary gut colonization by K. pneumoniae. This K. pneumoniae strain may then expand in the gut flora due to the selective pressure of the antibiotic reaching the gut lumen. Factors such as the concentration of the antimicrobial in the intestinal tract, the duration of the antimicrobial therapy, and the associated degree of disruption of the microflora may influence the likelihood that K. pneumoniae-resistant strains will or will not emerge at the gut level (3, 21). In previous studies on rodent models of Escherichia coli thigh infection and K. pneumoniae lung infection, we showed that the bacterial load at the start of antimicrobial treatment plays an important role in the enrichment of resistant strains at the infection site (4, 8). With a low inoculum, an early start of antimicrobial treatment with marbofloxacin, an expanded-spectrum fluoroquinolone extensively used in veterinary medicine, prevented mutant enrichment at the infection site, whereas a late start of the antimicrobial treatment with a high inoculum led to the enrichment of the resistant mutant subpopulation. Moreover, we showed that the emergence of resistance was dependent on the total marbofloxacin dose and dosage regimen.
The aim of the present study was to assess, in a K. pneumoniae experimental infection model, the impact of different marbofloxacin dosage regimens on the commensal intestinal flora and to test the hypothesis that the critical site of emergence of resistance of a targeted lung pathogen during antibiotic treatment may be not the lung itself but the gut flora. With this as our aim, we developed a model of lung infection in gnotobiotic rats, with two inoculum sizes of K. pneumoniae, each treated with two different marbofloxacin dosage regimens. We chose to work with a gnotobiotic model harboring a Gram-positive and a Gram-negative bacterial population in the intestine.
The Escherichia coli and Enterococcus faecium strains used for the establishment of dixenic gut flora in rats came from pig samples from French slaughterhouses (AFSSA, Fougères, France), and K. pneumoniae ATCC 43 816 was used for the establishment of the lung infection.
Marbofloxacin powder was kindly provided by Vetoquinol, Lure, France.
Germfree male OFA rats (Charles River, L'Arbresle, France), at weights of 170 to 200 g, were housed individually in different sterile isolators, with a 12-h light/12-h dark cycle. Rats were fed ad libitum with an irradiated rodent chow (R03-40; UAR, Villemoisson, France) and were supplied with sterilized distilled water.
The germfree status of the rats was checked immediately after their reception and during the period of acclimatization (see “Bacteriological procedures”). After about 1 week, the rats were inoculated intragastrically with 1 ml of a saline (0.9% NaCl) suspension of a E. coli (109 CFU/ml) strain and 1 ml of a saline (0.9% NaCl) suspension of E. faecium (109 CFU/ml).
The experimental lung infection was produced as previously described (1, 8). Briefly, the trachea was cannulated, and the lungs were inoculated with 0.05 ml of a saline suspension (0.9% NaCl) of K. pneumoniae containing 2 × 106 CFU/ml (105 CFU total; group A) or 2 × 109 CFU/ml (108 CFU total; group B).
All animal procedures were conducted in accordance with accepted standards of animal care under agreement A 31909 for animal experimentation from the French Ministry of Agriculture.
Subcutaneous marbofloxacin treatment (Marbocyl; Vetoquinol, Lure, France) was started 4 h (group A) or 24 h (group B) after the lung infection. There were two modalities of treatment, as follows: doses were administered in one single administration (groups A1 and B1, 10 and 7 animals tested, respectively) or the same total dose was fractionated into 4 daily administrations over 4 days (groups A2 and B2, 8 and 7 animals tested, respectively). The total marbofloxacin dose was 16 mg/kg for group A and 64 mg/kg for group B. Stool samples were collected at days 0, 4, and 7 after the first marbofloxacin administration for bacterial analyses. The animals were sacrificed 7 days after the first marbofloxacin administration by an intraperitoneal injection of sodium pentobarbital (Dolethal; Vetoquinol, France). The lungs were aseptically removed and homogenized in 10 ml of 0.9% NaCl before bacteriological analysis.
MICs were determined in triplicate for the bacteria by using a broth micro dilution method, according to CLSI reference methods (2a).
Stool samples were diluted 10-fold in distilled water and homogenized. The germfree status was verified on Schaedler agar supplemented with sheep blood, brain heart infusion (BHI) agar supplemented with sheep blood, and malt extract agar. During the study, 100 μl of fecal homogenates collected at days 0, 4, and 7 after the start of marbofloxacin treatment was plated on Slanetz and Bartley medium for the E. faecium strain, on MacConkey agar for the E. coli strain, and on MacConkey agar supplemented with 0.3 μg/ml of marbofloxacin for resistant E. coli, and colonies were counted after 24 h of incubation at 37°C for E. coli and 48 h for E. faecium. The lowest level of detection was 100 CFU/g feces.
Lung homogenates were plated on drug-free Mueller Hinton agar plates containing 10% activated charcoal and 10% MgSO4 to enumerate total bacterial counts of K. pneumoniae and on MacConkey agar supplemented with 0.3 μg/ml marbofloxacin to enumerate bacterial counts of resistant K. pneumoniae. Colonies were counted after overnight incubation at 37°C. The lowest level of detection was 100 CFU/lung, and bacteria were considered eradicated below this level.
Two satellite groups (groups C and D) of conventional male OFA rats (Charles River, L'Arbresle, France), at weights of 250 to 270 g, were inoculated with 0.05 ml of an inoculum of 2 × 106 CFU/ml (group C) or 2 × 1010 CFU/ml (group D) of K. pneumoniae. Four hours after the inoculation, group C was given 4 or 16 mg/kg of marbofloxacin by subcutaneous administration. For group D, subcutaneous marbofloxacin administration was 24 h after the K. pneumoniae inoculation and was 4, 16, or 100 mg/kg. The total amount of excreted feces was collected over 48 h. Two to four rats were included per treatment group. Stool samples were stored at −20°C until assayed for marbofloxacin by a high-performance liquid chromatography method with fluorescence detection (λexc [wavelength excitation] = 295 nm; λem [wavelength emission] = 500 nm) (Agilent 1100), adapted from Schneider et al. (18). Briefly, for each sample time, feces were pooled, mixed with 2.5% trichloroacetic acid, and centrifuged. The supernatant was added to 1 ml of dichloromethane and mixed for 10 s. A total of 0.2 ml of a mixture of MeOH (2% HCl)-H2O (90:10 dilution) was added to the organic layer, and 100 μl of the supernatant was injected into a C18e column (5 μm 125- by 4-mm Lichrospher; Merck) and eluted with a phosphoric acid (0.01 M)-triethylamine (0.004 M) (pH 2)-acetonitrile gradient. The calibration curve of marbofloxacin was established over the concentration range from 550 to 5,000 ng/ml using a linear regression model. The accuracy varied from 94.8 to 113.87%, and the intraday and interday precision was lower than 10.51% and 7.7%, respectively. The limit of quantification was 550 ng/ml. The samples were diluted to ensure that the concentrations fell within the range of the calibration curve.
The MIC of marbofloxacin was 0.032 μg/ml for both K. pneumoniae and E. coli and 2 μg/ml for E. faecium.
Bacterial infections have been shown to alter the pharmacokinetics of drugs (5), including fluoroquinolones (7, 13). For this reason, we evaluated the amount of marbofloxacin excreted in the feces in our model of K. pneumoniae lung infection. We observed that the amount of marbofloxacin excreted in the feces of rats was proportional to the dose within each animal group (A or B) but differed between groups as follows: the percentage of the marbofloxacin dose excreted in feces of the animals infected with the high K. pneumoniae inoculum was 4-fold less than in feces of those infected with the low inoculum (mean of 5% versus 23%, respectively) (Table (Table1).1). Consequently, taking into account the fact that group B received a four-times-higher marbofloxacin dose than group A (64 versus 16 mg/kg), the amount of marbofloxacin excreted in the feces was approximately the same in the two groups.
The amount of K. pneumoniae in each rat's lung and the percentage of animals with resistant K. pneumoniae at the end of the experiment for the two initial inoculum sizes and the different dosing regimens are shown in Table Table2.2. All the rats infected with the low K. pneumoniae inoculum survived (group A), and 7 days after the start of the antimicrobial treatment, bacteria were not detected in the lungs of all the rats (except two), whatever the dosing regimen. For the rats infected with the high K. pneumoniae inoculum (group B), the survival rate differed slightly according to the dosage regimen and was higher for animals treated with the fractionated marbofloxacin regimen (group B2) than for animals treated with the one-shot dose (group B1). Nevertheless, the total bacterial population in the surviving animals at 7 days after the start of antimicrobial treatment was almost the same for the two dosing regimens.
More importantly, whatever the dosage regimen, there was no K. pneumoniae resistant to marbofloxacin at the end of the trial, neither in the lungs of animals infected with the low K. pneumoniae inoculum (groups A1 and A2) nor in those infected with the high inoculum (groups B1 and B2). Theses results differ from our previous results showing the emergence of K. pneumoniae resistant to 0.3 μg/ml marbofloxacin in the animals infected by a high inoculum and treated with 64 mg/kg of marbofloxacin (8). However, the present study was longer, and we observed rat mortality within the high-inoculum group (group B) during the 64 mg/kg marbofloxacin treatment, in contrast to the results of the previous study. This mortality was probably due to a higher sensitivity to the infection of gnotobiotic animals than conventional rats. The animals that died during the treatment were possibly carriers of resistant K. pneumoniae in the lungs (data not checked).
For all animals, whatever the dosage regimen, the E. faecium population remained unchanged during and after the marbofloxacin treatment (Fig. (Fig.11 A). Unfortunately, an evaluation of a resistant E. faecium population was not carried out due to an interaction between marbofloxacin and Slanetz and Bartley culture medium.
For the E. coli population, a decrease in the number of isolates was observed during the treatment in all groups, depending on both the marbofloxacin dose and the dosing regimen. This decrease was followed by an increase to pretreatment levels after termination of the therapy (Fig. (Fig.1B).1B). Seven days after the initiation of marbofloxacin treatment, a resistant E. coli subpopulation (MIC ranging from 0.256 to 2 μg/ml) was found in all treatment groups. The percentage of rats harboring resistant intestinal E. coli was between 10 and 17% for groups A1, A2, and B2, and only group B1 (high dose, single shot) showed a higher percentage at 50% (Table (Table3).3). The fact that the amount of marbofloxacin excreted in the feces was approximately the same in the two groups (groups A and B) could explain why the numbers of animals with resistant E. coli in their feces were rather similar for the different groups, whatever the marbofloxacin total dose and regimens (single or fractionated administration) (Table (Table3).3). Nevertheless, it was difficult to correlate the emergence of resistant E. coli with the marbofloxacin concentration in the feces since marbofloxacin could be highly bound to fecal matter (12).
However, the most important result of this study was the intestinal colonization by the targeted pathogenic bacteria, K. pneumoniae, present at the infection site. This K. pneumoniae colonization was established throughout the study, and 7 days after the start of antimicrobial treatment, this colonization was observed in the majority of surviving animals for the groups treated with a single administration (60% and 100% for the low- and high-dose groups, respectively) (Table (Table33 and Fig. Fig.1C).1C). Deglutition by the infected animals of K. pneumoniae moving out of the lungs to the mucus escalator was probably the cause of the colonization of the intestinal tract by K. pneumoniae.
More importantly, intestinal colonization by K. pneumoniae was accompanied by the emergence in the gut of a K. pneumoniae subpopulation resistant to the fluoroquinolone. Fecal K. pneumoniae resistant to 0.3 μg/ml marbofloxacin (MIC ranging from 0.512 to 2 μg/ml) was detected in 57% (8/14) of rats receiving a single dose (16 or 64 mg/kg), whatever the dose level (Table (Table3).3). For animals receiving the fractionated administration of marbofloxacin, only 1 out of 14 rats carried a resistant K. pneumoniae subpopulation in the gut (Table (Table3).3). It is noteworthy that the emergence of resistant K. pneumoniae in the intestinal tract occurred in animals harboring no resistant K. pneumoniae in their lungs. This intestinal colonization by resistant K. pneumoniae occurred in 54% of all surviving rats (15/28) and seemed to be slightly more frequent in groups infected with the high inoculum (8/10, i.e., 80%) (Table (Table3)3) than in groups infected with the low inoculum (7/18, i.e., 39%) (Table (Table3).3). The most likely reason is that the probability of colonizing the intestinal tract increases with the inoculum size in the lung (larger K. pneumoniae population and more mucus production).
Furthermore, whatever the size of K. pneumoniae pulmonary inoculum, we observed a higher percentage of intestinal colonization by resistant K. pneumoniae when the treatment was a single-dose administration of marbofloxacin rather that a fractionated administration of the same total dose (8/14, i.e., 57%, for single administration versus 1/14, i.e., 7%, for fractionated administration) (Table (Table3).3). In a previous study, we showed that the antibiotic exposure played an important role in the selection of resistant bacteria at the infection site (8), and in agreement with the present results, we also showed that the plasma antibiotic exposure with fractionated administration limited the enrichment of the resistant subpopulation compared to that with the single administration of the same total marbofloxacin dose.
However, the design of the present study did not enable us to document whether the emergence of the resistant K. pneumoniae subpopulation occurred in the digestive tract or in the lungs, but some clues suggest the emergence of resistance in the gut. Indeed, we observed resistant K. pneumoniae in the guts of animals harboring no resistant K. pneumoniae in their lungs. Moreover, for animals infected with the low inoculum and treated by the single administration of 16 mg/kg of marbofloxacin, the size of the K. pneumoniae population in the gut was higher than the population inoculated into the lungs (8.1 ± 1.1 log10 CFU/g of feces versus 5 log10 CFU/lung) (Table (Table3),3), increasing the likelihood of resistant mutants appearing in this intestinal bacterial population. Therefore, the scenario of the intestinal colonization by wild-type K. pneumoniae followed by the appearance in the gut of resistant K. pneumoniae mutants secondarily selected by marbofloxacin intestinal exposure seems more probable than the gut colonization by resistant K. pneumoniae of pulmonary origin.
We are not in a position to estimate the exact antibiotic exposure in the gut, but it is likely that the dosing regimens that are optimized to achieve clinical success while minimizing the emergence of resistance at the lung infection site may actually be inappropriate in terms of gut colonization and intestine-located emergence of resistance.
For the sake of simplicity, the present study was carried out with an artificial intestinal ecosystem in diassociated gnotobiotic rats, i.e., a system lacking anaerobes that can play an important role in colonization resistance (20). However, this animal model appears to be relevant to the study of intestinal colonization by resistant bacteria given that anaerobic flora could be decreased in patients by some antimicrobial treatments. Therefore, such a model represents an alternative to classical models of intestinal colonization by resistant pathogenic bacteria, where anaerobic flora are eradicated by clindamycin treatment before colonization challenge (6, 15). To our knowledge, this study is the first using a model of natural gut colonization by pathogenic bacteria from a nonintestinal infection site and showing the emergence of resistant strains in this new intestinal population during fluoroquinolone treatment of a lung infection. Indeed, usually studies on intestinal colonization use orogastric gavage of bacteria already resistant to the antibiotic (6, 14).
In conclusion, the results of the present study highlight that a fluoroquinolone treatment leading to the bacterial cure and prevention of resistance emerging at the infection site might at the same time lead to the emergence of resistant pathogenic bacteria in the digestive tract secondary to intestinal colonization by wild-type pathogens. This could generate a reservoir of resistant pathogens for secondary infections (16, 23). In addition, it was shown that such an event is influenced, and thus might be manageable, by the dosage regimen. This is a difficult challenge because antibiotic exposure to prevent the emergence of resistance in a pathogen not only should apply to the target site of infection but also should be extended to the gut flora when colonization occurs.
We thank Jacqueline Manceau for doing analytical assays and Annick Brault, Mireille Bruneau, Pamela Louâpre, Stephane Marteau, Catherine Poirier, and Jean-Guy Rolland for technical support.
We received grants from the Institut National de la Recherche Agronomique (INRA) and from the Agence Française de Sécurité Sanitaire des Aliments (AFSSA). There are no potential conflicts of interest.
Published ahead of print on 10 May 2010.