Correlation between Fecal Concentrations of Ciprofloxacin and Fecal Counts of Resistant Enterobacteriaceae in Piglets Treated with Ciprofloxacin: toward New Means To Control the Spread of Resistance?

TEXT

Increasing resistance to fluoroquinolones in Gram-negative bacilli is of major concern, because it compromises their therapeutic use (6, 7). Colonic flora is the reservoir of many of the Enterobacteriaceae species with clinical significance (2) which are exposed to antibiotic residues during fluoroquinolone treatments (9). It has been shown that administration of enrofloxacin, a fluoroquinolone used in animals, could promote the emergence of quinolone-resistant enterobacteria in fecal flora of pigs (12). We also found that administration of oral ciprofloxacin is associated with the emergence of quinolone-resistant strains of Enterobacteriaceae in fecal flora from human volunteers (3). However, rates of emergence of resistance were not significantly different even though the volunteers were exposed to different antibiotic dosages. Of note, detection of resistance was only qualitative in that study, i.e., absence versus presence of detectable resistant bacteria, and not quantitative, expressing the densities of resistant bacteria in the feces. Indeed, several lines of evidence suggest that quantitative aspects of emergence of resistance should also be analyzed. For instance, in neutropenic patients, Gram-negative bacillus bacteremias are caused by the predominant fecal clone of this type (11). Also, the density of a given Escherichia coli strain in a subject's fecal flora can influence to some extent its ability to cause urinary tract infection (8). In order to explore further the dynamics of fluoroquinolone resistance in the fecal flora during treatments, we assessed here in a prospective randomized study in piglets the relationship between fecal concentrations of ciprofloxacin and amounts of excreted ciprofloxacin-resistant Enterobacteriaceae.

Twenty-nine piglets from a single farm were included 4 weeks after birth. They were born from sows that were treated at the time of parturition with 2 g oxytetracycline given intramuscularly but had not received directly any antimicrobials. Piglets were housed in individual boxes for 21 days before the start of treatment (day 1 [D1]) and were randomly assigned to oral treatment with placebo (9 piglets), 1.5 mg ciprofloxacin/kg of body weight/day (10 piglets), or 15 mg ciprofloxacin/kg/day (10 piglets) from D1 to D5. Ciprofloxacin suspension (Ciflox; Bayer) or mineral water (placebo group) was administered once a day at least 4 h before food to animals fasted for at least 12 h. The protocol was approved by the local ethical committee.

Fecal samples were recovered from piglets after anal stimulation on mornings before treatment (at D−1 and D1), during treatment (at D3 and D5, before ciprofloxacin administration), and after treatment (at D7 and D9). Samples were stored frozen at −80°C until microbiological analysis and ciprofloxacin assay were performed blinded of the treatment group. Fecal concentrations of ciprofloxacin were measured by microbiological assay (5) after 10-fold dilution in HCl 0.1 N and then further dilutions in phosphate buffer, pH 8.0. Counts of total and resistant Enterobacteriaceae were obtained after plating serial dilutions of fecal samples on Drigalski agar (Bio-Rad Laboratories, Marnes-la-Coquette, France) supplemented or not with 20 mg/liter of nalidixic acid or with 2 mg/liter of ciprofloxacin. However, since there was no significant differences in the counts obtained on the two types of media (data not shown), only counts on ciprofloxacin containing agar were used for further statistical analysis. Individual area under the curve (AUC) values from D1 to D9 of ciprofloxacin fecal concentrations (AUC_CIP), of log counts (AUC_RES), and of percentages (AUC_%RES) of ciprofloxacin-resistant Enterobacteriaceae above pretreatment baseline values were computed by the trapezoidal approach. These criteria were compared across groups by nonparametric analysis of variance (ANOVA) and pairwise Wilcoxon tests if global tests were statistically significant (with the significance level at 0.05). All statistical analyses were performed using SAS 9.1 (SAS, Cary, NC).

Concentrations of ciprofloxacin increased with the dose administered and peaked at 84.8 ± 57.9 versus 11.6 ± 12.6 μg/g of feces at D5 in piglets treated with 15 and 1.5 mg/kg/day of ciprofloxacin, respectively (P < 0.001) (Fig. 1A). Fecal antibiotic activity was detectable to some extent in all animals 2 days after cessation of treatment (D7), but no activity was detected in any animal at D9. AUC_CIP was significantly higher for the group treated with 15 mg/kg/day of ciprofloxacin than that for the group treated with 1.5 mg/kg/day (P < 0.0005) (Table 1).

• Open in new tab
• Download powerpoint

Fig 1

Mean (and 95% confidence intervals) of ciprofloxacin concentrations (A) and of log counts of ciprofloxacin-resistant Enterobacteriaceae (B) as well as of percentages of Enterobacteriaceae resistant to ciprofloxacin (C) in fecal samples from piglets treated with placebo (n = 9) and 1.5 mg/kg/day (n = 10) and 15 mg/kg/day (n = 10) of oral ciprofloxacin from day 1 to day 9.

Most of the piglets (25/29, 86%) had ciprofloxacin-resistant Enterobacteriaceae detected in fecal samples before treatment. Administration of ciprofloxacin promoted the increase of both absolute counts (Fig. 1B) and percentages of ciprofloxacin-resistant enterobacteria (Fig. 1C) in the 2 treatment groups compared with the placebo group. AUC_RES and AUC_%RES were significantly different between the groups treated with 1.5 mg/kg/day and 15 mg/kg/day and the placebo group (both P < 0.01 for AUC_RES and both P < 0.0005 for AUC_%RES). AUC_RES and AUC_%RES were also significantly higher for the group treated with 15 mg/kg/day than that for the group treated with 1.5 mg/kg/day (P < 0.05 and P < 0.001, respectively) (Table 1).

There were significant correlations between AUC_CIP and AUC_RES as well as between AUC_CIP and AUC_%RES (both P < 0.0001, Spearman tests). These relationships were adequately described using an E_max model AUC_RES = AUC_RES₀ + (AUC_RES_max × AUC_CIP)/(AUC_CIP₅₀ + AUC_CIP), where AUC_RES₀, AUC_RES_max, and AUC_CIP₅₀ are, respectively, the baseline, the maximal effect, and the value of AUC_CIP to obtain 50% of the maximal effect (Fig. 2).

• Open in new tab
• Download powerpoint

Fig 2

Relationship between individual AUC of fecal concentrations of ciprofloxacin (AUC_CIP) versus AUC of log counts of fecal ciprofloxacin-resistant Enterobacteriaceae (AUC_RES) (A) and versus AUC of percentages of ciprofloxacin-resistant Enterobacteriaceae (AUC_%RES) (B) for 29 piglets in 3 treatment groups. Each symbol represents the AUC_CIP and AUC_RES or AUC_%RES values for one piglet. The black curve among the symbols is the mean curve predicted by the E_max model.

We showed a significant positive correlation between the AUC of fecal ciprofloxacin concentrations and the AUC of log ciprofloxacin-resistant Enterobacteriaceae counts from D1 of treatment to D9 posttreatment. Rapid increase of ciprofloxacin-resistant enterobacteria from the beginning of ciprofloxacin administration probably resulted from the selection of resistant bacteria which were already present in low counts before treatment in the fecal flora of most (86%) of our piglets. Indeed, in a previous study in human volunteers (3), only 6 of 48 (12.5%) had quinolone-resistant enterobacteria detected before treatment, and the increased prevalence of resistance was observed only after the end of the treatment without significant dose relationship. The piglet study described here was much more fitted to evidence the pharmacokinetic/pharmacodynamic relationship since the doses varied by a factor of 10, starting from a very low dose to a clinical dose (from 1.5 to 15 mg/kg/day), whereas they varied only by a factor of 3 (250 mg twice/day to 750 mg twice/day) in the volunteers. Median ciprofloxacin fecal concentrations at steady state were also much lower in piglet than in human volunteer groups of treatment (7 and 73 versus 845 and 1,938 μg/g of feces, respectively). An increase in the percentage of ciprofloxacin-resistant Bacteroides fragilis strains in feces of human flora-associated mice treated with various doses of ciprofloxacin has been also reported as dose related (10). However, our results differ in the absence of correlation between antibiotic dosage and the percentages of fecal quinolone-resistant enterobacteria during treatment that was observed by others in pigs receiving several doses of enrofloxacin, varying in a range of 1 to 6 (12). The dynamic of emergence of resistant bacteria is dependent on several parameters, among which are the concentrations of free and active antibiotic reaching the colonic flora, MICs of bacterial populations, and barrier effects exerted in the colonic ecosystem. This may account for the nonlinear correlation between fecal antibiotic concentrations and the level of resistant bacteria we observed and, also, for differences in results from similar studies.

We used AUCs to characterize the fecal pharmacokinetic/pharmacodynamic relationship for ciprofloxacin, taking into consideration the whole time course of concentrations and bacterial counts. We believe that AUCs of fecal counts are the most relevant endpoints, because they describe, better than single time point values, the total amount of resistant bacteria excreted, which is actually what one wants to decrease to control the spread of resistance.

Indeed, the link we showed between intestinal concentration of ciprofloxacin and the amount of ciprofloxacin-resistant enterobacteria excreted by the animals may be of utmost importance in the current context of the spread of bacterial resistance. It has been shown in a rat model that the administration of a compound consisting of activated charcoal entrapped within zinc-pectinate beads could remove residual colonic ciprofloxacin (4). Thus, if such a removal could be obtained in animals or humans receiving therapeutic doses of quinolones without impairing the pharmacokinetics of the drug in plasma, this would open new avenues within the scope of the recently signaled eco-evo drugs (1) to help control the emergence and spread of quinolone resistance in gut-originating Gram-negative bacteria.