Ciprofloxacin Dosage and Emergence of Resistance in Human Commensal Bacteria

Abstract

Background

Although optimization of the fluoroquinolone dosage increases the efficacy of this class of drugs against bacterial infections, its impact on the emergence of resistance in commensal bacteria is unknown.

Methods

Six different 14-day dosages of oral ciprofloxacin were randomly assigned to 48 healthy volunteers. Individual pharmacokinetic and pharmacodynamic parameters combining antibiotic exposure in plasma, saliva, and stool specimens and ciprofloxacin minimum inhibitory concentrations (MICs) and mutant prevention concentrations against viridans group streptococci in the pharyngeal flora and Escherichia coli in the fecal flora were estimated. Their links with the emergence of resistance to nalidixic acid or ciprofloxacin in the fecal flora and to levofloxacin in the pharyngeal flora 7, 14, or 42 days after ciprofloxacin initiation were investigated.

Results

Resistance emerged in the fecal and pharyngeal flora of 25% and 33% of the subjects, respectively, mainly when local concentrations of ciprofloxacin were less than the MIC. No variable that integrated pharmacokinetic data and pharmacodynamic parameters was found to differ significantly between the subjects in whom resistance emerged and those in whom it did not. Probabilities of the emergence of resistance were not significantly different across the different antibiotic dosages.

Conclusions

Selection of resistant commensals during ciprofloxacin therapy is a frequent ecological side effect that is not preventable by dosage optimization.

Trial registration

Clinical Trials.gov identifier: NCT00190151.

Bacterial resistance to antibiotics is an increasing therapeutic problem, both in the community and the hospital, and involves all antibiotics, including fluoroquinolones [1]. Decreased susceptibility to fluoroquinolones arises mainly by single-step mutations in gyrA and parC, which encode the fluoroquinolones targets, the topoisomerase enzymes [2], conferring cross-resistance to all fluoroquinolones [3]. Accumulation of multiple mutations in several genes confers an increasing level of resistance associated with clinical failure [4, 5]. How-ever, even low-level resistance can generate therapeutic failure [6].

Resistance to fluoroquinolones can result from direct selection at the site of infection or from selection in commensal bacteria followed by horizontal gene transfer to pathogens [7], which might even be more frequent because there are many more bacteria among commensal flora than within any infectious focus [8]. In addition, differences in local antibiotic concentrations, as compared with the plasma concentration, can affect selection of resistant bacteria in different sites [9, 10]. Also, resistant commensal bacteria may be selected after any antibiotic treatment, whereas resistant pathogens can emerge only in infected patients [11].

Pharmacokinetic and pharmacodynamic parameters give rationale to antibacterial dosing [12, 13]. The minimum inhibitory concentration (MIC) and, more recently, the mutant prevention concentration (MPC) are parameters used to investigate the relationship between antibiotic exposure and efficacy or prevention of resistance development [14]. The ratio of the area under the concentration-time curve (AUC) to the MIC and the ratio of the peak concentration to the MIC for the infecting bacteria have been linked to treatment success with fluoroquinolones [15–18] but not to emergence of resistant pathogens, probably because the inocula at the foci of infection are limited and emergence of resistance is uncommon [15,19].

To investigate whether optimization of dosages might prevent the emergence of resistance in commensal flora, we studied the relationship between antibiotic pharmacokinetic and pharmacodynamic parameters in plasma, saliva, and stool specimens and the emergence of resistant strains in the pharyngeal and fecal flora of healthy volunteers who received various dosages of ciprofloxacin, the reference fluoroquinolone.

Subjects, Materials, and Methods

Enrollment Criteria and Treatment Groups

Subjects

Healthy volunteers aged 18–45 years were selected on the basis of normal findings of a thorough general examination (interview and physical examination), normal intestinal transit (1 formed stool per day), normal electrocardiography findings with a QTc interval of < 450 ms, and normal findings of a biological work-up (blood count, blood biochemistry, liver tests, urinalysis, and serological tests for hepatitis viruses and human immunodeficiency virus) [20]. Women of child-bearing age were included if they were using effective contraception and had a negative result of a preinclusion pregnancy test. The volunteers had no known allergy to a fluoroquinolone and had taken no antibacterial or antifungal drug, theophylline, steroids, vitamin K antagonists, or barrier antacids during the 3 months before inclusion. Subjects who were unlikely to adhere to the study protocol, had donated blood (quantity, >500 mL), or had participated in another study during the previous 3 months were not included. Caffeine consumption was limited and stable during the 2-week treatment period. Volunteers were advised to avoid exposure to sunlight throughout the treatment period. The study design was approved by the local ethics committee of Paris Bichat-Cergy Pontoise. All participants gave written informed consent before entering the study.

Regimens

Volunteers were randomly assigned into 6 groups of 8 individuals each, receiving either 250 mg of oral ciprofloxacin every 12 h, 500 mg every 24 h, 500 mg every 12 h, 750 mg every 24 h, 750 mg every 12 h, or 1000 mg every 24 h for a total of 14 days. These regimens were chosen to generate pharmacokinetic variability across a range of clinically relevant total daily doses. Each intake was observed and its time recorded.

Microbiologic Follow-up Study

Samples

Fecal and pharyngeal samples were collected before initiation of treatment and on days 7, 14, and 42; stored at −80°C; and blinded until analysis [20]. The analysis focused on viridans group streptococci (VGS) in the pharyngeal flora and Escherichia coli in the fecal flora for the following reasons: these bacterial species are present in all subjects, they are involved in various clinical infections (bacteremia, endocarditis, and urinary tract infections), and they are recognized sources of horizontal gene transfer within the commensal flora. For each target species, we determined the susceptibility to quinolones in the global population (dominant flora) and the emergence of quinolone-resistant subpopulations (subdominant flora).

Detection of ciprofloxacin susceptibility in the dominant flora

We used a procedure specifically designed to estimate susceptibility of the dominant flora as a whole to fluoroquinolones. Saliva samples were inoculated on colimycin-nalidixic acid blood agar (Biomérieux) and fecal samples on Drigalski agar. After growth, we isolated 10 separate colonies from each plate, identified as VGS and E. coli, using standard techniques, to obtain a representative sample of the dominant flora. These 10 colonies were mixed, and susceptibility to fluoroquinolones was tested for the mixture as described elsewhere, using MICs in duplicate by the agar dilution method [21] and MPCs in triplicate [14, 22]. Geometric means of these replicates were used in the analysis.

Detection of resistance in the subdominant flora

Resistance of each target bacterial species in the subdominant flora was detected by plating the fecal samples on Drigalski agar with 16 mg/L of nalidixic acid, to detect a first-step E. coli mutant, or with 1 mg/L of ciprofloxacin, to detect mutants resistant to ciprofloxacin. Pharyngeal samples were plated on colimycin-nalidixic acid blood agar with 2 mg/L of levofloxacin because VGS are naturally resistant to ciprofloxacin but susceptible to levofloxacin [23]. MICs of the colonies growing on selective media were determined by the agar dilution method [21].

End points

Resistance to nalidixic acid and ciprofloxacin among E. coli from the fecal flora and resistance to levofloxacin among VGS from the pharyngeal flora were determined according to Clinical and Laboratory Standards Institute break points [24]. Emergence of resistance was defined by the detection of resistant strains at day 7, 14, or 42 in subjects in whom only susceptible strains were detected and resistant strains were not detected before treatment.

Pharmacokinetic Follow-up Study

Serum and saliva samples were taken from each volunteer before and 1, 3, 6, and 12 h after receipt of the first ciprofloxacin dose; at trough on days 8 and 14; and again 1, 3, 6, and 12 h after the last dose [10]. Stool samples were collected on days 0, 7, 14, and 42. All samples were blinded and stored at −80°C until analysis. Ciprofloxacin concentrations were determined by liquid chromatography with fluorimetric detection after deproteinization or stool extraction in acidic medium, as described elsewhere [25].

Statistical Analysis

A population pharmacokinetic analysis with, as previously described [26–28], a 1-compartment model with first-order absorption was used to analyze plasma and saliva concentrations and estimate the maximal concentration (peak) and AUC from 0 to 24 h at steady state for each volunteer, taking into account the dosing schedule. AUC/MIC, AUC/MPC, peak/MIC, peak/MPC, AUC above the MIC, AUC above MPC, AUC between the MIC and MPC, time above MIC, time above MPC, and time between the MIC and MPC were determined using individual MICs and MPCs of ciprofloxacin. In feces, concentration/MIC and concentration/MPC ratios were determined using the average of the concentrations measured at days 7 and 14.

Confidence intervals (CIs) for the percentages of subjects in whom resistance emerged among bacteria were estimated using the binomial distribution. For each target flora, the volunteers were divided into 2 groups (regardless of the dosage) according to the emergence or nonemergence of resistance. Variables integrating pharmacokinetic and pharmacodynamic parameters were compared between the 2 groups, using the Mann-Whitney nonparametric test. Logistic regression analysis was also performed for each flora to model the link between the probability of emergence of resistance and the logarithm of AUC/MIC.

The number of subjects was determined on the basis of the assumption that quinolone resistance would emerge in the fecal flora in one-third of the volunteers [29–31]. With this proportion and the mean AUC/MIC ratios reported previously to be significantly associated with respect to emergence of resistance at the site of infection [32], the power of this study was 90%, with a type I error of 5%.

Results

Subjects

Eighty subjects were screened, and 48 (28 men and 20 women) who fulfilled the predefined criteria for healthy subjects were selected. The median age was 28.9 years (range, 19.5–43.9 years), and the median weight was 65 kg overall (range, 47–93 kg), 71 kg (range, 53–93 kg) among men, and 55 kg (range, 47–85 kg) among women. One subject who developed tendinitis at day 8 while receiving 750 mg twice daily and stopped therapy was excluded from the microbiological and pharmacokinetic follow-up analysis. The remaining 47 subjects had no treatment-related adverse events.

Microbiologic study

Before treatment, the median MICs and MPCs of ciprofloxacin against the dominant flora were 0.016 mg/L (range, 0.005–0.5 mg/L) and 0.25 mg/L (range, 0.06–2.8 mg/L), respectively, in the fecal flora and 2 mg/L (range, 0.71–45.25 mg/L) and 22.6 mg/L (range, 8–512 mg/L), respectively, in the pharyngeal flora (figure 1). In the fecal flora, 1 subject had no detectable E. coli, whereas 6 (13%) initially had strains resistant to nalidixic acid, including 1 who had strain with resistance to both ciprofloxacin (MIC, 32 mg/L) and nalidixic acid (MIC, >1024 mg/L) and 5 who had strains with resistance to nalidixic acid only (MICs, 64 to >1024 mg/ L; ciprofloxacin MICs, 0.01–0.5 mg/L). All of the strains resistant to ciprofloxacin were also resistant to nalidixic acid. Before treatment, one subject had VGS strains in the pharyngeal flora that were resistant to levofloxacin (MIC, 16 mg/L) and another had no available pharyngeal sample.

Figure 1

Number of subjects harboring strains of Escherichia coli in fecal flora that were resistant to nalidixic acid (Nal) or ciprofloxacin (Cip) (top left) or strains of viridans group streptococci in pharyngeal flora that were resistant to levofloxacin (bottom left) and their corresponding minimum inhibitory concentrations (MICs) for Nal (white circles) and Cip (black circles) (top right) and levofloxacin (bottom right) in healthy subjects receiving various regimens of Cip for 14 days. Strains that were resistant to Cip also appear among strains resistant to Nal. There were 48 subjects at the start of therapy and 47 subjects later, because 1 subject discontinued therapy. In addition, 1 patient did not have a pharyngeal sample at day 0.

During treatment, susceptible E. coli were not detected in the flora of any stool specimens recovered from volunteers. Resistance to ciprofloxacin was detected in 3 subjects (6%) at days 7 and 14. By contrast, at day 42, 14 subjects (30%) were colonized by strains resistant to nalidixic acid (MICs, 64 to >1024 mg/L), including 4 with strains that were resistance to ciprofloxacin (MICs, 32–64 mg/L). Overall, resistance to nalidixic acid or ciprofloxacin that was not initially detected before therapy emerged in 10 of 40 subjects (25% [95% CI, 13%–40%]) during or after therapy.

In the pharyngeal flora, VGS resistant to levofloxacin was detected in 8 subjects (17%) on day 7, in 10 (21%) on day 14, and in 4 (10%) on day 42. MICs of levofloxacin against the resistant strains ranged between 4 and 64 mg/L. Overall, resistance to levofloxacin that was not initially detected before therapy emerged in 15 of 45 volunteers (33% [95% CI, 20%–49%]) during or after receipt of ciprofloxacin therapy.

Antibiotic concentrations and pharmacokinetic studies

The median peak concentrations of ciprofloxacin in plasma specimens at steady state were 1.35 mg/L (range, 1.13–1.77 mg/L) for subjects receiving 250 mg twice daily and 4.26 mg/L (range, 2.89–5.66 mg/L) for those receiving 1000 mg once daily; the median AUCs were 11.77 mg/L-h (range, 11.56–12.08 mg/L-h) for subjects receiving 250 mg twice daily to 36.27 mg/L-h (range, 33.23–45.49 mg/L-h) for those receiving 750 mg twice daily (figure 2). Thus, the median concentrations of ciprofloxacin in plasma were greater than the median MICs against the fecal dominant flora during most of the time for subjects receiving any of the regimens and greater than the median MPCs for subjects receiving any of the twice daily regimens. For the pharyngeal flora, the median ciprofloxacin concentrations in plasma were far less than the median MPCs and just greater than the median MICs at peak, except for subjects receiving 250 mg twice daily.

Figure 2

Mean concentration profiles of ciprofloxacin at steady state in plasma and saliva specimens from healthy subjects receiving ciprofloxacin for 14 days, according to dosage, and distribution of the concentration in fecal specimens on days 7 and 14. Horizontal lines represent median minimum inhibitory concentrations (MICs; solid lines) and mutant prevention concentrations (MPCs; dotted lines) of ciprofloxacin against the dominant florain each commensal flora.

The median concentrations of ciprofloxacin in stool specimens at steady state were 845 mg/L (range, 455–1265 mg/L) for subjects receiving 250 mg twice daily to 1938 mg/L (range, 1100–3235 mg/L) for those receiving 500 mg twice daily, far greater than the median MIC and MPC of ciprofloxacin against the fecal dominant flora. Ciprofloxacin was undetectable in stool specimens at day 42 for all subjects.

The median peak concentrations of ciprofloxacin in saliva specimens at steady state were 0.48 mg/L (range, 0.32–0.84 mg/L) for subjects receiving 250 mg twice daily and 1.79 mg/L (range, 1.05–2.35 mg/L) for those receiving 1000 mg once daily, never reaching the median MIC of ciprofloxacin against the pharyngeal dominant flora and far less than its median MPC. Ciprofloxacin was undetectable in saliva specimens at day 42 for all subjects.

Relationship between antibiotic exposure and emergence of resistance

The distribution of the ciprofloxacin AUC/MIC in plasma specimens, as well as the ciprofloxacin concentration/MIC in stool specimens, were not significantly different in subjects in whom resistance to nalidixic acid or ciprofloxacin did or did not emerge in fecal flora (table 1 and figure 3). Other variables integrating plasma pharmacokinetic and pharmacodynamic parameters of ciprofloxacin failed to show any significant difference (table 2). Logistic regression analysis did not evidence any significant link between the probability of emergence of resistance in fecal flora and antibiotic exposure, as measured by the ciprofloxacin AUC/MIC in plasma specimens or by the ciprofloxacin concentration/MIC in stool specimens (table 1).

Table 1

Observed Percentage of Subjects in Whom Ciprofloxacin Resistance Emerged in Escherichia coli from the Fecal Flora and in Viridans Group Streptococci from the Pharyngeal Flora Recovered from Healthy Subjects

			P
Parameter of ciprofloxacin exposure	Subjects at risk, no.	Subjects with resistant isolates, no. (%)	Group comparisona	Logistic regressionb
Fecal flora
AUC/MIC in plasma, h			.35	.36
361–721	10	2 (20)
722–1112	10	2 (20)
1113–2004	10	3 (30)
2005–4236	10	3 (30)
Overall	40	10 (25)
Concentration/MIC in stool			.93	.86
20×103–49×103	10	2 (20)
50×103–77×103	10	2 (30)
77×103–117×103	10	3 (30)
118×103–358×103	10	2 (20)
Overall	40	10 (25)
Pharyngeal flora
AUC/MIC in plasma, h			.86	.27
0.37–5.80	11	3 (27)
5.81–8.54	11	4 (36)
8.55–12.22	11	4 (36)
12.24–22.79	12	4 (33)
Overall	45	15 (33)
AUC/MIC in saliva, h			.45	.16
0.06 –0.87	11	4 (36)
0.88–1.47	11	4 (36)
1.48–2.21	11	4 (36)
2.22–6.82	12	3 (25)
Overall	45	15 (33)

NOTE. Parameter ranges are expressed as quartiles of the distribution. AUC, area under the concentration-time curve; MIC, minimum inhibitory concentration.

^a

Denotes comparison between subjects with and subjects without emergence of ciprofloxacin-resistant bacteria after exposure to ciprofloxacin.

^b

Denotes comparison between the probability of the emergence of ciprofloxacin-resistant bacteria after exposure to ciprofloxacin.

Figure 3

Distribution of the ratio of the ciprofloxacin area under the concentration-time curve (AUC) from 0 to 24 h to the minimum inhibitory concentration (MIC) in plasma (top and bottom left) and saliva (bottom right) specimens and of the fecal concentrations of ciprofloxacin/MIC (top right) in healthy subjects receiving various regimens of ciprofloxacin for 14 days and in whom no resistant strains were detected at day 0. Results are presented according to the emergence or nonemergence of Escherichia coli strains resistant to nalidixic acid or ciprofloxacin in fecal flora (top; n = 40) or of strains of viridans group streptococci that were resistant to levofloxacin in pharyngeal flora (bottom; n = 45) during days 7–42 after initiation of the ciprofloxacin regimen. The solid line represents the median value in each group. Each circle represents a subject.

Table 2

Comparison of Variables Integrating Ciprofloxacin Pharmacokinetic Parameters in Plasma or Concentrations in Stool and Pharmacodynamic Parameters of Ciprofloxacin against Escherichia coli from the Dominant Fecal Flora between Groups of Subjects with or without Emergence of Resistance in E. coli 7–42 days after Initiating a Ciprofloxacin Regimen

	Emergence of resistance, median value (range)
Variable, by specimen type	Yes (n = 10)	No (n = 30)	P
Plasma
AUC/MIC, h	1516.0 (524.8–3137.0)	1094.0 (360.5–4236.0)	.35
AUC/MPC, h	66.9 (22.0–375.9)	89.5 (32.2–374.2)	.41
Peak/MIC	186.9 (61.6–443.2)	148.4 (29.5–715.0)	.38
Peak/MPC	9.3 (3.9–46.3)	13.3 (3.8–42.2)	.40
AUC > MIC, mg/L-h	13.5 (11.0–34.3)	21.1 (10.3–45.2)	.38
AUC > MPC, mg/L-h	8.4 (5.5–31.1)	16.3 (5.4–39.6)	.18
MIC < AUC < MPC, mg/L-h	4.2 (1.7–8.0)	3.2 (0.4–7.1)	.57
Time > MIC, h	24.0 (22.7–24.0)	24.0 (13.3–24.0)	.15
Time > MPC, h	15.0 (7.4–24.0)	16.0 (7.8–24.0)	.64
MIC < Time < MPC, h	8.9 (0.0–16.6)	7.4 (0.0–14.1)	.16
Stool
Concentration/MIC	80 × 103 (22 × 103–358 × 103)	75 × 103 (20 × 103–340 × 103)	.93
Concentration/MPC	48 × 102 (8 × 102–223 × 102)	7 × 103 (1 × 103–46 × 103))	.30

NOTE. AUC, area under the concentration-time curve; MIC, minimum inhibitory concentration; MPC, mutant prevention concentration.

Similar results were obtained for pharyngeal flora. The distributions of the ciprofloxacin AUC/MIC in plasma and saliva specimens did not differ significantly between subjects in whom levofloxacin resistance emerged in the pharyngeal flora and subjects in whom it did not (table 1 and figure 3), nor were there significant differences in any other variables that integrated pharmacokinetic data in plasma or saliva with pharmacodynamic parameters of ciprofloxacin (table 3). The probability of emergence of resistance in pharyngeal flora was not significantly linked to antibiotic exposure, measured as the ciprofloxacin AUC/MIC in plasma or saliva specimens (table 1).

Table 3

Comparison of Variables Integrating Ciprofloxacin Pharmacokinetic Parameters in Plasma or Saliva and Pharmacodynamic Parameters of Ciprofloxacin against Viridans Group Streptococci (VGS) from the Dominant Pharyngeal Flora between Groups of Subjects with or without Emergence of Resistance in VGS 7–42 Days after Initiating a Ciprofloxacin Regimen

	Emergence of resistance, median value (range)
Variable, by specimen type	Yes (n = 15)	No (n =30)	P
Plasma
AUC/MIC, h	9.47 (0.37–12.79)	8.47 (2.95–22.79)	0.86
AUC/MPC, h	0.76 (0.10–2.98)	0.55 (0.10–2.01)	0.53
Peak/MIC	1.45 (0.04–3.90)	1.29 (0.34–4.00)	0.97
Peak/MPC	0.10 (0.01–0.46)	0.09 (0.01–0.80)	0.58
AUC > MIC, mg/L-h	2.30 (0.00–8.46)	0.71 (0.00–12.52)	0.78
AUC > MPC, mg/L-h	0.00 (0.00–0.00)	0.00 (0.00–0.00)	NA
MIC < AUC < MPC, mg/L-h	2.30 (0.00–8.46)	0.71 (0.00–12.52)	0.78
Time > MIC, h	3.96 (0.00–4.82)	2.22 (0.00–9.26)	0.72
Time > MPC, h	0.00 (0.00–0.00)	0.00 (0.00–0.00)	NA
MIC < Time < MPC, h	3.96 (0.00–4.82)	2.22 (0.00–9.26)	0.72
Saliva
AUC/MIC, h	1.46 (0.063–3.01)	1.57 (0.53–6.82)	0.45
AUC/MPC, h	0.11 (0.01–0.53)	0.09 (0.01–0.60)	0.77
Peak/MIC	0.39 (0.02–1.12)	0.46 (0.14–1.66)	0.71
Peak/MPC	0.03 (0.00–0.23)	0.03 (0.00–0.10)	0.73
AUC > MIC, mg/L-h	0.00 (0.00–0.06)	0.00 (0.00–0.42)	0.70
AUC > MPC, mg/L-h	0.00 (0.00–0.00)	0.00 (0.00–0.00)	NA
MIC < AUC < MPC, mg/L-h	0.00 (0.00–0.06)	0.00 (0.00–0.42)	0.70
Time > MIC, h	0.00 (0.00–0.41)	0.00 (0.00–0.94)	0.68
Time > MPC, h	0.00 (0.00–0.00)	0.00 (0.00–0.00)	NA
MIC < Time < MPC, h	0.00 (0.00–0.41)	0.00 (0.00–0.94)	0.68

NOTE. AUC, area under the concentration-time curve; MIC, minimum inhibitory concentration; MPC, mutant prevention concentration; NA, not applicable.

Discussion

We found a high rate of emergence of resistance to quinolones in commensal flora from healthy volunteers during and after a 14-day course of oral ciprofloxacin, both in fecal E. coli strains and pharyngeal VGS strains. Because this was observed in healthy volunteers without recent antibiotic exposure, these rates are probably the minimal to be expected in patients. Indeed, higher rates of ciprofloxacin resistance, ranging from 32% to 40%, have already been reported in fecal Enterobacteriaceae in cancer or leukemia patients exposed to fluoroquinolones [28, 29]. This can be explained by 2 reasons. First, antibiotic exposure of the commensal flora is more frequent in patients than in healthy volunteers, and resistance may accumulate over time. Second, interpatient pharmacokinetic variability is expected to be greater in heavier, sicker, and older patients from the medical wards than that observed in healthy volunteers.

Emergence of resistance was mainly observed for fecal E. coli strains against nalidixic acid and for pharyngeal VGS strains against levofloxacin. Few patients had E. coli strains in the fecal flora that developed resistance to ciprofloxacin, and our study was underpowered to specifically examine differences in ciprofloxacin resistance.

Although our study was not specifically designated to analyze time-effect relationship, it showed that prevalences of resistance were not different between days 7 and 14, suggesting that the impact on commensal flora was already achieved at day 7 (figure 1).

Different kinetic patterns of the emergence of resistant bacteria were observed in the 2 commensal flora. Resistant E. coli strains in the fecal flora were detected mainly 4 weeks after the end of therapy, whereas resistant VGS strains in the pharyngeal flora were recovered primarily during the 2-week treatment period (figure 1). Pharmacokinetic and pharmacodynamic parameters could account for these differences. In stool specimens, ciprofloxacin concentrations were very high; indeed, they were several thousand times greater than the initial MPC on the dominant flora, explaining that E. coli virtually disappeared from the stool during therapy and that selection of resistance was unlikely. Such disappearance has previously been reported with different quinolones [33]. It has also been shown that ciprofloxacin persists in the feces of volunteers for several days after the end of oral treatment [34]. Therefore, resistance was probably selected in the fecal flora when ciprofloxacin concentrations decreased below the MPC and the MIC [14, 19]. This occurred between days 14 and 42, when ciprofloxacin could no longer be detected in stool specimens. In contrast, pharyngeal concentrations of ciprofloxacin during therapy were close to but less than the initial MIC against VGS during the entire treatment period, regardless of the regimen received, explaining why resistance could be selected during treatment and why it then decreased as antibiotic selective pressure vanished (figure 1). Indeed, subinhibitory concentrations of fluoroquinolones favor selection of resistance and induce genetic transformability, increasing the rate of mutation and genetic exchange in response to antibiotics [19, 35].

Depending on the target flora, we based the detection of quinolone resistance on different surrogate markers, according to their clinical relevance. Resistance to nalidixic acid, a first-generation quinolone that can be prescribed for the treatment of uncomplicated urinary tract infections, was used for E. coli because it is indicative of a single first-step mutation in the target gene gyrA and is associated with reduced activity of fluoroquinolones [29–31]. Resistance to levofloxacin was used to detect emergence of resistance in pharyngeal VGS because VGS are naturally resistant to ciprofloxacin and, therefore, have no MIC break points. In addition, levofloxacin is widely recommended for the treatment of respiratory infections [23]. Emergence of resistance to levofloxacin in pharyngeal VGS strains obviates the selection of cross-resistance among fluoroquinolones, as shown for Streptococcus pneumoniae [3].

A major issue in the use of fluoroquinolones is the role of dosage in the optimization of efficacy. We did not identify any significant difference in antibiotic exposure in plasma, saliva, and stool specimens between subjects in whom resistance was selected and those in whom it was not. Similarly, the probability of emergence of resistance was comparable regardless of the ciprofloxacin regimen, expressed as the ratio of the AUC to the MIC against the dominant flora. This was observed despite the fact that large ranges of ciprofloxacin dosages were investigated, from the minimum to the maximum total daily dose that can be administered therapeutically. A lack of power is here unlikely, and the duration of treatment was sufficient to allow the selection of resistant mutants. Therefore, these results indicate that, although optimization of the ciprofloxacin dosage is useful to increase efficacy at the focus of infection [15–18], it is not likely to be helpful to decrease the risk for selection of resistant strains in commensal flora. This may be explained by the different pharmacokinetic patterns of ciprofloxacin in saliva and stool specimens, as compared with the pattern in plasma specimens, and by the fact that variations in dosages had little impact on local ciprofloxacin concentrations in saliva and stool specimens, as compared with the respective MICs and MPCs of ciprofloxacin against the dominant flora at these sites (figure 2). Indeed, for all ciprofloxacin regimens tested, concentrations of ciprofloxacin were always far greater than the MPC for Enterobacteriaceae in stool specimens and less than the MIC for VGS in saliva specimens (figure 2). Overall, the comparable rates of emergence of resistance in the fecal and pharyngeal flora, regardless of antibiotic exposure, suggest a random phenomenon occurring at subinhibitory concentrations (table 1). Thus, selection of resistant commensals during ciprofloxacin therapy should be considered as an ecological side effect that is not preventable by optimizing antibiotic dosage. Other strategies to prevent such an event are warranted.

Acknowledgments

We thank Agnès Certain (pharmacist), for her helpful technical assistance; Isabelle Poirier, Emmanuelle Royer, and Patricia Servant (nurses); and Jean-Claude Soulé (head of the gastroenterology department), who allowed the nurses of his department to supervise the intake of ciprofloxacin on Sundays (all from Bichat hospital, Paris, France).

Author notes

Potential conflicts of interest: none reported.

Presented in part: 47th Interscience Conference on Antibiotic Agents and Chemotherapy, September 2007, Chicago, IL (abstract A-585).

Financial support: Direction de la Recherche Clinique, Assistance Publique-Hôpitaux de Paris (Contrat d'Investigation et de Recherche Clinique).