INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Streptococcus pneumoniae is a significant cause of morbidity and mortality and is one of the primary pathogens implicated in community-acquired pneumonia (2, 3, 5, 17, 36). Newer fluoroquinolones, such as levofloxacin, trovafloxacin, and clinafloxacin, have been shown to possess greater activity than older agents of this class against gram-positive organisms, including S. pneumoniae (3, 6, 8, 13, 19, 23, 34). The newer fluoroquinolones thus are potential alternatives for therapy of infections caused by multidrug-resistant pneumococci. With the increasing use of these drugs, however, there is growing concern regarding the development of quinolone resistance among gram-positive bacteria (10, 21).

Animal models, in vitro pharmacodynamic models, and limited human data have demonstrated that the primary pharmacodynamic parameters which most closely correlate with therapeutic efficacy of fluoroquinolones are area under the concentration-time curve from 0 to 24 h (AUC_0-24)/MIC and maximum concentration of drug in serum (C_max)/MIC ratios (1, 16, 28, 37). However, there are limited data correlating pharmacodynamic parameters with the development of resistance (39).

Ciprofloxacin-resistant strains of S. pneumoniae have been reported, and most troubling is the possibility of cross-resistance to the newer fluoroquinolones (11, 21, 26, 27, 32). These compounds have improved activity against topoisomerase IV, which is thought to be the primary target for most fluoroquinolones in gram-positive organisms such as S. pneumoniae and Staphylococcus aureus. In vitro data on the susceptibility of penicillin-susceptible and -resistant pneumococci to fluoroquinolones have demonstrated that both ciprofloxacin and ofloxacin MICs tend to cluster around their susceptibility breakpoints of 2 and 4 µg/ml, respectively, whereas the MICs of the newer quinolones are much lower: levofloxacin, 0.5 to 2 µg/ml; gatifloxacin, 0.25 to 0.5 µg/ml; moxifloxacin, 0.015 to 0.06 µg/ml; and trovafloxacin, 0.06 to 0.5 µg/ml (13, 22, 34, 41).

The purpose of the present study was to evaluate pharmacodynamic relationships such as AUC/MIC and C_max/MIC by comparing the activities of several newer fluoroquinolones with that of ciprofloxacin against ciprofloxacin-resistant S. pneumoniae using an in vitro pharmacodynamic model. Additionally, we characterized the impact of ciprofloxacin resistance in S. pneumoniae on fluoroquinolone pharmacodynamics in relationship to killing and the emergence of higher-level resistance.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains. The parent isolate, S. pneumoniae 79, was a penicillin- and erythromycin-resistant clinical isolate (MICs of 2 and 12 µg/ml, respectively). Additional study strains consisted of two laboratory-derived, ciprofloxacin-resistant mutants of strain 79. These mutants were produced from the parent strain by serial passage in the presence of ciprofloxacin (21). Briefly, strain 79 was streaked onto agar plates containing increasing multiples of the ciprofloxacin MIC. After several passages, two mutants (R919 and R921) for which ciprofloxacin MICs were apparently increased were recovered and used in further studies.

Antimicrobial agents. Gatifloxacin was supplied by Bristol-Myers Squibb, New Brunswick, N.J.; grepafloxacin was supplied by Glaxo Wellcome, Research Triangle Park, N.C.; ciprofloxacin and moxifloxacin were supplied by Bayer Corporation, West Haven, Conn.; and trovafloxacin was supplied by Pfizer Inc., Groton, Conn. Levofloxacin for injection was commercially purchased from Ortho-McNeil, Raritan, N.J. One lot number of each drug was used throughout the study.

Susceptibility testing. MICs were determined by broth microdilution using Mueller-Hinton broth (Difco Laboratories, Detroit, Mich.) supplemented with calcium (25 mg/ml), magnesium (12.5 mg/ml), and 5% lysed horse blood (Rockland, Inc., Gilbertsville, Pa.) (SMHB-LHB) according to NCCLS guidelines (31). MICs also were determined in Todd-Hewitt broth supplemented with 0.5% yeast extract (Difco Laboratories) (THB-Y). In addition, trovafloxacin MICs were determined in the presence of albumin (4 g/dl) to evaluate the effect of protein binding on the activity of the drug. The possible presence of an efflux mechanism of resistance was investigated by determining the MICs of several common substrates of multidrug efflux pumps (ethidium bromide, acriflavine, benzalkonium chloride, cetrimide, and tetraphenylphosphonium bromide) and by evaluating the effect of reserpine (10 µg/ml) on selected MICs (7, 24, 30, 42). MIC plates were incubated in candle jars at 37°C for 24 h in the presence of approximately 3% CO₂.

Sequence determination of the QRDRs of gyrA, gyrB, parC, and parE. The quinolone resistance-determining regions (QRDRs) of gyrA, gyrB, parC, and parE were amplified from bacterial strains using PCR. Primers and conditions were employed as previously described (33). PCR products were sequenced using the dideoxy chain termination method (37).

In vitro pharmacodynamic models. THB-Y was utilized for the in vitro pharmacodynamic models, and tryptic soy agar supplemented with 5% sheep blood (TSA-SB) was used to determine colony counts (20). The in vitro model consisted of a 250-ml one-compartment glass chamber with ports for the addition and/or removal of media, antibiotics, and samples (19, 20). Prior to each experiment, colonies from an overnight growth on TSA-SB were added to THB-Y as necessary to obtain a suspension of ~1 × 10⁸ CFU/ml. A 2.5-ml volume of this suspension was added to the chamber to produce a starting inoculum of ~1 × 10⁶ CFU/ml. Fresh stock solutions of fluoroquinolones were prepared on the first day of the experiment and were stored at 2 to 8°C between dosage administration times. Dosing regimens included ciprofloxacin at 400 mg every 12 h (peak concentration, 5 µg/ml), gatifloxacin at 400 mg every 24 h (peak concentration, 3.5 µg/ml), grepafloxacin at 600 mg every 24 h (peak concentration, 2.5 µg/ml), levofloxacin at 500 mg every 24 h (peak concentration, 6 µg/ml), moxifloxacin at 400 mg every 24 h (peak concentration, 5 µg/ml), and trovafloxacin at 200 mg every 24 h (peak concentration, 3 µg/ml) for 48 h. Antibiotics were administered as a bolus into the models over 30 s using a hypodermic syringe. A peristaltic pump (Masterflex; Cole-Parmer Instrument Company, Chicago, Ill.) was used to displace antibiotic-containing media to simulate the half-lives of ciprofloxacin (3 h), gatifloxacin (8 h), grepafloxacin (15 h), levofloxacin (6 h), moxifloxacin (10 h), and trovafloxacin (12 h) (16, 35, 38). Each model apparatus was placed in a water bath and maintained at 37°C for the entire study period. Model experiments were performed in duplicate to ensure reproducibility.

Pharmacokinetic analysis. Samples (0.5 ml) from each model experiment were obtained at 0, 0.5, 2, 4, 8, 24, 28, 32, and 48 h for the determination of antibiotic concentrations and were stored at 70°C until analysis. Concentrations of ciprofloxacin, gatifloxacin, grepafloxacin, levofloxacin, moxifloxacin, and trovafloxacin were determined by bioassay using Klebsiella pneumoniae ATCC 10031 as the indicator organism. Blank 1/4-in. paper disks were spotted with 20 µl of samples or standards, which then were placed in triplicate on Mueller-Hinton agar plates preswabbed with a 0.5-McFarland standard suspension of indicator organism. Concentrations of fluoroquinolones used as standards were 5, 1.25, and 0.3125 µg/ml. Plates were incubated for 18 to 24 h at 37°C. All plates achieved a correlation coefficient of 0.95. For all fluoroquinolones, the between-day coefficient of variation was 3.1 to 6.8%. Antibiotic half-lives were calculated from the slopes of the drug concentration versus time plots. The AUC obtained from the drug concentration versus time plot was calculated by the linear trapezoidal rule using the PK Analyst programs (Micromath, Salt Lake City, Utah).

Pharmacodynamic analysis. Samples (0.5 ml) were removed from each model at the same time points used for the pharmacokinetic analysis, except that samples also were obtained at 1 and 6 h. Each sample was serially diluted in cold 0.9% sodium chloride, and bacterial counts were determined by placing 20-µl spots of dilutions onto TSA-SB. Colonies were counted following incubation for 24 h at 37°C. It has been determined previously that these methods have a limit of detection of 2 log₁₀ CFU/ml (9). The total reduction in log₁₀ CFU/ml over 48 h was determined by plotting time-kill curves, and the time to achieve a 99.9% reduction in the initial inoculum was determined by visual inspection of these curves. To avoid antibiotic carryover, all samples were diluted sufficiently prior to plating, such that antibiotic concentrations were reduced below the MICs of the drugs for each organism. The AUC_0-24/MIC and C_max/MIC ratios were calculated for each fluoroquinolone. The development of raised MICs of each antibiotic during experiments was investigated by plating 100 µl of the 24- and 48-h samples onto TSA-SB containing two and four times the MIC of the appropriate fluoroquinolone. These plates were examined for growth after incubation for 48 h at 37°C. Changes in MICs also were checked for each model at 48 h via Etest (AB Biodisk, Solna, Sweden).

Statistical analysis. Changes in log₁₀ CFU/ml at 48 h plus time to 99.9% killing were compared by analysis of variance with Tukey's post hoc test for multiple comparisons. Relationships between AUC_0-24/MIC, C_max/MIC, and log₁₀ CFU/ml at 48 h were determined by Pearson's correlation. A P value of 0.05 was considered significant.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Susceptibility testing. MIC data for study strains are summarized in Table 1. The ciprofloxacin MICs for strains R919 and R921 were unchanged following seven serial passages on antibiotic-free media. Overall, there were no differences in MICs determined in SMHB-LHB or THB-Y. Trovafloxacin was the most active fluoroquinolone against R919 and R921, followed by grepafloxacin, moxifloxacin, gatifloxacin, and finally levofloxacin. Ciprofloxacin and norfloxacin MICs were increased 8- to 16-fold and 4- to 8-fold for R919 and R921, respectively. The presence of albumin did not affect the MIC results observed for trovafloxacin.

QRDR sequencing. Strain R919 was found to have a mutation resulting in an S114G substitution in GyrA. No mutations resulting in amino acid substitutions were found in the gyrB QRDR of R919 or the gyrA and gyrB QRDRs of R921. All strains were found to have mutations resulting in a K137N substitution in ParC and an I460V substitution in ParE.

Pharmacodynamic and pharmacokinetic studies. The results of model experiments are shown in Fig. 1. All drugs but ciprofloxacin achieved 99.9% killing in approximately 28 h against isolate R919. Regrowth at 48 h was minimal with gatifloxacin, grepafloxacin, moxifloxacin, and trovafloxacin; each maintained 99.9% killing at this time point. Against isolate R921, all but ciprofloxacin and levofloxacin achieved 99.9% killing by 28 h, with gatifloxacin, moxifloxacin, and trovafloxacin killing to the limit of detection (2 log₁₀ CFU/ml) at 48 h. Modest, statistically insignificant regrowth was observed with gatifloxacin at 48 h. Although levofloxacin initially showed activity against both R919 and R921, by 48 h significant regrowth was seen (P < 0.05). Ciprofloxacin had no inhibitory or killing effect against either R919 or R921.

View larger version (22K): [in this window] [in a new window]   FIG. 1.   In vitro pharmacodynamic results. (A) R919; (B) R921. Data are means (± standard deviations) of multiple (four to six) determinations.

Resistance. Samples from each model were evaluated for resistance at 24 and 48 h by plating aliquots onto solid media containing two and four times the MIC of the appropriate drug. Increases in MICs of ciprofloxacin (32 µg/ml) were observed at 24 and 48 h for both R919 and R921. Slight increases in MICs (one- to twofold) of levofloxacin were observed at 48 h for both mutants. No changes in MICs of any of the other fluoroquinolones were observed.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The incidence of penicillin resistance in S. pneumoniae has increased significantly in the United States over the last 2 decades (4, 12, 17, 40). Resistance to other antimicrobial agents, such as cephalosporins, macrolides, and trimethoprim-sulfamethoxazole, also is increasing (17, 40). The emergence of multidrug-resistant S. pneumoniae has prompted the need for the development of alternative treatments. The newer fluoroquinolones are viable options, as they have greater activity against gram-positive organisms. In fact, recent treatment guidelines for community-acquired pneumonia list newer fluoroquinolones such as levofloxacin, grepafloxacin, and trovafloxacin as alternatives for treatment of this condition (5).

There has been a rapid emergence of resistance to the fluoroquinolones in gram-positive bacteria, especially in S. aureus (11, 13, 24, 42). Reported resistance mechanisms in that organism include alterations in topoisomerase IV and DNA gyrase and/or active efflux of the drug (21-26, 33, 42). Similar resistance mechanisms can be found in S. pneumoniae. Furthermore, ciprofloxacin has not proven to be consistently reliable for treating infections caused by S. pneumoniae. Ciprofloxacin-intermediate (MIC, 1 to 2 µg/ml) and -resistant (MIC, >2 µg/ml) strains have already been reported (3, 10, 13, 14, 21, 33, 42).

To date, the incidence of fluoroquinolone-resistant S. pneumoniae has been relatively low (<2%). However, an increase in the usage of fluoroquinolones for therapy of community-acquired pneumonia and other infections may result in increased resistance. Two recent reports from Hong Kong and Canada found resistance to be present in 12.1 and 1.7% of strains, respectively. Interestingly, resistance was shown not only to the older fluoroquinolones, such as ciprofloxacin, but also to levofloxacin and trovafloxacin (10, 21).

Strain R919 was found to have a mutation resulting in an S114G substitution in GyrA. This substitution has been described previously in ciprofloxacin-resistant clinical isolates of S. pneumoniae, but its relation to the resistance phenotype is uncertain, as no genetic analysis of it has been done (15). All strains, including the ciprofloxacin-susceptible parent strain, had single amino acid substitutions in both ParC and ParE. The fact that these substitutions were found in the susceptible parent strain indicates that they are not responsible for the raised fluoroquinolone MICs observed in R919 and R921. MIC data revealed that both R919 and R921 have multidrug resistance phenotypes reversible by reserpine. This indicates the likely presence of a multidrug efflux pump which, based on the nearly complete reversal of resistance by reserpine, is the main resistance mechanism in R919 and R921. The multidrug resistance phenotypes of R919 and R921 may represent activity of PmrA, a recently described S. pneumoniae multidrug efflux pump (18). The fact that reserpine resulted in a decrease in MICs of various compounds in strain 79 as well indicates that this pump is active in wild-type susceptible strains.

Our experiments were designed to evaluate the efficacy of the newer fluoroquinolones against ciprofloxacin-resistant strains of S. pneumoniae. As predicted, ciprofloxacin was not effective against either R919 or R921. Although levofloxacin displayed activity against R919 for up to 28 h, significant regrowth was noted at 48 h and minimal short-term activity was seen against R921. Gatifloxacin, grepafloxacin, moxifloxacin, and trovafloxacin each had significant activity against both isolates, with greater than 99.9% killing being observed at 48 h. Both isolates displayed increases in the level of resistance to ciprofloxacin and levofloxacin upon exposure to either drug, with increases in MICs seen as early as 24 h for ciprofloxacin.

Most of the definitive studies with respect to the pharmacodynamic effects of the fluoroquinolones have concentrated on gram-negative infections, where an AUC_0-24/MIC ratio of >100 or a C_max/MIC ratio of >8 appears to be predictive of a good therapeutic response (16, 28, 36, 39). There are limited data, however, examining this relationship in gram-positive bacteria. Andes and Craig evaluated 19 publications examining the use of fluoroquinolones in various models of experimental endocarditis (1). The data suggested that a 24-h AUC/MIC ratio of 100 may be the best predictor of successful outcome. In contrast, a study by Wright et al. examining pharmacodynamic outcome parameters for levofloxacin and ciprofloxacin versus S. pneumoniae in an in vitro pharmacodynamic model found no significant difference in the relationship of response to AUC_0-24/MIC or C_max/MIC. Ciprofloxacin AUC_0-24/MIC ratios of 17.4 and 8.7 were associated with regrowth at 24 h. AUC_0-24/MIC ratios of >34.8 were not associated with regrowth (D. H. Wright, M. L. Peterson, L. B. Hovde, G. Brown, and J. C. Rotschafer, Abstr. 38th Intersci. Conf. Antimicrob. Agents Chemother., abstr. A-111a, 1998). Previously, an attempt by our laboratory to describe a relationship between the AUC_0-24/MIC ratio for fluoroquinolones and S. pneumoniae was not successful (20). This likely was due to the use of fluoroquinolone-sensitive strains in our experiment, where the AUC_0-24/MIC ratio was optimized even with some of the older fluoroquinolones. Evaluating strains with lower susceptibility to fluoroquinolones increases the ability to find a correlation. In the present study, we did observe a significant (P < 0.05) correlation with AUC_0-24/MIC and C_max/MIC ratios in relationship to killing and resistance; however, a specific value for these two parameters was not determined. It should be noted that our pharmacodynamic experiment did not model pneumococcal pneumonia and that the contribution of immune factors, such as neutrophils, was absent. Therefore, our results should be extrapolated with caution. Significant regrowth and resistance were observed with an AUC_0-24/MIC ratio of 31.7 and C_max/MIC ratio of 3.1, and bacterial counts remained at the limit of detection for AUC_0-24/MIC ratios of 82. We did not have AUC_0-24/MIC ratios that ranged between >31.7 and 75.9. Therefore, we cannot pinpoint the exact AUC_0-24/MIC ratio needed to prevent regrowth and resistance. It is also important to note that the significance of regrowth in our model is unknown. Although there was a significant relationship between the AUC_0-24/MIC and C_max/MIC ratios and regrowth, it is not known whether the regrowth could be extrapolated to clinical situations, especially in light of the fact that for most of the time there were no changes in MICs from baseline for the organisms. In order to determine more definitively a specific AUC_0-24/MIC and/or C_max/MIC ratio predictive of outcome, studies examining numerous isolates for which MICs are close to the NCCLS breakpoints are needed.