Mechanism of Fluoroquinolone Resistance Is an Important Factor in Determining the Antimicrobial Effect of Gemifloxacin against Streptococcus pneumoniae in an In Vitro Pharmacokinetic Model

Abstract

Antibacterial effect and emergence of resistance to gemifloxacin and levofloxacin were studied in an in vitro pharmacokinetic model of infection. A panel of Streptococcus pneumoniae strains with known mechanisms of resistance were used; two strains had no known resistance mechanism, two had efflux pumps, three had gyrA plus parC mutations, and one had only a parC mutation. Gemifloxacin MICs were in the range of 0.016 to 0.25 mg/liter, and levofloxacin MICs ranged from 1 to 16 mg/liter. Antimicrobial effect was measured by area under the bacterial-kill curve up to 72 h, and emergence of resistance was determined by population analysis profile before and during drug exposure. The area under the curve (AUC)/MIC ratios for gemifloxacin and levofloxacin were 35 to 544 and 3 to 48, respectively. As expected on the basis of these AUC/MIC ratio differences, antibacterial effect was much greater for gemifloxacin than levofloxacin. In the gemifloxacin simulations, mechanism of resistance as well as MIC determined the antibacterial effect, as indicated by gemifloxacin’s greater effect against efflux strains compared to those with gyrA or parC mutations despite similar MICs. This was not true of levofloxacin. Emergence of resistance was not easily demonstrated with either agent, and mechanism of resistance did not have any impact on it.

Fluoroquinolone resistance in Streptococcus pneumoniae is, at present, an uncommon occurrence (British Society for Antimicrobial Chemotherapy Working Party, Abstr. 41st Intersci. Conf. Antimicrob. Agents Chemother., 2001). However, it is well described in terms of its genetic basis and phenotypic expression, being related to either mutations in the quinolone resistance-determining regions (QRDR) of the genome or the presence of efflux pumps in the bacterial membrane. High-level fluoroquinolone resistance (ciprofloxacin MIC of 64 mg/liter) in S. pneumoniae is commonly associated, as determined by DNA sequence analysis, with mutations in parC and gyrA, although changes in parE and gyrB also occur. However, such strains are more susceptible to advanced-generation fluoroquinolones, such as gemifloxacin [V. J. Heaton, C. Goldsmith, J. Ambler, and L. M. Fisher, J. Antimicrob. Chemother. 44(Suppl. A):140, abstr. P452, 1999]). Efflux-mediated resistance, detected phenotypically by determination of norfloxacin MICs with or without an efflux inhibitor, was associated with ciprofloxacin MICs of ≥2 mg/liter but with gemifloxacin MICs of 0.06 mg/liter [N. Brenwald, M. J. Gill, F. Boswell, and R. Wise, J. Antimicrob. Chemother. 44(Suppl. A):145, abstr. P477, 1999]. The impact of the two types of fluoroquinolone resistance in gram-positive bacteria on the pharmacodynamics of antibacterial effects of and emergence of resistance to fluoroquinolones has not been fully explored. In an in vitro fibrin clot pharmacokinetic model with Staphylococcus aureus, inhibition of the NorA efflux pump by omeprazole increased the antibacterial activity of ciprofloxacin and decreased the emergence of resistance. There was little effect on the antibacterial activity of levofloxacin and none on the emergence of resistance (1).

When S. pneumoniae was used in an in vitro model, the presence of efflux pumps reduced the antibacterial effects of levofloxacin, moxifloxacin, and sparfloxacin and promoted the emergence of resistance (K. J. Madaras-Kelly, C. Daniels, M. Hegbloom, C. Nielson, and T. Kurtz, Abstr. 40th Intersci. Conf. Antimicrob. Agents Chemother., abstr. 296, 2000). In contrast, in a neutropenic murine thigh infection model with S. pneumoniae resistant to ciprofloxacin as a result of efflux or gyrase parC and parE mutations, the gemifloxacin area-under-the-curve (AUC)/MIC ratio needed to produce a net bacteriostatic effect was lower for the efflux pump-containing strains than for the other strains tested (D. Andes and W. A. Craig, Abstr. 39th Intersci. Conf. Antimicrob. Agents Chemother., abstr. 2032, 1999).

To further define the impact of the mechanism of fluoroquinolone resistance on the pharmacodynamic effects of this drug class on S. pneumoniae, we compared the antibacterial effects of and emergence of resistance to gemifloxacin or levofloxacin for a panel of S. pneumoniae strains with no known mechanism of resistance and with efflux pumps or mutations in the QRDR.

MATERIALS AND METHODS

Model.

A New Brunswick (Hatfield, Hertfordshire, England) Bioflo 1000 in vitro model was used to simulate the total serum drug concentrations associated with the oral administration of gemifloxacin at 320 mg every 24 h or levofloxacin at 500 mg every 24 h. The apparatus, which has been described before, consists of a single central chamber connected to a dosing chamber, which is in turn attached to a reservoir containing broth. The central chamber is connected to a collecting vessel for overflow (8). The contents of the dosing chamber and the central chamber were diluted with brain heart infusion broth (Oxoid, Basingstoke, England) by using a peristaltic pump (Ismatec; Bennett & Co., Weston-super-Mare, England) at flow rates of 35 ml/h for gemifloxacin and 39 ml/h for levofloxacin. The temperature was maintained at 37°C, and the broth in the dosing and central chambers was agitated by a magnetic stirrer.

Media.

75% Brain heart infusion broth was used in all experiments. Magnesium chloride (1%; BDH, Poole, Dorset, England) was incorporated into nutrient agar plates (Merck, Dorset, England) containing 5% whole horse blood (TCS Microbiology, Buckingham, England) to neutralize the fluoroquinolones before viable counts were determined.

Strains.

S. pneumoniae strains with defined mechanisms of resistance to fluoroquinolones were provided by G. Woodnutt, GlaxoSmithKline Pharmaceuticals, Collegeville, Pa. These isolates were as follows (designations are given as Southmead Hospital number-gemifloxacin MIC [milligrams per liter]-levofloxacin MIC [milligrams per liter]-GlaxoSmithKline number-resistance mechanism): SMH 21811-0.016-1.0-L/SB701226-none, SMH 21810-0.06-2-L/SB130-32-none, SMH 21851-0.06-2-L/SB611259-efflux pump, SMH 21849-0.06-8-L/SB507103-gyrA and parC mutations, SMH 21813-0.06-2-L/SB309350-parC mutation, SMH 21850-0.12-2-L/SB209163-efflux pump, SMH 21815-0.12-8-L/SB1-43C-gyrA and parC mutations, and SMH 21814-0.25-16-L/SB134-26-gyrA and parC mutations. The strains were clinical isolates from a variety of epidemiology programs studying resistance (3).

Efflux pumps were detected by inhibition by reserpine (strains 21851 and 21850). These two strains had wild-type QRDR. The various target substitutions in the QRDR were as follows: strain 21849, parC (Arg 95 → Cys) and gyrA (Ser 81 → Phe); strain 21813, parC (Ser 79 → Tyr); strain 21815, parC (Ser 79 → Phe) and gyrA (Ser 81 → Phe); and strain 21814, parC (Ser 79 → Phe) and gyrA (Ser 81 → Tyr).

Antibiotics.

Gemifloxacin was obtained from GlaxoSmithKline Pharmaceuticals, and levofloxacin was obtained from Aventis Pharma, Paris, France. Stock solutions were prepared according to British Society of Antimicrobial Chemotherapy guidelines (2) and stored at −70°C.

Pharmacokinetics, bacterial killing curves, and emergence of resistance.

The in vitro activities of the two fluoroquinolones at changing concentrations against the strains described were tested in the model. The target maximum concentration of gemifloxacin in serum was 1.1 mg/liter at 1.5 h, the half-life was 7 h, and the AUC was 8.7 mg/liter · h for a 320-mg dose. The target maximum concentration of levofloxacin was 5.0 mg/liter at 1.5 h, the half-life was 6.5 h, and the AUC was 48 mg/liter · h for a 500-mg dose. Total drug concentrations were modeled for each drug, and in all simulations, three doses were used—that is, the model was run for 72 h.

For all experiments, 100 μl of an overnight broth suspension of the test strain was inoculated into the central culture chamber (volume, 360 ml), and the model was run for 18 h to allow the organism growth to reach equilibrium at a density of about 10⁷ to 10⁸ CFU/ml. Gemifloxacin or levofloxacin was then added to the dosing chamber, and single samples were taken from the central chamber throughout the 72-h test period at 0, 2, 4, 6, 12, 24, 26, 28, 30, 36, 48, 60, and 72 h for the assessment of viable bacterial counts. The emergence of resistance was assessed at 0, 24, 48, and 72 h. Approximately 0.5 ml of broth was withdrawn from the central chamber and immediately plated on antibiotic-free and antibiotic-containing media. The bacteria were quantified by using a Spiral Plater (Don Whitley Spiral Systems, Shipley, West Yorkshire, England); the minimum detection level was 2 × 10² CFU/ml.

Additional aliquots were stored at −70°C for the measurement of gemifloxacin by using a bioassay described previously (4). All samples were assayed with Escherichia coli NCTC 10418 as the indicator. All standards and samples were prepared as needed in the same concentrations of brain heart infusion and ISB as those used in the simulations. The detection limit was 0.03 mg/liter, and the coefficient of variation was 11.1%.

Levofloxacin was assayed by high-pressure liquid chromatography. Samples were mixed with equal volumes of methanol, allowed to stand for 5 min, and centrifuged at 25,000 × g for 5 min. A 20-μl portion of the supernatant was injected into the column. The stationary phase was Spherisorb 50DS in a stainless steel column (25 cm by 4.6 mm; HPLC Technology, Macclesfield, United Kingdom). The mobile phase was 0.16% orthophosphoric acid, adjusted to pH 3 with tetrabutylammonium hydroxide; 50 ml of acetonitrile was added to the IL solution after pH adjustment. The flow rate was 1.5 ml/min. Detection was done by using a fluorescence excitation wavelength of 310 nm and an emission wavelength of 467 nm (model LC 240 chromatograph; Perkin-Elmer, Beaconfield, United Kingdom). The detection limit was 0.05 mg/liter, and the coefficient of variation was 8.9%.

Drug resistance was assessed before fluoroquinolone exposure; after 24, 48, and 72 h of exposure, samples were plated on 5% whole blood agar containing gemifloxacin or levofloxacin at 0.5, 1, 2, 4, 8, and 16 times the MIC for the strain in order to quantify the fluoroquinolone-resistant subpopulation.

All pharmacokinetic simulations and killing curve and drug resistance determinations were performed at least in triplicate.

Measurement of antibacterial effect and drug resistance.

The antibacterial effect was assessed by calculating the log change in viable counts between time zero and 24 h (Δ24), 48 h (Δ48), and 72 h (Δ72). The maximum reduction (Δ_max) was also recorded, as was the time taken for the inoculum to fall by 99.9% from its value at time zero (T99.9). In addition, the area under the bacterial killing curve (AUBKC, measured as log CFU per milliliter · h) was calculated by using the log linear trapezoidal rule for the period from 0 to 72 h (AUBKC_0-72) after standardization of the inoculum.

For drug resistance studies, the measure of resistance was the AUC for the population analysis profile (PAP). This value was calculated by using the log linear trapezoidal rule at 0, 24, 48, and 72 h (see Fig. 1 in reference 10a).

FIG. 1.

Target (open symbols) and measured (closed symbols) concentrations of gemifloxacin (squares) and levofloxacin (circles) in the pharmacokinetic model. Error bars indicate standard deviations.

RESULTS

Pharmacokinetic curves.

The measured gemifloxacin and levofloxacin concentrations were in good agreement with the target concentrations (Fig. 1).

Antibacterial effects.

The initial inoculum in all experiments was between log 7.4 and 8.5 units. The antibacterial effects of gemifloxacin or levofloxacin on the eight strains of S. pneumoniae tested are shown in Tables 1 and 2. The two strains with no known mechanism of resistance and low gemifloxacin MICs (0.016 and 0.06 mg/liter) were cleared from the model, with large reductions in viable counts and small AUBKC values. For strains for which the MIC was 0.06 mg/liter, the pattern of killing of S. pneumoniae strains with an efflux pump was more similar to that of the strains with no mechanism of resistance than to that of the strains with gyrA or parC mutations.

TABLE 1.

Antibacterial effect of gemifloxacin on S. pneumoniae^a

Strain	Gemifloxacin MIC (mg/liter)	AUC/MIC (h)	Resistance mechanism	Log CFU/ml				T99.9 (h)	AUBKC0-72 (log CFU/ml · h)
				Δ24	Δ48	Δ72	Δmax
21811	0.016	544	None	6.2 ± 0.3	6.2 ± 0.3	6.2 ± 0.3	6.2 ± 0.3	11 ± 2	69.2 ± 9.5
21810	0.06	145	None	2.1 ± 1.0	4.0 ± 1.4	3.3 ± 2.6	5.3 ± 0.3	26 ± 2	209.3 ± 4.1
21851	0.06	145	Efflux	3.1 ± 1.0	5.0 ± 1.2	5.2 ± 0.9	5.8 ± 0.3	15 ± 12	112.1 ± 37.8
21849	0.06	145	parC + gyrA	0.7 ± 0.8	0.9 ± 0.1	0.9 ± 0.1	1.2 ± 0.3	>72	371.2 ± 21.9
21813	0.06	145	parC	0.3 ± 0.3	0.5 ± 0.4	0.6 ± 0.3	1.2 ± 0.5	>72	390.9 ± 2.0
21850	0.12	72	Efflux	2	4.4	3.8	5.0 ± 1.1	21 ± 8	141.0 ± 46.2
21815	0.12	72	parC + gyrA	0.2 ± 0.4	0.5 ± 0.1	0.4 ± 0.4	0.8 ± 0.9	>72	400.0 ± 11.3
21814	0.25	35	parC + gyrA	0 ± 0.2	0.1 ± 0.1	0 ± 0.2	0.3 ± 0.3	>72	431.8 ± 10.3

^aData are reported as mean and standard deviation.

TABLE 2.

Antibacterial effect of levofloxacin on S. pneumoniae^a

Strain	Levofloxacin MIC (mg/liter)	AUC/MIC (h)	Resistance mechanism	Log CFU/ml				T99.9 (h)	AUBKC0-72 (log CFU/ml · h)
				Δ24	Δ48	Δ72	Δmax
21811	1	48	None	0.6 ± 0.6	2.6 ± 2.7	4.7 ± 2.9	4.7 ± 2.9	52 ± 18	282 ± 95
21810	2	24	None	1.6 ± 2.6	0.3 ± 0.3	0.1 ± 1	2.2 ± 3.0	>72	375 ± 75
21851	2	24	Efflux	0.2 ± 0.2	0.2 ± 0.3	0.2 ± 0.7	0.3 ± 0.1	>72	422 ± 15
21849	8	6	parC + gyrA	0.1 ± 0.1	1.1 ± 0.2	0.4 ± 0.6	1.1 ± 0.2	>72	398 ± 2
21813	2	24	parC	0.2 ± 0.1	0.3 ± 0.1	0.2 ± 0.1	0.3	>72	419 ± 3
21850	2	24	Efflux	0.3 ± 0.3	0.2 ± 0.1	0.6	0.5 ± 0.1	>72	393 ± 39
21815	8	6	parC + gyrA	0.1 ± 0.1	0 ± 0.2	0 ± 0.2	0.9 ± 0.1	>72	417 ± 11
21814	16	3	gyrA	0.2 ± 0.3	0.1 ± 0.2	0 ± 0.2	0.3 ± 0.2	>72	423 ± 18

^aSee Table 1, footnote a.

The pattern seen with levofloxacin was different (Table 2), as in no simulation did clearance of the pathogen from the model occur. No differences between strains based on mechanism of resistance were detectable.

Emergence of resistance.

The emergence of resistance was studied by using the PAP (see Fig. 1 in reference 10a). The AUC for the PAP was taken as the primary measure of emergence of resistance (Tables 3 and 4). For strains with gemifloxacin MICs of 0.016 or 0.06 mg/liter and no resistance mechanism or efflux pump, the AUC for the PAP was reduced over the period of simulation, as bacteria were cleared from the model. There was a trend toward a smaller AUC for the PAP for strain 21849 (MIC, 0.06 mg/liter; parC and gyrA mutations), strain 21813 (MIC, 0.06 mg/liter; parC mutation), and strains 21815 and 21814 (MICs, 0.12 and 0.25 mg/liter; parC and gyrA mutations). For levofloxacin, there was little clearance of the strains at an MIC of ≥2 mg/liter; however, despite this result, the AUC for the PAP did not show any trend toward increasing resistance (Table 4).

TABLE 3.

Emergence of resistance to gemifloxacin in strains of S. pneumoniae with defined resistance mechanisms^a

Strain	MIC (mg/liter)	Resistance mechanism	AUC for PAP (log CFU/ml × MIC) at the following time (h):
			0	24	48	72
21811	0.016	None	6.6 ± 1.0	<3	<3	<3
21810	0.06	None	10 ± 0.5	6.4 ± 2.8	<3	<3
21851	0.06	Efflux	21.5 ± 4.2	16.4 ± 12.7	<3	<3
21849	0.06	parC + gyrA	12.9 ± 2.3	13.4 ± 1.3	17.1 ± 9.8	22.8 ± 20.3
21813	0.06	parC	14.5 ± 2.0	33.9 ± 10.5	33.2 ± 14.1	39.1 ± 11.1
21850	0.12	Efflux	13.8	9.5	5.1	11.5
21815	0.12	parC + gyrA	17.1 ± 3.1	17.1 ± 5.9	17.7 ± 5.9	25.7 ± 2.4
21814	0.25	parC + gyrA	15.2	20.7	18.5	21.5

^aSee Table 1, footnote a.

TABLE 4.

Emergence of resistance to levofloxacin in strains of S. pneumoniae with defined resistance mechanismsa

Strain	MIC (mg/liter)	Resistance mechanism	AUC for PAP (log CFU/ml × MIC) at the following time (h):
			0	24	48	72
21811	1	None	3.5 ± 0.3	3.3 ± 0.1	<3	<3
21810	2	None	8.1 ± 0.1	6.2 ± 2.7	6.1 ± 3.1	15.3 ± 5.7
21851	2	Efflux	12.9 ± 0.3	10.8 ± 2.5	12.8 ± 0.1	7.8 ± 0.1
21849	8	parC + gyrA	9.9 ± 1.7	9.4 ± 3.1	4.6 ± 1.8	7.7 ± 1.1
21813	2	parC	8.2 ± 0.5	8.7 ± 0.8	7.7 ± 0.1	8.8 ± 1.2
21850	2	Efflux	8.2 ± 0.1	7.7 ± 0.2	7.6 ± 0.6	7.2 ± 0.6
21815	8	parC + gyrA	14.4 ± 0.4	13.3 ± 2.0	9.5 ± 0.1	14.7 ± 0.6
21814	16	parC + gyrA	8.1 ± 0.2	8.1 ± 0.3	8.6 ± 0.2	8.1 ± 0.2

DISCUSSION

The antibacterial effect of gemifloxacin was markedly greater than that of levofloxacin against the panel of S. pneumoniae strains used. However, for both agents, the MICs for the selected strains were above the normally recorded MICs at which 90% of the strains are inhibited of 0.03 to 0.06 mg/liter for gemifloxacin and 1.0 to 2.0 mg/liter for levofloxacin (5, 11). The strains therefore represent an abnormally resistant and challenging population and are unlike strains likely to be encountered in routine clinical practice. Levofloxacin resulted in little bacterial clearance from the model over the 72-h simulation, as may be expected, given that the AUC/MIC ratios for seven of eight strains were in the range of 3 to 24. For only one strain, the most sensitive (levofloxacin MIC, 1.0 mg/liter), was the AUC/MIC ratio in the range of 40 to 50, a value previously associated with pneumococcal clearance in some in vitro pharmacokinetic models (6, 7, 14). In this model, a 4.5- to 5-log reduction in counts was achieved by 72 h and was associated with an AUC/MIC ratio of 48. In contrast, gemifloxacin had AUC/MIC ratios in the range of 35 to 544 and a much more marked antibacterial effect than levofloxacin. However, the strain that was most resistant to gemifloxacin (AUC/MIC ratio, 35) was cleared less well from the model than the strain that was most susceptible to levofloxacin (AUC/MIC ratio, 48). Previously, it was shown with this model system that the AUC/MIC ratio best predicts the antibacterial effect on S. pneumoniae; the greater effect seen here with gemifloxacin is certainly related to its larger AUC/MIC ratios (9, 10).

In addition, gemifloxacin had markedly different antibacterial effects on strains for which MICs were similar but which had different mechanisms of resistance. A comparison of antibacterial effects on strains for which the gemifloxacin MIC was 0.06 or 0.12 mg/liter indicated less of an effect against strains with parC or both parC and gyrA mutations than against strains with no known mechanism of resistance or efflux pump. Gemifloxacin appeared to have a greater effect against one strain with an efflux pump than against a strain for which the MIC was equivalent but which had no detectable mechanism of resistance. Previous animal data obtained with gemifloxacin in a neutropenic mouse thigh model support these observations. The AUC/MIC ratio required for a net bacteriostatic effect against 13 S. pneumoniae strains with gyrA, parC, or parE mutations was 150, while the equivalent value for efflux pump-containing strains was 27 (Andes and Craig, 39th ICAAC). Along with our data, these results indicate that there is a mismatch between the MIC and the antibacterial effect of gemifloxacin on S. pneumoniae for which MICs are increased, due to either the presence of efflux pumps or mutations in the QRDR. In short, there is a greater antibacterial effect on S. pneumoniae strains containing efflux pumps than would be predicted on the basis of MICs alone; the mechanism of resistance, independent of the MIC, is a factor in determining the antibacterial effect.

Our in vitro pharmacokinetic model and the reported animal data are at variance with another in vitro pharmacokinetic study of an efflux pump-containing strain of S. pneumoniae and two strains not containing efflux pumps. It was shown that for levofloxacin, moxifloxacin, and sparfloxacin, the antibacterial effect was lower against the efflux pump-positive strain. However, the strains were not matched for MIC; hence, the contributions of MIC increases and efflux pumps could not be easily separated (Madaras-Kelly et al., 40th ICAAC).

Despite the lack of clearance from the model with levofloxacin, there was no detectable emergence of resistance. Similar observations for levofloxacin were made previously with in vitro and animal pharmacokinetic models using different techniques (13, 14; N. L. Jumbe, A. Louie, W. Liu, M. Deziel, M. H. Miller, and G. L. Drusano, Abstr. 40th Intersci.Conf. Antimicrob. Agents Chemother., abstr. 291, 2000), and we observed a similar phenomenon with moxifloxacin in this in vitro model. It was previously suggested that the emergence of resistance is more likely in strains with efflux pumps than in those without efflux pumps, and some data from in vitro pharmacokinetic models support this notion (Madaras-Kelly et al., 40th ICAAC). MICs for S. pneumoniae were measured before and after exposure to levofloxacin, moxifloxacin, and sparfloxacin, and an increase in the MIC was noted in 1 of 12 simulations with two strains not containing efflux pumps but in 4 of 6 simulations with an efflux pump-containing strain. In contrast, these data indicate that for both gemifloxacin and levofloxacin, the emergence of resistance is not determined by the initial MIC or the mechanism of resistance. However, some caution must be exercised, as the MIC will have an impact on the emergence of resistance if it is low enough to allow complete eradication from the model; in addition, as the findings obtained with S. pneumoniae were predominantly negative, other target pathogens, such as Pseudomonas aeruginosa or S. aureus, may prove better paradigms for studying the pharmacodynamics of the emergence of resistance.

In conclusion, our data confirm that the emergence of fluoroquinolone resistance in S. pneumoniae is difficult to detect in in vitro pharmacokinetic models even with a large bacterial inoculum, low AUC/MIC ratios, and long periods of drug exposure. However, they clearly show that for S. pneumoniae, the MIC alone does not predict the antibacterial effect of gemifloxacin, and the mechanism of resistance is an important factor. This work illustrates the need to determine the influence of the mechanism of resistance on factors other than the MIC when one is trying to assess their likely pharmacodynamic implications and subsequent clinical impact.