Postantibiotic Effects and Bactericidal Activities of Levofloxacin and Gatifloxacin at Concentrations Simulating Those of Topical Ophthalmic Administration against Fluoroquinolone-Resistant and Fluoroquinolone-Sensitive Methicillin-Resistant Staphylococcus aureus Strains

Abstract

The bactericidal activities and postantibiotic effects (PAEs) of levofloxacin and gatifloxacin at concentrations corresponding to those in antibiotic eye drops against methicillin-resistant Staphylococcus aureus strains were determined. Levofloxacin and gatifloxacin at concentrations simulating those in eye drops showed lower bactericidal activities and shorter PAEs against fluoroquinolone-resistant strains than against fluoroquinolone-sensitive strains.

Ocular infections caused by methicillin-resistant Staphylococcus aureus (MRSA) strains, which are clinically serious and difficult to treat, have been reported (18, 27-30, 34). The prevalence of levofloxacin (LVX) resistance in MRSA strains isolated from patients with ocular infections is 80 to 90% in Japan and the United States (12, 14, 17). Topical treatments may have selected for strains with high-level fluoroquinolone resistance (1, 8, 10, 14, 17), which is associated with multiple mutations in the quinolone resistance-determining regions (QRDRs) of MRSA strains isolated at ophthalmologic clinics (10). Although topical fluoroquinolone eye drops contain much higher concentrations (0.3 to 1.5%; equivalent to 3,000 to 15,000 μg/ml) than their MICs even against strains with high-level fluoroquinolone resistance, the antibiotic concentrations in tear films decline rapidly (by about 1/100 at 30 min) (31), and treatment failures have been noted (6, 33). We performed time-kill and postantibiotic effect (PAE) studies with LVX and gatifloxacin (GAT) at concentrations simulating those of topical ophthalmic administration against fluoroquinolone-resistant and -sensitive MRSA strains to examine the efficacies of these drugs. We examined nine typical Japanese health care facility-associated MRSA strains: staphylococcal cassette chromosome mec type IIa (3, 13, 22, 24) strains with high-level LVX resistance (LHR; MICs, ≥64 μg/ml), low-level LVX resistance (LLR; MICs, 4 to 32 μg/ml), or LVX sensitivity (LS; MICs, ≤1 μg/ml). Cation-adjusted Muller-Hinton II broth (CAMHB; BD, Sparks, MD) and Mueller-Hinton agar (BD) were used for determination of the MICs, time-kill profile analyses, PAE assays, and viable count assays. MICs were measured by broth microdilution methods and were interpreted according to the recommendations of the Clinical and Laboratory Standards Institute (4). To elucidate the contribution of efflux mediated by NorA to fluoroquinolone resistance, the MICs of LVX and GAT were determined in the presence or absence of 20 μg/ml reserpine, which is a multidrug transport inhibitor (21). The QRDRs and norA promoter regions were sequenced as reported previously (10, 20, 21, 35). Time-kill assays were performed according to previously published guidelines (19). LVX and GAT were used at 4×, 8×, and 16× the MICs and at 2,048 μg/ml, which is similar to the concentrations of these two drugs used in eye drops. The initial inoculum (time zero) was adjusted to 1 × 10⁶ CFU/ml. Samples were taken at 0 min, 5 min, and 30 min and at 1, 2, 4, 6, and 24 h. Serial 10-fold dilutions in sterile saline were plated, and the colonies were counted after 48 h of incubation at 37°C (19). Bactericidal activity was defined as a >3-log₁₀ decrease in the numbers of CFU/ml (19). The PAEs of LVX and GAT were determined by the filter method (5). Strains grown in culture to 1 × 10⁶ CFU/ml in the exponential phase were exposed to 3,000 μg/ml LVX or 2,000 μg/ml GAT for 5 min; these antibiotic concentrations were derived from the areas under the concentration-time curves of LVX and GAT in tears after the administration of a single eye drop (25, 31). After removal of the drugs and six washes with saline, the membrane-trapped bacteria were resuspended in prewarmed CAMHB and reincubated (5). Control bacteria not exposed to drug were washed and resuspended in a similar manner. Viable counts were determined every 1 h until the cultures in the test tubes were turbid. The PAE was defined as described previously (5).

The MIC profiles and the mechanisms of fluoroquinolone resistance in the MRSA strains used are shown in Table 1. The level of fluoroquinolone resistance was dependent upon multiplication of the mutation sites in the QRDRs. The correlation between changes in the norA promoter region and the effects of reserpine on the MICs of LVX and GAT suggested that the NorA efflux system did not confer fluoroquinolone resistance on the strains used, as reported previously (9, 10, 21). The results of the time-kill assays for LVX and GAT are presented in Fig. 1 and 2. At up to 60 min of exposure, LVX and GAT showed no detectable bactericidal activities at 4× to 16× the MICs. Only GAT was bactericidal against the LLR strains at 120 min of exposure. At a high concentration (2,048 μg/ml), close to that present in eye drops, LVX completely killed (>99.9%) only one of three LHR strains, zero of three LLR strains, and two of three LS strains, whereas GAT completely killed two of three LHR strains, two of three LLR strains, and three of three LS strains at 60 min. All the strains were killed completely by LVX and GAT at 4×, 8×, and 16× the MICs and at 2,048 μg/ml after 24 h of exposure (data not shown). Overall, GAT was more bactericidal than LVX for the MRSA strains. Similar findings have been obtained in previous studies, and it has been suggested that the differences observed are attributable to the C-8 methoxy moiety (2, 11). The PAEs of LVX and GAT are presented in Fig. 3. The PAEs of both drugs against the LHR and LLR strains were shorter in duration than those against the LS strains (Fig. 3A). There was a linear correlation between the PAEs of both drugs and their MICs (Fig. 3B). Since LVX and GAT at sub-MICs show PAEs against S. aureus strains (15, 23), the infrequent ophthalmic administration of both drugs may be effective against ocular infections caused by LS strains. The shorter PAEs and lower bactericidal activities against LHR and LLR strains compared to those against LS strains may affect clinical outcomes. The maintenance of adequate concentrations of fluoroquinolones (preferably GAT) by frequent administration may be more clinically efficacious against fluoroquinolone-resistant strains. Indeed, the frequent administration of eye drops containing GAT has successfully been used to treat moderately or highly resistant S. aureus strains (MICs, 12 μg/ml and >64 μg/ml, respectively) in a rabbit keratitis model (26). However, high concentrations of fluoroquinolones can cause corneal perforations in humans (7, 16, 32). We did not investigate how the growth-inhibitory effects of long-duration supra-MIC and sub-MIC dosages of antibiotics following brief exposure to a high concentration of the antibiotic affected the antibacterial responses. Further studies are required to determine the usefulness of these two antibiotics in topical formulations (eye drops) in different in vitro simulation models that reproduce the pharmacokinetic-pharmacodynamic properties of the antibiotics in the human eye when they are administered on traditional dosing schedules.

TABLE 1.

Fluoroquinolone susceptibilities and mutation profiles of MRSA strains

Strain	MIC (μg/ml)					Type of LVX resistance	Amino acid or nucleotide change ina:
	OXA	LVX	LVX + reserpine	GAT	GAT + reserpine		gyrA	gyrB	grlA	norAb
JCSC 6763	512	512	512	128	128	LHR	Ser84Leu + Ser85Pro		Ser80Phe + Glu84Lys	In362-IS256
JCSC 6764	512	256	256	32	32	LHR	Ser84Leu	Asp437Glu	Ser80Tyr + Glu84Lys	T389C
JCSC 6765	256	64	64	16	16	LHR	Ser84Leu		Ser80Tyr + Glu84Lys	T389C
JCSC 6766	512	8	8	2	2	LLR	Glu88Lys		Ser80Phe
JCSC 6767	256	8	8	2	2	LLR	Glu88Lys		Ser80Phe
JCSC 6768	256	4	2	2	1	LLR	Glu88Lys		Ser80Phe
JCSC 6769	512	0.5	0.5	0.125	0.125	LS			Glu84Gly
JCSC 6770	256	0.25	0.25	0.125	0.125	LS
JCSC 6822	128	0.25	0.25	0.125	0.125	LS				In346-A

^aNo changes were detected in grlB.

^bPortion of the norA gene from nucleotides 305 to 476 encompassing the −35 and −10 sequences in the promoter region representing the transcriptional start site described by Yoshida et al. (35)

FIG. 1.

Killing activities (log₁₀ CFU/ml) of LVX and GAT against LS, LLR, and LHR MRSA strains at exposure times of 5 to 120 min. Each test group contained three different strains. The data are presented as means with 1 standard deviation (bars). P values for comparisons of LVX and GAT treatments are as follows: *, P < 0.002; **, P < 0.02; ***, P < 0.03; ****, P < 0.05. The dashed lines designate nearly complete bactericidal activity (>99.9% killing) for each drug.

FIG. 2.

Time-kill assays for 2,048 μg/ml LVX and GAT against LS, LLR, and LHR MRSA strains. The dashed lines designate nearly complete bactericidal activity (>99.9% killing) for each drug.

FIG. 3.

PAEs (A) and relationship between PAEs and MICs (B) of LVX (3,000 μg/ml) and GAT (2,000 μg/ml) against LS, LLR, and LHR MRSA strains after 5 min of exposure to the antibiotics. Each test group comprised three different strains. The data are presented as means with 1 standard deviation (bars).