To compare the bactericidal activity of besifloxacin, moxifloxacin and gatifloxacin and determine the contribution of the preservative benzalkonium chloride (BAK) to bactericidal activity.
Time–kill experiments were performed against four species (n = 12) with besifloxacin, moxifloxacin and gatifloxacin, in the presence or absence of BAK, at t = 0, 5, 15, 30, 45, 60, 120 and 360 min, according to standard CLSI methods.
In the presence of BAK, bactericidal activity was observed within 5 min, regardless of the fluoroquinolone tested. The bactericidal activity of BAK was unaffected by the concurrent presence of besifloxacin and rapid killing (within 5 to 15 min) was not observed at BAK concentrations below 50 mg/L. However, when tested without BAK, besifloxacin was bactericidal in as little as 45 min, while moxifloxacin and gatifloxacin required at least 120 min; besifloxacin kill rates against fluoroquinolone-susceptible and -resistant strains were at least 2- to 4-fold faster than those of gatifloxacin or moxifloxacin.
Besifloxacin was the most rapidly bactericidal fluoroquinolone tested, followed by gatifloxacin and moxifloxacin, both of which had similar activity. Our studies demonstrate that the previously reported rapid in vitro killing by gatifloxacin formulations was probably due to the concurrent presence of 50 mg/L BAK, which is much higher than the 3.2 mg/L BAK observed in human tears 1 min after instillation of ophthalmic gatifloxacin solutions [Friedlaender MH, Breshears D, Amoozgar B et al. The dilution of benzalkonium chloride (BAK) in the tear film. Adv Ther 2006; 23: 835–41].
Bacterial infections of the ocular surface are frequently treated empirically with broad-spectrum fluoroquinolones, such as gatifloxacin, moxifloxacin and the newly approved besifloxacin. While moxifloxacin ophthalmic solution, 0.5%, is preservative-free, gatifloxacin ophthalmic solution, 0.3%, and besifloxacin ophthalmic suspension, 0.6%, both contain benzalkonium chloride (BAK) at 0.005% and 0.01%, respectively, as a preservative. BAK is a quaternary ammonium salt that can disrupt membrane lipid bilayers.1 Due to its biocidal activity and safety profile, BAK is widely used as a preservative in ophthalmic products typically at concentrations of 0.015–0.05%. Although some authors have studied the combined effect of BAK and fluoroquinolones that are used ophthalmically, the actual contribution of BAK to the in vitro and in vivo activity of fluoroquinolones has been debated.2,3
Previous work from our laboratories investigated the bactericidal activity of besifloxacin against Staphylococcus aureus, Staphylococcus epidermidis, Streptococcus pneumoniae and Haemophilus influenzae. The results showed that, for 91.6% of isolates tested, besifloxacin had the lowest MICs and was the most rapidly bactericidal agent when compared with moxifloxacin and ciprofloxacin. However, those studies did not include gatifloxacin or BAK, and measured bactericidal activity only 2, 4, 6 and 24 h after the addition of drug.4
Here we report the results from time–kill experiments performed to compare the bactericidal activity of besifloxacin with that of moxifloxacin and gatifloxacin at timepoints as early as 5 min. In addition, the contribution of BAK to bacterial kill kinetics was evaluated.
Materials and methods
Twelve bacterial isolates were selected from the Eurofins Medinet strain collection to include fluoroquinolone-susceptible, fluoroquinolone-resistant or fluoroquinolone-non-susceptible clinical isolates.4 All strains were ocular clinical isolates, except for the fluoroquinolone-resistant S. pneumoniae and the fluoroquinolone-non-susceptible H. influenzae isolates. Staphylococcal isolates included methicillin-susceptible and -resistant strains, while all S. pneumoniae isolates were penicillin susceptible and all H. influenzae isolates were β-lactamase negative.
MIC determination was done by the broth microdilution method according to CLSI reference methods.5 Besifloxacin and BAK were obtained from Bausch & Lomb, Inc. (Rochester, NY, USA), while moxifloxacin and gatifloxacin were purchased from TREK Diagnostic Systems, Inc. (Cleveland, OH, USA).
Time–kill assays were performed according to CLSI methods.6 To ensure an impartial comparison of besifloxacin, moxifloxacin and gatifloxacin, a fluoroquinolone concentration of 4× MIC was used in each experiment; the 1× MIC for each strain is indicated in Figure 1. BAK concentrations were 50 and 100 mg/L (0.005% and 0.01%, respectively) or 2× (8 mg/L), 4×, 8× and 16× MIC as indicated in the legends to Figures 1 and 2, respectively. Growth-phase bacterial cultures at 5 × 105 to 5 × 106 cfu/mL were treated with one of the fluoroquinolones, BAK or a combination of the two. After 0, 5, 15, 30, 45, 60, 120 or 360 min of exposure, samples were serially diluted, plated on solid medium, incubated for 20–24 h and surviving cells enumerated. Bactericidal activity was defined as a ≥3 log decrease in cfu/mL relative to the bacterial concentration at 0 min. Cultures not exposed to any drugs served as no-treatment controls.
Besifloxacin was bactericidal for all four isolates of S. aureus within 45–60 min (Figure 1a–d). Moxifloxacin and gatifloxacin were similar to each other in their kill rate, required 120 min or longer to reach a ≥3 log decrease in viable cells, and were not always bactericidal within the 6 h study period.
Bactericidal rates for S. epidermidis ranged from 45–120 min for besifloxacin to 120–360 min for moxifloxacin and gatifloxacin (Figure 1e–h). The latter two drugs had kill kinetics similar to each other for methicillin-susceptible S. epidermidis isolates, while gatifloxacin was faster than moxifloxacin against methicillin-resistant S. epidermidis strains. Except for the pair of fluoroquinolone-susceptible/fluoroquinolone-resistant methicillin-susceptible S. epidermidis isolates, there was no apparent correlation between the rate of staphylococcal killing by any of the three drugs tested here and the methicillin- or fluoroquinolone-resistance phenotype.
The cfu count for the fluoroquinolone-susceptible isolate of S. pneumoniae was reduced to baseline after 120 min by besifloxacin, while the two comparator drugs did not show bactericidal activity at up to 6 h (Figure 1i). None of the three fluoroquinolones achieved a ≥3 log decrease in cfu/mL against the fluoroquinolone-resistant isolate within 360 min (Figure 1j).
Against H. influenzae, besifloxacin was bactericidal within 120 min for the fluoroquinolone-susceptible and the fluoroquinolone-non-susceptible isolate (Figure 1k and l). Moxifloxacin and gatifloxacin both showed bactericidal activity against the fluoroquinolone-susceptible and fluoroquinolone-non-susceptible isolates after 6 h.
To determine the effect of BAK, the time–kill experiments described above were also performed in the presence of BAK at concentrations found in the ophthalmic formulations. The 12 test strains were incubated in the presence of 50 mg/L BAK, 100 mg/L BAK or 4× MIC of any of the three fluoroquinolones plus 50 or 100 mg/L BAK (Figure 1). Regardless of the bacterial species, antibacterial resistance phenotype or fluoroquinolone tested, the addition of BAK led to a very rapid drop in bacterial viability below the lowest detectable levels of 102 cfu/mL within 5 min.
To test if the bactericidal activity of BAK is concentration-dependent, we performed time–kill experiments with the fluoroquinolone-susceptible methicillin-susceptible S. aureus and fluoroquinolone-resistant methicillin-resistant S. aureus isolates in the presence of 2×, 4×, 8× and 16× MIC of BAK (where 1× MIC = 4 mg/L). Besifloxacin was added, where applicable, at 4× MIC to determine whether the fluoroquinolone influences the rate of BAK-induced cell death. The results for the fluoroquinolone-susceptible methicillin-susceptible S. aureus isolate show that the bactericidal activity of BAK increased with its concentration (Figure 2); similar results were obtained for the fluoroquinolone-resistant methicillin-resistant S. aureus isolate (data not shown). While BAK at 4× MIC required 6 h to reduce the number of viable cells to the lowest detectable limits, only 45 or 15 min were required in the presence of 8× or 16× MIC of BAK, respectively. These results confirm earlier experiments (Figure 1a) that show rapid bactericidal activity of BAK at concentrations of 50 mg/L or higher. The data also show that the presence of 4× MIC besifloxacin does not increase or decrease the rate of bacterial cell death due to BAK at the concentrations tested here.
Previous studies have indicated that besifloxacin is more potent and faster in its bactericidal activity than moxifloxacin and ciprofloxacin.4 The results presented here show that, in general, besifloxacin kills bacteria more rapidly and at lower concentrations than both gatifloxacin and moxifloxacin, which are very similar to each other in their time–kill kinetics.
The addition of 50 mg/L BAK to either one of the three fluoroquinolones resulted in a decrease in viable cells to the lowest limit of detection within 5 min. Therefore, the rapid in vitro bactericidal action of gatifloxacin ophthalmic solution, 0.3%, when compared with the preservative-free moxifloxacin ophthalmic solution, 0.5%, that has been reported by Callegan et al.2 can be attributed to the action of BAK, not the fluoroquinolone.
While Kowalski et al.3 examined the in vitro bactericidal activity of BAK with an initial timepoint of 1 h, our data and those reported by Callegan et al.2 demonstrate that BAK-induced killing occurs within 5–15 min. In addition, BAK concentrations in human tears dropped to 3.2 mg/L within 1 min following the instillation of a 35 μL drop of gatifloxacin ophthalmic solution.7 Taken together, these studies indicate the importance of including early timepoints when assessing the potential in vivo relevance of BAK's bactericidal activity.
Previous studies have examined potential interactions between fluoroquinolones and BAK using various methods.3,8 However, our own study was not designed or intended to address this issue conclusively. The results presented here demonstrate that the rate of BAK-induced killing is concentration-dependent and does not increase in the presence of a fluoroquinolone. This finding is not surprising considering that BAK kills rapidly by non-specifically disrupting membranes and proteins, while killing by fluoroquinolones requires an active metabolism.1,9 Therefore, a cell already affected by the lethal action of BAK would not be expected to be able to initiate fluoroquinolone-induced cell death.
While BAK is diluted within minutes on the surface of the eye, the residence time of besifloxacin is on the scale of several hours. Proksch et al.10 reported a total besifloxacin concentration of 610 mg/L in tear fluid after the instillation of a single drop of besifloxacin ophthalmic suspension, 0.6%, to healthy human subjects. After 24 h, the total besifloxacin concentration was ∼1.6 mg/L, which is well above the MIC for all isolates tested here, with the exception of the fluoroquinolone-non-susceptible isolate of H. influenzae.
Two clinical bacterial conjunctivitis trials that compared besifloxacin ophthalmic suspension, 0.6%, with a vehicle control (lacking besifloxacin but otherwise containing the same components, including BAK) showed statistically significantly higher microbial eradication and clinical resolution rates for the besifloxacin-treatment groups.11,12 Although the clinical relevance of BAK's in vitro bactericidal activity remains to be determined, the in vitro bactericidal activity reported here for besifloxacin is consistent with the clinically confirmed superiority of besifloxacin ophthalmic suspension to its BAK-containing vehicle.
This work was supported by a research grant from Bausch & Lomb, Inc., Rochester, NY.
W. H., C. K. H., C. M. S. and T. W. M. are employees of Bausch & Lomb, Inc., Rochester, and C. M. P. is employed by Eurofins Medinet, Inc. None of the authors owns stock or options in Bausch & Lomb, Inc. or Eurofins Medinet, Inc.
Parts of this study were presented in poster form at the Fiftieth Interscience Conference on Antimicrobial Agents and Chemotherapy, Boston, MA, 2010 (Poster A1-670).
We would like to thank Joe Blondeau for editorial comments.