Materials and Methods

Animals—Twenty-four adult mixed-breed horses were used. All study horses were determined to be free of ocular disease via complete ophthalmic examination (including Schirmer tear test I, slit-lamp biomicroscopy, and indirect ophthalmoscopy). All study procedures were reviewed and approved by the Institutional Animal Care and Use Committee of the University of Tennessee.

Experimental protocol—Horses were randomly assigned to 1 of 3 groups. Group 1 (n = 8) received antimicrobials topically at 0 (first application), 2, 4, and 6 hours, and tear, cornea, and aqueous humor samples were collected at 8 hours after the first application. Group 2 (n = 8) received antimicrobials topically at 0, 2, 4, 6, and 10 hours, and samples were collected at 14 hours. Group 3 (n = 8) received antimicrobials topically at 0, 2, 4, 6, 10, and 14 hours, and samples were collected at 18 hours.

Only 1 eye of each horse was used. A 1-mL syringe attached to a 21-gauge lacrimal cannula was used to deliver 0.1 mL of ciprofloxacin (0.3%)^a and moxifloxacin (0.5%)^b solutions into the ventral conjunctival fornix, with a 5-minute waiting period separating administration of each. Horses were preassigned to alternating groups prior to enrollment into the study to receive either ciprofloxacin or moxifloxacin first for the duration of the study.

Sample collection—Tear samples were collected by placing a Schirmer tear test strip^c in the ventral conjunctival fornix for 30 seconds. Tear volume was determined by weighing the tear strip before and after tear collection. After tear sample collection, horses were euthanized for reasons unrelated to the present study. After euthanasia, aqueous humor samples were collected (0.2 mL) by aqueocentesis with a 1-mL syringe and a 25-gauge needle. Axial corneal buttons, including the epithelium, were collected with the aid of an 8-mm biopsy punch^d and corneal section scissors immediately after aqueous humor collection. All samples were placed in separate vials, weighed, and frozen at −80°C until analyzed.

HPLC—Ciprofloxacin and moxifloxacin concentrations were measured via HPLC with fluorometric detection. The system consisted of a separation module,^e C18 column (5-μm particle size; 3.9 × 150 mm),^f guard column,^g fluorescence detector,^h and computer equipped with software.ⁱ The mobile phase was a mixture of acetonitrile and 0.025M H₃PO₄ (pH, 3) with tetrabutylammonium hydroxide. The mixture was pumped at a starting gradient of 93% H₃PO₄ and 7% acetonitrile and was adjusted to 80% H₃PO₄ and 22% acetonitrile over 3.5 minutes. The mixture was held for 8 minutes in these conditions, then returned to initial conditions over 4 minutes. The flow rate was 1.0 mL/min throughout the analysis. The fluorescence detector was set at an excitation wavelength of 287 nm and an emission wavelength of 480 nm, with the gain at 1×. The column was maintained at ambient temperature (approx 23°C).

Previously frozen tear strips were minced in tubes, and 1 mL of mobile phase was added, followed by 50 μL of pipemidic acid (300 μg/mL; internal standard). Tubes were vortex-mixed for 30 minutes, and 20 μL of supernatant was injected onto the HPLC. Previously frozen corneas were thawed, weighed, and placed in 16 × 100-mm tubes. Fifty microliters of pipemidic acid and 1 mL of mobile phase were added to each tube. Each sample was homogenized and centrifuged at 1,600 × g for 10 minutes, and then 20 μL of supernatant was injected.

Aqueous humor samples were thawed and placed in microcentrifuge tubes, to which 50 μL of internal standard was added. Tubes were mixed with a vortex machine, and 20 μL of the solution was injected onto the HPLC. The retention time of moxifloxacin was 9.05 minutes, and that of ciprofloxacin was 6.8 minutes; pipemidic acid eluted at 2.8 minutes. The mean recovery of moxifloxacin and ciprofloxacin was 99%. The limit of quantification for all samples was 0.01 μg/mL.

Standard curves for tear strips were prepared by fortifying appropriate control samples with ciprofloxacin and moxifloxacin to produce a linear concentration range of 0.025 to 25 μg/mL. Intraday and interday assay precision was determined with 4 sets of control samples at 3 concentrations. The intraday coefficient of variation for assays was 0.37% to 0.66%; the interday coefficient of variation was 2.9% to 9.6%.

Statistical analysis—Data were assessed for normality by means of Kolmogorov-Smirnov test. To determine the effect of administration order (ie, 0.3% ciprofloxacin solution first vs 0.5% moxifloxacin solution first) on drug tissue concentration for each horse group, a Student t test was used for normally distributed and a Mann-Whitney rank sum test was used for nonnormally distributed data. A 1-way ANOVA was used to determine the effect of the 3 administration protocols on tissue drug concentrations when data were normally distributed, and a Kruskal-Wallis test was performed when they were not. Statistical significance was set at a value of P < 0.05 for all tests. A commercially available software program^j was used to perform all analyses.

Results

One corneal sample belonging to a horse in group 2, which received the 0.5% moxifloxacin before the 0.3% ciprofloxacin solution first, was lost during processing. The concentration of moxifloxacin in the cornea was significantly (P = 0.003) higher in group 1 (received antimicrobials at 0, 2, 4, and 6 hours) than it was in groups 2 (received antimicrobials at 0, 2, 4, 6, and 10 hours) and 3 (received antimicrobials at 0, 2, 4, 6, 10, and 14 hours) when measured at 8, 14, and 18 hours, respectively (Table 1). All aqueous humor concentrations of ciprofloxacin were less than the limit of quantification for the system used. Aqueous humor concentrations of moxifloxacin were less than detectable concentrations in 2 of the samples: 1 sample in group 2 and the other in group 3. Mean corneal concentrations of moxifloxacin (0.959 μg/mL) were significantly (P = 0.001) higher in group 2 when ciprofloxacin was administered first, compared with when moxifloxacin was administered first (0.542 μg/mL).

Significant differences in moxifloxacin concentration due to drug administration order were not detected in any other group or tissue, although the concentration was numerically (although not significantly [P = 0.583]) higher in group 3 when it was given before ciprofloxacin. Administration did not significantly affect the ciprofloxacin concentration either, but the corneal concentration was numerically (although not significantly [P > 0.350]) higher when ciprofloxacin was given before moxifloxacin. The power of all statistical comparisons, assuming a 10-fold difference as significant, was < 0.4.

Discussion

The results of the study reported here indicated that topical ocular application of 0.3% ciprofloxacin and 0.5% moxifloxacin solutions resulted in both drugs reaching clinically useful concentrations within tears when administered according to the regimens used. In vitro susceptibility of equine bacterial keratitis isolates to ciprofloxacin or moxifloxacin is unknown, with the exception of S equi (MIC for ciprofloxacin that will limit 100% of bacterial growth = 1.0 μg/mL).²² However, the S equi data, along with recently published MIC₉₀ values for the human ophthalmic pathogens Staphylococcus aureus (ciprofloxacin MIC₉₀ = 0.5 μg/mL; moxifloxacin MIC₉₀ = 0.094 μg/mL),²³ Streptococcus pneumoniae (ciprofloxacin MIC₉₀ = 0.75 μg/mL; moxifloxacin MIC₉₀ = 0.094 μg/mL),²³ and P aeruginosa (ciprofloxacin MIC₉₀90 = 0.125 μg/mL; moxifloxacin MIC₉₀ = 0.75 μg/mL),²⁴ suggest that if the antimicrobial susceptibility patterns of equine bacterial isolates were similar to those of humans, the tissue drug concentrations reached in tears by use of any of the 3 dosing regimens in our study should be effective in the treatment of most horses with bacterial keratitis. However, corneal concentrations of ciprofloxacin were quite similar to the ciprofloxacin MIC values for the most common gram-positive pathogens, whereas moxifloxacin concentrations were 8 to 19 times the moxifloxacin MIC₉₀ values for gram-positive organisms.

Considering the concentration of moxifloxacin was only approximately twice that of ciprofloxacin, the increased tissue concentration-to-MIC₉₀ ratio for moxifloxacin mostly reflects a lower MIC₉₀ for this drug. However, the higher concentration of moxifloxacin within the stroma is not surprising considering that moxifloxacin has greater lipophilicity and aqueous solubility than ciprofloxacin.⁸ Because fluoroquinolones are dose dependent in their activity, the greater lipophilicity and resultant higher concentrations in the cornea suggest that moxifloxacin is a better choice than ciprofloxacin in the treatment of corneal stromal abscesses, in which there is an intact epithelium. In the present study, the corneas of all horses possessed intact epithelial surfaces. It is likely that corneal stromal concentrations would be higher had the epithelium been disrupted.

The corneal drug concentrations of ciprofloxacin achieved in our study were generally lower than those obtained in human clinical studies. The highest mean corneal ciprofloxacin concentration was obtained in group 1 (with 4 applications, 2 hours apart) and was 0.836 μg/g of corneal tissue. Similar topical ciprofloxacin preparations applied in humans undergoing ocular surgery resulted in 0.6,¹³ 5.28,¹⁵ and 9.92¹⁴ μg/g of corneal tissue. The dosing schemes used in the human studies varied widely and differed from the protocols used in our study. More importantly, the interval to sample collection was usually shorter in the human studies than in the present study, and the human subjects had diseased rather than healthy eyes. These variations in study approaches make it difficult to draw any meaningful conclusions regarding interspecies comparisons; however, anatomic differences between the 2 species may also play a role in the apparent species differences.

Aqueous humor concentrations of ciprofloxacin were not detectable in the present study. Aqueous humor concentrations of ciprofloxacin in human clinical studies^12,14,25 have ranged from 0.135 to 1.13 μg/mL. A clinical study²⁶ in dogs achieved a concentration of 0.36 μg/mL. In our study, aqueous humor concentrations of moxifloxacin ranged from 0.02 to 0.06 μg/mL. Human aqueous humor concentrations of moxifloxacin ranged from 0.88 to 2.28 μg/mL, which was similar to the 1.42 to 1.75 μg/mL range obtained in rabbits.⁸ Again, these differences may reflect differences in methodology, subject health status, or species differences in anatomy and physiology (eg, corneal thickness, anterior chamber volume, and tear turnover rate).

In the present study, 0.3% ciprofloxacin and 0.5% moxifloxacin solutions were applied to the same eye of each horse, with 5 minutes separating the 2 treatments. Applying the 2 drugs to the same eye at approximately the same time allowed for the use of paired comparisons, which substantially decreased the sample size needed to detect a significant difference if one truly existed.¹³ Although the horses in our study were euthanized for reasons unrelated to our study, in situations in which collection of a small tissue sample could involve the destruction of large mammals such as horses, it is ethically important to keep the sample size to a minimum. In our study, the benefit of applying the 2 solutions to the same eye was not entirely fulfilled. The differences in tissue drug concentrations are relevant only in light of their respective MICs for the pathogens of interest. Once MIC values for specific equine corneal pathogens become available for ciprofloxacin and moxifloxacin, a decision regarding the theoretical efficacy of each drug can be made with the aid of our data.

Several potential problems are associated with applying the 2 solutions to the same eye. First, drug interactions are possible. Drug interactions in our study were avoided in part by applying each drug 5 minutes apart. In humans, an instilled drug is almost completely eliminated from the cul-de-sac within 5 minutes because of tear turnover.²⁷ However, this does not account for possible interactions between the drugs once they enter the tissue. Additionally, there are interactions between the drugs and the tissues that may alter their absorption. Little variation in epitheliotoxicity exists among fluoroquinolones, with the major differences relating to the presence of preservatives within a preparation.^28,29 The 0.3% ciprofloxacin formulation used in the present study contains 0.006% BAC, whereas the 0.5% moxifloxacin formulation does not. Although this concentration of BAC is very low, it can alter the corneal epithelium, and this effect increases with exposure time.³⁰ In a single-drop protocol, it would be expected that the presence of BAC in one solution but not the other would cause a considerable first-drug effect, by which the drug applied second diffuses more readily into the cornea than when it is applied first. However, because the study protocol involved repeated drug application and the effects of BAC are long lasting,³¹ the epithelial surface encountered by both drugs should be similar after the first application. Our results indicted there was a significantly higher concentration of moxifloxacin in cornea tissue when moxifloxacin was administered second in group 2. This difference was not detected in any other group, and there was not a pattern that would suggest a first-drug effect. It is also important to consider that these comparisons were made by use of 4 data points (4 receiving moxifloxacin first and 4 receiving ciprofloxacin first), and because 1 corneal sample was lost, the comparison for corneal tissue was made by use of only 3 data points in group 2. Given the available data and the significant finding in 1 group only, it may be that there was a first-drug effect, but we suspect it was a weak one.

Although it is not common to apply 2 fluoroquinolones to the same cornea in clinical situations, it is common practice to use other combinations of antimicrobials as well as atropine in horses with keratitis. Commercially available formulations of ophthalmic atropine typically contain 0.01% BAC as a preservative. Because of the severity of the anterior uveitis that occurs secondary to corneal ulceration in horses, it is not uncommon for atropine to be topically applied as frequently as every 6 hours. This concentration of BAC causes epithelial changes even at short exposure times.³⁰ Thus, it is likely that all horses treated for keratitis have an altered corneal epithelial surface. Consequently, although the concentration of moxifloxacin within the corneas in our study may have been influenced by the effect of BAC in the 0.3% ciprofloxacin formulation on the corneal epithelium, such an effect would be noticed clinically when atropine is used concurrently. Therefore, the corneal surface exposed to moxifloxacin in our study was probably more clinically relevant than had the moxifloxacin been given alone.

The intervals between drug administrations used in the present study were chosen to approximate those commonly used clinically in horses with complicated ulcerative bacterial keratitis. The dosing groups corresponded to 3 time points of a clinical situation in which there is a loading phase of drug administered every 2 hours for 4 doses followed by every 4 hours for doses administered thereafter. The data indicated there was no significant difference in tissue ciprofloxacin concentrations at any of the 3 time points, suggesting that after 4 doses administered every 2 hours, decreasing the frequency to every 4 hours thereafter should not decrease drug effectiveness. For moxifloxacin, however, there was a significant decrease in corneal drug concentrations after the first time point, even though values at all time points were higher than published MIC₉₀ values for human corneal pathogens. This suggests that if moxifloxacin is used for deep stromal disease in horses, drug administration at 2-hour intervals may be effective against some intermediately susceptible Staphylococcus spp and Streptococcus spp, whereas administration at 4-hour intervals may result in periods with tissue concentrations less than the MIC for these bacterial species.³²

Considered together, the data obtained in the present study suggested that tear concentrations of topically applied fluoroquinolones are significantly higher than published MIC₉₀ values for the most common corneal pathogens in horses. In general corneal concentrations of topical fluoroquinolones were close to the MIC₉₀s for these pathogens. However, because of the high variability in corneal concentrations that were achieved, fluoroquinolone concentrations may not reach therapeutic values in some horses even after administration every 2 hours when the ocular epithelium is intact.