Concentration and Bioavailability of Ciprofloxacin and Teicoplanin in the Cornea

Abstract

purpose. To investigate the concentration and bioavailability of ciprofloxacin and teicoplanin in the cornea.

methods. A biological assay was developed with corneal tissue used as a carrier for the antimicrobial. Concentration and biological activity were determined with a chemical assay and zone of inhibition (ZOI) around corneal samples with epithelial and endothelial surfaces in contact with the indicator organism. Patients undergoing penetrating keratoplasty received ciprofloxacin 0.3% or teicoplanin 1%.

results. There were good correlations between antimicrobial concentration and ZOI, when either filter paper or corneal discs were used (R ² > 92%). Of 33 patients, the mean (median) concentration of ciprofloxacin in the cornea was 1.37 mg/L (0.46 mg/L) and 1.89 mg/L (1.44 mg/L; bioassay) in the epithelial and endothelial orientations, respectively, and 14.87 mg/L (7.41) in the cornea and 0.51 mg/L (0.42) in the aqueous (chemical assay). For teicoplanin, the mean (median) concentration in the cornea was 9.58 mg/L (0 mg/L) in the epithelial and 4.78 mg/L (0 mg/L) in the endothelial orientations (bioassay). In the chemical assay, teicoplanin could not be detected in the cornea or aqueous at the lower limit of detection of 3.6 mg/L.

conclusions. The ZOI produced by corneal tissue provides a potential bioassay of antimicrobial activity and concentration. Although in contrast to teicoplanin ciprofloxacin shows good corneal penetration, with high endothelial-to-epithelial levels, only approximately 10% of measured levels in a chemical assay are available, according to a bioassay. Teicoplanin shows relatively poor corneal penetration through intact epithelium. These methods may be useful in evaluating the biological activity across the cornea of antimicrobials introduced into ophthalmic practice to deal with changing bacterial resistance.

The effectiveness of a topical antimicrobial agent in the treatment of microbial keratitis is in part dependent on the physicochemical properties of the antimicrobial solution and the structure of the cornea. ^{1 2 3 4 5 6 7 8 9 10 11} The concentration of the agent within the cornea is used as a measure of these properties. The amount of drug in the cornea has been determined both directly—that is, measuring the concentration of the drug in a homogenized cornea removed at surgery ^{12 13 14} —and indirectly, by measuring the concentration in the aqueous humor. ^{15 16 17 18} The concentration of an antimicrobial, however, does not necessarily equate to the activity and bioavailability of the drug, which may be less than 10% of the instilled amount. ^{1 3 19} In particular, protein binding, ²⁰ pH of the local environment and interaction with other chemicals are likely to affect the antimicrobial action of the agent. ^{1 3 19 20} These factors may be of even greater importance in the inflamed eye, particularly in cases of microbial keratitis, where there may be greater protein binding. Thus, while concentration of the drug is important, the biological activity of the drug in situ needs to be measured if the antimicrobial effectiveness is to be determined. ¹⁶  

Cahane et al. ¹⁹ measured corneal concentration of ciprofloxacin by using a bioassay that required that the cornea be digested. Although this assay provides an indication of activity, it does not necessarily determine the activity across the cornea; and, by digesting the cornea, the antimicrobial is extracted, thus potentially removing the biological factors that determine its activity in vivo. 

The fluoroquinolones such as ciprofloxacin are effective agents that are used to treat microbial keratitis. ^{15 16 20 21 22 23 24 25} Their effectiveness against bacteria such as streptococcus and strains of staphylococcus may be limited, ^{2 26 27 28} although the newer generation drugs, such as moxifloxacin and gatifloxacin, have enhanced activity against Gram-positive bacteria. ^{29 30} These agents, however, are not a panacea for the treatment of microbial keratitis, particularly with the emergence of resistant strains of staphylococci, streptococci, and Enterobacteriaceae. ^{30 31 32 33 34} New antimicrobials are usually developed to treat systemic or nonophthalmic infections before being introduced into ophthalmic practice—for example, vancomycin, teicoplanin, and linezolid. ^{4 27 28 34 35 36} Because chemical assays may not reflect biological activity within a tissue, an additional biological assay is needed to evaluate the activity of antimicrobials that are potentially usable in ophthalmic practice. 

The purposes of this study were therefore to develop a model to measure and compare the concentration and biological activity across the cornea of topically delivered ciprofloxacin and teicoplanin, by using chemical and biological assays. Ciprofloxacin and teicoplanin were selected, as they are very different antimicrobial agents. There is a wealth of published data on the penetration of ciprofloxacin into the cornea. It is representative of the fluoroquinolones and is still commonly used in many parts of the world for the treatment of microbial keratitis. ^{33 37 38} In contrast, teicoplanin has a much higher molecular mass (2 kDa) than ciprofloxacin (370 Da). It differs in its water solubility and octanol/water partition coefficient ³⁹ and tends to be highly protein bound, ²⁰ which suggests a poorer penetration into and through the cornea. 

Method

The concentration of antimicrobial in corneal and aqueous samples from patients was determined with a chemical assay. The size of the zone of inhibition (ZOI) around a sample of cornea was used as a measure of the biological activity, to determine the antimicrobial’s biological activity and, indirectly, its concentration. The concentration of the antimicrobial was then extrapolated from standardized curves of concentration and ZOI by using both filter paper and corneal samples. Although it is well established that the ZOI around a circular filter paper disc provides a good determinant of the antimicrobial concentration, it was necessary to determine whether the same holds true for a biological tissue such as cornea and secondly for a corneal quadrant, given that the corneas removed were divided into quadrants for analysis. This question was investigated first by comparing the ZOI of quadrants and circular discs of filter paper and second the behavior of corneal discs and quadrants of cornea discs with the same sized filter paper. Finally, because of the potential difference in behavior of a filter paper disc, depending on the orientation of the cornea, the corneal tissue was oriented with respect to the epithelium and endothelium in contact with the agar plate containing the indicator organism. 

Ethics approval was obtained from the local research and ethics committee, and the research adhered to the tenets of the Declaration of Helsinki. Informed consent was obtained from the subjects after explanation of the nature and possible consequences of the study, and the research was approved by the Department of Research and Development of the Royal Liverpool University Hospital. 

Preparation of Agar Plates with the Indicator Organism

Bacterial cultures were prepared on agar (Isosensitest, Oxoid; Unipath, Basingstoke, UK), Escherichia coli (ATCC 25922; American Type Culture Collection, Manassas, VA) for ciprofloxacin and Staphylococcus aureus “Oxford” (NCTC 6571; National Collection of Type Cultures, London, UK) for teicoplanin. These cultures had been spread onto the agar with a sterile swab and a rotary plater, adhering to standard methodology defined by the British Society for Antimicrobial Chemotherapy ⁴⁰ (BSAC) to produce a semiconfluent growth. Cultures were incubated for 16 to 18 hours in air at 37°C. 

Preparation of Standard Concentration Curves

Antimicrobial Concentration.

Ten different concentrations of ciprofloxacin (Bayer Pharmaceuticals, West Haven, CT; 0.06–20 mg/L) and teicoplanin (Aventis Pharma, Bridgewater, NJ; 0.1–50 mg/L) were prepared with sterile water from three starting solutions: 200, 20, and 1 μg mL⁻¹. These solutions were used for both the filter paper discs and corneal specimens. 

Filter Paper Discs: Quarter and Whole Discs.

Twenty microliters of each antimicrobial concentration was adsorbed onto four filter paper discs (6-mm diameter; Oxoid; Unipath) and allowed to air dry. Two discs for each concentration were divided into quarters (QD) with a sterile number 11 blade, and the remaining two were left whole (FD). For each set of discs of a given concentration, two plates were prepared with the same indicator organism. Each plate had one FD and four QD placed in a pentagonal shape, aiming to ensure that the distance between the pieces and the edge of the plates was equal. All the plates were incubated at 37°C for 16 to 18 hours. 

Porcine Corneas.

The central cornea was removed from non–heat-treated porcine eyes with a 6-mm trephine. These corneal discs were placed in dilutions of ciprofloxacin and teicoplanin, blotted dry, and then placed either endothelial or epithelial side down on the agar plate and incubated for 16 hours at 37°C. Six corneas (three endothelium down and three epithelium down) together with three filter paper discs were used for each concentration. The average of the three ZOI for each concentration and orientation (epithelium or endothelium in contact with plate) were used for the analysis. 

Human Corneas and Aqueous.

Fifteen human corneas were used (Manchester Eye Bank, Manchester, UK); two 6-mm diameter discs were trephined from each cornea, producing 30 discs. Each disc was washed three times in sterile normal saline before being suspended in the latter for 30 minutes to remove storage solution. Each disc was then divided into quarters, thereby producing eight quarter segments from each original cornea. This division was intended to match the quadrants of filter paper discs and corneas from the patients undergoing penetrating keratoplasty. One segment from each cornea was placed on a culture of E. coli (ATCC 25922) for 16 to 18 hours in air at 37°C, to determine whether any residual antimicrobial activity remained. There were no detectable ZOIs from any of the segments. 

While the control quarter segment from each cornea was being incubated overnight, the other seven quarter segments from each original cornea were allowed to soak overnight in normal saline. Each quarter segment was first blotted onto filter paper to remove excess saline and then soaked in differing concentrations of antimicrobial. Each corneal segment was taken at random from the original 15 corneas. After 15 minutes, the segment was then removed from the solution and again blotted onto filter paper to remove excess antibiotic solution. It was then placed onto the agar plate (Isosensitest; Oxoid) containing a bacterial culture of E. coli (ATCC 25922) for ciprofloxacin and S. aureus (NCTC 6571) for teicoplanin. ⁴⁰  

Collection and Preparation of Human Corneas, Tear Film, and Aqueous

Patients undergoing penetrating keratoplasty were included. They were excluded if the epithelium of the cornea was not intact or if they were on treatment with topical drugs. Topical antimicrobials were commenced 4.5 to 5 hours before surgery, comprising eight doses at 30-minute intervals, with the last dose administered approximately 40 to 50 minutes before the commencement of surgery. Control patients received no topical drops. Tear film samples were collected form both the treated and untreated eye by placing a 6-mm filter paper disc against the lower bulbar conjunctiva and then placing the disc on an agar plate containing the indicator organism. Samples were collected 15 and 30 minutes after the last application of the eye drop and then at the commencement of surgery before preparation of the eye with 5% povidone iodine. These discs were then place on the prepared agar plates. Approximately 40 to 50 μL of aqueous was removed at the commencement of surgery and kept stored at −80°C for analysis. 

After removal, the cornea was divided into four quadrants, two of which, in addition to an aqueous sample, were retained for histologic analysis and high-pressure liquid chromatography (HPLC). The remaining two segments were placed on the prepared agar plates and incubated at 37°C for 16 hours. To counter the effect of the iodine 5% ocular surface preparation, we used agar plates containing the neutralizing agent (3%) Tween 80. 

Measurement of ZOI

In all instances, the plates were incubated for 16 to 18 hours at 37°C in air. The ZOI was measured to the nearest 0.5 mm with a metric ruler. The ZOI for the QD (QDs) and corneal quadrants was measured from the apex of the quadrant radially across the quadrant to the edge of the ZOI. For the full disc (FD), the ZOI was measured as a diameter, across the whole of the ZOI around the disc. This measurement for the FD was then halved to give the radius of the ZOI, to obtain results comparable to the QD measurements. Where no ZOI could be detected, a reading of 3 mm was taken, which is the length of a quadrant, given that the diameter of the original corneal disc trephined was 6 mm. 

Measurement of the Chemical Concentration of the Antimicrobial

The concentrations of teicoplanin and ciprofloxacin were determined in the cornea by storing one quadrant of each cornea from patients undergoing penetrating keratoplasty and 40 to 50 μL of aqueous at −80°C. Corneal antimicrobial concentrations were assayed by HPLC or fluorescence polarization immunoassay (FPIA), by using methods based on those reported by Diamond et al. ²⁴ In brief, corneal samples were weighed, a volume of phosphate-buffered saline equivalent to the weight of the sample (1 g ≡ 1 mL) was added, and the samples were homogenized. After centrifugation (10,000g) for 2 minutes, the resultant supernatant was assayed for ciprofloxacin by HPLC, ²⁴ or for teicoplanin by FPIA. For FPIA, a commercially available kit was used, as instructed by the manufacturer (Seradyne Inc., Indianapolis, IN). The limits of detection were ciprofloxacin 0.8 mg/L and teicoplanin 3.6 mg/L. 

Results

Antimicrobial Concentrations and ZOIs

When filter paper discs (full and QDs) were used, there was a very good correlation (coefficient of determination R ² > 0.92) between the size of the ZOI and the concentration of antimicrobial for both ciprofloxacin (FD: R ² = 0.92, P < 0.0001, QD: R ² = 0.93, P < 0.0001) and teicoplanin (FD: R ² = 0.98, P < 0.0001, QD: R ² = 0.96, P < 0.0001; Fig. 1 ). 

The average difference in zone size between FD and QD was 0.75 mm (P = 0.008) and 0.27 mm (P = 0.009) for ciprofloxacin and teicoplanin, respectively. Although the difference appeared to increase with increasing concentration, it did not show any significant linear correlation for either ciprofloxacin (P = 0.25) or teicoplanin (P = 0.06). The difference in ZOI between full and QDs according to the concentration of ciprofloxacin and teicoplanin is shown in Figure 2 . It was not possible, however, to be certain that the difference in ZOI between an FD and QD was constant to enable the use of a correction factor. It was therefore, necessary to construct a concentration curve based on the values obtained with a QD. 

For both antimicrobials, there were also good correlations between the ZOI and antimicrobial concentration in pig corneas (R ² = 0.92, R ² = 0.96, P < 0.0001; Fig. 3 ). There was no significant difference in the ZOI between endothelial and epithelial orientations within the range of concentrations used (P = 0.73). 

Although there was a good correlation between a corneal (pig) and filter paper disc (FD) (R ² = 0.97), there was a significant difference (P = 0.01). In particular the difference between the corneal and filter paper disc, showed a nonconstant positive bias (P = 0.01), increasing with increasing concentration of ciprofloxacin (Fig. 4) . 

A constant correction factor could not therefore be used to determine the antimicrobial concentration in a cornea from the predicted value when a filter paper disc was used. It was therefore necessary to calculate the standardized curves with the corneas. In addition, because we could not determine whether there would be a difference between porcine and human corneas, standardized curves from human corneas were established for both ciprofloxacin and teicoplanin. Furthermore, because there was a difference between QD and FD (using filter paper discs), it was necessary to construct the standardized curves of the QD of human corneas. 

There were high correlations between the ZOI and antimicrobial concentration for both ciprofloxacin (R ² = 0.99) and teicoplanin (R ² = 0.99) with human corneal QD and FD (Fig. 5) . The concentration of antimicrobial could be determined from the following linear equations, which were used to indirectly determine the antimicrobial concentrations in the quarter corneal segments from the study population after penetrating keratoplasty.  

\[\mathrm{Log}\ {[}\mathrm{teicoplanin}{]}{=}0.44(\mathrm{ZOI\ of}\ 3.26)\]

 

\[\mathrm{Log}\ {[}\mathrm{ciprofloxacin}{]}{=}0.25(\mathrm{ZOI\ of}\ 7.24)\]

 

Tear Film Cornea and Aqueous Concentrations

Thirty-three patients were included. Because of the known penetration of ciprofloxacin into the cornea but the unknown behavior of teicoplanin in the cornea, 20 of the 33 patients received teicoplanin, 9 received ciprofloxacin, and 4 acted as control subjects. Indications for corneal transplantation included pseudophakic bullous keratopathy (n = 8), corneal scarring due to previous microbial keratitis (n = 2), interstitial keratitis (n = 1), herpetic keratitis (HSK; n = 6), keratoconus (n = 6), Fuchs dystrophy (n = 5), macular corneal dystrophy (n = 1), and lattice corneal dystrophy (n = 1); regraft due to endothelial failure (n = 2); and previous HSK (n = 1). The corneal discs diameters ranged from 7.25 to 8 mm, giving a maximum difference of 0.37 mm in the radii of the corneal quadrants (Table 1) . There was no obvious association between indication for keratoplasty and the inhibition zones. Neither the tear nor corneal samples of control patients showed any measurable inhibition zones (Table 1) . The same was true of the tear/conjunctiva samples from the fellow untreated eyes, which acted as the internal control (data not shown). 

Ciprofloxacin

The concentration of ciprofloxacin was 0.3% or 3000 mg/L. The concentration (indirectly determined from the ZOI) in the tear film at 15 minutes, 30 minutes, and at surgery after application of the last drop, were, 6.7, 0.4, and 0.9 mg/L, respectively—400- to 8000-fold dilutions (Table 1) . Three samples did not produce any measurable ZOI at 15 minutes, 30 minutes, and at surgery. The mean and median concentrations in the cornea with ZOI was 1.37 mg/L (2000 dilution) and 0.46 mg/L with epithelial orientation, and 1.89 mg/L (1500 dilution) 1.44 mg/L with endothelial orientation, respectively. HPLC showed that the mean corneal and aqueous concentrations were 14.87 mg/L (7.41) and 0.51 mg/L (0.42), respectively—200- and 7000-fold dilutions. Three of nine corneal and two of nine aqueous samples for HPLC were lost. The ZOI-to-HPLC ratio of concentration was 0.11 (0.25) and 0.10 (0.19) for epithelial and endothelial orientations, respectively. The ratio of the concentration of ciprofloxacin in the tear film at surgery to that at 15 minutes after the last drop was 10%; epithelium to tear film, 150%; and endothelium to epithelium, 140%. 

Teicoplanin

The concentration of administered teicoplanin was 1% or 10,000 mg/L (10 mg/mL). Between 85% (17/20) and 70% (14/20) of tear samples in the teicoplanin group produced measurable inhibition zones 15 and 30 minutes after application of the last drop, respectively (Table 1) . The concentration in the tear film at 15 minutes, 30 minutes, and surgery after application of the last drop, were 70, 66, and 29 mg/L, respectively—150- to 300-fold dilutions. Fifty-five percent (11/20) of corneas with epithelium down and 65% (13/20) of those with endothelium down did not display a measurable ZOI. The mean and median concentrations in the cornea was 9.58 and 0 mg/L epithelium and 4.78 and 0 mg/L endothelium—1000- and 2000-fold dilutions, respectively. With FPIA, teicoplanin could not be detected in the cornea or aqueous. The ratio of the concentration of teicoplanin in the tear film at surgery to that at 15 minutes after the last drop was 30%; epithelium to tear film, 30%; and endothelium to epithelium, 50%. 

Discussion

The corneal penetration of a topically applied antimicrobial drop is determined by the physicochemical properties of the antimicrobial solution and the structure of the cornea itself. In the healthy human eye, administration of drug topically involves immediate mixing of drug with the tear film with the induction of reflex lacrimation. The loss of drug from the precorneal area is a net effect of corneal and noncorneal absorption and tear secretion and drainage. The cornea contains many small aqueous pathways for low-molecular-mass molecules, but a limited number or larger paracellular aqueous pathways through which high-molecular-mass molecules can penetrate. To enter the cornea, the antimicrobial must diffuse directly through the hydrophobic intact corneal epithelium, since the aqueous paracellular pathways are closed off by tight intercellular junctions. ^{9 11} Thus, lipid soluble molecules diffuse through the cornea more readily than hydrophilic ones. The molecular mass of hydrophilic drugs is one of the most important factors in their corneal penetration. ^{41 42} The intact hydrophobic epithelium is thus relatively impervious to polar and hydrophilic compounds with molecular mass greater than 60 to 100 Da. ^{41 42 43} These barriers and the corneal permeability are likely to be significantly altered if the cornea is ulcerated or otherwise damaged. Although previous studies have measured corneal ¹⁹ and aqueous ^{15 16} antimicrobial levels with a bioassay, digested cornea has been used in these assays. Although it provides some indication of activity, this method does not determine the activity across the cornea. In addition, by digesting the cornea, the antimicrobial is extracted, thus removing the biological factors, which affect its activity in vivo. 

It is well established that the ZOI produced by an antimicrobial relates to the concentration of the agent. This fact underlies the basis of this study and under the conditions used, it is apparent that the ZOI provides a good indication of the concentration of antimicrobial. That is, between 92% to 98% of the change in ZOI was associated with a change in antimicrobial concentration. It was apparent that the size of the filter paper disc used to produce the ZOI is important, with FDs producing significantly larger net ZOI than QDs for both ciprofloxacin and teicoplanin. In addition, because it was not possible to be certain that the difference in ZOI between an FD and a QD was constant enough to enable its use as a correction factor, it was necessary to construct a concentration curve with values obtained with a QD. 

Similar to that with a filter paper disc, there was a good correlation between the ZOI and antimicrobial concentration in pig corneas. This result indicates that the ZOI in corneal discs also provides a good indication of the concentration of antimicrobial. Although there was a good correlation between corneal tissue and a filter paper disc, there was a significant difference. In particular the difference between the corneal and filter paper disc showed a nonconstant positive bias, increasing with increasing concentration of ciprofloxacin. A constant correction factor could not therefore be used to determine the antimicrobial concentration in a cornea from the predicted value of a filter paper disc. It was therefore necessary to construct standardized curves in corneas. In addition, because it was unclear whether there would be a difference between porcine and human corneas, standardized curves with human corneas were established for both ciprofloxacin and teicoplanin. Furthermore, because of the noted difference between the ZOI filter paper quarters and FDs, it was also necessary to construct the standardized curves with quarter segments of human corneas. There were very high correlations between the ZOI and antimicrobial concentration for both ciprofloxacin and teicoplanin in human corneal discs and there was no significant difference in the ZOI between the endothelial and epithelial orientations within the range of concentrations used. 

The concentration of administered ciprofloxacin was 0.3% or 3000 mg/L. The concentration in the tear film at 30 minutes and at surgery was 0.4 and 0.9 mg/L, respectively—that is, on average below the break points for ciprofloxacin. The concentration of administered teicoplanin was 1% or 10,000 mg/L. Twenty to 30% of the tear samples in the teicoplanin group did not produce measurable inhibition zones. The concentrations in the tear film at 30 minutes and at surgery were 66 and 29 mg/L, respectively, which suggests that teicoplanin has a reasonable retention in the tear film but relatively poor penetration into the intact cornea. 

The ratio between the biological assay and chemical assay for the corneal concentration of ciprofloxacin, was approximately 0.1, 10% of the concentration measured by HPLC. For teicoplanin, between 55% and 65% of the corneas did not display a measurable ZOI. The mean concentration of teicoplanin in the cornea, when using the ZOI, was 9.58 mg/L in the epithelial orientation and 4.78 mg/L in the endothelial orientation, but with median concentrations of 0 mg/L. Although these levels would be adequate to treat wild-type strains of staphylococci (average MICs of 0.5–2 μg/mL), ⁴⁴ or Streptococcus pneumoniae and the viridans group streptococci (MICs of 0.25–4 μg/mL), ⁴ these levels may not be sufficient to treat reported developing resistance among the coagulase-negative staphylococci (MIC ≥ 8 μg/mL), or Staphylococcus haemolyticus strains with MIC ≥ 64 μg/mL. ⁴⁵ Only half of the corneas showed teicoplanin activity through an intact epithelium. 

Teicoplanin was detected in the cornea or aqueous by FPIA. In 11/20 instances, this corresponded with the biological assay that failed to detect teicoplanin activity on the corneal tissue; however, in the remaining 9/20 extrapolated biological concentrations were above the expected lower limit of detection for FPIA. Negative results were most likely attributable to small samples (corneal quadrants) which when eluted into buffer resulted in a significant dilution to concentrations below the lower limit of detection for the assay. In the corneas of those patients in whom it was present, the teicoplanin zones and concentrations in the epithelium were greater than those in the endothelium, reflecting a reduced transcorneal penetration of teicoplanin and possible retention by the epithelium. The absence of detectable teicoplanin in the aqueous or cornea would not support a loss of teicoplanin through the cornea into the aqueous or retention within the cornea. The variance of aqueous and corneal ciprofloxacin levels determined by both the chemical assay and bioassay is in keeping with the suggestion of McDermott et al., ¹³ and Yalvac et al., ¹⁸ that there is a variation among patients in the efficiency of eluting the drug from the human corneal stroma. Although the differences in the sizes of the corneal discs (7.25–8 mm) probably accounts also for the variation in measured levels of antimicrobial, this effect is likely to be very small, given that the maximum difference in radii of the corneal quadrants was 0.37 mm. Although the epithelium was intact in all the corneas of the patients undergoing penetrating keratoplasty, the differences in their histology and function—for example, scarring and vascularization to endothelial failure and edema—were likely to have contributed significantly to the variability of measured corneal concentrations. 

The slightly higher concentration of ciprofloxacin in an endothelial orientation, is consistent with previous studies that show good ciprofloxacin penetration into the anterior chamber. ^{17 18} The factors that contribute to this high permeability include the low molecular mass and its lipophilicity, allowing it to cross the through the hydrophobic intact corneal epithelium via the transcellular route. In contrast, the high polarity and high water solubility of teicoplanin prevents it from mixing well with the lipid phase constituents of the epithelium. Additional factors such as the effect of the preserving agent for ciprofloxacin, benzalkonium chloride, has been shown to be an absorption promoter that facilitates drug penetration across the epithelial barrier. ⁴¹ For example, Podder et al. ⁴⁶ reported that benzalkonium chloride increased the ocular absorption of timolol after instillation in rabbits by approximately 80%. The effects of some absorption promoters increase with an increase in the molecular mass of hydrophilic drugs, possibly by increasing the contribution of the few large paracellular pathways, which suggests that penetration of teicoplanin, being a very large molecule, could be enhanced by using a drop formulation containing an absorption promoter. This, however, may not apply in the ulcerated state, as the corneal epithelium of patients with ulcerative keratitis is already disrupted, making the paracellular pathways accessible. That is, loss or disruption of the epithelial layer greatly increases corneal permeability, particularly of highly water-soluble drugs, and penetration is further enhanced during inflammation. ¹ Thus, it is likely that teicoplanin levels in corneas of patients with corneal ulcers would be higher than those found in this study. 

The main methods that have been be used to assay the teicoplanin and ciprofloxacin concentration in serum and other fluids including aqueous humor are HPLC, FPIA, and bioassay with Bacillus subtilis. ^{47 48 49 50} It thus remains unclear whether the differences between the measured levels of ciprofloxacin and teicoplanin reflect differences in the bioavailability or the use of different assays. The effectiveness of a topical antimicrobial agent in the treatment of a microbial keratitis is determined by its physicochemical properties and the structure of the cornea. Although the concentration of the antimicrobial within the cornea is a measure of these properties, the concentration of an antimicrobial does not necessarily equate to the activity and bioavailability of the drug, which may be less than 10% of the instilled amount. ^{1 3} The bioavailability is determined by other factors such as protein binding, ²⁰ pH of the local environment, and interaction with other chemicals, which may affect the antimicrobial action of the agent. ^{1 3} It is likely that these factors are of even greater importance in cases of microbial keratitis, where there may be greater protein binding. Thus, while concentration of the drug is important, the biological activity of the drug in situ, needs to be measured if the antimicrobial effectiveness is to be determined. We found that using the inferred concentration from ZOI, to the concentration measured by HPLC, that the bioavailability was approximately 10% in keeping with previous reports. ^{1 3} There are, however, significant limitations with the inference of concentrations determined with the bioassay. In the bioassay used in this study, ZOI of growth around a quadrant of cornea served as an indicator of antimicrobial activity. In such systems small changes in the agar (depth or different manufacturer batch) density of the indicator organism, application of the corneal segment, and incubation conditions can alter the ZOI to make the antimicrobial agent appear more or less active. In addition, measuring zones of inhibition is subjective and liable to intra- and interobserver error. Although these systems are used commonly in microbiologic laboratories to determine antimicrobial susceptibilities for clinical treatment and control samples are used to standardize the method, these controls may not be adequate when using a biological system to infer concentrations of antimicrobials in tissue after topical administration. 

We did not perform a similar measure for aqueous levels and as the aqueous environment is likely to be very different from the cornea, it would not be correct to infer a similar aqueous bioavailability. Nevertheless, based on the report of Kim et al., ¹⁵ the biological activity in the aqueous appears to be equivalent to levels measured with HPLC. They measured aqueous penetration and biological activity of moxifloxacin 0.5% ophthalmic solution and gatifloxacin 0.3% solution in patients who had undergone cataract surgery patients. ¹⁵ Mean concentrations in aqueous humor obtained via HPLC analysis for moxifloxacin (which has higher lipophilicity and ocular penetration in comparison to the other fluoroquinolones ^{15 51} ) and gatifloxacin were 1.80 and 0.48 μg/mL respectively, while the microbiologic dilution bioassay of the aqueous humor samples based on inhibitory activity gave equivalent estimated concentrations for moxifloxacin and gatifloxacin of 2.1 and 0.4 μg/mL, respectively. ¹⁵  

Engel et al. ¹⁶ compared 16 different bioassays for the measurement of antimicrobials in fluids such as aqueous humor. They found that sensitivities in water for ciprofloxacin were 0.12 mg/L but these improved fourfold when the fluid for dilution was aqueous or normal saline as used. Although there have been no studies using the method we report, Engel et al. ¹⁶ reported that the best results were obtained when the appropriate indicator organism was applied to the surface of the agar, as in this study. 

Further evaluation of the proposed bioassay necessitates performing chemical assays in parallel to the bioassay using known concentrations of antimicrobials. Despite these limitations, it is apparent that measurement of antimicrobials in the cornea by a chemical assay may not fully reflect the bioavailability of the agent. 

Teicoplanin persisted in the tear film for longer than ciprofloxacin maintaining levels well above the break point. Conversely, whereas the levels of teicoplanin were higher on the epithelial side in those patients in whom it was detected (11/20), the levels of ciprofloxacin on the endothelial surface relative to teicoplanin were much higher. In particular, there was a significant linear correlation between the concentration of ciprofloxacin in the tear film at surgery and concentration at the endothelial side of the cornea (R ² = 0.75, P = 0.002) and a nonsignificant trend for the epithelial side (R ² = 0.35, P = 0.09). The result suggests that the reduced permeability of the cornea to teicoplanin contributes to greater levels in the tear film, particularly as in nine of our patients it was not found in the cornea. In contrast the relative permeability of the cornea to ciprofloxacin contributed to the lower levels in the tear film. Although there are no published data measuring teicoplanin in the cornea, Cahane et al. ¹⁹ measured vancomycin levels in the cornea with a bioassay. Patients received topical vancomycin (33 mg/mL) up to 15 minutes before undergoing penetrating keratoplasty. The corneas were digested, and cumulative concentrations of vancomycin were determined by a bioassay technique using B. subtilis (ATCC 6633) on an antibiotic medium (No. 5; Difco, Detroit, MI). They reported a corneal tissue concentration of 46.70 (SEM 4.11) μg/g corneal tissue. ¹⁹ Although these levels are 5- to 10-fold higher than we found, their patients received the last drop 15 minutes before surgery and the concentration of vancomycin was threefold higher than that of teicoplanin (10 mg/mL) in our study. Nevertheless, it suggests that not all the teicoplanin in the cornea is detected by using the bioassay we describe. It remains to be determined whether any retained teicoplanin in the cornea is in is in fact bound or in a biologically active state, which would not be determined if the cornea is digested. 

The ZOI produced by corneal tissue provides a potential bioassay of antimicrobial activity and concentration. Although in contrast to teicoplanin ciprofloxacin shows good corneal penetration, with high endothelial-to-epithelial levels, only approximately 10% of the measured levels using a chemical assay are available according to a bioassay. Teicoplanin shows relatively poor corneal penetration through an intact epithelium. The proposed methods may be useful to evaluate biological activity across the cornea of other antimicrobials introduced into ophthalmic practice to deal with changing bacteria resistance patterns. 

 Submitted for publication November 23, 2008; revised December 27, 2008; accepted April 20, 2009.

 Disclosure: S.B. Kaye, None; T. Neal, None; S. Nicholson, None; J. Szkurlat, None; S. Bamber, None; A.C. Baddon, None; S. Anderson, None; K. Seddon, None; N. Dwyer, None; A.M. Lovering, None; G. Smith, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Concentration of antimicrobial and ZOI using a whole (FD) or quarter (QD) disc of filter paper. (A) Ciprofloxacin FD and QD; (B) teicoplanin FD and QD.

Figure 2.

 

Differences in ZOI between FDs and QDs. (A) Ciprofloxacin, (B) teicoplanin.

Figure 3.

 

ZOI (endothelial orientation) and antimicrobial concentration using pig corneas. ZOI = 12.04 + 8.69 Log [Cip] mg/L.

Figure 4.

 

Differences in ZOI produced by cornea (pig) and filter paper discs.

Figure 5.

 

ZOI and antimicrobial concentration found with the use of QDs of human corneas. Zone size is the diameter of the ZOI. (A) Teicoplanin, (B) ciprofloxacin.

Table 1.

 

Concentration of Teicoplanin and Ciprofloxacin in Patients Undergoing Penetrating Keratoplasty, Determined by the Two Assays

Patient	Indication	Chemical Assay			Bioassay
								Cornea			Tear Film
				Aqueous	Cornea Whole	Epithelium	Endothelium	15 min	30 min	Surgery
	Controls
1	PBK	N	n	0.00	0.00	0.00	0.00	0.00
2	KC	N	n	0.00	0.00	0.00	0.00	0.00
3	KC	N	n	0.00	0.00	N/A	N/A	0.00
4	PBK	N	n	0.00	0.00	0.00	N/A	0.00
	Teicoplanin
1	Fuchs	N	n	45.17	9.80	0.00	212.89	0.00
2	PBK	N	n	8.63	8.63	99.32	26.15	38.29
3	PBK	N	n	16.31	7.60	67.84	31.65	21.62
4	PBK	N	n	0.00	0.00	67.84	N/A	N/A
5	RG (HSK)	N	n	0.00	0.00	26.15	N/A	N/A
6	Scar (MK)	N	n	0.00	0.00	31.65	N/A	99.32
7	HSK	N	n	0.00	0.00	21.62	21.62	21.62
8	PBK	N	n	7.60	12.64	145.41	82.08	0.00
9	KC	N	n	0.00	0.00	212.89	N/A	0.00
10	PBK	N	n	0.00	0.00	99.32	46.33	0.00
11	Scar (MK)	N	n	0.00	0.00	31.65	0.00	N/A
12	HSK	N	n	0.00	0.00	0.00	0.00	N/A
13	RG (EF-IK)	N	n	66.19	39.77	99.32	212.89	21.62
14	RG (EF-PBK)	N	n	5.89	0.00	175.94	21.62	31.65
15	PBK	N	n	14.36	8.63	67.84	67.84	N/A
16	KC	N	n	0.00	0.00	46.33	46.33	N/A
17	KC	N	n	0.00	0.00	0.00	14.76	0.00
18	HSK	N	n	0.00	0.00	46.33	21.62	0.00
19	Fuchs	N	n	3.54	0.00	99.32	31.65	99.32
20	HSK	N	n	23.90	8.63	67.84	212.89	67.84
	Mean			9.58	4.78	70.33	65.64	28.66
	SD			17.49	9.35	57.95	76.14	35.84
	Median			0	0	67.84	31.65	21.62
	Ciprofloxacin
1	Fuchs	1.11	20.9	0.61	1.92	0.00	0.00	0.00
2	HSK	0.46	N/A	0.34	1.44	3.61	1.89	0.20
3	Fuchs	N/A	N/A	1.55	1.55	0.00	0.00	0.14
4	Fuchs	1.1	7.7	4.86	3.65	4.98	0.10	0.08
5	MD	0.31	24.1	0.00	0.00	0.00	N/A	0.14
6	HSK	0.1	18.3	0.46	0.46	13.11	0.00	0.99
7	KC	0.19	5.7	0.00	0.00	N/A	0.00	0.00
8	HSK	N/A	N/A	4.52	8.02	24.99	0.99	6.88
9	Lattice	0.28	12.5	0.00	0.00	N/A	N/A	0.00
	Mean	0.51	14.87	1.37	1.89	6.67	0.37	0.94
	SD	0.42	7.41	1.94	2.59	9.33	0.70	2.25
	Median			0.46	1.44	3.61	0.00	0.14

 Tear film samples were collected at 15 minutes and 30 minutes following application and at commencement of surgery. For teicoplanin, the level was not detected (n) at 3.6 mg/L. Tear film levels from fellow eyes were zero and are not shown. N/A, specimen not available; HSK, herpetic keratitis; Fuchs, Fuchs dystrophy; MD, macular dystrophy; KC, keratoconus; Lattice, lattice dystrophy; EF, endothelial failure; PBK, pseudophakic bullous keratopathy; IK, interstitial keratitis; scar and regraft (RG), original diagnosis and indication.