Pharmacokinetics of Meropenem for Use in Bacterial Keratitis

Abstract

Purpose: To investigate the toxicity and corneal pharmacokinetics of meropenem as a potential antimicrobial for bacterial keratitis.

Methods: Corneal epithelial cell and keratocyte toxicity was investigated using methyl thiazolyl tetrazolium (MTT) and LIVE/DEAD assays. The penetration of meropenem through the human cornea was measured using an artificial anterior chamber. In one group of corneas, the epithelial and endothelial layers were removed and in a second group these layers were left intact. We applied 50 μL (10 mg/mL) meropenem to the corneal surface and collected samples in the anterior chamber from 45 minutes up to 24 hours. Meropenem concentrations were estimated with a bioassay and HPLC.

Results: Meropenem had significantly higher cellular metabolic activity (MTT assay) at both 5 mg/mL and 2.5 mg/mL compared with moxifloxacin (P = 0.029 and P = 0.018, respectively), with 96% cell viability (LIVE/DEAD assay). The measured values for meropenem concentrations in corneal and aqueous samples were significantly higher using a bioassay than with HPLC (P = 0.004). For both intact and denuded corneas, the concentrations in the anterior chamber increased from 0.48 μg/mL (SD 0.89) and 0.89 μg/mL (SD 0.81) to 6.35 μg/mL (SD 0.81) and 13.48 μg/mL (SD 14.82) using HPLC, and from 0.68 μg/mL (SD 1.50) and 1.31 μg/mL (SD 1.55) to 47.03 μg/mL (SD 5.51) and 43.69 μg/mL (SD 27.22) measured with a bioassay.

Conclusions: Meropenem has very low toxicity in vitro. It has good corneal penetration, achieving anterior chamber concentrations above MIC₉₀ for bacteria such as Staphylococcus aureus, Pseudomonas aeruginosa, streptococci, coagulase-negative staphylococci, and the Enterobacteriaceae.

Bacterial keratitis is a serious condition associated with significant visual morbidity.^1–5 There has been a trend in several countries toward an increase in antimicrobial resistance of bacteria causing keratitis.^6–8 Emerging resistance has been noted to the latest fourth-generation fluoroquinolones moxifloxacin and gatifloxacin.^7,8 In addition to other potential treatments, therefore, there is a need to evaluate antimicrobials that can be delivered topically to treat keratitis. 

We previously reported the susceptibility of bacterial isolates, determined by minimum inhibitory concentrations (MIC), from 772 cases of keratitis to commonly used and potentially novel antimicrobials.⁹ Meropenem was identified as a potential broad-spectrum antimicrobial for the treatment of keratitis with excellent in vitro activity. The MIC₉₀ values of meropenem were 0.08 μg/mL to Staphylococcus aureus, 0.25 μg/mL to Pseudomonas aeruginosa, 0.054 μg/mL to streptococci, and 0.094 μg/mL to Enterobacteriaceae.⁹ Meropenem is a member of the carbapenem class of antimicrobials, which also includes imipenem, ertapenem, and doripenem. It is a β-lactam antimicrobial that has an exceptional range of activity against both gram-positive and -negative pathogens, including extended-spectrum beta-lactamase and AmpC-producing Enterobacteriaceae. It has particular efficacy in the treatment of bacterial meningitis and skin, bone, gastrointestinal, respiratory, and urinary tract infections.¹⁰ Meropenem is currently approved by the US Food and Drug Administration to treat skin infections, intra-abdominal infections, and bacterial meningitis.¹¹ There are also a small number of case reports supporting the safe use of intravitreal meropenem to treat endophthalmitis resistant to other antimicrobials.^12–14 To our knowledge, however, there are no reports of the ocular pharmacokinetics of topically applied meropenem. This study investigates the in vitro toxicity and pharmacokinetics across the cornea of meropenem following topical delivery. 

Methods

Cell Culture and Treatment

Cultures of human keratocytes (HKs) were established from the rims of donor corneo-scleral discs that had been used for corneal transplantation using cells from passages 5 to 10 as previously described.¹⁵ Human corneal epithelial cells (HCEs), immortalized by the SV40 virus, were kindly provided by Araki-Sasaki, PhD (Kinki Central Hospital, Hyogo, Japan).¹⁶ The HCEs and HKs were maintained at 37°C in 5% CO₂ in a 1:1 mixture of Dulbecco's Modified Eagle's Medium (DMEM) and Nutrient Mixture F-12 (Sigma, Dorset, UK) supplemented with 10% fetal bovine serum (Biosera, East Sussex, UK), 1% L-glutamine, and 1% penicillin and streptavidin (Sigma). Powders of moxifloxacin (Bayer AG, Berlin, Germany) and meropenem (APP pharmaceuticals, Rome, Italy) were weighed and reconstituted into suspensions using DMEM/F12 media. 

Methyl Thiazolyl Tetrazolium Assay

The Cell Titer 96 Cell Proliferation Assay Kit (Promega, Southampton, UK) was used to measure cell viability and toxicity, according to manufacturer's instructions. The HCEs and HKs were seeded at 5000 cells per well in 96-well plates. Once confluence was achieved, cells were incubated at 37°C overnight in DMEM/F12 media without any supplements. After incubation, media was discarded and cells were treated with 5 mg/mL and 2.5 mg/mL of meropenem or moxifloxacin and remained in culture for 1 hour; 15 μL methyl thiazolyl tetrazolium (MTT) dye solution was then added followed by further incubation for 4 hours and 100 μL Stop Solution. The plates were then kept at 4°C overnight in a moistened chamber. Each experiment was performed in triplicate and blank controls (culture medium only), positive controls (cells treated with the cytotoxin dimethyl sulfoxide [DMSO]), and negative controls (cells plus medium) were included. Cellular metabolic activity was determined by reading absorbance at 570 nm with a reference filter of 690 nm, using an automated microplate reader (Bio-Rad, Hemel Hempstead, UK). Cell viability was inferred by expressing the absorbance as percentages in relation to negative controls. Values were expressed as the mean percentages of control values ± SD from three independent experiments. 

LIVE/DEAD Assay

The LIVE/DEAD Viability/Cytotoxicity Kit (Invitrogen, Paisley, UK) was used. Twenty-four–well plates were seeded with 50,000 HKs or HCEs per well. Once confluence was achieved, the cells were incubated overnight in DMEM/F12 media without any supplements. Each experiment included cells treated with meropenem 10 mg/mL for 1 hour, positive controls (cells treated with DMSO), and negative controls (cells plus medium). Three wells were used for each parameter and the experiment was repeated on three separate plates. Following incubation, 100 μL of the LIVE/DEAD assay reagents (Invitrogen, Paisley, UK) containing 2 μM calcein acetomethoxy (AM) and 4 μM ethidium homodimer-1 was added to each well and incubated at 37°C with CO₂ for 30 minutes. Cells were examined with an Axiovert 200 fluorescent microscope (Carl Zeiss, Cambridge, UK) with associated monochrome camera. Green and red filters were used to photograph three images per well, of live and dead cells, respectively. The number of cells was determined using an in-house automated programme written in Matlab R14 (The Mathworks, Inc., Natick, MA, USA). In brief, the original image was preprocessed by an adaptive thresholding technique to improve the contrast and uneven illumination. A selective enhancement filter initially described by Li et al.¹⁷ was used to enhance the cells, in the preprocessed image. Regions of interest (ROIs) were detected by thresholding the enhanced image and further refined by removing smaller objects. The number of cells was determined by labelling the remaining ROIs in the image. 

Corneal and Anterior Chamber Model

Human cadaver corneo-scleral discs were provided by the Manchester eye bank. Consent had been obtained to use the donated eyes for transplantation and research. Two groups of corneas were used to compare drug penetration across the whole cornea or the corneal stroma; in one group the corneas were left intact and in the other the epithelium and endothelium were removed under the dissecting microscope using cellulose eye spears (Beaver-Visitec, Oxfordshire, UK). Following trephination, a central 9-mm disc from each cornea was washed in PBS and mounted in an artificial anterior chamber (PermeGear, Hellertown, PA, USA) containing 5 mL PBS at room temperature (Fig. 1). Fifty microliters of a 10-mg/mL (500 μg) solution of meropenem was placed on the external surface of the cornea. Prior to collecting each sample from the anterior chamber, to ensure even mixing of meropenem in the anterior chamber, fluid was repeatedly aspirated and injected via the sampling port using a needle attached to a 20-mL syringe before sampling. Twenty-microliter samples were collected from the artificial anterior chamber at 45 minutes, 1.5 hours, 3.5 hours, and 24 hours and stored at −20°C before processing. After 24 hours, each cornea was washed with PBS to remove meropenem from the surfaces, homogenized in 2 mL PBS, centrifuged at 10,000g for 5 minutes, and the supernatants transferred into microtubes. Concentrations of meropenem were determined from the aqueous and corneal homogenates using a bioassay and HPLC. Each measurement was performed in triplicate. 

Figure 1

 

PermeGear artificial anterior chamber used in meropenem penetration experiments.

Bioassay

A disc diffusion bioassay was performed to estimate the concentration of meropenem in corneal homogenate and the anterior chamber samples.^18,19 A standard curve was constructed using known concentrations of meropenem (0.01 to 313 μg/mL) as follows. Escherichia coli (ATCC 25922) was seeded on agar plates supplemented with horse blood and incubated for 18 to 20 hours at 37°C in air¹⁹ (http://bsac.org.uk/wp-content/uploads/2012/02/BSAC-disc-susceptibility-testing-method-Jan-2015.pdf). Bacterial colonies were removed from the plates and added to sterile water to a 0.5 McFarland turbidity standard. The suspension was diluted 1 in 100 in sterile water, 10 μL of which was spread evenly onto IsoSensitest agar (ISA) plates (Oxoid, Basingstoke, UK). Blank antimicrobial sensitivity discs were inoculated with 10 μL of different concentrations of meropenem and the plates incubated for 18 to 20 hours in air at 37°C. Following incubation, the diameter of the zone of inhibition (ZOI) surrounding the disk was measured to the nearest millimeter by one observer (Fig. 2).^18,19 Experiments were performed in triplicate for 16 concentrations of meropenem, from 0.01 to 313 μg/mL. A standard curve was constructed using a logarithm transformation of the data. To estimate the meropenem concentration in the anterior chamber fluid and corneal homogenate, 10 μL of the sample in question was placed onto each of three blank discs that were subsequently placed on ISA plates inoculated with E. coli. Plates were incubated for 18 to 20 hours and ZOIs were measured as described above. The mean ZOI from the three measurements was used to determine the meropenem concentration from the standard curve. Positive and negative controls were run in parallel with the test specimen. 

Figure 2

 

Example of meropenem bioassay. Six-millimeter discs of filter paper containing meropenem were placed on an agar plate inoculated with E. coli. The ZOI is the area around the disc devoid of bacterial colonies, the diameter of which is measured to the nearest millimeter. For example, for the disc at top left, the ZOI is 32 mm.

High-Performance Liquid Chromatography

Quantitation of meropenem in experimental samples was carried out using an isocratic reverse-phase HPLC (RP-HPLC) based on a method by Mendez et al.²⁰ Samples were centrifuged at 10,000g for 5 minutes and 20 μL applied to a Phenomenex Luna (Cheshire, UK) 3u C18 100 Å RP-HPLC column (150 × 2.00 mm) equilibrated in 30 mM sodium phosphate, 12% acetonitrile, pH 3.0. Meropenem was eluted isocratically in the same buffer and typically emerged with a 5- to 6-minute residence time. Two HPLC systems were used for the analyses: a Beckman System Gold HPLC system (Beckman Coulter, Fullerton, CA, USA) with elution monitored at 300 nm or an ESA HPLC system (ESA, Inc., Chelmsford, MA, USA) fitted with an auto sampler and electrochemical detector with the electrode set at 600 mV. A standard curve of 0 − 50 μg/L meropenem in Hanks' balanced salt solution (HBSS) was used for calibration in both cases. For the analysis of meropenem metabolites generated during the experiments, a gradient RP-HPLC separation was used. Samples were applied to the same column as above but which was equilibrated in 0.1% trifluoroacetic acid (TFA). Separation was carried out using a 30-minute gradient of 0% to 60% (vol/vol) acetonitrile in 0.1% TFA, using a flow rate of 0.2 mL/min. Elution was monitored at 220 nm. 

Statistical Methods

The Mann-Whitney U test (SPSS version 21; IBM SPSS Statistics, IBM Corporation, Chicago, IL, USA) was used to compare cell viability between the drugs tested. Kruskal-Wallis test, SPSS (version 22) was used to compare HPLC with bioassay and intact to nonintact corneas. A Bonferroni correction was made for multiple statistical tests. 

Results

In Vitro Toxicity of Meropenem and Moxifloxacin

The MTT assays of HCEs and HKs showed that cells exposed to meropenem had significantly higher cellular metabolic activity at both 2.5 mg/mL and 5 mg/mL compared with cells exposed to moxifloxacin (P = 0.029 and P = 0.018, respectively) (Fig. 3). The LIVE/DEAD assay of HCEs showed the viability of cells treated with meropenem was 96%, compared with 95% in the untreated control cells (P = 0.52). 

Figure 3

 

An MTT assay of cultured HKs and HCEs incubated for 1 hour with meropenem (MP) and moxifloxacin (MX) at 2.5 mg/mL and 5 mg/mL. Values are expressed as percentage of control (viability in absence of drug). Results are shown as mean ± SD (n = 6).

Concentration of Meropenem in the Anterior Chamber and Corneal Tissue

The standard curve used for the meropenem bioassay produced a very high coefficient of determination, R² = 0.99 (P < 0.01) (Fig. 4). The concentrations of meropenem in the cornea and artificial anterior are presented in the Table (HPLC and bioassay). The mean anterior chamber concentrations exceeded the MIC₉₀ values of meropenem to S. aureus (0.08 μg/mL), P. aeruginosa (0.25 μg/mL), streptococci (0.054 μg/mL), and to the Enterobacteriaceae (0.094 μg/mL)⁹ at all four time points. The concentration of meropenem in the anterior chamber through all corneas (intact and denuded) increased over all four time points, from 0.75 μg/mL at 45 minutes to 10.94 μg/mL at 24 hours using HPLC, and from 1.09 μg/mL at 45 minutes to 44.89 μg/mL at 24 hours using a bioassay (Fig. 5). Although the anterior chamber concentration increased with time using both HPLC and a bioassay, there was a significant linear decrease in the ratio of HPLC to bioassay concentration over the four time points, that is, from 0.69 at 45 minutes to 0.24 at 24 hours (P = 0.001). Twenty-four hours after the application of one drop (50 μL of 10 mg/mL) of meropenem, approximately 3 to 5 μg was retained within the cornea and approximately 200 μg was in the 5-mL volume of the artificial anterior chamber. 

Figure 4

 

The standard curve used for meropenem bioassay. Escherichia coli (ATCC 25922) was seeded on agar plates supplemented with horse blood and incubated for 18 to 20 hours at 37°C in air. Diameter (mm) of ZOI measured against the logarithm of the concentration (μg/mL) of meropenem.

Table

 

Meropenem Concentration in the Anterior Chamber at Different Time Points and in the Cornea

 

Figure 5

 

Concentration of meropenem in the anterior chamber after the application of 50 μL meropenem (10 mg/mL) onto 10 intact corneas (solid lines) and 18 denuded corneas (dashed lines). Measurements made using bioassay (gray lines) and HPLC (black lines) at 45 minutes, 90 minutes, 210 minutes, and 24 hours. The MIC₉₀ of E. coli (0.094 μg/mL) is indicated on the plot as a dotted black line. Standard deviations and minimum and maximum concentrations are included in the Table.

The mean concentration of meropenem in the anterior chamber over all the four time points measured using HPLC was significantly higher through denuded corneas compared with intact corneas (P = 0.007), with a mean ratio of 0.42 (SD 0.12). When we selectively compared the anterior chamber concentration of meropenem through intact to denuded corneas at each individual time point, however, there was only a significant difference at 90 minutes (P = 0.012) as measured using HPLC (Table). In contrast, the concentration of meropenem in the anterior chamber over all the four time points was similar through intact and denuded corneas using a bioassay (P = 0.082) with a mean ratio of 0.80 (SD 0.23). The concentration of meropenem in the anterior chamber was above the MIC₉₀ of S. aureus, P. aeruginosa, streptococci, and the Enterobacteriaceae for all denuded corneas at 90 minutes (HPLC and bioassay), but for intact corneas it was only above in 80% (HPLC) and 70% (bioassay) of all samples at 210 minutes. After 210 minutes, it was above the MIC₉₀ of all these bacteria for all intact corneas. Of interest, the ratio of the anterior chamber to cornea concentration for intact corneas (1.29, SD 0.27) was higher than denuded corneas (0.43, SD 0.84) measured using HPLC, and very much higher using a bioassay for both intact (23.03, SD 1.37) and denuded (1.16, SD 0.95) corneas. The concentration of meropenem was significantly higher using a bioassay than HPLC (P = 0.004), which may reflect active metabolites of meropenem. Figure 6 demonstrates HPLC meropenem metabolite analysis. The presence of a single peak of meropenem can be seen in Figure 6A, which analyzes a sample containing freshly made meropenem. The presence of other peaks were observed when samples are analyzed from later time points (Figs. 6B, 6C). 

Figure 6

 

Meropenem metabolites during penetration experiments using isocratic RP-HPLC. Aqueous samples from these experiments were analyzed using a gradient RP-HPLC separation with elution monitored at 220 nm. Fresh meropenem (10 mg/mL) (A) and aqueous samples at 3.5 hours (B) and 24 hours (C). Peaks not evident in control samples obtained from using cornea exposed to HBSS alone.

Discussion

Topically applied antimicrobials are the mainstay for the treatment of bacterial keratitis. They are, however, rapidly lost from the ocular surface and therefore the bioavailability of topically applied drugs is typically less than 5%.²¹ To reach adequate tissue concentrations in the cornea, a high drug concentration coupled with frequent application is required. An assessment of corneal cellular toxicity is therefore an important requirement for analyzing the safety of all topically administered drugs. 

In this study, we used both the LIVE/DEAD²² and MTT assays¹⁶ to assess the toxicity of meropenem. The MTT assay is a measure of mitochondrial cell function, and hence is an indirect indicator of cell viability.¹⁶ Saarinen-Savolainen et al.²³ showed that the MTT assay in immortalized corneal cell lines is an accurate and reliable method to assess the toxicity of topically applied drugs to the cornea. Our MTT assay showed that the cellular toxicity of meropenem on HKs and HCEs was negligible and significantly lower than that of moxifloxacin. The LIVE/DEAD assay distinguishes live cells by the presence of intracellular esterase activity, determined by the enzymatic conversion of the nonfluorescent calcein AM to the intensely fluorescent calcein. Using this assay, there were similar low levels of cell toxicity in both meropenem treated and untreated (control) cells. The low level of toxicity, seen in both MTT and LIVE/DEAD assays, parallels the good systemic safety profile of meropenem.^10,11,24 As with other topical delivery studies, we did not assess the effect of meropenem on the endothelium, which would be necessary if the drug were to be given via an intracameral rather than topical route. 

The corneal pharmacokinetics of topically applied antimicrobials is commonly performed in animals such as rabbits.^25,26 There are, however, several structural, physiological, and biochemical differences between the human and the rabbit eye.²⁷ Rabbits have a nictitating membrane, blink infrequently, have a larger corneal surface area, and a thinner cornea, all of which affect the pharmacokinetics of topically applied agents. The method used in this study of investigating the corneal penetration with ex vivo human corneas using an artificial anterior chamber allows testing of human tissue without the constraints of in vivo tests. Although an epithelial defect is almost a universal feature of bacterial keratitis, we assessed the penetration of meropenem both with and without an epithelial layer. In addition, because the stroma is the nidus of infection in keratitis, we measured the penetration of meropenem both into and across the stroma. 

We used both HPLC and a bioassay to analyze meropenem concentrations for this study. Although biological assays are easy and economical to perform, a number of variables can affect the sensitivity.²⁸ Performing the assay in triplicate mitigates some of the intrinsic variability; however, manual measurements of ZOI limit this assay. Chemical assays are sensitive but may not reflect the activity of the compound, particularly in the tissue, which may be affected, for example, by pH and protein binding within the tissue or fluid.^29,30 A further important difference when comparing the data between the two methods relates to their ability to measure active metabolites. The isocratic RP-HPLC method used here to quantify meropenem in multiple samples was designed specifically to detect meropenem and (not necessarily) different chemical species that result from its breakdown products. The extra peaks observed after gradient RP-HPLC analysis with general UV detection of the older aqueous samples (Figs. 5B, 5C), may therefore represent metabolites of meropenem. A bioassay will detect any active substances against the test bacteria, irrespective of their chemical characteristics.³¹ Although there is intrinsic variability in a bioassay, this was reduced in part by using a standard strain of E. coli (ATCC 25922). The significantly higher concentrations of meropenem detected with the bioassay compared with HPLC, particularly at later time points, may reflect the presence of biologically active metabolites after prolonged contact with the corneas. Further studies to investigate the metabolic products of meropenem are indicated. 

An important factor that determines the efficacy of an antimicrobial is its ability to penetrate to the target ocular tissues at concentrations greater than the MIC. The MIC₉₀ (defined as the antimicrobial concentration that would inhibit the growth of 90% of the tested bacterial isolates) is considered to be the minimum tissue concentration of an antimicrobial required for it to be considered efficacious. Sueke et al.⁹ used isolates from patients with keratitis to determine the MICs of meropenem. In our experiments, the concentration of meropenem in the aqueous of the artificial chamber detected by both HPLC and bioassay following application of 50 μL of a 10 mg/mL concentration of meropenem on the cornea, exceeded the MIC₉₀ of S. aureus, P. aeruginosa, streptococci, and the Enterobacteriaceae for all denuded corneas after 45 minutes and after 210 minutes for intact corneas. The ability of an antimicrobial to penetrate the cornea is dependent in part on the physicochemical structure of the drug.²¹ The corneal epithelia possess a significant barrier to the ocular penetration of hydrophilic drugs.^32,33 Meropenem is a hydrophilic compound with a molecular weight of 383.^34,35 The current study shows an increase in meropenem penetration across corneas with their epithelium removed compared with intact corneas. These results concur with Shih and Lee,³⁶ who showed that the penetration of the hydrophilic drug atenolol was impeded by rabbit corneal epithelium compared with the lipophilic drugs timolol and levobunolol. In the setting of bacterial keratitis, however, where the epithelium is often absent, the penetration of a hydrophilic drug may therefore still be effective. 

There are several limitations to this study. The variability in aqueous concentrations of meropenem between corneas may have been due to differences in their anatomic structure, donor factors, and/or duration of storage in the eye bank. Although an artificial chamber is an attractive model, it differs in several aspects from the human eye in vivo. Particularly, a high concentration of meropenem was maintained on the corneal surface in the experimental model, whereas in patients there is likely to be a significant dilution from the tear film and loss via lacrimal drainage, although this would be compensated by the repeated application of drops. Clinically, aqueous turnover would also tend to dilute the concentration of meropenem in the anterior chamber compared with the steady state in our model. In terms of the corneal concentration of meropenem, we measured only the remaining concentration in the cornea after 24 hours. It would be expected, given the increase in aqueous concentration over this period, that the gradient concentration in the cornea was lowest after 24 hours. It would, therefore, be necessary to measure the concentration of meropenem in the cornea at earlier time points to further determine the pharmacokinetics across the cornea. It is of interest that although the concentrations of meropenem were lower in the intact compared with the denuded corneas, the ratio of the anterior chamber to cornea concentration was higher in the intact corneas. Although speculative, this may reflect possible transport of meropenem or its metabolites by the endothelium from the cornea into the anterior chamber. 

In conclusion, in this experimental in vitro model we have demonstrated that meropenem is safe and after topical administration penetrates the cornea, is biologically active, and can achieve levels in the aqueous well above the MIC₉₀ for bacteria, such as S. aureus, P. aeruginosa, streptococci, coagulase-negative staphylococci, and the Enterobacteriaceae.⁹ 

Acknowledgments

Disclosure: H. Sueke, None; S. Kaye, None; M.C. Wilkinson, None; S. Kennedy, None; V. Kearns, None; Y. Zheng, None; P. Roberts, None; S. Tuft, None; T. Neal, None