Enhanced Efficacy of Cefiderocol With PolyTrim Combination Therapy in a Rabbit Model of Extensively Drug-Resistant Pseudomonas aeruginosa Keratitis

Abstract

Purpose

The 2023 Pseudomonas aeruginosa keratitis outbreak linked to contaminated artificial tears underscored the need for new therapies against extensively drug-resistant ocular pathogens. In vitro data suggested additive or synergistic activity between the siderophore-cephalosporin cefiderocol (FDC) and either moxifloxacin (MOX) or polymyxin B sulfate (PB). Here, we tested whether combining FDC with the commercial formulations of MOX (0.5%) or PolyTrim (PT; PB 10,000 U/mL + trimethoprim 0.1%) improved in vivo outcomes.

Methods

New Zealand White rabbits were injected intrastromally with 5000 CFU of Pseudomonas aeruginosa strain CDC1270. After 16 hours, established infections were treated every 30 minutes for 8 hours with FDC, followed 5 minutes later by MOX or PT. Additional groups received monotherapies or saline. Bacterial burden (CFU/cornea) and anterior chamber cultures were determined, and ocular inflammation was assessed.

Results

FDC + MOX reduced bacterial burden by 2.8 log₁₀ CFU, outperforming FDC and MOX monotherapies, although no eyes were sterilized. PT monotherapy produced a significant 3.2 log₁₀ reduction in CFU. The FDC + PT combination yielded the most potent activity, with almost 6.9 log₁₀ reduction and sterilization of 10 of 12 corneas. Notably, anterior chamber invasion occurred in saline and MOX groups but in none of the FDC-containing groups, and perforations were absent in all FDC and PT treatments.

Conclusions

FDC + PT was superior to monotherapies and to FDC + MOX, achieving frequent sterilization and preventing intraocular spread and corneal perforations. These data support PT alone or in combination with FDC as promising therapeutic candidates for the treatment of XDR P. aeruginosa keratitis.

P seudomonas aeruginosa is the leading cause of keratitis among Gram-negative bacteria.^1–3 Though typically treatable with antibiotics, resistant organisms are associated with worse patient outcomes.^4–7 Multi- and extensively drug-resistant (XDR) P. aeruginosa is rare in many regions of the globe but is endemic in countries such as India.^8–10 The recent artificial tears P. aeruginosa keratitis outbreak in the United States was caused by a sequence type 1203 organism that was extensively drug-resistant.^11–14 Strikingly, it was resistant to many first-line keratitis therapies, including several aminoglycoside, cephalosporin, and fluoroquinolone class antibiotics.

This represented a gap in treatment that we have been attempting to address. The XDR P. aeruginosa outbreak strain was found to be susceptible to the novel siderophore-cephalosporin class antibiotic, cefiderocol (FDC).^11,15 FDC is a parenteral antibiotic that was approved by the US Food and Drug Administration (FDA) in 2019 and by the European Medicines Agency (EMA) in 2020 for the treatment of complicated urinary tract infections, hospital-acquired bacterial pneumonia, and ventilator-associated bacterial pneumonia caused by Gram-negative bacteria in adult patients with limited treatment options.¹⁶ This antibiotic uses a Trojan horse mechanism to overcome prokaryotic permeability barriers by mimicking bacterial iron scavenging molecules.¹⁷ FDC on its own was effective in treating experimental P. aeruginosa keratitis caused by the outbreak strain.^15,18 While repurposing FDC for ocular use is promising, it has been reported that bacteria can become resistant to FDC.¹⁹ We therefore sought antibiotic combinations that could improve FDC function and reduce concerns for resistance development. We reported that both moxifloxacin (MOX) and polymyxin B (PB) had additive to synergistic effects when each was separately coupled with FDC in vitro.^20,21

The goals of this study were to test whether the combination of FDC with MOX or PB would improve the function of FDC in vivo using a rabbit keratitis model and clinical formulations of the drugs. To enhance the clinical relevance of the study, the commercially available ophthalmic preparation of polymyxin B, known as PolyTrim (PT), was used. It contains polymyxin B sulfate (10,000 U/mL), trimethoprim sulfate (1 mg/mL), and benzalkonium chloride 0.04 mg/mL. Although PB, trimethoprim, and benzalkonium chloride each possess antibacterial activity, PB is the most active against Gram-negative bacteria at the concentrations supplied.

Previous work with FDC as a monotherapy in our rabbit keratitis model demonstrated that it achieved greater efficacy if the corneal epithelium was removed prior to treatment to both better mimic a corneal ulcer and increase penetration into the stroma, as compared to animal subjects with intact corneal epithelium.^15,18 In this study, a more stringent test was employed in which the dosing intervals were lengthened and the corneal epithelium was left intact to increase the challenge to FDC, thereby allowing clearer assessment of whether combination therapy provided benefits beyond FDC monotherapy.

Methods

Bacterial Strain and Minimum Inhibitory Concentrations

CDC1270 is an XDR P. aeruginosa keratitis isolate from the 2023 outbreak, obtained from the US Centers for Disease Control and Prevention (CDC). Minimum inhibitory concentrations (MICs) were previously determined for the CDC1270 strain for FDC (0.125 µg/mL), MOX (90 µg/mL, which would be considered 128 µg/mL by standard dilution intervals), and PB (3 µg/mL or 4 µg/mL by standard dilution intervals).^15,20

Antibiotics

One-gram vials of the parenteral form of cefiderocol (Fetroja) were kindly provided by Shionogi (Florham Park, NJ, USA) and were refrigerated until use. The 50-mg/mL FDC formulations were freshly prepared in saline, as previously described,^15,18 on the days of the rabbit trials and were kept on ice during dosing. The pH of the FDC 50-mg/mL formulation was measured as pH 5.5 with no buffering.¹⁵ The commercial formulations of moxifloxacin ophthalmic solution 0.5% (Apotex, Weston, FL, USA, or Lupin Pharmaceuticals, Baltimore, MD, USA) were used in the study. The pH of moxifloxacin ophthalmic solution 0.5% is 6.8, as stated on the box. The MOX and FDC solutions (37-µL drops) were instilled using a Rainin EDP electronic pipette set in the multidispense mode. MOX was kept at room temperature during dosing. The 10-mL bottles of commercially available polymyxin B sulfate and trimethoprim ophthalmic solution, USP (10,000 Units/mL polymyxin B/1 mg/mL trimethoprim) (SANDOZ, manufactured by Alcon Laboratories, Fort Worth, TX, USA), were obtained from the inpatient pharmacy at UPMC Presbyterian Hospital. The pH of PT was measured as 6.0. PT drops were instilled using the commercial dropper bottles kept at room temperature during dosing.

Animals

Male and female New Zealand White (NZW) specific pathogen-free (SPF) rabbits weighing 1.1 to 1.4 kg were obtained from Charles River Laboratories Canadian rabbitry. All studies conformed to the ARVO Statement on the Use of Animals in Ophthalmic and Vision Research and were approved by the University of Pittsburgh's Institutional Animal Care and Use Committee (IACUC Protocol #23053154).

Experimental Design

A total of 48 NZW male or female rabbits were used in the experiments. Twenty-four rabbits were used for each cefiderocol combination treatment experiment (FDC + MOX and FDC + PT). The 24 rabbits for each combination therapy experiment were divided into four groups of six rabbits. The experiments were divided into duplicate trials using 12 rabbits each for convenience and to determine reproducibility between trials. The rabbits were acclimated to our facility for 6 days. On the day of the experiment, the rabbits were anesthetized with intramuscular (IM) injections of ketamine (40 mg/kg) and xylazine (4 mg/kg). Their corneas were anesthetized with topical proparacaine 0.5% and intrastromally injected in the central corneal stroma with 25 µL of phosphate-buffered saline containing approximately 5000 CFU of the extensively drug resistant P. aeruginosa (XDRPA) strain CDC1270. The rabbits received analgesia in the form of IM injections of ketoprofen (1.5 mg/kg) prior to recovery from anesthesia.

Sixteen hours postinfection, the rabbits were divided into four groups. The groups varied depending on the combination drug used with cefiderocol. The groups were as follows: (1) combination therapy (cefiderocol 50 mg/mL + MOX or PT), (2) cefiderocol 50 mg/mL alone, (3) combination drug alone (MOX or PT), and (4) saline (negative control). Topical treatment started at this time. The dosing regimen consisted of one drop every 30 minutes for 8 hours (17 total doses). Prior to the final topical treatments, the rabbits’ eyes were examined using a slit lamp and graded using the modified MacDonald–Shadduck scoring system²² and photographed using the camera on the slit lamp.

Following slit-lamp examination, the rabbits were anesthetized with ketamine and xylazine as described above and euthanized with an intravenous overdose of Euthasol solution (390 mg/mL pentobarbital sodium, 50 mg/mL phenytoin sodium). The euthanasia procedure followed the 2020 American Veterinary Medical Association (AVMA) Guidelines for Euthanasia. Prior to harvesting the corneas, the eyes were proptosed, and anterior chamber taps were taken at the limbus using a 23-gauge needle attached to a 1-cc syringe to determine whether there was bacterial spread into the anterior chambers from the corneas. Care was taken not to penetrate the corneas near the areas that were infected. The syringes were kept on ice until processing. Corneal buttons were harvested and processed as previously described.¹⁵ The bacterial burden from the corneas was determined using the EddyJet 2 spiral plating system (Neutec Group, Farmingdale, NY, USA) on trypticase soy agar with 5% sheep's blood plates (Remel, Lenexa, KS, USA). Plates were incubated overnight at 32°C so the colonies would not grow too large to be accurately counted. The following morning, bacterial burden was calculated using the Flash and Grow colony counting system (Neutec Group).

The anterior chamber taps were expelled from the syringes into Eppendorf tubes. Then, 50 µL aqueous humor was added to 50 µL phosphate-buffered saline, and the entire 100 µL was inoculated onto blood agar plates. These plates were also incubated overnight at 32°C. The outcome is positive or negative bacterial spread into the anterior chambers.

Statistical Analysis

The modified MacDonald–Shaddock scores were combined for each parameter to produce a total ocular score for each eye. The median ± interquartile range (IQR) total ocular scores were calculated and analyzed nonparametrically using Kruskal–Wallis ANOVA with Dunn's multiple comparisons test. The corneal colony counts + 1 from each experimental group were log₁₀ transformed and analyzed using Kruskal–Wallis ANOVA with Dunn's multiple comparisons test for differences among the treatment groups. The ratio of eyes with positive XDRPA growth in their anterior chambers was analyzed using Fisher’s exact test. Analysis was performed using GraphPad Prism (GraphPad Software, La Jolla, CA, USA) and Minitab (State College, PA, USA) software.

Results

Evaluation of FDC + MOX for the Treatment of Experimental XDR P. aeruginosa Keratitis

Figure 1A shows a schematic of the experimental approach. Inflammation scores were similar across all groups (median and interquartile ranges for scores for control, FDC, MOX, and FDC + MOX: 10.5 ± 3.8, 11.5 ± 2.8, 12.0 ± 4.8, 13.0 ± 4.8, P = 0.36 Kruskal–Wallis), with most scores stemming from conjunctival inflammation. Figure 1B shows representative images.

Figure 1.

Efficacy of antibacterial treatment combination with FDC and MOX. NZW rabbits were infected with XDR P. aeruginosa. (A) Timeline of the experiment. (B) Representative images of eyes after treatment. Fluorescein was used to observe ulceration. (C) CFU enumeration. The medians with 95% confidence intervals are shown. Asterisks indicate significant differences from the saline group by Kruskal–Wallis with Dunn's posttest. **P < 0.01, ***P < 0.001, ****P < 0.0001.

Saline-treated eyes showed more than a greater than 3 log₁₀ increase in P. aeruginosa CFU over the course of the 24-hour experiment compared to the inoculum (Fig. 1C). FDC monotherapy produced a ∼1.5 log₁₀ reduction in bacterial burden (P < 0.001 vs. saline). MOX monotherapy yielded a nonsignificant 0.3 log₁₀ reduction consistent with a high MOX MIC (90 µg/mL). The FDC + MOX combination therapy group had a 2.8 log₁₀ reduction of the XDR P. aeruginosa (P < 0.0001 vs. saline); however, it did not sterilize (0/12) (Table). Perforations occurred in 1 of 24 saline-treated and 2 of 12 MOX-treated corneas, 0 of 24 in FDC-treated eyes, and 0 of 12 in FDC + MOX-treated eyes. Overall, anterior chamber invasion was detected in 9 of 23 saline-treated eyes across both experiments (with 60–1260 CFU/eye); the aqueous humor from 4 of 10 MOX-treated eyes was positive for P. aeruginosa, whereas no eyes from FDC monotherapy or FDC + MOX had detectable bacteria in the anterior chamber (P < 0.05).

Table.

Sterilization, Aqueous Invasion, and Perforation Outcomes

Treatment Groups	Number of Corneas Sterilized* (P Value vs. Saline)	Number of Eyes With PA in AC † (P Value vs. Saline)	Number of Perforated Eyes
Saline	0/24	9/23	1/24
FDC	0/24 (1.000 NS)	0/24 (0.0006)	0/24
MOX	0/12 (1.000 NS)	4/10‡ (1.00 NS)	2/12
PT	4/12 (0.0084)	1/12 (0.1126 NS)	0/12
FDC + MOX	0/12 (1.000 NS)	0/12 (0.0146)	0/12
FDC + PT	10/12 (<0.0001)	0/12 (0.0146)	0/12

AC, anterior chamber; NS, not significant.

P values determined by Fisher's exact test versus saline group.

^*Below the limit of detection, 20 CFU.

^†Above the limit of detection, 40 CFU.

^‡Aqueous humor was not obtained from subjects with corneal perforation.

Evaluation of FDC + PT for the Treatment of Experimental XDR P. aeruginosa Keratitis

Figure 2 shows results from the combination of FDC + PT using the same experimental protocol as above (Fig. 1A). Clinical inflammation scores were again comparable across groups (median score ± interquartile range: saline 11.0 ± 2.0, FDC 10.0 ± 3.8, PT 11.0 ± 3.0, and FDC + PT 10.0 ± 1.8, P = 0.03 Kruskal–Wallis, no differences by Dunn's multiple comparison test). Figure 2A depicts representative eyes after the last dose.

Figure 2.

Efficacy of antibacterial treatment combination with FDC and PT. NZW rabbits were infected with XDR P. aeruginosa. (A) Representative images of eyes after treatment. Fluorescein was used to observe ulceration over the course of the 24-hour experiment. (B) CFU enumeration. The medians with 95% confidence intervals are shown. Asterisks indicate significant differences from the saline group by Kruskal–Wallis with Dunn's posttest. **P < 0.01, ****P < 0.0001.

FDC monotherapy yielded a median 2.2 log₁₀ reduction, which was not significantly different from saline treatment. PT monotherapy achieved a 3.2 log₁₀ bactericidal reduction (P < 0.01). The FDC + PT combination produced the largest effect, with a >6.9 log₁₀ reduction in colony counts compared with the saline treatment group (P < 0.0001) and sterilization of 10 of 12 eyes (Fig. 2C; Table). The number of eyes sterilized with FDC + PT was significantly greater than all other groups, including PT monotherapy (P < 0.05). No PT-treated eyes perforated. Anterior chamber invasion occurred in 1 of 12 PT-treated eyes and 0 of 12 FDC + PT-treated eyes (both P < 0.05 vs. saline).

Discussion

This study follows through on two in vitro studies showing promising results from the combination of FDC with either MOX or polymyxin B against keratitis isolates of P. aeruginosa, including the XDR outbreak strain. While FDC on its own worked well in rabbits with abraded corneal epithelium to mimic the normal corneal ulceration that occurs during severe bacterial keratitis, it was less effective when the corneal epithelium was left intact.^15,18 In this study, a similar limited effect of FDC alone was observed as the corneal epithelium remained intact, resulting in a more modest reduction of 1.6 to 2.3 log₁₀ CFU. Nevertheless, FDC alone was associated with no eyes (n = 24) with corneal perforation or bacteria in the anterior chamber.

MOX on its own showed minimal activity against the XDR P. aeruginosa strain used, consistent with the high MOX MIC of this strain (90 µg/mL). The modest additional benefit of FDC + MOX over FDC alone did not translate into sterilization or superior clinical metrics in this model, suggesting that more Gram-negative targeted fluoroquinolones like ciprofloxacin may be better partners for future testing.

In contrast to MOX, PT (PB + trimethoprim) demonstrated bactericidal activity as a monotherapy and enhanced activity with FDC (almost 7 log₁₀ reduction), which may be due to PB-mediated outer membrane disruption, further enhancing FDC target access, as has been shown for other antibiotics.²³ Additionally, the FDC + PT treatment group showed no corneal perforations, no bacteria in the aqueous humor, and no tolerability concerns. The surprising antibacterial activity of PT monotherapy is most likely due to the relatively low MIC of the CDC1270 strain (3.0 µg/mL) to PB. Based on Clinical & Laboratory Standards Institute (CLSI) susceptibility breakpoints for PB and P. aeruginosa, the CDC1270 strain has low-level resistance PB (susceptible, ≤2 µg/mL; resistant, ≥4 µg/mL). We have previously stated this as intermediate susceptibility, as the precise measurement is 3.0 µg/mL, but this would be reported to have a MIC of 4.0 µg/mL using the standard doubling dilution system, which is considered resistant. The result of our study demonstrated that the concentration of PB achieved in the cornea with frequent topical dosing exceeded the MIC, effectively overcoming the low-level PB resistance indicated by in vitro testing. We have previously demonstrated that topical antibiotics can successfully treat infecting microbes that are considered resistant by MIC values and CLSI breakpoints in our experimental keratitis model.^24–26 The results of the current study would have to be confirmed by clinical studies in patients. We chose to use PT rather than a compounded formulation of PB eye drops. This choice is clinically relevant because clinicians often must start antimicrobial therapy for bacterial keratitis before the causative organism has been identified. Given its commercial availability and broad-spectrum coverage from the combination of polymyxin B and trimethoprim, PT is more likely to be included in empiric treatment of patients with suspected bacterial keratitis than compounded PB. Additionally, if an XDR outbreak strain is identified, PT is much more readily accessible than compounded PB and thus a more practical clinical alternative. Unfortunately, there were no reports of topical PB monotherapy use in patients with keratitis during the 2023 XDR P. aeruginosa keratitis outbreak. This may have been because the CDC did not report susceptibility of the CDC1270 strain to PB, although it showed intermediate susceptibility to a similar antibiotic, colistin. The use of topical PB (PT) alone or in combination with topical FDC may have provided a positive outcome for patients with keratitis during that outbreak. A further examination of PT and PT with FDC against isolates with higher MIC values to PB is warranted.

Limitations of the study include the short treatment time frame that may underestimate efficacy; the lack of corneal epithelium removal, which may have hindered penetration; and the use of a single outbreak strain. Future work can include longer time frame experiments, epithelial debridement to mimic extensive corneal ulceration, evaluation of resistance development, and more elaborate inflammation analysis. The use of PT introduces additional variables, as benzalkonium chloride and/or trimethoprim may also enhance FDC activity. Indeed, we have demonstrated that benzalkonium chloride can benefit the efficacy of gatifloxacin.²⁷ Therefore, we cannot definitively conclude that the FDC + PB synergy observed in vitro is solely responsible for the excellent in vivo outcomes provided by the combination of FDC + PT observed here. Interexperiment variability in the FDC monotherapy group was also observed, with median 1.5 and 2.2 log₁₀ reductions being measured, which could be due to various differences between rabbits over the course of the several months during which the studies were conducted and/or dose instillation by different researchers.

An additional limitation of the study is the use of only one P. aeruginosa strain. While P. aeruginosa keratitis isolates are genetically and phenotypically heterogeneous, the focus of the current study was on the extensively drug-resistant antibiotic profile of the CDC1270 strain. This strain of sequence type 1203 contained Verona integron-mediated metallo-β-lactamase and the Guiana extended-spectrum β-lactamase that contributed to its extensively drug-resistant antibiotic profile, which had not been formerly seen in the United States.^12,13 The additional use of a genetically dissimilar P. aeruginosa strain with a similar antibiotic resistance profile would have been ideal to demonstrate the efficacy of the combination therapies. However, we do not possess any other extensively drug-resistant or multidrug-resistant P. aeruginosa keratitis strains in our collection. Furthermore, the use of an additional P. aeruginosa strain with a lesser antibiotic resistance profile most likely would not have provided additional information regarding combination therapy since cefiderocol alone was highly effective against an antibiotic-susceptible P. aeruginosa keratitis strain.^15,17

In conclusion, FDC combined with PT markedly improved microbiological outcomes compared to monotherapies and the FDC + MOX pairing. Most notably, it frequently sterilized corneas despite the short time frame and prevented intraocular spread. These findings provide a “proof of concept” that combination therapy with FDC + PT may provide enhanced antimicrobial activity for the treatment of antibiotic-resistant P. aeruginosa keratitis and prioritizes FDC + PT for translational development for the treatment of antibiotic-resistant P. aeruginosa keratitis.