A Comparative Treatment Study of Intravitreal Voriconazole and Liposomal Amphotericin B in an Aspergillus fumigatus Endophthalmitis Model

Abstract

Purpose: To compare the effects of voriconazole (VCZ) and liposomal amphotericin B (Amp-B) in an experimental model of Aspergillus fumigatus endophthalmitis.

Methods: Thirty guinea pigs received an intravitreal injection of A. fumigatus to induce endophthalmitis. The animals were randomly divided into three groups, including control (0.02 mL balanced salt solution intravitreal injection) and experimental (20 μg VCZ/0.02 mL or 20 μg liposomal Amp-B/0.02 mL intravitreal injection) groups. Corneal opacity, aqueous flare, and vitreous opacity were graded, and electroretinographic examinations were performed at multiple time points. At 28 days post treatment, histopathology was performed to examine the retinal architecture.

Results: The inflammation in the VCZ and liposomal Amp-B groups was milder than that in the control group. Corneal opacity, aqueous flare, and vitreous opacity scores, as well as electroretinographic recording, showed significantly less inflammation in the VCZ group compared with the liposomal Amp-B group during the early and middle stages of endophthalmitis (P < 0.05). Normal histologic structure of the retina was observed in eyes treated with VCZ and liposomal Amp-B.

Conclusions: Both intravitreal VCZ and liposomal Amp-B were effective treatments for A. fumigatus-induced endophthalmitis in guinea pigs. Voriconazole was superior to liposomal Amp-B at doses similar to the initial therapy for acute infections. Further experimental and clinical studies are required to confirm the efficacy of these two antifungal drugs.

Chinese Abstract

Fungal endophthalmitis is an ocular infection associated  with a difficult diagnosis, few therapeutic options, and an unfavorable prognosis. The major pathogens are Blastomyces dermatitidis, Aspergillus spp., and Candida spp.¹ Endophthalmitis can be exogenous or endogenous depending upon the origin of infection. The predisposing factors for exogenous fungal endophthalmitis are trauma and intraocular surgery. Endogenous fungal endophthalmitis is associated with broad-spectrum antibiotic and prolonged corticosteroid therapies.² 

Voriconazole (VCZ) is a second-generation, synthetic triazole with broad-spectrum antifungal activity against many medically important pathogens, including Aspergillus spp., Candida spp., and Fusarium spp.^3–5 Through the inhibition of cytochrome P-450-mediated 14a-demethylation, VCZ destroys the cell wall, resulting in cell lysis.⁶ In a study by Pfaller et al.,⁷ VCZ exhibited promising in vitro activity against 239 clinical isolates of Aspergillus spp. Several studies have concluded that intravitreal VCZ could be a treatment strategy for fungal endophthalmitis.^8–12 

Amphotericin B (Amp-B) is the treatment of choice for fungal endophthalmitis. Amphotericin B increases fungal membrane permeability and cell death by binding ergosterol.¹³ Liposomal Amp-B, consisting of Amp-B and phospholipid, was developed to decrease the side effects of Amp-B, including nephrotoxic effects and infusion-related reactions.¹⁴ Intravitreal administration of liposomal Amp-B is one of the most effective treatments for fungal endophthalmitis. High intraocular concentrations can be administered due to the low ocular toxity.^15–17 

However, the efficacy of intravitreal VCZ versus liposomal Amp-B for fungal endophthalmitis is unknown. In this study, we investigated a guinea pig model to determine the effects of VCZ compared to liposomal Amp-B for the treatment of exogenous Aspergillus fumigatus endophthalmitis. 

Materials and Methods

In Vitro Studies

The M27-A3 microdilution method of the Clinical Laboratory Standards Institute was performed to test the antifungal susceptibility of VCZ and liposomal Amp-B. Checkerboard analysis was performed using 10-fold dilutions of both drugs (ranging from 16–0.0313 μg/mL) in 96-well microplates. The minimum inhibitory concentration (MIC) values were recorded following a 96-hour incubation. 

Animal Model

Guinea pigs (0.4–0.45 kg in weight) were obtained from the Animal Center of the College of Basic Sciences, Jilin University. The use of these animals adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. A strain of A. fumigatus (JLMR054) that was isolated from a patient with fungal endophthalmitis was donated by the Fungal Department of the College of Basic Sciences, Jilin University. Using a 30-gauge needle and a 1.0-mL plastic syringe, 0.02 mL Aspergillus suspension (1.0 × 10⁶ CFU/mL) was injected directly into the vitreous cavity of 30 guinea pigs at the pars plana, approximately 1.5 mm posterior to the limbus. Voriconazole and liposomal Amp-B were supplied in powder form by Livzon Pharmaceutical Group, Inc. (Zhuhai, China) and New Pioneer, Inc. (Shanghai, China), respectively. The powder was suspended in 100% dimethylsulfoxide prior to use. The animals were randomly divided into three groups. Group A (control group) received an intravitreally administered balanced salt solution (BSS) 24 hours after Aspergillus injection. In group B, 20 μg/0.02 mL VCZ was intravitreally injected 24 hours after Aspergillus inoculation. The same injection was repeated every 24 hours, for a total of three injections. In group C, 20 μg/0.02 mL liposomal Amp-B was intravitreally injected 24 hours after Aspergillus inoculation. The same injection was repeated every 24 hours, for a total of three injections. All procedures were performed in the right eye following adequate anesthesia with an intraperitoneal injection of 10% chloral hydrate solution. Levofloxacin 0.5% ophthalmic solution (Santen Co., Osaka, Japan) was applied to the ocular surface 3 days prior to the injection, and ofloxacin ophthalmic ointment was applied following the injection to prevent infections. Mydriasis was achieved with phenylephrine hydrochloride 0.5% ophthalmic solution and tropicamide 0.5% ophthalmic solution (Santen Co.). 

In Vivo Studies

On days 1, 3, 5, 7, 10, 14, and 28 post treatment, the severity of inflammation was evaluated by an observer blinded to the treatment groups with a method similar to the one used by Yang et al.¹⁸ The cornea, anterior chamber, iris, and vitreous were examined via slit-lamp microscopy and indirect ophthalmoscopy. Corneal opacity, aqueous flare, and vitreous opacity were graded on a scale from 0 to 4, as listed in Table 1. 

Table 1

 

Clinical Inflammation Grading Scale

 

Smear and Culture

On days 1, 7, 14, and 28 after drug injection, approximately 0.2 mL vitreous fluid was aspirated from one randomly chosen eye in each group. Cultures and smears for fungi were performed with these collected samples. 

Histopathology

On day 28, two guinea pigs in each group were chosen randomly for histopathologic analyses. The whole eye was placed in fixation solution followed by 10% formalin and embedded in paraffin. The sections were stained with hematoxylin-eosin and were evaluated for histopathologic changes of the intraocular structure. 

Electroretinogram Recordings

Flash electroretinography (ERG) was performed to monitor the functional changes within the retina in vivo. Two guinea pigs from each group were adapted to the dark for at least 30 minutes prior to scotopic ERGs, and all of the procedures were performed under dim red light. The animals' pupils were dilated after complete anesthetization as described above. For the ERG recordings, electrodes were placed on the center of the cornea. The reference and ground electrodes were attached to the forehead and ear-back region, respectively. Data from both eyes were acquired automatically using an Electrophysiological Test Unit (RETIport, Roland Consult, Brandenburg, Germany). The amplitudes and b-wave latencies were averaged from 10 stimulus presentations. 

Statistical Analysis

The Statistical Package for the Social Sciences, version 13.0 (Chicago, IL, USA), was applied for data processing. The Kruskal-Wallis test (P ≤ 0.05) was used to evaluate the therapeutic effects, and the Mann-Whitney test (P ≤ 0.05) was used to compare the results between each group. 

Results

In Vitro Findings

The antifungal susceptibility testing of VCZ and liposomal Amp-B was repeated three times. The MIC of VCZ against A. fumigatus was 0.125 μg/mL, and liposomal Amp-B demonstrated effective inhibition of A. fumigatus growth at 0.25 μg/mL. These results are consistent with the results previously reported by the Clinical and Laboratory Standards Institute (CLSI) M27-A3. 

Clinical Observation

The grades of corneal opacity, anterior aqueous flare, and vitreous opacity in the three treatment groups at selected intervals are presented in Figure 1. Inflammation in the control group progressed throughout the 28 days following A. fumigatus inoculation. During the first 3 days, there was a mild inflammatory reaction in the control group. Two weeks later, the inflammatory response was exacerbated, and we observed neovascularization in the cornea and iris as well as dense anterior chamber exudates (Fig. 2A) or hypopyon (Fig. 2B), pupil seclusion, lens opacification, and condensations in the vitreous cavity (Fig. 2C). Clinical deterioration was observed 4 weeks later (two cases of ocular atrophy with panendophthalmitis [Fig. 2D] and one case of corneal perforation). 

Figure 1

 

Clinical grading of corneal opacity (A), aqueous flare (B), and vitreous opacity (C) in the different groups at selected intervals.

Figure 2

 

Clinical observations (control group). (A) Anterior chamber exudates; (B) corneal neovascularization and hypopyon; (C) vitreous condensation; (D) panendophthalmitis.

However, the inflammatory reactions in both the VCZ- and liposomal Amp-B–injected groups gradually subsided during the observation period. In the VCZ group, we observed a mild inflammatory reaction in all affected eyes with grade 0 to 1 corneal opacity, grade 0 to 1 aqueous flare, and grade 0 to 2 vitreous opacity. Two eyes presented with corneal opacity of grade 2 on day 5, and one eye presented with vitreous opacity of grade 3 on day 10. In the liposomal Amp-B group, we observed corneal opacity at grades 0 to 2, aqueous flare grades at 0 to 2, and vitreous opacity grades at 0 to 2 (except that one eye presented with corneal opacity of grade 3 on day 10; one eye showed aqueous flare of grade 4 on day 14; and one eye had a vitreous opacity of grade 3 on day 5). At the end of the observation period (day 28), all of the eyes in the two groups demonstrated reduced corneal edema, the absence of an anterior chamber reaction, and a clear vitreous. 

The clinical inflammation scores and statistical results for corneal opacity, aqueous flare, and vitreous opacity at selected intervals are presented in Tables 2, 3, and 4, respectively. The VCZ group demonstrated significantly less inflammation in both the anterior segment and vitreous cavity than the control group throughout the observation period (except at day 1). Simultaneously, the liposomal Amp-B group demonstrated significant differences in corneal opacity and aqueous flare from day 3 to day 28 compared with the control group. The liposomal Amp-B group demonstrated significant improvements in vitreous opacity beginning on day 5 following treatment. In contrast, the VCZ group demonstrated significantly less anterior chamber inflammation than the liposomal Amp-B–injected group on days 7 and 10. After 2 weeks, the inflammation in the anterior segment between the two groups was not significantly different. Regarding the vitreous opacity, the VCZ group experienced less vitreous inflammation than the liposomal Amp-B group on days 5 and 7 (P < 0.05). After this time point, no significant difference in vitreous opacity was observed between the groups. 

Table 2

 

Corneal Opacity at Selected Intervals

 

Table 3

 

Aqueous Flare at Selected Intervals

 

Table 4

 

Vitreous Opacity at Selected Intervals

 

Microbiologic Analysis

Filamentous hyphae and fruiting bodies were detected in vitreous sample smears from the three groups on days 1 and 7, respectively. On days 14 and 28, massive hyphae were observed in the vitreous specimens, which displayed white, cheesy condensation. The fungal cultures were considered positive in the control group (Fig. 3). Although positive hyphae were observed in smears from the VCZ and liposomal Amp-B injection groups on day 7, the number of hyphae elements tended to be lower. On days 14 and 28, the VCZ- and liposomal Amp-B-treated groups were not culture positive and did not have hyphae. 

Figure 3

 

Photographs of Aspergillus hyphae. (A) Hyphae with periodic acid-Schiff staining with obvious branches and diaphragms. (B) Dense gray-black fungi in the culture dish.

Histopathologic Examination

At 28 days, two eyes were randomly selected from each group for histopathologic examination. In the eyes injected with BSS, the retinal tissue of each layer was severely disorganized and could not be identified (Fig. 4A). The vitreous cavity was infiltrated by inflammatory cells, and we discovered neovascularization in the corneal stroma. In contrast, in the eyes injected with VCZ (Fig. 4B) and liposomal Amp-B (Fig. 4C), no obvious abnormalities in the retinal architecture were detected. The vitreous cavity was clear, and the corneal tissue appeared normal. 

Figure 4

 

Histopathologic examination of eyes 28 days following intravitreal injection of BSS (control), VCZ, and liposomal Amp-B. (A) Control group: The retinal tissue of each layer could not be identified, and the vitreous cavity was infiltrated by inflammatory cells. (B, C) The VCZ- and liposomal Amp-B–treated groups exhibited explicit retinal structures with fewer inflammatory cells.

Electroretinography

Scotopic ERG was performed to examine whether the retinal inflammation induced by A. fumigatus infection led to functional changes. To analyze rod function, both the amplitudes and b-wave latencies were recorded. The control group demonstrated an obvious reduction in b-wave amplitudes from day 1 to day 28. However, the amplitudes in the VCZ injection group and the liposomal Amp-B injection group (Fig. 5B) did not differ. Compared with the VCZ group, the b-wave amplitude was significantly decreased from day 3. A decreasing trend was observed over the following 28 days in the control group (P < 0.05). In contrast, we observed an obvious difference between the control group and the liposomal Amp-B group from day 7 (P < 0.05). On days 3, 7, and 14, the mean amplitude of the liposomal Amp-B group was significantly lower than that of the VCZ group (P < 0.05). Four weeks later, there was no significant difference between the two groups (Table 5). 

Figure 5

 

Scotopic b-wave latency (A) and amplitude (B). Each data point indicates the average of 10 stimuli. Zero represents 1 day before treatment.

Table 5

 

B-wave Amplitudes at Selected Intervals

 

To characterize rod photoreceptor function more directly, b-wave latencies were also measured (Fig. 5A). An obvious delay in latency was observed during the disease course in the control group compared with the two inflammation-controlled groups. The latencies between the control group and the VCZ group were significantly different on days 3, 7, 14, and 28. The latencies between the control group and the liposomal Amp-B group were significantly different on days 7, 14, and 28. Compared with the VCZ group, the liposomal Amp-B group displayed a significant delay in latency on days 3, 7, and 14. After 28 days, the two groups were similar (Table 6). 

Table 6

 

B-wave Latencies at Selected Intervals

 

Discussion

We performed a randomized, comparative study to evaluate the effects of VCZ and liposomal Amp-B in an experimental endophthalmitis model. A. fumigatus is the most common fungal pathogen in endophthalmitis and causes a rapid, progressive infection. In this trial, fungal endophthalmitis in guinea pigs was induced via an intravitreal injection of 1.0 × 10⁶ CFU/mL A. fumigatus. Three days following incubation with A. fumigatus, all eyes demonstrated slight to mild diffuse inflammatory reactions. The inflammation was exacerbated on day 5 (gradual enlargement of corneal edema, anterior purulent exudation, and vitreous haze). Ten days later, we observed corneal neovascularization and anterior chamber hypopyon. Clinical deterioration was observed 4 weeks later (two cases of ocular atrophy with panendophthalmitis). The retinal histopathologic findings suggested severe inflammation with destruction of the retinal architecture. The clinical observations, histopathologic studies, positive cultures, and smears of vitreous specimens demonstrated that the A. fumigatus endophthalmitis was successfully modeled. 

Several studies indicate that intravitreal injection of Amp-B or VCZ effectively treats fungal endophthalmitis.^8,15–17 This study compared the in vivo antifungal effects of VCZ and liposomal Amp-B against A. fumigatus in exogenous endophthalmitis in guinea pig eyes. 

In a guinea pig model of invasive A. fumigatus, both VCZ and liposomal Amp-B demonstrated antifungal effects. No significant differences were detected between the VCZ group and the liposomal Amp-B group regarding clinical degradation and ERG results on days 1 and 3. The differences in corneal opacity (P < 0.05 on days 7 and 10), aqueous flare (P < 0.05 on days 7, 10, and 14), and vitreous opacity (P < 0.05 on days 5 and 7), as well as ERG recordings of b-wave amplitudes (P < 0.05 on days 3, 7, and 14) and latency (P < 0.05 on day 14), were significantly different between the two groups. This result indicates that liposomal Amp-B was less effective than VCZ at similar doses. According to the clinical grading and ERG examinations, treatment with VCZ during the early and middle stages led to better responses than liposomal Amp-B. These results are consistent with a previously published comparison of VCZ and Amp-B for the treatment of invasive Aspergillus.¹⁹ We acknowledge that the histopathology and electrophysiology were analyzed in only two animals from each group. Thus, the results could be more conclusive if the histopathology and electrophysiology were examined in a larger number of animals. There were no significant differences in the late stage. In this study, VCZ and liposomal Amp-B did not cause retinal toxicity (ERG or histopathologic studies) when the intravitreal dose was 20 μg/0.02 mL. Our results suggested that VCZ and liposomal Amp-B were equally effective during the late stage of exogenous A. fumigatus infection with better outcomes and no recurrence. In contrast, initial therapy with VCZ was superior to liposomal Amp-B. 

The initial inferiority of liposomal Amp-B could be explained by the membrane-stabilizing effect of cholesterol. Liposomal Amp-B consists of Amp-B and unilamellar liposomes containing cholesterol.²⁰ When the fungal cells attach to liposomal Amp-B, as an active antifungal ingredient, Amp-B will be released from the liposome.²¹ Amphotercercin B increases the membrane permeability and leakage of cellular components. These actions cause fungal cell death through ergosterol binding, which is an integral component of the fungal cell wall.¹³ Liposomal Amp-B may have a delayed efficacy due to the cholesterol contained within the liposomes.¹⁵ Alternatively, this difference could be partially due to the contact between liposomal Amp-B and its target. Intravitreal administration of liposomal Amp-B away from the lesion could result in low drug availability, which might contribute to the decreased effects of liposomal Amp-B for acute infections. In contrast, VCZ inhibits ergosterol biosynthesis in the fungal cell wall and subsequent cell death through direct inhibition of lanosterol demethylation.⁴ Additionally, the MIC values of the two drugs were different. Therefore, VCZ was superior to liposomal Amp-B for the treatment of acute A. fumigatus infection. 

In summary, both intravitreal VCZ and liposomal Amp-B were effective antifungal agents for exogenous A. fumigatus endophthalmitis. However, VCZ was superior to liposomal Amp-B at a similar dose (as an initial therapy) for acute infections. Further experimental and clinical studies are required to confirm the efficacy of these two antifungal drugs. 

Acknowledgments

We thank all members of the Pathology Laboratory of the College of Basic Sciences at Jilin University for their support. We also thank Hong Zhang for his excellent assistance with the statistical analysis. 

Supported by the National Natural Science Foundation of China (Grant No. 31071222) and the Frontier Interdiscipline Program of Norman Bethune Health Center of Jilin University (Grant No. 2013106023). 

Disclosure: J. Zhao, None; Y. Cheng, None; X. Song, None; C. Wang, None; G. Su, None; Z. Liu, None