Comparative Toxicity and Concentrations of Intravitreal Amphotericin B Formulations in a Rabbit Model

Abstract

purpose. To determine the toxicity of various doses of intravitreal amphotericin B deoxycholate, amphotericin B lipid complex (ABLC), and liposomal amphotericin B (L-AmB).

methods. Fifty-two rabbits were divided into two treatment groups (groups A and B). Thirteen treatments were administered intravitreally to the 104 rabbit eyes. Treatments included a control plus 10, 20, 30, and 50 μg amphotericin B deoxycholate, ABLC, and L-AmB. Eye examinations were performed before injection and on day 11 for group A and on day 18 for group B. At death, on days 13 and 21 in groups A and B, respectively, vitreous humor was aspirated and concentrations of amphotericin B were determined by high performance liquid chromatography (HPLC), followed by enucleation for histologic studies.

results. Significantly more eyes treated with ABLC showed development of vitreal opacities than developed in eyes treated with amphotericin B deoxycholate or L-AmB (P < 0.05). Vitreal band formation was significantly higher in ABLC-treated eyes than in those treated with L-AmB, (P = 0.039). Vitreal inflammation was greater in eyes treated with L-AmB (75%), amphotericin B deoxycholate (78%), and ABLC (91%) than with the control (50%; P = 0.08). Retinal ganglion cell loss was greater in eyes treated with amphotericin B deoxycholate (81%), L-AmB (91%), and ABLC (97%) than with the control (38%; P = 0.003). Amphotericin B concentrations were measurable for all doses of the three formulations.

conclusions. Based on histologic data, increasing doses of all three agents appear to be associated with increasing toxicity, however based on ophthalmologic data, L-AmB appears to be less toxic than either amphotericin B deoxycholate or ABLC.

Endophthalmitis is a severe, sight-threatening infection of the intraocular fluids or tissues typically caused by bacteria or fungi. ¹ Fungal endophthalmitis is most commonly due to endogenous Candida albicans, although exogenous fungal endophthalmitis due to Fusarium, Acremonium, and Aspergillus sp has been reported. ² Amphotericin B deoxycholate is the treatment of choice for fungal endophthalmitis. ³ Direct intravitreal injection of amphotericin B is used because of the poor intraocular concentrations obtained with intravenous or subconjunctival administration and because of concerns about the nephrotoxicity associated with systemic exposure to amphotericin B. ⁴ Doses of 5 to 10 μg of intravitreal amphotericin B are recommended and generally well tolerated, although doses as low as 1.0 μg have caused marked retinal damage. ⁵  

Recently, lipid-based formulations of amphotericin B have become commercially available. These formulations are significantly less nephrotoxic and can be safely administered systemically at doses up to five times that of amphotericin B deoxycholate. Studies using investigator-prepared, noncommercially available, lipid-based amphotericin administered intravitreally have shown reduced ocular toxicity compared with amphotericin B deoxycholate. ^{6 7 8} However, there are no data on the safety of the commercially available lipid-associated amphotericin B products administered intravitreally. The objective of this study was to determine the toxicity of various intravitreal doses of amphotericin B deoxycholate, amphotericin B lipid complex (ABLC), and liposomal amphotericin B (L-AmB) through ophthalmic and histologic examination. The secondary objective of this study was to compare the amphotericin B concentrations 14 and 21 days after intravitreal injection of various doses of amphotericin B deoxycholate, ABLC, and L-AmB. 

Materials and Methods

Randomization

Sixty healthy New Zealand White rabbits weighing 2 to 3 kg were used. All procedures were performed in accordance with the Animal Care Committee at the University of Illinois at Chicago and the ARVO Statement for the Use of Animals in Ophthalmic and Visual Research. Fifty-two rabbits were randomly assigned to study groups A and B. The rabbits in group A were killed on day 13, and the rabbits in group B were killed on day 21. The 52 eyes in each group were randomly assigned to receive 1 of 13 different intravitreal treatments (four eyes per treatment group). The treatments were amphotericin B deoxycholate, ABLC, and L-AmB in doses of 10, 20, 30, and 50 μg each as well as a control of sterile water for injection (SWFI). Eight rabbits were killed to obtain untreated control vitreous specimens. 

Injections and Enucleation

Commercially available amphotericin B deoxycholate (Novaplus, Irving, TX), ABLC (Abelcet; The Liposome Company, Princeton, NJ), and L-AmB (AmBisome; Fujisawa Heathcare, Deerfield, IL) were obtained from the inpatient pharmacy at the University of Illinois at Chicago Medical Center. The lipid content of ABLC is a 7:3 molar ratio of dimyristoyl phosphatidylcholine and dimyristoyl phosphatidylglycerol. The lipid content of L-AmB is a 10:5:4 molar ratio of hydrogenated soy phosphatidylcholine, cholesterol, and distearoyl phosphatidylglycerol. ABLC was available as a suspension at a concentration of 5 mg/mL. Amphotericin B deoxycholate and L-AmB were available in a powder form and SWFI for a final concentration of 5 mg/mL. The dilutions were made on the morning of injection (day 1). The appropriate concentration of each amphotericin B formulation was added to SWFI to make final doses of 10, 20, 30, and 50 μg. Dilutions of both ABLC and L-AmB were filtered using a 5-μm filter. Eight syringes containing 0.1 mL of treatment were made for each of the 13 different treatments, for a total of 104 syringes. 

On the day of injection, the rabbits were sedated with a 1-mL mixture of ketamine (King Pharmaceutical, Inc., Bristol, TN) and acepromazine (Boehringer Ingelheim Vetmedica, Inc., St. Joseph, MO)—10 mL of a 100 mg/mL concentration of ketamine and 2.5 mL of a 10 mg/mL concentration of acepromazine administered intramuscularly in the hindquarter. Each rabbit also received 2 to 3 drops of a topical anesthetic, 0.5% proparacaine (Alcon Laboratories, Inc., Fort Worth, TX). After sedation, 0.1 mL of one of the 13 treatments was injected into each of the 104 eyes. Treatment was administered with the bevel of the needle positioned upward in the midvitreous of the eyes, slowly and under direct visualization. 

Rabbits were killed on days 13 and 21 in groups A and B, respectively. After death, approximately 0.3 mL of vitreous was withdrawn and immediately frozen at −70°C until analysis for amphotericin B concentrations. The eyes were then enucleated and stored in 15 mL 10% formalin until histologic examination. 

Ophthalmoscopic Studies

All eyes were examined the day before treatment was administered (day 0), on day 11 (group A), and on day 18 (group B). Eyes were dilated with 2 drops of 1% tropicamide 30 minutes before examination. Eyes were examined through indirect ophthalmoscopy for vitreal opacities and vitreal bands. 

Histologic Studies

After enucleation, the eyes were fixed in 10% neutral-buffered formaldehyde. The globes were opened vertically and the calottes processed and embedded in paraffin. Five-micrometer sections of the bisected globe were cut and evaluated for changes by a pathologist blinded to the study regimens. The anterior and posterior segments were evaluated for toxicity in the vitreous, nerve fiber layer, retinal ganglion cell layer, bipolar cell layer, photoreceptor layer, and retinal pigment epithelium. The same pathologist, for evaluator consistency, randomly reevaluated 25 slides. 

Initial examination of sections from approximately half of the eyes, in a random masked fashion, revealed a range of pathologic findings that included loss of retinal ganglion cells, vitreous inflammation, and focal damage to the retinal nerve fiber layer with preservation of the other outer retinal layers. It was then decided to grade the first two findings while evaluating damage to other retinal layers, if present. The vitreous was evaluated for inflammation and graded as follows: 0, indicating absence of inflammation; 0.5, indicating trace inflammation; 1+, indicating mild inflammation; 2+, indicating moderate inflammation; and 3+, indicating severe inflammation. Retinal ganglion cell loss was graded in a masked fashion by examining the superior and inferior retina. Retinal ganglion cell counts in each retinal hemisphere of normal retina were examined and retinal ganglion cell numbers in each microscopic field evaluated. The test retinas (controls and treated) were then graded in a masked fashion. The grades included: 0, indicating no cell loss; 0.5, indicating trace cell loss (<10% retinal ganglion cell loss); 1+, indicating mild cell loss (<20% retinal ganglion cell loss); 2+, indicating moderate cell loss (loss of >40% of retinal ganglion cell); and 3+, indicating severe cell loss (loss of >75% of retinal ganglion cell). The retinal nerve fiber layer was evaluated for necrosis and was graded for the presence or absence of focal necrosis. 

Assay of Amphotericin B in the Vitreous Humor

Amphotericin B concentrations in plasma and tissue were determined using reversed-phase HPLC. The blank vitreous from eight rabbits were used as the control. Vitreous samples were directly extracted with methanol. Vitreous amphotericin B concentrations were determined using minor modifications to a previously published assay. ⁹ Standard curves were linear with r ² > 0.99. The lower limit of quantitation was 0.025 μg/mL. Inter- and intraday variability and precision were less than 7.5%. 

Statistical Analysis

Data were analyzed on computer (SAS, ver. 8.01; SAS Institute Inc., Cary, NC). Presence of ophthalmic and histologic data were compared by nonparametric analyses. Changes in histologic grading scales controlling for dose or medication type were compared using the Cochran-Mantel-Haenszel test. P < 0.05 was considered significant. 

Results

Ophthalmoscopic Examination

At baseline examination (day 0) all eyes were negative for cataracts, vitreal opacities, or vitreal bands. Posterior subcapsular cataracts were observed in 75% of the 104 eyes examined in groups A and B on days 11 and 18, respectively. Cataract formation did not differ between control eyes or eyes injected with the amphotericin B formulations. 

The results of the ophthalmoscopic examinations are shown in Figure 1 . There was no significant difference between the eyes examined on days 11 and 18. Thus, these data were combined, resulting in eight eyes per dose per formulation and control. Vitreal opacity (Fig. 1a) or band formation (Fig. 1b) was not evident in any eyes in the control group or the L-AmB group at all doses. In the amphotericin B deoxycholate treatment group, opacities and bands were evident at the 30- and 50-μg doses. In the ABLC treatment group, vitreal opacities and/or bands were evident at all doses. Significantly more eyes treated with ABLC showed development of vitreal opacities versus those treated with amphotericin B deoxycholate (P = 0.039) or L-AmB (P < 0.0001). Significantly more eyes treated with amphotericin B deoxycholate showed vitreal opacities versus those treated with L-AmB (P = 0.039). Significantly more eyes in the ABLC group had vitreal bands compared with eyes treated with L-AmB (P = 0.039). No other differences were observed in band formation. 

Histologic Examination

Histologic examination in some of the enucleated specimens revealed various degrees of vitreous inflammation, loss of retinal ganglion cells, and focal necrosis of the nerve fiber layer. There were no pathologic changes observed in the outer retinal layers or the retinal pigment epithelium. 

The results of the histologic changes are summarized in Figures 2 and 3 . There was no significant difference between the eyes enucleated at days 13 versus 21; thus, these data were combined, resulting in eight eyes per dose per formulation and control. Figure 2 depicts the number of eyes that exhibited vitreal inflammation (Fig. 2a) , retinal ganglion cell loss (Fig. 2b) , and focal necrosis of the nerve fiber layer (Fig. 2c) . Figure 3 depicts the median grade for those eyes that had vitreal inflammation (Fig. 3a) and retinal ganglion cell loss (Fig. 3b) . By histologic examination, the vitreous inflammation was mainly seen in the anterior vitreous, and the inflammatory infiltrate was primarily composed of lymphocytes and macrophages. Overall, less vitreal inflammation was observed in eyes treated with control (50%) than with ABLC (91%), amphotericin B deoxycholate (78%), or L-AmB (75%; P = 0.08). Median grades for vitreal inflammation for the 10-, 20-, and 30-μg doses of each formulation were trace to mild (0.5–1.25) with higher median grades at 50 μg (1–2). In general, increasing doses of any formulation was associated with an increasing median grade of vitreal inflammation (P < 0.001). 

Retinal ganglion cell loss was observed in most of the eyes treated with an amphotericin B formulation, regardless of dose (Fig. 4) . Loss of retinal ganglion cells was noted, in both the peripheral and central retina. Overall, the average retinal ganglion cell loss was higher in the ABLC and L-AmB group versus the amphotericin B deoxycholate group. Retinal ganglion cell loss was present in more eyes treated with ABLC (97%), L-AmB (91%), or deoxycholate (81%) than in those treated with the control (38%; P = 0.003). Most retinal ganglion cell degeneration was found to be mild to moderate (0.75–2.0) with the exception of the 50-μg dose ABLC (median score, 2.75). Overall, increasing doses were associated with increased retinal ganglion cell damage (P < 0.001). 

By histology, focal axonal swelling, condensation, and vacuolization suggestive of necrosis of the nerve fiber layer was present mainly in the in the peripheral retina and was associated with focal inflammatory changes. Focal necrosis of the nerve fiber layer was rare and did not significantly differ between treatment groups. Focal necrosis was observed in two eyes each in the amphotericin B deoxycholate and L-AmB treatment groups and in four eyes in the ABLC treatment group (Figs. 5 6) . None of the eyes in the control group showed focal necrosis of the nerve fiber layer. 

The median grade of vitreal inflammation and retinal ganglion cell loss for each dose of each formulation is described in Figures 3a and 3b , respectively. Because it is well documented in the literature that intravitreal doses of amphotericin B deoxycholate greater than 10 μg cause toxicity in the eye, ¹⁰ a subset analysis was performed to compare the median grade of vitreal inflammation and retinal ganglion cell loss for each dose of each amphotericin B formulation with 10 μg amphotericin B deoxycholate. There was no statistically significant difference between amphotericin B deoxycholate 10 μg and any of the other doses of the other formulations based on an ANOVA analysis with Tukey subanalysis. 

Amphotericin B Concentrations

The mean concentrations determined by HPLC for each of the treatment regimens are listed in Table 1 . Concentrations were measurable at days 13 and 21 but were highly variable. 

Discussion

Amphotericin B deoxycholate is the treatment of choice for fungal endophthalmitis, based mainly on case reports and anecdotal evidence. Intravitreal administration of amphotericin B has been shown to be the most effective route. ¹¹ Intravitreal doses of 5 and 10 μg of amphotericin B have been used since the 1970s based primarily on rabbit studies. ¹⁰  

The reduced toxicity observed with systemic administration of lipid-based amphotericin products sparked interest in the use of intravitreal lipid-based amphotericin B for fungal endophthalmitis. An early toxicity study with a noncommercially available lipid-based amphotericin product, containing a 3:5:1 molar ratio of egg phosphatidylcholine, cholesterol, and tocopherol succinate, suggested that liposomal intercalation may reduce the toxicity of intravitreal amphotericin B. ⁷ A rabbit model was used to determine the ocular toxicity of a homemade liposome-intercalated amphotericin B formulation compared with amphotericin B deoxycholate. Forty eyes were randomly assigned to receive intravitreal doses of amphotericin B deoxycholate or homemade liposomal amphotericin B of up to 20 μg or a control. Vitreal band formation was evident, by direct ophthalmoscopy, only in the eyes treated with amphotericin B deoxycholate. Vitreal bodies were evident in the majority of eyes given liposomal amphotericin B or empty liposomes potentially signifying a toxic reaction or residual liposomes. However, significantly more eyes treated with amphotericin B deoxycholate showed retinal damage on histologic examination. In another study, a rabbit model was used to compare the safety and efficacy of up to 40 μg of a homemade lipid-based amphotericin B formulation containing a 4:1:3 molar ratio of phosphatidylcholine, phosphatidylglycerol, and cholesterol with amphotericin B deoxycholate in experimentally induced C. albicans endophthalmitis. ⁶ Similar to the earlier study, it was concluded that reduced toxicity occurred at higher doses of liposome-bound amphotericin than amphotericin B deoxycholate. However, higher doses of the lipid formulation were associated with decreased efficacy in this study. 

Our study was performed to determine the toxicity of intravitreal administration of two commercially available lipid-based amphotericin B formulations. Although cataract formation was observed in the majority (75%) of eyes, this was determined to be a result of injection technique, because cataract formation did not differ in control and amphotericin B–treated eyes. Cataract formation has been documented in other studies at doses of amphotericin B deoxycholate as low as 1.0 μg. ⁵ Vitreal opacities or bands were not observed in any control or L-AmB–treated eyes compared with 13% and 31% of the eyes injected with amphotericin B deoxycholate or ABLC, respectively. Development of opacity was not observed with L-AmB, whereas it was observed starting at the 10-μg dose in eyes treated with ABLC and starting at 30 μg for amphotericin B deoxycholate. Band formation was observed in 9% of the amphotericin B deoxycholate-treated eyes versus 13% of the ABLC-treated eyes. Similar to opacities, band formation was observed starting at the 10-μg dose of ABLC and the 30-μg dose of amphotericin B deoxycholate. 

On histologic examination, our results were similar to that of other studies which found vitreal inflammation and retinal necrosis or atrophy. ^{5 7 11} Vitreal inflammation was observed in more than two thirds of the eyes treated with either amphotericin B deoxycholate, ABLC, or L-AmB. The severity of vitreal inflammation was found to be mild regardless of the formulation used; however, median grade increased with increasing doses. When all concentrations of each formulation were compared with the 10-μg dose of amphotericin deoxycholate, statistical significance was not found. The location of the chronic inflammatory response suggests that this mild inflammation may be related to increased drug concentration close to the injection site. An increase of the median grade of inflammation with increasing drug concentration suggests that this response may be related to the drug vehicle and/or liposome or the drug itself. Similar vitritis has been described after intravitreal amphotericin B injections. ^{5 7 11} Nevertheless, the inflammatory response was mild, with no evidence of acute inflammation, necrosis, or abscess, and is an acceptable risk if efficacy is enhanced as ascertained by future studies. 

Retinal ganglion cell loss was found to be similar between the various treatment groups (81%–97%). In general, cell loss was noted to be mild to moderate with severity increasing with increasing doses. When all concentrations of each formulation were compared with the 10-μg dose of amphotericin B deoxycholate, statistical significance in retinal ganglion cell loss was not found. Development of limited retinal ganglion cell loss is most likely clinically insignificant when the devastation of fungal endophthalmitis is taken into account. Also in the central nervous system, neuronal degeneration has been described after intrathecal administration of amphotericin B. ¹² Of note, retinal ganglion cell loss was observed in three of the eight control eyes studied. The degree of retinal ganglion cell loss was very mild (<0.5) and could be within the error margin of the semiquantitative method of retinal ganglion cell evaluation used in this study. 

Focal necrosis of the nerve fiber layer was relatively rare and was mainly in the peripheral retina and thought to be a result of technique rather than toxicity of the various formulations. Similar reports of focal retinal degeneration have been described after intravitreal injection of amphotericin B. ^{5 10}  

Although highly variable, concentrations of amphotericin B were measurable up to day 21 with all formulations. Because of the variability, specific concentrations could not be correlated with toxicity. 

A limitation of our study was that the model did not differentiate the reasons for the differences between formulations in regard to toxicity. These differences could be due to the amount of free drug that is released in the vitreous, the type of vehicle that is used for each of the formulations, or a combination of both of these factors. This model also did not assess the efficacy of these various doses and formulations. Further studies should be performed to examine these questions. 

Based on histologic examination, it appears that the number of toxic changes increases as the dose increases of each formulation. However, based on ophthalmologic data, L-AmB appears to be less toxic than either amphotericin B deoxycholate or ABLC. Whether the higher doses of the lipid-based formulations offer some reduction in toxicity in conjunction with enhanced efficacy has yet to be determined. Further studies evaluating the efficacy of the higher doses of these formulations should be performed. 

 Preliminary analysis of these data was presented at the 40th Annual Meeting of the Interscience Conference on Antimicrobial Agents and Chemotherapy (ICAAC), Toronto, Ontario, Canada, September 17–20, 2000.

 Supported by National Eye Institute Core Grant EY001792.

 Submitted for publication October 4, 2002; revised December 12, 2002; accepted December 22, 2002.

 Disclosure: J.P. Cannon, Elan Pharmaceuticals, Inc. (R); R. Fiscella, None; S. Pattharachayakul, None; K.W. Garey, None; F. De Alba, None; S. Piscitelli, None; D.P. Edward, None; L.H. Danziger Elan Pharmaceuticals, Inc. (R)

 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.

 

Ophthalmologic examination. The number of eyes at each dose of each formulation and the control group with vitreal opacities (a) and vitreal bands (b) present on ophthalmologic examination.

Figure 2.

 

Histologic examination. The number of eyes at each dose of each formulation and the control group with vitreal inflammation (a), retinal ganglion cell loss (b), and focal necrosis of the nerve fiber layer present on histologic examination.

Figure 3.

 

Histologic examination, median grade. The median grade of vitreal inflammation (a) and retinal ganglion cell loss (b) for each dose of each formulation and the control group.

Figure 4.

 

Micrographs of rabbit retinal sections. Top: normal rabbit retina treated with vehicle treated control with intact inner and outer retinal layers. Middle: moderate (2+) retinal ganglion cells loss, mild nerve fiber layer atrophy was present. The remaining retinal layers were unremarkable. Artifactitious changes were visible in the outer segments (30 μg ABLC). Bottom: Severe loss of retinal ganglions cells (3+) with preserved outer retinal layers (50 μg ABLC). Hematoxylin and eosin; original magnification ×25.

Figure 5.

 

Microphotograph showing axonal swelling (arrows), vacuolization (✶), condensation, and loss of retinal ganglion cells suggestive of retinal nerve fiber layer necrosis (50 μg ABLC). Hematoxylin and eosin; original magnification, ×25.

Figure 6.

 

Top: Microphotograph showing mild (a trace to 1+) vitreous inflammation with a lymphocytic infiltrate (10 μg ABLC). Bottom: Severe localized vitreous inflammation (3+) with underlying loss of retinal ganglion cells (50 μg ABLC). Hematoxylin and eosin; original magnification, ×25.

Table 1.

 

Intravitreal Concentrations of Amphotericin B

Treatment	Day	Dose (μg)
				10	20	30	50
Amphotericin B	13	0.545 ± 0.56	2.733 ± 0.94	1.021 ± 0.529	3.191 ± 3.50
Deoxycholate	21	0.189 ± 0.19	2.252 ± 3.20	1.334 ± 1.86	0.770 ± 0.48
ABLC	13	0.392 ± 0.54	2.632 ± 4.91	0.128 ± 0.09	2.812 ± 4.58
	21	1.057 ± 0.80	1.431 ± 0.98	1.669 ± 2.71	0.539 ± 0.81
L-AmB	13	0.057 ± 0.06	0.817 ± 0.70	2.538 ± 3.46	4.426 ± 5.24
	21	0.264 ± 0.48	0.284 ± 0.10	2.663 ± 1.96	2.225 ± 2.58